EXTERNAL REVIEW DRAFT - DO NOT CITE OR QUOTE EPA/63 5/R-09/011A
www.epa.gov/iris
vvEPA
TOXICOLOGICAL REVIEW
OF
Trichloroethylene
(CAS No. 79-01-6)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
October 2009
NOTICE
This document is an EXTERNAL REVIEW draft. This information is distributed solely for
the purpose of pre-dissemination peer review under applicable information quality guidelines. It
has not been formally disseminated by U.S. EPA. It does not represent and should not be
construed to represent any Agency determination or policy. It is being circulated for review of
its technical accuracy and science policy implications.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document is a preliminary draft for review purposes only. This information is
distributed solely for the purpose of pre-dissemination peer review under applicable information
quality guidelines. It has not been formally disseminated by EPA. It does not represent and
should not be construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 ii DRAFT—DO NOT CITE OR QUOTE
-------
GUIDE TO READERS OF THIS DOCUMENT
Due to the length of the TCE toxicological review, it is recommended that
Chapters 1 and 6 be read prior to Chapters 2-5.
Chapter 1 is the standard introduction to an IRIS Toxicological Review, describing the
purpose of the assessment and the guidelines used in its development.
Chapter 2 is an exposure characterization that summarizes information about TCE
sources, releases, media levels and exposure pathways for the general population (occupational
exposure is also discussed to a lesser extent).
Chapter 3 describes the toxicokinetics and physiologically based pharmacokinetic
(PBPK) modeling of TCE and metabolites (PBPK modeling details are in Appendix A).
Chapter 4 is the hazard characterization of TCE. Section 4.1 summarizes the evaluation
of epidemiologic studies of cancer and TCE (qualitative details in Appendix B; meta-analyses in
Appendix C). Each of the Sections 4.2-4.9 provides self-contained summary and syntheses of
the epidemiologic and laboratory studies on TCE and metabolites, organized by tissue/type of
effects, in the following order: genetic toxicity, central nervous system (CNS), kidney, liver,
immune system, respiratory tract, reproduction and development, and other cancers. Additional
details are provided in Appendix D for CNS effects and Appendix E for liver effects.
Section 4.10 summarizes the available data on susceptible lifestages and populations.
Section 4.11 describes the overall hazard characterization, including the weight of evidence for
noncancer effects and for carcinogenicity.
Chapter 5 is the dose-response assessment of TCE. Section 5.1 describes the dose-
response analyses for noncancer effects, and Section 5.2 describes the dose-response analyses for
cancer. Additional computational details are described in Appendix F for noncancer dose-
response analyses, Appendix G for cancer dose-response analyses based on rodent bioassays, and
Appendix H for cancer dose-response analyses based on human epidemiologic data.
Chapter 6 is the summary of the major conclusions in the characterization of TCE hazard
and dose response.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 iii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS of TOXICOLOGICAL REVIEW for TRICHLOROETHYLENE
(CAS No. 79-01-6)
LIST OF TABLES xv
LIST OF FIGURES xxvii
LIST OF ABBREVIATIONS AND ACRONYMS xxx
FOREWORD xxxvii
AUTHORS, CONTRIBUTORS, AND REVIEWERS xxxviii
ACKNOWLEDGMENTS xlii
EXECUTIVE SUMMARY xliii
1. INTRODUCTION 1-1
2. EXPOSURE CHARACTERIZATION 2-1
2.1. ENVIRONMENTAL SOURCES 2-2
2.2. ENVIRONMENTAL FATE 2-6
2.2.1. Fate in Terrestrial Environments 2-6
2.2.2. Fate in the Atmosphere 2-6
2.2.3. Fate in Aquatic Environments 2-7
2.3. EXPOSURE CONCENTRATIONS 2-7
2.3.1. Outdoor Air—Measured Levels 2-7
2.3.2. Outdoor Air—Modeled Levels 2-10
2.3.3. Indoor Air 2-11
2.3.4. Water 2-13
2.3.5. Other Media 2-15
2.3.6. Biological Monitoring 2-16
2.4. EXPOSURE PATHWAYS AND LEVELS 2-17
2.4.1. General Population 2-17
2.4.1.1. Inhalation 2-17
2.4.1.2. Ingestion 2-18
2.4.1.3. Dermal 2-20
2.4.1.4. Exposure to TCE Related Compounds 2-21
2.4.2. Potentially Highly Exposed Populations 2-22
2.4.2.1. Occupational Exposure 2-22
2.4.2.2. Consumer Exposure 2-23
2.4.3. Exposure Standards 2-24
2.5. EXPOSURE SUMMARY 2-24
3. TOXICOKINETICS 3-1
3.1. ABSORPTION 3-2
3.1.1. Oral 3-2
3.1.2. Inhalation 3-4
3.1.3. Dermal 3-11
3.2. DISTRIBUTION AND BODY BURDEN 3-11
3.3. METABOLISM 3-19
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 iv DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
3.3.1. Introduction 3-19
3.3.2. Extent of Metabolism 3-20
3.3.3. Pathways of Metabolism 3-23
3.3.3.1. Cytochrome P450-Dependent Oxidation 3-23
3.3.3.2. Glutathione (GSH) Conjugation Pathway 3-40
3.3.3.3. Relative Roles of the Cytochrome P450 (CYP) and
Glutathione (GSH) Pathways 3-54
3.4. TRICHLOROETHYLENE (TCE) EXCRETION 3-57
3.4.1. Exhaled Air 3-57
3.4.2. Urine 3-59
3.4.3. Feces 3-61
3.5. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODELING
OF TRICHLOROETHYLENE (TCE) AND ITS METABOLITES 3-62
3.5.1. Introduction 3-62
3.5.2. Previous Physiologically Based Pharmacokinetic (PBPK) Modeling of
Trichloroethylene (TCE) for Risk Assessment Application 3-62
3.5.3. Development and Evaluation of an Interim "Harmonized"
Trichloroethylene (TCE) Physiologically Based Pharmacokinetic
(PBPK) Model 3-64
3.5.4. Physiologically Based Pharmacokinetic (PBPK) Model for
Trichloroethylene (TCE) and Metabolites Used for This Assessment 3-67
3.5.4.1. Introduction 3-67
3.5.4.2. Updated Physiologically Based Pharmacokinetic (PBPK)
Model Structure 3-67
3.5.4.3. Specification of Physiologically Based Pharmacokinetic
(PBPK) Model Parameter Prior Distributions 3-68
3.5.4.4. Dose Metric Predictions 3-72
3.5.5. Bayesian Estimation of Physiologically Based Pharmacokinetic
(PBPK) Model Parameters, and Their Uncertainty and Variability 3-72
3.5.5.1. Updated Pharmacokinetic Database 3-72
3.5.5.2. Updated Hierarchical Population Statistical Model 3-81
3.5.5.3. Use of Interspecies Scaling to Update Prior Distributions in
the Absence of Other Data 3-82
3.5.5.4. Implementation 3-84
3.5.6. Evaluation of Updated Physiologically Based Pharmacokinetic (PBPK)
Model 3-85
3.5.6.1. Convergence 3-85
3.5.6.2. Evaluation of Posterior Parameter Distributions 3-87
3.5.6.3. Comparison of Model Predictions With Data 3-96
3.5.6.4. Summary Evaluation of Updated Physiologically Based
Pharmacokinetic (PBPK) Model 3-112
3.5.7. Physiologically Based Pharmacokinetic (PBPK) Model Dose Metric
Predictions 3-113
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 v DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
3.5.7.1. Characterization of Uncertainty and Variability 3-113
3.5.7.2. Implications for the Population Pharmacokinetics of
Trichloroethylene (TCE) 3-128
3.5.7.3. Overall Evaluation of Physiologically Based
Pharmacokinetic (PBPK) Model-Based Internal Dose
Predictions 3-135
4. HAZARD CHARACTERIZATION 4-1
4.1. EPIDEMIOLOGIC STUDIES ON CANCER AND TRICHLOROETHYLENE
(TCE)—METHODOLOGICAL OVERVIEW 4-1
4.2. GENETIC TOXICITY 4-29
4.2.1. Trichloroethylene (TCE) 4-30
4.2.1.1. DNA Binding Studies 4-30
4.2.1.2. Bacterial Systems—Gene Mutations 4-32
4.2.1.3. Fungal and Yeast Systems—Gene Mutations, Conversions
and Recombination 4-35
4.2.1.4. Mammalian Systems Including Human Studies 4-37
4.2.1.5. Summary 4-49
4.2.2. Trichloroacetic Acid (TCA) 4-50
4.2.2.1. Bacterial Systems—Gene Mutations 4-50
4.2.2.2. Mammalian Systems 4-52
4.2.2.3. Summary 4-56
4.2.3. Dichloroacetic Acid (DCA) 4-57
4.2.3.1. Bacterial and Fungal Systems—Gene Mutations 4-57
4.2.3.2. Mammalian Systems 4-61
4.2.3.3. Summary 4-62
4.2.4. Chloral Hydrate 4-63
4.2.4.1. DNA Binding Studies 4-63
4.2.4.2. Bacterial and Fungal Systems—Gene Mutations 4-70
4.2.4.3. Mammalian Systems 4-71
4.2.4.4. Summary 4-74
4.2.5. Dichlorovinyl Cysteine (DCVC) and S-Dichlorovinyl Glutathione
(DCVG) 4-74
4.2.6. Trichloroethanol (TCOH) 4-79
4.2.7. Synthesis and Overall Summary 4-80
4.3. CENTRAL NERVOUS SYSTEM (CNS) TOXICITY 4-84
4.3.1. Alterations in Nerve Conduction 4-85
4.3.1.1. Trigeminal Nerve Function: Human Studies 4-85
4.3.1.2. Nerve Conduction Velocity—Human Studies 4-90
4.3.1.3. Trigeminal Nerve Function: Laboratory Animal Studies 4-90
4.3.1.4. Discussion and Conclusions: Trichloroethylene (TCE)-
Induced Trigeminal Nerve Impairment 4-91
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 vi DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
4.3.2. Auditory Effects 4-93
4.3.2.1. Auditory Function: Human Studies 4-93
4.3.2.2. Auditory Function: Laboratory Animal Studies 4-95
4.3.2.3. Summary and Conclusion of Auditory Effects 4-99
4.3.3. VestibularFunction 4-101
4.3.3.1. Vestibular Function: Human Studies 4-101
4.3.3.2. Vestibular Function: Laboratory Animal Data 4-101
4.3.3.3. Summary and Conclusions for the Vestibular Function
Studies 4-102
4.3.4. Visual Effects 4-103
4.3.4.1. Visual Effects: Human Studies 4-103
4.3.4.2. Visual Effects: Laboratory Animal Data 4-105
4.3.4.3. Summary and Conclusion of Visual Effects 4-107
4.3.5. Cognitive Function 4-108
4.3.5.1. Cognitive Effects: Human Studies 4-108
4.3.5.2. Cognitive Effects: Laboratory Animal Studies 4-110
4.3.5.3. Summary and Conclusions of Cognitive Function Studies 4-112
4.3.6. PsychomotorEffects 4-113
4.3.6.1. Psychomotor Effects: Human Studies 4-113
4.3.6.2. Psychomotor Effects: Laboratory Animal Data 4-116
4.3.6.3. Summary and Conclusions for Psychomotor Effects 4-121
4.3.7. Mood Effects and Sleep Disorders 4-122
4.3.7.1. Effects onMood: Human Studies 4-122
4.3.7.2. Effects on Mood: Laboratory Animal Findings 4-122
4.3.7.3. Sleep Disturbances 4-123
4.3.8. Developmental Neurotoxicity 4-123
4.3.8.1. Human Studies 4-123
4.3.8.2. Animal Studies 4-124
4.3.8.3. Summary and Conclusions for the Developmental
Neurotoxicity Studies 4-128
4.3.9. Mechanistic Studies of Trichloroethylene (TCE) Neurotoxicity 4-128
4.3.9.1. Dopamine Neuron Disruption 4-128
4.3.9.2. Neurochemical and Molecular Changes 4-130
4.3.10. Potential Mechanisms for Trichloroethylene (TCE)-Mediated
Neurotoxicity 4-134
4.3.11. Overall Summary and Conclusions—Weight of Evidence 4-137
4.4. KIDNEY TOXICITY AND CANCER 4-141
4.4.1. Human Studies of Kidney 4-141
4.4.1.1. Nonspecific Markers of Nephrotoxicity 4-141
4.4.1.2. End-Stage Renal Disease 4-147
4.4.2. Human Studies of Kidney Cancer 4-147
4.4.2.1. Studies of Job Titles and Occupations with Historical
Trichloroethylene (TCE) Usage 4-148
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 vii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
4.4.2.2. Cohort and Case-Controls Studies of Trichloroethylene
(TCE) Exposure 4-159
4.4.2.3. Examination of Possible Confounding Factors 4-163
4.4.2.4. Susceptible Populations—Kidney Cancer and
Trichloroethylene (TCE) Exposure 4-166
4.4.2.5. Meta-Analysis for Kidney Cancer 4-167
4.4.3. Human Studies of Somatic Mutation of von Hippel-Lindau (VHL)
Gene 4-172
4.4.4. Kidney Noncancer Toxicity in Laboratory Animals 4-177
4.4.5. Kidney Cancer in Laboratory Animals 4-184
4.4.5.1. Inhalation Studies of Trichloroethylene (TCE) 4-184
4.4.5.2. Gavage and Drinking Water Studies of Trichloroethylene
(TCE) 4-185
4.4.5.3. Conclusions: Kidney Cancer in Laboratory Animals 4-187
4.4.6. Role of Metabolism in Trichloroethylene (TCE) Kidney Toxicity 4-188
4.4.6.1. In Vivo Studies of the Kidney Toxicity of Trichloroethylene
(TCE) Metabolites 4-188
4.4.6.2. In Vitro Studies of Kidney Toxicity of Trichloroethylene
(TCE) and Metabolites 4-196
4.4.6.3. Conclusions as to the Active Agents of Trichloroethylene
(TCE)-Induced Nephrotoxicity 4-197
4.4.7. Mode(s) of Action for Kidney Carcinogenicity 4-198
4.4.7.1. Hypothesized Mode of Action: Mutagenicity 4-198
4.4.7.2. Hypothesized Mode of Action: Cytotoxicity and
Regenerative Proliferation 4-202
4.4.7.3. Additional Hypothesized Modes of Action with Limited
Evidence or Inadequate Experimental Support 4-204
4.4.7.4. Conclusions About the Hypothesized Modes of Action 4-206
4.4.8. Summary: Trichloroethylene (TCE) Kidney Toxicity, Carcinogenicity,
and Mode-of-Action 4-208
4.5. LIVER TOXICITY AND CANCER 4-210
4.5.1. Liver Noncancer Toxicity in Humans 4-210
4.5.2. Liver Cancer in Humans 4-217
4.5.3. Experimental Studies of Trichloroethylene (TCE) in Rodents—
Introduction 4-231
4.5.4. Trichloroethylene (TCE)-Induced Liver Noncancer Effects 4-233
4.5.4.1. Liver Weight 4-234
4.5.4.2. Cytotoxicity 4-238
4.5.4.3. Measures of DNA Synthesis, Cellular Proliferation, and
Apoptosis 4-245
4.5.4.4. Peroxisomal Proliferation and Related Effects 4-248
4.5.4.5. Oxidative Stress 4-250
4.5.4.6. Bile Production 4-251
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 viii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
4.5.4.7. Summary: Trichloroethylene (TCE)-Induced Noncancer
Effects in Laboratory Animals 4-253
4.5.5. Trichloroethylene (TCE)-Induced Liver Cancer in Laboratory Animals.... 4-254
4.5.5.1. Negative or Inconclusive Studies of Mice and Rats 4-254
4.5.5.2. Positive Trichloroethylene (TCE) Studies of Mice 4-262
4.5.5.3. Summary: Trichloroethylene (TCE)-Induced Cancer in
Laboratory Animals 4-264
4.5.6. Role of Metabolism in Liver Toxicity and Cancer 4-264
4.5.6.1. Pharmacokinetics of Chloral Hydrate (CH), Trichloroacetic
Acid (TCA), and Dichloroacetic Acid (DCA) From
Trichloroethylene (TCE) Exposure 4-265
4.5.6.2. Comparisons Between Trichloroethylene (TCE) and
Trichloroacetic Acid (TCA), Dichloroacetic Acid (DCA),
and Chloral Hydrate (CH) Noncancer Effects 4-265
4.5.6.3. Comparisons of Trichloroethylene (TCE)-Induced
Carcinogenic Responses With Trichloroacetic Acid (TCA),
Dichloroacetic Acid (DCA), and Chloral Hydrate (CH)
Studies 4-282
4.5.6.4. Conclusions Regarding the Role of Trichloroacetic Acid
(TCA), Dichloroacetic Acid (DCA), and Chloral Hydrate
(CH) in Trichloroethylene (TCE)-Induced Effects in the
Liver 4-307
4.5.7. Mode of Action (MOA) for Trichloroethylene (TCE) Liver
Carcinogenicity 4-308
4.5.7.1. Mutagenicity 4-308
4.5.7.2. Peroxisome Proliferator Activated Receptor Alpha (PPARa)
Receptor Activation 4-310
4.5.7.3. Additional Proposed Hypotheses and Key Events with
Limited Evidence or Inadequate Experimental Support 4-316
4.5.7.4. Mode of Action (MOA) Conclusions 4-325
4.6. IMMUNOTOXICITY AND CANCERS OF THE IMMUNE SYSTEM 4-331
4.6.1. Human Studies 4-331
4.6.1.1. Noncancer Immune-Related Effects 4-331
4.6.1.2. Cancers of the Immune System, Including Childhood
Leukemia 4-343
4.6.2. Animal Studies 4-373
4.6.2.1. Immunosuppression 4-373
4.6.2.2. Hypersensitivity 4-381
4.6.2.3. Autoimmunity 4-384
4.6.2.4. Cancers of the Immune System 4-397
4.6.3. Summary 4-400
4.6.3.1. Noncancer Effects 4-400
4.6.3.2. Cancer 4-401
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 ix DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
4.7. RESPIRATORY TRACT TOXICITY AND CANCER 4-403
4.7.1. Epidemiologic Evidence 4-403
4.7.1.1. Chronic Effects: Inhalation 4-403
4.7.1.2. Cancer 4-403
4.7.2. Laboratory Animal Studies 4-415
4.7.2.1. Respiratory Tract Animal Toxicity 4-415
4.7.2.2. Respiratory Tract Cancer 4-423
4.7.3. Role of Metabolism in Pulmonary Toxicity 4-427
4.7.4. Mode of Action for Pulmonary Carcinogenicity 4-432
4.7.4.1. Mutagenicity via Oxidative Metabolism 4-432
4.7.4.2. Cytotoxicity Leading to Increased Cell Proliferation 4-434
4.7.4.3. Additional Hypothesized Modes of Action with Limited
Evidence or Inadequate Experimental Support 4-43 5
4.7.4.4. Conclusions About the Hypothesized Modes of Action 4-436
4.7.5. Summary and Conclusions 4-438
4.8. REPRODUCTIVE AND DEVELOPMENTAL TOXICITY 4-440
4.8.1. Reproductive Toxicity 4-440
4.8.1.1. Human Reproductive Outcome Data 4-440
4.8.1.2. Animal Reproductive Toxicity Studies 4-444
4.8.1.3. Discussion/Synthesis of noncancer reproductive toxicity
findings 4-460
4.8.2. Cancers of the Reproductive System 4-466
4.8.2.1. Human Data 4-467
4.8.2.2. Animal studies 4-480
4.8.2.3. Mode of Action for Testicular Tumors 4-482
4.8.3. Developmental Toxicity 4-483
4.8.3.1. Human Developmental Data 4-483
4.8.3.2. Animal Developmental Toxicology Studies 4-506
4.8.3.3. Discussion/Synthesis of Developmental Data 4-530
4.9. OTHER SITE-SPECIFIC CANCERS 4-547
4.9.1. Esophageal Cancer 4-547
4.9.2. Bladder Cancer 4-556
4.9.3. Central Nervous System and Brain Cancers 4-562
4.10. SUSCEPTIBLE LIFESTAGES AND POPULATIONS 4-563
4.10.1.Lifestages 4-567
4.10.1.1. Early Lifestages 4-568
4.10.1.2. Later Lifestages 4-577
4.10.2. Other Susceptibility Factors 4-578
4.10.2.1. Gender 4-578
4.10.2.2. Genetic Variability 4-583
4.10.2.3. Race/Ethnicity 4-585
4.10.2.4. Pre-Existing Health Status 4-585
4.10.2.5. Lifestyle Factors and Nutrition Status 4-586
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 x DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
4.10.3. Uncertainty of Database for Susceptible Populations 4-589
4.11. HAZARD CHARACTERIZATION 4-589
4.11.1. Characterization of Noncancer Effects 4-589
4.11.1.1. Neurotoxicity 4-589
4.11.1.2. Kidney Toxicity 4-594
4.11.1.3. Liver Toxicity 4-595
4.11.1.4. Immunotoxicity 4-597
4.11.1.5. Respiratory Tract Toxicity 4-598
4.11.1.6. Reproductive Toxicity 4-599
4.11.1.7. Developmental Toxicity 4-600
4.11.2. Characterization of Carcinogenicity 4-604
4.11.2.1. Summary Evaluation of Epidemiologic Evidence of
Trichloroethylene (TCE) and Cancer 4-604
4.11.2.2. Summary of Evidence for Trichloroethylene (TCE)
Carcinogenicity in Rodents 4-612
4.11.2.3. Summary of Additional Evidence on Biological Plausibility 4-614
4.11.3. Characterization of Factors Impacting Susceptibility 4-620
5. DOSE-RESPONSE ASSESSMENT 5-1
5.1. DOSE-RESPONSE ANALYSES FOR NONCANCER ENDPOINTS 5-1
5.1.1. Modeling Approaches and Uncertainty Factors for Developing
Candidate Reference Values Based on Applied Dose 5-3
5.1.2. Candidate Critical Effects by Effect Domain 5-7
5.1.2.1. Candidate Critical Neurological Effects on the Basis of
Applied Dose 5-7
5.1.2.2. Candidate Critical Kidney Effects on the Basis of Applied
Dose 5-11
5.1.2.3. Candidate Critical Liver Effects on the Basis of Applied
Dose 5-15
5.1.2.4. Candidate Critical Body Weight Effects on the Basis of
Applied Dose 5-16
5.1.2.5. Candidate Critical Immunological Effects on the Basis of
Applied Dose 5-16
5.1.2.6. Candidate Critical Respiratory Tract Effects on the Basis of
Applied Dose 5-19
5.1.2.7. Candidate Critical Reproductive Effects on the Basis of
Applied Dose 5-19
5.1.2.8. Candidate Critical Developmental Effects on the Basis of
Applied Dose 5-25
5.1.2.9. Summary of cRfCs, cRfDs, and Candidate Critical Effects 5-30
5.1.3. Application of Physiologically Based Pharmacokinetic (PBPK) Model
to Inter- and Intraspecies Extrapolation for Candidate Critical Effects 5-33
5.1.3.1. Selection of Dose Metrics for Different Endpoints 5-33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xi DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
5.1.3.2. Methods for Inter- and Intraspecies Extrapolation Using
Internal Doses 5-45
5.1.3.3. Results and Discussion of p-RfCs and p-RfDs for Candidate
Critical Effects 5-62
5.1.4. Uncertainties in cRfCs and cRfDs 5-63
5.1.4.1. Qualitative Uncertainties 5-63
5.1.4.2. Quantitative Uncertainty Analysis of Physiologically Based
Pharmacokinetic (PBPK) Model-Based Dose Metrics for
Lowest-Observed-Adverse-Effect Level (LOAEL) or No-
Ob served-Adverse-Effect Level (NOAEL)-Based Point of
Departures (PODs) 5-66
5.1.5. Summary of Noncancer Reference Values 5-76
5.1.5.1. Preferred Candidate Reference Values (cRfCs, cRfD, p-
cRfCs and p-cRfDs) for Candidate Critical Effects 5-76
5.1.5.2. Reference Concentration 5-82
5.1.5.3. Reference Dose 5-85
5.2. DOSE-RESPONSE ANALYSIS FOR CANCER ENDPOINTS 5-88
5.2.1. Dose-Response Analyses: Rodent Bioassays 5-88
5.2.1.1. Rodent Dose-Response Analyses: Studies and Modeling
Approaches 5-88
5.2.1.2. Rodent Dose-Response Analyses: Dosimetry 5-96
5.2.1.3. Rodent Dose-Response Analyses: Results 5-109
5.2.1.4. Uncertainties in Dose-Response Analyses of Rodent
Bioassays 5-119
5.2.2. Dose-Response Analyses: Human Epidemiologic Data 5-130
5.2.2.1. Inhalation Unit Risk Estimate for Renal Cell Carcinoma
Derived from Charbotel et al. (2006) Data 5-130
5.2.2.2. Adjustment of the Inhalation Unit Risk Estimate for
Multiple Sites 5-137
5.2.2.3. Route-to-Route Extrapolation Using Physiologically Based
Pharmacokinetic (PBPK) Model 5-141
5.2.3. Summary of Unit Risk Estimates 5-144
5.2.3.1. Inhalation Unit Risk Estimate 5-144
5.2.3.2. Oral Unit Risk Estimate 5-146
5.2.3.3. Application of Age-Dependent Adjustment Factors 5-147
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND
DOSE RESPONSE 6-1
6.1. HUMAN HAZARD POTENTIAL 6-1
6.1.1. Exposure 6-1
6.1.2. Toxicokinetics and Physiologically-Based Pharmacokinetic (PBPK)
Modeling 6-2
6.1.3. Noncancer Toxicity 6-3
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
6.1.3.1. Neurological Effects 6-4
6.1.3.2. Kidney Effects 6-5
6.1.3.3. Liver Effects 6-6
6.1.3.4. Immunological Effects 6-7
6.1.3.5. Respiratory Tract Effects 6-8
6.1.3.6. Reproductive Effects 6-8
6.1.3.7. Developmental Effects 6-9
6.1.4. Carcinogenicity 6-11
6.1.5. Susceptibility 6-17
6.2. DOSE-RESPONSE ASSESSMENT 6-18
6.2.1. Noncancer Effects 6-18
6.2.1.1. Background and Methods 6-18
6.2.1.2. Uncertainties and Application of Uncertainty Factors (UFs) 6-19
6.2.1.3. Candidate Critical Effects and Reference Values 6-23
6.2.1.4. Noncancer Reference Values 6-28
6.2.2. Cancer 6-31
6.2.2.1. Background and Methods 6-31
6.2.2.2. Inhalation Unit Risk Estimate 6-32
6.2.2.3. Oral Unit Risk Estimate 6-34
6.2.2.4. Uncertainties in Cancer Dose-Response Assessment 6-35
6.2.2.5. Application of Age-Dependent Adjustment Factors 6-40
6.3. OVERALL CHARACTERIZATION OF TCE HAZARD AND DOSE
RESPONSE 6-41
REFERENCES R-l
APPENDIX A: PBPK MODELING OF TCE AND METABOLITES-DETAILED
METHODS AND RESULTS A-l
APPENDIX B: SYSTEMATIC REVIEW OF EPIDEMIOLOGIC STUDIES ON
CANCER AND TRICHLOROETHYLENE (TCE) EXPO SURE B-1
APPENDIX C: MET A-ANALYSIS OF CANCER RESULTS FROM
EPIDEMIOLOGICAL STUDIES C-l
APPENDIX D: NEUROLOGICAL EFFECTS OF TRICHLOROETHYLENE D-l
APPENDIX E: ANALYSIS OF LIVER AND COEXPOSURE ISSUES FOR
THE TCE TOXICOLOGICAL REVIEW E-l
APPENDIX F: TCE NONCANCER DOSE-RESPONSE ANALYSES F-l
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xiii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
APPENDIX G: TCE CANCER DOSE-RESPONSE ANALYSES WITH RODENT
CANCER BIOASSAY DATA G-l
APPENDIX H: LIFETABLE ANALYSIS AND WEIGHTED LINEAR REGRESSION
BASED ON RESULTS FROM CHARBOTEL ET AL H-l
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xiv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
2-1. TCE metabolites and related parent compounds 2-1
2-2. Chemical properties of TCE 2-2
2-3. Properties and uses of TCE related compounds 2-3
2-4. Toxics Release Inventory (TRI) releases of TCE 2-4
2-5. Concentrations of trichloroethylene in ambient air 2-8
2-6. TCE ambient air monitoring data 2-9
2-7. Mean TCE air levels across monitors by land setting and use (1985 to 1998) 2-9
2-8. Concentrations of trichloroethylene in water based onpre-1990 studies 2-13
2-9. Levels in food 2-16
2-10. TCE levels in whole blood by population percentile 2-17
2-11. Modeled 1999 annual exposure concentrations for trichloroethylene 2-18
2-12. Preliminary estimates of TCE intake from food ingestion 2-20
2-13. Preliminary intake estimates of TCE and TCE-related chemicals 2-21
2-14. Years of solvent use in industrial degreasing and cleaning operations 2-23
2-15. TCE standards 2-24
3-1. Blood:air PC values for humans 3-5
3-2. Blood:airPC values for rats and mice 3-6
3-3. Air and blood concentrations during exposure to TCE in humans 3-7
3-4. Retention of inhaled TCE vapor in humans 3-8
3-5. Uptake of TCE in human volunteers following 4 hour exposure to 70 ppm 3-8
3-6. Concentrations of TCE in maternal and fetal blood at birth 3-13
3-7. Distribution of TCE to rat tissues following inhalation exposure 3-14
3-8. Tissue:blood partition coefficient values for TCE 3-16
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
3-9. Age-dependence of tissue:air partition coefficients in rats 3-17
3-10. Predicted maximal concentrations of TCE in rat blood following a 6-hour
inhalation exposure 3-17
3-11. Tissue distribution of TCE metabolites following inhalation exposure 3-18
3-12. Binding of 14C from [14C]TCE in rat liver and kidney at 72 hours after oral
administration of 200 mg/kg [14C]TCE 3-19
3-13. In vitro TCE oxidative metabolism in hepatocytes and microsomal fractions 3-26
3-14. In vitro kinetics of trichloroethanol and trichloroacetic acid formation from
chloral hydrate in rat, mouse, and human liver homogenates 3-29
3-15. In vitro kinetics of DCA metabolism in hepatic cytosol of mice, rats, and
humans 3-31
3-16. TCOH and TCA formed from CH in vitro in lysed whole blood of rats and
mice or fractionated blood of humans 3-33
3-17. Reported TCA plasma binding parameters 3-34
3-18. Partition coefficients for TCE oxidative metabolites 3-35
3-19. Urinary excretion of trichloroacetic acid by various species exposed to
trichloroethylene 3-37
3-20. P450 isoform kinetics for metabolism of TCE to CH in human, rat, and mouse
recombinant P450s 3-38
3-21. P450 isoform activities in human liver microsomes exhibiting different affinities
for TCE 3-39
3-22. Comparison of peak blood concentrations in humans exposed to 100 ppm TCE
for 4 hours 3-43
3-23. GSH conjugation of TCE in liver and kidney cellular fractions in humans, male
F344rats, and male B6C3F1 mice 3-44
3-24. Kinetics of TCE metabolism via GSH conjugation in male F344 rat kidney and
human liver and kidney cellular and subcellular fractions 3-45
3-25. GGT activity in liver and kidney subcellular fractions of mice, rats, and humans 3-50
3-26. Multispecies comparison of whole-organ activity levels of GGT and dispeptidase 3-51
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xvi DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
3-27. Comparison of hepatic in vitro oxidation and conjugation of TCE 3-55
3-28. Estimates of DCVG in blood relative to inhaled TCE dose in humans exposed to
50 and 100 ppm 3-56
3-29. Concentrations of TCE in expired breath from inhalation-exposed humans 3-58
3-30. Conclusions from evaluation of Hack et al. (2006), and implications for PBPK
model development 3-65
3-31. Discussion of changes to the Hack et al. (2006) PBPK model implemented for
this assessment 3-70
3-32. PBPK model-based dose metrics 3-73
3-33. Rodent studies with pharmacokinetic data considered for analysis 3-74
3-34. Human studies with pharmacokinetic data considered for analysis 3-78
3-35. Parameters for which scaling from mouse to rat, or from mouse and rat to human,
was used to update the prior distributions 3-83
3-36. Physiological parameters: prior and posterior combined uncertainty and
variability 3-88
3-37. Distribution parameters: prior and posterior combined uncertainty and
variability 3-90
3-38. Absorption parameters: prior and posterior combined uncertainty and
variability 3-92
3-39. TCE metabolism parameters: prior and posterior combined uncertainty and
variability 3-93
3-40. Metabolite metabolism parameters: prior and posterior combined uncertainty and
variability 3-94
3-41. Estimates of the residual error 3-98
3-42. Summary comparison of updated PBPK model predictions and in vivo data in
mice 3-100
3-43. Summary comparison of updated PBPK model predictions and in vivo data
used for "calibration" in rats 3-103
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xvii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
3-44. Summary comparison of updated PBPK model predictions and in vivo data
used for "out-of-sample" evaluation in rats 3-106
3-45. Summary comparison of updated PBPK model predictions and in vivo data
used for "calibration" in humans 3-109
3-46. Summary comparison of updated PBPK model predictions and in vivo data
used for "out-of-sample" evaluation in humans 3-111
3-47. Posterior predictions for representative internal doses: mouse 3-123
3-48. Posterior predictions for representative internal doses: rat 3-124
3-49. Posterior predictions for representative internal doses: human 3-125
3-50. Degree of variance in dose metric predictions due to incomplete convergence,
combined uncertainty and population variability, uncertainty in particular human
population percentiles, model fits to in vivo data 3-136
4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR)
studies assessing cancer and TCE exposure 4-2
4-2. Case-control epidemiologic studies examining cancer and TCE exposure 4-9
4-3. Geographic-based studies assessing cancer and TCE exposure 4-19
4-4. Standards of epidemiologic study design and analysis use for identifying cancer
hazard and TCE exposure 4-21
4-5. Summary of criteria for meta-analysis study selection 4-25
4-6. TCE genotoxicity: bacterial assays 4-33
4-7. TCE genotoxicity: fungal and yeast systems 4-36
4-8. TCE genotoxicity: mammalian systems—gene mutations and chromosome
aberrations 4-38
4-9. TCE genotoxicity: mammalian systems—micronucleus, sister chromatic
exchanges 4-43
4-10. TCE genotoxicity: mammalian systems—unscheduled DNA synthesis, DNA
strand breaks/protein crosslinks, cell transformation 4-47
4-11. Genotoxicity of Trichloroacetic acid—bacterial systems 4-51
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xviii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
4-12. TCA Genotoxicity—mammalian systems 4-54
4-13. Genotoxicity of dichloroacetic acid 4-58
4-14. Genotoxicity of dichloroacetic acid—mammalian systems 4-59
4-15. Chloral hydrate genotoxicity: bacterial, yeast and fungal systems 4-64
4-16. Chloral hydrate genotoxicity: mammalian systems—all genetic endpoints,
in vitro 4-66
4-17. Chloral hydrate genotoxicity: mammalian systems—all genetic damage, in vivo 4-68
4-18. TCE GSH conjugation metabolites genotoxicity 4-76
4-19. Genotoxicity of trichloroethanol 4-80
4-20. Summary of human trigeminal nerve and nerve conduction velocity studies 4-86
4-21. Summary of animal trigeminal nerve studies 4-91
4-22. Summary of human auditory function studies 4-94
4-23. Summary of animal auditory function studies 4-97
4-24. Summary of mammalian sensory studies—vestibular and visual systems 4-102
4-25. Summary of human visual function studies 4-104
4-26. Summary of animal visual system studies 4-106
4-27. Summary of human cognition effect studies 4-109
4-28. Summary of animal cognition effect studies 4-111
4-29. Summary of human choice reaction time studies 4-114
4-30. Summary of animal psychomotor function and reaction time studies 4-117
4-31. Summary of animal locomotor activity studies 4-119
4-32. Summary of human developmental neurotoxicity associated with TCE
exposures 4-124
4-33. Summary of mammalian in vivo developmental neurotoxicity studies—oral
exposures 4-126
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xix DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
4-34. Summary of animal dopamine neuronal studies 4-130
4-35. Summary of neurophysiological, neurochemical, and neuropathological effects
with TCE exposure 4-132
4-36. Summary of in vitro ion channel effects with TCE exposure 4-134
4-37. Summary of human kidney toxicity studies 4-143
4-38. Summary of human studies on TCE exposure and kidney cancer 4-149
4-39. Summary of case-control studies on kidney cancer and occupation or job title 4-156
4-40. Summary of human studies on somatic mutations of the VHL genea 4-174
4-41. Summary of renal toxicity and tumor findings in gavage studies of
trichloroethylenebyNTP(1990) 4-179
4-42. Summary of renal toxicity and tumor findings in gavage studies of
trichloroethylenebyNCI(1976) 4-181
4-43. Summary of renal toxicity findings in gavage studies of trichloroethylene by
Maltonietal. (1988) 4-181
4-44. Summary of renal toxicity and tumor incidence in gavage studies of
trichloroethylene by NTP (1988) 4-182
4-45. Summary of renal toxicity and tumor findings in inhalation studies of
trichloroethylene by Maltoni etal. (1988) 4-183
4-46. Summary of renal tumor findings in inhalation studies of trichloroethylene by
Henschler et al. (1980) and Fukuda etal. (1983) 4-185
4-47. Summary of renal tumor findings in gavage studies of trichloroethylene by
Henschler et al. (1984) and Van Duuren et al. (1979) 4-187
4-48. Summary of histological changes in renal proximal tubular cells induced by
chronic exposure to TCE, DCVC, and TCOH 4-190
4-49. Summary of human liver toxicity studies 4-212
4-50. Selected results from epidemiologic studies of TCE exposure and cirrhosis 4-215
4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer 4-219
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xx DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
4-52. Summary of liver tumor findings in gavage studies of trichloroethylene by
NTP(1990) 4-255
4-53. Summary of liver tumor findings in gavage studies of trichloroethylene by
NCI (1976) 4-256
4-54. Summary of liver tumor incidence in gavage studies of trichloroethylene by
NTP(1988) 4-257
4-55. Summary of liver tumor findings in inhalation studies of trichloroethylene by
Maltonietal. (1988) 4-258
4-56. Summary of liver tumor findings in inhalation studies of trichloroethylene by
Henschler et al. (1980) and Fukuda et al. (1983) 4-259
4-57. Summary of liver tumor findings in gavage studies of trichloroethylene by
Henschler etal. (1984) 4-260
4-58. Studies of immune parameters and trichloroethylene in humans 4-333
4-59. Case-control studies of autoimmune diseases with measures of trichloroethylene
exposure 4-341
4-60. Incidence cohort studies of TCE exposure and lymphopoietic and hematopoietic
cancer risk 4-347
4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and
hematopoietic cancer risk 4-350
4-62. Case-control studies of TCE exposure and lymphopoietic cancer or leukemia 4-358
4-63. Geographic-based studies of TCE and non-Hodgkin lymphoma or leukemia in
adults 4-362
4-64. Selected results from epidemiologic studies of TCE exposure and childhood
leukemia 4-365
4-65. Summary of TCE immunosuppression studies 4-374
4-66. Summary of TCE hypersensitivity studies 4-382
4-67. Summary of autoimmune-related studies of TCE and metabolites in mice and
rats 4-385
4-68. Malignant lymphomas incidence in mice exposed to TCE in gavage and
inhalation exposure studies 4-398
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxi DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
4-69. Leukemia incidence in rats exposed to TCE in gavage and inhalation exposure
studies 4-399
4-70. Selected results from epidemiologic studies of TCE exposure and lung cancer 4-404
4-71. Selected results from epidemiologic studies of TCE exposure and laryngeal
cancer 4-411
4-72. Animal toxicity studies of trichloroethylene 4-416
4-73. Animal carcinogenicity studies of trichloroethylene 4-424
4-74. Human reproductive effects 4-445
4-75. Summary of mammalian in vivo reproductive toxicity studies—inhalation
exposures 4-447
4-76. Summary of mammalian in vivo reproductive toxicity studies—oral exposures 4-449
4-77. Summary of adverse female reproductive outcomes associated with TCE
exposures 4-461
4-78. Summary of adverse male reproductive outcomes associated with TCE
exposures 4-463
4-79. Summary of human studies on TCE exposure and prostate cancer 4-469
4-80. Summary of human studies on TCE exposure and breast cancer 4-472
4-81. Summary of human studies on TCE exposure and cervical cancer 4-475
4-82. Histopathology findings in reproductive organs 4-481
4-83. Testicular tumors in male rats exposed to TCE, adjusted for reduced survival 4-482
4-84. Developmental studies in humans 4-484
4-85. Summary of mammalian in vivo developmental toxicity studies—inhalation
exposures 4-507
4-86. Ocular defects observed 4-508
4-87. Summary of mammalian in vivo developmental toxicity studies—oral exposures 4-509
4-88. Types of congenital cardiac defects observed in TCE-exposed fetuses 4-517
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
4-89. Types of heart malformations per 100 fetuses 4-518
4-90. Congenital cardiac malformations 4-52
4-91. Summary of adverse fetal and early neonatal outcomes associated with TCE
exposures 4-531
4-92. Summary of studies that identified cardiac malformations associated with TCE
exposures 4-533
4-93. Events in cardiac valve formation in mammals and birds 4-536
4-94. Summary of other structural developmental outcomes associated with TCE
exposures 4-540
4-95. Summary of developmental neurotoxicity associated with TCE exposures 4-542
4-96. Summary of developmental immunotoxicity associated with TCE exposures 4-544
4-97. Summary of childhood cancers associated with TCE exposures 4-546
4-98. Selected observations from case-control studies of TCE exposure and
esophageal cancer 4-548
4-99. Summary of human studies on TCE exposure and esophageal cancer 4-551
4-100. Summary of human studies on TCE exposure and bladder cancer 4-558
4-101. Summary of human studies on TCE exposure and brain cancer 4-564
4-102. Estimated lifestage-specific daily doses for TCE in water 4-570
5-1. Neurological effects in studies suitable for dose-response, and corresponding
cRfCs and cRfDs 5-8
5-2. Kidney, liver, and body weight effects in studies suitable for dose-response, and
corresponding cRfCs and cRfDs 5-12
5-3. Immunological effects in studies suitable for dose-response, and corresponding
cRfCs and cRfDs 5-17
5-4. Reproductive effects in studies suitable for dose-response, and corresponding
cRfCs and cRfDs 5-21
5-5. Developmental effects in studies suitable for dose-response, and corresponding
cRfCs and cRfDs 5-26
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxiii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
5-6. Ranges of cRfCs based on applied dose for various noncancer effects associated
with inhalation TCE exposure 5-31
5-7. Ranges of cRfDs based on applied dose for various noncancer effects associated
with oral TCE exposure 5-32
5-8. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical neurological
effects 5-49
5-9. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical kidney effects 5-51
5-10. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical liver effects 5-53
5-11. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical immunological
effects 5-54
5-12. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical reproductive effects5-56
5-13. cRfCs and cRfDs and p-cRfCs and p-cRfDs for candidate critical developmental
effects 5-59
5-14. Comparison of "sensitive individual" HECs or HEDs for neurological effects
based on PBPK modeled internal dose metrics at different levels of confidence
and sensitivity, at the NOAEL or LOAEL 5-69
5-15. Comparison of "sensitive individual" HECs or HEDs for kidney and liver
effects based on PBPK modeled internal dose metrics at different levels of
confidence and sensitivity, at the NOAEL or LOAEL 5-70
5-16. Comparison of "sensitive individual" HECs or HEDs for immunological effects
based on PBPK modeled internal dose metrics at different levels of confidence
and sensitivity, at the NOAEL or LOAEL 5-72
5-17. Comparison of "sensitive individual" HECs or HEDs for reproductive effects
based on PBPK modeled internal dose metrics at different levels of confidence
and sensitivity, at the NOAEL or LOAEL 5-73
5-18. Comparison of "sensitive individual" HECs or HEDs for developmental effects
based on PBPK modeled internal dose metrics at different levels of confidence
and sensitivity, at the NOAEL or LOAEL 5-75
5-19. Lowest p-cRfCs or cRfCs for different effect domains 5-77
5-20. Lowest p-cRfDs or cRfDs for different effect domains 5-79
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxiv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
5-21. Lowest p-cRfCs for candidate critical effects for different types of effect based
on primary dose metric 5-81
5-22. Lowest p-cRfDs for candidate critical effects for different types of effect based
on primary dose metric 5-81
5-23. Summary of critical studies, effects, PODs, and UFs supporting the RfC 5-84
5-24. Summary of critical studies, effects, PODs, and UFs supporting the RfD 5-87
5-25. Inhalation bioassays 5-89
5-26. Oral bioassays 5-90
5-27. Specific dose-response analyses performed and dose metrics used 5-94
5-28. Mean PBPK model predictions for weekly internal dose in humans exposed
continuously to low levels of TCE via inhalation (ppm) or orally (mg/kg/d) 5-109
5-29. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor
type (inhalation) 5-110
5-30. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor
type (oral) 5-112
5-31. Comparison of survival-adjusted results for 3 oral male rat data sets 5-116
5-32. Inhalation: most sensitive bioassay for each sex/species combination 5-120
5-33. Oral: most sensitive bioassay for each sex/species combination 5-120
5-34. Summary of PBPK model-based uncertainty analysis of unit risk estimates for
each sex/species/bioassay/tumor type (inhalation) 5-127
5-35. Summary of PBPK model-based uncertainty analysis of unit risk estimates for
each sex/species/bioassay/tumor type (oral) 5-128
5-36. Results from Charbotel et al. on relationship between TCE exposure and RCC 5-131
5-37. Extra risk estimates for RCC incidence from various levels of lifetime exposure
to TCE, using linear cumulative exposure model 5-133
5-38. EC01, LEC01, and unit risk estimates for RCC incidence, using linear
cumulative exposure model 5-134
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
5-39. Relative contributions to extra risk for cancer incidence from TCE exposure
for multiple tumor types 5-140
5-40. Route-to-route extrapolation of site-specific inhalation unit risks to oral slope
factors 5-144
5-41. Estimates of age-specific water ingestion rates 5-151
5-42. Sample calculation for total lifetime cancer risk based on the kidney unit risk
estimate, adjusting for potential risk at multiple sites and for potential increased
early-life susceptibility and assuming a constant lifetime exposure to 1 ug/mL
of TCE in drinking water 5-152
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxvi DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
2-1. Molecular structure of TCE 2-2
2-2. Source contribution to TCE emissions 2-5
2-3. Annual emissions of TCE 2-6
2-4. Modeled ambient air concentrations of TCE 2-11
3-1. Gas uptake data from closed chamber exposure of rats to TCE 3-10
3-2. Disposition of [14C]TCE administered by oral gavage in mice 3-21
3-3. Disposition of [14C]TCE administered by oral gavage in rats 3-22
3-4. Scheme for the oxidative metabolism of TCE 3-24
3-5. Scheme for GSH-dependent metabolism of TCE 3-41
3-6. Interorgan TCE transport and metabolism via the GSH pathway 3-48
3-7. Overall structure of PBPK model for TCE and metabolites used in this
assessment 3-69
3-8. Schematic of how posterior predictions were generated for comparison with
experimental data 3-97
3-9. Comparison of urinary excretion data for NAcDCVC and predictions from the
Hacketal. and the updated PBPK models 3-107
3-10. Comparison of DCVG concentrations in human blood and predictions from the
updated 3-112
3-11. PBPK model predictions for the fraction of intake that is metabolized under
continuous inhalation and oral exposure conditions in mice, rats, and humans 3-114
3-12. PBPK model predictions for the fraction of intake that is metabolized by
oxidation under continuous inhalation and oral exposure conditions in mice,
rats, and humans 3-115
3-13. PBPK model predictions for the fraction of intake that is metabolized by GSH
conjugation under continuous inhalation and oral exposure conditions in mice,
rats, and humans 3-116
3-14. PBPK model predictions for the fraction of intake that is bioactivated DCVC in
the kidney under continuous inhalation and oral exposure conditions in rats and
humans 3-117
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxvii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES (continued)
3-15. PBPK model predictions for fraction of intake that is oxidized in the respiratory
tract under continuous inhalation and oral exposure conditions in mice, rats, and
humans 3-118
3-16. PBPK model predictions for the fraction of intake that is "untracked" oxidation
of TCE in the liver under continuous inhalation and oral exposure conditions
in mice, rats, and humans 3-119
3-17. PBPK model predictions for the weekly AUC of TCE in venous blood per unit
exposure under continuous inhalation and oral exposure conditions in mice, rats,
and humans 3-120
3-18. PBPK model predictions for the weekly AUC of TCOH in blood per unit
exposure under continuous inhalation and oral exposure conditions in mice,
rats, and humans 3-121
3-19. PBPK model predictions for the weekly AUC of TCA in the liver per unit
exposure under continuous inhalation and oral exposure conditions in mice,
rats, and humans 3-122
4-1. Meta-analysis of kidney cancer and overall TCE exposure 4-169
4-2. Meta-analysis of kidney cancer and TCE exposure—highest exposure groups 4-171
4-3. Relative risk estimates of liver and biliary tract cancer and overall TCE exposure.... 4-227
4-4. Meta-analysis of liver cancer and TCE exposure—highest exposure groups 4-229
4-5. Comparison of average fold-changes in relative liver weight to control and
exposure concentrations of 2 g/L or less in drinking water for TCA and DCA in
maleB6C3Fl mice for 14-30 days 4-267
4-6. Comparisons of fold-changes in average relative liver weight and gavage dose
of male B6C3F1 mice for 10-28 days of exposure and in male B6C3F1 and
Swiss mice 4-269
4-7. Comparison of fold-changes in relative liver weight for data sets in male
B6C3F1, Swiss, and NRMI mice between TCE studies and studies of direct
oral TCA administration to B6C3 Fl mice 4-271
4-8. Fold-changes in relative liver weight for data sets in male B6C3F1, Swiss, and
NRMI mice reported by TCE studies of duration 28-42 days using internal dose
metrics predicted by the PBPK model described in Section 3.5 4-27
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxviii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES (continued)
4-9. Dose-response relationship, expressed as percent incidence and fold-increase
over controls, for TCE hepatocarcinogenicity in NCI 4-287
4-10. Dose-response relationship, expressed as incidence and fold-increase over
controls, for TCE hepatocarcinogenicity in Maltoni et al 4-287
4-11. Dose-response data for hepatocellular carcinomas incidence and multiplicity,
induced by DCA from DeAngelo et al 4-288
4-12. Reported incidences of hepatocellular carcinomas and adenomas plus
carcinomas in various studies inB6C3Fl mice 4-290
4-13. Reported incidence of hepatocellular carcinomas induced by DCA and TCA
in 104-week studies 4-292
4-14. Effects of dietary control on the dose-response curves for changes in liver
tumor incidences induced by CHin diet 4-296
4-15. Meta-analysis of lymphoma and overall TCE exposure 4-370
4-16. Meta-analysis of lymphoma and TCE exposure—highest exposure groups 4-371
5-1. Flow-chart of the process used to derive the RfD and RfC for noncancer effects 5-2
5-2. Flow-chart for dose-response analyses of rodent noncancer effects using PBPK
model-based dose metrics 5-46
5-3. Schematic of combined interspecies, intraspecies, and route-to-route
extrapolation from a rodent study LOAEL or NOAEL 5-47
5-4. Flow-chart for uncertainty analysis of HECs and HEDs derived using PBPK
model-based dose metrics 5-67
5-5. Flow-chart for dose-response analyses of rodent bioassays using PBPK model-
based dose metrics 5-108
5-6. Flow-chart for uncertainty analysis of dose-response analyses of rodent bio
assays using PBPK model-based dose metrics 5-126
5-7. Flow-chart for route-to-route extrapolation of human site-specific cancer
inhalation unit risks to oral slope factors 5-142
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxix DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
1,2-DCVC
[14C]TCE
17-P-HSD
SepiPGF
8-OHdG
ADAF
ADME
AIC
ALL
ALT
ANA
ANCA
ASD
ASPEN
AST
ATF-2
ATSDR
AUC
AV
AVC
AZDHS
BAER
BAL
BMD
BMDL
BMDS
BMI
BMR
BSO
BW
S-(l,2-dichlorovinyl)-L-cysteine
[14C]-radio labeled TCE
17-p-hydroxy steroid dehydrogenase
8-epiprostaglandin F2alpha
8-hydroxy-2' deoxyguanosine
age-dependent adjustment factor
absorption, distribution, metabolism, and excretion
Akaike Information Criteria
acute lymphoblastic leukemia
alanine aminotransferase
antinuclear antibodies
antineutrophil-cytopiasmic antibody
autism spectrum disorder
Assessment System for Population Exposure Nationwide
aspartate aminotrasferase
activating transcription factor 2
Agency for Toxic Substances and Disease Registry
area-under-the-curve
atrioventricular
atrioventricular canal
Arizona Department of Health Services
brainstem auditory-evoked response
bronchoalveolar lavage
benchmark dose
benchmark dose lower bound
BenchMark Dose Software
body mass index
benchmark response
buthionine-(S,R)-sulfoximine
body weight
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxx DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
CADHS
CH
CI
CLL
CNS
C02
CoA
cRfCs
cRfDs
CRT
CYP
DBF
DBF
DCA
DCAC
DCE
DCVC
DCVG
DCVT
DEHA
DEHP
DHEAS
DNP
ECso
ECC
ECX
EEG
ERG
FAA
FDVE
California Department of Health Services
chloral hydrate
confidence interval
chronic lymphocytic leukemia
central nervous system
carbon dioxide
coenzyme A
candidate RfCs
candidate RfDs
choice reaction time
cytochrome
D-type peroxisomal bifunctional protein
dibutyl phthalate
dichloroacetic acid
dichloroacetyl chloride
dichloroethane
dichlorovinyl cysteine
S-dichlorovinyl glutathione
S-(l,2-dichlorovinyl) thiol
di(2-ethylhexyl) adipate
di(2-ethylhexyl) phthalate
dehydroepiandrosterone sulphate
dinitrophenol
median effective concentrations
extrahepatic cholangiocarcinoma
effective concentration corresponding to an extra risk of x%
el ectroencephal ograph
electroretinogram
fumarylacetoacetate
fluoromethyl-2,2-difluoro-1 -(trifluoromethyl)vinyl ether
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxi DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
FFVC
FMO
FOB
FSH
G-CSF
G6PDH
GA
GABA
GD
GGT
GI
GIS
GSD
GSH
GST
GT
H&E
H2O
H202
HAP
HCC
HC1
HDL-C
HEC
HED
HH
HPT
i.a.
i.p.
i.v.
(£',Z)-S-(l-fluoro-2-fluoromethoxy-2-(trifluoromethyl)vinyl)-Lcysteine
flavin mono-oxygenase
functional observational battery
follicle-stimulating hormone
granulocyte colony stimulating factor
glucose 6p dehydrogenase
glomerular antigen
gamma-amino butyric acid
gestation day
y-glutamyl transpeptidase or y-transpeptidase
gastro-intestinal
geographic information system
geometric standard deviation
glutathione
glutathione-S-transferase
glutamyl transferase
hematoxylin and eosin
water
hydrogen peroxide
hazardous air pollutant
hepatocellular carcinoma
hydrochloric acid
high density lipoprotein-cholesterol
human equivalent concentration
human equivalent dose
Hamberger and Hamilton
hypothalamic-pituitary-testis
intra-arterial
intraperitoneal
intravenous
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
IARC
ICC
ICD
ICRP
idPOD
IDR
IGF-II
IL
IRIS
IUGR
LDH
LEC
LECX
LH
LOAEL
LOH
LORR
MADPH
MA
MAA
MCA
MCMC
MCP
MLE
MMPI
MNU
MOA
MSW
NAC
NAcDCVC
International Agency for Research on Cancer
intrahepatic cholangiocarcinoma
International Classification of Disease
The International Commission on Radiological Protection
internal dose points of departure
incidence density ratio
insulin-like growth factor-II (gene)
interleukin
Integrated Risk Information System
intrauterine growth restriction
lactate dehydrogenase
lowest effective concentration
lowest effective concentration corresponding to an extra risk of x%
luteinizing hormone
lowest observed adverse effect level
loss of heterozygosity
loss of righting reflex
Massachusetts Department of Public Health
maleylacetone
maleylacetoacetate
monochloroacetic acid
Markov chain Monte Carlo
methylclofenapate
maximum likelihood estimate
Minnesota Multiphasic Personal Inventory
methyl nitrosourea
mode of action
multistage Weibull
N-acetylcysteine
N-acetyl-S-(l,2-dichlorovinyl)-L-cysteine
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxiii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
NAG
NAT
NCI
NHANES
NHL
NK
NOAEL
NOEC
NOEL
NPL
NPMC
NRC
NSATA
NTP
NYS DOH
OECD
OFT
OP
OR
p.v.
PB
PBPK
PCE
PCEs
PCNA
PCO
PCR
p-cRfC
p-cRfD
PFU
N-acetyl-p-D-glucosaminidase
N-acetyl transferase
National Cancer Institute
National Health and Nutrition Examination Survey
non-Hodgkin's lymphoma
natural killer
no-observed-adverse-effect level
no-observed-effect concentration
no-observed-effect level
National Priorities List
nonpurified rat peritoneal mast
National Research Council
National-Scale Air Toxics Assessment
National Toxicology Program
New York State Department of Health
Organization for Economic Co-operation and Development
outflow tract
oscillatory potential
odds ratio
intraperivenous
TCE blood-air partition coefficient
physiologically based pharmacokinetics
perchl oroethy 1 ene
polychromatic erythrocytes
proliferating cell nuclear antigen
palmitoyl-CoA oxidation
polymerase chain reaction
PBPK model-based candidate RfCs
PBPK model-based candidate RfDs
plaque-forming units
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxiv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
PND
P02
POD
PPARa
QC
RBL-2H3
RCC
RfC
RfD
ROS
RR
RRp
RT
S9
SBA
SCEs
S-D
SD
SDH
SEER
SES
SGA
SHBG
SIR
SMR
SRBC
SRT
SSB
TaClo
TEARS
postnatal day
partial pressure oxygen
point of departure
peroxisome proliferator activated receptor alpha
quality control
rat basophilic leukemia
renal cell carcinoma
inhalation reference concentration
oral reference dose
reactive oxygen species
relative risk
pooled RR
reaction time
metabolic activation system
serum bile acids
sister chromatid exchanges
Sprague-Dawley
standard deviation
sorbitol dehydrogenase
Surveillance, Epidemiology, and End Results
socio-economic status
small for gestational age
sex-hormone binding globulin
standardized incidence ratio
standardized mortality ratio
sheep red blood cells
simple reaction time
single-strand breaks
tetrahydro-beta-carbolines
thiobarbiturate acid-reactive substances
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS (continued)
TCA
TCAA
TCAH
TCE
TCOG
TCOH
TRI
TSEP
TTC
TWA
UA
UCL
UF
U.S. EPA
USGS
U-TCA
U-TTC
VEGF
VEP
VHL
VOC
VSCCs
W
YFF
trichloroacetic acid
tri chl oroacetal dehy de
trichloroacetaldehyde hydrate
tri chl oroethy 1 ene
tri chl oroethanol -glucuroni de conj ugate
trichloroethanol
Toxics Release Inventory
trigeminal somatosensory evoked potential
total trichloro compounds
time-weighted average
University of Arizona
upper confidence limit
uncertainty factor
U.S. Environmental Protection Agency
United States Geological Survey
urinary-TCA
urinary total trichloro-compounds
vascular endothelial growth factor
visual evoked potential
von Hippel-Lindau
volatile organic compound
voltage sensitive calcium channels
wakefulness
fluorescent Y-bodies
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxvi DRAFT—DO NOT CITE OR QUOTE
-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard and dose-response assessment in IRIS pertaining to chronic exposure to
trichloroethylene. It is not intended to be a comprehensive treatise on the chemical or
toxicological nature of trichloroethylene.
The intent of Chapter 6, Major Conclusions in the Characterization of Hazard and Dose
Response, is to present the major conclusions reached in the derivation of the reference dose,
reference concentration and cancer assessment, where applicable, and to characterize the overall
confidence in the quantitative and qualitative aspects of hazard and dose response. For other
general information about this assessment or other questions relating to IRIS, the reader is
referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxvii DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGER
Weihsueh A. Chiu
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
AUTHORS
AmbujaBale
National Center for Environmental Assessment—Immediate Office
U.S. Environmental Protection Agency
Washington, DC
Stanley Bar one
National Center for Environmental Assessment—Immediate Office
U.S. Environmental Protection Agency
Washington, DC
Rebecca Brown
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Jane C. Caldwell
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Chao Chen
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Weihsueh A. Chiu
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Glinda Cooper
National Center for Environmental Assessment—Immediate Office
U.S. Environmental Protection Agency
Washington, DC
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxviii DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Ghazi Dannan
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Marina Evans
National Health and Environmental Effects Research Laboratory
(on detail to National Center for Environmental Assessment—Washington Office)
U.S. Environmental Protection Agency
Research Triangle Park, NC
John Fox
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
KathrynZ. Guyton
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Maureen R. Gwinn
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Jennifer Jinot
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Nagalakshmi Keshava
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
John Lipscomb
National Center for Environmental Assessment—Cincinnati Office
U.S. Environmental Protection Agency
Cincinnati, OH
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xxxix DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Susan Makris
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Miles Okino
National Exposure Research Laboratory—Las Vegas Office
U.S. Environmental Protection Agency
Las Vegas, NV
Fred Power
National Exposure Research Laboratory—Las Vegas Office
U.S. Environmental Protection Agency
Las Vegas, NV
John Schaum
National Center for Environmental Assessment—Washington Office
Office of Research and Development
Washington, DC
Cheryl Siegel Scott
National Center for Environmental Assessment—Washington Office
Office of Research and Development
Washington, DC
REVIEWERS
This document has been reviewed by U.S. EPA scientists, reviewers from other Federal
agencies, and the public, and peer reviewed by independent scientists external to U.S. EPA. A
summary and U.S. EPA's disposition of the comments received from the independent external
peer reviewers and from the public is included in Appendix I.
INTERNAL EPA REVIEWERS
Daniel Axelrad
National Center for Environmental Economics
Robert Benson
U.S. EPA Region 8
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xl DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Ted Birner
National Center for Environmental Assessment—Immediate Office
Nancy Chiu
Office of Water
David Farrar
National Center for Environmental Assessment—Cincinnati Office
Lynn Flowers
National Center for Environmental Assessment—Immediate Office
Stiven Foster
Office of Solid Waste and Emergency Response
Susan Griffin
U.S. EPA Region 8
Samantha Jones
National Center for Environmental Assessment—Immediate Office
Leonid Kopylev
National Center for Environmental Assessment—Washington Office
Allan Marcus
National Center for Environmental Assessment—Immediate Office
Gregory Miller
Office of Children's Health Protection and Environmental Education
Deirdre Murphy
Office of Air Quality Planning and Standards
Marian Olsen
U.S. EPA Region 2
Peter Preuss
National Center for Environmental Assessment—Immediate Office
Kathleen Raffaele
National Center for Environmental Assessment—Washington Office
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xli DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
William Sette
Office of Solid Waste and Emergency Response
Bob Sonawane
National Center for Environmental Assessment—Washington Office
Suryanarayana Vulimiri
National Center for Environmental Assessment—Washington Office
Nina Ching Y. Wang
National Center for Environmental Assessment—Cincinnati Office
Paul White
National Center for Environmental Assessment—Washington Office
Marcia Bailey
U.S. EPA Region 10
ACKNOWLEDGMENTS
Drafts of Section 3.3 (TCE metabolism) were prepared for the U.S. EPA by Syracuse
Research Corporation under contract. Additional support, including literature searches and
retrievals and drafts of Appendix D were prepared for the U.S. EPA by the Oak Ridge Institute
for Science and Education (ORISE) through interagency agreement number DW-89939822-01-0
with the U.S. Department of Energy (DOE). ORISE is managed by Oak Ridge Associated
Universities under a contract with DOE. The PBPK modeling sections of this report are
dedicated to the memory of Fred Power (1938-2007). His keen analytical mind will be greatly
missed, but his gentle heart and big smile will be missed even more.
Additionally, we gratefully acknowledge Terri Konoza of NCEA for her management of
the document production process, and the following individuals for document production
support:
IntelliTech Systems, Inc. ECFlex, Inc.
Chris Broyles Heidi Glick
Debbie Kleiser Lana Wood
Stacey Lewis
Linda Tackett
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xlii DRAFT—DO NOT CITE OR QUOTE
-------
EXECUTIVE SUMMARY
There is substantial potential for human exposure to trichloroethylene (TCE), as it has a
widespread presence in ambient air, indoor air, soil, and groundwater. At the same time, humans
are likely to be exposed to a variety of compounds that are either metabolites of TCE or which
have common metabolites or targets of toxicity. Once exposed, humans, as well as laboratory
animal species, rapidly absorb TCE, which is then distributed to tissues via systemic circulation,
extensively metabolized, and then excreted primarily in breath as unchanged TCE or carbon
dioxide, or in urine as metabolites.
Based on the available human epidemiologic data and experimental and mechanistic
studies, it is concluded that TCE poses a potential human health hazard for noncancer toxicity to
the central nervous system, the kidney, the liver, the immune system, the male reproductive
system, and the developing fetus. The evidence is more limited for TCE toxicity to the
respiratory tract and female reproductive system. Following U.S. Environmental Protection
Agency (U.S. EPA, 2005a) Guidelines for Carcinogen Risk Assessment, TCE is characterized as
carcinogenic in humans by all routes of exposure. This conclusion is based on convincing
evidence of a causal association between TCE exposure in humans and kidney cancer. The
human evidence of carcinogenicity from epidemiologic studies of TCE exposure is compelling
for non-Hodgkins Lymphoma but less convincing than for kidney cancer, and more limited for
liver and biliary tract cancer. Further support for the characterization of TCE as carcinogenic in
humans by all routes of exposure is derived from positive results in multiple rodent cancer
bioassays in rats and mice of both sexes, similar toxicokinetics between rodents and humans,
mechanistic data supporting a mutagenic mode of action (MOA) for kidney tumors, and the lack
of mechanistic data supporting the conclusion that any of the MOA(s) for TCE-induced rodent
tumors are irrelevant to humans.
As TCE toxicity and carcinogenicity are generally associated with TCE metabolism,
susceptibility to TCE health effects may be modulated by factors affecting toxicokinetics,
including lifestage, gender, genetic polymorphisms, race/ethnicity, pre-existing health status,
lifestyle, and nutrition status. In addition, while these some of these factors are known risk
factors for effects associated with TCE exposure, it is not known how TCE interacts with known
risk factors for human diseases.
For noncancer effects, the most sensitive types of effects, based either on human
equivalent concentrations/doses or on candidate inhalation reference concentrations (RfCs)/oral
reference doses (RfDs), appear to be developmental, kidney, and immunological (adult and
developmental) effects. The neurological and reproductive effects appear to be about an order of
magnitude less sensitive, with liver effects another two orders of magnitude less sensitive. The
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xliii DRAFT—DO NOT CITE OR QUOTE
-------
preferred RfC estimate of 0.001 ppm (1 ppb or 5 ug/m3) is based on route-to-route extrapolated
results from oral studies for the critical effects of heart malformations (rats), immunotoxicity
(mice), and toxic nephropathy (rats, mice), and an inhalation study for the critical effect of
increased kidney weight (rats). Similarly, the preferred RfD estimate for noncancer effects of
0.0004 mg/kg/d is based on the critical effects of heart malformations (rats), adult
immunological effects (mice), developmental immunotoxicity (mice), and toxic nephropathy
(rats). There is high confidence in these preferred noncancer reference values, as they are
supported by moderate- to high-confidence estimates for multiple effects from multiple studies.
For cancer, the preferred estimate of the inhalation unit risk is 2 x 10~2 per ppm [4 x 10~6
per jig/m3], based on human kidney cancer risks reported by Charbotel et al. (2006) and
adjusted, using human epidemiologic data, for potential risk for tumors at multiple sites. The
preferred estimate of the oral unit risk for cancer is 5 x 10~2 per mg/kg/d, resulting from
physiologically-based pharmacokinetic model-based route-to-route extrapolation of the
inhalation unit risk estimate based on the human kidney cancer risks reported in Charbotel et al.
(2006) and adjusted, using human epidemiologic data, for potential risk for tumors at multiple
sites. There is high confidence in these unit risks for cancer, as they are based on good quality
human data, as well as being similar to unit risk estimates based on multiple rodent bioassays.
Because there is both sufficient weight of evidence to conclude that TCE operates through a
mutagenic MOA for kidney tumors and a lack of TCE-specific quantitative data on early-life
susceptibility, the default age-dependent adjustment factors (ADAFs) can be applied for the
kidney cancer component of the unit risks for cancer; however, the application of ADAFs is
likely to have a minimal impact on the total cancer risk except when exposures are primarily
during early life.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 xliv DRAFT—DO NOT CITE OR QUOTE
-------
1 1 INTRODUCTION
2
3
4 This document presents background information and justification for the Integrated Risk
5 Information System (IRIS) Summary of the hazard and dose-response assessment of
6 trichloroethylene. IRIS Summaries may include oral reference dose (RfD) and inhalation
7 reference concentration (RfC) values for chronic and other exposure durations, and a
8 carcinogenicity assessment.
9 The RfD and RfC, if derived, provide quantitative information for use in risk assessments
10 for health effects known or assumed to be produced through a nonlinear (presumed threshold)
11 mode of action. The RfD (expressed in units of mg/kg/d) is defined as an estimate (with
12 uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
13 population (including sensitive subgroups) that is likely to be without an appreciable risk of
14 deleterious effects during a lifetime. The inhalation RfC (expressed in units of ppm or ug/m3) is
15 analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
16 inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
17 effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
18 values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
19 acute (<24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
20 lifetime) exposure durations, all of which are derived based on an assumption of continuous
21 exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
22 derived for chronic exposure duration.
23 The carcinogenicity assessment provides information on the carcinogenic hazard
24 potential of the substance in question and quantitative estimates of risk from oral and inhalation
25 exposure may be derived. The information includes a weight-of-evidence judgment of the
26 likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
27 effects may be expressed. Quantitative risk estimates may be derived from the application of a
28 low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
29 the estimate of risk per mg/kg/d of oral exposure. Similarly, an inhalation unit risk is a plausible
30 upper bound on the estimate of risk per ppm or ug/m3 in air breathed.
31 Development of these hazard identification and dose-response assessments for
32 trichloroethylene has followed the general guidelines for risk assessment as set forth by the
33 National Research Council (NRC, 1983). U.S. EPA Guidelines and Risk Assessment Forum
34 Technical Panel Reports that may have been used in the development of this assessment include
35 the following: Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA,
36 1986a), Guidelines for Mutagenicity Risk Assessment (U. S. EPA, 1986b), Recommendations for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 1 -1 DRAFT—DO NOT CITE OR QUOTE
-------
1 and Documentation of Biological Values for Use in Risk Assessment (U.S. EPA, 1988),
2 Guidelines for Developmental Toxicity Risk Assessment (U.S. EPA, 1991), Interim Policy for
3 Particle Size and Limit Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994a), Methods
4 for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry
5 (U.S. EPA, 1994b), Use of the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA,
6 1995), Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996), Guidelines for
1 Neurotoxicity Risk Assessment (U.S. EPA, 1998), Science Policy Council Handbook: Risk
8 Characterization (U.S. EPA, 2000a), Benchmark Dose Technical Guidance Document (U.S.
9 EPA, 2000b), Supplementary Guidance for Conducting Health Risk Assessment of Chemical
10 Mixtures (U.S. EPA, 2000c), A Review of the Reference Dose and Reference Concentration
11 Processes (U.S. EPA, 2002), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
12 Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
13 (U.S. EPA, 2005b), Science Policy Council Handbook: Peer Review (U.S. EPA, 2006a), and A
14 Framework for Assessing Health Risks of Environmental Exposures to Children (U. S. EPA,
15 2006b).
16 The literature search strategy employed for this compound was based on the Chemical
17 Abstracts Service Registry Number and at least one common name. Any pertinent scientific
18 information submitted by the public to the IRIS Submission Desk was also considered in the
19 development of this document. The relevant literature was reviewed through April, 2009.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 1 -2 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
2. EXPOSURE CHARACTERIZATION
The purpose of this exposure characterization is to summarize information about
trichloroethylene (TCE) sources, releases, media levels, and exposure pathways for the general
population (occupational exposure is also discussed to a lesser extent). It is not meant as a
substitute for a detailed exposure assessment for a particular risk assessment application. While
this section primarily addresses TCE, it also includes some information on a number of related
compounds. These related compounds include metabolites of TCE and other parent compounds
that produce similar metabolites as shown in Table 2-1. The first column in this table lists the
principal TCE metabolites in humans (trichloroethanol, trichloroethanol-glucuronide and
trichloroacetic acid) as well as a number of minor ones (Agency for Toxic Substances and
Disease Registry [ATSDR], 1997a). The subsequent columns list parent compounds that can
produce some of the same metabolites. The metabolic reaction pathways are much more
complicated than implied here and it should be understood that this table is intended only to
provide a general understanding of which parent compounds lead to which TCE metabolites.
Exposure to the TCE-related compounds can alter or enhance TCE's metabolism and toxicity by
generating higher internal metabolite concentrations than would result from TCE exposure by
itself. This characterization is based largely on earlier work by Wu and Schaum (2000, 2001),
but also provides updates in a number of areas.
Table 2-1. TCE metabolites and related parent compounds*
TCE metabolites
Oxalic acid
Chloral
Chloral hydrate
Monochloroacetic acid
Dichloroacetic acid
Trichloroacetic acid
Trichloroethanol
Trichloroethanol-
glucuronide
Parent compounds
Tetrachloro-
ethylene
X
X
X
X
X
X
X
1,1-Dichloro-
ethane
X
X
1,1,1-Tri-
chloroethane
X
X
X
X
1,1,1,2-Tetra-
chloroethane
X
X
X
X
X
X
1,2-Dichloro-
ethylene
X
X
24
25
26
* X indicates that the parent compound can produce the corresponding metabolite (Hazardous Substances Data
Bank, http://toxnet.nlm.nih.gov./cgi-bin/sis/htmlgen?HSDB).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-1 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
2.1. ENVIRONMENTAL SOURCES
TCE is a stable, colorless liquid with a chloroform-like odor and chemical formula
C2C13H as diagrammed in Figure 2-1 (Lewis, 2001). Its chemical properties are listed in
Table 2-2.
H
Cl
Cl
Figure 2-1. Molecular structure of TCE.
10
11
Table 2-2. Chemical properties of TCE
Property
Molecular weight
Boiling point
Melting point
Density
Solubility
Vapor pressure
Vapor density
Henry's Law Constant
Octanol/water partition
coefficient
Air concentration conversion
Value
131.39
87.2°C
-84.7°C
1.4642 at 20°C
1,280 mg/L water at 25°C
69.8mmHG@25°C
4.53 (air = 1)
9.85 x 10"3 atm-cu m/mol @ 25°C
log Kow = 2.61
1 ppb = 5.38 ug/m3
Reference
Lide, 1998
Lide, 1998
Lide, 1998
Merck Index, 1996
Hotvath et al., 1999
Boublik et al., 1984
Merck Index, 1996
Leighton, 1981
Hansch, 1995
HSDB, 2002
12
13
14
15
16
17
18
19
20
21
Trichloroethylene has been produced commercially since the 1920s in many countries by
chlorination of ethylene or acetylene. Its use in vapor degreasing began in the 1920s. In the
1930s, it was introduced for use in dry cleaning. This use was largely discontinued in the 1950s
and was replaced with tetrachloroethylene (ATSDR, 1997a). More recently, 80-90% of
trichloroethylene production worldwide is used for degreasing metals (International Agency for
Research on Cancer [IARC], 1995). It is also used in adhesives, paint-stripping formulations,
paints, lacquers, and varnishes (SRI, 1992). A number of past uses in cosmetics, drugs, foods,
and pesticides have now been discontinued including use as an extractant for spice oleoresins,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-2 DRAFT—DO NOT CITE OR QUOTE
-------
2
O
4
5
6
7
8
9
10
11
natural fats and oils, hops and decaffeination of coffee (IARC, 1995), and as a carrier solvent for
the active ingredients of insecticides and fungicides, and for spotting fluids (WHO, 1985;
ATSDR, 1997a). The production of TCE in the United States peaked in 1970 at 280 million kg
(616 million pounds) and declined to 60 million kg (132 million pounds) in 1998 (United States
Geological Survey [USGS], 2006). In 1996, the United States imported 4.5 million kg
(10 million pounds) and exported 29.5 million kg (65 million pounds) (Chemical Marketing
Reporter, 1997). Table 2-3 summarizes the basic properties and principal uses of the TCE
related compounds.
Table 2-3. Properties and uses of TCE related compounds
Tetrachloroethylene
1,1,1 -Trichloroethane
1 ,2-Dichloroethylene
1,1,1,2-
Tetrachloroethane
1 , 1 -Dichloroethane
Chloral
Chloral hydrate
Monochloroacetic
acid
Dichloroacetic acid
Trichloroacetic acid
Oxalic acid
Dichlorovinyl
cysteine
Trichloroethanol
Water
solubility
(mg/L)
150
4,400
3,000-6,000
1,100
5,500
High
High
High
High
High
220,000
Not available
Low
Vapor
pressure
(mmHG)
18.5 @25°C
124 @25°C
273-395
@30°C
14 @25°C
234 @25°C
35 @20°C
NA
1 @43°C
<1 @20°C
1 @50°C
0.54@105°C
Not available
NA
Uses
Dry cleaning, degreasing, solvent
Solvents, degreasing
Solvents, chemical intermediates
Solvents, but currently not
produced in United States
Solvents, chemical intermediates
Herbicide production
Pharmaceutical production
Pharmaceutical production
Pharmaceuticals, not widely used
Herbicide production
Scouring/cleaning agent,
degreasing
Not available
Anesthetics and chemical
intermediate
Sources
a
a
a
a,b
a
a
a
a
a
a
b
c
12
13
14
15
aWu and Schaum (2001).
bHSDB (2003).
"Lewis (2001).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-3 DRAFT—DO NOT CITE OR QUOTE
-------
2
O
4
5
6
7
8
9
10
11
Releases of TCE from nonanthropogenic activities are negligible (HSDB, 2002). Most of
the TCE used in the United States is released to the atmosphere, primarily from vapor degreasing
operations (ATSDR, 1997a). Releases to air also occur at treatment and disposal facilities, water
treatment facilities, and landfills (ATSDR, 1997a). TCE has also been detected in stack
emissions from municipal and hazardous waste incineration (ATSDR, 1997a). TCE is on the list
for reporting to U.S. Environmental Protection Agency (U.S. EPA)'s Toxics Release Inventory
(TRI). Reported releases into air predominate over other types and have declined over the period
1994 to 2004 (see Table 2-4).
Table 2-4. Toxics Release Inventory (TRI) releases of TCE (pounds/year)
Year
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
On-site
fugitive air
15,018,818
12,498,086
10,891,223
9,276,150
6,769,810
5,861,635
5,485,493
4,968,282
4,761,104
3,963,054
3,040,460
2,733,983
2,816,241
On-site
stack air
15,929,943
13,784,853
10,995,228
8,947,909
6,504,289
4,784,057
4,375,516
3,453,451
3,436,289
3,121,718
3,144,980
2,893,168
2,795,184
Total on-
site air
emissions
30,948,761
26,282,939
21,886,451
18,224,059
13,274,099
10,645,692
9,861,009
8,421,733
8,197,393
7,084,772
6,185,440
5,627,152
5,611,425
On-site
surface
water
discharges
1,671
1,477
541
568
882
1,034
593
406
579
595
216
533
482
Total on-site
underground
injection
288
550
1,291
986
593
0
47,877
98,220
140,190
90,971
123,637
86,817
0
Total on-
site
releases
to land
4,070
3,577
9,740
3,975
800
148,867
9,607
12,609
230
150,642
2
4,711
77,339
Total off-
site
disposal
or other
releases
96,312
74,145
89,527
182,423
136,766
192,385
171,952
133,531
139,398
66,894
71,780
60,074
90,758
Total on-
and off-
site
disposal
or other
releases
31,051,102
26,362,688
21,987,550
18,412,011
13,413,140
10,987,978
10,091,038
8,666,499
8,477,790
7,393,873
6,381,075
5,779,287
5,780,004
12
13
14
15
16
17
18
19
20
Source: U.S. EPA TRI Explorer, http://www.epa.gov/triexplorer/trends.htm.
Under the National-Scale Air Toxics Assessment (NSATA) program, U.S. EPA has
developed an emissions inventory for TCE (U.S. EPA, 2007a). The inventory includes sources
in the United States plus the Commonwealth of Puerto Rico and the U.S. Virgin Islands. The
types of emission sources in the inventory include large facilities, such as waste incinerators and
factories and smaller sources, such as dry cleaners and small manufacturers. Figures 2-2 and 2-3
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-4 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
show the results of the 1999 emissions inventory for TCE. Figure 2-2 shows the percent
contribution to total emissions by source category. A variety of sources have TCE emissions
with the largest ones identified as halogenated solvent cleaners and metal parts and products.
Figure 2-3 shows a national map of the emission density (tons/sq mi-yr) for TCE. This map
shows the highest densities in the far west and northeastern regions of the United States.
Emissions range from 0 to 4.12 tons/mi2-yr.
Trichloroethylene Emissions
1999
2° > Aerospace industries
2% lion a Steel Manufacturing
2% Consumer and Commercial Products Use
4% Dry Cleaning
6% & Products (Surface Coating)
Municipal Landfills
2^ Puip ami Paper Production
2J;. Prinlis3g, Coating & Dyeing Of
19% Categories (293 categories)
Solvent
Figure 2-2. Source contribution to TCE emissions.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-5 DRAFT—DO NOT CITE OR QUOTE
-------
4
5
6
9
10
11
12
13
14
15
16
17
18
19
20
21
1999 County Emission Densities
Trichbroethylene — United States Counties
Distribution of U.S. Emission Densities
HlghsetlnU.S. —
-------
1 reaction with hydroxyl radicals, the estimated half-life of trichloroethylene in the atmosphere is
2 on the order of 1 to 11 days with production of phosgene, dichloroacetyl chloride, and formyl
3 chloride. Under smog conditions, degradation is more rapid (half-life on the order of hours)
4 (HSDB, 2002; Howard et al., 1991).
5
6 2.2.3. Fate in Aquatic Environments
7 The dominant fate of trichloroethylene released to surface waters is volatilization
8 (predicted half-life of minutes to hours). Bioconcentration, biodegradation, and sorption to
9 sediments and suspended solids are not thought to be significant (HSDB, 2002).
10 Trichloroethylene is not hydrolyzed under normal environmental conditions. However, slow
11 photo-oxidation in water (half-life of 10.7 months) has been reported (HSDB, 2002; Howard et
12 al., 1991).
13 2.3. EXPOSURE CONCENTRATIONS
14 TCE levels in the various environmental media result from the releases and fate processes
15 discussed in Sections 2.1 and 2.2. No statistically based national sampling programs have been
16 conducted that would allow estimates of true national means for any environmental medium. A
17 substantial amount of air and groundwater data, however, has been collected as well as some
18 data in other media, as described below.
19
20 2.3.1. Outdoor Air—Measured Levels
21 TCE has been detected in the air throughout the United States. According to ATSDR
22 (1997a), atmospheric levels are highest in areas concentrated with industry and population, and
23 lower in remote and rural regions. Table 2-5 shows levels of TCE measured in the ambient air at
24 a variety of locations in the United States.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-7 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 2-5. Concentrations of trichloroethylene in ambient air
4
5
6
7
8
9
10
11
12
13
14
15
16
Area
Rural
Whiteface Mountain, NYa
Badger Pass, CAa
Reese River, NVa
Jetmar, KSa
All rural sites
Urban and Suburban
New Jersey a
New York City, NYa
Los Angeles, CAa
Lake Charles, LAa
Phoenix, AZa
Denver, COa
St. Louis, MOa
Portland, ORa
Philadelphia, PAa
Southeast Chicago, ILb
East St. Louis, ILb
District of Columbia0
Urban Chicago, ILd
Suburban Chicago, ILd
300 cities in 42 states6
Several Canadian Citiesf
Several United States Citiesf
Phoenix, AZg
Tucson, AZg
All urban/suburban sites
Year
1974
1977
1977
1978
1974-1978
1973-79
1974
1976
1976-78
1979
1980
1980
1984
1983-1984
1986-1990
1986-1990
1990-1991
pre-1993
pre-1993
pre-1986
1990
1990
1994-1996
1994-1996
1973-1996
Concentration (ug/m3)
Mean
0.5
0.06
0.06
0.07
9.1
3.8
1.7
8.6
2.6
1.07
0.6
1.5
1.9
1.0
2.1
1.94
0.82-1.16
0.52
2.65
0.28
6.0
0.29
0.23
Range
<0.3-1.9
0.005-0.09
0.005-0.09
0.04-0.11
0.005-1.9
ND-97
0.6-5.9
0.14-9.5
0.4-11.3
0.06-16.7
0.15-2.2
0.1-1.3
0.6-3.9
1.6-2.1
1-16.65
0-1.53
0-1.47
0-97
aIARC (1995).
bSweet (1992).
"Handler (1992).
dScheff(1993).
eShah(1988).
fBunce (1994).
gZielinska (1998).
More recent ambient air measurement data for TCE were obtained from U.S. EPA's Air
Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html (U.S.
EPA, 2007b). These data were collected from a variety of sources including state and local
environmental agencies. The data are not from a statistically based survey and cannot be
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-8 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
assumed to provide nationally representative values. The most recent data (2006) come from
258 monitors located in 37 states. The means for these monitors range from 0.03 to 7.73 ug/m3
and have an overall average of 0.23 ug/m3. Table 2-6 summarizes the data for the years
1999-2006. The data suggest that levels have remained fairly constant since 1999 at about
0.3 ug/m3. Table 2-7 shows the monitoring data organized by land setting (rural, suburban, or
urban) and land use (agricultural, commercial, forest, industrial, mobile, and residential). Urban
air levels are almost 4 times higher than rural areas. Among the land use categories, TCE levels
are highest in commercial/industrial areas and lowest in forest areas.
Table 2-6. TCE ambient air monitoring data (ug/m3)
Year
1999
2000
2001
2002
2003
2004
2005
2006
Number of
monitors
162
187
204
259
248
256
313
258
Number of
states
20
28
31
41
41
37
38
37
Mean
0.30
0.34
0.25
0.37
0.35
0.32
0.43
0.23
Standard
deviation
0.53
0.75
0.92
1.26
0.64
0.75
1.05
0.55
Median
0.16
0.16
0.13
0.13
0.16
0.13
0.14
0.13
Range
0.01-4.38
0.01-7.39
0.01-12.90
0.01-18.44
0.02-6.92
0.00-5.78
0.00-6.64
0.03-7.73
12
13
14
15
16
17
18
Source: U.S. EPA's Air Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html.
Table 2-7. Mean TCE air levels across monitors by land setting and use
(1985 to 1998)
Mean
concentration
(ug/m3)
n
Rural
0.42
93
Subur-
ban
1.26
500
Urban
1.61
558
Agricul-
tural
1.08
31
Com-
mercial
1.84
430
Forest
0.1
17
Indus-
trial
1.54
186
Mobile
1.5
39
Resi-
dential
0.89
450
19
20
21
22
23
Source: U.S. EPA's Air Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-9 DRAFT—DO NOT CITE OR QUOTE
-------
1 2.3.2. Outdoor Air—Modeled Levels
2 Under the National-Scale Air Toxics Assessment program, U.S. EPA has compiled
3 emissions data and modeled air concentrations/exposures for the Criteria Pollutants and
4 Hazardous Air Pollutants (U.S. EPA, 2007a). The results of the 1999 emissions inventory for
5 TCE were discussed earlier and results presented in Figures 2-2 and 2-3. A computer simulation
6 model known as the Assessment System for Population Exposure Nationwide (ASPEN) is used
7 to estimate toxic air pollutant concentrations (U.S. EPA, 2005). This model is based on the
8 U.S. EPA's Industrial Source Complex Long Term model which simulates the behavior of the
9 pollutants after they are emitted into the atmosphere. ASPEN uses estimates of toxic air
10 pollutant emissions and meteorological data from National Weather Service Stations to estimate
11 air toxics concentrations nationwide. The ASPEN model takes into account important
12 determinants of pollutant concentrations, such as
13
14 • rate of release;
15 • 1 ocati on of rel ease;
16 • the height from which the pollutants are released;
17 • wind speeds and directions from the meteorological stations nearest to the release;
18 • breakdown of the pollutants in the atmosphere after being released (i.e., reactive decay);
19 • settling of pollutants out of the atmosphere (i.e., deposition) and
20 • transformation of one pollutant into another (i.e., secondary formation).
21
22 The model estimates toxic air pollutant concentrations for every census tract in the continental
23 United States, the Commonwealth of Puerto Rico and the U.S. Virgin Islands. Census tracts are
24 land areas defined by the U.S. Bureau of the Census and typically contain about 4,000 residents
25 each. Census tracts are usually smaller than 2 square miles in size in cities but much larger in
26 rural areas.
27 Figure 2-4 shows the results of the 1999 ambient air concentration modeling for TCE.
28 The county median air levels range from 0 to 3.79 |ig/m3 and an overall median of 0.054 |ig/m3.
29 They have a pattern similar to the emission densities shown in Figure 2-3. These NSATA
30 modeled levels appear lower than the monitoring results presented above. For example, the 1999
31 air monitoring data (Table 2-6) indicates a median outdoor air level of 0.16 ug/m3 which is about
32 3 times as high as the modeled 1999 county median (0.054 |ig/m3). However, it should be
33 understood that the results from these two efforts are not perfectly comparable. The modeled
34 value is a median of county levels for the entire United States which includes many rural areas.
35 The monitors cover many fewer areas (n= 162 for 1999) and most are in nonrural locations. A
36 better analysis is provided by U.S. EPA (2007) which presents a comparison of modeling results
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-10 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
from NSATA to measured values at the same locations. For 1999, it was found that
formaldehyde levels were underestimated at 79% of the sites (n = 92). Thus, while the NSATA
modeling results are useful for understanding geographic distributions, they may frequently
underestimate ambient levels.
1999 Estimated County Median Ambient Concentrations
Trichbroethylene — United States Counties
Distribution of U.S. Ambient Concentrations
HlghsetlnU.S.
95
SO
Pereentile 75
50
£5
Lows* In U.S.
County Median Ambient Pollutant Concentration
( rnlercgrarns / cubic meter)
Source: U.S. EPA / QAQPS
1999 NATA National—Scale Air Toxfcs Assessment
7
8
9
10
11
12
13
14
15
16
17
18
19
Figure 2-4. Modeled ambient air concentrations of TCE.
2.3.3. Indoor Air
TCE can be released to indoor air from use of consumer products that contain it (i.e.,
adhesives and tapes), vapor intrusion (migration of volatile chemicals from the subsurface into
overlying buildings) and volatilization from the water supply. Where such sources are present, it
is likely that indoor levels will be higher than outdoor levels. A number of studies have
measured indoor levels of TCE:
• The 1987 U.S. EPA Total Exposure Assessment Methodology study (U.S. EPA, 1987)
showed that the ratio of indoor to outdoor TCE concentrations for residences in
Greensboro, NC, was about 5:1.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-11 DRAFT—DO NOT CITE OR QUOTE
-------
1 • In two homes using well water with TCE levels averaging 22 to 128 ug/L, the TCE levels
3
2 in bathroom air ranged from <500 to 40,000 ug /m when the shower ran less than 30
3 minutes (Andelman et al., 1985).
4 • Shah and Singh (1988) report an average indoor level of 7.2 ug/m3 based on over 2,000
5 measurements made in residences and workplaces during 1981-1984 from various
6 locations across the United States.
7 • Hers et al. (2001) provides a summary of indoor air TCE measurements at locations in
8 United States, Canada, and Europe with a range of <1 to 165 ug/m3.
9 • Sapkota et al. (2005) measured TCE levels inside and outside of the Baltimore Harbor
10 Tunnel toll booths during the summer of 2001. Mean TCE levels were 3.11 ug/m3
11 indoors and 0.08 ug/m3 outdoors based on measurements on 7 days. The authors
12 speculated that indoor sources, possibly dry cleaning residues on uniforms, were the
13 primary source of the indoor TCE.
14 • Sexton et al. (2005) measured TCE levels inside and outside residences in
15 Minneapolis/St. Paul metropolitan area. Two day samples were collected over three
16 seasons in 1999. Mean TCE levels were 0.5 ug/m3 indoors (n = 292), 0.2 ug/m3 outdoors
17 (n = 132) and 1.0 ug/m3 based on personal sampling (n = 288).
18 • Zhu et al. (2005) measured TCE levels inside and outside of residences in Ottawa,
19 Canada. 75 homes were randomly selected and measurements were made during the
20 winter of 2002/2003. TCE was above detection limits in the indoor air of 33% of the
21 residences and in the outdoor air of 19% of the residences. The mean levels were
22 0.06 ug/m3 indoors and 0.08 ug/m3 outdoors. Given the high frequency of nondetects, a
23 more meaningful comparison can be made on basis of the 75th percentiles: 0.08 ug/m3
24 indoors and 0.01 ug/m3 outdoors.
25
26 TCE levels measured indoors have been directly linked to vapor intrusion at two sites in New
27 York:
28
29 • TCE vapor intrusion has occurred in buildings/residences near a former Smith Corona
30 manufacturing facility located in Cortlandville, NY. An extensive sampling program
31 conducted in 2006 has detected TCE in groundwater (1-13 ug/L), soil gas (6-97 ug/m3),
32 subslab gas (2-1,600 ug/m3), and indoor air (1-17 ug/m3) (NYSDEC, 2006a).
33 • Evidence of vapor intrusion of TCE has also been reported in buildings and residences in
34 Endicott, NY. Sampling in 2003 showed total volatile organic compounds (VOCs) in
35 soil gas exceeding 10,000 ug/m3 in some areas. Indoor air sampling detected TCE levels
36 ranging from 1 to 140 ug/m3 (NYSDEC, 2006b).
37
38 Little et al. (1992) developed attenuation coefficients relating contaminants in soil gas
39 (assumed to be in chemical equilibrium with the groundwater) to possible indoor levels as a
40 result of vapor intrusion. On this basis they estimated that TCE groundwater levels of 540 ug/L,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-12 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
(a high contamination level) could produce indoor air levels of 5 to 500 ug/m3. Vapor intrusion
is likely to be a significant source only in situations where residences are located near soils or
groundwater with high contamination levels. U.S. EPA (2002) recommends considering vapor
intrusion when volatiles are suspected to be present in groundwater or soil at a depth of
<100 feet. Hers et al. (2001) concluded that the contribution of VOCs from subsurface sources
relative to indoor sources is small for most chemicals and sites.
2.3.4. Water
A number of early (pre-1990) studies measured TCE levels in natural water bodies
(levels in drinking water are discussed later in this section) as summarized in Table 2-8.
According to IARC (1995), the reported median concentrations of TCE in 1983-1984 were
0.5 ug/L in industrial effluents and 0.1 ug/L in ambient water. Results from an analysis of the
U.S. EPA STORET Data Base (1980-1982) showed that TCE was detected in 28% of 9,295
surface water reporting stations nationwide (ATSDR, 1997a). A more recent search of the
STORET database for TCE measurements nationwide during 2008 in streams, rivers and lakes
indicated 3 detects (0.03 to 0.04 ug/L) out of 150 samples (STORET Database,
http ://www. epa.gov/storet/dbtop. html).
Table 2-8. Concentrations of trichloroethylene in water based on pre-1990
studies
Water type
Industrial effluent
Surface waters
Rainwater
Groundwater
Drinking water
Location
U.S.
U.S.
Portland, OR
MN
NJ
NY
PA
MA
AZ
U.S.
U.S
U.S.
MA
NJ
CA
CA
NC
ND
Year
83
83
84
83
76
80
80
76
76
77
78
84
84
85
84
84
84
Mean
(Hg/L)
0.006
23.4
66
5
5
Median
(Hg/L)
0.5
0.1
Range
(Ug/L)
0.002-0.02
0.2-144
<1,530
<3,800
<27,300
<900
8.9-29
0.2-49
0-53
0.5-210
max. 267
max. 67
8-12
Number of
samples
NR
NR
NR
NR
NR
NR
NR
NR
NR
1130
486
486
48
48
Ref.
IARC, 1995
IARC, 1995
Ligockietal., 1985
Sabeletal, 1984
Burmasteretal., 1982
Burmasteretal., 1982
Burmasteretal., 1982
Burmasteretal., 1982
IARC, 1995
IARC, 1995
IARC, 1995
IARC, 1995
IARC, 1995
Cohnetal., 1994
U.S. EPA, 1987
U.S. EPA, 1987
U.S. EPA, 1987
U.S. EPA, 1987
22 NR = Not Reported.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-13 DRAFT—DO NOT CITE OR QUOTE
-------
1 ATSDR (1997a) has reported that TCE is the most frequently reported organic
2 contaminant in groundwater and the one present in the highest concentration in a summary of
3 ground water analyses reported in 1982. It has been estimated that between 9 and 34% of the
4 drinking water supply sources tested in the United States may have some trichloroethylene
5 contamination. This estimate is based on available Federal and State surveys (ATSDR, 1997a).
6 Squillace et al. (2004) reported TCE levels in shallow groundwater based on data from
7 the National Water Quality Assessment Program managed by USGS. Samples from 518 wells
8 were collected from 1996 to 2002. All wells were located in residential or commercial areas and
9 had a median depth of 10 m. The authors reported that approximately 8.3% of the well levels
10 were above the detection limit (level not specified), 2.3% were above 0.1 ug/L and 1.7% were
11 above 0.2 ug/L.
12 As part of the Agency's first Six-Year Review, EPA obtained analytical results for over
13 200,000 monitoring samples reported at 23,035 public water systems in 16 states (U.S. EPA,
14 2003). Approximately 2.6% of the systems had at least one sample exceed a minimum reporting
15 level of 0.5 ug/L; almost 0.65% had at least one sample that exceeds the MCL of 5 ug/L. Based
16 on average system concentrations estimated by U.S. EPA, 54 systems (0.23%) had an average
17 concentration that exceeded the MCL. U.S. EPA's statistical analysis to extrapolate the sample
18 result to all systems regulated for TCE resulted in an estimate of 154 systems with average TCE
19 concentrations that exceed the MCL.
20 TCE concentrations in ground water have been measured extensively in California. The
21 data were derived from a survey of water utilities with more than 200 service connections. The
22 survey was conducted by the California Department of Health Services (CA DHS, 1986). From
23 January 1984 through December 1985, untreated water from wells in 819 water systems were
24 sampled for organic chemical contamination. The water systems use a total of 5,550 wells,
25 2,947 of which were sampled. TCE was found in 187 wells at concentrations up to 440 ug/L,
26 with a median concentration among the detects of 3.0 ug/L. Generally, the wells with the highest
27 concentrations were found in the heavily urbanized areas of the state. Los Angeles County
28 registered the greatest number of contaminated wells (149).
29 A second California study collected data on TCE levels in public drinking water
30 (Williams et al., 2002). The data were obtained from the CA DHS. The data spanned the years
31 1995 to 2001 and the number of samples for each year ranged from 3,447 to 4,226. The percent
32 of sources that were above the detection limit ranged from 9.6 to 11.7 per year (detection limits
33 not specified). The annual average detected concentrations ranged from 14.2 to 21.6 ug/L.
34 Although not reported, the overall average concentration of the samples (assuming an average of
35 20 ug/L among the samples above the detection limit, 10% detection rate and 0 for the
36 nondetects) would be about 2 ug/L.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 The USGS (2006) conducted a national assessment of 55 VOCs, including
2 trichloroethylene, in ground water. A total of 3,500 water samples were collected during
3 1985-2001. Samples were collected at the well head prior to any form of treatment. The types
4 of wells sampled included 2,400 domestic wells and 1,100 public wells. Almost 20% of the
5 samples contained one or more of the VOCs above the assessment level of 0.2 ug/L. The
6 detection frequency increased to over 50% when a subset of samples was analyzed with a low
7 level method that had an assessment level of 0.02 ug/L. The largest detection frequencies were
8 observed in California, Nevada, Florida, the New England States and Mid-Atlantic states. The
9 most frequently detected VOCs (>1% of samples) include TCE, tetrachloroethylene,
10 1,1,1-trichloroethane (methyl chloroform), 1,2 dichloroethylene, and 1,1-dichloroethane.
11 Findings specific to TCE include the following:
12
13 • Detection frequency was 2.6% at 0.2 ug/L and was 3.8% at 0.02 ug/L.
14 • The median concentration was 0.15 ug/L with a range of 0.02 to 100 ug/L.
15 • The number of samples exceeding the MCL (5 ug/L) was 6 at domestic wells and 9 at
16 public wells.
17
18 USGS (2006) also reported that four solvents (TCE, tetrachloroethylene,
19 1,1,1-trichloroethane and methylene chloride) occurred together in 5% of the samples. The most
20 frequently occurring two-solvent mixture was TCE and tetrachloroethylene. The report stated
21 that the most likely reason for this co-occurrence is the reductive dechlorination of
22 tetrachloroethylene to TCE.
23
24 2.3.5. Other Media
25 Levels of TCE were found in the sediment and marine animal tissue collected in
26 1980-1981 near the discharge zone of a Los Angeles County waste treatment plant.
27 Concentrations were 17 ug/L in the effluent, <0.5 ug/kg in dry weight in sediment, and
28 0.3-7 ug/kg wet weight in various marine animal tissue (IARC, 1995). TCE has also been found
29 in a variety of foods. FDA has limits on TCE use as a food additive in decaffeinated coffee and
30 extract spice oleoresins (see Table 2-15). Table 2-9 summarizes data from two sources:
31
32 • IARC (1995) reports average concentrations of TCE in limited food samples collected in
33 the United States.
34 • Fleming-Jones and Smith (2003) measured VOC levels in over 70 foods collected from
35 1996 to 2000 as part of the FDA's Total Diet Program. All foods were collected directly
36 from supermarkets. Analysis was done on foods in a ready-to-eat form. Sample sizes for
37 most foods were in the 2-5 range.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-15 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 2-9. Levels in food
IARC (1995)
Cheese 3.8 ug/kg
Butter and Margarine 73.6 ug/kg
Peanut Butter 0.5 ug/kg
Cereals 3 ug/kg
Grain-based Foods 0.9 ug/kg
Fleming-Jones and Smith (2003)
Cheese 2-3 ug/kg
Butter 7-9 ug/kg
Margarine 2-21 ug/kg
Cheese Pizza 2 ug/kg
Nuts 2-5 ug/kg
Peanut Butter 4-70 ug/kg
Ground Beef 3-6 ug/kg
Beef Frankfurters 2-105 ug/kg
Hamburger 5-9 ug/kg
Cheeseburger 7 ug/kg
Chicken Nuggets 2-5 ug/kg
Bologna 2-20 ug/kg
Pepperoni Pizza 2 ug/kg
Banana 2 ug/kg
Avocado 2-75 ug/kg
Orange 2 ug/kg
Chocolate Cake 3-57 ug/kg
Blueberry Muffin 3-4 ug/kg
Sweet Roll 3 ug/kg
Chocolate Chip Cookies 2-4 ug/kg
Apple Pie 2-4 ug/kg
Doughnuts 3 ug/kg
Tuna 9-11 ug/kg
Cereal 3 ug/kg
Popcorn 4-8 ug/kg
French Fries 3 ug/kg
Potato Chips 4-140 ug/kg
Coleslaw 3 ug/kg
4
5
6
7
8
9
10
11
2.3.6. Biological Monitoring
Biological monitoring studies have detected TCE in human blood and urine in the United
States and other countries such as Croatia, China, Switzerland, and Germany (IARC, 1995).
Concentrations of TCE in persons exposed through occupational degreasing operations were
most likely to have detectable levels (IARC, 1995). In 1982, 8 of 8 human breastmilk samples
from 4 United States urban areas had detectable levels of TCE. The levels of TCE detected,
however, are not specified (HSDB, 2002; ATSDR, 1997a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-16 DRAFT—DO NOT CITE OR QUOTE
-------
1 The Third National Health and Nutrition Examination Survey (NHANES III) examined
2 TCE concentrations in blood in 677 nonoccupationally exposed individuals. The individuals
3 were drawn from the general U.S. population and selected on the basis of age, race, gender and
4 region of residence (IARC, 1995; Ashley et al., 1994). The samples were collected during 1988
5 to 1994. TCE levels in whole blood were below the detection limit of 0.01 ug/L for about 90%
6 of the people sampled (see Table 2-10). Assuming that nondetects equal half of the detection
7 limit, the mean concentration was about 0.017 ug/L.
9
10
Table 2-10. TCE levels in whole blood by population percentile
Percentiles
Concentration
(ug/L)
10
ND
20
ND
30
ND
40
ND
50
ND
60
ND
70
ND
80
ND
90
0.012
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
ND = Nondetect, i.e., below detection limit of 0.01 ug/L.
Data from IARC (1995) and Ashley (1994).
2.4. EXPOSURE PATHWAYS AND LEVELS
2.4.1. General Population
Because of the pervasiveness of TCE in the environment, most people are likely to have
some exposure via one or more of the following pathways: ingestion of drinking water,
inhalation of outdoor/indoor air, or ingestion of food (ATSDR, 1997a). As noted earlier, the
NHANES survey suggests that about 10% of the population has detectable levels of TCE in
blood. Each pathway is discussed below.
2.4.1.1. Inhalation
As discussed earlier, U.S. EPA has estimated emissions and modeled air concentrations
for the Criteria Pollutants and Hazardous Air Pollutants under the National-Scale Air Toxics
Assessment program (U.S. EPA, 2007a). This program has also estimated inhalation exposures
on a nationwide basis. The exposure estimates are based on the modeled concentrations from
outdoor sources and human activity patterns (U.S. EPA, 2005). Table 2-11 shows the 1999
results for TCE.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-17 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 2-11. Modeled 1999 annual exposure concentrations (ug/m ) for
trichloroethylene
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Percentile
5
10
25
50
75
90
95
Mean
Exposure concentration (ug/m3)
Rural areas
0.030
0.034
0.038
0.044
0.053
0.070
0.097
0.058
Urban areas
0.048
0.054
0.065
0.086
0.122
0.189
0.295
0.130
Nationwide
0.038
0.043
0.056
0.076
0.113
0.172
0.262
0.116
Percentiles and mean are based on census tract values.
Source: http://www.epa.gov/ttn/atw/nata/ted/exporisk.html#indb.
These modeled inhalation exposures would have a geographic distribution similar to that
of the modeled air concentrations as shown in Figure 2-4. Table 2-11 indicates that TCE
inhalation exposures in urban areas are generally about twice as high as rural areas. While these
modeling results are useful for understanding the geographic distribution of exposures, they
appear to underestimate actual exposures. This is based on the fact that, as discussed earlier, the
modeled ambient air levels are generally lower than measured values. Also, the modeled
exposures do not consider indoor sources. Indoor sources of TCE make the indoor levels higher
than ambient levels. This is particularly important to consider since people spend about 90% of
their time indoors (U.S. EPA, 1997). A number of measurement studies were presented earlier
that showed higher TCE levels indoors than outdoors. Sexton et al. (2005) measured TCE levels
in Minneapolis/St. Paul area and found means of 0.5 ug/m3 indoors (n = 292) and 1.0 ug/m3
based on personal sampling (n = 288). Using 1.0 ug/m3 and an average adult inhalation rate of
13 m3 air/day (U.S. EPA, 1997) yields an estimated intake of 13 ug/day. This is consistent with
ATSDR (1997a), which reports an average daily air intake for the general population of 11 to
33 ug/day.
2.4.1.2. Ingestion
The median value from the nationwide survey of domestic and public wells by USGS for
1985-2001 is 0.15 ug/L. This value was selected for exposure estimation purposes because it
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-18 DRAFT—DO NOT CITE OR QUOTE
-------
1 was the most current and most representative of the national population. Using this value and an
2 average adult water consumption rate of 1.4 L/d (this is from U.S. EPA, 1997, but note that
3 U.S. EPA (2004) indicates a mean per capita daily average total water ingestion from all sources
4 of 1.233 L) yields an estimated intake of 0.2 ug/day. This is lower than the ATSDR (1997a)
5 estimate water intake for the general population of 2 to 20 ug/day. The use of the USGS survey
6 to represent drinking water is uncertain in two ways. First, the USGS survey measured only
7 groundwater and some drinking water supplies use surface water. Second, the USGS measured
8 TCE levels at the well head, not the drinking water tap. Further discussion about the possible
9 extent and magnitude of TCE exposure via drinking water is presented below.
10 According to ATSDR (1997a), TCE is the most frequently reported organic contaminant
11 in ground water (ATSDR, 1997a), and between 9 and 34% of the drinking water supply sources
12 tested in the United States may have some TCE contamination. Approximately 90% of the
13 155,000 public drinking water systems1 in the United States are ground water systems. The
14 drinking water standard for TCE only applies to community water systems (CWSs) and
15 approximately 78% of the 51,972 CWSs in the United States are ground water systems
16 (U.S. EPA, 2008). Although commonly detected in water supplies, the levels are generally low
17 because, as discussed earlier, MCL violations for TCE in public water supplies are relatively rare
18 for any extended period (U.S. EPA, 1998). The USGS (2006) survey found that the number of
19 samples exceeding the MCL (5 ug/L) was 6 at domestic wells (n = 2,400) and 9 at public wells
20 (n = 1,100). Private wells, however, are often not closely monitored and if located near TCE
21 disposal/contamination sites where leaching occurs, may have undetected contamination levels.
22 About 10% of Americans (27 million people) obtain water from sources other than public water
23 systems, primarily private wells (U.S. EPA, 1995). TCE is a common contaminant at Superfund
24 sites. It has been identified in at least 861 of the 1,428 hazardous waste sites proposed for
25 inclusion on the U.S. EPA National Priorities List (NPL) (ATSDR, 1997a). Studies have shown
26 that many people live near these sites: 41 million people live less than 4 miles from one or more
27 of the nation's NPL sites, and on average 3,325 people live within 1 mile of any given NPL site
28 (ATSDR, 1996b).
29 Table 2-12 presents preliminary estimates of TCE intake from food. They are based on
30 average adult food ingestion rates and food data from Table 2-9. This approach suggests a total
31 ingestion intake of about 5 ug/d. It is important to consider this estimate as preliminary because
32 it is derived by applying data from very limited food samples to broad classes of food.
1 Public water systems (PWSs) are defined as systems which provide water for human consumption through pipes or
other constructed conveyances to at least 15 service connections or serves an average of at least 25 people for at
least 60 days a year. U.S. EPA further specifies three types of PWSs, including Community Water System
(CWS)—a PWS that supplies water to the same population year-round.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-19 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 2-12. Preliminary estimates of TCE intake from food ingestion
Fruit
Vegetables
Fish
Meat
Dairy products
Grains
Sweets
Total
Consumption
rate (g/kg-d)
3.4
4.3
2.1
8
4.1
0.5
Consumption
rate (g/d)
238
301
20
147
560
287
35
Concentration
in food (ug/kg)
2
3
10
5
3
3
3
Intake
(Hg/d)
0.48
0.90
0.20
0.73
1.68
0.86
0.10
4.96
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
"Consumption rates are per capita averages from U.S. EPA (1997).
bConsumption rates in g/d assume 70 kg body weight.
2.4.1.3. Dermal
TCE in bathing water and consumer products can result in dermal exposure. A modeling
study has suggested that a significant fraction of the total dose associated with exposure to
volatile organics in drinking water results from dermal absorption (Brown et al., 1984).
U.S. EPA (2004) used a prediction model based on octanol-water partitioning and molecular
weight to derive a dermal permeability coefficient for TCE in water of 0.012 cm/hour. U.S. EPA
used this value to compute the dermally absorbed dose from a 35 minute shower and compared it
to the dose from drinking 2 L of water at the same concentration. This comparison indicated that
the dermal dose would be 17% of the oral dose. Much higher dermal permeabilities were
reported by Nakai et al. (1999) based on human skin in vitro testing. For dilute aqueous
solutions of TCE, they measured a permeability coefficient of 0.12 cm/hour (26°C). Nakai et al.
(1999) also measured a permeability coefficient of 0.018 cm/hour for tetrachloroethylene in
water. Poet et al. (2000) measured dermal absorption of TCE in humans from both water and
soil matrices. The absorbed dose was estimated by applying a physiologically based
pharmacokinetic model to TCE levels in breath. The permeability coefficient was estimated to
be 0.015 cm/hour for TCE in water and 0.007 cm/hour for TCE in soil (Poet et al., 2000).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-20 DRAFT—DO NOT CITE OR QUOTE
-------
1 2.4.1.4. Exposure to TCERelated Compounds
2 Table 2-13 presents adult exposure estimates that have been reported for the TCE related
3 compounds. This table was originally compiled by Wu and Schaum (2001). The exposure/dose
4 estimates are taken directly from the listed sources or derived based on monitoring data
5 presented in the source documents. They are considered "preliminary" because they are
6 generally based on very limited monitoring data. These preliminary estimates suggest that
7 exposures to most of the TCE related compounds are comparable to or greater than TCE itself.
9
10
Table 2-13. Preliminary intake estimates of TCE and TCE-related chemicals
Chemical
Trichloroethylene (TCE)
Tetrachloroethylene
(PERC)
1,1,1 -Trichloroethane
1 ,2-Dichloroethylene
Cis- 1 ,2-Dichloroethylene
1,1, 1 ,2-Tetrachloroethane
1 , 1 -Dichloroethane
Chloral
Monochloroacetic acid
Dichloroacetic acid
Trichloroacetic acid
Population
General
General
Occupational
General
General
Occupational
General
General
General
General
General
General
General
General
General
General
General
General
General
Media
Air
Water
Air
Air
Water
Air
Air
Water
Air
Water
Air
Water
Air
Air
Water
Water
Water
Water
Water
Range of estimated
adult exposures
(jig/day)
11-33
2-20b
2,232-9,489
80-200
0.1-0.2
5,897-219,685
10.8-108
0.38-4.2
1-6
2.2
5.4
0.5-5.4
142
4
2.47-469.38
0.02-36.4
2-2.4
10-266
8.56-322
Range of adult doses
(mg/kg/d)
1.57E-04-4.71E-04
2.86E-05-2.86E-04
3.19E-02-1.36E-01
1.14E-03-2.86E-03
1.43E-06-2.86E-06
8.43E-02-3.14
1.54E-04-1.54E-03
5.5E-06-6.00E-05
1.43E-05-8.57E-05
3.14E-05
7.71E-05
7.14E-06-7.71E-05
2.03E -03
5.71E-05
3.53E-05-6.71E-03
2.86E-07-5.20E-04
2.86E-05-3.43E-05
1.43E-04-3.80E-03
1.22E-03-4.60E-03
Data sources"
ATSDR, 1997a
ATSDR, 1997a
ATSDR, 1997a
ATSDR, 1997b
ATSDR, 1997b
ATSDR, 1997b
ATSDR, 1995
ATSDR, 1995
ATSDR, 1996a
ATSDR, 1996a
HSDB, 1996
HSDB, 1996
HSDB, 2002
ATSDR, 1990
ATSDR, 1990
HSDB, 1996
U.S. EPA, 1994
IARC, 1995
IARC, 1995
11
12
13
14
15
aOriginally compiled in Wu and Schaum (2001).
bNew data from USGS (2006) suggests much lower water intakes, i.e., 0.2 ug/d.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-21 DRAFT—DO NOT CITE OR QUOTE
-------
1 2.4.2. Potentially Highly Exposed Populations
2 Some members of the general population may have elevated TCE exposures. ATSDR
3 (1997'a) has reported that TCE exposures may be elevated for people living near waste facilities
4 where TCE may be released, residents of some urban or industrialized areas, people exposed at
5 work (discussed further below) and individuals using certain products (also discussed further
6 below). Because TCE has been detected in breast milk samples of the general population,
7 infants who ingest breast milk may be exposed, as well. Increased TCE exposure is also a
8 possible concern for bottle-fed infants because they ingest more water on a bodyweight basis
9 than adults (the average water ingestion rate for adults is 21 mL/kg-d and for infants under one
10 year old it is 44 mL/kg-d—U.S. EPA, 1997). Also, because TCE can be present in soil, children
11 may be exposed through activities such as playing in or ingesting soil.
12
13 2.4.2.1. Occupational Exposure
14 Occupational exposure to TCE in the United States has been identified in various
15 degreasing operations, silk screening, taxidermy, and electronics cleaning (IARC, 1995). The
16 major use of trichloroethylene is for metal cleaning or degreasing (IARC, 1995). Degreasing is
17 used to remove oils, greases, waxes, tars, and moisture before galvanizing, electroplating,
18 painting, anodizing, and coating. The five primary industries using TCE degreasing are furniture
19 and fixtures; electronic and electric equipment; transport equipment; fabricated metal products;
20 and miscellaneous manufacturing industries (IARC, 1995). Additionally, TCE is used in the
21 manufacture of plastics, appliances, jewelry, plumbing fixtures, automobile, textiles, paper, and
22 glass (IARC, 1995).
23 Table 2-13 lists the primary types of industrial degreasing procedures and the years that
24 the associated solvents were used. Vapor degreasing has the highest potential for exposure
25 because vapors can escape into the work place. Hot dip tanks, where trichloroethylene is heated
26 to close to its boiling point of 87°C, are also major sources of vapor that can create exposures as
27 high as vapor degreasers. Cold dip tanks have a lower exposure potential, but they have a large
28 surface area which enhances volatilization. Small bench-top cleaning operations with a rag or
29 brush and open bucket have the lowest exposure potential. In combination with the vapor
30 source, the size and ventilation of the workroom are the main determinants of exposure intensity
31 (NRC, 2006).
32 Occupational exposure to TCE has been assessed in a number of epidemiologic studies.
33 Studies of aircraft workers show short term peak exposures in the hundreds of ppm
34 (>540 mg/m3) and long term exposures in the low tens of ppm (>54 mg/m3) (Spirtas et al., 1991;
35 Blair et al., 1998; Garabrant et al., 1988; Morgan et al., 1998; and Boice et al., 1998). Similar
36 exposures have been reported for cardboard/paperboard workers (Henschler et al., 1995; Sinks et
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-22 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
al., 1992) and uranium processors (Ritz, 1999). ATSDR (1997a) reports that the majority of
published worker exposure data show time-weighted average (TWA) concentrations ranging
from <50 ppm to 100 ppm (<270-540 mg/m3). National Institute of Occupational Safety and
Health conducted a survey of various industries from 1981 to 1983 and estimated that
approximately 401,000 U.S. employees in 23,225 plants in the United States were potentially
exposed to TCE during this timeframe (IARC, 1995; ATSDR, 1997a).
Occupational exposure to TCE has likely declined since the 1950's and 1960's due to
decreased usage, better release controls and improvements in worker protection. Reductions in
TCE use are illustrated in Table 2-14, which shows that by about 1980 common degreasing
operations had substituted other solvents for TCE.
Table 2-14 Years of solvent use in industrial degreasing and cleaning
operations
Years
-1934-1954
-1955-1968
-1969-1978
~1979-1990s
Vapor degreasers
Trichloroethylene
(poorly controlled)
Trichloroethylene
(poorly controlled,
tightened in 1960s)
Trichloroethylene,
(better controlled)
1,1,1 -Trichloroethane
(replaced
trichloroethylene)
Cold dip tanks
Stoddard solvent*
Trichloroethylene
(replaced some
Stoddard solvent)
Trichloroethylene,
Stoddard solvent
1,1,1-
Trichloroethane
(replaced
trichloroethylene),
Stoddard solvent
Rag or brush and bucket on bench top
Stoddard solvent (general use), alcohols
(electronics shop), carbon tetrachloride
(instrument shop).
Stoddard solvent, trichloroethylene
(replaced some Stoddard solvent),
perchloroethylene, 1,1,1 -trichloroethane
(replaced carbon tetrachloride, alcohols,
ketones).
Trichloroethylene, perchloroethylene,
1,1,1 -trichloroethane, alcohols, ketones,
Stoddard solvent.
1,1,1 -Trichloroethane, perchloroethylene,
alcohols, ketones, Stoddard solvent.
15
16
17
18
19
20
21
22
23
24
* A mixture of straight and branched chain paraffins (48%), naphthenes (38%), and aromatic hydrocarbons (14%).
Source: Stewart and Dosemeci (2005).
2.4.2.2. Consumer Exposure
Consumer products reported to contain TCE include wood stains, varnishes, and finishes;
lubricants; adhesives; typewriter correction fluids; paint removers; and cleaners (ATSDR,
1997a). Use of TCE has been discontinued in some consumer products (i.e., as an inhalation
anesthetic, fumigant, and an extractant for decaffeinating coffee) (ATSDR, 1997a).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-23 DRAFT— DO NOT CITE OR QUOTE
-------
1
2
3
4
5
2.4.3. Exposure Standards
Table 2-15 summarizes the federal regulations limiting TCE exposure.
Table 2-15. TCE standards
Standard
OSHA Permissible Exposure Limit: Table Z-2
8 -hour time -weighted average.
OSHA Permissible Exposure Limit: Table Z-2
Acceptable ceiling concentration (this cannot
be exceeded for any time period during an
8-hour shift except as allowed in the
maximum peak standard below).
OSHA Permissible Exposure Limit: Table Z-2
Acceptable maximum peak above the
acceptable ceiling concentration for an 8 -hour
shift. Maximum Duration: 5 minutes in any
2 hours.
MCL under the Safe Drinking Water Act.
FDA Tolerances for
decaffeinated ground coffee
decaffeinated soluble (instant) coffee
extract spice oleoresins.
Value
lOOppm
(538 mg/m3)
200 ppm
(1076 mg/m3)
300 ppm
(16 14 mg/m3)
5 ppb (5 ug/L)
25 ppm (25 ug/g)
10 ppm (10 ug/g)
30 ppm (30 ug/g)
Reference
29 CFR 1910.1000 (7/1/2000)
29 CFR 1910.1000 (7/1/2000)
29 CFR 1910.1000 (7/1/2000)
40 CFR 141. 161
21 CFR 173.290 (4/1/2000)
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
2.5. EXPOSURE SUMMARY
TCE is a volatile compound with moderate water solubility. Most TCE produced today
is used for metal degreasing. The highest environmental releases are to the air. Ambient air
monitoring data suggests that levels have remained fairly constant since 1999 at about 0.3 ug/m3.
Indoor levels are commonly 3 or more times higher than outdoor levels due to releases from
building materials and consumer products. TCE is among the most common groundwater
contaminants and the median level based on a large survey by USGS for 1985-2001 is
0.15 ug/L. It has also been detected in a wide variety of foods in the 1-100 ug/kg range. None
of the environmental sampling has been done using statistically based national surveys.
However, a substantial amount of air and groundwater data have been collected allowing
reasonably well supported estimates of typical daily intakes by the general population:
inhalation—13 ug/day and water ingestion—0.2 ug/day. The limited food data suggests an
intake of about 5 ug/day, but this must be considered preliminary.
Much higher exposures have occurred to various occupational groups. For example, past
studies of aircraft workers have shown short term peak exposures in the hundreds of ppm
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 (>540,000 ug/m3) and long term exposures in the low tens of ppm (>54,000 ug/m3).
2 Occupational exposures have likely decreased in recent years due to better release controls and
3 improvements in worker protection.
4 Preliminary exposure estimates were presented for a variety of TCE related compounds
5 which include metabolites of TCE and other parent compounds that produce similar metabolites.
6 Exposure to the TCE related compounds can alter or enhance TCE's metabolism and toxicity by
7 generating higher internal metabolite concentrations than would result from TCE exposure by
8 itself. The preliminary estimates suggest that exposures to most of the TCE related compounds
9 are comparable to or greater than TCE itself.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 2-25 DRAFT—DO NOT CITE OR QUOTE
-------
1 3. TOXICOKINETICS
2
3
4 Trichloroethylene (TCE) is a lipophilic compound that readily crosses biological
5 membranes. Exposures may occur via the oral, dermal, and inhalation route, with evidence for
6 systemic availability from each route. TCE is rapidly and nearly completely absorbed from the
7 gut following oral administration, and studies with animals indicate that exposure vehicle may
8 impact the time-course of absorption: oily vehicles may delay absorption whereas aqueous
9 vehicles result in a more rapid increase in blood concentrations.
10 Following absorption to the systemic circulation, TCE distributes from blood to solid
11 tissues by each organ's solubility. This process is mainly determined by the blood:tissue
12 partition coefficients, which are largely established by tissue lipid content. Adipose partitioning
13 is high, adipose tissue may serve as a reservoir for TCE, and accumulation into adipose tissue
14 may prolong internal exposures. TCE attains high concentrations relative to blood in the brain,
15 kidney, and liver—all of which are important target organs of toxicity. TCE is cleared via
16 metabolism mainly in three organs: the kidney, liver, and lungs.
17 The metabolism of TCE is an important determinant of its toxicity. Metabolites are
18 generally thought to be responsible for toxicity—especially for the liver and kidney. Initially,
19 TCE may be oxidized via cytochrome P450 (CYP) xenobiotic metabolizing isozymes or
20 conjugated with glutathione by glutathione S-transferase enzymes. While CYP2E1 is generally
21 accepted to be the CYP form most responsible for TCE oxidation at low concentrations, others
22 forms may also contribute, though their contributions may be more important at higher, rather
23 than lower, environmentally-relevant exposures.
24 Once absorbed, TCE is excreted primarily either in breath as unchanged TCE or carbon
25 dioxide (CO2), or in urine as metabolites. Minor routes of elimination include excretion of
26 metabolites in saliva, sweat, and feces. Following oral administration or upon cessation of
27 inhalation exposure, exhalation of unmetabolized TCE is a major elimination pathway. Initially,
28 elimination of TCE upon cessation of inhalation exposure demonstrates a steep concentration-
29 time profile: TCE is rapidly eliminated in the minutes and hours postexposure, and then the rate
30 of elimination via exhalation decreases. Following oral or inhalation exposure, urinary
31 elimination of parent TCE is minimal, with urinary elimination of the metabolites trichloroacetic
32 acid and trichloroethanol accounting for the bulk of the absorbed dose of TCE.
33 Sections 3.1-3.4 below describe the absorption, distribution, metabolism, and excretion
34 of TCE and its metabolites in greater detail. Section 3.5 then discusses physiologically based
35 pharmacokinetic modeling of TCE and its metabolites.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-1 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.1. ABSORPTION
2 Trichloroethylene is a low-molecular-weight lipophilic solvent; these properties explain
3 its rapid transfer from environmental media into the systemic circulation after exposure. As
4 discussed below, it is readily absorbed into the bloodstream following exposure via oral
5 ingestion and inhalation, with more limited data indicating dermal penetration.
6
7 3.1.1. Oral
8 Available reports on human exposure to TCE via the oral route are largely restricted to
9 case reports of occupational or intentional (suicidal) ingestions and suggest significant gastric
10 absorption (e.g., Perbellini et al., 1991; Yoshida et al., 1996; Briining et al., 1998). Clinical
11 symptoms attributable to TCE or metabolites were observed in these individuals within a few
12 hours of ingestion (such as lack of consciousness), indicating absorption of TCE. In addition,
13 TCE and metabolites were measured in blood or urine at the earliest times possible after
14 ingestion, typically upon hospital admission, while urinary excretion of TCE metabolites was
15 followed for several days following exposure. Therefore, based on these reports, it is likely that
16 TCE is readily absorbed in the gastrointestinal tract; however, the degree of absorption cannot be
17 confidently quantified because the ingested amounts are not known.
18 Experimental evidence in mice and rats supports rapid and extensive absorption of TCE,
19 although variables such as stomach contents, vehicle, and dose may affect the degree of gastric
20 absorption. D'Souza et al. (1985) reported on bioavailability and blood kinetics in fasted and
21 nonfasted male Sprague-Dawley rats following intragastric administration of TCE at 5-25 mg/kg
22 in 50% polyethylene glycol (PEG 400) in water. TCE rapidly appeared in peripheral blood (at
23 the initial 0.5 minutes sampling) of fasted and nonfasted rats with peak levels being attained
24 shortly thereafter (6-10 minutes), suggesting that absorption is not diffusion limited, especially
25 in fasted animals. The presence of food in the gastro-intestinal (GI) tract, however, seems to
26 influence TCE absorption based on findings in the nonfasted animals of lesser bioavailability
27 (60-80% vs. 90% in fasted rats), smaller peak blood levels (2-3 fold lower than nonfasted
28 animals), and a somewhat longer terminal half-life (ti/2) (174 vs. 112 minutes in fasted rats).
29 Studies by Prout et al. (1985) and Dekant et al. (1986a) have shown that up to 98% of
30 administered radiolabel was found in expired air and urine of rats and mice following gavage
31 administration of [14C]-radio labeled TCE ([14C]TCE). Prout et al. (1985) and Green and Prout
32 (1985) compared the degree of absorption, metabolites, and routes of elimination among two
33 strains each of male rats (Osborne-Mendel and Park Wistar) and male mice (B6C3F1 and Swiss-
34 Webster) following a single oral administration of 10, 500, or 1,000 [14C]TCE. Additional dose
35 groups of Osborne-Mendel male rats and B6C3F1 male mice also received a single oral dose of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 2,000 mg/kg [14C]TCE. At the lowest dose of 10 mg/kg, there were no major differences
2 between rats and mice in routes of excretion with most of the administered radiolabel (nearly
3 60-70%) being in the urine. At this dose, the expired air from all groups contained 1-4% of
4 unchanged TCE and 9-14% CC>2. Fecal elimination of the radiolabel ranged from 8.3% in
5 Osborne-Mendel rats to 24.1% in Park Wistar rats. However, at doses between 500 and 2,000
6 mg/kg, the rat progressively excreted a higher proportion of the radiolabel as unchanged TCE in
7 expired air such that 78% of the administered high dose was found in expired air (as unchanged
8 TCE) while only 13% was excreted in the urine.
9 Following exposure to a chemical by the oral route, distribution is determined by delivery
10 to the first organ encountered in the circulatory pathway—the liver (i.e., the first-pass effect),
11 where metabolism and elimination may limit the proportion that may reach extrahepatic organs.
12 Lee et al. (1996) evaluated the efficiency and dose-dependency of presystemic elimination of
13 TCE in male Sprague-Dawley rats following administration into the carotid artery, jugular vein,
14 hepatic portal vein, or the stomach of TCE (0.17, 0.33, 0.71, 2, 8, 16, or 64 mg/kg) in a 5%
15 aqueous Alkamus emulsion (polyethoxylated vegetable oil) in 0.9% saline. The first-pass
16 elimination, decreased from 57.5 to <1% with increasing dose (0.17-16 mg/kg) which implied
17 that hepatic TCE metabolism may be saturated at doses above 16 mg/kg in the male rat. At
18 doses of 16 mg/kg or higher, hepatic first-pass elimination was almost nonexistent indicating
19 that, at relatively large doses, virtually all of TCE passes through the liver without being
20 extracted (Lee et al., 1996). In addition to the hepatic first-pass elimination findings, pulmonary
21 extraction, which was relatively constant (at nearly 5-8% of dose) over the dose range, also
22 played a role in eliminating TCE.
23 In addition, oral absorption appears to be affected by both dose and vehicle used. The
24 majority of oral TCE studies have used either aqueous solution or corn oil as the dosing vehicle.
25 Most studies that relied on an aqueous vehicle delivered TCE as an emulsified suspension in
26 Tween 80® or PEG 400 in order to circumvent the water solubility problems. Lee et al.
27 (2000a, b) used Alkamuls (a polyethoxylated vegetable oil emulsion) to prepare a 5% aqueous
28 emulsion of TCE that was administered by gavage to male Sprague-Dawley rats. The findings
29 confirmed rapid TCE absorption but reported decreasing absorption rate constants (i.e., slower
30 absorption) with increasing gavage dose (2-432 mg/kg). The time to reach blood peak
31 concentrations increased with dose and ranged between 2 and 26 minutes postdosing. Other
32 pharmacokinetics data, including area under the blood concentration time curve (AUC) and
33 prolonged elevation of blood TCE levels at the high doses, indicated prolonged GI absorption
34 and delayed elimination due to metabolic saturation occurring at the higher TCE doses.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 A study by Withey et al. (1983) evaluated the effect of dosing TCE with corn oil versus
2 pure water as a vehicle by administering four volatile organic compounds separately in each
3 dosing vehicle to male Wistar rats. Based on its limited solubility in pure water, the dose for
4 TCE was selected at 18 mg/kg (administered in 5 mL/kg). Times to peak in blood reported for
5 TCE averaged 5.6 minutes when water was used. In comparison, the time to peak in blood was
6 much longer (approximately 100 minutes) when the oil vehicle was used and the peaks were
7 smaller, below the level of detection, and not reportable.
8 Time-course studies reporting times to peak in blood or other tissues have been
9 performed using both vehicles (Withey et al., 1983; Larson and Bull, 1992 a, b; D'Souza et al.,
10 1985; Green and Prout, 1985; Dekant et al., 1984). Related data for other solvents (Kim et al.,
11 1990; Dix et al., 1997; Lilly et al., 1994; Chieco et al., 1981) confirmed differences in TCE
12 absorption and peak height between the two administered vehicles. One study has also evaluated
13 the absorption of TCE from soil in rats (Kadry et al., 1991) and reported absorption within 16
14 hours for clay and 24 hours for sandy soil. In summary, these studies confirm that TCE is
15 relatively quickly absorbed from the stomach, and that absorption is dependent on vehicle used.
16
17 3.1.2. Inhalation
18 Trichloroethylene is a lipophilic volatile compound that is readily absorbed from inspired
19 air. Uptake from inhalation is rapid and the absorbed dose is proportional to exposure
20 concentration and duration, and pulmonary ventilation rate. Distribution into the body via
21 arterial blood leaving the lungs is determined by the net dose absorbed and eliminated by
22 metabolism in the lungs. Metabolic clearance in the lungs will be further discussed in
23 Section 3.3, below. In addition to metabolism, solubility in blood is the major determinant of the
24 TCE concentration in blood entering the heart and being distributed to the each body organ via
25 the arterial blood. The measure of TCE solubility in each organ is the partition coefficient, or the
26 concentration ratio between both organ phases of interest. The blood-to-air partition coefficient
27 (PC) quantifies the resulting concentration in blood leaving the lungs at equilibrium with
28 alveolar air. The value of the blood-to-air partition coefficient is used in physiologically based
29 pharmacokinetic (PBPK) modeling (see Section 3.5). The blood-to-air partition has been
30 measured in vitro using the same principles in different studies and found to range between
31 8.1-11.7 in humans and somewhat higher values in mice and rats (13.3-25.8) (see
32 Tables 3-1-3-2, and references therein).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-4 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-1 Blood:air PC values for humans
Blood: air partition
coefficient
8.1 + 1.8
8.11
9.13 + 1.73 [6.47-11]
9.5
9.77
9.92
11.15 + 0.74
[10.1-12.1]
11.2+1.8 [7.9-15]
11.0+ 1.6 [6.6-13. 5]
11.7+ 1.9 [6.7-16.8]
10.6 + 2.3 [3-14.4]
Reference/notes
Fiserova-Bergerova et al., 1984; mean + SD (SD converted from SE
based on n = 5)
Gargas et al., 1989; (n = 3-15)
Fisher et al., 1998; mean + SD [range] of females (n = 6)
Sato and Nakajima, 1979; (n = 1)
Koizumi, 1989
Satoetal., 1977; (w=l)
Fisher et al., 1998; mean + SD [range] of males (n = 7)
Mahle et al., 2007; mean + SD; 20 male pediatric patients aged 3-7
years [range; USAF, 2004]
Mahle et al., 2007; mean + SD; 18 female pediatric patients aged
3-17years [range; USAF, 2004]
Mahle et al., 2007; mean + SD; 32 male patients aged 23-82 years
[range; USAF, 2004]
Mahle et al., 2007; mean + SD; 27 female patients aged 23-82 years
[range; USAF, 2004]
3
4
SD = standard deviation, SE = standard error.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-5 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-2 Blood:air PC values for rats and mice
Blood: air partition
coefficient
Reference/notes
Rat
15 + 0.5
17.5
20.5 + 2.4
20.69 + 3.3
21.9
25.8
25.82+1.7
13.3 + 0.8 [11.6-15]
13.4+1.8 [11.8-17.2]
17.5 + 3.6 [11. 7-23.1]
21.8+ 1.9 [16.9-23. 5]
Fisher et al., 1989; mean + SD (SD converted from SE based on
w = 3)
Rodriguez et al., 2007
Barton et al., 1995; mean + SD (SD converted from SE based on
w = 4)
Simmons et al., 2002; mean + SD (n = 7-10)
Gargas et al., 1989 (n = 3-15)
Koizumi, 1989 (pooled n = 3)
Sato et al., 1977; mean + SD (n = 5)
Mahle et al., 2007; mean + SD; 10 PND 10 male rat pups [range;
USAF, 2004]
Mahle et al., 2007; mean + SD; 10 PND 10 female rat pups [range;
USAF, 2004]
Mahle et al., 2007; mean + SD; 9 adult male rats [range; USAF, 2004]
Mahle et al., 2007; mean + SD; 11 aged male rats [range; USAF,
2004]
Mouse
13.4
14.3
15.91
Fisher et al., 1991; male
Fisher et al., 1991; female
Abbas and Fisher, 1997
4
5
6
7
9
10
11
12
13
SD = standard deviation, SE = standard error, PND = postnatal day.
TCE enters the human body by inhalation quickly and at high concentrations may lead to
death (Coopman et al., 2003), unconsciousness, and acute kidney damage (Carrieri et al., 2007).
Controlled exposure studies in humans have shown absorption of TCE to approach a steady state
within a few hours after the start of inhalation exposure (Monster et al., 1976; Fernandez et al.,
1977; Vesterberg et al., 1976; Vesterberg and Astrand, 1976). Several studies have calculated
the net dose absorbed by measuring the difference between the inhaled concentration and the
exhaled air concentration. Soucek and Vlachova (1959) reported between 58-70% absorption of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-6 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
the amount inhaled for 5-hour exposures between 93-158 ppm. Bartonicek (1962) obtained an
average retention value of 58% after 5 hours of exposure to 186 ppm. Monster et al. (1976) also
took into account minute ventilation measured for each exposure, and calculated between
37-49% absorption in subjects exposed to 70 and 140 ppm. The impact of exercise, the increase
in workload, and its effect on breathing has also been measured in controlled inhalation
exposures. Astrand and Ovrum (1976) reported 50-58% uptake at rest and 25-46% uptake
during exercise from exposure at 100 or 200 ppm (540 or 1,080 mg/m3, respectively) of TCE for
30 minutes (see Table 3-3). These authors also monitored heart rate and pulmonary ventilation.
In contrast, Jakubowski and Wieczorek (1988) calculated about 40% retention in their human
volunteers exposed to TCE at 9 ppm (mean inspired concentration of 48-49 mg/m3) for 2 hours
at rest, with no change in retention during increase in workload due to exercise (see Table 3-4).
Table 3-3. Air and blood concentrations during exposure to TCE in humans
(Astrand and Ovrum, 1976)
TCE
cone.
(mg/m3)
540
540
540
540
540
540
1,080
1,080
1,080
1,080
1,080
1,080
Work
load
(watt)
0
0
50
50
50
50
0
0
50
50
100
150
Exposure
series
I
II
I
II
II
II
I
III
I
III
III
III
TCE concentration in
Alveolar
air
(mg/m3)
124 + 9
127+11
245 + 12
218 + 7
234+12
244+16
280+18
212 + 7
459 + 44
407 + 30
542 + 33
651 + 53
Arterial
blood
(mg/kg)
1.1 + 0.1
1.3 + 0.1
2.7 + 0.2
2.8 + 0.1
3.1 + 0.3
3.3 + 0.3
2.6 + 0.0
2.1 + 0.2
6.0 + 0.2
5.2 + 0.5
7.5 + 0.7
9.0+1.0
Venous
blood
(mg/kg)
0.6 + 0.1
0.5 + 0.1
1.7 + 0.4
1.8 + 0.3
2.2 + 0.4
2.2 + 0.4
1.4 + 0.3
1.2 + 0.1
3.3 + 0.8
2.9 + 0.7
4.8+1.1
7.4+1.1
Uptake as
%of
amount
available
53 + 2
52 + 2
40 + 2
46+1
39 + 2
37 + 2
50 + 2
58 + 2
45 + 2
51 + 3
36 + 3
25 + 5
Amount
taken up
(mg)
79 + 4
81+7
160 + 5
179 + 2
157 + 2
147 + 9
156+9
186 + 7
702 + 31
378 + 18
418 + 39
419 + 84
16
17
18
19
Series I consisted of 30-minute exposure periods of rest, rest, SOW and 50W; Series II consisted of
30-minute exposure periods of rest, SOW, SOW, SOW; Series III consisted of 30-minute exposure
periods of rest, SOW, 100W, 150W.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-7 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-4 Retention of inhaled TCE vapor in humans (Jakubowski and
Wieczorek, 1988)
Workload
Rest
25 W
SOW
75 W
Inspired concentration
(mg/m3)
48 + 3*
49 + 1.3
49 + 1.6
48 + 1.9
Pulmonary
ventilation (m3/hour)
0.65 + 0.07
1.30 + 0.14
1.53+0.13
1.87 + 0.14
Retention
0.40 + 0.05
0.40 + 0.05
0.42 + 0.06
0.41 + 0.06
Uptake
(mg/h)
12+1.1
25 + 2.9
31+2.8
37 + 4.8
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
*Mean + standard deviation, n = 6 adult males.
W = watts.
Environmental or occupational settings may results from a pattern of repeated exposure
to TCE. Monster et al. (1979) reported 70-ppm TCE exposures in volunteers for 4 hours for
5 consecutive days, averaging a total uptake of 450 mg per 4 hours exposure (see Table 3-5). In
dry-cleaning workers, Skender et al. (1991) reported initial blood concentrations of 0.38 |imol/L,
increasing to 3.4 jimol/L 2 days after. Results of these studies support rapid absorption of TCE
via inhalation.
Table 3-5. Uptake of TCE in human volunteers following 4 hour exposure to
70 ppm (Monster et al., 1979)
A
B
C
D
E
Mean
BW
(kg)
80
82
82
67
90
MV (L/min)
9.8 + 0.4
12.0 + 0.7
10.9 + 0.8
11.8 + 0.8
11.0 + 0.7
% Retained
45 + 0.8
44 + 0.9
49+1.2
35 + 2.6
46+1.1
Uptake
(mg/day)
404 + 23
485 + 35
493 + 28
385 + 38
481+25
Uptake (mg/kg/d)
5.1
5.9
6.0
5.7
5.3
5.6 + 0.4
20
21
22
23
BW = body weight.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-8 DRAFT—DO NOT CITE OR QUOTE
-------
1 Direct measurement of retention after inhalation exposure in rodents is more difficult
2 because exhaled breath concentrations are challenging to obtain. The only available data are
3 from Dallas et al. (1991), who designed a nose-only exposure system for rats using a facemask
4 equipped with one-way breathing valves to obtain measurements of TCE in inspired and exhaled
5 air. In addition, indwelling carotid artery cannulae were surgically implanted to facilitate the
6 simultaneous collection of blood. After a 1-hour acclimatization period, rats were exposed to 50-
7 or 500-ppm TCE for 2 hours and the time course of TCE in blood and expired air was measured
8 during and for 3 hours following exposure. When air concentration data were analyzed to reveal
9 absorbed dose (minute volume multiplied by the concentration difference between inspired and
10 exhaled breath), it was demonstrated that the fractional absorption of either concentration was
11 more than 90% during the initial 5 minutes of exposure. Fractional absorption then decreased to
12 69 and 71% for the 50 and 500-ppm groups during the second hour of exposure. Cumulative
13 uptake appeared linear with respect to time over the 2-hour exposure, resulting in absorbed doses
14 of 8.4 mg/kg and 73.3 mg/kg in rats exposed to 50 and 500 ppm, respectively. Given the 10-fold
15 difference in inspired concentration and the 8.7-fold difference in uptake, the authors interpreted
16 this information to indicate that metabolic saturation occurred at some concentration below
17 500 ppm. In comparing the absorbed doses to those developed for the 70-ppm-exposed human
18 (see Monster et al., 1979), Dallas et al. (1991) concluded that on a systemic dose (mg/kg) basis,
19 rats receive a much higher TCE dose from a given inhalation exposure than do humans. In
20 particular, using the results cited above, the absorption per ppm-hour was 0.084 and
21 0.073 mg/kg-ppm-hour at 50 and 500 ppm in rats (Dallas et al., 1991) and
22 0.019 mg/kg-ppm-hour at 70 ppm in humans (Monster et al., 1979)—a difference of around
23 4-fold. However, rats have about a 10-fold higher alveolar ventilation rate per unit body weight
24 than humans (Brown et al., 1997), which more than accounts for the observed increase in
25 absorption.
26 Other experiments, such as closed-chamber gas uptake experiments or blood
27 concentration measurements following open-chamber (fixed concentration) experiments,
28 measure absorption indirectly but are consistent with significant retention. Closed-chamber
29 gas-uptake methods (Gargas et al., 1988) place laboratory animals or in vitro preparations into
30 sealed systems in which a known amount of TCE is injected to produce a predetermined
31 chamber concentration. As the animal retains a quantity of TCE inside its body, due to
32 metabolism, the closed-chamber concentration decreases with time when compared to the start of
33 exposure. Many different studies have made use of this technique in both rats and mice to
34 calculate total TCE metabolism (i.e., Andersen, 1987; Fisher et al., 1991; Simmons et al., 2002).
35 This inhalation technique is combined with PBPK modeling to calculate metabolic parameters,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-9 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
and the results of these studies are consistent with rapid absorption of TCE via the respiratory
tract. Figure 3-1 shows and example from Simmons et al. (2002), in Long Evans rats, that
demonstrates an immediate decline in chamber concentrations of TCE indicating absorption,
with multiple initial concentrations needed for each metabolic calculation. At concentrations
below metabolic saturation, a secondary phase of uptake appears, after 1 hour from starting the
exposure, indicative of metabolism. At concentrations above 1,000 ppm, metabolism appears
saturated, with time course curves having a flat phase after absorption. At intermediate
concentrations, between 100-1,000 ppm, the secondary phase of uptake appears after
distribution as continued decreases in chamber concentration as metabolism proceeds. Using a
combination of experiments that include both metabolic linear decline and saturation obtained by
using different initial concentrations, both components of metabolism can be estimated from the
gas uptake curves, as shown in Figure 3-1.
10000 q
? :
Q.
Q.
C
O
*• 1000,
:_
*•
c
o
o
c
o
0
o 100 --
J3 :
E :
n
f
o
UJ
0
"- 10 1
+ 3000 ppm
a 1000 ppm
++ A 500 ppm
++++++++++| o 100 ppm
+++++++++ +++++++++++
n
n
A n Dn
A A Dnr"i i— 1 1—!
A ^ l~ln
A A . 1— 1 D l~l|~| n i— i
A A l— l D n r~|
^AA ^Dn
o AAAAAA Dl=l
^>k A A yi
°°Oo/s AA^A
^O^s ^AA
^O/s ^
o^*
O A
^O/s
v
i i i i i
01234567
Time (hr)
Figure 3-1 Gas uptake data from closed-chamber exposure of rats to TCE.
Symbols represent measured chamber concentrations. Source: Simmons et
al. (2002).
Several other studies in humans and rodents have measured blood concentrations of TCE
or metabolites and urinary excretion of metabolites during and after inhalation exposure (e.g.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-10 DRAFT—DO NOT CITE OR QUOTE
-------
1 Fisher et al., 1998, 1991, 1990; Filser and Bolt, 1979). While qualitatively indicative of
2 absorption, blood concentrations are also determined by metabolism, distribution, and excretion,
3 so comparisons between species may reflect similarities or differences in any of the absorption,
4 distribution, metabolism, and excretion (ADME) processes.
5
6 3.1.3. Dermal
7 Skin membrane is believed to present a diffusional barrier for entrance of the chemical
8 into the body, and TCE absorption can be quantified using a permeability rate or permeability
9 constant, though not all studies performed such a calculation. Absorption through the skin has
10 been shown to be rapid by both vapor and liquid TCE contact with the skin. Human dermal
11 absorption of TCE vapors was investigated by Kezic et al. (2000). Human volunteers were
12 exposed to 3.18 x 104 ppm around each enclosed arm for 20 minutes. Adsorption was found to
13 be rapid (within 5 minutes), reaching a peak in exhaled breath around 30 minutes, with a
14 calculated dermal penetration rate averaging 0.049 cm/hour for TCE vapors.
15 With respect to dermal penetration of liquid TCE, Nakai et al. (1999) used surgically
16 removed skin samples exposed to TCE in aqueous solution in a chamber designed to measure the
17 difference between incoming and outgoing [14C]TCE. The in vitro permeability constant
18 calculated by these researchers averaged 0.12 cm/hour. In vivo, Sato and Nakajima (1978)
19 exposed adult male volunteers dermally to liquid TCE for 30 minutes, with exhaled TCE
20 appearing at the initial sampling time of 5 minutes after start of exposure, with a maximum
21 observed at 15 minutes. In Kezic et al. (2001), human volunteers were exposed dermally for
22 3 minutes to neat liquid TCE, with TCE detected in exhaled breath at the first sampling point of
23 3 minutes, and maximal concentrations observed at 5 minutes. Skin irritancy was reported in all
24 subjects, which may have increased absorption. A dermal flux of 430 + 295 (mean + standard
25 error [SE]) nmol/cm2/minute was reported in these subjects, suggesting high interindividual
26 variability.
27 Another species where dermal absorption for TCE has been reported is in guinea pigs.
28 Jakobson et al. (1982) applied liquid TCE to the shaved backs of guinea pigs and reported peak
29 blood TCE levels at 20 minutes after initiation of exposure. Bogen et al. (1992) estimated
30 permeability constants for dermal absorption of TCE in hairless guinea pigs between
31 0.16-0.47 mL/cm2/hour across a range of concentrations (19-100,000 ppm).
32
33 3.2. DISTRIBUTION AND BODY BURDEN
34 TCE crosses biological membranes and quickly results in rapid systemic distribution to
35 tissues—regardless of the route of exposure. In humans, in vivo studies of tissue distribution are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-11 DRAFT—DO NOT CITE OR QUOTE
-------
1 limited to tissues taken from autopsies following accidental poisonings or from surgical patients
2 exposed environmentally, so the level of exposure is typically unknown. Tissue levels reported
3 after autopsy show wide systemic distribution across all tested tissues, including the brain,
4 muscle, heart, kidney, lung, and liver (Ford et al., 1995; De Baere et al., 1997; Dehon et al.,
5 2000; Coopman et al., 2003). However, the reported levels themselves are difficult to interpret
6 because of the high exposures and differences in sampling protocols. In addition, human
7 populations exposed environmentally show detectable levels of TCE across different tissues,
8 including the liver, brain, kidney, and adipose tissues (McConnell et al., 1975; Pellizzari et al.,
9 1982; Kroneld, 1989).
10 In addition, TCE vapors have been shown to cross the human placenta during childbirth
11 (Laham, 1970), with experiments in rats confirming this finding (Withey and Karpinski, 1985).
12 In particular, Laham (1970) reported determinations of TCE concentrations in maternal and fetal
13 blood following administration of TCE vapors (concentration unreported) intermittently and at
14 birth (see Table 3-6). TCE was present in all samples of fetal blood, with ratios of
15 concentrations in fetal maternal blood ranging from approximately 0.5 to approximately 2. The
16 concentration ratio was less than 1.0 in six pairs, greater than 1 in 3 pairs, and approximately 1 in
17 1 pair; in general, higher ratios were observed at maternal concentrations below
18 2.25 mg/100 mL. Because no details of exposure concentration, duration, or time postexposure
19 were given for samples taken, these results are of minimal quantitative value, but they do
20 demonstrate the placental transfer of TCE in humans. Withey and Karpinski (1985) exposed
21 pregnant rats to TCE vapors (302, 1,040, 1,559, or 2,088 ppm for 5 hours) on gestation Day 17
22 and concentrations of TCE in maternal and fetal blood were determined. At all concentrations,
23 TCE concentration in fetal blood was approximately one-third the concentration in
24 corresponding maternal blood. Maternal blood concentrations approximated 15, 60, 80, and
25 110 jig/gram blood. When the position along the uterine horn was examined, TCE
26 concentrations in fetal blood decreased toward the tip of the uterine horn.
27 TCE appears to also distribute to mammary tissues and is excreted in milk.
28 Pellizzari et al. (1982) conducted a survey of environmental contaminants in human milk using
29 samples from cities in the northeastern region of the United States and one in the southern
30 region. No details of times postpartum, milk lipid content, or TCE concentration in milk or
31 blood are reported, but TCE was detected in 8 milk samples taken from 42 lactating women.
32 Fisher et al. (1990) exposed lactating rats to 600-ppm TCE for 4 hours and collected milk
33 immediately following the cessation of exposure. TCE was clearly detectable in milk, and, from
34 a visual interpretation of the graphic display of their results, concentrations of TCE in milk
35 approximated 110 |ig/mL milk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-12 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-6 Concentrations of TCE in maternal and fetal blood at birth
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
TCE concentration in
blood (mg/100 mL)
Maternal
4.6
3.8
8
5.4
7.6
3.8
2
2.25
0.67
1.05
Fetal
2.4
2.2
5
3.6
5.2
o o
J.J
1.9
O
1
2
Ratio of
concentrations
fetalrmaternal
0.52
0.58
0.63
0.67
0.68
0.87
0.95
1.33
1.49
1.90
Source: Laham (1970).
In rodents, detailed tissue distribution experiments have been performed using different
routes of administration (Savolainen et al., 1977; Pfaffenberger et al., 1980; Abbas and Fisher,
1997; Greenberg et al., 1999; Simmons et al., 2002; Keys et al., 2003). Savolainen et al. (1977)
exposed adult male rats to 200-ppm TCE for 6 hours/day for a total of 5 days. Concentrations of
TCE in the blood, brain, liver, lung, and perirenal fat were measured 17 hours after cessation of
exposure on the fourth day and after 2, 3, 4, and 6 hours of exposure on the fifth day (see
Table 3-7). TCE appeared to be rapidly absorbed into blood and distributed to brain, liver, lungs,
and perirenal fat. TCE concentrations in these tissues reached near-maximal values within
2 hours of initiation of exposure on the fifth day. Pfaffenberger et al. (1980) dosed rats by
gavage with 1 or 10 mg TCE/kg/day in corn oil for 25 days to evaluate the distribution from
serum to adipose tissue. During the exposure period, concentrations of TCE in serum were
below the limit of detection (1 |ig/L) and were 280 and 20,000 ng per gram of fat in the 1 and
10 mg/day dose groups, respectively. Abbas and Fisher (1997) and Greenberg et al. (1999)
measured tissue concentrations in the liver, lung, kidney, and fat of mice administered TCE by
gavage (300-2,000 mg/kg) and by inhalation exposure (100 or 600 ppm for 4 hours). In a study
to investigate the effects of TCE on neurological function, Simmons et al. (2002) conducted
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-13 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
pharmacokinetic experiments in rats exposed to 200, 2,000, or 4,000 ppm TCE vapors for 1 hour.
Time-course data were collected on blood, liver, brain, and fat. The data were used to develop a
PBPK model to explore the relationship between internal dose and neurological effect. Keys et
al. (2003), exposed groups of rats to TCE vapors of 50 or 500 ppm for 2 hours and sacrificed at
different time points during exposure. In addition to inhalation, this study also includes oral
gavage and intra-arterial dosing, with the following time course measured: liver, fat, muscle,
blood, GI, brain, kidney, heart, lung, and spleen. These pharmacokinetic data were presented
with an updated PBPK model for all routes.
Table 3-7. Distribution of TCE to rat tissues3 following inhalation exposure
(Savolainen et al., 1977)
Exposure
on 5th day
Ob
2
O
4
6
Tissue (concentration in nmol/gram tissue)
Cerebrum
0
9.9 + 2.7
7.3 + 2.2
7.2+ 1.7
7.4 + 2.1
Cerebellum
0
11.7 + 4.2
8.8 + 2.1
7.6 + 0.5
9.5+2.5
Lung
0.08
4.9 _ 0.3
5.5 + 1.4
5.8+ 1.1
5.6 + 0.5
Liver
0.04
3.6
5.5 + 1.7
2.5 + 1.4
2.4 + 0.2
Perirenal fat
0.23 + 0.09
65.9 + 1.2
69.3+3.3
69.5 + 6.3
75.4+ 14.9
Blood
0.35 + 0.1
7.5 + 1.6
6.6 + 0.9
6.0 + 0.2
6.8 + 1.2
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
aData presented as mean of 2 determinations + range.
bSample taken 17 hours following cessation of exposure on Day 4.
Besides the route of administration, another important factor contributing to body
distribution is the individual solubility of the chemical in each organ, as measured by a partition
coefficient. For volatile compounds, partition coefficients are measured in vitro using the vial
equilibration technique to determine the ratio of concentrations between organ and air at
equilibrium. Table 3-8 reports values developed by several investigators from mouse, rat, and
human tissues. In humans, partition coefficients in the following tissues have been measured:
brain, fat, kidney, liver, lung, and muscle; but the organ having the highest TCE partition
coefficient is fat (63-70), while the lowest is the lung (0.5-1.7). The adipose tissue also has the
highest measured value in rodents, and is one of the considerations needed to be accounted for
when extrapolating across species. However, the rat adipose partition coefficient value is
smaller (23-36), when compared to humans, that is, TCE is less lipophilic in rats than humans.
For the mouse, the measured fat partition coefficient averages 36, ranging between rats and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 humans. The value of the partition coefficient plays a role in distribution for each organ and is
2 computationally described in computer simulations using a PBPK model. Due to its high
3 lipophilicity in fat, as compared to blood, the adipose tissue behaves as a storage compartment
4 for this chemical, affecting the slower component of the chemical's distribution. For example
5 Monster et al. (1979) reported that, following repeated inhalation exposures to TCE, TCE
6 concentrations in expired breath postexposure were highest for the subject with the greatest
7 amount of adipose tissue (adipose tissue mass ranged 3.5-fold among subjects). The intersubject
8 range in TCE concentration in exhaled breath increased from approximately 2-fold at 20 hours to
9 approximately 10-fold 140 hours postexposure. Notably, they reported that this difference was
10 not due to differences in uptake, as body weight and lean body mass were most closely
11 associated with TCE retention. Thus, adipose tissue may play an important role in postexposure
12 distribution, but does not affect its rapid absorption.
13 Mahle et al. (2007) reported age-dependent differences in partition coefficients in rats,
14 (see Table 3-9) that can have implications as to life-stage-dependent differences in tissue TCE
15 distribution. To investigate the potential impact of these differences, Rodriguez et al. (2007)
16 developed models for the postnatal Day 10 rat pup; the adult and the aged rat, including
17 age-specific tissue volumes and blood flows; and age-scaled metabolic constants. The models
18 predict similar uptake profiles for the adult and the aged rat during a 6-hour exposure to
19 500 ppm; uptake by the postnatal day (PND) 10 rat was higher (see Table 3-10). The effect was
20 heavily dependent on age-dependent changes in anatomical and physiological parameters
21 (alveolar ventilation rates and metabolic rates); age-dependent differences in partition coefficient
22 values had minimal impact on predicted differences in uptake.
23 Finally, TCE binding to tissues or cellular components within tissues can affect overall
24 pharmacokinetics. The binding of a chemical to plasma proteins, for example, affects the
25 availability of the chemical to other organs and the calculation of the total half-life. However,
26 most studies have evaluated binding using [14C]TCE, from which one cannot distinguish binding
27 of TCE from binding of TCE metabolites. Nonetheless, several studies have demonstrated
28 binding of TCE-derived radiolabel to cellular components (Moslen et al., 1977; Mazzullo et al.,
29 1992). Bolt and Filser (1977) examined the total amount irreversibly bound to tissues following
30 9-, 100-, and 1,000-ppm exposures via inhalation in closed chambers. The largest percent of in
31 vivo radioactivity taken up occurred in the liver; albumin is the protein favored for binding (see
32 Table 3-11). Bannerjee and van Duuren (1978) evaluated the in vitro binding of TCE to
33 microsomal proteins from the liver, lung, kidney, and stomachs in rats and mice. In both rats and
34 mice, radioactivity was similar in stomach and lung, but about 30% lower in kidney and liver.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-15 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-8 Tissuerblood partition coefficient values for TCE
Species/
tissue
TCE partition coefficient
Tissue:blood
Tissue:air
References
Human
Brain
Fat
Kidney
Liver
Lung
Muscle
2.62
63.8-70.2
1.3-1.8
3.6-5.9
0.48-1.7
1.7-2.4
21.2
583-674.4
12-14.7
29.4-54
4.4-13.6
15.3-19.2
Fiserova-Bergerova et al., 1984
Sato et al., 1977; Fiserova-Bergerova et al., 1984; Fisher et al.,
1998
Fiserova-Bergerova etal., 1984; Fisher etal., 1998
Fiserova-Bergerova etal., 1984; Fisher etal., 1998
Fiserova-Bergerova etal., 1984; Fisher etal., 1998
Fiserova-Bergerova etal., 1984; Fisher etal., 1998
Rat
Brain
Fat
Heart
Kidney
Liver
Lung
Muscle
Spleen
Testis
Milk
0.71-1.29
22.7-36.1
1.1
1.0-1.55
1.03-2.43
1.03
0.46-0.84
1.15
0.71
7.10
14.6-33.3
447-661
28.4
17.7-40
20.5-62.7
26.6
6.9-21.6
29.7
18.3
N.R.
Sato et al., 1977; Simmons et al., 2002; Rodriguez et al., 2007
Gargas et al., 1989; Sato et al., 1977; Simmons et al., 2002;
Rodriguez et al. 2007; Fisher et al., 1989, Koizumi, 1989;
Barton etal., 1995
Sato etal. 1977
Sato et al., 1977; Barton et al., 1995; Rodriguez et al., 2007
Gargas et al., 1989; Sato et al., 1977; Simmons et al., 2002;
Rodriguez et al., 2007; Fisher et al., 1989; Koizumi, 1989;
Barton etal., 1995
Sato etal., 1977
Gargas et al., 1989; Sato et al., 1977; Simmons et al., 2002;
Rodriguez et al., 2007; Fisher et al., 1989; Koizumi, 1989;
Barton etal., 1995
Sato etal., 1977
Sato etal., 1977
Fisher etal., 1990
Mouse
Fat
Kidney
Liver
Lung
Muscle
36.4
2.1
1.62
2.6
2.36
578.8
32.9
23.2
41.5
37.5
Abbas and Fisher, 1997
Abbas and Fisher, 1997
Fisher etal., 1991
Abbas and Fisher, 1997
Abbas and Fisher, 1997
N.R. = not reported.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-16 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-9 Age-dependence of tissuerair partition coefficients in rats
Age
PND10 male
PND 10 female
Adult male
Aged male
Liver
22.1+2.3
21.2+1.7
20.5 + 4.0
34.8 + 8.7^
Kidney
15.2 + 1.3
15.0 + 1.1
17.6 + 3. 9a
19.9 + 3.4a
Fat
398.7 + 89.2
424.5 + 67.5
631.4 + 43. la
757.5+ 48. 3°*
Muscle
43.9 + 11.0
48.6 + 17.3
12.6 + 4.3
26.4 + 10.3^
Brain
11.0 + 0.6
11.6+1.2
17.4 + 2.6
25.0 + 2.0a'b
3
4
5
6
7
8
9
10
11
12
13
14
"Statistically significant (p < 0.05) difference between either the adult or aged partition coefficient and the PND 10
male partition coefficient.
bStatistically significant (p < 0.05) difference between aged and adult partition coefficient.
Data are mean + standard deviation; n = 10, adult male and pooled male and female litters; 11, aged males.
Source: Mahle et al. (2007).
Table 3-10. Predicted maximal concentrations of TCE in rat blood
following a 6-hour inhalation exposure (Rodriguez et al., 2007)
Age
PND 10
Adult
Aged
Exposure concentration
50 ppm
Predicted peak
concentration
(mg/L) in:a
Venous
blood
3.0
0.8
0.8
Brain
2.6
1.0
1.2
Predicted
time to reach
90% of steady
state (hour)b
4.1
3.5
6.7
500 ppm
Predicted peak
concentration
(mg/L) in:a
Venous
blood
33
22
21
Brain
28
23
26
Predicted
time to reach
90% of steady
state (hour)b
4.2
11.9
23.3
15
16
17
"During a 6 hour exposure.
'Under continuous exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-17 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-11. Tissue distribution of TCE metabolites following inhalation
exposure
Tissue*
Lung
Liver
Spleen
Kidney
Small
intestine
Muscle
Percent of radioactivity taken up/g tissue
TCE = 9 ppm,
n = 4
Total
metabolites
0.23 + 0.026
0.77 + 0.059
0.14 + 0.015
0.37 + 0.005
0.41 + 0.058
0.11 + 0.005
Irreversibly
bound
0.06 + 0.002
0.28 + 0.027
0.05 + 0.002
0.09 + 0.007
0.05 + 0.010
0.014 + 0.001
TCE = 100 ppm,
n = 4
Total
metabolites
0.24 + 0.025
0.68 + 0.073
0.15 + 0.001
0.40 + 0.029
0.38 + 0.062
0.11 + 0.013
Irreversibly
bound
0.06 + 0.006
0.27 + 0.019
0.05 + 0.004
0.09 + 0.007
0.07 + 0.008
0.012 + 0.001
TCE = 1,000 ppm,
n = 3
Total
metabolites
0.22 + 0.055
0.88 + 0.046
0.15 + 0.006
0.39 + 0.045
0.28 + 0.015
0.10 + 0.011
Irreversibly
bound
0.1 + 0.003
0.48 + 0.020
0.08 + 0.003
0.14 + 0.016
0.09 + 0.015
0.027 + 0.003
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
*Male Wistar rats, 250 g.
n = number of animals.
Values shown are means + standard deviation.
Source: Bolt and Filser (1977).
Based on studies of the effects of metabolizing enzyme induction on binding, there is
some evidence that a major contributor to the observed binding is from TCE metabolites rather
than from TCE itself. Dekant et al. (1986a) studied the effect of enzyme modulation on the
binding of radiolabel from [14C]TCE by comparing tissue binding after administration of
200 mg/kg via oral gavage in corn oil between control (naive) rats and rats pretreated with
phenobarbital (a known inducer of CYP2B family) or arochlor 1254 (a known inducer of both
CYP1A and CYP2B families of isoenzymes) (see Table 3-12). The results indicate that
induction of total cytochromes P-450 content by 3- to 4-fold resulted in nearly 10-fold increase
in radioactivity (decays per minute; DPM) bound in liver and kidney. By contrast, Mazzullo et
al. (1992) reported that, phenobarbital pretreatment did not result in consistent or marked
alterations of in vivo binding of radiolabel to DNA, RNA, or protein in rats and mice at 22 hours
after an intraperitoneal (i.p.) injection of [14C]TCE. On the other hand, in vitro experiments by
Mazzullo et al. (1992) reported reduction of TCE-radiolabel binding to calf thymus DNA with
introduction of a CYP inhibitor into incubations containing rat liver microsomal protein.
Moreover, increase/decrease of glutathione (GSH) levels in incubations containing lung
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-18 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
cytosolic protein led to a parallel increase/decrease in TCE-radiolabel binding to calf thymus
DNA.
Table 3-12 Binding of 14C from [14C]TCE in rat liver and kidney at 72 hours
after oral administration of 200 mg/kg [14C]TCE (Dekant et al., 1986a)
Tissue
Liver
Kidney
DPM/gram tissue
Untreated
850 + 100
680+100
Phenobarbital
9,300 + 1,100
5,700 + 900
Arochlor 1254
8,700 + 1,000
7,300 + 800
3.3. METABOLISM
This section focuses on both in vivo and in vitro studies of the biotransformation of
trichloroethylene, identifying metabolites that are deemed significant for assessing toxicity and
carcinogenicity. In addition, metabolism studies may be used to evaluate the flux of parent
compound through the known metabolic pathways. Sex-, species-, and interindividual
differences in the metabolism of TCE are discussed, as are factors that possibly contribute to this
variability. Additional discussion of variability and susceptibility is presented in Section 4.10.
3.3.1. Introduction
The metabolism of TCE has been studied mostly in mice, rats, and humans and has been
extensively reviewed (U.S. EPA, 1985, 2001; Lash et al., 2000a; IARC, 1995). It is now well
accepted that TCE is metabolized in laboratory animals and in humans through at least two
distinct pathways: (1) oxidative metabolism via the cytochrome P450 mixed-function oxidase
system and (2) GSH conjugation followed by subsequent further biotransformation and
processing, either through the cysteine conjugate beta lyase pathway or by other enzymes (Lash
et al., 2000b). While the flux through the conjugative pathway is less, quantitatively, than the
flux through oxidation (Bloemen et al., 2001), GSH conjugation is an important route
lexicologically, giving rise to relatively potent toxic biotransformation products
(Elfarra et al., 1986a, b).
Information about metabolism is important because, as discussed extensively in
Chapter 4, certain metabolites are thought to cause one or more of the same acute and chronic
toxic effects, including carcinogenicity, as TCE. Thus, in many of these cases, the toxicity of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-19 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCE is generally considered to reside primarily in its metabolites rather than in the parent
2 compound itself.
O
4 3.3.2. Extent of Metabolism
5 TCE is extensively metabolized in animals and humans. The most comprehensive
6 mass-balance studies are in mice and rats (Dekant et al., 1984; Dekant et al., 1986a, b; Green and
7 Prout, 1985; Prout et al., 1985) in which [14C]TCE is administered by oral gavage at doses of 2
8 to 2,000 mg/kg, the data from which are summarized in Figure 3-2 and Figure 3-3. In both mice
9 and rats, regardless of sex and strain, there is a general trend of increasing exhalation of
10 unchanged TCE with dose, suggesting saturation of a metabolic pathway. The increase is
11 smaller in mice (from 1-6% to 10-18%) than in rats (from 1-3% to 43-78%), suggesting
12 greater overall metabolic capacity in mice. The dose at which apparent saturation occurs appears
13 to be more sex- or strain-dependent in mice than in rats. In particular, the marked increase in
14 exhaled TCE occurred between 20 and 200 mg/kg in female NMRI mice, between 500 and
15 1,000 mg/kg in B6C3F1 mice, and between 10 and 500 mg/kg in male Swiss-Webster mice.
16 However, because only one study is available in each strain, interlot or interindividual variability
17 might also contribute to the observed differences. In rats, all three strains tested showed marked
18 increase in unchanged TCE exhaled between 20 and 200 mg/kg or 10 and 500 mg/kg.
19 Recovered urine, the other major source of excretion, had mainly trichloroacetic acid (TCA),
20 trichloroethanol (TCOH), and trichloroethanol-glucuronide conjugate (TCOG), but revealed no
21 detectable TCE. The source of radioactivity in feces was not analyzed, but it is presumed not to
22 include substantial TCE given the complete absorption expected from the corn oil vehicle.
23 Therefore, at all doses tested in mice, and at doses <200 mg/kg in rats, the majority of orally
24 administered TCE is metabolized. Pretreatment of rats with P450 inducers prior to a 200 mg/kg
25 dose did not change the pattern of recovery, but it did increase the amount recovered in urine by
26 10-15%, with a corresponding decrease in the amount of exhaled unchanged TCE (Dekant et al.,
27 1986a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-20 DRAFT—DO NOT CITE OR QUOTE
-------
>
•a
u
100
90
80
70
60
50
40
30
20
10
0
1
mg/kg 20 200
mg/kg mg/kg
F/NMRI
10 500 1000 2000
mg/kg mg/kg mg/kg mg/kg
M/B6C3F1
10 500 1000
mg/kg mg/kg mg/kg
M/Swiss-Webster
DCage
wash
• Carcass
DCO2
Exhaled
D Feces
E3 Urine
DICE
Exhaled
Mouse Sex/Strain and Dose
4
Figure 3-2. Disposition of [ C]TCE administered by oral gavage in mice
(Dekant et al., 1984,1986a; Green and Prout, 1985; Prout et al., 1985).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-21 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
I
V
f
IB
o
100
90 -
80 -
70 -
60 -
50 -
40
30 -
20 -
10
0
I
mg/kg 20 200
mg/kg mg/kg
F/Wistar
10 500 1000 2000
mg/kg mg/kg mg/kg mg/kg
M/Osborne-Mendel
10 500 1000
mg/kg mg/kg mg/kg
M/Alderley-Park Wistar
DCage
wash
• Carcass
IDC02
Exhaled
D Feces
E3 Urine
DICE
Exhaled
Rat Sex/Strain and Dose
rl4.
Figure 3-3. Disposition of [ C]TCE administered by oral gavage in rats
(Dekant et al., 1984,1986a; Green and Prout, 1985; Prout et al., 1985).
Comprehensive mass balance studies are not available in humans, but several studies
have measured or estimated recovery of TCE in exhaled breath and/or TCA and TCOH in urine
following controlled inhalation exposures to TCE (Monster et al., 1976; Opdam, 1989; Soucek
and Vlachova, 1960). Opdam (1989) only measured exhaled breath, and estimated that, on
average, 15-20% of TCE uptake (retained dose) was exhaled after exposure to 5.8-38 ppm for
29-62 minutes. Soucek and Vlachova (1960) and Bartonicek (1962) did not measure exhaled
breath but did report 69-73% of the retained dose excreted in urine as TCA and TCOH
following exposure to 93-194 ppm (500-1,043 mg/m3) for 5 hours. Soucek and Vlachova
(1960) additionally reported 4% of the retained dose excreted in urine as monochloroacetic acid
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 (MCA). Monster et al. (1976) reported that an average of 10% of the retained TCE dose was
2 eliminated unchanged following 6 hour exposures to 70-140 ppm (376-752 mg/m3) TCE, along
3 with an average of 57% of the retained dose excreted in urine as TCA and free or conjugated
4 TCOH. The differences among these studies may reflect a combination of interindividual
5 variability and errors due to the difficulty in precisely estimating dose in inhalation studies, but
6 in all cases less than 20% of the retained dose was exhaled unchanged and greater than 50% was
7 excreted in urine as TCA and TCOH. Therefore, it is clear that TCE is extensively metabolized
8 in humans. Unlike the rodent studies, no saturation was evident in any of these human recovery
9 studies even though the metabolic capacity may not have been saturated at the exposure levels
10 that were tested.
11
12 3.3.3. Pathways of Metabolism
13 As mentioned in Section 3.3.1, TCE metabolism in animals and humans has been
14 observed to occur via two major pathways: P450-mediated oxidation and GSH conjugation.
15 Products of the initial oxidation or conjugation step are further metabolized to a number of other
16 metabolites. For P450 oxidation, all steps of metabolism occur primarily in the liver, although
17 limited oxidation of TCE has been observed in the lungs of mice, as discussed below. The GSH
18 conjugation pathway also begins predominantly in the liver, but lexicologically significant
19 metabolic steps occur extrahepatically—particularly in the kidney (Lash et al., 1995, 1998,
20 1999b, 2006). The mass-balance studies cited above found that at exposures below the onset of
21 saturation, >50% of TCE intake is excreted in urine as oxidative metabolites (primarily as TCA
22 and TCOH), so TCE oxidation is generally greater than TCE conjugation. This is discussed in
23 detail in Section 3.3.3.3.
24
25 3.3.3.1. Cytochrome P450-Dependent Oxidation
26 Oxidative metabolism by the cytochrome P450, or CYP-dependent, pathway is
27 quantitatively the major route of TCE biotransformation (U.S. Environmental Protection Agency
28 [U.S. EPA], 1985; IARC, 1995; Lash et al., 2000a, b). The pathway is operative in humans and
29 rodents and leads to several metabolic products, some of which are known to cause toxicity and
30 carcinogenicity (U.S. EPA, 1985; IARC, 1995). Although several of the metabolites in this
31 pathway have been clearly identified, others are speculative or questionable. Figure 3-4 depicts
32 the overall scheme of TCE P450 metabolism.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-23 DRAFT—DO NOT CITE OR QUOTE
-------
H (TCE) Cl
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
P450
OHCH,
(W-(Hydroxyacetyl)- N (CH2)2OH
aminoethanol) |-|
CI3C-
CI2CH II
(DCAC) Cl
EHR
CO
(TCOG) Q-gluc H (Glyoxylic°H (MCA) OH
acid)
Figure 3-4. Scheme for the oxidative metabolism of TCE.
Adapted from: Lash et al. (2000a, b), Clewell et al. (2000), Cummings et al.
(2001), Forkert et al. (2006), and long et al. (1998).
In brief, TCE oxidation via P450, primarily CYP2E1 (Guengerich et al., 1991), yields an
oxygenated TCE-P450 intermediate and TCE oxide. The TCE-P450 complex is a transition state
that goes on to form chloral. In the presence of water, chloral rapidly equilibrates with chloral
hydrate (CH), which undergoes reduction and oxidation by alcohol dehydrogenase and aldehyde
dehydrogenase or aldehyde oxidase to form TCOH and TCA, respectively (Miller and
Guengerich 1983; Green and Prout, 1985; Dekant et al., 1986a). Table 3-13 summarizes
available in vitro measurements of TCE oxidation, as assessed by the formation of CH, TCOH,
and TCA. Glucuronidation of TCOH forms TCOG, which is readily excreted in urine.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 Alternatively, TCOG can be excreted in bile and passed to the small intestine where it is
2 hydrolyzed back to TCOH and reabsorbed (Bull, 2000). TCA is poorly metabolized but may
3 undergo dechlorination to form dichloroacetic acid (DCA). However, TCA is predominantly
4 excreted in urine, albeit at a relatively slow rate as compared to TCOG. Like the TCE-P450
5 complex, TCE oxide also seems to be a transient metabolite. Recent data suggest that it is
6 transformed to dichloroactyl chloride, which subsequently decomposes to form DCA (Cai and
7 Guengerich, 1999). As shown in Figure 3-4, several other metabolites, including oxalic acid and
8 TV-(hydroxyacetyl) aminoethanol, may form from the TCE oxide or the TCE-O-P450
9 intermediate and have been detected in the urine of rodents and humans following TCE
10 exposure. Pulmonary excretion of CC>2 has been identified in exhaled breath from rodents
11 exposed to 14C-labeled TCE and is thought to arise from metabolism of DCA. The following
12 sections provide details as to pathways of TCE oxidation, including discussion of inter- and
13 intraspecies differences in metabolism.
14
15 3.3.3.1.1. Formation of trichloroethylene oxide. In previous studies of halogenated alkene
16 metabolism, the initial step was the generation of a reactive epoxides (Anders and Jackobson,
17 1985). Early studies in anesthetized human patients (Powell, 1945), dogs (Butler, 1949), and
18 later reviews (e.g., Goeptar et al., 1995) suggest that the TCE epoxide may be the initial reaction
19 product of TCE oxidation.
20 Epoxides can form acyl chlorides or aldehydes, which can then form aldehydes,
21 carboxylic acids, or alcohols, respectively. Thus, the appearance of CH, TCA, and TCOH as the
22 primary metabolites was considered consistent with the oxidation of TCE to the epoxide
23 intermediate (Powell, 1945; Butler, 1949). Following in vivo exposures to 1,1 -dichloroethylene,
24 a halocarbon very similar in structure to TCE, mouse liver cytosol and microsomes and lung
25 Clara cells exhibited extensive P450-mediated epoxide formation (Forkert, 1999a, b; Forkert et
26 al., 1999; Dowsley et al., 1996). Indeed, TCE oxide inhibits purified CYP2E1 activity (Cai and
27 Guengerich, 2001) similarly to TCE inhibition of CYP2E1 in human liver microsomes
28 (Lipscomb et al., 1997).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-25 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-13. In vitro TCE oxidative metabolism in hepatocytes and
microsomal fractions
In vitro
system
Human
hepatocytes
Human liver
microsomal
protein
Rat liver
microsomal
protein
Rat kidney
microsomal
protein
Mouse liver
microsomal
protein
KM
uM in
medium
210+159
(45-403)
16.7 + 2.45
(13.3-19.7)
30.9 + 3.3
(27.0-36.3)
51.1 + 3.77
(46.7-55.7)
24.6
12 + 3
(9-14)
26+ 17
(13-45)
55.5
72 + 82
42 + 21
940
35.4
378 + 414
161+29
VMAX
nmol TCE
oxidized/min/mg
MSP* or 106
hepatocytes
0.268 + 0.215
(0.101-0.691)
1.246 + 0.805
(0.490-3.309)
1.442 + 0.464
(0.890-2.353)
2.773 + 0.577
(2.078-3.455)
1.44
0.52 + 0.17
(0.37-0.79)
0.33 + 0.15
(0.19-0.48)
4.826
0.96 + 0.65
2.91+0.71
0.154
5.425
8.6 + 4.5
26.06 + 7.29
1,000 x
VMAX/KM
2.45 + 2.28
(0.46-5.57)
74.1+44.1
(38.9-176)
47.0+16.0
(30.1-81.4)
54.9+14.1
(37.3-69.1)
58.5
48 + 23
(26-79)
15 + 10
(11-29)
87.0
24 + 21
80 + 34
0.164
153
42 + 29
163 + 37
Source
Lipscomb etal., 1998a
Lipscomb et al., 1997 (Low KM)
Lipscomb et al., 1997 (Mid KM)
Lipscomb et al., 1997 (High
KM)
Lipscomb etal., 1998b (pooled)
Elfarra et al., 1998 (males, high
affinity)
Elfarra et al., 1998 (females,
high affinity)
Lipscomb etal., 1998b (pooled)
Elfarra et al., 1998 (males, high
affinity)
Elfarra et al., 1998 (females,
high affinity)
Cummings et al., 2001
Lipscomb etal., 1998b (pooled)
Elfarra et al., 1998 (males)
Elfarra et al., 1998 (females)
4
5
6
7
8
9
10
* MSP = Microsomal protein.
Notes: Results presented as mean + standard deviation (minimum-maximum). KM for human hepatocytes
converted from ppm in headspace to uM in medium using reported hepatocyte:air partition coefficient (Lipscomb et
al., 1998a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-26 DRAFT—DO NOT CITE OR QUOTE
-------
1 Conversely, cases have been made against TCE oxide as an obligate intermediate. Using
2 liver microsomes and reconstituted P450 systems (Miller and Guengerich, 1983, 1982) or
3 isolated rat hepatocytes (Miller and Guengerich, 1983), it has been suggested that chlorine
4 migration and generation of a TCE-O-P450 complex (via the heme oxygen) would better explain
5 the observed destruction of the P450 heme, an outcome not likely to be epoxide-mediated.
6 Miller and Guengerich (1982) found CYP2E1 to generate an epoxide but argued that the
7 subsequent production of chloral was not likely related to the epoxide. Green and Prout (1985)
8 argued against epoxide (free form) formation in vivo in mice and rats, suggesting that the
9 expected predominant metabolites would be carbon monoxide, CC>2, MCA, and DCA, rather than
10 the observed predominant appearance of TCA and TCOH and its glucuronide (TCOG).
11 It appears likely that both a TCE-O-P450 complex and a TCE oxide are formed, resulting
12 in both CH and dichloroacetyl chloride, respectively, though it appears that the former
13 predominates. In particular, it has been shown that dichloroacetyl chloride can be generated
14 from TCE oxide, dichloracetyl chloride can be trapped with lysine (Cai and Guengerich, 1999),
15 and that dichloracetyl-lysine adducts are formed in vivo (Forkert et al., 2006). Together, these
16 data strongly suggest TCE oxide as an intermediate metabolite, albeit short-lived, from TCE
17 oxi dati on in vivo.
18
19 3.3.3.1.2. Formation of chloral hydrate (CH), trichloroethanol (TCOH) and trichloroacetic
20 acid (TCA). CH (in equilibrium with chloral) is a major oxidative metabolite produced from
21 TCE as has been shown in numerous in vitro systems, including human liver microsomes and
22 purified P450 CYP2E1 (Guengerich et al., 1991) as well as recombinant rat, mouse, and human
23 P450s including CYP2E1 (Forkert et al., 2005). However, in rats and humans, in vivo circulating
24 CH is generally absent from blood following TCE exposure. In mice, CH is detectable in blood
25 and tissues but is rapidly cleared from systemic circulation (Abbas and Fisher, 1997). The low
26 systemic levels of CH are because of its rapid transformation to other metabolites.
27 CH is further metabolized predominantly to TCOH (Sellers et al., 1972), a reaction
28 thought to be catalyzed by alcohol dehydrogenase (Shultz and Weiner, 1979) and/or CYP2E1
29 (Ni et al., 1996). The role for alcohol dehydrogenase was suggested by the observation that
30 ethanol inhibited CH reduction to TCOH (Larson and Bull, 1989; Muller et al., 1975; Sellers et
31 al., 1972). For instance, Sellers et al. (1972) reported that coexposure of humans, to ethanol and
32 CH resulted in a higher percentage of urinary TCOH (24% of CH metabolites) compared to TCA
33 (19%). When ethanol was absent, 10 and 11% of CH was metabolized to TCOH and TCA,
34 respectively. However, because ethanol can be oxidized by both alcohol dehydrogenase and
35 CYP2E1, there is some ambiguity as to whether these observations involve competition with one
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 or the other of these enzymes. For instance, Ni et al. (1996) reported that CYP2E1 expression
2 was necessary for metabolism of CH to mutagenic metabolites in a human lymphoblastoid cell
3 line, suggesting a role for CYP2E1. Furthermore, Ni et al. (1996) reported that cotreatment of
4 mice with CH and pyrazole, a specific CYP2E1 inducer, resulted in enhanced liver microsomal
5 lipid peroxidation, while treatment with DPEA, an inhibitor of CYP2E1, suppressed lipid
6 peroxidation, suggesting CYP2E1 as a primary enzyme for CH metabolism in this system.
7 Lipscomb et al. (1996) suggested that two enzymes are likely responsible for CH reduction to
8 TCOH based on observation of bi-phasic metabolism for this pathway in mouse liver
9 microsomes. This behavior has also been observed in mouse liver cytosol, but was not observed
10 in rat or human liver microsomes. Moreover, CH metabolism to TCOH increased significantly
11 both in the presence of NADH in the 700x g supernatant of mouse, rat, and human liver
12 homogenate as well as with the addition of NADPH in human samples, suggesting two enzymes
13 may be involved (Lipscomb et al., 1996).
14 TCOH formed from CH is available for oxidation to TCA (see below) or glucuronidation
15 via UDP-glucuronyltransferase to TCOG, which is excreted in urine or in bile (Stenner et al.,
16 1997). Biliary TCOG is hydrolyzed in the gut and available for reabsorption to the liver as
17 TCOH, where it can be glucuronidated again or metabolized to TCA. This enterohepatic
18 circulation appears to play a significant role in the generation of TCA from TCOH and in the
19 observed lengthy residence time of this metabolite, compared to TCE. Using jugular-, duodenal -
20 , and bile duct-cannulated rats, Stenner et al. (1997) showed that enterohepatic circulation of
21 TCOH from the gut back to the liver and subsequent oxidation to TCA was responsible for 76%
22 of TCA measured in the systemic blood.
23 Both CH and TCOH can be oxidized to TCA, and has been demonstrated in vivo in mice
24 (Larson and Bull, 1992a; Dekant et al., 1986a; Green and Prout, 1985), rats (Stenner et al., 1997;
25 Pravecek et al., 1996; Templin et al., 1995b; Larson and Bull, 1992a; Dekant et al., 1986a; Green
26 and Prout, 1985), dogs (Templin et al., 1995a), and humans (Sellers et al., 1978). Urinary
27 metabolite data in mice and rats exposed to 200 mg/kg TCE (Larson and Bull, 1992a;
28 Dekant et al., 1986a) and humans following oral CH exposure (Sellers et al., 1978) show greater
29 TCOH production relative to TCA production. However, because of the much longer urinary
30 half-life in humans of TCA relative to TCOH, the total amount of TCA excreted may be similar
31 to TCOH (Monster et al., 1976; Fisher et al., 1998). This is thought to be primarily due to
32 conversion of TCOH to TCA, either directly or via "back-conversion" of TCOH to CH, rather
33 than due to the initial formation of TCA from CH (Marshall and Owens, 1955).
34 In vitro data are also consistent with CH oxidation to TCA being much less than CH
35 reduction to TCOH. For instance, Lipscomb et al. (1996) reported 1,832-fold differences in KM
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-28 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
values and 10-195-fold differences in clearance efficiency (VMAX/KM) for TCOH and TCA in all
three species (see Table 3-14). Clearance efficiency of CH to TCA in mice is very similar to
humans but is 13-fold higher than rats. Interestingly, Bronley-DeLancey et al. (2006) recently
reported that similar amounts of TCOH and TCA were generated from CH using cryopreserved
human hepatocytes. However, the intersample variation was extremely high, with measured
VMAX ranging from 8-fold greater TCOH to 5-fold greater TCA and clearance (VMAX/KM)
ranging from 13-fold greater TCOH to 17-fold greater TCA. Moreover, because a comparison
with fresh hepatocytes or microsomal protein was not made, it is not clear to what extent these
differences are due to population heterogeneity or experimental procedures.
Table 3-14. In vitro kinetics of trichloroethanol and trichloroacetic acid
formation from chloral hydrate in rat, mouse, and human liver homogenates
Species
Rat
Moused
High affinity
Low affinity
Human
TCOH
Ka
m
0.52
0.19
0.12
0.51
1.34
V b
' max
24.3
11.3
6.3
6.1
34.7
V]VLAX/Km
46.7
59.5
52.5
12.0
25.9
TCA
Ka
m
16.4
3.5
nae
na
23.9
V b
' max
4
10.6
na
na
65.2
V]viAx/Km
0.24
3.0
na
na
2.7
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
aKm presented as mM CH in solution.
bVmax presented as nmoles/mg supernatant protein/min.
'Clearance efficiency represented by VMAX/KM-
dMouse kinetic parameters derived for observations over the entire range of CH exposure as well as discrete, bi-
phasic regions for CH concentrations below (high affinity) and above (low affinity) 1.0 mM.
ena = not applicable.
Source: Lipscomb et al. (1996).
The metabolism of CH to TCA and TCOH involves several enzymes including CYP2E1,
alcohol dehydrogenase, and aldehyde dehydrogenase enzymes (Guengerich et al., 1991; Miller
and Guengerich, 1983; Ni et al., 1996; Shultz and Weiner, 1979; Wang et al., 1993). Because
these enzymes have preferred cofactors (NADPH, NADH, and NAD+), cellular cofactor ratio
and redox status of the liver may have an impact on the preferred pathway
(Kawamoto et al., 1988; Lipscomb et al., 1996).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.3.3.1.3. Formation of dichloroacetic acid (DCA) and other products. As discussed above,
2 DCA could hypothetically be formed via multiple pathways. The work reviewed by Guengerich
3 (2004) has suggested that one source of DCA may be through a TCE oxide intermediary. Miller
4 and Guengerich (1983) report evidence of formation of the epoxide, and Cai and Guengerich
5 (1999) report that a significant amount (about 35%) of DCA is formed from aqueous
6 decomposition of TCE oxide via hydrolysis in an almost pH-independent manner. Because this
7 reaction forming DCA from TCE oxide is a chemical process rather than a process mediated by
8 enzymes, and because evidence suggests that some epoxide was formed from TCE oxidation,
9 Guengerich (2004) notes that DCA would be an expected product of TCE oxidation (see also
10 Yoshioka et al. [2002]). Alternatively, dechlorination of TCA and oxidation of TCOH have been
11 proposed as sources of DCA (Lash et al., 2000a). Merdink et al. (2000) investigated
12 dechlorination of TCA and reported trapping a DCA radical with the spin-trapping agent phenyl-
13 tert-butyl nitroxide, identified by gas chromatography/mass spectroscopy, in both a chemical
14 Fenton system and rodent microsomal incubations with TCA as substrate. Dose-dependent
15 catalysis of TCA to DCA was observed in cultured microflora from B6C3F1 mice (Moghaddam
16 et al., 1996). However, while antibiotic-treated mice lost the ability to produce DCA in the gut,
17 plasma DCA levels were unaffected by antibiotic treatment, suggesting that the primary site of
18 murine DCA production is other than the gut (Moghaddam et al., 1997).
19 However, direct evidence for DCA formation from TCE exposure remains equivocal. In
20 vitro studies in human and animal systems have demonstrated very little DCA production in the
21 liver (James et al., 1997). In vivo, DCA was detected in the blood of mice (Templin et al., 1993;
22 Larson and Bull, 1992a) and humans (Fisher et al., 1998) and in the urine of rats and mice
23 (Larson and Bull, 1992b) exposed to TCE by aqueous oral gavage. However, the use of strong
24 acids in the analytical methodology produces ex vivo conversion of TCA to DCA in mouse blood
25 (Ketcha et al., 1996). This method may have resulted in the appearance of DCA as an artifact in
26 human plasma (Fisher et al., 1998) and mouse blood in vivo (Templin et al., 1995b). Evidence
27 for the artifact is suggested by DCA AUCs that were larger than would be expected from the
28 available TCA (Templin et al., 1995a). After the discovery of these analytical issues, Merdink et
29 al. (1998) reevaluated the formation of DCA from TCE, TCOH, and TCA in mice, with
30 particular focus on the hypothesis that DCA is formed from dechlorination of TCA. They were
31 unable to detect blood DCA in naive mice after administration of TCE, TCOH, or TCA. Low
32 levels of DCA were detected in the blood of children administered therapeutic doses of CH
33 (Henderson et al., 1997), suggesting TCA or TCOH as the source of DCA. Oral TCE exposure
34 in rats and dogs failed to produce detectable levels of DCA (Templin et al., 1995a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-30 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Another difficulty in assessing the formation of DC A is its rapid metabolism at low
exposure levels. Degradation of DCA is mediated by glutathione-S-transferase (GST)-zeta
(Saghir and Schultz, 2002; Tong et al., 1998), apparently occurring primarily in the hepatic
cytosol. DCA metabolism results in suicide inhibition of the enzyme, evidenced by decreased
DCA metabolism in DCA-treated animals (Gonzalez-Leon et al., 1999) and humans (Shroads et
al., 2008) and loss of DCA metabolic activity and enzymatic protein in liver samples from
treated animals (Schultz et al., 2002). This effect has been noted in young mice exposed to DCA
in drinking water at doses approximating 120 mg/kg/d (Schultz et al., 2002). The experimental
data and pharmacokinetic model simulations of several investigators (Jia et al., 2006; Keys et al.,
2004; Li et al., 2008; Merdink et al., 1998; Shroads et al., 2008) suggest that several factors
prevent the accumulation of measurable amounts of DCA: (1) its formation as a short-lived
intermediate metabolite, and (2) its rapid elimination relative to its formation from TCA. While
DCA elimination rates appear approximately one order of magnitude higher in rats and mice than
in humans (James et al., 1997) (see Table 3-15), they still may be rapid enough so that even if
DCA were formed in humans, it would be metabolized too quickly to appear in detectable
quantities in blood.
Table 3-15. In vitro kinetics of DCA metabolism in hepatic cytosol
of mice, rats, and humans
Species
Mouse
Rat
Human
VMAX
(nmol/min/mg protein)
13.1
11.6
0.37
KM
("M)
350
280
71
VMAX/KM
37.4
41.4
5.2
Source: James et al. (1997).
A number of other metabolites, such as oxalic acid, MCA, glycolic acid, and glyoxylic
acid, are formed from DCA (Lash et al., 2000a; Saghir and Schultz, 2002). Unlike other
oxidative metabolites of TCE, DCA appears to be metabolized primarily via hepatic cytosolic
proteins. Since P450 activity resides almost exclusively in the microsomal and mitochondrial
cell fractions, DCA metabolism appears to be independent of P450. Rodent microsomal and
mitochondrial metabolism of DCA was measured to be <10% of cytosolic metabolism
(Lipscomb et al., 1995). DCA in the liver cytosol from rats and humans is transformed to
glyoxylic acid via a GSH-dependent pathway (James et al., 1997). In rats, the KM for GSH was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 0.075 mM with a VMAX for glyoxylic acid formation of 1.7 nmol/mg protein/minute. While this
2 pathway may not involve GST (as evidenced by very low GST activity in this study), Tong et al.
3 (1998) showed GST-zeta, purified from rat liver, to be involved in metabolizing DC A to
4 glyoxylic acid, with a VMAX of 1,334 nmol/mg protein/minute and KM of 71.4 uM for glyoxylic
5 acid formation and a GSH KM of 59 uM.
6
7 3.3.3.1.4. Tissue distribution of oxidative metabolism and metabolites. Oxidative metabolism
8 of TCE, irrespective of the route of administration, occurs predominantly in the liver, but TCE
9 metabolism via the P450 (CYP) system also occurs at other sites because CYP isoforms are
10 present to some degree in most tissues of the body. For example, both the lung and kidneys
11 exhibit cytochrome P450 enzyme activities (Green et al., 1997a, b; Forkert et al., 2005;
12 Cummings et al., 2001). Green et al. (1997b) detected TCE oxidation to chloral in microsomal
13 fractions of whole-lung homogenates from mice, rats, and humans, with the activity in mice the
14 greatest and in humans the least. The rates were slower than in the liver (which also has a higher
15 microsomal protein content as well as greater tissue mass) by 1.8-, 10-, and >10-fold in mice,
16 rats, and humans, respectively. While qualitatively informative, these rates were determined at a
17 single concentration of about 1 mM TCE. A full kinetic analysis was not performed, so
18 clearance and maximal rates of metabolism could not be determined. With the kidney,
19 Cummings et al. (2001) performed a full kinetic analysis using kidney microsomes, and found
20 clearance rates (VMAX/KM) for oxidation were more than 100-fold smaller than average rates that
21 were found in the liver (see Table 3-13). In human kidney microsomes, Amet et al. (1997)
22 reported that CYP2E1 activity was weak and near detection limits, with no CYP2E1 detectable
23 using immunoblot analysis. Cummings and Lash (2000) reported detecting oxidation of TCE in
24 only one of 4 kidney microsome samples, and only at the highest tested concentration of 2 mM,
25 with a rate of 0.13 nmol/minute/mg protein. This rate contrasts with the VMAX values for human
26 liver microsomal protein of 0.19-3.5 nmol/minute/mg protein reported in various experiments
27 (see Table 3-13, above). Extrahepatic oxidation of TCE may play an important role for
28 generation of toxic metabolites in situ. The roles of local metabolism in kidney and lung toxicity
29 are discussed in detail in Sections 4.4 and 4.7, respectively.
30 With respect to further metabolism beyond oxidation of TCE, CH has been shown to be
31 metabolized to TCA and TCOH in lysed whole blood of mice and rats and fractionated human
32 blood (Lipscomb et al., 1996) (see Table 3-16). TCOH production is similar in mice and rats and
33 is approximately 2-fold higher in rodents than in human blood. However, TCA formation in
34 human blood is 2- or 3-fold higher than in mouse or rat blood, respectively. In human blood,
35 TCA is formed only in the erythrocytes. TCOH formation occurs in both plasma and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-32 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
erythrocytes, but 4-fold more TCOH is found in plasma than in an equal volume of packed
erythrocytes. While blood metabolism of CH may contribute further to its low circulating levels
in vivo., the metabolic capacity of blood (and kidney) may be substantially lower than liver.
Regardless, any CH reaching the blood may be rapidly metabolized to TCA and TCOH.
Table 3-16. TCOH and TCA formed from CH in vitro in lysed whole blood
of rats and mice or fractionated blood of humans (nmoles formed in 400 uL
samples over 30 minutes)
TCOH
TCA
Rat
45.4 + 4.9
0.14 + 0.2
Mouse
46.7+1.0
0.21+0.3
Human
Erythrocytes
15.7 + 1.4
0.42 + 0.0
Plasma
4.48 + 0.2
not detected
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Source: Lipscomb et al. (1996).
DCA and TCA are known to bind to plasma proteins. Schultz et al. (1999) measured
DC A binding in rats at a single concentration of about 100 jiM and found a binding fraction of
less than 10%. However, these data are not greatly informative for TCE exposure in which DCA
levels are significantly lower, and limitation to a single concentration precludes fitting to
standard binding equations from which the binding at low concentrations could be extrapolated.
Templin et al. (1993, 1995a, b), Schultz et al. (1999), Lumpkin et al. (2003), and Yu et al. (2003)
all measured TCA binding in various species and at various concentration ranges. Of these,
Templin et al. (1995a, b) and Lumpkin et al. (2003) measured levels in humans, mice, and rats.
Lumpkin et al. (2003) studied the widest concentration range, spanning reported TCA plasma
concentrations from experimental studies. Table 3-17 shows derived binding parameters.
However, these data are not entirely consistent among researchers; 2- to 5-fold differences in
BMAX and Kd are noted in some cases, although some differences existed in the rodent strains and
experimental protocols used. In general, however, at lower concentrations, the bound fraction
appears greater in humans than in rats and mice. Typical human TCE exposures, even in
controlled experiments with volunteers, lead to TCA blood concentrations well below the
reported Kd (see Table 3-17, below), so the TCA binding fraction should be relatively constant.
However, in rats and mice, experimental exposures may lead to peak concentrations similar to,
or above, the reported Kd (e.g., Templin et al., 1993; Yu et al., 2000), meaning that the bound
fraction should temporarily decrease following such exposures.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-33 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-17 Reported TCA plasma binding parameters
A
BMAX
("M)
Ka
("M)
A+
BMAX/KJ
Concentration
range (uM
bound+free)
Human
Templin et al., 1995a
Lumpkin et al., 2003
-
-
1,020
708.9
190
174.6
5.37
4.06
3-1,224
0.06-3,065
Rat
Templin et al., 1995a
Yu et al., 2000
Lumpkin et al., 2003
-
0.602
-
540
312
283.3
400
136
383.6
1.35
2.90
0.739
3-1,224
3.8-1,530
0.06-3,065
Mouse
Templin et al., 1993
Lumpkin et al., 2003
-
-
310
28.7
248
46.1
1.25
0.623
3-1,224
0.06-1,226
4
5
6
7
8
9
10
11
12
13
14
15
16
Notes: Binding parameters based on the equation Cbound = A x Cfree + BMAX x Cfree/(Kd + Cfree), where Cbouad is the
bound concentration, Cfree is the free concentration, and A = 0 for Templin et al. (1993, 1995a) and Lumpkin et al.
(2003). The quantity A+ BMAX/Kd is the ratio of bound-to-free at low concentrations.
Limited data are available on tissue:blood partitioning of the oxidative metabolites CH,
TCA, TCOH and DC A, as shown in Table 3-18. As these chemicals are all water soluble and
not lipophilic, it is not surprising that their partition coefficients are close to 1 (within about
2-fold). It should be noted that the TCA tissue:blood partition coefficients reported in
Table 3-18 were measured at concentrations 1.6-3.3 M, over 1,000-fold higher than the reported
Kd. Therefore, these partition coefficients should reflect the equilibrium between tissue and free
blood concentrations. In addition, only one in vitro measurement has been reported of
blood:plasma concentration ratios for TCA: Schultz et al. (1999) reported a value of 0.76 in rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-34 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-18 Partition coefficients for TCE oxidative metabolites
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Species/tissue
Tissuerblood partition coefficient
CH
TCA
TCOH
DCA
Human3
Kidney
Liver
Lung
Muscle
-
-
-
-
0.66
0.66
0.47
0.52
2.15
0.59
0.66
0.91
-
-
-
-
Mouseb
Kidney
Liver
Lung
Muscle
0.98
1.42
1.65
1.35
0.74
1.18
0.54
0.88
1.02
1.3
0.78
1.11
0.74
1.08
1.23
0.37
a Fisher etal. (1998).
b Abbas and Fisher (1997).
Note: TCA and TCOH partition coefficients have not been reported for rats.
3.3.3.1.5. Species-, sex-, and age-dependent differences of oxidative metabolism. The ability
to describe species- and sex-dependent variations in TCE metabolism is important for species
extrapolation of bioassay data and identification of human populations that are particularly
susceptible to TCE toxicity. In particular, information on the variation in the initial oxidative
step of CH formation from TCE is desirable, because this is the rate-limiting step in the eventual
formation and distribution of the putative toxic metabolites TCA and DCA (Lipscomb et al.,
1997).
Inter- and intraspecies differences in TCE oxidation have been investigated in vitro using
cellular or subcellular fractions, primarily of the liver. The available in vitro metabolism data on
TCE oxidation in the liver (see Table 3-13) show substantial inter and intraspecies variability.
Across species, microsomal data show that mice apparently have greater capacity (VMAX) than
rat or humans, but the variability within species can be 2- to 10-fold. Part of the explanation may
be related to CYP2E1 content. Although liver P450 content is similar across species, mice and
rats exhibit higher levels of CYP2E1 content (0.85 and 0.89 nmol/mg protein, respectively)
(Nakajima et al., 1993; Davis et al., 2002) than humans (approximately 0.25-0.30 nmol/mg
protein) (Elfarra et al., 1998; Davis et al., 2002). Thus, the data suggest that rodents would have
a higher capacity than humans to metabolize TCE, but this is difficult to verify in vivo because
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-35 DRAFT—DO NOT CITE OR QUOTE
-------
1 very high exposure concentrations in humans would be necessary to assess the maximum
2 capacity of TCE oxidation.
3 With respect to the KM of liver microsomal TCE oxidative metabolism, where KM is
4 indicative of affinity (the lower the numerical value of KM, the higher the affinity), the trend
5 appears to be mice and rats have higher KM values (i.e., lower affinity) than humans, but with
6 substantial overlap due to interindividual variability. Note that, as shown in Table 3-13, the
7 ranking of rat and mouse liver microsomal KM values between the two reports by Lipscomb et al.
8 (1998b) and Elfarra et al. (1998) is not consistent. However, both studies clearly show that KM is
9 the lowest (i.e., affinity is highest) in humans. Because clearance at lower concentrations is
10 determined by the ratio VMAX to KM, the lower apparent KM in humans may partially offset the
11 lower human VMAX, and lead to similar oxidative clearances in the liver at environmentally
12 relevant doses. However, differences in activity measured in vitro may not translate into in vivo
13 differences in metabolite production, as the rate of metabolism in vivo depends also on the rate of
14 delivery to the tissue via blood flow (e.g., Lipscomb et al., 2003). The interaction of enzyme
15 activity and blood flow is best investigated using PBPK models and is discussed, along with
16 descriptions of in vivo data, in Section 3.5.
17 Data on sex- and age-dependence in oxidative TCE metabolism are limited but suggest
18 relatively modest differences in humans and animals. In an extensive evaluation of
19 CYP-dependent activities in human liver microsomal protein and cryopreserved hepatocytes,
20 Parkinson et al. (2004) identified no age or gender-related differences in CYP2E1 activity. In
21 liver microsomes from 23 humans, the KM values for females was lower than males, but VMAX
22 values were very similar (Lipscomb et al., 1997). Appearance of total trichloro compounds in
23 urine following intrapertoneal dosing with TCE was 28% higher in female rats than in males
24 (Verma and Rana, 2003). The oxidation of TCE in male and female rat liver microsomes was
25 not significantly different; however, pregnancy resulted in a decrease of 27-39% in the rate of
26 CH production in treated microsomes from females (Nakajima et al., 1992b). Formation of CH
27 in liver microsomes in the presence of 0.2 or 5.9 mM TCE exhibited some dependency on age of
28 rats, with formation rates in both sexes of 1.1-1.7 nmol/mg protein/minute in 3-week-old
29 animals and 0.5-1.0 nmol/mg protein/minute in 18-week-old animals (Nakajima et al., 1992b).
30 Fisher et al. (1991) reviewed data available at that time on urinary metabolites to
31 characterize species differences in the amount of urinary metabolism accounted for by TCA (see
32 Table 3-19). They concluded that TCA seemed to represent a higher percentage of urinary
33 metabolites in primates than in other mammalian species, indicating a greater proportion of
34 oxidation leading ultimately to TCA relative to TCOG.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-36 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-19 Urinary excretion of trichloroacetic acid by various species
exposed to trichloroethylene (based on data reviewed in Fisher et al., 1991)
Species
Baboon3-0
Chimpanzee3
Monkey,
Rhesus3-0
Mice, NMRIb
Mice, B6C3F13
Rabbit,
Japanese
White3-0
Rat, Wistarb
Rat, Osborne-
Mendel3
Rat, Holtzman3
Percentage of
urinary
excretion of TCA
Male
16
24
19
—
7-12
0.5
—
6-7
7
Female
—
22
—
8-20
—
"
14-17
—
—
Dose route
Intramuscular
injection
Intramuscular
injection
Intramuscular
injection
Oral
intubation
Oral
intubation
Intraperitoneal
injection
Oral
intubation
Oral
intubation
Intraperitoneal
injection
TCE dose
(mgTCE/kg)
50
50
50
2-200
10-2,000
200
2-200
10-2,000
lOmgTCE/rat
References, comments
Mueller et al., 1982
Mueller et al., 1982
Mueller et al., 1982
Dekantetal., 1986a
Green and Prout, 1985
Nomiyama and
Nomiyama, 1979
Dekantetal., 1986a
Green and Prout, 1985
Nomiyama and
Nomiyama, 1979
4
5
6
7
8
9
10
11
12
13
""Percentage urinary excretion determined from accumulated amounts of TCOH and TCA in urine 3 to 6 days
postexposure.
Percentage urinary excretion determined from accumulated amounts of TCOH, dichloroacetic acid, oxalic acid, and
7V-(hydroxyacetyl)aminoethanol in urine 3 days postexposure.
°Sex is not specified.
Note: Human data tabulated in Fisher et al. (1991) from Nomiyama and Nomiyama (1971) was not included here
because it was relative to urinary excretion of total trichloro-compounds, not as fraction of intake as was the case for
the other data included here.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-37 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
3.3.3.1.6. CFP isoforms and genetic polymorphisms. A number of studies have identified
multiple P450 isozymes as having a role in the oxidative metabolism of TCE. These isozymes
include CYP2E1 (Nakajima et al., 1992a, 1990, 1988; Guengerich and Shimada, 1991;
Guengerich et al., 1991), CYP3A4 (Shimada et al., 1994), CYP1A1/2, CYP2C11/6
(Nakajima et al., 1993, 1992a), CYP2F, and CYP2B1 (Forkert et al., 2005). Recent studies in
CYP2E1-knockout mice have shown that in the absence of CYP2E1, mice still have substantial
capacity for TCE oxidation (Kim and Ghanayem, 2006; Forkert et al., 2006). However,
CYP2E1 appears to be the predominant (i.e., higher affinity) isoform involved in oxidizing TCE
(Nakajima et al., 1992a; Guengerich and Shimada, 1991; Guengerich et al., 1991; Forkert et al.,
2005). In rat liver, CYP2E1 catalyzed TCE oxidation more than CYP2C11/6 (Nakajima et al.,
1992a). In rat recombinant-derived P450s, the CYP2E1 had a lower KM (higher affinity) and
higher VMAX/KM ratio (intrinsic clearance) than CYP2B1 or CYP2F4 (Forkert et al., 2005).
Interestingly, there was substantial differences in KM between rat and human CYP2Els and
between rat CYP2F4 and mouse CYP2F2, suggesting that species-specific isoforms have
different kinetic behavior (see Table 3-20).
Table 3-20. P450 isoform kinetics for metabolism of TCE to CH in human,
rat, and mouse recombinant P450s
Experiment
Human rCYP2El
RatrCYP2El
RatrCYP2Bl
Rat rCYP2F4
Mouse rCYP2F2
KM
uM
196 + 40
14 + 3
131+36
64 + 9
114 + 17
VMAX
pmol/min/pmol P450
4 + 0.2
11+0.3
9 + 0.5
17 + 0.5
13+0.4
VMAX/KM
0.02
0.79
0.07
0.27
0.11
20
21
22
23
24
25
26
27
28
29
Source: Forkert et al. (2005)
The presence of multiple P450 isoforms in human populations affects the variability in
individuals' ability to metabolize TCE. Studies using microsomes from human liver or from
human lymphoblastoid cell lines expressing CYP2E1, CYP1A1, CYP1A2, or CYP3A4 have
shown that CYP2E1 is responsible for greater than 60% of oxidative TCE metabolism
(Lipscomb et al., 1997). Similarities between metabolism of chlorzoxazone (a CYP2E1
substrate) in liver microsomes from 28 individuals (Peter et al., 1990) and TCE metabolism
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-38 DRAFT— DO NOT CITE OR QUOTE
-------
1 helped identify CYP2E1 as the predominant (high affinity) isoform for TCE oxidation.
2 Additionally, Lash et al. (2000a) suggested that, at concentrations above the KM value for
3 CYP2E1, CYP1A2 and CYP2A4 may also metabolize TCE in humans; however, their
4 contribution to the overall TCE metabolism was considered low compared to that of CYP2E1.
5 Given the difference in expression of known TCE-metabolizing P450 isoforms (see Table 3-21)
6 and the variability in P450-mediated TCE oxidation (Lipscomb et al., 1997), significant
7 variability may exist in individual human susceptibility to TCE toxicity.
9
10
11
Table 3-21. P450 isoform activities in human liver microsomes exhibiting
different affinities for TCE
Affinity group
Low KM
Mid KM
High KM
CYP isoform activity (pmol/min/mg protein)
CYP2E1
520 + 295
820 + 372
1,317 + 592
CYP1A2
241 + 146
545 + 200
806 + 442
CYP3A4
2.7 + 2.7
2.9 + 2.8
1.8 + 1.1
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Activities of CYP1A2, CYP2E1, and CYP3 A4 were measured with phenacetin, chlorzoxazone, and testosterone as
substrates, respectively. Data are means + standard deviation from 10, 9, and 4 samples for the low-, mid-, and
high-KM groups, respectively. Only CYP3A4 activities are not significantly different (p < 0.05) from one another
by Kruskal-Wallis one-way analysis of variance.
Source: Lash et al. (2000a).
Differences in content and/or intrinsic catalytic properties (KM, VMAX) of specific
enzymes among species, strains, and individuals may play an important role in the observed
differences in TCE metabolism and resulting toxicities. Lipscomb et al. (1997) reported
observing three statistically distinct groups of KM values for TCE oxidation using human
microsomes. The mean ± standard deviation [SD] (uM TCE) for each of the three groups was
16.7 + 2.5 (n = 10), 30.9 + 3.3 (n = 9), and 51.1 + 3.8 (n = 4). Within each group, there were no
significant differences in sex or ethnicity. However, the overall observed KM values in female
microsomes (21.9 + 3.5 uM, n = 10) were significantly lower than males (33.1 + 3.5 uM,
n= 13). Interestingly, in human liver microsomes, different groups of individuals with different
affinities for TCE oxidation appeared to also have different activities for other substrates not
only with respect to CYP2E1 but also CYP1A2 (Lash et al., 2000a) (see Table 3-21). Genetic
polymorphisms in humans have been identified in the CYP isozymes thought to be responsible
for TCE metabolism (Pastino et al., 2000), but no data exist correlating these polymorphisms
with enzyme activity. It is relevant to note that repeat polymorphism (Hu et al., 1999) or
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 polymorphism in the regulatory sequence (McCarver et al., 1998) were not involved in the
2 constitutive expression of human CYP2E1; however, it is unknown if these types of
3 polymorphisms may play a role in the inducibility of the respective gene.
4 Individual susceptibilities to TCE toxicity may also result from variations in enzyme
5 content, either at baseline or due to enzyme induction/inhibition, which can lead to alterations in
6 the amounts of metabolites formed. Certain physiological and pathological conditions or
7 exposure to other chemicals (e.g., ethanol and acetominophen) can induce, inhibit, or compete
8 for enzymatic activity. Given the well established (or characterized) role of the liver to
9 oxidatively metabolize TCE (by CYP2E1), increasing the CYP2E1 content or activity (e.g., by
10 enzyme induction) may not result in further increases in TCE oxidation. Indeed, Kaneko et al.
11 (1994) reported that enzyme induction by ethanol consumption in humans increased TCE
12 metabolism only at high concentrations (500 ppm, 2,687 mg/m3) in inspired air. However, other
13 interactions between ethanol and the enzymes that oxidatively metabolize TCE metabolites can
14 result in altered metabolic fate of TCE metabolites. In addition, enzyme inhibition or
15 competition can decrease TCE oxidation and subsequently alter the TCE toxic response via, for
16 instance, increasing the proportion undergoing GSH conjugation (Lash et al., 2000a). TCE itself
17 is a competitive inhibitor of CYP2E1 activity (Lipscomb et al., 1997), as shown by reduced
18 /7-nitrophenol hydroxylase activity in human liver microsomes, and so may alter the toxicity of
19 other chemicals metabolized through that pathway. On the other hand, suicidal CYP heme
20 destruction by the TCE-oxygenated CYP intermediate has also been shown (Miller and
21 Guengerich, 1983).
22
23 3.3.3.2. Glutathione (GSH) Conjugation Pathway
24 Historically, the conjugative metabolic pathways have been associated with xenobiotic
25 detoxification. This is true for GSH conjugation of many compounds. However, several
26 halogenated alkanes and alkenes, including TCE, are bioactivated to cytotoxic metabolites by the
27 GSH conjugate processing pathway (mercapturic acid) pathways (Elfarra et al., 1986a, b). In the
28 case of TCE, production of reactive species several steps downstream from the initial GSH
29 conjugation is believed to cause cytotoxicity and carcinogenicity, particularly in the kidney.
30 Since the GSH conjugation pathway is in competition with the P450 oxidative pathway for TCE
31 biotransformation, it is important to understand the role of various factors in determining the flux
32 of TCE through each pathway. Figure 3-5 depicts the present understanding of TCE metabolism
33 via GSH conjugation.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-40 DRAFT—DO NOT CITE OR QUOTE
-------
H (TCE) Cl
1GST
Cl SG
H Cl
GGT
CGDP O
NHf Np
A/y -SSL. /vv
ar* u • / v K
« v« n f*i f* u
2 2 ' Acylase CI2 C2 H 1
x O- n
FMO-3 / (DCVC) (NAcDCVC)
P450./ 1
y/ P-lyase
^ o ^ r ^
II NH?+ C'\ /S"
/S\^/ H> \
J LH ,00^
V ^ -/
(DCVCS)
1
CYP3A
r ° ^\
o A
II NH
A/V
CI2 C2 H |
O-
V J
^~^ -^
(NAcDCVCS)
2 Figure 3-5 Scheme for GSH-dependent metabolism of TCE.
3
4 Adapted from: Lash et al. (2000a); Cummings and Lash (2000); NRC (2006).
5
6
7 3.3.3.2.1. Formation ofS-(l,2-dichlorovinyl)glutathione
8 (DCVG). The conjugation of TCE to GSH produces S-(l,2
or S-(2,2-dichlorovinyl)glutathione
i-dichlorovinyl)glutathione or its
9 isomer S-(2,2-dichlorovinyl)glutathione (DCVG). There is some uncertainty as to which
10 GST isoforms mediate TCE conjugation. Lash and colleagues studied TCE conjugation in renal
1 1 tissue preparations, isolated renal tubule cells from male F344 rats and purified GST alpha-class
12 isoforms 1-1, 1-2 and 2-2 (Cummings et al., 2000a; Cummings and Lash, 2000; Lash et al.,
13 2000b). The results demonstrated high conjugative activity in renal cortex and in proximal
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-41 DRAFT—DO NOT CITE OR QUOTE
-------
1 tubule cells. Although the isoforms studied had similar VMAX values, the KM value for GST 2-2
2 was significantly lower than the other forms, indicating that this form will catalyze TCE
3 conjugation at lower (more physiologically relevant) substrate concentrations. In contrast, using
4 purified rat and human enzymes, Hissink et al. (2002) reported in vitro activity for DCVG
5 formation only for mu- and pi-class GST isoforms, and none towards alpha-class isoforms;
6 however, the rat mu-class GST 3-3 was several folds more active than the human mu-class
7 GST Ml-1. Although GSTs are present in tissues throughout the body, the majority of TCE
8 GSH conjugation is thought to occur in the liver (Lash et al., 2000a). Using in vitro studies with
9 renal preparations, it has been demonstrated that GST catalyzed conjugation of TCE is increased
10 following the inhibition of CYP-mediated oxidation (Cummings et al., 2000b).
11 In F344 rats, following gavage doses of 263-1,971 mg/kg TCE in 2 mL corn oil, DCVG
12 was observed in the liver and kidney of females only, in blood of both sexes (Lash et al., 2006),
13 and in bile of males (Dekant et al., 1990). The data from Lash et al. (2006) are difficult to
14 interpret because the time courses seem extremely erratic, even for the oxidative metabolites
15 TCOH and TCA. Moreover, a comparison of blood levels of TCA and TCOH with other studies
16 in rats at similar doses reveals differences of over 1,000-fold in reported concentrations. For
17 instance, at the lowest dose of 263 mg/kg, the peak blood levels of TCE and TCA in male F344
18 rats were 10.5 and 1.6 |ig/L, respectively (Lash et al., 2006). By contrast, Larson and Bull
19 (1992a) reported peak blood TCE and TCA levels in male Sprague-Dawley rats over 1,000-fold
20 higher—around 10 and 13 mg/L, respectively—following oral doses of 197 mg/kg as a
21 suspension in 1% aqueous Tween 80®. The results of Larson and Bull (1992a) are similar to Lee
22 et al. (2000a), who reported peak blood TCE levels of 20-50 mg/L after male Sprague-Dawley
23 rats received oral doses of 144-432 mg/kg in a 5% aqueous Alkamus emulsion (polyethoxylated
24 vegetable oil), and to Stenner et al. (1997), who reported peak blood levels of TCA in male F344
25 rats of about 5 mg/L at a slightly lower TCE oral dose of 100 mg/kg administered to fasted
26 animals in 2% Tween 80®. Thus, while useful qualitatively as an indicator of the presence of
27 DCVG in rats, the quantitative reliability of reported concentrations, for metabolites of either
28 oxidation or GSH conjugation, may be questionable.
29 In humans, DCVG was readily detected at in human blood following onset of a 4-hour
30 TCE inhalation exposure to 50 or 100 ppm (269 or 537 mg/m3; Lash et al., 1999a). At 50 ppm,
31 peak blood levels ranged from 2.5 to 30 uM, while at 100 ppm, the mean (+ SE, n = 8) peak
32 blood levels were 46.1 + 14.2 uM in males and 13.4 + 6.6 uM in females. Although on average,
33 male subjects had 3-fold higher peak blood levels of DCVG than females, DCVG blood levels.
34 in half of the male subjects were similar to or lower than those of female subjects. This suggests
35 a polymorphism in GSH conjugation of TCE rather than a true gender difference (Lash et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-42 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
1999a) as also has been indicated by Hissink et al. (2002) for the human mu-class GST Ml-1
enzyme. Interestingly, as shown in Table 3-22, the peak blood levels of DCVG are similar on a
molar basis to peak levels of TCE, TCA, and TCOH in the same subjects, as reported in
Fisher etal. (1998).
Table 3-22 Comparison of peak blood concentrations in humans exposed to
100 ppm (537 mg/m3) TCE for 4 hours (Fisher et al., 1998; Lash et al., 1999a)
Chemical species
TCE
TCA
TCOH
DCVG
Peak blood concentration (mean + SD, uM)
Males
23 + 11
56 + 9.8
21 + 5.0
46.1 + 14.2
Females
14 + 4.7
59+12
15 + 5.6
13.4 + 6.6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Tables 3-23 and 3-24 summarize DCVG formation from TCE conjugation from in vitro
studies of liver and kidney cellular and subcellular fractions in mouse, rat, and human. Tissue-
distribution and species-and gender-differences in DCVG formation are discussed below.
3.3.3.2.2. Formation ofS-(l,2-dichlorovinyl) cysteine or S-(2,2-dichlorovinyl) cysteine
(DCVC). The cysteine conjugate, isomers S-(l,2-dichlorovinyl) cysteine (1,2-DCVC) or
S-(2,2-dichlorovinyl) cysteine (2,2-DCVC), is formed from DCVG in a two-step sequence.
DCVG is first converted to the cysteinylglycine conjugate
S-(l,2-dichlorovinyl)-L-cysteinylglycine or its isomer S-(2,2-dichlorovinyl)-L-cysteinylglycine
by y-glutamyl transpeptidase (GGT) in the renal brush border (Elfarra and Anders, 1984; Lash et
al., 1988).
Cysteinylglycine dipeptidases in the renal brush border and basolateral membrane
convert DCVG to DCVC via glycine cleavage (Goeptar et al., 1995; Lash et al., 1998). This
reaction can also occur in the bile or gut, as DCVG excreted into the bile is converted to DCVC
and reabsorbed into the liver where it may undergo further acetylation.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-43 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-23. GSH conjugation of TCE (at 1-2 mM) in liver and kidney
cellular fractions in humans, male F344 rats, and male B6C3F1 mice
Species and cellular/subcellular fraction (TCE
concentration)
DCVG formation
(nmol/hour/mg protein or 106 cells)
Male
Female
Human
Hepatocytes (0.9 mM) [pooled]
Liver cytosol (1 mM) [individual samples]
Liver cytosol (2 mM) [pooled]
Liver microsomes (1 mM) [individual samples]
Liver microsomes (1 mM) [pooled]
Kidney cytosol (2 mM) [pooled]
Kidney microsomes (1 mM) [pooled]
11+3
156+16
174+13
346
108 + 24
83 + 11
146
42
320
Rat
Liver cytosol (2 mM)
Liver microsomes (2 mM)
Kidney cortical cells (2 mM)
Kidney cytosol (2 mM)
Kidney microsomes (2 mM)
7.30 + 2.8
10.3 + 2.8
0.48 + 0.02
0.45 + 0.22
not detected
4.86 + 0.14
7.24 + 0.24
0.65 + 0.15
0.32 + 0.02
0.61+0.06
Mouse
Liver cytosol (2 mM)
Liver microsomes (2 mM)
Kidney cytosol (2 mM)
Kidney microsomes (2 mM)
24.5 + 2.4
40.0 + 3.1
5.6 + 0.24
5.47+1.41
21.7 + 0.9
25.6 + 0.8
3.7 + 0.48
16.7 + 4.7
4
5
Mean + SE. Source: Lash et al. (1999a, 1998, 1995); Cummings and Lash (2000); Cummings et al. (2000b).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-44 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-24 Kinetics of TCE metabolism via GSH conjugation in male F344
rat kidney and human liver and kidney cellular and subcellular fractions
Tissue and cellular fraction
KM
(uM TCE)
VMAX
(nmol
DCVG/min/mg
protein or 106
hepatocytes)
1,000 x
VMAX/KM
Rat
Kidney proximal tubular cells: low affinity
Kidney proximal tubular cells: high affinity
2,910
460
0.65
0.47
0.22
1.0
Human
Liver hepatocytes*
Liver cytosol: low affinity
Liver cytosol: high affinity
Liver microsomes: low affinity
Liver microsomes: high affinity
Kidney proximal tubular cells: low affinity
Kidney proximal tubular cells: high affinity
Kidney cytosol
Kidney microsomes
37-106
333
22.7
250
29.4
29,400
580
26.3
167
0.16-0.26
8.77
4.27
3.1
1.42
1.35
0.11
0.81
6.29
2.4-4.5
2.6
190
12
48
0.046
0.19
31
38
*Kinetic analyses of first 6 to 9 (out of 10) data points from Figure 1 from Lash et al. (1999a) using Lineweaver-
Burk or Eadie-Hofstee plots and linear regression (R2 = 0.50-0.95). Regression with best R2 used first 6 data
points and Eadie-Hofstee plot, with resulting KM and VMAX of 106 and 0.26, respectively.
Source: Lash et al. (1999a); Cummings and Lash (2000); Cummings et al. (2000b).
3.3.3.2.3. Formation ofNAcDCVC. N-acetylation of DCVC can either occur in the kidney, as
demonstrated in rat kidney microsomes (Duffel and Jakoby, 1982), or in the liver (Birner et al.,
1997). Subsequent release of DCVC from the liver to blood may result in distribution to the
kidney resulting in increased internal kidney exposure to the acetylated metabolite over and
above what the kidney already is capable of generating. In the kidney, NAcDCVC may undergo
deacetylation, which is considered a rate-limiting-step in the production of proximal tubule
damage (Wolfgang et al., 1989; Zhang and Stevens, 1989). As a polar mercapturtae, NAcDCVC
may be excreted in the urine as evidenced by findings in mice (Birner et al., 1993), rats
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-45 DRAFT—DO NOT CITE OR QUOTE
-------
1 (Bernauer et al., 1996; Commandeur and Vermeulen, 1990), and humans who were exposed to
2 TCE (Bernauer et al., 1996; Birner et al., 1993), suggesting a common glutathione-mediated
3 metabolic pathway for DCVC among species.
4
5 3.3.3.2.4. Beta lyase metabolism ofS-(l,2-dichlorovinyl) cysteine (DCVC). The enzyme
6 cysteine conjugate p-lyase catalyzes the breakdown of DCVC to reactive nephrotoxic
7 metabolites (Goeptar et al., 1995). This reaction involves removal of pyruvate and ammonia and
8 production of S-(l,2-dichlorovinyl) thiol (DCVT), an unstable intermediate, which rearranges to
9 other reactive alkylation metabolites that form covalent bonds with cellular nucleophiles
10 (Goeptar et al., 1995; Dekant et al., 1988). The rearrangement of DCVT to enethiols and their
11 acetylating agents has been described in trapping experiments (Dekant et al., 1988) and proposed
12 to be responsible for nucleophilic adduction and toxicity in the kidney. The quantification of
13 acid-labile adducts was proposed as a metric for TCE flux through the GSH pathway. However,
14 the presence of analytical artifacts precluded such analysis. In fact, measurement of acid-labile
15 adduct products resulted in higher values in mice than in rats (Eyre et al., 1995a, b).
16 DCVC metabolism to reactive species via a P-lyase pathway has not been directly
17 observed in vivo in animals or humans. However, P-lyase activity in humans and rats (reaction
18 rates were not reported) was demonstrated in vivo using a surrogate substrate,
19 2-(fluoromethoxy)-l,l,3,3,3-pentafluoro-l-propene (Iyer et al., 1998). P-lyase-mediated
20 reactive adducts have been described in several extrarenal tissues, including rat and human liver
21 and intestinal microflora (Larsen and Stevens, 1986; Tomisawa et al., 1984, 1986; Stevens,
22 1985a; Stevens and Jakoby, 1983; Dohn and Anders, 1982; Tateishi et al., 1978) and rat brain
23 (Alberati-Giani et al., 1995; Malherbe et al., 1995).
24 In the kidneys, glutamine transaminase K appears to be primarily responsible for P-lyase
25 metabolism of DCVC (Perry et al., 1993; Lash et al., 1990a, 1986; Jones et al., 1988;
26 Stevens et al., 1988, 1986). P-lyase transformation of DCVC appears to be regulated by 2-keto
27 acids. DCVC toxicity in isolated rat proximal tubular cells was significantly increased with the
28 addition of a-keto-y-methiolbutyrate or phenylpyruvate (Elfarra et al., 1986b). The presence of
29 a-keto acid cofactors is necessary to convert the inactive form of the P-lyase enzyme (containing
30 pyridoxamine phosphate) to the active form (containing pyridoxal phosphate) (Goeptar et al.,
31 1995).
32 Both low- and high-molecular-weight enzymes with P-lyase activities have been
33 identified in rat kidney cytosol and mitochondria (Abraham et al., 1995a, b; Stevens et al., 1988;
34 Lashetal., 1986). While glutamine transaminase K and kynureninase-associated P-lyase
35 activities have been identified in rat liver (Alberati-Giani et al., 1995; Stevens, 1985a), they are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-46 DRAFT—DO NOT CITE OR QUOTE
-------
1 quite low compared to renal glutamine transaminase K activity and do not result in
2 hepatotoxicity in DCVG-or DCVC-treated rats (Elfarra and Anders, 1984). Similar isoforms of
3 p-lyase have also been reported in mitochondrial fractions of brain tissue (Cooper, 2004).
4 The kidney enzyme L-a-hydroxy (L-amino) acid oxidase is capable of forming an
5 iminium intermediate and keto acid analogues (pyruvate or S-(l,2-dichlorovinyl)-2-oxo-
6 3-mercaptopropionate) of DCVC, which decomposes to dichlorovinylthiol (Lash et al., 1990b;
7 Stevens et al., 1989). In rat kidney homogenates, this enzyme activity resulted in as much as
8 35% of GSH pathway-mediated bioactivation. However, this enzyme is not present in humans,
9 an important consideration for extrapolation of renal effects across species.
10
11 3.3.3.2.5. SulfoxidationofS-(l,2-dichlorovinyl)cysteine(DCVC)andNAcDCVC. A second
12 pathway for bioactivation of TCE S-conjugates involves sulfoxidation of either the cysteine or
13 mercapturic acid conjugates (Sausen and Elfarra, 1990; Park et al., 1992; Lash et al., 1994, 2003;
14 Werner et al., 1995a, b, 1996; Birner et al., 1998; Krause et al., 2003). Sulfoxidation of DCVC
15 was mediated mainly by flavin monooxygenase 3 (FMO3), rather than CYP, in rabbit liver
16 microsomes (Ripp et al., 1997) and human liver microsomes (Krause et al., 2003). Krause et al.,
17 (2003) also reported DCVC sulfoxidation by human cDNA-expressed FMO3, as well as
18 detection of FMO3 protein in human kidney samples. While Krause et al. (2003) were not able
19 to detect sulfoxidation in human kidney microsomes, the authors noted FMO3 expression in the
20 kidney was lower and more variable than that in the liver.
21 Sulfoxidation of NAcDCVC, by contrast, was found to be catalyzed predominantly, if not
22 exclusively, by CYP3A enzymes (Werner et al., 1996), whose expressions are highly
23 polymorphic in humans. Sulfoxidation of other haloalkyl mercapturic acid conjugates has also
24 been shown to be catalyzed by CYP3A (Werner et al., 1995a, b; Altuntas et al., 2004). While
25 Lash et al. (2000a) suggested that this pathway would be quantitatively minor because of the
26 relatively low CYP3 A levels in the kidney, no direct data exist to establish the relative
27 toxicological importance of this pathway relative to bioactivation of DCVC by P-lyase or FMO3.
28 However, the contribution of CYP3A in S-conjugate sulfoxidation to nephrotoxicity in vivo was
29 recently demonstrated by Sheffels et al. (2004) with fluoromethyl-2,2-difluoro-
30 l-(trifluoromethyl)vinyl ether (FDVE). In particular, in vivo production and urinary excretion of
31 FDVE-mercapturic acid sulfoxide metabolites were unambiguously established by mass
32 spectrometry, and CYP inducers/inhibitors increased/decreased nephrotoxicity in vivo while
33 having no effect on urinary excretion of metabolites produced through P-lyase (Sheffels et al.,
34 2004). These data suggest that, by analogy, sulfoxidation of NAcDCVC may be an important
35 bioactivating pathway.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-47 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
12
13
14
3.3.3.2.6. Tissue distribution of glutathione (GSH) metabolism. The sites of enzymatic
metabolism of TCE to the various GSH pathway-mediated metabolites are significant in
determining target tissue toxicity along this pathway. Figure 3-6 presents a schematic of
interorgan transport and metabolism of TCE along the glutathione pathway. TCE is taken up
either by the liver or kidney and conjugated to DCVG. The primary factors affecting TCE flux
via this pathway include high hepatic GST activity, efficient transport of DCVG from the liver to
the plasma or bile, high renal brush border and low hepatic GGT activities, and the capability for
GSH conjugate uptake into the renal basolateral membranes with limited or no uptake into liver
cell plasma membranes.
TCE
DCVG
DCVC
••Metabolism
NAcDCVC
>
Blood/
Plasma
Rest of
Body
Liver
|
Small
Intestine
Kidney
h Pilr
W
Blood/
Plasma
Rest of
Body
Liver
i ' E
^ i i
i i
i i
1
•-) -
i |
i i
i i
i i
1 I ^
i i
!
Blood/
Plasma
Rest of
Body
Liver
IT II
»
w
d flow
flow
|
i ' !
Small
Intestine
Kidney
*
Urine
1
i i
i i
!
1
i
w
\
DCVT
I \ \\
Small
Intestine
III
Kidney
" j
i
i
1
!
i
i
1
i
™
1
i
i
w
»
Blood/
Plasma
Rest of
Body
Liver
|
Small
Intestine
1
Kidney
nr^m
DCVCS NAcDCVCS
w
Figure 3-6. Interorgan TCE transport and metabolism via the GSH
pathway. See Figure 3-5 for enzymes involved in metabolic steps. Source:
Lash et al. (2000a, b); NRC (2006).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-48 DRAFT—DO NOT CITE OR QUOTE
-------
1 As discussed previously, GST activity is present in many different cell types. However,
2 the liver is the major tissue for GSH conjugation. GST activities in rat and mouse cytosolic
3 fractions were measured using l-chloro-2,4-dinitrobenzene, a GST substrate that is nonspecific
4 for particular isoforms (Lash et al., 1998). Specific activities (normalized for protein content) in
5 whole kidney cytosol were slightly less than those in the liver (0.64 compared to 0.52 mU/mg
6 protein for males and females). However, the much larger mass of the liver compared to the
7 kidney indicates that far more total GST activity resides in the liver. This is consistent with in
8 vitro data on TCE conjugation to DCVG, discussed previously (see Tables 3-23 and 3-24). For
9 instance, in humans, rats, and mice, liver cytosol exhibits greater DCVG production than kidney
10 cytosol. Distinct high- and low-affinity metabolic profiles were observed in the liver but not in
11 the kidney (see Table 3-24). In microsomes, human liver and kidney had similar rates of DCVG
12 production, while for rats and mice, the production in the liver was substantially greater.
13 According to studies by Lash et al. (1998, 1999b), the activity of GGT, the first step in
14 the conversion of DCVG to DCVC, is much higher in the kidney than the liver of mice, rats, and
15 humans, with most of the activity being concentrated in the microsomal, rather than the
16 cytosolic, fraction of the cell (see Table 3-25). In rats, this activity is quite high in the kidney but
17 is below the level of detection in the liver while the relative kidney to liver levels in humans and
18 mice were higher by 18- and up to 2,300-fold, respectively. Similar qualitative findings were
19 also reported in another study (Hinchman and Ballatori, 1990) when total organ GGT levels were
20 compared in several species (see Table 3-26). Cysteinylglycine dipeptidase was also
21 preferentially higher in the kidney than the liver of all tested species although the interorgan
22 differences in this activity (1-9-folds) seemed to be less dramatic than for GGT (see Table 3-26).
23 High levels of both GGT and dipeptidases have also been reported in the small intestine of rat
24 (Kozak and Tate, 1982) and mouse (Habib et al., 1996, 1998), as well as GGT in the human
25 jejunum (Fairman et al., 1977). No specific human intestinal Cysteinylglycine dipeptidase has
26 been identified; however, a related enzyme (EC 3.4.13.11) from human kidney microsomes has
27 been purified and studied (Adachi et al., 1989) while several human intestinal dipeptidases have
28 been characterized including a membrane dipeptidase (EC 3.4.13.19) which has a wide dipeptide
29 substrate specificity including Cysteinylglycine (Hooper et al., 1994; Ristoff and Larsson, 2007).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-49 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-25 GGT activity in liver and kidney subcellular fractions of mice,
rats, and humans
Species
Mouse
Rat
Human
Sex
Male
Female
Male
Female
Male
Tissue
Liver
Kidney
Liver
Kidney
Liver
Kidney
Liver
Kidney
Liver
Kidney
Cellular fraction
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Cytosol
Microsomes
Activity (mil/nig)
0.07 + 0.04
0.05 + 0.04
1.63+0.85
92.6+15.6
0.10 + 0.10
0.03 + 0.03
0.79 + 0.79
69.3 + 14.0
<0.02
<0.02
<0.02
1,570+100
<0.02
<0.02
<0.02
1,840 + 40
8.89 + 3.58
29
13.2+1.0
960 + 77
4
5
Source: Lashetal. (1998, 1999b).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-50 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-26 Multispecies comparison of whole-organ activity levels of GGT
and dispeptidase
Species
Rat
Mouse
Rabbit
Guinea pig
Pig
Macaque
Whole organ enzyme activity (umol substrate/organ)
Kidney
GGT
1,010 + 41
60.0 + 4.2
1,119+186
148 + 13
3,800 + 769
988
Dispeptidase
20.2 + 1.1
3.0 + 0.3
112 + 17
77+10
2,428 + 203
136
Liver
GGT
7.1 + 1.4
0.47 + 0.05
71.0 + 9.1
46.5 + 4.2
1,600 + 255
181
Dispeptidase
6.1+0.4
1.7 + 0.2
12.6+1.0
13.2+1.5
2,178 + 490
71
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Source: Hinchman and Ballatori (1990).
3.3.3.2.7. Sex- and species-dependent differences in glutathione (GSH) metabolism. Diverse
sex and species differences appear to exist in TCE metabolism via the glutathione pathway. In
rodents, rates of TCE conjugation to GSH in male rats and mice are higher than females (see
Table 3-23). Verma and Rana (2003) reported 2-fold higher GST activity values in liver cytosol
of female rats, compared to males, given 15 intraperitoneal injections of TCE over 30 days
period. This effect may be due to sex-dependent variation in induction, as GST activities in male
and female controls were similar. DCVG formation rates by liver and kidney subcellular
fractions were much higher in both sexes of mice than in rats and, except for mouse kidney
microsomes, the rates were generally higher in males than in females of the same species(see
Table 3-23).
In terms of species differences, comparisons at 1-2 mM TCE concentrations (see
Table 3-23) suggest that, in liver and kidney cytosol, the greatest DCVG production rate was in
humans, followed by mice and then rats. However, different investigators have reported
considerably different rates for TCE conjugation in human liver and kidney cell fractions. For
instance, values in Table 3-23 from Lash et al. (1999a) are between two and five orders of
magnitude higher than those reported by Green et al. (1997a). (The rates of DCVG formation by
liver cytosol from male F344 rat, male B6C3F1 mouse, and human were 1.62, 2.5, and
0.19 pmol/minute/mg protein, respectively, while there were no measurable activity in liver
microsomes or subcellular kidney fractions [Green et al., 1997a]). The reasons for such
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-51 DRAFT—DO NOT CITE OR QUOTE
-------
1 discrepancies are unclear but may be related to different analytical methods employed such as
2 detection of radiolabled substrate vs. derivatized analytes (Lash et al., 2000a).
3 Expression of GGT activity does not appear to be influenced by sex (see Table 3-25); but
4 species differences in kidney GGT activity are notable with rat subcellular fractions exhibiting
5 the highest levels and mice and humans exhibiting about 4-6% and 50%, respectively, of rat
6 levels (Lash et al., 1999a, 1998). Table 3-26 shows measures of whole-organ GGT and
7 dispeptidase activities in rats, mice, guinea pigs, rabbits, pigs, and monkeys. These data show
8 that the whole kidney possesses higher activities than liver for these enzymes, despite the
9 relatively larger mass of the liver.
10 As discussed above, the three potential bioactivating pathways subsequent to the
11 formation of DCVC are catalyzed by p-lyase, FMO3 or CYP3A. Lash et al. (2000a) compared
12 in vitro P-lyase activities and kinetic constants (when available) for kidney of rats, mice, and
13 humans. They reported that variability of these values spans up to two orders of magnitude
14 depending on substrate, analytical method used, and research group. Measurements of rat,
15 mouse, and human P-lyase activities collected by the same researchers following
16 tetrachloroethylene exposure (Green et al., 1990) resulted in higher KM and lower VMAX values
17 for mice and humans than rats. Further, female rats exhibited higher KM and lower VMAX values
18 than males
19 With respect to FMO3, Ripp et al. (1999) found that this enzyme appeared catalytically
20 similar across multiple species, including humans, rats, dogs, and rabbits, with respect to several
21 substrates, including DCVC, but that there were species differences in expression. Specifically,
22 in male liver microsomes, rabbits had 3-fold higher methionine S-oxidase activity than mice and
23 dogs had 1.5-fold higher activity than humans and rats. Species differences were also noted in
24 male and female kidney microsomes; rats exhibited 2- to 6-fold higher methionine S-oxidase
25 activity than the other species. Krause et al. (2003) detected DCVC sulfoxidation in incubations
26 with human liver microsomes but did not in an incubation with a single sample of human kidney
27 microsomes. However, FMO3 expression in the 26 human kidney samples was found to be
28 highly variable, with a range of 5-6-fold (Krause et al., 2003).
29 No data on species differences in CYP3A-mediated sulfoxidation of NAcDCVC are
30 available. However, Altuntas et al. (2004) examined sulfoxidation of cysteine and mercapturic
31 acid conjugates of FDVE (fluoromethyl-2,2-difluoro-l-(trifluoromethyl)vinyl ether) in rat and
32 human liver and kidney microsomes. They reported that the formation of sulfoxides from the
33 mercapturates 7V-Ac-FFVC and (Z)-TV-Ac-FFVC (FFVC is (E,Z)-S-(l-fluoro-2-fluoromethoxy-
34 2-(trifluoromethyl)vinyl-Lcysteine) were greatest in rat liver microsomes, and 2- to 30-fold
35 higher than in human liver microsomes (which had high variability). Sulfoxidation of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-52 DRAFT—DO NOT CITE OR QUOTE
-------
1 TV-Ac-FFVC could not be detected in neither rat nor human kidney microsomes, but
2 sulfoxidation of (Z)-TV-Ac-FFVC was detected in both rat and human kidney microsomes at rates
3 comparable to human liver microsomes. Using human- and rat-expressed CYP3 A, Altuntas et
4 al. (2004) reported that rates of sulfoxidation of (Z)-TV-Ac-FFVC were comparable in human
5 CYP3 A4 and rat CYP3 Al and CYP3 A2., but that only rat CYP3 Al and A2 catalyzed
6 sulfoxidation of 7V-Ac-FFVC. As the presence or absence of the species differences in
7 mercapturate sulfoxidation appear to be highly chemical-specific, no clear inferences can be
8 made as to whether species differences exist for sulfoxidation of NAcDCVC
9 Also relevant to assess the flux through the various pathways are the rates of
10 7V-acetylation and de-acetylation of DCVC. This is demonstrated by the results of Elfarra and
11 Hwang (1990) using S-(2-benzothiazolyl)-L-cysteine as a marker for p-lyase metabolism in rats,
12 mice, hamsters, and guinea pigs. Guinea pigs exhibited about 2-fold greater flux through the
13 P-lyase pathway, but this was not attributable to higher P-lyase activity. Rather, guinea pigs
14 have relatively low 7V-acetylation and high deacetylation activities, leading to a high level of
15 substrate recirculation (Lau et al., 1995). Thus, a high 7V-deacetylase:7V-acetylase activity ratio
16 may favor DCVC recirculation and subsequent metabolism to reactive species. In human,
17 Wistar rat, Fischer rat, and mouse cytosol, deacetylation rates for NAcDCVC varied less than
18 3-fold (0.35, 0.41, 0.61, and 0.94 nmol DCVC formed/minute/mg protein in humans, rats, and
19 mice) (Birner et al., 1993). However, similar experiments have not been carried out for
20 7V-acetylation of DCVC, so the balance between its 7V-acetylation and de-acetylation has not been
21 established.
22
23 3.3.3.2.8. Human variability and susceptibility in glutathione (GSH) conjugation. Knowledge
24 of human variability in metabolizing TCE through the glutathione pathway is limited to in vitro
25 comparisons of variance in GST activity rates. Unlike CYP-mediated oxidation, quantitative
26 differences in the polymorphic distribution or activity levels of GST isoforms in humans are not
27 presently known. However, the available data (Lash et al., 1999a, b) do suggest that significant
28 variation in GST-mediated conjugation of TCE exists in humans. In particular, at a single
29 substrate concentration of 1 mM, the rate of GSH conjugation of TCE in human liver cytosol
30 from 9 male and 11 females spanned a range of 2.4-fold (34.7-83.6 nmol DCVG
31 formed/20-minute/mg protein) (Lash et al., 1999b). In liver microsomes from 5 males and
32 15-females, the variation in activity was 6.5-fold (9.9-64.6 nmol DCVG formed/20 minute/mg
33 protein). No sex-dependent variation was identified. Despite being less pronounced than the
34 known variability in human CYP-mediated oxidation, the impact on risk assessment of the
35 variability in GSH conjugation to TCE is currently unknown especially in the absence of data on
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-53 DRAFT—DO NOT CITE OR QUOTE
-------
1 variability for N-acetylation and bioactivation via p-lyase, FMO3, or CYP3 A in the human
2 kidney.
O
4 3.3.3.3. Relative Roles of the Cytochrome P450 (CYP) and Glutathione (GSH) Pathways
5 In vivo mass balance studies in rats and mice, discussed above, have shown
6 unequivocally that in these species, CYP oxidation of TCE predominates over GSH conjugation.
7 In these species, at doses from 2 to 2,000 mg/kg of [14C]TCE, the sum of radioactivity in exhaled
8 TCE, urine, and exhaled CC>2 constitutes 69-94% of the dose, with the vast majority of the
9 radioactivity in urine (95-99%) attributable to oxidative metabolites (Dekant et al., 1986a, 1984;
10 Green and Prout, 1985; Prout et al., 1995). The rest of the radioactivity was found mostly in
11 feces and the carcass. More rigorous quantitative limits on the amount of GSH conjugation
12 based on in vivo data such as these can be obtained using PBPK models, discussed in
13 Section 3.5.
14 Comprehensive mass-balance studies are unavailable in humans. DCVG and DCVC in
15 urine have not been detected in any species, while the amount of urinary NAcDCVC from
16 human exposures is either below detection limits or very small from a total mass balance point of
17 view (Birner et al., 1993; Bernauer et al., 1996; Lash et al., 1999b; Bloemen et al., 2001). For
18 instance, the ratio of primary oxidative metabolites (TCA + TCOH) to NAcDCVC in urine of
19 rats and humans exposed to 40-160 ppm (215 to 860 mg/m3) TCE heavily favored oxidation,
20 resulting in ratios of 986-2,562:1 in rats and 3,292-7,163:1 in humans (Bernauer et al., 1996).
21 Bloemen et al. (2001) reported that at most 0.05% of an inhaled TCE dose would be excreted as
22 NAcDCVC, and concluded that this suggested TCE metabolism by GSH conjugation was of
23 minor importance. While it is a useful biomarker of exposure and an indicator of GSH
24 conjugation, NAcDCVC may capture only a small fraction of TCE flux through the GSH
25 conjugation pathway due to the dominance of bioactivating pathways (Lash et al., 2000a).
26 A number of lines of evidence suggest that the amount of TCE conjugation to GSH in
27 humans, while likely smaller than the amount of oxidation, may be much more substantial than
28 analysis of urinary mercapturates would suggest. In Table 3-27, in vitro estimates of the VMAX,
29 KM, and clearance (VMAX/KM) for hepatic oxidation and conjugation of TCE are compared in a
30 manner that accounts for differences in cytosolic and microsomal partitioning and protein
31 content. Surprisingly, the range of in vitro kinetic estimates for oxidation and conjugation of
32 TCE substantially overlap, suggesting similar flux through each pathway, though with high
33 interindividual variation. The microsomal and cytosolic protein measurements of GSH
34 conjugation should be caveated by the observation by Lash et al. (1999a) that GSH conjugation
35 of TCE was inhibited by -50% in the presence of oxidation. Note that this comparison cannot be
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-54 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
made in rats and mice because in vitro kinetic parameters for GSH conjugation in the liver are
not available in those species (only activity at 1 or 2 mM have been measured).
Table 3-27 Comparison of hepatic in vitro oxidation and conjugation of
TCE
Cellular or
subcellular
fraction
Hepatocytes
Liver
microsomes
Liver
cytosol
VMAX
(nmol TCE
metabolized/min/g tissue)
Oxidation
10.0-68.4
6.1-111
-
-
GSH
conjugation
16-25
45
380
KM
(uM in blood)
Oxidation
22.1-198
2.66-11. la
71.0-297b
-
-
GSH
conjugation
16-47
5.9a
157b
4.5a
22.7b
VMAX/KM
(mL/min/g tissue)
Oxidation
0.087-1.12
1.71-28.2a
0.064-1. 06b
-
-
GSH
conjugation
0.55-1.0
7.6a
0.29b
84a
16.7b
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Note: When biphasic metabolism was reported, only high affinity pathway is shown here.
Conversion assumptions for VMAX:
Hepatocelluarity of 99 million cells/g liver (Barter et al., 2007);
Liver microsomal protein content of 32 mg protein/g tissue (Barter et al., 2007); and
Liver cytosolic protein content of 89 mg protein/g tissue (based on rats: Prasanna et al., 1989;
van Bree et al., 1990).
Conversion assumptions for KM:
For hepatocytes, KM in headspace converted to KM in blood using blood:air partition coefficient of 9.5
(reported range of measured values 6.5-12.1, Table 3-1);
For microsomal protein, option (a) assumes KM in medium is equal to KM in tissue, and converts to
KM in blood by using a liverblood partition coefficient of 5 (reported ranges of measured values
3.6-5.9, Table 3-8), and option (b) converts KM in medium to KM in air using the measured
microsomal protein:air partition coefficient of 1.78 (Lipscomb et al., 1997), and then converts to
KM in blood by using the blood:air partition coefficient of 9.5; and
For cytosolic protein, option (a) assumes KM in medium is equal to KM in tissue, and converts to KM
in blood by using a liverblood partition coefficient of 5 (reported ranges of measured values
3.6-5.9, Table 3-8), and option (b) assumes KM in medium is equal to KM in blood, so no
conversion is necessary.
Furthermore, as shown earlier in Table 3-22, the human in vivo data of Lash et al.
(1999a) show blood concentrations of DCVG similar, on a molar basis, to that of TCE, TCA, or
TCOH, suggesting substantial conjugation of TCE. In addition, these data give a lower limit as
to the amount of TCE conjugated. In particular, by multiplying the peak blood concentration of
DCVG by the blood volume, a minimum amount of DCVG in the body at that time can be
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-55 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
derived (i.e., assuming the minimal empirical distribution volume equal to the blood volume).
As shown in Table 3-28, this lower limit amounts to about 0.4-3.7% of the inhaled TCE dose.
Since this is the minimum amount of DCVG in the body at a single time point, the total amount
of DCVG formed is likely to be substantially greater owing to possible distribution outside of the
blood as well as the metabolism and/or excretion of DCVG. Lash et al. (1999) found levels of
urinary mercapturates were near or below the level of detection of 0.19 uM, results that are
consistent with those of Bloemen et al. (2001), who reported urinary concentrations below
0.04 uM at 2- to 4-fold lower cumulative exposures. Taken together, these results confirm the
suggestion by Lash et al. (2000a) that NAcDCVC is a poor quantitative marker for the flux
through the GSH pathway.
Table 3-28 Estimates of DCVG in blood relative to inhaled TCE dose in
humans exposed to 50 and 100 ppm (269 and 537 mg/m3; Fisher et al., 1998;
Lash et al., 1999)
Sex exposure
Estimated inhaled TCE dose
(mmol)a
Estimated peak amount of DCVG in
blood (mmol)b
Males
50 ppm x 4 hours
100 ppm x 4 hours
3.53
7.07
0.11+0.08
0.26 + 0.08
Females
50 ppm x 4 hours
100 ppm x 4 hours
2.36
4.71
0.010 + 0
0.055 + 0.027
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
alnhaled dose estimated by (50 or 100 ppm)/(24,450 ppm/mM) x (240 min) x QP, where alveolar ventilation rate QP
is 7.2 L/min for males and 4.8 L/min for females. QP is calculated as (VT - VD) x fR with the following
respiratory parameters: tidal volume VT (0.75 L for males, 0.46 L for females), dead space VD (0.15 L for males,
0.12 L for females), and respiration frequency fR (12 min"1 for males, 14 min"1 for females) (assumed sitting,
awake from The International Commission on Radiological Protection [ICRP], 2002).
bPeak amount of DCVG in blood estimated by multiplying the peak blood concentration by the estimated blood
volume: 5.6 L in males and 4.1 L in females (ICPJ3, 2002).
In summary, TCE oxidation is likely to be greater quantitatively than conjugation with
GSH in mice, rats, and humans. However, the flux through the GSH pathway, particularly in
humans, may be greater by an order of magnitude or more than the <0.1% typically excreted of
NAcDCVC in urine. This is evidenced both by a direct comparison of in vitro rates of oxidation
and conjugation, as well as by in vivo data on the amount of DCVG in blood. PBPK models can
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-56 DRAFT—DO NOT CITE OR QUOTE
-------
1 be used to more quantitatively synthesize these data and put more rigorous limits on relative
2 amount TCE oxidation and conjugation with GSH. Such analyses are discussed in Section 3.5.
O
4 3.4. TRICHLOROETHYLENE (TCE) EXCRETION
5 This section discusses the major routes of excretion of TCE and its metabolites in exhaled
6 air, urine, and feces. Unmetabolized TCE is eliminated primarily via exhaled air. As discussed
7 in Section 3.3, the majority of TCE absorbed into the body is eliminated by metabolism. With
8 the exception of CC>2, which is eliminated solely via exhalation, most TCE metabolites have low
9 volatility and, therefore, are excreted primarily in urine and feces. Though trace amounts of TCE
10 metabolites have also been detected in sweat and saliva (Bartonicek et al., 1962), these excretion
11 routes are likely to be relatively minor.
12
13 3.4.1. Exhaled Air
14 In humans, pulmonary elimination of unchanged trichloroethylene and other volatile
15 compounds is related to ventilation rate, cardiac output, and the solubility of the compound in
16 blood and tissue, which contribute to final exhaled air concentration of TCE. In their study of
17 the impact of workload on TCE absorption and elimination, Astrand and Ovrum (1976)
18 characterized the postexposure elimination of TCE in expired breath. TCE exposure (540 or
19 1,080 mg/m3; 100 or 200 ppm) was for a total of 2 hours, at workloads from 0 to 150 Watts.
20 Elimination profiles were roughly equivalent among groups, demonstrating a rapid decline in
21 TCE concentrations in expired breath postexposure (see Table 3-29).
22 The lung clearance of TCE represents the volume of air from which all TCE can be
23 removed per unit time, and is a measure of the rate of excretion via the lungs. Monster et al.
24 (1976) reported lung clearances ranging from 3.8 to 4.9 L/minute in four adults exposed at rest to
25 70 ppm and 140 ppm of trichloroethylene for four hours. Pulmonary ventilation rates in these
26 individuals at rest ranged from 7.7-12.3 L/minute. During exercise, when ventilation rates
27 increased to 29-30 L/minute, lung clearance was correspondingly higher, 7.7-12.3 L/minute.
28 Under single and repeated exposure conditions, Monster et al. (1976, 1979) reported from
29 7-17% of absorbed TCE excreted in exhaled breath.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-57 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-29. Concentrations of TCE in expired breath from inhalation-
exposed humans (Astrand and Ovrum, 1976)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Time
postexposure
0
30
60
90
120
300
420
19 hours
Alveolar air
I*
459 + 44
70 + 5
40 + 4
35 + 9
31 + 8
8 + 1
5 + 0.5
2 + 0.3
II
244+16
51 + 3
28 + 2
21 + 1
16+1
9 + 2
4 + 0.5
2 + 0.2
III
651+53
105 + 18
69 + 8
55+2
45 + 1
14 + 2
8+1.3
4 + 0.5
* Roman numerals refer to groups assigned different workloads.
Concentrations are in mg/m3 for expired air.
Pulmonary elimination of unchanged trichloroethylene at the end of exposure is a
first-order diffusion process across the lungs from blood into alveolar air, and it can be thought
of as the reversed equivalent of its uptake from the lungs. Exhaled pulmonary excretion occurs
in several distinct (delayed) phases corresponding to release from different tissue groups, at
different times. Sato et al. (1977) detected 3 first-order phases of pulmonary excretion in the
first 10 hours after exposure to 100 ppm for 4 hours, with fitted half-times of pulmonary
elimination of 0.04 hour, 0.67 hour, and 5.6 hours, respectively. Opdam (1989) sampled alveolar
air up to 20-310 hours after 29-62 minute exposures to 6-38 ppm, and reported terminal half-
lives of 8-44 hours at rest. Chiu et al. (2007) sampled alveolar air up to 100 hours after 6-hour
exposures to 1 ppm and reported terminal half-lives of 14-23 hours. The long terminal half-time
of TCE pulmonary excretion indicates that a considerable time is necessary to completely
eliminate the compound, primarily due to the high partitioning to adipose tissues (see
Section 3.2).
As discussed above, several studies (Dekant et al., 1986a, 1984; Green and Prout, 1985;
Prout et al., 1985) have investigated the disposition of [14C]TCE in rats and mice following
gavage administrations (see Section 3.3.2). These studies have reported CC>2 as an exhalation
excretion product in addition to unchanged TCE. With low doses, the amount of TCE excreted
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-58 DRAFT—DO NOT CITE OR QUOTE
-------
1 unchanged in exhaled breath is relatively low. With increasing dose in rats, a disproportionately
2 increased amount of radiolabel is expired as unchanged TCE. This may indicate saturation of
3 metabolic activities in rats at doses 200 mg/kg and above, which is perhaps only minimally
4 apparent in the data from mice. In addition, exhaled air TCE concentration has been measured
5 after constant inhalation exposure for 2 hours to 50 or 500 ppm in rats (Dallas et al., 1991), and
6 after dermal exposure in rats and humans (Poet, 2000). Exhaled TCE data from rodents and
7 humans have been integrated into the PBPK model presented in Section 3.5.
8 Finally, TCOH is also excreted in exhaled breath, though at a rate about 10,000-fold
9 lower than unmetabolized TCE (Monster et al., 1976, 1979).
10
11 3.4.2. Urine
12 Urinary excretion after TCE exposure consists predominantly of the metabolites TCA
13 and TCOH, with minor contributions from other oxidative metabolites and GSH conjugates.
14 Measurements of unchanged TCE in urine have been at or below detection limits (e.g.,
15 Fisher et al., 1998; Chiu et al., 2007). The recovery of urinary oxidative metabolites in mice,
16 rats, and humans was addressed earlier (see Section 3.3.2) and will not be discussed here.
17 Because of their relatively long elimination half-life, urinary oxidative metabolites have
18 been used as an occupational biomarker of TCE exposure for many decades
19 (Ikeda and Imamura, 1973; Carrieri, 2007). Ikeda and Imamura (1973) measured total trichloro
20 compounds (TTC), TCOH and TCA, in urine over three consecutive postexposure days for
21 4 exposure groups totaling 24 adult males and one exposure group comprising 6 adult females.
22 The elimination half-life for TTC ranged 26.1 to 48.8 hours in males and was 50.7 hours in
23 females. The elimination half-life for TCOH was 15.3 hours in the only group of males studied
24 and was 42.7 hours in females. The elimination half-life for TCA was 39.7 hours in the only
25 group of males studied and was 57.6 hours in females. These authors compared their results to
26 previously published elimination half-lives for TTC, TCOH, and TCA. Following experimental
27 exposures of groups of 2 to 5 adults, elimination half-lives ranged 31-50 hours for TTC;
28 19-29 hours for TCOH; and 36-55 hours for TCA (Bartonicek, 1962; Stewart et al., 1970;
29 Nomiyama and Nomiyama, 1971; Ogata et al., 1971). The urinary elimination half-life of TCE
30 metabolites in a subject who worked with and was addicted to sniffing TCE for 6-8 years
31 approximated 49.7 hours for TCOH, 72.6 hours for TCA, and 72.6 hours for TTC (Ikeda et al.,
32 1971).
33 The quantitative relationship between urinary concentrations of oxidative metabolites and
34 exposure in an occupational setting was investigated by Ikeda (1977). This study examined the
35 urinary elimination of TCE and metabolites in urine of 51 workers from 10 workshops. The
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-59 DRAFT—DO NOT CITE OR QUOTE
-------
1 concentration of TCA and TCOH in urine demonstrated a marked concentration-dependence,
2 with concentrations of TCOH being approximately twice as high as those for TCA. Urinary
3 half-life values were calculated for 6 males and 6 females from 5 workshops; males were
4 intermittently exposed to 200 ppm and females were intermittently exposed to 50 ppm
5 (269 mg/m3). Urinary elimination half-lives for TTC, TCOH and TCA were 26.1, 15.3, and
6 39.7 hours; and 50.7, 42.7 and 57.6 hours in males and females, respectively, which were similar
7 to the range of values previously reported. These authors estimated that urinary elimination of
8 parent TCE during exposure might account for one-third of the systemically absorbed dose.
9 Importantly, urinary TCA exhibited marked saturation at exposures higher than 50 ppm.
10 Because neither TTC nor urinary TCOH (in the form of the glucuronide TCOG) showed such an
11 effect, this saturation cannot be due to TCE oxidation itself, but must rather be from one of the
12 metabolic processes forming TCA from TCOH. Unfortunately, since biological monitoring
13 programs usually measure only urinary TCA, rather than TTC, urinary TCA levels above around
14 150 mg/L cannot distinguish between exposures at 50 ppm and at much higher concentrations.
15 It is interesting to attempt to extrapolate on a cumulative exposure basis the Ikeda (1977)
16 results for urinary metabolites obtained after occupational exposures at 50 ppm to the controlled
17 exposure study by Chiu et al. (2007) at 1.2 ppm for 6 hours (the only controlled exposure study
18 for which urinary concentrations, rather than only cumulative excretion, are available). Ikeda
19 (1977) reported that measurements were made during the second half of the week, so one can
20 postulate a cumulative exposure duration of 20-40 hours. At 50 ppm, Ikeda (1977) report a
21 urinary TCOH concentration of about 290 mg/L, so that per ppm-hour, the expected urinary
22 concentration would be 290/(50 x 20-40) = 0.145-0.29 mg/L-ppm-hour. The cumulative
23 exposure in Chiu et al. (2007) is 1.2 x 6 = 7.2 ppm-hour, so the expected urinary TCOH
24 concentration would be 7.2 x (0.145-0.29) = 1.0-2.1 mg/L. This estimate is somewhat
25 surprisingly consistent with the actual measurements of Chiu et al. (2007) during the first day
26 postexposure, which ranged from 0.8-1.2 mg/L TCOH in urine.
27 On the other hand, extrapolation of TCA concentrations was less consistent. At 50 ppm,
28 Ikeda (1977) report a urinary TCA concentration of about 140 mg/L, so that per ppm-hour, the
29 expected urinary concentration would be 140/(50 x 20-40) = 0.07-0.14 mg/L-ppm-hour. The
30 cumulative exposure in Chiu et al. (2007) is 1.2 x 6 = 7.2 ppm-hour, so the expected urinary
31 TCA concentration would be 7.2 x (0.07-0.14) = 0.5-1.0 mg/L, whereas Chiu et al. (2007)
32 reported urinary TCA concentrations on the first day after exposure of 0.03-0.12 mg/L.
33 However, as noted in Chiu et al. (2007), relative urinary excretion of TCA was 3- to 10-fold
34 lower in Chiu et al. (2007) than other studies at exposures 50-140 ppm, which may explain part
35 of the discrepancies. However, this may be due in part to saturation of many urinary TCA
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-60 DRAFT—DO NOT CITE OR QUOTE
-------
1 measurements, and, furthermore, interindividual variance, observed to be substantial in Fisher et
2 al. (1998), cannot be ruled out.
3 Urinary elimination kinetics have been reported to be much faster in rodents than in
4 humans. For instance, adult rats were exposed to 50, 100, or 250 ppm (269, 537, or
5 1,344 mg/m3) via inhalation for 8 hours or were administered an i.p. injection (1.47 g/kg) and the
6 urinary elimination of total trichloro compounds was followed for several days (Ikeda and
7 Imamura, 1973). These authors calculated urinary elimination half-lives of 14.3-15.6 hours for
8 female rats and 15.5-16.6 hours for male rats; the route of administration did not appear to
9 influence half-life value. In other rodent experiments using orally administered radiolabeled
10 TCE, urinary elimination was complete within one or two days after exposure (Dekant et al.,
11 1986a, 1984; Green and Prout, 1985; Prout et al., 1985).
12
13 3.4.3. Feces
14 Fecal elimination accounts for a small percentage of TCE as shown by limited
15 information in the available literature. Bartonicek (1962) exposed 7 human volunteers to
16 1.042 mg TCE/L air for 5 hours and examined TCOH and TCA in feces on the third and seventh
17 day following exposure. The mean amount of TCE retained during exposure was 1,107 mg,
18 representing 51-64% (mean 58%) of administered dose. On the third day following TCE
19 exposure, TCOH and TCA in feces demonstrated mean concentrations of 17.1 and
20 18.5 mg/100 grams feces, similar to concentrations in urine. However, because of the 10-fold
21 smaller daily rate of excretion of feces relative to urine, this indicates fecal excretion of these
22 metabolites is much less significant than urinary excretion. Neither TCOH nor TCA was
23 detected in feces on the seventh day following exposure.
24 In rats and mice, total radioactivity has been used to measure excretion in feces after oral
25 gavage TCE administration in corn oil, but since the radiolabel was not characterized it is not
26 possible to determine whether the fecal radiolabel in feces represented unabsorbed parent
27 compound, excreted parent compound, and/or excreted metabolites. Dekant et al. (1984)
28 reported mice eliminated 5% of the total administered TCE, while rats eliminated 2% after oral
29 gavage. Dekant et al. (1986a) reported a dose response related increase in fecal elimination with
30 dose, ranging between 0.8-1.9% in rats and 1.6-5% in mice after oral gavage in corn oil. Due to
31 the relevant role of CYP2E1 in the metabolism of TCE (see Section 3.3.3.1.6), Kim and
32 Ghanayem (2006) compared fecal elimination in both wild-type and CYP2E1 knockouts mice
33 and reported fecal elimination ranging between 4.1-5.2% in wild-type and 2.1-3.8% in
34 knockout mice exposed by oral gavage in aqueous solution.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-61 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.5. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODELING OF
2 TRICHLOROETHYLENE (TCE) AND ITS METABOLITES
3 3.5.1. Introduction
4 PBPK models are extremely useful tools for quantifying the relationship between
5 external measures of exposure and internal measures of lexicologically relevant dose. In
6 particular, for the purposes of this assessment, PBPK models are evaluated for the following:
7 (1) providing additional quantitative insights into the ADME of TCE and metabolites described
8 in the sections above; (2) cross-species pharmacokinetic extrapolation of rodent studies of both
9 cancer and noncancer effects, (3) exposure-route extrapolation; and (4) characterization of
10 human pharmacokinetic variability. The following sections first describe and evaluate previous
11 and current TCE PBPK modeling efforts, then discuss the insights into ADME (1, above), and
12 finally present conclusions as to the utility of the model to predict internal doses for use in dose-
13 response assessment (2-4, above).
14
15 3.5.2. Previous Physiologically Based Pharmacokinetic (PBPK) Modeling of
16 Trichloroethylene (TCE) for Risk Assessment Application
17 TCE has an extensive number of both in vivo pharmacokinetic and PBPK modeling
18 studies (see Chiu et al., 2006, supplementary material, for a review). Models previously
19 developed for occupational or industrial hygiene applications are not discussed here but are
20 reviewed briefly in Clewell et al. (2000). Models designed for risk assessment applications have
21 focused on descriptions of TCE and its major oxidative metabolites TCA, TCOH, and TCOG.
22 Most of these models were extensions of the "first generation" of models developed by Fisher
23 and coworkers (Allen and Fisher, 1993; Fisher et al., 1991) in rats, mice, and humans. These
24 models, in turn, are based on a Ramsey and Andersen (1984) structure with flow-limited tissue
25 compartments and equilibrium gas exchange, saturable Michaelis-Menten kinetics for oxidative
26 metabolism, and lumped volumes for the major circulating oxidative metabolites TCA and
27 TCOH. Fisher and coworkers updated their models with new in vivo and in vitro experiments
28 performed in mice (Abbas and Fisher, 1997; Greenberg et al., 1999) and human volunteers
29 (Fisher et al., 1998) and summarized their findings in Fisher (2000). Clewell et al. (2000) added
30 enterohepatic recirculation of TCOG and pathways for local oxidative metabolism in the lung
31 and GST metabolism in the liver. While Clewell et al. (2000) does not include the updated
32 Fisher data, they have used a wider set of in vivo and in vitro mouse, rat, and human data than
33 previous models. Finally, Bois (2000a, b) performed re-estimations of PBPK model parameters
34 for the Fisher and Clewell models using a Bayesian population approach (Gelman et al., 1996,
35 and discussed further below).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 As discussed in Rhomberg (2000), the choice as to whether to use the Fisher, Clewell,
2 and Bois models for cross-species extrapolation of rodent cancer bioassays led to quantitative
3 results that differed by as much as an order of magnitude. There are a number of differences in
4 modeling approaches that can explain their differing results. First, the Clewell et al. (2000)
5 model differed structurally in its use of single-compartment volume-of-distribution models for
6 metabolites as opposed to the Fisher (2000) models' use of multiple physiologic compartments.
7 Also, the Clewell et al. (2000) model, but not the Fisher models, includes enterohepatic
8 recirculation of TCOH/TCOG (although reabsorption was set to zero in some cases). In addition
9 to structural differences in the models, the input parameter values for these various models were
10 calibrated using different subsets of the overall in vivo database (see Chiu et al., 2006,
11 supplementary material, for a review). The Clewell et al. (2000) model is based primarily on a
12 variety of data published before 1995; the Fisher (2000) models were based primarily on new
13 studies conducted by Fisher and coworkers (after 1997); and the Bois (2000a, b) re-estimations
14 of the parameters for the Clewell et al. (2000) and Fisher (2000) models used slightly different
15 data sets than the original authors. The Bois (2000a, b) re-analyses also led to somewhat
16 different parameter estimates than the original authors, both because of the different data sets
17 used as well as because the methodology used by Bois allowed many more parameters to be
18 estimated simultaneously than were estimated in the original analyses.
19 Given all these methodological differences, it is not altogether surprising that the
20 different models led to different quantitative results. Even among the Fisher models themselves,
21 Fisher (2000) noted some inconsistencies, including differing estimates for metabolic parameters
22 between mouse gavage and inhalation experiments. These authors included possible
23 explanations for these inconsistencies: the impact of corn oil vehicle use during gavage
24 (Staats et al., 1991) and the impact of a decrease in ventilation rate in mice due to sensory
25 irritation during the inhalation of solvents (e.g., Stadler and Kennedy, 1996).
26 As discussed in a report by the National Research Council (NRC, 2006), several
27 additional PBPK models relevant to TCE pharmacokinetics have been published since 2000 and
28 are reviewed briefly here. Poet et al. (2000) incorporated dermal exposure to TCE in PBPK
29 models in rats and humans, and published in vivo data in both species from dermal exposure
30 (Thrall et al., 2000; Poet et al., 2000). Albanese et al. (2002) published a series of models with
31 more complex descriptions of TCE distribution in adipose tissue but did not show comparisons
32 with experimental data. Simmons et al. (2002) developed a PBPK model for TCE in the
33 Long-Evans rat that focused on neurotoxicity endpoints and compared model predictions with
34 experimentally determined TCE concentrations in several tissues—including the brain. Keys et
35 al. (2003) investigated the lumping and unlumping of various tissue compartments in a series of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-63 DRAFT—DO NOT CITE OR QUOTE
-------
1 PBPK models in the rat and compared model predictions with TCE tissue concentrations in a
2 multitude of tissues. Although none of these TCE models included metabolite descriptions, the
3 experimental data was available for either model or evaluation. Finally, Keys et al. (2004)
4 developed a model for DCA in the mouse that included a description of suicide inhibition of
5 GST-zeta, but this model was not been linked to TCE.
6
7 3.5.3. Development and Evaluation of an Interim "Harmonized" Trichloroethylene (TCE)
8 Physiologically Based Pharmacokinetic (PBPK) Model
9 Throughout 2004, U.S. EPA and the U.S. Air Force jointly sponsored an integration of
10 the Fisher, Clewell, and Bois modeling efforts (Hack et al., 2006). In brief, a single interim
11 PBPK model structure combining features from both the Fisher and Clewell models was
12 developed and used for all 3 species of interest (mice, rats, and humans). An effort was made to
13 combine structures in as simple a manner as possible; the evaluation of most alternative
14 structures was left for future work. The one level of increased complexity introduced was
15 inclusion of species- and dose-dependent TCA plasma binding, although only a single in vitro
16 study of Lumpkin et al. (2003) was used as parameter inputs. As part of this joint effort, a
17 hierarchical Bayesian population analysis using Markov chain Monte Carlo (MCMC) sampling
18 (similar to the Bois [2000a, b] analyses) was performed on the revised model with a
19 cross-section of the combined database of kinetic data to provide estimates of parameter
20 uncertainty and variability (Hack et al., 2006). Particular attention was given to using data from
21 each of the different efforts, but owing to time and resource constraints, a combined analysis of
22 all data was not performed. The results from this effort suggested that a single model structure
23 could provide reasonable fits to a variety of data evaluated for TCE and its major oxidative
24 metabolites TCA, TCOH, and TCOG. However, in many cases, different parameter values—
25 particularly for metabolism—were required for different studies, indicating significant
26 interindividual or interexperimental variability. In addition, these authors concluded that
27 dosimetry of DCA, conjugative metabolites, and metabolism in the lung remained highly
28 uncertain (Hack et al., 2006).
29 Subsequently, U.S. EPA conducted a detailed evaluation of the Hack et al. (2006) model
30 that included (1) additional model runs to improve convergence; (2) evaluation of posterior
31 distributions for population parameters; and (3) comparison of model predictions both with the
32 data used in the Hack et al. (2006) analysis as well as with additional data sets identified in the
33 literature. Appendix A provides the details and conclusions of this evaluation, briefly
34 summarized in Table 3-30, along with their pharmacokinetic implications.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-64 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-30. Conclusions from evaluation of Hack et al. (2006), and implications for PBPK model development
to
vo £;•
I
I
§
***.
£3'
1
TO'
i
^§
H I
O >
HH Oq
H TO
O
H
W
Conclusion from evaluation of Hack et al. (2006) model
For some model parameters, posterior distributions were somewhat inconsistent
with the prior distributions.
• For parameters with strongly informative priors (e.g., tissue volumes and
flows), this may indicate errors in the model.
• For many parameters, the prior distributions were based on visual fits to the
same data. If the posteriors are inconsistent, then that means they priors
were "inappropriately" informative, and, thus, the same data was used
twice.
A number of data sets involve TCE (i.a., portal vein), TCA (oral, i.v.), and
TCOH (oral, i.v.) dosing routes that are not currently in the model, but could be
useful for calibration.
TCE concentrations in blood, air, and tissues well-predicted only in rats, not in
mice and humans. Specifically:
• In mice, the oral uptake model could not account for the time-course of
several data sets. Blood TCE concentrations after inhalation consistently
over-predicted.
• In rats, tissue concentrations measured in data not used for calibration were
accurately predicted.
• In humans, blood and air TCE concentrations were consistently over-
predicted in the majority of (but not all) data sets.
Total metabolism appears well-predicted in rats and mice based on closed
chamber data, but required significantly different VMAX values between dose
groups. Total recovery in humans (60-70%) is less than the model would
predict. In all three species, the ultimate disposition of metabolism is uncertain.
In particular, there are uncertainties in attributing the "missing" metabolism to
• GSH pathway (e.g., urinary mercapturates may only capture a fraction of
the total flux; moreover, in Bernauer et al. (1996), excretion was still on-
going at end of collection period; model does not accurately depict time-
course of mercapturate excretion).
• Other hepatic oxidation (currently attributed to DCA).
• Extrahepatic systemic metabolism (e.g., kidney).
• Presystemic metabolism in the lung.
• Additional metabolism of TCOH or TCA (see below).
Implications for PBPK model parameters, structure, or data
Re-evaluation of all prior distributions
• Update priors for parameters with independent data (physiological
parameters, partition coefficients, in vitro metabolism), looking across
all available data sets.
• For priors without independent data (e.g., many metabolism
parameters), use less informative priors (e.g., log-uniform distributions
with wide bounds) so as prevent bias.
Evaluate modifications to the model structure, as discussed below.
• Additional dosing routes can be added easily.
• In mice, uptake from the stomach compartment (currently zero), but
previously included in Abbas and Fisher (1997), may improve the
model fit.
• In mice and humans, additional extrahepatic metabolism, either
presystemic (e.g., in the lung) or postsystemic (e.g., in the kidney)
and/or a wash-in/wash-out effect may improve the model fit.
• Calibration of GSH pathway may be improved by utilizing in vitro data
on liver and kidney GSH metabolism, adding a DCVG compartment to
improve the prediction of the time-course for mercapturate excretion,
and/or using the Lash et al. (1999b) blood DCVG in humans
(necessitating the addition of a DCVG compartment).
• Presystemic lung metabolism can only be evaluated if added to the
model (in vitro data exists to estimate the VMAX for such metabolism).
In addition, a wash-in/wash-out effect (e.g., suggested by Greenberg et
al., 1999) can be evaluated using a continuous breathing model that
separately tracks inhaled and exhaled air, with adsorption/desporption in
the respiratory tract.
• Additional elimination pathways for TCOH and TCA can be added for
evaluation.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-30. Conclusions from evaluation of Hack et al. (2006), and implications for PBPK model development
(continued)
Conclusion from evaluation of Hack et al. (2006) model
Implications for PBPK model parameters, structure, or data
ON
^§
TCA blood/plasma concentrations well predicted following TCE exposures in
all species. However, there may be inaccuracies in the total flux of TCA
production, as well as its disposition.
• In TCA dosing studies, the majority (>50%), but substantially <100%, was
recovered in urine, suggesting significant metabolism of TCA. Although
urinary TCA was well predicted in mice and humans (but not in rats), if
TCA metabolism is significant, then this means that the current model
underestimates the flux of TCE metabolism to TCA.
• An improved TCOH/TCOG model may also provide better estimates of
TCA kinetics (see below).
TCOH/TCOG concentrations and excretion were inconsistently predicted,
particularly after TCOH dosing.
• In mice and rats, first-order clearance for TCOH glucuronidation was
predicted to be greater than hepatic blood flow, which is consistent with a
first pass effect that is not currently accounted for.
• In humans, the estimated clearance rate for TCOH glucuronidation was
substantially smaller than hepatic blood flow. However, the presence of
substantial TCOG in blood (as opposed to free TCOH) in the Chiu et al.
(2007) data are consistent with greater glucuronidation than predicted by
the model.
• In TCOH dosing studies, substantially <100% was recovered in urine as
TCOG and TCA, suggesting another metabolism or elimination pathway.
• Additional elimination pathways for TCOH and TCA can be added for
evaluation.
• The addition of a liver compartment for TCOH and TCOG would
permit hepatic first-pass effects to be accounted for, as appears
necessary for mice and rats.
H I
O >
HH Oq
H TO
i.a. = intra-arterial, i.v. = intravenous.
O
H
W
-------
1 3.5.4. Physiologically Based Pharmacokinetic (PBPK) Model for Trichloroethylene (TCE)
2 and Metabolites Used for This Assessment
3 3.5.4.1. Introduction
4 Based on the recommendations of the NRC (2006) as well as additional analysis and
5 evaluation of the Hack et al. (2006) PBPK model, an updated PBPK model for TCE and
6 metabolites was developed for use in this risk assessment. The updated model is reported in
7 Evans et al. (2009) and Chiu et al. (2009), and the discussion below provides some details in
8 additional to the information in the published articles.
9 This updated model included modification of some of aspects of the Hack et al. (2006)
10 PBPK model structure, incorporation of additional in vitro and in vivo data for estimating model
11 parameters, and an updated hierarchical Bayesian population analysis of PBPK model
12 uncertainty and variability. In the subsections below, the updated PBPK model, and baseline
13 parameter values are described, and the approach and results of the analysis of PBPK model
14 uncertainty and variability. Appendix A provides more detailed descriptions of the model and
15 parameters, including background on hierarchical Bayesian analyses, model equations, statistical
16 distributions for parameter uncertainty and variability, data sources for these parameter values,
17 and the PBPK model code. Additional computer codes containing input files to the MCSim
18 program are available electronically.
19
20 3.5.4.2. Updated Physiologically Based Pharmacokinetic (PBPK) Model Structure
21 The updated TCE PBPK model is illustrated in Figure 3-7, with the major changes from
22 the Hack et al. (2006) model described here. The TCE submodel was augmented by the addition
23 of kidney and venous blood compartments, and an updated respiratory tract model that included
24 both metabolism and the possibility of local storage in the respiratory tissue. In particular, in the
25 updated lung, separate processes describing inhalation and exhalation allowed for adsorption and
26 desorption from tracheobronchial epithelium (wash-in/wash-out), with the possibility of local
27 metabolism as well. In addition, conjugative metabolism in the kidney was added, motivated by
28 the in vitro data on TCE conjugation described in Sections 3.3.3.2-3.3.3.3. With respect to
29 oxidation, a portion of the lung metabolism was assumed to produce systemically available
30 oxidative metabolites, including TCOH and TCA, with the remaining fraction assumed to be
31 locally cleared. This is clearly a lumping of a multistep process, but the lack of data precludes
32 the development of a more sequential model. TCE oxidation in the kidney was not included
33 because it was not likely to constitute a substantial flux of total TCE oxidation given the much
34 lower CYP activity in the kidney relative to the liver (Cummings et al., 1999, 2000) and the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-67 DRAFT—DO NOT CITE OR QUOTE
-------
1 greater tissue mass of the liver.l In addition, liver compartments were added to the TCOH and
2 TCOG submodels to account properly for first-pass hepatic metabolism, which is important for
3 consistency across routes of exposure. Furthermore, additional clearance pathways of TCOH
4 and TCA was added to their respective submodels. With respect to TCE conjugation, in humans,
5 an additional DCVG compartment was added between TCE conjugation and production of
6 DCVC. In addition, it should be noted that the urinary clearance of DCVC represents a lumping
7 of TV-acetylation of DCVC, deacetylation of NAcDCVC, and urinary excretion NAcDCVC, and
8 that the bioactivation of DCVC represents a lumping of thiol production from DCVC by beta-
9 lyase, sulfoxidation of DCVC by FMO3, and sulfoxidation of NAcDCVC by CYP3A. Such
10 lumping was used because these processes are not individually identifiable given the available
11 data.
12
13 3.5.4.3. Specification of Physiologically Based Pharmacokinetic (PBPK) Model Parameter
14 Prior Distributions
15 Point estimates for PBPK model parameters ("baseline values"), used as central estimates
16 in the prior distributions for population mean parameters in the hierarchical Bayesian statistical
17 model (see Appendix A), were developed using standard methodologies to ensure biological
18 plausibility, and were a refinement of those used in Hack et al. (2006). Because the Bayesian
19 parameter estimation methodology utilizes the majority of the useable in vivo data on TCE
20 pharmacokinetics, all baseline parameter estimates were based solely on measurements
21 independent of the in vivo data. This avoids using the same data in both the prior and the
22 likelihood. These parameters were, in turn, given truncated normal or lognormal distributions
23 for the uncertainty in the population mean. If no independent data were available, as is the case
24 for many "downstream" metabolism parameters, then no baseline value was specified, and a
25 noninformative prior was used. Section 3.5.5.4, below, discusses the updating of these
26 noninformative priors using interspecies scaling.
1 The extraction ratio for kidney oxidation is likely to be very low, as shown by the following calculation in rats and
humans. In rats, the in vitro kidney oxidative clearance (VMAX/KM) rate (Table 3-13, converting units) is
1.64 x 10~7 L/min/mg microsomal protein. Converting units using 16 mg microsomal protein to g tissue (Bong et
al., 1985) gives a clearance rate per unit tissue mass of 2.6 x 10~6 L/min/g kidney. This is more than a 1000-fold
smaller than the kidney specific blood flow rate of 6.3 * 10"3 L/min/g kidney (Brown et al., 1997). In humans, an in
vitro clearance rate of 6.5 * 10"8 L/min/mg microsomal protein is derived from the only detectable in vitro oxidation
rate from Cummings and Lash (2000) of 0.13 nmol/minute/mg protein at 2 mM. Using the same conversion from
microsomal protein to tissue mass gives a clearance rate of 1.0 x 10"6 L/min/g kidney, more than 1000-fold smaller
than the kidney specific blood flow of 3.25 x 10"3 L/min/g kidney (Brown et al., 1997). No data on kidney
metabolism are available in mice, but the results are likely to be similar. Therefore, even accounting for
uncertainties of up to an order of magnitude in the in vitro-to-in vivo conversion, kidney oxidation should contribute
negligibly to total metabolism of TCE.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-68 DRAFT—DO NOT CITE OR QUOTE
-------
_Conjugatjpn_
' Cqnjugatbn_
Oxidative Metabolism
Lung
Conjugative Metabolism
(rat and human only)
Legend
Input (exposure/dose)
J "Dynamic" Compartment (solved by DDEs)
[ "Static" Compartment (at local steady-state)
Transformation or Excretion
1 1
2 Figure 3-7. Overall structure of PBPK model for TCE and metabolites used
3 in this assessment. Boxes with underlined labels are additions or modifications
4 of the Hack et al. (2006) model, which are discussed in Table 3-31.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-69 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-31 Discussion of changes to the Hack et al. (2006) PBPK model
implemented for this assessment
Change to Hack et al.
(2006) PBPK model
Discussion
TCE respiratory tract
compartments and
metabolism
In vitro data indicate that the lung (at least in the mouse) has a significant
capacity for oxidizing TCE. However, in the Hack et al. (2006) model,
respiratory metabolism was blood flow-limited. The model structure used was
inconsistent with other PBPK models in which the same mechanism for
respiratory metabolism is assumed (e.g., styrene, Sarangapani et al. [2003]).
In these models, the main source of exposure in the respiratory tract tissue is
from the respiratory lumen—not from the tracheobronchial blood flow. In
addition, a wash-in/wash-out effect has also been postulated. The current
structure, which invokes a "continuous breathing" model with separate
"inhaled" and "exhaled" respiratory lumens, can accommodate both
respiratory metabolism due to exposure from the respiratory lumen as well as a
wash-in/wash-out effect in which there is temporary storage in the respiratory
tract tissue.
Moreover, preliminary analyses indicated that these changes to the model
structure allowed for a substantially better fit to mouse closed chamber data
under the requirement that all the dose levels are modeled using the same set
of parameters.
TCE kidney
compartment
In vitro data indicate that the kidney has a significant capacity for conjugating
TCE with GSH.
TCE venous blood
compartment
Many PBPK models have used a separate blood compartment. It was believed
to be potentially important and feasible to implement here because (1) TCE
blood concentrations were often not well predicted by the Hack et al. (2006)
model; (2) the TCA submodel has a plasma compartment, which is a fraction
of the blood volume based on the blood volume; (3) adequate independent
information on blood volume is available; and (4) the updated model was to
include the intravenous route of exposure.
TCOH and TCOG
liver compartments
In mice and rats, the Hack et al. (2006) model estimated a rate of TCOH
glucuronidation that exceeded hepatic blood flow (all glucuronidation is
assumed to occur in the liver), indicated a significant first-pass effect.
Therefore, a separate liver compartment is necessary to account properly for
hepatic first-pass.
TCOH and TCA
"other" elimination
pathways
Mass-balance studies with TCOH and TCA dosing indicated that, although the
majority of TCOH and TCA are excreted in urine, the amount is still
substantially less than 100%. Therefore, additional elimination of TCOH and
TCA must exist and should be accounted for.
DCVG compartment
(human model only)
Blood DCVG data in humans exist as part of the Fisher et al. (1998)
experiments, reported in Lash et al. (1999b), and a DCVG compartment is
necessary in order to utilize those data.
4
5
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-70 DRAFT—DO NOT CITE OR QUOTE
-------
1 In keeping with standard practice, many of the PBPK model parameters were "scaled" by
2 body or organ weights, cardiac output, or allometrically by an assumed (fixed) power of body
3 weight. Metabolic capacity and cardiac output were scaled by the % power of body weight and
4 rate coefficients were scaled by the—1A power of body weight, in keeping with general
5 expectations as to the relationship between metabolic rates and body size (U.S. EPA, 1992; West
6 et al., 2002) So as to ensure a consistent model structure across species as well as improve the
7 performance of the MCMC algorithm, parameters were further scaled to the baseline point-
8 estimates where available, as was done by Hack et al. (2006). For example, to obtain the actual
9 liver volume in liters, a point estimate is first obtained by multiplying the fixed, species-specific
10 baseline point estimate for the fractional liver volume by a fixed body weight (measured or
11 species-specific default) with density of 1 kg per liter assumed to convert from kg to liters.
12 Then, any deviation from this point estimate is represented by multiplying by a separate "scaled"
13 parameter VLivC that has a value of 1 if there is no deviation from the point estimate. These
14 "scaled" parameters are those estimated by the MCMC algorithm, and for which population
15 means and variances are estimated.
16 Baseline physiological parameters were re-estimated based on the updated tissue lumping
17 (e.g., separate blood and kidney compartments) using the standard references ICRP (2002) and
18 Brown et al. (1997). For a few of these parameters, such as hematocrit and respiratory tract
19 volumes in rodents, additional published sources were used as available, but no attempt was
20 made to compile a comprehensive review of available measurements. In addition, a few
21 parameters, such as the slowly perfused volume, were calculated rather than sampled in order to
22 preserve total mass or flow balances.
23 For chemical-specific distribution and metabolism parameters, in vitro data from various
24 sources were used. Where multiple measurements had been made, as was the case for many
25 partition coefficients, TCA plasma protein binding parameters, and TCE metabolism, different
26 results were pooled together, with their uncertainty reflected appropriately in the prior
27 distribution. Such in vitro measurements were available for most chemical partition coefficients,
28 except for those for TCOG (TCOH used as a proxy) and DCVG. There were also such data to
29 develop baseline values for the oxidative metabolism of TCE in the liver (VMAX and KM), the
30 relative split in TCE oxidation between formation of TCA and TCOH, and the VMAX for TCE
31 oxidation in the lung. All other metabolism parameters were not given baseline values and
32 needed to be estimated from the in vivo data.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-71 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.5.4.4. Dose Metric Predictions
2 The purpose of this PBPK model is to make predictions of internal dose in rodents used
3 in toxicity studies or in humans in the general population, and not in the groups or individuals for
4 which pharmacokinetic data exist. Therefore, to evaluate its predictive utility for risk
5 assessment, a number of dose metrics were selected for simulation in a "generic" mouse, rat, or
6 human, summarized in Table 3-32. The parent dose metric was AUC in blood. TCE metabolism
7 dose metrics (i.e., related to the amount metabolized) included both total metabolism,
8 metabolism splits between oxidation versus conjugation, oxidation in the liver versus the lung,
9 the amount of oxidation in the liver to products other than TCOH and TC A, and the amount of
10 TCA produced. These metabolism rate dose metrics are scaled by body weight in the case of
11 TCA produced, by the metabolizing tissue volume and by body weight to the % power in the
12 cases of the lung and "other" oxidation in the liver, and by body weight to the 3/4 power only in
13 other cases. With respect to the oxidative metabolites, liver concentrations of TCA and blood
14 concentrations of free TCOH were used. With respect to conjugative metabolites, the dose
15 metrics considered were total GSH metabolism scaled by body weight to the 3/4 power, and the
16 amount of DCVC bioactivated (rather than excreted in urine) per unit body weight to the %
17 power and per unit kidney mass.
18 All dose metrics are converted to daily or weekly averages based on simulations lasting
19 10 weeks for rats and mice and 100 weeks for humans. These simulation times were the shortest
20 for which additional simulation length did not add substantially to the average (i.e., less than a
21 few percent change with a doubling of simulation time).
22
23 3.5.5. Bayesian Estimation of Physiologically Based Pharmacokinetic (PBPK) Model
24 Parameters, and Their Uncertainty and Variability
25 3.5.5.1. Updated Pharmacokinetic Database
26 An extensive search was made for data not previously considered in the PBPK modeling
27 of TCE and metabolites, with a few studies identified or published subsequent to the review by
28 Chiu et al. (2006). The studies considered for analysis are listed in Tables 3-33-3-34, along with
29 an indication of whether and how they were used.2
2 Additional in vivo data on TCE or metabolites published after the PBPK modeling was completed (reported in
Sweeney et al., 2009; Liu et al., 2009; and Kim et al., 2009) was evaluated separately, and discussed in Appendix A.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-72 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-32. PBPK model-based dose metrics
Abbreviation
ABioactDCVCBW34
ABioactDCVCKid
AMetGSHBW34
AMetLivlBW34
AMetLivOtherBW34
AMetLivOtherLiv
AMetLngBW34
AMetLngResp
AUCCBld
AUCCTCOH
AUCLivTCA
TotMetabBW34
TotOxMetabBW34
TotTCAInBW
Description
Amount of DCVC bioactivated in the kidney (mg) per unit body weight'74 (kg'/4)
Amount of DCVC bioactivated in the kidney (mg) per unit kidney mass (kg)
Amount of TCE conjugated with GSH (mg) per unit body weight'74 (kg3/4)
Amount of TCE oxidized in the liver per unit body weight'74 (kg3/4)
Amount of TCE oxidized to metabolites other than TCA and TCOH in the liver
(mg) per unit body weight'74 (kg3/4)
Amount of TCE oxidized to metabolites other than TCA and TCOH in the liver
(mg) per unit liver mass (kg)
Amount of TCE oxidized in the respiratory tract (mg) per unit body weight'74 (kg'74)
Amount of TCE oxidized in the respiratory tract (mg) per unit respiratory tract
tissue mass (kg)
Area under the curve of the venous blood concentration of TCE (mg-h/L)
Area under the curve of the blood concentration of TCOH (mg-h/L)
Area under the curve of the liver concentration of TCA (mg-h/L)
Total amount of TCE metabolized (mg) per unit body weight'74 (kg'74)
Total amount of TCE oxidized (mg) per unit body weight'74 (kg'/4)
Total amount of TCA produced (mg) per unit body weight (kg)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-73 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-33 Rodent studies with pharmacokinetic data considered for analysis
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
s
Reference
Species
(strain)
Sex
TCE exposures
Other
exposures
Calibration
Validation
Not
used
Comments
Mouse studies
Abbas et a\.,
1996
Abbas and
Fisher, 1997
Abbas et a\.,
1997
Barton et a\.,
1999
Birneretal., 1993
Fisher and Allen,
1993
Fisher et al.,
1991
Green and Prout,
1985
Greenberg et al.,
1999
Larson and Bull,
1992a
Larson and Bull,
1992b
Merdink et al.,
1998
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(NMRI)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
Mouse
(B6C3F1)
M
M
M
M
M+F
M+F
M+F
M
M
M
M
M
~
Oral (corn oil)
—
—
Gavage
Gavage (corn
oil)
Inhalation
Gavage (corn
oil)
Inhalation
"
Oral (aqueous)
i.v.
CH i.v.
~
TCOH, TCA
i.v.
DCA i.v. and
oral (aqueous)
~
~
~
TCA i.v.
~
DCA, TCA
oral (aqueous)
~
CH i.v.
V*
V
V
V*
V
V*
V
V
V
V
V
V
CH not in model.
DCA not in model.
Only urine concentrations
available, not amount.
Only data on TCA dosing
was used, since DCA is
not in the model.
Only data on TCE dosing
was used, since CH is not
in the model.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
s
i
Table 3-33. Rodent studies with pharmacokinetic data considered for analysis (continued)
Reference
Proutetal., 1985
Templin et al.,
1993
Species
(strain)
Mouse
(B6C3F1,
Swiss)
Mouse
(B6C3F1)
Sex
M
M
TCE exposures
Gavage (corn
oil)
Oral (aqueous)
Other
exposures
TCA oral
Calibration
V*
V*
Validation
Not
used
Comments
Rat studies
Andersen et al.,
1987
Barton et al.,
1995
Bernauer et al.,
1996
Birneretal., 1993
Birneretal., 1997
Bruckner et al.,
unpublished
Dallas et al.,
1991
Rat (F344)
Rat (S-D)
Rat (Wistar)
Rat (Wistar,
F344)
Rat (Wistar)
Rat (S-D)
Rat (S-D)
M
M
M
M+F
M+F
M
M
Inhalation
Inhalation
Inhalation
Gavage (ns)
Inhalation
Inhalation
~
"
~
~
DCVC i.v.
~
V*
V
V*
V
V
V
V
Initial chamber
concentrations
unavailable, so not used.
Only urine concentrations
available, not amount.
Single dose, route does
not recapitulate how
DCVC is formed from
TCE, excreted NAcDCVC
~1 00-fold greater than that
from relevant TCE
exposures (Bernauer et
al., 1996).
Not published, so not used
for calibration. Similar to
Keys et al. (2003) data.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
s
Table 3-33. Rodent studies with pharmacokinetic data considered for analysis (continued)
Reference
D'Souza et a\.,
1985
Fisher et a\.,
1989
Fisher et a\.,
1991
Green and Prout,
1985
Hissink et a\.,
2002
Jakobson et a\.,
1986
Kaneko et a\.,
1994
Keyset al., 2003
Kimmerle and
Eben, 1973a
Larson and Bull,
1992a
Larson and Bull,
1992b
Lash et al., 2006
Species
(strain)
Rat (S-D)
Rat (F344)
Rat (F344)
Rat
(Osborne-
Mendel)
Rat (Wistar)
Rat (S-D)
Rat (Wistar)
Rat (S-D)
Rat (Wistar)
Rat (F344)
Rat (S-D)
Rat (F344)
Sex
M
F
M+F
M
M
F
M
M
M
M
M
M+F
TCE exposures
\.v., oral
(aqueous)
Inhalation
Inhalation
Gavage (corn
oil)
Gavage (corn
oil), i.v.
Inhalation
Inhalation
Inhalation,
oral (aqueous),
i.a.
Inhalation
"
Oral (aqueous)
Gavage (corn
oil)
Other
exposures
~
"
TCA gavage
(aqueous)
~
Various
pretreatments
(oral)
Ethanol
pretreatment
(oral)
"
~
DCA, TCA
oral (aqueous)
~
~
Calibration
V
V*
V
V
V
V
V
V
V*
Validation
V
V
Not
used
V
V
Comments
Only TCE blood
measurements, and >10-
fold greater than other
similar studies.
Experiment with blood
only data not used for
calibration.
Pretreatments not
included. Only blood TCE
data available.
Pretreatments not
included.
Only TCA dosing data
used, since DCA is not in
the model.
Highly inconsistent with
other studies.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
- a
o g"
vo 5'
I
Co
1
TO'
s
Table 3-33. Rodent studies with pharmacokinetic data considered for analysis (continued)
Reference
Lee etal., 1996
Lee et al.,
2000a, b
Merdink et al.,
1999
Poet etal., 2000
Proutetal., 1985
Saghiret al.,
2002
Simmons et al.,
2002
Stenner et al.,
1997
Templin et al.,
1995
Thrall et al., 2000
Yu etal. ,2000
Species
(strain)
Rat (S-D)
Rat (S-D)
Rat (F344)
Rat (F344)
Rat
(Osborne-
Mendel,
Wistar)
Rat (F344)
Rat (Long-
Evans)
Rat (F344)
Rat (F344)
Rat (F344)
Rat (F344)
Sex
M
M
M
M
M
M
M
M
M
M
M
TCE exposures
Arterial, venous,
portal, stomach
injections
Stomach
injection, i.v.,
p.v.
~
Dermal
Gavage (corn
oil)
~
Inhalation
intraduodenal
Oral (aqueous)
i.v., i.p.
~
Other
exposures
p-nitrophenol
pretreatment
(i.a.)
CH, TCOH i.v.
—
DCA i.v., oral
(aqueous)
~
TCOH, TCA
i.v.
~
with toluene
TCA i.v.
Calibration
V
V
V*
V
V
V*
V
Validation
V
V
Not
used
V
V
V
Comments
Only blood TCE data
available.
Pretreatments not
included. Only
experiments with blood
and liver data used for
calibration.
TCOH dosing used; CH
not in model.
Dermal exposure not in
model.
DCA not in model
Only exhaled breath data
available from i.v. study.
i.p. dosing not in model.
o §
H I
O >
HH Oq
H ^
O
H
W
*Part or all of the data in the study was used for calibration in Hack et al. (2006).
i.a. = intra-arterial, i.p. = intraperitoneal, i.v. = intravenous, p.v. = intraperivenous.
-------
Table 3-34 Human studies with pharmacokinetic data considered for analysis
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
s
00 o
Reference
Bartonicek, 1962
Bernaueretal., 1996
Bloemen et al., 2001
Chiuetal.,2007
Ertleetal., 1972
Fernandez et al.,
1977
Fisher etal., 1998
Kimmerle and Eben,
1973b
Lapareet al., 1995
Lash etal., 1999b
Monster etal., 1976
Monster etal., 1979
Mulleretal., 1972
Species
(number of
individuals)
Human
(n = 8)
Human
Human
(n = 4)
Human
(n = 6)
Human
Human
Human
(n = 17)
Human
(n = 12)
Human
(n = 4)
Human
Human
(n = 4)
Human
Human
Sex
M+F
M
M
M
M
M
M+F
M+F
M+F
M+F
M
M
ns
TCE
exposures
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Other
exposures
"
~
"
~
CH oral
~
~
~
"
~
-
~
Calibration
Va
V
va
V
V
vb
Validation
V
V
V
vb
va
Not
used
V
V
Comments
Sparse data, so not included for
calibration to conserve
computational resources.
Grouped data, but unique in that
includes NAcDCVC urine data.
Sparse data, so not included for
calibration to conserve
computational resources.
Very similar to Muller data.
Complex exposure patterns, and
only grouped data available for
urine, so used for validation.
Grouped only, but unique in that
DCVG blood data available (same
individuals as Fisher et al. [1 998]),
Experiments with exercise not
included.
Grouped data only.
Same data also included in Muller
etal. (1975).
^§
H I
O >
HH Oq
H TO
o
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
s
Table 3-34. Human studies with pharmacokinetic data considered for analysis (continued)
Reference
Mulleretal., 1974
Mulleretal., 1975
Paycoketal., 1945
Poet etal., 2000
Satoetal., 1977
Stewart et a I., 1970
Treibig etal., 1976
Vesterberg and
Astrand, 1976
Species
(number of
individuals)
Human
Human
Human
(n = 3)
Human
Human
Human
Human
Human
Sex
M
M
ns
M+F
M
ns
ns
M
TCE
exposures
Inhalation
Inhalation
~
Dermal
Inhalation
Inhalation
Inhalation
Inhalation
Other
exposures
CH, TCA,
TCOH oral
Ethanol oral
TCA i.v.
~
-
-
-
~
Calibration
V
V
Validation
Va
Va
V
va
va
Not
used
V
Comments
TCA and TCOH dosing data used
for calibration, since it is rare to
have metabolite dosing data.
TCE dosing data used for
validation, since only grouped
data available. CH not in model.
Grouped data only.
Dermal exposure not in model.
All experiments included exercise,
so were not included.
i
^§
aPart or all of the data in the study was used for calibration in Hack et al. (2006).
bGrouped data from this study was used for calibration in Hack et al. (2006), but individual data was used here.
H I
O >
HH Oq
H TO
O
H
W
-------
1 The least amount of data was available for mice, so an effort was made to include as
2 many studies as feasible for use in calibrating the PBPK model parameters. Exceptions include
3 mouse studies with CH or DCA dosing, since those metabolites are not included in the PBPK
4 model. In addition, the Birner et al. (1993) data only reported urine concentrations, not the
5 amount excreted in urine. Because there is uncertainty as to total volume of urine excreted, and
6 over what time period, these data were not used. Moreover, many other studies had urinary
7 excretion data, so this exclusion should have minimal impact. Several data sets not included by
8 Hack et al. (2006) were used here. Of particular importance was the inclusion of TCA and
9 TCOH dosing data from Abbas et al. (1997), Green and Prout (1985), Larson and Bull (1992a),
10 and Templin et al. (1993).
11 A substantial amount of data are available in rats, so some data that appeared to be
12 redundant was excluded from the calibration set and saved for comparison with posterior
13 predictions (a "validation" set). In particular, those used for "validation" are one closed-chamber
14 experiment (Andersen et al., 1987), several data sets with only TCE blood data (D'Souza et al.,
15 1985; Jakobson et al., 1986; Lee et al., 1996, and selected time courses from Fisher et al. [1991]
16 and Lee et al. [2000a, b]), and one unpublished data set (Bruckner et al., unpublished). The
17 Andersen et al. (1987) data was selected randomly from the available closed chamber data, while
18 the other data sets were selected because they unpublished or because they more limited in scope
19 (e.g., TCE blood only) and so were not as efficient for use in the computationally-intensive
20 calibration stage. As with the mouse analyses, TCA and TCOH dosing data were incorporated to
21 better calibrate those pathways.
22 The human pharmacokinetic database of controlled exposure studies is extensive but also
23 more complicated. For the majority of the studies, only grouped or aggregated data were
24 available, and most of those data were saved for "validation" since there remained a large
25 number of studies for which individual data were available. However, some data that may be
26 uniquely informative are only available in grouped form, in particular DCVG blood
27 concentrations, NAcDCVC urinary excretion, and data from TCA and TCOH dosing. In
28 addition, several human data sets, while having individual data, involved sparse collection at
29 only one or a few time points per exposure (Bartonicek, 1962; Bloemen et al., 2001) and were
30 subsequently excluded to conserve computational resources. Lapare et al. (1995), which
31 involved multiple, complex exposure patterns over the course of a month and was missing the
32 individual urine data, was also excluded due to the relatively low amount of data given the large
33 computational effort required to simulate it. Several studies also investigated the effects of
34 exercise during exposure on human TCE toxicokinetics. The additional parameters in a model
35 including exercise would need to characterize the changes in cardiac output, alveolar ventilation,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-80 DRAFT—DO NOT CITE OR QUOTE
-------
1 and regional blood flow as well as their inter-individual variability, and would have further
2 increased the computational burden. Therefore, it was decided that such data would be excluded
3 from this analysis. Even with these exclusions, data on a total of 42 individuals, some involving
4 multiple exposures, were included in the calibration.
5
6 3.5.5.2. Updated Hierarchical Population Statistical Model
1 Generally, only aggregated pharmacokinetic data (arithmetic mean and standard
8 deviation or standard error) are available from rodent studies. In the Hack et al. (2006) model,
9 each simulation was treated as a separate observational unit, so different dosing levels within the
10 same study were treated separately and assigned different PBPK model parameters. However,
11 the dose-response data are generally also only separated by sex and strain, and otherwise
12 aggregated, so the variability that is of interest is interstudy (e.g., lot-to-lot), interstrain, and
13 intersex variability, rather than interindividual variability. In addition, any particular lot of
14 animals within a study, which are generally inbred and kept under similarly controlled
15 conditions, are likely to be relatively homogeneous. Therefore, in the revised model, for rodents,
16 different animals of the same sex and strain in the same study (or series of studies conducted
17 simultaneously) were treated as identical, and grouped together. Thus, the predictions from the
18 population model in rodents simulate "average" pharmacokinetics for a particular "lot" of
19 rodents of a particular species, strain, and sex.
20 In humans, however, interindividual variability is of interest, and, furthermore,
21 substantial individual data are available in humans. However, in some studies, the same
22 individual was exposed more than once, and, so, those data should be grouped together (in the
23 Hack et al. [2006] model, they were treated as different "individuals"). Because the primary
24 interest here is chronic exposure, and because it would add substantially to the computational
25 burden, interoccasion variability—changes in pharmacokinetic parameters in a single individual
26 over time—is not addressed. Thus, the predictions from the population model in humans are the
27 "average" across different occasions for a particular individual (adult).
28 As discussed in Section 3.3.3.1, sex and (in rodents) strain differences in oxidative
29 metabolism were modest or minimal. While some sex-differences have been noted in GSH
30 metabolism (see Sections 3.3.3.2.7-3.3.3.2.8), almost all of the available in vivo data is in males,
31 making it more difficult to statistically characterize that difference with PBPK modeling.
32 Therefore, within a species, different sexes and (in rodents) strains were considered to be drawn
33 from a single, species-level population.
34 Figure A-l in Appendix A illustrates the hierarchical structure. Informative prior
35 distributions reflecting the uncertainty in the population mean and variance, detailed in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-81 DRAFT—DO NOT CITE OR QUOTE
-------
1 Appendix A, were updated from those used in Hack et al. (2006) based on an extensive analysis
2 of the available literature. Section 3.5.5.3, next, discusses specification of prior distributions in
3 the case where no data independent of the calibration data exist.
4
5 3.5.5.3. Use of Interspecies Scaling to Update Prior Distributions in the Absence of Other
6 Data
7 For many metabolic parameters, little or no in vitro or other prior information is available
8 to develop prior distributions. Initially, for such parameters, noninformative priors in the form of
9 log-uniform distributions with a range spanning at least 104 were specified. However, in the
10 time available for analysis (up to about 100,000 iterations), only for the mouse did all these
11 parameters achieve adequate convergence. This suggests that some of these parameters are
12 poorly identified for the rat and human. Additional preliminary runs indicated replacing the log-
13 uniform priors with lognormal priors and/or requiring more consistency between species could
14 improve identifiability sufficiently for adequate convergence. However, an objective method of
15 "centering" the lognormal distributions that did not rely on the in vivo data (e.g., via visual fitting
16 or limited optimization) being calibrated against was necessary in order to minimize potential
17 bias.
18 Therefore, the approach taken was to consider three species sequentially, from mouse to
19 rat to human, and to use interspecies scaling to update the prior distributions across species. This
20 sequence was chosen because the models are essentially "nested" in this order, the rat model
21 adds to the mouse model the "downstream" GSH conjugation pathways, and the human model
22 adds to the rat model the intermediary DCVG compartment. Therefore, for those parameters
23 with little or no independent data only, the mouse posteriors were used to update the rat priors,
24 and both the mouse and rat posteriors were used to update the human priors. Table 3-35 contains
25 a list of the parameters for which this scaling was used to update prior distributions. The scaling
26 relationship is defined by the "scaled parameters" listed in Appendix A (see Section A.4.1,
27 Table A-4), and generally follows standard practice. For instance, VMAX and clearance rates
28 scale by body weight to the 3A power, whereas KM values are assumed to not scale, and rate
29 constants (inverse time units) scale by body weight to the -Vi power.
30
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-82 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-35 Parameters for which scaling from mouse to rat, or from mouse and rat to human, was used to
update the prior distributions
Parameter with no or highly uncertain a priori data
Respiratory lumen-Missue diffusion flow rate
TCOG body/blood partition coefficient
TCOG liver/body partition coefficient
Fraction of hepatic TCE oxidation not to TCA+TCOH
VMAX for hepatic TCE GSH conjugation
KM for hepatic TCE GSH conjugation
VMAX for renal TCE GSH conjugation
KM for renal TCE GSH conjugation
VMAX for Tracheo-bronchial TCE oxidation
KM for Tracheo-bronchial TCE oxidation
Fraction of respiratory oxidation entering systemic circulation
VMAX for hepatic TCOH-^TCA
KM for hepatic TCOH^TCA
VMAX for hepatic TCOH-^TCOG
KM for hepatic TCOH^TCOG
Rate constant for hepatic TCOH->other
Rate constant for TCA plasma->urine
Rate constant for hepatic TCA->other
Rate constant for TCOG liver->bile
Lumped rate constant for TCOG bile-»TCOH liver
Rate constant for TCOG->urine
Lumped rate constant for DCVC-»Urinary NAcDCVC
Rate constant for DCVC bioactivation
Mouse
^Rat
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
Rat-»
Human
V
V
Mouse+
Rat-»
Human
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
Comments
No a priori information
Prior centered on TCOH data, but highly uncertain
Prior centered on TCOH data, but highly uncertain
No a priori information
Rat data on at 1 and 2 mM. Human data at more
concentrations, so VMAX and KM can be estimated
Rat data on at 1 and 2 mM. Human data at more
concentrations, so VMAX and KM can be estimated
Prior based on activity at a single concentration
No a priori information
No a priori information
No a priori information
No a priori information
No a priori information
No a priori information
No a priori information
Prior centered at GFR, but highly uncertain
No a priori information
No a priori information
No a priori information
Prior centered at GFR, but highly uncertain
Not included in mouse model
Not included in mouse model
OO
^§
H I
O >
HH Oq
H TO
O
H
W
See Appendix A, Table A-4 for scaling relationships.
-------
1 The scaling model is given explicitly as follows. If 0, are the "scaled" parameters
2 (usually also natural-log-transformed) that are actually estimated, and A is the "universal"
3 (species-independent) parameter, then 9, = A + eh where et is the species-specific "departure"
4 from the scaling relationship, assumed to be normally distributed with variance oe2. Therefore,
5 the mouse model gives an initial estimate of "A," which is used to update the prior distribution
6 for 0r = A + er in the rat. The rat and mouse together then give a "better" estimate of A, which is
7 used to update the prior distribution for Qh = A + eh in the human, with the assumed distribution
8 for &h- The mathematical details are given in Appendix A, but two key points in this model are
9 worth noting here:
10
11 • It is known that interspecies scaling is not an exact relationship, and that, therefore, in
12 any particular case it may either over- or underestimate. Therefore, the variance in the
13 new priors reflect a combination of (1) the uncertainty in the "previous" species'
14 posteriors as well as (2) a "prediction error" that is distributed lognormally with
15 geometric standard deviation (GSD) of 3.16-fold, so that the 95% confidence range about
16 the central estimate spans 100-fold. This choice was dictated partially by practicality, as
17 larger values of the GSD used in preliminary runs did not lead to adequate convergence
18 within the time available for analysis.
19 • The rat posterior is a product of its prior (which is based on the mouse posterior) and its
20 likelihood. Therefore, using the rat and mouse posteriors together to update the human
21 priors would use the mouse posterior "twice." Therefore, the rat posterior is
22 disaggregated into its prior and its likelihood using a lognormal approximation (since the
23 prior is lognormal), and only the (approximate) likelihood is used along with the mouse
24 posterior to develop the human prior.
25
26 With this methodology for updating the prior distributions, adequate convergence was
27 achieved for the rat and human after 110,000-140,000 iterations (discussed further below).
28
29 3.5.5.4. Implementation
30 The PBPK model was coded in for use in the MCSim software (version 5.0.0), which was
31 developed particularly for implementing MCMC simulations. As a quality control (QC) check,
32 results were checked against the original Hack et al. (2006) model, with the original structures
33 restored and parameter values made equivalent, and the results were within the error tolerances
34 of the ordinary differential equation (ODE) solver after correcting an error in the Hack et al.
35 (2006) model for calculating the TCA liver plasma flow. In addition, the model was translated to
36 MatLab (version 7.2.0.232) with simulation results checked and found to be within the error
37 tolerances of the ODE solver (odelSs). Mass balances were also checked using the baseline
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-84 DRAFT—DO NOT CITE OR QUOTE
-------
1 parameters, as well as parameters from preliminary MCMC simulations, and found to be within
2 the error tolerances of the ODE solver. Appendix A contains the MCSim model code.
O
4 3.5.6. Evaluation of Updated Physiologically Based Pharmacokinetic (PBPK) Model
5 3.5.6.1. Convergence
6 As in previous similar analyses (Gelman et al., 1996; Bois 2000a, b; Hack et al., 2006;
7 David et al., 2006), the potential scale reduction factor ",R" is used to determine whether different
8 independent MCMC chains have converged to a common distribution. The R diagnostic is
9 calculated for each parameter in the model, and represents the factor by which the standard
10 deviation or other measure of scale of the posterior distribution (such as a confidence interval
11 [CI]) may potentially be reduced with additional samples (Gelman et al., 2004). This
12 convergence diagnostic declines to 1 as the number of simulation iterations approaches infinity,
13 so values close to 1 indicate approximate convergence, with values of 1.1 and below commonly
14 considered adequate (Gelman et al., 2004). However, as an additional diagnostic, the
15 convergence of model dose metric predictions was also assessed. Specifically, dose metrics for a
16 number of generic exposure scenarios similar to those used in long-term bioassays were
17 generated, and their natural log (due to their approximate lognormal posterior distributions) was
18 assessed for convergence using the potential scale reduction factor "R" This is akin to the idea
19 of utilizing sensitivity analysis so that effort is concentrated on calibrating the most sensitive
20 parameters for the purpose of interest. In addition, predictions of interest which do not
21 adequately converge can be flagged as such, so that the statistical uncertainty associated with the
22 limited sample size can be considered.
23 The mouse model had the most rapid reduction in potential scale reduction factors.
24 Initially, four chains of 42,500 iterations each were run, with the first 12,500 discarded as
25 "burn-in" iterations. The initial decision for determining "burn-in" was determined by visual
26 inspection. At this point, evaluating the 30,000 remaining iterations, all the population
27 parameters except for the VMAX for DCVG formation had R < 1.2, with only the first-order
28 clearance rate for DCVG formation and the VMAX and KM for TCOH glucuronidation having
29 R > 1.1. For the samples used for inference, all of these initial iterations were treated as "burn-
30 in" iterations, and each chain was then restarted and run for an additional
31 68,700-71,400 iterations (chains were terminated at the same time, so the number of iterations
32 per chains was slightly different). For these iterations, all values of R were <1.03. Dose metric
33 predictions calculated for exposure scenarios 10-600 ppm either continuously or 7 hour/day,
34 5 day/week and 10-3,000 mg/kg/d either continuously or by gavage 5 day/week. These
35 predictions were all adequately converged, with all values of R < 1.03.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-85 DRAFT—DO NOT CITE OR QUOTE
-------
1 As discussed above, for parameters with little or no a priori information, the posterior
2 distributions from the mouse model were used to update prior distributions for the rat model,
3 accounting for both the uncertainty reflected in the mouse posteriors as well as the uncertainty in
4 interspecies extrapolation. Four chains were run to 111,960-128,000 iterations each (chains
5 were terminated at the same time and run on computers with slightly different processing speeds,
6 so the number of iterations per chains was slightly different). As is standard, about the first
7 "half of the chains—i.e., the first 64,000 iterations—were discarded as "burn-in" iterations, and
8 the remaining iterations were used for inferences. For these remaining iterations, the diagnostic
9 R was < 1.1 for all population parameters except the fraction of oxidation not producing TC A or
10 TCOH (R = 1.44 for population mean, R = 1.35 for population variance), the KM for TCOH ->
11 TCA (R= 1.19 for population mean), the VMAX and Km for TCOH glucuronidation (R = 1.23 and
12 1.12, respectively for population mean, and R = 1.13 for both population variances), and the rate
13 of "other" metabolism of TCOH (R = 1.29 for population mean and R = 1.18 for population
14 variance). Due to resource constraints, chains needed to be stopped at this point. However,
15 these are similar to the degree of convergence reported in Hack et al. (2006). Dose metric
16 predictions calculated for two inhalation exposure scenarios (10-600 ppm continuously or
17 7 hours/day, 5 day/week) and two oral exposure scenarios (10-3,000 mg/kg/d continuously or by
18 gavage 5 day/week).
19 All dose metric predictions hadR < 1.04, except for the amount of "other" oxidative
20 metabolism (i.e., not producing TCA or TCOH), which had R = 1.12-1.16, depending on the
21 exposure scenario. The poorer convergence of this dose metric is expected given that a key
22 determining parameter, the fraction of oxidation not producing TCA or TCOH, had the poorest
23 convergence among the population parameters.
24 For the human model, a set of four chains was run for 74,160-84,690 iterations using
25 "preliminary" updated prior distributions based on the mouse posteriors and preliminary runs of
26 the rat model. Once the rat chains were completed, final updated prior distributions were
27 calculated and the last iteration of the preliminary runs were used as starting points for the final
28 runs. The center of the final updated priors shifted by less than 25% of the standard deviation of
29 either the preliminary or revised priors, so that the revised median was between the 40th
30 percentile and 60th percentile of the preliminary median, and vice versa. The standard deviations
31 changed by less than 5%. Therefore, the use of the preliminary chains as a starting point should
32 introduce no bias, as long as an appropriate burn-in period is used for the final runs.
33 The final chains were run for an additional 59,140-61,780 iterations, at which point, due
34 to resource constraints, chains needed to be stopped. After the first 20,000 iterations, visual
35 inspection revealed the chains were no longer dependent on the starting point. These iterations
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-86 DRAFT—DO NOT CITE OR QUOTE
-------
1 were therefore discarded as "burn-in" iterations, and for the remaining -40,000 iterations used
2 for inferences. All population mean parameters had R< l.l except for the respiratory tract
3 diffusion constant (R = 1.20), the liverblood partition coefficient for TCOG (R = 1.23), the rate
4 of TCE clearance in the kidney producing DC VG (R = 1.20), and the rate of elimination of
5 TCOG in bile (R = 1.46). All population variances also had R < 1.1 except for the variance for
6 the fraction of oxidation not producing TCOH or TCA (R = 1.10). Dose metric predictions were
7 assessed for continuous exposure scenarios at 1-60 ppm in air or 1-300 mg/kg/d orally. These
8 predictions were all adequately converged with all values of R < 1.02.
9
10 3.5.6.2. Evaluation of Posterior Parameter Distributions
11 Posterior distributions of the population parameters need to be checked as to whether
12 they appear reasonable given the prior distributions. Inconsistency between the prior and
13 posterior distributions may indicate insufficiently broad (i.e., due to overconfidence) or
14 otherwise incorrectly specified priors, a misspecification of the model structure (e.g., leading to
15 pathological parameter estimates), or an error in the data. As was done with the evaluation of
16 Hack et al. (2006) in Appendix A, parameters were flagged if the interquartile regions of their
17 prior and posterior distributions did not overlap.
18 Appendix A contains detailed tables of the "sampled" parameters, and their prior and
19 posterior distributions. Because these parameters are generally scaled one or more times to
20 obtain a physically meaningful parameter, they are difficult to interpret. Therefore, in
21 Tables 3-36-3-40, the prior and posterior distributions for the PBPK model parameters obtained
22 after scaling are summarized. Note that because these model parameters are at the individual
23 (for humans) or sex/species/study unit (for rodents) level, they were generated using the
24 uncertainty distributions for the population mean and variance, and hence the distributions reflect
25 both uncertainty in the population characteristics as well as variability in the population.
26 Furthermore, they account for correlations among the population-level parameters.
27 The prior and posterior distributions for most physiological parameters were similar (see
28 Table 3-36). The posterior distribution was substantially narrower (i.e., less uncertainty) than the
29 prior distribution only in the case of the diffusion rate from the respiratory lumen to the
30 respiratory tissue, which also was to be expected given the very wide, noninformative prior for
31 that parameter.
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-87 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-36. Physiological parameters: prior and posterior combined uncertainty and variability
to
VO C?'
I
I
§
***.
£3'
1
TO'
Parameter description
Cardiac output (L/h)
Alveolar ventilation (L/h)
Scaled fat blood flow
Scaled gut blood flow
Scaled liver blood flow
Scaled slowly perfused
blood flow
Scaled rapidly perfused
blood flow
Scaled kidney blood flow
Respiratory lumen:tissue
diffusive clearance rate
(L/h)
Fat fractional
compartment volume
Gut fractional
compartment volume
Liver fractional
compartment volume
Rapidly perfused
fractional compartment
volume
PBPK
parameter
QC
QP
QFatC
QGutC
QLivC
QSIwC
QRapC
QKidC
DResp
VFatC
VGutC
VLivC
VRapC
Mouse
Prior
median
(2.5%, 97.5%)
0.84
(0.49, 1 .4)
2.1
(0.99, 4.4)
0.07
(0.012, 0.13)
0.14
(0.098, 0.18)
0.02
(0.014,0.026)
0.22
(0.1,0.33)
0.46
(0.31,0.61)
0.091
(0.038,0.14)
0.02
(0.000027, 16)
0.07
(0.014, 0.13)
0.049
(0.037, 0.06)
0.055
(0.031,0.079)
0.1
(0.082,0.12)
Posterior
median
(2.5%, 97.5%)
1
(0.46, 1.7)
2.1
(0.84, 4.5)
0.073
(0.015,0.13)
0.16
(0.11, 0.19)
0.021
(0.014,0.026)
0.21
(0.1,0.33)
0.44
(0.3, 0.59)
0.09
(0.038,0.14)
2.5
(0.8, 7.2)
0.089
(0.029,0.13)
0.048
(0.037, 0.06)
0.046
(0.03, 0.073)
0.1
(0.082,0.12)
Rat
Prior
median
(2.5%, 97.5%)
5.4
(3.7, 7.9)
10
(4.3, 25)
0.07
(0.012,0.13)
0.15
(0.11,0.2)
0.021
(0.015,0.027)
0.34
(0.15,0.52)
0.28
(0.073, 0.49)
0.14
(0.11, 0.17)
10
(0.4, 100)
0.07
(0.013,0.13)
0.032
(0.024, 0.04)
0.034
(0.023, 0.045)
0.088
(0.069,0.11)
Posterior
median
(2.5%, 97.5%)
6.4
(3.5,9.1)
7.6
(3.4, 19)
0.081
(0.023, 0.13)
0.17
(0.12, 0.2)
0.022
(0.015,0.027)
0.31
(0.15, 0.5)
0.28
(0.074, 0.45)
0.14
(0.11,0.17)
21
(6.6, 74)
0.068
(0.016, 0.12)
0.031
(0.025, 0.039)
0.033
(0.023, 0.044)
0.088
(0.07,0.11)
Human
Prior
median
(2.5%, 97.5%)
390
(230, 670)
370
(170,780)
0.05
(0.0082, 0.092)
0.19
(0.13, 0.25)
0.064
(0.012,0.12)
0.22
(0.094, 0.35)
0.28
(0.11, 0.46)
0.19
(0.15, 0.23)
570
(35, 3,900)
0.2
(0.038, 0.36)
0.02
(0.017,0.023)
0.025
(0.015,0.035)
0.088
(0.075, 0.1)
Posterior
median
(2.5%, 97.5%)
340
(190,720)
440
(170, 1,100)
0.044
(0.0076, 0.09)
0.16
(0.12,0.22)
0.039
(0.0087, 0.091)
0.17
(0.085, 0.3)
0.39
(0.23, 0.51)
0.19
(0.15,0.23)
270
(63, 930)
0.16
(0.036, 0.31)
0.02
(0.017,0.023)
0.026
(0.016,0.035)
0.088
(0.076, 0.099)
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-36. Physiological parameters: prior and posterior combined uncertainty and variability (continued)
Parameter description
Fractional volume of
respiratory lumen
Fractional volume of
respiratory tissue
Kidney fractional
compartment volume
Blood fractional
compartment volume
Slowly perfused fractional
compartment volume
Plasma fractional
compartment volume
TCA body fractional
compartment volume [not
incl. blood+liver]
TCOH/G body fractional
compartment volume [not
incl. liver]
PBPK
parameter
VRespLumC
VRespEffC
VKidC
VBIdC
VSIwC
VPIasC
VBodC
VBodTCOHC
Mouse
Prior
median
(2.5%, 97.5%)
0.0047
(0.0037,
0.0056)
0.0007
(0.00056,
0.00084)
0.017
(0.014,0.02)
0.049
(0.038, 0.06)
0.55
(0.48, 0.62)
0.025
(0.012,0.041)
0.79
(0.76, 0.81)
0.83
(0.81,0.86)
Posterior
median
(2.5%, 97.5%)
0.0047
(0.0038,
0.0056)
0.0007
(0.00056,
0.00084)
0.017
(0.014,0.02)
0.049
(0.039, 0.059)
0.54
(0.48, 0.61)
0.022
(0.012,0.036)
0.79
(0.77, 0.81)
0.84
(0.82, 0.86)
Rat
Prior
median
(2.5%, 97.5%)
0.0047
(0.0031,
0.0062)
0.0005
(0.00034,
0.00066)
0.007
(0.0051,
0.0089)
0.074
(0.058, 0.09)
0.59
(0.53, 0.66)
0.039
(0.019,0.062)
0.79
(0.77, 0.81)
0.87
(0.85, 0.88)
Posterior
median
(2.5%, 97.5%)
0.0047
(0.0033,
0.0061)
0.0005
(0.00035,
0.00066)
0.007
(0.0052,
0.0088)
0.074
(0.059, 0.09)
0.6
(0.54, 0.66)
0.04
(0.023, 0.059)
0.79
(0.77, 0.81)
0.87
(0.86, 0.88)
Human
Prior
median
(2.5%, 97.5%)
0.0024
(0.0015,
0.0033)
0.00018
(0.00011,
0.00025)
0.0043
(0.003, 0.0056)
0.077
(0.06, 0.094)
0.44
(0.28, 0.61)
0.043
(0.033, 0.055)
0.75
(0.73, 0.77)
0.83
(0.82, 0.84)
Posterior
median
(2.5%, 97.5%)
0.0024
(0.0016,
0.0032)
0.00018
(0.00012,
0.00024)
0.0043
(0.0031,
0.0055)
0.078
(0.062, 0.092)
0.48
(0.32, 0.61)
0.044
(0.035, 0.054)
0.75
(0.74, 0.77)
0.83
(0.82, 0.84)
OO
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-37. Distribution parameters: prior and posterior combined uncertainty and variability
Parameter description
TCE blood:air partition
coefficient
TCE fat:blood partition
coefficient
TCE gut:blood partition
coefficient
TCE livenblood partition
coefficient
TCE rapidly perfused :blood
partition coefficient
TCE respiratory tissue:air
partition coefficient
TCE kidney:blood partition
coefficient
TCE slowly perfused: blood
partition coefficient
TCA blood:plasma
concentration ratio
Free TCA body:blood
plasma partition coefficient
Free TCA livenblood
plasma partition coefficient
Protein TCA dissociation
constant (umole/L)
Maximum binding
concentration (umole/L)
PBPK
parameter
PB
PFat
PGut
PLiv
PRap
PResp
PKid
PSIw
TCAPIas
PBodTCA
PLivTCA
kDissoc
BMAX
Mouse
Prior
median
(2.5%, 97.5%)
15
(8.2, 27)
36
(17,74)
1.9
(0.72, 5.1)
1.7
(0.65, 4.5)
1.9
(0.72, 5)
2.6
(0.98, 6.8)
2.1
(0.8, 5.6)
2.4
(0.92, 6.4)
0.8
(0.35, 19)
0.82
(0.21, 19)
1.1
(0.3, 25)
110
(5.8, 2,000)
95
(4.1,2,200)
Posterior
median
(2.5%, 97.5%)
14
(7.5, 29)
35
(18,71)
1.5
(0.71,3.8)
2.2
(0.82, 4.7)
1.8
(0.77, 4.5)
2.5
(1.1,6.2)
2.7
(0.9,6.1)
2.2
(0.96, 5.6)
1.1
(0.65, 2.6)
0.89
(0.4, 2.5)
1.1
(0.48, 3.1)
130
(11,1 ,600)
140
(9.3, 2,200)
Rat
Prior
median
(2.5%, 97.5%)
22
(12,41)
27
(13,56)
1.4
(0.53, 3.7)
1.5
(1,2.2)
1.3
(0.5, 3.4)
1
(0.38, 2.6)
1.3
(0.63, 2.7)
0.58
(0.28, 1.2)
0.79
(0.53, 1.1)
0.7
(0.12,3.9)
0.92
(0.16, 5.1)
280
(62, 1 ,200)
330
(50,2,100)
Posterior
median
(2.5%, 97.5%)
19
(11,34)
31
(17,57)
1.2
(0.55, 2.7)
1.5
(1.1,2.1)
1.3
(0.56, 3)
1
(0.45, 2.3)
1.2
(0.66, 2.3)
0.72
(0.37, 1.3)
0.78
(0.61,0.97)
0.77
(0.24, 2.7)
1.2
(0.31,4)
270
(76, 860)
320
(68, 1,400)
Human
Prior
median
(2.5%, 97.5%)
9.6
(5.9, 16)
67
(41, 110)
2.6
(0.99, 6.8)
4.1
(1.5,11)
2.6
(0.99, 6.8)
1.3
(0.5, 3.5)
1.6
(0.98, 2.6)
2.1
(1,4.4)
0.78
(0.53, 18)
0.5
(0.15, 10)
0.63
(0.2, 13)
180
(160,210)
840
(530, 1 ,300)
Posterior
median
(2.5%, 97.5%)
9.3
(6.2, 14)
57
(41,87)
2.8
(1.2,6.1)
4.1
(2, 8.3)
2.4
(1 , 6.2)
1.3
(0.64, 2.7)
1.6
(1.1,2.3)
2.4
(0.96, 4.9)
0.64
(0.54, 2.7)
0.43
(0.2, 1.7)
0.54
(0.26, 2.3)
180
(160,200)
740
(520, 1,100)
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-37. Distribution parameters: prior and posterior combined uncertainty and variability (continued)
Parameter description
TCOH body:blood partition
coefficient
TCOH livenbody partition
coefficient
TCOG body:blood partition
coefficient
TCOG livenbody partition
coefficient
DCVG effective volume of
distribution
PBPK
parameter
PBodTCOH
PLivTCOH
PBodTCOG
PLivTCOG
VDCVG
Mouse
Prior
median
(2.5%, 97.5%)
1.1
(0.49, 2.5)
1.3
(0.58, 2.9)
1.1
(0.015,84)
1.3
(0.017, 100)
—
Posterior
median
(2.5%, 97.5%)
0.89
(0.48, 1.9)
1.9
(0.74, 3.4)
0.47
(0.13, 1.6)
1.3
(0.36, 4.6)
—
Rat
Prior
median
(2.5%, 97.5%)
1.1
(0.2, 5.9)
1.3
(0.24,7.1)
0.47
(0.021, 15)
1.3
(0.052, 33)
—
Posterior
median
(2.5%, 97.5%)
1
(0.26, 3.8)
1.2
(0.28, 5.6)
1.9
(0.09, 19)
9.7
(2.4, 47)
—
Human
Prior
median
(2.5%, 97.5%)
0.9
(0.4, 2)
0.6
(0.26, 1.3)
0.75
(0.03, 18)
1.7
(0.092, 29)
64
(4.8, 37,000)
Posterior
median
(2.5%, 97.5%)
1.5
(0.76, 2.4)
0.64
(0.34, 1.1)
0.69
(0.014,44)
3.1
(0.074, 43)
6.1
(4.8, 7.8)
i
PB = TCE blood-air partition coefficient.
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-38. Absorption parameters: prior and posterior combined uncertainty and variability
Parameter description
TCE stomach absorption
coefficient (/h)
TCE stomach-duodenum
transfer coefficient (/h)
TCE duodenum absorption
coefficient (/h)
TCA stomach absorption
coefficient (/h)
TCOH stomach absorption
coefficient (/h)
PBPK
parameter
kAS
kTSD
kAD
kASTCA
kASTCOH
Mouse
Prior
median
(2.5%, 97.5%)
1.6
(0.0022, 890)
1.3
(0.019,99)
0.78
(0.0012,460)
0.7
(0.0011,450)
0.79
(0.0012,490)
Posterior
median
(2.5%, 97.5%)
1.8
(0.052, 75)
5.2
(0.05, 98)
0.26
(0.0078, 15)
3.9
(0.016,300)
0.83
(0.0028, 160)
Rat
Prior
median
(2.5%, 97.5%)
1.3
(0.0022, 890)
1.5
(0.019, 100)
0.71
(0.0011,490)
0.77
(0.0012,470)
0.64
(0.0012,470)
Posterior
median
(2.5%, 97.5%)
2.4
(0.014,310)
3
(0.047, 94)
0.19
(0.0057, 5.3)
1.4
(0.032, 84)
0.72
(0.0064, 110)
Human
Prior
median
(2.5%, 97.5%)
—
-
—
0.69
(0.0012,480)
0.82
(0.0012,490)
Posterior
median
(2.5%, 97.5%)
—
-
—
4.4
(0.011,490)
7.7
(0.022, 460)
vo
i
^§
"*4-
o§
H I
O
H
W
-------
Table 3-39. TCE metabolism parameters: prior and posterior combined uncertainty and variability
to
vo £;•
I
I
§
***.
£3'
1
TO'
Parameter description
VMAX for hepatic TCE
oxidation (mg/h)
KM for hepatic TCE
oxidation (mg/L)
Fraction of hepatic TCE
oxidation not to
TCA+TCOH
Fraction of hepatic TCE
oxidation to TCA
VMAX for hepatic TCE GSH
conjugation (mg/h)
KM for hepatic TCE GSH
conjugation (mg/L)
VMAX for renal TCE GSH
conjugation (mg/h)
KM for renal TCE GSH
conjugation (mg/L)
VMAX fortracheo-bronchial
TCE oxidation (mg/h)
KM fortracheo-bronchial
TCE oxidation (mg/L)
Fraction of respiratory
metabolism to systemic
circ.
PBPK
parameter
VMAX
KM
FracOther
FracTCA
VMaxDCVG
KMDCVG
VMaxKidDCVG
KMKidDCVG
VMaxClara
KMClara
FracLungSys
Mouse
Prior
median
(2.5%, 97.5%)
4.3
(0.72, 27)
35
(2.3, 520)
0.47
(0.0015, 1)
0.07
(0.00021,0.66)
4.8
(0.0072, 3,300)
220
(0.0043,
8,200,000)
0.3
(0.00046, 200)
180
(0.0043,
7,600,000)
0.3
(0.016,6)
1.1
(0.0014,670)
0.51
(0.0014, 1)
Posterior
median
(2.5%, 97.5%)
2.4
(0.7, 10)
2.7
(0.69, 23)
0.023
(0.0025, 0.19)
0.13
(0.052, 0.31)
0.65
(0.0084, 640)
2,500
(0.11,
3,700,000)
0.029
(0.0011,22)
220
(0.11,430,000)
0.45
(0.012,6.1)
0.011
(0.0017, 0.18)
0.79
(0.15, 1)
Rat
Prior
median
(2.5%, 97.5%)
6
(1,36)
21
(0.81,610)
0.026
(0.0014,0.54)
0.22
(0.024, 0.74)
2.3
(0.012, 1,500)
1,700
(1,4,000,000)
0.038
(0.00024, 13)
480
(0.34, 760,000)
0.19
(0.005,4.1)
0.015
(0.0013,0.67)
0.81
(0.036, 1)
Posterior
median
(2.5%, 97.5%)
5.4
(1.8, 17)
0.72
(0.35, 4)
0.28
(0.017,0.87)
0.047
(0.0072,0.14)
6.5
(0.15,330)
6,700
(87, 780,000)
0.0025
(0.00042,
0.02)
0.27
(0.02, 3.6)
0.2
(0.0056, 2.3)
0.025
(0.0034, 0.84)
0.75
(0.049, 0.99)
Human
Prior
median
(2.5%, 97.5%)
430
(72, 2,500)
3.8
(0.11, 140)
0.12
(0.0058,
0.77)
0.18
(0.011,0.78)
96
(0.0066,
1,200,000)
2.9
(0.17,50)
170
(0.018,
1,800,000)
2.6
(0.15,48)
25
(0.84, 490)
0.022
(0.0016,0.6)
0.75
(0.042, 0.99)
Posterior
median
(2.5%, 97.5%)
180
(59, 930)
0.16
(0.017,3.8)
0.1
(0.0064, 0.67)
0.034
(0.0081,0.21)
320
(8.5, 12,000)
3.4
(0.16,77)
2.1
(0.035, 120)
0.78
(0.22, 7)
17
(0.74, 160)
0.27
(0.0029, 65)
0.96
(0.81,0.99)
^§
H I
O >
HH Oq
H TO
O
H
W
-------
Table 3-40. Metabolite metabolism parameters: prior and posterior combined uncertainty and variability
to
vo £;•
I
I
§
***.
£3'
1
TO'
Parameter description
VMAX for hepatic
TCOH^TCA (mg/h)
KM for hepatic TCOH^TCA
(mg/L)
VMAx for hepatic
TCOH^TCOG (mg/h)
KM for hepatic
TCOH^TCOG (mg/L)
Rate constant for hepatic
TCOH^other (/h)
Rate constant for TCA
plasma— >urine (/h)
Rate constant for hepatic
TCA^other (/h)
Rate constant for TCOG
liver— >bile (/h)
Lumped rate constant for
TCOG bile^TCOH liver (/h)
Rate constant for
TCOG^urine (/h)
Rate constant for hepatic
DCVG^DCVC (/h)
Lumped rate constant for
DCVC^urinary NAcDCVC
(/h)
Rate constant for DCVC
bioactivation (/h)
PBPK
parameter
VMaxTCOH
KMTCOH
VMaxGluc
KMGluc
kMetTCOH
kUrnTCA
kMetTCA
kBile
kEHR
kUrnTCOG
kDCVG
kNAT
kKidBioact
Mouse
Prior
median
(2.5%, 97.5%)
0.066
(0.000012,450)
0.85
(0.00017,6,000)
0.085
(0.000012,430)
1.1
(0.0015,670)
0.27
(0.000038, 1,500)
25
(0.3, 2,000)
0.26
(0.00036, 160)
0.25
(0.00035, 160)
0.23
(0.00034, 160)
0.67
(0.000089, 4,800)
—
"
—
Posterior
median
(2.5%, 97.5%)
0.12
(0.03, 0.52)
0.92
(0.2,4.1)
4.8
(1 .4, 25)
34
(2.7, 200)
8.7
(1 .3, 36)
3.1
(0.59, 15)
1.5
(0.45, 5)
2.4
(0.5, 13)
0.036
(0.0024,0.16)
12
(0.62, 420)
—
"
—
Rat
Prior
median
(2.5%, 97.5%)
0.67
(0.023, 21)
0.94
(0.029, 33)
27
(0.8,910)
28
(0.73, 580)
4.5
(0.14, 160)
1.9
(0.16,54)
0.82
(0.026, 24)
1.3
(0.04, 44)
0.016
(0.00045, 0.69)
10
(0.078, 1,200)
—
0.13
(0.00021,92)
0.14
(0.00021 , 90)
Posterior
median
(2.5%, 97.5%)
0.71
(0.14,3.8)
19
(1.8, 130)
11
(1.3, 120)
6.1
(0.25, 54)
2.5
(0.25, 31)
0.98
(0.29, 3.5)
0.47
(0.11, 1.7)
12
(1.7,64)
1.8
(0.12, 11)
9.1
(0.27, 540)
—
0.003
(0.00048, 0.022)
0.0087
(0.00091,0.057)
Human
Prior
median
(2.5%, 97.5%)
42
(0.61,3,300)
4.8
(0.23, 100)
820
(11,56,000)
11
(0.46, 250)
0.79
(0.036, 18)
0.26
(0.031,4.9)
0.16
(0.0079, 3.2)
1.1
(0.053, 20)
0.076
(0.0031, 1.8)
2.6
(0.027, 230)
0.034
(0.000053, 22)
0.00085
(0.00005, 0.034)
0.0021
(0.000072, 0.09)
Posterior
median
(2.5%, 97.5%)
9
(0.83, 110)
2.2
(0.29, 21)
890
(89, 5,800)
130
(20, 490)
0.26
(0.0046, 6.9)
0.12
(0.032, 0.45)
0.1
(0.011,0.67)
2.6
(0.55, 11)
0.054
(0.016,0.19)
2.2
(0.0067, 640)
2.5
(1.1,5.9)
0.00011
(0.000038,
0.00099)
0.023
(0.0036, 0.095)
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 For distribution parameters (see Table 3-37), there were only relatively minor changes
2 between prior and posterior distributions for TCE and TCOH partition coefficients. The
3 posterior distributions for several TCA partition coefficients and plasma binding parameters
4 were substantially narrower than their corresponding priors, but the central estimates were
5 similar, meaning that values at the high and low extremes were not likely. For TCOG as well,
6 partition coefficient posterior distributions were substantially narrower, which was expected
7 given the greater uncertainty in the prior distributions (TCOH partition coefficients were used as
8 a proxy). Again, posterior distributions indicated that the high and low extremes were not likely.
9 Finally, posterior distribution for the distribution volume for DCVG was substantially narrower
10 than the prior distribution, which only provided a lower bound given by the blood volume. In
11 this case, the upper bounds were substantially lower in the posterior.
12 Posterior distributions for oral absorption parameters (see Table 3-38) in mice and rats
13 (there were no oral studies in humans) were also informed by the data, as reflected in their being
14 substantially more narrow than the corresponding priors. Finally, with a few exceptions, TCE
15 and metabolite kinetic parameters (see Tables 3-39-3-40) showed substantially narrower
16 posterior distributions than prior distributions, indicating that they were fairly well specified by
17 the in vivo data. The exceptions were the VMAX for hepatic oxidation in humans (for which there
18 was substantial in vitro data) and the VMAX for respiratory metabolism in mice and rats (although
19 the posterior distribution for the KM for this pathway was substantially narrower than the
20 corresponding prior).
21 In terms of general consistency between prior and posterior distributions, in only a few
22 cases did the interquartile regions of the prior and posterior distributions not overlap. In most of
23 these cases, including the diffusion rate from respiratory lumen to tissue, the KMs for renal TCE
24 GSH conjugation and respiratory TCE oxidation, and several metabolite kinetic parameters, the
25 prior distributions themselves were noninformative. For a noninformative prior, the lack of
26 overlap would only be an issue if the posterior distributions were affected by the truncation limit,
27 which was not the case here. The only other parameter for which there was a lack of
28 interquartile overlap between the prior and posterior distribution was the KM for hepatic TCE
29 oxidation in mice and in rats, though the prior and posterior 95% confidence intervals did
30 overlap within each species. As discussed Section 3.3, there is some uncertainty in the
31 extrapolation of in vitro KM values to in vivo values (within the same species). In addition, in
32 mice, it has been known for some time that KM values appear to be discordant among different
33 studies (Abbas and Fisher, 1997; Greenberg et al., 1999; Fisher et al., 1991).
34 In sum, the Bayesian analysis of the updated PBPK model and data exhibited no major
35 inconsistencies in prior and posterior parameter distributions. The most significant issue was the
36 KM for hepatic oxidative metabolism, for which the posterior estimates were low compared to,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-95 DRAFT—DO NOT CITE OR QUOTE
-------
1 albeit somewhat uncertain, in vitro estimates, and it could be argued that a wider prior
2 distribution would have been better. However, the central estimates were not at or near the
3 truncation boundary, so it is unlikely that wider priors would change the results substantially.
4 Therefore, there were no indications based on this evaluation of prior and posterior distributions
5 either that prior distributions were overly restrictive or that model specification errors led to
6 pathological parameter estimates.
7
8 3.5.6.3. Comparison of Model Predictions With Data
9 As with the Hack et al. (2006) model, initially the sampled group- or individual-specific
10 parameters were used to generate predictions for comparison to the calibration data (see
11 Figure 3-8). Thus, the predictions for a particular data set are conditioned on the posterior
12 parameter distributions for same data set. Because these parameters were "optimized" for each
13 experiment, these group- or individual-specific predictions should be accurate by design—and,
14 on the whole, were so. In addition, the "residual error" estimate for each measurement (see
15 Table 3-41) provides some quantitative measure of the degree to which there were deviations due
16 to intrastudy variability and model misspecification, including any difficulties fitting multiple
17 dose levels in the same study using the same model parameters.
18 Next, only samples of the population parameters (means and variances) were used, and
19 "new" groups or individuals were sampled from appropriate distribution using these population
20 means and variances (see Figure 3-8). That is, the predictions were only conditioned on the
21 population-level parameters distributions, representing an "average" over all the data sets, and
22 not on the specific predictions for that data set. These "new" groups or individuals then
23 represent the predicted population distribution, incorporating variability in the population as well
24 as uncertainty in the population means and variances. Because of the limited amount of mouse
25 data, all available data for that species was utilized for calibration, and there was no data
26 available for "out-of-sample" evaluation (often referred to as "validation data," but this term is
27 not used here due to ambiguities as to its definition). In rats, several studies that contained
28 primarily blood TCE data, which were abundant, were used for out-of-sample evaluation. In
29 humans, there were substantial individual and aggregated (group mean) data that was available
30 for out-of-sample evaluation, as computational intensity limited the number of individuals that
31 could be used in the MCMC-based calibration.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-96 DRAFT—DO NOT CITE OR QUOTE
-------
MCMC outputs
Posterior
Posterior I2
Posterior group-
specific
e.
Posterior
prediction
ulation
9i
Group/
Individuc
Eij
tij
Experiment]
3l i
Posterior group^specific
prediction
Yii
Ji IL.
4
5
6
7
Figure 3-8. Schematic of how posterior predictions were generated for
comparison with experimental data. Two sets of posterior predictions were
generated: population predictions (diagonal hashing) and group-specific
predictions (vertical hashing). (Same as Figure A-2 in Appendix A)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-97 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-41. Estimates of the residual error
Measurement
abbreviation
RetDose
CAIvPPM
ClnhPPM
CMixExh
CArt
CVen
CBIdMix
CFat
CGut
CKid
CLiv
CMus
AExhpost
CTCOH
CLivTCOH
CPIasTCA
CBIdTCA
CLivTCA
AUrnTCA
AUrnTCA_collect
ABileTCOG
CTCOG
CTCOGTCOH
CLivTCOGTCOH
AUrnTCOGTCOH
AUrnTCOGTCOH_c
ollect
CDCVGmol
AUrnNDCVC
AUrnTCTotMole
TotCTCOH
Measurement description
Retained TCE dose (mg)
TCE concentration in alveolar air (ppm)
TCE concentration in closed chamber (ppm)
TCE concentration in mixed exhaled air (mg/L)
TCE concentration in arterial blood (mg/L)
TCE concentration in venous blood (mg/L)
TCE concentration in mixed arterial and venous
blood (mg/L)
TCE concentration in fat (mg/L)
TCE concentration in gut (mg/L)
TCE concentration in kidney (mg/L)
TCE concentration in liver (mg/L)
TCE concentration in muscle (mg/L)
Amount of TCE exhaled postexposure (mg)
Free TCOH concentration in blood (mg/L)
Free TCOH concentration in liver (mg/L)
TCA concentration in plasma (mg/L)
TCA concentration in blood (mg/L)
TCA concentration in liver (mg/L)
Cumulative amount of TCA excreted in urine (mg)
Cumulative amount of TCA collected in urine
(noncontinuous sampling) (mg)
Cumulative amount of bound TCOH excreted in
bile (mg)
Bound TCOH concentration in blood
Bound TCOH concentration in blood in free TCOH
equivalents
Bound TCOH concentration in liver in free TCOH
equivalents (mg/L)
Cumulative amount of total TCOH excreted in
urine (mg)
Cumulative amount of total TCOH collected in
urine (noncontinuous sampling) (mg)
DCVG concentration in blood (mmol/L)
Cumulative amount of NAcDCVC excreted in
urine (mg)
Cumulative amount of TCA+total TCOH excreted
in urine (mmol)
Total TCOH concentration in blood (mg/L)
GSD for "residual" error
(median estimate)
Mouse
-
-
1.18
-
-
2.68
1.61
2.49
-
2.23
1.71
-
1.23
1.54
1.59
1.40
1.49
1.34
1.34
-
-
-
1.49
1.63
1.26
-
-
-
-
1.85
Rat
-
-
1.11-1.12
1.5
1.17-1.52
1 .22-4.46
1.5
1.85-2.66
1.86
1.47
1.67-1.78
1.65
1.12-1.17
1.14-1.64
-
1.13-1.21
1.13-1.59
1.67
1.18-1.95
-
2.13
2.76
-
-
1.12-2.27
-
-
1.17
1.12-1.54
1.49
Human
1.13
1.44-1.83
-
-
-
1 .62-2.95
-
-
-
-
-
-
-
1.14-2.1
-
1.12-1.17
1.12-1.49
-
1.11-1.54
2-2.79
-
-
-
-
1.11-1.13
1.3-1.63
1.53
1.17
-
1.2-1.69
3
4
Values higher than 2-fold are in bold.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-98 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.5.6.3.1. Mouse model and data. Table 3-42 provides an evaluation of the predictions of the
2 mouse model for each data set, with figures showing data and predictions in Appendix A. With
3 exception of the remaining over-prediction of TCE in blood following inhalation exposure, the
4 parent PBPK model (for TCE) appears to now be robust in mice. Most of the problems
5 previously encountered with the Abbas and Fisher (1997) gavage data were solved by allowing
6 absorption from both of the stomach and duodenal compartments. Notably, the addition of
7 possible wash-in/wash-out, respiratory metabolism, and extrahepatic metabolism (i.e., kidney
8 GSH conjugation) was insufficient to remove the long-standing discrepancy of PBPK models
9 over-predicting TCE blood levels, suggesting another source of model or experimental error is
10 the cause. However, the availability of tissue concentration levels of TCE somewhat ameliorates
11 this limitation.
12 In terms of TCA and TCOH, the overall mass balance and metabolic disposition to these
13 metabolites also appeared to be robust, as urinary excretion following dosing with TCE, TCOH,
14 as well as TCA could be modeled accurately. This improvement over the Hack et al. (2006)
15 model was likely due in part to the addition of nonurinary clearance ("untracked" metabolism) of
16 TCA and TCOH. Also, the addition of a liver compartment for TCOH and TCOG, so that first-
17 pass metabolism could be properly accounted for, was essential for accurate simulation of the
18 metabolite pharmacokinetics both from i.v. dosing of TCOH and from exposure to TCE.
19 These conclusions are corroborated by the estimated "residual" errors, which include
20 intrastudy variability, interindividual variability, and measurement and model errors. The
21 implied GSD for this error in each in vivo measurement is presented in Table 3-41. As expected,
22 the venous blood TCE concentration had the largest residual error, with a GSD of 2.7, reflecting
23 largely the difficulty in fitting TCE blood levels following inhalation exposure. In addition, the
24 fat and kidney TCE concentrations also are somewhat uncertain, with a GSD for the residual
25 error of 2.5 and 2.2, respectively, while other residual errors had GSD of less than 2-fold. These
26 tissues were only measured in two studies, Abbas and Fisher (1997) and Greenberg et al. (1999),
27 and the residual error reflects the difficulties in simultaneously fitting the model to the different
28 dose levels with the same parameters.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-99 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-42. Summary comparison of updated PBPK model predictions and
in vivo data in mice
Study
Exposure(s)
Discussion
Abbas and
Fisher, 1997
TCE gavage
(corn oil)
Generally, model predictions were quite good, especially with
respect to tissue concentrations of TCE, TCA, and TCOH. There
were some discrepancies in TCA and TCOG urinary excretion and
TCA and TCOG concentrations in blood due to the requirement
(unlike in Hack et al. [2006]) that all experiments in the same study
utilize the same parameters. Thus, for instance, TCOG urinary
excretion was accurately predicted at 300 mg/kg, underpredicted at
600 mg/kg, over-predicted at 1,200 mg/kg, and underpredicted again
at 2,000 mg/kg, suggesting significant intraexperimental variability
(not addressed in the model).
Population predictions were quite good, with the almost all of the
data within the 95% Cl of the predictions, and most within the
interquartile region.
Abbas et al.,
1997
TCOH, TCA
i.v.
Both group-specific and population predictions were quite good.
Urinary excretion, which was over-predicted by the Hack et al.
(2006) model, was accurately predicted due to the allowance of
additional "untracked" clearance. In the case of population
predictions, almost all of the data were within the 95% Cl of the
predictions, and most within the interquartile region.
Fisher and Allen,
1993
TCE gavage
(corn oil)
Both group-specific and population predictions were quite good.
Some discrepancies in the time-course of TCE blood concentrations
were evidence across doses in the group-specific predictions, but
not in the population predictions, suggesting significant intragroup
variability (not addressed in the model).
Fisher et al.,
1991
TCE
inhalation
Blood TCE levels during and following inhalation exposures
were still over-predicted at the higher doses. However, there was
the stringent requirement (absent in Hack et al. [2006]) that the
model utilize the same parameters for all doses and in both the
closed and open chamber experiments. Moreover, the Hack et al.
(2006) model required significant differences in the parameters for
the different closed chamber experiments, while predictions here
were accurate utilizing the same parameters across different initial
concentrations. These conclusions were the same for group-specific
and population predictions (e.g., TCE blood levels remained over-
predicted in the later case).
Green and Prout,
1985
TCE gavage
(corn oil)
Both group-specific and population predictions were adequate,
though the data collection was sparse. In the case of population
predictions, almost all of the data were within the 95% Cl of the
predictions, and about half within the interquartile region.
Greenberg et al.,
1999
TCE
inhalation
Model predictions were quite good across a wide variety of
measures that included tissue concentrations of TCE, TCA, and
TCOH. However, as with the Hack et al. (2006) predictions, TCE
blood levels were over-predicted by up to 2-fold. Population
predictions were quite good, with the exception of TCE blood levels.
Almost all of the other data was within the 95% Cl of the predictions,
and most within the interquartile region.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-100 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-42. Summary comparison of updated PBPK model predictions and
in vivo data in mice (continued)
Study
Larson and Bull,
1992a
Larson and Bull,
1992b
Merdink et al.,
1998
Proutetal., 1985
Templin et al.,
1993
Exposure(s)
TCE gavage
(aqueous)
TCA gavage
(aqueous)
TCE i.v.
TCE gavage
(corn oil)
TCE gavage
(aqueous)
Discussion
Both group-specific and population predictions were quite good,
though the data collection was somewhat sparse. In the case of
population predictions, all of the data were within the 95% Cl of the
predictions,
Both group-specific and population predictions were quite good.
In the case of population predictions, most of the data were within
the interquartile region.
Both group-specific and population predictions were quite good,
though the data collection was somewhat sparse. In the case of
population predictions, all of the data were within the 95% Cl of the
predictions,
Both group-specific and population predictions were adequate,
though there was substantial scatter in the data due to the use of
single animals at each data point.
Both group-specific and population predictions were quite good.
With respect to population predictions, almost all of the other data
was within the 95% Cl of the predictions, and most within the
interquartile region.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
In terms of total metabolism, closed-chamber data were fit accurately with the updated
model. While the previous analyses of Hack et al. (2006) allowed for each chamber experiment
to be fit with different parameters, the current analysis made the more restrictive assumption that
all experiments in a single study utilize the same parameters. Furthermore, the accuracy of
closed chamber predictions did not require the very high values for cardiac output that were used
by Fisher et al. (1991), confirming the suggestion (discussed in Appendix A) that additional
respiratory metabolism would resolve this discrepancy. The accurate model means that
uncertainty with respect to possible wash-in/wash-out, respiratory metabolism, and extrahepatic
metabolism could be well characterized. For instance, the absence of in vivo data on GSH
metabolism in mice means that this pathway remains relatively uncertain; however, the current
model should be reliable for estimating lower and upper bounds on the GSH pathway flux.
3.5.6.3.2. Rat model and data. A summary evaluation of the predictions of the rat model as
compared to the data are provided in Tables 3-43 and 3-44, with figures showing data and
predictions in Appendix A. Similar to previous analyses (Hack et al., 2006), the TCE submodel
for the rat appears to be robust, with blood and tissue concentrations accurately predicted.
Unlike in the mouse, some data consisting of TCE blood and tissue concentrations were used for
"out-of-sample evaluation" (sometimes loosely termed "validation"). These data were generally
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-101 DRAFT—DO NOT CITE OR QUOTE
-------
1 well simulated; most of the data within the 95% confidence interval of posterior predictions.
2 This provides additional confidence in the predictions for the parent compound.
3 In terms of TCA and TCOH, as with the mouse, the overall mass balance and metabolic
4 disposition to these metabolites also appeared to be robust: urinary excretion following dosing
5 with TCE, TCOH, as well as TCA, could be modeled accurately, and, secondly, the residual
6 errors did not indicate substantial mis-fit (GSD < 1.25). This improvement over the Hack et al.
7 (2006) model was likely due in part to the addition of nonurinary clearance ("untracked"
8 metabolism) of TCA and TCOH. In addition, the addition of a liver compartment for TCOH and
9 TCOG, so that first-pass metabolism could be properly accounted for, was essential for accurate
10 simulation of the metabolite pharmacokinetics both from i.v. dosing of TCOH and from TCE
11 exposure. Blood and plasma concentrations of TCA and TCOH were fairly well simulated, with
12 GSD for the residual error of 1.2-1.3, but a bit more discrepancy was evident with TCA liver
13 concentrations. However, TCA liver concentrations were only available in one study (Yu et al.,
14 2000), and the data show a change in the ratio of liver to blood concentrations at the last time
15 point, which may be the source of the added residual error.
16 In terms of total metabolism, as with the mouse, closed-chamber data were fit accurately
17 with the updated model (residual error GSD of about 1.11). In addition, the data on NAcDCVC
18 urinary excretion was well predicted (residual error GSD of 1.18), in particular the fact that
19 excretion was still ongoing at the end of the experiment (see Figure 3-9, panels A and B). Thus,
20 there is greater confidence in the estimate of the flux through the GSH pathway than there was
21 from the Hack et al. (2006) model. However, the overall flux is still estimated indirectly, and
22 there remains some ambiguity as to the relative contributions respiratory wash-in/wash-out,
23 respiratory metabolism, extrahepatic metabolism, DCVC bioactivation versus 7V-acetylation, and
24 oxidation in the liver producing something other than TCOH or TCA. Therefore, there remain a
25 large range of possible values for the flux through the GSH conjugation and other indirectly
26 estimated pathways that are nonetheless consistent with all the available in vivo data. The use of
27 noninformative priors for the metabolism parameters for which there were no in vitro data means
28 that a fuller characterization of the uncertainty in these various metabolic pathways could be
29 achieved. Thus, the model should be reliable for estimating lower and upper bounds on several
30 of these pathways.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-102 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for "calibration" in rats
Study
Exposure(s)
Discussion
Bernauer et al.
1996
TCE inhalation
Posterior fits to these data were adequate, especially with the
requirement that all predictions for dose levels utilize the same PBPK
model parameters. Predictions of TCOG and TCA urinary excretion
was relatively accurate, though the time-course of TCA excretion
seemed to proceed more slowly with increasing dose, an aspect not
captured in by model. Importantly, unlike the Hack et al. (2006)
results, the time-course of NAcDCVC excretion was quite well
simulated, with the excretion rate remaining non-negligible at the last
time point (48 h). It is likely that the addition of the DCVG submodel
between TCE and DCVC, along with prior distributions that accurately
reflected the lack of reliable independent (e.g., in vitro) data on
bioactivation, allowed for the better fit.
Dallas et al.,
1991
TCE inhalation
These data, consisting of arterial blood and exhaled breath
concentrations of TCE, were accurately predicted by the model using
both group-specific and population sampled parameters. In the case
of population predictions, most of the data were within the 95% Cl of
the predictions.
Fisher et al.,
1989
TCE inhalation
These data, consisting of closed chamber TCE concentrations,
were accurately simulated by the model using both group-specific
and population sampled parameters. In the case of population
predictions, most of the data were within the 95% Cl of the
predictions.
Fisher et al.,
1991
TCE inhalation
These data, consisting of TCE blood, and TCA blood and urine
time-courses, were accurately simulated by the model using both
group-specific and population sampled parameters. In the case of
population predictions, most of the data were within the 95% Cl of the
predictions.
Green and
Prout, 1985
TCE gavage
(corn oil)
TCA i.v.
TCA gavage
(aqueous)
For TCE treatment, these data, consisting of one time point each
in urine for TCA, TCA +TCOG, and TCOG, were accurately simulated
by both group-specific and population predictions.
For TCA i.v. treatment, the single datum of urinary TCA+TCOG at
24 h was at the lower 95% Cl in the group-specific simulations, but
accurately predicted with the population sampled parameters,
suggesting intrastudy variability is adequately accounted for by
population variability.
For TCA gavage treatment, the single datum of urinary
TCA+TCOG at 24 h was accurately simulated by both group-specific
and population predictions.
Hissink et al.,
2002
TCE gavage
(corn oil)
TCE i.v.
These data, consisting of TCE blood, and TCA+TCOG urinary
excretion time-courses, were accurately simulated by the model using
group-specific parameters. In the case of population predictions,
TCA+TCOH urinary excretion appeared to be somewhat under-
predicted.
Kaneko et al.
1994
TCE inhalation
These data, consisting of TCE blood and TCA and TCOG urinary
excretion time-courses, were accurately predicted by the model using
both group-specific and population sampled parameters. In the case
of population predictions, TCA+TCOH urinary excretion appeared to
be somewhat underpredicted, However, all of the data were within
the 95% Cl of the predictions.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-103 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for "calibration" in rats (continued)
Study
Exposure(s)
Discussion
Keys et al.,
2003
TCE inhalation,
gavage
(aqueous), i.a.
These data, consisting of TCE blood, gut, kidney, liver, muscle
and fat concentration time-courses, were accurately predicted by the
model using both group-specific and population sampled parameters.
In the case of population predictions, most of the data were within the
95% Cl of the predictions.
Kimmerle and
Eben, 1973a
TCE inhalation
Some inaccuracies were noted in group-specific predictions,
particularly with TCA and TCOG urinary excretion, TCE exhalation
postexposure, and TCE venous blood concentrations. In the case of
TCA excretion, the rate was underpredicted at the lowest dose (49
mg/kg) and over-predicted at 330 ppm. In terms of TCOG urinary
excretion, the rate was over-predicted at 175 ppm and
underpredicted at 330 ppm. Similarly for TCE exhaled postexposure,
there was some over-prediction at 175 ppm and some
underprediction at 300 ppm. Finally, venous blood concentrations
were over-predicted at 3,000 ppm. However, for population
predictions, most of the data were within with 95% confidence region.
Larson and
Bull, 1992a
TCA gavage
(aqueous)
These data, consisting of TCA plasma time-courses, were
accurately predicted by the model using both group-specific and
population sampled parameters. In the case of population
predictions, all of the data were within the 95% Cl of the predictions.
Larson and
Bull, 1992b
TCE gavage
(aqueous)
These data, consisting of TCE, TCA, and TCOH in blood, were
accurately predicted by the model using both group-specific and
population sampled parameters. In the case of population
predictions, all of the data were within the 95% Cl of the predictions.
Lee et al.,
2000a
TCE i.v., p.v.
These data, consisting of TCE concentration time course in
mixed arterial and venous blood and liver, were predicted using both
the group specific and population predictions. In both cases, most of
the data were within the 95% Cl of the predictions.
Merdink et al.
1999
TCOH i.v.
TCOH blood concentrations were accurately predicted using
group-specific parameters. However, population-based parameters
seemed to lead to some under-prediction, though most of the data
were within the 95% Cl of the predictions.
Prout et al.,
1985
TCE gavage
(corn oil)
Most of these data were accurately predicted using both group-
specific and population-sampled parameters. However, at the
highest two doses (1,000 and 2,000 mg/kg), there were some
discrepancies in the (very sparsely collected) urinary excretion
measurements. In particular, using group-specific parameters,
TCA+TCOH urinary excretion was under-predicted at 1,000 mg/kg
and over-predicted at 2,000 mg/kg. Using population-sampled
parameters, this excretion was underpredicted in both cases, though
not entirely outside of the 95% Cl.
Simmons et al.,
2002
TCE inhalation
Most of these data were accurately predicted using both group-
specific and population-sampled parameters. In the open chamber
experiments, there was some scatter in the data that did not seem to
be accounted for in the model. The closed chamber data were
accurately fit.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-104 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for "calibration" in rats (continued)
Study
Exposure(s)
Discussion
Stenner et al.,
1997
TCE
intraduodenal
TCOH i.v.
TCOH i.v., bile-
cannulated
These data, consisting of TCA and TCOH in blood and TCA and
TCOG in urine, were generally accurately predicted by the model
using both group-specific and population sampled parameters.
However, using group-specific parameters, the amount of TCOG in
urine was over-predicted for 100 TCOH mg/kg i.v. dosing, though
total TCOH in blood was accurately simulated. In addition, in bile-
cannulated rats, the TCOG excretions at 5 and 20 mg/kg i.v. were
underpredicted, while the amount at 100 mg/kg was accurately
predicted. On the other hand, in the case of population predictions,
all of the data were within the 95% Cl of the predictions, and mostly
within the interquartile region, even for TCOG urinary excretion. This
suggests that intrastudy variability may be a source of the poor fit in
using the group-specific parameters.
Templin et al.,
1995
TCE oral
(aqueous)
These data, consisting of TCE, TCA, and TCOH in blood, were
accurately predicted by the model using both group-specific and
population sampled parameters. In the case of population
predictions, all of the data were within the 95% Cl of the predictions.
Yuetal.,2000
TCA i.v.
These data, consisting of TCA in blood, liver, plasma, and urine,
were generally accurately predicted by the model using both group-
specific and population sampled parameters. The only notable
discrepancy was at the highest dose of 50 mg/kg, in which the rate of
urinary excretion from 0-6 h appeared to more rapid than the model
predicted. However, all of the data were within the 95% Cl of the
predictions based on population-sampled parameters.
1
2
i.a. = infra-arterial, i.v. = intravenous, p.v. = intraperivenous.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-105 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-44. Summary comparison of updated PBPK model predictions and
in vivo data used for "out-of-sample" evaluation in rats
Study
Andersen et al.,
1987
Bruckner et al.,
unpublished
Fisher et al.,
1991
Jakobson et al.,
1986
Lee etal., 1996
Lee et al.,
2000a, b
Exposure(s)
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE i.a., i.v.,
p.v., gavage
TCE gavage
Discussion
These closed chamber data were well within the 95% Cl of the
predictions based on population-sampled parameters.
These data on TCE in blood, liver, kidney, fat, muscle, gut,
and venous blood, were generally accurately predicted based on
population-sampled parameters. The only notable exception was
TCE in the kidney during the exposure period at the 500 ppm
level, which were somewhat under-predicted (though levels
postexposure were accurately predicted).
These data on TCE in blood were well within the 95% Cl of
the predictions based on population-sampled parameters.
These data on TCE in arterial blood were well within the 95%
Cl of the predictions based on population-sampled parameters.
Except at some very early time-points (<0.5 h), these data on
TCE in blood were well within the 95% Cl of the predictions based
on population-sampled parameters.
These data on TCE in blood were well within the 95% Cl of
the predictions based on population-sampled parameters.
4
5
i.a. = intra-arterial, i.v. = intravenous, p.v. = intraperivenous.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-106 DRAFT—DO NOT CITE OR QUOTE
-------
cn
3.
a
o
a
I
CM
Hack et al. (2006) model (ral)
\n
d
q
d
A
/ i
i i
0 10
+
A
A
t
1 1 1
20 30 40
Time (h)
•a
O ~
Q
|
&
|
CM_
CM"
^_
O
CM_
—
<
Z
o
CM -
O _
CO
O -
D
\
0
I
10
I
20
I
30
I
40
I
50
Time (h)
Figure 3-9. Comparison of urinary excretion data for NAcDCVC and
predictions from the Hack et al. (2006) and the updated PBPK models. Data
are from Bernauer et al. (1996) for (A and B) rats or (C and D) humans exposed
for 6 h to 40 (o), 80 (A), or 160 (+) ppm in air (thick horizontal line denotes the
exposure period). Predictions from Hack et al. (2006) and the corresponding data
(A and C) are only for the 1,2 isomer, whereas those from the updated model (B
and D) are for both isomers combined. Parameter values used for each prediction
are a random sample from the group- or individual-specific parameters from the
rat and human MCMC chains (the last iteration of the first chain was used in each
case). Note that in the Hack et al. (2006) model, each dose group had different
model parameters, whereas in the updated model, all dose groups are required to
have the same model parameters. See files linked to Appendix A for comparisons
with the full distribution of predictions.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-107 DRAFT—DO NOT CITE OR QUOTE
-------
1 3.5.6.3.3. Human model. Table 3-45-3-46 provide a summary evaluation of the predictions of
2 the model as compared to the human data, with figures showing data and predictions in
3 Appendix A. With respect to the TCE submodel, blood and exhaled air measurements appeared
4 more robust than previously found from the Hack et al. (2006) model. TCE blood concentrations
5 from most studies were well predicted. However, those from Chiu et al. (2007) were
6 consistently over-predicted, and a few of those from Fisher et al. (1998) were consistently
7 underpredicted. Alveolar or mixed exhaled breath concentrations of TCE from all studies except
8 Fisher et al. (1998) were well predicted, though the discrepancy appeared smaller than that
9 originally reported by Fisher et al. (1998) for their PBPK model. In addition, the majority of the
10 "out-of-sample" evaluation data consisted of TCE in blood or breath, and were generally well
11 predicted, lending confidence to the model predictions for the parent compound.
12 In terms of TCA and TCOH, as with the mouse and rat, the overall mass balance and
13 metabolic disposition to these metabolites also appeared to be robust, as urinary excretion
14 following TCE exposure could be modeled accurately. However, data from Chiu et al. (2007)
15 indicated substantial interoccasion variability, as the same individual exposed to the same
16 concentration on different occasions sometimes had substantial differences in urinary excretion.
17 Since Chiu et al. (2007) was the only calibration study for which this urine collection was
18 intermittent, this interoccasion variability was also reflected in the larger residual error (GSD of
19 1.55 and 1.59 for TCA and TCOH, respectively—Table 3-41) for intermittent urine collection as
20 compared to cumulative collection (respective residual error GSD of 1.36 and 1.11). Blood and
21 plasma concentrations of TCA and free TCOH were fairly well simulated, with GSD for the
22 residual error of 1.1-1.4, though total TCOH in blood had greater residual error with GSD of
23 about 1.6. This partially reflects the "sharper" peak concentrations of total TCOH in the Chiu et
24 al. (2007) data relative to the model predictions. In addition, TCA and TCOH blood and urine
25 data were available from several studies for "out-of-sample" evaluation and were generally well
26 predicted by the model, lending further confidence to the model predictions for these
27 metabolites.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-108 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-45. Summary comparison of updated PBPK model predictions and
in vivo data used for "calibration" in humans
Reference
Exposure(s)
Discussion
Bernauer et al.,
1996
TCE inhalation
These data, consisting of TCA, TCOG and NAcDCVC
excreted in urine, were accurately predicted by the model using both
individual-specific and population sampled parameters. The posterior
NAcDCVC predictions were an important improvement over the
predictions of Hack et al. (2006), which predicted much more rapid
excretion than observed. The fit improvement is probably a result of
the addition of the DCVG submodel between TCE and DCVC, along
with the broader priors on DCVC excretion and bioactivation.
Interestingly, in terms of population predictions, the NAcDCVC
excretion data from this study were on the low end, though still within
the95%CI.
Chiu etal.,
2007
TCE inhalation
Overall, posterior predictions were quite accurate across most
of the individuals and exposure occasions. TCE alveolar breath
concentrations were well simulated for both individual-specific and
population-generated simulations, though there was substantial
scatter (intraoccasion variability). However, TCE blood concentrations
were consistently over-predicted in most of the experiments, both
using individual-specific and population-generated parameters. This
was not unexpected, as Chiu et al. (2007) noted the TCE blood
measurements to be lower by about 2-fold relative to previously
published studies. As discussed in Chiu et al. (2007), wash-in/wash-
out and extrahepatic (including respiratory) metabolism were not
expected to be able to account for the difference, and indeed all these
processes were added to the current model without substantially
improving the discrepancy.
With respect to metabolite data, TCA and total TCOH in blood
were relatively accurately predicted. There was individual
experimental variability observed for both TCA and TCOH in blood at
six hours (end of exposure). The population-generated simulations
over-predicted TCA in blood, while they were accurate in predicting
blood TCOH. Predictions of free TCOH in blood also showed over-
prediction for individual experiments, with variability at the end of
exposure timepoint. However, TCOH fits were improved for the
population-generated simulations. TCA and TCOG urinary excretion
was generally well simulated, with simulations slightly under- or over-
predicting the individual experimental data in some cases.
Fisher et al.
1998
TCE inhalation
The majority of the predictions for these data were quite
accurate. Interestingly, in contrast to the predictions for Chiu et al.
(2007), TCE blood levels were somewhat underpredicted in a few
cases, both from using individual-specific and population-generated
predictions. These two results together suggest some unaccounted-
for study-to-study variance, though interindividual variability cannot be
discounted as the data from Chiu et al. (2007) were from individuals in
the Netherlands and that from Fisher et al. (1998) were from
individuals in the United States. As reported by Fisher et al. (1998),
TCE in alveolar air was somewhat over-predicted in several cases,
however, the discrepancies seemed smaller than originally reported
for the Fisher etal. (1998) model.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-109 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-45. Summary comparison of updated PBPK model predictions and
in vivo data used for "calibration" in humans (continued)
Reference
Fisher et al.,
1998
(continued)
Kimmerle and
Eben, 1973b
Monster et al.,
1976
Mulleret al.,
1974
Paykocet al.,
1945
Exposure(s)
TCE inhalation
(continued)
TCE inhalation
TCE inhalation
TCA,
TCOH oral
TCA i.v.
Discussion
With respect to metabolite data, TCOH and TCA in blood and
TCOG and TCA in urine were generally well predicted, though data for
some individuals appeared to exhibit inter- and/or intraoccasion
variability. For example, in one case in which the same individual
(female) was exposed to both 50 and 100 ppm, the TCOH blood data
was over-predicted at the higher one exposure. In addition, in one
individual, initial individual-specific simulations for TCA in urine were
underpredicted but shifted to over-predictions towards the end of the
simulations. The population-generated results over-predicted TCA in
urine for the same individual. Given the results from Chiu et al.
(2007), interoccasion variability is likely to be the cause, though some
dose-related effect cannot be ruled out.
Finally, DCVG data was well predicted in light of the high
variability in the data and availability of only grouped data or data from
multiple individual who cannot be matched to the appropriate TCE and
oxidative metabolite data set. In all cases, the basic shape (plateau
and then sharp decline) and order of magnitude of the time-course
were well predicted, Furthermore, the range of the data was well-
captured by the 95% Cl of the population-generated predictions.
These data were well fit by the model, using either individual-
specific or population-generated parameters.
The data simulated in this case were exhaled alveolar TCE,
TCE in venous blood, TCA in blood, TCA in urine, and TCOG in urine.
Both using individual-specific and population-generated simulations,
all fits are within the 95% Cl. The one exception was the retained
dose for a male exposed to 65 ppm, which was outside the 95% Cl for
the population-generated results.
The data measured after oral TCA was timecourse TCA
measured in plasma and urine. Individual-specific predictions were
accurate, but both data sets were over-predicted in the population-
generated simulations.
The data measured after oral TCOH was timecourse TCOH in
blood, TCOG in urine, TCA in plasma, and TCA in urine. Individual-
specific predictions were accurate, but the population-generated
simulations over-predicted TCOH in blood and TCOG in urine. The
population-based TCA predictions were accurate.
These results indicate that "unusual" parameter values were
necessary in the individual-specific simulations to give accurate
predictions.
These data were well fit by the model, using either individual-
specific or population-generated parameters.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-110 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 3-46. Summary comparison of updated PBPK model predictions and
in vivo data used for "out-of-sample" evaluation in humans
Reference
Bartonicek, 1962
Bloemen et al., 2001
Fernandez et al., 1977
Lapare et al., 1995
Monster etal., 1979
Mulleretal., 1974,
1975
Sato etal., 1977
Stewart et a I., 1970
Treibig etal., 1976
Exposure(s)
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
TCE inhalation
Discussion
While these data were mostly within the 95% Cl of the
predictions, they tended to be at the high end for all the
individuals in the study.
These data were all well within the 95% Cl of the predictions.
These data were all well within the 95% Cl of the predictions.
These data were all well within the 95% Cl of the predictions.
These data were all well within the 95% Cl of the predictions.
Except for TCE in alveolar air, which was over-predicted during
exposure, these data were all well within the 95% Cl of the
predictions.
These data were all well within the 95% Cl of the predictions.
These data were all well within the 95% Cl of the predictions.
Except for TCE in alveolar air, these data were all well within
the 95% Cl of the predictions.
4
5
6 In terms of total metabolism, no closed-chamber data exist in humans, but alveolar breath
7 concentrations were generally well simulated, suggesting that total metabolism may be fairly
8 robust. In addition, as with the rat, the data on NAcDCVC urinary excretion was well predicted
9 (residual error GSD of 1.12), in particular the fact that excretion was still ongoing at the end of
10 the experiment (48 hrs after the end of exposure). Thus, there is greater confidence in the
11 estimate of the flux through the GSH pathway than there was from the Hack et al. (2006) model,
12 in which excretion was completed within the first few hours after exposure (see Figure 3-9,
13 panels C and D). If only urinary data were available, as is the case for the rat, the overall flux
14 would still estimated indirectly, and there would remain some ambiguity as to the relative
15 contributions respiratory wash-in/wash-out, respiratory metabolism, extrahepatic metabolism,
16 DCVC bioactivation versus TV-acetylation, and oxidation in the liver producing something other
17 than TCOH or TCA. However, unlike in the rat, the blood DCVG data, while highly variable,
18 nonetheless provide substantial constraints (at least a strong lower bound) on the flux of GSH
19 conjugation, and is well fit by the model (see Figure 3-10). Importantly, the high residual error
20 GSD for blood DCVG reflects the fact that only grouped or unmatched individual data were
21 available, so in this case, the residual error includes interindividual variability, which is not
22 included in the other residual error estimates. For the other indirectly estimated pathways, there
23 remain a large range of possible values that are nonetheless consistent with all the available in
24 vivo data. The use of noninformative priors for the metabolism parameters for which there were
25 no in vitro data means that a fuller characterization of the uncertainty in these various metabolic
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-111 DRAFT—DO NOT CITE OR QUOTE
-------
1 pathways could be achieved. Thus, as with the rat, the model should be reliable for estimating
2 lower and upper bounds on several of these pathways.
O
o
^D
.a
c
5
O
in
o _
o
o
\
0
\
2
\
4
Time fh)
6
8
4 Figure 3-10. Comparison of DCVG concentrations in human blood and
5 predictions from the updated model. Data are mean concentrations for males
6 (A) and females (o) reported in Lash et al. (1999b) for humans exposed for
7 4 hours to 100 ppm TCE in air (thick horizontal line denotes the exposure period).
8 Data for oxidative metabolites from the same individuals were reported in Fisher
9 et al. (1998) but could not be matched with the individual DCVG data (Lash
10 2007, personal communication). The vertical error bars are standard errors of the
11 mean as reported in Lash et al. (1999b) (n = 8, so standard deviation is 80.5-fold
12 larger). Lines are PBPK model predictions for individual male (solid) and female
13 (dashed) subjects. Parameter values used for each prediction are a random sample
14 from the individual-specific parameters from the human MCMC chains (the last
15 iteration of the 1st chain was used). See files linked to Appendix A for
16 comparisons with the full distribution of predictions.
17
18
19 3.5.6.4. Summary Evaluation of Updated Physiologically Based Pharmacokinetic (PBPK)
20 Model
21 Overall, the updated PBPK model, utilizing parameters consistent with the available
22 physiological and in vitro data from published literature, provides reasonable fits to an extremely
23 large database of in vivo pharmacokinetic data in mice, rats, and humans. Posterior parameter
24 distributions were obtained by MCMC sampling using a hierarchical Bayesian population
25 statistical model and a large fraction of this in vivo database. Convergence of the MCMC
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-112 DRAFT—DO NOT CITE OR QUOTE
-------
1 samples for model parameters was good for mice, and adequate for rats and humans. In addition,
2 in rats and humans, the model produced predications that are consistent with in vivo data from
3 many studies not used for calibration (insufficient studies were available in mice for such "out of
4 sample" evaluation).
5
6 3.5.7. Physiologically Based Pharmacokinetic (PBPK) Model Dose Metric Predictions
7 3.5.7.1. Characterization of Uncertainty and Variability
8 Since it is desirable to characterize the contributions from both uncertainty in population
9 parameters and variability within the population, so the following procedure is adopted. First,
10 500 sets of population parameters (i.e., population mean and variance for each parameter) are
11 extracted from the posterior MCMC samples—these represent the uncertainty in the population
12 parameters. To minimize autocorrelation, they were obtained by "thinning" the chains to the
13 appropriate degree. From each of these sets of population parameters, 100 sets of "individual,"
14 or "study group" in the case of rodents, parameters were generated by Monte Carlo—each of
15 these represents the population variability, given & particular set of population parameters. Thus
16 a total of 50,000 individuals (or study groups, for rodents), representing 100 (variability) each for
17 500 different populations (uncertainty), were generated.
18 Each set was run for a variety of generic exposure scenarios. The combined distribution
19 of all 50,000 individuals reflects both uncertainty and variability—i.e., the case in which one is
20 trying to predict the dosimetry for a single random study (for rodents) or individual (for humans).
21 In addition, for each dose metric, the mean predicted internal dose was calculated from set of the
22 500 sets of 100 individuals, resulting in a distribution for the uncertainty in the population mean.
23 Comparing the combined uncertainty and variability distribution with the uncertainty distribution
24 in the population mean gives a sense of how much of the overall variation is due to uncertainty
25 versus variability.
26 Figures 3-11-3-19 show the results of these simulations for a number of representative
27 dose metrics across species continuously exposed via inhalation or orally. For display purposes,
28 dose metrics have been scaled by total intake (resulting in a predicted "fraction" metabolized) or
29 exposure level (resulting in an internal dose per ppm for inhalation or per mg/kg/d for oral
30 exposures). In these figures, the thin error bars representing the 95% confidence interval for
31 overall uncertainty and variability, and the thick error bars representing the 95% confidence
32 interval for the uncertainty in the population mean. The interpretation of these figures is that if
33 the thick error bars are much smaller (or greater) than the think error bars, then variability (or
34 uncertainty) contributes the most to overall uncertainty and variability.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-113 DRAFT—DO NOT CITE OR QUOTE
-------
Fraction Metabolized
B
Fraction Metabolized
1
2
3
4
5
6
7
8
9
10
CO
0
CD
d
(Nj
CD
O
CD
CO
CD
CD
CD
CN
CD
O
CD
0.1 1 10 100 1000
Continuous inhalation ( ppm)
0.1
10
100 1000
Continuous oral ( mg/kg-d )
Figure 3-11. PBPK model predictions for the fraction of intake that is
metabolized under continuous inhalation (A) and oral (B) exposure
conditions in mice (white), rats (diagonal hashing), and humans (horizontal
hashing). Bars and thin error bars represent the median estimate and 95%
confidence interval for a random rodent group or human individual, and reflect
combined uncertainty and variability. Circles and thick error bars represent the
median estimate and 95% confidence interval for the population mean, and reflect
uncertainty only.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-114 DRAFT—DO NOT CITE OR QUOTE
-------
Fraction Oxidized
B
Fraction Oxidized
1
2
3
4
5
6
7
8
9
10
CO
0
CD
d
(Nj
CD
O
CD
CO
CD
CD
CD
CN
CD
O
CD
0.1 1 10 100 1000
Continuous inhalation ( ppm)
0.1
10
100 1000
Continuous oral ( mg/kg-d )
Figure 3-12. PBPK model predictions for the fraction of intake that is
metabolized by oxidation (in the liver and lung) under continuous inhalation
(A) and oral (B) exposure conditions in mice (white), rats (diagonal hashing),
and humans (horizontal hashing). Bars and thin error bars represent the median
estimate and 95% confidence interval for a random rodent group or human
individual, and reflect combined uncertainty and variability. Circles and thick
error bars represent the median estimate and 95% confidence interval for the
population mean, and reflect uncertainty only.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-115 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
o -
o -=
o -
o -
o -1
Fraction Conjugated
* :--'
MRH IVRH IVRH IVRH IVRH
I
,-1
\
\
\
o -
o -=
o —
o -
o —'
3
H
H •
IVF
Fr
«
RH IVF
actionC
H '
RH IVF
Conjugate
* H
"fZrr-fM^
RH ^
*
/IR
;d
^
H
'*''*'*'"
H ^
H
^
V •
JR\-
10 1 10 10 10
Continuous inhalation (ppm)
10 1 10 10 10
Continuous oral (mg/kg-d)
Figure 3-13. PBPK model predictions for the fraction of intake that is
metabolized by GSH conjugation (in the liver and kidney) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats
(dashed line), and humans (solid line). X-values are slightly offset for clarity
Open circles (connected by lines) and thin error bars represent the median
estimate and 95% confidence interval for a random rodent group or human
individual, and reflect combined uncertainty and variability. Filled circles and
thick error bars represent the median estimate and 95% confidence interval for the
population mean, and reflect uncertainty only.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-116 DRAFT—DO NOT CITE OR QUOTE
-------
A Fraction bioactivated in kidney
1
2
3
4
5
6
7
o -
o -
o -
o -
o -1
RH
RH
RH
RH
RH
10 1 10 10 10
Continuous inhalation (ppm)
o -
o —
o —
o -
o —'
3 Fraction bioactivated in kidney
RH
RH
RH
RH
RH
102 103
10 1 10
Continuous oral (mg/kg-d)
Figure 3-14. PBPK model predictions for the fraction of intake that is
bioactivated DCVC in the kidney under continuous inhalation (A) and oral
(B) exposure conditions in rats (dashed line) and humans (solid line).
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% confidence interval for a
random rodent group or human individual, and reflect combined uncertainty and
variability. Filled circles and thick error bars represent the median estimate and
95% confidence interval for the population mean, and reflect uncertainty only.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-117 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
A Fraction lung oxidation
Ti
o -=
o -
o -
O -
o -=
o -1
v
MRH MRH MRH MRH MRH
I
,-1
T
1
o —
o -
o —
o -
o —
o —'
Fraction lung oxidation
IVRH IVRH IVRH MRH IVRH
r
\
\
\
\
10 1 10 10 10
Continuous inhalation (ppm)
10 1 10 10 10
Continuous oral (mg/kg-d)
Figure 3-15. PBPK model predictions for fraction of intake that is oxidized
in the respiratory tract under continuous inhalation (A) and oral (B)
exposure conditions in mice (dotted line), rats (dashed line), and humans
(solid line). X-values are slightly offset for clarity. Open circles (connected by
lines) and thin error bars represent the median estimate and 95% confidence
interval for a random rodent group or human individual, and reflect combined
uncertainty and variability. Filled circles and thick error bars represent the
median estimate and 95% confidence interval for the population mean, and reflect
uncertainty only.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-118 DRAFT—DO NOT CITE OR QUOTE
-------
A Fraction 'other1 liver oxidation
1
2
3
4
5
6
7
8
9
10
o -
o -
o -
o -1
MRH IVRH IVRH IVRH IVRH
I
,-1
\
\
\
\
o —
o —
o —
o —'
3 Fraction 'other' liver oxidation
IVRH IVRH IVRH MRH IVRH
r
\
\
\
\
10 1 10 10 10
Continuous inhalation (ppm)
10 1 10 10 10
Continuous oral (mg/kg-d)
Figure 3-16. PBPK model predictions for the fraction of intake that is
"untracked" oxidation of TCE in the liver under continuous inhalation (A)
and oral (B) exposure conditions in mice (dotted line), rats (dashed line), and
humans (solid line) X-values are slightly offset for clarity. Open circles
(connected by lines) and thin error bars represent the median estimate and 95%
confidence interval for a random rodent group or human individual, and reflect
combined uncertainty and variability. Filled circles and thick error bars represent
the median estimate and 95% confidence interval for the population mean, and
reflect uncertainty only.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-119 DRAFT—DO NOT CITE OR QUOTE
-------
o -
Q.
Q.
0)
Q.
o -
o -1
AUCTCE in blood
per ppm
MRH IVRH IVRH IVRH IVRH
I I I I I
1CT1 1 101 102 103
Continuous inhalation (ppm)
o -
2L ^
5
o —
o —'
AUCTCE in blood
per mg/kg-d
IVRH IVRH IVRH MRH IVRH
r
\
\
\
\
10 1 10 10 10
Continuous oral (mg/kg-d)
4
5
6
7
8
9
10
11
12
Figure 3-17. PBPK model predictions for the weekly AUC of TCE in venous
blood (mg-hour/L-week) per unit exposure (ppm or mg/kg/d) under
continuous inhalation (A) and oral (B) exposure conditions in mice (dotted
line), rats (dashed line), and humans (solid line). X-values are slightly offset
for clarity. Open circles (connected by lines) and thin error bars represent the
median estimate and 95% confidence interval for a random rodent group or
human individual, and reflect combined uncertainty and variability. Filled circles
and thick error bars represent the median estimate and 95% confidence interval
for the population mean, and reflect uncertainty only.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-120 DRAFT—DO NOT CITE OR QUOTE
-------
Q.
Q.
0)
Q.
O -=
O -=
O -
o -=
o -1
AUCTCOH in blood
perppm
O -3
O -=
O -=
£ -o
5
MRH IVRH IVRH IVRH IVRH
I I I I I
1CT1 1 101 102 103
Continuous inhalation (ppm)
o -
o —'
AUCTCOH in blood
per mg/kg-d
IVRH IVRH IVRH MRH IVRH
r
\
\
\
\
10 1 10 10 10
Continuous oral (mg/kg-d)
4
5
6
7
8
9
10
11
12
Figure 3-18 PBPK model predictions for the weekly AUC of TCOH in blood
(mg-hour/L-week) per unit exposure (ppm or mg/kg/d) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats
(dashed line), and humans (solid line). X-values are slightly offset for clarity
Open circles (connected by lines) and thin error bars represent the median
estimate and 95% confidence interval for a random rodent group or human
individual, and reflect combined uncertainty and variability. Filled circles and
thick error bars represent the median estimate and 95% confidence interval for the
population mean, and reflect uncertainty only.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-121 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
O)
o -
O -
o -.
o -1
AUCTCA in liver
A per ppm ^0
H
H
-1 •
H
H ' '
H
^ '..
[<:;
i '.
S
IVRH IVRH IVRH IVRH IVF
*¥•
mg-h/kg-wk per mg/kg-d
19?
10 10 10
1 1 1 1 Illll 1 III Illll 1 1 1 III III 1 1
?H T '
AUCTCA in liver
3 per mg/kg-d
•i
H
"
~~--~.
'Si
: 4 i
t _ ^V
"* "• ^ H
IVRH IVRH IVRH MRH IVF
3 -
JH
10"1 1 101 102 103
Continuous inhalation (ppm)
10 1 10 10 10
Continuous oral (mg/kg-d)
Figure 3-19 PBPK model predictions for the weekly AUC of TCA in the
liver (mg-hour/L-week) per unit exposure (ppm or mg/kg/d) under
continuous inhalation (A) and oral (B) exposure conditions in mice (dotted
line), rats (dashed line), and humans (solid line). X-values are slightly offset
for clarity. Open circles (connected by lines) and thin error bars represent the
median estimate and 95% confidence interval for a random rodent group or
human individual, and reflect combined uncertainty and variability. Filled circles
and thick error bars represent the median estimate and 95% confidence interval
for the population mean, and reflect uncertainty only.
For application to human health risk assessment, the uncertainty in and variability among
rodent internal dose estimates both contribute to uncertainty in human risk estimates. Therefore,
it is appropriate to combine uncertainty and variability when applying rodent dose metric
predictions to quantitative risk assessment. The median and 95% confidence interval for each
dose metric at some representative exposures in rodents are given in Tables 3-47-3-48, and the
confidence interval in these tables includes both uncertainty in the population mean and variance
as well as variability in the population. On the other hand, for use in predicting human risk, it is
often necessary to separate, to the extent possible, interindividual variability from uncertainty,
and this disaggregation is summarized in Table 3-49.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-122 DRAFT—DO NOT CITE OR QUOTE
-------
Table 3-47. Posterior predictions for representative internal doses: mouse
to
vo £;•
I
I
§
***.
£3'
1
TO'
-2
•••2
i
Dose metric
ABioactDCVCBW
34
ABioactDCVCKid
AMetGSHBW34
AMetLivl BW34
AMetLivOtherBW
34
AMetLivOtherLiv
AMetl_ngBW34
AMetLngResp
AUCCBId
AUCCTCOH
AUCLivTCA
TotMetabBW34
TotOxMetabBW34
TotTCAInBW
Posterior predictions for mouse dose metrics: median (2.5%, 97.5%)
100ppm, 7h/d, 5d/wk
0.304(0.000534, 12.4)
43.7(0.0774, 1780)
0.684(0.0307, 17.6)
170(61.2,403)
3.81 (0.372, 38.4)
196(19,2,070)
187(7.75,692)
638,000
(26,500,2,510,000)
96.9(45,211)
87.9 (9.9, 590)
1,880(444,7,190)
377(140,917)
375(139,916)
272 (88.9, 734)
600 ppm, 7 h/d, 5 d/wk
2.35 (0.00603, 37)
336(0.801,5,240)
5.15(0.285,44.9)
878 (342, 2,030)
20(1.86, 192)
1,030(96.5, 10,100)
263(10.9,2,240)
918,000
(36,800, 7,980,000)
822 (356, 2,040)
480(42.1,4,140)
5,070(1,310, 18,600)
1 ,260 (475, 3,480)
1,250(451,3,450)
729(267, 1,950)
300 mg/kg/d, 5 d/wk
0.676(0.00193, 18.4)
96.8(0.281,2,550)
1.66(0.0718,24.5)
400(125,610)
8.38 (0.773, 80.1)
437(39.5,4,180)
38.5(3.49, 147)
134,000
(12,500,514,000)
110(6.95,411)
132(14.4,670)
2,260 (520, 8,750)
472(165,617)
465(161,616)
334(106,875)
1,000 mg/kg/d, 5 d/wk
2.81 (0.0086, 42.4)
393(1.23,6,170)
6.37 (0.567, 49.4)
874(233, 1,960)
20(1.55,202)
1,020(82.1, 10,400)
127(8.59,484)
433,000
(30,200, 1,690,000)
592(56, 1,910)
389 (34, 2,600)
4,660(939, 18,900)
1,110(303,2,010)
1,100(294,2,010)
694(185, 1,910)
Units
mg/wk-kg374
mg/wk-kg tissue
mg/wk-kg374
mg/wk-kg374
mg/wk-kg374
mg/wk-kg tissue
mg/wk-kg374
mg/wk-kg tissue
mg-h/L-wk
mg-h/L-wk
mg-h/L-wk
mg/wk-kg374
mg/wk-kg374
mg/wk-kg
^§
H I
O >
HH Oq
H TO
o
Note: Mouse body weight is assumed to be 0.03 kg. Predictions are weekly averages over 10 weeks of the specified exposure protocol. Confidence interval
reflects both uncertainties in population parameters (mean, variance) as well as population variability.
H
W
-------
Table 3-48. Posterior predictions for representative internal doses: rat
to
vo £;•
I
I
§
***.
£3'
1
TO'
Dose metric
ABioactDCVCBW
34
ABioactDCVCKid
AMetGSHBW34
AMetLivl BW34
AMetLivOtherBW
34
AMetLivOtherLiv
AMetl_ngBW34
AMetLngResp
AUCCBId
AUCCTCOH
AUCLivTCA
TotMetabBW34
TotOxMetabBW34
TotTCAInBW
Posterior predictions for rat dose metrics: median (2.5%,97.5%)
100ppm, 7h/d, 5d/wk
0.341 (0.0306, 2.71)
67.8(6.03,513)
0.331 (0.0626,2.16)
176(81.1,344)
45.5 (2.52, 203)
1,870(92.1,8,670)
15(0.529, 173)
41,900(1,460,496,000)
86.7 (39.2, 242)
83.6(1.94, 1,560)
587 (53.7, 4,740)
206(103,414)
206(103,414)
31.7(3.92, 174)
600 ppm, 7 h/d, 5 d/wk
2.3(0.175,22.6)
450 (35.4, 4,350)
2.27(0.315, 19.3)
623 (271 , 1 ,270)
160(7.84,749)
6,660(313,31,200)
24.5(0.819,227)
67,900 (2,350, 677,000)
1,160(349,2,450)
446(6, 10,900)
2,030(186, 13,400)
682 (288, 1 ,430)
677 (285, 1 ,430)
110(13.8,490)
300 mg/kg/d, 5 d/wk
2.15(0.17,20.2)
420(31.6,3,890)
2.13(0.293, 16)
539(176, 1,060)
134(6.83,659)
5,490 (280, 27,400)
15.1 (0.527, 115)
40,800(1,500,325,000)
670(47.8, 1,850)
304(4.71,7,590)
1,730(124, 11,800)
572(199, 1,080)
568(191, 1,080)
90.1 (10.4,417)
1,000 mg/kg/d, 5 d/wk
8.89(0.711,84.1)
1,720(134, 15,800)
8.84(1.35,69.3)
951 (273, 2,780)
238(11.3, 1390)
9,900 (492, 59,600)
32.1 (1.01,311)
85,700 (2,660, 877,000)
3,340 (828, 8,430)
685(8.14,32,500)
3,130(200,21,000)
1,030(302,2,920)
1,010(286,2,910)
164(17.3,800)
Units
mg/wk-kg374
mg/wk-kg tissue
mg/wk-kg374
mg/wk-kg374
mg/wk-kg374
mg/wk-kg tissue
mg/wk-kg374
mg/wk-kg tissue
mg-h/L-wk
mg-h/L-wk
mg-h/L-wk
mg/wk-kg374
mg/wk-kg374
mg/wk-kg
^§
Note: Rat body weight is assumed to be 0.3 kg. Predictions are weekly averages over 10 weeks of the specified exposure protocol. Confidence interval reflects
both uncertainties in population parameters (mean, variance) as well as population variability.
H I
O >
HH Oq
H TO
o
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Table 3-49. Posterior predictions for representative internal doses: human
Dose metric
Posterior predictions for human dose metrics:
2.5% population: median (2.5%, 97.5%)
50% population: median (2.5%, 97.5%)
97.5% population: median (2.5%, 97.5%)
Female
0.001 ppm continuous
Male
0.001 ppm continuous
Female
0.001 mg/kg/d continuous
Male
0.001 mg/kg/d continuous
ABioactDCVCBW
34
0.000256 (6.97e-5, 0.000872)
0.00203 (0.00087, 0.00408)
0.0119(0.00713,0.0177)
0.000254 (6.94e-5, 0.000879)
0.00202(0.000859,0.00413)
0.012(0.00699,0.0182)
0.000197 (6.13e-5, 0.000502)
0.00262(0.0012,0.00539)
0.021 (0.0118,0.0266)
0.0002 (6.24e-5, 0.000505)
0.00271 (0.00125,0.00559)
0.022(0.0124,0.0277)
ABioactDCVCKid
0.02 (0.00549, 0.0709)
0.16(0.0671,0.324)
0.95(0.56, 1.45)
0.0207 (0.00558, 0.0743)
0.163(0.0679,0.342)
0.979(0.563, 1.51)
0.0152(0.0048,0.0384)
0.207 (0.0957, 0.43)
1.68(0.956,2.26)
0.016(0.00493,0.0407)
0.22(0.102,0.459)
1.81 (1.03,2.43)
i
^§
H I
O >
HH Oq
H TO
O
H
W
AMetGSHBW34
0.000159(4.38e-05,
0.000539)
0.00126(0.000536,0.00253)
0.00736(0.00442,0.011)
0.000157 (4.37e-05, 0.00054)
0.00125(0.000528,0.00254)
0.00736(0.00434,0.0112)
0.000121 (3.826-05,
0.000316)
0.00161 (0.000748,0.00331)
0.013(0.00725,0.0164)
0.000123(3.826-05,
0.000323)
0.00167(0.000777,0.00343)
0.0136(0.00759,0.0171)
AMetLivl BW34
0.00161 (0.000619,0.00303)
0.00637(0.00501,0.00799)
0.0157(0.0118,0.0206)
0.00157(0.000608,0.00292)
0.00619(0.00484,0.00779)
0.0152(0.0115,0.02)
0.00465(0.00169,0.0107)
0.0172(0.0153,0.0183)
0.0192(0.019,0.0193)
0.00498(0.00184,0.0112)
0.018(0.0161, 0.0191)
0.02(0.0198,0.0201)
AMetLivOtherBW
34
4.98e-5 (8.59e-6, 0.000222)
0.000671 (0.000134,0.00159)
0.00507 (0.00055, 0.00905)
4.87e-5 (8.33e-6, 0.000214)
0.000652 (0.000129, 0.00153)
0.00491 (0.000531,0.00885)
0.000143 (2.35e-5, 0.000681)
0.00166(0.00035,0.00365)
0.00993(0.00109,0.0153)
0.00015 (2.49e-5, 0.000713)
0.00173(0.000365,0.00382)
0.0103(0.00113,0.0159)
AMetLivOtherLiv
0.000748 (0.000138, 0.00335)
0.0104(0.00225,0.0237)
0.0805(0.00871,0.147)
0.00065(0.000119,0.00288)
0.00898(0.00193,0.0203)
0.0691 (0.00751,0.127)
0.00214(0.000354,0.00979)
0.0253 (0.00564, 0.0543)
0.157(0.0188,0.251)
0.00197(0.00033,0.00907)
0.0234 (0.00526, 0.0503)
0.146(0.0173,0.232)
AMetl_ngBW34
6.9e-6(6.13e-7, 7.99e-5)
0.00122(0.000309,0.0032)
0.0123(0.00563,0.0197)
7.25e-6 (6.44e-7, 8.39e-5)
0.00127(0.000325,0.00329)
0.0124(0.00582,0.0199)
7.54e-8 (6.59e-9, 7.85e-7)
1.51e-5(3.44e-6, 4.6e-5)
0.000396 (0.000104, 0.00097)
7.05e-8(6.1e-9, 7.25e-7)
1.39e-5(3.21e-6, 4.24e-5)
0.000366 (9.54e-5, 0.000906)
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table 3-49. Posterior predictions for representative internal doses: human (continued)
Dose metric
AMetLngResp
AUCCBId
AUCCTCOH
AUCLivTCA
TotMetabBW34
TotOxMetabBW34
Posterior predictions for human dose metrics:
2.5% population: median (2.5%, 97.5%)
50% population: median (2.5%, 97.5%)
97.5% population: median (2.5%, 97.5%)
Female
0.001 ppm continuous
0.0144(0.00116,0.155)
2.44(0.613,6.71)
25.8(12.4,42.3)
0.00151 (0.00122,0.00186)
0.00285(0.00252,0.00315)
0.00444 (0.00404, 0.00496)
0.00313(0.00135,0.00547)
0.0181 (0.0135,0.0241)
0.082(0.0586,0.118)
0.0152(0.00668,0.0284)
0.126(0.0784,0.194)
0.754(0.441, 1.38)
0.0049 (0.00383, 0.00595)
0.0107(0.00893,0.0129)
0.0246(0.0185,0.0326)
0.00273(0.00143,0.00422)
0.00871 (0.0069,0.0111)
0.0224(0.0158,0.0309)
Male
0.001 ppm continuous
0.0146(0.00118,0.157)
2.44(0.621,6.65)
25.3(12.2,41.2)
0.00158(0.00127,0.00191)
0.00295 (0.00262, 0.00326)
0.00456(0.00416,0.00507)
0.00305(0.00134,0.00532)
0.0179(0.0133,0.0238)
0.0812(0.0585, 0.117)
0.0137(0.00598,0.0258)
0.114(0.0704,0.177)
0.699(0.408, 1.3)
0.00482 (0.0038, 0.00585)
0.0105(0.00877,0.0127)
0.0244(0.0183,0.0324)
0.00269(0.00143,0.00415)
0.00857(0.00675,0.011)
0.0222(0.0155,0.0308)
Female
0.001 mg/kg/d continuous
0.00015(1.27e-05,0.00153)
0.0313(0.00725,0.0963)
0.813(0.216,2.13)
4.33e-05 (3.36-05, 6.23e-05)
0.000229(0.000122,
0.000436)
0.00167(0.000766,0.00324)
0.00584(0.00205,0.0122)
0.0333 (0.025, 0.0423)
0.115(0.0872,0.163)
0.029(0.0116,0.0524)
0.227(0.138,0.343)
1.11 (0.661, 1.87)
0.0163(0.0136,0.0181)
0.0191 (0.0188,0.0193)
0.0194(0.0194,0.0194)
0.0049(0.00183,0.0108)
0.0173(0.0154,0.0183)
0.0192(0.019,0.0193)
Male
0.001 mg/kg/d continuous
0.0001 34 (1 .1 5e-05, 0.001 37)
0.0279 (0.00644, 0.086)
0.716(0.189, 1.9)
3.846-05 (2.896-05, 5.61 e-05)
0.000204(0.000109,
0.000391)
0.00153(0.000693,0.00303)
0.00615(0.00213,0.0127)
0.035 (0.0264, 0.0445)
0.122(0.0919,0.172)
0.0279(0.0114,0.0501)
0.219(0.133,0.33)
1.09(0.64, 1.88)
0.0173(0.0147,0.019)
0.0199(0.0196,0.0201)
0.0202 (0.0202, 0.0202)
0.00516(0.00194,0.0114)
0.018(0.0161,0.0191)
0.02(0.0198,0.0201)
^§
H I
O >
HH Oq
H TO
o
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 3-49. Posterior predictions for representative internal doses: human (continued)
Dose metric
TotTCAInBW
Posterior predictions for human dose metrics:
2.5% population: median (2.5%, 97.5%)
50% population: median (2.5%, 97.5%)
97.5% population: median (2.5%, 97.5%)
Female
0.001 ppm continuous
0.000259(0.000121,
0.000422)
0.00154(0.00114,0.00202)
0.00525 (0.00399, 0.00745)
Male
0.001 ppm continuous
0.000246(0.000114,
0.000397)
0.00146(0.00109,0.00193)
0.00499 (0.0038, 0.0071)
Female
0.001 mg/kg/d continuous
0.000501 (0.000189,
0.000882)
0.00286 (0.00222, 0.00357)
0.00659 (0.00579, 0.00724)
Male
0.001 mg/kg/d continuous
0.000506 (0.000192, 0.00089)
0.00289 (0.00222, 0.0036)
0.00662(0.00581,0.00726)
Note: Human body weight is assumed to be 70 kg for males, 60 kg for females. Predictions are weekly averages over 100 weeks of continuous exposure (dose
metric units same as previous tables). Each row represents a different population percentile (2.5, 50, and 97.5%), and the confidence interval in each entry
reflects uncertainty in population parameters (mean, variance).
^§
H I
O ^
HH Oq
H TO
O
H
W
-------
1 3.5.7.2. Implications for the Population Pharmacokinetics of Trichloroethylene (TCE)
2 3.5.7.2.1. Results. The overall uncertainty and variability in key toxicokinetic predictions, as a
3 function of dose and species, is shown in Figures 3-11-3-19. As expected, TCE that is inhaled
4 or ingested is substantially metabolized in all species, predominantly by oxidation
5 (Figures 3-11-3-12). At higher exposures, metabolism becomes saturated and the fraction
6 metabolized declines. Mice on average have a greater capacity to oxidized TCE than rats or
7 humans, and this is reflected in the predictions at the two highest levels for each route. The
8 uncertainty in the predictions for the population means for total and oxidative metabolism is
9 relatively modest, therefore, the wide confidence interval for combined uncertainty and
10 variability largely reflects intergroup (for rodents) or interindividual (for humans) variability. Of
11 particular note is the high variability in oxidative metabolism at low doses in humans, with the
12 95% confidence interval spanning from 0.1-0.7 for inhalation and 0.2-1.0 for ingestion.
13 Predictions of GSH conjugation and renal bioactivation of DCVC are highly uncertain in
14 rodents, spanning more than 1,000-fold in mice and 100-fold in rats (Figures 3-13-3-14). In
15 both mice and rats, the uncertainty in the population mean virtually overlaps with the combined
16 uncertainty and variability, reflecting the lack of GSH-conjugate specific data in mice (the
17 bounds are based on mass balance) and the availability of only urinary NAcDCVC excretion in
18 one study in rats. In humans, however, the blood concentrations of DCVG from Lash et al.
19 (1999b) combined with the urinary NAcDCVC data from Bernauer et al. (1996) were able to
20 better constrain GSH conjugation and bioactivation of DCVC, with 95% confidence intervals on
21 the population mean spanning only 3-fold or so. However, substantial variability is predicted
22 (reflecting variability in the measurements of Lash et al., 1999b), for the error bars for the
23 population mean are substantially smaller than that for overall uncertainty and variability. Of
24 particular note is the prediction of one or two orders of magnitude more GSH conjugation and
25 DCVC bioactivation, on average, in humans than in rats, although importantly, the 95%
26 confidence intervals for the predicted population means do overlap.
27 Predictions for respiratory tract oxidative metabolism were, as expected, greatest in mice,
28 followed by rats and then humans (see Figure 3-15). In addition, due to the "presystemic" nature
29 of the respiratory tract metabolism model as well as the hepatic first-pass effect, substantially
30 more metabolism was predicted from inhalation exposures as compared to oral exposures.
31 Interestingly, the population means appeared to be fairly well constrained despite the lack of
32 direct data, suggesting that overall mass balance is an important constraint for the presystemic
33 respiratory tract metabolism modeled here.
34 Some constraints were also placed on "other" hepatic oxidation—i.e., through a pathway
35 that does not result in chloral formation and subsequent formation of TCA and TCOH
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-128 DRAFT—DO NOT CITE OR QUOTE
-------
1 (Figure 3-16). The 95% confidence interval for overall uncertainty and variability spanned about
2 100-fold, a large fraction of that due to uncertainty in the population mean. Interestingly, a
3 higher rate per kg tissue was predicted for rats than for mice or humans, although importantly,
4 the 95% confidence intervals for the population means overlap among all three species.
5 The AUC of TCE in blood (see Figure 3-17) showed the expected nonlinear behavior
6 with increasing dose, with the nonlinearity more pronounced with oral exposure, as would be
7 expected by hepatic first-pass. Notably, the predicted AUC of TCE in blood from inhalation
8 exposures corresponds closely with cross-species ppm-equivalence, as is often assumed. For low
9 oral exposures (<1 mg/kg-d), cross-species mg/kg-d equivalence appears to be fairly accurate
10 (within 2-fold), implying the usual assumption of mg/kg 4-d equivalence would be somewhat less
11 accurate, at least for humans. Interestingly, the AUC of TCOH in blood (see Figure 3-18) was
12 relatively constant with dose, reflecting the parallel saturation of both TCE oxidation and TCOH
13 glucuronidation. In fact, in humans, the mean AUC for TCOH in blood increases up to 100 ppm
14 or 100 mg/kg/d, due to saturation of TCOH glucuronidation, before decreasing at 1,000 ppm or
15 1,000 mg/kg-d, due to saturation of TCE oxidation.
16 The predictions for the AUC for TCA in the liver showed some interesting features (see
17 Figure 3-19). The predictions for all three species with within an order of magnitude of each
18 other, with a relatively modest uncertainty in the population mean (reflecting the substantial
19 amount of data on TCA). The shape of the curves, however, differs substantially, with humans
20 showing saturation at much lower doses than rodents, especially for oral exposures. In fact, the
21 ratio between the liver TCA AUC and the rate of TCA production, although differing between
22 species, is relatively constant as a function of dose within species (not shown). Therefore, the
23 shape of the curves largely reflect saturation in the production of TCA from TCOH, not in the
24 oxidation of TCE itself, for which saturation is predicted at higher doses, particularly via the oral
25 route (see Figure 3-12). In addition, while for the same exposure (ppm or mg/kg/d TCE) more
26 TCA (on a mg/kg/d basis) is produced in mice relative to rats and humans, humans and rats have
27 longer TCA half-lives even though plasma protein binding of TCA is on average greater.
28
29 3.5.7.2.2. Discussion. This analysis substantially informs four of the major areas of
30 pharmacokinetic uncertainty previously identified in numerous reports (reviewed in Chiu et al.,
31 2006): GSH conjugation pathway, respiratory tract metabolism, alternative pathways of TCE
32 oxidation including DCA formation, and the impact of plasma binding on TCA kinetics
33 particularly in the liver. In addition, the analysis helps identify data that have the potential to
34 further reduce the uncertainties in TCE toxicokinetics and risk assessment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-129 DRAFT—DO NOT CITE OR QUOTE
-------
1 With respect to the first, previous estimates of the degree of TCE GSH conjugation and
2 subsequent bioactivation of DCVC in humans were based on urinary excretion data alone
3 (Bernauer et al., 1996; Birner et al., 1993). For instance, Bloemen et al. (2001) concluded that
4 due to the low yield of identified urinary metabolites through this pathway (<0.05% as compared
5 to 20-30% in urinary metabolites of TCE oxidation), GSH conjugation of TCE is likely of minor
6 importance. However, as noted by Lash et al. (2000a, b), urinary excretion is a poor quantitative
7 marker of flux through the GSH pathway because it only accounts for the portion detoxified, and
8 not the portion bioactivated (a limitation acknowledged by Bloemen et al., 2001). A
9 re-examination of the available in vitro data on GSH conjugation by Chiu et al. (2006) suggested
10 that the difference in flux between TCE oxidation and GSH conjugation may not be as large as
11 suggested by urinary excretion data. For example, the formation rate of DCVG from TCE in
12 freshly isolated hepatocytes was similar in order of magnitude to the rate measured for oxidative
13 metabolites (Lipscomb et al., 1998; Lash et al., 1999a). A closer examination of the only other
14 available human in vivo data on GSH conjugation, the DCVG blood levels reported in Lash et al.
15 (1999b) also suggests a substantially greater flux through this pathway than inferred from urinary
16 data. In particular, the peak DCVG blood levels reported in this study were comparable on a
17 molar basis to peak blood levels of TCOH, the major oxidative metabolite, in the same subjects,
18 as previously reported by Fisher et al. (1998). A lower bound estimate of the GSH conjugation
19 flux can be derived as follows. The reported mean peak blood DCVG concentrations of 46 uM
20 in males exposed to 100 ppm TCE for 4 hrs (Lash et al., 1999b), multiplied by a typical blood
21 volume of 5 1 (ICRP, 2002), yields a peak amount of DCVG in blood of 0.23 mmoles. In
22 comparison, the retained dose from 100 ppm exposure for 4 hours is 4.4 mmol, assuming
23 retention of about 50% (Monster et al., 1976) and minute-volume of 9 L/minute (ICRP, 2002).
24 Thus, in these subjects, about 5% of the retained dose is present in blood as DCVG at the time of
25 peak blood concentration. This is a strong lower bound on the total fraction of retained TCE
26 undergoing GSH conjugation because DCVG clearance is ongoing at the time of peak
27 concentration, and DCVG may be distributed to tissues other than blood. It should be reiterated
28 that only grouped DCVG blood data were available for PBPK model-based analysis; however,
29 this should only result in an underestimation of the degree of variation in GSH conjugation.
30 Finally, this hypothesis of a significant flux through the human GSH conjugation pathway is
31 consistent with the limited available total recovery data in humans in which only 60-70% of the
32 TCE dose is recovered as TCE in breath and excreted urinary metabolites (reviewed in Chiu et
33 al., 2007).
34 Thus, there is already substantial qualitative and semi-quantitative evidence to suggest a
35 substantially greater flux through the GSH conjugation pathway than previously estimated based
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-130 DRAFT—DO NOT CITE OR QUOTE
-------
1 on urinary excretion data alone. The scientific utility of applying a combination of PBPK
2 modeling and Bayesian statistical methods to this question comes from being able to
3 systematically integrate these different types of data—in vitro and in vivo, direct (blood DCVG)
4 and indirect (total recovery, urinary excretion)—and quantitatively assess their consistency and
5 implications. For example, the in vitro data discussed above on GSH conjugation were used for
6 developing prior distributions for GSH conjugation rates, and were not used in previous PBPK
7 models for TCE. Then, both the direct and indirect in vivo data were used to the extent possible
8 either in the Bayesian calibration or model evaluation steps.
9 Several other aspects of the predictions related to GSH conjugation of TCE are worthy of
10 note. Predictions for rats and mice remain more uncertain due to their having less direct
11 toxicokinetic data, but are better constrained by total recovery studies. For instance, the total
12 recovery of 60-70% of dose in exhaled breath and oxidative metabolites in human studies is
13 substantially less than the >90% reported in rodent studies (also noted by Goeptar et al., 1995).
14 In addition, it has been suggested that "saturation" of the oxidative pathway for volatiles in
15 general, and TCE in particular, may lead to marked increases in flux through the GSH
16 conjugation pathway (Slikker et al., 2004a, b; Goeptar et al., 1995), but the PBPK model predicts
17 only a modest, at most ~2-fold, change in flux. This is because there is evidence that both
18 pathways are saturable in the liver for this substrate at similar exposures and because GSH
19 conjugation also occurs in the kidney. Therefore, the available data are not consistent with
20 toxicokinetics alone causing substantially nonlinearites in TCE kidney toxicity or cancer, or in
21 any other effects associated with GSH conjugation of TCE.
22 Finally, the present analysis suggests a number of areas where additional data can further
23 reduce uncertainty in and better characterize the TCE GSH conjugation pathway. The Bayesian
24 analysis predicts a relatively low distribution volume for DCVG in humans, a hypothesis that
25 could be tested experimentally. In addition, corroboration of the DCVG blood levels reported in
26 Lash et al. (1999b) in future studies would further increase confidence in the predictions.
27 Moreover, it would be useful in such studies to be able to match individuals with respect to
28 toxicokinetic data on oxidative and GSH conjugation metabolites so as to better characterize
29 variability. A consistent picture as to which GST isozymes are involved in TCE GSH
30 conjugation, along with data on variability in isozyme polymorphisms and activity levels, can
31 further inform the extent of human variability. In rodents, more direct data on GSH metabolites,
32 such as reliably-determined DCVG blood concentrations, preferably coupled with simultaneous
33 data on oxidative metabolites, would greatly enhance the assessment of GSH conjugation flux in
34 laboratory animals. Given the large apparent variability in humans, data on inter-strain
35 variability in rodents may also be useful.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-131 DRAFT—DO NOT CITE OR QUOTE
-------
1 With respect to oxidative metabolism, as expected, the liver is the major site of oxidative
2 metabolism in all three species, especially after oral exposure, where >85% of total metabolism
3 is oxidation in the liver in all three species. However, after inhalation exposure, the model
4 predicts a greater proportion of metabolism via the respiratory tract than previous models for
5 TCE. This is primarily because previous models for TCE respiratory tract metabolism (Clewell
6 et al., 2000; Hack et al., 2006) were essentially flow-limited—i.e., the amount of respiratory tract
7 metabolism (particularly in mice) was determined primarily by the (relatively small) blood flow
8 to the tracheobronchial region. However, the respiratory tract structure used in the present model
9 is more biologically plausible, is more consistent with that of other volatile organics metabolized
10 in the respiratory tract (e.g., styrene), and leads to a substantially better fit to closed chamber data
11 in mice.
12 Consistent with the qualitative suggestions from in vitro data, the analysis here predicts
13 that mice have a greater rate of respiratory tract oxidative metabolism as compared to rats and
14 humans. However, the predicted difference of 50-fold or so on average between mice and
15 humans is not as great as the 600-fold suggested by previous reports (Green et al., 1997; Green,
16 2000; NRC, 2006). The suggested factor of 600-fold was based on multiplying the Green et al.
17 (1997) data on TCE oxidation in lung microsomes from rats versus mice (23-fold lower) by a
18 factor for the total CYP content of human lung compared to rat lung (27-fold lower) (Wheeler et
19 al. [1990], incorrectly cited as being from Raunio et al. [1998]). However, because of the
20 isozyme-specificity of TCE oxidation, and the differing proportions of different isozymes across
21 species, total CYP content may not be the best measure of inter-species differences in TCE
22 respiratory tract oxidative metabolism. Wheeler et al. (1992) reported that CYP2E1 content of
23 human lung microsomes is about 10-fold lower than that of human liver microsomes. Given that
24 Green et al. (1997) report that TCE oxidation by human liver microsomes is about 3-fold lower
25 than that in mouse lung microsomes, this suggests that the mouse-to-human comparison TCE
26 oxidation in lung microsomes would be about 30-fold. Moreover, the predicted amount of
27 metabolism corresponds to about the detection limit reported by Green et al. (1997) in their
28 experiments with human lung microsomes, suggesting overall consistency in the various results.
29 Therefore, the 50-fold factor predicted by our analysis is biologically plausible given the
30 available in vitro data. More direct in vivo measures of respiratory tract metabolism would be
31 especially beneficial to reduce its uncertainty as well as better characterize its human variability.
32 TCA dosimetry is another uncertainty that was addressed in this analysis. In particular,
33 the predicted inter-species differences in liver TCA AUC are modest, with a range of 10-fold or
34 so across species, due to the combined effects of inter-species differences in the yield of TCA
35 from TCE, plasma protein binding, and elimination half-life. This result is in contrast to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-132 DRAFT—DO NOT CITE OR QUOTE
-------
1 previous analyses which did not include TCA protein binding (Clewell et al., 2000; Fisher,
2 2000), which predicted significantly more than an order of magnitude difference in TCA AUC
3 across species. In addition, in order to be consistent with available data, the model requires some
4 metabolism or other clearance of TCA in addition to urinary excretion. That urinary excretion
5 does not represent 100% of TCA clearance is evident empirically, as urinary recovery after TCA
6 dosing is not complete even in rodents (Abbas et al., 1997; Yu et al., 2000). Additional
7 investigation into possible mechanisms, including metabolism to DCA or enterohepatic
8 recirculation with fecal excretion, would be beneficial to provide a stronger biological basis for
9 this empirical finding.
10 With respect to "untracked" oxidative metabolism, this pathway appears to be a relatively
11 small contribution to total oxidative metabolism. While it is temping to use this pathway as a
12 surrogate for DCA production through from the TCE epoxide (Cai and Guengerich, 1999), one
13 should be reminded that DCA may be formed through multiple pathways (see Section 3.3).
14 Therefore, this pathway at best represents a lower bound on DCA production. In addition, better
15 quantitative markers of oxidative metabolism through the TCE epoxide pathway (e.g.,
16 dichloroacetyl lysine protein adducts, as reported in Forkert et al., 2006) are needed in order to
17 more confidently characterize its flux.
18 In a situation such as TCE in which there is large database of studies coupled with
19 complex toxicokinetics, the Bayesian approach provides a systematic method of simultaneously
20 estimating model parameters and characterizing their uncertainty and variability. While such an
21 approach is not necessarily needed for all applications, such as route-to-route extrapolation (Chiu
22 and White, 2006), as discussed in Barton et al. (2007), characterization of uncertainty and
23 variability is increasingly recognized as important for risk assessment while representing a
24 continuing challenge for both PBPK modelers and users. If there is sufficient reason to
25 characterize uncertainty and variability in a highly transparent and objective manner, there is no
26 reason why our approach could not be applied to other chemicals. However, such an endeavor is
27 clearly not trivial, though the high level of effort for TCE is partially due to the complexity of its
28 metabolism and the extent of its toxicokinetic database.
29 It is notable that, with experience, the methodology for the Bayesian approach to PBPK
30 modeling of TCE has evolved significantly from that of Bois (2000a, 2000b), to Hack et al.
31 (2006), to the present analysis. Part of this evolution has been a more refined specification of the
32 problem being addressed, showing the importance of "problem formulation" in risk assessment
33 applications of PBPK modeling. The particular hierarchical population model for each species
34 was specified based on the intended use of the model predictions, so that relevant data can be
35 selected for analysis (e.g., excluding most grouped human data in favor of individual human
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-133 DRAFT—DO NOT CITE OR QUOTE
-------
1 data) and data can be appropriately grouped (e.g., in rodent data, grouping by sex and strain
2 within a particular study). Thus, the predictions from the population model in rodents are the
3 "average" for a particular "lot" of rodents of a particular species, strain, and sex. This is in
4 contrast to the Hack et al. (2006) model, in which each dose group was treated as a separate
5 "individual." As discussed above, this previous population model structure led to the unlikely
6 result that different dose groups within a closed chamber study had significantly different VMAX
7 values. In humans, however, interindividual variability is of interest, and furthermore,
8 substantial individual data are available in humans. Hack et al. (2006) mixed individual- and
9 group-level data, depending on the availability from the published study, but this approach likely
10 underestimates population variability due to group means being treated as individuals. In
11 addition, in some studies, the same individual was exposed more than once, and in Hack et al.
12 (2006), these were treated as different "individuals." In this case, actual interindividual
13 variability may be either over- or underestimated, depending on the degree of interoccasion
14 variability. While it is technically feasible to include interoccasion variability, it would have
15 added substantially to the computational burden and reduced parameter identifiability. In
16 addition, a primary interest for this risk assessment is chronic exposure, so the predictions from
17 the population model in humans are the "average" across different occasions for a particular
18 individual (adult).
19 The second aspect of this evolution is the drive towards increased objectivity and
20 transparency. For instance, available information, or the lack thereof, is formally codified and
21 explicit either in prior distributions or in the data used to generate posterior distributions, and not
22 both. Methods at minimizing subjectivity (and hence improving reproducibility) in parameter
23 estimation include: (1) clear separation between the in vitro or physiologic data used to develop
24 prior distributions and the in vivo data used to generate posterior distributions; (2) use of
25 noninformative distributions, first updated using a probabilistic model of interspecies-scaling
26 that allows for prediction error, for parameters lacking in prior information; and (3) use of a
27 more comprehensive database of physiologic data, in vitro measurements, and in vivo data for
28 parameter calibration or for out-of-sample evaluation ("validation"). These measures increase
29 the confidence that the approach employed also provides adequate characterization of the
30 uncertainty in metabolic pathways for which available data was sparse or relatively indirect, such
31 as GSH conjugation in rodents and respiratory tract metabolism. Moreover, this approach yields
32 more confident insights into what additional data can reduce these uncertainties than approaches
33 that rely on more subjective methods.
34 Like all analyses, this one has a number of limitations and opportunities for refinement,
35 both biological and statistical. One would be the inclusion of a CH submodel, so that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-134 DRAFT—DO NOT CITE OR QUOTE
-------
1 pharmacokinetic data, such as that recently published by Merdink et al. (2008), could be
2 incorporated. In addition, our probabilistic analysis is still dependent on a model structure
3 substantially informed by deterministic analyses that test alternative model structures (Evans et
4 al., submitted), as probabilistic methods for discrimination or selection among complex,
5 nonlinear models such as that for TCE toxicokinetics have not yet been widely accepted.
6 Therefore, additional refinement of the respiratory tract model may be possible, though more
7 direct in vivo data would likely be necessary to strongly discriminating among models.
8 Furthermore, additional model changes that may be of utility to risk assessment, such as
9 development of models for different lifestages (including childhood and pregnancy), would
10 likely require additional in vivo or in vitro data, particularly as to metabolism, to ensure model
11 identifiability. Finally, improvements are possible in the statistical and population models and
12 analyses, such as incorporation of interoccasion variability (Bernillon and Bois, 2000),
13 application of more sophisticated "validation" methods (such as cross-validation), and more
14 rigorous treatment of grouped data (Chiu and Bois, 2007).
15
16 3.5.7.3. Overall Evaluation of Physiologically Based Pharmacokinetic (PBPK) Model-Based
17 Internal Dose Predictions
18 The utility of the PBPK model developed here for making predictions of internal dose
19 can be evaluated based on four different components: (1) the degree to which the simulations
20 have converged to the true posterior distribution; (2) the degree of overall uncertainty and
21 variability; (3) for humans, the degree of uncertainty in the population; and (4) the degree to
22 which the model predictions are consistent with in vivo data that are informative to a particular
23 dose metric. Table 3-50 summarizes these considerations for each dose metric prediction. Note
24 that this evaluation does not consider in any way the extent to which a dose metric may be the
25 appropriate choice for a particular toxic endpoint.
26 Overall, the least uncertain dose metrics are the fluxes of total metabolism
27 (TotMetabBW34), total oxidative metabolism (TotOxMetabBW34), and hepatic oxidation
28 (AMetLivlBW34). These all have excellent posterior convergence (R diagnostic < 1.01),
29 relatively low uncertainty and variability (GSD < 2), and relatively low uncertainty in human
30 population variability (GSD for population percentiles <2). In addition, the PBPK model
31 predictions compare well with the available in vivo pharmacokinetic data.
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-135 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
Table 3-50. Degree of variance in dose metric predictions due to incomplete convergence (columns 2-4),
combined uncertainty and population variability (columns 5-7), uncertainty in particular human population
percentiles (columns 8-10), model fits to in vivo data (column 11). The GSD is the geometric standard deviation,
which is a "fold-change" from the central tendency.
Dose metric
abbreviation
ABioactDCVCBW
34,
ABioactDCVCKid
AMetGSHBW34
AMetLivl BW34
AMetLivOtherBWS
4,
AMetLivOtherLiv
AMetLngBW34,
AMetLngResp
AUCCBId
AUCCTCOH
AUCLivTCA
TotMetabBW34
TotOxMetabBW34
TotTCAInBW
Convergence: Rfor
generic scenarios
Mouse
—
<1.011
<1.000
<1.004
<1.001
<1.001
<1.001
<1.000
<1.001
<1.001
<1.002
Rat
<1.016
<1.024
<1.003
<1.151
<1.003
<1.004
<1.029
<1.005
<1.004
<1.003
<1.002
Human
<1.015
<1.015
<1.004
<1.012
<1.002
<1.005
<1.002
<1.002
<1.004
<1.004
<1.001
GSD for combined
uncertainty and
variability
Mouse
—
<9.09
<2.02
<3.65
<4.65
<3.04
<3.35
<2.29
<1.92
<1.94
<1.96
Rat
<3.92
<3.28
<1.84
<3.36
<4.91
<3.16
<8.78
<3.18
<1.82
<1.85
<2.69
Human
<3.77
<3.73
<1.97
<3.97
<10.4
<3.32
<5.84
<2.90
<1.81
<1.96
<2.30
GSD for uncertainty in
human population
percentiles
1-5%
<2.08
<2.08
<1.82
<2.63
<4.02
<1.20
<1.73
<1.65
<1.13
<1.77
<1.68
25-75%
<1.64
<1.64
<1.16
<1.92
<2.34
<1.43
<1.20
<1.30
<1.12
<1.15
<1.19
95-99%
<1.30
<1.29
<1.16
<2.05
<1.83
<1.49
<1.23
<1.40
<1.18
<1.20
<1.19
Comments regarding model fits to
in vivo data
Good fits to urinary NAcDCVC, and
blood DCVG.
Good fits to urinary NAcDCVC, and
blood DCVG.
Good fits to oxidative metabolites.
No direct in vivo data.
No direct in vivo data, but good fits to
closed chamber.
Generally good fits, but poor fit to a
few mouse and human studies
Good fits across all three species.
Good fits to rodent data.
Good fits to closed chamber.
Good fits to closed chamber and
oxidative metabolites.
Good fits to TCA data.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 Predictions for TCE in blood (AUCCBld) are somewhat more uncertain. Although
2 convergence was excellent across species (R < 1.01), overall uncertainty and variability was
3 about 3-fold. In humans, the uncertainty in human population variability was relatively low
4 (GSD for population percentiles <1.5). TCE blood level predictions were somewhat high in
5 comparison to the Chiu et al. (2006) study at 1 ppm, though the predictions were better for most
6 of the other studies at higher exposure levels. In mice, TCE blood levels were somewhat over-
7 predicted in open-chamber inhalation studies. In both mice and rats, there were some cases in
8 which fits were inconsistent across dose groups if the same parameters were used across dose
9 groups, indicating unaccounted-for dose-related effects or intrastudy variability. However, in
10 both rats and humans, TCE blood (humans and rats) and tissue (rats only) concentrations from
11 studies not used for calibration (i.e., saved for "out-of-sample" evaluation/'Validation") were
12 well simulated, adding confidence to the parent compound dose metric predictions.
13 For the TCA dose metric predictions (TotTCAInBW, AUCLivTCA) convergence in all
14 three species was excellent (R < 1.01). Overall uncertainty and variability was intermediate
15 between dose metrics for metabolism and that for TCE in blood, with GSD of about 2 to 3-fold.
16 Uncertainty in human population percentiles was relatively low (GSD of 1.2 to 1.7). While liver
17 TCA levels were generally well fit, the data was relatively sparse. Plasma and blood TCA levels
18 were generally well fit, though in mice, there were again some cases in which fits were
19 inconsistent across dose groups if the same parameters were used across dose groups, indicating
20 unaccounted-for dose-related effects or intrastudy variability. In humans, the accurate
21 predictions for TCA blood and urine concentrations from studies used for "out of sample"
22 evaluation lends further confidence to dose metrics involving TCA.
23 The evaluation of TCOH in blood followed a similar pattern. Convergence in all three
24 species was good, though the rat model had slightly worse convergence (R ~ 1.03) than the
25 mouse and humans (R < 1.01). In mice, overall uncertainty and variability was slightly more
26 than for TCE in blood. There much higher overall uncertainly and variability in the rat
27 predictions (GSD of almost 9) that likely reflects true interstudy variability. The
28 population-generated predictions for TCOH and TCOG in blood and urine were quite wide, with
29 some in vivo data both at the upper and lower ends of the range of predictions. In humans, the
30 overall uncertainty and variability was intermediate between mice and rats (GSD = 5.8). As with
31 the rats, this likely reflects true population heterogeneity, as the uncertainty in human population
32 percentiles was relatively low (GSD of around 1.2-1.7-fold). For all three species, fits to in vivo
33 data are generally good. In mice, however, there were again some cases in which fits were
34 inconsistent across dose groups if the same parameters were used across dose groups, indicating
35 unaccounted-for dose-related effects or intrastudy variability. In humans, the accurate
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-137 DRAFT—DO NOT CITE OR QUOTE
-------
1 predictions for TCOH blood and urine concentrations from studies used for "out of sample"
2 evaluation lends further confidence to those dose metrics involving TCOH.
3 GSH metabolism dose metrics (ABioactDCVCBW34, ABioactDCVCKid,
4 AMetGSHBW34) had the greatest overall uncertainty in mice but was fairly well characterized
5 in rats and humans. In mice, there was no in vivo data informing this pathway except for the
6 indirect constraint of overall mass balance. So although convergence was adequate (R<1.02),
7 the uncertainty/variability was very large, with a GSD of 9-fold for the overall flux (the amount
8 of bioactivation was not characterized because there are no data constraining downstream GSH
9 pathways). For rats, there were additional constraints from (well-fit) urinary NAcDCVC data,
10 which reduced the overall uncertainty and variability substantially (GSD < 4-fold). In humans,
11 in addition to urinary NAcDCVC data, DCVG blood concentration data was available, though
12 only at the group level. However, these data, both of which were well fit, in addition to the
13 greater amount of in vitro metabolism data, allowed for the flux through the GSH pathway and
14 the rate of DCVC bioactivation to be fairly well constrained, with overall uncertainty and
15 variability having GSD < 4-fold, and uncertainty in population percentiles no more than about
16 2-fold.
17 The final two dose metrics, respiratory metabolism (AMetLngBW34, AMetLngResp)
18 and "other" oxidative metabolism (AMetLivOtherBW34, AMetLivOtherLiv), also lacked direct
19 in vivo data and were predicted largely on the basis of mass balance and physiological
20 constraints. Respiratory metabolism had good convergence (R < 1.01), helped by the availability
21 of closed chamber data in rodents. In rats and mice, overall uncertainty and variability was
22 rather uncertain (GSD of 4~5-fold), but the overall uncertainty and variability was much greater
23 in humans, with a GSD of about 10-fold. This largely reflects the significant variability across
24 individuals as well as substantial uncertainty in the low population percentiles (GSD of 4-fold).
25 However, the middle (i.e., "typical" individuals) and upper percentiles (i.e., the individuals at
26 highest risk) are fairly well constrained with a GSD of around 2-fold. For the "other" oxidative
27 metabolism dose metric, convergence was good in mice and humans (R < 1.02), but less than
28 ideal in rats (R ~ 1.15). In rodents, the overall uncertainty and variability were moderate, with a
29 GSD around 3.5-fold, slightly higher than that for TCE in blood. The overall uncertainty and
30 variability in this metric in humans had a GSD of about 4-fold, slightly higher than for GSH
31 conjugation metrics. However, uncertainty in the middle and upper population percentiles had
32 GSDs of only about 2-fold, similar to that for respiratory metabolism.
33 Overall, as shown in Table 3-50, the updated PBPK model appears to be most reliable for
34 the fluxes of total, oxidative, and hepatic oxidative metabolism. In addition, dose metrics related
35 to blood levels of TCE and oxidative metabolites TCOH and TCA had only modest uncertainty.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-138 DRAFT—DO NOT CITE OR QUOTE
-------
1 In the case of TCE in blood, for some data sets, model predictions over-predicted the in vivo
2 data, and, in the case of TCOH in rats, substantial interstudy variability was evident. For GSH
3 metabolism, dose-metric predictions for rats and humans had only slightly greater uncertainty
4 than the TCE and metabolism metrics. Predictions for mice were much more uncertain,
5 reflecting the lack of GSD-specific in vivo data. Finally, for "other" oxidative metabolism and
6 respiratory oxidative metabolism, predictions also had somewhat more uncertainty than the TCE
7 and metabolism metrics, though uncertainty in middle and upper human population percentiles
8 was modest.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 3-139 DRAFT—DO NOT CITE OR QUOTE
-------
1 4. HAZARD CHARACTERIZATION
2
3
4 This chapter presents the hazard characterization of trichloroethylene (TCE) health
5 effects. Because of the number of studies and their relevance to multiple endpoints, the
6 evaluation of epidemiologic studies of cancer and TCE is summarized in Section 4.1 (endpoint-
7 specific results are presented in subsequent sections). Genotoxicity data are discussed in
8 Section 4.2. Due to the large number of endpoints and studies in the toxicity database,
9 subsequent sections (see Sections 4.3-4.10) are organized by tissue/organ system. Each section
10 is further organized by noncancer and cancer endpoints, discussing data from human
11 epidemiologic and laboratory experimental studies. In cases where there is adequate
12 information, the role of metabolism in toxicity, comparisons of toxicity between TCE and its
13 metabolites, and carcinogenic mode of action (MOA) are also discussed. Finally, Section 4.11
14 summarizes the overall hazard characterization and the weight of evidence for noncancer and
15 carcinogenic effects.
16
17 4.1. EPIDEMIOLOGIC STUDIES ON CANCER AND TRICHLOROETHYLENE
18 (TCE)—METHODOLOGICAL OVERVIEW
19 This brief overview of the epidemiologic studies on cancer and TCE below provides
20 background to the discussion contained in Sections 4.4-4.10. Over 50 epidemiologic studies on
21 cancer and TCE exposure (see Tables 4-1-4-3) were examined to assess their ability to inform
22 weight of evidence evaluation, i.e., to inform the cancer hazard from TCE exposure, according to
23 15 standards of study design (see Table 4-4), conduct, and analysis. The analysis of
24 epidemiologic studies on cancer and TCE serves to document essential design features, exposure
25 assessment approaches, statistical analyses, and potential sources of confounding and bias. This
26 analysis, furthermore, supports the discussion of site-specific cancer observations in
27 Sections 4.4-4.9. In those sections, study findings are presented with an assessment and
28 discussion of their observations according to a study's weight of evidence and the potential for
29 alternative explanations, including bias and confounding. Tables containing observed findings
30 for site-specific cancers are also found in Sections 4.4-4.9. Full details of the weight-of
31 evidence-review to identify a cancer hazard and study selections for meta-analysis may be found
32 in Appendix B.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-1 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing
cancer and TCE exposure
Reference
Description
Study group (TV)
Comparison group (TV)
Exposure assessment and other information
Aircraft and aerospace workers
Radican et al.
(2008), Blair
etal. (1998)
Civilian aircraft-maintenance
workers with at least 1 yr in
1952-1956 at Hill Air Force Base,
UT. Vital status (VS) to 1990
(Blair et al. 1998) or 2000 (Radican
et al., 2008); cancer incidence
1973-1990 (Blair et al., 1998).
14,457 (7,204 ever exposed to
TCE).
Incidence (Blair et al., 1998) and
mortality rates (Blair et al., 1998;
Radican et al., 2008) of
nonchemical exposed subjects.
Most subjects (n = 10,718) with potential exposure to 1 to 25
solvents. Cumulative TCE assigned to individual subjects using
JEM. Exposure-response patterns assessed using cumulative
exposure, continuous or intermittent exposures, and peak exposure.
TCE replaced in 1968 with 1,1,1-trichloroethane and was
discontinued in 1978 in vapor degreasing activities. Median TCE
exposures were about 10 ppm for rag and bucket; 100-200 ppm
for vapor degreasing. Poisson regression analyses controlled for
age, calendar time, sex (Blair et al., 1998) or Cox proportional hazard
model for age and race.
Krishnadasan
et al. (2007)
Nested case-control study within a
cohort of 7,618 workers employed
for between 1950 and 1992, or who
had started employment before
1980 at Boeing/Rockwell/
Rocketdyne (Santa Susana Field
Laboratory, [the UCLA cohort of
Morgensternetal., 1997]). Cancer
incidence 1988-1999.
326 prostate cancer cases, 1,805
controls.
Response rate:
Cases, 69%; Controls, 60%.
JEM for TCE, hydrazine, PAHs, benzene, mineral oil constructed
from company records, walk-through, or interviews. Lifestyle factors
obtained from living subjects through mail and telephone surveys.
Conditional logistic regression controlled for cohort, age at diagnosis,
physical activity, SES and other occupational exposure (benzene,
PAHs, mineral oil, hydrazine).
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I Co
I
TO
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Zhao et al.
(2005); Ritz
etal. (1999)
Aerospace workers with >2 yrs of
employment at Rockwell/
Rocketdyne (now Boeing) and who
worked at Santa Susana Field
Laboratory, Ventura, CA, from
1950-1993 (the UCLA cohort of
Morgensternetal. [1997]). Cancer
mortality as of December 31, 2001.
Cancer incidence 1988-2000 for
subjects alive as of 1988.
6,044 (2,689 with high cumulative
exposure to TCE). Mortality rates of
subjects in lowest TCE exposure
category.
5,049 (2,227 with high cumulative
exposure to TCE). Incidence rates of
subjects in lowest TCE exposure
category.
JEM for TCE, hydrazine, PAHs, mineral oil, and benzene. IH
ranked each job title ranked for presumptive TCE exposure as high
(3), medium (2), low (1), or no (0) exposure for 3 time periods
(1951-1969, 1970-1979, 1980-1989). Cumulative TCE score: low
(up to 3), medium (over 3 up to 12), high (over 12) assigned to
individual subjects using JEM. Cox proportional hazard, controlled
for time, since 1st employment, SES, age at diagnosis and hydrazine.
Boice et
al.
(2006a)
Aerospace workers with >6 months
employment at Rockwell/
Rocketdyne (Santa Susana Field
Laboratory and nearby facilities)
from 1948-1999 (IEI cohort, IEI
[2005]). VS to 1999.
41,351, 1,642 male hourly test stand
mechanics (1,111 with potential TCE
exposure).
Mortality rates of United States
population and California
population. Internal referent groups
including male hourly
nonadministrative Rocketdyne
workers; male hourly,
nonadministrative SSFL workers;
and test stand mechanics with no
potential exposure to TCE.
Potential TCE exposure assigned to test stands workers only whose
tasks included the cleaning or flushing of rocket engines (engine
flush) (n = 639) or for general utility cleaning (n = 472); potential
for exposure to large quantities of TCE was much greater during
engine flush than when TCE used as a utility solvent. JEM for TCE
and hydrazine without semiquantitative intensity estimates.
Exposure to other solvents not evaluated due to low potential for
confounding (few exposed, low exposure intensity, or not
carcinogenic). Exposure metrics included employment duration,
employment decade, years worked with potential TCE exposure, and
years worked with potential TCE exposure via engine cleaning,
weighted by number of tests. Lifetable (SMR); Cox proportional
hazard controlling for birth year, hire year, and hydrazine exposure.
H I
O >
HH Oq
H TO
O
H
W
Boice et
al. (1999)
Aircraft-manufacturing workers
with at least 1 yr >1960 at
Lockheed Martin (Burbank, CA).
VS to 1996.
77,965 (2,267 with potential routine
TCE exposures and 3,016 with
routine or intermittent TCE
exposure).
Mortality rates of United States
population (routine TCE exposed
subjects) and non-exposed internal
referents (routine and intermittent
TCE exposed subjects).
12% with potential routine mixed solvent exposure and 30% with
route or intermittent solvent exposure. JEM for potential TCE
exposure on (1) routine basis or (2) intermittent or routine basis
without semiquantitative intensity estimate. Exposure-response
patterns assessed by any exposure or duration of exposure and
internal control group. Vapor degreasing with TCE before 1966 and
PCE, afterwards. Lifetable analyses; Poisson regression analysis
adjusting for birth date, starting employment date, finishing
employment date, sex and race.
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Morgan et al.
(1998)
Aerospace workers with >6 months
1950-1985 at Hughes (Tucson,
AZ). VS to 1993.
20,508 (4,733 with TCE exposures).
Mortality rates of United States
population for overall TCE exposure;
mortality rates of all-other cohort
subjects (internal referents).
TCE exposure intensity assigned using JEM. Exposure-response
patterns assessed using cumulative exposure (low versus high) and
job with highest TCE exposure rating (peak, medium/high exposure
versus no/low exposure). "High exposure" job classification
defined as >50 ppm. Vapor degreasing with TCE 1952-1977, but
limited IH data <1975. Limited IH data before 1975 and medium/
low rankings likely misclassified given temporal changes in
exposure intensity not fully considered (NRC, 2006).
Costa et
al. (1989)
Aircraft manufacturing workers
employed 1954-1981at plant in
Italy. VS to 1981.
8,626 subjects
Mortality rates of the Italian
population.
No exposure assessment to TCE and job titles grouped into one of
four categories: blue- and white-collar workers, technical staff, and
administrative clerks. Lifetable (SMR).
Garabrant
etal.
(1988)
Aircraft manufacturing workers >4
yrs employment and who had
worked at least 1 d at San Diego,
CA, plant 1958-1982. VS to 1982.
14,067
Mortality rates of United States
population.
TCE exposure assessment for 70 of 14,067 subjects; 14 cases of
esophageal cancer and 56 matched controls. For these 70 subjects,
company work records identified 37% with job title with potential
TCE exposure without quantitative estimates. Lifetable (SMR).
Cohorts Identified From Biological Monitoring (U-TCA)
Hansen et al.
(2001)
Workers biological monitored using
U-TCA and air-TCE, 1947-1989.
Cancer incidence from 1964-1996.
803 total
Cancer incidence rates of the Danish
population.
712 with U-TCA, 89 with air-TCE measurement records, 2 with
records of both types. U-TCA from 1947-1989; air TCE
measurements from 1974. Historic median exposures estimated
from the U-TCA concentrations were: 9 ppm for 1947 to 1964,
5 ppm for 1965 to 1973, 4 ppm for 1974 to 1979, and 0.7 ppm for
1980 to 1989. Air TCE measurements from 1974 onward were
19 ppm (mean) and 5 ppm (median). Overall, median TCE
exposure to cohort as extrapolated from air TCE and U-TCA
measurements was 4 ppm (arithmetic mean, 12 ppm). Exposure
metrics: year 1st employed, employment duration, mean exposure,
cumulative exposure. Exposure metrics: employment duration,
average TCE intensity, cumulative TCE, period 1st employment.
Lifetable analysis (SIR).
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Anttila et al.
(1995)
Axelson et al.
(1994)
Description
Workers biological monitored using
U-TCA, 1965-1982. VS 1965-
1991 and cancer incidence 1967-
1992.
Workers biological monitored using
U-TCA, 1955-1975. VS to 1986
and cancer incidence 1958-1987.
Study group (TV)
Comparison group (TV)
3,974 total (3,089 with U-TCA
measurements).
Mortality and cancer incidence rates
of the Finnish population.
1,4,21 males
Mortality and cancer incidence rates
of Swedish male population.
Exposure assessment and other information
Median U-TCA, 63 umol/L for females and 48 umol/L for males;
mean U-TCA was 100 umol/L. Average 2.5 U-TCA measurements
per individual. Using the Ikeda et al. (1972) relationship for TCE
exposure to U-TCA, TCE exposures were roughly 4 ppm
(median) and 6 ppm (mean). Exposure metrics: years since
1st measurement. Lifetable analysis (SMR, SIR).
Biological monitoring for U-TCA from 1955 and 1975. Roughly %
of cohort had U-TCA concentrations equivalent to <20 ppm
TCE. Exposure metrics: duration exposure, mean U-TCA.
Lifetable analysis (SMR, SIR).
Other Cohorts
Clapp and
Hoffman
(2008)
Sung et al.
(2007, 2008)
Chang et al.
(2005),
Chang et al.
(2003)
Deaths between 1969-2001 among
employees >5 yrs employment
duration at an IBM facility
(Endicott, NY).
Female workers 1st employed 1973-
1997 at an electronics (RCA)
manufacturing factory (Taoyuan,
Taiwan). Cancer incidence 1979-
2001 (Sung et al., 2007).
Childhood leukemia 1979-2001
among first born of female subjects
in Sung et al. (2007, 2008).
Male and female workers employed
1978-1997 at electronics factory as
studied by Sung et al. (2007). VS
from 1985-1997 and cancer
incidence 1979-1997.
360 deaths
Proportion of deaths among New
York residents during 1979 to 1998.
63,982 females and 40,647 females
with 1st live born offspring.
Cancer incidence rates of Taiwan
population (Sung et al., 2007).
Childhood leukemia incidence rates
of first born live births of Taiwan
population (Sung et al., 2007).
86,868 total
Incidence (Chang et al., 2005) or
mortality (Chang et al., 2003) rates
Taiwan population.
No exposure assessment to TCE. PMR analysis.
No exposure assessment. Chlorinated solvents including TCE and
PCE found in soil and groundwater at factory site. Company records
indicated TCE not used 1975-1991 and PCE 1975-1991 and PCE
after 1981. No information for other time periods. Exposure-
response using employment duration. Lifetable analysis (SMR, SIR)
(Chang et al., 2003, 2005; Sung et al., 2007) or Poisson regression
adjusting for maternal age, education, sex, and birth year (Sung et
al., 2008).
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
ATSDR
(2004)
Workers 1952-1980 at the View-
Master factory (Beaverton, OR).
616 deaths 1989-2001
Proportion of deaths between
1989-2001 in Oregon population.
No exposure information on individual subjects. TCE and other
VOCs detected in well water at the time of the plant closure in 1998
were TCE, 1,220-1,670 ug/L; 1,1-DCE, up to 33 ug/L; and, PCE up
to 56 ug/L. PMR analysis.
Raaschou-
Nielsen et al.
(2003)
Blue-collar workers employed
>1968 at 347 Danish TCE-using
companies. Cancer incidence
through 1997.
40,049 total (14,360 with presumably
higher level exposure to TCE).
Cancer incidence rates of the Danish
population.
Employers had documented TCE usage. Blue-collar versus white-
collar workers and companies with <200 workers were variables
identified as increasing the likelihood for TCE exposure. Subjects
from iron and metal, electronics, painting, printing, chemical, and
dry cleaning industries. Median exposures to trichloroethylene
were 40-60 ppm for the years before 1970,10-20 ppm for 1970
to 1979, and approximately 4 ppm for 1980 to 1989. Exposure
metrics: employment duration, year 1st employed, and # employees
in company. Lifetable (SIR).
Ritz (1999a)
Male uranium-processing plant
workers >3 months employment
1951-1972 at DOE facility
(Fernald, OH). VS 1951-1989,
cancer.
3,814 white males monitored for
radiation (2,971 with potential TCE
exposure).
Mortality rates of the United States
population; Non-TCE exposed
internal controls for TCE exposure-
response analyses.
JEM for TCE, cutting fluids, kerosene, and radiation generated by
employees and industrial hygienists. Subjects assigned potential
TCE according to intensity: light (2,792 subjects), moderate
(179 subjects), heavy (no subjects). Lifetable (SMR) and
conditional logistic regression adjusted for pay status, date first hire,
radiation.
Henschler et
al. (1995)
Male workers > 1 yr 1956-1975 at
cardboard factory (Arnsberg region,
Germany). VS to 1992.
169 exposed; 190 unexposed
Mortality rates from German
Democratic Republic (broad
categories) or renal cell carcinoma
incidence rates from Danish
population, German Democratic, or
non-TCE exposed subjects.
Walk-through surveys and employee interviews used to identify
work areas with TCE exposure. TCE exposure assigned to renal
cancer cases using workman's compensation files. Lifetable (SMR,
SIR) or Mantel-Haenszel.
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Greenland et
al. (1994)
Cancer deaths, 1969-1984, among
pensioned workers employed
<1984 at GE transformer
manufacturing plant (Pittsfield,
MA), and who had job history
record; controls were noncancer
deaths among pensioned workers.
512 cases, 1,202 controls.
Response rate:
Cases, 69%;
Controls, 60%.
Industrial hygienist assessment from interviews and position
descriptions. TCE (no/any exposure) assigned to individual subjects
using JEM. Logistic regression.
Sinks et al.
(1992)
Workers employed 1957-1980 at a
paperboard container
manufacturing and printing plant
(Newnan, GA). VS to 1988.
Kidney and bladder cancer
incidence through 1990.
2,050 total
Mortality rates of the United States
population, bladder and kidney
cancer incidence rates from the
Atlanta-SEER registry for the years
1973-1977.
No exposure assessment to TCE; analyses of all plant employees
including white- and blue-collar employees. Assignment of work
department in case-control study based upon work history; Material
Safety Data Sheets identified chemical usage by department.
Lifetable (SMR, SIR) or conditional logistic regression adjusted for
hire date and age at hire, and using 5- and 10-year lagged
employment duration.
Blair et al.
(1989)
Workers employed 1942-1970 in
U.S. Coast. VS to 1980.
3,781 males of whom 1,767 were
marine inspectors (48%).
Mortality rates of the United States
population. Mortality rates of marine
inspectors also compared to that of
noninspectors.
No exposure assessment to TCE. Marine inspectors worked in
confined spaces and had exposure potential to multiple chemicals.
TCE was identified as one of 10 potential chemical exposures.
Lifetable (SMR) and directly adjusted relative risks.
Shannon et
al. (1988)
Workers employed >6 mos at GE
lamp manufacturing plant,
1960-1975. Cancer incidence from
1964—1982.
1,870 males and females, 249 (13%)
in coiling and wire-drawing area.
Cancer incidence rates from Ontario
Cancer Registry.
No exposure assessment to TCE. Workers in CWD had potential
exposure to many chemicals including metals and solvents. A 1955-
dated engineering instruction sheet identified trichloroethylene used
as degreasing solvent in CWD. Lifetable (SMR).
Shindell and
Ulrich(1985)
Workers employed >3 months at a
TCE manufacturing plant 1957-
1983. VSto 1983.
2,646 males and females
Mortality rates of the United States
population.
No exposure assessment to TCE; job titles categorized as either
white- or blue-collar. Lifetable analysis (SMR).
-------
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio (PMR) studies assessing cancer and
TCE exposure (continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Wilcosky et
al. (1984)
Respiratory, stomach, prostate,
lymphosarcoma, and lymphatic
leukemia cancer deaths 1964-1972
among 6,678 active and retired
production workers at a rubber
plant (Akron, OH); controls were a
20% age-stratified random sample
of the cohort.
183 cases (101 respiratory,
33 prostate, 30 stomach, 9
lymphosarcoma and 10 lymphatic
leukemia cancer deaths).
JEM without quantitative intensity estimates for 20 exposures
including TCE. Exposure metric: ever held job with potential TCE
exposure.
DCE = dichloroethylene, CWD = coiling and wire drawing; DOE = U.S. Department of Energy, IEI = International Epidemiology Institute, JEM = job-exposure
matrix, NRC = National Research Council, PCE = perchloroethylene, PMR = proportionate mortality ratio, SIR = standardized incidence ratio, SMR =
standardized mortality ratio, SSFL = Santa Susanna Field Laboratory, U-TCA = urinary trichloroacetic acid, UCLA = University of California, Los Angeles,
VOCs = volatile organic compounds, VS = vital status.
-------
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure
Reference
Population
Study group (TV)
Comparison group (N)
Response rates
Exposure assessment and other information
Bladder
Pesch et al.
(2000a)
Siemiatycki et
al. (1994),
Siemiatycki
(1991)
Histologically confirmed
urothelial cancer (bladder, ureter,
renal pelvis) cases from German
hospitals (5 regions) in
1991-1995; controls randomly
selected from residency registries
matched on region, sex, and age.
Male bladder cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and random digit
dialing (ROD).
l,035cases
4,298 controls
Cases, 84%; Controls, 71%
484 cases
533 population controls;
740 other cancer controls
Cases, 78%; Controls, 72%
Occupational history using job title or self-reported exposure. JEM and ITEM to
assign exposure potential to metals and solvents (chlorinated solvents, TCE, PCE).
Lifetime exposure to TCE exposure examined as 30th, 60th, and 90th percentiles
(medium, high, and substantial) of exposed control exposure index. Duration used
to examine occupational title and job task duties and defined as 30th, 60th, and 90th
percentiles (medium, long, and very long) of exposed control durations.
Logistic regression with covariates for age, study center, and smoking.
JEM to assign 294 exposures including TCE on semiquantitative scales categorized
as any or substantial exposure. Other exposure metrics included exposure duration
in occupation or job title.
Logistic regression adjusted for age, ethnic origin, socioeconomic status, smoking,
coffee consumption, and respondent status [occupation or job title] orMantel-
Haenszel stratified on age, income, index for cigarette smoking, coffee
consumption, and respondent status (TCE).
Brain
DeRoos et al.
(2001)
Olshan et al.
(1999)
Neuroblastoma cases in children
of <19 yrs selected from
Children's Cancer Group and
Pediatric Oncology Group with
diagnosis in 1992-1994;
population controls (RDD)
matched to control on birth date.
504 cases
504 controls
Cases, 73%; Controls, 74%
Telephone interview with parent using questionnaire to assess parental occupation
and self -reported exposure history and judgment-based attribution of exposure to
chemical classes (halogenated solvents) and specific solvents (TCE). Exposure
metric was any potential exposure.
Logistic regression with covariate for child's age and material race, age, and
education.
-------
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Heineman et
al. (1994)
White, male cases, age >30 yrs,
identified from death certificates
in 1978-1981; controls identified
from death certificates and
matched for age, year of death and
study area.
300 cases
386 controls
Cases, 74%; Controls, 63%
In-person interview with next-of-kin; questionnaire assessing lifetime occupational
history using job title and JEM of Gomez et al. (1994). Cumulative exposure
metric (low, medium or and high) based on weighted probability and duration.
Logistic regression with covariates for age and study area.
Colon and Rectum
Goldberg et al.
(2001),
Siemiatycki
(1991)
Male colon cancer cases, 35-75
yrs, from 16 large.
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and random digit
dialing (ROD).
497 cases
533 population controls and
740 cancer controls
Cases, 82%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales); potential TCE
exposure defined as any or substantial exposure.
Logistic regression adjusted for age, ethnic origin, birthplace, education, income,
parent's occupation, smoking, alcohol consumption, tea consumption, respondent
status, heating source socioeconomic status, smoking, coffee consumption, and
respondent status (occupation, some chemical agents) or Mantel-Haenszel stratified
on age, income, index for cigarette smoking, coffee consumption, and respondent
status (TCE).
Dumas et al.
(2000),
Simeiatycki
(1991)
Male rectal cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
292 cases
533 population controls and
740 other cancer controls
Cases, 78%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales); potential TCE
exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, respondent status, cigarette
smoking, beer consumption and body mass index (TCE) or Mantel-Haenszel
stratified on age, income, index for cigarette smoking, coffee consumption, ethnic
origin, and beer consumption (TCE).
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Fredriksson et
al. (1989)
Population
Colon cancer cases aged 30-75
yrs identified through the Swedish
Cancer Registry among patients
diagnosed in 1980-1983;
population-based controls were
frequency -matched on age and sex
and were randomly selected from
a population register.
Study group (TV)
Comparison group (N)
Response rates
329 cases
658 controls
Not available
Exposure assessment and other information
Mailed questionnaire assessing occupational history with telephone interview
follow-up. Serf-reported exposure to TCE defined as any exposure.
Mantel-Haenszel stratified on age, sex, and physical activity.
Esophagus
Parent et al.
(2000a),
Siemiatycki
(1991)
Male esophageal cancer cases,
35-75 yrs, diagnosed in 19 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
292 cases
533 population controls;
740 subjects with other
cancers
Cases, 78%; controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales); potential TCE
exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, respondent status, cigarette
smoking, beer consumption and body mass index (solvents) or Mantel-Haenszel
stratified on age, income, index for cigarette smoking, coffee consumption, ethnic
origin, and beer consumption (TCE).
Lymphoma
Wang et al.
(2009)
Cases among females aged 21 and
84 yrs with NHL in 1996-2000
and identified from Connecticut
Cancer Registry; population-based
female controls (1) if <65 yrs of
age, having Connecticut address
stratified by 5-yr age groups
identified from random digit
dialing or (2) >65 yrs of age, by
random selection from Centers for
Medicare and Medicaid Service
files.
601 cases
717 controls
Cases, 72%; Controls, 69%
(<65 yrs), 47% (>65 yrs)
In-person interview with using questionnaire assessment specific jobs held for
>1 yr. Intensity and probability of exposure to broad category of organic solvents
and to individual solvents, including TCE, estimated using JEM (Gomez et al,
1994; Dosemeci et al., 1994) and assigned blinded. Exposure metric of any
exposure, exposure intensity (low, medium/high), and exposure probability (low,
medium/high).
Logistic regression adjusted for age, family history of hematopoietic cancer,
alcohol consumption and race.
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Constantini et
al. (2008),
Miligi et al.
(2006)
Cases aged 20-74 with NHL,
including CLL, all forms of
leukemia, or multiple myeloma
(MM) in 1991-1993 and
identified through surveys of
hospital and pathology
departments in study areas and in
specialized hematology centers in
8 areas in Italy; population-based
controls stratified by 5-yr age
groups and by sex selected
through random sampling of
demographic or of National Health
Service files.
1,428 NHL + CLL, 586
Leukemia,
263, MM
1,278 controls (leukemia
analysis)
1,100 controls (MM
analysis)
Cases, 83%; Controls, 73%
In-person interview primarily at interviewee's home (not blinded) using
questionnaire assessing specific jobs, extra occupational exposure to solvents and
pesticides, residential history, and medical history. Occupational exposure
assessed by job-specific or industry-specific questionnaires. JEM used to assign
TCE exposure and assessed using intensity (2 categories) and exposure duration
(2 categories). All NHL diagnoses and 20% sample of all cases confirmed by
panel of 3 pathologists.
Logistic regression with covariates for sex, age, region, and education. Logistic
regression for specific NHL included an additional covariate for smoking.
Seidler et al.
(2007)
Mester et al.
(2006)
Becker et al.
(2004)
NHL and Hodgkin's disease cases
aged 18-80 yrs identified through
all hospitals and ambulatory
physicians in six regions of
Germany between 1998 and 2003;
population controls were
identified from population
registers and matched on age, sex,
and region.
710 cases
710 controls
Cases, 87%; Controls, 44%
n-person interview using questionnaire assessing personal characteristics, lifestyle,
medical history, UV light exposure, and occupational history of all jobs held for
>1 yr. Exposure of a prior interest were assessed using job task-specific
iupplementary questionnaires. JEM used to assign cumulative quantitative TCE
;xposure metric, categorized according to the distribution among the control persons
50th and 90th percentile of the exposed controls).
Conditional logistic regression adjusted for age, sex, region, smoking and alcohol
;onsumption.
H I
O ^
HH Oq
H TO
O
H
W
Persson and
Fredriksson
(1999)
Combined
analysis of
NHL cases in
Persson et al.
(1993),
Persson et al.
(1989)
Histologicallly confirmed cases of
B-cell NHL, age 20-79 yrs,
identified in two hospitals in
Sweden: Oreboro in 1964-1986
(Persson et al., 1989) and in
Linkoping between 1975-1984
(Persson et al., 1993); controls
were identified from previous
studies and were randomly
selected from population registers.
NHL cases, 199
479 controls
Cases, 96% (Oreboro),
90% (Linkoping);
controls, not reported
Mailed questionnaire to assess self reported occupational exposures to TCE and
other solvents.
Unadjusted Mantel-Haenszel chi-square.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Nordstrom et
al. (1998)
Histologically-confirmed cases in
males of hairy-cell leukemia
reported to Swedish Cancer
Registry in 1987-1992 (includes
one case latter identified with an
incorrect diagnosis date);
population-based controls
identified from the National
Population Registry and matched
(1:4 ratio) to cases for age and
county.
Ill cases
400 controls
Cases, 91%; Controls, 83%
Mailed questionnaire to assess self reported working history, specific exposure,
and leisure time activities.
Univariate analysis for chemical-specific exposures (any TCE exposure).
Fritschi and
Siemiatycki,
1996a),
Siemiatycki
(1991)
Male NHL cases, age 35-75 yrs,
diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and ROD.
215 cases
533 population controls
(Group 1) and
1,900 subjects with other
cancers (Group 2)
Cases, 83%; Controls, 71%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales). Exposure metric
defined as any or substantial exposure.
Logistic regression adjusted for age, proxy status, income, and ethnicity (solvents)
or Mantel-Haenszel stratified by age, body mass index, and cigarette smoking
(TCE).
H I
O ^
HH Oq
H TO
O
H
W
Hardell et al.
(1994, 1981)
Histologically-confirmed cases of
NHL in males, age 25-85 yrs,
admitted to Swedish (Umea)
hospital between 1974-1978;
living controls (1:2 ratio) from the
National Population Register,
matched to living cases on sex,
age, and place of residence;
deceased controls from the
National Registry for Causes of
Death, matched (1:2 ratio) to dead
cases on sex, age, place of
residence, and year of death.
105 cases
335 controls
Response rate not available
Self-administered questionnaire assessing serf-reported solvent exposure; phone
follow-up with subject, if necessary.
Unadjusted Mantel-Haenszel chi-square.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Co
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Persson et al.
(1993),
Persson et al.
(1989)
Histologicallly confirmed cases of
Hodgkin's disease, age 20-80 yrs,
identified in two hospitals in
Sweden: Oreboro in 1964-1986
(Persson et al., 1989) and in
Linkoping between 1975-1984
(Persson et al., 1993); controls
randomly selected from
population registers.
54 cases (1989 study);
3 leases (1993 study)
275 controls (1989 study);
204 controls (1993 study)
Response rate not available
Mailed questionnaire to assess self reported occupational exposures to TCE and
other solvents.
Logistic regression with adjustment for age and other exposure; unadjusted
Mantel-Haenszel chi-square.
Childhood Leukemia
Shu et al.
(2004, 1999)
Childhood leukemia cases, <15
yrs, diagnosed between 1989 and
1993 by a Children's Cancer
Group member or affiliated
institute; population controls
(random digit dialing), matched
for age, race, and telephone area
code and exchange.
1,842 cases
1,986 controls
Cases, 92%; controls, 77%
Telephone interview with mother, and whenever available, fathers using
questionnaire to assess occupation using job-industry title and self-reported
exposure history. Questionnaire included questions specific for solvent, degreaser
or cleaning agent exposures.
Logistic regression with adjustment for maternal or paternal education, race, and
family income. Analyses of paternal exposure also included age and sex of the
index child.
H I
O >
HH Oq
H TO
O
H
W
Costas et al.
(2002), MA
DPH (1997)
Childhood leukemia (<19 yrs age)
diagnosed in 1969-1989 and who
were resident of Woburn. MA;
controls randomly selected from
Woburn public School records,
matched for age.
19 cases
37 controls
Cases, 91%; Controls, not
available
Questionnaire administered to parents separately assessing demographic and
lifestyle characteristics, medical history information, environmental and
occupational exposure and use of public drinking water in the home. Hydraulic
mixing model used to infer delivery of TCE and other solvents water to residence.
Logistic regression with composite covariate, a weighted variable of individual
covariates.
McKinney et
al. (1991)
Incident childhood leukemia and
non-Hodgkin's lymphoma cases,
1974-1988, ages not identified,
from three geographical areas in
England; controls randomly
selected from children of residents
in the three areas and matched for
sex and birth health district.
109 cases
206 controls
Cases, 72%; Controls, 77%
In-person interview with questionnaire with mother to assess maternal occupational
exposure history, and with father and mother, as surrogate, to assess paternal
occupational exposure history. No information provided in paper whether
interviewer was blinded as to case and control status.
Matched pair design using logistic regression for univariate and multivariate
analysis.
-------
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Lowengart et
al. (1987)
Population
Childhood leukemia cases aged
<10 yrs and identified from the
Los Angeles (CA) Cancer
Surveillance Program in
1980-1984; controls selected from
RDD or from friends of cases and
matched on age, sex, and race.
Study group (TV)
Comparison group (N)
Response rates
123 cases
123 controls
Cases, 79%; Controls,
not available
Exposure assessment and other information
Telephone interview with questionnaire to assess parental occupational and serf-
reported exposure history.
Matched (discordant) pair analysis.
Melanoma
Fritschi and
Siemiatycki
(1996b),
Siemiatycki
(1991)
Male melanoma cases, age 35-75
yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
103 cases
533 population controls and
533 other cancer controls
Cases, 78%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales); potential TCE
exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, and ethic origin (TCE) or Mantel-
Haenszel stratified on age, income, index for cigarette smoking, and ethnic origin
(TCE).
Pancreas
Kernan et al.
(1999)
Pancreatic cancer deaths from
1984-1993 in 24 U.S. states; age-,
sex-, race-, and state-matched
noncancer deaths, excluding other
pancreatic diseases and
pancreatitis, controls.
63,097 cases
252,386 population controls
Response rates not
identified
Exposure surrogate assigned for 1 1 1 chlorinated hydrocarbons, including TCE, and
2 broad chemical categories using usual occupation on death certificate and job-
exposure-matrix of Gomez et al. (1994).
Race and sex-specific mortality odds ratios from logistic regression analysis
adjusted for age, marital status, metropolitan area, and residential status.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
1
Co
I
TO
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (TV)
Comparison group (N)
Response rates
Exposure assessment and other information
Prostate
Aronson et al.
(1996),
Siemiatycki
(1991)
Male prostate cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
449 cases
533 population controls
(Group 1) and
other cancer cases from
same study (Group 2)
Cases, 81%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales).
Logistic regression adjusted for age, ethnic origin, socioeconomic status, Quetlet,
and respondent status (occupation) or Mantel-Haenszel stratified on age, income,
index for cigarette smoking, ethnic origin, and respondent status (TCE).
Renal Cell
Charbotel et
al. (2006,
2009)
Briining et al.
(2003)
Pesch et al.
(2000b)
Cases from Arve Valley region in
France identified from local
urologists files and from area
teaching hospitals; age- and sex-
matched controls chosen from file
of same urologist as who treated
case or recruited among the
patients of the case's general
practitioner.
Histologically-confirmed cases
1992-2000 from German
hospitals (Arnsberg); hospital
controls (urology department)
serving area, and local geriatric
department, for older controls,
matched by sex and age.
Histologically-confirmed cases
from German hospitals (5 regions)
in 1991-1995; controls randomly
selected from residency registries
matched on region, sex, and age.
87 cases
316 controls
Cases, 74%; controls, 78%
134 cases
401 controls
Cases, 83%; Controls, not
available
935 cases
4,298 controls
Cases, 88%; Controls, 71%
Telephone interview with case or control, or, if deceased, with next-of-kin (22%
cases, 2% controls). Questionnaire assessing occupational history, particularly,
employment in the screw cutting jobs, and medical history. Semiquantitative TCE
exposure assigned to subjects using a task/TCE-Exposure Matrix designed using
information obtained from questionnaires and routine atmospheric monitoring of
work shops or biological monitoring (U-TCA) of workers carried out since the
1960s. Cumulative exposure, cumulative exposure with peaks, and TWA.
Conditional logistic regression with covariates for tobacco smoking and body mass
index.
In-person interviews with case or next-of-kin; questionnaire assessing occupational
history using job title. Exposure metrics included longest job held, JEM of Pannett
et al. (1985) to assign cumulative exposure to TCE and PCE, and exposure
duration..
Logistic regression with covariates for age, sex, and smoking.
In-person interview with case or next-of-kin; questionnaire assessing occupational
history using job title (JEM approach), self-reported exposure, or job task (JTEM
approach) to assign TCE and other exposures.
Logistic regression with covariates for age, study center, and smoking.
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Co
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Parent et al.
(2000b),
Siemiatycki
(1991)
Male renal cell carcinoma cases,
age 35-75 yrs, diagnosed in 16
large Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
142 cases
533 population controls
(Group 1) and
other cancer controls
(excluding lung and bladder
cancers) (Group 2)
Cases, 82%; Controls, 71%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (about 300 exposures on semiquantitative scales); TCE
defined as any or substantial exposure.
Mantel-Haenszel stratified by age, body mass index, and cigarette smoking (TCE)
or logistic regression adjusted for respondent status, age, smoking, and body mass
index (occupation, job title).
Dosemeci et
al. (1999)
Histologically-confirmed cases,
1988-1990, white males and
females, 20-85 yrs, from
Minnesota Cancer Registry;
controls stratified for age and sex
using RDD, 21-64 yrs, or from
HCFA records, 64-85 yrs.
438 cases
687 controls
Cases, 87%; Controls, 86%
In-person interviews with case or next-of-kin; questionnaire assessing occupational
history of TCE using job title and JEM of Gomez et al. (1994). Exposure metric
was any TCE exposure.
Logistic regression with covariates for age, smoking, hypertension, and body mass
index.
Vamvakas et
al. (1998)
Cases who underwent
nephrectormy in 1987-1992 in a
hospital in Arnsberg region of
Germany; controls selected
accident wards from nearby
hospital in 1992.
58 cases
84 controls
Cases, 83%; Controls, 75%
In-person interview with case or next-of-kin; questionnaire assessing occupational
history using job title or self-reported exposure to assign TCE and PCE exposure.
Logistic regression with covariates for age, smoking, body mass index,
hypertension, and diuretic intake.
Multiple or Other Sites
H I
O >
HH Oq
H TO
O
Lee et al.
(2003)
Liver, lung, stomach, colorectal
cancer deaths in males and
females between 1966-1997 from
two villages in Taiwan; controls
were cardiovascular and cerebral-
vascular disease deaths from same
underlying area as cases.
53 liver,
39 stomach,
26 colorectal,
41 lung cancer cases
286 controls
Response rate not reported
Residence as recorded on death certificate.
Mantel-Haenszel stratified by age, sex, and time period.
H
W
-------
Table 4-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Siemiatycki
(1991)
Male cancer cases, 1979-1985,
35-75 yrs, diagnosed in
16 Montreal-area hospitals,
histologically confirmed; cancer
controls identified concurrently;
age-matched, population-based
controls identified from electoral
lists and ROD.
857 lung and
117 pancreatic cancer cases
533 population controls
(Group 1) and other cancer
cases from same study
(Group 2)
Cases, 79% (lung), 71%
(pancreas); Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job titles
and self-reported exposures); analyzed and coded by a team of chemists and
industrial hygienists (294 exposures on semiquantitative scales); TCE defined as
any or substantial exposure.
Mantel-Haenszel stratified on age, income, index for cigarette smoking, ethnic
origin, and respondent status (lung cancer) and age, income, index for cigarette
smoking, and respondent status (pancreatic cancer).
HCFA = Health Care Financing Administration, JEM = job-expo sure matrix, ITEM = job-task-expo sure matrix, NCI = National Cancer Institute, PCE =
perchloroethylene, RDD = random digit dialing, U-TCA = urinary trichloroacetic acid, UV = ultra-violet.
-------
Table 4-3. Geographic-based studies assessing cancer and TCE exposure
to
VO Co*'
I
I
§
***.
£3'
1
TO'
1
Co
I
TO
Reference
Description
Analysis approach
Exposure assessment
Broome County, NY Studies
ATSDR
(2006a, 2008)
Total, 22 site-specific, and
childhood cancer incidence from
1980-2001 among residents in
2 areas in Endicott, NY.
SIR among all subjects (ATSDR, 2006a) or
among white subjects only (ATSDR, 2008) with
expected numbers of cancers derived using age-
specific cancer incidence rates for New York
State, excluding New York City. Limited
assessment of smoking and occupation using
medical and other records in lung and kidney
cancer subjects (ATSDR, 2008).
Two study areas, Eastern and Western study areas, identified
based on potential for soil vapor intrusion exposures as
defined by the extent of likely soil vapor contamination.
Contour lines of modeled VOC soil vapor contamination
levels based on exposure model using GIS mapping and soil
vapor sampling results taken in 2003. The study areas were
defined by 2000 Census block boundaries to conform to
model predicted areas of soil vapor contamination. TCE was
the most commonly found contaminant in indoor air in
Eastern study area at levels ranging from 0.18 to 140 ug/m3 ,
with tetrachloroethylene, cis- 1 ,2-dichloroethene, 1,1,1-
trichloroethane, 1,1-dichloroethylene, 1,1-dichloroethane,
and Freon 113 detected at lower levels. PCE was most
common contaminant in indoor air in Western study area
with other VOCs detected at lower levels.
Maricopa County, AZ Studies
Aickin et al.
(1992) Aickin
(2004)
Cancer deaths, including leukemia,
1966-1986, and childhood (<19 yrs
old) leukemia incident cases
(1965-1986), Maricopa County,
AZ.
Standardized mortality RR from Poisson
regression modeling. Childhood leukemia
incidence data evaluated using Bayes methods
and Poisson regression modeling.
Location of residency in Maricopa County, AZ, at the time
of death as surrogate for exposure. Some analyses examined
residency in West Central Phoenix and cancer. Exposure
information is limited to TCE concentration in two drinking
water wells in 1982.
Pima County, AZ Studies
AZDHS
(1990, 1995)
Cancer incidence in children
(<19 yrs old) and testicular cancer
in 1970-1986 and 1987-1991,
Pima County, AZ.
Standardized incidence RR from Poisson
regression modeling using method of Aickin et al.
(1992). Analysis compares incidence in Tucson
Airport Area to rate for rest of Pima County.
Location of residency in Pima, County, AZ, at the time of
diagnosis or death as surrogate for exposure. Exposure
information is limited to monitoring since 1981 and include
VOCs in soil gas samples (TCE, PCE, 1,1-dichloroethylene,
1,1,1-trichloroacetic acid); PCBs in soil samples, and TCE in
municipal water supply wells.
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-3. Geographic-based studies assessing cancer and TCE exposure (continued)
Reference
Description
Analysis approach
Exposure assessment
Other
Coyle et al.
(2005)
Morgan and
Cassady
(2002)
Vartiainen et
al. (1993)
Conn et al.
(1994)
Fagliano et
al. (1990)
Mallin (1990)
Isacson et al.
(1985)
Incident breast cancer cases among
men and women, 1995-2000,
reported to Texas Cancer Registry.
Incident cancer cases, 1988-1989,
among residents of 13 census tracts
in Redlands area, San Bernardino
County, CA.
Total cancer and site-specific
cancer cases (lymphoma sites and
liver) from 1953-1991 in two
Finnish municipalities.
Incident leukemia and NHL cases,
1979-1987 ,from75 municipalities
and identified from the New Jersey
State Cancer Registry. Histological
type classified using WHO scheme
and the classification of NIH
Working Formulation Group for
grading NHL.
Incident bladder cancer cases and
deaths, 1978-1985, among
residents of 9 NW Illinois counties.
Incident bladder, breast, prostate,
colon, lung and rectal cancer cases
reported to Iowa cancer registry
between 1969-1981.
Correlation study using rank order statistics of
mean average annual breast cancer rate among
women and men and atmospheric release of 12
hazardous air pollutants.
SIR for all cancer sites and 16 site-specific
cancers; expected numbers using incidence rates
of site-specific cancer of a four-county region
between 1988-1992.
SIR with expected number of cancers and site-
specific cancers derived from incidence of the
Finnish population.
Logistic regression modeling adjusted for age.
SIR and SMR by county of residence and zip
code; expected numbers of bladder cancers using
age-race-sex specific incidence rates from SEER
or bladder cancer mortality rates of the United
States population from 1978-1985.
Age-adjusted site-specific cancer incidence in
Iowa towns with populations of 1,000-10,000
and who were serviced by a public drinking
water supply.
Reporting to EPA Toxic Release Inventory the number of
pounds released for 12 hazardous air pollutants, (carbon
tetrachloride, formaldehyde, methylene chloride, styrene,
tetrachloroethylene, trichloroethylene, arsenic, cadmium,
chromium, cobalt, copper, and nickel).
TCE and perchlorate detected in some county wells; no
information on location of wells to residents, distribution of
contaminated water, or TCE exposure potential to individual
residents in studied census tracts.
Monitoring data from 1992 indicated presence of TCE,
tetrachloroethylene and 1,1,1,-trichloroethane in drinking
water supplies in largest towns in municipalities. Residence
in town used to infer exposure to TCE.
Monitoring data from 1984-1985 on TCE, THM, and VOCs
concentrations in public water supplies, and historical
monitoring data conducted in 1978-1984.
Exposure data are lacking for the study population with the
exception of noting one of two zip code areas with observed
elevated bladder cancer rates also had groundwater supplies
contaminated with TCE, PCE and other solvents.
Monitoring data of drinking water at treatment plant in each
Iowa municipality with populations of 1,000-10,000 used to
infer TCE and other volatile organic compound
concentrations in finished drinking water supplies.
to
I
TO
H I
O >
HH Oq
H TO
O
H
W
GIS = geographic information system, NW = Northwestern, PCE = perchloroethylene, RR = rate ratio, SEER = Surveillance, Epidemiology, and End Results,
SIR = standardized incidence ratio, SMR = standardized mortality ratio, VOCs = volatile organic compounds, WHO = World Health Organization.
-------
Table 4-4. Standards of epidemiologic study design and analysis use for identifying cancer hazard and TCE
exposure.
Category A: Study Design
Clear articulation of study objectives or hypothesis. The ideal is a clearly stated hypothesis or study objectives and the study is designed to achieve the identified
objectives.
Selection and characterization in cohort studies of exposure and control groups and of cases and controls (case-control studies) is adequate. The ideal is for
selection of cohort and referents from the same underlying population and differences between these groups to be due to TCE exposure or level of TCE exposure and
not to physiological, health status, or lifestyle factors. Controls or referents are assumed to lack or to have background exposure to TCE. These factors may lead to a
downward bias including one of which is known as "healthy worker bias," often introduced in analyses when mortality or incidence rates from a large population
such as the United States population are used to derive expected numbers of events. The ideal in case-control studies is cases and controls are derived from the same
population and are representative of all cases and controls in that population. Any differences between controls and cases are due to exposure to TCE itself and not
to confounding factors related to both TCE exposure and disease. Additionally, the ideal is for controls to be free of any disease related to TCE exposure. In this
latter case, potential bias is toward the null hypothesis.
Category B: Endpoint Measured
Levels of health outcome assessed. Three levels of health outcomes are considered in assessing the human health risks associated with exposure to TCE:
biomarkers of effects and susceptibility, morbidity, and mortality. Both morbidity as enumerated by incidence and mortality as identified from death certificates are
useful indicators in risk assessment for hazard identification. The ideal is for accurate and predictive indicator of disease. Incidence rates are generally considered to
provide an accurate indication of disease in a population and cancer incidence is generally enumerated with a high degree of accuracy in cancer registries. Death
certifications are readily available and have complete national coverage but diagnostic accuracy is reduced and can vary by specific diagnosis. Furthermore,
diagnostic inaccuracies can contribute to death certificates as a poor surrogate for disease incidence. Incidence, when obtained from population-based cancer
registries, is preferred for identifying cancer hazards.
Changes in diagnostic coding systems for lymphoma, particularly non-Hodgkin's lymphoma. Classification of lymphomas today is based on morphologic,
immunophenotypic, genotypic, and clinical features using the World Health Organization (WHO) classification, introduced in 2001, and incorporation of WHO
terminology into International Classification of Disease (ICD)-0-3. ICD Versions 7 and earlier had rubrics for general types of lymphatic and hematopoietic cancer,
but no categories for distinguishing specific types of cancers, such as acute leukemia. Epidemiologic studies based on causes of deaths as coded using these older
ICD classifications typically grouped together lymphatic neoplasms instead of examining individual types of cancer or specific cell types. Before the use of
immunophenotyping, these grouping of ambiguous diseases such as non-Hodgkin's lymphoma and Hodgkin's lymphoma may be have misclassified. With the
introduction of ICD-10 in 1990, lymphatic tumors coding, starting in 1994 with the introduction of the Revised European-American Lymphoma classification, the
basis of the current WHO classification, was more similar to that presently used. Misclassification of specific types of cancer, if unrelated to exposure, would have
attenuated estimate of relative risk and reduced statistical power to detect associations. When the outcome was mortality, rather than incidence, misclassification
would be greater because of the errors in the coding of underlying causes of death on death certificates (IOM, 2003). Older studies that combined all lymphatic and
hematopoietic neoplasms must be interpreted with care.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-4. Standards of epidemiologic study design and analysis use for identifying cancer hazard and TCE
exposure (continued).
Category C: TCE-Exposure Criteria
Adequate characterization of exposure. The ideal is for TCE exposure potential known for each subject and quantitative assessment [job-exposure-matrix
approach] of TCE exposure assessment for each subject as a function of job title, year exposed, duration, and intensity. The assessment approach is accurate for
assigning TCE intensity [TCE concentration or a time-weighted-average] to individual study subjects and estimates of TCE intensity are validated using monitoring
data from the time period. For the purpose of this report, the objective for cohort and case-controls studies is to differentiate TCE-exposed subjects from subjects
with little or no TCE exposure. A variety of dose metrics may be used to quantify or classify exposures for an epidemiologic study. They include precise summaries
of quantitative exposure, concentrations of biomarkers, cumulative exposure, and simple qualitative assessments of whether exposure occurred (yes or no). Each
method has implicit assumptions and potential problems that may lead to misclassification. Studies in which it was unclear that the study population was actually
exposed to TCE are excluded from analysis.
Category D: Follow-up (Cohort)
Loss to follow-up. The ideal is complete follow-up of all subjects; however, this is not achievable in practice, but it seems reasonable to expect loss to follow-up not
to exceed 10%. The bias from loss to follow-up is indeterminate. Random loss may have less effect than if subjects who are not followed have some significant
characteristics in common.
to
to
Follow-up period allows full latency period for over 50% of the cohort. The ideal to follow all study subjects until death. Short of the ideal, a sufficient follow-up
period to allow for cancer induction period or latency over 15 or 20 yrs is desired for a large percentage of cohort subjects.
Category E: Interview Type (Case-control)
I
TO
Interview approach. The ideal interviewing technique is face-to-face by trained interviewers with more than 90% of interviews with cases and control subjects
conduced face-to-face. The effect on the quality of information from other types of data collection is unclear, but telephone interviews and mail-in questionnaires
probably increase the rate of misclassification of subject information. The bias is toward the null hypothesis if the proportion of interview by type is the same for
case and control, and of indeterminate direction otherwise.
H I
O ^
HH Oq
H TO
O
H
W
Blinded interviewer. The ideal is for the interviewer to be unaware whether the subject is among the cases or controls and the subject to be unaware of the purpose
and intended use of the information collected. Blinding of the interviewer is generally not possible in a face-to-face interview. In face-to-face and telephone
interviews, potential bias may arise from the interviewer expects regarding the relationship between exposure and cancer incidence. The potential for bias from face-
to-face interviews is probably less than with mail-in interviews. Some studies have assigned exposure status in a blinded manner using a job-exposure matrix and
information collected in the unblinded interview. The potential for bias in this situation is probably less with this approach than for nonblinded assignment of
exposure status.
Category F: Proxy Respondents
Proxy respondents. The ideal is for data to be supplied by the subject because the subject generally would be expected to be the most reliable source; less than 10%
of either total cases or total controls for case-control studies. A subject may be either deceased or too ill to participate, however, making the use of proxy responses
unavoidable if those subjects are to be included in the study. The direction and magnitude of bias from use of proxies is unclear, and may be inconsistent across
studies.
-------
Table 4-4. Standards of epidemiologic study design and analysis use for identifying cancer hazard and TCE
exposure (continued).
Category G: Sample Size
The ideal is for the sample size is large enough to provide sufficient statistical power to ensure that any elevation of effect in the exposure group, if present, would be
found, and to ensure that the confidence bounds placed on relative risk estimates can be well characterized.
Category H: Analysis Issues
Control for potentially confounding factors of importance in analysis. The ideal in cohort studies is to derive expected numbers of cases based on age-sex- and
time-specific cancer rates in the referent population and in case-control studies by matching on age and sex in the design and then adjusting for age in the analysis of
data. Age and sex are likely correlated with exposure and are also risk factors for cancer development. Similarly, other factors such as cigarette smoking and
alcohol consumption are risk factors for several site-specific cancers reported as associative with TCE exposure. To be a confounder of TCE, exposure to the other
factor must be correlated, and the association of the factor with the site-specific cancer must be causal. The expect effect from controlling for confounders is to
move the estimated relative risk estimate closer to the true value.
Statistical methods are appropriate. The ideal is that conclusions are drawn from the application of statistical methods that are appropriate to the problem and
accurately interpreted.
Evaluation of exposure-response. The ideal is an examination of a linear exposure-response as assessed with a quantitative exposure metric such as cumulative
exposure. Some studies, absent quantitative exposure metrics, examine exposure response relationships using a semiquantitative exposure metric or by duration of
exposure. A positive dose-response relationship is usually more convincing of an association as causal than a simple excess of disease using TCE dose metric.
However, a number of reasons have been identified for a lack of linear exposure-response finding and the failure to find such a relationship mean little from an
etiological viewpoint.
Documentation of results. The ideal is for analysis observations to be completely and clearly documented and discussed in the published paper, or provided in
supplementary materials accompanying publication.
-------
1 Twenty-three of the studies identified in a systematic review were selected for inclusion
2 in the meta-analysis through use of the following meta-analysis inclusion criteria: (1) cohort or
3 case-control designs; (2) evaluation of incidence or mortality; (3) adequate selection in cohort
4 studies of exposure and control groups and of cases and controls in case-control studies; (4) TCE
5 exposure potential inferred to each subject and quantitative assessment of TCE exposure
6 assessment for each subject by reference to industrial hygiene records indicating a high
7 probability of TCE use, individual biomarkers, job exposure matrices, water distribution models,
8 or obtained from subjects using questionnaire (case-control studies); and (5) relative risk
9 estimates for kidney cancer, liver cancer, or lymphoma adjusted, at minimum, for possible
10 confounding of age, sex, and race (see Table 4-5). This evaluation is summarized below,
11 separately for cohort and case-control studies. Appendix C contains a full discussion of the
12 meta-analysis, its analytical methodology, including sensitivity analyses, and findings.
13 The cohort studies (Wilcosky et al., 1984; Shindell and Ulrich, 1985; Garabrant et al.,
14 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Sinks et al., 1992; Axelson et
15 al., 1994; Greenland et al., 1994; Anttila et al., 1995; Henschler et al., 1995; Ritz, 1999; Blair et
16 al., 1998; Morgan et al., 1998; Boice et al., 1999, 2006; Hansen et al., 2001; Raaschou-Nielsen et
17 al., 2003; Chang et al., 2003; ATSDR, 2004; Chang et al., 2005; Zhao et al., 2005; Krishnadasan
18 et al., 2007; Sung et al., 2007, 2008; Clapp and Hoffman, 2008; Radican et al., 2008) (see
19 Table 4-1), with data on the incidence or morality of site-specific cancer in relation to TCE
20 exposure, range in size (803 [Hansen et al., 2001] to 86,868 [Chang et al., 2003, 2005]), and
21 were conducted in Denmark, Sweden, Finland, Germany, Taiwan, and the United States (see
22 Table 4-1). Three case-control studies nested within cohorts (Wilcosky et al., 1984; Greenland
23 et al., 1994; Krishnadasan et al., 2007) are considered as cohort studies because the summary
24 risk estimate from a nested case-control study, the odds ratio, was estimated from incidence
25 density sampling. This is considered an unbiased estimate of the hazard ratio, similar to a
26 relative risk estimate from a cohort study, if, as is the case for these studies, controls are selected
27 from the same source population as the cases, the sampling rate is independent of exposure
28 status, and the selection probability is proportional to time-at-risk (Rothman et al., 2008). Cohort
29 and nested case-control study designs are analytical epidemiologic studies and are generally
30 relied on for identifying a causal association between human exposure and adverse health effects
31 (U.S. EPA, 2005a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-24 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-5. Summary of criteria for meta-analysis study selection
Decision
outcome
Studies
Primary reason(s)
Studies recommended for meta-analysis:
Siemiatycki, 1991; Axelson et al., 1994;
Hardell, 1994; Greenland etal, 1994;
Anttila et al., 1995; Morgan et al., 1998;
Nordstrom etal., 1998; Boice etal., 1999,
2006a; Dosemeci et al., 1999; Persson and
Fredriksson, 1999; Pesch et al., 2000b;
Hansen et al., 2001; Briining et al., 2003;
Raaschou-Nielsen et al., 2003; Zhao et al.,
2005; Miligi et al., 2006; Seidler et al.,
2007; Charbotel et al., 2006, 2009; Radican
et al., 2008 (Blair et al., 1998-incidence);
Wang etal., 2009
Analytical study designs of cohort or case-control;
Evaluation of incidence or mortality;
Adequate selection in cohort studies of exposure and control
groups and of cases and controls in case-control studies;
TCE exposure potential inferred to each subject and
quantitative assessment of TCE exposure assessment for
each subject by reference to industrial hygiene records
indicating a high probability of TCE use, individual
biomarkers, job exposure matrices, water distribution
models, or obtained from subjects using questionnaire (case-
control studies);
Relative risk estimates for kidney cancer, liver cancer, or
lymphoma adjusted, at minimum, for possible confounding
of age, sex, and race).
Studies not recommended for meta-analysis:
ATSDR, 2004; Clapp and Hoffman, 2008;
Cohnetal., 1994
Wilcosky etal., 1984; Isacson et al., 1985;
Shindell and Ullrich, 1985; Garabrant et al.,
1988; Shannon et al., 1988; Blair et al.,
1989; Costa etal., 1989; AZDHS, 1990,
1995; Mallin, 1990; Aickin et al., 1992;
Sinks et al., 1992; Vartiainen et al., 1993;
Morgan and Cassady, 2002; Lee et al.,
2003; Aickin, 2004; Chang et al., 2003,
2005; Coyle et al., 2005; ATSDR, 2006a,
2008; Sung et al., 2007, 2008
Lowengart et al., 1987; Fredriksson et al.,
1989; McKinney et al., 1991; Heineman et
al., 1994; Siemiatycki et al., 1994; Aronson
et al., 1996; Fritchi and Siemiatycki, 1996b;
Dumas et al., 2000; Kernan et al., 1999; Shu
et al., 1999, 2004; Parent et al., 2000a;
Pesch et al., 2000a; DeRoos et al., 2001;
Goldberg et al., 2001; Costas et al., 2002;
Krishnadasan et al., 2007
Ritz, 1999a
Henschleretal., 1995
Weakness with respect to analytical study design (i.e.,
geographic -based, ecological or proportional mortality ratio
design).
TCE exposure potential not assigned to individual subjects
using job exposure matrix, individual biomarkers, water
distribution models, or industrial hygiene data from other
process indicating a high probability of TCE use (cohort
studies).
Cancer incidence or mortality reported for cancers other than
kidney, liver, or lymphoma.
Subjects monitored for radiation exposure with likelihood for
potential confounding;
Cancer mortality and TCE exposure not reported for kidney
cancer and all hemato- and lymphopoietic cancer reported as
broad category.
Incomplete identification of cohort and index kidney cancer
cases included in case series.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-25 DRAFT—DO NOT CITE OR QUOTE
-------
1 While all of these cohort studies are considered in the overall weight of evidence, ten of
2 them met all five meta-analysis inclusion criteria: the cohorts of Blair et al. (1998) and its
3 follow-up by Radican et al. (2008); Morgan et al. (1998), Boice et al. (1999, 2006), and Zhao et
4 al. (2005), of aerospace workers or aircraft mechanics; and Axelson et al. (1994), Anttila et al.
5 (1995), Hansen et al. (2001), and Raaschou-Nielsen et al. (2003) of Nordic workers in multiple
6 industries with TCE exposure; and Greenland et al. (1994) of electrical manufacturing workers.
7 Subjects or cases and controls in these studies are considered to sufficiently represent the
8 underlying population, and the bias associated with selection of referent populations is
9 considered minimal. The exposure-assessment approaches included detailed job-exposure
10 matrix, biomonitroing data, or use of industrial hygiene data on TCE exposure pattens and
11 factors that affect such exposure, with high probability of TCE exposure potential to individual
12 subjects. The statistical analyses methods were appropriate and well documented, the measured
13 endpoint was an accurate indicator of disease, and the follow-up was sufficient for cancer
14 latency. These studies are also considered as high-quality studies for identifying kidney, liver
15 and lymphoma cancer hazard. The remaining cohort studies less satisfactorily meet identified
16 criteria or standards of epidemiologic design and analysis, having deficiencies in multiple criteria
17 (Wilcosky et al., 1984; Shindell and Ulrich, 1985; Garabrant et al., 1988; Costa et al., 1989;
18 Sinks et al., 1992; Henschler et al., 1995; Ritz, 1999; Chang et al., 2003, 2005; ATSDR, 2004;
19 Sung et al., 2007, 2008; Clapp and Hoffman, 2008). Krishnandansen et al. (2007), who reported
20 on prostate cancer, met four of the five meta-analysis inclusion criteria except that for reporting a
21 relative risk estimate cancer of the kidney, liver or lymphoma, the site-specific cancers examined
22 using meta-analysis.
23 The case-control studies on TCE exposure are of several site-specific cancers, including
24 bladder (Siemiatycki, 1991; Siemiatycki et al., 1994; Pesch et al., 2000a); brain (Heineman et al.,
25 1994; DeRoos et al., 2001); childhood lymphoma or leukemia (Lowengart et al., 1987;
26 McKinney et al., 1991; Shu et al., 1999; 2004; Costas et al., 2002); colon cancer (Siemiatycki,
27 1991; Goldberg et al., 2001); esophageal cancer (Siemiatycki, 1991; Parent et al., 2000a); liver
28 cancer (Lee et al., 2003); lung (Siemiatycki, 1991); adult lymphoma or leukemia (Hardell et al.,
29 1994 [non-Hodgkin's lymphoma (NHL), Hodgkin lymphoma]; leukemia (Siemiatycki, 1991;
30 Fritschi and Siemiatycki, 1996a; Nordstrom et al., 1998 [hairy cell leukemia]; Persson and
31 Fredriksson, 1999 [NHL]; Miligi et al., 2006 [NHL and chronic lymphocytic leukemia (CLL)];
32 Seidler et al., 2007 [NHL, Hodgkin lymphoma]; Costantini et al., 2008 [leukemia types, CLL
33 included with NHL in Miligi et al., 2006]); melanoma (Siemiatycki, 1991; Fritchi and
34 Siemiatycki, 1996b); rectal cancer (Siemiatycki, 1991; Dumas et al., 2000); renal cell carcinoma,
35 a form of kidney cancer (Siemiatycki, 1991; Parent et al. (2000b); Vamvakas et al., 1998;
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-26 DRAFT—DO NOT CITE OR QUOTE
-------
1 Dosemeci et al., 1999; Pesch et al., 2000b; Briining et al., 2003; Charbotel et al., 2006);
2 pancreatic cancer (Siemiatyck, 1991); and prostate cancer (Siemiatycki, 1991; Aronson et al.,
3 1996) (see Table 4-2). No case-control studies of reproductive cancers (breast or cervix) and
4 TCE exposure were found in the peer-reviewed literature.
5 While all of these case-control studies are considered in the overall weight of evidence,
6 thirteen of them met the meta-analysis inclusion criteria identified in Section B.2.9 (Siemiatycki,
7 1991; Hardell et al., 1994; Nordstrom et al., 1998; Dosemeci et al., 1999; Persson and
8 Fredriksson, 1999; Pesch et al., 2000b; Briining et al., 2003; Miligi et al., 2006; Charbotel et al.,
9 2006, 2009; Seidler et al., 2007; Constantini et al., 2008, Wang et al., 2009). They were of
10 analytical study design, cases and controls were considered to represent underlying populations
11 and selected with minimal potential for bias; exposure assessment approaches included
12 assignment of TCE exposure potential to individual subjects using information obtained from
13 face-to-face, mailed, or telephone interviews; analyses methods were appropriate, well-
14 documented, included adjustment for potential confounding exposures, with relative risk
15 estimates and associated confidence intervals reported for kidney cancer, liver cancer or
16 lymphoma.
17 These studies were also considered, to varying degrees, as high-quality studies for
18 weight-of evidence characterization of hazard. Both Briining et al. (2003) and Charbotel et al.
19 (2006, 2009) had a priori hypotheses for examining renal cell carcinoma and TCE exposure.
20 Strengths of both studies are in their examination of populations with potential for high exposure
21 intensity and in areas with high frequency of TCE usage and their assessment of TCE potential.
22 An important feature of the exposure assessment approach of Charbotel et al. (2006) is their use
23 of a large number of studies on biological monitoring of workers in the screw-cutting industry a
24 predominant industry with documented TCE exposures as support. Other studies were either
25 large multiple-center studies (Pesch et al., 2000a, b; Miligi et al., 2006; Constantini et al., 2008;
26 Wang et al., 2009) or reporting from one location of a larger international study (Dosemeci et al.,
27 1999; Seidler et al., 2007). In contrast to Briining et al. (2003) and Charbotel et al. (2006, 2009),
28 two studies conducted in geographical areas with widespread TCE usage and potential for
29 exposure to higher intensity, in these other studies, a lower exposure prevalence to TCE is found
30 (any TCE exposure: 15% of cases [Dosemeci et al.,1999]; 6% of cases [Miligi et al., 2006]; 13%
31 of cases [Seidler et al., 2007]; 13% of cases [Wang et al., 2008]) and most subjects were
32 identified as exposed to TCE probably had minimal contact (3% of cases with moderate/high
33 TCE exposure [Miligi et al., 2006]; 1% of cases with high cumulative TCE [Seidler et al., 2007];
34 2% of cases with high intensity, but of low probability TCE exposure [Wang et al., 2008]). This
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 pattern of lower exposure prevalence and intensity is common to community-based population
2 case-control studies (Teschke et al., 2002).
3 Thirteen case-control studies did not meet specific meta-analysis inclusion criterion
4 (Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and Siemiatycki, 1996b; Dumas et al.,
5 2000; Parent et al., 2000a; Goldberg et al., 2001; Vamvakas et al., 1998; Kernan et al., 1999; Shu
6 et al., 1999, 2004; Pesch et al., 2000a; Costas et al., 2002; Lee et al., 2003). Ten of twelve
7 studies reported relative risk estimates for site-specific cancers other than kidney, liver, and
8 lymphomas (Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and Siemiatycki, 1996b;
9 Kernan et al., 1999; Dumas et al., 2000; Parent et al., 2000a; Pesch et al., 2000a; Goldberg et al.,
10 2001; Shu et al., 1999, 2004; Costas et al., 2002). Vamvakas et al. (1998) has been subject of
11 considerable controversy (Bloemen and Tomenson, 1995; Swaen, 1995; McLaughlin and Blot,
12 1997; Green and Lash, 1999; Cherrie et al., 2001; Mandel, 2001) with questions raised on
13 potential for selection bias related to the study's controls. This study was deficient in the
14 criterion for adequacy of case and control selection. Briining et al. (2003), a study from the same
15 region as Vamvakas et al. (1998), is considered a stronger study for identifying cancer hazard
16 since it addresses many of the deficiencies of Vamvakas et al. (1998). Lee et al. (2003) in their
17 study of hepatocellular cancer assigns one level of exposure to all subjects in a geographic area,
18 and inherent measurement error and misclassification bias because not all subjects are exposed
19 uniformly. Additionally, statistical analyses in this study did not control for hepatitis viral
20 infection, a known risk factor for hepatocellular cancer and of high prevalence in the study area.
21 The geographic-based studies (Isacson et al., 1985; AZ DHS, 1990, 1995; Mallin, 1990;
22 Aicken et al., 1992, 2004; Vartianinen et al., 1993; Cohn et al., 1994, Morgan and Cassady,
23 2002; ATSDR, 2006, 2008) with data on cancer incidence are correlation studies to examine
24 cancer outcomes of residents in communities with TCE and other chemicals detected in
25 groundwater wells or in municipal drinking water supplies (see Table 4-3). These studies did not
26 meet all five meta-analysis inclusion criteria. The geographic-base studies are not of analytical
27 designs such as cohort and case-control designs. Another deficiency in all studies is their low
28 level of detail to individual subjects for TCE. One level of exposure to all subjects in a
29 geographic area is assigned without consideration of water distribution networks, which may
30 influence TCE concentrations delivered to a home, or a subject's ingestion rate to estimate TCE
31 exposure to individual study subjects. Some inherent measurement error and misclassification
32 bias is likely in these studies because not all subjects are exposed uniformly. Additionally, in
33 contrast to case-control studies, the geographic-based studies, including ATSDR (2008), had
34 limited accounting for other potential risk factors. These studies are of low sensitivity for weight-
35 of evidence characterization of hazard compared to high-quality cohort and case-control studies.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.2. GENETIC TOXICITY
2 This section discusses the genotoxic potential of TCE and its metabolites. A summary is
3 provided at the end of each section for TCE or its metabolite for their mutagenic potential in
4 addition to an overall synthesis summary at the end of the genotoxicity section. The liver and
5 kidney are subjects of study for the genotoxic potential of TCE and its metabolites, and are
6 discussed more in-depth in sections 4.4.3, 4.4.7, 4.5.6.2.7, 4.5.7, E.2.3, and E.2.4.
7 The application of genotoxicity data to predict potential carcinogenicity is based
8 on the principle that genetic alterations are found in all cancers. Genotoxicity is the ability of
9 chemicals to alter the genetic material in a manner that permits changes to be transmitted during
10 cell division. Although most tests for mutagenicity detect changes in DNA or chromosomes,
11 some specific modifications of the epigenome including proteins associated with DNA or RNA,
12 can also cause transmissible changes. Changes that occur due to the modifications in the
13 epigenome are discussed in endpoint-specific Sections 4.3-4.9 as well as Sections E.3.l-E.3.4.
14 Genetic alterations can occur through a variety of mechanisms including gene mutations,
15 deletions, translocations, or amplification; evidence of mutagenesis provides mechanistic support
16 for the inference of potential for carcinogenicity in humans.
17 Evaluation of genotoxicity data entails a weight of evidence approach that includes
18 consideration of the various types of genetic damage that can occur. In acknowledging that
19 genotoxicity tests are by design complementary evaluations of different mechanisms of
20 genotoxicity, a recent IPCS publication (Eastmond et al., 2009) notes that "multiple negative
21 results may not be sufficient to remove concern for mutagenicity raised by a clear positive result
22 in a single mutagenicity assay." These considerations inform the present approach. In addition,
23 consistent with U.S. EPA's Guidelines on Carcinogenic Risk Assessment and Supplemental
24 Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens) (2005a, b), the
25 approach does not address relative potency (e.g., among TCE metabolites, or of such metabolites
26 with other known genotoxic carcinogens) per se, nor does it consider quantitative issues related
27 to the probable production of these metabolites in vivo. Instead, the analysis of genetic toxicity
28 data presented here focuses on the identification of a genotoxic hazard of these metabolites; a
29 quantitative analysis of TCE metabolism to reactive intermediates, via PBPK modeling, is
30 presented in Section 3.5.
31 TCE and its known metabolites trichloroacetic acid (TCA), dichloroacetic acid (DCA),
32 chloral hydrate (CH), trichloroethanol (TCOH), S-(l,2-dichlorovinyl)-L-cysteine (1,2-DCVC)
33 and S-dichlorovinyl glutathione (DCVG) have been studied to varying degrees for their
34 genotoxic potential. The following section summarizes available data on genotoxicity for both
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCE and its metabolites for each potential genotoxic endpoints, when available, in different
2 organisms.
3 4.2.1. Trichloroethylene (TCE)
4 4.2.1.1. DNA Binding Studies
5 Covalent binding of TCE to DNA and protein in cell-free systems has been studied by
6 several investigators. Incubation of 14C-TCE with salmon sperm DNA in the presence of
7 microsomal preparations from B6C3F1 mice resulted in dose-related covalent binding of TCE to
8 DNA. The binding was enhanced when the microsomes were taken from mice pretreated with
9 phenobarbital, which induces cytochrome P450 (CYP) enzymes, suggesting the binding may be
10 related to an oxidative metabolite, or when l,2-epoxy-3,3,3-trichloropropane, an inhibitor of
11 epoxide hydrolase, was added to the incubations (Banerjee and Van Duuren, 1978). In addition,
12 covalent binding of 14C-TCE with microsomal proteins was detected after incubation with
13 microsomal preparations from mouse lung, liver, stomach and kidney and rat liver (Banerjee and
14 Van Duuren, 1978). Furthermore, incubation of 14C-TCE with calf thymus DNA in the presence
15 of hepatic microsomes from phenobarbital-pretreated rats yielded significant covalent binding
16 (DiRenzoetal., 1982).
17 A number of studies have also examined the role TCE metabolism in covalent binding.
18 Miller and Guengerich (1983) used liver microsomes from control, b-naphthoflavone- and
19 phenobarbital-induced B6C3F1 mice, Osborne-Mendel rats and human liver microsomes.
20 Significant covalent binding of TCE metabolites to calf thymus DNA and proteins was observed
21 in all experiments. Phenobarbital treatment increased the formation of chloral and TCE oxide
22 formation, DNA and protein adducts. In contrast, b-naphthoflavone treatment did not induce the
23 formation of any microsomal metabolite suggesting that the forms of CYP induced by
24 phenobarbital are primarily involved in TCE metabolism while the b-naphthoflavone-inducible
25 forms of CYP have only a minor role in TCE metabolism. TCE metabolism (based on TCE-
26 epoxide and DNA-adduct formation) was 2.5-3-fold higher in mouse than in rat microsomes due
27 to differences in rates and clearance of metabolism (discussed in Section 3.3.3.1). The levels of
28 DNA and protein adducts formed in human liver microsomal system approximated those
29 observed in liver microsomes prepared from untreated rats. It was also shown that whole
30 hepatocytes of both untreated mice and phenobarbital-induced rats and mice could activate TCE
31 into metabolites able to covalently bind extracellular DNA. A study by Cai and Guengerich
32 (2001) postulate TCE oxide (an intermediate in the oxidative metabolism of TCE in rat and
33 mouse liver microsomes) is responsible for the covalent binding of TCE with protein, and to a
34 lesser extent, DNA. The authors used mass spectrometry to analyze the reaction of TCE oxide
35 (synthesized by m-chloroperbenzoic acid treatment of TCE) with nucleosides, oligonucleotides
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-30 DRAFT—DO NOT CITE OR QUOTE
-------
1 and protein to understand the transient nature of the inhibition of enzymes in the context of
2 adduct formation. Protein amino acid adducts were observed during the reaction of TCE oxide
3 with the model peptides. The majority of these adducts were unstable under physiological
4 conditions. Results using other peptides also indicate that adducts formed from the reaction of
5 TCE oxide with macromolecules and their biological effects are likely to be relatively short-
6 lived.
7 Studies have been conducted using in vitro and in vivo systems to understand the DNA
8 and protein binding capacity of TCE. Binding of TCE was observed in calf thymus DNA. In a
9 study in male mice, after repeated intraperitoneal (i.p.) injections of 14C-TCE, radioactivity was
10 detected in the DNA and RNA of all organs studied (kidney, liver, lung, spleen, pancreas, brain
11 and testis) (Bergman, 1983). However, in vivo labeling was shown to be due to metabolic
12 incorporation of Cl fragments, particularly in guanine and adenine, rather than to DNA-adduct
13 formation. In another study (Stott et al., 1982), following i.p. injection of 14C-TCE in male
14 Sprague-Dawley rats (10-100 mg/kg) and B6C3F1 mice (10-250 mg/kg), high liver protein
15 labeling was observed while very low DNA labeling was detected. Stott et al. (1982) also
16 observed very low levels of DNA binding (0.62 ± 0.43 alkylation/106 nucleotides) in mice
17 administered 1,200 mg/kg of TCE. In addition, a dose-dependent binding of TCE to hepatic
18 DNA and protein at low doses in mice was demonstrated by Kautiainen et al. (1997). In their
19 dose-response study (doses between 2 |ig/kg and 200 mg/kg BW), the highest level of protein
20 binding (2.4 ng/g protein) was observed 1 hour after the treatment followed by a rapid decline,
21 indicating pronounced instability of the adducts and/or rapid turnover of liver proteins. Highest
22 binding of DNA (120 pg/g DNA) was found between 24 and 72 hours following treatment.
23 Dose-response curves were linear for both protein and DNA binding. In this study, the data
24 suggest that TCE does bind to DNA and proteins in a dose-dependent fashion, however, the type
25 and structure of adducts were not determined.
26 Mazzullo et al. (1992) reported that TCE was covalently bound in vivo to DNA, RNA
27 and proteins of rat and mouse organs 22 hours after i.p. injection. Labeling of proteins from
28 various organs of both species was higher than that of DNA. Bioactivation of TCE to its
29 intermediates using various microsomal fractions was dependent on CYP enzyme induction and
30 the capacity of these intermediates to bind to DNA. It appeared that mouse lung microsomes
31 were more efficient in forming the intermediates than rat lung microsomes, although no other
32 species specific differences were found (Mazzullo et al., 1992) This also supports the results
33 described by Miller and Guengerich (1983). The authors suggest some binding ability of TCE to
34 interact covalently with DNA (Mazzullo et al., 1992).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 In summary, studies report that TCE exposure in vivo can lead to binding to nucleic acids
2 and proteins, and some authors have suggested that such binding is likely due to conversion to
3 one or more reactive metabolites.
4 4.2.1.2. Bacterial Systems—Gene Mutations
5 Gene mutation studies (Ames assay) in various Salmonella typhimurium (S. typhimuriuni)
6 strains of bacteria exposed to TCE both in the presence and absence of stabilizing agent have
7 been conducted by different laboratories (Henschler et al., 1977; Simmon et al., 1977; Waskell,
8 1978; Baden et al., 1979; Crebelli et al., 1982; Shimada et al., 1985; Mortelmans et al., 1986;
9 McGregor et al., 1989) (see Table 4-6). It should be noted that these studies have tested TCE
10 samples of different purities using various experimental protocols. In all in vitro assays,
11 volatization is a concern when TCE is directly administered.
12 Waskell (1978) studied the mutagenicity of several anesthetics and their metabolites.
13 Included in their study was TCE (and its metabolites) using the Ames assay. The study was
14 conducted both in the presence and absence of S9 and caution was exercised to perform the
15 experiment under proper conditions (incubation of reaction mixture in sealed dessicator vials).
16 This study was performed in both TA98 and TA100 S. typhimurium strains at a dose range of
17 0.5-10% between 4 and 48 hours. No change in revertant colonies was observed in any of the
18 doses or time courses tested. No information either on the presence or absence of stabilizers in
19 TCE obtained commercially nor its effect on cytotoxicity was provided in the study.
20 In other studies highly purified, epoxide free TCE samples were not mutagenic in
21 experiments with and without exogenous metabolic activation by S9 in S. typhimurium strain
22 TA100 using the plate incorporation assay (Henschler et al., 1977). Furthermore, no mutagenic
23 activity was found in several other strains including TA1535, TA1537, TA97, TA98, and
24 TAlOOusing the preincubation protocol (Mortelmans et al., 1986). Simmon et al. (1977)
25 observed a less than 2-fold but reproducible and dose-related increase in his+ revertants in plates
26 inoculated with S. typhimurium TA100 and exposed to a purified, epoxide-free TCE sample.
27 The authors observed no mutagenic response in strain TA1535 with S9 mix and in either
28 TA1535 or TA100 without rat or mouse liver S9. Similar results were obtained by Baden et al.
29 (1979), Bartsch et al. (1979) and Crebelli et al. (1982). In all these studies purified, epoxide-free
30 TCE samples induced slight but reproducible and dose-related increases in his+ revertants in
31 S. typhimurium TA100 only in the presence of S9. No mutagenic activity was detected without
32 exogenous metabolic activation or when liver S9 from naive rats, mice and hamsters (Crebelli et
33 al., 1982) was used for activation. Therefore, a number of these studies showed positive results
34 in TA100 with metabolic activation, but not in other strains or without metabolic activation.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-32 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-6. TCE genotoxicity: bacterial assays
to
VO Co*'
I
I
§
***.
S'
1
TO'
Co
I
Test system/endpoint
S. typhimurium (TA1 00)
S. typhimurium (TA1535, TA100)
S. typhimurium (TA98, TA100)
S. typhimurium (TA100, TA1535)
S. typhimurium (TA1 00)
S. typhimurium (TA1 00)
S. typhimurium (TA1535, TA100)
S. typhimurium (TA98, TA100,
TA1535, TA1537, TA97)
S. typhimurium (TA98, TA100,
TA1 535)
S. typhimurium (TA98, TA100,
TA1 535)
S. typhimurium
S. typhimurium
S. typhimurium (YG7108)
E. co//(K12)
Doses tested
0.1-10 ul_ (epoxide-free)
1-2.5% (epoxide-free)
0.5-10%
1-3% (epoxide-free)
5-20% (v/v)
0.33-1.33% (epoxide-free)
1-5% (higher and lower
purity)
10-1 000 uL/plate
<1 0,000 |jg/plate
(unstabilized)
<1 0,000 |jg/plate (oxirane-
stabilized)
<1 0,000 |jg/plate
(epoxybutane stabilized)
<1 0,000 |jg/plate
(epichlorohydrin stabilized)
1000-3000 |jg/plate
0.9 mM (analytical grade)
With activation
—
+ (TA100)
-(TA1535)
—
+ (TA100)
+/-(TA1535)
+
- (higher purity)
+ (lower purity)
—
—
+
ND
ND
ND
+
Without
activation
—
—
—
-
—
—
ND
+
+
+
+
—
Comments
plate incorporation assay
the study was conducted
in sealed dessicator vials
negative under normal
conditions, but 2-fold
increase in mutations in a
preincubation assay
extensive cytotoxicity
preincubation protocol
vapor assay
vapor assay
preincubation assay
vapor assay
microcolony assay/
revertants
revertants at arg56 but not
nad113 or other loci
References
Henschler et al.,
1977
Simmon et al.,
1977
Waskell, 1978
Baden etal., 1979
Bartsch etal., 1979
Crebellietal., 1982
Shimada et al.,
1985
Mortelmans et al.,
1986
McGregor et al.,
1989
McGregor et al.,
1989
McGregor et al.,
1989
McGregor et al.,
1989
Emmertetal., 2006
Greimetal, 1975
H I
O >
HH Oq
H TO
O
ND = not determined .
H
W
-------
1 Shimada et al. (1985) tested a low-stabilized, highly purified TCE sample in an Ames
2 reversion test, modified to use vapor exposure, in S. typhimurium TA1535 and TA100. No
3 mutagenic activity was observed—either in the presence or absence of S9 mix. However, at the
4 same concentrations (1, 2.5, and 5%), a sample of lower purity, containing undefined stabilizers,
5 was directly mutagenic in TA100 (>5-fold) and TA1535 (>38-fold) at 5% concentration
6 regardless of the presence of S9. It should be noted that the doses used in this study resulted in
7 extensive killing of bacterial population, particularly at 5% concentration, more than 95%
8 toxicity was observed.
9 A series of studies evaluating TCE (with and without stabilizers) was conducted by
10 McGregor et al. (1989). The authors tested high purity and oxirane-stabilized TCE samples for
11 their mutagenic potential in S. typhimurium strains TA1535, TA98, and TA100. Preincubation
12 protocol was used to test stabilized TCE (up to 10,000 jig/plate). Mutagenic response was not
13 observed either in the presence or absence of metabolic activation. When TCE was tested in a
14 vapor delivery system without the oxirane stabilizers, no mutagenic activity was observed.
15 However, TA1535 and TA100 produced a mutagenic response both in the presence and absence
16 of S9 when exposed to TCE containing 0.5-0.6% 1,2-epoxybutane. Furthermore, exposure to
17 epichlorohydrin also increased the frequency of mutants.
18 Emmert et al. (2006) used a CYP2E1-competent bacterial strain (S. typhimurium
19 containing YG7108pin3ERbs plasmid) in their experiments. TCE was among several other
20 compounds investigated and was tested at concentrations of 1,000-3,000 jig/plate. TCE induced
21 toxicity and microcolonies at or above 1,000 jig per plate. A study on Escherichia coli (E. coll)
22 K12 strain was conducted by Greim et al. (1975) using analytical-grade TCE samples.
23 Revertants were scored at two loci: arg56, sensitive to base-pair substitution and nadns, reverted
24 by frameshift mutagens. In addition, forward mutations to 5-methyltryptophan resistance and
25 galactose fermentation were selected. Approximately 2-fold increase in arg+ colonies was
26 observed. No change in other sites was observed. No definitive conclusion can be drawn from
27 this study due to lack of information on reproducibility and dose-dependence.
28 In addition to the above studies, the ability of TCE to induce gene mutations in bacterial
29 strains has been reviewed and summarized by several authors (Fahrig et al., 1995; Crebelli and
30 Carere, 1989; Douglas et al., 1999; Moore and Harrington-Brock, 2000; Clewell and Andersen,
31 2004). In summary, TCE, in its pure form as a parent compound is unlikely to induce point
32 mutations in most bacterial strains. It is possible that some mutations observed in response to
33 exposure to technical grade TCE may be contributed by the contaminants/impurities such as
34 1,2-epoxybutane and epichlorohydrin, which are known bacterial mutagens. However, several
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-34 DRAFT—DO NOT CITE OR QUOTE
-------
1 studies of TCE reported low, but positive responses in the TA100 strain in the presence of S9
2 metabolic activation, even when genotoxic stabilizers were not present.
3
4 4.2.1.3. Fungal and Yeast Systems—Gene Mutations, Conversions and Recombination
5 Gene mutations, conversions, and recombinations have been studied to identify the effect
6 of TCE in fungi and yeast systems (see Table 4-7).
7 Crebelli et al. (1985) studied the mutagenicity of TCE in Aspergillus nidulans (A.
8 nidulans) both for gene mutations and mitotic segregation. No increase in mutation frequency
9 was observed when A. nidulans was plated on selective medium and then exposed to TCE
10 vapors. A small but statistically significant increase in mutations was observed when conidia of
11 cultures were grown in the presence of TCE vapors and then plated on selective media. Since
12 TCE required actively growing cells to exerts its genotoxic activity and previous studies
13 (Bignami et al., 1980) have shown activity in the induction ofmethGl suppressors by
14 trichloroethanol and chloral hydrate, it is possible that endogenous metabolic conversion of TCE
15 into trichloroethanol or chloral hydrate may have been responsible for the positive response.
16 To understand the cytochrome P450 mediated genotoxic activity of TCE, Callen et al.
17 (1980) conducted a study in two yeast strains (D7 and D4) CYP. The D7 strain in it log-phase
18 had a CYP concentration up to 5 times higher than a similar cell suspension of D4 strain. Two
19 different concentrations (15 and 22 mM) at two different time points (1 and 4 hours) were
20 studied. A significant increase in frequencies of mitotic gene conversion and recombination was
21 observed at 15 mM concentrations at 1-hour exposure period in the D7 strain, however, the
22 22 mM concentration was highly cytotoxic (only 0.3% of the total number of colonies survived).
23 No changes were seen in D4 strain, suggesting that metabolic activation via CYP played an
24 important role in both genotoxicity and cytotoxicity. However, marginal or no genotoxic activity
25 was observed when incubation of cells and test compounds were continued for 4 hours in either
26 strain, possibly because of increased cytotoxicity, or a destruction of the metabolic system.
27 Koch et al. (1988) studied the genotoxic effects of chlorinated ethylenes including TCE
28 in various yeast Saccharomyces cerevisiae strains. Strain D7 was tested (11.1, 16.6, and 22.2
29 mM TCE) both in stationary-phase cells without S9, stationary-phase cells with S9 and
30 logarithmic-phase cells using different concentrations. No significant change in mitotic gene
31 conversion or reverse mutation was observed in either absence or presence of S9. In addition,
32 there was an considerable increase in the induction of mitotic aneuploidy in Strain D61.M,
33 though no statistical analysis was performed.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-7. TCE genotoxicity: fungal and yeast systems
to
VO Co*'
I
I
§
^.
Co'
1
TO'
Co
I
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
Gene Conversions
S. cerevisiae D7 and D4
S. cerevisiae D7
S. pombe
S. cerevisiae D7
A. nidulans
15 and 22 mM; 1
and 4 h
11.1, 16.6, and
22.2 mM
0.2 to 200 mM
("pure" and
technical grade)
ND
"
"
+
no data
+ at 1 h, D7
strain;
-at4h, both
D7 and D4
"
"
-
+
gene conversion;
CYP content 5-fold greater
in D7 strain;
high cytotoxicity at 22 mM
both stationary and log
phase/production of
phototropic colonies
forward mutation, different
experiments with different
doses and time
forward mutation
Callen etal., 1980
Kochetal., 1988
Rossi etal., 1983
Bronzetti etal., 1980
Crebelli etal., 1985
Recombination
S. cerevisiae
S. cerevisiae D7 and D4
A. nidulans
15 and 22 mM;
1 and 4 h
+
ND
ND
-
+
+
gene conversion
gene cross over
Bronzetti etal., 1980
Callen etal., 1980
Crebellii etal., 1985
Mitotic aneuploidy
S. cerevisiae D61 .M
5.5, 11.1, and
16.6 mM
+
+
loss of dominant color
homolog
Kochetal., 1988
~ ND = not determined .
H ^
O h^.
h-jOg
O
H
W
-------
1 Rossi et al. (1983) studied the effect of TCE on yeast species S. pombe both using in vitro
2 and host mediated mutagenicity studies and the effect of two stabilizers, epichlorohydrin and
3 1,2-epoxybutane that are contained in the technical grade of TCE. The main goal of this study
4 was to evaluate genotoxic activity of TCE samples of different purity and if the effect is due to
5 the additives present in the TCE or TCE itself. Forward mutations at five loci (ade 1, 3, 4, 5, 9)
6 of the adenine pathway in the yeast, strain PI was evaluated. The stationary-phase cells were
7 exposed to 25 mM concentration of TCE for 2, 4, and 8 hours in the presence and absence of S9.
8 No change in mutation frequency was observed either in pure-grade samples or technical-grade
9 samples either in the presence or absence of S9 at any of the time-points tested. Interestingly,
10 this suggests that the stabilizers used in technical-grade TCE are not genotoxic in yeast. In a
11 follow-up experiment, the same authors studied the effect of different concentrations (0.22, 2.2
12 and 22.0 mM) in a host mediated assay using liver microsome preparations obtained from
13 untreated mice, from phenobarbital-pretreated and NF-pretreated mice and rats, which also
14 suggested that stabilizers were not genotoxic in yeast. This experiment is described in more
15 detail in Section 4.2.1.4.1.
16 Furthermore, TCE was tested for its ability to induce both point mutation and mitotic
17 gene conversion in diploid strain of yeast S. cerevisiae (strain D7) both with and without a
18 mammalian microsomal activation system. In a suspension test with D7, TCE was active only
19 with microsomal activation (Bronzetti et al., 1980).
20 These studies are consistent with those of bacterial systems in indicating that pure TCE as
21 a parent compound is not likely to cause mutations, gene conversions, or recombinations in
22 fungal or yeast systems. In addition, the data suggest that contaminants used as stabilizers in
23 technical grade TCE are not genotoxic in these systems, and that the observed genotoxic activity
24 in these systems is predominantly mediated by TCE metabolites.
25
26 4.2.1.4. Mammalian Systems Including Human Studies
27 4.2.1.4.1. Gene mutations (bacterial, fungal, or yeast with a mammalian host). Very few
28 studies have been conducted to identify the effect of TCE, particularly on gene (point) mutations
29 using mammalian systems (see Table 4-8). An overall summary of different endpoints using
30 mammalian systems will be provided at the end of this section. In order to assess the potential
31 mutagenicity of TCE and its possible contaminants, Rossi et al. (1983) performed genotoxicity
32 tests using two different host mediated assays with pure- and technical-grade TCE. Male mice
33 were administered with one dose of 2 g/kg of pure or technical grade TCE by gavage. Following
34 the dosing, for the intraperitoneal host-mediated assay, yeast cell suspensions (2 x io9 cells/mL)
35 were inoculated into the peritoneal cavity of the animals. Following 16 hours, animals were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-37 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-8. TCE genotoxicity: mammalian systems—gene mutations and chromosome aberrations
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
Gene mutations (forward mutations)
Schizosaccharomyces pombe
2g/kg, 4 and 16 h
ND
Host-mediated:
intravenous and
intraperitoneal injections
of yeast cells
Rossi etal., 1983
Gene mutations (mutations frequency)
lacZtransgenic mice
0,203, 1,153, or
3,141 ppm
No base
changes or
small
deletions
No base
changes or
small
deletions
Lung, liver, bone marrow,
spleen, kidney, testicular
germ cells used
Douglas etal., 1999
Chromosomal aberrations*
CHO
C57BL/6J mice
S-D rats
745-14,900
ug/mL
499-14,900
ug/mL
5, 50, 500, or
5,000 ppm (6 h)
5, 50, 500, or
5,000 ppm (6 h,
single and 4-d
exposure)
ND
—
-
—
ND
NA
NA
8-14 h
2 h exposure
Splenocytes
Peripheral blood
lymphocytes
Galloway et al., 1987
Galloway et al., 1987
Kligerman et al., 1994
Kligerman et al., 1994
*It should be noted that results of most chromosomal aberration assays report the combined incidence of multiple effects, including chromatid breaks,
isochromatid or chromosome breaks, chromatid exchanges, dicentric chromosomes, ring chromosomes, and other aberrations.
ND = not determined, NA = not applicable.
-------
1 sacrificed and yeast cells were recovered to detect the induction of forward mutations at five loci
2 (ade 1, 2, 4, 5, 9) of the adenine pathway. A second host-mediated assay was performed by
3 exposing the animals to 2 g/kg of pure or technical grade TCE and inoculating the cells into the
4 blood system. Yeast cells were recovered from livers following 4h of exposure. Forward
5 mutations in the five loci (ade 1,2,4,5,9} were not observed in host-mediated assay either with
6 pure or technical-grade TCE. Genotoxic activity was not detected when the mutagenic epoxide
7 stabilizers were tested for mutagenicity independently or in combination. To confirm the
8 sensitivity of the assay, the authors tested a positive control—N-nitroso-dimethyl-nitrosamine
9 (1 mg/kg) and found a mutation frequency of more than 20 times the spontaneous level. The
10 authors suggest that the negative result could have been due to an inadequate incubation time of
11 the sample with the yeast cells.
12 Male and female transgenic lac Z mice were exposed by inhalation to an actual
13 concentrations of 0, 203, 1,153, and 3,141 ppm TCE, 6 hours/day for 12 days (Douglas et al.,
14 1999). Following 14 and 60 days of last exposure, animals were sacrificed and the mutation
15 frequencies were determined in various organs such as bone marrow, kidney, spleen, liver, lung,
16 and testicular germ cells. No statistically significant increases in base-changes or small-deletions
17 were observed at any of the doses tested in male or female lung, liver, bone marrow, spleen, and
18 kidney, or in male testicular germ cells when the animals were sampled 60 days after exposure.
19 In addition, statistically significantly increased gene mutations were not observed in the lungs at
20 14 days after the end of exposure (Douglas et al., 1999). The authors acknowledge that lacZ
21 bacteriophage transgenic assay does not detect large deletions. The authors also acknowledge
22 that their hypothesis does not readily explain the increases in small deletions and base-change
23 mutations found in the von Hippel-Lindau tumor suppressor gene in renal cell carcinomas of the
24 TCE-exposed population. DCA, a TCE metabolite has been shown to increase lad mutations in
25 transgenic mouse liver, however, only after 60-weeks-of-exposure to high concentration
26 (> 1,000 ppm) in drinking water (Leavitt et al., 1997). DCA induced relatively small increase in
27 lac I mutations when the animals were exposed for 60 weeks, a significantly longer duration than
28 the TCE exposure in the Douglas et al. (1999) study (<2 weeks). Because a relatively small
29 fraction of TCE is metabolized to DCA (see Section 3.3), the mutagenic effect of DCA is
30 unlikely to have been detected in the experiments in Douglas et al. (1999). GSH conjugation,
31 which leads to the production of genotoxic metabolites (see Section 4.2.5), constitutes a
32 relatively small (and relatively uncertain) portion of TCE metabolism in mice, with little data on
33 the extent of renal DCVC bioactivation versus detoxification in mice (see Sections 3.3 and 3.5).
34 In addition, statistically significantly increased kidney tumors have not been reported in mice
35 with TCE treatment, and the increased incidence of kidney tumors in rats, while considered
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 biologically significant, are quite low and not always statistically significant (see Section 4.4).
2 Therefore, although Douglas et al. (1999) did not detect increased mutations in the kidney, these
3 results are not highly informative as to the role of mutagenicity in TCE-induced kidney tumors,
4 given the uncertainties in the production in genotoxic GSH conjugation metabolites in mice and
5 the low carcinogenic potency of TCE for kidney tumors in rodents relative to what is detectable
6 in experimental bioassays..
7
8 4.2.1.4.2. von Hippel-Lindau (VHL) gene mutations. Studies have been conducted to
9 determine the role of VHL gene mutations in renal cell carcinoma, with and without TCE
10 exposure, and are summarized here. Most of these studies are epidemiologic, comparing VHL
11 mutation frequencies of TCE-exposed to nonexposed cases from renal cell carcinoma
12 case-control studies, or to background mutation rates among other renal cell carcinoma case
13 series (described in Section 4.4.3). Inactivation of the VHL gene through mutations, loss of
14 heterozygosity and imprinting has been observed in about 70% of renal clear cell carcinomas
15 (Alimov et al., 2000; Kenck et al., 1996). Recent studies have also examined the role of other
16 genes or pathways in renal cell carcinoma subtypes, including c-myc activation and vascular
17 endothelial growth factor (VEGF) (Purge et al., 2007; Toma et al., 2008).
18 Several studies have examined the role of VHL gene inactivation in renal cell carcinoma,
19 including a recent study that measured not only mutations but also promoter hypermethylation
20 (Nickerson et al., 2008). This study focused on kidney cancer regardless of cause, and found that
21 91% of cc-renal cell carcinoma (RCC) exhibited alterations of the VHL gene, suggesting a role
22 for VHL mutations as an early event in cc-RCC. A recent analysis of current epidemiological
23 studies of renal cell cancer suggests VHL gene alterations as a marker of cc-RCC, but that
24 limitations of previous studies may make the results difficult to interpret (Chow and Devesa,
25 2008). Conflicting results have been reported in epidemiological studies of VHL mutations in
26 TCE-exposed cases and are described in detail in Section 4.5.2. Both Briining et al. (1997) and
27 Brauch et al. (1999, 2004) associated increased VHL mutation frequency in TCE-exposed renal
28 cell carcinoma cases. The two other available studies of Schraml et al. (1999) and
29 Charbotel et al. (2007) because of their limitations and lower mutation detection rate in the case
30 of Charbotel et al. (2007) neither add nor detract to the conclusions from the earlier studies.
31 Additional discussion of these data are in Section 4.4.3.
32 Limited animal studies have examined the role of TCE and VHL mutations, although
33 Mally et al. (2006) have recently conducted both in vitro and in vivo studies using the Eker rat
34 model (see Section 4.4.6.1.1). The Eker rat model (Tsc-2+") is at increased risk for the
35 development of spontaneous renal cell carcinoma and as such has been used to understand the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-40 DRAFT—DO NOT CITE OR QUOTE
-------
1 mechanisms of renal carcinogenesis (Stemmer et al., 2007; Wolf et al., 2000). One study has
2 demonstrated similar pathway activation in Eker rats as that seen in humans with VHL mutations
3 leading to renal cell carcinoma, suggesting Tsc-2 inactivation is analogous to inactivation of VHL
4 in human renal cell carcinoma (Liu et al., 2003). In Mally et al. (2006), male rats carrying the
5 Eker mutation were exposed to TCE (0, 100, 250, 500, or 1,000 mg/kg body weight [BW] by
6 gavage, 5 days a week) for 13 weeks to determine the renal effects (additional data from this
7 study on in vitro DCVC exposure are discussed below, Section 4.2.5). A significant increase in
8 labeling index in kidney tubule cells was observed, however, no enhancement of preneoplastic
9 lesions or tumor incidence was found in Eker rat kidneys compared to controls. In addition, no
10 VHL gene mutations in exons 1-3 were detected in tumors obtained from either control or TCE-
11 exposed Eker rats. Although no other published studies have directly examined VHL mutations
12 following exposure to TCE, two studies performed mutational analysis of archived formalin-
13 fixed paraffin embedded tissues from renal carcinomas from previous rat studies. These
14 carcinomas were induced by the genotoxic carcinogens potassium bromate (Shiao et al., 2002) or
15 7V-nitrosodimethylamine (Shiao et al., 1998). Limited mutations in the VHL gene were observed
16 in all samples, but, in both studies, these were found only in the clear cell renal carcinomas.
17 Limitations of these two studies include the small number of total samples analyzed, as well as
18 potential technical issues with DNA extraction from archival samples (see Section 4.4.3).
19 However, analyses of VHL mutations in rats may not be informative as to the potential
20 genotoxicity of TCE in humans because the VHL gene may not be the target for
21 nephrocarcinogenesis in rats to the extent that it appears to be in humans.
22
23 4.2.1.4.3. Chromosomal aberrations. A few studies were conducted to investigate the ability
24 of TCE to induce chromosomal aberrations in mammalian systems (see Table 4-8).
25 Galloway et al. (1987) studied the effect of TCE on chromosome aberrations in Chinese hamster
26 ovary cells. When the cells were exposed to TCE (499-14,900 |ig/mL) for 2 hours with
27 metabolic activation, S9, no chromosomal aberrations were observed. Furthermore, without
28 metabolic activation, no changes in chromosomal aberrations were found when the cells were
29 exposed to TCE concentrations of 745-14,900 |ig/mL for 8-14 hours. It should be noted that in
30 this study, liquid incubation method was used and the experiment was part of a larger study to
31 understand the genotoxic potential of 108 chemicals.
32 Three inhalation studies in mice and rats examined if TCE could induce cytogenetic
33 damage (Kligerman et al., 1994). In the first two studies, CD rats or C57B1/6 mice, were
34 exposed to 0-, 5-, 500-, or 5,000-ppm TCE for 6 hours. Peripheral blood lymphocytes (PEL) in
35 rats and splenocytes in mice were analyzed for induction of chromosomal aberrations, sister
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-41 DRAFT—DO NOT CITE OR QUOTE
-------
1 chromatid exchanges and micronucleus formation. The results of micronucleus and sister
2 chromatid exchanges will be discussed in the next sections (see Sections 4.2.1.4.4 and 4.2.1.4.5).
3 No significant increase in chromosomal aberrations was observed in binucleated peripheral
4 blood lymphocytes. In the third study, the authors exposed the same strain of rats for 6
5 hours/day over 4 consecutive days. No statistically significant concentration-related increases in
6 chromosomal aberrations were observed. The limited results of the above studies have not
7 reported TCE to cause chromosomal aberrations either in in vitro or in vivo mammalian systems.
8
9 4.2.1.4.4. Micronucleus induction. The appearance of micronuclei is another endpoint that can
10 demonstrate the genotoxic effect of a chemical. Several studies have been conducted to identify
11 if TCE can cause micronucleus formation (see Table 4-9).
12 Wang et al. (2001) investigated micronucleus formation by TCE administered as a vapor
13 in CHO-K1 cells in vitro. Cells were grown in culture media with an inner petri dish containing
14 TCE that would evaporate into the media containing cells. The concentration of TCE in cultured
15 medium was determined by gas chromatography. The actual concentration of TCE ranged from
16 0.8 and 1.4 ppm after a 24-hour treatment. A significant dose-dependent increase in micronuclei
17 formation was observed. A dose-dependent decrease in cell growth and cell number was also
18 observed. The authors did not test if the micronuclei formed was due to direct damage to the
19 DNA or spindle formation.
20 Robbiano et al. (2004) conducted an in vitro study on DNA damage and micronuclei
21 formation in rat and human kidney cells exposed to six carcinogenic chemicals including TCE.
22 The authors examined for the ability of TCE to induce DNA fragmentation and formation of
23 micronuclei in primary cultures of rat and human kidney cells derived from kidney cancer
24 patients with 1-4 mM TCE concentrations. A significant dose-dependent increase in the
25 frequency of micronuclei was obtained in primary kidney cells from both male rats and human of
26 both genders. The authors acknowledge that the significance of the results should be considered
27 in light of the limitations including (1) examination of TCE on cells from only three rats, (2)
28 considerable variation in the frequency of DNA lesions induced in the cells, and (3) the
29 possibility that kidney cells derived from kidney cancer patients may be more sensitive to DNA-
30 damaging activity due to a more marked expression of enzymes involved in the metabolic
31 activation of kidney procarcinogens and suppression of DNA repair processes. Never the less,
32 this study is important and provides information of the possible genotoxic effects of TCE.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-42 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-9. TCE genotoxicity: mammalian systems—micronucleus, sister chromatic exchanges
to
VO Lo'
I
I
§
***.
£3'
1
TO'
Co
I
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
Micronucleus
Human hepatoma HepG2 cells
Primary cultures of human and rat kidney
cells
Sprague-Dawley rats
CHO-K1 cells
Male CD-1 mice
C56BL/6J mice
S-D rats
0.5-4 mM, 24 h
1.0, 2.0, or4.0
mM
3,591 mg/kg
0.8-1.4 ppm
457 mg/kg
5, 50, 500, or
5,000 ppm
5, 50, 500, or
5,000 ppm
NA
NA
+
+
-
+
+
+
-
+
NA
NA
NA
dose-dependent
significant increase
dose-dependent
significant increase
bone marrow, correlated
with TCOH in urine
splenocytes
dose dependent;
peripheral blood
lymphocytes
Huetal., 2008
Robbiano et al., 2004
Robbianoetal., 2004
Wang etal., 2001
Hrelinetal., 1994
Kligerman et al., 1994
Kligerman et al., 1994
Sister chromatid exchanges
CHO
CHO
CHO
Human lymphocytes
S-D rats
Peripheral blood lymphocytes from
humans occupationally exposed
C57BL/6J mice
0.17%
1 7.9-700 ug/mL
49.7-14,900
ug/mL
178 ug/mL
5, 50, 500, or
5,000 ppm
occupational
exposure
5, 50, 500, or
5,000 ppm
-
ND
+
ND
-
-
-
ND
+
ND
+
NA
NA
NA
1 h (vapor)
25 h (liquid)
2h
peripheral blood
lymphocytes
splenocytes
White etal., 1979
Galloway et al., 1987
Galloway et al., 1987
Guetal., 1981a, b
Kligerman et al., 1994
Nagayaet al., 1989
Kligerman et al., 1994
H I
O >
HH Oq
H TO
O
H
W
ND = not determined, NA = not applicable.
-------
1 In the same study, Robbiano et al. (2004) administered rats a single oral dose of TCE
2 (3,591 mg/kg) corresponding to /^ LDso, which had been pre-exposed to folic acid for 48 hours
3 and the rats were euthanized 48 hours later following exposure to TCE. The frequency of
4 binucleated cells was taken as an index of kidney cell proliferation. A statistically significant
5 increase in the average frequency of micronucleus was observed.
6 Hu et al. (2008) studied the effect of TCE on micronuclei frequencies using human
7 hepatoma HepG2 cells. The cells were exposed to 0.5, 1, 2, and 4 mM TCE for 24 hours. TCE
8 caused a significant increase in micronuclei frequencies at all concentrations tested. It is
9 important to note that similar concentrations were used in Robbiano et al. (2004).
10 As described in the chromosomal aberration section (see Section 4.2.1.4.3), inhalation
11 studies were performed using male C57BL/6 mice and CD rats (Kligerman et al., 1994) to
12 determine if TCE could induce micronuclei. In the first and second study, rats or mice
13 respectively, were exposed to 0-, 5-, 500-, or 5,000-ppm TCE for 6 hours. Peripheral blood
14 lymphocytes in rats and splenocytes in mice were cultured and analyzed for induction of
15 micronuclei formation. Bone marrow polychromatic erythrocytes (PCEs) were also analyzed for
16 micronuclei. TCE caused a statistically significant increase in micronuclei formation at all
17 concentrations in rat bone marrow PCEs but not in mice. The authors note that TCE was
18 significantly cytotoxic at the highest concentration tested as determined by significant
19 concentration-related decrease in the ratio of PCEs/normochromatic erythrocytes. In the third
20 study, to confirm the results of the first study, the authors exposed rats to one dose of 5,000 ppm
21 for 6 hours. A statistical increase in bone marrow micronuclei-PCEs was observed confirming
22 the results of the first study.
23 Hrelia et al. (1994) treated male CD-I mice with TCE (457 mg/kg BW; i.p.) for 30 hours.
24 Bone marrow cells were harvested for determination of micronuclei frequencies in PCEs. An
25 increase in micronuclei frequency at 30 hours after treatment was observed. Linear regression
26 analysis showed that micronuclei frequency induced by TCE correlated with trichloroethanol
27 concentrations in urine, a marker of TCE oxidative metabolism (Hrelia et al., 1994).
28 In summary, based on the results of the above studies, TCE is capable of inducing
29 micronuclei in different in vitro and in vivo systems tested. Specific methods were not used that
30 could definitively identify the mechanism of micronuclei formation. These are important
31 findings that indicate TCE has genotoxic potential as measured by the micronucleus formation.
32
33 4.2.1.4.5. Sister chromatid exchanges (SCEs). Studies have been conducted to understand the
34 ability of TCE to induce SCEs both in vitro and in vivo systems (see Table 4-9). White et al.
35 (1979) evaluated the possible induction of SCE in CHO using a vapor exposure procedure by
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-44 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposing the cells to TCE (0.17%) for 1 hour in the presence of S9 metabolic activation. No
2 change in SCE frequencies were observed between the control and the treatment group.
3 However, in another study by Galloway et al. (1987) a dose-related increase in SCE frequency in
4 repeated experiments both with and without metabolic activation was observed. It should be
5 noted that in this study, liquid incubation was used, and the exposure times were 25 hours
6 without metabolic activation at a concentration between 17.9 to 700 |ig/mL and 2 hours in the
7 presence of S9 at a concentration of 49.7 to 14,900 |ig/mL. Due to the difference in the dose,
8 length of exposure and treatment protocol (vapor exposure vs. liquid incubation), no direct
9 comparison can be made. It should also be noted that inadequacy of dose selection and the
10 absence of positive control in the White et al. (1979) makes it difficult to interpret the study. In
11 another study (Gu et al., 1981a), a small but positive response was observed in assays with
12 peripheral lymphocytes.
13 No statistically significant increase in SCEs was found when male C57B1/6 mice or CD
14 rats were exposed to TCE at concentrations of 5,500, or 5,000 ppm for 6 hours (Kligerman et al.,
15 1994). Furthermore, in another study by Nagaya et al. (1989), lymphocytes of TCE-exposed
16 workers (n = 22) and matched controls (n = 22) were analyzed for SCEs. The workers had
17 constantly used TCE in their jobs although the exact exposure was not provided. The duration of
18 their employment ranged from 0.7 to 34 years, averaging about 10 years. It should be noted that
19 there were both smokers and non-smokers among the exposed population. If a subject had not
20 smoked for at least 2 years before the samples were taken, then they were considered as non-
21 smokers. There were 8 nonsmokers in the group. If they were classified as smokers, then they
22 smoked between 10-50 cigarettes per day. No significant increase in mean SCE frequencies
23 were found in exposed population compared to controls, though the study is relatively small.
24 In summary, induction of SCEs have been reported in several, though not all, paradigms
25 of TCE exposure, consistent with the structural damage to DNA/chromosomes indicated by
26 excess micronuclei formation.
27
28 4.2.1.4.6. Unscheduled DNA synthesis. In vitro studies are briefly described here, with
29 additional discussion of effects related to TCE-induced unscheduled DNA synthesis in the
30 context of the liver in Section E.2.4.1. Perocco and Prodi (1981) studied unscheduled DNA
31 synthesis in human lymphocytes cultured in vitro (see Table 4-10). Three doses of TCE (2.5,
32 5.0, and 10 |iL/mL) were used as final concentrations with and without S9. The results indicate
33 that there was an increase in UDS only in the presence of S9, and in addition, the increase was
34 maximal at the TCE concentration of 5 |iL/mL. Three chlorinated ethane and ethylene solvent
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-45 DRAFT—DO NOT CITE OR QUOTE
-------
1 products were examined for their genotoxicity in hepatocyte primary culture DNA repair assays
2 using vapor phase exposures. Rat hepatocytes primary cultures were initiated and exposed to
3 low-stabilized or standard stabilized TCE (0.1-2.5%) for 3 or 18 hours. Unscheduled DNA
4 synthesis or DNA repair was not observed using either low or standard stabilized TCE, even at
5 vapor phase doses up to those that produced extensive cell killing after 3 or 18 hour exposure
6 (Shimada et al., 1985). Costa and Ivanetich (1984) examined the ability of TCE to induce
7 unscheduled DNA synthesis hepatocytes isolated from phenobarbital treated rats. The UDS was
8 assessed only at the highest concentration that is tolerated by the hepatocytes (2.8 mM TCE).
9 These results indicate that TCE stimulated unscheduled DNA synthesis in isolated rodent
10 hepatocytes, and, importantly, in human lymphocytes in vitro.
11
12 4.2.1.4.7. DNA strand breaks. DNA damage in response to TCE exposure was studied using
13 comet assay in human hepatoma HepG2 cells (Hu et al., 2008; see Table 4-10). The cells were
14 exposed to 0.5, 1, 2, and 4 mM for 24 hours. TCE increased the DNA migration in a significant
15 dose-dependent manner at all tested concentrations suggesting TCE caused DNA strand breaks
16 and chromosome damage.
17 TCE (4-10 mmol/kg body wt) were given to male mice by i.p. injection. The induction
18 of single-strand breaks (SSB) in DNA of liver, kidney, and lung was studied by the DNA
19 unwinding technique. There was a linear increase in the level of single strand breaks in kidney
20 and liver DNA but not in lung DNA 1 hour after administration (Walles, 1986).
21 Robbiano et al. (2004) conducted an in vitro study on DNA damage in rat and human
22 kidney cells exposed to six carcinogenic chemicals including TCE in the comet assay. The
23 authors examined the ability of TCE to induce DNA fragmentation in primary cultures of rat and
24 human kidney cells with 1-4 mM TCE concentrations. TCE was dissolved in ethanol with a
25 maximum concentration of 0.3% and the rat cultures were exposed to 20 hours. Primary human
26 kidney cells were isolated from fragments of kidney discarded during the course of surgery for
27 carcinoma of both male and female donors with an average age of 64.2 years and were also
28 exposed to 20 hours. Significant dose-dependent increases in the ratio of treated/control tail
29 length (average 4-7 jiM compared to control) was observed as measured by comet assay in
30 primary kidney cells from both male rats and human of both genders.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-46 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Co
I
Table 4-10. TCE genotoxicity: mammalian systems—unscheduled DNA synthesis, DNA strand breaks/protein
crosslinks, cell transformation
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
Unscheduled DMA synthesis
Rat primary hepatocytes
Human lymphocytes
Phenobarbital induced rat hepatocytes
2.5,5, 10ul_/ml_
2.8 mM
ND
+/-
ND
-
+
increase was only in
certain doses and
maximum at 5 uL/mL
cone.
Shimada etal., 1985
Perocco and Prodi,
1981
Costa and Ivanetich,
1984
DMA strand breaks/protein crosslinks
Primary rat kidney cells
Primary cultures of human kidney cells
Sprague-Dawley rats
Sprague-Dawley rats
0.5, 1.0,2.0,
4.0 mM
1.0,2.0, 4.0 mM
3,591 mg/kg
500, 1,000, and
2,000 ppm
NA
ND
+
—
+
+
NA
NA
dose-dependent
significant increase
dose-dependent
significant increase
single oral administration
comet assay
Robbiano et al., 2004
Robbiano et al., 2004
Robbiano et al., 2004
Clay, 2008
Cell transformation
BALB/c 3T3 mouse cells
Rat embryo cells
Syrian hamster embryo cells
4,20, 100,250
ug/mL
5, 10, 25ug/ml_
NA
NA
NA
+
+
—
weakly positive compared
to other halogenated
compounds tested in the
same experiment
Tuetal., 1985
Price etal., 1978
Amacher and Zelljadt,
1983
H I
O >
HH Oq
H TO
O
ND = not determined, NA = not applicable.
H
W
-------
1 Clay et al. (2008) studied the DNA damage inducing capacity of TCE using the comet
2 assay in rat kidney proximal tubules. Rats were exposed by inhalation to a range of TCE
3 concentrations (500, 1,000, or 2,000 ppm) for 6 hours/day for 5 days. TCE did not induce DNA
4 damage (as measured by tail length and percent tail DNA and tail movement) in rat kidney
5 proximal tubules in any of the doses tested possibly due to study limitations (small number of
6 animals tested [n = 5] and limited exposure time [6 hours/day for only 5 days]). These results
7 are in contrast to the findings of Robbiano et al. (2004) which showed DNA damage and
8 increased micronuclei in the rat kidney 20 hours following a single dose (3,591 mg/kg BW) of
9 TCE. Therefore, based on the above studies, while several studies reported DNA damage
10 induced by TCE. The DNA damage reported by comet assay is consistent with results for other
11 markers of chromosomal damage or DNA structural damage such as excess micronuclei
12 formation and SCE induced by TCE exposure.
13
14 4.2.1.4.8. DNA damage related to oxidative stress. A detailed description of studies related to
15 lipid peroxidation of TCE is presented in conjunction with discussion of liver toxicity (see
16 Section 4.5, E.2.4.3, andE.3).
17
18 4.2.1.4.9. Cell transformation. In vitro cell transformation using BALB/C-3T3 cells was
19 conducted using TCE with concentrations varying from 0-250 |ig/mL in liquid phase exposed
20 for 72 hours (see Table 4-10). The cytotoxicity of TCE at the concentration tested in the
21 transformation assay was determined by counting cells from duplicate plates of each test
22 conditions at the end of the treatment period. A dose-dependent increase in Type III foci was
23 observed although no statistical analysis was conducted (Tu et al., 1985). In another study by
24 Amacher and Zelljadt (1983), Syrian hamster embryo cells were exposed to 5, 10, or 25 |ig/mL
25 of TCE. In this experiment, two different serums (horse serum and fetal bovine serum) were also
26 tested to understand the importance of serum quality in the transformation assay. Preliminary
27 toxicity assay was performed to select dose levels which had 50-90% cell survival. One week
28 after dosing, the cell colonies were fixed and counted for variability determination and
29 examination of individual colonies for the evidence of morphological transformation. No
30 significant change in morphological transformation was obtained. Furthermore, no significant
31 changes were seen in transformation colonies when tested in different serum. However, these
32 studies are of limited use for determining the genotoxic potential of TCE because they did not
33 examine the foci for mutations, for instance in oncogenes or tumor suppressor genes.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-48 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.2.1.5. Summary
2 Evidence from a number of different analyses and a number of different laboratories
3 using a fairly complete array of endpoints suggests that TCE, following metabolism, has the
4 potential to be genotoxic. A series of carefully controlled studies evaluating TCE itself (without
5 mutagenic stabilizers and without metabolic activation) found it to be incapable of inducing gene
6 mutations in most standard mutation bacterial assays (Waskell, 1978; Henschler et al., 1977;
7 Mortelmans et al., 1986; Simmon et al., 1977; Baden et al., 1979; Bartsch et al., 1979; Crebelli et
8 al., 1982; Shimada et al., 1985; Simmon et al., 1977; Baden et al., 1979). Therefore, it appears
9 that it is unlikely that TCE is a direct-acting mutagen, though TCE has shown potential to affect
10 DNA and chromosomal structure. Low, but positive responses were observed in the TA100
11 strain in the presence of S9 metabolic activation, even when genotoxic stabilizers were not
12 present, suggesting metabolites of TCE are genotoxic. TCE is also positive in some but not all
13 fungal and yeast systems (Crebelli et al., 1985; Koch et al., 1988; Rossi et al., 1983; Callen et al.,
14 1980). Data from human epidemiological studies support the possible mutagenic effect of TCE
15 leading to VHL gene damage and subsequent occurrence of renal cell carcinoma. Association of
16 increased VHL mutation frequency in TCE-exposed renal cell carcinoma cases has been
17 observed (Briining et al.,1997; Brauch et al., 1999, 2004).
18 TCE can lead to binding to nucleic acids and proteins (Di Renzo et al., 1982; Bergman,
19 1983; Miller and Guengerich, 1983; Mazzullo et al., 1992; Kautiainen et al., 1997), and such
20 binding appears to be due to conversion to one or more reactive metabolites. For instance,
21 increased binding was observed in samples bioactivated with mouse and rat microsomal fractions
22 (Banerjee and VanDuuren, 1978; Di Renzo et al., 1982; Miller and Guengerich, 1983;
23 Mazzullo et al., 1992). DNA binding is consistent with the ability to induce DNA and
24 chromosomal perturbations. Several studies report the induction of micronuclei in vitro and in
25 vivo from TCE exposure (Kligerman et al., 1994; Hrelia et al., 1994; Wang et al., 2001;
26 Robbiano et al., 2004; Hu et al., 2008). Reports of SCE induction in some studies are consistent
27 with DNA effects, but require further study (White et al., 1979; Gu et al., 1981a, b; Nagaya et al.,
28 1989; Kligerman et al., 1994).
29 Overall, evidence from a number of different analyses and a number of different
30 laboratories using various genetic endpoints indicates that TCE has a potential to induce damage
31 to the structure of the chromosome in a number of targets but has a more limited ability to induce
32 mutation in bacterial systems.
33 Below, the genotoxicity data for TCE metabolites TCA, DC A, TCOH, chloral hydrate,
34 DCVC, and DCVG are briefly reviewed. The contributions of these data are 2-fold. First, to the
35 extent that these metabolites may be formed in the in vitro and in vivo test systems for TCE, they
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-49 DRAFT—DO NOT CITE OR QUOTE
-------
1 provide insight into what agent or agents may contribute to the limited activity observed with
2 TCE in these genotoxicity assays. Second, because the in vitro systems do not necessarily fully
3 recapitulate in vivo metabolism, the genotoxicity of the known in vivo metabolites themselves
4 provide data as to whether one may expect genotoxicity to contribute to the toxicity of TCE
5 following in vivo exposure.
6
7 4.2.2. Trichloroacetic Acid (TCA)
8 The TCE metabolite TCA has been studied using a variety of genotoxicity assay for its
9 genotoxic potential (see International Agency for Research on Cancer [IARC, 2004] for
10 additional information). Evaluation of in vitro studies of TCA must consider toxicity and
11 acidification of medium resulting in precipitation of proteins, as TCA is commonly used as a
12 reagent to precipitate proteins.
13
14 4.2.2.1. Bacterial Systems—Gene Mutations
15 TCA has been evaluated in a number of in vitro test systems including the bacterial
16 assays (Ames) using different S. typhimurium strains such as TA98, TA100, TA104, TA1535,
17 and RSJ100 (Table 4-11). The majority of these studies did not report positive findings for
18 genotoxicity (Waskell, 1978; Shirasu et al., 1976; Nestmann et al., 1980; DeMarini et al., 1994;
19 Rapson et al., 1980; Moriya et al., 1983; Nelson et al., 2001; Kargalioglu et al., 2002) Waskell
20 (1978) studied the effect of TCA (0.45 mg/plate) on bacterial strains TA98 and TA100 both in
21 the presence and absence of S9. The author did not find any revertants at the maximum nontoxic
22 dose tested. Following exposure to TCA, Rapson et al. (1980) reported no change in mutagenic
23 activity in strain TA100 in the absence of S9. DeMarini et al. (1994) performed different studies
24 to evaluate the genotoxicity of TCA, including the Microscreen prophage-induction assay (TCA
25 concentrations 0 to 10 mg/mL) and use of the S. typhimurium TA100 strain using bag
26 vaporization technique (TCA concentrations 0-100 ppm), neither of which yielded positive
27 results. Nelson et al. (2001) reported no positive findings with TCA using a S. typhimurium
28 microsuspension bioassay (S. typhimurium strain TA104) following incubation of TCA for
29 various lengths of time, with or without rat cecal microbiota. Similarly, no activity was observed
30 in a study conducted by Kargalioglu et al. (2002) where S. typhimurium strains TA98, TA100,
31 and RSJ100 were exposed to TCA (0.1-100 mM) either in the presence or absence of S9
32 (Kargalioglu et al., 2002).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-50 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-11.. Genotoxicity of Trichloroacetic acid—bacterial systems
Test system/endpoint
A, Prophage induction, E. coli WP2s
SOS chromotest, Escherichia coli PQ37
S. typhimurium TA1535, 1536, 1537,
1538, reverse mutation
S. typhimurium TA100, 98, reverse
mutation
S. typhimurium TA100, 1535, reverse
mutation
S. typhimurium TA1537, 1538, 98, reverse
mutation
S. typhimurium TA100, reverse mutation
S. typhimurium TA100, 98, reverse
mutation
S. typhimurium TA100, reverse mutation
S. typhimurium TA100, reverse mutation,
liquid medium
S. typhimurium TA104, reverse mutation,
microsuspension
S. typhimurium TA100, RSJ100, reverse
mutation
S. typhimurium TA98, reverse mutation
S. typhimurium TA1535, SOS DNA repair
Doses
(LED or HID)a
10,000
10,000
20 ug/plate
450 ug/plate
4,000 ug/plate
2,000 ug/plate
520 ug/plate
5,000 ug/plate
600 ppm
1,750
250 ug/plate
16,300
13,100
Results'1
With
activation
-
-
NT
-
-
-
NT
-
-
+
-
-
-
+
Without
activation
-
-
-
-
-
-
-
-
-
+
-
-
-
-
Reference
DeMarini et al., 1994
Ciller etal., 1997
Shirasu et al., 1976
Waskell, 1978
Nestmann et al., 1980
Nestmann et al., 1980
Rapsonetal., 1980
Moriyaetal., 1983
DeMarini et al., 1994
Ciller etal., 1997
Nelson etal., 2001
Kargalioglu et al., 2002
Kargalioglu et al., 2002
Onoetal., 1991
3
4
5
6
7
8
9
10
11
12
13
14
15
aLED, lowest effective dose; HID, highest ineffective dose; doses are in ug/mL for in vitro tests unless specified.
bResults: +, positive; -, negative; NT, not tested.
Table adapted from IARC monograph (2004) and modified/updated for newer references.
TCA was also negative in other bacterial systems. The SOS chromotest (which measures
DNA damage and induction of the SOS repair system) mE. coli PQ37, +/- S9 (Giller et al.,
1997) evaluated the genotoxic activity of TCA ranging from 10 to 10,000 |ig/mL and did not
find any response. Similarly, TCA was not genotoxic in the Microscreen prophage-induction
assay inE. coli with TCA concentrations ranging from 0 to 10,000 |ig/mL, with and without S9
activation (DeMarini et al., 1994).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-51 DRAFT—DO NOT CITE OR QUOTE
-------
1 However, TCA induced a small increase in SOS DNA repair (an inducible error-prone
2 repair system) in S. typhimurium strain TA1535 in the presence of S9 (Ono et al., 1991).
3 Furthermore, Giller et al. (1997) reported that TCA demonstrated genotoxic activity in an Ames
4 fluctuation test in S. typhimurium TA100 in the absence of S9 at noncytotoxic concentrations
5 ranging from 1,750 to 2,250 |ig/mL. The addition of S9 decreased the genotoxic response, with
6 effects observed at 3,000-7,500 |ig/mL. Cytotoxic concentrations in the Ames fluctuation assay
7 were 2,500 and 10,000 |ig/mL without and with microsomal activation, respectively.
8
9 4.2.2.2. Mammalian Systems
10 4.2.2.2.1. Gene mutations. The mutagenicity of TCA has also been tested in cultured
11 mammalian cells (Table 4-12). Harrington-Brock et al. (1998) examined the potential of TCA to
12 induce mutations in L5178Y/TK+/" -3.7.2C mouse lymphoma cells. In this study, mouse
13 lymphoma cells were incubated in culture medium treated with TCA concentrations up to
14 2,150 |ig/mL in the presence of S9 metabolic activation and up to 3,400 |ig/mL in the absence of
15 S9 mixture. In the presence of S9, a doubling of mutant frequency was seen at concentrations of
16 2,250 |ig/mL and higher, including several concentrations with survival >10%. In the absence of
17 S9, TCA increased the mutant frequency by 2-fold or greater only at concentrations of
18 2,000 |ig/mL or higher. These results were obtained at <11% survival rates. The authors noted
19 that the mutants included both large-colony and small-colony mutants. The small-colony
20 mutants are indicative of chromosomal damage. It should be noted that no rigorous statistical
21 evaluation was conducted on these data.
22
23 4.2.2.2.2. Chromosomal aberrations. Mackay et al. (1995) investigated the ability of TCA to
24 induce chromosomal damage in an in vitro chromosomal aberration assay using cultured human
25 cells. The authors treated the cells with TCA as free acid, both in the presence and absence of
26 metabolic activation. TCA induced chromosomal damage in cultured human peripheral
27 lymphocytes at concentrations (2,000 and 3,500 |ig/mL) that significantly reduced the pH of the
28 medium. However, exposure of cells to neutralized TCA did not have any effect even at a
29 cytotoxic concentration of 5,000 |ig/mL. It is possible that the reduced pH was responsible for
30 the TCA-induced clastogenicity in this study. To further evaluate the role of pH changes in the
31 induction of chromosome damage, the authors isolated liver-cell nuclei from B6C3Fi mice and
32 suspended in a buffer at various pH levels. The cells were stained with chromatin-reactive
33 (fluorescein isothiocyanate) and DNA-reactive (propidium iodide) fluorescent dyes. A decrease
34 in chromatin staining intensity was observed with the decrease in pH, suggesting that pH
35 changes, independent of TCA exposure, can alter chromatin conformation. It was concluded by
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-52 DRAFT—DO NOT CITE OR QUOTE
-------
1 the authors that TCA-induced pH changes are likely to be responsible for the chromosomal
2 damage induced by un-neutralized TCA. In another in vitro study, Plewa et al. (2002) evaluated
3 the induction of DNA strand breaks induced by TCA (1-25 mM) in CHO cells and did not
4 observe any genotoxicity.
5
6 4.2.2.2.3. Micronucleus. Relative genotoxicity of TCA was tested in a mouse in vivo system
7 (Table 4-12) using three different cytogenetic assay (bone marrow chromosomal aberrations,
8 micronucleus and sperm-head abnormalities) (Bhunya and Behera, 1987) and for chromosomal
9 aberrations in chicken (Bhunya and Jena, 1996). TCA induced a variety of anomalies including
10 micronucleus in the bone marrow of mice and chicken. A small increase in the frequency of
11 micronucleated erythrocytes at 80 |ig/mL in a newt (Pleurodeles waltl larvae) micronucleus test
12 was observed in response to TCA exposure (Giller et al., 1997). Mackay et al. (1995)
13 investigated the ability of TCA to induce chromosomal DNA damage in the in vivo bone-marrow
14 micronucleus assay in mice. C57BL mice were given TCA intraperitoneally at doses of 0, 337,
15 675, or 1,080 mg/kg/d for males and 0, 405, 810, or 1,300 mg/kg/d for females for two
16 consecutive days, and bone-marrow samples were collected 6 and 24 hours after the last dose.
17 The administered doses represented 25, 50, and 80% of the median lethal dose, respectively. No
18 treatment-related increase in micronucleated polychromatic erythrocytes was observed.
19
20 4.2.2.2.4. Other DNA damage Studies. DNA unwinding assays have been used as indicators of
21 single strand breaks and are discussed in detail in Section E.2.3. Studies were conducted on the
22 ability of TCA to induce single-strand breaks (Chang et al., 1992; Styles et al., 1991; Nelson and
23 Bull, 1988; Nelson et al., 1989; Table 4-12). Nelson and Bull (1988) evaluated the ability of
24 TCA and other compounds to induce single-strand DNA breaks in vivo in Sprague-Dawley rats
25 and B6C3Fi mice. Single oral doses were administered to three groups of three animals, with an
26 additional group as a vehicle control. Animals were sacrificed after 4 hours, and 10% liver
27 suspensions were analyzed for single-strand DNA breaks by the alkaline unwinding assay.
28 Dose-dependent increases in single-strand DNA breaks were induced in both rats and mice, with
29 mice being more susceptible than rats. The lowest dose of TCA that produced significant SSBs
30 was 0.6 mmol/kg (98 mg/kg) in rats but 0.006 mmol/kg (0.98 mg/kg) in mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-53 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-12.. TCA Genotoxicity—mammalian systems (both in vitro and in vivo)
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Test system/endpoint
Gene mutation, mouse lymphoma L5178Y/TK+/- cells, in vitro
DMA strand breaks, B6C3F1 mouse and Fischer 344 rat hepatocytes, in
vitro
DMA strand breaks, human CCRF-CEM lymphoblastic cells, in vitro
DMA damage, Chinese hamster ovary cells, in vitro, comet assay
DMA strand breaks, B6C3F1 mouse liver, in vivo
DMA strand breaks, B6C3F1 mouse liver, in vivo
DMA strand breaks, B6C3F1 mouse liver, in vivo
DMA strand breaks, B6C3F1 mouse liver and epithelial cells from stomach
and duodenum, in vivo
DMA strand breaks, male B6C3F1 mice, in vivo
Micronucleus formation, Swiss mice, in vivo
Micronucleus formation, female C57BL/6JfBL1 0/Alpk mouse bone-marrow
erythrocytes, in vivo
Micronucleus formation, male C57BL/6JfBL10/Alpk mouse bone-marrow
erythrocytes, in vivo
Micronucleus formation, Pleurodeles waltl newt larvae peripheral
erythrocytes, in vivo
Chromosomal aberrations, Swiss mouse bone-marrow cells in vivo
Chromosomal aberrations, Swiss mouse bone-marrow cells in vivo
Chromosomal aberrations, Swiss mouse bone-marrow cells in vivo
Chromosomal aberrations, chicken Gallus domesticus bone marrow, in
vivo
Doses
(LED or HID)3
3,000
1,630
1,630
3mM
1.0, oral, xi
500, oral, xi
500, oral, 10
repeats
1,630, oral, xi
500 (neutralized)
125, i.p., x2
1,300, i.p., x2
1,080, i.p., x2
80
125, i.p., xi
100, i.p., x5
500, oral, xi
200, i.p., xi
Results"
With
activation
(+)
NT
NT
NT
Without
activation
?
-
-
-
+
+
-
-
-
+
-
-
+
+
+
+
+
Reference
Harrington-Brock etal., 1998
Chang etal., 1992
Chang etal., 1992
Plewa et al., 2002
Nelson and Bull, 1988
Nelson etal., 1989
Nelson etal., 1989
Chang etal., 1992
Styles etal., 1991
Bhunya and Behera, 1987
Mackay etal., 1995
Mackay etal., 1995
Ciller etal, 1997
Bhunya and Behera, 1987
Bhunya and Behera, 1987
Bhunya and Behera, 1987
Bhunya and Jena, 1996
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
p
"I
I
§
^.
£3'
a
1
^t>
>
_;.
Table 4-12. TCA Genotoxicity—mammalian systems (both in vitro and in vivo) (continued)
Test system/endpoint
Chromosomal aberrations, human lymphocytes, in vitro
Sperm morphology, Swiss mice, in vivo
Doses
(LED or HID)3
5,000,
(neutralized)
125, i.p., x5
Results"
With
activation
-
Without
activation
+
Reference
Mackay et al., 1995
Bhunya and Behera, 1987
•a
^ 1
(j\ ^
(j\ o
s
^
I
g&
r* TO
C ^
^rl «s
H O
O §
°§
i-?1 a-
•^H SL"
g|
O^
HH Oq
H TO
SI
^^
/o ^
B*
H
W
aLED, lowest effective dose; HID, highest ineffective dose; doses are in ug/mL for in vitro tests; mg/kg for in vivo tests unless specified.
bResults: + = positive; (+) = weakly positive; - = negative; NT = not tested; ? = inconclusive.
"§ Table adapted from IARC monograph (2004) and modified/updated for newer references.
-------
1 However, in a follow-up study, Nelson et al. (1989) male B6C3F1 mice were treated with
2 500 mg/kg TCA, and single strand breaks in whole liver homogenate were examined, and no
3 significant differences from controls were reported. Moreover, in the experiments in the same
4 study with DC A, increased single strand breaks were reported, but with no dose-response
5 between 10 and 500 mg/kg, raising concerns about the reliability of the DNA unwinding assay
6 used in these studies. For further details, see Section E.2.3. In an additional follow-up
7 experiment with a similar experimental paradigm, Styles et al. (1991) tested TCA for its ability
8 to induce strand breaks in male B6C3Fi mice in the presence and absence of liver growth
9 induction. The test animals were given 1, 2, or 3 daily doses of neutralized TCA (500 mg/kg) by
10 gavage and killed 1 hour after the final dose. Additional mice were given a single 500-mg/kg
11 gavage dose and sacrificed 24 hours after treatment. Liver nuclei DNA were isolated, and the
12 induction of single strand breaks was evaluated using the alkaline unwinding assay. Exposure to
13 TCA did not induce strand breaks under the conditions tested in this assay. In a study by Chang
14 et al. (1992), administration of single oral doses of TCA (1 to 10 mmol/kg) to B6C3Fi mice did
15 not induce DNA strand breaks in a dose-related manner as determined by the alkaline unwinding
16 assay. No genotoxic activity (evidence for strand breakage) was detected in F344 rats
17 administered by gavage up to 5 mmol/kg (817 mg/kg).
18 In summary, although Nelson and Bull (1988) report effects on DNA unwinding for TCE
19 and its metabolites with DCA having the highest activity and TCA the lowest, Nelson et al.
20 (1989), using the same assay, reported no effect for TCA and the same effect at 10 and
21 500 mg/kg for DCA in mice. Moreover, Styles et al.(1991) did not find a positive result for TCA
22 using the same paradigm as Nelson and Bull (1988) and Nelson et al. (1989). Furthermore,
23 Chang et al (1992) also did not find increased single strand breaks for TCA exposure in rats.
24 (see Section E.2.4.3).
25
26 4.2.2.3. Summary
27 In summary, TCA has been studied using a variety of genotoxicity assays, including the
28 recommended battery. No mutagenicity was reported in S. typhimurium strains in the presence
29 or absence of metabolic activation or in an alternative protocol using a closed system, except in
30 one study on strain TA100 using a modified protocol in liquid medium. This is largely
31 consistent with the results from TCE, which was negative in most bacterial systems except some
32 studies with the TA100 strain. Mutagenicity in mouse lymphoma cells was only induced at
33 cytotoxic concentrations. Measures of DNA-repair responses in bacterial systems have been
34 inconclusive, with induction of DNA repair reported in S. typhimurium but not in E. coll. TCA-
35 induced clastogenicity may be secondary to pH changes and not a direct effect of TCA.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-56 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.2.3. Dichloroacetic Acid (DCA)
2 DCA is another metabolite of TCE that has been studied using a variety of genotoxicity
3 assay for its genotoxic potential (Tables 4-13 and 4-14; see IARC [2004] for additional
4 information).
5
6 4.2.3.1. Bacterial and Fungal Systems—Gene Mutations
1 Studies were conducted to evaluate mutagenicity of DCA in different S. typhimurium and
8 E. coli strains (DeMarini et al., 1994; Ciller et al., 1997; Waskell, 1978; Herbert et al., 1980; Fox
9 et al., 1996; Kargalioglu et al., 2002; Nelson et al., 2001; Fox et al., 1996). DCA was mutagenic
10 in three strains of S. typhimurium: strain TA100 in three of five studies, strain RSJ100 in a single
11 study, and strain TA98 in two of three studies. DCA failed to induce point mutations in other
12 strains of S. typhimurium (T A104, T A15 3 5, T A15 3 7, and T A15 3 8) or in K coli strain WP2uvr A.
13 In one study, DCA caused a weak induction of SOS repair in E. coli strain PQ37 (Ciller et al.,
14 1997).
15 DeMarini et al. (1994), in the same study as described in the TCA section of this chapter,
16 also studied DCA as one of their compounds for analysis. In the prophage-induction assay using
17 E. coli, DCA, in the presence of S9, was genotoxic producing 6.6-7.2 plaque-forming units
18 (PFU)/mM and slightly less than 3-fold increase in PFU/plate in the absence of S9. In the
19 second set of studies, which involved the evaluation of DCA at concentrations of 0-600 ppm for
20 mutagenicity in S. typhimurium TA100 strain, DCA was mutagenic both in the presence and
21 absence of S9, producing 3-5 times increases in the revertants/plate compared to the
22 background. The lowest effective concentration for DCA without S9 was 100 ppm and 50 ppm
23 in the presence of S9. In the third and most important study, mutation spectra of DCA were
24 determined at the base-substitution allele hisG46 of S. typhimurium TA100. DCA-induced
25 revertants were chosen for further molecular analysis at concentrations that produced mutant
26 yields that were 2-5-fold greater than the background. The mutation spectra of DCA were
27 significantly different from the background mutation spectrum. Thus, despite the modest
28 increase in the mutant yields (3-5 times) produced by DCA, the mutation spectra confirm that
29 DCA is mutagenic. DCA primarily induced GC-AT transitions.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-57 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-13.. Genotoxicity of dichloroacetic acid (bacterial systems)
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
Test system/endpoint
A Prophage induction, E. coli WP2s
SOS ch remotest, £. co/; PQ37
S. typhimurium, DMA repair-deficient strains TS24, TA2322,
TA1 950
S. typhimurium TA100, TA1535, TA1537, TA1538, reverse
mutation
S. typhimurium TA100, reverse mutation
S. fyp/?;muriumTA100,TA1535, TA1537, TA98, reverse mutation
S. typhimurium TA100, reverse mutation, liquid medium
S. typhimurium RSJ100, reverse mutation
S. typhimurium TA1 04, reverse mutation, microsuspension
S. typhimurium TA98, reverse mutation
S. typhimurium TA98, reverse mutation
S. typhimurium TA1 00, reverse mutation
E. coli WP2uvrA, reverse mutation
Doses
(LED or HID)3
2,500
500
31,000
50
5,000
100
1,935
150 ug/plate
1 0 ug/plate
5,160
1,935
5,000
Results'1
With
activation
+
-
-
-
+
-
+
-
-
(+)
-
+
-
Without
activation
-
(+)
-
-
+
-
+
+
-
-
+
+
-
Reference
DeMarini etal., 1994
Ciller etal., 1997
Waskell, 1978
Herbert et al., 1980
DeMarini etal., 1994
Fox etal., 1996
Ciller etal., 1997
Kargalioglu etal., 2002
Nelson etal., 2001
Herbert et al., 1980
Kargalioglu etal., 2002
Kargalioglu etal., 2002
Fox etal., 1996
00
I
H I
O >
HH Oq
H TO
aLED, lowest effective dose; HfD, highest ineffective dose; doses are in ug/mL for in vitro tests unless specified.
bResults: + = positive; (+) = weakly positive; - = negative.
Table adapted from IARC monograph (2004) and modified/updated for newer references.
O
H
W
-------
Table 4-14.. Genotoxicity of dichloroacetic acid—mammalian systems
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Test system/endpoint
Gene mutation, mouse lymphoma cell line L5178Y/TK+/- in vitro
Gene mutation, mouse lymphoma cell line L5178Y/TK+/-3.7.2C in
vitro
DMA strand breaks and alkali-labile damage, Chinese hamster ovary
cells in vitro (single-cell gel electrophoresis assay)
DMA strand breaks, B6C3F1 mouse hepatocytes in vitro
DMA strand breaks, Fischer 344 rat hepatocytes in vitro
Micronucleus formation, mouse lymphoma L5178Y/TK+/-3.7.2C cell
line in vitro
Chromosomal aberrations, Chinese hamster ovary in vitro
Chromosomal aberrations, mouse lymphoma L5178Y/Tk+/- -3.7.2C
cell line in vitro
Aneuploidy, mouse lymphoma L5178Y/Tk+/-3.7.2C cell line in vitro
DMA strand breaks, human CCRF-CEM lymphoblastoid cells in vitro
DMA strand breaks, male B6C3F1 mouse liver in vivo
DMA strand breaks, male B6C3F1 mouse liver in vivo
DMA strand breaks, male B6C3F1 mouse liver in vivo
DMA strand breaks, male B6C3F1 mouse splenocytes in vivo
DMA strand breaks, male B6C3F1 mouse epithelial cells from stomach
and duodenum in vivo
DMA strand breaks, male B6C3F1 mouse liver in vivo
DMA strand breaks, alkali-labile sites, cross linking, male B6C3F1
mouse blood leukocytes in vivo (single-cell gel electrophoresis assay)
Doses
(LED or HID)3
5,000
400
3,225 ug/mL
2,580
1,290
800
5,000
600
800
1,290
13, oral, xi
10, oral, xi
1,290, oral, xi
1,290, oral, xi
1,290, oral, xi
5,000, dw,
x7-14d
3,500, dw, x28 d
Results'1
With
activation
-
NT
NT
NT
NT
NT
-
NT
NT
NT
Without
activation
-
+
-
-
-
-
-
+
-
-
+
+
-
-
-
-
+
Reference
Foxetal., 1996
Harrington-Brock et al., 1998
Plewa et al., 2002
Chang et al., 1992
Chang etal.,1992
Harrington-Brock et al., 1998
Foxetal., 1996
Harrington-Brock et al., 1998
Harrington-Brock et al., 1998
Chang et al., 1992
Nelson and Bull, 1988
Nelson etal., 1989
Chang etal., 1992
Chang etal., 1992
Chang etal., 1992
Chang etal., 1992
Fuscoeet al., 1996
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-14. Genotoxicity of dichloroacetic acid—mammalian systems (continued)
Test system/endpoint
DMA strand breaks, male Sprague-Dawley rat liver in vivo
DMA strand breaks, male Fischer 344 rat liver in vivo
DMA strand breaks, male Fischer 344 rat liver in vivo
Gene mutation, lacl transgenic male B6C3F1 mouse liver assay in vivo
Micronucleus formation, male B6C3F1 mouse peripheral erythrocytes
in vivo
Micronucleus formation, male B6C3F1 mouse peripheral erythrocytes
in vivo
Micronucleus formation, male B6C3F1 mouse peripheral erythrocytes
in vivo
Micronucleus formation, male and female Crl:CD (SD) BR rat bone-
marrow erythrocytes in vivo
Micronucleus formation, Pleurodeles waltl newt larvae peripheral
erythrocytes in vivo
Doses
(LED or HID)3
30, oral, xi
645, oral, xi
2,000, dw,
x30 weeks
1,000, dw,
x60 weeks
3,500, dw, xg d
3,500, dw, x28 d
3,500, dw, xio
weeks
1,100, i.v., x3
80 d
Results'1
With Without
activation activation
+
-
-
+
+
-
+
-
—
Reference
Nelson and Bull, 1988
Chang et al., 1992
Chang et al., 1992
Leavittetal., 1997
Fuscoe et al., 1996
Fuscoeetal., 1996
Fuscoe etal., 1996
Foxetal., 1996
Ciller etal., 1997
H I
"LED, lowest effective dose; HID, highest ineffective dose; doses are in ug/mL for in vitro tests; mg/kg for in vivo tests unless specified; dw = drinking-water (in
mg/L); d = day; w = week; i.v. = intravenous.
bResults: + = positive; - = negative; NT = not tested.
Table adapted from IARC monograph (2004) and modified/updated for newer references.
O
H
W
-------
1 Kargalioglu et al. (2002) analyzed the cytotoxicity and mutagenicity of the drinking
2 water disinfection by-products including DC A in S. typhimurium strains TA98, TA100, and
3 RSJ100 +/- S9. DCA was mutagenic in this test although the response was low when compared
4 to other disinfection by-products tested in strain TA100. This study was also summarized in a
5 review by Plewa et al. (2002). Nelson et al. (2001) investigated the mutagenicity of DCA using
6 a S. typhimurium microsuspension bioassay following incubation of DCA for various lengths of
7 time, with or without rat cecal microbiota. No mutagenic activity was detected for DCA with
8 S. typhimurium strain TA104.
9 Although limited data, it appears that DCA has mutagenic activity in the S. typhimurium
10 strains, particularly TA100.
11
12 4.2.3.2. Mammalian Systems
13 4.2.3.2.1. Gene mutations. The mutagenicity of DCA has been tested in mammalian systems,
14 particularly, mouse lymphoma cell lines in vitro (Fox et al., 1996; Harrington-Brock et al., 1998)
15 and lacl transgenic mice in vivo (Leavitt et al., 1997). Harrington-Brock et al. (1998) evaluated
16 DCA for it mutagenic activity in L5178Y/TK +/- (-) 3.7.2C mouse lymphoma cells. A dose-
17 related increase in mutation (and cytotoxic) frequency was observed at concentrations between
18 100 and 800 |ig/mL. Most mutagenic activity of DCA at the Tk locus was due to the production
19 of small-colony Tk mutants (indicating chromosomal mutations). Different pH levels were
20 tested in induction of mutant frequencies and it was determined that the mutagenic effect
21 observed was due to the chemical and not pH effects.
22 Mutation frequencies were studied in male transgenic B6C3F1 mice harboring the
23 bacterial lad gene administered DCA at either 1.0 or 3.5 g/L in drinking water (Leavitt et al.,
24 1997). No significant difference in mutant frequency was observed after 4 or 10 weeks of
25 treatment in both the doses tested as compared to control. However, at 60 weeks, mice treated
26 with 1.0 g/L DCA showed a slight increase (1.3-fold) in the mutant frequency over the control,
27 but mice treated with 3.5 g/L DCA had a 2.3-fold increase in the mutant frequency. Mutational
28 spectra analysis revealed that -33% had G:C-A:T transitions and 21% had G:C-T:A
29 transversions and this mutation spectra was different than that was seen in the untreated animals,
30 indicating that the mutations were likely induced by the DCA treatment. The authors conclude
31 that these results are consistent with the previous observation that the proportion of mutations at
32 T:A sites in codon 61 of the H-ras gene was increased in DCA-induced liver tumors in B6C3F1
33 mice (Leavitt etal., 1997).
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-61 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.2.3.2.2. Chromosomal aberrations and micronucleus. Harrington-Brock et al. (1998)
2 evaluated DCA for its potential to induce chromosomal aberrations in DCA-treated (0, 600, and
3 800 |ig/mL) mouse lymphoma cells. A clearly positive induction of aberrations was observed at
4 both concentrations tested. No significant increase in micronucleus was observed in DCA-
5 treated (0, 600, and 800 jig/mL) mouse lymphoma cells (Harrington-Brock et al., 1998).
6 However, no chromosomal aberrations were found in Chinese hamster ovary cells exposed to
7 DCA (Fox etal., 1996)
8 Fuscoe et al. (1996) investigated in vivo genotoxic potential of DCA in bone marrow and
9 blood leukocytes using the peripheral-blood-erythrocyte micronucleus assay (to detect
10 chromosome breakage and/or malsegregation) and the alkaline single cell gel electrophoresis
11 (comet) assay, respectively. Mice were exposed to DCA in drinking water, available ad libitum,
12 for up to 31 weeks. A statistically significant dose-related increase in the frequency of
13 micronucleated PCEs was observed following subchronic exposure to DCA for 9 days.
14 Similarly, a significant increased was also observed when exposed for >10 weeks particularly at
15 the highest dose of DCA tested (3.5 g/L). DNA cross-linking was observed in blood leukocytes
16 in mice exposed to 3.5 g/L DCA for 28 days. These data provide evidence that DCA may have
17 some potential to induce chromosome damage when animals were exposed to concentrations
18 similar to those used in the rodent bioassay.
19
20 4.2.3.2.3. Other DNA damage studies. Nelson and Bull (1988) and Nelson et al. (1989) have
21 been described above in Section 4.2.2.4 and E.2.3, with positive results for DNA unwinding for
22 DCA, though Nelson et al. (1989) reported the same response at 10 and 500 mg/kg in mice,
23 raising concerns about the reliability of the assay in these studies. Chang et al. (1992) conducted
24 both in vitro and in vivo studies to determine the ability of DCA to cause DNA damage. Primary
25 rat (Fischer 344) hepatocytes and primary mouse hepatocytes treated with DCA for 4 hours did
26 not in induce DNA single strand breaks as detected by alkaline DNA unwinding assay. No DNA
27 strand breaks were observed in human CCRF-CEM lymphoblastoid cells in vitro exposed to
28 DCA. Similarly, analysis of the DNA single strand breaks in mice killed 1 hour after a single
29 dose of 1, 5 or 10 mM/kg DCA did not cause DNA damage. None of the Fischer 344 rats killed
30 4 hours after a single gavage treatment (1-10 mM/kg) produced any detectable DNA damage.
31
32 4.2.3.3. Summary
33 In summary, DCA has been studied using a variety but limited number of genotoxicity
34 assays. Within the available data, DCA has been demonstrated to be mutagenic in the
35 S. typhimurium assay, particularly in strain TA100, the in vitro mouse lymphoma assay and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 in vivo cytogenetic and gene mutation assays. DCA can cause DNA strand breaks in mouse and
2 rat liver cells following in vivo administration by gavage.
O
4 4.2.4. Chloral Hydrate
5 Chloral hydrate has been evaluated for its genotoxic potential using a variety of
6 genotoxicity assays (Tables 4-15, 4-16, and 4-17). These data are particularly important because
7 it is known that a large flux of TCE metabolism leads to chloral hydrate as an intermediate, so a
8 comparison of their genotoxicity profiles is likely to be highly informative.
9 4.2.4.1. DNA Binding Studies
10 Limited analysis has been performed examining DNA binding potential of chloral
11 hydrate (Keller and Heck, 1988; Von Tungeln et al., 2002; Ni et al., 1995). Keller and Heck
12 (1988) conducted both in vitro and in vivo experiments using B6C3F1 mouse strain. The mice
13 were pretreated with 1,500 mg/kg TCE for 10 days and then given 800 mg/kg [14C] chloral. No
14 detectable covalent binding of 14C to DNA in the liver was observed. Another study with in vivo
15 exposures to nonradioactive chloral hydrate at a concentration of 1,000 and 2,000 nmol in mice
16 B6C3F1 demonstrated an increase in malondialdehyde-derived and 8-oxo-2'-deoxyguanosine
17 adducts in liver DNA (Von Tungeln et al., 2002). Ni et al. (1995) observed malondialdehyde
18 adducts in calf thymus DNA when exposed to chloral hydrate and microsomes from male
19 B6C3F1 mouse liver.
20 Keller and Heck (1988) investigated the potential of chloral to form DNA-protein cross-
21 links in rat liver nuclei using concentrations 25, 100, or 250 mM. No statistically significant
22 increase in DNA-protein cross-links was observed. DNA and RNA isolated from the [14C]
23 chloral-treated nuclei did not have any detectable 14C bound. However, the proteins from choral-
24 treated nuclei did have a concentration-related binding of 14C.
25
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-63 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-15.. Chloral hydrate genotoxicity: bacterial, yeast and fungal systems
to
VO C?'
I
I
§
***.
S'
1
TO'
I
Test system/endpoint
SOS chromotest, Escherichia coli PQ37
S. typhimurium TMQQ, TA1535, TA98, reverse mutation
S. typhimurium JA^OO, TA1537, TA1538, TA98, reverse mutation
S. typhimurium TMQQ, reverse mutation
S. typhimurium T MOD, reverse mutation
S. typhimurium T MOD, reverse mutation, liquid medium
S. typhimurium T MOD, TA104, reverse mutation
S. typhimurium TM 04, reverse mutation
S. typhimurium TM 535, reverse mutation
S. typhimurium TM 535, TA1537 reverse mutation
S. typhimurium TM 535, reverse mutation
S. typhimurium TA98, reverse mutation
S. typhimurium TA98, reverse mutation
A.nidulans, diploid strain 35X17, mitotic cross-overs
A. nidulans, diploid strain 30, mitotic cross-overs
A. nidulans, diploid strain NH, mitotic cross-overs
A. nidulans, diploid strain P1 , mitotic cross-overs
A. nidulans, diploid strain 35X17, nondisjunctions
A. nidulans, diploid strain 30, aneuploidy
A. nidulans, haploid conidia, aneuploidy, polyploidy
A. nidulans, diploid strain NH, nondisjunctions
A. nidulans, diploid strain P1 , nondisjunctions
Doses
(LED or HID)3
10,000
10,000
1,000
5,000 ug/plate
2,000 ug/plate
300
1 ,000 ug/plate
1 ,000 ug/plate
1,850
6,667
10,000
7,500
10,000 ug/plate
1,650
6,600
1,000
990
825
825
1650
450
660
Results"
With
activation
-
-
+
-
+
+
+
+
-
-
-
-
-
NT
NT
NT
NT
NT
NT
NT
NT
NT
Without
activation
-
-
+
-
+
-
+
+
-
-
-
-
+
-
-
-
-
+
+
+
+
+
Reference
Ciller etal., 1995
Waskell., 1978
Haworthetal., 1983
Leuschnerand Leuschner,
1991
Nietal., 1994
Ciller etal., 1995
Beland, 1999
Nietal., 1994
Leuschnerand Leuschner,
1991
Haworth etal., 1983
Beland, 1999
Haworth etal., 1983
Beland, 1999
Crebelli etal., 1985
Kafer, 1986
Kappas, 1989
Crebelli etal., 1991
Crebelli etal., 1985
Kafer, 1986
Kafer, 1 986
Kappas, 1989
Crebelli etal., 1991
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
^.
S?'
1
TO'
I
Table 4-15. Chloral hydrate genotoxicity: bacterial, yeast and fungal systems (continued)
Test system/endpoint
A. nidulans, haploid strain 35, hyperploidy
S. cerevisiae, meiotic recombination
S. cerevisiae, disomy in meiosis
S. cerevisiae, disomy in meiosis
S. cerevisiae, D61.M, mitotic chr. malsegregation
Drosophila melanogaster, somatic mutation wing spot test
Drosophila melanogaster, induction of sex-linked lethal mutation
Drosophila melanogaster, induction of sex-linked lethal mutation
Doses
(LED or HID)3
2,640
3,300
2,500
3,300
1,000
825
37.2 feed
67.5 inj
Results"
With
activation
NT
NT
NT
NT
NT
Without
activation
+
?
+
+
+
+
?
-
Reference
Crebellietal., 1991
Sora and Agostini Carbone,
1987
Sora and Agostini Carbone,
1987
Sora and Agostini Carbone,
1987
Albertini, 1990
Zordan etal., 1994
Beland, 1999
Beland, 1999
o §
H I
O >
HH Oq
H ^
O
aLED, lowest effective dose; IUD, highest ineffective dose; doses are in ug/mL for in vitro tests; inj = injection.
bResults: + = positive; - = negative; NT = not tested; ? = inconclusive.
^ Table adapted from IARC monograph (2004) and modified/updated for newer references.
H
W
-------
Table 4-16 Chloral hydrate genotoxicity: mammalian systems—all genetic endpoints, in vitro
to
VO Lo'
I
I
§
***.
£3'
1
TO'
I
TO
Test system/endpoint
DNA-protein cross-links, rat nuclei in vitro
DMA single-strand breaks, rat primary hepatocytes in vitro
Gene mutation, mouse lymphoma L5178Y/TK+/", in vitro
Sister chromatid exchange, CHO cells, in vitro
Micronucleus formation, (kinetochore-positive), Chinese hamster C1 cells, in vitro
Micronucleus formation, (kinetochore-negative), Chinese hamster C1 cells, in vitro
Micronucleus formation, (kinetochore-positive), Chinese hamster LUC2 cells, in
vitro
Micronucleus formation, (kinetochore-positive), Chinese hamster LUC2 cells, in
vitro
Micronucleus formation, Chinese hamster V79 cells, in vitro
Micronucleus formation, mouse lymphoma L5178Y/TK+/", in vitro
Micronucleus formation, mouse lymphoma L5178Y/TK+/", in vitro
Chromosomal aberrations, Chinese Hamster CHED cells, in vitro
Chromosomal aberrations, Chinese Hamster ovary cells, in vitro
Chromosomal aberrations, mouse lymphoma L5178Y/TK +/- cells line, in vitro
Aneuploidy, Chinese hamster CHED cells, in vitro
Aneuploidy, primary Chinese hamster embryonic cells, in vitro
Aneuploidy, Chinese hamster LUC2p4 cells, in vitro
Aneuploidy, mouse lymphoma L5178Y/TK+/" , in vitro
Tetraploidy and endoredupliation, Chinese hamster LUC2p4cells, in vitro
Cell transformation, Syrian hamster embryo cells (24-h treatment)
Cell transformation, Syrian hamster dermal cell line (24-h treatment)
DMA single-strand breaks, human lymphoblastoid cells, in vitro
Doses
(LED or
HID)3
41,250
1,650
1,000
100
165
250
400
400
316
1,300
500
20
1,000
1,250
10
250
250
1,300
500
350
50
1,650
Results"
With
activation
NT
NT
+
NT
NT
NT
NT
NT
NT
NT
NT
+
NT
NT
NT
NT
NT
NT
NT
NT
NT
Without
activation
-
-
(+)
+
+
-
+
+
+
-
+
+
+
(+)
+
+
+
-
+
+
+
-
Reference
Keller and Heck, 1988
Chang etal., 1992
Harrington-Brock etal., 1998
Beland, 1999
Degrassi and Tanzarella, 1988
Degrassi and Tanzarella, 1988
Parry etal., 1990
Lynch and Parry, 1993
Seelbach etal., 1993
Harrington-Brock et al., 1998
Nesslany and Marzin, 1999
Furnusetal., 1990
Beland, 1999
Harrington-Brock et al., 1998
Furnusetal., 1990
Natarajan etal., 1993
Warretal., 1993
Harrington-Brock et al., 1998
Warretal., 1993
Gibson etal., 1995
Parry etal., 1996
Chang etal., 1992
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-16. Chloral hydrate genotoxicity: mammalian systems—all genetic endpoints, in vitro (continued)
Test system/endpoint
Gene mutation, tk and hprt locus, human lymphoblastoid
Sister chromatid exchanges, human lymphocytes, in vitro
Micronucleus formation, human lymphocytes, in vitro
Micronucleus formation, human lymphoblastoid AHH-1 cell line, in vitro
Micronucleus formation, human lymphoblastoid MCL-5 cell line, in vitro
Micronucleus formation (kinetochore-positive), human diploid LEO fibroblasts, in
vitro
Aneuploidy (double Y induction), human lymphocytes, in vitro
Aneuploidy (hyperdiploidy and hypodiploidy), human lymphocytes in vitro
Polyploidy, human lymphocytes, in vitro
C-Mitosis, human lymphocytes, in vitro
Doses
(LED or
HID)3
1,000
54
100
100
500
120
250
50
137
75
Results'1
With
activation
NT
NT
-
NT
NT
NT
NT
NT
NT
NT
Without
activation
+
(+)
+
+
-
+
+
+
+
+
Reference
Beland, 1999
Guetal., 1981
Van Hummelen & Kirsch-
Volders, 1992
Parry etal., 1996
Parry etal., 1996
Bonatti et al., 1992
Vagnarelli etal., 1990
Sbrana etal., 1993
Sbrana etal., 1993
Sbrana etal., 1993
H I
O >
HH Oq
H TO
O
aLED, lowest effective dose; HID, highest ineffective dose; doses are in ug/mL for in vitro tests.
bResults: + = positive; (+) = weakly positive in an inadequate study; - = negative; NT = not tested.
Table adapted from IARC monograph (2004) and modified/updated for newer references.
H
W
-------
Table 4-17.. Chloral hydrate genotoxicity: mammalian systems—all genetic damage, in vivo
to
VO Lo'
I
I
§
***.
£3'
1
TO'
Test system/endpoint
DMA single-strand breaks, male Sprague-Dawley rat liver
DMA single-strand breaks, male Fischer 344 rat liver
DMA single-strand breaks, male B6C3F1 mouse liver
DMA single-strand breaks, male B6C3F1 mouse liver
Micronucleus formation, male and female NMRI mice, bone-marrow erythrocytes
Micronucleus formation, BALB/c mouse spermatids
Micronucleus formation, male BALB/c mouse bone-marrow erythrocytes and early
spermatids
Micronucleus formation, male BALB/c mouse bone-marrow erythrocytes
Micronucleus formation, male F1 mouse bone-marrow erythrocytes
Micronucleus formation, C57B1 mouse spermatids
Micronucleus formation, male Swiss CD-1 mouse bone-marrow erythrocytes
Micronucleus formation, B6C3F1 mouse spermatids after spermatogonial stem-cell
treatment
Micronucleus formation, B6C3F1 mouse spermatids after meiotic cell treatment
Micronucleus formation, male F1, BALB/c mouse peripheral-blood erythrocytes
Micronucleus formation, male B6C3F1 mouse bone-marrow erythrocytes
Micronucleus formation, infants, peripheral lymphocytes
Chromosomal aberrations, male and female F1 mouse bone marrow cells
Chromosomal aberrations, male and female Sprague-Dawley rat bone-marrow cells
Chromosomal aberrations, BALB/c mouse spermatogonia treated
Chromosomal aberrations, F1 mouse secondary spermatocytes
Chromosomal aberrations, male Swiss CD-1 mouse bone-marrow erythrocytes
Chromosomal aberrations, ICR mouse oocytes
Micronucleus formation, infants, peripheral lymphocytes
Doses
(LED or HID)3
300, oral
1650, oral
100, oral
825, oral
500, i.p.
83, i.p.
83, i.p.
200, i.p.
400, i.p.
41, i.p.
200, i.p.
165, i.p.
413, i.p.
200, i.p.
500, i.p., x3
50, oral
600, i.p.
1,000, oral
83, i.p.
82.7, i.p.
400, i.p.
600, i.p.
50, oral
Results"
+
-
+
-
-
-
+
+
-
+
+
+
-
+
+
-
-
+
-
+
Reference
Nelson and Bull, 1988
Chang etal., 1992
Nelson and Bull, 1988
Chang etal., 1992
Leuschnerand Leuschner, 1991
Russo and Levis, 1992
Russo and Levis, 1992
Russo etal., 1992
Leopard! etal., 1993
Allen etal., 1994
Marrazini et al., 1994
Nutleyetal., 1996
Nutleyetal., 1996
Grawe etal., 1997
Beland, 1999
Ikbal etal., 2004
Xu and Alder, 1990
Leuschnerand Leuschner, 1991
Russo and Levis, 1992b
Russo etal., 1984
Marrazini et al. 1994
Mailhes etal., 1993
Ikbal etal., 2004
00
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
^.
Co'
1
TO'
I
Table 4-17. Chloral hydrate genotoxicity: mammalian systems—all genetic damage, in vivo (continued)
Test system/endpoint
Polyploidy, male and female F1, mouse bone-marrow cells
Aneuploidy F1 mouse secondary spermatocytes
Aneuploidy, male F1 mouse secondary spermatocytes
Hyperploidy, male Swiss CD-1 mouse bone-marrow erythrocytes
Doses
(LED or HID)3
600, i.p.
200, i.p.
400, i.p.
200, i.p.
Results'1
-
+
+
Reference
Xu and Adler, 1990
Miller and Adler, 1992
Leopard! etal., 1993
Marrazini et al., 1994
H I
O >
HH Oq
H TO
O
aLED, lowest effective dose; HfD, highest ineffective dose; doses are in mg/kg bw for in vivo tests, i.p. = intraperitoneally.
bResults: + = positive; - = negative.
Table adapted from IARC monograph (2004) and modified/updated for newer references.
H
W
-------
1 4.2.4.2. Bacterial and Fungal Systems—Gene Mutations
2 Chloral hydrate induced gene mutations in S. typhimurium TA100 and TA104 strains, but
3 not in most other strains assayed. Four of six studies of chloral hydrate exposure in
4 S. typhimurium TA100 and two of two studies in S. typhimurium TA104 were positive for
5 revertants (Haworth et al., 1983; Ni et al., 1994; Ciller et al., 1995; Beland, 1999). Waskell
6 (1978) studied the effect of chloral hydrate along with TCE and its other metabolites. Chloral
7 hydrate was tested at different doses (1.0-13 mg/plate) in different S. typhimurium strains
8 (TA98, TA100, TA1535) for gene mutations using Ames assay. No revertant colonies were
9 observed in strains TA98 or TA1535 both in the presence and absence of S9 mix. Similar results
10 were obtained by Leuschner and Leuschner (1991). However, in TA100, a dose-dependent
11 statistically significant increase in revertant colonies was obtained both in the presence and
12 absence of S9. It should be noted that chloral hydrate that was purchased from Sigma was re-
13 crystallized from one to six times from chloroform and the authors describe this as crude chloral
14 hydrate. However, this positive result is consistent with other studies in this strain as noted
15 above. Furthermore, Ciller et al. (1995) studied chloral hydrate genotoxicity in three short-term
16 tests. Chloral-induced mutations in strain TA100 of S. typhimurium (fluctuation test). Similar
17 results were obtained by Haworth et al. (1983). These are consistent with several studies of
18 TCE, in which low, but positive responses were observed in the TA100 strain in the presence of
19 S9 metabolic activation, even when genotoxic stabilizers were not present.
20 A significant increase in mitotic segregation was observed in Aspergillus nidulans when
21 exposed to 5 and 10 mM chloral hydrate (Crebelli et al., 1985). Studies of mitotic crossing-over
22 in Aspergillus nidulans have been negative while these same studies were positive for
23 aneuploidy (Crebelli et al., 1985, 1991; Kafer, 1986; Kappas, 1989).
24 Two studies were conducted in Saccharomyces cerevisiae to understand the
25 chromosomal malsegregation as a result of exposure to chloral hydrate (Sora and Agostini, 1987;
26 Albertini, 1990). Chloral hydrate (1-25 mM) was dissolved in sporulation medium and the
27 frequencies of various meiotic events such as recombination, disomy were analyzed. Chloral
28 hydrate inhibited sporulation as a function of dose and increased diploid and disomic clones .
29 Chloral hydrate was also tested for mitotic chromosome malsegregation using Saccharomyces
30 cerevisiae D61.M (Albertini, 1990). The tester strain was exposed to a dose range of
31 1-8 mg/mL. An increase in the frequency of chromosomal malsegregation was observed as a
32 result of exposure to chloral hydrate.
33 Limited analysis of chloral hydrate mutagenicity has been performed in Drosophila
34 (Zordan et al., 1994; Beland, 1999). Of these two studies, chloral hydrate was positive in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-70 DRAFT—DO NOT CITE OR QUOTE
-------
1 somatic mutation wing spot test (Zordan et al., 1994), equivocal in the induction of sex-linked
2 lethal mutation when in feed but negative when exposed via injection (Beland, 1999).
3
4 4.2.4.3. Mammalian Systems
5 4.2.4.3.1. Gene mutations. Harrington-Brock (1998) noted that chloral hydrate-induced
6 concentration related cytotoxicity in TK+/- mouse lymphoma cell lines without S9 activation. A
7 nonstatistical increase in mutant frequency was observed in cells treated with chloral hydrate.
8 The mutants were primarily small colony TK mutants, indicating that most chloral hydrate-
9 induced mutants resulted from chromosomal mutations rather than point mutations. It should be
10 noted that in most concentrations tested (350-1,600 ug/mL), cytotoxicity was observed. Percent
11 cell survival ranged from 96 to 4%.
12
13 4.2.4.3.2. Micronucleus. Micronuclei induction following exposure to chloral hydrate is
14 positive in most test systems in both in vitro and in vivo assays, although some negative tests do
15 also exist (Harrington-Brock et al., 1998; Degrassi and Tanzarella, 1988; Beland, 1999; Lynch
16 and Parry, 1993; Seelbach et al., 1993; Marrazini et al., 1994; Nesslany and Marzin, 1999; Russo
17 and Levis, 1992a, b; Russo et al., 1992; Leopardi et al., 1993; Allen et al., 1994; Nutley et al.,
18 1996; Grawe et al., 1997; Giller et al., 1995; Leuschner and Leuschner, 1991; Van Hummelen
19 and Kirsch-Volders, 1992; Parry et al., 1996; Bonatti et al., 1992; Ikbal et al., 2004). Some
20 studies have attempted to make inferences regarding aneuploidy induction or clastogenicity as an
21 effect of chloral hydrate. Aneuploidy results from defects in chromosome segregration during
22 mitosis and is a common cytogenetic feature of cancer cells (see Section E.3.1.5).
23 Giller et al. (1995) studied chloral hydrate genotoxicity in three short-term tests. Chloral
24 hydrate caused a significant increase in the frequency of micrenucleated erythrocytes following
25 in vivo exposure of the amphibian Pleurodeles waltl newt larvae.
26 Chloral hydrate induced aneuploidy in vitro in multiple Chinese hamster cell lines
27 (Warr et al., 1993; Furnus et al., 1990; Natarajan et al., 1993) and human lymphocytes
28 (Vagnarelli et al., 1990; Sbrana et al., 1993) but not mouse lymphoma cells
29 (Harrington-Brock et al., 1998). In vivo studies performed in various mouse strains led to
30 increased aneuploidy in spermatocytes (Russo et al., 1984; Liang and Pacchierotti, 1988;
31 Miller and Adler, 1992) but not oocytes (Mailhes et al., 1988) or bone marrow cells (Xu and
32 Adler, 1990; Leopardi et al., 1993).
33 The potential of chloral hydrate to induce aneuploidy in mammalian germ cells has been
34 of particular interest since Russo et al. (1984) first demonstrated that chloral hydrate treatment of
35 male mice results in significant increase in frequencies of hyperploidy in metaphase II cells.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-71 DRAFT—DO NOT CITE OR QUOTE
-------
1 This hyperploidy was thought to have arisen from chromosomal nondisjunction in
2 premeiotic/meiotic cell division and may be a consequence of chloral hydrate interfering with
3 spindle formation (reviewed by Russo et al. [1984] and Liang and Brinkley [1985]). Chloral
4 hydrate also causes meiotic delay, which may be associated with aneuploidy (Miller and
5 Alder, 1992). Chloral hydrate has been shown to induce micronuclei but not structural
6 chromosomal aberrations in mouse bone-marrow cells. Micronuclei induced by nonclastogenic
7 agents are generally believed to represent intact chromosomes that failed to segregate into either
8 daughter-cell nucleus at cell division (Russo et al., 1992; Wang Xu and Adler, 1990).
9 Furthermore, chloral hydrate-induced micronuclei in mouse bone-marrow cells (Russo et al.,
10 1992) and in cultured mammalian cells (Degrassi and Tanzarella, 1988; Bonatti et al., 1992)
11 have shown to be predominantly kinetochore-positive in composition upon analysis with
12 immunofluorescent methods. The presence of a kinetochore in a micronucleus is considered
13 evidence that the micronucleus contains a whole chromosome lost at cell division (Degrassi and
14 Tanzarella, 1988; Hennig et al., 1988; Eastmond and Tucker, 1989). Therefore, both TCE and
15 chloral hydrate appear to increase the frequency of micronuclei.
16 Allen et al. (1994) treated male C57B1/6J mice were given a single intraperitoneal
17 injection of 0, 41, 83, or 165 mg/kg chloral hydrate. Spermatids were harvested at 22 hours, 11,
18 13.5, and 49 days following exposure (Allen et al., 1994). Harvested spermatids were processed
19 to identify both kinetochore-positive micronucleus (aneugen) and kinetochore-negative
20 micronucleus (clastogen). All chloral hydrate doses administered 49 days prior to cell harvest
21 were associated with significantly increased frequencies of kinetochore-negative micronuclei in
22 spermatids, however, dose dependence was not observed. This study is in contrast with other
23 studies (Degrassi and Tanzarella, 1988; Bonatti et al., 1992) who demonstrated predominantly
24 kinetochore-positive micronucleus.
25 The ability of chloral hydrate to induce aneuploidy and polyploidy was tested in human
26 lymphocyte cultures established from blood samples obtained from two healthy nonsmoking
27 donors (Sbrana et al., 1993). Cells were exposed for 72 and 96 hours at doses between 50 and
28 250 |ig/mL. No increase in percent hyperdiploid, tetraploid, or endoreduplicated cells were
29 observed when cells were exposed to 72 hours at any doses tested. However, at 96 hours of
30 exposure, significant increase in hyperdiploid was observed at one dose (150 ug/mL) and was
31 not dose dependent. Significant increase in tetraploid was observed at dose 137 mg/mL, again,
32 no dose dependence was observed.
33 Ikbal et al. (2004) assessed the genotoxic effects in cultured peripheral blood
34 lymphocytes of 18 infants (age range of 31-55 days) before and after administration of a single
35 dose of chloral hydrate (50 mg/kg of body weight) for sedation before a hearing test for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-72 DRAFT—DO NOT CITE OR QUOTE
-------
1 micronucleus frequency. A significant increase in micronuclei frequency was observed after
2 administration of chloral hydrate.
O
4 4.2.4.3.3. Chromosomal aberrations. Several studies have included chromosomal aberration
5 analysis in both in vitro and in vivo systems exposed to chloral hydrate and have resulted in
6 positive in in vitro studies—although not all studies had statistically significant increase
7 (Furnus et al., 1990; Beland, 1999; Harrington-Brock et al., 1998).
8 Analysis of chloral hydrate treated mouse lymphoma cell lines for chromosomal
9 aberrations resulted in a nonsignificant increase in chromosomal aberrations
10 (Harrington-Brock et al., 1998). However, it should be noted that the concentrations tested
11 (1,250 and 1,300 |ig/mL) were cytotoxic (with a cell survival of 11 and 7%, respectively).
12 Chinese hamster embryo cells were also exposed to 0.001, 0.002, and 0.003% chloral hydrate for
13 1.5 hours (Furnus et al., 1990). A nonstatistically significant increase in frequency of
14 chromosomal aberrations was observed only 0.002 and 0.003% concentrations, with the increase
15 not dose-dependent. In this study, it should be noted that the cells were only exposed for
16 1.5 hours to chloral hydrate and cells were allowed to grow for 48 hours (two cell cycles) to
17 obtain similar mitotic index before analyzing for chromosomal aberrations. No information on
18 cytotoxicity was provided except that higher doses decreased the frequency of mitotic cells at the
19 time of fixation.
20 In vivo chromosome aberration studies have mostly reported negative or null results (Xu
21 and Adler, 1990; Leuschner and Leuschner, 1991; Russo and Levis, 1992a, b; Liang and
22 Pacchierotti, 1988; Mailhes et al., 1993) with the exception of one study (Russo et al., 1984) in
23 an Fl cross of mouse strain between C57Bl/Cne x C3H/Cne.
24
25 4.2.4.3.4. Sister chromatid exchanges (SCEs). SCEs were assessed by Ikbal et al. (2004) in
26 cultured peripheral blood lymphocytes of 18 infants (age range of 31-55 days) before and after
27 administration of a single dose of chloral hydrate (50 mg/kg of body weight) for sedation before
28 a hearing test. The authors report a significant increase in the mean number of SCEs, from
29 before administration (7.03 ± 0.18 SCEs/cell) and after administration (7.90 ± 0.19 SCEs/cell) ,
30 with each of the 18 individuals showing an increase with treatment. Micronuclei were also
31 significantly increased. SCEs were also assessed by Gu et al. (1981a) in human lymphocytes
32 exposed in vitro with inconclusive results, although positive results were observed by Beland
33 (1999) in Chinese hamster ovary cells exposed in vitro with and without an exogenous metabolic
34 system.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-73 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.2.4.3.5. Cell Transformation. Chloral hydrate was positive in the two studies designed to
2 measure cellular transformation (Gibson et al., 1995; Parry et al., 1996). Both studies exposed
3 Syrian hamster cells (embryo and dermal) to chloral hydrate and induced cellular transformation.
4
5 4.2.4.4. Summary
6 Chloral hydrate has been reported to induce micronuclei formation, aneuploidy, and
7 mutations in multiple in vitro systems and in vivo. In vivo studies have limited results to an
8 increased micronuclei formation mainly in mouse spermatocytes. CH is positive to in some
9 studies in in vitro genotoxicity assays that detect point mutations, micronuclei induction,
10 chromosomal aberrations, and/or aneuploidy. The in vivo data exhibit mixed results (Xu and
11 Adler, 1990; Russo et al., 1992; Mailhes et al., 1993; Allen et al., 1994; Alder, 1993; Nutley et
12 al., 1996; Leuschner and Beuscher, 1998). Most of the positive studies show that chloral hydrate
13 induces aneuploidy. Based on the existing array of data, CH has the potential to be genotoxic,
14 particularly when aneuploidy is considered in the weight of evidence for genotoxic potential.
15 Some have suggested that chloral hydrate may act through a mechanism of spindle poisoning and
16 resulting in numerical changes in the chromosomes, but some data also suggest induction of
17 chromosomal aberrations. These results are consistent with TCE, albeit there are more limited
18 data on TCE for these genotoxic endpoints.
19
20 4.2.5. Dichlorovinyl Cysteine (DCVC) and S-Dichlorovinyl Glutathione (DCVG)
21 DCVC and DCVG have been studied for their genotoxic potential; however, since there
22 is limited number of studies to evaluate them based on each endpoint, particularly in mammalian
23 systems, the following section has been combined to include all the available studies for different
24 endpoints of genotoxicity. Study details can be found in Table 4-18.
25 DCVC and DCVG, cysteine intermediates of TCE formed by the GST pathway, are
26 capable of inducing point mutations as evidenced by the fact that they are positive in the Ames
27 assay. Dekant et al. (1986) demonstrated mutagenicity of DCVC in S. typhimurium strains
28 (TA100, TA2638, and TA98) using the Ames assay in the absence of S9. The effects were
29 decreased with the addition of a beta-lyase inhibitor aminooxyacetic acid, suggesting that
30 bioactivation by this enzyme plays a role in genotoxicity. Vamvakas et al. (1987) tested
31 7V-acetyl-S-(l,2-dichlorovinyl)-L-cysteine) (NAcDCVC) for mutagenicity following addition of
32 rat kidney cytosol and found genotoxic activity. Furthermore, Vamvakas (1988a), in another
33 experiment, investigated the mutagenicity of DCVG and DCVC in S. typhimurium strain
34 TA2638, using kidney subcellular fractions for metabolic activation and AOAA (a beta-lyase
35 inhibitor) to inhibit genotoxicity. DCVG and DCVC both exhibited direct-acting mutagenicity,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-74 DRAFT—DO NOT CITE OR QUOTE
-------
1 with kidney mitochondria, cytosol, or microsomes enhancing the effects for both compounds and
2 AOAA diminishing, but not abolishing the effects. Importantly, addition of liver subcellular
3 fractions did not enhance the mutagenicity of DCVG, consistent with in situ metabolism playing
4 a significant role in the genotoxicity of these compounds in the kidney.
5 While additional data are not available on DCVG or NAcDCVC, the genotoxicity of
6 DCVC is further supported by the predominantly positive results in other available in vitro and
7 in vivo assays. Jaffe et al. (1985) reported DNA strand breaks due to DCVC administered in
8 vivo, in isolated perfused kidneys, and in isolated proximal tubules of albino male rabbits.
9 Vamvakas et al. (1989) reported dose-dependent increases in unscheduled DNA synthesis in
10 LLC-PK1 cell clones at concentrations without evidence of cytotoxicity. In addition,
11 Vamvakas et al. (1996) reported that 7-week DCVC exposure to LLC-PK1 cell clones at
12 noncytotoxic concentrations induces morphological and biochemical de-differentiation that
13 persists for at least 30 passages after removal of the compound. This study also reported
14 increased expression of the proto-oncogene c-fos in the cells in this system. In a Syrian hamster
15 embryo fibroblast system, DCVC did not induce micronuclei, but demonstrated an unscheduled
16 DNA synthesis response (Vamvakas et al., 1988b).
17 Two more recent studies are discussed in more detail. Mally et al. (2006) isolated
18 primary rat kidney epithelial cells from Tsc-2Ek/+ (Eker) rats, and reported increased
19 transformation when exposed to 10 jiM DCVC, similar to that of the genotoxic renal carcinogens
20 TV-methyl-TV-nitro-jV-nitrosoguanidine (Horesovsky et al., 1994). The frequency was variable
21 but consistently higher than background. No loss-of-heterozygosity (LOH) of the Tsc-2 gene
22 was reported either in these DCVC transformants or in renal tumors (which were not increased in
23 incidence) from TCE-treated Eker rats, which Mally et al. (2006) suggested support a
24 nongenotoxic mechanism because a substantial fraction of spontaneous renal tumors in Eker rats
25 showed LOH at this locus (Kubo et al., 1994, Yeung et al., 1995) and because LOH was
26 exhibited both in vitro and in vivo with 2,3,4-tris(glutathion-S-yl)-hydroquinone treatment in
27 Eker rats (Yoon et al., 2001). However, 2,3,4-tris(glutathion-S-yl)-hydroquinone is not
28 genotoxic in standard mutagenicity assays (Yoon et al., 2001), and Kubo et al. (1994) also
29 reported that none of renal tumors induced by the genotoxic carcinogen 7V-ethyl-7V-nitrosourea
30 showed LOH. Therefore, the lack of LOH at the Tsc-2 locus induced by DCVC in vitro, or TCE
31 in vivo, reported by Mally et al. (2006) is actually more similar to the response from the
32 genotoxic carcinogen 7V-ethyl-7V-nitrosourea than the nongenotoxic carcinogen
33 2,3,4-tris(glutathion-S-yl)-hydroquinone. Therefore, these data do not substantially contradict
34 the body of evidence on DCVC genotoxicity.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-75 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
p
"I
I
§
^.
£3'
a
1
^t>
>
i
TO'
^5
I
^1
•
-------
Table 4-18. TCE GSR conjugation metabolites genotoxicity (continued)
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
DMA strand breaks
Male rabbit renal tissue (perfused
kidneys and proximal tubules)
Primary kidney cells from both male
rats and human
In vivo — male Sprague-Dawley rats
exposed to TCE or DCVC — comet
assay
0-100 mg/kg or
10uMto 10 mM
1-4mM;20h
exposure
TCE: 500-2,000
ppm, inhalation, 6
h/d,5d
DCVC: 1 or 10
mg/kg, single oral
dose for 16 h
ND
NA
+ (DCVC)
- (TCE)
+
+
NA
Dose dependent increase SB in both i.v.
and i.p. injections (i.v. injections were
done only for 1 0 and 20 mg/kg).
Perfusion of rabbit kidney (45 min
exposure) and proximal tubules (30 min
exposure) expt. Resulted in a dose
dependent difference in the amount of
single strand breaks.
Statistically significant increase in all
doses (1 , 2, or 4 mM) both in rats and
human cells.
No significant increase in tail length in
any of the TCE exposed groups. In
Expt. 1 . 2 h exposure — 1 or 10 mg to
DCVC resulted in significant increase
with no dose response, but not at 16 h.
In Expt. 2. ND for 1 mg, significant
increase at 10 mg.
Jaffeet.al., 1985
Robbiano, 2004
Clay, 2008
Micronucleus
Syrian hamster embryo fibroblasts
Primary kidney cells from both male
rats and human
Male Sprague-Dawley rats; proximal
tubule cells (in vivo)
1-4mM;20h
exposure
4 mM/kg TCE
exposure, single
dose
NA
NA
NA
-
+
+
No micronucleus formation.
Statistically significant increase in all
doses (1 , 2, and 4 mM) both in rats and
human cells.
Statistically significant increase in the
average frequency of micronucleated
kidney cells was observed.
Vamvakas et al.,
1988b
Robbiano, 2004
Robbiano et al.,
1998
-------
Table 4-18. TCE GSH conjugation metabolites genotoxicity (continued)
Test system/endpoint
Doses tested
With
activation
Without
activation
Comments
References
Cell transformation
Kidney tubular epithelial cell line
(LLC-PK1)
Rat kidney epithelial cells (in vitro)
1or5 uM; 7 wks
10uM;24h
exposure, 7 wks
post incubation
NA
NA
+
+
Induced morphological cell
transformation at both concentrations
tested. Furthermore, cells maintained
both biochemical and morphological
alterations remained stable for 30
passages
Cell transformation was higher than
control, however, cell survival percent
ranged from 39-64% indicating
cytotoxicity
Vamvakas et al.,
1996
Mallyetal.,2006
Gene expression
Kidney tubular epithelial cell line
(LLC-PK1)
Kidney tubular epithelial cell line
(LLC-PK1)
1or5 uM clones,
30, 60, 90 min
NA
NA
+
+
Increased c-fos expression in 1 and
5 uM exposed clones at three different
times tested
Expression of c-fos and c-myc increased
in a time-dependent manner
Vamvakas et al.,
1996
Vamvakas et al.,
1993
i.v. = intravenous, LDH = lactate dehydrogenase, LOH = loss of heterozygosity, ND = not determined, NA = not applicable.
-------
1 Finally, Clay (2008) evaluated the genotoxicity of DCVC in vivo using the comet assay
2 to assess DNA breakage in the proximal tubules of rat kidneys. Rats were exposed orally to a
3 single dose of DCVC (1 or 10 mg/kg). The animals were sacrificed either 2 or 16 hours after
4 dosing and samples prepared for detecting the DNA damage. DCVC (1 and 10 mg/kg) induced
5 no significant DNA damage in rat kidney proximal tubules at the 16-hour sampling time or after
6 1 mg/kg DCVC at the 2-hour sampling time. While Clay et al. (2008) concluded that these data
7 were insufficient to indicate a positive response in this assay, the study did report a statistically
8 significant increase in percent tail DNA 2 hours after treatment with 10 mg/kg DCVC, despite
9 the small number of animals at each dose (n = 5) and sampling time. Therefore, these data do
10 not substantially contradict the body of evidence on DCVC genotoxicity.
11 Overall, DCVC, and to a lesser degree DCVG and NAcDCVC, have demonstrated
12 genotoxicity based on consistent results in a number of available studies. While some recent
13 studies (Mally et al., 2006; Clay, 2008) have reported a lack of positive responses in some in vivo
14 measures of genotoxicity with DCVC treatment, due to a number of limitations discussed above,
15 these studies do not substantially contradict the body of evidence on DCVC genotoxicity. It is
16 known that these metabolites are formed in vivo following TCE exposure, specifically in the
17 kidney, so they have the potential to contribute to the genotoxicity of TCE, especially in that
18 tissue. Moreover, DCVC and DCVG genotoxic responses were enhanced when metabolic
19 activation using kidney subcellular fractions was used (Vamvakas et al., 1988a). Finally, the lack
20 of similar responses in in vitro genotoxicity assays with TCE, even with metabolic activation, is
21 likely the result of the small yield (if any) of DCVC under in vitro conditions, since in vivo,
22 DCVC is likely formed predominantly in situ in the kidney while S9 fractions are typically
23 derived from the liver. This hypothesis could be tested in experiments in which TCE is
24 incubated with subcellular fractions from the kidney, or from both the kidney and the liver (for
25 enhanced GSH conjugation).
26
27 4.2.6. Trichloroethanol (TCOH)
28 Limited studies are available on the effect of TCOH on genotoxicity (Table 4-19).
29 TCOH is negative in the S. typhimurium assay using the TA100 strain (Bignami et al., 1980;
30 DeMarini et al., 1994; Waskell, 1978). A study by Beland (1999) using S. typhimurium strain
31 TA104 did not induce reverse mutations without exogenous metabolic activation, however did
32 increase mutant frequency in the presence of exogenous metabolic activation at a dose above
33 2,500 jig/plate. TCOH has not been evaluated in the other recommended screening assays.
34 Therefore, the database is limited for the determination of TCOH genotoxicity.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-79 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-19.. Genotoxicity of trichloroethanol
Test system/endpoint
S. typhimuriumTMQQ, 98,
reverse mutation
S. typhimuriumTMQQ, reverse
mutation
S. typhimuriumTMQ4, reverse
mutation
S. typhimuriumJAlOO, 1535
reverse mutation
Sister chromatid exchanges
Doses
(LED or HID)3
7,500 ug/plate
0.5 ug/cm3 vapor
2,500 ug/plate
NA
NA
Results'1
With
activation
-
-
+
-
NA
Without
activation
-
-
-
-
+
Reference
Waskell, 1978
DeMarini etal., 1994
Beland, 1999
Bignami etal., 1980
Guetal., 1981 b
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
aLED, lowest effective dose; HID, highest ineffective dose.
bResults: + = positive; - = negative; NA = doses not available, results based on the abstract.
4.2.7. Synthesis and Overall Summary
Trichloroethylene and its metabolites (TCA, DCA, CH, DCVC, DCVG, and TCOH) have
been evaluated to varying degrees for their genotoxic activity in several of in vitro systems such
as bacteria, yeast, and mammalian cells and, also, in in vivo systems.
There are several challenges in interpreting the genotoxicity results obtained from TCE
exposure. For example, some studies in bacteria should be interpreted with caution if conducted
using technical grade TCE since it may contain known bacterial mutagens in trace amounts as
stabilizers (e.g., 1,2-epoxybutane and epichlorohydrin). Because of the volatile nature of TCE,
there could be false negative results if proper precautions are not taken to limit evaporation,
such as the use of a closed sealed system. The adequacy of the enzyme-mediated activation of
TCE in vitro tests is another consideration. For example, it is not clear if standard S9 fractions
can adequately recapitulate the complex in vivo metabolism of TCE to reactive intermediates,
which in some cases entails multiple sequential steps involving multiple enzyme systems (e.g.,
CYP, GST, etc.) and interorgan processing (as is described in more detail in Section 3.3). In
addition, the relative potency of the metabolites in vitro may not necessarily inform their relative
contribution to the overall mechanistic effects of the parent compound, TCE. Furthermore,
although different assays provided data relevant to different types of genotoxic endpoints, not all
effects that are relevant for carcinogenesis are encompassed. The standard battery of prokaryotic
as well as mammalian genotoxicity test protocols typically specify the inclusion of significantly
cytotoxic concentrations of the test compound.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-80 DRAFT—DO NOT CITE OR QUOTE
-------
1 With respect to potency, several TCE studies have been conducted along with numerous
2 other chlorinated compounds and the results interpreted as a comparison of the group of
3 compounds tested (relative potency). However, for the purposes of hazard characterization, such
4 comparisons are not informative - particularly if they are not necessarily correlated with in vivo
5 carcinogenic potency. Also, differentiating the effects of TCE with respect to its potency can be
6 influenced by many factors such as the type of cells, their differing metabolic capacities,
7 sensitivity of the assay, need for greater concentration to show any effect, interpretation of data
8 when the effects are marginal, and gradation of severity of effects.
9 Also, type of samples used, methodology used for the isolation of genetic material, and
10 duration of exposure can particularly influence the results of several studies. This is particularly
11 true for human epidemiological studies. For example, while some studies use tissues obtained
12 directly from the patients others use formalin fixed tissues sections to isolate DNA for mutation
13 detection. Type of fixing solution, fixation time, and period of storage of the tissue blocks often
14 affect the quality of DNA. Formic acid contained in the formalin solution or picric acid
15 contained in Bouin's solution is known to degrade nucleic acids resulting in either low yield or
16 poor quality of DNA. In addition, during collection of tumor tissues, contamination of
17 neighboring normal tissue can easily occur if proper care is not exercised. This could lead to the
18 'dilution effect' of the results, i.e., because of the presence of some normal tissue; frequency of
19 mutations detected in the tumor tissue can be lower than expected. Due to some of these
20 technical difficulties in obtaining proper material (DNA) for the detection of mutation, the results
21 of these studies should be interpreted cautiously.
22 The following synthesis, summary, and conclusions focus on the available studies that
23 may provide some insight into the potential genotoxicity of TCE considering the above
24 challenges when interpreting the mutagenicity data for TCE.
25 Overall, evidence from a number of different analyses and a number of different
26 laboratories using a fairly complete array of endpoints suggests that TCE, following metabolism,
27 has the potential to be genotoxic. TCE has a limited ability to induce mutation in bacterial
28 systems, but greater evidence of potential to bind or to induce damage in the structure of DNA or
29 the chromosome in a number of targets. A series of carefully controlled studies evaluating TCE
30 itself (without mutagenic stabilizers and without metabolic activation) found it to be incapable of
31 inducing gene mutations in most standard mutation bacterial assays (Waskell, 1978;
32 Henschler et al., 1977; Mortelmans et al., 1986; Simmon et al., 1977; Baden et al., 1979;
33 Bartsch et al., 1979; Crebelli et al., 1982; Shimada et al., 1985; Simmon et al., 1977; Baden et
34 al., 1979). Therefore, it appears that it is unlikely that TCE is a direct-acting mutagen, though
35 TCE has shown potential to affect DNA and chromosomal structure. TCE is also positive in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-81 DRAFT—DO NOT CITE OR QUOTE
-------
1 some but not all fungal and yeast systems (Crebelli et al., 1985; Koch et al., 1988; Rossi et al.,
2 1983; Callen et al., 1980). Data from human epidemiological studies support the possible
3 mutagenic effect of TCE leading to VHL gene damage and subsequent occurrence of renal cell
4 carcinoma. Association of increased VHL mutation frequency in TCE-exposed renal cell
5 carcinoma cases has been observed (Briining et al.,1997; Brauch et al., 1999, 2004).
6 TCE can lead to binding to nucleic acids and proteins (Di Renzo et al., 1982; Bergman,
7 1983; Miller and Guengerich, 1983; Mazzullo et al., 1992; Kautiainen et al., 1997), and such
8 binding appears to be due to conversion to one or more reactive metabolites. For instance,
9 increased binding was observed in samples bioactivated with mouse and rat microsomal fractions
10 (Banerjee and VanDuuren, 1978; Di Renzo et al., 1982; Miller and Guengerich, 1983;
11 Mazzullo et al., 1992). DNA binding is consistent with the ability to induce DNA and
12 chromosomal perturbations. Several studies report the induction of micronuclei in vitro and in
13 vivo from TCE exposure (Kligerman et al., 1994; Hrelia et al., 1994; Wang et al., 2001;
14 Robbiano et al., 2004; Hu et al., 2008). Reports of SCE induction in some studies are consistent
15 with DNA effects, but require further study (White et al., 1979; Gu et al., 1981a, b; Nagaya et al.,
16 1989; Kligerman et al., 1994).
17 TCA, an oxidative metabolite of TCE, exhibits little, if any genotoxic activity in vitro.
18 TCA did not induce mutations in S. typhimurium strains in the absence of metabolic activation or
19 in an alternative protocol using a closed system (Waskell, 1978; Rapson et al., 1980; DeMarini et
20 al., 1994; Giller et al., 1997; Nelson et al., 2001; Kargalioglu et al., 2002) but a mutagenic
21 response was induced in TA100 in the Ames fluctuation test (Giller et al., 1997). However, in
22 vitro experiments with TCA should be interpreted with caution if steps have not been taken to
23 neutralize pH changes caused by the compound (Mackay, 1995). Measures of DNA-repair
24 responses in bacterial systems have shown induction of DNA repair reported in S. typhimurium
25 but not in E. coli. Mutagenicity in mouse lymphoma cells was only induced at cytotoxic
26 concentrations (Harrington-Brock et al., 1998). TCA was positive in some genotoxicity studies
27 in vivo mouse, newt, and chick test systems (Bhunya and Behera, 1987; Bhunya and Jena, 1996;
28 Birner et al., 1994; Giller et al., 1997). DNA unwinding assays have either shown TCA to be
29 much less potent than DCA (Nelson and Bull, 1988) or negative (Nelson et al., 1989; Styles et
30 al., 1991). Due to limitations in the genotoxicity database, the possible contribution of TCA to
31 TCE genotoxicity is unclear.
32 DCA, a chloroacid metabolite of TCE, has also been studied using different types of
33 genotoxicity assays. Although limited studies are conducted for different genetic endpoints,
34 DCA has been demonstrated to be mutagenic in the S. typhimurium assays, in vitro
35 (DeMarini et al., 1994; Kargalioglu et al., 2002; Plewa et al., 2002) in some strains, mouse
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-82 DRAFT—DO NOT CITE OR QUOTE
-------
1 lymphoma assay, (Harrington-Brock et al., 1998) in vivo cytogenetic tests (Leavitt et al., 1997;
2 Fuscoe et al., 1996), the micronucleus induction test, the Big Blue mouse system, and other tests
3 (Bignami et al., 1980; Chang et al., 1989; DeMarini et al., 1994; Leavitt et al., 1997;
4 Fuscoe et al., 1996; Nelson and Bull, 1988; Nelson et al., 1989; Harrington-Brock et al., 1998).
5 DCA can cause DNA strand breaks in mouse and rat liver cells following in vivo mice and rats
6 (Fuscoe et al., 1996). Because of uncertainties as to the extent of DCA formed from TCE
7 exposure, inferences as to the possible contribution from DCA genotoxicity to TCE toxicity are
8 difficult to make.
9 Chloral hydrate is mutagenic in the standard battery of screening assays. Effects include
10 positive results in bacterial mutation tests for point mutations and in the mouse lymphoma assay
11 for mutagenicity at the Tk locus (Haworth et al., 1983). In vitro tests showed that CH also
12 induced micronuclei and aneuploidy in human peripheral blood lymphocytes and Chinese
13 hamster pulmonary cell lines. Micronuclei were also induced in Chinese hamster embryonic
14 fibroblasts. Several studies demonstrate that chloral hydrate induces aneuploidy (loss or gain of
15 whole chromosomes) in both mitotic and meiotic cells, including yeast (Singh and Sinha, 1976,
16 1979; Kafer, 1986; Gualandi, 1987; Sora and Agostini-Carbone, 1987), cultured mammalian
17 somatic cells (Degrassi and Tanzarella, 1988), and spermatocytes of mice (Russo et al., 1984;
18 Liang and Pacchierotti, 1988). Chloral hydrate was negative for sex-linked recessive lethal
19 mutations in Drosophila (Yoon et al., 1985). It induces SSB in hepatic DNA of mice and rats
20 (Nelson and Bull, 1988) and mitotic gene conversion in yeast (Bronzetti et al., 1984). Schatten
21 and Chakrabarti (1998) showed that chloral hydrate affects centrosome structure, which results
22 in the inability to reform normal microtubule formations and causes abnormal fertilization and
23 mitosis of sea urchin embryos. Based on the existing array of data, CH has the potential to be
24 genotoxic, particularly when aneuploidy is considered in the weight of evidence for genotoxic
25 potential. Chloral hydrate appears to act through a mechanism of spindle poisoning and resulting
26 in numerical changes in the chromosomes. These results are consistent with TCE, albeit there
27 are limited data on TCE for these genotoxic endpoints.
28 DCVC, and to a lesser degree DCVG, has demonstrated bacterial mutagenicity based on
29 consistent results in a number of available studies (Dekant et al., 1986; Vamvakas et al., 1987;
30 Vamvakas, 1988a). DCVC has demonstrated a strong, direct-acting mutagenicity both with and
31 without the presence of mammalian activation enzymes. It is known that these metabolites are
32 formed in vivo following TCE exposure, so they have the potential to contribute to the
33 genotoxicity of TCE. The lack of similar response in bacterial assays with TCE is likely the
34 result of the small yield (if any) of DCVC under in vitro conditions, since in vivo, DCVC is
35 likely formed predominantly in situ in the kidney (S9 fractions are typically derived from the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-83 DRAFT—DO NOT CITE OR QUOTE
-------
1 liver). DCVC and DCVG have not been evaluated extensively in other genotoxicity assays, but
2 the available in vitro and in vivo data are predominantly positive. For instance, several studies
3 have reported the DCVC can induce primary DNA damage in mammalian cells in vitro and in
4 vivo (Jaffe et al., 1985; Vamvakas et al., 1989; Clay, 2008). Long-term exposure to DCVC
5 induced de-differentiation of cells (Vamavakas et al., 1996). It has been shown to induce
6 expression of the protooncogene c-fos (Vamvakas et al., 1996) and cause cell transformation in
7 rat kidney cells (Mally et al., 2006). In LLC-PK1 cell clones, DCVC was reported in induce
8 unscheduled DNA synthesis, but not micronuclei (Vamvakas et al., 1988b). Finally, DCVC
9 induced transformation in kidney epithelial cells isolated from Eker rats carrying the
10 heterozygous Tsc-2 mutations (Mally et al., 2006). Moreover, the lack of LOH at the Tsc-2 locus
11 observed in exposed cells does not constitute negative evidence of DCVC genotoxicity, as none
12 of renal tumors induced in Eker rats by the genotoxic carcinogen TV-ethyl-TV-nitrosourea showed
13 LOH (Kuboetal., 1994).
14 In support of the importance of metabolism, there is some concordance between effects
15 observed from TCE and those from several metabolites. For instance, both TCE and chloral
16 hydrate have been shown to induce micronucleus in mammalian systems, but chromosome
17 aberrations have been more consistently observed with chloral hydrate than with TCE. The role
18 of TCA in TCE genotoxicity is less clear, as there is less concordance between the results from
19 these two compounds. Finally, several other TCE metabolites show at least some genotoxic
20 activity, with the strongest data from DCA, DCVG, and DCVC. While quantitatively smaller in
21 terms of flux as compared to TCA and TCOH (for which there is almost no genotoxicity data),
22 these metabolites may still be lexicologically important.
23 Thus, uncertainties with regard to the characterization of TCE genotoxicity remain,
24 particularly because not all TCE metabolites have been sufficiently tested in the standard
25 genotoxicity screening battery to derive a comprehensive conclusion. However, the metabolites
26 that have been tested particularly DCVC have predominantly resulted in positive data although
27 to a lesser extent in DCVG and NAcDCVC, supporting the conclusion that these compounds are
28 genotoxic, particularly in the kidney, where in situ metabolism produces and/or bioactivates
29 these TCE metabolites.
30
31 4.3. CENTRAL NERVOUS SYSTEM (CNS) TOXICITY
32 TCE exposure results in central nervous system (CNS) effects in both humans and
33 animals that can result from acute, subchronic, or chronic exposure. There are studies indicating
34 that TCE exposure results in CNS tumors and this discussion can be found in Section 4.9. The
35 studies discussed in this section focus on the most critical neurological effects that were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-84 DRAFT—DO NOT CITE OR QUOTE
-------
1 extracted from the neurotoxicological literature. Although there are several studies and reports
2 that have evaluated TCE as an anesthetic, those studies were not included in this section because
3 of the high exposure levels in comparison to the selected critical neurological effects described
4 below. The critical neurological effects are nerve conduction changes, sensory effects, cognitive
5 deficits, changes in psychomotor function, and changes in mood and sleep behaviors. The
6 selection criteria that were used to determine study importance included study design and
7 validity, pervasiveness of neurological effect, and for animal studies, the relevance of these
8 reported outcomes in humans. More detailed information on human and animal neurological
9 studies with TCE can be found in Appendix D.
10
11 4.3.1. Alterations in Nerve Conduction
12 4.3.1.1. Trigeminal Nerve Function: Human Studies
13 A number of human studies have been conducted that examined the effects of
14 occupational or drinking water exposures to TCE on trigeminal nerve function (see Table 4-20).
15 Many studies reported that humans exposed to TCE present trigeminal nerve function
16 abnormalities as measured by blink reflex and masseter reflex test measurements (Feldman et al.,
17 1988, 1992; Kilburn and Warshaw, 1993; Kilburn, 2002a; Ruitjen et al., 2001). The blink and
18 masseter reflexes are mediated primarily by the trigeminal nerve and changes in measurement
19 suggest impairment in nerve conduction. Other studies measured the trigeminal somatosensory
20 evoked potential (TSEP) following stimulation of the trigeminal nerve and reported statistically
21 significantly delayed response on evoked potentials among exposed subjects compared to
22 nonexposed individuals (Barret et al., 1982, 1984, 1987; Mhiri et al., 2004). Two studies which
23 also measured trigeminal nerve function did not find any effect (El-Ghawabi et al., 1973;
24 Rasmussen et al., 1993c) but the methods were not provided in either study (El-Ghawabi et al.,
25 1973; Rasmussen et al., 1993c) or an appropriate control group was not included
26 (Rasmussen et al., 1993c). These studies and results are described below and summarized in
27 detail in Table 4-20.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-85 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-20. Summary of human trigeminal nerve and nerve conduction
velocity studies
Reference
Subjects
Exposure
Effect
Barret etal..
1982
11 workers with chronic
TCE exposure
Controls: 20 unexposed
subjects
Presence of TCE and TCA found
through urinalysis. Atmospheric
TCE concentrations and duration of
exposure not reported in paper.
Following stimulation of the
trigeminal nerve, significantly higher
voltage stimuli was required to obtain
a normal response and there was a
significant increase in latency for
response and decreased response
amplitude.
Barret etal..
1984
188 factory workers.
No unexposed controls;
lowest exposure group
used as comparison
>150 ppm; n = 54 < 150 ppm;
«=134,
7h/dfor7yr
Trigeminal nerve and optic nerve
impairment, asthenia and dizziness
were significantly increased with
exposure.
Barret etal..
1987
104 degreaser machine
operators
Controls: 52 unexposed
subjects
Mean age 41.6yrs
Mean duration, 8.2 yrs, average daily
exposure 7 h/d.
Average TCOH range = 162-245
mg/g creatinine
Average TCA range = 93-131 mg/g
creatinine
Evoked trigeminal responses were
measured following stimulation of the
nerve and revealed increased latency
to respond, amplitude or both and
correlated with length of exposure
(p < 0.01) and with age (p < 0.05), but
not concentration.
El-Ghawabi
etal., 1973
30 money printing shop
workers
Controls: 20
nonexposed males
10 workers exposed to
inks not containing TCE
Mean TCE air concentrations ranged
from 41 ppm to 163 ppm. Exposure
durations:
Less than 1 yr: n = 3
1 yr: n = 1
2 yrs: n = 2
3 yrs: n = 11
4 yrs: n = 4
5 yrs or greater: n = 9
No effect on trigeminal nerve function
was noted.
Feldman et
al., 1988
21 Woburn,MA
residents;
27 controls
TCE maximum reported
concentration in well water was 267
ppb; other solvents also present.
Exposure duration ranged from
1-12 yrs.
Measurement of the blink reflex as
mediated by the trigeminal nerve
resulted in significant increases in the
latency of reflex components
(p< 0.001).
Feldman et
al., 1992
18 workers;
30 controls
TCE exposure categories of
"extensive", "occasional," and
"chemical other than TCE"
"extensive" = chronically exposed
(>1 yr) to TCE for 5 d/wk and >50%
workday.
"occupational" = chronically exposed
to TCE for 1-3 d/wk and >50%
workday.
The blink reflex as mediated by the
trigeminal was measured. The
"extensive" group revealed latencies
greater than 3 SD above the
nonexposed group mean on blink
reflex components.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-86 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-20. Summary of human trigeminal nerve and nerve conduction
velocity studies (continued)
Reference
Subjects
Exposure
Effect
Kilburn and
Warshaw,
1993
160 residents living in
Southwest Tucson with
TCE, other solvents, and
chromium in
groundwater
Control: 113 histology
technicians from a
previous study
(Kilburn etal., 1987;
Kilburn and Warshaw,
1992)
>500 ppb of TCE in well water
before 1981 and 25 to 100 ppb
afterwards
Duration ranged from 1 to 25 yrs
Significant impairments in sway speed
with eyes open and closed and blink
reflex latency (R-l) which suggests
trigeminal nerve impairment.
Kilburn,
2002a
236 residents near a
microchip plant in
Phoenix, AZ
Controls: 161 regional
referents from
Wickenburg, AZ and
67 referents in
northeastern Phoenix
0.2-10,000 ppb of TCE,
0.2-260,000 ppb TCA, 0.2-6,900
ppb 1,1-DCE, 0.2-1,600 1,2-DCE,
0.2-23,000 ppb PCE, O.02-330
ppb VC in well water
Exposure duration ranged from 2 to
37 yrs
Trigeminal nerve impairment as
measured by the blink reflex test; both
right and left blink reflex latencies
(R-l) were prolonged. Exposed group
mean 14.2 + 2.1 ms (right) or
13.9 + 2.1 ms (left) versus referent
group mean of 13.4 + 2.1 ms (right)
or!3.5 +2.1ms (left),/? = 0.0001
(right) and 0.008 (left).
Mhiri et al.,
2004
23 phosphate industry
workers
Controls: 23 unexposed
workers
Exposure ranged from 50-150 ppm,
for 6 hr/d for at least 2 yrs
Mean urinary trichloroethanol and
trichloroacetic acid levels were
79.3 ± 42 and 32.6 ± 22 mg/g
creatinine
TSEPs were recorded. Increase in the
TSEP latency was observed in 15 out
of 23 (65%) workers.
Rasmussen
etal., 1993c
96 Danish metal
degreasers
Age range: 19-68;
No unexposed controls;
low exposure group
used as comparison
Average exposure duration: 7.1 yrs.);
range of full-time degreasing:
1 month to 36 yrs. Exposure to TCE
ortoCFC113
1) Low exposure: n = 19, average
full-time exposure 0.5 yrs
2) Medium exposure: n = 36, average
full-time exposure 2.1 yrs
3) High exposure: n = 41, average
full-time exposure 11 yrs. TCA in
high exposure group = 7.7 mg/L
(max = 26.1 mg/L)
No statistically significant trend on
trigeminal nerve function, although
some individuals had abnormal
function.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-87 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-20. Summary of human trigeminal nerve and nerve conduction
velocity studies (continued)
Reference
Subjects
Exposure
Effect
Ruitjen et
al, 1991
3 1 male printing
workers. Mean age
44 yrs; mean duration
16yrs
Controls: 28 unexposed;
Mean age 45 yrs
Mean cumulative exposure =
704 ppm x yrs (SD 583, range:
160-2,150 ppmx yrs
Mean, 17 ppm at time of study;
historic TCE levels from 1976-1981,
mean of 35 ppm
Mean duration of 16 yrs
Measurement of trigeminal nerve
function by using the blink reflex
resulted in no abnormal findings.
Increased latency in the masseter
reflex is indicative of trigeminal nerve
impairment.
Triebig et
al., 1982
24 workers (20 males,
4 females)
occupationally
exposed — ages 17-56;
Controls:
144 individuals to
establish normal nerve
conduction parameters;
Matched group:
24 unexposed workers
(20 males, 4 females)
Exposure duration of 1 month to
258 months (mean 83 months). Air
exposures were between 5-70 ppm
No statistically significant difference
in nerve conduction velocities
between the exposed and unexposed
groups.
Triebig et
al., 1983
66 workers
occupationally exposed
Control: 66 workers not
exposed to solvents
Subjects were exposed to a mixture
of solvents, including TCE
Exposure-response relationship
observed between length of solvent
exposure and statistically significant
reduction in mean sensory ulnar nerve
conduction velocities.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
DCE = dichloroethylene, PCE = perchloroethylene, SD = standard deviation.
Integrity of the trigeminal nerve is commonly measured using blink and masseter
reflexes. Five studies (Barret et al., 1984; Feldman et al., 1988, 1992; Kilburn and Warshaw,
1993; Kilburn, 2002a) reported a significant increase in the latency to respond to the stimuli
generating the reflex. The latency increases in the blink reflex ranged from 0.4 ms (Kilburn,
2002a) to up to 3.44 ms (Feldman et al., 1988). The population groups in these studies were
exposed by inhalation occupationally (Barret et al., 1984) and through drinking water
environmentally (Feldman et al., 1988; Kilburn and Warshaw, 1993; Kilburn, 2002a).
Feldman et al. (1992) demonstrated persistence in the increased latency of the blink reflex
response. In one subject, exposure to TCE (levels not reported by authors) occurred through a
degreasing accident (high and acute exposure), and increased latency response times persisted
20 years after the accident. Another two subjects, evaluated at 9 months and 1 month following
a high occupational exposure (exposure not reported by authors), also had higher blink reflex
latencies with an average increase of 2.8 ms over the average response time in the control group
used in the study. Although one study (Ruitjen et al., 1991) did not find these increases in male
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-88 DRAFT—DO NOT CITE OR QUOTE
-------
1 printing workers exposed to TCE, this study did find a statistically significant average increase
2 of 0.32 ms (p < 0.05) in the latency response time in TCE-exposed workers on the masseter
3 reflex test, another test commonly used to measure the integrity of the trigeminal nerve.
4 Three studies (Barret et al., 1982, 1987; Mhiri et al., 2004) adopting TSEPs to measure
5 trigeminal nerve function found significant abnormalities in these evoked potentials. These
6 studies were conducted on volunteers who were occupationally exposed to TCE through metal
7 degreasing operations (Barret et al., 1982, 1987) or through cleaning tanks in the phosphate
8 industry (Mhiri et al., 2004). Barret et al. (1982) reported that in eight of the eleven workers, an
9 increased voltage ranging from a 25 to a 45 volt increase was needed to generate a normal TSEP
10 and two of workers had an increased TSEP latency. Three out of 11 workers had increases in
11 TSEP amplitudes. In a later study, Barret et al. (1987) also reported abnormal TSEPs (increased
12 latency and/or increased amplitude) in 38% of the degreasers that were evaluated. The
13 individuals with abnormal TSEPs were significantly older (45 vs. 40.1 years;/? < 0.05) and were
14 exposed to TCE longer (9.9 vs. 5.6 years;/? < 0.01). Mhiri et al. (2004) was the only study to
15 evaluate individual components of the TSEP and noted significant increases in latencies for all
16 TSEP potentials (Nl, PI, N2, P2, N3; p < 0.01) and significant decreases in TSEP amplitude
17 (PI,p < 0.02; N2,p < 0.05). A significant positive correlation was demonstrated between
18 exposure duration and increased TSEP latency (p < 0.02).
19 Two studies reported no statistically significant effect of TCE exposure on trigeminal
20 nerve function (El-Ghawabi et al., 1973; Rasmussen et al., 1993). El-Ghawabi et al. (1973)
21 conducted a study on 30 money printing shop workers occupationally exposed to TCE.
22 Trigeminal nerve involvement was not detected, but the authors did not include the experimental
23 methods that were used to measure trigeminal nerve involvement and did not provide any data as
24 to how this assessment was made. Rasmussen et al. (1993c) conducted an historical cohort study
25 on 99 metal degreasers, 70 exposed to TCE and 29 to the fluorocarbon, CFC113. It was reported
26 that 1 out of 21 people (5%) in the low exposure, 2 out of 37 (5%) in the medium exposure and 4
27 out of 41 (10%) in the high exposure group experienced abnormalities in trigeminal nerve
28 sensory function, with a linear trend test/?-value of 0.42. The mean urinary trichloroacetic acid
29 concentration was reported for the high exposure group only and was 7.7 mg/L (maximum
30 concentration, 26.1 mg/L). The trigeminal nerve function findings of high exposure group
31 subjects was compared to that of low exposure group since this study did not include an
32 unexposed or no TCE exposure group, and decreased the sensitivity of the study.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-89 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.3.1.2. Nerve Conduction Velocity—Human Studies
2 Two occupational studies assessed ulnar and median nerve function using tests of
3 conduction latencies (Triebig, 1982, 1983) (see Table 4-20). The ulnar nerve and median nerves
4 are major nerves located in the arm and forearm. Triebig (1982) studied twenty-four healthy
5 workers (20 males, 4 females) exposed to TCE occupationally (5-70 ppm) at three different
6 plants and did not find statistically significant differences in ulnar or median nerve conduction
7 velocities between exposed and unexposed subjects. This study has measured exposure data, but
8 exposures/responses are not reported by dose levels. The Triebig (1983) study is similar in
9 design to the previous study (Triebig, 1982) but of a larger number of subjects. In this study, a
10 dose-response relationship was observed between lengths of exposure to mixed solvents that
11 included TCE (at unknown concentration). A statistically significant reduction in nerve
12 conduction velocities was observed for the medium- and long-term exposure groups for the
13 sensory ulnar nerve as was a statistically significant reduction in mean nerve conduction velocity
14 observed between exposed and control subjects.
15
16 4.3.1.3. Trigeminal Nerve Function: Laboratory Animal Studies
17 There is little evidence that TCE disrupts trigeminal nerve function in animal studies.
18 Two studies demonstrated TCE produces morphological changes in the trigeminal nerve at a
19 dose of 2,500 mg/kg/d for 10 weeks (Barret et al., 1991, 1992). However, dichloroacetylene, a
20 degradation product formed during the volatilization of TCE was found to produce more severe
21 morphological changes in the trigeminal nerve and at a lower dose of 17 mg/kg/d (Barret et al.,
22 1991, 1992). Only one study (Albee et al., 2006) has evaluated the effects of TCE on trigeminal
23 nerve function and a subchronic inhalation exposure did not result in any significant functional
24 changes. A summary of these studies is provided in Table 4-21.
25 Barret et al. (1991, 1992) conducted two studies evaluating the effects of both TCE and
26 dichloroacetylene on trigeminal nerve fiber diameter and internodal length as well as several
27 markers for fiber myelination. Female Sprague-Dawley rats (n = 7/group) were dosed with
28 2,500 mg/kg TCE or 17 mg/kg/d dichloroacetylene by gavage for 5 days/week for 10 weeks.
29 TCE-dosed animals only exhibited changes in the smaller Class A fibers where internode length
30 increased marginally (<2%) and fiber diameter increased by 6%. Conversely, dichloroacetylene-
31 treated rats exhibited significant and more robust decreases in internode length and fiber
32 diameter in both fiber classes A (decreased 8%) and B (decreased 4%).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-90 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-21. Summary of animal trigeminal nerve studies
Reference
Barret et
al, 1991
Barret et
al., 1992
Albee et
al., 1997
Albee et
al., 2006
Exposure
route
Direct gastric
administration
Direct gastric
administration
Inhalation
Inhalation
Species/strain/
sex/number
Rat, Sprague-
Dawley, female,
7/group
Rat, Sprague-
Dawley, female,
7/group
Rat, Fischer 344,
male, 6
Rat, Fischer 344,
male and female,
10/sex/group
Dose level/
exposure
duration
0, 2.5 g/kg, acute
administration
17 mg/kg
dichloroacetylene
0, 2.5 g/kg; 1
dose/d, 5 d/wk, 10
wks
17 mg/kg
dichloroacetylene
0 or 300-ppm
dichloro-
acetylene, 2.25 h
0, 250, 800, or
2,500 ppm
NOAEL;
LOAEL*
LOAEL:
2.5 g/kg
LOAEL:
2.5 g/kg
LOAEL:
300 ppm
dichloro-
acetylene
NOAEL:
2,500 ppm
Effects
Morphometric analysis was
used for analyzing the
trigeminal nerve. Increase in
external and internal fiber
diameter as well as myelin
thickness was observed in the
trigeminal nerve after TCE
treatment.
Trigeminal nerve analyzed
using morphometric analysis.
Increased internode length and
fiber diameter in class A fibers
of the trigeminal nerve
observed with TCE treatment.
Changes in fatty acid
composition also noted.
Dichloroacetylene (TCE
byproduct) exposure impaired
the TSEP up to 4 d
postexposure.
No effect on TSEPs was noted
at any exposure level.
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
*NOAEL = no-observed-adverse-effect level, LOAEL = lowest-observed-adverse-effect-level.
Albee et al. (2006) evaluated the effects of a subchronic inhalation TCE exposure in
Fischer 344 rats (10/sex/group). Rats were exposed to 0, 250, 800, and 2,500 ppm TCE for
6 hours/day, 5 days/week for 13 weeks. TCE exposures were adequate to produce permanent
auditory impairment even though TSEPs were unaffected. While TCE appears to be negative in
disrupting the trigeminal nerve, the TCE breakdown product, dichloroacetylene, does impair
trigeminal nerve function. Albee et al. (1997) showed that a single inhalation exposure of rats to
300-ppm dichloroacetylene, for 2.25 hours, disrupted trigeminal nerve evoked potentials for at
least 4 days post exposure.
4.3.1.4. Discussion and Conclusions: Trichloroethylene (TCE)-Induced Trigeminal Nerve
Impairment
Epidemiologic studies of exposure to TCE found impairment of trigeminal nerve
function, assessed by the blink reflex test or the TSEP, in humans exposed occupationally by
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-91 DRAFT—DO NOT CITE OR QUOTE
-------
1 inhalation or environmentally by ingestion (see Table 4-20). Mean inhalational exposures
2 inferred from biological monitoring or from a range of atmospheric monitoring in occupational
3 studies was approximately 50 ppm to <150 ppm TCE exposure. Residence location is the
4 exposure surrogate in geographical-base studies of contaminated water supplies with several
5 solvents. Well water contaminant concentrations of TCE ranged from <0.2 ppb to 10,000 ppb
6 and do not provide an estimate of TCE concentrations in drinking water to studied individuals.
7 Two occupational studies, each including more than 100 subjects, reported statistically
8 significant dose-response trends based on ambient TCE concentrations, duration of exposure,
9 and/or urinary concentrations of the TCE metabolite TCA (Barret et al., 1984, 1987). Three
10 geographical-based studies of environmental exposures to TCE via contaminated drinking water
11 are further suggestive of trigeminal nerve function decrements; however, these studies are more
12 limited than occupational studies due to questions of subject selection. Both exposed subjects
13 who were litigants and control subjects who may not be representative of exposed (Kilburn and
14 Warshaw, 1993; Kilburn et al., 2002a); referents in Kilburn and Warshaw (1993) were histology
15 technicians and subjects in a previous study of formaldehyde and other solvent exposures and
16 neurobehavioral effects (Kilburn et al., 1987; Kilburn and Warshaw, 1992). Results were mixed
17 in a number of smaller studies. Two of these studies reported changes in trigeminal nerve
18 response (Mhiri et al., 2004; Barret et al., 1982), including evidence of a correlation with
19 duration of exposure and increased latency in one study (Mhiri et al., 2004). Ruitjen et al. (1991)
20 reported no significant change in the blink reflex, but did report an increase in the latency of the
21 masseter reflex, which also may reflect effects on the trigeminal nerve. Two other studies
22 reported no observed effect on trigeminal nerve impairment, but the authors failed to provide
23 assessment of trigeminal nerve function (El-Ghawabi et al., 1973, Rasmussen et al., 1993c) or
24 there was not a control (nonexposed) group included in the study (Rasmussen et al., 1993c).
25 Therefore, because of limitations in statistical power, the possibility of exposure
26 misclassification, and possible differences in measurement methods, these studies are not judged
27 to provide substantial evidence against a causal relationship between TCE exposure and
28 trigeminal nerve impairment. Overall, the weight of evidence supports a relationship between
29 TCE exposure and trigeminal nerve dysfunction in humans.
30 Impairment of trigeminal nerve function is observed in studies of laboratory animal
31 studies. Although one subchronic animal study demonstrated no significant impairment of
32 trigeminal nerve function following TCE exposure up to 2,500 ppm (no-observed-adverse-effect
33 level [NOAEL]; Albee et al., 2006), morphological analysis of the nerve revealed changes in its
34 structure (Barret et al., 1991, 1992). However, the dose at which an effect was observed by
35 Barret et al. (1991, 1992) was high (2,500 mg/kg/d—lowest-observed-adverse-effect level
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-92 DRAFT—DO NOT CITE OR QUOTE
-------
1 [LOAEL]) compared to any reasonable occupational or environmental setting, although no lower
2 doses were used. The acute or subchronic duration of these studies, as compared to the much
3 longer exposure duration in many of the human studies, may also contribute to the apparent
4 disparity between the epidemiologic and (limited) laboratory animal data.
5 The subchronic study of Barret et al. (1992) and the acute exposure study of Albee et al.
6 (1997) also demonstrated that dichloroacetylene, a (ex vivo) TCE degradation product, also
7 induces trigeminal nerve impairment, at much lower doses than TCE. It is possible that under
8 some conditions, coexposure to dichloroacetylene from TCE degradation may contribute to the
9 changes observed to be associated with TCE exposure in human studies, and this issue is
10 discussed further below in Section 4.3.10.
11 Overall evidence from numerous epidemiologic studies supports a conclusion that TCE
12 exposure induces trigeminal nerve impairment in humans. Laboratory animal studies provide
13 limited additional support, and do not provide strong contradictory evidence. Persistence of
14 these effects after cessation of exposure cannot be determined since exposure was ongoing in the
15 available human and laboratory animal studies.
16
17 4.3.2. Auditory Effects
18 4.3.2.1. Auditory Function: Human Studies
19 The TCE Subregistry from the National Exposure Registry developed by the ATSDR was
20 the subject of three studies (Burg et al., 1995, 1999; ATSDR, 2003). A fourth study (Rasmussen
21 et al., 1993c) of degreasing workers exposed to either TCE or CFC113 also indirectly evaluated
22 auditory function. These studies are discussed below and presented in detail in Table 4-22.
23 Burg et al. (1995, 1999) reviewed the effects of TCE on 4,281 individuals (TCE
24 Subregistry) residentially exposed to this solvent for more than 30 consecutive days. Face-to-
25 face interviews were conducted with the TCE Subregistry population and self-reported hearing
26 loss was evaluated based on personal assessment through the interview (no clinical evaluation
27 was conducted). TCE registrants that were 9 years old or younger had a statistically significant
28 increase in hearing impairment as reported by the subjects. The relative risk (RR) in this age
29 group for hearing impairments was 2.13 (95% confidence interval [CI]: 1.12-4.06) which
30 decreased to 1.12 (95% CI: 0.52-2.24) for the 10-17 age group and 0.32 (95% CI: 0.10-1.02)
31 for all older age groups. A statistically significant association (when adjusted for age and sex)
32 was found between duration of exposure, in these studies this was length of residency, and
33 reported hearing impairment. The odds ratio (OR) was 2.32 (95% CI: 1.18-4.56) for subjects
34 exposed to TCE >2 years and <5 years, 1.17 (95% CI: 0.55-2.49) for exposure >5 years and
35 <10 years, 2.46 (95% CI: 1.30-5.02) for exposure durations greater than 10 years.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-93 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-22. Summary of human auditory function studies
Reference
Subjects
Exposure
Effect
ATSDR, 2003
116 children, under
10 yrs of age, residing
near 6 Superfund sites.
Further study of children
in Burg etal. (1995,
1999)
Control: 182 children
TCE and other solvents in
ground water supplies.
Exposures were modeled
using tap water TCE
concentrations and GIS for
spatial interpolation, and
LaGrange for temporal
interpolation to estimate
exposures from gestation
to 1990 across the area of
subject residences.
Control = 0 ppb; low
exposure group = 0 < 23
ppb-yrs; and high exposure
group = >23 ppb-yrs
Auditory screening revealed increased
incidence of abnormal middle ear
function in exposed groups as indicated
from acoustic reflex test. Adjusted odds
ratios for right ear ipsilateral acoustic
reflects control, OR: 1.0, low exposure
group, OR: 5.1, p < 0.05; high exposure
group, OR: 7.2, p < 0.05. ORs adjusted
for age, sex, medical history and other
chemical contaminants. No significant
decrements reported in the pure tone
and typanometry screening.
Burg etal., 1995
From an NHIS TCE
subregistry of 4,281
(4,041 living and
240 deceased) residents
Environmentally exposed
to TCE and other solvents
via well water in Indiana,
Illinois, and Michigan
Increase in self-reported hearing
impairments for children <9 yrs.
Burg etal., 1999
3,915 white registrants
Mean age 34 yrs (SD =
19.9 yrs)
Cumulative TCE exposure
subgroups: <50 ppb,
n = 2,867; 50-500 ppb,
n = 870; 500-5,000 ppb,
n = 190; >5,000 ppb,
Exposure duration
subgroups: <2 yrs, 2-5 yrs,
5-10 yrs. , >10 yrs
A statistically significant
association (adjusted for age and sex)
between duration of exposure and
self-reported hearing impairment
was found.
Rasmussen et al..
1993b
96 Danish metal
degreasers. Age range:
19-68 yrs;
No unexposed controls;
low exposed group is
referent
Average exposure
duration: 7.1 yrs.); range
of full-time degreasing:
1 month to 36 yrs.
Exposure to TCE or and
CFC113
(1) Low exposure: n = 19,
average full-time exposure
0.5 yrs
(2) Medium exposure:
n = 36, average full-time
exposure 2.1 yrs
(3) High exposure: n = 41,
average full-time exposure
11 yrs. MeanU-TCAin
high exposure group = 7.7
mg/L (max = 26.1 mg/L);
Auditory impairments noted through
several neurological tests.
Significant relationship of exposure was
found with Acoustic-motor function
(p < 0.001), Paced Auditory Serial
Addition Test (p < 0.001), Rey Auditory
Verbal Learning Test (p< 0.001).
NHIS = National Health Interview Survey, U-TCA = urinary trichloroacetic acid.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-94 DRAFT—DO NOT CITE OR QUOTE
-------
1 ATSDR (2003) conducted a follow-up study to the TCE subregistry findings (Burg et al.,
2 1995, 1999) and focused on the subregistry children located in Elkhart, IN, Rockford, IL and
3 Battle Creek, MI using clinical tests for oral motor, speech, and hearing function. Exposures
4 were modeled using tap water TCE concentrations and geographic information system (GIS) for
5 spatial interpolation, and LaGrange for temporal interpolation to estimate exposures from
6 gestation to 1990 across the area of subject residences. Modeled data were used to estimate
7 lifetime exposures (ppb-years) to TCE in residential wells. The median TCE exposure for the
8 children was estimated from drinking water as 23 ppb/year of exposure (ranging from
9 0-702 ppb/year). Approximately 20% (ranged from 17-21% depending on ipsilateral or
10 contralateral test reflex) of the children in the TCE subregistry and 5-7% in the control group
11 exhibited an abnormal acoustic reflex (involuntary muscle contraction that measures movement
12 of the stapedius muscle in the middle ear following a noise stimulus) which was statistically
13 significant (p = 0.003). Abnormalities in this reflex could be an early indicator of more serious
14 hearing impairments. No significant decrements were reported in the pure tone and typanometry
15 screening.
16 Rasmussen et al. (1993b) used a psychometric test to measure potential auditory effects
17 of TCE exposure in an occupational study. Results from 96 workers exposed to TCE and other
18 solvents were presented in this study. Details of the exposure groups and exposure levels are
19 provided in Table 4-22. The acoustic motor function test was used for evaluation of auditory
20 function. Significant decrements (p < 0.05) in acoustic motor function performance scores
21 (average decrement of 2.5 points on a 10-point scale) was reported for TCE exposure.
22
23 4.3.2.2. Auditory Function: Laboratory Animal Studies
24 The ability of TCE to permanently disrupt auditory function and produce abnormalities in
25 inner ear histopathology has been demonstrated in several studies using a variety of test methods.
26 Two different laboratories have identified NOAELs following inhalation exposure for auditory
27 function of 1,600 ppm for 12 hours/day for 13 weeks in Long Evans rats (n = 6-10) (Rebert et
28 al., 1991) and 1,500 ppm for 18 hours/day, 5 days/week for 3 weeks in Wistar-derived rats
29 (n = 12) (Jaspers et al., 1993). The LOAELs identified in these and similar studies are
30 2,500-4,000 ppm TCE for periods of exposure ranging from 4 hours/day for 5 days to
31 12 hours/day for 13 weeks (e.g., Muijser et al., 2000; Rebert et al., 1995, 1993; Crofton et al.,
32 1994; Crofton and Zhao, 1997; Fechter et al., 1998; Boyes et al., 2000; Albee et al., 2006).
33 Rebert et al. (1993) estimated acute blood TCE levels associated with permanent hearing
34 impairment at 125 ug/mL by methods that probably underestimated blood TCE values (rats were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-95 DRAFT—DO NOT CITE OR QUOTE
-------
1 anaesthetized using 60% carbon dioxide [CO2]). A summary of these studies is presented in
2 Table 4-23.
3 Reflex modification was used in several studies to evaluate the auditory function in TCE-
4 exposed animals (Jaspers et al., 1993; Muijser et al., 2000; Fechter et al., 1998; Crofton and
5 Zhao, 1993; Crofton et al., 1994; Crofton and Zhou, 1997; Boyes et al., 2000; Yamamura et al.,
6 1983). These studies collectively demonstrate significant decreases in auditory function at mid-
7 frequency tones (8-20 kHz tones) for TCE exposures greater than 1,500 ppm after acute, short-
8 term, and chronic durations. Only one study (Yamamura et al., 1983) did not demonstrate
9 impairment in auditory function from TCE exposures as high as 17,000 ppm for 4 hours/day over
10 5 days. This was the only study to evaluate auditory function in guinea pigs, whereas the other
11 studies used various strains of rats. Despite the negative finding in Yamamura et al. (1983),
12 auditory testing was not performed in an audiometric sound attenuating chamber and extraneous
13 noise could have influenced the outcome. It is also important to note that the guinea pig has
14 been reported to be far less sensitive than the rat to the effects of ototoxic aromatic hydrocarbons
15 such as toluene.
16 Crofton and Zhao (1997) also presented a benchmark dose for which the calculated dose
17 of TCE would yield a 15 dB loss in auditory threshold. This benchmark response was selected
18 because a 15 dB threshold shift represents a significant loss in threshold sensitivity for humans.
19 The benchmark concentrations for a 15 dB threshold shift are 5,223 ppm for 1 day, 2,108 ppm
20 for 5 days, 1,418 ppm for 20 days and 1,707 ppm for 65 days of exposure. While more sensitive
21 test methods might be used and other definitions of a benchmark effect chosen with a strong
22 rationale, these data provide useful guidance for exposure concentrations that do yield hearing
23 loss in rats.
24
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-96 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-23. Summary of animal auditory function studies
Reference
Rebertetal.,
1991
Rebertetal.,
1993
Rebertetal.,
1995
Crofton et
al., 1994
Crofton and
Zhou, 1997;
Boyes et al.,
2000
Exposure
route
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Species/strain/
sex/number
Rat, Long Evans,
male, 10/group
Rat, F344, male,
4-5/group
Rat, Long Evans,
male, 9/group
Rat, Long Evans,
male, 9/group
Rat, Long Evans,
male, 7-8/group
Rat, Long Evans,
male, 9-12/group
Rat, Long Evans,
male, 8- 10/group
Rat, Long Evans,
male, 8- 10/group
Rat, Long Evans,
male, 8- 10/group
Dose level/
exposure
duration
Long Evans: 0,
1,600, and 3,200
ppm; 12 h/d, 12
wks
F344: 0, 2,000,
3,200 ppm; 12
h/d, 3 wks
0, 2,500, 3,000,
3,500 ppm; 8
h/d, 5d
0, 2,800 ppm; 8
h/d, 5d
0, 3,500 ppm
TCE; 8 h/d, 5 d
0, 4,000, 6,000,
8,000 ppm; 6 h
0, 1,600, 2,400,
and 3,200 ppm;
6 h/d, 5 d
0, 800, 1,600,
2,400, and 3,200
ppm; 6 h/d,
5 d/wk, 4 wks
0, 800, 1,600,
2,400, and 3,200
ppm; 6 h/d,
5 d/wk, 13 wks
NOAEL;
LOAEL a
Long Evans:
NOAEL:
1,600 ppm;
LOAEL:
3,200 ppm
F344:
LOAEL:
2,000 ppm
NOAEL:
2,500 ppm
LOAEL:
3,000 ppm
LOAEL:
2,800 ppm
LOAEL:
3,500 ppm
NOAEL:
6,000 ppm
LOAEL:
8,000 ppm
NOAEL:
2,400 ppm
LOAEL:
3,200 ppm
NOAEL:
2,400 ppm
LOAEL:
3,200 ppm
NOAEL:
1,600 ppm
LOAEL:
2,400 ppm
Effects
BAERs were measured.
Significant decreases in BAER
amplitude and an increase in
latency of appearance of the
initial peak (PI).
BAERs were measured 1-2 wks
postexposure to assess auditory
function. Significant decreases in
BAERs were noted with TCE
exposure.
BAER measured 2-14 days
postexposure at a 16 kHz tone.
Hearing loss ranged from
55-85 dB.
BAER measured and auditory
thresholds determined 5-8 wks
postexposure. Selective
impairment of auditory function
for mid-frequency tones (8 and
16 kHz).
Auditory thresholds as measured
by BAERs for the 16 kHz tone
increased with TCE exposure.
Measured 3-5 wks post exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-97 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-23. Summary of animal auditory function studies (continued)
Reference
Fechter et
al, 1998
Jaspers et
al., 1993
Muijseret
al., 2000
Albee et al.,
2006
Yamamura
etal., 1983
Exposure
route
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Species/strain/
sex/number
Rat, Long Evans,
male, 12/group
Rat, Wistar
derived WAG-
Rii/MBL, male,
12/group
Rat, Wistar
derived WAG-
Rii/MBL, male, 8
Rat, Fischer 344,
male and female,
10/sex/group
Guinea Pig, albino
Hartley, male,
7-10/group
Dose level/
exposure
duration
0, 4,000 ppm; 6
h/d, 5d
0, 1,500, and
3,000 ppm; 18
h/d, 5 d/wk, 3
wks
0, 3,000 ppm;
18 h/d, 5 d/wk,
3 wks
0, 250, 800,
2,500 ppm; 6
h/d, 5 d/wk, 13
wks
0, 6,000, 12,000,
17,000 ppm; 4
h/d, 5d
NOAEL;
LOAEL
LOAEL:
4,000 ppm
NOAEL:
1,500 ppm
LOAEL:
3,000 ppm
NOAEL:
800 ppm
LOAEL:
2,500 ppm
NOAEL:
17,000 ppm
Effects
Cochlear function measured 5-7
wks after exposure. Loss of
spiral ganglion cells noted. Three
wks postexposure, auditory
function was significantly
decreased as measured by
compound action potentials and
reflex modification.
Auditory function assessed
repeatedly 1-5 wks postexposure
for 5, 20, and 35 kHz tones; no
effect at 5 or 35 kHz; decreased
auditory sensitivity at 20 kHz,
3,000 ppm.
Auditory sensitivity decreased
with TCE exposure at 4, 8, 16, and
20 kHz tones. White noise
potentiated the decrease in
auditory sensitivity.
Mild frequency specific hearing
deficits; focal loss of cochlear
hair cells.
No change in auditory sensitivity
at any exposure level as measured
by cochlear action potentials and
microphonics. Study was
conducted in guinea pig and
species is less sensitive to
auditory toxicity than rats.
Studies were also not conducted
in a sound-isolation chamber and
effects may be impacted by
background noise.
2
3
4
5
6
7
Brainstem auditory-evoked potentials (BAERs) were also measured in several studies
(Rebert et al., 1991, 1993, 1995; Albee et al., 2006) following at exposures ranging from
3-13 weeks. Rebert et al. (1991) measured BAERs in male Long Evans rats (n = 10) and F344
rats (n = 4-5) following stimulation with 4, 8, and 16 kHz sounds. The Long-Evans rats were
exposed to 0, 1,600, or 3,200 ppm TCE, 12 hours/day for twelve weeks and the F344 rats were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-98 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposed to 0, 2,000, or 3,200 ppm TCE, 12 hours/day for three weeks. BAER amplitudes were
2 significantly decreased at all frequencies for F344 rats exposed to 2,000 and 3,000 ppm TCE and
3 for Long Evans rats exposed to 3,200 ppm TCE. These data identify a LOAEL at 2,000 ppm for
4 the F344 rats and a NOAEL at 1,600 ppm for the Long Evans rats. In subsequent studies Rebert
5 et al. (1993, 1995) again demonstrated TCE significantly decreases BAER amplitudes and also
6 significantly increases the latency of appearance. Similar results were obtained by Albee et al.
7 (2006) for male and female F344 rats exposed to TCE for 13 weeks. The NOAEL for this study
8 was 800 ppm based on ototoxicity at 2,500 ppm.
9 Notable physiological changes were also reported in a few auditory studies. Histological
10 data from cochleas in Long-Evans rats exposed to 4,000 ppm TCE indicated that there was a loss
11 in spiral ganglion cells (Fechter et al., 1998). Similarly, there was an observed loss in hair cells
12 in the upper basal turn of the cochlea in F344 rats exposed to 2,500-ppm TCE (Albee et al.,
13 2006).
14
15 4.3.2.3. Summary and Conclusion of Auditory Effects
16 Human and animal studies indicated that TCE produces decrements in auditory function.
17 In the human epidemiological studies (ATSDR, 2003; Burg et al., 1995, 1999; Rasmussen et al.,
18 1993c) it is suggested that auditory impairments result from both an inhalation and oral TCE
19 exposure. A LOAEL of approximately 23 ppb-years TCE (extrapolated from <23 ppb-years
20 group in the ATSDR, 2003) from oral intake is noted for auditory effects in children. The only
21 occupational study where auditory effects were seen reported mean urinary trichloroacetic acid
22 concentration, a nonspecific metabolite of TCE, of 7.7 mg/L for the high cumulative exposure
23 group only (Rasmussen et al., 1993c). A NOAEL or a LOAEL for auditory changes resulting
24 from inhalational exposure to TCE cannot be interpolated from average urinary trichloroacetic
25 acid (U-TCA) concentration of subjects in the high exposure group because of a lack of detailed
26 information on long-term exposure levels and duration (Rasmussen et al., 1993c). Two studies
27 (Burg et al., 1995, 1999) evaluated self-reported hearing effects in people included in the TCE
28 subregistry comprised of people residing near Superfund sites in Indiana, Illinois, and Michigan.
29 In Burg et al. (1995), interviews were conducted with the TCE exposed population and it was
30 found that children aged 9 years or younger had statistically significant hearing impairments in
31 comparison to nonexposed children. This significant increase in hearing impairment was not
32 observed in any other age group that was included in this epidemiological analysis. This lack of
33 effect in other age groups may suggest association with another exposure other than drinking
34 water ; however, it may also suggest that children may be more susceptible than adults. In a
35 follow-up analysis, Burg et al. (1999) adjusted the statistical analysis of the original data (Burg et
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-99 DRAFT—DO NOT CITE OR QUOTE
-------
1 al., 1995) for age and sex. When these adjustments were made, a statistically significant
2 association was reported self-reported for auditory impairment and duration of residence. These
3 epidemiological studies provided only limited information given their use of an indirect exposure
4 metric of residence location, no auditory testing of this studied population and self-reporting of
5 effects. ATSDR (2003) further tested the findings in the Burg studies (Burg et al., 1995, 1999)
6 by contacting the children that were classified as having hearing impairments in the earlier study
7 and conducting several follow-up auditory tests. Significant abnormalities were reported for the
8 children in the acoustic reflex test which suggested effects to the lower brainstem auditory
9 pathway with the large effect measure, the odds ratio, was reported for the high cumulative
10 exposure group. Strength of analyses was its adjustment for potential confounding effects of
11 age, sex, medical history and other chemical contaminants in drinking water supplies. The
12 ATSDR findings were important in that the results supported Burg et al. (1995, 1999).
13 Rasmussen et al. (1993b) also evaluated auditory function in metal workers with inhalation
14 exposure to either TCE or CFC113. Results from tasks including an auditory element suggested
15 that these workers may have some auditory impairment. However, the tasks did not directly
16 measure auditory function.
17 Animals strongly indicated that TCE produces deficits in hearing and provides biological
18 context to the epidemiological study observations. Although there is a strong association
19 between TCE and ototoxicity in the animal studies, most of the effects began to occur at higher
20 inhalation exposures. NOAELs for ototoxicity ranged from 800-1,600 ppm for exposure
21 durations of at least 12 weeks (Albee et al., 2006; Crofton and Zhou, 1997; Boyes et al., 2000;
22 Rebert et al., 1991). Inhalation exposure to TCE was the route of administration in all the animal
23 studies. These studies either used reflex modification audiometry (Jaspers et al., 1993; Crofton
24 et al., 1994; Crofton and Zhou, 1997; Muijser et al., 2000) procedures or measured brainstem
25 auditory evoked potentials (Rebert et al., 1991, 1993, 1995) to evaluate hearing in rats.
26 Collectively, the animal database demonstrates that TCE produces ototoxicity at mid-frequency
27 tones (4-24 kHz) and no observed changes in auditory function were observed at either the low
28 (<4 kHz) or high (>24 kHz) frequency tones. Additionally, deficits in auditory effects were
29 found to persist for at least 7 weeks after the cessation of TCE exposure (Rebert et al., 1991;
30 Jaspers et al., 1993; Crofton and Zhou, 1997; Fechter et al., 1998; Boyes et al., 2000). Decreased
31 amplitude and latency were noted in the BAERs (Rebert et al., 1991, 1993, 1995) suggesting that
32 TCE exposure affects central auditory processes. Decrements in auditory function following
33 reflex modification audiometry (Jaspers et al., 1993; Crofton et al., 1994; Crofton and Zhou,
34 1997; Muijser et al., 2000) combined with changes observed in cochlear histopathology (Fechter
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-100 DRAFT—DO NOT CITE OR QUOTE
-------
1 et al., 1998; Albee et al., 2006) suggest that ototoxicity is occurring at the level of the cochlea
2 and/or brainstem.
O
4 4.3.3. Vestibular Function
5 4.3.3.1. Vestibular Function: Human Studies
6 The earliest reports of neurological effects resulting from TCE exposures focused on
7 subjective vestibular system symptoms, such as headaches, dizziness, and nausea. These
8 symptoms are subjective and self-reported. However, as they have been reported extensively in
9 the literature, there is little doubt that these effects can be caused by exposures to TCE.,
10 occupational exposures (Grandjean et al., 1955; Liu et al., 1988; Rasmussen et al., 1986; Smith
11 et al., 1970), environmental exposures (Hirsch et al., 1996), and in chamber studies (Stewart et
12 al., 1970; Smith etal., 1970).
13 Kylin et al. (1967) exposed 12 volunteers to 1,000 ppm (5,500 mg/m3) TCE for two hours
14 in a 1.5 x 2 x 2 meters chamber. Volunteers served as their own controls since 7 of the 12 were
15 pretested prior to exposure and the remaining 5 were post-tested days after exposure. Subjects
16 were tested for optokinetic nystagmus, which was recorded by electronystogmography, that is,
17 "the potential difference produced by eye movements between electrodes placed in lateral angles
18 between the eyes." Venous blood was also taken from the volunteers to measure blood TCE
19 levels during the vestibular task. The authors concluded that there was an overall reduction in
20 the limit ("fusion limit") to reach optokinetic nystagmus when individuals were exposed to TCE.
21 Reduction of the "fusion limit" persisted for up to 2 hours after the TCE exposure was stopped
22 and the blood TCE concentration was 0.2 mg/100 mL.
23
24 4.3.3.2. Vestibular Function: Laboratory Animal Data
25 The effect of TCE on vestibular function was evaluated by either (1) promoting
26 nystagmus (vestibular system dysfunction) and comparing the level of effort required to achieve
27 nystagmus in the presence and absence of TCE or (2) using an elevated beam apparatus and
28 measuring the balance. Overall, it was found that TCE disrupts vestibular function as presented
29 below and summarized in Table 4-24.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-101 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-24. Summary of mammalian sensory studies—vestibular and visual
systems
Reference
Exposure route
Species/strain/
sex/number
Dose level/
exposure duration
NOAEL;
LOAEL
Effects
Vestibular system studies
Thamet
al, 1979
Thamet
al., 1984
Niklasson
etal., 1993
Umezu et
al., 1997
Intravenous
Intravenous
Inhalation
Intraperitoneal
Rabbit, strain
unknown, sex
unspecified, 19
Rat, Sprague-
Dawley, female,
11
Rat, strain
unknown, male
and female, 28
Mouse, ICR,
male, 116
1-5 mg/kg/min
80 ug/kg/min
0, 2,700, 4,200,
6,000, 7,200 ppm; 1
h
0, 250, 500, or 1,000
mg/kg, single dose
and evaluated 30 min
postadministration
...
LOAEL:
2,700 ppm
NOAEL:
250 mg/kg
LOAEL:
500 mg/kg
Positional nystagmus
developed once blood
levels reached 30 ppm.
Excitatory effects on the
vestibule-oculomotor
reflex. Threshold effect at
blood (TCE) of 120 ppm or
0.9mM/L.
Increased ability to produce
nystagmus.
Decreased equilibrium and
coordination as measured
by the Bridge test (staying
time on an elevated balance
beam).
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Niklasson et al. (1993) showed acute impairment of vestibular function in male- and
female-pigmented rats during acute inhalation exposure to TCE (2,700-7,200 ppm) and to
tricholoroethane (500-2,000 ppm). Both of these agents were able to promote nystagmus during
optokinetic stimulation in a dose related manner. While there were no tests performed to assess
persistence of these effects, Tham et al. (1979, 1984) did find complete recovery of vestibular
function in rabbits (n= 19) and female Sprague-Dawley rats (n= 11) within minutes of
terminating a direct arterial infusion with TCE solution.
The finding that trichloroethylene can yield transient abnormalities in vestibular function
is not unique. Similar impairments have also been shown for toluene, styrene, along with
trichloroethane (Niklasson et al., 1993) and by Tham et al. (1984) for a broad range of aromatic
hydrocarbons. The concentration of TCE in blood at which effects were observed for TCE (0.9
mM/L) was quite close to that observed for most of these other vestibulo-active solvents.
4.3.3.3. Summary and Conclusions for the Vestibular Function Studies
Studies of TCE exposure in both humans and animals reported abnormalities in vestibular
function. Headaches, dizziness, nausea, motor incoordination, among other subjective symptoms
are reported in occupational epidemiological studies of TCE exposure (Grandjean et al., 1955;
Liu et al., 1988; Rasmussen et al., 1986; Smith et al., 1970; Hirsch et al., 1996; Stewart et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-102 DRAFT—DO NOT CITE OR QUOTE
-------
1 1970). One human exposure study (Kylin et al., 1967) found that vestibular function was
2 affected following an acute exposure to 1,000-ppm TCE (LOAEL). Individuals had a decreased
3 threshold to reach nystagmus than when exposed to TCE than to air. Animal studies also
4 evaluated the threshold to reach nystagmus and reported that TCE decreased the threshold to
5 produce nystagmus in rats (LOAEL: 2,700 ppm; Tham et al., 1984; Niklasson et al., 1993) and
6 rabbits (Tham et al., 1983).
7
8 4.3.4. Visual Effects
9 4.3.4.1. Visual Effects: Human Studies
10 Visual impairment in humans has been demonstrated following exposures through
11 groundwater (Kilburn, 2002a; Reif et al., 2003), from occupational exposure through inhalation
12 (Rasmussen et al., 1993b; Troster and Ruff, 1990) and from a controlled inhalation exposure
13 study (Vernon and Ferguson, 1969). Visual functions such as color discrimination and
14 visuospatial learning tasks are impaired in TCE-exposed individuals. Additionally, an acute
15 exposure can impair visual depth perception. Details of the studies are provided below and
16 summarized in Table 4-25.
17 Geographical-based studies utilized color discrimination and contrast sensitivity tests to
18 determine the effect of TCE exposure on vision. In these studies it was reported that TCE
19 exposure significantly increased color discrimination errors (Kilburn, 2002a) or decreases in
20 contrast sensitivity tests approached statistical significance after adjustments for several possible
21 confounders (p = 0.06 or 0.07; Reif et al., 2003). Exposure in Kilburn (2002a) is poorly
22 characterized, and for both studies, TCE is one of several contaminants in drinking water
23 supplies; neither study provides an estimate of an individual's exposure to TCE.
24 Rasmussen et al. (1993b) evaluated visual function in 96 metal workers, working in
25 degreasing at various factories and with exposure to TCE or CFC113. Visual function was tested
26 through the visual gestalts test (visual perception) and a visual recall test. In the visual gestalts
27 test, the number of total errors significantly increased from the low group (3.4 errors) to the high
28 exposure group (6.5 errors;/* = 0.01). No significant changes were observed in the visual recall
29 task. Troster and Ruff (1990) presented case studies conducted on two occupationally exposed
30 workers to TCE. Both patients presented with a visual-spatial task and neither could complete
31 the task within the number of trials allowed suggesting visual function deficits as a measure of
32 impaired visuospatial learning.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-103 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-25. Summary of human visual function studies
Reference
Subjects
Exposure
Effect
Kilburn,
2002a
236 residents near a
microchip plant in
Phoenix, AZ
Controls: 67 local
referents from Phoenix,
AZ and 161 regional
referents from
Wickenburg, AZ
TCE, TCA, 1,1-DCE, 1,2-DCE,
PCE, and VC detected in well water
up to 260,000 ppm; TCE
concentrations in well water were
0.2-10,000 ppb. Exposure duration
ranged from 2-37 yrs.
Exposure duration ranged from 2 to
37 yrs.
Color discrimination errors were
increased among residents
compared to regional referents
(p < 0.01). No adjustment for
possible confounding factors.
Reifetal.,
2003
143 residents of the
Rocky Mountain
Arsenal community of
Denver
Referent group at lowest
concentration (<5 ppb).
Exposure modeling of TCE
concentrations in groundwater and
in distribution system to estimate
mean TCE concentration by census
block of residence. High exposure
group >15 ppb.
Medium exposure group >5 ppb and
<15 ppb.
Low exposure referent group <5
ppb.
Contrast sensitivity test
performances (C and D) was
marginally statistically significant
(p = 0.06 and 0.07, respectively).
No significant effects reported for
the Benton visual retention test.
Significant decrements (p = 0.02)
were reported in the Benton visual
retention test when stratified with
alcohol consumption.
Rasmussen
et al., 1993b
96 Danish metal
degreasers. Age range:
19-68; no unexposed
controls; low exposure
group was referent
Average exposure duration: 7.1
yrs); range of full-time degreasing:
1 month to 36 yrs. Exposure to
TCEorCFC113.
1)
^ Low exposure: n = 19, average
full-time expo 0.5 yrs.
2) Medium exposure: n = 36,
average full-time exposure 2.1 yrs.
3) high exposure: n = 41, average
full-time exposure 11 yrs. TCA in
high exposure group = 7.7 mg/L
(max = 26.1 mg/L).
Statistically significant relationship
of exposure was found with the
Visual Gestalts learning and
retention test (cognitive test)
indicating deficits in visual
performance.
Troster and
Ruff, 1990
2 occupationally TCE-
exposed workers
Controls: 2 groups of
n = 30 matched controls;
(all age and education
matched)
Exposure concentration unknown
Exposure duration, 3-8 months.
Both workers experienced impaired
visuospatial learning.
Vernon and
Ferguson,
1969
8 male volunteers age
range 21-30; self
controls
0, 100, 300, and 1,000 ppm of TCE
for2h.
Statistically significant effects on
visual depth perception as measured
by the Howard-Dolman test.
NOAEL: 300 ppm; LOAEL: 1,000
ppm; No significant changes in any
of the other visual test
measurements.
4
5
DCE = dichloroethylene.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-104 DRAFT—DO NOT CITE OR QUOTE
-------
1 In a chamber exposure study (Vernon and Ferguson, 1969), eight male volunteers (ages
2 21-30) were exposed to 0, 100, 300, and 1,000-ppm TCE for 2 hours. Each individual was
3 exposed to all TCE concentrations and a span of at least three days was given between
4 exposures. When the individuals were exposed to 1,000-ppm TCE (5,500 mg/m3), significant
5 abnormalities were noted in depth perception as measured by the Howard-Dolman test
6 (p< 0.01). There were no effects on the flicker fusion frequency test (threshold frequency at
7 which the individual sees a flicker as a single beam of light) or on the form perception illusion
8 test (volunteers presented with an illusion diagram).
9
10 4.3.4.2. Visual Effects: Laboratory Animal Data
11 Changes in visual function have been demonstrated in animal studies during acute
12 (Boyes et al., 2003, 2005) and subchronic exposure (Rebert et al., 1991; Blain et al., 1994). In
13 these studies, the effect of TCE on visual evoked responses to patterns (Boyes et al., 2003, 2005;
14 Rebert et al., 1991) or a flash stimulus (Rebert et al., 1991; Blain et al., 1994) were evaluated.
15 Overall, the studies demonstrated that exposure to TCE results in significant changes in the
16 visual evoked response, which is reversible once TCE exposure is stopped. Details of the studies
17 are provided below and are summarized in Table 4-26.
18 Boyes et al. (2003, 2005) exposed adult, male Long-Evans rats were to TCE in a head-
19 only exposure chamber while pattern onset/offset visual evoked potentials (VEPs) were
20 recorded. Exposure conditions were designed to provide concentration x time products of
21 0 ppm/hours (0 ppm for 4 hours) or 4,000 ppm/hours (see Table 4-26 for more details). VEP
22 amplitudes were depressed by TCE exposure during the course of TCE exposure. The degree of
23 VEP depression showed a high correlation with the estimated brain TCE concentration for all
24 levels of atmospheric TCE exposure.
25 In a subchronic exposure study, Rebert et al. (1991) exposed male Long Evans rats to
26 1,600- or 3,200-ppm TCE, for 12 weeks, 12 hours/day. No significant changes in flash evoked
27 potential measurements were reported following this exposure paradigm. Decreases in pattern
28 reversal visual evoked potentials (N1P1 amplitude) reached statistical significance following 6,
29 9, and 12 weeks of exposure. The drop in response amplitude ranged from approximately 20%
30 after 8 weeks to nearly 50% at Week 14 but recovered completely within 1 week postexposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-105 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-26. Summary of animal visual system studies
Reference
Rebert et
al, 1991
Boyes et
al., 2003
Boyes et
al., 2005
Blain et
al., 1994
Exposure route
Inhalation
Inhalation
Inhalation
Inhalation
Species/strain/
sex/number
Rat, Long
Evans, male,
10/group
Rat, Long
Evans, male,
9- 10/group
Rat, Long
Evans, male,
8- 10/group
Rabbit, New
Zealand albino,
male, 6-8/group
Dose level/
exposure duration
0, 1,600, and 3,200
ppm; 12 h/d, 12 wks
0 ppm, 4 h; 1,000
ppm, 4; 2,000 ppm,
2h;
3,000 ppm, 1.3 h
4,000 ppm, 1 h
0 ppm, 4 h;
500 ppm, 4 h;
1,000 ppm, 4 h;
2,000 ppm, 2 h;
3,000 ppm, 1.3 h
4,000 ppm, 1 h;
5,000 ppm, 0.8 h
0, 350, 700 ppm;
4 h/d, 4 d/wk, 12 wks
NOAEL;
LOAEL
NOAEL:
1,600 ppm
LOAEL:
1,000 ppm,
4h
LOAEL:
500 ppm,
4h
LOAEL:
350 ppm
Effects
Significant amplitude
decreases in pattern
reversal evoked potentials
(N1P1 amplitude) at 6, 9,
and 12 wks.
Visual function
significantly affected as
measured by decreased
amplitude (F2) in Fourier-
transformed visual evoked
potentials. Peak brain TCE
concentration correlated
with dose response.
Visual function
significantly affected as
measured by decreased
amplitude (F2) in Fourier-
transformed visual evoked
potentials. Peak brain TCE
concentration correlated
with dose response.
Significant effects noted in
visual function as measured
by ERG and OPs
immediately after exposure.
No differences in ERG or
OP measurements were
noted at 6 wks post-TCE
exposure.
3
4
5
6
7
8
9
10
11
12
This transient effect of TCE on the peripheral visual system has also been reported by
Blain (1994) in which New Zealand albino rabbits were exposed by inhalation to 350- and
700-ppm TCE 4 hours/day, 4 days/week for 12 weeks. Electroretinograms (ERG) and
oscillatory potentials (OPs) were recorded weekly under mesopic conditions. Recordings from
the 350- and 700-ppm exposed groups showed a significant increase in the amplitude of the a-
and b-waves (ERG). The amplitude of the OPs was significantly decreased at 350 ppm (57%)
and increased at 700 ppm (117%). These electroretinal changes returned to preexposure
conditions within six weeks after the inhalation stopped.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-106 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.3.4.3. Summary and Conclusion of Visual Effects
2 Changes in visual function are reported in human studies. Although central visual function
3 was not evaluated in the human studies (such as electroretinograms, evoked potential
4 measurements), clinical tests indicated deficits in color discrimination (Kilburn, 2002a), visual
5 depth perception (Vernon and Ferguson, 1969) and contrast sensitivity (Reif et al., 2003). These
6 changes in visual function were observed following both an acute exposure (Vernon and Ferguson,
7 1969) and residence in areas with groundwater contamination with TCE and other chemicals
8 (Kilburn, 2002a; Reif et al., 2003). The exposure assessment approach of Reif et al., who adopted
9 exposure modeling and information on water distribution patterns, is considered superior to that of
10 Kilburn (2002a) who used residence location as a surrogate for exposure. In the one acute,
11 inhalation study (Vernon and Ferguson, 1969), a NOAEL of 300 ppm and a LOAEL of 1,000 ppm
12 for 2 hours was reported for visual effects. A NOAEL is not available from the drinking water
13 studies since well water TCE concentration is a poor surrogate for an individual's TCE ingestion
14 (Kilburn, 2002a) and limited statistical analysis comparing high exposure group to low exposure
15 group (Reif et al., 2003).
16 Animal studies have also demonstrated changes in visual function. All of the studies
17 evaluated central visual function by measuring changes in evoked potential response following a
18 visual stimulus that was presented to the animal. Two acute exposure inhalation studies (Boyes et
19 al., 2003, 2005) exposed Long Evans rats to TCE based on a concentration x time schedule
20 (Haber's law) and reported decreases in visual evoked potential amplitude. All of the exposures
21 from these two studies resulted in decreased visual function with a LOAEL of 500 ppm for
22 4 hours. Another important finding that was noted is the selection of the appropriate dose metric
23 for visual function changes following an acute exposure. Boyes et al. (2003, 2005) found that
24 among other potential dose metrics, brain TCE concentration was best correlated with changes in
25 visual function as measured by evoked potentials under acute exposure conditions. Two
26 subchronic exposure studies (Rebert et al., 1991; Blain et al., 1994) demonstrated visual function
27 changes as measured by pattern reversal evoked potentials (Rebert et al., 1991) or
28 electroretinograms/oscillatory potentials (Blain et al., 1994). Unlike the other three visual function
29 studies conducted with rats, Blain et al. demonstrated these changes in rabbits. Significant changes
30 in ERGs and oscillatory potentials were noted following a 12-week exposure at 350 ppm (LOAEL)
31 in rabbits (Blain et al., 1994) and in rats exposed to 3,200-ppm TCE for 12 weeks there were
32 significant decreases in pattern reversal evoked potentials but no effect was noted in the 1,600-ppm
33 exposure group (Rebert et al., 1991). Both subchronic studies examined visual function following
34 an exposure-free period of either 2 weeks (Rebert et al., 1991) or 6 weeks (Blain et al., 1994) and
35 found that visual function returned to pre-exposure levels and the changes are reversible.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-107 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.3.5. Cognitive Function
2 4.3.5.1. Cognitive Effects: Human Studies
3 Effects of TCE on learning and memory have been evaluated in populations
4 environmentally exposed to TCE through well water, in workers occupationally exposed through
5 inhalation and under controlled exposure scenarios. Details of the studies are provided in
6 Table 4-27 and discussed briefly below. In the geographical-based studies (Kilburn and
7 Warshaw, 1993; Kilburn, 2002a), cognitive function was impaired in both studies and was
8 evaluated by testing verbal recall and digit span memory among other measures. In Arizona
9 residents involved in a lawsuit (Kilburn and Warshaw, 1993), significant impairments in all three
10 cognitive measures were reported; verbal recall (p = 0.001), visual recall (p = 0.03) and digit
11 span test (p = 0.07), although a question exists whether the referent group was comparable to
12 exposed subjects and the study's lack of consideration of possible confounding exposures in
13 statistical analyses. Significant decreases in verbal recall ability was also reported in another
14 environmental exposure study where 236 residents near a microchip plant with TCE
15 concentration in well water ranging from 0.2-10,000 ppb (Kilburn, 2002a).
16 Cognitive impairments are assessed in the occupational exposure and case studies
17 (Rasmussen, 1993a, b; Troster and Ruff, 1990). In metal degreasers occupationally exposed to
18 TCE and CFC113, significant cognitive performance decreases were noted in verbal recall
19 testing (p = 0.03) and verbal learning (p = 0.04; Rasmussen et al., 1993a). No significant effects
20 were found in the visual recall or digit span test for these workers. Troster and Ruff (1990)
21 reported decrements (no statistical analysis performed) in cognitive performance as measured in
22 verbal and visual recall tests that were conducted immediately after presentation (learning phase)
23 and one hour after original presentation (retention/memory phase) for two case studies.
24 Several controlled (chamber) exposure studies were conducted to cognitive ability during
25 TCE exposure and most did not find any significant decrements in the neurobehavioral
26 measurement. Only Salvini et al. (1971) found significant decrements in cognitive function. Six
27 males were exposed to 110 ppm (550 mg/m3) TCE for 4-hour intervals, twice per day.
28 Statistically significant results were observed for perception tests learning (p < 0.001), mental
29 fatigue (p < 0.01), subjects (p < 0.05); and choice reaction time (CRT) learning (p < 0.01),
30 mental fatigue (p < 0.01), subjects (p < 0.05). Triebig et al. (1977a, b) exposed 7 total subjects
31 (male and female) to 100 ppm TCE for 6 hours/day, 5 days/week and did not report any
32 decreases in cognition but details on the experimental procedures were not provided.
33 Additionally, Gamberale et al. (1976) found that subjects exposed to TCE as high as 194 ppm for
34 70 minutes did not exhibit any impairments on a short term memory test in comparison to an air
35 exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-108 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-27. Summary of human cognition effect studies
Reference
Subjects
Exposure
Effect
Kilburn and
Warshaw, 1993
170 residents living in
Southwest Tucson with
TCE, other solvents, and
chromium in
groundwater.
Control: 68 residential
referents matched to
subjects from 2 previous
studies of waste oil and
oil refinery exposures.
>500 ppb of TCE in well
water before 1981 and 25 to
100 ppb afterwards
Exposure duration ranged
from 1 to 25 yrs
Decreased performance in the digit span
memory test and story recall ability.
Kilburn, 2002a
236 residents near a
microchip plant;
Controls: 67 local
referents from Phoenix,
AZ and 161 regional
referents from
Wickenburg, AZ.
0.2-10,000 ppb of TCE,
0.2-260,000 ppb TCA,
0.2-6,900 ppb 1,1-DCE,
0.2-1,600 1,2-DCE,
0.2-23,000 ppb PCE,
0.02-330 ppb VCin well
water
Exposure duration ranged
from 2 to 37 yrs. Exposure
duration ranged from 2 to 37
yrs
Cognitive effects decreased as measured
by lower scores on Culture Fair 2A,
vocabulary, grooved pegboard
(dominant hand), trail making test, and
verbal recall (i.e., memory).
Rasmussen,
1993a, b
96 Danish metal
degreasers. Age range:
19-68; No external
controls.
Average exposure duration:
7.1 yrs.); range of full-time
degreasing: 1 month to 36 yrs
1) Low exposure: n = 19,
average full-time expo 0.5 yrs
2) Medium exposure: n = 36,
average full-time exposure 2.1
yrs
3) High exposure: n = 41,
average full-time exposure 11
yrs. TCA in high exposure
group = 7.7 mg/L (max = 26.1
mg/L)
Cognitive impairment (psycho-organic
syndrome) prevalent in exposed
individuals. The incidence of this
syndrome was 10.5% in the low
exposure, 39.5% for medium exposure,
and 63.4% for high exposure. Age is a
confounder. Dose-response with 9 of
15 tests; Controlling for confounds,
significant relationship of exposure was
found with Acoustic-motor function
(p < 0.001), Paced Auditory Serial
Addition Test (p < 0.001), Rey Auditory
Verbal-Learning Test (p < 0.001),
vocabulary (p < 0.001) and visual
gestalts (p < 0.001); significant age
effects. Age is a confounder.
Troster and
Ruff, 1990
2 occupationally TCE-
exposed workers.
Controls: 2 groups of
n = 30 matched controls;
(all age and education
matched.
Exposure concentration
unknown; Exposure duration,
3-8 months
Both TCE cases exhibited significant
deficits in verbal recall and visuospatial
learning.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-109 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-27. Summary of human cognition effect studies (continued)
2
3
4
5
Reference
Triebig, 1976
Triebig, 1977a
Triebig, 1977b
Salvinietal.,
1971
Gamberale et
al, 1976
Stewart et al.,
1970
Chalupa, 1960
Subjects
Controlled exposure
study 4 females, 3 males
Controls: 4 females, 3
males
7 men and 1 woman
occupationally exposed
with an age range from
23-38 yrs. No control
group.
Controlled exposure
study on 3 male and 4
female students
Control: 3 male and 4
female students
Controlled exposure
study 6 students, male
Self used as control
15 healthy men aged
20-31 yrs old
Controls: Within
Subjects (15 self-
controls)
130 (108 males, 22
females); Controls: 63
unexposed men
Case study - Six
subjects. Average age
38
Exposure
0, 100 ppm (550 mg/m3), 6
h/d, 5 d.
50 ppm (260 mg/m3).
Exposure duration not
reported
0, 100 ppm (550 mg/m3), 6
h/d, 5d
TCE concentration was 110
ppm for 4-hour intervals,
twice per day. 0 ppm control
exposure for all as self
controls
0 mg/m3, 540 mg/m3 (97
ppm), 1,080 mg/m3 (194
ppm), 70 min
TCA metabolite levels in urine
were measured: 60.8% had
levels up to 20 mg/L, and
82. 1% had levels up to 60
mg/L
No exposure data were
reported
Effect
There was no correlation seen between
exposed and unexposed subjects for any
measured psychological test results. No
methods description was provided.
The psychological tests showed no
statistically significant difference in the
results before or after the exposure-free
time period. No methods description
was provided.
No significantly different changes were
obtained. No methods description was
provided.
Statistically significant results were
observed for perception tests learning
(p < 0.001) and CRT learning
(p<0.01).
Repetition of the testing led to
a pronounced improvement in
performance as a result of the
training effect; No interaction effects
between exposure to TCE and training.
No significant effect on cognitive tests
noted, but more effort required to
perform the test in exposed group.
80% of those with pathological EEG
displayed memory loss; 30% of those
with normal EEGs displayed memory
loss.
DCE = dichloroethylene, EEG = electroencephalogram.
6 4.3.5.2. Cognitive Effects: Laboratory Animal Studies
1 Many reports have demonstrated significant differences in performance of learning tasks
8 such as the speed to complete the task. However, there is little evidence that learning and
9 memory function are themselves impaired by exposure. There are also limited data that suggest
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-110 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
alterations in the hippocampus of laboratory animals exposed to TCE. Given the important role
that this structure plays in memory formation, such data may be relevant to the question of
whether TCE impairs memory. The studies are briefly discussed below and details are provided
in Table 4-28.
Table 4-28. Summary of animal cognition effect studies
Reference
Kjellstrand
etal., 1980
Isaacson et
al, 1990
Kishietal.,
1993
Umezu et
al., 1997
Oshiro et
al., 2004
Exposure
route
Inhalation
Oral,
drinking
water
Inhalation
Intra-
peritoneal
Inhalation
Species/strain/
sex/number
Geibil,
Mongolian,
males and
females,
12/sex/dose
Rat, Sprague-
Dawley, male
weanlings,
12/dose
Rats, Wistar,
male, number
not specified
Mouse, ICR,
male, 6 exposed
to all treatments
(repeated
exposure)
Rat, Long
Evans, male, 24
Dose level/
exposure duration
0, 320 ppm; 9 months,
continuous (24 h/d) except
1-2 h/wk for cage cleaning
(1) 0 mg/kg/d, 8 wks
(2) 5.5 mg/d (47 mg/kg/d*),
4 wks + 0 mg/kg/d, 4 wks
(3) 5.5 mg/dd, 4 wks (47
mg/kg/db) + 0 mg/kg/d, 2
wks + 8.5 mg/dd (24
mg/kg/db), 2 wks
0, 250,500, 1,000, 2,000,
and 4,000 ppm, 4 hours
0, 125, 250, 500, and 1,000
mg/kg, single dose and
evaluated 30 min
postadministration
0, 1,600, and 2,400 ppm; 6
h/d, 5 d/wk, 4 wks
NOAEL;
LOAEL
NOAEL: 320
ppm
NOAEL: 5.5
mg/d, 4 wks —
spatial learning
LOAEL: 5.5
mg/d —
hippocampal
demyelination
LOAEL: 250
ppm
NOAEL: 500
mg/kg
LOAEL: 1,000
mg/kg
NOAEL: 2,400
ppm
Effects
No significant effect
on spatial memory
(radial arm maze).
Decreased latency to
find platform in the
Morris water maze
(Group #3);
Hippocampal
demyelination
observed in all TCE-
treated groups.
Decreased lever
presses and avoidance
responses in a shock
avoidance task.
Decreased response
rate in an operant
response — condition
avoidance task.
No change in reaction
time in signal
detection task and
when challenged with
amphetamine, no
change in response
from control.
9
10
11
12
13
14
15
*mg/kg/d conversion estimated from average male Sprague-Dawley rat body weight from ages 21-49 days (118 g)
for the 5.5 mg dosing period and ages 63-78 days (354 g) for the 8.5 mg dosing period.
Two studies (Kulig et al., 1987; Umezu et al., 1997) reported decreased performance in
operant-conditioning cognitive tasks for rodents. Kishi et al. (1993) acutely exposed Wistar rats
to TCE at concentrations of 250, 500, 1,000, 2,000, and 4,000 ppm for four hours. Rats exposed
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-111 DRAFT—DO NOT CITE OR QUOTE
-------
1 to 250 ppm TCE and higher showed a significant decrease both in the total number of lever
2 presses and in avoidance responses compared with controls. The rats did not recover their pre-
3 exposure performance until about 2 hours after exposure. Likewise, Umezu et al. (1997)
4 reported a depressed rate of operant responding in male ICR strain mice (n = 6, exposed to all
5 TCE doses, see Table 4-28) in a conditioned avoidance task that reached significance with i.p.
6 injections of 1,000 mg/kg. Increased responding during the signaled avoidance period at lower
7 doses (250 and 500 mg/kg) suggests an impairment in ability to inhibit responding or failure to
8 attend to the signal.
9 Although cognitive impairments are noted, two additional studies indicate no change in
10 cognition with continuous TCE exposure or improvements in cognitive tasks. No decrements in
11 cognitive function as measured by the radial arm maze were observed in Mongolian gerbils
12 exposed continuously by inhalation to 320 ppm TCE for 9 months (Kjellstrand et al., 1980).
13 Improved performance was noted in a Morris swim test for weanling rats orally dosed with
14 5.5 mg/day for 4 weeks followed by 2 weeks of no exposure and an additional 2 weeks of
15 8.5 mg/day (Isaacson et al., 1990). This improved performance occurred despite a loss in
16 hippocampal myelination.
17
18 4.3.5.3. Summary and Conclusions of Cognitive Function Studies
19 Human environmental and occupational exposure studies suggest impairments in
20 cognitive function. Kilburn and Warshaw (1993) and Kilburn (2002a) reported memory deficits
21 individuals although a question exists whether the referent group was comparable to exposed
22 subjects and these studies lack of consideration of possible confounding exposures in statistical
23 analyses. Significant impairments were found in visual and verbal recall and with the digit span
24 test. Similarly, in occupational exposure studies (Rasmussen et al., 1993a, b; Troster and Ruff,
25 1990), short term memory tests indicated that immediate memory and learning were impaired in
26 the absence of an effect on digit span performance. In controlled exposure and/or chamber
27 studies, two studies did not report any cognitive impairment (Stewart et al., 1970; Gamberale et
28 al., 1976) and one study (Salvini et al., 1971) reported significant impairments in learning
29 memory and complex choice reaction tasks. All of the controlled exposure studies were acute
30 and/or short-term exposure studies and the sensitivity of test procedures is unknown due to the
31 lack of methodologic information provided in the reports. Despite identified study deficiencies,
32 these studies collectively suggest cognitive function impairment.
33 The animal studies measured cognitive function through spatial memory and operant
34 responding tasks. In the two studies where spatial memory was evaluated, there was either no
35 effect at 320 ppm TCE (Kjellstrand et al., 1980) or improved cognitive performance in weanling
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-112 DRAFT—DO NOT CITE OR QUOTE
-------
1 rats at a dose of 5.5 mg/day for four weeks (Isaacson et al., 1990). Improved cognitive
2 performance was observed in weanling rats (Isaacson et al., 1990) and could be due to
3 continuing neurodevelopment as well as compensation from other possible areas in the brain
4 since there was a significant loss in hippocampal myelination. Significant decreases in operant
5 responding (avoidance/punished responding) during TCE exposure were reported in two studies
6 (Kishi et al., 1993; Umezu et al., 1997). When TCE exposure was discontinued operant
7 responding return to control levels and it is unclear if the significant effects are due to decreased
8 motor function or decreased cognitive ability.
9
10 4.3.6. Psychomotor Effects
11 There is considerable evidence in the literature for both animals and humans on
12 psychomotor testing although human and laboratory animal studies utilize very different
13 measures of motor behavior. Generally, the human literature employs a wide variety of
14 psychomotor tasks and assesses error rates and reaction time in the performance of the task. The
15 laboratory animal data, by contrast, tend to include unlearned naturalistic behaviors such as
16 locomotor activity, gait changes, and foot splay to assess neuromuscular ability.
17
18 4.3.6.1. Psychomotor Effects: Human Studies
19 The effects of TCE exposure on psychomotor response have been studied primarily as a
20 change in reaction time (RT) with studies on motor dyscoordination resulting from TCE
21 exposure providing subjective reporting.
22
23 4.3.6.1.1. Reaction time. Several studies have evaluated the effects of TCE on reaction time
24 using simple and choice reaction time tasks (simple reaction time [SRT] and CRT tasks). The
25 studies are presented below and summarized in more detail in Table 4-29.
26
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-113 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-29. Summary of human choice reaction time studies
Reference
Subjects
Exposure
Effect
Kilburn,
2002a
236 residents near a microchip
plant in Phoenix, AZ
Controls:
161 regional referents from
Wickenburg, AZ
67 referents from Phoenix, AZ
not residing near a plant
0.2-10,000 ppb of
TCE, chronic exposure
Simple and choice reaction times were
increased in the exposed group (p <
0.05).
Kilburn and
Warshaw,
1993
160 residents living in
Southwest Tucson with TCE and
other solvents in groundwater
Control: 68 residential referents
matched to subjects from 2
previous studies of waste oil and
oil refinery exposures
>500ppbofTCEin
well-water before 1981
and 25 to 100 ppb
afterwards
Exposure duration
ranged from 1 to 25 yrs
Mean simple reaction time was 67
milliseconds (msec) longer than the
referent group/) < 0.0001).
CRT of the exposed subjects was
between 93-100 msec longer in three
different trials (p < 0.0001) compared to
referents.
Reifetal.,
2003
143 residents of the Rocky
Mountain Arsenal community of
Denver
Referent group at lowest
concentration (<5 ppb)
High exposure group
>15 ppb
Medium exposure
group >5 ppb and <15
ppb
Low exposure referent
group <5 ppb
Significant increase in reaction time as
measured by the simple reaction time
test (p < 0.04) in only among subjects
who reported alcohol use (defined as
having at least one drink per month).
Kilburn and
Thornton,
1996
Group A: Registered voters from
Arizona and Louisiana with no
exposure to TCE: n = 264, aged
18-83. Group B volunteers
from California n = 29 (17 males
and 12 females) Group C:
exposed to TCE and other
chemicals for 5 yrs or more
w = 217
No exposure or
groundwater analyses
reported
Significant increase in simple and
choice reaction time in exposed group
compared to the unexposed populations.
Gamberale et
al., 1976
15 healthy men aged 20-31 yrs
old
Controls: Within subjects (15
self-controls)
0 mg/m3, 540 mg/m3
(97 ppm), 1,080 mg/m3
(194 ppm), 70 min.
No change in CRT or SRT. Increase in
time required to perform the RT-
Addition Test (task for adding numbers)
(p<0.05).
Gunetal.,
1978
4 female workers from one plant
exposed to TCE and 4 female
workers from another plant
exposed to TCE +
nonhalogenated hydrocarbon
solvent
Control: (n = 8) 4 unexposed
female workers from each plant
3-419 ppm, duration
not specified
TCE-only exposure increased reaction
time in comparison to controls. In TCE
+ solvent group, ambient TCE was
lower and mean reaction time shortened
in Session 2, then rose subsequently to
be greater than at the start.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-114 DRAFT—DO NOT CITE OR QUOTE
-------
1 Increases in reaction time were observed in environmental exposure studies by Kilburn
2 (2002a), Kilburn and Warshaw (1993), and Kilburn and Thornton (1996) as well as in an
3 occupational exposure study by Gun et al. (1978). All populations except that of Gun et al.
4 (1978) were exposed through groundwater contaminated as the result of environmental spills and
5 the exposure duration was for at least 1 year and exposure levels ranged from 0.2 to 10,000 ppb
6 for the three studies. Kilburn and Warshaw (1993) reported that SRT significantly increased
7 from 281 ± 55 msec to 348 ± 96 msec in individuals (p < 0.0001). CRT of the exposed subjects
8 was 93 msec longer (p < 0.0001) than referents. Kilburn and Thornton (1996) evaluated SRT
9 and CRT function and also found similar increases in reaction time. The average SRT and CRT
10 for the combined control groups were 276 msec and 532 msec, respectively. These reaction
11 times increased in the TCE exposure group where the average SRT was 334 msec and CRT was
12 619 msec. Similarly, Kilburn (2002a) compared reaction times between 236 TCE-exposed
13 persons and the 161 unexposed regional controls. SRTs significantly increased from
14 283 ± 63 msec in controls to 334 ± 118 msec in TCE exposed individuals (p < 0.0001).
15 Similarly, CRTs also increased from 510 ± 87 msec to 619 ± 153 msec with exposure to TCE
16 (p< 0.0001).
17 No effect on SRT was reported in a geographical-based study by Reif et al. (2003). SRTs
18 were 301 msec for the lowest exposure group and 316 msec for the highest exposure group
19 (p = 0.42). When the SRT data were analyzed individuals that consumed at least on alcoholic
20 drink per month (n = 80), a significant increase (18%, p < 0.04) in SRT times were observed
21 between the lowest exposure and the highest exposure groups. In TCE exposed individuals who
22 did not consume alcohol (n = 55), SRTs decreased from 321 msec in the lowest exposed group to
23 296 msec in the highest exposed group, but this effect was not statistically significantly different.
24 A controlled exposure (chamber study) of 15 healthy men aged 20-31 years old, were exposed to
25 0, 540, and 1,080 mg/m3 TCE for 70 minutes or served as his own control, reported no
26 statistically significant differences with the SRT or CRT tasks. However, in the RT-Addition
27 test the level of performance varied between the different exposure conditions (F(2.24) = 4.35;
28 p< 0.05) and between successive measurement occasions (F(2.24) = 19.25; p < 0.001).
29
30 4.3.6.1.2. Muscular dyscoordination. Three studies examined motor dyscoordination effects
31 from TCE exposure using subjective and self-reported individual assessment. Rasmussen et al.
32 (1993c) presented findings on muscular dyscoordination for 96 metal degreasers exposed to
33 either TCE or CFC113. A statistically significant increasing trend of dyscoordination with TCE
34 exposure was observed (p = 0.01) in multivariate regression analyses which adjusted for the
35 effects of age, neurological disease, arteriosclerotic disease, and alcohol abuse. Furthermore, a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-115 DRAFT—DO NOT CITE OR QUOTE
-------
1 greater number of abnormal coordination tests were observed in the higher exposure group
2 compared to the low exposure group (p = 0.003).
3 Gash et al. (2008) reported fine motor hand movement times in subjects who had filed
4 workman compensation claims were significantly slower (p < 0.0001) than age-matched
5 nonexposed controls. Exposures were based on self-reported information, and no information on
6 the control group is presented. Troster and Ruff (1990) reported a case study conducted on two
7 occupationally exposed workers to TCE. Mild deficits in motor speed were reported for both
8 cases. In the first case, manual dexterity was impaired in a male exposed to TCE (unknown
9 concentration) for eight months. In the second case study where a female was exposed to TCE
10 (low concentration; exact level not specified) for 3 months, there was weakness in the quadriceps
11 muscle as evaluated in a neurological exam and a decreased sensation to touch on one hand.
12 Both Gash et al. (2008) and Troster and Ruff (1990) provide very limited information given their
13 deficiencies related to lack of exposure data, self-reported information, and limited reporting of
14 referents and statistical analysis.
15
16 4.3.6.2. Psychomotor Effects: Laboratory Animal Data
17 Several animal studies have demonstrated that TCE exposure produces changes in
18 psychomotor function. At high doses (>2,000 mg/kg) TCE causes mice to lose their righting
19 reflex when the compound is injected intraperitoneally (Shih et al., 2001; Umezu et al., 1997).
20 At lower exposures (inhalation and oral), TCE produces alterations in neurobehavioral measures
21 including locomotor activity, gait, operant responding, and reactivity. The studies are described
22 in Sections 4.3.6.2.1-4.3.6.2.3 and summarized in Tables 4-30 and 4-31.
23
24 4.3.6.2.1. Loss of righting reflex. Umezu et al. (1997) studied disruption of the righting reflex
25 following acute injection (i.p.) of 2,000, 4,000, and 5,000 mg/kg TCE in male ICR mice. TCE
26 disrupted the righting reflex at doses of 2,000 mg/kg and higher. At 2,000 mg/kg, loss of
27 righting reflex (LORR) was observed in only 2/10 animals injected. At 4,000 mg/kg,
28 9/10 animals experienced LORR and 100% of the animals experienced LORR at 5,000 mg/kg.
29 Shih et al. (2001) reported impaired righting reflexes at exposure doses of 5,000 mg/kg
30 (i.p.) in male Mfl mice. Mice pretreated with dimethyl sulfoxide or disulfuram (CYP2E1
31 inhibitor) delayed LORR in a dose related manner. By contrast, the alcohol dehydrogenase
32 inhibitor, 4-metylpyradine did not delay LORR that resulted from 5,000 mg/kg TCE. These data
33 suggest that the anesthetic properties of TCE involve its oxidation via CYP2E1 to an active
34 metabolite.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-116 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-30. Summary of animal psychomotor function and reaction time
studies
Reference
Savolainen
etal., 1977
Kishietal.,
1993
Kuligetal.,
1987
Moser et al.,
1995
Bushnell,
1997
Shih et al.,
2001
Umezu et
al., 1997
Bushnell
and Oshiro,
2000
Exposure
route
Inhalation
Inhalation
Inhalation
Oral
Inhalation
Intra-
peritoneal
Intra-
peritoneal
Inhalation
Species/strain/
sex/number
Rat, Sprague-
Dawley, male,
10
Rats, Wistar,
male, number
not specified
Rat, Wistar,
male, 8/dose
Rat, Fischer
344, female,
8/dose
Rat, Long
Evans, male, 12
Mouse, MF1,
male, 6
Mouse, ICR,
male, 10/group
Mouse, ICR,
male,
6- 10/group
Rat, Long
Evans, male, 32
Dose level/
exposure duration
0, 200 ppm; 6 h/d,
4d
0, 250,500, 1,000,
2,000, and 4,000
ppm, 4 hours
0, 500, 1,000, and
1,500 ppm; 16 h/d,
5 d/wk, 18 wks
0, 150, 500, 1,500,
and 5,000 mg/kg, 1
dose
0, 50, 150, 500,
and 1,500 mg/kg/d,
14 d
0, 400, 800, 1,200,
1,600, 2,000, or
2,400 ppm, 1 h/test
day, 4 consecutive
test days, 2 wks
0, 5,000 mg/kg,
acute
0, 2,000, 4,000,
5,000 mg/kg— loss
of righting reflex
measure
0, 62.5, 125, 250,
500, and 1,000
mg/kg, single dose
and evaluated 30
min
postadministration
0, 2,000, 2,400
ppm; 70 min/d, 9 d
NOAEL; LOAEL
LOAEL: 200 ppm
LOAEL: 250 ppm
NOAEL: 1,500 ppm
NOAEL: 500 mg/kg
LOAEL: 1,500
mg/kg
NOAEL: 150
mg/kg/d
LOAEL: 500
mg/kg/d
NOAEL: 800 ppm
LOAEL: 1,200 ppm
LOAEL: 5,000
mg/kg
LOAEL: 2,000
mg/kg — loss of
righting reflex
NOAEL: 500 mg/kg
LOAEL: 1,000
mg/kg — operant
behavior
NOAEL: 125 mg/kg
LOAEL: 250
mg/kg — punished
responding
LOAEL: 2,000 ppm
Effects
Increased frequency of
preening, rearing, and
ambulation. Increased
preening time.
Decreased lever presses
and increased responding
when lever press coupled
with a 10-s electric shock
(decreased avoidance
response).
No change in spontaneous
activity, grip strength, or
hindlimb movement.
Decreased motor activity;
Neuro-muscular and
sensorimotor impairment.
Increased rearing activity
and decreased forelimb
grip strength.
Decreased sensitivity and
increased response time
in the signal detection
task.
Impairment of righting
reflex.
Loss of righting reflex.
Decreased responses
(lever presses) in an
operant response task for
food reward.
Increased responding
when lever press coupled
with a 20-V electric
shock (punished
responding).
Decreased performance
on the signal detection
task. Increased response
time and decreased
response rate.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-117 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-30. Summary of animal psychomotor function and reaction time
studies (continued)
Reference
Nunes et al.,
2001
Moser et al.,
2003
Albee et al.,
2006
Exposure
route
Oral
Oral
Inhalation
Species/strain/
sex/number
Rat, Sprague-
Dawley, male,
10/group
Rat, Fischer
344, female,
10/group
Rat, Fischer
344, male and
female,
10/sex/group
Dose level/
exposure duration
0, 2,000 mg/kg/d,
7d
0, 40, 200, 800,
and 1,200 mg/kg/d,
10 d
0, 250, 800, 2,500
ppm; 6 h/d, 5 d/wk,
13wks
NOAEL; LOAEL
LOAEL: 2,000
mg/kg/d
NOAEL: 2,500 ppm
Effects
Increased foot splay. No
change in any other FOB
parameter (e.g.,
piloerection, activity,
reactivity to handling).
Decreased motor activity;
Decreased sensitivity to
tail pinch; Increased
abnormality in gait;
Decreased grip strength;
Adverse changes in
several FOB parameters.
No change in any FOB
measured parameter.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-118 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-31. Summary of animal locomotor activity studies
Reference
Wolff and
Siegmund,
1978
Kuligetal.,
1987
Moser et al.,
1995
Waseem et al.,
2001
Moser et al.,
2003
Exposure
route
Intra-
peritoneal
Inhalation
Oral
Oral
Inhalation
Oral
Species/strain/
sex/number
Mouse, AB,
male, 18
Rat, Wistar,
male, 8/dose
Rat, Fischer
344, female,
8/dose
Rat, Wistar,
male, 8/group
Rat, Wistar,
male, 8/group
Rat, Fischer
344, female,
10/group
Dose level/
Exposure duration
0, 182 mg/kg,
tested 30 min after
injection
0, 500, 1,000, and
1,500 ppm; 16 h/d,
5 d/wk, 18 wks
0, 150, 500, 1,500,
and 5,000 mg/kg,
1 dose
0, 50, 150, 500, and
1,500 mg/kg/d, 14 d
0, 350, 700, and
1,400 ppm in
drinking water for
90 d
0, 376 ppm for up
to 180 d; 4 h/d,
5 d/wk
0, 40, 200, 800, and
1,200 mg/kg/d, 10 d
NOAEL;
LOAEL
LOAEL:
182 mg/kg
NOAEL:
500 ppm
LOAEL:
1,000 ppm
NOAEL:
500 mg/kg
LOAEL:
1,500 mg/kg
NOAEL:
150 mg/kg/d
LOAEL:
500 mg/kg/d
NOAEL:
1,400 ppm
LOAEL:
376 ppm
Effects
Decreased spontaneous motor
activity.
No change in spontaneous
activity, grip strength or
hindlimb movement.
Increased latency time in the
two-choice visual
discrimination task (cognitive
disruption and/or motor activity
related effect).
Decreased motor activity;
Neuro-muscular and
sensorimotor impairment.
Increased rearing activity.
No significant effect on
spontaneous locomotor
activity.
Changes in locomotor activity
and vary by timepoint when
measured over the 180-d
period.
Decreased motor activity;
Decreased sensitivity;
Increased abnormality in gait;
Adverse changes in several
FOB parameters.
3
4
5
6
7
8
9
10
11
12
4.3.6.2.2. Activity, sensory-motor and neuromuscular function. Changes in sensory-motor
and neuromuscular activity was reported in three studies (Kishi et al., 1993; Moser et al., 1995;
Moser et al., 2003). Kishi et al. (1993) exposed male Wistar rats to 250, 500, 1,000, 2,000, and
4,000 ppm TCE for 4 hours. Rats exposed to 250-ppm TCE showed a significant decrease both
in the total number of lever presses and in avoidance responses at 140 minutes of exposure
compared with controls. Moser et al. (1995) evaluated the effects of acute and short-term
(14 day) administration of TCE in adult female Fischer 344 rats (n = 8-10/dose) on activity
level, neuromuscular function and sensorimotor function as part of a larger functional
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-119 DRAFT—DO NOT CITE OR QUOTE
-------
1 observational battery (FOB) testing. The NOAEL levels identified by the authors are 500 mg/kg
2 (10% of the limit dose) for the acute treatment and 150 mg/kg (3% of the limit dose) for the
3 14-day study. In the acute study, TCE produced the most significant effects in motor activity
4 (activity domain), gait (neuromuscular domain), and click response (sensorimotor domain). In
5 the 14-day study, only the activity domain (rearing) and neuromuscular domain (forelimb grip
6 strength) were significantly different (p < 0.05) from control animals. In a separate 10-day study
7 (Moser et al., 2003), TCE administration significantly (p < 0.05) reduced motor activity, tail
8 pinch responsiveness, reactivity to handling, hind limb grip strength and body weight.
9 Significant increases (p < 0.05) in piloerection, gait scores, lethality, body weight loss, and
10 lacrimation was also reported in comparison to controls.
11 There are also two negative studies which used adequate numbers of subjects in their
12 experimental design but used lower doses than did Moser et al. (2003). Albee et al. (2006)
13 exposed male and female Fischer 344 rats (n = 10/sex) to TCE by inhalation at exposure doses of
14 250, 800, and 2,500 ppm, for 6 hours/day, 5 days/week, for 13 weeks. The FOB was performed
15 monthly although it is not certain how much time elapsed from the end of exposure until the
16 FOB test was conducted. No treatment related differences in grip strength or landing foot splay
17 were demonstrated in this study. Kulig et al. (1987) also failed to show significant effects of
18 TCE inhalation exposure on markers of motor behavior. Wistar rats (n = 8) exposed to 500,
19 1,000, and 1,500 ppm, for 16 hours/day, 5 days/week, for 18 weeks failed to show changes in
20 spontaneous activity, grip strength, or coordinated hind limb movement. Measurements were
21 made every three weeks during the exposure period and occurred between 45 and 180 minutes
22 following the previous TCE inhalation exposure.
23
24 4.3.6.2.3. Locomotor activity. The data, with regard to locomotor activity, are inconsistent.
25 Several studies showed that TCE exposure can decrease locomotor activity including Wolff and
26 Siegmund (1978) where AB mice (n = 18) were treated acutely with a dose of 182 mg/kg, i.p. at
27 one of 4 time points during a 24-hour day. Moser et al. (1995, 2003) reported reduced locomotor
28 activity in female Fischer 344 rats (n = 8-10) gavaged with TCE over an acute
29 (LOAEL = 5,000 mg/kg TCE) or subacute period (LOAEL = 500 but no effect at 5,000 mg/kg).
30 In the Moser et al. (2003), it appears that 200-mg/kg TCE yielded a significant reduction in
31 locomotor activity and that the degree of impairment at this dose represented a maximal effect on
32 this measure. That is, higher doses of TCE appear to have produced equivalent or slightly less of
33 an effect on this behavior. While this study identifies a LOAEL of 200-mg/kg TCE by gavage
34 over a 10-day period, this is a much more lower dose effect than that reported in Moser et al.
35 (1995). Both studies (Moser et al., 1995, 2003) demonstrate a depression in motor activity that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-120 DRAFT—DO NOT CITE OR QUOTE
-------
1 occurs acutely following TCE administration. Kulig et al. (1987) demonstrated that rats had
2 increased response latency to a two choice visual discrimination following 1,000- and 1,500-ppm
3 TCE exposures for 18 weeks. However, no significant changes in grip strength, hindlimb
4 movement, or any other motor activity measurements were noted.
5 There are also a few studies (Fredriksson et al., 1993; Waseem et al., 2001) generally
6 conducted using lower exposure doses that failed to demonstrate impairment of motor activity or
7 ability following TCE exposure. Waseem et al. (2001) failed to demonstrate changes in
8 locomotor activity in male Wistar rats (n = 8) dosed with TCE (350, 700, and 1,400 ppm) in
9 drinking water for 90 days. Wistar rats (n = 8) exposed to 500, 1,000, and 1,500 ppm for
10 16 hours/day, 5 days/week, for 18 weeks failed to show changes in spontaneous activity. No
11 changes in locomotor activity were observed for 17-day-old male NMRI mice that were dosed
12 postnatally with 50 or 290 mg/kg/d from Day 10 to 16 (Fredriksson et al., 1993). However,
13 rearing activity was significantly decreased in the NMRI mice at Day 60.
14
15 4.3.6.3. Summary and Conclusions for Psychomotor Effects
16 In human studies, psychomotor effects such as reaction time and muscular
17 dyscoordination have been examined following TCE exposure. In the reaction time studies,
18 statistically significant increases in CRT and SRT were reported in the Kilburn studies (Kilburn,
19 2002a; Kilburn and Warshaw, 1993; Kilburn and Thornton, 1996). All of these studies were
20 geographically based and it was suggested that the results were used for litigation and the
21 differences between exposed and referent groups on other factors influencing reaction speed time
22 may introduce a bias to the findings. Additionally, in these studies exposure to TCE and other
23 chemicals occurred through drinking water for at least 1 year and TCE concentrations in well
24 water ranged from 0.2 ppb to 10,000 ppb. Reif et al. (2003) whose exposure assessment
25 approach included exposure modeling of water distribution system to estimate TCE
26 concentrations in tap water at census track of residence found that residents with drinking water
27 containing TCE (up to >15 ppb—the highest level not specified) and other chemicals did not
28 significantly increase CRTs or SRTs. Inhalation studies also demonstrated increased reaction
29 times. An acute exposure chamber study (Gamberale et al., 1976) tested for CRT, SRT, and RT-
30 addition following a 70-minute exposure to TCE. A concentration-dependent significant
31 decrease in performance was observed with the RT-addition test and not for CRT or SRT tasks.
32 An occupational exposure study on 8 female workers exposed to TCE (Gun et al., 1978) also
33 reported increased reaction time in the females exposed to TCE-only. Muscular dyscoordination
34 for humans following TCE exposure has been reported in a few studies as a subjective
35 observation. The studies indicated that exposure resulted in decreased motor speed and dexterity
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-121 DRAFT—DO NOT CITE OR QUOTE
-------
1 (Troster and Ruff, 1990; Rasmussen et al., 1993c) and self-reported faster asymptomatic fine
2 motor hand movements (Gash et al., 2008).
3 Animal studies evaluated psychomotor function by examining locomotor activity, operant
4 responding, changes in gait, loss of righting reflex, and general motor behavior (see Tables 4-30
5 and 4-31 for references). Overall, the studies demonstrated that TCE causes loss of righting
6 reflex at injection doses of 2,000 mg/kg or higher (Umezu et al., 1997; Shih et al., 2001).
7 Regarding general psychomotor testing, significant decreases in lever presses and avoidance
8 were observed at inhalation exposures as low as 250 ppm for 4 hours (LOAEL; Kishi et al.,
9 1993). Following subchronic inhalation exposures, no significant changes in psychomotor
10 activity were noted at up to 2,500 ppm for 13 weeks (Albee et al., 2006) or at 1,500 ppm for
11 18 weeks (Kulig et al., 1987). In the oral administration studies (Moser et al., 1995, 2003),
12 psychomotor effects were evaluated using an FOB. More psychomotor domains were
13 significantly affected by TCE treatment in the acute study in comparison to the 14-day study, but
14 a lower NOAEL (150 mg/kg/d) was reported for the 14-day study in comparison to the acute
15 study (500 mg/kg; Moser et al., 1995). Upon closer examination of the data, a biphasic effect in
16 one measure of the FOB (rearing) was resulting in the lower NOAEL for the 14-day study and
17 doses that were higher and lower than the NOAEL did not produce a statistically significant
18 increase in the number of rears. Therefore, it can be surmised that acute exposure to TCE results
19 in significant changes in psychomotor function. However, there may be some tolerance to these
20 psychomotor changes in increased exposure duration to TCE as evidenced by the results noted in
21 the short-term and subchronic exposure studies.
22
23 4.3.7. Mood Effects and Sleep Disorders
24 4.3.7.1. Effects on Mood: Human Studies
25 Reports of mood disturbance (depression, anxiety) resulting from TCE exposure are
26 numerous in the human literature. These symptoms are subjective and difficult to quantify.
27 Studies by Gash et al. (2008), Kilburn and Warshaw (1993), Kilburn (2002a, 2002b),
28 McCunney et al. (1988), Mitchell et al. (1969), Rasmussen and Sabroe (1986), and Troster and
29 Ruff (1990) reported mood disturbances in humans. Reif et al. (2003) and Triebig (1976, 1977)
30 reported no effect on mood following TCE exposures.
31
32 4.3.7.2. Effects on Mood: Laboratory Animal Findings
33 It is difficult to obtain comparable data of emotionality in laboratory studies. However,
34 Moser et al. (2003) and Albee et al. (2006) both report increases in handling reactivity among
35 rats exposed to TCE. In the Moser study, female Fischer 344 rats received TCE by oral gavage
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-122 DRAFT—DO NOT CITE OR QUOTE
-------
1 for periods of 10 days at doses of 0, 40, 200, 800, and 1,200 while Albee et al. (2006) exposed
2 Fischer 344 rats to TCE by inhalation at exposure doses of 250, 800, and 2,500 ppm for
3 6 hours/day, 5 days/week, for 13 weeks.
4
5 4.3.7.3. Sleep Disturbances
6 Arito et al. (1994) exposed male Wistar rats to 50-, 100-, and 300-ppm TCE for
7 8 hour/day, 5 days/week, for 6 weeks and measured electroencephalographic (EEG) responses.
8 EEG responses were used as a measure to determine the number of awake (wakefulness hours)
9 and sleep hours. Exposure to all the TCE levels significantly decreased amount of time spent in
10 wakefulness (W) during the exposure period. Some carry over was observed in the 22 hours post
11 exposure period with significant decreases in wakefulness seen at 100-ppm TCE. Significant
12 changes in W-sleep elicited by the long-term exposure appeared at lower exposure levels. These
13 data seem to identify a low dose effect of TCE and established a LOAEL of 50 ppm for sleep
14 changes.
15
16 4.3.8. Developmental Neurotoxicity
17 4.3.8.1. Human Studies
18 In humans, CNS birth defects were observed in a few studies (ATSDR, 2001; Bove,
19 1996; Bove et al., 1995; Lagakos et al., 1986). Postnatally, observed adverse effects in humans
20 include delayed newborn reflexes following exposure to TCE during childbirth (Beppu, 1968),
21 impaired learning or memory (Bernad et al., 1987, abstract; White et al., 1997); aggressive
22 behavior (Bernad et al., 1987, abstract); hearing impairment (Burg and Gist, 1999); speech
23 impairment (Burg and Gist, 1999; White et al., 1997); encephalopathy (White et al., 1997);
24 impaired executive and motor function (White et al., 1997); attention deficit (Bernad et al., 1987,
25 abstract; White et al., 1997), and autism spectrum disorder (Windham et al., 2006). The human
26 developmental neurotoxicity studies are discussed in more detail in Section 4.8.2.1.2, and
27 summarized in Table 4-32.
28
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-123 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-32. Summary of human developmental neurotoxicity associated with
TCE exposures
Finding
CNS defects, neural tube defects
Delayed newborn reflexes
Impaired learning or memory
Aggressive behavior
Hearing impairment
Speech impairment
Encephalopathy
Impaired executive function
Impaired motor function
Attention deficit
Autism spectrum disorder (ASD)
Species
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Citations
ATSDR, 2001
Bove, 1996; Bove et al., 1995
Lagakos et al., 1986
Beppu, 1968
Bernad et al., 1987, abstract
White etal., 1997
Bernad et al., 1987, abstract
Burg and Gist, 1999
Burg and Gist, 1999
White etal., 1997
White etal., 1997
White etal., 1997
White etal., 1997
White et al., 1997
Bernad et al., 1987, abstract
Windham et al., 2006
4
5
6 4.3.8.2. Animal Studies
1 There are a few studies demonstrating developmental neurotoxicity following
8 trichloroethylene exposure (range of exposures) to experimental animals. These studies
9 collectively suggest that developmental neurotoxicity result from TCE exposure, however, some
10 types of effects such as learning and memory measures have not been evaluated. Most of the
11 studies demonstrate either spontaneous motor activity changes (Taylor et al., 1985) or
12 neurochemical changes such as decreased glucose uptake and changes in the specific gravity of
13 the cortex and cerebellum (Westergren et al., 1984; Noland-Grebec et al., 1986; Isaacson and
14 Taylor, 1989). In addition, in most of these studies there is no assessment of the exposure to
15 TCE or metabolites in the pups/offspring. Details of the studies are presented below and
16 summarized in Table 4-33.
17 Taylor et al. (1985) administered TCE to female Sprague-Dawley rats in their drinking
18 water from 14 days before breeding throughout gestation and until pups were weaned at 21 days.
19 Measured TCE concentrations in the dams ranged from 312-646 mg/L, 625-1,102 mg/L, and
20 1,250-1,991 mg/L in the low, mid, and high-dose groups as measured from the drinking water.
21 Pups were evaluated for exploratory activity at 28, 60, or 90 days. No significant differences
22 were noted between control and treated pups at 28 days. At 60 days, all TCE-treated animals
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-124 DRAFT—DO NOT CITE OR QUOTE
-------
1 had significantly increased exploratory activity in comparison to age-matched controls, but only
2 the high group had increased activity at 90 days. A significant increase in spontaneous motor
3 activity (as measured by a wheel-running task) was noted in only the high dose TCE
4 (1,250-1,991 mg/L) group during the onset of the darkness period. This study demonstrated that
5 both spontaneous and open field activities are significantly affected by developmental TCE
6 exposure.
7 Spontaneous behavioral changes were also investigated in another study by Fredriksson
8 et al. (1993). Male and female NMRI pups (mice) were orally administered 50 or 290 mg/kg/d
9 for 7 days starting at postnatal Day 10. Spontaneous motor activity was investigated in male
10 mice at ages 17 and 60 days. TCE-treated animals tested at Day 17 did not demonstrate changes
11 in any spontaneous activity measurements in comparison to control animals. Both doses of TCE
12 (50 and 290 mg/kg/d) significantly decreased rearing in 60 day-old male mice.
13 Westergren et al. (1984) examined the brain specific gravity of litters from mice exposed to
14 TCE. NMRI mice (male and female) were exposed to 150-ppm TCE (806.1 mg/m3) for 30 days
15 prior to mating. Exposure in males continued until the end of mating and females were exposed
16 until the litters were born. Brains were removed from the offspring at either postnatal Days 1,
17 10, 20-22, or 29-31. At postnatal Days 1 and 10, significant decreases were noted in the
18 specific gravity of the cortex. Significant decreases in the specific gravity of the cerebellum
19 were observed at postnatal Day 10 (decrease from 1.0429 ± 0.00046 to 1.0405 ± 0.00030) and
20 20-22 (decrease from 1.0496 ± 0.00014 to 1.0487 ± 0.00060). Cerebellum measurements were
21 not reported for postnatal Day 29-31 animals. Neurobehavioral assessments were not conducted
22 in this study. Additionally, decreased brain specific gravity is suggestive of either decreased
23 brain weight or increased brain volume (probably from edema) or a combination of the two
24 factors and is highly suggestive of an adverse neurological effect. The effects of TCE on the
25 cortical specific gravity were not persistent since cortices from postnatal Day 29-31 animals did
26 not exhibit any significant changes. It is unclear if the effects on the cerebellum were persistent
27 since results were not reported for the postnatal Day 29-31 animals. However, the magnitude of
28 the change in the specific gravity of the cerebellum is decreased from postnatal Day 10 to
29 postnatal Day 20-22 suggesting that the effect may be reversible given a longer recovery period
30 from TCE.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-125 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-33. Summary of mammalian in vivo developmental neurotoxicity
studies—oral exposures
Reference
Fredriksson
etal., 1993
George et
al, 1986
Isaacson
and Taylor,
1989
Noland-
Gerbec et
al., 1986
Taylor et
al., 1985
Blossom et
al., 2008
Species/strain/
sex/number
Mouse, NMRI, male
pups, 12 pups from
3-4 different
litters/group
Rat, F334, male and
female, 20 pairs/
treatment group,
40 controls/sex
Rat, Sprague-Dawley,
females, 6 dams/group
Rat, Sprague-Dawley,
females, 9-11 dams/
group
Rat, Sprague-Dawley,
females, no. dams/
group not reported
Mouse, MRL +/+,
dams and both sexes
offspring, 8 litters/
group; 3-8 pups/group
Dose level/
exposure duration
0, 50, or 290 mg/kg/d
PND 10-16
0,0.15, 0.30, or 0.60%
microencapsulated TCE in
diet
Breeders exposed 1 wk
premating, then for 13 wk;
pregnant $s throughout
pregnancy (i.e., 18 wk
total)
0,312, or 625 mg/L
(0,4.0, or 8.1 mg/d)b
Dams (and pups) exposed
from 14 d prior to mating
until end of lactation
0, 3 12 mg/L
(Avg. total intake of dams:
825 mg TCE over 6 Id.)
Dams (and pups) exposed
from 14 d prior to mating
until end of lactation
0,312, 625, and
1,250 mg/L in drinking
water
Dams (and pups) exposed
from 14 d prior to mating
until end of lactation
Drinking water, from GD 0
to PND 42; 0 or QJ
mg/mL; maternal dose =
25.7 mg/kg/d; offspring
PND 24-42 dose =
3 1.0 mg/kg/d
NOAEL;
LOAEL a
LOAEL:
50 mg/kg/d
LOAEL:
0.15%
LOAEL:
3 12 mg/L
LOAEL:
3 12 mg/L
LOAEL:
3 12 mg/L
LOAEL:
3 1 mg/kg/d
for offspring
Effects
Rearing activity sig. -i- at both
dose levels on PND 60.
Open field testing in pups: a
sig. dose-related trend toward t
time required for male and
female pups to cross the first
grid in the test device.
Sig. -i- myelinated fibers in the
stratum lacunosum-moleculare
of pups. Reduction in myelin
in the CA1 region of the
hippocampus.
Sig. ^ uptake of 3H-2-DG in
whole brains and cerebella
(no effect in hippocampus) of
exposed pups at 7, 11, and 16
d, but returned to control
levels by 2 1 d.
Exploratory behavior sig. t in
60- and 90-d old male rats at
all treatment levels.
Locomotor activity (measured
through the wheel-running
tasks) was higher in rats from
dams exposed to 1,250 mg/L
TCE.
Righting reflex, bar holding,
and negative geotaxis were not
impaired. Significant
association between impaired
nest quality and TCE exposure.
Lower GSH levels and
GSH:GSSG ratios with TCE
exposure.
4
5
6
7
aLOEL (lowest-observed-effect level) are based upon reported study findings.
bDose conversions provided by study author(s).
GD = gestation day.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-126 DRAFT—DO NOT CITE OR QUOTE
-------
1 The effect of TCE on glucose uptake in the brain was evaluated in rat pups exposed to
2 TCE during gestation and through weaning. The primary source of energy utilized in the CNS is
3 glucose. Changes in glucose uptake in the brain are a good indicator for neuronal activity
4 modification. Noland-Grebec et al. (1986) administered 312 mg/L TCE through drinking water
5 to female Sprague-Dawley rats from 2 weeks before breeding and up until pups reached 21 days
6 of age. To measure glucose uptake, 2-deoxyglucose was administered intraperitoneally to male
7 pups at either postnatal Day 7, 11, 16, or 21. Significant decreases in glucose uptake were noted
8 in whole brain and cerebellum at all postnatal days tested. Significant decreases in glucose
9 uptake were also observed in the hippocampus except for animals tested at postnatal Day 21.
10 The observed decrease in glucose uptake suggests decreased neuronal activity.
11 Female Sprague-Dawley rats (70 days old) were administered TCE in drinking water at a
12 level of either 4.0 or 8.1 mg/day for 14 days prior to mating and continuing up through lactation
13 (Isaacson and Taylor, 1989). Only the male pups were evaluated in the studies. At postnatal
14 Day 21, brains were removed from the pups, sectioned, and stained to evaluate the changes in
15 myelin. There was a significant decrease (40% decrease) in myelinated fibers in the CA1 region
16 of the hippocampus of the male pups. This effect appeared to be limited to the CA1 region of the
17 hippocampus since other areas such as the optic tract, fornix, and cerebral peduncles did not have
18 decreases in myelinated fibers.
19 Neurological changes were found in pups exposed to TCE in a study conducted by the
20 National Toxicology Program (NTP) in Fischer 344 rats (George et al., 1986). TCE was
21 administered to rats at dietary levels of 0, 0.15, 0.30, or 0.60%. No intake calculations were
22 presented for the rat study and therefore, a dose rate is unavailable for this study. Open field
23 testing revealed a significant (p < 0.05) dose-related trend toward an increase in the time required
24 for male and female Fl weanling pups (postnatal day [PND] 21) to cross the first grid in the
25 testing device, suggesting an effect on the ability to react to a novel environment.
26 Blossom et al. (2008) treated male and female MRL +/+ mice with 0 or 0.1 mg/mL TCE
27 in the drinking water. Treatment was initiated at the time of mating, and continued in the
28 females (8/group) throughout gestation and lactation. Behavioral testing consisted of righting
29 reflex on PNDs 6, 8, and 10; bar-holding ability on PNDs 15 and 17; and negative geotaxis on
30 PNDs 15 and 17. Nest building was assessed and scored on PND 35, the ability of the mice to
31 detect and distinguish social odors was examined with an olfactory habituation/dishabituation
32 method at PND 29, and a resident intruder test was performed at PND 40 to evaluate social
33 behaviors. Righting reflex, bar holding, and negative geotaxis were not impaired by treatment.
34 There was a significant association between impaired nest quality and TCE exposure in tests of
35 nest-building behavior; however, TCE exposure did not have an effect on the ability of the mice
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-127 DRAFT—DO NOT CITE OR QUOTE
-------
1 to detect social and nonsocial odors using habituation and dishabituation methods. Resident
2 intruder testing identified significantly more aggressive activities (i.e., wrestling and biting) in
3 TCE-exposed juvenile male mice as compared to controls, and the cerebellar tissue from the
4 male TCE-treated mice had significantly lower GSH levels and GSH:GSSG ratios, indicating
5 increased oxidative stress and impaired thiol status, which have been previously reported to be
6 associated with aggressive behaviors (Franco et al., 2006). Histopathological examination of the
7 brain did not identify alterations indicative of neuronal damage or inflammation.
8
9 4.3.8.3. Summary and Conclusions for the Developmental Neurotoxicity Studies
10 Gestational exposure to TCE in humans has resulted in several developmental
11 abnormalities. These changes include neuroanatomical changes such as neural tube defects
12 (ATSDR, 2001; Bove et al., 1995, 1996; Lagakos et al., 1986) and encephalopathy (White et al.,
13 1997). Clinical neurological changes such as impaired cognition (Bernad et al., 1987; White et
14 al., 1997), aggressive behavior (Bernad et al., 1987), and speech and hearing impairment (Burg
15 and Gist, 1999; White et al., 1997) are also observed when TCE exposure occurs in utero.
16 In animal studies, anatomical and clinical developmental neurotoxicity is also observed.
17 Following inhalation exposures of 150 ppm to mice during mating and gestation, the specific
18 gravity of offspring brains was significantly decreased at postnatal time points through the age of
19 weaning; this effect did not persist to 1 month of age (Westergren et al., 1984). In studies
20 reported by Taylor et al. (1985), Isaacson and Taylor (1989), and Noland-Gerbec et al. (1986),
21 312 mg/L exposures in drinking water that were initiated 2 weeks prior to mating and continued
22 to the end of lactation resulted in (a) significant increase in exploratory behavior at postnatal
23 Days 60 and 90, (b) reductions in myelination in the CA1 hippocampal region of offspring at
24 weaning, and (c) significantly decreased uptake of 2-deoxyglucose in the rat brain at postnatal
25 Day 21. Gestational exposures to mice (Fredriksson et al., 1993) resulted in significantly
26 decreased rearing activity on postnatal Day 60, and dietary exposures during the course of a
27 continuous breeding study in rats (George et al., 1986) found a significant trend toward increased
28 time to cross the first grid in open field testing. In a study by Blossom et al. (2008), male mice
29 exposed gestationally to TCE exhibited lower GSH levels and lower GSH:GSSG ratios which is
30 also observed in mice that have more aggressive behaviors (Franco et al., 2006).
31
32 4.3.9. Mechanistic Studies of Trichloroethylene (TCE) Neurotoxicity
33 4.3.9.1. Dopamine Neuron Disruption
34 There are very recent laboratory animal findings resulting from short-term TCE
35 exposures that demonstrate vulnerability of dopamine neurons in the brain to this chlorinated
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-128 DRAFT—DO NOT CITE OR QUOTE
-------
1 hydrocarbon. The key limitation of these laboratory animal studies is that only 1 dosing regimen
2 was included in each study. Moreover, there has been no systematic body of data to show that
3 other chlorinated hydrocarbons such as tetrachloroethylene or aromatic solvents similarly target
4 this cell type. Confidence in the limited data regarding dopamine neuron death and in vivo TCE
5 exposure would be greatly enhanced by identifying a dose-response relationship. If indeed TCE
6 can target dopamine neurons it would be anticipated that human exposure to this agent would
7 result in elevated rates of parkinsonism. There are no systematic studies of this potential
8 relationship in humans although one limited report attempted to address this possibility.
9 Difficulties in subject recruitment into that study limit the weight that can be given to the results.
10 Endogenously formed chlorinated tetrahydro-beta-carbolines (TaClo) have been
11 suggested to contribute to the development of Parkinson-like symptoms (Bringmann et al., 1992,
12 1995; Reiderer et al., 2002; Kochen et al., 2003). TaClo can be formed endogenously from
13 metabolites of TCE such as trichloroacetaldehyde. TaClo has been characterized as a potent
14 neurotoxicant to the dopaminergic system. Some research groups have hypothesized that
15 Parkinson-like symptoms resulting from TCE exposure may occur through the formation of
16 TaClo, but not enough evidence is available to determine if this mechanism occurs.
17
18 4.3.9.1.1. Dopamine neuron disruption: human studies. There are no human studies that
19 present evidence of this effect. Nagaya et al. (1990) examined serum dopamine p-hydroxylase
20 activity without differences observed in mean activities between control and exposed subjects.
21 In the study, 84 male workers exposed to TCE were compared to 83 male age-matched controls.
22 The workers had constantly used TCE in their jobs and their length of employment ranged from
23 0.1 to 34 years.
24
25 4.3.9.1.2. Dopamine neuron disruption: animal studies. There are limited data from mice and
26 rats that suggest the potential for TCE to disrupt dopamine neurons in the basal ganglia (see
27 Table 4-34). Gash et al. (2008) showed that TCE gavage in Fischer 344 rats (n = 9) at an
28 exposure level of 1,000 mg/kg/d, 5 days/week, for 6 weeks yielded degeneration of dopamine
29 neurons in the substantia nigra and alterations in dopamine turnover as reflected in a shift in
30 dopamine metabolite to parent compound ratios. Guehl et al. (1999) reported similar findings in
31 OF1 mice (n = 10) that were injected i.p. with 400 mg/kg/d TCE 5 days/week for 4 weeks. Each
32 of these studies evaluated only a single dose level of TCE so that establishing a dose-response
33 relationship is not possible. Consequently, these data are of limited utility in risk assessment
34 because they do not establish the potency of TCE to damage dopamine neurons. They are
35 important, however, in identifying a potential permanent impairment that might occur following
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-129 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
4
5
TCE exposure at relatively high exposure doses. They also identify a potential mechanism by
which TCE could produce CNS injury.
Table 4-34. Summary of animal dopamine neuronal studies
Reference
Guehl et
al, 1999
Gash et al.,
2008
Exposure
route
Intraperitoneal
Administration
Oral gavage
Species/strain/
sex/number
Mouse, OF1,
male, 10
Rat, Fischer 344,
male, 9/group
Dose level/
exposure
duration
0 and 400
mg/kg; 5 d/wk,
4 wks
0 and 1,000
mg/kg; 5 d/wk,
6 wks
NOAEL;
LOAEL
LOAEL:
400 mg/kg
LOAEL:
1,000
mg/kg
Effects
Significant dopaminergic
neuronal death in substantia
nigra.
Degeneration of dopamine-
containing neurons in
substantia nigra.
Change in dopamine
metabolism.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
4.3.9.1.3. Summary and conclusions of dopamine neuron studies. Only two animal studies
have reported changes in dopamine neuron effects from TCE exposure (Gash et al., 2008;
Guehl et al., 1999). Both studies demonstrated toxicity to dopaminergic neurons in the
substantia nigra in rats or mice. LOAELs of 400 mg/kg (mice; Guehl et al., 1999) and
1,000 mg/kg (rats; Gash et al., 2008) were reported for this effect. Dopaminergic neuronal
degeneration following TCE exposure has not been studied in humans. However, there were no
changes in serum dopamine p-hydroxylase activity in TCE-exposed and control individuals
(Nagaya et al., 1990). Loss of dopaminergic neurons in the substantia nigra also occurs in
patients with Parkinson's disease and the substantia nigra is an important region in helping to
control movements. As a result, loss of dopaminergic neurons in the substantia nigra may be one
of the potential mechanisms involved in the clinical psychomotor effects that are observed
following TCE exposure.
4.3.9.2. Neurochemical and Molecular Changes
There are limited data obtained only from laboratory animals that TCE exposure may
have consequences on GABAergic (gamma-amino butyric acid [GABA]) and glutamatergic
neurons (Briving et al., 1986; Shih et al., 2001; see Table 4-35). However, the data obtained are
limited with respect to brain region examined, persistence of effect, and whether there might be
functional consequences to these changes. The data of Briving et al. (1986) demonstrating
changes in cerebellar high affinity uptake for GABA and glutamate following chronic low level
(50 and 150 ppm) TCE exposure do not appear to be reflected in the only other brain region
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-130 DRAFT—DO NOT CITE OR QUOTE
-------
1 evaluated (hippocampus). However, glutamate levels were increased in the hippocampus. The
2 data of Shih et al. (2001) are indirect in that it shows an altered response to GABAergic
3 antagonist drugs in mice treated by acute injection with 250, 500, 1,000, and 2,000 mg/kg TCE.
4 However, these data do show some dose dependency with significant findings observed with
5 TCE exposure as low as 250 mg/kg.
6 The development and physiology of the hippocampus has also been evaluated in two
7 different studies (Isaacson and Taylor, 1989; Ohta et al., 2001). Isaacson and Taylor (1989)
8 found a 40% decrease in myelinated fibers from hippocampi dissected from neonatal Sprague-
9 Dawley rats (n = 2-3) that were exposed to TCE (4 and 8.1 mg/day) in utero and during the
10 preweaning period. Ohta et al. (2001) injected male ddY mice with 300 mg/kg TCE and found a
11 significant reduction in response to titanic stimuli in excised hippocampal slices. Both of these
12 studies demonstrated that there is some interaction with TCE and the hippocampal area in the
13 brain.
14 Impairment of sciatic nerve regeneration was demonstrated in mice and rats exposed to
15 TCE (Kjellstrand et al., 1987). Under heavy anesthesia, the sciatic nerve of the animals was
16 artificially crushed to create a lesion. Prior to the lesion, some animals were pre-exposed to TCE
17 for 20 days and then for an additional 4 days after the lesion. Another set of animals were only
18 exposed to TCE for 4 days following the sciatic nerve lesion. For mice, regeneration of the
19 sciatic nerve in comparison to air-exposed animals was 20 and 33% shorter in groups exposed to
20 150- and 300-ppm TCE for 4 days, respectively. This effect did not significantly increase in
21 mice pre-exposed to TCE for 20 days, and the regeneration was 30% shorter in the 150-ppm
22 group and 22% shorter in the 300-ppm group. Comparatively, a 10% reduction in sciatic nerve
23 regeneration length was observed in rats exposed to TCE for 20 days prior to the lesion plus the
24 4 days after the sciatic nerve lesion.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-131 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-35. Summary of neurophysiological, neurochemical, and
neuropathological effects with TCE exposure
Reference
Exposure
route
Species/strain/
sex/number
Dose level/
exposure duration
NOAEL; LOAEL
Effects
Neurophysiological studies
Shih et al.,
2001
Ohtaetal.,
2001
Intra-
peritoneal
Intra-
peritoneal
Mouse, MF1,
male, 6/group
Mouse, ddY,
male, 5/group
0, 250 500, 1,000, or
2,000 mg/kg, 15 min;
followed by tail
infusion of PTZ (5
mg/mL), picrotoxin
(0.8 mg/mL),
bicuculline (0.06
mg/mL), strychnine
(0.05 mg/mL), 4-AP
(2 mg/mL), or
NMDA (8 mg/mL)
0, 300, or 1,000
mg/kg, sacrificed 24
hours after injection
—
LOAEL: 300
mg/kg
Increased threshold
for seizure appearance
with TCE
pretreatment for all
convulsants. Effects
strongest on the
GABAA antagonists,
PTZ, picrotoxin, and
bicuculline suggesting
GAB AA receptor
involvement. NMDA
and glycine Re
involvement also
suggested.
Decreased response
(LTP response) to
tetanic stimulation in
the hippocampus.
Neurochemical studies
Briving et al.,
1986
Subramoniam
etal., 1989
Inhalation
Oral
Gerbils,
Mongolian,
male and
female, 6/group
Rat, Wistar,
female,
0, 50, or 150 ppm,
continuous, 24 h/d,
12 months
0 or 1,000 mg/kg, 2
or 20 hours
0 or 1,000 mg/kg/d, 5
d/wk, 1 yr
NOAEL: 50 ppm;
LOAEL: 150 ppm
for glutamate levels
in hippocampus
NOAEL: 150 ppm
for glutamate and
GABA uptake in
hippocampus
LOAEL: 50 ppm
for glutamate and
GABA uptake in
cerebellar vermis
—
Increased glutamate
levels in the
hippocampus.
Increased glutamate
and GABA uptake in
the cerebellar vermis.
PI and PIP2 decreased
by 24 and 17% at 2 h.
PI and PIP2 increased
by 22 and 3 8% at 20
h.
PI, PIP, and PIP2
reduced by 52,23, and
45% in 1 yr study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-132 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-35. Summary of neurophysiological, neurochemical, and
neuropathological effects with TCE exposure (continued)
Reference
Haglid et al.,
1981
Exposure
route
Inhalation
Species/strain/
sex/number
Gerbil,
Mongolian,
male and
female,
6-7/group
Dose level/
exposure duration
0, 60, or 320 ppm, 24
h/d, 7 d/wk, 3 months
NOAEL; LOAEL
LOAEL: 60 ppm,
brain protein
changes
NOAEL: 60 ppm;
LOAEL: 320 ppm,
brain DNA changes
Effects
(1) Decreases in total
brain soluble protein
whereas increase in
SI 00 protein.
(2) Elevated DNA in
cerebellar vermis and
sensory motor cortex.
Neuropathological studies
Kjellstrand et
al., 1987
Isaacson and
Taylor, 1989
Inhalation
Oral
Mouse, NMRI,
male
Rat, Sprague-
Dawley, female
Rat, Sprague-
Dawley,
females, 6
dams/group
0, 150, or 300 ppm,
24 h/d, 4 or 24 d
0, 300 ppm, 24 h/d, 4
or24d
0,312, or 625 mg/L.
(0,4.0, or 8.1 mg/d)
Dams (and pups)
exposed from 14 d
prior to mating until
end of lactation.
LOAEL: 150 ppm,
4 and 24 d
NOAEL: 300 ppm,
4d
LOAEL: 300 ppm,
24 d
LOAEL: 3 12 mg/L
Sciatic nerve
regeneration was
inhibited in both mice
and rats.
Significant -i-
myelinated fibers in
the stratum
lacunosum-moleculare
of pups. Reduction in
myelin in the
hippocampus.
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
PTZ = pentylenetetrazole.
There are also a few in vitro studies (summarized in Table 4-36) that have demonstrated
that TCE exposure alters the function of inhibitory ion channels such as GABAA and glycine
receptors (Krasowski and Harrison, 2000; Beckstead et al., 2000), and serotonin receptors
(Lopreato et al., 2003). Krasowski and Harrison (2000) and Beckstead et al. (2000) were able to
demonstrate that human GABAA and glycine receptors could be potentiated by TCE when a
receptor agonist was coapplied. Krasowski and Harrison (2000) conducted an additional
experiment in order to determine if TCE was interacting with the receptor or perturbating the
cellular membrane (bilipid layer). Specific amino acids on the GABAA and glycine receptors
were mutated and in the presence of a receptor agonist (GABA for GABAA and glycine for
glycine receptors) and in these mutated receptors TCE-mediated potentiation was significantly
decreased or abolished suggesting that there was an interaction between TCE and these
receptors. Lopreato et al. (2003) conducted a similar study with the 5HT3A serotonin receptor
and found that when TCE was coapplied with serotonin, there was a potentiation in receptor
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-133 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
response. Additionally, TCE has been demonstrated to alter the function of voltage sensitive
calcium channels (VSCCs) by inhibiting the calcium mediated-current at a holding potential of-
70 mV and shifting the activation of the channels to a more hyperpolarizing potential (Shafer et
al., 2005).
Table 4-36. Summary of in vitro ion channel effects with TCE exposure
Reference
Cellular
system
Neuronal channel/
receptor
Concentrations
Effects
In vitro studies
Shafer et al.,
2005
Beckstead et
al., 2000
Lopreato et
al., 2003
Krasowski and
Harrison,
2000
PC12 cells
Xenopus
oocytes
Xenopus
oocytes
Human
embryonic
kidney 293
cells
vscc
Human
recombinant:
glycine receptor
al,
GAB AA receptors,
alpl,
-------
1 hippocampus are not sufficient to conclude that TCE has significant demyelinating effects in this
2 key brain region. Indeed, the bulk of the evidence from studies of learning and memory function
3 (which would be tied to hippocampal function) suggests no clear impairments due to TCE.
4 Some researchers (Albee et al., 1997, 2006; Barret et al., 1991, 1992; Laureno, 1988,
5 1993) have indicated that changes in trigeminal nerve function may be due to dichloroacetylene
6 which is formed under nonbiological conditions of high alkalinity or temperature during
7 volatilization of TCE. In experimental settings, trigeminal nerve function (Albee et al., 1997)
8 and trigeminal nerve morphology (Barret et al., 1991, 1992) was found to be more altered
9 following a low exposure to dichloroacetylene in comparison to the higher TCE exposure.
10 Barret et al. (1991, 1992) also demonstrated that TCE administration results in morphological
11 changes in the trigeminal nerve. Thus, dichloroacetylene may contribute to trigeminal nerve
12 impairment may be plausible following an inhalation exposure under conditions favoring its
13 formation. Examples of such conditions include passing through a carbon dioxide scrubber
14 containing alkaline materials, application to remove a wax coating from a concrete-lined stone
15 floor, or mixture with alkaline solutions or caustic (Saunders, 1967; Greim et al., 1984;
16 Bingham et al., 2001). However, dichloroacetylene exposures have not been identified or
17 measured in human epidemiologic studies with TCE exposure, and thus, do not appear to be
18 common to occupational or residential settings (Lash and Green, 1993). Moreover, changes in
19 trigeminal nerve function have also been consistently reported in humans exposed to TCE
20 following an oral exposure (Kilburn, 2002a; across many human studies of occupational and
21 drinking water exposures under conditions with highly varying potentials for dichloroacetylene
22 formation (Barret et al, 1982, 1984, 1987; Feldman et al., 1988). As a result, the mechanism(s)
23 for trigeminal nerve function impairment following TCE exposure is unknown., 1992;
24 Kilburn and Warshaw, 1993; Kilburn, 2002a; Mihri et al., 2004; Ruitjen et al., 1991). The
25 varying dichloroacetylene exposure potential across these studies suggests TCE exposure, which
26 is common to all of them, as the most likely etiologic agent for the observed effects.
27 The clearest consequences of TCE are permanent impairment of hearing in animal
28 models and disruption of trigeminal nerve function in humans with animal models showing
29 comparable changes following administration of a TCE metabolite. With regard to hearing loss,
30 the effect of TCE has much in common with the effects of several aromatic hydrocarbons
31 including ethylbenzene, toluene, and/>-xylene. Many studies have attempted to determine how
32 these solvents damage the cochlea. Of the hypotheses that have been advanced, there is little
33 evidence to suggest oxidative stress, changes in membrane fluidity, or impairment of central
34 efferent nerves whose endings innervate receptor cells in the cochlea. Rather, for reasons that
35 are still uncertain these solvents seem to preferentially target supporting cells in the cochlea
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-135 DRAFT—DO NOT CITE OR QUOTE
-------
1 whose death then alters key structural elements of the cochlea resulting ultimately in hair cell
2 displacement and death. Recently, potential modes of action resulting in ototoxicity have been
3 speculated to be due to blockade of neuronal nicotinic receptors present on the auditory cells
4 (Campo et al., 2007) and potentially changes in calcium transmission (Campo et al., 2008) from
5 toluene exposure. Although these findings were reported following an acute toluene exposure, it
6 is speculated that this mechanism may be a viable mechanism for TCE -mediated ototoxicity.
7 A few studies have tried to relate TCE exposure with selective impairments of dopamine
8 neurons. Two studies (Gash et al., 2008; Guehl et al., 1999) demonstrated dopaminergic
9 neuronal death and/or degeneration following an acute TCE administration. However, the only
10 human TCE exposure study examining dopamine neuronal activity found no changes in serum
11 dopamine p-hydroxylase activity in comparison to nonexposed individuals (Nagaya et al., 1990).
12 It is thought that TaClo, which can be formed from TCE metabolites such as
13 trichloroacetaldehyde, may be the potent neurotoxicant that selectively targets the dopaminergic
14 system. More studies are needed to confirm the dopamine neuronal function disruption and if
15 this disruption is mediated through TaClo.
16 There is good evidence that TCE and certain metabolites such as choral hydrate have
17 CNS depressant properties and may account for some of the behavioral effects (such as
18 vestibular effects, psychomotor activity changes, central visual changes, sleep and mood
19 changes) that have been observed with TCE. Specifically, in vitro studies have demonstrated
20 that TCE exposure results in changes in neuronal receptor function for the GABAA, glycine, and
21 serotonin receptors (Krasowski and Harrison, 2000; Beckstead et al., 2000; Lopreato et al.,
22 2003). All of these inhibitory receptors that are present in the CNS are potentiated when
23 receptor-specific agonist and TCE are applied. These results are similar to other anesthetics and
24 suggest that some of the behavioral functions are mediated by modifications in ion channel
25 function. However, it is quite uncertain whether there are persistent consequences to such high
26 dose TCE exposure. Additionally, with respect to the GABAergic system, acute administration
27 of TCE increased the seizure threshold appearance and this effect was the strongest with
28 convulsants that were GABA receptor antagonists (Shih et al., 2001). Therefore, this result
29 suggests that TCE interacts with the GABA receptor and that was also verified in vitro
30 (Krasowski and Harrison, 2000; Beckstead et al., 2000).
31 Also, TCE exposure has been linked to decreased sensitivity to titanic stimulation in the
32 hippocampus (Ohta et al., 2001) as well as significant reduction in myelin in the hippocampus in
33 a developmental exposure (Isaacson and Taylor, 1990). These effects are notable since the
34 hippocampus is highly involved in memory and learning functions. Changes in the hippocampal
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-136 DRAFT—DO NOT CITE OR QUOTE
-------
1 physiology may correlate with the cognitive changes that were reported following TCE
2 exposure.
O
4 4.3.11. Overall Summary and Conclusions—Weight of Evidence
5 Both human and animal studies have associated TCE exposure with effects on several
6 neurological domains. The strongest neurological evidence of hazard in humans is for changes
7 in trigeminal nerve function or morphology and impairment of vestibular function. Fewer and
8 more limited evidence exists in humans on delayed motor function, and changes in auditory,
9 visual, and cognitive function or performance. Acute and subchronic animal studies show
10 morphological changes in the trigeminal nerve, disruption of the peripheral auditory system
11 leading to permanent function impairments and histopathology, changes in visual evoked
12 responses to patterns or flash stimulus, and neurochemical and molecular changes. Additional
13 acute studies reported structural or functional changes in hippocampus, such as decreased
14 myelination or decreased excitability of hippocampal CA1 neurons, although the relationship of
15 these effects to overall cognitive function is not established. Some evidence exists for motor-
16 related changes in rats/mice exposed acutely/subchronically to TCE, but these effects have not
17 been reported consistently across all studies.
18 Epidemiologic evidence supports a relationship between TCE exposure and trigeminal
19 nerve function changes, with multiple studies in different populations reporting abnormalities in
20 trigeminal nerve function in association with TCE exposure (Barret et al., 1982, 1984, 1987;
21 Feldman et al., 1988, 1992; Kilburn and Warshaw, 1993; Ruitjen et al., 2001; Kilburn, 2002a;
22 Mhiri et al., 2004). Of these, two well conducted occupational cohort studies, each including
23 more than 100 TCE-exposed workers without apparent confounding from multiple solvent
24 exposures, additionally reported statistically significant dose-response trends based on ambient
25 TCE concentrations, duration of exposure, and/or urinary concentrations of the TCE metabolite
26 TCA (Barret et al., 1984; Barret et al., 1987). Limited additional support is provided by a
27 positive relationship between prevalence of abnormal trigeminal nerve or sensory function and
28 cumulative exposure to TCE (most subjects) or CFC-113 (<25% of subjects) (Rasmussen et al.,
29 1993c). Test for linear trend in this study was not statistically significant and may reflect
30 exposure misclassification since some subjects included in this study did not have TCE exposure.
31 The lack of association between TCE exposure and overall nerve function in three small studies
32 (trigeminal: El-Ghawabi et al., 1973; ulnar and medial: Triebig et al., 1982, 1983) does not
33 provide substantial evidence against a causal relationship between TCE exposure and trigeminal
34 nerve impairment because of limitations in statistical power, the possibility of exposure
35 misclassification, and differences in measurement methods. Laboratory animal studies have also
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-137 DRAFT—DO NOT CITE OR QUOTE
-------
1 shown TCE-induced changes in the trigeminal nerve. Although one study reported no significant
2 changes in trigeminal somatosensory evoked potential in rats exposed to TCE for 13 weeks
3 (Albee et al., 2006), there is evidence of morphological changes in the trigeminal nerve
4 following short-term exposures in rats (Barret et al., 1991, 1992).
5 Human chamber, occupational, geographic based/drinking water, and laboratory animal
6 studies clearly established TCE exposure causes transient impairment of vestibular function.
7 Subjective symptoms such as headaches, dizziness, and nausea resulting from occupational
8 (Granjean et al., 1955; Liu et al., 1988; Rasmussen and Sabroe, 1986; Smith et al., 1970),
9 environmental (Hirsch et al., 1996), or chamber exposures (Stewart et al., 1970; Smith et al.,
10 1970) have been reported extensively. A few laboratory animal studies have investigated
11 vestibular function, either by promoting nystagmus or by evaluating balance (Niklasson et al.,
12 1993; Tham et al., 1979; Tham et al., 1984; Umezu et al., 1997).
13 In addition, mood disturbances have been reported in a number of studies, although these
14 effects also tend to be subjective and difficult to quantify (Gash et al., 2007; Kilburn and
15 Warshaw, 1993; Kilburn, 2002a, 2002b; McCunney et al., 1988; Mitchell et al., 1969;
16 Rasmussen and Sabroe, 1986; Troster and Ruff, 1990), and a few studies have reported no
17 effects from TCE on mood (Reif et al., 2003; Triebig et al., 1976, 1977a). Few comparable
18 mood studies are available in laboratory animals, although both Moser et al. (2003) and Albee et
19 al. (2006) report increases in handling reactivity among rats exposed to TCE. Finally,
20 significantly increased number of sleep hours was reported by Arito et al. (1994) in rats exposed
21 via inhalation to 50-300-ppm TCE for 8 hours/day for 6 weeks.
22 Four epidemiologic studies of chronic exposure to TCE observed disruption of auditory
23 function. One large occupational cohort study showed a statistically significant difference in
24 auditory function with cumulative exposure to TCE or CFC-113 as compared to control groups
25 after adjustment for possible confounders, as well as a positive relationship between auditory
26 function and increasing cumulative exposure (Rasmussen et al., 1993b). Of the three studies
27 based on populations from ATSDR's TCE Subregistry from the National Exposure Registry,
28 more limited than Rasmussen et al. (1993b) due to inferior exposure assessment, Burg et al.
29 (1995) and Burg and Gist (1999) reported a higher prevalence of self-reported hearing
30 impairments. The third study reported that auditory screening revealed abnormal middle ear
31 function in children less than 10-years-of-age, although a dose-response relationship could not be
32 established and other tests did not reveal differences in auditory function (ATSDR, 2003a).
33 Further evidence for these effects is provided by numerous laboratory animal studies
34 demonstrating that high dose subacute and subchronic TCE exposure in rats disrupts the auditory
35 system leading to permanent functional impairments and histopathology.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-138 DRAFT—DO NOT CITE OR QUOTE
-------
1 Studies in humans exposed under a variety of conditions, both acutely and chronically,
2 report impaired visual functions such as color discrimination, visuospatial learning tasks, and
3 visual depth perception in subjects with TCE exposure. Abnormalities in visual depth perception
4 were observed with a high acute exposure to TCE under controlled conditions (Vernon and
5 Ferguson, 1969). Studies of lower TCE exposure concentrations also observed visuofunction
6 effects. One occupational study (Rasmussen et al., 1993b) reported a statistically significant
7 positive relationship between cumulative exposure to TCE or CFC-113 and visual gestalts
8 learning and retention among Danish degreasers. Two studies of populations living in a
9 community with drinking water containing TCE and other solvents furthermore suggested
10 changes in visual function (Kilburn et al., 2002a; Reif et al., 2003). These studies used more
11 direct measures of visual function as compared to Rasmussen et al. (1993b), but their exposure
12 assessment is more limited because TCE exposure is not assigned to individual subjects
13 (Kilburn et al., 2002a), or because there are questions regarding control selection (Kilburn et al.,
14 2002a) and exposure to several solvents (Kilburn et al., 2002a; Reif et al., 2003).
15 Additional evidence of effects of TCE exposure on visual function is provided by a
16 number of laboratory animal studies demonstrating that acute or subchronic TCE exposure
17 causes changes in visual evoked responses to patterns or flash stimulus (Boyes et al., 2003, 2005;
18 Blain et al., 1994). Animal studies have also reported that the degree of some effects is
19 correlated with simultaneous brain TCE concentrations (Boyes et al., 2003, 2005) and that, after
20 a recovery period, visual effects return to control levels (Blain et al., 1994; Rebert et al., 1991).
21 Overall, the human and laboratory animal data together suggest that TCE exposure can cause
22 impairment of visual function, and some animal studies suggest that some of these effects may
23 be reversible with termination of exposure.
24 Studies of human subjects exposed to TCE either acutely in chamber studies or
25 chronically in occupational settings have observed deficits in cognition. Five chamber studies
26 reported statistically significant deficits in cognitive performance measures or outcome measures
27 suggestive of cognitive effects (Stewart et al., 1970; Gamberale et al., 1976; Triebig et al., 1976,
28 1977a; Gamberale et al., 1977). Danish degreasers with high cumulative exposure to TCE or
29 CFC-113 had a high risk (OR: 13.7, 95% CI: 2.0-92.0) for psychoorganic syndrome
30 characterized by cognitive impairment, personality changes, and reduced motivation, vigilance,
31 and initiative compared to workers with low cumulative exposure. Studies of populations living
32 in a community with contaminated groundwater also reported cognitive impairments
33 (Kilburn and Warshaw, 1993; Kilburn, 2002a), although these studies carry less weight in the
34 analysis because TCE exposure is not assigned to individual subjects and their methodological
35 design is weaker.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-139 DRAFT—DO NOT CITE OR QUOTE
-------
1 Laboratory studies provide some additional evidence for the potential for TCE to affect
2 cognition, though the predominant effect reported has been changes in the time needed to
3 complete a task, rather than impairment of actual learning and memory function (Kulig et al.,
4 1987; Kishi et al., 1993; Umezu et al., 1997). In addition, in laboratory animals, it can be
5 difficult to distinguish cognitive changes from motor-related changes. However, several studies
6 have reported structural or functional changes in the hippocampus, such as decreased
7 myelination (Issacson et al., 1990; Isaacson and Taylor, 1989) or decreased excitability of
8 hippocampal CA1 neurons (Ohta et al., 2001), although the relationship of these effects to
9 overall cognitive function is not established.
10 Two studies of TCE exposure, one chamber study of acute exposure duration and one
11 occupational study of chronic duration, reported changes in psychomotor responses. The
12 chamber study of Gamberale et al. (1976) reported a dose-related decrease in performance in a
13 choice reaction time test in healthy volunteers exposed to 100- and 200-ppm TCE for 70 minutes
14 as compared to the same subjects without exposure. Rasmussen et al. (1993c) reported a
15 statistically significant association with cumulative exposure to TCE or CFC-113 and
16 dyscoordination trend among Danish degreasers. Observations in a third study (Gun et al., 1978)
17 are difficult to judge given the author's lack of statistical treatment of data. In addition, Gash et
18 al. (2007) reported that 14 out of 30 TCE-exposed workers exhibited significantly slower fine
19 motor hand movements as measured through a movement analysis panel test. Studies of
20 population living in communities with TCE and other solvents detected in groundwater supplies
21 reported significant delays in simple and choice reaction times in individuals exposed to TCE in
22 contaminated groundwater as compared to referent groups (Kilburn, 2002a; Kilburn and
23 Warshaw, 1993; Kilburn and Thornton, 1996). Observations in these studies are more uncertain
24 given questions of the representativeness of the referent population, lack of exposure assessment
25 to individual study subjects, and inability to control for possible confounders including alcohol
26 consumption and motivation. Finally, in a presentation of 2 case reports, decrements in motor
27 skills as measured by the grooved pegboard and finger tapping tests were observed (Troster and
28 Ruff, 1990).
29 Laboratory animal studies of acute or subchronic exposure to TCE observed psychomotor
30 effects, such as loss of righting reflex (Umezu et al., 1997; Shih et al., 2001) and decrements in
31 activity, sensory-motor function, and neuromuscular function (Kishi et al., 1993; Moser et al.,
32 1995; Moser et al., 2003). However, two studies also noted an absence of significant changes in
33 some measures of psychomotor function (Kulig et al., 1987; Albee et al., 2006). In addition, less
34 consistent results have been reported with respect to locomotor activity in rodents. Some studies
35 have reported increased locomotor activity after an acute i.p. dosage (Wolff and Siegmund,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-140 DRAFT—DO NOT CITE OR QUOTE
-------
1 1978) or decreased activity after acute or short term oral gavage dosing (Moser et al., 1995,
2 2003). No change in activity was observed following exposure through drinking water (Waseem
3 et al., 2001), inhalation (Kulig et al., 1987) or orally during the neurodevelopment period
4 (Fredriksson et al., 1993).
5 Several neurochemical and molecular changes have been reported in laboratory
6 investigations of TCE toxicity. Kjellstrand et al. (1987) reported inhibition of sciatic nerve
7 regeneration in mice and rats exposed continuously to 150-ppm TCE via inhalation for 24 days.
8 Two studies have reported changes in GAB Aergic and glutamatergic neurons in terms of GAB A
9 or glutamate uptake (Driving et al., 1986) or response to GABAergic antagonistic drugs (Shih et
10 al., 2001) as a result of TCE exposure, with the Driving et al. (1986) conducted at 50 ppm for
11 12 months. Although the functional consequences of these changes is unclear, Tham et al.
12 (1979, 1984) described central vestibular system impairments as a result of TCE exposure that
13 may be related to altered GABAergic function. In addition, several in vitro studies have
14 demonstrated that TCE exposure alters the function of inhibitory ion channels such as receptors
15 for GABAA glycine, and serotonin (Krasowski and Harrison, 2000; Beckstead et al., 2000;
16 Lopreato et al., 2003) or of voltage-sensitive calcium channels (Shafer et al., 2005).
17
18 4.4. KIDNEY TOXICITY AND CANCER
19 4.4.1. Human Studies of Kidney
20 4.4.1.1. Nonspecific Markers of'Nephrotoxicity
21 Investigations of nephrotoxicity in human populations show that highly exposed workers
22 exhibit evidence of damage to the proximal tubule (NRC, 2006). The magnitude of exposure
23 needed to produce kidney damage is not clear. Observation of elevated excretion of urinary
24 proteins in the four studies (Briining et al., 1999a, b; Bolt et al., 2004; Green et al., 2004)
25 indicates the occurrence of a toxic insult among TCE-exposed subjects compared to unexposed
26 controls. Two studies are of subjects with previously diagnosed kidney cancer (Briining et al.,
27 1999a; Bolt et al., 2004), subjects in Briining et al. (1999b) and Green et al. (2004) are disease
28 free. Urinary proteins are considered nonspecific markers of nephrotoxicity and include
29 al-Microglobulin, albumin, and 7V-acetyl-p-D-glucosaminidase (NAG; Price et al., 1999, 1996;
30 Lybarger et al., 1999). Four studies measure al-microglobulin with elevated excretion observed
31 in the German studies (Briining et al., 1999a, b; Bolt et al., 2004) but not Green et al. (2004).
32 However, Green et al. (2004) found statistically significant group mean differences in NAG,
33 another nonspecific marker of tubular toxicity, in disease free subjects. Observations in Green et
34 al. (2004) provide evidence of tubular damage among workers exposed to trichloroethylene at
35 32 ppm (mean) (range, 0.5-252 ppm). Elevated excretion of NAG as a nonspecific marker of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-141 DRAFT—DO NOT CITE OR QUOTE
-------
1 tubular damage has also been observed with acute TCE poisoning (Carried et al., 2007). These
2 and other studies relevant to evaluating TCE nephrotoxicity are discussed in more detail below.
3 Biological monitoring of persons who previously experienced "high" exposures to
4 trichloroethylene (100-500 ppm) in the workplace show altered kidney function evidenced by
5 urinary excretion of proteins suggestive of renal tubule damage. Similar results were observed in
6 the only study available of subjects with TCE exposure at current occupational limits (NRC,
7 2006). Table 4-37 provides details and results from these studies. Briming etal. (1999a) report a
8 statistically significantly higher prevalence of elevated proteinuria suggestive of severe tubular
9 damage (n = 24, 58.5%, p < 0.01) and an elevated excretion of al-microglobin, another urinary
10 biomarker of renal tubular function, was observed in 41 renal cell carcinoma cases with prior
11 trichloroethylene exposure and with pending workman's compensation claims compared with the
12 nonexposed renal cell cancer patients (n = 14, 28%) and to hospitalized surgical patients n = 2,
13 2%). Statistical analyses did not adjust for differences in median systolic and diastolic blood
14 pressure that appeared higher in exposed renal cell carcinoma cases compared to nonexposed
15 controls. Similarly, severe tubular proteinuria is seen in 14 of 39 workers (35%) exposed to
16 trichloroethylene in the electrical department, fitters shop and through general degreasing
17 operations of felts and sieves in a cardboard manufacturing factory compared to no subjects of
18 46 nonexposed males office and administrative workers from the same factory (p < 0.01)
19 (Briining et al., 1999b). Furthermore, slight tubular proteinuria is seen in 20% of exposed
20 workers and in 2% of nonexposed workers (Briining et al., 1999b). Exposed subjects also had
21 statistically significantly elevated levels of al-microglobulin compared to unexposed controls.
22 Furthermore, subjects with tubular damage as indicated by urinary protein patterns had higher
23 GST-alpha concentrations than nonexposed subjects (p < 0.001). Both sex and use of spot or 24-
24 hour urine samples are shown to influence al-microglobulin (Andersson et al., 2008); however,
25 these factors are not considered to greatly influence observations given only males were subjects
26 and al-microglobulin levels in spot urine sample are adjusted for creatinine concentration.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-142 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-37. Summary of human kidney toxicity studies
Subjects
Effect
Exposure
TCE exposure was through
degreasing activities in metal
parts factory or semiconductor
industry.
U-total trichloro compounds:
Exposed, 83.4 mg/g Cr (range,
2-66.2 mg/g Cr.
Controls, N.D.
8.4 + 7.9 yrs mean
employment duration.
Reference
206 subjects-
104 male workers exposed
to TCE; 102 male controls
(source not identified)
Increased p2-microglobulin and total
protein in spot urine specimen.
p2-microglobulin:
Exposed, 129.0 + 113.3 mg/g
creatinine (Cr)
Controls, 113.6 + 110.6 mg/g Cr
Total protein:
Exposed, 83.4+113.2 mg/g
creatinine (Cr)
Controls, 54.0 + 18.6 mg/g Cr
Nagaya et al..
1989
29 metal workers
NAG in morning urine specimen, 0.17
+ 0.11U/mmolCr
Breathing zone monitoring, 3
ppm (median) and 5 ppm
(mean).
Selden et al..
1993
191 subjects-
41 renal cell carcinoma
cases pending cases
involving compensation
with TCE exposure;
50 unexposed renal cell
carcinoma cases from same
area as TCE-exposed cases;
100 nondiseased control
and hospitalized surgical
patients
Increased urinary proteins patterns,
al-microglobulin, and total protein in
spot urine specimen
Slight/severe tubular damage:
TCE RCC cases, 93%
Nonexposed RCC cases, 46%
Surgical controls, 11%
al-microglobulin (mg/g creatinine):
Exposed RCC cases, 24.6 + [SD] 13.9
Unexposed RCC cases, 11.3 + [SD]
9.8
Surgical controls, 5.5 + [SD] 6.8
All exposed RCC cases
exposed to 'high" and "very
high" TCE intensity.
18 yr mean exposure duration.
Briining et
al., 1999a
85 male workers employed
in cardboard manufacturing
factory (39 TCE exposed,
46) nonexposed office and
administrative controls)
Increased urinary protein patterns and
excretion of proteins in spot urine
specimen
Slight/severe tubular damage:
TCE exposed, 67%
Nonexposed, RCC cases, 9%
p< 0.001
al-microglobulin (mg/g creatinine):
Exposed, 16.2 + [SD] 10.3
Unexposed, 7.8 + [SD] 6.9
p< 0.001
GST-alpha (ug/g creatinine):
Exposed 6.0 ± [SD] 3.3
Unexposed, 2.0 + [SD] 0.57
p< 0.001
No group differences in total protein
or GST-pi
"High" TCE exposure to
workers in the fitters shop and
electrical department.
"Very high" TCE exposure to
workers through general
degreasing operations in
carton machinery section.
Briining et
al., 1999b
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-143 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-37. Summary of human kidney toxicity studies (continued)
Subjects
Effect
Exposure
Reference
99 renal cell carcinoma
cases and 298 hospital
controls (from Bruning et
al. [2003] and alive at the
time of interview)
Increased excretion of al-
microglobulin in spot urine specimen
Proportion of subjects with al-
microglobulin <5.0
mg/L:
Exposed cases, 15%
Unexposed cases, 51%
Exposed controls, 55%
Unexposed controls, 55%
p < 0.05, prevalence of exposed cases
compared to prevalences of either
exposed controls or unexposed
controls
Mean al-microglobulin:
Exposed cases, 18.1 mg/L
Unexposed cases, <5.0 mg/L
All exposed RCC cases
exposed to 'high" and "very
high" TCE intensity .
Boltetal.,
2004
124 subjects (70 workers
currently exposed to TCE
and 54 hospital and
administrative staff
controls)
Analysis of urinary proteins in spot
urine sample obtained 4 d after
exposure
Increased excretion of albumin, NAG,
and formate in spot urine specimen
Albumin (mg/g creatinine):3
Exposed, 9.71 + [SD] 11.6
Unexposed, 5.50 + [SD] 4.27
Total NAG (U/g creatinine):
Exposed, 5.27 + [SD] 3. 78
Unexposed, 2.41 + [SD] 1.91
Format (mg/g creatinine):
Exposed, 9.45 + [SD] 4.78
Unexposed, 5.55 + [SD] 3.00
No group mean differences in
GST-alpha, retinol binding protein,
al -microglobulin, p2-microglobulin,
total protein, and methylmalonic acid
Mean U-TCA of exposed
workers was 64 + [SD] 102
(Range, 1-505).
Mean U-TCOH of exposed
workers was 122 + [SD] 119
(Range, 1-639).
Mean TCE concentration to
exposed subjects was
estimated as 32 ppm (range,
0.5-252 ppm) and was
estimated by applying the
German occupational
exposure limit (maximale
arbeitsplatz konzentration,
MAK) standard to U-TCA and
assuming that the linear
relationship holds for
exposures above 100 ppm.
86% of subjects with exposure
to <50 ppm TCE.
Green etal.,
2004
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-144 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-37. Summary of human kidney toxicity studies (continued)
Subjects
Effect
Exposure
Reference
101 cases or deaths from
end-stage renal disease
(ESDR) among male and
female subjects in Hill Air
Force Base aircraft
maintenance worker cohort
of Blair etal. (1998)
TCE exposure:
Cox Proportional Hazard Analysis:
Ever exposed to TCE,b
1.86(1.02,3.39)
Logistic regression:13
No chemical exposure (referent
group): 1.0
<5 unit-year, 1.73 0.86, 3.48)
5-25 unit-year, 1.65 (0.82, 3.35)
>25 unit-year, 1.65 (0.82, 3.35)
Monotonic trend test,;? > 0.05
Indirect low-intermittent TCE
exposure, 2.47(1.17, 5.19)
Indirect peak/infrequent TCE
exposure 3.55 (1.25, 10.74)
Direct TCE exposure, "not
statistically significant" but hazard
ratio and confidence intervals were
not presented in paper
Cumulative TCE exposure
(intensity x duration)
identified using 3 categories,
<5 unit-year, 5-25 unit year,
>25 unit-year per job exposure
matrix of Stewart et al. (1991).
Radican et
al., 2006
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Tor a urine sample, 10-17 mg of albumin per g of creatinine is considered to be suspected albuminuria in males
(15-25 in females) (De Jong and Brenner 2004).
bHazard ratio and 95% confidence interval.
N.D. = not detectable, SD = standard deviation.
Bolt et al. (2004) measured al-microglobulin excretion in living subjects from the renal
cell carcinoma case-control study by Briining et al. (2003). Some subjects in this study were
highly exposed. Of the 134 with renal cell cancer, 19 reported past exposures that led to narcotic
effects and 18 of the 401 controls, experienced similar effects (OR: 3.71, 95% CI: 1.80-7.54)
(Briining et al., 2003). Bolt et al. (2004) found that al-microglobulin excretion increased in
exposed renal cancer patients compared with nonexposed patients controls. A lower proportion
of exposed cancer patients had normal al-microglobulin excretion, less than 5 mg/L, the
detection level for the assay and the level considered by these investigators as associated with no
clinical or subclinical tubule damage, and a higher proportion of high values, defined as
>45 mg/L, compared to cases who did not report TCE occupational exposure and to nonexposed
controls (p < 0.05). Exposed cases, additionally, had statistically significantly higher median
concentration of al-microglobulin compared to unexposed cases in creatinine-unadjusted spot
urine specimens (p < 0.05). Reduced clearance of creatinine attributable to renal cancer does not
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-145 DRAFT—DO NOT CITE OR QUOTE
-------
1 explain the lower percentage of normal values among exposed cases given findings of similar
2 prevalence of normal excretion among unexposed renal cell cases and controls.
3 In their study of 70 current employees (58 males, 12 females) of an electronic factory
4 with trichloroethylene exposure and 54 (50 males, 4 females) age-matched subjects drawn from
5 hospital or administrative staff, Green et al. (2004) found that urinary excretion of albumin, total
6 NAG and formate were increased in the exposed group compared with the unexposed group.l
1 No differences between exposed and unexposed subjects were observed in other urinary proteins,
8 including al-microglobulin, p2-microglobulin, and GST-alpha. Green et al. (2004) stated that
9 NAG is not an indicator of nephropathy, or damage, but rather is an indicator of functional
10 change in the kidney. Green et al. (2004) further concluded that increased urinary albumin or
11 NAG was not related to trichloroethylene exposure; analyses to examine the exposure-response
12 relationship found neither NAG or albumin concentration correlated to U-TCA or employment
13 duration (years). The National Research Council (NRC, 2006) did not consider U-TCA as
14 sufficiently reliable to use as a quantitative measure of TCE exposure, concluding that the data
15 reported by Green et al. (2004) were inadequate to establish exposure-response information
16 because the relationship between U-TCA and ambient TCE intensity is highly variable and
17 nonlinear, and conclusions about the absence of association between TCE and nephrotoxicity can
18 not be made based on U-TCA. Moreover, use of employment duration does not consider
19 exposure intensity differences between subjects with the same employment duration, and bias
20 introduced through misclassification of exposure may explain the Green et al. (2004) findings.
21 Selden et al. (1993) in their study of 29 metal workers (no controls) reported a correlation
22 between NAG and U-TCA (r = 0.48,/> < 0.01) but not with other exposure metrics of recent or
23 long-term exposure. Personal monitoring of worker breath indicated median and mean time-
24 weighted-average TCE exposures of 3 and 5 ppm, respectively. Individual NAG concentrations
25 were within normal reference values. Rasmussen et al. (1993), also, reported a positive
26 relationship (p = 0.05) between increasing urinary NAG concentration (adjusted for creatinine
27 clearance) and increasing duration in their study of 95 metal degreasers (no controls) exposed to
28 either TCE (70 subjects) or CFC113(25 subjects). Multivariate regression analyses which
29 adjusted for age were suggestive of an association between NAG and exposure duration
30 (p = 0.011). Mean urinary NAG concentration was higher among subjects with annual exposure
31 of >30 hours/week, defined as peak exposure, compared to subjects with annual exposure of less
1 Elevation of NAG in urine is a sign of proteinuria, and proteinuria is both a sign and a cause of kidney malfunction
(Zandi-Nejad et al., 2004). For a urine sample, 10-17 mg of albumin per g of creatinine is considered to be
suspected albuminuria in males (15-25 in females) (De Jong and Brenner, 2004).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-146 DRAFT—DO NOT CITE OR QUOTE
-------
1 than <30 hours/week (72.4 + 44.1 ug/g creatinine compared to 45.9 + 30.0 ug/g creatinine,
2 /?<0.01).
3 Nagaya et al. (1989) did not observe statistically significant group differences in urinary
4 p2-microglobulin and total protein in spot urine specimens of male degreasers and their controls,
5 nor were these proteins correlated with urinary total trichloro-compounds (U-TTC). The paper
6 lacks details on subject selection, whether urine collection was at start of work week or after
7 sufficient exposure, and presentation ofp-values and correlation coefficients. The presentation
8 of urinary protein concentrations stratified by broad age groups is less statistically powerful than
9 examination of this confounder using logistic regression. Furthermore, although valid for
10 pharmacokinetic studies, examination of renal function using U-TTC as a surrogate for TCE
11 exposure is uncertain, as discussed above for Green et al. (2004).
12 4.4.1.2. End-Stage Renal Disease
13 End-stage renal disease is associated with hydrocarbon exposure, a group that includes
14 trichloroethylene, 1,1,1-trichloroethane, and JP4 (jet propellant 4), in the one study examining
15 this endpoint (Radican et al., 2006). Table 4-37 provides details and results from Radican et al.
16 (2006). This study assessed end-stage renal disease in a cohort of aircraft maintenance workers
17 at Hill Air Force Base (Blair et al., 1998) with strong exposure assessment to trichloroethylene
18 (NRC, 2006). Other occupational studies do not examine end-stage renal disease specifically,
19 instead reporting relative risks associated with deaths due to nephritis and nephrosis (Boice et al.,
20 1999, 2006; ATSDR, 2004), all genitourinary system deaths (Garabrant et al., 1988; Costa et al.,
21 1989; Ritz, 1999), or providing no information on renal disease mortality in the published paper
22 (Blair et al., 1998; Morgen et al., 1998; Chang et al., 2003).
23
24 4.4.2. Human Studies of Kidney Cancer
25 Cancer of the kidney and renal pelvis is the 6th leading cause of cancer in the United
26 States with an estimated 54,390 (33,130 men and 21,260 women) newly diagnosed cases and
27 13,010 deaths (Jemal et al., 2008; Ries et al., 2008). Age-adjusted incidence rates based on cases
28 diagnosed in 2001-2005 from 17 Surveillance, Epidemiology, and End Results (SEER)
29 geographic areas are 18.3 per 100,000 for men and 9.2 per 100,000 for women. Age-adjusted
30 mortality rates are much lower; 6.0 per 100,000 for men and 2.7 for women.
31 Cohort, case-control, and geographical studies have examined trichloroethylene and
32 kidney cancer, defined either as cancer of kidney and renal pelvis in cohort and geographic based
33 studies or as renal cell carcinoma, the most common type of kidney cancer, in case-control
34 studies. Appendix C identifies these studies' design and exposure assessment characteristics.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-147 DRAFT—DO NOT CITE OR QUOTE
-------
1 Observations in these studies are presented below in Table 4-38. Rate ratios for incidence
2 studies in Table 4-38 are, generally, larger than for mortality studies.
3 Additionally, a large body of evidence exists on kidney cancer risk and either job or
4 industry titles where trichloroethylene usage has been documented. TCE has been used as a
5 degreasing solvent in a number of jobs, task, and industries, some of which include metal,
6 electronic, paper and printing, leather manufacturing and aerospace/aircraft manufacturing or
7 maintenance industries and job title of degreaser, metal workers, electrical worker, and machinist
8 (IARC, 1995; Bakke et al., 2007). NRC (2006) identifies characteristics for kidney cancer case-
9 control studies that assess job title or occupation in their Table 3-8. Relative risks and 95%
10 confidence intervals reported in these studies are found in Table 4-39 below.
11
12 4.4.2.1. Studies of Job Titles and Occupations with Historical Trichloroethylene (TCE)
13 Usage
14 Elevated risks are observed in many of the cohort or case-control studies between kidney
15 cancer and industries or job titles with historical use of trichloroethylene (Partenen et al., 1991;
16 McCredie and Stewart, 1993; Schlehofer et al., 1995; Mandel et al., 1995; Pesch et al., 2000a;
17 Parent et al., 2000; Mattioli et al., 2002; Briining et al., 2003; Zhang et al., 2004; Charbotel et al.,
18 2006; Wilson et al., 2008). Overall, these studies, although indicating association with metal
19 work exposures and kidney cancer, are insensitive for identifying a TCE hazard. The use of job
20 title or industry as a surrogate for exposure to a chemical is subject to substantial
21 misclassification that will attenuate rate ratios due to exposure variation and differences among
22 individuals with the same job title. Several small case-control studies (Jensen et al., 1988;
23 Harrington et al., 1989; Sharpe et al., 1989; Auperin et al., 1994; Vamvakas et al., 1998;
24 Parent et al., 2000) have insufficient statistical power to detect modest associations due to their
25 small size and potential exposure misclassification (NRC, 2006). For these reasons, statistical
26 variation in the risk estimate is large and observation of statistically significantly elevated risks
27 associated with metal work in many of these studies is noteworthy. Some studies also examined
28 broad chemical grouping such as degreasing solvents or chlorinated solvents. Observations in
29 studies that assessed degreasing agents or chlorinated solvents reported statistically significant
30 elevated kidney cancer risk (Asal et al., 1998; Harrington et al., 1989; McCredie and Stewart,
31 1993; Mellemgaard et al., 1994; Schlehofer et al., 1995; Pesch et al., 2000a; Briining et al.,
32 2003). Observations of association with degreasing agents together with job title or occupations
33 where TCE has been used historically provide a signal and suggest an etiologic agent common to
34 degreasing activities.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-148 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
Exposure group
Relative risk (95% CI)
No. obs.
events
Reference
Cohort and PMR studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cum TCE score
Med cum TCE score
High TCE score
p for trend
Not reported
1.00a
1.87 (0.56, 6.20)
4.90 (1.23, 19.6)
p = 0.023
6
6
4
Zhao et al., 2005
TCE, 20 yrs exposure lagb
Low cum TCE score
Med cum TCE score
High TCE score
p for trend
1.00a
1.19(0.22,6.40)
7.40(0.47, 116)
;? = 0.120
6
7
3
All employees at electronics factory (Taiwan)
Males
Females
Females
1.06(0.45, 2.08) c
1.09(0.56, 1.91) c
1.10(0.62, 1.82) c
8
12
15
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Any exposure, males
Any exposure, females
1.2 (0.98, 1.46)
1.2 (0.97, 1.48)
1.2(0.55,2.11)
103
93
10
Chang et al., 2005
Sung etal., 2008
Raaschou-Nielsen et
al., 2003
Exposure lag time
20 yrs
1.3 (0.86, 1.88)
28
Employment duration
<1 yr
1-4.9 yrs
>5yrs
0.8 (0.5, 1.4)
1.2 (0.8, 1.7)
1.6(1.1,2.3)
16
28
32
Subcohort w/higher exposure
Any TCE exposure
1.4(1.0, 1.8)
53
Employment duration
1-4.9 yrs
>5yrs
Biologically monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
Cumulative exp (Ikeda)
<17 ppm-yr
>17 ppm-yr
1.1 (0.7, 1. if
1.7(1.1, 2.4)d
1.1 (0.3,2.8)
0.9 (0.2, 2.6)
2.4 (0.03, 14)
Not reported
23
30
4
3
1
Hansen etal., 2001
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-149 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Mean concentration (Ikeda)
<4ppm
4+ppm
Employment duration
<6.25 yrs
>6.25
Relative risk (95% CI)
Not reported
Not reported
No. obs.
events
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Not reported
Reference
Blair etal., 1998
Males, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.4(0.4,4.7)
1.3 (0.3, 4.7)
0.4(0.1,2.3
9
5
2
Females, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
3.6(0.5,25.6)
0
0
2
Biologically -monitored Finnish workers
All subjects
0.87 (0.32, 1.89)
6
Anttila et al., 1995
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
Not reported
Not reported
Cardboard manufacturing workers in Arnsberg, Germany
Exposed workers
7.97 (2.59, 8.59)e
5
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
1.16(0.42,2.52)
Not reported
6
Cardboard manufacturing workers, Atlanta area, GA
All subjects
All departments
Finishing department
3.7(1.4,8.1)
oo (3.0, w)f
16.6(1.7, 453. l)f
6
5
3
Henschleretal.,
1995
Axelson et al., 1994
Sinks etal., 1992
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-150 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Relative risk (95% CI)
No. obs.
events
Reference
Cohort and PMR studies — mortality
Computer manufacturing workers (IBM), NY
Males
Females
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Any exposure to TCE
Low cum TCE score
Med cum TCE score
High TCE score
p for trend
1.64(0.45, 4.21) 8
2.22 (0.89, 4.57)
Not reported
1.00a
1.43(0.49,4.16)
2.13 (0.50,8.32)
p = 03l
4
0
7
7
7
o
J
Clapp and Hoffman,
2008
Boiceetal.,2006
Zhao et al., 2005
TCE, 20 yrs exposure lagb
Low cum TCE score
Med cum TCE score
High TCE score
p for trend
1.00a
1.69 (0.29, 9.70)
1.82 (0.09, 38.6)
;? = 0.635
10
6
1
View-Master employees
Males
Females
2.76 (0.34, 9.96)g
6.21 (2.68, 12.23)g
2
8
United States Uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, 2-10 yrs duration"1
Light TCE exposure, >10 yrs duration"1
Mod TCE exposure, >2 yrs duration"1
Not reported
1.94 (0.59, 6.44)
0.76(0.14,400.0)
5
2
0
Aerospace workers (Lockheed)
Routine Exp
Routine-Intermittent3
0.99 (0.40, 2.04)
Not presented
7
11
ATSDR, 2004
Ritz, 1999 (as
reported in NRC,
2006)
Boiceetal., 1999
Duration of exposure
Oyrs
<1 yr
1-4 yrs
>5yrs
1.0
0.97 (0.37, 2.50)
0.19(0.02, 1.42)
0.69(0.22,2.12)
22
6
1
4
p for trend
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-151 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Relative risk (95% CI)
No. obs.
events
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)e
High intensity (>50 ppm) e
1.32 (0.57, 2.60)
0.47 (0.01, 2.62)
1.78 (0.72, 3.66)
8
1
7
Reference
Morgan etal., 1998
TCE subcohort (Cox analysis)
Never exposed
Ever exposed
1.00a
1.14 (0.51, 2.58)h
24
8
Peak
No/Low
Med/Hi
1.00a
1.89 (0.85, 4.23)h
24
8
Cumulative
Referent
Low
High
1.00a
0.31 (0.04, 2.36)h
1.59(0.68, 3.71)h
24
1
7
Aircraft maintenance workers (Hill AFB, Utah)
TCE subcohort
1.6(0.5, 5.1) a
15
Blair etal., 1998
Males, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
2.0(0.5,7.6)
0.4(0.1,4.0)
1.2(0.3,4.8)
8
1
4
Females, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
9.8 (0.6, 157)
3.5 (0.2, 56.4)
1.18(0.47,2.94)'
1.24 (0.41, 3.71)1
l.O1
1.87(0.59,5.97'
0.31(0.03,2.75)'
1.16(0.31,4.32)'
0.93(0.15,5.76)'
1.0a
2.86 (0.27, 29.85)'
0.97(0.10,9.50)'
0
1
1
18
16
10
1
5
2
0
1
1
Radican etal., 2008
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-152 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Relative risk (95% CI)
No. obs.
events
Cardboard manufacturing workers in Arnsberg, Germany
TCE exposed workers
Unexposed workers
Deaths reported to among GE pension fund (Pittsfield, MA)
3.28(0.40, 11.84)
(0.00, 5.00)
0.99 (0.30, 3.32)f
2
0
12
Cardboard manufacturing workers, Atlanta area, GA
1.4 (0.0, 7.7)
1
U. S. Coast Guard employees
Marine inspectors
Noninspectors
1.06(0.22,3.10)
1.03 (0.21,3.01)
3
3
Aircraft manufacturing plant employees (Italy)
All subjects
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
Not reported
0.93 (0.48, 1.64)
12
Reference
Henschleretal., 1995
Greenland et al., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrant et al., 1988
Case-control studies
Population of Arve Valley, France
Any TCE exposure
Any TCE exposure (High confidence exposure)
1.64 (0.95, 2.84)
1.88 (0.89, 3.98)
37
16
Charbotel etal., ,
2006, 2007, 2009
Cumulative TCE exposure
Referent/nonexposed
Low
Medium
High
Test for trend
1.00a
1.62 (0.75, 3.47)
1.15(0.47,2.77)
2.16(1.02,4.60)'
p = 0.04
49
12
9
16
Cumulative TCE exposure + peak
Referent/nonexposed
Low/medium, no peaks
Low/medium + peaks
High, no peaks
High + peaks
1.00a
1.35 (0.69, 2.63)
1.61 (0.36. 7.30)
1.76 (0.65, 4.73)
2.73 (1.06, 7.07)J
49
18
3
8
8
Cumulative TCE exposure, 10-yr lag
Referent/nonexposed
Low/medium, no peaks
Low/medium + peaks
High, no peaks
High + peaks
1.00a
1.44 (0.69, 2.80)
1.38 (0.32, 6.02)
1.50(0.53,4.21)
3.15(1.19,8.38)
49
19
3
7
8
Time-weighted-average TCE exposurek
Referent/nonexposed
1.00a
46
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-153 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Any TCE without cutting fluid
Any cutting fluid without TCE
<50 ppm TCE + cutting fluid
50+ ppm TCE + cutting fluid
Relative risk (95% CI)
1.62 (0.76, 3.44)
2.39(0.52, 11.03)
1.14(0.49,2,66)
2.70(1.02,7.17)
No. obs.
events
15
o
J
12
10
Population of Arnsberg Region, Germany
Longest job held-TCE/PERC (CAREX)
Self-assessed exposure to TCE
1.80(1.01,3.20)
2.47(1.36,4.49)
117
25
Reference
Bruningetal.,2003
Duration of self -assessed TCE exposure
0
<10yrs
10-20 yrs
>20 yrs
1.00a
3.78 (1.54, 9.28)
1.80 (0.67, 4.79)
2.69 (0.84, 8.66)
109
11
7
8
Population in 5 German Regions
Any TCE Exposure
Males
Females
Not reported
Not reported
Not reported
Peschetal.,2000a
TCE exposure (Job Task Exposure Matrix)
Males
Medium
High
Substantial
1.3 (1.0, 1.8)
1.1 (0.8, 1.5)
1.3 (0.8,2.1)
68
59
22
Females
Medium
High
Substantial
1.3 (0.7, 2.6)
0.8 (0.4, 1.9)
1.8 (0.6, 5.0)
11
7
5
Population of Minnesota
Dosemeci et al., 1999
Ever exposed to TCE, NCI JEM
Males
Females
Males + Females
1.04 (0.6, 1.7)
1.96(1.0,4.0)
1.30(0.9, 1.9)
33
22
55
Population of Arnsberg Region, Germany
Self-assessed exposure to TCE
10.80 (3.36, 34.75)
19
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
0.8 (0.4, 2.0)1
0.8 (0.2, 2.6)1
4
2
Vamvakas et al.,
1998
Siemiatycki et al.,
1991
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-154 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-38. Summary of human studies on TCE exposure and kidney cancer
(continued)
Exposure group
Relative risk (95% CI)
No. obs.
events
Reference
Geographic based studies
Residents in two study areas in Endicott, NY
Residents of 13 census tracts inRedlands, CA
1.90(1.06,3.13)
0.80(0.54, 1.12)m
15
54
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
ATSDR, 2006, 2008
Morgan and Cassidy,
2002
Vartiainen et al.,
1993
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
"Internal referents, workers not exposed to TCE.
bRelative risks for TCE exposure after adjustment for 1st employment, socioeconomic status, age at event, and all
other carcinogens, including hydrazine.
°Chang et al. (2005)—urinary organs combined.
dSIR for renal cell carcinoma.
eHenschler et al. (1995) Expected number of incident cases calculated using incidence rates from the Danish Cancer
Registry.
fOdds ratio from nested case-control analysis.
8Proportional mortality ratio.
hRisk ratio from Cox Proportional Hazard Analysis, stratified by age, sex and decade (Environmental Health
Strategies, 1997).
Tn Radican et al. (2008), kidney cancer defined as renal cell carcinoma (ICDA 8 code 189.0) and estimated relative
risks from Cox proportional hazard models were adjusted for age and sex.
J Analyses adjusted for age, sex, smoking and body mass index. The odds ratio, adjusted for age, sex, smoking, body
mass index and exposure to cutting fluids and other petroleum oils, for high cumulative TCE exposure was 1.96
(95% CI: 0.71, 5.37) and for high cumulative + peak TCE exposure was 2.63 (95% CI: 0.79, 8.83). The odds
ratio for, considering only job periods with high confidence TCE exposure assessment, adjusted for age, sex,
smoking and body mass index, for high cumulative dose plus peaks was 3.80 (95% CI: 1.27. 11.40).
kThe exposure surrogate is calculated for one occupational period only and is not the average exposure concentration
over the entire employment period.
1 90% confidence interval.
m99% confidence interval.
JEM = job-exposure matrix, NCI = National Cancer Institute, PERC = perchloroethylene.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-155 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table 4-39. Summary of case-control studies on kidney cancer and
occupation or job title
Case ascertainment area/exposure group
Relative risk
(95% CI)
No. exposed
cases
Swedish Cancer Registry Cases
Machine/electronics industry
Shop and construction metal work
Machine assembly
Metal plating work
Shop and construction metal work
1.30(1.08, 1.55)a[M]
1.75(1.04, 2.76) a[F]
1.19(1.00, 1.40)a[M]
1.62(0.94, 2.59) a[M]
2.70 (0.73, 6.92) a[M]
1.66(0.71, 3.26) a[F]
120
18
143
4
8
Arve Valley, France
Metal industry
Metal workers, job title
Metal industry, screw-cutting workshops
Machinery, electrical and transportation
equipment manufacture
1.02 (0.59, 1.76)
1.00 (0.56, 1.77)
1.39 (0.75, 2.58)
1.19(0.61,2.33)
28
25
22
15
Iowa Cancer Registry Cases
Assemblers
>10 yrs employment
2.5 (0.8, 7.6)
4.2(1.2, 15.3)
5
4
Arnsberg Region, Germany
Iron/steel
Occupations with contact to metals
Longest job held
Metal greasing/degreasing
1.15(0.29,4.54)
1.53 (0.97, 2.43)
1.14(0.66, 1.96)
5.57(2.33, 13.32)
o
5
46
24
15
Degreasing agents
Low exposure
High exposure
2.11(0.86,5.18)
1.01 (0.40, 2.54)
9
7
Bologna, Italy
Metal workers
Printers
Solvents
2.21 (0.99, 5.37)
1.55(0.17, 13.46)
0.79(0.31, 1.98) [M]
1.47(0.12, 17.46) [F]
37
7
17
3
Montreal, Canada
Metal fabricating and machining industry
Metal processors
Printing and publishing industry
Printers
Aircraft mechanics
1.0 (0.6, 1.8)
1.2 (0.4, 3.4)
1.1(0.4,3.0)
3.0(1.2,7.5)
2.8(1.0,8.4)
14
4
4
6
4
Reference
Wilson et al., 2008
Charbotel et al.,
2006
Zhang et al., 2004
Bruning et al., 2003
Mattiolietal.,2002
Parent et al., 2000
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-156 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-39. Summary of case-control studies on kidney cancer and
occupation or job title (continued)
Case ascertainment area/exposure group
Relative risk
(95% CI)
No. exposed
cases
5 Regions in Germany
Electrical and electronic equipment
assembler
Printers
Metal cleaning/degreasing, job task
3.2 (1.0, 10.3) [M]
2.7(1.3, 5.8) [F}
3.5(1.1, 11.2)[M]
2.1(0.4, 11.7)[F]
1.3 (0.7, 2.3) [M]
1.5 (0.3, 7.7) [F]
5
11
5
2
15
2
New Zealand Cancer Registry
Toolmakers and blacksmiths
Printers
1.48 (0.72, 3.03)
0.67 (0.25, 1.83)
No info
Minnesota Cancer Surveillance System
Iron or steel
1.6(1.2,2.2)
8
Rhein-Neckar-Odenwald Area, Germany
Reference
Pesch et al., 2000a
Delahunt et al.,
1995
Mandeletal., 1995
Schlehofer et al.,
1995
Metal
Industry
Occupation
1.63 (1.07, 2.48)
1.38(0.89,2.12)
71
Electronic
Industry
Occupation
Chlorinated solvents
Metal and metal compounds
0.51 (0.26, 1.01)
0.57(0.25, 1.33)
2.52(1.23,5.16)
1.47 (0.94, 2.30)
14
9
27
62
Danish Cancer Registry
Iron and steel
Solvents
1.4 (0.8, 2.4) [M]
1.0(0.1, 3.2) [F]
1.5 (0.9, 2.4) [M]
6.4 (1.8, 23) [F]
31
1
50
16
France
Machine fitters, assemblers, and precision
instrument makers
0.7 (0.3, 1.9)
16
Mellemgaard et al.,
1994
Auperin et al., 1994
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-157 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-39. Summary of case-control studies on kidney cancer and
occupation or job title (continued)
Case ascertainment area/exposure group
Relative risk
(95% CI)
No. exposed
cases
New South Wales, Australia
Iron and steel
Printing or graphics
Machinist or tool maker
Solvents
1.18(0.75, 1.85)b
2.39(1.26, 4.52) c
1.18(0.87, 2.08)b
0.82(0.32, 2.11)d
1.15(0.72, 1.86)b
1.83(0.92, 3.61) c
1.54(1. 11, 2.14)b
1.40(0.82, 2.40) c
52
19
29
6
48
16
109
24
Finnish Cancer Registry
Iron and metalware work
Machinists
Paper and pulp; printing/publishing
Nonchlorinated solvents
1.87 (0.94, 3.76)
2.33(0.83,6.51)
2.20 (1.02, 4.72) [M]
5.95 (1.21, 29.2) [F]
3.46 (0.91, 13.2) [M]
22
10
18
7
9
West Midlands UK Cancer Registry
Reference
McCredie and
Stewart, 1993
Partenen et al., 1991
Harrington et al.,
1989
Organic solvents
Ever exposed
Intermediate exposure
1.30(0.31,8.50)
1.54(0.69,4.10)
o
J
3
Montreal, Canada
Organic solvents
Degreasing solvents
1.68 (0.83, 2.22)
3.42 (0.92, 12.66)
33
10
Oklahoma
Metal degreasing
Machining
Painter, paint manufacture
1.7 (0.7, 3.8) [M]
1.7 (0.7, 4.3) [M]
1.3 (0.7, 2.6) [M]
19
13
22
Missouri Cancer Registry
Machinists
2.2 (0.5, 10.3)
o
J
Danish Cancer Registry
Iron and metal, blacksmith
Painter, paint manufacture
1.4(0.7, 2.9) d
1.8 (0.7, 4.6)
17
10
Sharpeetal., 1989
Asaletal., 1988
Brownson, 1988
Jensen etal., 1988
1
2
3
4
5
6
7
"Renal pelvis, Wilson et al. (2008).
bRenal cell carcinoma, McCredie and Stewart (1993).
"Renal pelvis, McCredie and Stewart (1993).
dRenal pelvis and ureter, Jensen et al. (1988).
UK = United Kingdom.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-158 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.4.2.2. Cohort and Case-Controls Studies of Trichloroethylene (TCE) Exposure
2 Cohort and case-controls studies that include job-exposure matrices for assigning TCE
3 exposure potential to individual study subjects show associations with kidney cancer, specifically
4 renal cell carcinoma, and trichloroethylene exposure. Support for this conclusion derives from
5 findings of increased risks in cohort studies (Henschler et al., 1995; Raaschou-Nielsen et al.,
6 2003; Zhao et al., 2005) and in case-control studies from the Arnsberg region of Germany
7 (Vamvakas et al., 1998; Pesch et al., 2000a; Briining et al., 2003), the Arve Valley region in
8 France (Charbotel et al., 2006, 2009), and the United States (Sinks et al., 1992; Dosemeci et al.,
9 1999).
10 A consideration of a study's statistical power and exposure assessment approach is
11 necessary to interpret observations in Table 4-38. Most cohort studies are underpowered to
12 detect a doubling of kidney cancer risks including the essentially null studies by Greenland et al.
13 (1994), Axelson et al. (1994 [incidence]), Anttila et al. (1995 [incidence]), Blair et al. (1998
14 [incidence and mortality]), Morgan et al. (1998), Boice et al. (1999) and Hansen et al. (2001).
15 Only the exposure duration-response analysis of Raaschou-Nielsen et al. (2003) had over 80%
16 statistical power to detect a doubling of kidney cancer risk (NRC, 2006), and they observed a
17 statistically significant association between kidney cancer and >5-year employment duration.
18 Rate ratios estimated in the mortality cohort studies of kidney cancer (e.g., Garabrant et al.,
19 1988; Sinks et al., 1992; Axelson et al., 1994; Greenland et al., 1994; Blair et al., 1998; Morgan
20 et al., 1998; Ritz, 1999; Boice et al., 1999, 2006) are likely underestimated to some extent
21 because their reliance on death certificates and increased potential of nondifferential
22 misclassification of outcome in these studies, although the magnitude is difficult to predict
23 (NRC, 2006). Cohort or PMR studies with more uncertain exposure assessment approaches,
24 e.g., studies of all subjects working at a factory (Garabrant et al., 1998; Costa et al., 1989;
25 ATSDR, 2004; Sung et al., 2007; Chang et al., 2003, 2005; Clapp and Hoffmann, 2008), do not
26 show association but are quite limited given their lack of attribution of higher or lower exposure
27 potentials; risks are likely diluted due to their inclusion of no or low exposed subjects.
28 Two studies were carried out in geographic areas with a high frequency and a high degree
29 of TCE exposure and were designed with a priori hypotheses to test for the effects of TCE
30 exposure on renal cell cancer risk (Briining et al., 2003; Charbotel et al., 2006, 2009) and for this
31 reason their observations have important bearing to the epidemiologic evidence evaluation. Both
32 studies found a 2-fold elevated risk with any TCE exposure after adjustment for several possible
33 confounding factors including smoking (2.47, 95% CI: 1.36, 4.49) for self-assessed exposure to
34 TCE (Briining et al., 2003); high cumulative TCE exposure (2.16, 95% CI: 1.02, 4.60) with a
35 positive and statistically significant trend test,/? = 0.04, (Charbotel et al., 2006). Furthermore,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-159 DRAFT—DO NOT CITE OR QUOTE
-------
1 renal cell carcinoma risk in Charbotel et al. (2005) increased to over 3-fold (95% CI: 1.19, 8.38)
2 in statistical analyses which considered a 10-year exposure lag period. An exposure lag period is
3 often adopted in analysis of cancer epidemiology to reduce exposure measurement biases
4 (Salvan et al., 1995). Most exposed cases in this study were exposed to TCE below any current
5 occupational standard (26 of 37 cases [70%]) had held a job with a highest time-weighted
6 average (TWA [<50 ppm]) (Charbotel et al., 2009). A subsequent analysis of Charbotel et al.
7 (2009) using an exposure surrogate defined as the highest TWA for any job held, an inferior
8 surrogate given TCE exposures in other jobs were not considered, reported an almost 3-fold
9 elevated risk (2.80, 95% CI: 1.12, 7.03) adjusted for age, sex, body mass index (BMI), and
10 smoking with exposure to TCE in any job to >50-ppm TWA (Charbotel et al., 2009).
11 Zhao et al. (2005) compared 2,689 TCE-exposed workers at a California aerospace
12 company to nonexposed workers from the same company as the internal referent population, and
13 found a monotonic increase in incidence of kidney cancer by increasing cumulative TCE
14 exposure. In addition, a 5-fold increased incidence was associated with high cumulative TCE
15 exposure. This relationship for high cumulative TCE exposure, lagged 20 years, was
16 accentuated with adjustment for other occupational exposures (RR = 7.40, 95% CI: 0.47, 116),
17 although the confidence intervals were increased. An increased confidence interval with
18 adjustments is not unusual in occupational studies, as exposure is usually highly correlated with
19 them, so that adjustments often inflate standard error without removing any bias (NRC, 2006).
20 Observed risks were lower for kidney cancer mortality and because of reliance on cause of death
21 on death certificates are likely underestimated because of nondifferential misclassification of
22 outcome (Percy et al., 1981). Boice et al. (2006), another study of 1,111 workers with potential
23 TCE exposure at this company and which overlaps with Zhao et al. (2005), found a 2-fold
24 increase in kidney cancer mortality (standardized mortality ration [SMR] = 2.22, 95% CI: 0.89,
25 4.57). This study examined mortality in a cohort whose definition date differs slightly from
26 Zhao et al. (2005), working between 1948-1999 with vital status as of 1999 (Boice et al., 2006)
27 compared to working between 1950-1993 with follow-up for mortality as of 2001 (Zhao et al.,
28 2005), and used a qualitative approach for TCE exposure assessment. Boice et al. (2006) is a
29 study of fewer subjects identified with potential TCE exposure, of fewer kidney cancer deaths [7
30 deaths; 10 incident cases, 10 deaths in Zhao et al. (2005)], of subjects with more recent
31 exposures, and with a inferior exposure assessment approach compared to Zhao et al. (2005); a
32 finding of a two-fold mortality increase (95% CI: 0.89, 4.57) is noteworthy given the
33 insensitivities.
34 Zhao et al. (2005) and Charbotel et al. (2006), furthermore, are two of the few studies to
3 5 conduct a detailed assessment of exposure that allowed for the development of a j ob-exposure
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-160 DRAFT—DO NOT CITE OR QUOTE
-------
1 matrix that provided rank-ordered levels of exposure to TCE and other chemicals. NRC (2006)
2 discussed the inclusion of rank-ordered exposure levels is a strength increasing precision and
3 accuracy of exposure information compared to more inferior exposure assessment approaches in
4 some other studies such as duration of exposure or a grouping of all exposed subjects.
5 The finding in Raaschou-Nielsen et al. (2003) of an elevated renal cell carcinoma risk
6 with longer employment duration is noteworthy given this study's use of a relatively insensitive
7 exposure assessment approach. One strength of this study is the presentation of incidence ratios
8 for a subcohort of higher exposed subjects, those with at least 1-year duration of employment
9 and first employment before 1980, as a sensitivity analysis for assessing the effect of possible
10 exposure misclassification bias. Renal cell carcinoma risk was higher in this subcohort
11 compared to the larger cohort and indicated some potential for misclassification bias in the
12 grouped analysis. For both the cohort and subcohort analyses, risk appeared to increase with
13 increasing employment duration, although formal statistical tests for trend are not presented in
14 the published paper.
15
16 4.4.2.2.1. Discussion of controversies on studies in theArnsberg region of Germany. Two
17 previous studies of workers in this region, a case-control study of Vamvakas et al. (1998) and
18 Henschler et al. (1995), a study prompted by a kidney cancer case cluster, observed strong
19 associations between kidney cancer and TCE exposure. A fuller discussion of the studies from
20 the Arnsberg region and their contribution to the overall weight of evidence on cancer hazard is
21 warranted in this evaluation given the considerable controversy (Bloemen and Tomenson, 1995;
22 Swaen, 1995; McLaughlin and Blot, 1997; Green and Lash, 1999; Cherrie et al., 2001; Mandel,
23 2001) surrounding Henschler et al. (1995) and Vamvakas et al. (1998).
24 Criticisms of Henschler et al. (1995) and Vamvakas et al. (1998) relate, in part, to
25 possible selection biases that would lead to inflating observed associations and limited inferences
26 of risk to the target population. Specifically, these include (1) the inclusion of kidney cancer
27 cases first identified from a cluster and the omission of subjects lost to follow-up from Henschler
28 et al. (1995); (2) use of a Danish population as referent, which may introduce bias due to
29 differences in coding cause of death and background cancer rate differences (Henschler et al.,
30 1995); (3) follow-up of some subjects outside the stated follow-up period (Henschler et al.,
31 1995); (4) differences between hospitals in the identification of cases and controls in Vamvakas
32 et al. (1998); (5) lack of temporality between case and control interviews (Vamvakas et al.,
33 1998); (6) lack of blinded interviews (Vamvakas et al., 1998); (7) age differences in Vamvakas
34 et al. (1998) cases and controls that may lead to a different TCE exposure potential; (8) inherent
35 deficiencies in Vamvakas et al. (1998) as reflected by its inability to identify other known kidney
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-161 DRAFT—DO NOT CITE OR QUOTE
-------
1 cancer risk factors; and, (9) exposure uncertainty, particularly unclear intensity of TCE exposure.
2 Overall, NRC (2006) noted that some of the points above may have contributed to an
3 underestimation of the true exposure distribution of the target population (points 5, 6, and 7),
4 other points would underestimate risk (points 3), and that these effects could not have explained
5 the entire excess risk observed in these studies (points 1, 2, and 4). The NRC (2006) furthermore
6 disagreed with the exposure uncertainty criticism (point 9), and concluded TCE exposures,
7 although of unknown intensity, were substantial and, clearly showed graded differences on
8 several scales in Vamvakas et al. (1998) consistent with this study's semi quantitative exposure
9 assessment.
10 Briining et al. (2003) was carried out in a broader region in southern Germany, which
11 included the Arnsberg region and a different set of cases and control identified from a later time
12 period than Vamvakas et al. (1998). The TCE exposure range in this study was similar to that in
13 Vamvakas et al. (1998), although at a lower exposure prevalence because of the larger and more
14 heterogeneous ascertainment area for cases and controls. For "ever exposed" to TCE,
15 Briining et al. (2003) observed a risk ratio of 2.47 (95% CI: 1.36, 4.49) and a 4-fold increase in
16 risk (95% CI: 1.80, 7.54) among subjects with any occurrence of narcotic symptom and a 6-fold
17 increase in risk (95% CI: 1.46, 23.99) for subjects who had daily occurrences of narcotic
18 symptoms; risks which are lower than observed in Vamvakas et al. (1998). The lower rate ratio
19 in Briining et al. (2003) might indicate bias in the Vamvakas et al. study or statistical variation
20 between studies related to the broader base population included in Briining et al. (2003).
21 Observational studies such as epidemiologic studies are subject to biases and
22 confounding which can be minimized but never completely eliminated through a study's design
23 and statistical analysis methods. While Briining et al. (2003) overcomes many of the
24 deficiencies of Henschler et al. (1995) and Vamvakas et al. (1998), nonetheless, possible biases
25 and measurement errors could be introduced through their use of prevalent cases and residual
26 noncases, use of controls from surgical and geriatric clinics, nonblinding of interviewers, a
27 2-year difference between cases and controls in median age, use or proxy or next-of-kin
28 interviews, and self-reported occupational history.
29 The impact of any one of the above points could either inflate or depress observed
30 associations. Biases related to a longer period for case compared to control ascertainment could
31 go in either direction. Next-of-kin interviewers for deceased cases, all controls being alive at the
32 time of interview, would be expected to underestimate risk if exposures were not fully reported
33 and thus, misclassified. On the other hand, the control subjects who were enrolled when the
34 interviews were conducted might not represent the true exposure distribution of the target
35 population through time and would lead to overestimate of risk. Selection of controls from
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-162 DRAFT—DO NOT CITE OR QUOTE
-------
1 clinics is not expected to greatly influence observed associations since these clinics specialized
2 in the type of care they provided (NRC, 2006). Briining et al. (2003) is not the only kidney case-
3 control study where interviewers were not blinded; in fact, only the study of Charbotel et al.
4 (2006) included blinding of interviewers. Blinding of interviewers is preferred to reduce
5 possible bias. Briining et al.'s use of frequency matching using 5-year age groupings is common
6 in epidemiologic studies and any biases introduced by age difference between cases and controls
7 is expected to be minimal because the median age difference was 3 years.
8 Despite these issues, the three studies of the Arnsberg region, with very high apparent
9 exposure and different base populations showed a significant elevation of risk and all have
10 bearing on kidney cancer hazard evaluations. The emphasis provided by each study for
11 identifying a kidney cancer hazard depends on its strengths and weaknesses. Briining et al.
12 (2003) overcomes many of the deficiencies in Henschler et al. (1995) and Vamvakas et al.
13 (1998). The finding of a statistically significantly approximately 3-fold elevated odds ratio with
14 occupational TCE exposure in Briining et al. (2003) strengthens the signal previously reported by
15 Henschler et al. (1995) and Vamvakas et al. (1998). A previous study of cardboard workers in
16 the United States (Sink et al., 1992), a study like Henschler et al. (1995) which was prompted by
17 a reported cancer cluster, had observed association with kidney cancer incidence, particularly
18 with work in the finishing department where TCE use was documented. Henschler et al. (1995),
19 Vamvakas et al. (1998) and Sinks et al. (1992) are less likely to provide a precise estimate of the
20 magnitude of the association given greater uncertainty in these studies compared to Briining et
21 al. (2003). For this reason, Briining et al. (2003) is preferred for meta-analysis treatment since it
22 is considered to better reflect risk in the target population than the two other studies. Another
23 study (Charbotel et al., 2006) of similar exposure conditions of a different base population and of
24 different case and control ascertainment methods as the Arnsberg region studies has become
25 available since the Arnsberg studies. This study shows a statistically significant elevation of risk
26 and high cumulative TCE exposure in addition to a positive trend with rank-order exposure
27 levels. Charbotel et al. (2006) adds evidence to observations from earlier studies on high TCE
28 exposures in Southern Germany and suggests that peak exposure may add to risk associated with
29 cumulative TCE exposure.
30
31 4.4.2.3. Examination of Possible Confounding Factors
32 Examination of potential confounding factors is an important consideration in the
33 evaluation of observations in the epidemiologic studies on TCE and kidney cancer. A known
34 risk factor for kidney cancer is cigarette smoking. Obesity, diabetes, hypertension and
35 antihypertensive medications, and analgesics are linked to kidney cancer, but causality has not
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-163 DRAFT—DO NOT CITE OR QUOTE
-------
1 been established (Moore et al., 2005; McLaughlin et al., 2006). On the other hand, fruit and
2 vegetable consumption is considered protective of kidney cancer risk (McLaughlin et al., 2006).
3 Studies by Asal et al. (1988), Partanen et al. (1991), McCredie and Stewart (1993), Auperin et al.
4 (1994), Chow et al. (1994), Mellemgaard et al. (1994), Mandel et al. (1995), Vamvakas et al.
5 (1998), Dosemeci et al. (1999), Pesch et al. (2000a), Briining et al. (2003), and Charbotel et al.
6 (2006) controlled for smoking and all studies except Pesch et al. (2000a) controlled for BMI.
7 Vamvakas et al. (1998) and Dosemeci et al. (1999) controlled for hypertension and or diuretic
8 intake in the statistical analysis. Because it is unlikely that exposure to trichloroethylene is
9 associated with smoking, body mass index, hypertension, or diuretic intake, these possible
10 confounders do not significantly affect the estimates of risk (NRC, 2006).
11 Direct examination of possible confounders is less common in cohort studies than in
12 case-control studies where information is obtained from study subjects or their proxies. Use of
13 internal controls, such as for Zhao et al. (2005), in general minimizes effects of potential
14 confounding due to smoking or socioeconomic status since exposed and referent subjects are
15 drawn from the same target population. Effect of smoking as a possible confounder may be
16 assessed indirectly through (1) examination of risk ratios for other smoking-related sites and
17 (2) examination of the expected contribution by these three factors to cancer risks. Lung cancer
18 risk in Zhao et al. (2005) was not elevated compared to referent subjects and this observation
19 suggests smoking patterns were similar between groups. Smoking was more prevalent in the
20 Raaschou-Nielsen et al. (2003) cohort than the background population as suggested by the
21 elevated risks for lung and other smoking-related sites; however, Raaschou-Nielsen et al. (2003)
22 do not consider smoking to fully explain the 20 and 40% excesses in renal cell carcinoma risk in
23 the cohort and subcohort. A high percentage of smokers in the cohort would be needed to
24 account for the magnitude of renal cell carcinoma excess. Specifically, Raaschou-Nielsen et al.
25 (2003) noted "a high smoking rate would be expected to generate a much higher excess risk of
26 lung cancer than was observed in this study."
27 The magnitude of confounding bias related to cigarette smoking in occupationally
28 employed populations to the observed lung, bladder and stomach cancer risk is minimal; less
29 than 20% for lung cancer and less than 10% for bladder and stomach cancers (Siemiatycki et al.,
30 1988; Leigh, 1996; Bang and Kim, 2001; Blair et al., 2007). Thus, in cohort studies lacking
31 direct adjustment for smoking and use of external referents, difference in cigarette smoking
32 between exposed and referent subjects is not sufficient to fully explain observed excess kidney
33 cancer risks associated with TCE, particularly, high TCE exposure. Information on possible
34 confounding due to BMI (obesity) and to diabetes is lacking in cohort studies; however, any
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-164 DRAFT—DO NOT CITE OR QUOTE
-------
1 uncertainties are likely small given the generally healthy nature of an employed population and
2 its favorable access to medical care.
3 Mineral oils such as cutting fluids or hydrazine common to some job titles with potential
4 TCE exposures (such as machinists, metal workers, and test stand mechanics) were included as
5 covariates in statistical analyses of Zhao et al. (2005), Boice et al. (2006) and Charbotel et al.
6 (2006, 2009). A TCE effect on kidney cancer incidence was still evident although effect
7 estimates were often imprecise due to lowered statistical power (Zhao et al., 2005; Charbotel et
8 al., 2006, 2009). Observed associations were similar in analyses including chemical coexposures
9 in both Zhao et al. (2005) and Charbotel et al. (2006, 2009) compared to chemical coexposure
10 unadjusted risks. The association or OR between high TCE score and kidney cancer incidence in
11 Zhao et al. (2005) was 7.71 (95% CI: 0.65, 91.4) after adjustment for other carcinogens including
12 hydrazine and cutting oils, compared to analyses unadjusted for chemical coexposures (4.90,
13 95% CI: 1.23, 19.6).
14 In Charbotel et al. (2006), exposure to TCE was strongly associated with exposure to
15 cutting fluids and petroleum oils (22 of the 37 TCE-exposed cases were exposed to both).
16 Statistical modeling of all factors significant at 10% threshold showed the OR for cutting fluids
17 to be almost equal to 1, whereas the OR for the highest level of TCE exposure was close to two
18 (Charbotel et al., 2006). Moreover, when exposure to cutting oils was divided into three levels, a
19 decrease in OR with level of exposure was found. In conditional logistic regression adjusted for
20 cutting oil exposure, the relative risk (OR) was similar to relative risks from unadjusted for
21 cutting fluid exposures (high cumulative TCE exposure: 1.96 [95% CI: 0.71-5.37] compared to
22 2.16 [95% CI: 1.02-4.60]; high cumulative and peak: 2.63 [95% CI: 0.79-8.83] compared to
23 2.73 [95% CI: 1.06-7.07] [Charbotel, 2006]). Charbotel et al. (2009) further examined TCE
24 exposure defined as the highest TWA in any job held, inferior to cumulative exposure given its
25 lack of consideration of TCE exposure potential in other jobs, either as exposure to TCE alone,
26 cutting fluids alone, or to both after adjusting for smoking, body mass index, age, sex, and
27 exposure to other oils (TCE alone: 1.62 [95% CI: 0.75, 3.44]); cutting fluids alone: 2.39
28 (95% CI: 0.52, 11.03); TCE >50-ppm TWA + cutting fluids: 2.70 (95% CI: 1.02, 7.17). There
29 were few cases exposed to cutting fluids alone (n = 3) or to TCE alone (n = 15), all of whom had
30 TCE exposure (in the highest exposed job held) of <35-ppm TWA, and the subgroup analyses
31 were of limited statistical power. A finding of higher risk for both cutting oil and TCE exposure
32 >50 ppm compared to cutting oil alone supports a TCE effect for kidney cancer. Adjustment for
33 cutting oil exposures, furthermore, did not greatly affect the magnitude of TCE effect measures
34 in the many analyses presented by Charbotel et al. (2006, 2009) suggesting cutting fluid
35 exposure as not greatly confounding TCE effect measures.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-165 DRAFT—DO NOT CITE OR QUOTE
-------
1 Boice et al. (2006) was unable to directly examine hydrazine exposure on TCE effect
2 measures because of a lack of model convergence in statistical analyses. Three of
3 7 TCE-exposed kidney cancer cases were identified with hydrazine exposure of 1.5 years or less
4 and the absence of exposure to the other 4 cases suggested confounding related to hydrazine was
5 unlikely to greatly modify observed association between TCE and kidney cancer.
6
7 4.4.2.4. Susceptible Populations—Kidney Cancer and Trichloroethylene (TCE) Exposure
8 Two studies of kidney cancer cases from the Arnsberg region in Germany have examined
9 the influence of polymorphisms of the glutathione-S-transferase metabolic pathway on renal cell
10 carcinoma risk and TCE exposure (Briining et al., 1997b; Wiesenhiitter et al., 2007). In their
11 study of 45 TCE-exposed male and female renal cell carcinoma cases pending legal
12 compensation and 48 unmatched male TCE-exposed controls, Briining et al. (1997b) observed a
13 higher prevalence of exposed cases homozygous and heterozygous for GST-MI positive, 60%,
14 than the prevalence for this genotype among exposed controls, 35%. The frequency of GST-MI
15 positive was lower among this control series than the frequency found in other European
16 population studies, 50% (Briining et al., 1997b). The prevalence of the GST-T1 positive
17 genotype was 93% among exposed cases and 77% among exposed controls. The prevalence of
18 GST-T1 positive genotype in the European population is 75% (Briining et al., 1997b).
19 Wiesenhiitter et al. (2007) compares the frequency of genetic polymorphism among
20 subjects from the renal cancer case-control study of Briining et al. (2003) and to the frequencies
21 of genetic polymorphisms in the areas of Dormund and Lutherstadt Wittenberg, Germany.
22 Wiesenhiitter et al. (2007) identified the genetic frequencies of GST-MI and GST-T1
23 phenotypes for 98 of the original 134 cases (73%) and 324 of the 401 controls (81%). The
24 prevalence of GST-MI positive genotype was 48% among all renal cell carcinoma cases, 40%
25 among TCE-exposed cases, and 52% among all controls. The prevalence of GST-T1 positive
26 genotypes was 81% among all cases and 81% among all controls. The prevalence of GST-T1
27 positive genotypes reported in this paper for all TCE-exposed cases was 20%. The numbers of
28 exposed (n = 4) and unexposed (n = 15) GST-T1 positive cases does not sum to the 79 cases with
29 the GST-T1 positive genotype identified in the table's first row; U.S. Environmental Protection
30 Agency (U.S. EPA) staff has written Professor Bolt requesting clarification of the data in Table 1
31 of Wiesenhiitter et al. (2007) (personal communication from Cheryl Siegel Scott to Professor
32 Herman Bolt, email dated August 05, 2008) [no reply received as of January, 2009 to request]).
33 Wiesenhiitter et al. (2007) noted background frequencies in the German population in the
34 expanded control group were 50% for GST-MI positive and 81% for GST-T1 positive
35 genotypes.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-166 DRAFT—DO NOT CITE OR QUOTE
-------
1 Observations in Briining et al. (1997b) and Wiesenhiitter et al. (2007) must be interpreted
2 cautiously. Few details are provided in these studies on selection criteria and not all subjects
3 from the Briining et al. (2003) case-control study are included. For GST-MI positive, the higher
4 prevalence among exposed cases in Briining et al. (1997b) compared Wiesenhiitter et al. (2007)
5 and the lower prevalence among controls compared to background frequency in the European
6 population may reflect possible selection biases. On the other hand, the broader base population
7 included in Briining et al. (2003) may explain the observed lower frequency of GST-MI positive
8 cases in Wiesenhiitter et al. (2007). Moreover, Wiesenhiitter et al. (2007) does not report
9 genotype frequencies for controls by exposure status and this information is essential to an
10 examination of whether renal cell carcinoma risk and TCE exposure may be modified by
11 polymorphism status.
12 Of the three larger (in terms of number of cases) studies that did provide results
13 separately by sex, Dosemeci et al. (1999) suggest that there may be a sex difference for TCE
14 exposure and renal cell carcinoma (OR: 1.04, [95% CI: 0.6, 1.7]) in males and 1.96 (95% CI:
15 1.0, 4.0 in females), while Raaschou-Nielsen (2003) report the same standardized incidence
16 ration (SIR = 1.2) for both sexes and crude ORs calculated from data from the Pesch et al.
17 (2000a) study (provided in a personal communication from Beate Pesch, Forschungsinstitut fiir
18 Arbeitsmedizin, to Cheryl Scott, U.S. EPA, 21 February 2008) are 1.28 for males and 1.23 for
19 females. Whether the Dosemeci et al. (1999) observations are due to susceptibility differences or
20 to exposure differences between males and females cannot be evaluated. Blair et al. (1998) and
21 Hansen et al. (2001) also present some results by sex, but these two studies have too few cases to
22 be informative about a sex difference for kidney cancer.
23
24 4.4.2.5. Meta-Analysis for Kidney Cancer
25 Meta-analysis (detailed methodology in Appendix C) was adopted as a tool for
26 examining the body of epidemiologic evidence on kidney cancer and TCE exposure and to
27 identify possible sources of heterogeneity. The meta-analyses of the overall effect of TCE
28 exposure on kidney cancer suggest a small, statistically significant increase in risk that was
29 stronger in a meta-analysis of the highest exposure group. There was no observable
30 heterogeneity across the studies for any of the meta-analyses and no indication of publication
31 bias. Thus, these findings of increased risks of kidney cancer associated with TCE exposure are
32 robust.
33 The meta-analysis of kidney cancer examines 14 cohort and case-control studies
34 identified through a systematic review and evaluation of the epidemiologic literature on TCE
35 exposure (Siemiatycki et al., 1991; Parent et al., 2000; Axelson et al., 1994; Anttila et al., 1995;
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-167 DRAFT—DO NOT CITE OR QUOTE
-------
1 Blair et al., 1998; Morgan et al., 1998; Boice et al., 1999; Dosemeci et al., 1999; Greenland et al.
2 1994; Pesch et al., 2000a; Hansen et al., 2001; Briining et al., 2003; Raaschou-Nielsen et al.,
3 2003; Zhao et al., 2005; Charbotel et al., 2006). Details of the systematic review and meta-
4 analysis of the TCE studies are fully discussed in Appendix B and C.
5 The pooled estimate from the primary random effects meta-analysis of the 14 studies was
6 1.25 (95% CI: 1.11, 1.41). The analysis was dominated by two (contributing almost 70% of the
7 weight) or three (almost 80% of the weight) large studies (Dosemeci et al., 1999; Pesch et al.,
8 2000a; Raaschou-Nielsen et al., 2003). Figure 4-1 arrays individual studies by their weight. No
9 single study was overly influential; removal of individual studies resulted in pooled RR (RRp)
10 estimates that were all statistically significant (p < 0.005) and that ranged from 1.22 (with the
11 removal of Briining et al. [2003]) to 1.27 (with the removal of Raaschou-Nielsen et al. [2003]).
12 Similarly, the overall RRp estimate was not highly sensitive to alternate RR estimate selections
13 nor was heterogeneity or publication bias apparent. Subgroup analyses were done examining the
14 cohort and case-control studies separately with the random effects model; the resulting RRp
15 estimates were 1.16 (95% CI: 0.96, 1.40) for the cohort studies and 1.41 (1.08, 1.83) for the case-
16 control studies. There was heterogeneity in the case-control subgroup, but it was not statistically
17 significant (p = 0.17).
18 Nine studies reported risks for higher exposure groups (Siemiatycki et al., 1991; Parent et
19 al., 2000; Blair et al., 1998; Morgan et al., 1998; Boice et al., 1999; Dosemeci et al., 1999; Pesch
20 et al., 2000a; Briining et al., 2003; Raaschou-Nielsen et al., 2003; Zhao et al., 2005; Charbotel et
21 al., 2006). Different exposure metrics were used in the various studies, and the purpose of
22 combining results across the different highest exposure groups was not to estimate an RRp
23 associated with some level of exposure. Instead, the focus on the highest exposure category was
24 meant to result in an estimate less affected by exposure misclassification. In other words, it is
25 more likely to represent a greater differential TCE exposure compared to people in the referent
26 group than the exposure differential for the overall (typically any versus none) exposure
27 comparison. Thus, if TCE exposure increases the risk of kidney cancer, the effects should be
28 more apparent in the highest exposure groups.
29 The RRp estimate from the random effects meta-analysis of the studies with results
30 presented for higher exposure groups was 1.59 (95% CI: 1.26, 2.01), higher than the RRp from
31 the overall kidney cancer meta-analysis. As with the overall analyses, the meta-analyses of the
32 highest-exposure groups were dominated by Pesch et al. (2000a) and Raaschou-Nielsen et al.
33 (2003), which provided about 70% of the weight. Axelson et al. (1994), Anttila et al. (1995) and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-168 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
vo *
I
§
***.
S'
1
I
H I
O >
HH Oq
H TO
TCE and Kidney Cancer
Study name Statistics for each study
Risk Lower Upper
ratio limit limit p-Value
Anttilo -IQQ^ n £7H n "3Q1 1 Q"37 n 77^n
Avalonn 1QQ4 1 1RH n ^91 9 ^£9 n 71 R9
Rnir'o -I QQQ H QQH H /79 9 H77 H Q7££
OroonlonH 1 QQ/ H QQH H 9QP Q 9QQ H QPf^Q
Ulonoon 9nn-1 1 1 HH H / 1 Q 9 QQ-1 H P/PP
^/lr\rnon 1 QQP i inr\i ih DD 1 -I/Q H ^H7 9 ^7f^ H 7/79
Raaschou-Nielsen 2003 RCC 1.200 0.950 1.516 0.1262
DoHi/-"3n 9HHP 1 1 PH H /79 9 Q^-1 H 79Q/
7hor\ 9Hn^ mr\r+ 9H \/ Ion 1 79H H Q77 7 P^Q H ARAC\
Bruning2003 2.470 1.359 4.488 0.0030
Ohorhntol 9HH7 hinh r^nnf ro-ovr\ 1 PPH H PPQ Q Q7f^ H HQP^
Dosemeci1999 1.300 0.895 1.889 0.1687
Pesch 2000 JTEM 1.240 1.030 1.492 0.0227
Qiomiotwr'L'i -I QCM H PHH H 9£7 9 9QQ H ^7HH
1.251 1.110 1.410 0.0002
0
1 0
Risk r
2 0
atio
5 '
and £
•-
-• —
4
;
)5% Cl
-•^-^
> J
5 1
0
random effects model; same for fixed
O
H
W
Figure 4-1. Meta-analysis of kidney cancer and overall TCE exposure (the pooled estimate is in the bottom
row). Symbol sizes reflect relative weights of the studies. The horizontal midpoint of the bottom diamond represents
the pooled RR estimate and the horizontal extremes depict the 95% Cl limits.)
-------
1 Hansen et al. (2001) do not report risk ratios for kidney cancer by higher exposure and a
2 sensitivity analysis was carried out to address reporting bias. The RRp estimate from the
3 primary random effects meta-analysis with null RR estimates (i.e., RR = 1.0) included for
4 Axelson et al. (1994), Anttila et al. (1995) and Hansen et al. (2001) to address reporting bias
5 associated with ever exposed was 1.53 (95% CI: 1.23, 1.91). Figure 4-2 arrays individual studies
6 by their weight. The inclusion of these 3 additional studies contributed less than 8% of the total
7 weight. No single study was overly influential; removal of individual studies resulted in RRp
8 estimates that were all statistically significant (p < 0.02) and that ranged from 1.43 (with the
9 removal of Raaschou-Nielsen et al. [2003]) to 1.58 (with the removal of Pesch et al. [2000a]).
10 Similarly, the RRp estimate was not highly sensitive to alternate RR estimate selections and
11 heterogeneity observed across the studies for any of the meta-analyses conducted with the
12 highest-exposure groups (all have/? < 0.002).
13 NRC (2006) deliberations on trichloroethylene commented on two prominent evaluations
14 of the then-current TCE epidemiologic literature using meta-analysis techniques, Wartenberg et
15 al. (2000) and Kelsh et al. (2005), submitted by Exponent-Health Sciences to NRC during their
16 deliberations. Wartenberg et al. (2000) reported an RRp of 1.7 (95% CI: 1.1, 2.7) for kidney
17 cancer incidence in the TCE subcohorts (Axelson et al., 1994; Anttila et al., 1995; Blair et al.,
18 1998; Henschler et al., 1995). For kidney cancer mortality in TCE subcohorts (Henschler et al.,
19 1995; Blair et al., 1998; Boice et al., 1999; Morgan et al., 1998; Ritz, 1999), Wartenberg et al.
20 (2000) reported an RRp of 1.2 (95% CI: 0.8, 1.7). Kelsh et al. (2005) examined a slightly
21 different grouping of cohort studies as did Wartenberg et al. (2000), presenting a pooled relative
22 risk estimate for kidney cancer incidence and mortality combined. The RRp for kidney cancer in
23 cohort studies (Axelson et al., 1994; Anttila et al., 1995; Blair et al., 1998; Morgan et al., 1998;
24 Boice et al., 1999; Hansen et al., 2001; Raaschou-Nielsen et al., 2003) was 1.29
25 (95% CI: 1.06-1.57) with no evidence of heterogeneity. Kelsh et al. (2005), also, presented
26 separately a pooled relative risk for renal cancer case-control studies and TCE. For case-control
27 studies (Siemiatycki et al., 1991; Greenland et al., 1994; Vamvakas et al., 1998; Dosemeci et al.,
28 1999; Pesch et al., 2000a; Briining et al., 2003), the RRp for renal cell carcinoma was 1.7
29 (95% CI: 1.0, 2.7) (interpolated from Figure 26 of NRC presentation) with evidence of
30 heterogeneity, and RRp of 1.2 (95% CI: 0.9, 1.4) (interpolated from Figure 26 of NRC
31 presentation) and no evidence of heterogeneity in a sensitivity analysis removing Vamvakas et
32 al. (1998) and Briining et al. (2003), two studies Kelsh et al. (2005) considered as "outliers."
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-170 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
vo *
I
§
***.
£3'
1
TO'
^
I
$
I
TCE and Kidney Cancer - highest exposure groups
Study namo
Study name
Boice 1999
and Kidney cancer - highest exposure groups
Stotiotioo for oooh otudy Riok ratio and 06% G\
ratio limit
Risk Lower
01382
study
Risk ratio and 95% Cl
limit p-Value
Upper
2003
n2003
fag
gftod conf re:exp
JfiaYb'oJer^QQy'good conf re:exp
^emfaty^l^l..
Antilla
1.531 1.225
1 0
.1 0.2 0.5
10
O
Oq
O
random effects model; same for fixed
Figure 4-2. Meta-analysis of kidney cancer and TCE exposure
estimates for Antilla, Axelson, and Hansen (see Appendix C text).
-highest exposure groups. With assumed null RR
H
W
-------
1 The present analysis was conducted according to NRC (2006) suggestions for
2 transparency, systematic review criteria, and examination of both cohort and case-control
3 studies. The present analysis includes the recently published study of Charbotel et al. (2006) and
4 an analysis that examines both the TCE subcohort and case-control studies together. As
5 discussed above, the pooled estimate from the primary random effects meta-analysis of the
6 14 studies was 1.25 (95% CI: 1.11, 1.41). Additionally, U.S. EPA examined kidney cancer risk
7 for higher exposure group. The RRp estimate from the random effects meta-analysis of the
8 studies with results presented for higher exposure groups was 1.59 (95% CI: 1.26, 2.01), higher
9 than the RRp from the overall kidney cancer meta-analysis, and 1.53 (95% CI: 1.23, 1.91) in the
10 meta-analysis with null RR estimates (i.e., RR = 1.0) to address possible reporting bias for three
11 studies.
12
13 4.4.3. Human Studies of Somatic Mutation of von Hippel-Lindau (VHL) Gene
14 Studies have been conducted to identify mutations in the VHL gene in renal cell
15 carcinoma patients, with and without TCE exposures (Wells et al., 2009; Charbotel et al., 2007;
16 Schraml et al., 1999; Brauch et al., 1999, 2004; Toma et al., 2008; Purge et al., 2007; Kenck et
17 al., 1996). Inactivation of the VHL gene through mutations, loss of heterozygosity (LOH) and
18 imprinting has been observed in about 70% of sporadic renal clear cell carcinomas, the most
19 common renal cell carcinoma subtype (Kenck et al., 1996). Other genes or pathways, including
20 c-myc activation and VEGF, have also been examined as to their role in various renal cell
21 carcinoma subtypes (Purge et al., 2007; Toma et al., 2008). Purge et al. (2007) reported that
22 there are molecularly distinct forms of RCC and possibly molecular differences between clear-
23 cell renal cell carcinoma subtypes. This study was performed using tissues obtained from
24 paraffin blocks. These results are supported by a more recent study which examined the genetic
25 abnormalities of clear cell renal cell carcinoma using frozen tissues from 22 cc-RCC patients and
26 paired normal tissues (Toma et al., 2008). This study found that 20 (91%) of the 22 cases had
27 LOH on chromosome 3p (harboring the VHL gene). Alterations in copy number were also found
28 on chromosome 9 (32% of cases), chromosome arm 14q (36% of cases), chromosome arm 5q
29 (45% of cases) and chromosome 7 (32% of cases), suggesting roles for multiple genetic changes
30 in RCC, and is also supported by genomes-wide single-nucleotide polymorphism analysis
31 (Toma et al., 2008).
32 Several papers link mutation of the VHL gene in renal cell carcinoma patients to TCE
33 exposure. These reports are based on comparisons of VHL mutation frequencies in TCE exposed
34 cases from renal cell carcinoma case-control studies or from comparison to background mutation
35 rates among renal cell carcinoma case series (see Table 4-40). Briining et al. (1997a) first
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-172 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported a high somatic mutation frequency (100%) in a series of 23 renal cell carcinomas cases
2 with medium to high intensity TCE exposure as determined by an abnormal SSCP pattern, with
3 most variations found in exon two. Only four samples were sequenced at the time of publication
4 and showed mutations in exon one, two and three (see Table 4-40). Some of the cases in this
5 study were from the case-control study of Vamvakas et al. (1998) (see Section 4.4.3 and
6 Appendix C).
7 Brauch et al. (1999, 2004) analyzed renal cancer cell tissues for mutations of the VHL
8 gene and reported increased occurrence of mutations in patients exposed to high concentrations
9 of TCE. In the first study (Brauch et al., 1999), an employer's liability or worker's
10 compensation registry was used to identify 44 renal cell carcinoma cases, 18 of whom were also
11 included in Briining et al. (1997a). Brauch et al. (1999) found multiple mutations in 42% of the
12 exposed patients who experienced any mutation and 57% showed loss of heterozygosity. A hot
13 spot mutation of cytosine to thymine at nucleotide 454 (C454T) was found in 39% of samples
14 that had a VHL mutation and was not found in renal cell cancers from nonexposed patients or in
15 lymphocyte DNA from either exposed or nonexposed cases or controls. As discussed above,
16 little information was given on how subjects were selected and whether there was blinding of
17 exposure status during the DNA analysis. In the second study, Brauch et al. (2004) investigated
18 21 of the 39 renal cell carcinoma patients identified as non-TCE exposed from Vamvakas et al.
19 (1998) for which tissue specimens were available. The earlier studies of Briining et al. (1997a)
20 or Brauch et al. (1999) included VHL sequencing of tissue specimens from TCE-exposed cases
21 from the renal cell carcinoma case-control study of Vamvakas et al. (1998). Brauch et al. (2004)
22 compared age at diagnosis and histopathologic parameters of tumors as well as somatic mutation
23 characteristics in the VHL tumor suppressor gene between the TCE-exposed and non-TCE
24 exposed renal cell carcinoma patient groups (TCE-exposed from their previous 1999 publication
25 to the non TCE-exposed cases newly sequenced in this study). Renal cell carcinoma did not
26 differ with respect to histopathologic characteristics in either patient group. Comparing results
27 from TCE-exposed and nonexposed patients revealed clear differences with respect to
28 (1) frequency of somatic VHL mutations, (2) incidence of C454T transition, and (3) incidence of
29 multiple mutations. The C454T hot spot mutation at codon 81 was exclusively detected in
30 tumors from TCE-exposed patients, as were multiple mutations. Also, the incidence of VHL
31 mutations in the TCE-exposed group was at least 2-fold higher than in the nonexposed group.
32 Overall, these finding support the view that the effect of TCE is not limited to clonal expansion
33 of cells mutated spontaneously or by some other agent.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-173 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-40. Summary of human studies on somatic mutations of the VHL gene"
to
o
vo
TCE exposure
status
Number of
subjects/
Number with
mutations (%)
Renal cell
carcinoma
subtype
Tissue type
analyzed
Assay
Number of
mutations
Type of
mutation
Missense
Nonmissense0
Briiningetal., 1997a
Exposed
23/23 (100%)
Unknown
Paraffin
SSCP,b sequencing13
23
1
3
Brauch et al., 1999
Exposed
44/33 (75%)
Unexposed
73/42 (58%)
Unknown
Paraffin, fresh
(lymphocyte)
SSCP, sequencing,
restriction enzyme
digestion
50
27
23
42
NA
NA
Schramletal., 1999
Exposed
9/3 (33%)
Clear cell 9 (75%)
Papillary 2 (18%)
Oncocytomas 1 (8%)
Unexposed
113/38
(34%)
Unknown
Paraffin
CGH, sequencing
4
1
3
50
Unknown
Unknown
Brauch etal., 2004
Exposed
17/14
(82%)
Unexposed
21/2 (10%)
Clear cell 37 (%)
Oncocytic adenoma 1 (%)
Bilateral metachronous 1
(%)
Paraffin
Sequencing
24
17
7
2
2
0
Charbotel etal., 2007
Exposed
25/2 (9%)
Unexposed
23/2 (8%)
Clear cell 5 1(75%)
Papillary 10 (10-15%)
Chromophobe 4 (5%)
Oncocytomas 4 (5%)
Paraffin, frozen tissues,
Bouin's fixative
Sequencing
2
1
1
2
1
1
'
I
I
§
Co
1
TO'
§
TO
"Adapted from NRC (2006) with addition of Schraml et al. (1999) and Charbotel et al. (2007).
bBy single stand conformation polymorphism (SSCP). Four (4) sequences confirmed by comparative genomic hybridization.
Includes insertions, frameshifts, and deletions.
O
Oq
O
H
W
-------
1 Brauch et al. (2004) were not able to analyze all RCCs from the Vamvakas study
2 (Vamvakas et al., 1998), in part because samples were no longer available. Using the data
3 described by Brauch et al. (2004) (VHL mutation found in 15 exposed and 2 nonexposed
4 individuals, and VHL mutation not found in 2 exposed and 19 unexposed individuals), the
5 calculated OR is 71.3. The lower bound of the OR including the excluded RCCs is derived from
6 the assumption that all 20 cases that were excluded were exposed but did not have mutations in
7 VHL (VHL mutations were found in 15 exposed and 2 unexposed individuals and VHL was not
8 found in 22 exposed and 18 unexposed individuals), leading to an OR of 6.5 that remains
9 statistically significant.
10 Charbotel et al. (2007) examines somatic mutations in the three VHL coding exons in
11 RCC cases from their case-control study (Charbotel et al., 2006). Of the 87 RCCs in the case-
12 control study, tissue specimens were available for 69 cases (79%) of which 48 were cc-RCC.
13 VHL sequencing was carried out for only the cc-RCC cases, 66% of the 73 cc-RCC cases in
14 Charbotel et al. (2006). Of the 48 cc-RCC cases available for VHL sequencing, 15 subjects were
15 identified with TCE exposure (31%), an exposure prevalence lower than 43% observed in the
16 case-control study. Partial to full sequencing of the VHL gene was carried out using polymerase
17 chain reaction (PCR) amplification and VHL mutation pattern recognition software of Beroud et
18 al. (1998). Full sequencing of the VHL gene was possible for only 26 RCC cases (36% of all
19 RCC cases). Single point mutations were identified in 4 cases (8% prevalence): 2 unexposed
20 cases, a G>C mutation in exon 2 splice site and a G>A in exon 1; one case identified with
21 low/medium exposure, T>C mutation in exon 2, and, one case identified with high TCE
22 exposure, T>C in exon 3. It should be noted that the two cases with T>C mutations were
23 smokers unlike the cases with G>A or G>C mutations. The prevalence of somatic VHL mutation
24 in this study is quite low compared to that observed in other RCC case series from this region;
25 around 50% (Bailly et al., 1995; Gallou et al., 2001). To address possible bias from
26 misclassification of TCE exposure, Charbotel et al. (2006) examined renal cancer risk for jobs
27 associated with a high level of confidence for TCE exposure. As would be expected if bias was
28 a result of misclassification, they observed a stronger association between higher confidence
29 TCE exposure and RCC, suggesting that some degree of misclassification bias is associated with
30 their broader exposure assessment approach. Charbotel et al. (2007) do not present findings on
31 VHL mutations for those subjects with higher level of confidence TCE exposure assignment.
32 Schraml et al. (1999) did not observe statistically significant differences in DNA
33 sequence or mutation type in a series of 12 renal cell carcinomas from subjects exposed to
34 solvents including varying TCE intensity and a parallel series of 113 clear cell carcinomas from
35 non-TCE exposed patients. Only 9 of the RCC were cc-RCC and were sequenced for mutations.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-175 DRAFT—DO NOT CITE OR QUOTE
-------
1 VHL mutations were observed in clear cell tumors only; four mutations in three TCE-exposed
2 subjects compared to 50 mutations in tumors of 38 nonexposed cases. Details as to exposure
3 conditions are limited to a statement that subjects had been exposed to high doses of solvents,
4 potential for mixed solvent exposures, and that exposure included a range of TCE
5 concentrations. Limitations of this study include having a wider range of TCE exposure
6 intensities as compared to the studies described above (Briining et al., 1997a; Brauch et al., 1999,
7 2004), which focused on patients exposed to higher levels of TCE, and the limited number of
8 TCE-exposed subjects analyzed, being the smallest of all available studies on RCC, TCE and
9 VHL mutation. For these reasons, Schraml et al. (1999) is quite limited for examining the
10 question of VHL mutations and TCE exposure.
11 A number of additional methodological issues need to be considered in interpreting these
12 studies. Isolation of DNA for mutation detection has been performed using various tissue
13 preparations, including frozen tissues, formalin fixed tissues and tissue sections fixed in Bouin's
14 solution. Ideally, studies would be performed using fresh or freshly frozen tissue samples to
15 limit technical issues with the DNA extraction. When derived from other sources, the quality
16 and quantity of the DNA isolated can vary, as the formic acid contained in the formalin solution,
17 fixation time and period of storage of the tissue blocks often affect the quality of DNA. Picric
18 acid contained in Bouin's solution is also known to degrade nucleic acids resulting in either low
19 yield or poor quality of DNA. In addition, during collection of tumor tissues, contamination of
20 neighboring normal tissue can easily occur if proper care is not exercised. This could lead to the
21 'dilution effect' of the results—i.e., because of the presence of some normal tissue, frequency of
22 mutations detected in the tumor tissue can be lower than expected. These technical difficulties
23 are discussed in these papers, and should be considered when interpreting the results.
24 Additionally, selection bias is possible given tissue specimens were not available for all RCC
25 cases in Vamvakas et al. (1998) or in Charbotel et al. (2006). Some uncertainty associated with
26 misclassification bias is possible given the lack of TCE exposure information to individual
27 subjects in Schraml et al. (1999) and in Charbotel et al. (2007) from their use of broader
28 exposure assessment approach compared to that associated with the higher confident exposure
29 assignment approach. A recent study by Nickerson et al. (2008) addresses many of these
30 concerns by utilizing more sensitive methods to look at both the genetic and epigenetic issues
31 related to VHL inactivation. This study was performed on DNA from frozen tissue samples and
32 used a more sensitive technique for analysis for mutations (endonuclease scanning) as well as
33 analyzing for methylation changes that may lead to inactivation of the VHL gene. This method
34 of analysis was validated on tissue samples with known mutations. Of the 205 cc-RCC samples
35 analyzed, 169 showed mutations in the VHL gene (82.4%). Of those 36 without mutation, 11
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-176 DRAFT—DO NOT CITE OR QUOTE
-------
1 were hypermethylated in the promoter region, which will also lead to inactivation of the VHL
2 gene. Therefore, this study showed inactivating alterations in the VHL gene (either by mutation
3 or hypermethylation) in 91% tumor samples analyzed.
4 The limited animal studies examining the role of VHL mutation following exposure to
5 chemicals including TCE are described below in Section 4.4.6.1.1. Conclusions as to the role of
6 VHL mutation in TCE-induced kidney cancer, taking into account both human and experimental
7 data, are presented below in Section 4.4.7.
8
9 4.4.4. Kidney Noncancer Toxicity in Laboratory Animals
10 Acute, subchronic, and chronic exposures to TCE cause toxicity to the renal tubules in
11 rats and mice of both sexes. Nephrotoxicity from acute exposures to TCE has only been reported
12 at relatively high doses, although histopathological changes have not been investigated in these
13 experiments. Chakrabarty and Tuchweber (1988) found that TCE administered to male F344
14 rats by intraperitoneal injection (723-2,890 mg/kg) or by inhalation (1,000-2,000 ppm for
15 6 hours) produced elevated urinary NAG, y-glutamyl transpeptidase (GOT), glucose excretion,
16 blood urea nitrogen (BUN), and high molecular weight protein excretion, characteristic signs of
17 proximal tubular, and possibly glomerular injury, as soon as 24 hours postexposure. In the
18 intraperitoneal injection experiments, inflammation was observed, although some inflammation
19 is expected due to the route of exposure, and nephrotoxicity effects were only statistically
20 significantly elevated at the highest dose (2,890 mg/kg). In the inhalation experiments, the
21 majority of the effects were statistically significant at both 1,000 and 2,000 ppm. Similarly, at
22 these exposures, renal cortical slice uptake of />-aminohippurate was inhibited, indicating
23 reduced proximal tubular function. Cojocel et al. (1989) found similar effects in mice
24 administered TCE by intraperitoneal injection (120-1,000 mg/kg) at 6 hours postexposure, such
25 as the dose-dependent increase in plasma BUN concentrations and decrease in/>-aminohippurate
26 accumulation in renal cortical slices. In addition, malondialdehyde (MDA) and ethane
27 production were increased, indicating lipid peroxidation.
28 Kidney weight increases have been observed following inhalation exposure to TCE in
29 both mice (Kjellstrand et al., 1983b) and rats (Woolhiser et al., 2006). Kjellstrand et al. (1983b)
30 demonstrated an increase in kidney weights in both male (20% compared to control) and female
31 (10% compared to control) mice following intermittent and continuous TCE whole-body
32 inhalation exposure (up to 120 days). This increase was significant in males as low as 75 ppm
33 exposure and in females starting at 150-ppm exposure. The latter study, an unpublished report
34 by Woolhiser et al. (2006), was designed to examine immunotoxicity of TCE but also contains
35 information regarding kidney weight increases in female Sprague Dawley (SD) rats exposed to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-177 DRAFT—DO NOT CITE OR QUOTE
-------
1 0-, 100-, 300-, and 1,000-ppm TCE for 6 hours/day, 5 days/week, for 4 weeks. Relative kidney
2 weights were significantly elevated (17.4% relative to controls) at 1,000-ppm TCE exposure.
3 However, the small number of animals and the variation in initial animal weight limit the ability
4 of this study to determine statistically significant increases.
5 Similarly, overt signs of subchronic nephrotoxicity, such as changes in blood or urinary
6 biomarkers, are also primarily a high dose phenomenon, although histopathological changes are
7 evident at lower exposures. Green et al. (1997b) reported administration of 2,000 mg/kg/d TCE
8 by corn oil gavage for 42 days in F344 rats caused increases of around 2-fold of control results in
9 urinary markers of nephrotoxicity such as urine volume and protein (both 1.8*), NAG (1.6*),
10 glucose (2.2x) and ALP (2.Ox), similar to the results of the acute study of Chakrabarty and
11 Tuchweber (1988), above. At lower dose levels, Green et al. (1998b) reported that plasma and
12 urinary markers of nephrotoxicity were unchanged. In particular, after 1-28 day exposures to
13 250 or 500 ppm TCE for 6 hours/day, there were no statistically significant differences in plasma
14 levels of BUN or in urinary levels of creatinine, protein, ALP, NAG, or GGT. However,
15 increased urinary excretion of formic acid, accompanied by changes in urinary pH and increased
16 ammonia, was found at these exposures. Interestingly, at the same exposure level of 500 ppm
17 (6 hours/day, 5 days/week, for 6 months), Mensing et al. (2002) reported elevated excretion of
18 low molecular weight proteins and NAG, biomarkers of nephrotoxicity, but after the longer
19 exposure duration of 6 months.
20 Numerous studies have reported histological changes from TCE exposure for subchronic
21 and chronic durations (Maltoni et al., 1988, 1986; Mensing et al., 2002; NTP, 1990, 1988). As
22 summarized in Table 4-41, in 13-week studies in F344 rats and B6C3F1 mice, NTP (1990)
23 reported relatively mild cytomegaly and karyomegaly of the renal tubular epithelial cells at the
24 doses 1,000-6,000 mg/kg/d (at the other doses, tissues were not examined). The NTP report
25 noted that "these renal effects were so minimal that they were diagnosed only during a
26 reevaluation of the tissues ... prompted by the production of definite renal toxicity in the 2-year
27 study." In the 6 month, 500-ppm inhalation exposure experiments of Mensing et al. (2002),
28 some histological changes were noted in the glomeruli and tubuli of exposed rats, but they
29 provided no detailed descriptions beyond the statement that "perivascular, interstitial infections
30 and glomerulonephritis could well be detected in kidneys of exposed rats."
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-178 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-41. Summary of renal toxicity and tumor findings in gavage studies
of trichloroethylene by NTP (1990)
Sex
Dose (mg/kg)a
Cytomegaly and
karyomegaly incidence
(severityb)
Adenoma
(overall;
terminal)
Adenocarcinoma
(overall;
terminal)
1/d, 5 d/wk, 13-wk study, F344/N rats
Male
Female
0, 125, 250, 500, 100
2,000
0, 62.5, 125, 250,
500
1,000
Tissues not evaluated
8/9 (Minimal/mild)
Tissues not evaluated
5/10 (Equivocal/minimal)
None reported
1/d, 5 d/wk, 13-wk study, B6C3Fi mice
Male
Female
0, 375, 750, 1,500
3,000
6,000
0, 375, 750, 1,500
3,000
6,000
Tissues not evaluated
7/1 Oc (Mild/moderate)
d
Tissues not evaluated
9/10 (Mild/moderate)
1/10 (Mild/moderate)
None reported
1/d, 5 d/wk, 103-wk study, F344/N rats
Male
Female
0
500
1,000
0
500
1,000
0% (0)
98% (2.8)
98% (3.1)
0% (0)
100% (1.9)
100% (2.7)
0/48; 0/33
2/49; 0/20
0/49; 0/16
0/50; 0/37
0/49; 0/33
0/48; 0/26
0/48; 0/33
0/49; 0/20
3/49; 3/16e
0/50; 0/37
0/49; 0/33
1/48; 1/26
1/d, 5 d/wk, 103-wk study, B6C3Fi mice
Male
Female
0
1,000
0
1,000
0% (0)
90% (1.5)
0% (0)
98% (1.8)
1/49; 1/33
0/50; 0/16
0/48; 0/32
0/49; 0/23
0/49; 0/33
1/50; 0/16
0/48; 0/32
0/49; 0/23
4
5
6
7
8
9
10
11
12
"Corn oil vehicle.
^Numerical scores reflect the average grade of the lesion in each group (1, slight; 2, moderate; 3, well marked; and
4, severe).
cObserved in four mice that died after 7-13 weeks and in three that survived the study.
All mice died during the first week.
ep = 0.028.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-179 DRAFT—DO NOT CITE OR QUOTE
-------
1 After 1-2 years of chronic TCE exposure by gavage (NCI, 1976; NTP, 1990, 1988) or
2 inhalation (Maltoni et al., 1988) (see Tables 4-41-4-45), both the incidence and severity of these
3 effects increases, with mice and rats exhibiting lesions in the tubular epithelial cells of the inner
4 renal cortex that are characterized by cytomegaly, karyomegaly, and toxic nephrosis. As with
5 the studies at shorter duration, these chronic studies reported cytomegaly and karyomegaly of
6 tubular cells. NTP (1990) specified the area of damage as the pars recta, located in the
7 corticomedullary region. It is important to note that these effects are distinct from the chronic
8 nephropathy and inflammation observed in control mice and rats (Lash et al., 2000b; Maltoni et
9 al., 1988; NCI, 1976).
10 These effects of TCE on the kidney appear to be progressive. Maltoni et al. (1988) noted
11 that the incidence and degree of renal toxicity increased with increased exposure time and
12 increased time from the start of treatment. As mentioned above, signs of toxicity were present in
13 the 13 week study (NTP, 1988), and NTP (1990) noted cytomegaly at 26 weeks. NTP (1990)
14 noted that as "exposure time increased, affected tubular cells continued to enlarge and additional
15 tubules and tubular cells were affected," with toxicity extending to the cortical area as kidneys
16 became more extensively damaged. NTP (1988, 1990) noted additional lesions that increased in
17 frequency and severity with longer exposure, such as dilation of tubules and loss of tubular cells
18 lining the basement membrane ("stripped appearance" [NTP, 1988] or flattening of these cells
19 [NTP, 1990]). NTP (1990) also commented on the intratubular material and noted that the
20 tubules were empty or "contained wisps of eosinophilic material."
21 With gavage exposure, these lesions were present in both mice and rats of both sexes, but
22 were on average more severe in rats than in mice, and in male rats than in female rats (NTP,
23 1990). Thus, it appears that male rats are most sensitive to these effects, followed by female rats
24 and then mice. This is consistent with the experiments of Maltoni et al. (1988), which only
25 reported these effects in male rats. The limited response in female rats or mice of either sex in
26 these experiments may be related to dose or strain. The lowest chronic gavage doses in the
27 National Cancer Institute (NCI, 1976) and NTP (1988, 1990) F344 rat experiments was
28 500 mg/kg/d, and in all these cases at least 80% (and frequently 100%) of the animals showed
29 cytomegaly or related toxicity. By comparison, the highest gavage dose in the Maltoni et al.
30 (1988) experiments (250 mg/kg/d) showed lower incidences of renal cytomegaly and
31 karyomegaly in male Sprague-Dawley rats (47% and 67%, overall and corrected incidences) and
32 none in female rats. The B6C3F1 mouse strain was used in the NCI (1976), NTP (1990), and
33 Maltoni et al. (1988) studies (see Tables 4-41-4-45). While the two gavage studies (NCI, 1976;
34 NTP, 1990) were consistent, reporting at least 90% incidence of cytomegaly and karyomegaly at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-180 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-42. Summary of renal toxicity and tumor findings in gavage studies
of trichloroethylene by NCI (1976)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Sex
Dose (mg/kg)a
Toxic nephrosis
(overall; terminal)
Adenoma or adenocarcinoma
(overall; terminal)15
1/d, 5 d/wk, 2-yr study, Osborn-Mendel rats
Males
Females
0
549
1,097
0
549
1,097
0/20; 0/2
46/50; 7/7
46/50; 3/3
0/20; 0/8
39/48; 12/12
48/50; 13/13
0/20; 0/2
l/50;c 0/7
0/50; 0/3
0/20; 0/8
0/48; 0/12
0/50; 0/13
1/d, 5 d/wk, 2-yr study, B6C3F1 mice
Males
Females
0
1,169
2,339
0
869
1,739
0/20; 0/8
48/50; 35/35
45/50; 20/20
0/20; 0/17
46/50; 40/40
46/47;e 39/39
0/20; 0/8
0/50; 0/35
l/50;d 1/20
0/20; 0/17
0/50; 0/40
0/47; 0/39
""Treatment period was 48 weeks for rats, 66 weeks for mice. Doses were changed several times during the study
based on monitoring of body weight changes and survival. Dose listed here is the time-weighted average dose
over the days on which animals received a dose.
bA few malignant mixed tumors and hamartomas of the kidney were observed in control and low dose male rats, but
are not counted here.
Tubular adenocarcinoma.
dTubular adenoma.
eOne mouse was reported with "nephrosis," but not "nephrosis toxic," and so was not counted here.
Table 4-43. Summary of renal toxicity findings in gavage studies of
trichloroethylene by Maltoni et al. (1988)
Sex
Dose (mg/kg)a
Megalonucleocytosisb (overall;
corrected0)
1/d, 4-5 d/wk, 52-wk exposure, observed for lifespan, Sprague-Dawley rats
Males
Females
0
50
250
0
50
250
0/20; 0/22
0/30; 0/24
14/30; 14/21
0/30; 0/30
0/30; 0/29
0/30; 0/26
"Olive oil vehicle.
bRenal tubuli megalonucleocytosis is the same as cytomegaly and karyomegaly of renal tubuli cells (Maltoni et al.,
1988).
Denominator for "corrected" incidences is the number of animals alive at the time of the first kidney lesion in this
experiment (39 weeks).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-181 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-44. Summary of renal toxicity and tumor incidence in gavage studies
of trichloroethylene by NTP (1988)
Sex
Dose (mg/kg)*
Cytomegaly
Toxic
Nephropathy
Adenoma
(overall;
terminal)
Adenocarcinoma
(overall;
terminal)
1/d, 5 d/wk, 2-yr study, ACT rats
Male
Female
0
500
1,000
0
500
1,000
0/50
40/49
48/49
0/48
43/47
42/43
0/50
18/49
18/49
0/48
21/47
19/43
0/50; 0/38
0/49; 0/19
0/49; 0/1 1
0/48; 0/34
2/47; 1/20
0/43; 0/19
0/50; 0/38
1/49; 0/19
0/49; 0/1 1
0/48; 0/34
1/47; 1/20
1/43; 0/19
1/d, 5 d/wk, 2-yr study, August rats
Male
Female
0
500
1,000
0
500
1,000
0/50
46/50
46/49
0/49
46/48
50/50
0/50
10/50
31/49
0/49
8/48
29/50
0/50; 0/21
1/50; 0/13
1/49; 1/16
1/49; 1/23
2/48; 1/26
0/50; 0/25
0/50; 0/21
1/50; 1/13
0/49; 0/16
0/49; 0/23
2/48; 2/26
0/50; 0/25
1/d, 5 d/wk, 2-yr study, Marshall rats
Male
Female
0
500
1,000
0
500
1,000
0/49
48/50
47/47
0/50
46/48
43/44
0/49
18/50
23/47
0/50
30/48
30/44
0/49; 0/26
1/50; 0/12
0/47; 0/6
1/50; 0/30
1/48; 1/12
0/44; 0/10
0/49; 0/26
0/50; 0/12
1/47; 0/6
0/50; 0/30
1/48; 0/12
1/44; 1/10
1/d, 5 d/wk, 2-yr study, Osborne-Mendel rats
Male
Female
0
500
1,000
0
500
1,000
0/50
48/50
49/50
0/50
48/50
49/49
0/50
39/50
35/50
0/50
30/50
39/49
0/50; 0/22
6/50; 5/17
1/50; 1/15
0/50; 0/20
0/50; 0/1 1
1/49; 0/7
0/50; 0/22
0/50; 0/17
1/50; 0/15
0/50; 0/20
0/50; 0/1 1
0/49; 0/7
4
5
*Corn oil vehicle.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-182 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-45. Summary of renal toxicity and tumor findings in inhalation
studies of trichloroethylene by Maltoni et al. (1988)a
Sex
Concentration
(ppm)
Meganucleocytosisb
(overall; corrected)
Adenoma
(overall;
corrected)
Adenocarcinoma
(overall;
corrected)
7 h/d, 5 d/wk, 2-yr exposure, observed for lifespan, Sprague-Dawley rats0
Male
Female
0
100
300
600
0
100
300
600
0/135; 0/122
0/130; 0/121
22/130; 22/1 16
101/130; 101/124
0/145; 0/141
0/130; 0/128
0/130; 0/127
0/130; 0/127
0/135; 0/122
1/130; 1/121
0/130; 0/1 16
1/130; 1/124
0/145; 0/141
1/130; 1/128
0/130; 0/127
0/130; 0/127
0/135; 0/122
0/130; 0/121
0/130; 0/1 16
4/130; 4/124
0/145; 0/141
0/130; 0/128
0/130; 0/127
1/130; 1/127
7 h/d, 5 d/wk, 78-wk exposure, observed for lifespan, B6C3F1 miced
Male
Female
0
100
300
600
0
100
300
600
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
0/90
1/90
0/90
0/90
1/90
0/90
0/90
0/90
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
aThree inhalation experiments in this study found no renal megalonucleocytosis, adenomas, or adenocarcinomas:
BT302 (8-week exposure to 0, 100, 600 ppm in Sprague-Dawley rats); BT303 (8-week exposure to 0, 100, or 600
ppm in Swiss mice); and BT305 (78-week exposure to 0, 100, 300, or 600 ppm in Swiss mice).
bRenal tubuli meganucleocytosis is the same as cytomegaly and karyomegaly of renal tubuli cells (Maltoni et al.,
1988).
'Combined incidences from experiments BT304 and BT304bis. Corrected incidences reflect number of rats alive at
47 weeks, when the first renal tubular megalonucleocytosis in these experiments appeared.
dFemale incidences are from experiment BT306, while male incidences are from experiment BT306bis, which was
added to the study because of high, early mortality due to aggressiveness and fighting in males in experiment
BT306. Corrected incidences not show, because only the renal adenocarcinomas appeared at 107 weeks in the
male and 136 in the female, when the most of the mice were already deceased.
all studied doses, whether dose accounts for the lack of kidney effects in Maltoni et al. (1988)
requires comparing inhalation and gavage dosing. Such comparisons depend substantially on the
internal dose metric, so conclusions as to whether dose can explain differences across studies
cannot be addressed without dose-response analysis using physiologically based
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-183 DRAFT—DO NOT CITE OR QUOTE
-------
1 pharmacokinetic (PBPK) modeling. Some minor differences were found in the multistrain NTP
2 study (1988), but the high rate of response makes distinguishing among them difficult. Soffritti
3 (personal communication with JC Caldwell, February 14, 2006) did note that the colony from
4 which the rats in Maltoni et al. (1986, 1988) experiments were derived had historically low
5 incidences of chronic progressive nephropathy and renal cancer.
6
7 4.4.5. Kidney Cancer in Laboratory Animals
8 4.4.5.1. Inhalation Studies of Trichloroethylene (TCE)
9 A limited number of inhalation studies examined the carcinogenicity of TCE, with no
10 statistically-significantly increases in kidney tumor incidence reported in mice or hamsters
11 (Fukuda et al., 1983; Henschler et al., 1980; Maltoni et al., 1988, 1986). The cancer bioassay by
12 (Maltoni et al., 1986, 1988) reported no statistically significant increase in kidney tumors in mice
13 or hamsters, but renal adenocarcinomas were found in male (4/130) and female (1/130) rats at
14 the high dose (600 ppm) after 2 years exposure and observation at natural death. In males, these
15 tumors seemed to have originated in the tubular cells, and were reported to have never been
16 observed in over 50,000 Sprague-Dawley rats (untreated, vehicle-treated, or treated with
17 different chemicals) examined in previous experiments in the same laboratory (Maltoni et al.,
18 1986). The renal adenocarcinoma in the female rat was cortical and reported to be similar to that
19 seen infrequently in historical controls. This study also demonstrated the appearance of
20 increased cytokaryomegaly or megalonucleocytosis, a lesion that was significantly and dose-
21 dependently increased in male rats only (see Table 4-45). Maltoni et al. (1986) noted that some
22 considerations supported either the hypothesis that these were precursor lesions of renal
23 adenocarcinomas cancer or the hypothesis that these are not precursors but rather the
24 morphological expression of TCE-induced regressive changes. The inhalation studies by Fukuda
25 et al. (1983) in Sprague-Dawley rats and female ICR mice, reported one clear cell carcinoma in
26 rats exposed to the highest concentration (450 ppm) but saw no increase in kidney tumors in
27 mice. This result was not statistically significant (see Table 4-46). One negative study
28 (Henschler et al., 1980) tested NMRI mice, Wistar rats, and Syrian hamsters of both sexes (60
29 animals per strain), and observed no significant increase in renal tubule tumors any of the species
30 tested. Benign adenomas were observed in male mice and rats, a single adenocarcinoma was
31 reported in male rats at the highest dose, and no renal adenocarcinomas reported in females of
32 either species (see Table 4-46). Renal cell carcinomas appear to be very rare in Wistar rats, with
33 historical control rates reported to be about 0.4% in males and 0.2% in females (Potericki and
34 Walsh, 1998), so these data are very limited in power to detect small increases in their incidence.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-184 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-46. Summary of renal tumor findings in inhalation studies of
trichloroethylene by Henschler et al. (1980)a and Fukuda et al. (1983)b
Sex
Concentration (ppm)
Adenomas
Adenocarcinomas
6 h/d, 5 d/wk, 18-month exposure, 30-month observation, Han:NMRI mice (Henschler et al.,
1980)
Males
Females
0
100
500
0
100
500
4/30
1/29
1/29
0/29
0/30
0/28
1/30
0/30
0/30
0/29
0/30
0/28
6 h/d, 5 d/wk, 18-month exposure, 36-month observation, Han:WIST rats (Henschler et al., 1980)
Males
Females
0
100
500
0
100
500
2/29
1/30
2/30
0/28
0/30
1/30
0/29
0/30
1/30
0/28
0/30
0/30
7 h/d, 5 d/wk, 2-yr study, Crj:CD (SD) rats (Fukuda et al., 1983)
Females
0
50
150
450
0/50
0/50
0/47
0/51
0/50
0/50
0/47
1/50
4
5
6
7
8
9
10
11
12
13
14
15
16
17
"Henschler et al. (1980) observed no renal tumors in control or exposed Syrian hamsters.
bFukuda et al. (1983) observed no renal tumors in control or exposed Crj:CD-l (ICR) mice.
4.4.5.2. Gavage and Drinking Water Studies of Trichloroethylene (TCE)
Several chronic gavage studies exposing multiple strains of rats and mice to 0-3,000
mg/kg TCE for at least 52 weeks have been conducted (see Tables 4-41-4-44, 4-47) (Henschler
et al., 1984; Maltoni et al., 1986; NCI, 1976; NTP, 1988, 1990; Van Duuren et al., 1979). Van
Duuren et al. (1979) examined TCE and 14 other halogenated compounds for carcinogenicity in
both sexes of Swiss mice. While no excess tumors were observed, the dose rate (0.5 mg once
per week, or an average dose rate of approximately 2.4 mg/kg/d for a 30 g mouse) is about 400-
fold lower than that in the other gavage studies. Inadequate design and reporting of this study
limit the ability to use the results as an indicator of TCE carcinogenicity. In the NCI (1976)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-185 DRAFT—DO NOT CITE OR QUOTE
-------
1 study, the results for Osborne-Mendel rats were considered by the authors to be inconclusive due
2 to significant early mortality. In rats of both sexes, no increase was seen in primary tumor
3 induction over that observed in controls. While both sexes of B6C3F1 mice showed a
4 compound-related increase in nephropathy, no increase in tumors over controls was observed.
5 The NCI study (1976) used technical grade TCE which contained two known carcinogenic
6 compounds as stabilizers (epichlorohydrin and 1,2-epoxybutane). However, a subsequent study
7 by Henschler et al. (1984) in mice reported no significant differences in systemic tumorigenesis
8 between pure, industrial, and stabilized TCE, suggesting that concentrations of these stabilizers
9 are too low to be the cause of tumors. A later gavage study by NTP (1988), using TCE stabilized
10 with diisopropylamine, observed an increased incidence of renal tumors in all four strains of rats
11 (ACI, August, Marshall, and Osborne-Mendel). All animals exposed for up to 2 years (rats and
12 mice) had non-neoplastic kidney lesions, even if they did not later develop kidney cancer (see
13 Table 4-44). This study was also considered inadequate by the authors because of chemically
14 induced toxicity, reduced survival, and incomplete documentation of experimental data. The
15 final NTP study (1990) in male and female F344 rats and B6C3F1 mice used epichlorohydrin-
16 free TCE. Only in the highest-dose group (1,000 mg/kg) of male F344 rats was renal carcinoma
17 statistically significant increased. The results for detecting a carcinogenic response in rats were
18 considered by the authors to be equivocal because both groups receiving TCE showed
19 significantly reduced survival compared to vehicle controls and because of a high rate (e.g., 20%
20 of the animals in the high-dose group) of death by gavage error. However, historical control
21 incidences at NTP of kidney tumors in F344 rats is very low,2 lending biological significance to
22 their occurrence in this study, despite the study's limitations. Cytomegaly and karyomegaly
23 were also increased, particularly in male rats. The toxic nephropathy observed in both rats and
24 mice and contributed to the poor survival rate (see Table 4-41). As discussed previously, this
25 toxic nephropathy was clearly distinguishable from the spontaneous chronic progression
26 nephropathy commonly observed in aged rats.
2 NTP (1990) reported a historical control incidence of 0.4% in males. The NTP web site reports historical control
rates of renal carcinomas for rats dosed via corn oil gavage on the NIH-07 diet (used before 1995, when the TCE
studies were conducted) to be 0.5% (2/400) for males and 0% (0/400) for females
(http://ntp-server.niehs.nih.gov/ntp/research/database searches/historical controls/path/r gavco.txt). In addition,
the 2 occurences in males came from the same study, with all other studies reporting 0/50 carcinomas.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-186 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-47. Summary of renal tumor findings in gavage studies of
trichloroethylene by Henschler et al. (1984)a and Van Duuren et al. (1979)h
Sex
(TCE dose)
Control or TCE Exposed
(Stabilizers if present)
Adenomas
Adenocarcinomas
5 d/wk, 18-month exposure, 24-month observation, Swiss mice (Henschler et al., 1984)
Males
(2.4g/kgbw)
Females
(1.8g/kgbw)
Control (none)
TCE (triethanolamine)
TCE (industrial)
TCE (epichlorohydrin (0.8%))
TCE (1,2-epoxybutane (0.8%))
TCE (both epichlorohydrin (0.25%)
and 1,2-epoxybutane (0.25%))
Control (none)
TCE (triethanolamine)
TCE (industrial)
TCE (epichlorohydrin (0.8%))
TCE (1,2-epoxybutane (0.8%))
TCE (both epichlorohydrin (0.25%)
and 1,2-epoxybutane (0.25%))
1/50
1/50
0/50
0/50
2/50
0/50
0/50
4/50
0/50
0/50
0/50
0/50
1/50
1/50
0/50
0/50
2/50
0/50
1/50
0/50
0/50
0/50
0/50
0/50
1 d/wk, 89-week exposure, Swiss rats (Van Duuren et al., 1979)
Males
(O.Smg)
Females
(O.Smg)
Control
TCE (unknown)
Control
TCE(unknown)
0/30
0/30
0/30
0/30
0/30
0/30
0/30
0/30
4
5
6
7
8
9
10
11
12
13
14
aHenschler et al. (1984) Due to poor condition of the animals resulting from the nonspecific toxicity of high doses of
TRI and/or the additives, gavage was stopped for all groups during weeks 35-40, 65 and 69-78, and all doses were
reduced by a factor of 2 from the 40th week on.
bVan Duuren et al. (1979) observed no renal tumors in control or exposed Swiss mice.
4.4.5.3. Conclusions: Kidney Cancer in Laboratory Animals
Chronic TCE carcinogenicity bioassays have shown evidence of neoplastic lesions in the
kidney in rats (mainly in males, with less evidence in females), treated via inhalation and gavage.
As discussed above, individual studies have a number of limitations and have shown limited
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-187 DRAFT— DO NOT CITE OR QUOTE
-------
1 increases in kidney tumors. However, given the rarity of these tumors as assessed by historical
2 controls and the repeatability of this result, these are considered biologically significant.
3
4 4.4.6. Role of Metabolism in Trichloroethylene (TCE) Kidney Toxicity
5 It is generally thought that one or more TCE metabolites rather than the parent compound
6 are the active moieties for TCE nephrotoxicity. As reviewed in Section 3.3, oxidation by CYPs,
7 of which CYP2EI is thought to be the most active isoform, results in the production of chloral
8 hydrate, trichloroacetic acid, dichloroacetic acid and trichloroethanol. The glutathione
9 conjugation pathway produces metabolites such as DCVG, DCVC, dichlorovinylthiol, and
10 NAcDCVC. Because several of the steps for generating these reactive metabolites occur in the
11 kidney, the GSH conjugation pathway has been thought to be responsible for producing the
12 active moiety or moieties of TCE nephrotoxicity. A comparison of TCE's nephrotoxic effects
13 with the effects of TCE metabolites, both in vivo and in vitro, thus, provides a basis for assessing
14 the relative roles of different metabolites. While most of the available data have been on
15 metabolites from GSH conjugation, such as DCVC, limited information is also available on the
16 major oxidative metabolites TCOH and TCA.
17
18 4.4.6.1. In Vivo Studies of the Kidney Toxicity of Trichloroethylene (TCE) Metabolites
19 4.4.6.1.1. Role of GSH conjugation metabolites of Trichloroethylene (TCE). In numerous
20 studies, DCVC has been shown to be acutely nephrotoxic in rats and mice. Mice receiving a
21 single dose of 1 mg/kg DCVC (the lowest dose tested in this species) exhibited karyolytic
22 proximal tubular cells in the outer stripe of the outer medulla, with some sloughing of cells into
23 the lumen and moderate desquamation of the tubular epithelium (Eyre et al., 1995b). Higher
24 doses in mice were associated with more severe histological changes similar to those induced by
25 TCE, such as desquamation and necrosis of the tubular epithelium (Darnerud et al., 1989;
26 Terracini and Parker, 1965a; Vaidya et al., 2003a, b). In rats, no histological changes in the
27 kidney were reported after single doses of 1, 5, and 10 mg/kg DCVC (Eyre et al., 1995a; Green
28 et al., 1997a), but cellular debris in the tubular lumen was reported at 25 mg/kg (Eyre et al.,
29 1995b) and slight degeneration and necrosis were seen at 50 mg/kg (Green et al., 1997). Green
30 et al. (1997) reported no histological changes were noted in rats after 10 doses of 0.1-5.0 mg/kg
31 DCVC (although increases in urinary protein and GGT were found), but some karyomegaly was
32 noted in mice after 10 daily doses of 1 mg/kg. Therefore, mice appear more sensitive than rats to
33 the nephrotoxic effects of acute exposure to DCVC, although the number of animals used at each
34 dose in these studies was limited (10 or less). Although the data are not sufficient to assess the
35 relatively sensitivity of other species, it is clear that multiple species, including rabbits, guinea
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-188 DRAFT—DO NOT CITE OR QUOTE
-------
1 pigs, cats, and dogs, are responsive to DCVC's acute nephrotoxic effects (Jaffe et al., 1984;
2 Krejci et al., 1991; Terracini and Parker, 1965b; Wolfgang et al., 1989b).
3 Very few studies are available at longer durations. Terracini and Parker (1965) gave
4 DCVC in drinking water to rats at a concentration of 0.01% for 12 weeks (approximately
5 10 mg/kg/d), and reported consistent pathological and histological changes in the kidney. The
6 progression of these effects was as follows: (1) during the first few days, completely necrotic
7 tubules, with isolated pyknotic cells being shed into the lumen; (2) after 1 week, dilated tubules
8 in the inner part of the cortex, lined with flat epithelial cells that showed thick basal membranes,
9 some with big hyperchromatic nuclei; (3) in the following weeks, increased prominence of
10 tubular cells exhibiting karyomegaly, seen in almost all animals, less pronounced tubular
11 dilation, and cytomegaly in the same cells showing karyomegaly. In addition, increased mitotic
12 activity was reported the first few days, but was not evident for the rest of the experiment.
13 Terracini and Parker (1965) also reported the results of a small experiment (13 male and
14 5 female rats) given the same concentration of DCVC in drinking water for 46 weeks, and
15 observed for 87 weeks. They noted renal tubular cells exhibiting karyomegaly and cytomegaly
16 consistently throughout the experiment. Moreover, a further group of 8 female rats given DCVC
17 in drinking water at a concentration of 0.001% (approximately 1 mg/kg/d) also exhibited similar,
18 though less severe, changes in the renal tubules. In mice, Jaffe et al. (1984) gave DCVC in
19 drinking water at concentrations of 0.001, 0.005, and 0.01% (estimated daily dose of 1-2, 7-13,
20 and 17-22 mg/kg/d), and reported similar effects in all dose groups, including cytomegaly,
21 nuclear hyperchromatism, and multiple nucleoli, particularly in the pars recta section of the
22 kidney. Thus, effects were noted in both mice and rats under chronic exposures at doses as low
23 as 1-2 mg/kg/d (the lowest dose tested). Therefore, while limited, the available data do not
24 suggest differences between mice and rats to the nephrotoxic effects of DCVC under chronic
25 exposure conditions, in contrast to the greater sensitivity of mice to acute and subchronic DCVC-
26 induced nephrotoxicity.
27 Importantly, as summarized in Table 4-48, the histological changes and their location in
28 these subchronic and chronic experiments with DCVC are quite similar to those reported in
29 chronic studies of TCE, described above, particularly the prominence of karyomegaly and
30 cytomegaly in the pars recta section of the kidney. Moreover, the morphological changes in the
31 tubular cells, such as flattening and dilation, are quite similar. Similar pathology is not observed
32 with the oxidative metabolites alone (see Section 4.4.6.1.2).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-189 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
vo *
I
§
***.
S'
1
TO'
il
I
TO
Table 4-48. Summary of histological changes in renal proximal tubular cells induced by chronic exposure to
TCE, DCVC, and TCOH
Effects
Karyomegaly
Cytomegaly
Cell necrosis/
hyperplasia
Morphology/
content of
tubules
TCE
Enlarged, hyperchromatic nuclei, irregular to
oblong in shape. Vesicular nuclei containing
prominent nucleoli.
Epithelial cells were large, elongated and flattened.
Stratified epithelium that partially or completely
filled the tubular lumens. Cells in mitosis were
variable in number or absent. Cells had abundant
eosinophilic or basophilic cytoplasm.
Some tubules enlarged/dilated to the extent that
they were difficult to identify. Portions of
basement membrane had a stripped appearance.
Tubules were empty or contained "wisps of
eosinophilic material."
DCVC
Enlarged, hyperchromatic nuclei with
and multiple nucleoli. Nuclear pyknosis
and karyorrhexis.
Epithelial cells were large, elongated
and flattened cells.
Thinning of tubular epithelium, frank
tubular necrosis, re-epitheliation.
Tubular atrophy, interstitial fibrosis and
destruction of renal parenchyma. More
basophilic and finely vacuolated.
Tubular dilation, denuded tubules.
Thick basal membrane. Focal areas of
dysplasia, intraluminal casts.
TCOH
None reported.
No report of enlarged cells.
No flattening or loss of
epithelium reported. Increased
tubular cell basophilia, followed
by increased cellular
eosinophilia, tubular cell
vacuolation.
No tubular dilation reported.
Intratubular cast formation.
Sources: NCI (1976); NTP (1988, 1990); Maltoni et al. (1988); Terracini and Parker (1965); Jaffe et al. (1985); Green et al. (2003).
O
Oq
O
H
W
-------
1 Additionally, it is important to consider whether sufficient DCVC may be formed from
2 TCE exposure to account for TCE nephrotoxicity. While direct pharmacokinetic measurements,
3 such as the excretion of NAcDCVC, have been used to argue that insufficient DCVC would be
4 formed to be the active moiety for nephrotoxicity (Green et al., 1997), as discussed in Chapter 3,
5 urinary NAcDCVC is a poor marker of the flux through the GSH conjugation pathway because
6 of the many other possible fates of metabolites in that pathway. In another approach, Eyre et al.
7 (1995b) using acid-labile adducts as a common internal dosimeter between TCE and DCVC, and
8 reported that a single TCE dose of 400 mg/kg in rats (similar to the lowest daily doses in the NCI
9 and NTP rat bioassays) and 1,000 mg/kg (similar to the lowest daily doses in the NCI and NTP
10 mouse bioassays) corresponded to a single equivalent DCVC dose of 6 and 1 mg/kg/d in rats and
11 mice, respectively. These equivalent doses of DCVC are greater or equal to those in which
12 nephrotoxicity has been reported in these species under chronic conditions. Therefore, assuming
13 that this dose correspondence is accurate under chronic conditions, sufficient DCVC would be
14 formed from TCE exposure to explain the observed histological changes in the renal tubules.
+/-
15 The Eker rat model (Tsc-2 ) is at increased risk for the development of spontaneous
16 renal cell carcinoma and as such has been used to understand the mechanisms of renal
17 carcinogenesis (Stemmer et al., 2007; Wolf et al., 2000). One study has demonstrated similar
18 pathway activation in Eker rats as that seen in humans with VHL mutations leading to renal cell
19 carcinoma, suggesting Tsc-2 inactivation is analogous to inactivation of VHL in human renal cell
20 carcinoma (Liu et al., 2003). Although the Eker rat model is a useful tool for analyzing
21 progression of renal carcinogenesis, it has some limitations in analysis of specific genetic
22 changes, particularly given the potential for different genetic changes depending on type of
23 exposure and tumor. The results of short-term assays to genotoxic carcinogens in the Eker rat
24 model (Morton et al., 2002; Stemmer et al., 2007) reported limited preneoplastic and neoplastic
25 lesions which may be related to the increased background rate of renal carcinomas in this animal
26 model.
27 Recently, Mally et al. (2006) exposed male rats carrying the Eker mutation to TCE
28 (0-1,000 mg/kg BW) by corn oil gavage and demonstrated no increase in renal preneoplastic
29 lesions or tumors. Primary Eker rat kidney cells exposed to DCVC in this study did induce an
30 increase in transformants in vitro but no DCVC-induced vhl or Tsc-2 mutations were observed.
31 In vivo exposure to TCE (5 days/week for 13 weeks), decreased body weight gain and increased
32 urinary excretion at the two highest TCE concentrations analyzed (500 and 1,000 mg/kg BW)
33 but did not change standard nephrotoxicity markers (GGT, creatinine and urinary protein).
34 Renal tubular epithelial cellular proliferation as measured by BrdU incorporation was
35 demonstrated at the three highest concentrations of TCE (250, 500 and 1,000 mg/kg/d). A
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-191 DRAFT—DO NOT CITE OR QUOTE
-------
1 minority of these cells also showed karyomegaly at the two higher TCE concentrations.
2 Although renal cortical tumors were demonstrated in all TCE exposed groups, these were not
3 significantly different from controls (13 weeks). These studies were complemented with in vitro
4 studies of DCVC (10-50 uM) in rat kidney epithelial (RKE) cells examining proliferation at 8,
5 24, and 72 hours and cellular transformation at 6-7 weeks. Treatment of RKE cells from
6 susceptible rats with DCVC gave rise to morphologically transformed colonies consistently
7 higher than background (Mally et al., 2006). Analyzing ten of the renal tumors from the TCE
8 exposed rats and nine of the DCVC transformants from these studies for alterations to the VHL
9 gene that might lead to inactivation found no alterations to VHL gene expression or mutations.
10 One paper has linked the VHL gene to chemical-induced carcinogenesis. Shiao et al.
11 (1998) demonstrated VHL gene somatic mutations in 7V-nitrosodimethylamine-induced rat kidney
12 cancers that were of the clear cell type. The clear cell phenotype is rare in rat kidney cancers,
13 but it was only the clear cell cancers that showed VHL somatic mutation (three of eight tumors
14 analyzed). This provided an additional link between VHL inactivation and clear cell kidney
15 cancer. However, this study examined archived formalin fixed paraffin embedded tissues from
16 previous experiments. As described previously (see Section 4.4.2), DNA extraction from this
17 type of preparation creates some technical issues. Similarly, archived formalin-fixed paraffin
18 embedded tissues from rats exposed to potassium bromide were analyzed in a later study by
19 Shiao et al. (2002). This later study examined the VHL gene mutations following exposure to
20 potassium bromide, a rat renal carcinogen known to induce clear cell renal tumors. Clear cell
21 renal tumors are the most common form of human renal epithelial neoplasms, but are extremely
22 rare in animals. Although F344 rats exposed to potassium bromide in this study did develop
23 renal clear cell carcinomas, only two of nine carried the same C to T mutation at the core region
24 of the Spl transcription-factor binding motif in the VHL promoter region, and one of four
25 untreated animals had a C to T mutation outside the conserved core region. Mutation in the VHL
26 coding region was only detected in one tumor, so although the tumors developed following
27 exposure to potassium bromide were morphologically similar to those found in humans; no
28 similarities were found in the genetic changes.
29 Elfarra et al. (1984) found that both DCVG and DCVC administered to male F344 rats by
30 intraperitoneal injections in isotonic saline resulted in elevations in BUN and urinary glucose
31 excretion. Furthermore, inhibition of renal GGT activity with acivicin protected rats from
32 DCVG-induced nephrotoxicity. In addition, both the B-lyase inhibitor AOAA and the renal
33 organic anion transport inhibitor probenecid provided protection from DCVC, demonstrating a
34 requirement for metabolism of DCVG to the cysteine conjugate by the action of renal GGT and
35 dipeptidase, uptake into the renal cell by the organic anion transporter, and subsequent activation
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-192 DRAFT—DO NOT CITE OR QUOTE
-------
1 by the B-lyase. This conclusion was supported further by showing that the -methyl analog of
2 DCVC, which cannot undergo a B-elimination reaction due to the presence of the methyl group,
3 was not nephrotoxic.
4 Korrapati et al. (2005) builds upon a series of investigations of hetero- (by HgCb) and
5 homo-(by DCVC, 15 mg/kg) protection against a lethal dose of DCVC (75 mg/kg). Priming, or
6 preconditioning, with pre-exposure to either HgCb or DCVC of male Swiss-Webster mice was
7 said to augment and sustain cell division and tissue repair, hence protecting against the
8 subsequent lethal DCVC dose(Vaidya et al., 2003a, b, c). Korrapati et al. (2005) showed that a
9 lethal dose of DCVC downregulates phosphorylation of endogenous retinoblastoma protein
10 (pRb), which is considered critical in renal proximal tubular and mesangial cells for the passage
11 of cells from Gl to S-phase, thereby leading to a block of renal tubule repair. Priming, in
12 contrast, upregulated P-pRB which was sustained even after the administration of a lethal dose of
13 DCVC, thereby stimulating S-phase DNA synthesis, which was concluded to result in tissue
14 repair and recovery from acute renal failure and death. These studies are more informative about
15 the mechanism of autoprotection than on the mechanism of initial injury caused by DCVC. In
16 addition, the priming injury (not innocuous, as it caused 25-50% necrosis and elevated blood
17 urea nitrogen) may have influenced the toxicokinetics of the second DCVC injection.
18
19 4.4.6.1.2. Role of oxidative metabolites of Trichloroethylene (TCE). Some investigators
20 (Green et al., 1998, 2003; Dow and Green, 2000) have proposed that TCE nephrotoxicity is
21 related to formic acid formation. They demonstrated that exposure to either trichloroethanol or
22 trichloroacetic acid causes increased formation and urinary excretion of formic acid (Green et al.,
23 1998). The formic acid does not come from trichloroethylene. Rather, trichloroethylene (or a
24 metabolite) has been proposed to cause a functional depletion of vitamin 812, which is required
25 for the methionine salvage pathway of folate metabolism. Vitamin Bi2 depletion results in folate
26 depletion. Folate is a cofactor in one-carbon metabolism and depletion of folate allows formic
27 acid to accumulate, and then to be excreted in the urine (Dow and Green, 2000).
28 TCE (1 and 5 g/L), TCA (0.25, 0.5 and 1 g/L) and TCOH (0.5 and 1.0 g/L) exposure in
29 male Fisher rats substantially increased excretion of formic acid in urine, an effect suggested as a
30 possible explanation for TCE-induced renal toxicity in rats (Green et al., 1998a). Green et al.
31 (2003a) reported tubular toxicity as a result of chronic (1 year) exposure to TCOH (0, 0.5, and
32 1.0 g/L). Although TCOH causes tubular degeneration in a similar region of the kidney as TCE,
33 there are several dissimilarities between the characteristics of nephrotoxicity between the two
34 compounds, as summarized in Table 4-48. In particular, Green et al. (1998) did not observe
35 TCOH causing karyomegaly and cytomegaly. These effects were seen as early as 13 weeks after
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-193 DRAFT—DO NOT CITE OR QUOTE
-------
1 the commencement of TCE exposure (NTP, 1990), with 300 ppm inhalation exposures to TCE
2 (Maltoni et al., 1988), as well as at very low chronic exposures to DCVC (Terracini and Parker,
3 1965; Jaffe et al., 1984). In addition, Green et al. (2003) reported neither flattening nor loss of
4 the tubular epithelium nor hyperplasia, but suggested that the increased early basophilia was due
5 to newly divided cells, and therefore, represented tubular regeneration in response to damage.
6 Furthermore, they noted that such changes were seen with the spontaneous damage that occurs in
7 aging rats. However, several of the chronic studies of TCE noted that the TCE-induced damage
8 observed was distinct from the spontaneous nephropathy observed in rats. A recent in vitro
9 study of rat hepatocytes and primary human renal proximal tubule cells from two donors
10 measured formic acid production following exposure to CH (0.3-3 mM, 3-10 days) (Lock et al.,
11 2007). This study observed increased formic acid production at day 10 in both human renal
12 proximal tubule cell strains, but a similar level of formic acid was measured when CH was added
13 to media alone. The results of this study are limited by the use of only two primary human cell
14 strains, but suggest exposure to CH does not lead to significant increases in formic acid
15 production in vivo.
16 Interestingly, it appears that the amount of formic acid excreted reaches a plateau at a
17 relatively low dose. Green et al. (2003) added folic acid to the drinking water of the group of
18 rats receiving the lower dose of TCOH (18.3 mg/kg/d) in order to modulate the excretion of
19 formic acid in that dose group, and retain the dose-response in formic acid excretion relative to
20 the higher-dose group (54.3 mg/kg/d). These doses of TCOH are much lower than what would
21 be expected to be formed in vivo at chronic gavage doses. For instance, after a single 500-mg/kg
22 dose of TCE (the lower daily dose in the NTP rat chronic bioassays), Green and Prout (1985)
23 reported excretion of about 41% of the TCE gavage dose in urine as TCOH or trichloroethanol-
24 glucuronide conjugate (TCOG) in 24 hours. Thus, using the measure of additional excretion
25 after 24 hours and the TCOH converted to TC A as a lower bound as to the amount of TCOH
26 formed by a single 500 mg/kg dose of TCE, the amount of TCOH would be about 205 mg/kg,
27 almost 4-fold greater than the high dose in the Green et al. (2003) study. By contrast, these
28 TCOH doses are somewhat smaller than those expected from the inhalation exposures of TCE.
29 For instance, after 6 hour exposure to 100 and 500 ppm TCE (similar to the daily inhalation
30 exposures in Maltoni et al., 1988), male rats excreted 1.5 and 4.4 mg of TCOH over 48 hours,
31 corresponding to 5 and 15 mg/kg for a rat weighing 0.3 kg (Kaneko et al., 1994). The higher
32 equivalent TCOH dose is similar to the lower TCOH dose used in Green et al. (2003), so it is
33 notable that while Maltoni et al. (1988) reported a substantial incidence of cytomegaly and
34 karyomegaly after TCE exposure (300 and 600 ppm), none was reported in Green et al. (2003).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-194 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCOH alone does not appear sufficient to explain the range of renal effects observed
2 after TCE exposure, particularly cytomegaly, karyomegaly, and flattening and dilation of the
3 tubular epithelium. However, given the studies described above, it is reasonable to conclude that
4 TCOH may contribute to the nephrotoxicity of TCE, possibly due to excess formic acid
5 production, because (1) there are some similarities between the effects observed with TCE and
6 TCOH and (2) the dose at which effects with TCOH are observed overlap with the approximate
7 equivalent TCOH dose from TCE exposure in the chronic studies.
8 Dow and Green (2000) noted that TCA also induced formic acid accumulation in rats,
9 and suggested that TCA may therefore, contribute to TCE-induced nephrotoxicity. However,
10 TCA has not been reported to cause any similar histologic changes in the kidney. Mather et al.
11 (1990) reported an increase of kidney-weight to body-weight ratio in rats after 90 days of
12 exposure to trichloroacetic acid in drinking water at 5,000 ppm (5 g/L) but reported no
13 histopathologic changes in the kidney. DeAngelo et al. (1997) reported no effects of
14 trichloroacetic acid on kidney weight or histopathology in rats in a 2-year cancer bioassay.
15 Dow and Green (2000) administered TCA at quite high doses (1 and 5 g/L in drinking water),
16 greater than the subsequent experiments of Green et al. (2003) with TCOH (0.5 and 1 g/L in
17 drinking water), and reported similar amounts of formic acid produced (about 20 mg/day for
18 each compound). However, cytotoxicity or karyomegaly did not appear to be analyzed.
19 Furthermore, much more TCOH is formed from TCE exposure than TCA. Therefore, if TCA
20 contributes substantially to the nephrotoxicity of TCE, its contribution would be substantially
21 less than that of TCOH. Lock et al. (2007) also measured formic acid production in human renal
22 proximal tubule cells exposed to 0.3-3 mM CH for 10 days CH. This study measured
23 metabolism of CH to TCOH and TCA as well as formic acid production and subsequent
24 cytotoxicity. Increased formic acid was not observed in this study, and limited cytotoxicity was
25 observed. However, this study was performed in human renal proximal tubular cells from only
26 two donors, and there is potential for large interindividual variability in response, particularly
27 with CYP enzymes.
28 In order to determine the ability of various chlorinated hydrocarbons to induce
29 peroxisomal enzymes, Goldsworthy and Popp (1987) exposed male Fisher-344 rats and male
30 B6C3F1 mice to TCE (1,000 mg/kg BW) and TCA (500 mg/kg BW) by corn oil gavage for
31 10 consecutive days. Peroxisomal activation was measured by palmitoyl CoA oxidase activity
32 levels. TCE led to increased peroxisomal activation in the kidneys of both rats (300% of control)
33 and mice (625% of control), while TCA led to an increase only in mice (280% of control). A
34 study by Zanelli et al. (1996) exposed Sprague-Dawley rats to TCA for 4 days and measured
35 both renal and hepatic peroxisomal and cytochrome P450 enzyme activities. TCA-treated rats
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-195 DRAFT—DO NOT CITE OR QUOTE
-------
1 had increased activity in CYP 4A subfamily enzymes and peroxisomal palmitoyl-CoA oxidase.
2 Both of these acute studies focused on enzyme activities and did not further analyze resulting
3 histopathology.
4
5 4.4.6.2. In Vitro Studies of Kidney Toxicity of Trichloroethylene (TCE) and Metabolites
6 Generally, it is believed that TCE metabolites are responsible for the bulk of kidney
7 toxicity observed following exposure. In particular, studies have demonstrated a role for DCVG
8 and DCVC in kidney toxicity. The work by Lash and colleagues (Cummings et al., 2000a, b;
9 Cummings and Lash, 2000; Lash et al., 2000a) examined the effect of trichloroethylene and its
10 metabolites in vitro. Trichloroethylene and DCVC are toxic to primary cultures of rat proximal
11 and distal tubular cells (Cummings et al., 2000b) while the TCE metabolites DCVG and DCVC
12 have been demonstrated to be cytotoxic to rat and rabbit kidney cells in vitro (Groves et al.,
13 1993; Hassall et al., 1983; Lash et al., 2000a, 2001; Wolfgang et al., 1989a). Glutathione-related
14 enzyme activities were well maintained in the cells, whereas CYP activities were not. The
15 enzyme activity response to DCVC was greater than the response to trichloroethylene; however,
16 the proximal and distal tubule cells had similar responses even though the proximal tubule is the
17 target in vivo. The authors attributed this to the fact that the proximal tubule is exposed before
18 the distal tubule in vivo and to possible differences in uptake transporters. They did not address
19 the extent to which transporters were maintained in the cultured cells.
20 In further studies, Lash et al. (2001) assessed the toxicity of trichloroethylene and its
21 metabolites DCVC and DCVG using in vitro techniques (Lash et al., 2001) as compared to in
22 vivo studies. Experiments using isolated cells were performed only with tissues from
23 Fischer 344 rats, and lactate dehydrogenase release was used as the measure of cellular toxicity.
24 The effects were greater in males. DCVC and trichloroethylene had similar effects, but DCVG
25 exhibited increased efficacy compared with trichloroethylene and DCVC.
26 In vitro mitochondrial toxicity was assessed in renal cells from both Fischer 344 rats and
27 B6C3F1 mice following exposure to both DCVC and DCVG (Lash et al., 2001). Renal
28 mitochondria from male rats and mice responded similarly; a greater effect was seen in cells
29 from the female mice. These studies show DCVC to be slightly more toxic than
30 trichloroethylene and DCVG, but species differences are not consistent with the effects observed
31 in long-term bioassays. This suggests that in vitro data be used with caution in risk assessment,
32 being mindful that in vitro experiments do not account for in vivo pharmacokinetic and metabolic
33 processes.
34 In LLC-PK1 cells, DCVC causes loss of mitochondrial membrane potential,
35 mitochondrial swelling, release of cytochrome c, caspase activation, and apoptosis (Chen et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-196 DRAFT—DO NOT CITE OR QUOTE
-------
1 2001). Thus, DCVC is toxic to mitochondria, resulting in either apoptosis or necrosis. DCVC-
2 induced apoptosis also has been reported in primary cultures of human proximal tubule cells
3 (Lash etal., 2001).
4 DCVC was further studied in human renal proximal tubule cells for alterations in gene
5 expression patterns related to proposed modes of action in nephrotoxicity (Lock et al., 2006). In
6 cells exposed to subtoxic levels of DCVC to better mimic workplace exposures, the expression
7 of genes involved with apoptosis (caspase 8, FADD-like regulator) was increased at the higher
8 dose (1 uM) but not at the lower dose (0.1 uM) of DCVC exposure. Genes related to oxidative
9 stress response (SOD, NFkB, p53, c-Jun) were altered at both subtoxic doses, with genes
10 generally upregulated at 0.1 uM DCVC being downregulated at 1 uM DCVC. The results of this
11 study support the need for further study, and highlight the involvement of multiple pathways and
12 variability of response based on different concentrations.
13 Lash et al. (2007) examined the effect of modulation of renal metabolism on toxicity of
14 TCE in isolated rat cells and microsomes from kidney and liver. Following exposure to
15 modulating chemicals, lactate dehydrogenase (LDH) was measured as a marker of cytotoxicity,
16 and the presence of specific metabolites was documented (DCVG, TCA, TCOH, and CH).
17 Inhibition of the CYP stimulated an increase of GSH conjugation of TCE and increased
18 cytotoxicity in kidney cells. This modulation of CYP had a greater effect on TCE-induced
19 cytotoxicity in liver cells than in kidney cells. Increases in GSH concentrations in the kidney
20 cells led to increased cytotoxicity following exposure to TCE. Depletion of GSH in hepatocytes
21 exposed to TCE, however, led to an increase in hepatic cytotoxicity. The results of this study
22 highlight the role of different bioactivation pathways needed in both the kidney and the liver,
23 with the kidney effects being more affected by the GSH conjugation pathways metabolic
24 products.
25 In addition to the higher susceptibility of male rats to TCE-induced
26 nephrocarcinogenicity and nephrotoxicity, isolated renal cortical cells from male F344 rats are
27 more susceptible to acute cytotoxicity from TCE than cells from female rats. TCE caused a
28 modest increase in LDH release from male rat kidney cells but had no significant effect on LDH
29 release from female rat kidney cells. These results on male susceptibility to TCE agree with the
30 in vivo data.
31
32 4.4.6.3. Conclusions as to the Active Agents of Trichloroethylene (TCE)-Induced
33 Nephrotoxicity
34 In summary, the TCE metabolites DCVC, TCOH, and TCA have all been proposed as
35 possible contributors to the nephrotoxicity of TCE. Both in vivo and in vitro data strongly
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-197 DRAFT—DO NOT CITE OR QUOTE
-------
1 support the conclusion that DCVC and related GSH conjugation metabolites are the active agents
2 of TCE-induced nephrotoxicity. Of these, DCVC induces effects in renal tissues, both in vivo
3 and in vitro, that are most similar to those of TCE, and formed in sufficient amounts after TCE
4 exposure to account for those effects. A role for formic acid due to TCOH or TCA formation
5 from TCE cannot be ruled out, as it is known that substantial TCOH and TCA are formed from
6 TCE exposure, that formic acid is produced from all three compounds, and that TCOH exposure
7 leads to toxicity in the renal tubules. However, the characteristics of TCOH-induced
8 nephrotoxicity do not account for the range of effects observed after TCE exposure while those
9 of DCVC-induced nephrotoxicity do. Also, TCOH does not induce the same pathology as TCE
10 or DCVC. TCA has also been demonstrated to induce peroxisomal proliferation in the kidney
11 (Goldsworthy and Popp, 1987), but this has not been associated with kidney cancer. Therefore,
12 although TCOH and possibly TCA may contribute to TCE-induced nephrotoxicity, their
13 contribution is likely to be small compared to that of DCVC.
14
15 4.4.7. Mode(s) of Action for Kidney Carcinogenicity
16 This section will discuss the evidentiary support for several hypothesized modes of action
17 for kidney carcinogen!city, including mutagenicity, cytotoxicity and regenerative proliferation,
18 peroxisome proliferation, a2|i-related nephropathy and formic acid-related nephropathy,
19 following the framework outlined in the Cancer Guidelines (U.S. EPA, 2005a, b).3
20
21 4.4.7.1. Hypothesized Mode of Action: Mutagenicity
22 One hypothesis is that TCE acts by a mutagenic mode of action in TCE-induced renal
23 carcinogenesis. According to this hypothesis, the key event leading to TCE-induced kidney
24 tumor formation constitute the following: TCE GSH conjugation metabolites (e.g., DCVG,
25 DCVC, NAcDCVC, and/or other reactive metabolites derived from subsequent beta-lyase, flavin
26 monooxygenases [FMO], or CYP metabolism) derived from the GSH-conjugation pathway, after
27 being either produced in situ in or delivered systemically to the kidney, cause direct alterations to
3 As recently reviewed (Guyton et al, 2008) the approach to evaluating mode of action information described in
U.S. EPA's Cancer Guidelines (2005a, b) considers the issue of human relevance of a hypothesized mode of action
in the context of hazard evaluation. This excludes, for example, consideration of toxicokinetic differences across
species; specifically, the Cancer Guidelines state, "the toxicokinetic processes that lead to formation or distribution
of the active agent to the target tissue are considered in estimating dose but are not part of the mode of action." In
addition, information suggesting quantitative differences in the occurrence of a key event between test species and
humans are noted for consideration in the dose-response assessment, but is not considered in human relevance
determination. In keeping with these principles, a formal analysis of the dose-response of key events in the
hypothesized modes of action is not presented unless it would aid in the overall weight of evidence analysis for
carcinogenicity, as presented in Section 4.11.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-198 DRAFT—DO NOT CITE OR QUOTE
-------
1 DNA (e.g., mutation, DNA damage, and/or micronuclei induction). Mutagenicity is a well-
2 established cause of carcinogenicity.
3
4 Experimental Support for the Hypothesized Mode of Action. Evidence for the hypothesized
5 mode of action for TCE includes (1) the formation of GSH-conjugation pathway metabolites in
6 the kidney demonstrated in TCE toxicokinetics studies; and (2) the genotoxicity of these GSH-
7 conjugation pathway metabolites demonstrated in most existing in vitro and in vivo assays of
8 gene mutations (i.e., Ames test) and in assays of unscheduled DNA synthesis, DNA strand
9 breaks, and micronuclei using both "standard" systems and renal cells/tissues.4 Additional
10 relevant data come from analyses of VHL mutations in human kidney tumors and studies using
11 the Eker rat model. These lines of evidence are elaborated below.
12 Toxicokinetic data are consistent with these genotoxic metabolites either being delivered
13 to or produced in the kidney. As discussed in Chapter 3, following in vivo exposure to TCE, the
14 metabolites DCVG, DCVC, and NAcDCVC have all been detected in the blood, kidney, or urine
15 of rats, and DCVG in blood and NAcDCVC in urine have been detected in humans (Birner et al.,
16 1993; Bernauer et al., 1996; Lash et al., 1999a, 2006). In addition, in vitro data have shown
17 DCVG formation from TCE in cellular and subcellular fractions from the liver, from which it
18 would be delivered to the kidney via systemic circulation, and from the kidney (see
19 Tables 3-23-3-24, and references therein). Furthermore, in vitro data in both humans and
20 rodents support the conclusion that DCVC is primarily formed from DCVG in the kidney itself,
21 with subsequent in situ transformation to NAcDCVC by 7V-Acetyl transferase or to reactive
22 metabolites by beta-lyase, FMO, or CYPs (see Sections 3.3.3.2.2-3.3.3.2.5). Therefore, it is
23 highly likely that both human and rodent kidneys are exposed to these TCE metabolites. .
24
25
26
27
28
4 The U.S. EPA Cancer Guidelines (2005a ,b) note reliance on "evaluation of in vivo or in vitro short-term testing
results for genetic endpoints" and evidence that "the carcinogen or a metabolite is DNA-reactive and/or has the
ability to bind to DNA"as part of this weight of evidence supporting a mutagenic mode of action. While evidence
from hypothesis-testing experiments that mutation is an early step in the carcinogenic process is considered if
available, it is not required for determination of a mutagenic mode of action; rather, reliance on short-term
genotoxicity tests is emphasized. Thus, such tests are the focus of this analysis, which also includes an analysis of
other available data from humans and animals. In keeping with these principles, a formal analysis of the temporal
concordance of key events in the hypothesized modes of action is not presented unless it would aid in the overall
weight of evidence analysis for carcinogenicity, as presented in Section 4.11.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-199 DRAFT—DO NOT CITE OR QUOTE
-------
1 As discussed in Section 4.2.1.4.2, DCVG, DCVC, and NAcDCVC have been
2 demonstrated to be genotoxic in most available in vitro assays.5 In particular, DCVC was
3 mutagenic in the Ames test in three of the tested strains of S. typhimurium (TA100, TA2638,
4 TA98) (Dekant et al., 1986; Vamvakas et al., 1988a), and caused dose-dependent increases in
5 unscheduled DNA synthesis in the two available assays: porcine kidney tubular epithelial cell
6 line (Vamvakas et al., 1996) and Syrian hamster embryo fibroblasts (Vamvakas et al., 1988b).
7 DCVC has also been shown to induce DNA strand breaks in both available studies (Jaffe et al.,
8 1985; Robbiano et al., 2004), and induce micronucleus formation in primary kidney cells from
9 rats and humans (Robbiano et al., 2004) but not in Syrian hamster embryo fibroblasts
10 (Vamvakas et al., 1988b). Only one study each is available for DCVG and 7V-AcDCVC, but
11 notably both were positive in the Ames test (Vamvakas et al., 1988a; Vamvakas et al., 1987).
12 Although the number of test systems was limited, these results are consistent.
13 These in vitro results are further supported by studies reporting kidney-specific
14 genotoxicity after in vivo administration of TCE or DCVC. In particular, Robbiano et al. (1998)
15 reported increased numbers of micronucleated cells in the rat kidney following oral TCE
16 exposure. Oral exposure to DCVC in both rabbits (Jaffe et al., 1985) and rats (Clay, 2008)
17 increased DNA strand breaks in the kidney. However, in one inhalation exposure study in rats,
18 TCE did not increase DNA breakage in the rat kidney, possibly due to study limitations (limited
19 exposure time [6 hours/day for only 5d] and small number of animals exposed [n = 5] [Clay,
20 2008]). One study of TCE exposure in the Eker rat, a rat model heterozygous for the tumor
21 suppressor gene Tsc-2, reported no significant increase in kidney tumors as compared to controls
22 (Mally et al., 2006). Inactivation of Tsc-2 in this rat model is associated with spontaneous renal
23 cell carcinoma with activation of pathways similar to that of VHL inactivation in humans
24 (Liu et al., 2003). TCE exposure for 13-weeks (corn oil gavage) led to increased nephrotoxicity
25 but no significant increases in preneoplastic or neoplastic lesions as compared to controls
26 (Mally et al., 2006). This lack of increased incidence of neoplastic or preneoplastic lesions
27 reported by Mally et al. (2006) in the tumor-prone Eker rat is similar to lack of significant short-
5 Evaluation of genotoxicity data entails a weight of evidence approach that includes consideration of the various
types of genetic damage that can occur. In acknowledging that genotoxicity tests are by design complementary
evaluations of different mechanisms of genotoxicity, a recent IPCS publication (Eastmond et al., 2009) notes that
"multiple negative eresults may not be sufficient to remove concern for mutagenicity raised by a clear positive result
in a single mutagenicity assay.". These considerations inform the present approach. In addition, consistent with
U.S. EPA's Cancer Guidelines (2005a, b), the approach does not address relative potency (e.g., among TCE
metabolites, or of such metabolites with other known genotoxic carcinogens) per se, nor does it consider
quantitative issues related to the probable production of these metabolites in vivo. Instead, the analysis of genetic
toxicity data presented in Section 4.2 and summarized here focuses on the identification of a genotoxic hazard of
these metabolites; a quantitative analysis of TCE metabolism to reactive intermediates, via PBPK modeling, is
presented in Section 3.5.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-200 DRAFT—DO NOT CITE OR QUOTE
-------
1 term response exhibited by other genotoxic carcinogens in the Eker rat (Morton et al., 2002;
2 Stemmer et al., 2007) and may be related to the increased background rate of renal carcinomas in
3 this animal model. Mally et al. (2006) also exposed primary kidney epithelial cells from the
4 Eker rat to DCVC in vitro and demonstrated increased transformation similar to that of other
5 renal carcinogens (Horesovsky et al., 1994).
6 As discussed in Section 4.2.1.4.1, although Douglas et al. (1999) did not detect increased
7 mutations in the kidney of lacZ transgenic mice exposed to TCE for 12 days, these results are not
8 highly informative as to the role of mutagenicity in TCE-induced kidney tumors, given the
9 uncertainties in the production in genotoxic GSH conjugation metabolites in mice and the low
10 carcinogenic potency of TCE for kidney tumors in rodents relative to what is detectable in
11 experimental bioassays. Limited, mostly in vitro, toxicokinetic data do not suggest mice have
12 less GSH conjugation or subsequent renal metabolism/bioactivation (see Section 3.3.3.2.7), but
13 quantitatively, the uncertainties in the flux through these pathways remain significant (see
14 Section 3.5). In additional, similar to other genotoxic renal carcinogens analyzed by NTP, there
15 is limited evidence of mouse kidney tumors following TCE exposure. However, given the
16 already low incidences of kidney tumors observed in rats, a relatively small difference in potency
17 in mice would be undetectable in available chronic bioassays. Notably, of seven chemicals
18 categorized as direct-acting genotoxic carcinogens that induced rat renal tumors in NTP studies,
19 only two also led to renal tumors in the mouse (tris[2,3-dibromopropyl]phosphate and
20 ochratoxin A) (Reznik et al., 1979; Kanisawa and Suzuki, 1978), so the lack of detectable
21 response in mouse bioassays does not preclude a genotoxic MO A.
22 VHL inactivation (via mechanisms such as deletion, silencing or mutation) observed in
23 human renal clear cell carcinomas, is the basis of a hereditary syndrome of kidney cancer
24 predisposition, and is hypothesized to be an early and causative event in this disease (e.g.,
25 Nickerson et al., 2008). Therefore, specific actions of TCE metabolites that produce or select for
26 mutations of the VHL suppressor gene could lead to kidney tumorigenesis. Several studies have
27 compared VHL mutation frequencies in cases with TCE exposures with those from control or
28 background populations. Briining et al. (1997a) and Brauch et al. (1999, 2004) reported
29 differences between TCE-exposed and nonexposed renal cell carcinoma patients in the frequency
30 of somatic VHL mutations, the incidence of a hot spot mutation of cytosine to thymine at
31 nucleotide 454, and the incidence of multiple mutations. These data suggest that kidney tumor
32 genotype data in the form of a specific mutation pattern may potentially serve to discriminate
33 TCE-induced tumors from other types of kidney tumors in humans. If validated, this would also
34 suggest that TCE-induced kidney tumors are dissimilar from those occurring in unexposed
35 individuals. Thus, while not confirming a mutation MO A, these data suggest that TCE-induced
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-201 DRAFT—DO NOT CITE OR QUOTE
-------
1 tumors may be distinct from those induced spontaneously in humans. However, it has not been
2 examined whether a possible linkage exists between VHL loss or silencing and mutagenic TCE
3 metabolites.
4 By contrast, Schraml et al. (1999) and Charbotel et al. (2007) reported that TCE-exposed
5 renal cell carcinoma patients did not have significantly higher incidences of VHL mutations
6 compared to nonexposed patients. However, details as to the exposure conditions were lacking
7 in Schraml et al. (1999). In addition, the sample preparation methodology employed by
8 Charbotel et al. (2007) and others (Briining et al., 1997a; Brauch et al., 1999) often results in
9 poor quality and/or low quantity DNA, leading to study limitations (less than 100% of samples
10 were able to be analyzed). Therefore, further investigations are necessary to either confirm or
11 contradict the validity of the genetic biomarkers for TCE-related renal tumors reported by
12 Briining et al. (1997a) and Brauch et al. (1999, 2004).
13 In addition, while exposure to mutagens is certainly associated with cancer induction (as
14 discussed with respect to the liver in Appendix E, Sections E.3.1 and E.3.2), examination of end-
15 stage tumor phenotype or genotype has limitations concerning determination of early key events.
16 The mutations that are observed with the progression of neoplasia are associated with increased
17 genetic instability and an increase in mutation rate. Further, inactivation of the VHL gene also
18 occurs through other mechanisms in addition to point mutations, such as loss of heterozygosity
19 or hypermethylation (Kenck et al., 1996; Nickerson et al., 2008) not addressed in these studies.
20 Recent studies examining the role of other genes or pathways suggest roles for multiple genes in
21 renal cell carcinoma development (Purge et al., 2007; Toma et al., 2008). Therefore, the
22 inconsistent results with respect to VHL mutation status do not constitute negative evidence for a
23 mutational MOA and the positive studies are suggestive of a TCE-induced kidney tumor
24 genotype.
25 In sum, the predominance of positive genotoxicity data in the database of available
26 studies of TCE metabolites derived from GSH conjugation (in particular the evidence of kidney-
27 specific genotoxicity following in vivo exposure to TCE or DCVC), coupled with the
28 toxicokinetic data consistent with the in situ formation of these GSH-conjugation metabolites of
29 TCE in the kidney, is consistent with the hypothesis that a mutagenic MOA is operative in TCE-
30 induced kidney tumors. Available data on the VHL gene in humans add biological plausibility to
31 these conclusions.
32
33 4.4.7.2. Hypothesized Mode of Action: Cytotoxicity and Regenerative Proliferation
34 Another hypothesis is that TCE acts by a cytotoxicity mode of action in TCE-induced
35 renal carcinogenesis. According to this hypothesis, the key events leading to TCE-induced
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-202 DRAFT—DO NOT CITE OR QUOTE
-------
1 kidney tumor formation comprise the following: the TCE GSH-conjugation metabolite DCVC,
2 after being either produced in situ in or delivered systemically to the kidney, causes cytotoxicity,
3 leading to compensatory cellular proliferation and subsequently increased mutations and clonal
4 expansion of initiated cells.
5
6 Experimental Support for the Hypothesized Mode of Action. Evidence for the hypothesized
7 MOA consist primarily of (1) the demonstration of nephrotoxicity following TCE exposure at
8 current occupational limits in human studies and chronic TCE exposure in animal studies; (2) the
9 relatively high potential of the TCE metabolite DCVC to cause nephrotoxicity; and (3)
10 toxicokinetic data demonstrating that DCVC is formed in the kidney following TCE exposure.
11 Data on nephrotoxicity of TCE and DCVC are discussed in more detail below, while the
12 toxicokinetic data were summarized previously in the discussion of mutagenicity. However,
13 there is a lack of experimental support linking TCE nephrotoxicity and sustained cellular
14 proliferation to TCE-induced nephrocarcinogenicity.
15 There is substantial evidence that TCE is nephrotoxic in humans and laboratory animals
16 and that its metabolite DCVC is nephrotoxic in laboratory animals. Epidemiological studies
17 have consistently demonstrated increased excretion of nephrotoxicity markers (NAG, protein,
18 albumin) at occupational (Green et al., 2004) and higher (Bolt et al., 2004; Briining et al.,
19 1999a, b) levels of TCE exposure. However, direct evidence of tubular toxicity, particularly in
20 renal cell carcinoma cases, is not available. These studies are supported by the results of
21 multiple laboratory animal studies. Chronic bioassays have reported very high (nearly 100%)
22 incidences of nephrotoxicity of the proximal tubule in rats (NTP, 1988, 1990) and mice (NCI,
23 1976; NTP, 1990) at the highest doses tested. In vivo studies examining the effect of TCE
24 exposure on nephrotoxicity showed increased proximal tubule damage following intraperitoneal
25 injection and inhalation of TCE in rats (Chakrabarty and Tuchweber, 1988) and intraperitoneal
26 injection in mice (Cojocel et al., 1989). Studies examining DCVC exposure in rats
27 (Terracini and Parker, 1965; Elfarra et al., 1986) and mice (Jaffe et al., 1984; Darnerud et al.,
28 1989) have also shown increases in kidney toxicity. The greater potency for kidney cytotoxicity
29 for DCVC compared to TCE was shown by in vitro studies (Lash et al., 1995, 1986; Stevens et
30 al., 1986). These studies also further confirmed the higher susceptibility of male rats or mice to
31 DCVC-induced cytotoxicity. Cytokaryomegaly (an effect specific to TCE and not part of the
32 chronic progressive nephropathy or the pathology that occurs in aging rat kidneys) was observed
33 in the majority of rodent studies and may or may not progress to carcinogenesis. Finally, as
34 discussed extensively in Section 4.4.6.1, a detailed comparison of the histological changes in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-203 DRAFT—DO NOT CITE OR QUOTE
-------
1 kidney caused by TCE and its metabolites supports the conclusion that DCVC is the predominant
2 moiety responsible for TCE-induced nephrotoxicity.
3 Because it is known that not all cytotoxins are carcinogens (i.e., cytotoxicity is not a
4 specific predictor of carcinogenicity), additional experimental support is required to link
5 nephrotoxicity to carcinogenicity. Clearly, cytotoxicity occurs at doses below those causing
6 carcinogenicity, as the incidence of nephrotoxicity in chronic bioassays is an order of magnitude
7 higher than that of renal tumors. However, there are multiple mechanisms by which TCE has
8 been hypothesized to induce cytotoxicity, including oxidative stress, disturbances in calcium ion
9 homeostasis, mitochondrial dysfunction, and protein alkylation (Lash et al., 2000a). Some of
10 these effects may therefore, have ancillary consequences related to tumor induction which are
11 independent of cytotoxicity per se. Under the hypothesized MO A, cytotoxicity leads to the
12 induction of repair processes and compensatory proliferation that could lead to an increased
13 production or clonal expansion of cells previously initiated by mutations occurred spontaneously,
14 from coexposures, or from TCE or its metabolites. Data on compensatory cellular proliferation
15 and the subsequent hypothesized key events in the kidney are few, with no data from rat strains
16 used in chronic bioassays. In rats carrying the Eker mutation, Mally et al. (2006) reported
17 increased DNA synthesis as measured by BrdU incorporation in animals exposed to the high
18 dose of TCE (1,000 mg/kg/d) for 13 weeks, but there was no evidence of clonal expansion or
19 tumorigenesis in the form of increased preneoplastic or neoplastic lesions as compared to
20 controls. While chronic nephrotoxicity was reported in the same bioassays showing increased
21 kidney tumor incidences, the use of such data to inform MOA is indirect and associative.
22 Moreover, chronic animal studies with reduced (in female rats) or absent (in mice of both sexes)
23 carcinogenic response have also demonstrated cytotoxicity (NTP, 1990, NCI, 1976). Therefore,
24 in both rodent and human studies of TCE, data demonstrating a causal link between tubular
25 toxicity and the induction of kidney tumors are lacking.
26
27 4.4.7.3. Additional Hypothesized Modes of Action with Limited Evidence or Inadequate
28 Experimental Support
29 Along with metabolites derived from GSH conjugation of TCE, oxidative metabolites are
30 also present and could induce toxicity in the kidney. After TCE exposure, the oxidative
31 metabolite and peroxisome proliferator TCA is present in the kidney and excreted in the urine as
32 a biomarker of exposure. Hypotheses have also been generated regarding the roles of
33 a2|i-globulin or formic acid in nephrotoxicity induced by TCE oxidative metabolites TCA or
34 TCOH. However, the available data are limited or inadequate for supporting these hypothesized
35 MO As.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-204 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.4.7.3.1. Peroxisomeproliferation. Although not as well studied as the effects of glutathione
2 metabolites in the kidney, there is evidence that oxidative metabolites affect the kidney after
3 TCE exposure. Both TCA and DCA are peroxisome proliferator activated receptor alpha
4 (PPARa) agonists although most activity has been associated with TCA production after TCE
5 exposure. Exposure to TCE has been found to induce peroxisome proliferation not only in the
6 liver but also the kidney. Peroxisome proliferation in the kidney has been evaluated by only one
7 study of TCE (Goldsworthy and Popp, 1987), using increases in cyanide-insensitive palmitoyl-
8 CoA oxidation (PCO) activity as a marker. Increases in renal PCO activity were observed in rats
9 (3.0-fold) and mice (3.6-fold) treated with TCE at 1,000 mg/kg/d for 10 days, with smaller
10 increases in both species from TCA treatment at 500 mg/kg/d for 10 days. However, no
11 significant increases in kidney/body weight ratios were observed in either species. There was no
12 relationship between induction of renal peroxisome proliferation and renal tumors (i.e., a similar
13 extent of peroxisome proliferation-associated enzyme activity occurred in species with and
14 without TCE-induced renal tumors). However, the increased peroxisomal enzyme activities due
15 to TCE exposure are indicative of oxidative metabolites being present and affecting the kidney.
16 Such metabolites have been associated with other tumor types, especially liver, and whether
17 coexposures to oxidative metabolites and glutathi one metabolites contribute to kidney
18 tumorigenicity has not been examined.
19
20 4.4.7.3.2. a2fi-Globulin-related nephropathy. Induction of a2|i-globulin nephropathy by TCE
21 has been investigated by Goldsworthy et al. (1988), who reported that TCE did not induce
22 increases in this urinary protein, nor did it stimulate cellular proliferation in rats. In addition,
23 whereas kidney tumors associated with a2|i-globulin nephropathy are specific to the male rat, as
24 discussed above, nephrotoxicity is observed in both rats and mice and kidney tumor incidence is
25 elevated (though not always statistically significant) in both male and female rats. TCOH was
26 recently reported to cause hyaline droplet accumulation and an increase in a2|i-globulin, but
27 these levels were insufficient to account for the observed nephropathy as compared to other
28 exposures (Green et al., 2003b). Therefore, it is unlikely that a2|i-globulin nephropathy
29 contributes significantly to TCE-induced renal carcinogenesis.
30
31 4.4.7.3.3. Formic acid-related nephrotoxicity. Another MO A hypothesis proposes that TCE
32 nephrotoxicity is mediated by increased formation and urinary excretion of formic acid mediated
33 by the oxidative metabolites TCA or TCOH (Green et al., 1998, 2003; Dow and Green, 2000).
34 The subsequent hypothesized key events are the same as those for DCVC-induced cytotoxicity,
35 discussed above (see Section 4.4.7.2). As discussed extensively in Section 4.4.6.1.2, these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-205 DRAFT—DO NOT CITE OR QUOTE
-------
1 oxidative metabolites do not appear sufficient to explain the range of renal effects observed after
2 TCE exposure, particularly cytomegaly, karyomegaly, and flattening and dilation of the tubular
3 epithelium. Although TCOH and possibly TCA may contribute to the nephrotoxicity of TCE,
4 perhaps due to excess formic acid production, these metabolites do not show the same range of
5 cytotoxic effects observed following TCE exposure (see Table 4-48). Therefore, without
6 specific evidence linking the specific nephrotoxic effects caused by TCOH or TCA to
7 carcinogenesis, and in light of the substantial evidence that DCVC itself can adequately account
8 for the nephrotoxic effects of TCE, the weight of evidence supports a conclusion that
9 cytotoxicity mediated by increased formic acid production induced by oxidative metabolites
10 TCOH and possibly TCA is not responsible for the majority of the TCE-induced cytotoxicity in
11 the kidneys, and therefore, would not be the major contributor to the other hypothesized key
12 events in this MOA, such as subsequent regenerative proliferation.
13
14 4.4.7.4. Conclusions About the Hypothesized Modes of Action
15 4.4.7.4.1. 1. Is the hypothesized mode of action sufficiently supported in the test animals?
16 4.4.7.4.1.1. Mutasenicity. The predominance of positive genotoxicity data in the database of
17 available studies of TCE metabolites derived from GSH conjugation (in particular the evidence
18 of kidney-specific genotoxi city following in vivo exposure to TCE or DCVC), coupled with the
19 toxicokinetic data consistent with the in situ formation of these GSH-conjugation metabolites of
20 TCE in the kidney, supports the conclusion that a mutagenic MOA is operative in TCE-induced
21 kidney tumors.
22
23 4.4.7.4.1.2. Cytotoxicity. As reviewed above, in vivo and in vitro studies have shown a
24 consistent nephrotoxic response to TCE and its metabolites in proximal tubule cells from male
25 rats. Therefore, it has been proposed that cytotoxicity seen in this region of the kidney is a
26 precursor to carcinogenicity. However, it has not been determined whether tubular toxicity is a
27 necessary precursor of carcinogenesis, and there is a lack of experimental support for causal
28 links, such as compensatory cellular proliferation or clonal expansion of initiated cells, between
29 nephrotoxicity and kidney tumors induced by TCE. Nephrotoxicity is observed in both mice and
30 rats, in some cases with nearly 100% incidence in all dose groups, but kidney tumors are only
31 observed at low incidences in rats at the highest tested doses. Therefore, nephrotoxicity alone
32 appears to be insufficient, or at least not rate-limiting, for rodent renal carcinogenesis, since
33 maximal levels of toxicity are reached before the onset of tumors.
34
35 4.4.7.4.1.3. Additional hypotheses. The kidney is also exposed to oxidative metabolites that
36 have been shown to be carcinogenic in other target organs. TCA is excreted in kidney after its
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-206 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolism from TCE and also can cause peroxisome proliferation in the kidney, but there are
2 inadequate data to define a MOA for kidney tumor induction based on peroxisome proliferation.
3 TCE induced little or no a2|i-globulin and hyaline droplet accumulation to account for the
4 observed nephropathy, so available data do not support this hypothesized MOA. The production
5 of formic acid following exposure to TCE and its oxidative metabolites TCOH and TCA may
6 also contribute to nephrotoxicity; however, the available data indicate that TCOH and TCA are
7 minor contributors to TCE-induced nephrotoxicity, and therefore, do not support this
8 hypothesized MOA. Because these additional MOA hypotheses are either inadequately defined
9 or are not supported by the available data, they are not considered further in the conclusions
10 below.
11
12 4.4.7.4.2. 2. Is the hypothesized mode of action relevant to humans?
13 4.4.7.4.2.1. Mutagenicity. The evidence discussed above demonstrates that TCE GSH-
14 conjugation metabolites are mutagens in microbial as well as test animal species. Therefore, the
15 presumption that they would be mutagenic in humans. Available data on the VHL gene in
16 humans add biological plausibility to this hypothesis. The few available data from human
17 studies concerning the mutagenicity of TCE and its metabolites suggest consistency with this
18 MOA, but are not sufficiently conclusive to provide direct supporting evidence for a mutagenic
19 MOA. Therefore, this MOA is considered relevant to humans.
20
21 4.4.7.4.2.2. Cytotoxicity. Although data are inadequate to determine that the MOA is
22 operative, none of the available data suggest that this MOA is biologically precluded in humans.
23 Furthermore, both animal and human studies suggest that TCE causes nephrotoxicity at
24 exposures that also induce renal cancer, constituting positive evidence of the human relevance of
25 this hypothesized MOA.
26
27 4.4.7.4.3. 3. Which populations or lifestages can be particularly susceptible to the
2 8 hypothesized mode of action ?
29 4.4.7.4.3.1. Mutasenicity. The mutagenic MOA is considered relevant to all populations and
30 lifestages. According to U.S. EPA's Cancer Guidelines (U.S. EPA, 2005a) and Supplemental
31 Guidance (U.S. EPA, 2005b), there may be increased susceptibility to early-life exposures for
32 carcinogens with a mutagenic mode of action. Therefore, because the weight of evidence
33 supports a mutagenic mode of action for TCE carcinogenicity and in the absence of chemical-
34 specific data to evaluate differences in susceptibility, early-life susceptibility should be assumed
35 and the age-dependent adjustment factors (ADAFs) should be applied, in accordance with the
36 Supplemental Guidance.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-207 DRAFT—DO NOT CITE OR QUOTE
-------
1 In addition, because the MOA begins with GSH-conjugation metabolites being delivered
2 systemically or produced in situ in the kidney, toxicokinetic differences—i.e., increased
3 production or bioactivation of these metabolites—may render some individuals more susceptible
4 to this MOA. Toxicokinetic-based susceptibility is discussed further in Section 4.10.
5 In rat chronic bioassays, TCE-treated males have higher incidence of kidney tumors than
6 similarly treated females. However, the basis for this sex-difference is unknown, and whether it
7 is indicative of a sex difference in human susceptibility to TCE-induced kidney tumors is
8 likewise unknown. The epidemiologic studies generally do not show sex differences in kidney
9 cancer risk. Lacking exposure-response information, it is not known if the sex-difference in one
10 renal cell carcinoma case-control study (Dosemeci et al., 1999) may reflect exposure differences
11 or susceptibility differences.
12
13 4.4.7.4.3.2. Cytotoxicity. Populations which may be more susceptible based on the
14 toxicokinetics of the production of GSH conjugation metabolites and the sex differences
15 observed in rat chronic bioassays are the same as for a mutagenic MOA. No data are available
16 as to whether other factors may lead to different populations or lifestages being more susceptible
17 to a cytotoxic MOA for TCE-induced kidney tumors. For instance, it is not known how the
18 hypothesized key events in this MOA interact with known risk factors for human renal cell
19 carcinoma.
20 The weight of evidence sufficiently supports a mutagenic MOA for TCE in the kidney,
21 based on supporting data that GSH-metabolites are genotoxic and produced in sufficient
22 quantities in the kidney to lead to tumorigenesis. Cytotoxicity and regenerative proliferation
23 were considered as an alternate MOA, however, there are inadequate data to support a causal
24 association between Cytotoxicity and kidney tumors. Further, hypothesized MO As relating to
25 peroxisomal proliferation, a2u-globulin nephropathy and formic acid-related nephrotoxicity
26 were considered and rejected due to limited evidence and/or inadequate experimental support.
27
28 4.4.8. Summary: Trichloroethylene (TCE) Kidney Toxicity, Carcinogenicity, and Mode-
29 of-Action
30 Human studies have shown increased levels of proximal tubule damage in workers
31 exposed to high levels of TCE (NRC, 2006). These studies analyzed workers exposed to TCE
32 alone or in mixtures and reported increases in various urinary biomarkers of kidney toxicity
33 (p2-microglublin, total protein, NAG, al-microglobulin) (Nagaya et al., 1989; Selden et al.,
34 1993; Briining et al. 1999a, b; Bolt et al., 2004; Green et al., 2004; Radican et al., 2006).
35 Laboratory animal studies examining TCE exposure provide additional support, as multiple
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-208 DRAFT—DO NOT CITE OR QUOTE
-------
1 studies by both gavage and inhalation exposure show that TCE causes renal toxicity in the form
2 of cytomegaly and karyomegaly of the renal tubules in male and female rats and mice. By
3 gavage, incidences of these effects under chronic bioassay conditions approach 100%, with male
4 rats appearing to be more sensitive than either female rats or mice of either sex based on the
5 severity of effects. Under chronic inhalation exposures, only male rats exhibited these effects.
6 Further studies with TCE metabolites have demonstrated a potential role for DCVC, TCOH, and
7 TCA in TCE-induced nephrotoxicity. Of these, DCVC induces the renal effects that are most
8 like TCE, and it is formed in sufficient amounts following TCE exposure to account for these
9 effects.
10 Kidney cancer risk from TCE exposure has been studied related to TCE exposure in
11 cohort, case-control and geographical studies. These studies have examined TCE in mixed
12 exposures as well as alone. Elevated risks are observed in many of the cohort and case-control
13 studies examining kidney cancer incidence in industries or job titles with historical use of TCE
14 (see Tables 4-38 and 4-39), particularly among subjects ever exposed to TCE (Dosemeci et al.,
15 1999; Briining et al., 2003; Raaschou-Nielsen et al., 2003) or subjects with TCE surrogate for
16 high exposure (Briining et al., 2003; Raaschou-Nielsen et al., 2003; Zhao et al., 2005; Charbotel
17 et al., 2006). Although there are some controversies related to deficiencies of the
18 epidemiological studies (Vamvakas et al., 1998; Henschler et al., 1995), many of these are
19 overcome in later studies (Briining et al., 2003; Charbotel et al., 2006). A meta-analysis of the
20 overall effect of TCE exposure on kidney cancer, additionally, suggests a small, statistically
21 significant increase in risk (pooled RR = 1.25 95% CI: 1.11, 1.41) with a pooled relative risk
22 estimate in the higher exposure group of 1.53, (95% CI: 1.23, 1.91), robust in sensitivity to
23 alternatives and lacking observed statistical heterogeneity among studies meeting explicitly-
24 defined inclusion criteria.
25 In vivo laboratory animal studies to date suggest a small increase in renal tubule tumors
26 in male rats and, to a lesser extent, in female rats, with no increases seen in mice or hamsters.
27 These results are based on limited studies of both oral and inhalation routes, some of which were
28 deemed insufficient to determine carcinogenicity based on various experimental issues.
29 However, because of the rarity of kidney tumors in rodents, the repeatability of this finding
30 across strains and studies supports their biological significance despite the limitations of
31 individual studies and relatively small increases in reported tumor incidence.
32 Some but not all human studies have suggested a role for VHL mutations in TCE-induced
33 kidney cancer (Briining et al., 1997a; Brauch et al., 1999, 2004; Schraml et al., 1999; Charbotel
34 et al., 2007). Certain aspects of these studies may explain some of these discrepant results. The
35 majority of these studies have examined paraffinized tissue that may lead to technical difficulties
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-209 DRAFT—DO NOT CITE OR QUOTE
-------
1 in analysis, as paraffin extractions yield small quantities of often low-quality DNA. The
2 chemicals used in the extraction process itself may also interfere with enzymes required for
3 further analysis (PCR, sequencing). Although these studies do not clearly show mutations in all
4 TCE-exposed individuals, or in fact in all kidney tumors examined, this does not take into
5 account other possible means of VHL inactivation, including silencing or loss, and other potential
6 targets of TCE mutagenesis were not systematically examined. A recent study by Nickerson et
7 al. (2008) analyzed both somatic mutation and promoter hypermethylation of the VHL gene in
8 cc-RCC frozen tissue samples using more sensitive methods. The results of this study support
9 the hypothesis that VHL alterations are an early event in clear cell RCC carcinogenesis, but these
10 alterations may not be gene mutations. No experimental animal studies have been performed
11 examining vhl inactivation following exposure to TCE, although one in vitro study examined vhl
12 mutation status following exposure to the TCE-metabolite DCVC (Mally et al., 2006). This
13 study found no mutations following DCVC exposure, although this does not rule out a role for
14 DCVC in vhl inactivation by some other method or vhl alterations caused by other TCE
15 metabolites.
16 Although not encompassing all of the actions of TCE and its metabolites that may be
17 involved in the formation and progression of neoplasia, available evidence supports the
18 conclusion that a mutagenic MOA mediated by the TCE GSH-conjugation metabolites
19 (predominantly DCVC) is operative in TCE-induced kidney cancer. This conclusion is based on
20 substantial evidence that these metabolites are genotoxic and are delivered to or produced in the
21 kidney, including evidence of kidney-specific genotoxicity following in vivo exposure to TCE or
22 DCVC. Cytotoxicity caused by DCVC leading to compensatory cellular proliferation is also a
23 potential MOA in renal carcinogenesis, but available evidence is inadequate to conclude that this
24 MOA is operative, either together with or independent of a mutagenic MOA. The additional
25 MOA hypotheses of peroxisome proliferation, accumulation of a2u-globulin, and cytotoxicity
26 mediated by TCE-induced excess formic acid production are not supported by the available data.
27
28 4.5. LIVER TOXICITY AND CANCER
29 4.5.1. Liver Noncancer Toxicity in Humans
30 The complex of chronic liver disease is a spectrum of effects and comprises nonalcoholic
31 fatty liver disease (nonalcoholic steatohepatitis) and cirrhosis, more rare anomalies ones such as
32 autoimmune hepatitis, primary biliary cirrhosis, and primary sclerosing cholangitis, and
33 hepatocellular and cholangiocarcinoma (intrahepatic bile duct cancer) (Juran and Lazaridis,
34 2006). Chronic liver disease and cirrhosis, excluding neoplasia, is the 12th leading cause of death
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-210 DRAFT—DO NOT CITE OR QUOTE
-------
1 in the United States in 2005 with 27,530 deaths (Kung et al., 2008) with a morality rate of 9.0
2 per 100,000 (Jemal et al., 2008).
3 Eight studies reported on liver outcomes and TCE exposure and are identified in
4 Table 4-49. Three studies are suggestive of effects on liver function tests in metal degreasers
5 occupationally exposed to trichloroethylene (Nagaya et al., 1993; Rasmussen et al., 1993; Xu et
6 al., 2009). Nagaya et al. (1993) in their study of 148 degreasers in metal parts factories,
7 semiconductor factors, or other factories, observed total mean serum cholesterol concentration,
8 mean serum high density lipoprotein-cholesterol (HDL-C) concentrations to increase with
9 increasing TCE exposure, as defined by U-TTC), although a statistically significant linear trend
10 was not found. Nagaya et al. (1993) estimated subjects in the low exposure group had TCE
11 exposure to 1 ppm-, 6-ppm TCE in the moderate exposure group, and 210-ppm TCE in the high
12 exposure group. No association was noted between serum liver function tests and U-TTC, a
13 finding not surprising given individuals with a history of hepatobiliary disease were excluded
14 from this study. Nagaya et al. (1993) follows 13 workers with higher U-TTC concentrations
15 over a 2-year period; serum HDL-C and two hepatic function enzymes, GOT and aspartate
16 aminotrasferase (AST) concentrations were highest during periods of high level exposure, as
17 indicated from U-TTC concentrations. Similarly, in a study of 95 degreasers, 70 exposed to
18 trichloroethylene exposure and 25 to CFC113 (Rasmussen et al., 1993), mean serum GGT
19 concentration for subjects with the highest TCE exposure duration was above normal reference
20 values and were about 3-fold higher compared to the lowest exposure group. Rasmussen et al.
21 (1993) estimated mean urinary TCE concentration in the highest exposure group as 7.7 mg/L
22 with past exposures estimated as equivalent to 40-60 mg/L. Multivariate regression analysis
23 showed a small statistically nonsignificant association due to age and a larger effect due to
24 alcohol abuse that reduced and changed direction of a TCE exposure affect. The inclusion of
25 CFC113 exposed subjects introduces a downward bias since liver toxicity is not associated with
26 CFC113 exposure (U.S. EPA, 2008) and would underestimate any possible TCE effect. Xu et al.
27 (2009) reported symptoms and liver function tests of 21 metal degreasers with severe
28 hypersensitivity dermatitis (see last paragraph in this section for discussion of other liver effects
29 in hypersensitivity dermatitis cases). TCE concentration of agent used to clean metal parts
30 ranged from 10.2 to 63.5% with workplace ambient monitoring time-weighted-average TCE
31 concentrations of 18 to 683 mg/m3 (3 to 127 ppm). Exposure was further documented by urinary
32 TCA levels in 14 of 21 cases above the recommended occupation level of 50 mg/L. The
33 prevalence of elevated liver enzymes among these subjects was 90% (19 cases) for alanine
34 aminotrasferase, 86% (18 cases) for asparatate aminotrasferase, and 76% (16 cases) for total
35 bilirubin (Xu et al., 2009).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-211 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-49. Summary of human liver toxicity studies
Subjects
148 male metal degreasers in
metal parts, semiconductor and
other factories
95 workers (70 TCE exposed,
25 CFC113 exposed) selected
from a cohort of 240 workers at
72 factors engaged in metal
degreasing with chlorinated
solvents
21 metal degreasers with
severe hypersensitivity
dermatitis
5 healthy workers engaged in
decreasing activities in steel
industry and 5 healthy workers
from clerical section of same
company
22 workers at a factory
manufacturing small appliances
4,489 males and female
residents from 15 Superfund
site and identified from
ATSDR Trichloroethylene
Exposure Subregistry
Case reports from 8 countries
of individuals with
idiosyncratic generalized skin
disorders
Deaths in California between
1979-1981 due to cirrhosis
Effect
Serum liver function enzyme
(HDL-C, AST, and GOT)
concentrations did not
correlated with TCE
exposure assesses in a
prevalence study but did
correlate with TCE
concentration over a 2-yr
follow-up period
Increased serum GOT
concentration with increasing
cumulative exposure
High prevalence of serum
liver function enzymes above
normal levels: ALT, 19 or 21
cases; AST, 18 of 21 cases,
and T-Bili, 16 of 21 cases
Total serum bile acid
concentration increased
between pre- and
postexposure (2-d period)
Increased in several bile
acids
Liver problems diagnosed
with past year
Hepatitis in 46 to 94% of
cases; other liver effects
includes hepatomegaly and
elevated liver function
enzymes; and in rare cases,
acute liver failure
SMRof211(95%CI: 136,
287) for white male sheet
metal workers and
SMR = 174 (95% CI:
150-197) for metal workers
Exposure
U-TTC levels obtained from spot
urine sample obtained during
working hours used to assign
exposure category included the
following:
High: 209 + 99 mg/g Cr
Medium: 35+27 mg/g Cr
Low: 5+2 mg/g Cr
Note: this study does not include
an unexposed referent group
4 groups (cumulative number of
years exposed over a working life):
I: 0.6 (0-0.99)
II: 1.9 (1-2.8)
III: 4.4 (2.9-6.7)
IV: 14.4 (6.8-35.6)
TWA mean ambient TCE
concentration occupational setting
of cases, 18 mg/m3to 683 mg/m3
14 of 21 cases withU-TCE above
recommended occupational level
of 50 mg/L
8-h TWA mean personal air: 8.9 +
3.2 ppm postexposure
Regular exposure to <5 ppm TCE;
peak exposure for 2 workers to
>250 00m
Residency in community with
Superfund site identified with TCE
and other chemicals
If reported, TCE, from <50 mg/m3
to more than 4,000 mg/m3.
Symptoms developed within 2-5
wks of initial exposure, with some
intervals up to 3 months
Occupational title on death
certificate
Reference
Nagaya et al.,
1993
Rasmussen et
al., 1993
Xu et al.,
2009
Neghab et
al., 1997
Driscoll et
al., 1992
Davis et al.,
2006
Kamijima et
al., 2007
Leigh and
Jiang, 1993
3
4
ALT = alanine aminotrasferase.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-212 DRAFT—DO NOT CITE OR QUOTE
-------
1 Two studies provide evidence of plasma or serum bile acids changes among TCE-
2 exposed degreasers. Neghab et al. (1997) in a small prevalence study of 10 healthy workers
3 (5 unexposed controls and 5 exposed) observed statistically significantly elevated total serum
4 bile acids, particularly deoxycholic acid and the subtotal of free bile acids, among TCE subjects
5 at postexposure compared to their pre-exposure concentrations and serum bile acid levels
6 correlated well with TCE exposure (r = 0.94). Total serum bile acid concentration did not
7 change in control subjects between pre- and postexposure, nor did enzyme markers of liver
8 function in either unexposed or exposed subjects differ between pre and postexposure period.
9 However, the statistical power of this study is quite limited and the prevalence design does not
10 include subjects who may have left employment because of possible liver problems. The paper
11 provides minimal details of subject selection and workplace exposure conditions, except that
12 pre-exposure testing was carried out on the 1st work day of the week (pre-exposure), repeated
13 sampling after 2 days (postexposure), and a postexposure 8-hour time-weighted-average TCE
14 concentration of 9 ppm for exposed subjects; no exposure information is provided for control
15 subjects. Driscoll et al. (1992) in a study of 22 subjects (6 unexposed and 16 exposed) employed
16 at a factory manufacturing small appliances reported statistically significant group differences in
17 logistic regression analyses controlling for age and alcohol consumption in mean fasting plasma
18 bile acid concentrations. Other indicators of liver function such as plasma enzyme levels were
19 statistically significant different between exposed and unexposed subjects. Laboratory samples
20 were obtained at the start of subject's work shift. Exposure data are not available on the
21 22 subjects and assignment of exposed and unexposed was based on work duties. Limited
22 personal monitoring from other nonparticipating workers at this facility indicated TCE exposure
23 as low, less than 5 ppm, with occasional peaks over 250 ppm although details are lacking
24 whether these data represent exposures of study subjects.
25 Davis et al. (2006) in their analysis of subjects from the TCE subregistry of ATSDR's
26 National Exposure Registry examined the prevalence of subjects reporting liver problems
27 (defined as seeking treatment for the problem from a physician within the past year) using rates
28 for the equivalent health condition from the National Health Interview Survey (a nationwide
29 multipurpose health survey conducted by the National Center for Health Statistics, Centers for
30 Disease Control and Prevention). The TCE subregistry is a cohort of exposed persons from
31 15 sites in 5 states. The shortest time interval from inclusion in the exposure registry and last
32 follow-up was 5 years for one site and 10 years for seven sites. Excess in past-year liver
33 disorders relative to the general population persisted for much of the lifetime of follow-up.
34 SMRs for liver problems were 3rd follow-up, SMR = 2.23 (99% CI: 1.13, 3.92); 4th follow-up,
35 SMR = 3.25 (99% CI: 1.82, 5.32); and, 5th follow-up, SMR = 2.82 (99% CI: 1.46, 4.89).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-213 DRAFT—DO NOT CITE OR QUOTE
-------
1 Examination by TCE exposure, duration or cumulative exposure to multiple organic solvents did
2 not show exposure-response patterns. Overall, these observations are suggestive of liver
3 disorders as associated with potential TCE exposure, but whether TCE caused these conditions is
4 not possible to determine given the study's limitations. These limitations include a potential for
5 misclassification bias, the direction of which could dampen observations in a negative direction,
6 and lack of adjustment in statistical analyses for alcohol consumption, which could bias
7 observations in a positive direction.
8 Evaluation in epidemiologic studies of risk factors for cirrhosis other than alcohol
9 consumption and Hepatitis A, B, and C is quite limited. NRC (2006) cited a case report of
10 cirrhosis developing in an individual exposed occupationally to TCE for 5 years from a hot-
11 process degreaser and to 1,1,1-trichloroethane for 3 months thereafter (Thiele et al., 1982). One
12 cohort study on cirrhosis deaths in California between 1979 and 1981 and occupational risk
13 factors as assessed using job title observed elevated risks with occupational titles of sheet metal
14 workers and metalworkers and cirrhosis among white males who comprised the majority of
15 deaths (Leigh and Jiang, 1993). This analysis lacks information on alcohol patterns by
16 occupational title in addition to specific chemical exposures. Few deaths attributable to cirrhosis
17 are reported for nonwhite male and for both white and nonwhite female metalworkers with
18 analyses examining these individuals limited by low statistical power. Some but not all
19 trichloroethylene mortality studies report risk ratios for cirrhosis (see Table 4-50). A statistically
20 significant deficit in cirrhosis mortality is observed in three studies (Morgan et al., 1998;
21 Boice et al., 1999, 2006) and with risk ratios including a risk of 1.0 in the remaining studies
22 (Garabrant et al., 1988; Blair et al., 1989, 1998; Ritz, 1999; ATSDR, 2004). These results do not
23 rule out an effect of TCE on liver cirrhosis since disease misclassification may partly explain
24 observations. Available studies are based on death certificates where a high degree of
25 underreporting, up to 50%, is know to occur (Blake et al., 1988).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-214 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-50. Selected results from epidemiologic studies of TCE exposure and
cirrhosis
Study
population
Exposure group
Relative risk (95% CI)
No. obs.
events
Reference
Cohort and PMR-mortality
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
0.39(0.16,0.80)
Not reported
7
Boice et al., 2006
Zhao et al., 2005
View-master workers
Males
Females
0.76(0.16,2.22)
1.51(0.72,2.78)
3
10
ATSDR, 2003, 2004
Electronic workers (Taiwan)
Primary liver, males
Primary liver, females
Not reported
Not reported
Chang et al., 2005, 2003
Uranium-processing workers
Any TCE exposure
Light TCE exposure, >2 yrs duration
Mod TCE exposure, >2 yrs duration
0.91 (0.63, 1.28)
Not reported
Not reported
33
Ritz, 1999
Aerospace workers (Lockheed)
TCE routine exposure
TCE routine-intermittent
0.61 (0.39,0.91)
Not reported
23
13
Boice etal., 1999
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
0.55 (0.30, 0.93)
0.95 (0.43, 1.80)
0.32(0.10,0.74)
14
9
5
Morgan etal., 1998,
2000
Aircraft maintenance workers (Hill AFB, Utah)
TCE subcohort
1.1 (0.6, 1.9)a
44
Blair etal., 1998
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0"
0.6 (0.2, 1.3)
0.8(0.3,1.9)
1.2(0.6,2.4)
10
9
17
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-215 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-50. Selected results from epidemiologic studies of TCE exposure and
cirrhosis (continued)
Study
population
Aircraft
maintenance
workers
(continued)
Exposure group
Relative risk (95% CI)
No. obs.
events
Reference
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Deaths reported to GE pension fund (Pittsfield, MA)
1.0a
2.4(1.4, 13.7)
1.8(0.2,15.0)
0.6(0.1,4.8)
1.04(0.56, 1.93) a'b
0.87 (0.43, 1.73)
1.0 a'b
0.56 (0.23, 1.40)
1.07 (0.45, 2.53)
1.06 (0.48, 2.38)
1.79 (0.54, 5.93)
1.00a
3.30(0.88, 12.41)
2.20 (0.26, 18.89)
0.59(0.97,5.10)
Not reported
6
1
1
37
31
8
10
13
6
4
1
1
U. S. Coast Guard employees
Marine inspectors
Noninspectors
1.36(0.79,2.17)
0.53 (0.23, 1.05)
17
8
Aircraft manufacturing plant employees (Italy)
All subjects
Not reported
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
0.86(0.67, 1.11)
63
Radican et al., 2008
Greenland et al., 1994
Blair etal. (1989)
Costa etal., 1989
Garabrant et al., 1988
2
O
4
5
6
7
8
10
11
12
13
aReferent group are subjects from the same plant or company, or internal referents.
^Numbers of cirrhosis deaths in Radican et al. (2009) are fewer than Blair et al. (1998) because Radican et al. (2008)
excluded cirrhosis deaths due to alcohol.
A number of case reports exist of liver toxicity including hepatitis accompanying
immune-related generalized skin diseases described as a variation of erythema multiforme,
Stevens-Johnson syndrome, toxic epiderma necrolysis patients, and hypersensitivity syndrome
(Section 4.6.1.2 describes these disorders and evidence on TCE) (Kamijima et al., 2007).
Kamijima et al. (2007) reported hepatitis was seen in 92-94% of cases presenting with an
immune-related generalized skin diseases of variation of erythema multiforme, Stevens-Johnson
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-216 DRAFT—DO NOT CITE OR QUOTE
-------
1 syndrome, and toxic epiderma necrolysis patients, but the estimates within the hypersensitivity
2 syndrome group were more variable (46-94%). Many cases developed with a short time after
3 initial exposure and presented with jaundice, hepatomegaly or hepatosplenomegaly, in addition,
4 to hepatitis. Hepatitis development was of a nonviral etiology, as antibody liters for Hepatitis A,
5 B, and C viruses were not detectable, and not associated with alcohol consumption (Huang et al.,
6 2002; Kamijima et al., 2007). Liver failure was moreover a leading cause of death among these
7 subjects. Kamijima et al. (2007) note the similarities between specific skin manifestations and
8 accompanying hepatic toxicity and case presentations of TCE-related generalized skin diseases
9 and conditions that have been linked to specific medications (e.g., carbamezepine, allupurinol,
10 antibacterial sulfonamides), possibly in conjunction with reactivation of specific latent viruses.
11 However, neither cytomegalovirus or Epstein-Barr viruses are implicated in the few reports
12 which did include examination of viral antibodies.
13
14 4.5.2. Liver Cancer in Humans
15 Primary hepatocellular carcinoma and cholangiocarcinoma (intrahepatic and extrahepatic
16 bile ducts) are the most common primary hepatic neoplasms (El-Serag, 2007; Blehacz and
17 Gores, 2008). Primary hepatocellular carcinoma is the 5th most common of cancer deaths in
18 males and 9th in females (Jemal et al., 2008). Age-adjusted incidence rates of hepatocellular
19 carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) are increasing, with a 2-fold
20 increase in HCC over the past 20 years. This increase has not attributable to an expanded
21 definition of liver cancer to include primary or secondary neoplasms since International
22 Classification of Disease (ICD)-9, incorrect classification of hilar cholangiocarcinomas in ICD-O
23 as ICC, or to improved detection methods (Welzel et al., 2006; El-Serag, 2007). It is estimated
24 that 21,370 Americans will be diagnosed in 2008 with liver and intrahepatic bile cancer; age-
25 adjusted incidence rates for liver and intrahepatic bile duct cancer for all races are 9.9 per
26 100,000 for males and 3.5 per 100,000 for females (Ries et al., 2008). Survival for liver and
27 biliary tract cancers remains poor and age-adjusted mortality rates are just slightly lower than
28 incidence rates. While hepatitis B and C viruses and heavy alcohol consumption are believed
29 major risk factors for HCC and intrahepatic cholangiocarcinoma, these risk factors cannot fully
30 account for roughly 10 and 20% of HCC cases (Kulkarni et al., 2004). Cirrhosis is considered a
31 premalignant condition for HCC, however, cirrhosis is not a sufficient cause for HCC since 10 to
32 25% of HCC cases lack evidence of cirrhosis at time of detection (Chiesa et al., 2000; Fattovich
33 et al., 2004; Kumar et al., 2007). Nonalcoholic steatohepatitis reflecting obesity and metabolic
34 syndrome is recently suggested as contributing to liver cancer risk (El-Serag, 2007).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-217 DRAFT—DO NOT CITE OR QUOTE
-------
1 All cohort studies, except Zhao et al. (2005), present risk ratios (SIRs or SMRs) for liver
2 and biliary tract cancer. More rarely reported in cohort studies are risk ratios for primary liver
3 cancer (hepatocellular carcinoma or HCC) or for gallbladder and extrahepatic bile duct cancer.
4 Four community studies also presented risk ratios for liver and biliary tract cancer including a
5 case-control study of primary liver cancer of residents of Taiwanese community with solvent-
6 contaminated drinking water wells (Vartiainen et al., 1993; Morgan and Cassidy, 2002; Lee et
7 al., 2003; ATSDR, 2006). Several population case-control studies examine liver cancer and
8 organic solvents or occupational job titles with possible TCE usage (Stemhagen et al., 1983;
9 Hardell et al., 1984; Hernberg et al., 1984, 1988; Austin et al., 1987; Dossing et al., 1997;
10 Heinemann et al., 2000; Porru et al., 2001; Weiderpass et al., 2003; Ji and Hemminki, 2005;
11 Kvam et al., 2005; Lindbohm et al., 2009); however, the lack of detailed exposure assessment to
12 TCE, specifically in the population case-control studies as well as in geographic-based studies,
13 or, too few exposed cases and controls in those studies that do present some information limits
14 their usefulness for evaluating hepatobiliary or gall bladder cancer and TCE exposure.
15 Table 4-51 presents observations from cohort, case-control, and community studies on liver and
16 biliary tract cancer, primary liver, and gallbladder and extrahepatic bile duct cancer and
17 trichloroethylene.
18 Excess liver cancer incidence is observed in most high quality studies (Axelson et al.,
19 1994; Anttila et al., 1995; Hansen et al., 2001; Raaschou-Nielsen et al., 2003) as is mortality in
20 studies which assess TCE exposure by job exposure matrix approaches (Blair et al., 1998;
21 Morgan et al., 1998; Ritz, 1999; ATSDR, 2004; Boice et al., 2006; Radican et al., 2008). Risks
22 for primary liver cancer and for gallbladder and biliary tract cancers in females were statistically
23 significantly elevated only in Raaschou-Nielsen et al. (2003), the study with the largest number
24 of observed cases without suggestion of exposure duration-response patterns. Cohort studies
25 with more uncertain exposure assessment approaches, e.g., studies of all subjects working at a
26 factory (Garabrant et al., 1998; Blair et al., 1989; Costa et al., 1989; Chang et al., 2003, 2005), do
27 not show association but are quite limited given their lacking attribution of who may have higher
28 or lower exposure potentials. Ritz (1999), the exception, found evidence of an exposure-
29 response relationship; mortality from hepatobiliary cancer was found to increase with degree and
30 duration of exposure and time since first exposure with a statistically significant but imprecise
31 (wide confidence intervals) liver cancer risk for those with the highest exposure and longest time
32 since first exposure. This observation is consistent with association with TCE, but with
33 uncertainty given one TCE exposed case in the highest exposure group and correlation between
34 TCE, cutting fluids, and radiation exposures.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-218 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer
to
vo £3'
I
I
§
***.
£3'
1
TO'
Study
population
Exposure group
Liver and intrahepatic bile
ducts
Relative risk
(95% CI)
No. obs.
events
Primary liver
Relative risk
(95% CY)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% CY)
No. obs.
events
Reference
Cohort and PMR studies — incidence
Aerospace workers (Rocketdyne)
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
Not reported
Not reported
Not reported
Danish blue-collar workers with TCE exposure
Males + females
Males + females
Males, any exposure
<1 yr employment duration
1-4.9 yrs employment duration
>5 yrs employment duration
Females, any exposure
<1 yr employment duration
1-4.9 yrs employment duration
>5 yrs employment duration
1.3 (1.0, 1.6)a
1.4(1.0, 1.8)b
1.1(0.8, 1.5)b
1.2(0.7, 2.1)b
0.9(0.5, 1.6)b
1.1(0.6, 1.7)b
2.8(1.6, 4.6)b
2.5(0.7, 6.5) b
4.5(2.2, 8.3) b
1.1(0.1, 3.8)b
82
57
41
13
13
15
16
4
10
2
1.1 (0.7, 1.6)
1.3 (0.6, 2.5)
1.0(0.5, 1.9)
1.1(0.5,2.1)
2.8(1.1,5.8)
2.8 (0.3, 10.0)
4.1(1.1, 10.5)
1.3(0.0,7.1)
27
9
9
9
7
2
4
1
1.1 (0.6, 1.9)
1.1(0.3,2.9)
0.8(0.2,2.1)
1.4(0.5,3.1)
2.8(1.3,5.3)
2.3 (0.3, 8.4)
4.8(1.7, 10.4)
0.9 (0.0, 5.2)
14
4
4
6
9
2
6
1
Biologically -monitored Danish workers
Males + females
Males
Females
2.1(0.7, 5.0)b
2.6(0.8, 6.0) b
5
5
0(0.4
exp)
1.7 (0.2, 6.0)
1.8 (0.2, 6.6)
2
2
0(0.1
exp)
2.5 (0.5, 7.3)
3.3 (0.7, 9.7)
3
3
0(0.3
exp)
Zhao et al., 2005
Raaschou-
Nielson et al.,
2003
Hansenetal.,
2001
to
VO
I
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Exposure group
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-yr
Mean concentration (Ikeda)
<4ppm
4+ppm
Employment duration
<6.25 yr
>6.25
Liver and intrahepatic bile
ducts
Relative risk
(95% CI)
Not reported
Not reported
Not reported
No. obs.
events
Primary liver
Relative risk
(95% CI)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% CI)
No. obs.
events
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Not reported
9
Not reported
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
1.0C
0.6(0.1,3.1)
0.6(0.1,3.8)
1.1(0.2,4.8)
3
2
4
0
1.03
1.2(0.1,2.1)
1.0(0.1, 16.7)
2.6 (0.3, 25.0)
2
1
3
0
Biologically -monitored Finnish workers
All subjects
1.89(0.86, 3.59) b
9
2.27 (0.74, 5.29)
5
1.56(0.43,4.00)
4
Mean air-TCE (Ikeda extrapolation from U-TCA)
<6ppm
6+ppm
Not reported
1.64 (0.20, 5.92)
2.74 (0.33, 9.88)
2
2
Reference
Blair etal., 1998
Anttila et al.,
1995
to
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Exposure group
Liver and intrahepatic bile
ducts
Relative risk
(95% Cl)
No. obs.
events
Primary liver
Relative risk
(95% Cl)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% Cl)
No. obs.
events
Biologically -monitored Swedish workers
Males
Females
1.41(0.38, 3.60)b
Not reported
4
Reference
Axelson et al.,
1994
Cohort and PMR-mortality
Computer manufacturing workers (IBM), NY
Not reported
1
Aerospace workers (Rocketdyne)
Any TCE (utility/eng flush)
Low cumulative TCE score
Med cumulative TCE score
High TCE score
p for trend
1.28 (0.35, 3.27)
Not reported
4
View-Master workers
Males
Females
2.45(0.50, 7.12)d
3
0
(2.61 exp)
1.01 (0.03, 5.63) d
1
0
(1.66 exp)
8.41(1.01, 30.4)d
2
0
(0.95 exp)
Electronic workers (Taiwan)
Primary liver, males
Primary liver, females
Not reported
Not reported
0
(0.69 exp)
0
(0.57 exp)
Clapp and
Hoffman, 2008
Boice etal.,
2006
Zhao etal., 2005
ATSDR, 2003,
2004
Chang et al.,
2005, 2003
to
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Exposure group
Liver and intrahepatic bile
ducts
Relative risk
(95% Cl)
No. obs.
events
Primary liver
Relative risk
(95% Cl)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% Cl)
No. obs.
events
Uranium-processing workers
Any TCE exposure
Light TCE exposure, >2 yrs
ration
Mod TCE exposure, >2 yrs
duration
Light TCE exposure, >5 yrs
duration
Mod TCE exposure, >5 yrs
duration
Not reported
0.93(0.19, 4.53)e
4.97(0.48, 51.1) e
2.86 (0.48, 17.3)f
12.1(1.03, 144)f
3
1
3
1
Aerospace workers (Lockheed)
TCE routine exposure
0.54(0.15, 1.38)
4
TCE routine-intermittent
Oyrs
Any exposure
5yrs
p for trend
1.00C
Not reported
0.53 (0.18, 1.60)
0.52(0.15, 1.79)
0.94 (0.36, 2.46)
>0.20
22
13
4
3
6
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)e
High intensity (>50 ppm)e
0.98(0.36,2.13)
1.32 (0.27, 3.85)
0.78(0.16,2.28)
6
3
3
Reference
Ritz, 1999
Boice et al.,
1999
Morgan et al.,
1998, 2000
to
I
TO
Co
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Exposure group
Liver and intrahepatic bile
ducts
Relative risk
(95% Cl)
No. obs.
events
Primary liver
Relative risk
(95% Cl)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% Cl)
No. obs.
events
TCE subcohort (Cox analysis)
Never exposed
Ever exposed
1.00C
1.48 (0.56, 3.91)g'h
14
6
Cumulative
Low
High
2.12(0.59, 7.66)h
1.19(0.34, 4.16) h
3
3
Peak
No/low
Medium/high
1.00C
0.98(0.29, 3. 35)h
17
3
Aircraft maintenance workers (Hill AFB, Utah)
TCE subcohort
1.3(0.5, 3.4) c
15
1.7 (0.2, 16.2)3
4
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0C
1.1(0.3,4.1)
0.9 (0.2, 4.3)
0.7 (0.2, 3.2)
6
3
3
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
1.0C
1.6 (0.2, 18.2)
2.3 (0.3, 16.7)
1.12(0.57, 2.19)c>1
1
0
2
31
1.25 (0.3 l,4.97)c''
8
Reference
Blair etal., 1998
Radicanetal.,
2008
to
I
TO
Co
H I
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Aircraft
maintenance
workers
(continued)
Exposure group
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Deaths reported to GE pension fund
(Pittsfield, MA)
Liver and intrahepatic bile
ducts
Relative risk
(95% Cl)
1.36(0.59, 3. ll)c
1.0C
1.17(0.45,3.09)
1.16(0.39,3.46)
1.72 (0.68, 4.38)
0.74(0.18, 2.97) c
1.03
0.69 (0.08, 5.74)
0.98 (0.20, 4.90)
0.54 (0.11, 2.63)1
No. obs.
events
28
10
6
12
o
5
i
0
2
9
Primary liver
Relative risk
(95% Cl)
2.72(0.34, 21.88) c
1.03
3.28(0.37,29,45)
4.05 (0.45, 36.41)
No. obs.
events
8
4
0
4
0
Gallbladder and extrahepatic bile ducts
Relative risk
(95% Cl)
No. obs.
events
U. S. Coast Guard employees
Marine inspectors
Noninspectors
1.12(0.23,3.26)
Not reported
3
0 (2 exp)
Aircraft manufacturing plant employees (Italy)
All subjects
0.70 (0.23, 1.64)
5
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
0.94 (0.40, 1.86)
8
Reference
Greenland etal.,
1994
Blair etal., 1989
Costa et al.,
1989
Garabrant etal.,
1988
Case-control studies
Residents of community with contaminated drinking water (Taiwan)
Village of residency, males
Upstream
Downstream
1.00
2.57(1.21,5.46)
26
Lee etal., 2003
to
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
Table 4-51. Selected results from epidemiologic studies of TCE exposure and liver cancer (continued)
Study
population
Exposure group
Liver and intrahepatic bile
ducts
Relative risk
(95% CI)
No. obs.
events
Primary liver
Relative risk
(95% CI)
No. obs.
events
Gallbladder and extrahepatic bile ducts
Relative risk
(95% CI)
No. obs.
events
Reference
Geographic studies
Residents in two study areas in Endicott, NY
Residents in 13 census tracts in Redlands, CA
0.71 (0.09, 2.56)
1.29 (0.74, 2.05)k
<6
28
Finnish residents
Residents of Hausjarvi
Residents of Huttula
0.76(0.3, 1.4)
0.6 (0.2, 1.3)
7
6
ATSDR, 2006
Morgan and
Cassidy, 2002
Vartiainen et al.,
1993
aICD-7, 155 and 156; Primary liver (155.0), gallbladder, and biliary passages (155.1), and liver secondary and unspecified (156).
bICD-7, 155; Primary liver, gallbladder, and biliary passages.
Internal referents, workers without TCE exposure.
dProportional mortality ratio (PMR).
eLogistic regression analysis with a 0-year lag for TCE exposure.
fLogistic regression analysis with a 15-year lag for TCE exposure.
gRisk ratio from Cox Proportional Hazard Analysis, stratified by age, sex, and decade (Environmental Health Strategies, 1997).
hMorgen et al. (1998) do not identify if SIR is for liver and biliary passage or primary liver cancer; identified as primary liver in NRC (2006).
'Radican et al. (2008) provide results for TCE exposure for follow-up through 1990, comparing the Poisson model rate ratios as reported by Blair et al. (1998)
with Cox model hazard ratios. Relative risk from Cox model adjusted for age and gender for liver and intrahepatic bile duct cancer was 1.2 (95% CI: 0.5, 3.4)
and for primary liver cancer was 1.3 (95% CI: 0.1, 12.0).
JOdds ratio.
k99% confidence intervals.
exp = exposures.
-------
1 Observations in these studies provide some evidence of susceptibility of liver, gallbladder
2 and biliary tract; observations consistent with pharmacokinetic processing of TCE and the
3 extensive intra- and extrahepatic recirculation of metabolites. Magnitude of risk of gallbladder
4 and biliary tract cancer is slightly higher than that for primary liver cancer in Raaschou-Nielsen
5 et al. (2003), the study with the most cases. Observations in Blair et al. (1998), Hansen et al.
6 (2001), and Radican et al. (2008), three smaller studies, suggest slightly larger risk ratios for
7 primary liver cancer compared to gallbladder and biliary tract cancer. Overall, these studies are
8 not highly informative for cross-organ comparison of relative magnitude of susceptibility.
9 The largest geographic studies (Morgan and Cassidy, 2002; Lee et al., 2003) are also
10 suggestive of association with the risk ratio (mortality odds ratio) in Lee et al. (2003) as
11 statistically significantly elevated. The geographic studies do not include a characterization of
12 TCE exposure to individual subjects other than residency in a community with groundwater
13 contamination by TCE with potential for exposure misclassification bias dampening
14 observations; these studies lack characterization of TCE concentrations in drinking water and
15 exposure characteristics such as individual consumption patterns. For this reason, observations
16 in Morgan and Cassidy (2002) and Lee et al. (2003) are noteworthy, particularly if positive bias
17 leading to false positive finding is considered minimal, and the lack of association with liver
18 cancer in the two other community studies (Vartiainen et al., 1993; ATSDR, 2006) does not
19 detract from Morgan and Cassidy (2002) or Lee et al. (2003). Lee et al. (2003), however, do not
20 address possible confounding related to hepatitis viral infection status, a risk factor for liver
21 cancer, or potential misclassification due to the inclusion of secondary liver cancer among the
22 case series, factors which may amplify observed association.
23 Meta-analysis is adopted as a tool for examining the body of epidemiologic evidence on
24 liver cancer and TCE exposure, to identify possible sources of heterogeneity and as an additional
25 means to identify cancer hazard. The meta-analyses of the overall effect of TCE exposure on
26 liver (and gall bladder/biliary passages) cancer suggest a small, statistically significant increase
27 in risk. The pooled estimate from the primary random effects meta-analysis of the 9 (all cohort)
28 studies is 1.33 (95% CI: 1.09, 1.64) (see Figure 4-3). The study of Raaschou-Nielsen et al.
29 (2003) contributes about 57% of the weight; its removal from the analysis does not noticeably
30 change the RRp estimate, but the estimate is no longer statistically significant (RRp = 1.31; 95%
31 CI: 0.96, 1.79). The pooled estimate was not overly influenced by any other single study, nor
32 was it overly sensitive to individual RR estimate selections. There is no evidence of publication
33 bias in this data set, and no observable heterogeneity across the study results.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-226 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Liver Cancer
Study name
Statistics for each study
Risk ratio and 95% Cl
Anttila 1995
Axelson 1994
Boice 1999
Boice 2006
Greenland 1994
Hansen 2001
Morgan 1998 unpub RR 1.481
Raaschou-Nielsen 2003 1.350
Radican 2008
Risk
ratio
1.890
1.410
0.540
1.280
0.540
2.100
1.481
1.350
1.120
1.334
Lower
limit
0.983
0.529
0.203
0.480
0.110
0.874
0.561
1.030
0.571
1.088
Upper
limit
3.632
3.757
1.439
3.410
2.640
5.045
3.909
1.770
2.195
1.636
p-Value
0.056
0.492
0.218
0.622
0.447
0.097
0.428
0.030
0.741
0.006
0.1 0.2 0.5 1
5 10
random effects model; same for fixed
Figure 4-3. Relative risk estimates of liver and biliary tract cancer and
overall TCE exposure. The pooled estimate is in the bottom row. Symbol sizes
reflect relative weights of the studies. The horizontal midpoint of the bottom
diamond represents the pooled RR estimate and the horizontal extremes depict the
95% CI limits.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-227 DRAFT—DO NOT CITE OR QUOTE
-------
1 Examination of sites individually (i.e., primary liver and intrahepatic bile ducts separate
2 from the combined liver and gallbladder/biliary passage grouping) resulted in the RRp estimate
3 for liver cancer alone (for the 3 studies for which the data are available; for the other studies,
4 results for the combined grouping were used) slightly lower than the one based entirely on
5 results from the combined cancer categories (1.31; 95% CI: 1.02, 1.67). This result is driven by
6 the fact that the risk ratio estimate from the large Raaschou-Nielsen et al. (2003) study decreased
7 from 1.35 for liver and gall bladder/biliary passage cancers combined to 1.28 for liver cancer
8 alone.
9 The RRp estimate from the random effects meta-analysis of liver cancer in the highest
10 exposure groups in the 6 studies which provide risk estimates associated with highest exposure
11 primary liver cancer is 1.32 (95% CI: 0.93, 1.86), slightly lower than the RRp estimate for liver
12 and gallbladder/biliary cancer and any TCE exposure of 1.33 (95% CI: 1.09, 1.64), and not
13 statistically significant (see Figure 4-4). Again, the RRp estimate of the highest-exposure groups
14 is dominated by one study (Raaschou-Nielsen et al., 2003). Two studies lack reporting of liver
15 cancer risk associated with highest exposure, so consideration of reporting bias (considered the
16 primary analysis) let to a result of 1.28 (95% CI: 0.93, 1.77), similar to that estimated in the more
17 restricted set of studies presenting risk ratios association with highest exposure groups in
18 published papers.
19 Different exposure metrics are used in the various studies, and the purpose of combining
20 results across the different highest exposure groups is not to estimate an RRp associated with
21 some level of exposure, but rather to examine impacts of combining RR estimates that should be
22 less affected by exposure misclassification. In other words, the highest exposure category is
23 more likely to represent a greater differential TCE exposure compared to people in the referent
24 group than the exposure differential for the overall (typically any versus none) exposure
25 comparison. Thus, if TCE exposure increases the risk of liver and gallbladder/biliary cancer, the
26 effects should be more apparent in the highest exposure groups. The findings of a lower RRp
27 associated with highest exposure group reflects observations in Radican et al. (2008) and
28 Raaschou-Nielsen et al. (2003), the study contributing greatest weight to the meta-analysis, that
29 RR estimates for the highest-exposure groups, although greater than 1.0, are less than the RR
30 estimates with any TCE exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-228 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Liver Cancer - highest exposure groups
Study name
Statistics for each study
Risk ratio and 95% Cl
Anttila 1995
Axelson 1994 est
Boice 1999
Morgan 1998
Raaschou-Nielsen 2003 1.200
Radican 2008
Hansen 2001
Zhao 2005
Risk
ratio
2
3.
0.
1.
1.
1.
1.
1.
1.
.740
.700
.940
.190
.200
.490
.000
.000
.281
Lower
limit
0
.685
0.521
0
.360
0.340
0.
0.
0.
0
0
.746
.666
.323
.084
.925
Upper
limit p-Value
10.
26.
2
4.
1.
3.
3.
11.
1.
.956
.267
.457
.162
.930
.332
.098
.857
.774
0
0
0
0
0
0
1
1
0
.154
.191
.900
.785
.452
.331
.000
.000
.136
random effects model; same for fixed
Figure 4-4. Meta-analysis of liver cancer and TCE exposure—highest
exposure groups. With assumed null RR estimates for Hansen and Zhao (see
Appendix C text).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-229 DRAFT—DO NOT CITE OR QUOTE
-------
1 Thus, while the finding of an elevated and statistically significant RRp for liver and
2 gallbladder/biliary cancer and any TCE exposure provides evidence of association, the statistical
3 significance of the pooled estimates is dependent on one study, which provides the majority of
4 the weight in the meta-analyses. Furthermore, combining results from the highest-exposure
5 groups yields lower RRp estimates than for an overall effect. These results do not rule out an
6 effect of TCE on liver cancer, because the liver cancer results are relatively underpowered with
7 respect to numbers of studies and number of cases; overall, the meta-analysis provides only
8 minimal support for association between TCE exposure and liver and gallbladder/biliary cancer.
9 NRC (2006) deliberations on trichloroethylene commented on two prominent evaluations
10 of the then-current TCE epidemiologic literature using meta-analysis techniques, Wartenberg et
11 al. (2000) and Kelsh et al. (2005), submitted by Exponent-Health Sciences to NRC during their
12 deliberations and published afterwards in the open literature as Alexander et al. (2007) with the
13 substitution of the recently published study of Boice et al. (2006) for Ritz (1999) which Kelsh et
14 al. (2005) included in their NRC presentation. NRC (2006) found weaknesses in the techniques
15 used in Wartenberg et al. (2000) and the Exponent analyses. U.S. EPA staff conducted their
16 analysis according to NRC (2006) suggestions for transparency, systematic review criteria, and
17 examination of both cohort and case-control studies. The U.S. EPA analysis of liver cancer
18 considered a similar set of studies as Alexander et al. (2007) although treatment of these studies
19 differs between analyses. Alexander et al. (2007) in their Table 2, for example, present pooled
20 relative risk estimates, grouping of studies with differing exposure potentials, for example,
21 including the large cohort of Boice et al. (1999) of 77,965 subjects, 2.267 (3%) identified with
22 TCE exposure, with biomarker studies (Axelson et al., 1994; Anttila et al., 1995; Hansen et al.,
23 2001), whereas studies in the U.S. EPA analysis were identified using a systematic review and
24 objective criteria. Alexander et al. (2007) lacks a defined rationale for grouping studies with
25 subjects of different TCE exposure potentials, particularly studies with well-defined TCE
26 exposure assessment with large cohorts which include both TCE-exposed and non-TCE
27 exposure subjects. The inclusion of studies whose subjects have little to no TCE exposure over
28 background levels has the potential to introduce misclassification bias and dampen observed risk
29 ratios, a likely alternative explanation for observed inconsistency across occupational groups
30 reported by the authors. Additionally, Alexander et al. (2007) lacks quantitative examination of
31 liver cancer risk in the higher TCE exposure groups without explanation given their meta-
32 analysis of NHL did present such an examination (Mandel et al., 2006). A third difference
33 between the U.S. EPA and previous meta-analyses is their treatment of Ritz (1999), included in
34 Wartenberg et al. (2000), Kelsh et al. (2005), and Alexander et al. (2007), but not in this analysis.
35 In spite the weaknesses in past meta-analyses, pooled liver and gall bladder/biliary tract cancer
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-230 DRAFT—DO NOT CITE OR QUOTE
-------
1 risk estimates for overall TCE exposure for TCE subcohorts is of a similar magnitude as that
2 observed in U.S. EPA's updated and expanded analysis, Wartenberg et al. (2000), 1.1 (95% CI:
3 0.3, 4.8) for incidence and 1.1 (95% CI: 0.7, 1.7) for mortality, Kelsh et al. (2005), 1.32 (95%
4 CI: 1.05, 1.66) and Alexander et al. (2007), 1.30 (95% CI: 1.09-1.55).
5
6 4.5.3. Experimental Studies of Trichloroethylene (TCE) in Rodents—Introduction
7 The previous sections have described available human data for TCE-induced noncancer
8 effects (e.g., disturbances in bile production) and whether an increased risk of liver cancer in
9 humans has been established from analysis of the epidemiological literature. A primary concern
10 for effects on the liver comes from a large database in rodents indicating that, not only TCE, but
11 a number of its metabolites are capable of inducing hepatocellular adenomas and carcinomas in
12 rodent species. Thus, many of rodent bioassays have focused on the study of liver cancer for
13 TCE and its metabolites and possible early effects specifically that may be related to tumor
14 induction.
15 This section describes the hazard data for TCE effects in the rodent liver and inferences
16 from studies of its metabolites. For more detailed descriptions of the issues providing context for
17 these data in terms the state of the science of liver physiology (see Section E. 1), cancer (see
18 Section E.3), liver cancer (see Section E.3), and the MOA of liver cancer and other TCE-induced
19 effects (see Section E.3.4), please see Appendix E. A more comprehensive review of individual
20 studies of TCE-induced liver effects in laboratory animals is also provided in Section E.2 that
21 includes detailed analyses of the strengths and the limitations of these studies. Issues have been
22 raised regarding the relevance of mouse liver tumor data to human liver cancer risk that are
23 addressed in Sections E.3.2 and E.3.3. Given that activation of the PPARa receptor has received
24 great attention as a potential MOA for TCE induced liver tumors, the current status of that
25 hypothesis is reviewed in Section E.3.4.1. Finally, comparative studies of TCE metabolites and
26 the similarities and differences of such study results are described in summary sections of
27 Appendix E (i.e., Section E.2.4) as well as discussions of proposed MO As for TCE-induced liver
28 cancer (i.e., Sections E.2.4 and E.3.4.2).
29 A number of acute and subchronic studies have been undertaken to describe the early
30 changes in the rodent liver after TCE administration with the majority using the oral gavage
31 route of administration. Several key issues affect the interpretation of these data. The few
32 drinking water studies available for TCE have recorded significant loss of TCE through
33 volatilization in drinking water solutions and thus, this route of administration is generally not
34 used. Some short-term studies of TCE have included detailed examinations while others have
35 reported primarily liver weight changes as a marker of TCE response. The matching and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-231 DRAFT—DO NOT CITE OR QUOTE
-------
1 recording of age, but especially initial and final body weight, for control and treatment groups is
2 of particular importance for studies using liver weight gain as a measure of TCE response as
3 differences in these parameters affect TCE-induced liver weight gain. Most data are for TCE
4 exposures of at least 10 days to 42 days. For many of the subchronic inhalation studies
5 (Kjellstrand et al., 1981, 1983a, b), issues associated with whole body exposures make
6 determination of dose levels more difficult. The focus of the long-term studies of TCE is
7 primarily detection and characterization of liver tumor formation.
8 For gavage experiments, death due to gavage errors and specifically from use of this
9 route of administration, especially at higher TCE exposure concentrations, has been a recurring
10 problem, especially in rats. Unlike inhalation exposures, the effects of vehicle can also be an
11 issue for background liver effects in gavage studies. Concerns regarding effects of oil vehicles,
12 especially corn oil, have been raised (Kim et al., 1990; Charbonneau et al., 1991). Several oral
13 studies in particular document that use of corn oil as the vehicle for TCE gavage dosing induces
14 a different pattern of toxicity, especially in male rodents (see Merrick et al., 1989;
15 Section E.2.2.1). Several studies also report the effects of corn oil on hepatocellular DNA
16 synthesis and indices of lipid peroxidation (Channel et al., 1998; Rusyn et al., 1999). For
17 example, Rusyn et al. (1999) report that a single dose of dietary corn oil increases hepatocyte
18 DNA synthesis 24 hours after treatment by ~3.5-fold of control, activates of NF-KB to a similar
19 extent ~2 hours after treatment almost exclusively in Kupffer cells, and induces an ~3-4-fold
20 increase of control NF-KB in hepatocytes after 8 hours and an increase in TNFa mRNA between
21 8 and 24 hours after a single dose in female rats.
22 In regard to studies that have used the i.p. route of administration, as noted by
23 Kawamoto et al. (1988), injection of TCE may result in paralytic ileus and peritonitis and that
24 subcutaneous treatment paradigm will result in TCE not immediately being metabolized but
25 retained in the fatty tissue. Wang and Stacey (1990) state that "intraperitoneal injection is not
26 particularly relevant to humans" and suggest that intestinal interactions require consideration in
27 responses such as increase serum bile acid.
28 While studies of TCE metabolites have been almost exclusively conducted via drinking
29 water, and thus, have avoided vehicle effects and gavage error, they have issues of palatability at
30 high doses and decreased drinking water consumption as a result that not only raises issues of the
31 resulting internal dose of the agent but also of effects of drinking water reduction.
32 Although there are data for both mice and rats for TCE exposure and studies of its
33 metabolites, the majority of the available information has been conducted in mice. This is
34 especially the case for long-term studies of DCA and TCA in rats. There is currently one study
35 each available for TCA and DCA in rats and both were conducted with such few numbers of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-232 DRAFT—DO NOT CITE OR QUOTE
-------
1 animals that the ability to detect and discern whether there was a treatment-related effect are very
2 limited (DeAngelo et al., 1997, 1996; Richmond et al., 1995).
3 With regard to the sensitivity of studies used to detect a response, there are issues
4 regarding not only the number of animals used but also the strain and weight of the animals. For
5 some studies of TCE strains were used that have less background rate of liver tumor
6 development and carcinogenic response. As for the B6C3F1 mouse, the strain most used in the
7 bioassays of TCE metabolites, the susceptibility of the B6C3F1 to hepatocarcinogenicity has
8 made the strain a sensitive biomarker for a variety of hepatocarcinogens. Moreover, Leakey et
9 al. (2003b) demonstrated that increased body weight at 45 weeks of life is an accurate predictor
10 of large background tumor rates. Unfortunately a 2-year study of chloral hydrate (George et al.,
11 2000) and the only available 2-year study of TCA (DeAngelo et al., 2008), which used the same
12 control animals, were both conducted in B6C3F1 mice that grew very large (-50 g) and prone to
13 liver cancer (64% background incidence of hepatocellular adenomas and carcinomas) and
14 premature mortality. Thus, these bioassays are of limited value for determination of the dose-
15 response for carcinogenicity.
16 Finally, as discussed below, the administration of TCE to laboratory animals as well as
17 environmental exposure of TCE in humans are effectively coexposure studies. TCE is
18 metabolized to a number of hepatoactive as well as hepatocarcinogenic agents. A greater
19 variability of response is expected than from exposure to a single agent making it particularly
20 important to look at the TCE database in a holistic fashion rather than the results of a single
21 study, especially for quantitative inferences. This approach is particularly useful given that the
22 number of animals in treatment groups in a variety of TCE and TCE metabolite studies have
23 been variable and small for control and treatment groups. Thus, their statistical power was not
24 only limited for detection of statistically significant changes but also in many cases to be able to
25 determine whether there is not a treatment related effect (i.e., Type II error for power
26 calculation). Section E.2.4.2 provides detailed analyses of the database for liver weight
27 induction by TCE and its metabolites in mice and the results of those analyses are described
28 below. Specifically, the relationship of liver weight induction, but also other endpoints such as
29 peroxisomal enzyme activation and increases in DNA synthesis to liver tumor responses are also
30 addressed as well.
31
32 4.5.4. Trichloroethylene (TCE)-Induced Liver Noncancer Effects
33 A number of effects have been studied as indicators of TCE effects on the liver but also
34 as proposed events whose sequellae could be associated with resultant liver tumors after chronic
35 TCE exposure in rodents. Similar effects have been studied in rodents exposed to TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-233 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolites which may be useful for not only determining whether such effects are associated
2 with liver tumors induced by these metabolites but also if they are similar to what has been
3 observed for TCE.
4
5 4.5.4.1. Liver Weight
6 Increases in liver weight in mice, rats, and gerbils have been reported as a result of acute
7 and short-term, and subchronic TCE treatment by inhalation and oral routes of exposure
8 (Nunes et al., 2001; Tao et al., 2000; Tucker et al., 1982; Goldsworthy and Popp, 1987;
9 Elcombe et al., 1985; Dees and Travis, 1993; Nakajima et al., 2000; Berman et al., 1995;
10 Melnick et al., 1987; Laughter et al., 2004; Merrick et al., 1989; Goel et al., 1992;
11 Kjellstrand et al., 1981, 1983a, b; Buben and O'Flaherty, 1985). The extent of TCE-induced
12 liver weight gain is dependent on species, strain, gender, nutrition status, duration of exposure,
13 route of administration, vehicle used in oral studies, and the concentration of TCE administered.
14 Of great importance to the determination of the magnitude of response is whether the dose of
15 TCE administered also affects whole body weight, and thus, liver weight and the percent
16 liver/body weight ratio. Therefore, studies which employed high enough doses to induce whole
17 body weight loss generally showed a corresponding decrease in percent liver/body weight at
18 such doses and "flattening" of the dose-response curve, while studies which did not show
19 systemic toxicity reported liver/body weight ratios generally proportional to dose. Chronic
20 studies, carried out for longer durations, that examine liver weight are few and often confounded
21 by the presence of preneoplastic foci or tumors that also affect liver weight after an extended
22 period of TCE exposure. The number of studies that examine liver weight changes in the rat are
23 much fewer than for mouse. Overall, the database for mice provides data for examination of the
24 differences in TCE-induced effects from differing exposure levels, durations of exposure,
25 vehicle, strain, and gender. One study provided a limited examination of TCE-induced liver
26 weight changes in gerbils.
27 TCE-induced increases in liver weight have been reported to occur quickly.
28 Kjellstrand et al. (1981) reported liver weight increases after 2 days inhalation exposure in
29 NMRI mice, Laughter et al. (2004) reported increased liver weight in SV129 mice in their 3-days
30 study (see below), and Tao et al. (2000) reported a increased in percent liver/body weight ratio in
31 female B6C3Flmice for after 5 days. Elcombe et al. (1985) and Dees and Travis (1993) reported
32 gavage results in mice and rats after 10 days exposure to TCE which showed TCE-induced
33 increases in liver weight. Tucker et al. (1982) reported that 14 days of exposure to 24 mg/kg and
34 240 mg/kg TCE via gavage to induce a dose-related increase in liver weight in male CD-I mice
35 but did not show the data.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-234 DRAFT—DO NOT CITE OR QUOTE
-------
1 For mice, the inhalation studies of Kjellstrand et al provided the most information on the
2 affect of duration of exposure, dose of exposure, strain tested, gender, initial weight, and
3 variability in response between experiments on TCE-induced liver weight increases. These
4 experiments also provided results that were independent of vehicle effect. Although the
5 determination of the exact magnitude of response is limited by experimental design,
6 Kjellstrand et al. (1981) reported that in NMRI mice, continuous TCE inhalation exposure
7 induced increased percent liver/body weight by 2 days and that by 30 days (the last recorded data
8 point) the highest percent liver/body weight ratio was reported (~1.75-fold over controls) in both
9 male and female mice. Kjellstrand et al. (1983b) exposed seven different strains of mice (wild,
10 C57BL, DBA, B6CBA, A/sn, NZB, NMRI) to 150-ppm TCE for 30 days and demonstrated that
11 strain, gender, and toxicity, as reflected by changes in whole body weight, affected the percent
12 liver/body weight ratios induced by 30 days of continuous TCE exposure. In general for the
13 7 strains of mice examined, female mice had the less variable increases in TCE-induced liver
14 weight gain across duplicate experiments than male mice. For instance, in strains that did not
15 exhibit changes in body weight (reflecting systemic toxicity) in either gender (wild-type and
16 DBA), 150-ppm TCE exposure for 30 days induced 1.74- to 1.87-fold of control percent
17 liver/body weight ratios in female mice and 1.45- to 2.00-fold of control percent liver/body
18 weight ratios in male mice. The strain with the largest TCE-induced increase in percent
19 liver/body weight increase was the NZB strain (~2.08-fold of control for females and 2.34- to
20 3.57-fold of control for males). Kjellstrand et al. (1983b) provided dose-response information
21 for the NMRI strain of mice (A Swiss-derived strain) that indicated dose-related increases in
22 percent liver/body weight ratios between 37- and 300-ppm TCE exposure for 30 days. The
23 150-ppm dose was reported to induce a 1.66- and 1.69-fold increases in percent liver/body
24 weight ratios in male and female mice, respectively. Interestingly, they also reported similar
25 liver weight increases among groups with the same cumulative exposure, but with different daily
26 exposure durations (1 hour/day at 3,600 ppm to 24 hours/day at 150 ppm for 30 days).
27 Not only have most gavage experiments have been carried out in male mice, which
28 Kjellstrand et al. (1983a) had demonstrated to have more variability in response than females,
29 but also vehicle effects were noted to occur in experiments that examined them. Merrick et al.
30 (1989) reported that corn oil induced a similar increase in percent liver/body weight ratios in
31 female mice fed TCE in emulphor and corn oil for 4 weeks, male mice TCE administered in the
32 corn oil vehicle induced a greater increase in liver weight than emulphor but less mortality at a
33 high does.
34 Buben and O'Flaherty (1985) treated male, outbred Swiss-Cox mice for 6 weeks at doses
35 ranging from 100 to 3,200 mg/kg/d, and reported increased liver/body-weight ratios at all tested
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-23 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 doses (1.12- to 1.75-fold of controls). Given the large strain differences observed by Kjellstrand
2 et al. (1983b), the use of predominantly male mice, and the effects of vehicle in gavage studies,
3 interstudy variability in dose-response relationships is not surprising.
4 Dependence of PPARa activation for TCE-liver weight gain has been investigated in
5 PPARa null mice by both Nakajima et al. (2000) and Laughter et al. (2004). Nakajima et al.
6 (2000) reported that after 2 weeks of 750 mg/kg TCE exposure to carefully matched SV129
7 wild-type or PPARa-null male and female mice (n = 6 group), there was a reported 1.50-fold
8 increase in wild-type and 1.26-fold of control percent liver/body weight ratio in PPARa-null
9 male mice. For female mice, there was ~1.25-fold of control percent liver/body weight ratios for
10 both wild-type and PPARa-null mice. Thus, TCE-induced liver weight gain was not dependent
11 on a functional PPARa receptor in female mice and some portion of it may have been in male
12 mice. Both wild-type male and female mice were reported to have similar increases in the
13 number of peroxisome in the pericentral area of the liver and TCE exposure and, although
14 increased 2-fold, were still only -4% of cytoplasmic volume. Female wild-type mice were
15 reported to have less TCE-induced elevation of very long chain acyl-CoA synthetase, D-type
16 peroxisomal bifunctional protein, mitochondrial trifunctional protein a subunits a and P, and
17 cytochrome P450 4A1 than males mice, even though peroxisomal volume was similarly elevated
18 in male and female mice. The induction of PPARa protein by TCE treatment was also reported
19 to be slightly less in female than male wild-type mice (2.17- vs. 1.44-fold of control induction,
20 respectively). Thus, differences between genders in this study were for increased liver weight
21 were not associated with differences in peroxisomal volume in the hepatocytes but there was a
22 gender-related difference in induction of enzymes and proteins associated with PPARa.
23 The study of Laughter et al. (2004) used SV129 wild-type and PPARa-null male mice
24 treated with 3 daily doses of TCE in 0.1% methyl cellulose for either 3 days (1,500 mg/kg TCE)
25 or 3 weeks (0, 10, 50, 125, 500, 1,000, or 1,500 mg/kg TCE 5 days a week). However, the
26 paradigm is not strictly comparable to other gavage paradigms due to the different dose vehicle
27 and the documented impacts of vehicles such as corn oil on TCE-induced effects. In addition, no
28 initial or final body weights of the mice were reported and thus, the influence of differences in
29 initial body weight on percent liver/body weight determinations could not be ascertained. While
30 control wild-type and PPARa-null mice were reported to have similar percent liver/body weight
31 ratios (i.e., -4.5%) at the end of the 3-day study, at the end of the 3-week experiment the percent
32 liver/body weight ratios were reported to be larger in the control PPARa-null male mice (5.1%).
33 TCE treatment for 3 days was reported for percent liver/body weight ratio to be 1.4-fold of
34 control in the wild-type mice and 1.07-fold of control in the null mice. After 3 weeks of TCE
35 exposure at varying concentrations, wild-type mice were reported to have percent liver/body
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-236 DRAFT—DO NOT CITE OR QUOTE
-------
1 weight ratios that were within -2% of control values with the exception of the 1,000 mg/kg and
2 1,500 mg/kg treatment groups (-1.18- and 1.30-fold of control, respectively). For the PPARa-
3 null mice the variability in percent liver/body weight ratios were reported to be greater than that
4 of the wild-type mice in most of the TCE groups and the baseline levels of percent liver/body
5 weight ratio for control mice 1.16-fold of that of wild-type mice. TCE exposure was apparently
6 more toxic in the PPARa-null mice. Decreased survival at the 1,500 mg/kg TCE exposure level
7 resulted in the prevention of recording of percent liver/body weight ratios for this group. At
8 1,000 mg/kg TCE exposure level, there was a reported 1.10-fold of control percent liver/body
9 weight ratio in the PPARa-null mice. None of the increases in percent liver/body weight in the
10 null mice were reported to be statistically significant by Laughter et al. (2004). However, the
11 power of the study was limited due to low numbers of animals and increased variability in the
12 null mice groups. The percent liver/body weight ratio after TCE treatment reported in this study
13 was actually greater in the PPARa-null mice than the wild-type male mice at the 1,000 mg/kg
14 TCE exposure level (5.6 ± 0.4% vs. 5.2 ± 0.5%, for PPARa-null and wild-type mice,
15 respectively) resulting in a 1.18-fold of wild-type and 1.10-fold of PPARa-null mice. Although
16 the results reported in Laughter et al. (2004) for DC A and TCA were not conducted in
17 experiments that used the same paradigm, the TCE-induced increase in percent liver/body weight
18 more closely resembled the dose-response pattern for DCA than for DCA wild-type SV129 and
19 PPARa-null mice.
20 No study examined strain differences among rats, and cross-study comparisons are
21 confounded by heterogeneity in the age of animals, dosing regimen, and other design
22 characteristics that may affect the degree of response. For rats, TCE-induced percent liver/body
23 weight ratios were reported to range from 1.16- to 1.46-fold of control values depending on the
24 study paradigm. The studies which employed the largest range of exposure concentrations
25 (Melnick et al., 1987; Berman et al., 1995) examined 4 doses in the rat. In general, there was a
26 dose-related increase in percent liver/body weight in the rat, especially at doses that did not cause
27 concurrent decreased survival or significant body weight loss. For gerbils, Kjellstrand et al.
28 (1981) reported a similar value of-1.25-fold of control percent liver/body weight as for S-D rats
29 exposed to 150 ppm TCE continuously for 30 days. Woolhiser et al. (2006) also reported
30 inhalation TCE exposure to increase the percent liver/body weight ratios in female Sprague-
31 Dawley rats although this strain appeared to be less responsive that others tested for induction of
32 hepatomegaly from TCA exposure and to also be less prone to spontaneous liver cancer.
33 The size of the liver is under tight control and after cessation of a mitogenic stimulus or
34 one inducing hepatomegaly, the liver will return to its preprogrammed size (see Appendix E).
35 The increase in liver weight from TCE-exposure also appears to be reversible. Kjellstrand et al.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-237 DRAFT—DO NOT CITE OR QUOTE
-------
1 (1981) reported a reduction in liver weight gain increases after cessation of TCE exposure for 5
2 or 30 days in male and female mice. However, experimental design limitations precluded
3 discernment of the magnitude of decrease. Kjellstrand et al. (1983b) reported that mice exposed
4 to 150 ppm TCE for 30 days and then examined 120 days after the cessation of exposure, had
5 liver weights were 1.09-fold of control for TCE-exposed female mice and the same as controls
6 for TCE-exposed male mice. However, the livers were not the same as untreated liver in terms
7 of histopathology. The authors reported that "after exposure to 150 ppm for 30 days, followed
8 by 120 days of rehabilitation, the morphological picture was similar to that of the air-exposure
9 controls except for changes in cellular and nuclear sizes." Qualitatively, the reduction in liver
10 weight after treatment cessation is consistent with the report of Elcombe et al. (1985) in Alderly
11 Park mice. The authors report that the reversibility of liver effects after the administration of
12 TCE to Alderly Park mice for 10 consecutive days. Effects upon liver weight, DNA
13 concentration, and tritiated thymidine incorporation 24 and 48 hours after the last dose of TCE
14 were reported to still be apparent. However, 6 days following the last dose of TCE, all of these
15 parameters were reported to return to control values with the authors not showing the data to
16 support this assertion. Thus, cessation of TCE exposure would have resulted in a 75% reduction
17 in liver weight by 4 days in mice exposed to the highest TCE concentration. Quantitative
18 comparisons are not possible because Elcombe et al. (1985) did not report data for these results
19 (e.g., how many animals, what treatment doses, and differences in baseline body weights) and
20 such a large decrease in such a short period of time needs to be verified.
21
22 4.5.4.2. Cytotoxicity
23 Acute exposure to TCE appears to induce low cytotoxicity below subchronically lethal
24 doses. Relatively high doses of TCE appear necessary to induce cytotoxicity after a single
25 exposure with two available studies reported in rats. Okino et al. (1991) reported small increases
26 in the incidence of hepatocellular necrosis in male Wistar rats exposed to 2,000 ppm (8 hours)
27 and 8,000 ppm (2 hours), but not at lower exposures. In addition, "swollen" hepatocytes were
28 noted at the higher exposure when rats were pretreated with ethanol or Phenobarbital. Serum
29 transaminases increased only marginally at the 8,000-ppm exposure, with greater increases with
30 pretreatments. Berman et al. (1995) reported hepatocellular necrosis, but not changes in serum
31 markers of necrosis, after single gavage doses of 1,500 and 5,000 mg/kg TCE in female F344
32 rats. However, they did not report any indications of necrosis after 14 days of treatment at
33 50-1,500 mg/kg/d nor the extent of necrosis.
34 At acute and subchronic exposure periods to multiple doses, the induction of cytotoxicity,
35 though usually mild, appears to differ depending on rodent species, strain, dosing vehicle and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-23 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 duration of exposure, and the extent of reporting to vary between studies. For instance,
2 Elcombe et al. (1985) and Dees and Travis(1993), which used the B6C3F1 mouse strain and corn
3 oil vehicle, reported only slight or mild necrosis after 10 days of treatment with TCE at doses up
4 to 1,500 mg/kg/d. Elcombe et al. (1985) also reported cell hypertrophy in the centrilobular
5 region. Dees and Travis (1993) reported some loss of vacuolization in hepatocytes of mice
6 treated at 1,000 mg/kg/d. Laughter et al. (2004) reported that "wild-type" SV129 mice exposed
7 to 1,500 mg/kg TCE exposure for 3 weeks exhibited mild granuloma formation with calcification
8 or mild hepatocyte degeneration but gave not other details or quantitative information as to the
9 extent of the lesions or what parts of the liver lobule were affected. The authors noted that
10 "wild-type mice administered 1,000 and 1,500 mg/kg exhibited centrilobular hypertrophy" and
11 that "the mice in the other groups did not exhibit any gross pathological changes" after TCE
12 exposure. Channel et al. (1998) reported no necrosis in B6C3F1 mice treated by
13 400-1,200 mg/kg/d TCE by corn oil gavage for 2 days to 8 weeks.
14 However, as stated above, Merrick et al. (1989) reported that corn oil resulted in more
15 hepatocellular necrosis, as described by small focal areas of 3-5 hepatocytes, in male B6C3F1
16 mice than use of emulphor as a vehicle for 4-week TCE gavage exposures. Necrotic hepatocytes
17 were described as surrounded by macrophages and polymorphonuclear cells. The authors
18 reported that visible necrosis was observed in 30-40% of male mice administered TCE in corn
19 oil but not that there did not appear to be a dose-response. For female mice, the extent of
20 necrosis was reported to be 0 for all control and TCE treatment groups using either vehicle.
21 Serum enzyme activities for alanine aminotransferase (ALT), AST, and LDH (markers of liver
22 toxicity) showed that there was no difference between vehicle groups at comparable TCE
23 exposure levels for male or female mice. Except for LDH levels in male mice exposed to TCE
24 in corn oil there was not a correlation with the extent of necrosis and the patterns of increases in
25 ALT and AST enzyme levels.
26 Ramdhan et al. (2008) assessed TCE-induced hepatotoxicity by measuring plasma ALT
27 and AST activities and histopathology in Sv/129 mice treated by inhalation exposure, which are
28 not confounded by vehicle effects. Despite high variability and only six animals per dose group,
29 all three measures showed statistically significant increases at the high dose of 2,000 ppm
30 (8 hours/day for 7 days), although a nonstatistically significant elevation is evident at the low
31 dose of 1,000 ppm. Even at the highest dose, cytotoxicity was not severe, with ALT and AST
32 measures increased 2-fold or less and an average histological score less than 2 (range 0-4).
33 Kjellstrand et al. (1983b) exposed male and female NRMI mice to 150 ppm for 30 to
34 120 days. Kjellstrand et al. (1983b) reported more detailed light microscopic findings from their
35 study and stated that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-239 DRAFT—DO NOT CITE OR QUOTE
-------
1 After 150 ppm exposure for 30 days, the normal trabecular arrangement of the
2 liver cells remained. However, the liver cells were generally larger and often
3 displayed a fine vacuolization of the cytoplasm. The nucleoli varied slightly to
4 moderately in size and shape and had a finer, granular chromatin with a varying
5 basophilic staining intensity. The Kupffer cells of the sinusoid were increased in
6 cellular and nuclear size. The intralobular connective tissue was infiltrated by
7 inflammatory cells. There was not sign of bile stasis. Exposure to TCE in higher
8 or lower concentrations during the 30 days produced a similar morphologic
9 picture. After intermittent exposure for 30 days to a time-weighted-average
10 concentration of 150 ppm or continuous exposure for 120 days, the trabecular
11 cellular arrangement was less well preserved. The cells had increased in size and
12 the variations in size and shape of the cells were much greater. The nuclei also
13 displayed a greater variation in basophilic staining intensity, and often had one or
14 two enlarged nucleoli. Mitosis was also more frequent in the groups exposed for
15 longer intervals. The vacuolization of the cytoplasm was also much more
16 pronounced. Inflammatory cell infiltration in the interlobular connective tissue
17 was more prominent. After exposure to 150 ppm for 30 days, followed by
18 120 days of rehabilitation, the morphological picture was similar to that of the air-
19 exposure controls except for changes in cellular and nuclear sizes.
20
21 Although not reporting comparisons between male and female mice in the results section
22 of the paper for TCE-induced histopathological changes, the authors stated in the discussion
23 section that "However, liver mass increase and the changes in liver cell morphology were similar
24 in TCE-exposed male and female mice." Kjellstrand et al. (1983b) did not present any
25 quantitative data on the lesions they describe, especially in terms of dose-response. Most of the
26 qualitative description presented was for the 150-ppm exposure level and the authors suggest that
27 lower concentrations of TCE give a similar pathology as those at the 150-ppm level, but do not
28 present data to support that conclusion. Although stating that Kupffer cells were reported to be
29 increased in cellular and nuclear size, no differential staining was applied light microscopy
30 sections to distinguish Kupffer from endothelial cells lining the hepatic sinusoid in this study.
31 Without differential staining such a determination is difficult at the light microscopic level.
32 Indeed, Goel et al. (1992) describe proliferation of "sinusoidal endothelial cells" after
33 1,000 and 2,000 mg/kg/d TCE exposure for 28 days in male Swiss mice. They reported that
34 histologically, "the liver exhibits swelling, vacuolization, widespread degeneration/necrosis of
35 hepatocytes as well as marked proliferation of endothelial cells of hepatic sinusoids at 1,000 and
36 2,000 mg/kg TCE doses." Only one figure is given, at the light microscopic level, in which it is
37 impossible to distinguish endothelial cells from Kupffer cells and no quantitative measures or
38 proliferation were examined or reported to support the conclusion that endothelial cells are
39 proliferating in response to TCE treatment. Similarly, no quantitative analysis regarding the
40 extent or location of hepatocellular necrosis was given. The presence or absence of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-240 DRAFT—DO NOT CITE OR QUOTE
-------
1 inflammatory cells were not noted by the authors as well. In terms of white blood cell count, the
2 authors note that it is slightly increased at 500 mg/kg/d but decreased at 1,000 and 2,000 mg/kg/d
3 TCE, perhaps indicating macrophage recruitment from blood to liver and kidney, which was also
4 noted to have pathology at these concentrations of TCE.
5 The inflammatory cell infiltrates described in the Kjellstrand et al. (1983b) study are
6 consistent with invasion of macrophages and well as polymophorphonuclear cells into the liver,
7 which could activate resident Kupffer cells. Although not specifically describing the changes as
8 consistent with increased polyploidization of hepatocytes, the changes in cell size and especially
9 the continued change in cell size and nuclear staining characteristics after 120 days of cessation
10 of exposure are consistent with changes in polyploidization induced by TCE. Of note is that in
11 the histological description provided by the authors, although vacuolization is reported and
12 consistent with hepatotoxicity or lipid accumulation, which is lost during routine histological
13 slide preparation, there is no mention of focal necrosis or apoptosis resulting from these
14 exposures to TCE.
15 Buben and O'Flaherty (1985) reported liver degeneration "as swollen hepatocytes" and to
16 be common with treatment of TCE to Male Swiss-Cox mice after 6 weeks. They reported that
17 "Cells had indistinct borders; their cytoplasm was clumped and a vesicular pattern was apparent.
18 The swelling was not simply due to edema, as wet weight/dry weight ratios did not increase."
19 Karyorrhexis (the disintegration of the nucleus) was reported to be present in nearly all
20 specimens and suggestive of impending cell death. No Karyorrhexis, necrosis, or polyploidy
21 was reported in controls, but a low score Karyorrhexis was given for 400 mg/kg TCE and a
22 slightly higher one given for 1,600 mg/kg TCE. Central lobular necrosis reported to be present
23 only at the 1,600 mg/kg TCE exposure level and assigned a low score. Polyploidy was described
24 as characteristic in the central lobular region but with low score for both 400 mg/kg and
25 1,600 mg/kg TCE exposures. The authors reported that "hepatic cells had two or more nuclei or
26 had enlarged nuclei containing increased amounts of chromatin, suggesting that a regenerative
27 process was ongoing" and that there were no fine lipid droplets in TCE exposed animals. The
28 finding of "no polyploidy" in control mouse liver in the study of Buben and O'Flaherty (1985) is
29 unexpected given that binucleate and polyploid hepatocytes are a common finding in the mature
30 mouse liver. It is possible that the authors were referring to unusually high instances of
31 "polyploidy" in comparison to what would be expected for the mature mouse. The score given
32 by the authors for polyploidy did not indicate a difference between the two TCE exposure
33 treatments and that it was of the lowest level of severity or occurrence. No score was given for
34 centrolobular hypertrophy although the DNA content and liver weight changes suggested a dose-
35 response. The "Karyrrhexis" described in this study could have been a sign of cell death
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-241 DRAFT—DO NOT CITE OR QUOTE
-------
1 associated with increased liver cell number or dying of maturing hepatocytes associated with the
2 increased ploidy, and suggests that TCE treatment was inducing polyploidization. Consistent
3 with enzyme analyses, centrilobular necrosis was only seen at the highest dose and with the
4 lowest qualitative score, indicating that even at the highest dose there was little toxicity.
5 At high doses, Kaneko et al. (2000) reported sporadic necrosis in male Mrl-lpr/lpr mice,
6 which are "genetically liable to autoimmune disease," exposed to 500 to 2,000 ppm, 4 hours/day,
7 6 days/week, for 8 weeks (n = 5). Dose-dependent mild inflammation and associated changes
8 were reported to be found in the liver. The effects on hepatocytes were reported to be minimal
9 by the authors with 500-ppm TCE inducing sporadic necrosis in the hepatic lobule. Slight
10 mobilization and activation of sinusoid lining cells were also noted. These pathological features
11 were reported to increase with dose.
12 NTP (1990), which used the B6C3F1 mouse strain, reported centrilobular necrosis in
13 6/10 male and 1/10 female B6C3F1 mice treated at a dose of 6,000 mg/kg/d for up to 13 weeks
14 (all the male mice and 8 of the 10 female mice died in the first week of treatment). At
15 3,000 mg/kg/d exposure level, although centrilobular necrosis was not observed, 2/10 males had
16 multifocal areas of calcification in their livers, which the authors suggest is indicative of earlier
17 hepatocellular necrosis. However, only 3/10 male mice at this dose survived to the end of the
18 13-week study.
19 For the NTP (1990) 2-year study, B6C3F1 mice were reported to have no treatment-
20 related increase in necrosis in the liver. A slight increase in the incidence of focal necrosis was
21 noted TCE-exposed male mice (8 vs. 2%) with a slight reduction in fatty metamorphosis in
22 treated male mice (0 treated vs. 2 control animals) and in female mice a slight increase in focal
23 inflammation (29 vs. 19% of animals) and no other changes. Therefore, this study did not show
24 concurrent evidence of liver toxicity with TCE-induced neoplasia after 2 years of TCE exposure
25 in mice.
26 For the more limited database in rats, there appears to be variability in reported TCE
27 induced cytotoxicity and pathology. Nunes et al. (2001) reported no gross pathological changes
28 in rats gavaged with corn oil or with corn oil plus 200 mg/kg TCE for 7 days. Goldsworthy and
29 Popp (1987) gave no descriptions of liver histology given in this report for TCE-exposed animals
30 or corn-oil controls. Kjellstrand et al. (1981) gave also did not give histological descriptions for
31 livers of rats in their inhalation study.
32 Elcombe et al. (1985) provided a description of the histopathology at the light
33 microscopy level in Osborne-Mendel rats, and Alderly Park rats exposed to TCE via gavage for
34 10 days. However, they did not provide a quantitative analysis or specific information regarding
35 the variability of response between animals within group and there was no indication by the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-242 DRAFT—DO NOT CITE OR QUOTE
-------
1 authors regarding how many rats were examined by light microscopy. Hematoxylin and eosin
2 sections from Osborne-Mendel rats were reported to show that
3 Livers from control rats contained large quantities of glycogen and isolated
4 inflammatory foci, but were otherwise normal. The majority of rats receiving
5 1,500 mg/kg body weight TCE showed slight changes in centrilobular
6 hepatocytes. The hepatocytes were more eosinophilic and contained little
7 glycogen. At lower doses these effects were less marked and were restricted to
8 fewer animals. No evidence of treatment-related hepatotoxicity (as exemplified
9 by single cell or focal necrosis) was seen in any rat receiving TCE. H&E
10 [hematoxylin and eosin] sections from Alderly Park Rats showed no signs of
11 treatment-related hepatotoxicity after administration of TCE. However, some
12 signs of dose-related increase in centrilobular eosinophilia were noted.
13
14 Thus, both mice and rats were reported to exhibit pericentral hypertrophy and
15 eosinophilia as noted from the histopathological examination in Elcombe et al. (1985).
16 Berman et al. (1995) reported that for female rats exposed to TCE for 14 days
17 hepatocellular necrosis was noted to occur in the 1,500 and 5,000 mg/kg groups in 6/7 and
18 6/8 female rats, respectively but not to occur in lower doses. The extent of necrosis was not
19 noted by the authors for the two groups exhibiting a response after 1 day of exposure. Serum
20 enzyme levels, indicative of liver necrosis, were not presented and because only positive results
21 were presented in the paper, presumed to be negative. Therefore, the extent of necrosis was not
22 of a magnitude to affect serum enzyme markers of cellular leakage.
23 Melnick et al. (1987) reported that the only treatment-related lesion observed
24 microscopically in rats from either dosed-feed or gavage groups was individual cell necrosis of
25 the liver with the frequency and severity of this lesion similar at each dosage levels of TCE
26 microencapsulated in the feed or administered in corn oil. The severity for necrosis was only
27 mild at the 2.2 and 4.8 g/kg feed groups and for the 6 animals in the 2.8 g/kg group corn oil
28 group. The individual cell necrosis was reported to be randomly distributed throughout the liver
29 lobule with the change to not be accompanied by an inflammatory response. The authors also
30 reported that there was no histologic evidence of cellular hypertrophy or edema in hepatic
31 parenchymal cells. Thus, although there appeared to be TCE-treatment related increases in focal
32 necrosis after 14 days of exposure, the extent was mild even at the highest doses and involved
33 few hepatocytes.
34 For the 13-week NTP study (1990), only control and high dose F344/N rats were
35 examined histologically. Pathological results were reported to reveal that 6/10 males and
36 6/10 female rats had pulmonary vasculitis at the highest concentration of TCE. This change was
37 also reported to have occurred in 1/10 control male and female rats. Most of those animals were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-243 DRAFT—DO NOT CITE OR QUOTE
-------
1 also reported to have had mild interstitial pneumonitis. The authors report that viral liters were
2 positive during this study for Sendai virus.
3 Kumar et al. (2001) reported that male Wistar rats exposed to 376 ppm, 4 hours/day,
4 5 days/week for 8-24 weeks showed evidence of hepatic toxicity. The authors stated that, "after
5 8 weeks of exposure enlarged hepatocytes, with uniform presence of fat vacuoles were found in
6 all of the hepatocytes affecting the periportal, midzonal, and centriolobular areas, and fat
7 vacuoles pushing the pyknosed nuclei to one side of hepatocytes. Moreover, congestion was not
8 significant. After exposure of 12 and 24 weeks, the fatty changes became more progressive with
9 marked necrosis, uniformly distributed in the entire organ." No other description of pathology
10 was provided in this report. In regard to the description of fatty change, the authors only did
11 conventional H&E staining of sections with no precautions to preserve or stain lipids in their
12 sections. However, as noted below, the NCI study also reports long-term TCE exposure in rats
13 to result in hepatocellular fatty metamorphosis. The authors provided a table with histological
14 scoring of simply + or—for minimal, mild or moderate effects and do not define the criteria for
15 that scoring. There is also no quantitative information given as to the extent, nature, or location
16 of hepatocellular necrosis. The authors report "no change was observed in glutamic oxoacetate
17 transaminase and glutamic pyruvated transaminase levels of liver in all the three groups. The
18 GSH level was significantly decreased while "total sulphydryl" level was significantly increased
19 during 8, 12, and 24 weeks of TCE exposure. The acid and alkaline phosphatases were
20 significantly increased during 8, 12, and 24 weeks of TCE exposure." The authors present a
21 series of figures that are poor in quality to demonstrate histopathological TCE-induced changes.
22 No mortality was observed from TCE exposure in any group despite the presence of liver
23 necrosis.
24 Thus, in this limited database that spans durations of exposure from days to 24 weeks and
25 uses differing routes of administration, generally high doses for long durations of exposure are
26 required to induce hepatotoxicity from TCE exposure in the rat. The focus of 2-year bioassays in
27 rats has been the detection of a cancer response with little or no reporting of noncancer pathology
28 in most studies. Henschler et al. (1984) and Fukuda et al. (1983) do not report noncancer
29 histopathology, but do both report rare biliary cell derived tumors in rats in relatively insensitive
30 assays. For male rats, noncancer pathology in the NCI (1976) study was reported to include
31 increased fatty metamorphosis after TCE exposure and angiectasis or abnormally enlarged blood
32 vessels. Angiectasis can be manifested by hyperproliferation of endothelial cells and dilatation
33 of sinusoidal spaces. For the NTP (1990) study there was little reporting of non-neoplastic
34 pathology or toxicity and no report of liver weight at termination of the study. In the NTP
35 (1988) study, the 2 year study of TCE exposure reported no evidence of TCE-induced liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-244 DRAFT—DO NOT CITE OR QUOTE
-------
1 toxicity described as non-neoplastic changes in ACT, August, Marshal, and Osborne-Mendel rats.
2 Interestingly, for the control animals of these four strains there was, in general, a low background
3 level of focal necrosis in the liver of both genders. Obviously, the negative results in this
4 bioassay for cancer are confounded by the killing of a large portion of the animals accidently by
5 experimental error but TCE-induced overt liver toxicity was not reported.
6 In sum, the cytotoxic effects in the liver of TCE treatment appear include little or no
7 necrosis in the rodent liver, but rather, a number of histological changes such as mild focal
8 hepatocyte degeneration at high doses, cellular "swelling" or hypertrophy, and enlarged nuclei.
9 Histological changes consistent with increased polyploidization and specific descriptions of
10 TCE-induced polyploidization have been noted in several experiments. Several studies note
11 proliferation of nonparenchymal cells after TCE exposure as well. These results are more
12 consistently reported in mice, but also have been reported in some studies at high doses in rats,
13 for which fewer studies are available. In addition, the increase in cellular and nuclear sizes
14 appeared to persist after cessation of TCE treatment. In neither rats nor mice is there evidence
15 that TCE treatment results in marked necrosis leading to regenerative hyperplasia.
16
17 4.5.4.3. Measures of DNA Synthesis, Cellular Proliferation, and Apoptosis
18 The increased liver weight observed in rodents after TCE exposure may result from either
19 increased numbers of cells in the liver, increased size of cells in the liver, or a combination of
20 both. Studies of TCE in rodents have studied whole liver DNA content of TCE-treated animals
21 to determine whether the concentration of DNA per gram of liver decreases as an indication of
22 hepatocellular hypertrophy (Buben and O'Flaherty, 1985; Dees and Travis, 1993; Elcombe et al.,
23 1985). While the slight decreases observed in some studies are consistent with hypertrophy, the
24 large variability in controls and lack of dose-response limits the conclusions that can be drawn
25 from these data. In addition, multiple factors beyond hypertrophy affect DNA concentration in
26 whole-liver homogenates, including changes in ploidy and the number of hepatocytes and
27 nonparenchymal cells.
28 The incorporation of tritiated thymidine or BrdU has also been analyzed in whole liver
29 DNA and in individual hepatocytes as a measure of DNA synthesis. Such DNA synthesis can
30 occur from either increased numbers of hepatocytes in the liver or by increased polyploidization.
31 Section E. 1.1 describes polyploidization in human and rodent liver and its impacts on liver
32 function, while Sections E.3.1.2 and E.3.3.1 discuss issues of target cell identification for liver
33 cancer and changes in ploidy as a key even in liver cancer using animals models, respectively.
34 Along with changes in cell size (hypertrophy), cell number (cellular proliferation), and the DNA
35 content per cell (cell ploidy), the rate of apoptosis has also been noted or specifically examined
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-245 DRAFT—DO NOT CITE OR QUOTE
-------
1 in some studies of TCE and its metabolites. All of these phenomena have been identified in
2 proposed hypotheses as key events possibly related to carcinogenicity. In particular, changes in
3 cell proliferation and apoptosis have been postulated to be part of the MOA for PPARa-agonists
4 by Klaunig et al. (2003) (see Section E.3.4).
5 In regard to early changes in DNA synthesis, the data for TCE are very limited
6 Mirsalis et al. (1989) reported measurements of in vivo-in vitro hepatocyte DNA repair and
7 S-phase DNA synthesis in primary hepatocytes from male Fischer-344 rats and male and female
8 B6C3F1 mice administered single doses of TCE by gavage in corn oil. They reported negative
9 results 2-12 hours after treatment from 50-1,000 mg/kg TCE in rats and mice (male and female)
10 for unscheduled DNA synthesis and repair using 3 animals per group. After 24 and 48 hours of
11 200 or 1,000 mg/kg TCE in male mice (n = 3) and after 48 hours of 200 (n = 3) or 1,000 (n = 4)
12 mg/kg TCE in female mice, similar values of 0.30 to 0.69% of hepatocytes were reported as
13 undergoing DNA synthesis in primary culture. Only the 1,000 mg/kg TCE dose in male mice at
14 48 hours was reported to give a result considered to be positive (-2.2% of hepatocytes) but no
15 statistical analyses were performed on these measurements. These results are limited by both the
16 number of animals examined and the relevance of the paradigm.
17 As noted above, TCE treatment in rodents has been reported to result in hepatocellular
18 hypertrophy and increased centrilobular eosinophilia. Elcombe et al. (1985) reported a small
19 decrease in DNA content with TCE treatment (consistent with hepatocellular hypertrophy) that
20 was not dose-related, increased tritiated thymidine incorporation in whole mouse liver DNA that
21 was that was treatment but not dose-related (i.e., a 2-, 2-, and 5-fold of control in mice treated
22 with 500, 1,000, and 1,500 mg/kg TCE), and slightly increased numbers of mitotic figures that
23 were treatment but not dose-related and not correlated with DNA synthesis as measured by
24 thymidine incorporation. Elcombe et al., reported no difference in response between 500 and
25 1,000 mg/kg TCE treatments for tritiated thymidine incorporation. Dees and Travis (1993) also
26 reported that incorporation of tritiated thymidine in DNA from mouse liver was elevated after
27 TCE treatment with the mean peak level of tritiated thymidine incorporation occurred at
28 250 mg/kg TCE treatment level and remaining constant for the 500 and 1,000 mg/kg treated
29 groups. Dees and Travis (1993) specifically report that mitotic figures, although very rare, were
30 more frequently observed after TCE treatment, found most often in the intermediate zone, and
31 found in cells resembling mature hepatocytes. They reported that there was little tritiated
32 thymidine incorporation in areas near the bile duct epithelia or close to the portal triad in liver
33 sections from both male and female mice. Channel et al. (1998) reported proliferating cell
34 nuclear antigen (PCNA) positive cells, a measure of cells that have undergone DNA synthesis,
35 was elevated only on Day 10 (out of the 21 studied) and only in the 1,200 mg/kg/d TCE exposed
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-246 DRAFT—DO NOT CITE OR QUOTE
-------
1 group with a mean of-60 positive nuclei per 1,000 nuclei for 6 mice (-6%). Given that there
2 was little difference in PCNA positive cells at the other TCE doses or time points studied, the
3 small number of affected cells in the liver could not account for the increase in liver size reported
4 in other experimental paradigms at these doses. The PCNA positive cells as well as "mitotic
5 figures" were reported to be present in centrilobular, midzonal, and periportal regions with no
6 observed predilection for a particular lobular distribution. No data were shown regarding any
7 quantitative estimates of mitotic figures and whether they correlated with PCNA results. Thus,
8 whether the DNA synthesis phases of the cell cycle indicated by PCNA staining were
9 indentifying polyploidization or increased cell number cannot be determined.
10 For both rats and mice, the data from Elcombe et al. (1985) showed that tritiated
11 thymidine incorporation in total liver DNA observed after TCE exposure did not correlate with
12 mitotic index activity in hepatocytes. Both Elcombe et al. (1985) and Dees and Travis (1993)
13 reported a small mitotic indexes and evidence of periportal hepatocellular hypertrophy from TCE
14 exposure. Neither mitotic index or tritiated thymidine incorporation data support a correlation
15 with TCE-induced liver weight increase in the mouse, but rather the increase to be most likely
16 due to hepatocellular hypertrophy. If higher levels of hepatocyte replication had occurred
17 earlier, such levels were not sustained by 10 days of TCE exposure. These data suggest that
18 increased tritiated thymidine levels were targeted to mature hepatocytes and in areas of the liver
19 where greater levels of polyploidization occur (see Section E.I.I). Both Elcombe et al. (1985)
20 and Dees and Travis (1993) show that tritiated thymidine incorporation in the liver was ~2-fold
21 greater than controls between 250-1,000 mg/kg TCE, a result consistent with a doubling of
22 DNA. Thus, given the normally quiescent state of the liver, the magnitude of this increase over
23 control levels, even if a result of proliferation rather than polyploidization, would be confined to
24 a very small population of cells in the liver after 10 days of TCE exposure.
25 Laughter et al. (2004) reported that there was an increase in DNA synthesis after aqueous
26 gavage exposure to 500 and 1,000 mg/kg TCE given as 3 boluses a day for 3 weeks with BrdU
27 given for the last week of treatment. An examination of DNA synthesis in individual
28 hepatocytes was reported to show that 1 and 4.5% of hepatocytes had undergone DNA synthesis
29 in the last week of treatment for the 500 and 1,000 mg/kg doses, respectively. Again, this level
30 of DNA synthesis is reported for a small percentage of the total hepatocytes in the liver and not
31 reported to be a result of regenerative hyperplasia.
32 Finally, Dees and Travis (1993) and Channel et al. (1998) reported evaluating changes in
33 apoptosis with TCE treatment. Dees and Travis (1993) enumerated identified by either
34 hematoxylin and eosin or feulgen staining in male and female mice after 10 days of TCE
35 treatment by. Only 0 or 1 apoptosis was observed per 100 high power (400x) fields in controls
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-247 DRAFT—DO NOT CITE OR QUOTE
-------
1 and all dose groups except for those given 1,000 mg/kg/d, in which 8 or 9 apoptoses per
2 100 fields were reported. None of the apoptoses were in the intermediate zones where mitotic
3 figures were observed, and all were located near the central veins. This is the same region where
4 one would expect endogenous apoptoses as hepatocytes "stream" from the portal triad toward the
5 central vein (Schwartz-Arad, 1989). In addition, this is the same region where Buben and
6 O'Flaherty (1985) noted necrosis and polyploidy. By contrast Channel et al. (1998) reported no
7 significant differences in apoptosis at any treatment dose (400 to 1,200 mg/kg/d) examined after
8 any time from 2 days to 4 weeks.
9
10 4.5.4.4. Peroxisomal Proliferation and Related Effects
11 Numerous studies have reported that TCE administered to mice and rats by gavage leads
12 to proliferation of peroxisomes in hepatocytes. Some studies have measured changes in the
13 volume and number of peroxisomes as measures of peroxisome proliferation while others have
14 measured peroxisomal enzyme activity such catalase and cyanide-insensitive PCO. Like liver
15 weight, the determination of a baseline level of peroxisomal volume, number, or enzyme activity
16 can be variable and have great effect on the ability to determine the magnitude of a treatment-
17 related effect.
18 Elcombe et al. (1985) reported increases in the percent of the cytoplasm occupied by
19 peroxisomes in B6C3F1 and Alderley Park mice treated for 10 days at 500 to 1,500 mg/kg/d.
20 Although the increase over controls appeared larger in the B6C3F1 strain, this is largely due to
21 the 2-fold smaller control levels in that strain, as the absolute percentage of peroxisomal volume
22 was similar between strains after treatment. All these results showed high variability, as
23 evidenced from the reported standard deviations. Channel et al. (1998) found a similar absolute
24 percentage of peroxisomal volume after 10 days treatment in the B6C3F1 mouse at
25 1,200 mg/kg/d TCE but with the percentage in vehicle controls similar to the Alderley-Park mice
26 in the Elcombe et al. (1985) study. Interestingly, Channel et al. (1998) found that the increase in
27 peroxisomes peaked at 10 days, with lower values after 6 and 14 days of treatment.
28 Furthermore, the vehicle control levels also varied almost 2-fold depending on the number of
29 days of treatment. Nakajima et al. (2000), who treated male wild-type SV129 mice at
30 750 mg/kg/d for 14 days, found even higher baseline values for the percentage of peroxisomal
31 volume, but with an absolute level after treatment similar to that reported by Channel et al.
32 (1998) in B6C3F1 mice treated at 1,200 mg/kg/d TCE for 14 days. Nakajima et al. (2000) also
33 noted that the treatment-related increases were smaller for female wild-type mice, and that there
34 were no increases in peroxisomal volume in male or female PPARa-null mice, although vehicle
35 control levels were slightly elevated (not statistically significant). Only Elcombe et al. (1985)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-248 DRAFT—DO NOT CITE OR QUOTE
-------
1 examined peroxisomal volume in rats, and reported smaller treatment-related increases in two
2 strains (OM and AP), but higher baseline levels. In particular, at 1,000 mg/kg/d, after 10 days
3 treatment, the percent peroxisomal volume was similar in OM and AP rats, with similar control
4 levels as well. While the differences from treatment were not statistically significant, only five
5 animals were used in each group, and variability, as can be seen by the standard deviations, was
6 high, particularly in the treated animals.
7 The activities of a number of different hepatic enzymes have also been as markers for
8 peroxisome proliferation and/or activation of PPARa. The most common of these are catalase
9 and cyanide-insensitive PCO. In various strains of mice (B6C3F1, Swiss albino, SV129 wild-
10 type) treated at doses of 500 to 2,000 mg/kg/d for 10 to 28 days, increases in catalase activity
11 have tended to be more modest (1.3- to 1.6-fold of control) as compared to increases in PCO
12 (1.4- to 7.9-fold of control) (Elcombe et al., 1985; Goel et al., 1992; Goldsworthy and Popp,
13 1987; Laughter et al., 2004; Nakajima et al., 2000; Watanabe and Fukui, 2000). In rats, Elcombe
14 et al. (1985) reported no increases in catalase or PCO activity in Alderley-Park rats treated at
15 1,000 mg/kg/d TCE for 10 days. In F344 rats, Goldsworthy and Popp (1987) and Melnick et al.
16 (1987) reported increases of up to 2-fold in catalase and 4.1-fold in PCO relative to controls
17 treated at 600 to 4,800 mg/kg/d for 10 to 14 days. The changes in catalase were similar to those
18 in mice at similar treatment levels, with 1.1 - to 1.5-fold of control enzyme activities at doses of
19 1,000 to 1,300 mg/kg/d (Elcombe et al., 1985; Melnick et al., 1987). However, the changes in
20 PCO were smaller, with 1.1- to 1.8-fold of control activity at these doses, as compared to 6.3- to
21 7.9-fold of control in mice (Goldsworthy and Popp, 1987; Melnick et al., 1987).
22 In SV129 mice, Nakajima et al. (2000) and Laughter et al. (2004) investigated the
23 dependence of these changes on PPARa by using a null mouse. Nakajima et al. reported that
24 neither male nor female wild-type or PPARa null mice had significant increases in catalase after
25 14 days of treatment at 750 mg/kg/d. However, given the small number of animals (4 per group)
26 and the relatively small changes in catalase observed in other (wild-type) strains of mice, this
27 study had limited power to detect such changes. Several other markers of peroxisome
28 proliferation, including acyl-CoA oxidase and CYP4A1 (PCO was not investigated), were
29 induced by TCE in male wild-type mice, but not in male null mice or female mice of either type.
30 Unfortunately, none of these markers have been investigated using TCE in female mice of any
31 other strain, so it is unclear whether the lack of response is characteristic of female mice in
32 general, or just in this strain. Interestingly, as noted above, liver/body weight ratio increases
33 were observed in both sexes of the null mice in this study. Laughter et al. (2004) only quantified
34 activity of the peroxisome proliferation marker PCO in their study, and found in null mice a
35 slight decrease (0.8-fold of control) at 500 mg/kg/d TCE and an increase (1.5-fold of control) at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-249 DRAFT—DO NOT CITE OR QUOTE
-------
1 1,500 mg/kg/d TCE after 3 weeks of treatment, with neither statistically significant (4-5 mice
2 per group). However, baseline levels of PCO were almost 2-fold higher in the null mice, and the
3 treated wild-type and null mice differed in PCO activity by only about 1.5-fold.
4 In sum, oral administration of TCE for up to 28 days causes proliferation of peroxisomes
5 in hepatocytes along with associated increases in peroxisomal enzyme activities in both mice and
6 rats. Male mice tend to be more sensitive in that at comparable doses, rats and female mice tend
7 to exhibit smaller responses. For example, for peroxisomal volume and PCO, the fold-increase
8 in rats appears to be lower by 3- to 6-fold than that in mice, but, for catalase, the changes were
9 similar between mice in F344 rats. No inhalation or longer-term studies were located, and only
10 one study examined these changes at more than one time-point. Therefore, little is known about
11 the route-dependence, time course, and persistence of these changes. Finally, two studies in
12 PPARa-null mice (Laughter et al., 2004; Nakajima et al., 2000) found diminished responses in
13 terms of increased peroxisomal volume and peroxisomal enzyme activities as compared to
14 wild-type mice, although there was some confounding due to baseline differences between null
15 and wild-type control mice in several measures.
16
17 4.5.4.5. Oxidative Stress
18 Several studies have attempted to study the possible effects of "oxidative stress" and
19 DNA damage resulting from TCE exposures. The effects of induction of metabolism by TCE, as
20 well as through coexposure to ethanol, have been hypothesized to in itself increase levels of
21 "oxidative stress" as a common effect for both exposures (see Sections E.3.4.2.3 and E.4.2.4).
22 Oxidative stress has been hypothesized to be a key event or MOA for peroxisome proliferators as
23 well, but has been found to neither be correlated with cell proliferation nor carcinogenic potency
24 of peroxisome proliferators (see Section E.3.4.1.1). As a MOA, it is not defined or specific as
25 the term "oxidative stress" is implicated as part of the pathophysiologic events in a multitude of
26 disease processes and is part of the normal physiologic function of the cell and cell signaling.
27 In regard to measures of oxidative stress, Rusyn et al. (2006) noted that although an
28 overwhelming number of studies draw a conclusion between chemical exposure, DNA damage,
29 and cancer based on detection of SOHdG, a highly mutagenic lesion, in DNA isolated from
30 organs of in vivo treated animals, a concern exists as to whether increases in SOHdG represent
31 damage to genomic DNA, a confounding contamination with mitochondrial DNA, or an
32 experimental artifact. As noted in Sections E.2.1.1 and E.2.2.11, studies of TCE which employ
33 the i.p. route of administration can be affected by inflammatory reactions resulting from that
34 routes of administration and subsequent toxicity that can involve oxygen radical formation from
35 inflammatory cells. Finally, as described in Section E.2.2.8, the study by Channel et al. (1998)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-250 DRAFT—DO NOT CITE OR QUOTE
-------
1 demonstrated that corn oil as vehicle had significant effects on measures of "oxidative stress"
2 such as TEARS.
3 The TEARS results presented by Channel et al. (1988) indicate suppression of TEARS
4 with increasing time of exposure to corn oil alone with data presented in such a way for SOHdG
5 and total free radical changes that the pattern of corn oil administration was obscured. It was not
6 apparent from that study that TCE exposure induced oxidative damage in the liver.
7 Toraason et al. (1999) measured SOHdG and a "free radical-catalyzed isomer of
8 arachidonic acid and marker of oxidative damage to cell membranes, 8-Epi-prostaglandin F2a
9 (8-epiPGF)", excretion in the urine and TEARS (as an assessment of malondialdehyde and
10 marker of lipid peroxidation) in the liver and kidney of male Fischer rats exposed to single i.p.
11 injections in of TCE in Alkamuls vehicle. Using this paradigm, 500-mg/kg TCE was reported to
12 induce Stage II anesthesia and a 1,000 mg/kg TCE to induce Level III or IV (absence of reflex
13 response) anesthesia and burgundy colored urine with 2/6 rats at 24 hours comatose and
14 hypothermic. The animals were sacrificed before they could die and the authors suggested that
15 they would not have survived another 24 hours. Thus, using this paradigm there was significant
16 toxicity and additional issues related to route of exposure. Urine volume declined significantly
17 during the first 12 hours of treatment and while water consumption was not measured, it was
18 suggested by the authors to be decreased due to the moribundity of the rats. Given that this study
19 examined urinary markers of "oxidative stress" the effects on urine volume and water
20 consumption, as well as the profound toxicity induced by this exposure paradigm, limit the
21 interpretation of the study. The issues of bias in selection of the data for this analysis, as well as
22 the issues stated above for this paradigm limit interpretation of these data while the authors
23 suggest that evidence of oxidative damage was equivocal.
24
25 4.5.4.6. Bile Production
26 Effects of TCE exposure in humans and in experimental animals is presented in
27 Section E.2.6. Serum bile acids (SEA) have been suggested as a sensitive indicator of
28 hepatotoxicity to a variety of halogenated solvents with an advantage of increased sensitivity and
29 specificity over conventional liver enzyme tests that primarily reflect the acute perturbation of
30 hepatocyte membrane integrity and "cell leakage" rather than liver functional capacity (i.e.,
31 uptake, metabolism, storage, and excretion functions of the liver) (Bai et al., 1992b; Neghab et
32 al., 1997). While some studies have reported negative results, a number of studies have reported
33 elevated SEA in organic solvent-exposed workers in the absence of any alterations in normal
34 liver function tests. These variations in results have been suggested to arise from failure of some
35 methods to detect some of the more significantly elevated SEA and the short-lived and reversible
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-251 DRAFT—DO NOT CITE OR QUOTE
-------
1 nature of the effect (Neghab et al., 1997). Neghab et al. (1997) have reported that occupational
2 exposure to l,l,2-trichloro-l,2,2-trifluoroethane and trichloroethylene has resulted in elevated
3 SBA and that several studies have reported elevated SBA in experimental animals to chlorinated
4 solvents such as carbon tetrachloride, chloroform, hexachlorobutadiene, tetrachloroethylene,
5 1,1,1-trichloroethane, and trichloroethylene at levels that do not induce hepatotoxicity (Bai et al.,
6 1992a, b; Hamdan and Stacey, 1993; Wang and Stacey, 1990). Toluene, a nonhalogenated
7 solvent, has also been reported to increase SBA in the absence of changes in other hepatobiliary
8 functions (Neghab and Stacey, 1997). Thus, disturbance in SBA appears to be a generalized
9 effect of exposure to chlorinated solvents and nonchlorinated solvents and not specific to TCE
10 exposure.
11 Wang and Stacey (1990) administered TCE in corn oil via i.p. injection to male
12 Sprague-Dawley rats with liver enzymes and SBA examined 4 hours after the last TCE
13 treatment. The limitations of i.p injection experiments have already been discussed. While
14 reporting no overt liver toxicity there was, generally, a reported dose-related increase in cholic
15 acid, chenodeoxycholic acid, deoxycholic acid, taurocholic acid, tauroursodeoxycholic acid with
16 cholic acid and taurochlolic acid increased at the lowest dose. The authors report that
17 "examination of liver sections under light microscopy yielded no consistent effects that could be
18 ascribed to trichloroethylene." In the same study a rats were also exposed to TCE via and using
19 this paradigm, cholic acid and taurocholic acid were also significantly elevated but the large
20 variability in responses between rats and the low number of rats tested in this paradigm limit its
21 ability to determine quantitative differences between groups. Nevertheless, without the
22 complications associated with i.p. exposure, inhalation exposure of TCE at relatively low
23 exposure levels that were not associated with other measures of toxicity were associated with
24 increased SBA level.
25 Hamdan et al. (1993) administered TCE in corn oil (1 mmol/kg) in male Sprague-Dawley
26 rats and followed the time-course of SBA elevation, TCE concentration, and trichloroethanol in
27 the blood up to 16 hours. Liver and blood concentration of TCE were reported to peak at 4 hours
28 while those of trichloroethanol peaked at 8 hours after dosing. TCE levels were not detectable
29 by 16 hours in either blood or liver while those of trichloroethanol were still elevated.
30 Elevations of SBA were reported to parallel those of TCE with cholic acid and taurochloate acid
31 reported to show the highest levels of bile acids. The authors state that liver injury parameters
32 were checked and found unaffected by TCE exposure but did not show the data. Thus, it was
33 TCE concentration and not that of its metabolite that was most closely related to changes in SBA
34 and after a single exposure and the effect appeared to be reversible. In an in vitro study by Bai
35 and Stacey (1993), TCE was studied in isolated rat hepatocytes with TCE reported to cause a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-252 DRAFT—DO NOT CITE OR QUOTE
-------
1 dose-related suppression of initial rates of cholic acid and taurocholic acid but with no significant
2 effects on enzyme leakage and intracellular calcium contents, further supporting a role for the
3 parent compound in this effect.
4
5 4.5.4.7. Summary: Trichloroethylene (TCE)-InducedNoncancer Effects in Laboratory
6 Animals
1 In laboratory animals, TCE leads to a number of structural changes in the liver, including
8 increased liver weight, small transient increases in DNA synthesis, cytomegaly in the form of
9 "swollen" or enlarged hepatocytes, increased nuclear size probably reflecting polyploidization,
10 and proliferation of peroxisomes. Liver weight increases proportional to TCE dose are
11 consistently reported across numerous studies, and appear to be accompanied by periportal
12 hepatocellular hypertrophy. There is also evidence of increased DNA synthesis in a small
13 portion of hepatocytes at around 10 days in vivo exposure. The lack of correlation of
14 hepatocellular mitotic figures with whole liver DNA synthesis or DNA synthesis observed in
15 individual hepatocytes supports the conclusion that cellular proliferation is not the predominant
16 cause of increased DNA synthesis. The lack of correlation of whole liver DNA synthesis and
17 those reported for individual hepatocytes suggests that nonparenchymal cells also contribute to
18 such synthesis. Indeed, nonparenchymal cell activation or proliferation has been noted in several
19 studies. Moreover, the histological descriptions of TCE exposed liver are consistent with and in
20 some cases specifically note increased polyploidy after TCE exposure. Interestingly, changes in
21 TCE-induced hepatocellular ploidy, as indicated by histological changes in nuclei, have been
22 noted to remain after the cessation of exposure. In regard to apoptosis, TCE has been reported to
23 either not change apoptosis or to cause a slight increase at high doses. Some studies have also
24 noted effects from dosing vehicle alone (such as corn oil in particular) not only on liver
25 pathology, but also on DNA synthesis.
26 Available data also suggest that TCE does not induce substantial cytotoxicity, necrosis, or
27 regenerative hyperplasia, as only isolated, focal necroses and mild to moderate changes in serum
28 and liver enzyme toxicity markers having been reported. Data on peroxisome proliferation,
29 along with increases in a number of associated biochemical markers, show effects in both mice
30 and rats. These effects are consistently observed across rodent species and strains, although the
31 degree of response at a given mg/kg/d dose appears to be highly variability across strains, with
32 mice on average appearing to be more sensitive.
33 In addition, like humans, laboratory animals exposed to TCE have been observed to have
34 increased serum bile acids, though the toxicologic importance of these effects is unclear.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-253 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.5.5. Trichloroethylene (TCE)-Induced Liver Cancer in Laboratory Animals
2 For 2-year or lifetime studies of TCE exposure a consistent hepatocarcinogenic response
3 has been observed using mice of differing strains and genders and from differing routes of
4 exposure. However, some rat studies have been confounded by mortality from gavage error or
5 the toxicity of the dose of TCE administered. In some studies, a relative insensitive strain of rat
6 has been used. However, in general it appears that the mouse is more sensitive than the rat to
7 TCE-induced liver cancer. Three studies give results the authors consider to be negative for
8 TCE-induced liver cancer in mice, but have either design and/or reporting limitations, or are in
9 strains and paradigms with apparent low ability for liver cancer induction or detection. Findings
10 from these studies are shown in Tables 4-52 through 4-57, and discussed below.
11
12 4.5.5.1. Negative or Inconclusive Studies of Mice and Rats
13 Fukuda et al. (1983) reported a 104-week inhalation bioassay in female Crj:CD-l (ICR)
14 mice and female Crj:CD (SD) rats exposed to 0-, 50-, 150-, and 450-ppm TCE (n = 50). There
15 were no reported incidences of mice or rats with liver tumors for controls indicative of relatively
16 insensitive strains and gender used in the study for liver effects. While TCE was reported to
17 induce a number of other tumors in mice and rats in this study, the incidence of liver tumors was
18 less than 2% after TCE exposure. Of note is the report of cystic cholangioma reported in 1 group
19 of rats.
20 Henschler et al. (1980) exposed NMRI mice and WIST random bred rats to 0-, 100-, and
21 500-ppm TCE for 18 months (n = 30). Control male mice were reported to have one
22 hepatocellular carcinoma and 1 hepatocellular adenoma with the incidence rate unknown. In the
23 100-ppm TCE exposed group, 2 hepatocellular adenomas and 1 mesenchymal liver tumor were
24 reported. No liver tumors were reported at any dose of TCE in female mice or controls. For
25 male rats, only 1 hepatocellular adenomas at 100 ppm was reported. For female rats no liver
26 tumors were reported in controls, but 1 adenoma and 1 cholangiocarcinoma was reported at
27 100-ppm TCE and at 500-ppm TCE, 2 cholangioadenomas, a relatively rare biliary tumor, was
28 reported. The difference in survival in mice, did not affect the power to detect a response, as was
29 the case for rats. However, the low number of animals studied, abbreviated exposure duration,
30 low survival in rats, and absent background response (suggesting low intrinsic sensitivity to this
31 endpoint) suggest a study of limited ability to detect a TCE carcinogenic liver response. Of note
32 is that despite their limitations, both Fukuda et al. (1983) and Henschler et al. (1980) report rare
33 biliary cell derived tumors in TCE-exposed rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-254 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-52. Summary of liver tumor findings in gavage studies of
trichloroethylene by NTP (1990)a
Sex
Dose (mg/kg)b
Adenoma
(overall; terminal0)
Adenocarcinoma
(overall; terminal0)
1/d, 5 d/wk, 103-wk study, F344/N rats
Male
Female
0
500
1,000
0
500
1,000
NAd
NA
NA
NA
NA
NA
0/49
0/49
1/49
0/50
1/48
1/48
1/d, 5 d/wk, 103-wk study, B6C3Fi mice
Male
Female
0
1,000
0
1,000
7/48; 6/33
14/50; 6/16
4/48; 4/32
16/49; ll/23e
8/48; 6/33
31/50; 14/16f
2/48; 2/32
13/49; 8/23g
4
5
6
7
8
9
10
11
"Liver tumors not examined in 13-week study, so data shown only for 103-week study.
bCorn oil vehicle.
°Terminal values not available for rats.
Data not available.
e/> 0.003.
/? 0.001.
gp <0.002.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-255 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
4
5
6
7
Table 4-53. Summary of liver tumor findings in gavage studies of
trichloroethylene by NCI (1976)
Sex
Dose (mg/kg)a
Hepatocarcinoma
1/d, 5 d/wk, 2-yr study, Osborn-Mendel rats
Males
Females
0
549
1,097
0
549
1,097
0/20
0/50
0/50
0/20
1/48
0/50
1/d, 5 d/wk, 2-yr study, B6C3F1 mice
Males
Females
0
1,169
2,339
0
869
1,739
1/20
26/50b
31/48b
0/20
4/50
1 l/47b
""Treatment period was 48 weeks for rats, 66 weeks for mice. Doses were changed several times during the
study based on monitoring of body weight changes and survival. Dose listed here is the time-weighted
average dose over the days on which animals received a dose.
V
-------
1
2
Table 4-54. Summary of liver tumor incidence in gavage studies of
trichloroethylene by NTP (1988)
Sex
Dose (mg/kg)*
Adenoma
Adenocarcinoma
1/d, 5 d/wk, 2-yr study, ACT rats
Male
Female
0
500
1,000
0
500
1,000
0/50
0/49
0/49
0/49
0/46
0/39
1/50
1/49
1/49
2/49
0/46
0/39
1/d, 5 d/wk, 2-yr study, August rats
Male
Female
0
500
1,000
0
500
1,000
0/50
0/50
0/48
0/48
0/48
0/50
0/50
1/50
1/48
2/48
0/48
0/50
1/d, 5 d/wk, 2-yr study, Marshall rats
Male
Female
0
500
1,000
0
500
1,000
1/49
0/50
0/47
0/49
0/48
0/46
1/49
0/50
1/47
0/49
0/48
0/46
1/d, 5 d/wk, 2-yr study, Osborne-Mendel rats
Male
Female
0
500
1,000
0
500
1,000
1/50
1/50
1/49
0/50
0/48
0/49
1/50
0/50
2/49
0/50
2/48
2/49
4
5
*Corn oil vehicle.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-257 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-55. Summary of liver tumor findings in inhalation studies of
trichloroethylene by Maltoni et al. (1988)a
4
5
6
7
8
9
10
Sex
Concentration (ppm)
Hepatoma
7 h/d, 5 d/wk, 8-wk exposure, observed for lifespan, Swiss mice
Male
Female
0
100
600
0
100
600
1/100
3/60
4/72
1/100
1/60
0/72
7 h/d, 5 d/wk, 78-wk exposure, observed for lifespan, Swiss mice
Male
Female
0
100
300
600
0
100
300
600
4/90
2/90
8/90
13/90
0/90
0/90
0/90
1/90
7 h/d, 5 d/wk, 78-wk exposure, observed for lifespan, B6C3F1 miceb
Male
Female
0
100
300
600
0
100
300
600
1/90
1/90
3/90
6/90
3/90
4/90
4/90
9/90
aThree inhalation experiments in this study found no hepatomas: BT302 (8-week exposure to 0, 100, 600 ppm
in Sprague-Dawley rats); BT303 (8-week exposure to 0, 100, or 600 ppm in Swiss mice); and BT304
(78-week exposure to 0, 100, 300, or 600 ppm in Sprague-Dawley rats).
bFemale incidences are from experiment BT306, while male incidences are from experiment BT306bis,
which was added to the study because of high, early mortality due to aggressiveness and fighting in males
in experiment BT306.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-258 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-56. Summary of liver tumor findings in inhalation studies of
trichloroethylene by Henschler et al. (1980)a and Fukuda et al. (1983)
Sex
Concentration (ppm)
Adenomas
Adenocarcinomas
6 h/d, 5 d/wk, 18-mo exposure, 30-mo observation, Han:NMRI mice (Henschler et al., 1980)
Males
Females
0
100
500
0
100
500
1/30 b
2/29 b
0/29
0/29
0/30
0/28
1/30
0/30
0/30
0/29
0/30
0/28
6 h/d, 5 d/wk, 18-mo exposure, 36-mo observation, Han:WIST rats (Henschler et al., 1980)
Males
Females
0
100
500
0
100
500
1/29
1/30
0/30
0/28
1/30
2/30
0/29
0/30
0/30
0/28
1/30
0/30
7 h/d, 5 d/wk, 2-yr study, Crj:CD (SD) rats (Fukuda et al., 1983)
Females
0
50
150
450
0/50
1/50
0/47
0/51
0/50
0/50
0/47
1/50
7 h/d, 5 d/wk, 2-yr study, Crj:CD (ICR) mice (Fukuda et al., 1983)
Females
0
50
150
450
0/49
0/50
0/50
1/46
0/49
0/50
0/50
0/46
4
5
6
"Henschler et al. (1980) observed no liver tumors in control or exposed Syrian hamsters.
bOne additional hepatic tumor of undetermined class not included.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-259 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-57. Summary of liver tumor findings in gavage studies of
trichloroethylene by Henschler et al. (1984)a
Sex
(TCE cone.)
TCE (Stabilizers if present)
Benignb
Malignant0
5 d/wk, 18-mo exposure, 24-mo observation, Swiss mice (Henschler et al., 1984)
Males
(2.4g/kgBW)
Females
(1.8g/kgBW)
Control (none)
TCE (triethanolamine)
TCE (industrial)
TCE (epichlorohydrin (0.8%))
TCE (1,2-epoxybutane (0.8%))
TCE (both epichlorohydrin (0.25%)
and 1,2-epoxybutane (0.25%))
Control (none)
TCE (triethanolamine)
TCE (industrial)
TCE (epichlorohydrin (0.8%))
TCE (1,2-epoxybutane (0.8%))
TCE (both epichlorohydrin (0.25%)
and 1,2-epoxybutane (0.25%))
5/50
7/50
9/50
3/50
4/50
5/50
1/50
7/50
9/50
3/50
2/50
4/50
0/50
0/50
0/50
1/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
aHenschler et al. (1984) Due to poor condition of the animals resulting from the nonspecific toxicity of high doses of
TRI and/or the additives, gavage was stopped for all groups during weeks 35-40, 65 and 69-78, and all doses were
reduced by a factor of 2 from the 40th week on.
blncludes hepatocellular adenomas, hemangioendothelioma, cholangiocellular adenoma.
Includes hepatocellular carcinoma, malignant hemangiosarcoma, cholangiocellular carcinoma.
Cone. = concentration.
Van Duuren et al. (1979), exposed mice to 0.5 mg/mouse to TCE via gavage once a week
in 0.1 mL trioctanion (n = 30). Inadequate design and reporting of this study limit that ability to
use the results as an indicator of TCE carcinogenicity.
The NCI (1976) study of TCE was initiated in 1972 and involved the exposure of
Osborn-Mendel rats to varying concentrations of TCE. A low incidence of liver tumors was
reported for controls and carbon tetrachloride positive controls in rats from this study. The
authors concluded that due to mortality, "the test is inconclusive in rats." They note the
insensitivity of the rat strain used to the positive control of carbon tetrachloride exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-260 DRAFT—DO NOT CITE OR QUOTE
-------
1 The NTP (1990) study of TCE exposure in male and female F344/N rats, and B6C3F1
2 mice (500 and 1,000 mg/kg for rats) is limited in the ability to demonstrate a dose-response for
3 hepatocarcinogenicity. For rats, the NTP (1990) study reported no treatment-related non-
4 neoplastic liver lesions in males and a decrease in basophilic cytological change reported from
5 TCE-exposure in female rats. The results for detecting a carcinogenic response in rats were
6 considered to be equivocal because both groups receiving TCE showed significantly reduced
7 survival compared to vehicle controls and because of a high rate (e.g., 20% of the animals in the
8 high-dose group) of death by gavage error.
9 The NTP (1988) study of TCE exposure in four strains of rats to "diisopropylamine-
10 stabilized TCE" was also considered inadequate for either comparing or assessing TCE-induced
11 liver carcinogenesis in these strains of rats because of chemically induced toxicity, reduced
12 survival, and incomplete documentation of experimental data. TCE gavage exposures of 0, 500,
13 or 1,000 mg/kg/d (5 days/week, for 103 weeks) male and female rats was also marked by a large
14 number of accidental deaths (e.g., for high-dose male Marshal rats 25 animals were accidentally
15 killed).
16 Maltoni et al. (1986) reported the results of several studies of TCE via inhalation and
17 gavage in mice and rats. A large number of animals were used in the treatment groups but the
18 focus of the study was detection of a neoplastic response with only a generalized description of
19 tumor pathology phenotype given and limited reporting of non-neoplastic changes in the liver.
20 Accidental death by gavage error was reported not to occur in this study. In regards to effects of
21 TCE exposure on rat survival, "a nonsignificant excess in mortality correlated to TCE treatment
22 was observed only in female rats (treated by ingestion with the compound)".
23 For rats, Maltoni et al. (1986) reported 4 liver angiosarcomas (1 in a control male rat,
24 1 both in a TCE-exposed male and female at 600 ppm TCE for 8 weeks, and 1 in a female rat
25 exposed to 600-ppm TCE for 104 weeks), but the specific results for incidences of hepatocellular
26 "hepatomas" in treated and control rats were not given. Although the Maltoni et al. (1986)
27 concluded that the small number was not treatment related, the findings were brought forward
28 because of the extreme rarity of this tumor in control Sprague-Dawley rats, untreated or treated
29 with vehicle materials. In rats treated for 104 weeks, there was no report of a TCE treatment-
30 related increase in liver cancer in rats. This study only presented data for positive findings so it
31 did not give the background or treatment-related findings in rats for liver tumors in this study.
32 Thus, the extent of background tumors and sensitivity for this endpoint cannot be determined.
33 Of note is that the Sprague-Dawley strain used in this study was also noted in the Fukuda et al.
34 (1983) study to be relatively insensitive for spontaneous liver cancer and to also be negative for
35 TCE-induced hepatocellular liver cancer induction in rats. However, like Fukuda et al. (1983)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-261 DRAFT—DO NOT CITE OR QUOTE
-------
1 and Henschler et al. (1980), that reported rare biliary tumors in insensitive strains of rat for
2 hepatocellular tumors, Maltoni et al. (1986) reported a relatively rare tumor type, angiosarcoma,
3 after TCE exposure in a relatively insensitive strain for "hepatomas." As noted above, many of
4 the rat studies were limited by premature mortality due to gavage error or premature mortality
5 (Henschler et al., 1980; NCI, 1976; NTP, 1990, 1988), which was reported not occur in
6 Maltoni etal. (1986).
7
8 4.5.5.2. Positive Trichloroethylene (TCE) Studies of Mice
9 In the NCI (1976) study of TCE exposure in B6C3F1 mice, TCE was reported to increase
10 incidence of hepatocellular carcinomas in both doses and both genders of mice (-1,170 and
11 2,340 mg/kg for males and 870 and 1,740 mg/kg for female mice). Hepatocellular carcinoma
12 diagnosis was based on histologic appearance and metastasis to the lung. The tumors were
13 described in detail and to be heterogeneous "as described in the literature" and similar in
14 appearance to tumors generated by carbon tetrachloride. The description of liver tumors in this
15 study and tendency to metastasize to the lung are similar to descriptions provided by
16 Maltoni et al. (1986) for TCE-induced liver tumors in mice via inhalation exposure.
17 The NTP (1990) study of TCE exposure in male and female B6C3F1 mice (1,000 mg/kg
18 for mice) reported decreased latency of liver tumors, with animals first showing carcinomas at
19 57 weeks for TCE-exposed animals and 75 weeks for control male mice. The administration of
20 TCE was also associated with increased incidence of hepatocellular carcinoma (tumors with
21 markedly abnormal cytology and architecture) in male and female mice. Hepatocellular
22 adenomas were described as circumscribed areas of distinctive hepatic parenchymal cells with a
23 perimeter of normal appearing parenchyma in which there were areas that appeared to be
24 undergoing compression from expansion of the tumor. Mitotic figures were sparse or absent but
25 the tumors lacked typical lobular organization. Hepatocellular carcinomas had markedly
26 abnormal cytology and architecture with abnormalities in cytology cited as including increased
27 cell size, decreased cell size, cytoplasmic eosinophilia, cytoplasmic basophilia, cytoplasmic
28 vacuolization, cytoplasmic hyaline bodies, and variations in nuclear appearance. Furthermore, in
29 many instances several or all of the abnormalities were present in different areas of the tumor
30 and variations in architecture with some of the hepatocellular carcinomas having areas of
31 trabecular organization. Mitosis was variable in amount and location. Therefore, the phenotype
32 of tumors reported from TCE exposure was heterogeneous in appearance between and within
33 tumors. However, because it consisted of a single-dose group in addition to controls, this study
34 is limited of limited utility for analyzing the dose-response for hepatocarcinogenicity. There was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-262 DRAFT—DO NOT CITE OR QUOTE
-------
1 also little reporting of non-neoplastic pathology or toxicity and no report of liver weight at
2 termination of the study.
3 Maltoni et al. (1986) reported the results of several studies of TCE in mice. A large
4 number of animals were used in the treatment groups but the focus of the study was detection of
5 a neoplastic response with only a generalized description of tumor pathology phenotype given
6 and limited reporting of non-neoplastic changes in the liver. There was no accidental death by
7 gavage error reported to occur in mice but, a "nonsignificant" excess in mortality correlated to
8 TCE treatment was observed in male B6C3F1 mice. TCE-induced effects on body weight were
9 reported to be absent in mice except for one experiment (BT 306 bis) in which a slight nondose
10 correlated decrease was found in exposed animals. "Hepatoma" was the term used to describe
11 all malignant tumors of hepatic cells, of different subhistotypes, and of various degrees of
12 malignancy and were reported to be unique or multiple, and have different sizes (usually
13 detected grossly at necropsy) from TCE exposure. In regard to phenotype tumors were described
14 as usual type observed in Swiss and B6C3F1 mice, as well as in other mouse strains, either
15 untreated or treated with hepatocarcinogens and to frequently have medullary (solid), trabecular,
16 and pleomorphic (usually anaplastic) patterns. Swiss mice from this laboratory were reported to
17 have a low incidence of hepatomas without treatment (1%). The relatively larger number of
18 animals used in this bioassay (n = 90 to 100), in comparison to NTP standard assays, allows for a
19 greater power to detect a response.
20 TCE exposure for 8 weeks via inhalation at 100 or 600 ppm may have been associated
21 with a small increase in liver tumors in male mice in comparison to concurrent controls during
22 the life span of the animals. In Swiss mice exposed to TCE via inhalation for 78 weeks, there a
23 reported increase in hepatomas associated with TCE treatment that was dose-related in male but
24 not female Swiss mice. In B6C3F1 mice exposed via inhalation to TCE for 78 weeks, increases
25 in hepatomas were reported in both males and females. However, the experiment in males was
26 repeated with B6C3F1 mice from a different source, since in the first experiment more than half
27 of the mice died prematurely due to excessive fighting. Although the mice in the two
28 experiments in males were of the same strain, the background level of liver cancer was
29 significantly different between mice from the different sources (1/90 versus 19/90), though the
30 early mortality may have led to some censoring. The finding of differences in response in
31 animals of the same strain but from differing sources has also been reported in other studies for
32 other endpoints. However, for both groups of male B6C3F1 mice the background rate of liver
33 tumors over the lifetime of the mice was no greater than about 20%.
34 There were other reports of TCE carcinogenicity in mice from chronic exposures that
35 were focused primarily on detection of liver tumors with limited reporting of tumor phenotype or
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-263 DRAFT—DO NOT CITE OR QUOTE
-------
1 non-neoplastic pathology. Herren-Freund et al. (1987) reported that male B6C3 Fl mice given
2 40 mg/L TCE in drinking water had increased tumor response after 61 weeks of exposure.
3 However, concentrations of TCE fell by about /^ at this dose of TCE during the twice a week
4 change in drinking water solution so the actual dose of TCE the animals received was less than
5 40 mg/L. The percent liver/body weight was reported to be similar for control and TCE-exposed
6 mice at the end of treatment. However, despite difficulties in establishing accurately the dose
7 received, an increase in adenomas per animal and an increase in the number of animals with
8 hepatocellular carcinomas were reported to be associated with TCE exposure after 61 weeks of
9 exposure and without apparent hepatomegaly. Anna et al. (1994) reported tumor incidences for
10 male B6C3F1 mice receiving 800 mg/kg/d TCE via gavage (5 days/week for 76 weeks). All
11 TCE-treated mice were reported to be alive after 76 weeks of treatment. Although the control
12 group contained a mixture of exposure durations (76-134 weeks) and concurrent controls had a
13 very small number of animals, TCE-treatment appeared to increase the number of animals with
14 adenomas, the mean number of adenomas and carcinomas, but with no concurrent TCE-induced
15 cytotoxicity.
16
17 4.5.5.3. Summary: Trichloroethylene (TCE)-Induced Cancer in Laboratory Animals
18 Chronic TCE bioassays have consistently reported increased liver tumor incidences in
19 both sexes of B6C3F1 mice treated by inhalation and gavage exposure in a number of bioassays.
20 The only inhalation study of TCE in Swiss mice also showed an effect in males. Data in the rat,
21 while not reporting statistically significantly increased risks, are not entirely adequate due to low
22 numbers of animals, inadequate reporting, use of insensitive bioassays, increased systemic
23 toxicity, and/or increased mortality. Notably, several studies in rats noted a few very rare types
24 of liver or biliary tumors (cystic cholangioma, cholangiocarcinoma, or angiosarcomas) in treated
25 animals.
26
27 4.5.6. Role of Metabolism in Liver Toxicity and Cancer
28 It is generally thought that TCE oxidation by CYPs is necessary for induction of
29 hepatotoxicity and hepatocarcinogenicity (Bull, 2000). Direct evidence for this hypothesis is
30 limited, e.g., the potentiation of hepatotoxicity by pretreatment with CYP inducers such as
31 ethanol and phenobarbital (Nakajima et al., 1988; Okino et al., 1991). Rather the presumption
32 that CYP-mediated oxidation is necessary for TCE hepatotoxicity and hepatocarcinogenicity is
33 largely based on similar effects (e.g., increases in liver weight, peroxisome proliferation, and
34 hepatocarcinogenicity) having been observed with TCE's oxidative metabolites. The discussion
35 below focuses the similarities and differences between the major effects in the liver of TCE and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-264 DRAFT—DO NOT CITE OR QUOTE
-------
1 of the oxidative metabolites CH, TCA, and DCA. In addition, CH is largely converted to TCOH,
2 TCA, and possibly DCA.
O
4 4.5.6.1. Pharmacokinetics of Chloral Hydrate (CH), Trichloroacetic Acid (TCA), and
5 Dichloroacetic Acid (DCA) From Trichloroethylene (TCE) Exposure
6 As discussed in Chapter 3, in vivo data confirm that CH and TCA, are oxidative
7 metabolites of TCE. In addition, there are indirect data suggesting the formation of DCA.
8 However, direct in vivo evidence of the formation of DCA is confounded by its rapid clearance
9 at low concentrations, and analytical artifacts in its detection in vivo that have yet to be entirely
10 resolved. PBPK modeling (see Section 3.5) predicts that the proportions of TCE metabolized to
11 CH and TCA varies considerably in mice (ranging from 15-97 and 4-38%, respectively) and
12 rats (ranging 7-75 and 0.5-22%, respectively). Therefore, a range of smaller concentrations of
13 TCA or CH may be relevant for comparisons with TCE-induced liver effects. For example, for
14 1,000 mg/kg/d oral doses of TCE, the relevant comparisons would be approximately
15 0.25-1.5 g/L in drinking water for TCA and CH. For DCA a corresponding range is harder to
16 determine and has been suggested to be an upper limit of about 12% (Barton et al., 1999).
17
18 4.5.6.2. Comparisons Between Trichloroethylene (TCE) and Trichloroacetic Acid (TCA),
19 Dichloroacetic Acid (DCA), and Chloral Hydrate (CH) Noncancer Effects
20 4.5.6.2.1. Hepatomegaly—qualitative and quantitative comparisons. As discussed above,
21 TCE causes hepatomegaly in rats, mice, and gerbils under both acute and chronic dosing. Data
22 from a few available studies suggest that oxidative metabolism is important for mediating these
23 effects. Buben and O'Flaherty (1985) collected limited pharmacokinetic data in a sample of the
24 same animals for which liver weight changes were being assessed. While liver weight increases
25 had similarly strong correlations with applied dose and urinary metabolites for doses up to
26 1,600 mg/kg/d (R2 of 0.97 for both), above that dose, the linear relationship was maintained with
27 urinary metabolites but not with applied dose. Ramdhan et al. (2008) conducted parallel
28 experiments at TCE 1,000 and 2,000 ppm (8 hours/day, 7 days) in wild-type and cyp2el-null
29 mice, which did not exhibit increased liver/body weight ratios with TCE treatment and excreted
30 2-fold lower amounts of oxidative metabolites TCA and TCOH in urine as compared to wild-
31 type mice. However, among control mice, those with the null genotype had 1.32-fold higher
32 absolute liver weights and 1.18-fold higher liver/body weight ratios than wild-type mice,
33 reducing the sensitivity of the experiment, particularly with only 6 mice per dose group.
34 With respect to oxidative metabolites themselves, data from CH studies are not
35 informative—either because data were not shown (Sanders et al., 1982) or, because at the time
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-265 DRAFT—DO NOT CITE OR QUOTE
-------
1 points measured, liver weight increases are substantially confounded by foci and carcinogenic
2 lesions (Leakey et al., 2003a). TCA and DCA have both been found to cause hepatomegaly in
3 mice and rats, with mice being more sensitive to this effect. DCA also increases liver/body
4 weight ratios in dogs, but TCE and TCA have not been tested in this species (Cicmanec et al.,
5 1991).
6 As noted above, TCE-induced changes in liver weight appear to be proportional to the
7 exposure concentration across route of administration, gender and rodent species. As an
8 indication of the potential contribution of TCE metabolites to this effect, a quantitative
9 comparison of the shape of the dose-response curves for liver weight induction for TCE and its
10 metabolites is informative. The analysis below was reported in Evans et al. (2009).
11 A number of short-term (<4 weeks) studies of TCA and DCA in drinking water have
12 attempted to measure changes in liver weight induction, with the majority of these studies being
13 performed in male B6C3F1 mice. Studies conducted from 14 to 30 days show a consistent
14 increase in percent liver/body weight induction by TCA or DCA. However, as stated in many of
15 the discussions of individual studies (see Appendix E), there is a limited ability to detect a
16 statistically significant change in liver weight change in experiments that use a relatively small
17 number of animals or do not match control and treatment groups for age and weight. The
18 experiments of Buben and O'Flaherty used 12-14 mice per group giving it a greater ability to
19 detect a TCE-induced dose response. However, many experiments have been conducted with
20 4-6 mice per dose group. For example, the data from DeAngelo et al. (2008) for TCA-induced
21 percent liver/body weight ratio increases in male B6C3F1 mice were only derived from
22 5 animals per treatment group after 4 weeks of exposure. The 0.05 and 0.5 g/L exposure
23 concentrations were reported to give a 1.09- and 1.16-fold of control percent liver/body weight
24 ratios which were consistent with the increases noted in the cross-study database above.
25 However, a power calculation shows that the Type II error (which should be >50% and thus,
26 greater than the chances of "flipping a coin") was only a 6 and 7% and therefore, the designed
27 experiment could accept a false null hypothesis. In addition, some experiments took greater care
28 to age and weight match the control and treatment groups before the start of treatment.
29 Therefore, given these limitations and the fact that many studies used a limited range of
30 doses, an examination of the combined data from multiple studies (Parrish et al., 1996; Sanchez
31 and Bull, 1990; Carter et al., 1995; Kato-Weinstein et al., 2001; DeAngelo et al., 1989, 2008) can
32 best inform/discern differences in DCA and TCA dose-response relationships for liver weight
33 induction (described in more detail in Section E.2.4.2). The dose-response curves for similar
34 concentrations of DCA and TCA are presented in Figure 4-5 for durations of exposure from
35 14-28 days in the male B6C3F1 mouse, which was the most common sex and strain used. As
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-266 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
noted in Appendix E, there appears to be a linear correlation between dose in drinking water and
liver weight induction up to 2 g/L of DC A. However, the shape of the dose-response curve for
TCA appears to be quite different. Lower concentrations of TCA induce larger increase that
does DC A, but the TCE response reaches an apparent plateau while that of DC A continues to
increase the response. TCA studies did not show significant duration-dependent difference in
liver weight induction in this duration range. Short duration studies (10-42 days) were selected
because (1) in chronic studies, liver weight increases are confounded by tumor burden,
(2) multiple studies are available, and (3) TCA studies do not show significant duration-
dependent differences in this duration range.
1.0
Concentration of DCA or TCA (g/l)
Figure 4-5. Comparison of average fold-changes in relative liver weight to
control and exposure concentrations of 2 g/L or less in drinking water for
TCA and DCA in male B6C3F1 mice for 14-30 days (Parrish et al.,1996;
Sanchez and Bull, 1990; Carter et al., 1995; Kato-Weinstein et al., 2001;
DeAngelo et al., 1989, 2008).
Of interest is the issue of how the dose-response curves for TCA and DCA compare to
that of TCE in a similar model and dose range. Since TCA and DCA have strikingly different
dose-response curves, which one if either best fits that of TCE and thus, can give insight as to
which is causative agent for TCE's effects in the liver? The carcinogenicity of chronic TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-267 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure has been predominantly studies in two mouse strains, Swiss and B6C3F1, both of
2 which reportedly developed liver tumors. Rather than administered in drinking water, oral TCE
3 studies have been conducted via oral gavage and generally in corn oil for 5 days of exposure per
4 week. Factors adding to the increased difficulty in establishing the dose-response relationship
5 for TCE across studies and for comparisons to the DCA and TCA database include vehicle
6 effects, the difference between daily and weekly exposures, the dependence of TCE effects in the
7 liver on its metabolism to a variety of agents capable inducing effects in the liver, differences in
8 response between strains, and the inherent increased variability in use of the male mouse model.
9 Despite difference in exposure route, etc., a consistent pattern of dose-response emerges from
10 combining the available TCE data. The effects of oral exposure to TCE from 10-42 days on
11 liver weight induction is shown below in Figure 4-6 using the data of Elcombe et al. (1985),
12 Dees and Travis (1993), Goel et al. (1992), Merrick et al. (1987), Goldsworthy and Popp (1987),
13 and Buben and O'Flaherty (1985). Oral TCE administration in male B6C3F1 and Swiss mice
14 appeared to induce a dose-related increase in percent liver/body weight that was generally
15 proportional to the increase in magnitude of dose, though as expected, with more variability than
16 observed for a similar exercise for DCA or TCA in drinking water. Some of the variability is
17 due to the inclusion of the 10 day studies, since as discussed in Section E.2.4.2, there was a
18 greater increase in TCE-induced liver weight at 28-42 days of exposure Swiss mice than the
19 10-day data in B6C3F1 mice, and Kjellstrand et al. (1981) noted that TCE-induced liver weight
20 increases are still increasing at 10 days inhalation exposure. A strain difference is not evident
21 between the Swiss and B6C3F1 males, as both the combined TCE data and that for only B6C3F1
22 mice show similar correlation with the magnitude of dose and magnitude of percent liver/body
23 weight increase. The correlation coefficients for the linear regressions presented for the B6C3F1
24 data are R2 =0.861 and for the combined data sets is R2 = 0.712. Comparisons of the slopes of
25 the dose-response curves suggest a greater consistency between TCE and DCA than between
26 TCE and TCA. There did not appear to be evidence of a plateau with higher TCE doses, and the
27 degree of fold-increase rises to higher levels with TCE than with TCA in the same strain of
28 mouse.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-268 DRAFT—DO NOT CITE OR QUOTE
-------
2.0
1.8 -
.°> 1.6-
CU
T3 1.4 -
1.2 -
1.0 I • •
0 500 1000 1500 2000 2500 3000
Concentration of TCE (mg/kg/day)
2.0
2
3
4
5
6
7
8
9
10
11
12
1.8 -
.™ 1.6-
0)
T3 1.4-
O
1.2 -
1.0
• B6C3F1 and Swiss
Plot 2 Regr
500 1000 1500 2000 2500
Concentration of TCE (mg/kg/day)
3000
Figure 4-6. Comparisons of fold-changes in average relative liver weight and
gavage dose of (top panel) male B6C3F1 mice for 10-28 days of exposure
(Merrick et al., 1989; Elcombe et al., 1985; Goldsworthy and Popp, 1987;
Dees and Travis, 1993) and (bottom panel) in male B6C3F1 and Swiss mice.
A more direct comparison would be on the basis of dose rather than drinking water
concentration. The estimations of internal dose of DC A or TCA from drinking water studies,
while varying considerably (DeAngelo et al., 1989, 2008), nonetheless suggest that the doses of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-269 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCE used in the gavage experiments were much higher than those of DC A or TCA. However,
2 only a fraction of ingested TCE is metabolized to DC A or TCA, as, in addition to oxidative
3 metabolism, TCE is also cleared by GSH conjugation and by exhalation. While DCA dosimetry
4 is highly uncertain (see Sections 3.3 and 3.5), the mouse PBPK model, described in Section 3.5
5 was calibrated using extensive in vivo data on TCA blood, plasma, liver, and urinary excretion
6 data from inhalation and gavage TCE exposures, and makes robust predictions of the rate of
7 TCA production. If TCA were predominantly responsible for TCE-induced liver weight
8 increases, then replacing administered TCE dose (e.g., mg TCE/kg/day) by the rate of TCA
9 produced from TCE (mg TCA/kg/day) should lead to dose-response curves for increased liver
10 weight consistent with those from directly administered TCA. Figure 4-7 shows this comparison
11 using the PBPK model-based estimates of TCA production for 4 TCE studies from 28-42 days
12 in the male NMRI, Swiss, and B6C3F1 mice (Kjellstrand et al., 1983b; Buben and O'Flaherty,
13 1985; Merrick et al., 1989; Goel et al., 1992) and 4 oral TCA studies in B6C3F1 male mice at
14 2 g/L or lower drinking water exposure (DeAngelo et al., 1989, 2008; Parrish et al., 1996;
15 Kato-Weinstein et al., 2001) from 14-28 days of exposure. The selection of the 28-42 day data
16 for TCE was intended to address the decreased opportunity for full expression of response at
17 10 days. PBPK modeling predictions of daily internal doses of TCA in terms of mg/kg/d via
18 produced via TCE metabolism would be are indeed lower than the TCE concentrations in terms
19 of mg/kg/d given orally by gavage. The predicted internal dose of TCA from TCE exposure
20 studies are of a comparable range to those predicted from TCA drinking water studies at
21 exposure concentrations in which palpability has not been an issue for estimation of internal
22 dose. Thus, although the TCE data are for higher exposure concentrations, they are predicted to
23 produce comparable levels of TCA internal dose estimated from direct TCA administration in
24 drinking water.
25 Figure 4-7 clearly shows that for a given amount of TCA produced from TCE, but going
26 through intermediate metabolic pathways, the liver weight increases are substantially greater
27 than, and highly inconsistent with, that expected based on direct TCA administration. In
28 particular, the response from direct TCA administration appears to "saturate" with increasing
29 TCA dose at a level of about 1.4-fold, while the response from TCE administration continues to
30 increase with dose to 1.75-fold at the highest dose administered orally in Buben and O'Flaherty
31 (1985) and over 2-fold in the inhalation study of Kjellstrand et al. (1983b). Because TCA liver
32 concentrations are proportional to the dose TCA, and do not depend on whether it is
33 administered in drinking water or internally produced in the liver, the results of the comparison
34 using the TCA liver dose metric are identical.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-270 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
2.5
» TCE Studies [28-42 d]
O TCA Studies [14-28 d]
- — -Linear (TCA Studies [14-28 d])
Linear (TCE Studies [28-42 d])
•a
i=
si
CO
£
O
1.5
100
400
200 300
mg TCA/kg-d
(produced [TCE studies] or administered [TCA studies])
500
Figure 4-7. Comparison of fold-changes in relative liver weight for data sets
in male B6C3F1, Swiss, and NRMI mice between TCE studies (Kjellstrand et
al., 1983b; Buben and O'Flaherty, 1985; Merrick et al., 1989; Goel et al.,
1992 [duration 28-42 days]) and studies of direct oral TCA administration to
B6C3 Fl mice (DeAngelo et al., 1989; Parrish et al., 1996; Kato-Weinstein et
al., 2001; DeAngelo et al., 2008 [duration 14-28 days]). Abscissa for TCE
studies consists of the median estimates of the internal dose of TCA predicted
from metabolism of TCE using the PBPK model described in Section 3.5 of the
TCE risk assessment. Lines show linear regression with intercept fixed at unity.
All data were reported fold-change in mean liver weight/body weight ratios,
except for Kjellstrand et al. (1983b), with were the fold-change in the ratio of
mean liver weight to mean body weight. In addition, in Kjellstrand et al. (1983b),
some systemic toxicity as evidence by decreased total body weight was reported
in the highest-dose group.
Furthermore, while as noted previously, oral studies appear to report a linear relationship
between TCE exposure concentration and liver weight induction, the inclusion of inhalation
studies on the basis of internal dose led to a highly consistent dose-response curve for among
TCE study. Therefore, it is unlikely that differing routes of exposure can explain the
inconsistencies in dose-response.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-271 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Additional analyses do, however, support a role for oxidative metabolism in TCE-
induced liver weight increases, and that the parent compound TCE is not the likely active moiety
(suggested previously by Buben and O'Flaherty [1985]). In particular, the same studies are
shown in Figure 4-8 using PBPK-model based predictions of the area-under-the-curve (AUC) of
TCE in blood and total oxidative metabolism, which produces chloral, TCOH, DC A, and other
metabolites in addition to TCA. The dose-response relationship between TCE blood levels and
liver weight increase, while still having a significant trend, shows substantial scatter and a low
R2 of 0.43. On the other hand, using total oxidative metabolism as the dose metric leads to
substantially more consistency dose-response across studies, and a much tighter linear trend with
an R2 of 0.90 (see Figure 4-8). A similar consistency is observed using liver-only oxidative
metabolism as the dose metric, with R2 of 0.86 (not shown). Thus, while the slope is similar
between liver weight increase and TCE concentration in the blood and liver weight increase and
rate of total oxidative metabolism, the data are a much better fit for total oxidative metabolism.
CD
CO
CO
CD
iq
c\i
p
c\i
R =0.426^
iq
c\i
CD
CO
CD
CD p
O CNJ
\ i \ i i r
0 100 200 300 400 500
Daily AUC TCE in Blood (mg-hr/l)
R2=0.8955
500
1000
1500
Daily TCE Oxidized (mg/kg-d)
Figure 4-8. Fold-changes in relative liver weight for data sets in male
B6C3F1, Swiss, and NRMI mice reported by TCE studies of duration
28-42 days (Kjellstrand et al., 1983b; Buben and O'Flaherty, 1985; Merrick
et al., 1989; Goel et al., 1992) using internal dose metrics predicted by the
PBPK model described in Section 3.5: (A) dose metric is the median estimate
of the daily AUC of TCE in blood, (B) dose metric is the median estimate of
the total daily rate of TCE oxidation. Lines show linear regression Use of
liver oxidative metabolism as a dose metric gives results qualitatively similar to
(B), with R2 = 0.86.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-272 DRAFT—DO NOT CITE OR QUOTE
-------
1 Although the qualitative similarity to the linear dose-response relationship between DCA
2 and liver weight increases is suggestive of DCA being the predominant metabolite responsible
3 for TCE liver weight increases, due to the highly uncertain dosimetry of DCA derived from
4 TCE, this hypothesis cannot be tested on the basis of internal dose. Similarly, another TCE
5 metabolite, chloral hydrate, has also been reported to induce liver tumors in mice, however, there
6 are no adequate comparative data to assess the nature of liver weight increases induced by this
7 TCE metabolite (see Section E.2.5 and Section 4.5.1.2.4 below). Whether its formation in the
8 liver after TCE exposure correlates with TCE-induced liver weight changes cannot be
9 determined.
10
11 4.5.6.2.2. Cytotoxicity. As discussed above, TCE has sometimes been reported to cause
12 minimal/mild focal hepatocellular necrosis or other signs of hepatic injury, albeit of low
13 frequency and mostly at doses >1,000 mg/kg/d (Dees and Travis, 1993; Elcombe et al., 1985) or
14 at exposures >1,000 ppm in air (Ramdhan et al., 2008) from 7-10 days of exposure. Data from
15 available studies are supportive of a role for oxidative metabolism in TCE-induced cytotoxicity
16 in the liver, though they are not informative as to the actual active moiety(ies). Buben and
17 O'Flaherty (1985) noted a strong correlation (R-squared of between glucose-6-phosphatase
18 inhibition and total urinary oxidative metabolites). Ramdhan et al. (2008) conducted parallel
19 experiments at TCE 1,000 and 2,000 ppm (8 hours/day, 7 days) in wild-type and cyp2el-null
20 mice, the latter of which did not exhibit hepatotoxicity (assessed by serum ALT, AST, and
21 histopathology) and excreted 2-fold lower amounts of oxidative metabolites TCA and TCOH in
22 urine as compared to wild-type mice. In addition, urinary TCA and TCOH excretion was
23 correlated with serum ALT and AST measures, though the R-squared values (square of the
24 reported correlation coefficients) were relatively low (0.54 and 0.67 for TCOH and TCA,
25 respectively).
26 With respect to CH (166 mg/kg/d) and DCA (-90 mg/kg/d), Daniel et al. (1992) reported
27 that after drinking water treatment, hepatocellular necrosis and chronic active inflammation were
28 reported to be mildly increased in both prevalence and severity in all treated groups after
29 104 weeks of exposure. The histological findings, from interim sacrifices (n = 5), were
30 considered by the authors to be unremarkable and were not reported. TCA has not been reported
31 to induce necrosis in the liver under the conditions tested. Relatively high doses of DCA (>1 g/L
32 in drinking water) appear to result in mild focal necrosis with attendant reparative proliferation at
33 lesion sites, but no such effects were reported at lower doses (<0.5 g/L in drinking water) more
34 relevant for comparison with TCE (DeAngelo et al., 1999; Sanchez and Bull, 1990; Stauber et
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-273 DRAFT—DO NOT CITE OR QUOTE
-------
1 al., 1998). Enlarged nuclei and changes consistent with increased ploidy, are further discussed
2 below in the context of DNA synthesis.
3 4.5.6.2.3. DNA synthesis andpolyploidization. The effects on DNA synthesis and
4 polyploidization observed with TCE treatment have similarly been observed with TCA and
5 DC A. With respect to CH, George et al. (2000) reported that CH exposure did not alter DNA
6 synthesis in rats and mice at any of the time periods monitored (all well past 2 weeks), with the
7 exception of 0.58 g/L chloral hydrate at 26 weeks slightly increasing hepatocyte labeling
8 (~2-3-fold of controls) in rats and mice but the percent labeling still representing 3% or less of
9 hepatocytes.
10 In terms of whole liver or hepatocyte label incorporation, the most comparable exposure
11 duration between TCE, TCA, and DCA studies is the 10- and 14-day period. Several studies
12 have reported that in this time period, peak label incorporation into individual hepatocytes and
13 whole liver for TCA and DCA have already passed (Styles et al., 1991; Sanchez and Bull, 1990;
14 Pereira, 1996; Carter et al., 1995). A direct time-course comparison is difficult, since data at
15 earlier times for TCE are more limited.
16 There are conflicting reports of DNA synthesis induction in individual hepatocytes for up
17 to 14 days of DCA or TCA exposure. In particular, Sanchez and Bull (1990) reported tritiated
18 thymidine incorporation in individual hepatocytes up to 2 g/L exposure to DCA or TCA induced
19 little increase in DNA synthesis except in instances and in close proximity to areas of
20 proliferation/necrosis for DCA treatment after 14 days of exposure in male mice. The largest
21 percentage of hepatocytes undergoing DNA synthesis for any treatment group was less than 1%
22 of hepatocytes. However, they reported treatment- and exposure duration-changes in hepatic
23 DNA incorporation of tritiated thymidine for DCA and TCA. For TCA treatment, the largest
24 increases over control levels for hepatic DNA incorporation (at the highest dose) was a 3-fold
25 increase after 5 days of treatment and a 2-fold increase over controls after 14 days of treatment.
26 For DCA whole-liver tritiated thymidine incorporation was only slightly elevated at necrogenic
27 concentrations and decreased at the 0.3 g/L non-necrogenic level after 14 days of treatment. In
28 contrast to Sanchez and Bull (1990), Stauber and Bull (1997) reported increased tritiated
29 thymidine incorporation for individual hepatocytes after 14 days of treatment with 2 g/L DCA or
30 TCA in male mice. They used a more extended period of tritiated thymidine exposure of
31 3-5 days and so these results represent aggregate DNA synthesis occurring over a more extended
32 period of time. A "1-day labeling index" was reported as less than 1% for the highest level of
33 increased incorporation. However, after 14 days, the labeling index was reported to be increased
34 by ~3.5-fold for TCA and ~5.5-fold for DCA over control values. After 28 days, the labeling
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-274 DRAFT—DO NOT CITE OR QUOTE
-------
1 index was reported to be decreased ~2.3-fold by DCA and increased ~2.5-fold after treatment
2 with TCA. Pereira (1996) reported that for female B6C3F1 mice, 5-day incorporation of BrDU,
3 as a measure of DNA synthesis, was increased at 0.86 g/L and 2.58 g/L DCA treatment for
4 5 days (~2-fold at the highest dose) but that by Day 12 and 33 levels had fallen to those of
5 controls. For TCA exposures, 0.33 g/L, 1.10 g/L and 3.27 g/L TCA all gave a similar ~3-fold
6 increase in BrdU incorporation by 5 days, but that by 12 and 33 days were not changed from
7 controls. Nonetheless, what is consistent is that these data report that, similar to TCE-exposed
8 mice at 10 days of exposure, cells undergoing DNA synthesis in DCA- or TCA-exposed mice for
9 up to 14 days of exposure to be confined to a very small population of cells in the liver. Thus,
10 these data are consistent with hypertrophy being primarily responsible for liver weight gains as
11 opposed to increases in cell number in mice.
12 Interestingly, a lack of correlation between whole liver label incorporation and that in
13 individual hepatocytes has been reported by several studies of DCA (Sanchez and Bull, 1990;
14 Carter et al., 1995). For example, Carter et al. (1995) reported no increase in labeling of
15 hepatocytes in comparison to controls for any DCA treatment group from 5 to 30 days of DCA
16 exposure. Rather than increase hepatocyte labeling, DCA induced no change from days 5 though
17 15 but significantly decreased levels between days 20 and 30 for 0.5 g/L that were similar to
18 those observed for the 5 g/L exposures. However, for whole liver DNA tritiated thymidine
19 incorporation, Carter et al. (1995) reported 0.5g/L DCA treatments to show trends of initial
20 inhibition of DNA tritiated thymidine incorporation followed by enhancement of labeling that
21 was not statistically significant from 5 to 30 days of exposure. Examination of individual
22 hepatocytes does not include the contribution of nonparenchymal cell DNA synthesis that would
23 be detected in whole liver DNA. As noted above, proliferation of the nonparenchymal cell
24 compartment of the liver has been noted in several studies of TCE in rodents, and thus, this is
25 one possible reason for the reported discrepancy.
26 Another possible reason for this inconsistency with DCA treatment is polyploidization, as
27 was suggested above for TCE. Although this was not examined for DCA or TCA exposure by
28 Sanchez and Bull (1990), Carter et al. (1995) reported that hepatocytes from both 0.5 and 5 g/L
29 DCA treatment groups had enlarged, presumably polyploidy nuclei, with some hepatocyte nuclei
30 labeled in the mid-zonal area. There were statistically significant changes in cellularity, nuclear
31 size, and multinucleated cells during 30 days exposure to DCA. The percentage of
32 mononucleated cells hepatocytes was reported to be similar between control and DCA treatment
33 groups at 5- and 10-day exposure. However, at 15 days and beyond DCA treatments were
34 reported to induce increases in mononucleated hepatocytes with later time periods to also
35 showing DCA-induced increases nuclear area, consistent with increased polyploidization without
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-275 DRAFT—DO NOT CITE OR QUOTE
-------
1 mitosis. The consistent reporting of an increasing number of mononucleated cells between 15
2 and 30 days could be associated with clearance of mature hepatocytes as suggested by the report
3 of DCA-induced loss of cell nuclei. The reported decrease in the numbers of binucleate cells in
4 favor of mononucleate cells is not typical of any stage of normal liver growth (Brodsky and
5 Uryvaeva, 1977). The pattern of consistent increase in percent liver/body weight induced by
6 0.5 g/L DCA treatment from days 5 though 30 was not consistent with the increased numbers of
7 mononucleate cells and increase nuclear area reported from Day 20 onward. Specifically, the
8 large differences in liver weight induction between the 0.5 g/L treatment group and the 5 g/L
9 treatment groups at all times studied also did not correlate with changes in nuclear size and
10 percent of mononucleate cells. Thus, increased liver weight was not a function of cellular
11 proliferation, but probably included both aspects of hypertrophy associated with polyploidization
12 and increased glycogen deposition (see below) induced by DCA. Carter et al. (1995) suggested
13 that although there is evidence of DCA-induced cytotoxicity (e.g., loss of cell membranes and
14 apparent apoptosis), the 0.5 g/L exposure concentration has been shown to increase
15 hepatocellular lesions after 100 weeks of treatment without concurrent peroxisome proliferation
16 or cytotoxicity (DeAngelo et al., 1999).
17 In sum, the observation of TCE-treatment related changes in DNA content, label
18 incorporation, and mitotic figures are generally consistent with patterns observed for both TCA
19 and DCA. In all cases, hepatocellular proliferation is confined to a very small fraction of
20 hepatocytes, and hepatomegaly observed with all three treatments probably largely reflects
21 cytomegaly rather than cell proliferation. Moreover, label incorporation likely largely reflects
22 polyploidization rather than hepatocellular proliferation, with a possible contribution from
23 nonparenchymal cell proliferation. As with TCE, histological changes in nuclear sizes and
24 number also suggest a significant degree of treatment-related polyploidization, particularly for
25 DCA.
26
27 4.5.6.2.4. Apoptosis. As for apoptosis, Both Elcombe et al. (1985) and Dees and Travis (1993)
28 reported no changes in apoptosis other than increased apoptosis only at a treatment level of
29 1,000-mg/kg TCE. Dees and Travis (1993) reported that increased apoptoses from TCE
30 exposure "did not appear to be in proportion to the applied TCE dose given to male or female
31 mice." Channel et al. (1998) reported that there was no significant difference in apoptosis
32 between TCE treatment and control groups with data not shown. However, the extent of
33 apoptosis in any of the treatment groups, or which groups and timepoints were studied for this
34 effect cannot be determined. While these data are quite limited, it is notable that peroxisome
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-276 DRAFT—DO NOT CITE OR QUOTE
-------
1 proliferators have been suggested inhibit, rather than increase, apoptosis as part of their
2 carcinogenic MOA (Klaunig et al., 2003).
3 However, for TCE metabolites, DCA has been most studied, though it is clear that age
4 and species affect background rates of apoptosis. Snyder et al. (1995), in their study of DCA,
5 report that control mice were reported to exhibit apoptotic frequencies ranging from -0.04 to
6 0.085%, that over the 30-day period of their study the frequency rate of apoptosis declined, and
7 suggest that this pattern is consistent with reports of the livers of young animals undergoing
8 rapid changes in cell death and proliferation. They reported rat liver to have a greater the
9 estimated frequency of spontaneous apoptosis (~0.1%) and therefore, greater than that of the
10 mouse. Carter et al. (1995) reported that after 25 days of 0.5 g/L DCA treatment apoptotic
11 bodies were reported as well as fewer nuclei in the pericentral zone and larger nuclei in central
12 and midzonal areas. This would indicate an increase in the apoptosis associated with potential
13 increases in polyploidization and cell maturation. However, Snyder et al. (1995) report that mice
14 treated with 0.5 g/L DCA over a 30-day period had a similar trend as control mice of decreasing
15 apoptosis with age. The percentage of apoptotic hepatocytes decreased in DC A-treated mice at
16 the earliest time point studied and remained statistically significantly decreased from controls
17 from 5 to 30 days of exposure. Although the rate of apoptosis was very low in controls,
18 treatment with 0.5g/L DCA reduced it further (-30-40% reduction) during the 30-day study
19 period. The results of this study not only provide a baseline of apoptosis in the mouse liver,
20 which is very low, but also to show the importance of taking into account the effects of age on
21 such determinations. The significance of the DCA-induced reduction in apoptosis reported in
22 this study, from a level that is already inherently low in the mouse, for the MOA for induction of
23 DCA-induce liver cancer is difficult to discern.
24
25 4.5.6.2.5. Glycogen accumulation. As discussed in Sections E.3.2 and E.3.4.2.1, glycogen
26 accumulation has been described to be present in foci in both humans and animals as a result
27 from exposure to a wide variety of carcinogenic agents and predisposing conditions in animals
28 and humans. The data from Elcombe et al. (1985) included reports of TCE-induced pericentral
29 hypertrophy and eosinophilia for both rats and mice but with "fewer animals affected at lower
30 doses." In terms of glycogen deposition, Elcombe report "somewhat" less glycogen pericentrally
31 in the livers of rats treated with TCE at 1,500 mg/kg than controls with less marked changes at
32 lower doses restricted to fewer animals. They do not comment on changes in glycogen in mice.
33 Dees and Travis (1993) reported TCE-induced changes to "include an increase in eosinophilic
34 cytoplasmic staining of hepatocytes located near central veins, accompanied by loss of
35 cytoplasmic vacuolization." Since glycogen is removed using conventional tissue processing
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-277 DRAFT—DO NOT CITE OR QUOTE
-------
1 and staining techniques, an increase in glycogen deposition would be expected to increase
2 vacuolization and thus, the report from Dees and Travis is consistent with less not more glycogen
3 deposition. Neither study produced a quantitative analysis of glycogen deposition changes from
4 TCE exposure. Although not explicitly discussing liver glycogen content or examining it
5 quantitatively in mice, these studies suggest that TCE-induced liver weight increases did not
6 appear to be due to glycogen deposition after 10 days of exposure and any decreases in glycogen
7 were not necessarily correlated with the magnitude of liver weight gain either.
8 For TCE and TCA 500 mg/kg treatments in mice for 10 days, changes in glycogen were
9 not reported in the general descriptions of histopathological changes (Elcombe et al., 1985;
10 Styles et al., 1991; Dees and Travis, 1993) or were specifically described by the authors as being
11 similar to controls (Nelson et al., 1989). However, for DCA, glycogen deposition was
12 specifically noted to be increased with treatment, although no quantitative analyses was
13 presented that could give information as to the nature of the dose-response (Nelson et al., 1989).
14 In regard to cell size, although increased glycogen deposition with DCA exposure was
15 noted by Sanchez and Bull (1990) to occur to a similar extent in B6C3F1 and Swiss Webster
16 male mice despite differences in DCA-induced liver weight gain. Lack of quantitative analyses
17 of that accumulation in this study precludes comparison with DCA-induced liver weight gain.
18 Carter et al. (1995) reported that in control mice there was a large variation in apparent glycogen
19 content and also did not perform a quantitative analysis of glycogen deposition. The variability
20 of this parameter in untreated animals and the extraction of glycogen during normal tissue
21 processing for light microscopy make quantitative analyses for dose-response difficult unless
22 specific methodologies are employed to quantitatively assess liver glycogen levels as was done
23 by Kato-Weinstein et al. (2001) and Pereira et al. (2004).
24 Bull et al. (1990) reported that glycogen deposition was uniformly increased from 2 g/L
25 DCA exposure with photographs of TCA exposure showing slightly less glycogen staining than
26 controls. However, the abstract and statements in the paper suggest that there was increased
27 PAS positive material from TCA treatment that has caused confusion in the literature in this
28 regard. Kato-Weinstein et al. (2001) reported that in male B6C3F1 mice exposed to DCA and
29 TCA, the DCA treatment increased glycogen and TCA decreased glycogen content of the liver
30 by using both chemical measurement of glycogen in liver homogenates and by using ethanol-
31 fixed sections stained with PAS, a procedure designed to minimize glycogen loss.
32 Kato-Weinstein et al. (2001) reported that glycogen rich and poor cells were scattered
33 without zonal distribution in male B6C3F1 mice exposed to 2 g/L DCA for 8 weeks. For TCA
34 treatments, they reported centrilobular decreases in glycogen and -25% decreases in whole liver
35 by 3 g/L TCA. Kato-Weinstein et al. (2001) reported whole liver glycogen to be increased
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-278 DRAFT—DO NOT CITE OR QUOTE
-------
1 ~1.50-fold of control (90 vs. 60 mg glycogen/g liver) by 2 g/L DCA after 8 weeks exposure male
2 B6C3F1 mice with a maximal level of glycogen accumulation occurring after 4 weeks of DCA
3 exposure. Pereira et al. (2004) reported that after 8 weeks of exposure to 3.2 g/L DCA liver
4 glycogen content was 2.20-fold of control levels (155.7 vs. 52.4 mg glycogen/g liver) in female
5 B6C3F1 mice. Thus, the baseline level of glycogen content reported by (-60 mg/g) and the
6 increase in glycogen after DCA exposure was consistent between Kato-Weinstein et al. (2001)
7 and Pereira et al. (2004). However, the increase in liver weight reported by Kato-Weinstein et al.
8 (2001) of 1.60-fold of control percent liver/body weight cannot be accounted for by the 1.50-fold
9 of control glycogen content. Glycogen content only accounts for 5% of liver mass so that 50%
10 increase in glycogen cannot account for the 60% increase liver mass induced by 2 g/L DCA
11 exposure for 8 weeks reported by Kato-Weinstein (2001). Thus, DCA-induced increases in liver
12 weight are occurring from other processes as well. Carter et al. (2003) and DeAngelo et al.
13 (1999) reported increased glycogen after DCA treatment at much lower doses after longer
14 periods of exposure (100 weeks). Carter reported increased glycogen at 0.5 g/L DCA and
15 DeAngelo et al. (1999) at 0.03 g/L DCA in mice. However, there is no quantitation of that
16 increase.
17
18 4.5.6.2.6. Peroxisomeproliferation and related effects. TCA and DCA have both been
19 reported to induce peroxisome proliferation or increase in related enzyme markers in rodent
20 hepatocytes (DeAngelo et al., 1989, 1997; Mather et al., 1990; Parrish et al., 1996). Between
21 TCA and DCA, both induce peroxisome proliferation in various strains of mice, but it clear that
22 TCA and DCA are weak PPARa agonists and that DCA is weaker than TCA in this regard
23 (Nelson et al., 1989) using a similar paradigm.
24 George et al. (2000) reported that CH exposure did not hepatic PCO activity in rats and
25 mice at any of the time periods monitored. It is notable that the only time at which DNA
26 synthesis index was (slightly) increased, at 26 weeks, there remained a lack of induction of PCO.
27 A number of measures that may be related to peroxisome proliferation were investigated in
28 Leakey et al. (2003a). Of the enzymes associated with PPARa agonism (total CYP, CYP2B
29 isoform, CYP4A, or lauric acid p-hydroxylase activity), only CYP4A and lauric acid
30 P-hydroxylase activity were significantly increased at 15 months of exposure in the dietary -
31 restricted group administered the highest dose (100 mg/kg CH) with no other groups reported
32 showing a statistically significant increased response (n = 12/group). There is an issue of
33 interpretation of peroxisomal enzyme activities and other enzymes associated with PPARa
34 receptor activation to be a relevant event in liver cancer induction at a time period in which
35 tumors or foci are already present. Although not statistically significant, the 100 mg/kg CH
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-279 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure group of ad libitum-fed mice also had an increase in CH-induced increases of CYP4A
2 and lauric acid p-hydroxylase activity. Seng et al. (2003) described CH toxicokinetics and
3 peroxisome proliferation-associated enzymes in mice at doses up to 1,000 mg/kg/d for 2 weeks
4 with dietary control or caloric restriction. Lauric acid P-hydroxylase and PCO activities were
5 reported to be induced only at doses >100 mg/kg in all groups, with dietary-restricted mice
6 showing the greatest induction. Differences in serum levels of TCA, the major metabolite
7 remaining 24 hours after dosing, were reported not to correlate with hepatic lauric acid
8 P-hydroxylase activities across groups.
9 Direct quantitative inferences regarding the magnitude of response in these studies in
10 comparison to TCE, however, are limited by possible variability and confounding. In particular,
11 many studies used cyanide-insensitive PCO as a surrogate for peroxisome proliferation, but the
12 utility of this marker may be limited for a number of reasons. First, several studies have shown
13 that this activity is not well correlated with the volume or number of peroxisomes that are
14 increased as a result of exposure to TCE or it metabolites (Nakajima et al., 2000; Elcombe et al.,
15 1985; Nelson et al., 1989). In addition, this activity appears to be highly variable both as a
16 baseline measure and in response to chemical exposures. Laughter et al. (2004) presented data
17 showing WY-14,643 induced increases in PCO activity that varied up to 6-fold between different
18 experiments in wild-type mice. They also showed that, in some instances, PCO activity in
19 untreated PPARa-null mice was up to 6-fold greater than that in wild-type mice. Parrish et al.
20 (1996) noted that control values between experiments varied as much as a factor of 2-fold for
21 PCO activity and thus, their data were presented as percent of concurrent controls. Furthermore,
22 Melnick et al. (1987) reported that corn oil administration alone can elevate PCO (as well as
23 catalase) activity, and corn oil has also been reported to potentiate the induction of PCO activity
24 of TCA in male mice (DeAngelo et al., 1989). Thus, quantitative inferences regarding the
25 magnitude of response in these studies are limited by a number of factors. For example, in the
26 studies reported in DeAngelo et al. (2008) a small number of animals was studied for PCO
27 activity at interim sacrifices (n = 5). PCO activity varied 2.7-fold as baseline controls. Although
28 there was a 10-fold difference in TCA exposure concentration, the increase in PCO activity at
29 4 weeks was 1.3-, 2.4-, and 5.3-fold of control. More information on the relationship of PCO
30 enzyme activity and its relationship to carcinogenicity is discussed in Section E.3.4 and below.
31
32 4.5.6.2.7. Oxidative stress. Very limited data are available as to oxidative stress and related
33 markers induced by the oxidative metabolites of TCE. As discussed in Appendix E, above, there
34 are limited data that do not indicate significant oxidative stress and associated DNA damage
35 associated with acute and subacute TCE treatment. In regard to DC A and TCA, Larson and Bull
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-280 DRAFT—DO NOT CITE OR QUOTE
-------
1 (1992) exposed male B6C3F1 mice or Fischer 344 rats to single doses TCA or DCA in distilled
2 water by oral gavage (n = 4). In the first experiment, TEARS was measured from liver
3 homogenates and assumed to be malondialdehyde. The authors stated that a preliminary
4 experiment had shown that maximal TEARS was increased 6 hours after a dose of DCA and
5 9 hours after a dose of TCA in mice and that by 24 hours TEARS concentrations had declined to
6 control values. Time-course information in rats was not presented. A dose of 100 mg/kgDCA
7 (rats or mice) or TCA (mice) did not elevate TEARS concentrations over that of control liver
8 with this concentration of TCA not examined in rats. For TCA, there was a slight dose-related
9 increase in TEARS over control values starting at 300 mg/kg in mice with the increase in
10 TEARS increasing at a rate that was lower than the magnitude of increase in dose. Of note, is
11 the report that the induction of TEARS in mice is transient and has subsided within 24 hours of a
12 single dose of DCA or TCA, that the response in mice appeared to be slightly greater with DCA
13 than TCA at similar doses, and that for DCA, there was similar TEARS induction between rats
14 and mice at similar dose levels.
15 Austin et al. (1996) appears to a follow-up publication of the preliminary experiment
16 cited in Larson and Bull (1992). Male B6C3F1 mice were treated with single doses of DCA or
17 TCA via gavage with liver examined for SOHdG. The authors stated that in order to conserve
18 animals, controls were not employed at each time point. There was a statistically significant
19 increase over controls in SOHdG for the 4- and 6-hour time points for DCA (-1.4- and 1.5-fold
20 of control, respectively) but not at 8 hours in mice. For TCA, there was a statistically significant
21 increase in SOHdG at 8 and 10 hours for TCA (-1.4- and 1.3-fold of control, respectively).
22 Consistent results as to low, transient increases in markers of "oxidative stress" were also
23 reported by Parrish et al. (1996), who in addition to examining oxidative stress alone, attempted
24 to examine its possible relationship to PCO and liver weight in male B6C3F1 mice exposed to
25 TCA or DCA for 3 or 10 weeks (n = 6). The dose-related increase in PCO activity at 21 days for
26 TCA was reported to not be increased similarly for DCA. Only the 2.0 g/L dose of DCA was
27 reported to induce a statistically significant increase at 21-days of exposure of PCO activity over
28 control (-1.8-fold of control). After 71 days of treatment, TCA induced dose-related increases in
29 PCO activities that were approximately twice the magnitude as that reported at 21 days.
30 Treatments with DCA at the 0.1 and 0.5 g/L exposure levels produced statistically significant
31 increase in PCO activity of-1.5- and 2.5-fold of control, respectively. The administration of
32 1.25 g/L clofibric acid in drinking water, used as a positive control, gave ~6-7-fold of control
33 PCO activity at 21 and 71 days exposure. Parrish et al. (1996) reported that laurate hydroxylase
34 activity was reported to be elevated significantly only by TCA at 21 days and to approximately
35 the same extent (-1.4- to 1.6-fold of control) increased at all doses tested and at 71 days both the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-281 DRAFT—DO NOT CITE OR QUOTE
-------
1 0.5 and 2.0 g/L TCA exposures to a statistically significant increase in laurate hydroxylase
2 activity (i.e., 1.6- and 2.5-fold of control, respectively). No change was reported after DCA
3 exposure. Laurate hydroxylase activity within the control values varying 1.7-fold between 21
4 and 71 days experiments. Levels of SOHdG in isolated liver nuclei were reported to not be
5 altered from 0.1, 0.5, or 2.0 g/L TCA or DCA after 21 days of exposure and this negative result
6 was reported to remain even when treatments were extended to 71 days of treatment. The
7 authors noted that the level of SOHdG increased in control mice with age (i.e., ~2-fold increase
8 between 71-day and 21-day control mice). Thus, the increases in PCO activity noted for DCA
9 and TCA were not associated with SOHdG levels (which were unchanged) and also not with
10 changes laurate hydrolase activity observed after either DCA or TCA exposure. Of note, is that
11 the authors report taking steps to minimize artifactual responses for their SOHdG determinations.
12 The authors concluded that their data suggest that peroxisome proliferative properties of TCA
13 were not linked to oxidative stress or carcinogenic response.
14
15 4.5.6.3. Comparisons of Trichloroethylene (TCE)-Induced Carcinogenic Responses With
16 Trichloroacetic Acid (TCA), Dichloroacetic Acid (DCA), and Chloral Hydrate (CH)
17 Studies
18 4.5.6.3.1. Studies in rats. As discussed above, data on TCE carcinogenicity in rats, while not
19 reporting statistically significantly increased risks, are not entirely adequate due to low numbers
20 of animals, increased systemic toxicity, and/or increased treatment-related or accidental
21 mortality. Notably, several studies in rats noted a few very rare types of liver or biliary tumors
22 (cystic cholangioma, cholangiocarcinoma, or angiosarcomas) in treated animals. For TCA, DCA
23 and CH, there are even fewer studies in rats, so there is a very limited ability to assess the
24 consistency or lack thereof in rat carcinogenicity among these compounds.
25 For TCA, the only available study in rats (DeAngelo et al., 1997) has been frequently
26 cited in the literature to indicate a lack of response in this species for TCA-induced liver tumors.
27 However, this study does report an apparent dose-related increase in multiplicity of adenomas
28 and an increase in carcinomas over control at the highest dose. The use by DeAngelo et al.
29 (1997) of a relatively low number of animals per treatment group (n = 20-24) limits this study's
30 ability to determine a statistically significant increase in tumor response. Its ability to determine
31 an absence of treatment-related effect is similarly limited. In particular, a power calculation of
32 the study shows that for most endpoints (incidence and multiplicity of all tumors at all exposure
33 DCA concentrations), the Type II error, which should be >50%, was less than 8%. The only
34 exception was for the incidence of adenomas and adenomas and carcinomas for the 0.5 g/L
35 treatment group (58%), at which, notably, there was a reported increase in reported adenomas or
36 adenomas and carcinomas combined over control (15 vs. 4%). Therefore, the likelihood of a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-282 DRAFT—DO NOT CITE OR QUOTE
-------
1 false null hypothesis was not negligible. Thus, while suggesting a lower response than for mice
2 for liver tumor induction, this study is inconclusive for determining of whether TCA induces a
3 carcinogenic response in the liver of rats.
4 For DC A, there are two reported long-term studies in rats (DeAngelo et al, 1996;
5 Richmond et al., 1995) that appear to have reported the majority of their results from the same
6 data set and which consequently were subject to similar design limitations and DCA-induced
7 neurotoxicity in this species. DeAngelo et al. (1996) reported increased hepatocellular adenomas
8 and carcinomas in male F344 rats exposed to DCA for 2 years. However, the data from
9 exposure concentrations at a 5 g/L dose had to be discarded and the 2.5 g/L DCA dose had to be
10 continuously lowered during the study due to neurotoxicity. There was a DCA-induced
11 increased in adenomas and carcinomas combined reported for the 0.5 g/L DCA (24.1 vs. 4.4%
12 adenomas and carcinomas combined in treated vs. controls) and an increase at a variable dose
13 started at 2.5 g/L DCA and continuously lowered (28.6 vs. 3.0% adenomas and carcinomas
14 combined in treated vs. controls). Only combined incidences of adenomas and carcinomas for
15 the 0.5 g/L DCA exposure group was reported to be statistically significant by the authors
16 although the incidence of adenomas was 17.2 vs. 4% in treated vs. control rats. Hepatocellular
17 tumor multiplicity was reported to be increased in the 0.5 g/L DCA group (0.31 adenomas and
18 carcinomas/animal in treated vs. 0.04 in control rats) but was reported by the authors to not be
19 statistically significant. At the starting dose of 2.5 g/L that was continuously lowered due to
20 neurotoxicity, the increased multiplicity of hepatocellular carcinomas was reported by the
21 authors to be to be statistically significant (0.25 carcinomas/animals vs. 0.03 in control) as well
22 as the multiplicity of combined adenomas and carcinomas (0.36 adenomas and
23 carcinomas/animals vs. 0.03 in control rats). Issues that affect the ability to determine the nature
24 of the dose-response for this study include (1) the use of a small number of animals (n = 23,
25 n = 21, and n = 23 at final sacrifice for the 2.0 g/L NaCl control, 0.05 g/L and 0.5 g/L treatment
26 groups) that limit the power of the study to both determine statistically significant responses and
27 to determine that there are not treatment-related effects (i.e., power) (2) apparent addition of
28 animals for tumor analysis not present at final sacrifice (i.e., 0.05 and 0.5 g/L treatment groups),
29 and (3) most of all, the lack of a consistent dose for the 2.5 g/L DCA exposed animals.
30 Similar issues are present for the study of Richmond et al. (1995) which was conducted
31 by the same authors as DeAngelo et al. (1996) and appeared to be the same data set. There was a
32 small difference in reports of the results between the two studies for the same data for the 0.5 g/L
33 DCA group in which Richmond et al. (1995) reported a 21% incidence of adenomas and
34 DeAngelo et al. (1996) reported a 17.2% incidence. The authors did not report any of the results
35 of DCA-induced increases of adenomas and carcinomas to be statistically significant. The same
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-283 DRAFT—DO NOT CITE OR QUOTE
-------
1 issues discussed above for DeAngelo et al. (1996) apply to this study. Similar to the DeAngelo
2 et al. (1997) study of TCA in rats, the use in these DCA studies (DeAngelo et al., 1996;
3 Richmond et al., 1995) of relatively small numbers of rats limits the detection of treatment-
4 related effects and the ability to determine whether there was no treatment related effects
5 (Type II error), especially at the low concentrations of DCA exposure.
6 For CH, George et al. (2000) exposed male F344/N rats to CH in drinking water for
7 2 years. Groups of animals were sacrificed at 13, 26, 52, and 78 weeks following the initiation
8 of dosing, with terminal sacrifices at Week 104. Only a few animals received a complete
9 pathological examination. The number of animals surviving >78 weeks and the number
10 examined for hepatocellular proliferative appeared to differ (42-44 animals examined but 32-35
11 surviving till the end of the experiment). Only the lowest treatment group had increased liver
12 tumors which were marginally significantly increased.
13 Leuschner and Beuscher (1998) examined the carcinogenic effects of CH in male and
14 female Sprague-Dawley rats (69-79 g, 25-29 days old at initiation of the experiment)
15 administered 0, 15, 45, and 135 mg/kg CH in unbuffered drinking water 7 days/week
16 (n = 50/group) for 124 weeks in males and 128 weeks in females. Two control groups were
17 noted in the methods section without explanation as to why they were conducted as two groups.
18 The authors report no substance-related influence on organ weights and no macroscopic evidence
19 of tumors or lesions in male or female rats treated with CH for 124 or 128 weeks. However, no
20 data are presented on the incidence of tumors in either treatment or control groups. The authors
21 did report a statistically significant increase in the incidence of hepatocellular hypertrophy in
22 male rats at the 135 mg/kg dose (14/50 animals vs. 4/50 and 7/50 in Controls I and II). For
23 female rats, the incidence of hepatocellular hypertrophy was reported to be 10/50 rats (Control I)
24 and 16/50 (Control II) rats with 18/50, 13/50 and 12/50 female rats having hepatocellular
25 hypertrophy after 15, 45, and 135 mg/kg CH, respectively. The lack of reporting in regard to
26 final body weights, histology, and especially background and treatment group data for tumor
27 incidences, limit the interpretation of this study. Whether this paradigm was sensitive for
28 induction of liver cancer cannot be determined.
29 Therefore, given the limitations in the available studies, a comparison of rat liver
30 carcinogenicity induced by TCE, TCA, DCA, and CH reveals no strong inconsistencies, but nor
31 does it provide much insight into the relative importance of different TCE metabolites in liver
32 tumor induction.
33
34 4.5.6.3.2. Studies in mice. Similar to TCE, the bioassay data in mice for DCA, TCA, and CH
35 are much more extensive and have shown that all three compounds induce liver tumors in mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-284 DRAFT—DO NOT CITE OR QUOTE
-------
1 Several 2-year bioassays have been reported for CH (Daniel et al., 1992; George et al., 2000;
2 Leakey et al., 2003a). For many of the DCA and TCA studies, the focus was not carcinogenic
3 dose-response but rather investigation of the nature of the tumors and potential MO As in relation
4 to TCE. As a result, studies often employed relatively high concentrations of DCA or TCA
5 and/or were conducted for a year or less. As shown previously in Section 4.5.4.2.1, the dose-
6 response curves for increased liver weight for TCE administration in male mice are more similar
7 to those for DCA administration and TCE oxidative metabolism than for direct TCA
8 administration (inadequate data were available for CH). An analogous comparison for DCA-,
9 TCA-, and CH-induced tumors would be informative, ideally using data from 2-year studies.
10
11 4.5.6.3.2.1. Trichloroethylene (TCE) carcinogenicitv dose-response data. Unfortunately, the
12 database for TCE, while consistently showing an induction of liver tumors in mice, is very
13 limited for making inferences regarding the shape of the dose-response curve. For many of these
14 experiments multiplicity was not given only liver tumor incidence. NTP (1990), Bull et al.
15 (2002), Anna et al. (1994) conducted gavage experiments in which they only tested one dose of
16 -1,000 mg/kg/d TCE. NCI (1976) tested two doses that were adjusted during exposure to an
17 average of 1,169 and 2,339 mg/kg/d in male mice with only 2-fold dose spacing in only 2 doses
18 tested. Maltoni et al. (1986) conducted inhalation experiments in two sets of B6C3F1 mice and
19 one set of Swiss mice at 3 exposure concentrations that were 3-fold apart in magnitude between
20 the low and mid-dose and 2-fold apart in magnitude between the mid- and high-dose. However,
21 for one experiment in male B6C3F1 mice (BT306), the mice fought and suffered premature
22 mortality and for two the experiments in B6C3F1 mice, although using the same strain, the mice
23 were obtained from differing sources with very different background liver tumor levels. For the
24 Maltoni et al. (1988) study a general descriptor of "hepatoma" was used for liver neoplasia rather
25 than describing hepatocellular adenomas and carcinomas so that comparison of that data with
26 those from other experiments is difficult. More importantly, while the number of adenomas and
27 carcinomas may be the same between treatments or durations of exposure, the number of
28 adenomas may decrease as the number of carcinomas increase during the course of tumor
29 progression. Such information is lost by using only a hepatoma descriptor.
30 Given the limited database, it would be useful if different studies could be combined to
31 yield a more comprehensive dose-response curve, as was done for liver weight, above. However,
32 this is probably not appropriate for several reasons. First, only NTP (1990) was performed with
33 dosing duration and time of sacrifice both being the "standard" 104 weeks. NCI (1976), Maltoni
34 et al. (1986), Anna et al. (1994), and Bull et al. (2002) all had shorter dosing periods and either
35 longer (Maltoni et al., 1986) or shorter (the other three studies) observation times. Therefore,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-285 DRAFT—DO NOT CITE OR QUOTE
-------
1 because of potential dose-rate effects and differences in the degree of expression of TCE-induced
2 tumors, it is difficult to even come up with a comparable administered dose metric across studies.
3 Moreover, the background tumor incidences are substantially different across experiments, even
4 controlling for mouse strain and sex. For example, across gavage studies in male B6C3F1 mice,
5 the incidence of hepatocellular carcinomas ranged from 1.2 to 16.7% (NCI, 1976; Anna et al.,
6 1994; NTP, 1990) and the incidence of adenomas ranged from 1.2 to 14.6% (Anna et al., 1994;
7 NTP, 1990) in control B6C3F1 mice. After -1,000 mg/kg/d TCE treatment, the incidence of
8 carcinomas ranged from 19.4 to 62% (Bull et al., 2002; NCI, 1976; Anna et al., 1994; NTP,
9 1990), with three of the studies (NCI, 1976; Anna et al., 1994; NTP, 1990) reporting a range of
10 incidences between 42.8 to 62.0%). The incidence of adenomas ranged from 28 to 66.7% (Bull et
11 al., 2002; Anna et al., 1994; NTP, 1990). In the Maltoni et al. (1986) inhalation study as well,
12 male B6C3F1 mice from two different sources had very different control incidences of hepatomas
13 (-2% versus about -20%).
14 Therefore, only data from the same experiment in which more than a single exposed dose
15 group was used provide reliable data on the dose-response relationship for TCE
16 hepatocarcinogenicity, and incidences from these experiments are shown in Figures 4-9 and
17 4-10. Except for one of the two Maltoni et al. (1986) inhalation experiments in male B6C3F1
18 mice, all of these data sets show relatively proportional increases with dose, albeit with
19 somewhat different slopes as may be expected across strains and sexes. Direct comparison is
20 difficult, since the "hepatomas" reported by Maltoni et al. (1986) are much more heterogeneous,
21 including neoplastic nodules, adenomas, and carcinomas, than the carcinomas reported by NCI
22 (1976). Nonetheless, although the data limitations preclude a conclusive statement, these data
23 are generally consistent with the linear relationship observed with TCE-induced liver weight
24 changes.
25
26 4.5.6.3.2.2. Dichloroacetic acid (DCA) carcinosenicity dose-response data. With respect to
27 DCA, Pereira (1996) reported that for 82 week exposure to DCA in female B6C3F1 mice, DCA
28 exposure concentrations of 0, 2, 6.67, and 20 mmol/L (0, 0.26, 0.86, and 2.6 g/L) led to close
29 proportionally increasing adenoma prevalences of 2.2, 6, 25, and 84.2%, though adenoma
30 multiplicity increased more than linearly between the highest two doses. Unfortunately, too few
31 carcinomas were observed at these doses and duration to meaningfully inform the shape of the
32 dose-response relationship. More useful is DeAngelo et al. (1999), which reported on a study of
33 DCA hepatocarcinogenicity in male B6C3F1 mice over a lifetime exposure. DeAngelo et al.
34 (1999) used 0.05 g/L, 0.5 g/L, 1.0 g/L, 2.0 g/L and 3.5 g/L exposure concentrations of DCA in
35 their 100-week dirking water study. The number of animals at final sacrifice was generally low
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-286 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
100%
0%
500 1000 1500 2000 2500
mg/kg-d (oral gavage)
-6HNCI76/B6C3F1 / F / oral
-A-NCI76/B6C3F1 / M /oral
• NTP90/B6C3F1 / F / oral
* NTP90 / B6C3F1 / M / oral
. Bull02 / B6C3F1 / M / oral (aqueous)
• Anna94 / B6C3F1 / M / oral (corn oil controls)
25 -,
500 1000 1500 2000 2500
mg/kg-d (oral gavage)
-NCI76/B6C3F1 / F / oral
-NCI76/B6C3F1 / M / oral
Figure 4-9. Dose-response relationship, expressed as (A) percent incidence
and (B) fold-increase over controls, for TCE hepatocarcinogenicity in NCI
(1976). For comparison, incidences of carcinomas for NTP (1990), Anna et al.
(1994), and Bull et al. (2002) are included, but without connecting lines since they
are not appropriate for assessing the shape of the dose-response relationship.
40%
200 400 600
ppm (7 hr/d, 5 d/wk)
800
200 400 600
ppm (7 hr/d, 5 d/wk)
800
9
10
11
12
13
14
-9- Maltoni86 / B6C3F1 / F / inhal
-&- Maltoni86 / B6C3F1 / M / inhal [BT306]
-A- Maltoni86 / B6C3F1 / M / inhal [BT306bis]
-B- Maltoni86 / Swiss / M / inhal
-9-Maltoni86 / B6C3F1 / F / inhal
-A- Maltoni86 / B6C3F1 / M / inhal [BT306]
-A- Maltoni86 / B6C3F1 / M / inhal [BT306bis]
-B- Maltoni86 / Swiss / M / inhal
Figure 4-10. Dose-response relationship, expressed as (A) incidence and (B)
fold-increase over controls, for TCE hepatocarcinogenicity in Maltoni et al.
(1986). Note that the BT306 experiment reported excessive mortality due to
fighting, and so the paradigm was repeated in experiment BT306bis using mice
from a different source.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-287 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
in the DC A treatment groups and variable. The multiplicity or number of hepatocellular
carcinomas/animals was reported to be significantly increased over controls in a dose-related
manner at all DCA treatments including 0.05 g/L DCA, and a no-observed-effect level (NOEL)
reported not to be observed by the authors. Between the 0.5 g/L and 3.5 g/L exposure
concentrations of DCA the magnitude of increase in multiplicity was similar to the increases in
magnitude in dose. The incidence of hepatocellular carcinomas were reported to be increased at
all doses as well but not reported to be statistically significant at the 0.05 g/L exposure
concentration. However, given that the number of mice examined for this response (n = 33), the
power of the experiment at this dose was only 16.9% to be able to determine that there was not a
treatment related effect. Indeed, Figure 4-11 replots the data from DeAngelo et al. (1999) with
an abscissa drawn to scale (unlike the figure in the original paper, which was not to scale),
suggests even a slightly greater than linear effect at the lowest dose (0.05 g/L, or 8 mg/kg/d) as
compared to the next lowest dose (0.5 g/L, or 84 mg/kg/d), though of course the power of such a
determination is limited. The authors did not report the incidence or multiplicity of adenomas
for the 0.05 g/L exposure group in the study or the incidence or multiplicity of adenomas and
carcinomas in combination. For the animals surviving from 79 to 100 weeks of exposure, the
incidence and multiplicity of adenomas peaked at 1 g/L while hepatocellular carcinomas
continued to increase at the higher doses. This would be expected where some portion of the
adenomas would either regress or progress to carcinomas at the higher doses.
100%
0%
0 100 200 300 400 500
DCA mg/kg-d
100 200 300 400 500
DCA mg/kg-d
Figure 4-11. Dose-response data for hepatocellular carcinomas (HC) (A)
incidence and (B) multiplicity, induced by DCA from DeAngelo et al. (1999).
Drinking water concentrations were 0, 0.05, 0.5, 1, 2, and 3.5 g/L, from which daily
average doses were calculated using observed water consumption in the study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-288 DRAFT—DO NOT CITE OR QUOTE
-------
1 Associations of DC A carcinogenicity with various noncancer, possibly precursor, effects
2 was also investigated. Importantly, the doses that induced tumors in DeAngelo et al. (1999)
3 were reported to not induce widespread cytotoxicity. An attempt was also made to relate
4 differing exposure levels to subchronic changes and peroxisomal enzyme induction.
5 Interestingly, DeAngelo et al. (1999) reported that peroxisome proliferation was significantly
6 increased at 3.5 g/L DCA only at 26 weeks, not correlated with tumor response, and to not be
7 increased at either 0.05 g/L or 0.5 g/L treatments. The authors concluded that DCA-induced
8 carcinogenesis was not dependent on peroxisome proliferation or chemically sustained
9 proliferation, as measured by DNA synthesis. Slight hepatomegaly was present by 26 weeks in
10 the 0.5 g/L group and decreased with time. By contrast, increases in both percent liver/body
11 weight and the multiplicity of hepatocellular carcinomas increased proportionally with DCA
12 exposure concentration after 79-100 weeks of exposure. DeAngelo et al. (1999) presented a
13 figure comparing the number of hepatocellular carcinomas/animal at 100 weeks compared with
14 the percent liver/body weight at 26 weeks that showed a linear correlation (r2 = 0.9977) while
15 peroxisome proliferation and DNA synthesis did not correlate with tumor induction profiles.
16 The proportional increase in liver weight with DCA exposure was also reported for shorter
17 durations of exposure as noted previously. Therefore, for DCA, both tumor incidence and liver
18 weight appear to increase proportionally with dose.
19
20 4.5.6.3.2.3. Trichloroacetic acid (TCA) carcinosenicity dose-response data. With respect to
21 TCA, Pereira (1996) reported that for 82 week exposure to TCA in female B6C3F1 mice, TCA
22 exposure concentrations of 0, 2, 6.67, and 20 mmol/L (0, 0.33, 1.1, and 3.3 g/L) led to increasing
23 incidences and multiplicity of adenomas and of carcinomas (Figure 4-12). DeAngelo et al.
24 (2008) reported the results of three experiments exposing male B6C3F1 mice to neutralized TCA
25 in drinking water (incidences also in Figure 4-12). Rather than using 5 exposure levels that were
26 generally 2-fold apart, as was done in DeAngelo et al. (1999) for DCA, DeAngelo et al. (2008)
27 studied only 3 doses of TCA that were an order of magnitude apart which limits the elucidation
28 of the shape of the dose-response curve. In addition, the 104-week data, DeAngelo et al. (2008)
29 contained 2 studies, each conducted in a separate laboratories—the two lower doses were studied
30 in one study and the highest dose in another. The first 104-week study was conducted using
31 2 g/L NaCl, or 0.05, 0.5, or 5 g/L TCA in drinking water for 60 weeks (Study #1) while the other
32 two were conducted for a period of 104 weeks (Study #2 with 2.5 g/L neutralized acetic acid or
33 4.5 g/L TCA exposure groups and Study #3 with deionized water, 0.05 g/L TCA and 0.5 g/L
34 TCA exposure groups). In addition, a relatively small number of animals were used for the
35 determination of a tumor response (n ~ 30 at final necropsy).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-289 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
100%
o
0%
2 4
TCA concentration (g/l)
DeAngelo et al. (2008) 60 wk (Study #1, M)
Pereira(1996)82 wk(F)
DeAngelo et al. (2008) 104 wk (Study #2, M)
DeAngelo et al. (2008) (Study #3, M)
2 4
TCA concentration (g/l)
DeAngelo et al. (2008) 60 wk (Study #1, M)
Pereira(1996)82wk(F)
DeAngelo et al. (2008) 104 wk (Study #2, M)
DeAngelo et al. (2008) (Study #3, M)
Figure 4-12. Reported incidences of hepatocellular carcinomas (HC) and
adenomas plus carcinomas (HA+HC) in various studies in B6C3F1 mice
(Pereira, 1996; DeAngelo et al., 2008). Combined HA + HC were not reported
in(Pereira, 1996).
In Study #1, the incidence data for adenomas observed at 60 weeks at 0.05 g/L, 0.5 g/L
and 5.0 g/L TCA were 2.1-, 3.0- and 5.4-fold of control values, with similar fold increases in
multiplicity. As shown by Pereira (1996), 60 weeks does not allow for full tumor expression, so
whether the dose-response relationship is the same at 104 weeks is not certain. For instance,
Pereira (1996) examined the tumor induction in female B6C3F1 mice and demonstrated that foci,
adenoma, and carcinoma development in mice are dependent on duration of exposure (period of
observation in controls). In control female mice a 360- vs. 576-day observation period showed
that at 360 days no foci or carcinomas and only 2.5% of animals had adenomas whereas by 576
days of observation, 11% had foci, 2% adenomas, and 2% had carcinomas. For DCA and TCA
treatments, foci, adenomas, and carcinoma incidence and multiplicity did not reach full
expression until 82 weeks at the 3 doses employed. Although the numbers of animals were
relatively low and variable at the two highest doses (18-28 mice) there were 50-53 mice studied
at the lowest dose level and 90 animals studied in the control group.
Therefore, the 104-week DeAngelo et al. (2008) data from Studies #2 and #3 would
generally be preferred for elucidating the TCA dose-response relationship. However, Study #2
was only conducted at one dose, and although Study #3 used lower doses, it exhibited
extraordinarily high control incidences of liver tumors. In particular, while the incidence of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-290 DRAFT—DO NOT CITE OR QUOTE
-------
1 adenomas and carcinomas was 12% in Study #2, it was reported to be 64% in Study #3. The
2 mice in Study #3 were of very large size (weighing -50 g at 45 weeks) as compared to Study #1,
3 Study #2, or most other bioassays in general, and the large background rate of tumors reported is
4 consistent with the body-weight-dependence observed by Leakey et al. (2003b).
5 To put into context the 64% incidence data for carcinomas and adenomas reported in
6 DeAngelo et al. (2008) for the control group of Study #3, other studies cited in this review for
7 male B6C3F1 mice show a much lower incidence in liver tumors with (1) NCI (1976) study of
8 TCE reporting a colony control level of 6.5% for vehicle and 7.1% incidence of hepatocellular
9 carcinomas for untreated male B6C3F1 mice (n = 70-77) at 78 weeks, (2) Herren-Freund et al.
10 (1987) reporting a 9% incidence of adenomas in control male B6C3F1 mice with a multiplicity
11 of 0.09 ± 0.06 and no carcinomas (n = 22) at 61 weeks, (3) NTP (1990) reporting an incidence of
12 14.6% adenomas and 16.6% carcinomas in male B6C3F1 mice after 103 weeks (n = 48), and
13 (4) Maltoni et al. (1986) reporting that B6C3F1 male mice from the "NCI source" had a
14 1.1% incidence of "hepatoma" (carcinomas and adenomas) and those from "Charles River Co."
15 had a 18.9% incidence of "hepatoma" during the entire lifetime of the mice (n = 90 per group).
16 The importance of examining an adequate number of control or treated animals before
17 confidence can be placed in those results in illustrated by Anna et al. (1994) in which at
18 76 weeks 3/10 control male B6C3F1 mice that were untreated and 2/10 control animals given
19 corn oil were reported to have adenomas but from 76 to 134 weeks, 4/32 mice were reported to
20 have adenomas (multiplicity of 0.13 ± 0.06) and 4/32 mice were reported to have carcinomas
21 (multiplicity of 0.12 ± 0.06). Thus, the reported combined incidence of carcinomas and
22 adenomas of 64% reported by DeAngelo et al. (2008) for the control mice of Study # 3, not only
23 is inconsistent and much higher than those reported in Studies #1 and #2, but also much higher
24 than reported in a number of other studies of TCE.
25 Therefore, this large background rate and the increased mortality for these mice limit
26 their use for determining the nature of the dose-response for TCA liver carcinogenicity. At the
27 two lowest doses of 0.05 g/L and 0.5 g/L TCA from Study #3, the differences in the incidences
28 and multiplicities for all tumors were 2-fold at 104 weeks. However, there was no difference in
29 any of the tumor results (i.e., adenoma, carcinoma, and combinations of adenoma and carcinoma
30 incidence and multiplicity) between the 4.5 g/L dose group in Study #2 and the 0.5 g/L dose
31 group in Study #3 at 104 weeks. By contrast, at 60 weeks of exposure, but within the same study
32 (Study #1), there was a 2-fold increase in multiplicity for adenomas, and for adenomas and
33 carcinomas combined between the 0.5 and 5.0 g/L TCA exposure groups. These results are
34 consistent with the two highest exposure levels reaching a plateau of response after a long
35 enough duration of exposure for full expression of the tumors (i.e., -90% of animals having liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-291 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
tumors at the 0.5 g/L and 5 g/L exposures). However, whether such a plateau would have been
observed in mice with a more "normal" body weight, and hence a lower background tumor
burden cannot be determined.
Because of the limitations of different studies, it is difficult to discern whether the liver
tumor dose-response curves of TCA and DCA are different in a way analogous to that for liver
weight (see Figure 4-13). Certainly, it is clear that at the same concentration in drinking water or
estimated applied dose, DCA is more potent than TCA, as DCA induces nearly 100% incidence
of carcinomas at a lower dose than TCA. Therefore, like with liver weight gains, DCA has a
steeper dose-response function than TCA. However, the evidence for a "plateau" in tumor
response at high doses with TCA, as was observed for liver weight, is equivocal, as it is
confounded by the highly varying background tumor rates and the limitations of the available
study paradigms.
0)
o
O
200 400
mg/kg-d
600
DeAngelo et al. (2008) (TCA Study #2)
DeAngelo et al. (2008) (TCA Study #3)
DeAngelo et al. (1999) (DCA)
Figure 4-13. Reported incidence of hepatocellular carcinomas induced by
DCA and TCA in 104-week studies (DeAngelo et al., 1999, 2008). Only
carcinomas were reported in DeAngelo et al. (1999), so combined adenomas and
carcinomas could not be compared.
DeAngelo et al. (2008) attempt to identify a NOEL for tumorigenicity using tumor
multiplicity data and estimated TCA dose. However, it is not an appropriate descriptor for these
data, especially given that "statistical significance" of the tumor response is the determinant used
by the authors to support the conclusions regarding a dose in which there is no TCA-induced
effect. Due to issues related to the appropriateness of use of the concurrent control in Study #3,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-292 DRAFT—DO NOT CITE OR QUOTE
-------
1 only the 60-week experiment (i.e., Study # 1) is useful for the determination of tumor dose-
2 response. Not only is there not allowance for full expression of a tumor response at the 60-week
3 time point but a power calculation of the 60-week study shows that the Type II error, which
4 should be >50% and thus, greater than the chances of "flipping a coin," was 41 and 71% for
5 incidence and 7 and 15% for multiplicity of adenomas for the 0.05 and 0.5 g/L TCA exposure
6 groups. For the combination of adenomas and carcinomas, the power calculation was 8 and 92%
7 for incidence and 6 and 56% for multiplicity at 0.05 and 0.5 g/L TCA exposure. Therefore, the
8 designed experiment could accept a false null hypothesis, especially in terms of tumor
9 multiplicity, at the lower exposure doses and erroneously conclude that there is no response due
10 to TCA treatment.
11 In terms of correlations with other noncancer, possibly precursor effects, DeAngelo et al.
12 (2008) also reported that PCO activity, which varied 2.7-fold as baseline controls, was 1.3-, 2.4-,
13 and 5.3-fold of control for the 0.05, 0.5, and 5 g/L TCA exposure groups in Study #1 at 4 weeks
14 was for adenomas incidence 2.1-, 3.0-, and 5.4-fold of control and not similar at the lowest dose
15 level at 60 weeks. However, it is not clear whether the similarly between PCO and
16 carcinogenicity at 60 weeks would persist for tumor incidence at 104 weeks. DeAngelo et al.
17 (2008) report a regression analyses that compare "percent of hepatocellular neoplasia," indicated
18 by tumor multiplicity, with TCA dose, represented by estimations of the TCA dose in mg/kg/d,
19 and with PCO activity for the 60-week and 104-week data. Whether adenomas and carcinomas
20 combined or individual tumor type were used in these analysis was not reported by the authors.
21 However, it would be preferable to compare "precursor" levels of PCO at earlier time points,
22 rather than at a time when there was already a significant tumor response. In addition, linear
23 regression analyses of these data are difficult to interpret because of the wide dose spacing of
24 these experiments. In such a situation, for a linear regression, control and 5 g/L exposure levels
25 will basically determine the shape of the dose-response curve since the 0.05 g/L and 0.5 g/L
26 exposure levels are so close to the control (0) value. Thus, dose response appears to be linear
27 between control and the 5.0 g/L value with the two lowest doses not affectively changing the
28 slope of the line (i.e., "leveraging" the regression). Moreover, at the 5 g/L dose level, there is
29 potential for effects due to palatability, as reported in one study in which drinking water
30 consumption declined at this concentration (DeAngelo et al., 2008). Thus, the value of these
31 analyses is limited by (1) use of data from Study # 3 in a tumor prone mouse that is not
32 comparable to those used in Studies #1 and #2, (2) the appropriateness of using PCO values from
33 later time points and the variability in PCO control values, (3) the uncertainty of the effects of
34 palatability on the 5 g/L TCA results which were reported in one study to reduce drinking water
35 consumption, and (4) the dose-spacing of the experiment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-293 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.5.6.3.2.4. Chloral hydrate (CH) carcinogenic dose-response. Although a much more limited
2 database in rodents than for TCA or DCA, there is evidence that chloral hydrate is also a rodent
3 liver hepatocarcinogen (see also Section E.2.5 and Caldwell and Keshava [2006]).
4 Daniel et al. (1992) exposed adult male B6C3F1 28-day-old mice to 1 g/L CH in drinking
5 water for 30 and 60 weeks (n = 5 for interim sacrifice) and for 104 weeks (n = 40). The
6 concentration of CH was 1 g/L and estimated to provide a 166-mg/kg/d dose. It is not clear from
7 the report what control group better matched the CH group, as the mean initial body weights of
8 the groups as well as the number of animals varied considerably in each group (i.e.,
9 -40% difference in mean body weights at the beginning of the study). Liver tumors were
10 increased by CH treatment. The percent incidence of liver carcinomas and adenomas in the
11 surviving animals was 15% in control and 71% in CH-treated mice and the incidence of
12 hepatocellular carcinoma reported to be 46% in the CH-treated group. The number of
13 tumors/animals was also significantly increased with CH treatment. However, because this was
14 a single dose study, a comparison with the dose-response relationship with TCE, TCA, or DCA
15 is not feasible.
16 George et al. (2000) exposed male B6C3F1 mice to CH in drinking water for 2 years.
17 Groups of animals were sacrificed at 26, 52, and 78 weeks following the initiation of dosing,
18 with terminal sacrifices at Week 104. Only a few animals received a complete pathological
19 examination. Preneoplastic foci and adenomas were reported to be increased in the livers of all
20 CH treatment groups at 104 weeks. The percent incidence of hepatocellular adenomas was
21 reported to be 21.4, 43.5, 51.3, and 50% in control, 13.5, 65.0 and 146.6 mg/kg/d CH treatment
22 groups, respectively. The percent incidence of hepatocellular carcinomas was reported to be
23 54.8, 54.3, 59.0 and 84.4% in these same groups. The resulting percent incidence of
24 hepatocellular adenomas and carcinomas was reported to be 64.3, 78.3, 79.5 and 90.6%. Of
25 concern is the reporting of a 64% incidence of hepatocellular carcinomas and adenomas in the
26 control group of mice for this experiment, which is the same as that for another study published
27 by this same laboratory (DeAngelo et al., 2008). DeAngelo et al. (2008) did not identify them as
28 being contemporaneous studies or sharing controls, but a comparison of the control data
29 published by DeAngelo et al. (2008) for TCA and that published by George et al. (2000) for the
30 CH studies shows them to be the same data set. Therefore, as discussed above, this data set was
31 derived from B6C3F1 mice that were large (-50 g) and resultantly tumor prone, making
32 determinations of the dose-response of CH from this experiment difficult. Therefore, for the
33 purposes of comparison of dose-response relationships, this study has the same limitations as the
34 DeAngelo et al. (2008) study, discussed above.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-294 DRAFT—DO NOT CITE OR QUOTE
-------
1 Leakey et al. (2003a) studied the effects of CH exposure (0, 25, 50, and 100 mg/kg/d,
2 5 days/week, 104-105 weeks via gavage) in male B6C3F1 mice with dietary control used to
3 manipulate body growth (n = 48 for 2-year study and n = 12 for the 15-month interim study).
4 Dietary control was reported to decrease background liver tumor rates (decreased by 15-20%)
5 and was reported to be associated with decreased variation in liver-to-body weight ratios, thereby
6 potentially increasing assay sensitivity. In dietary-controlled groups and groups fed ad libitum,
1 liver adenomas and carcinomas (combined) were reported to be increased with CH treatment.
8 With dietary restriction there was a more discernable CH tumor-response with overall tumor
9 incidence reduced, and time-to-tumor increased by dietary control in comparison to ad libitum
10 fed mice. Incidences of hepatocellular adenoma and carcinoma overall rates were reported to be
11 33, 52, 49, and 46% for control, 25, 50, and 100 mg/kg adlibitum-fed mice, respectively. For
12 dietary controlled mice the incidence rates were reported to be 22.9, 22.9, 29.2, and 37.5% for
13 controls, 25, 50, and 100 mg/kg CH, respectively. Body weights were matched and carefully
14 controlled in this study. These data are shown in Figure 4-14, relative to control incidences. It is
15 evident from these data that dietary control significantly changes the apparent shape of the dose-
16 response curve, presumably by reducing variability between animals. While the ad libitum dose
17 groups had an apparent "saturation" of response, this was not evident with the dietary controlled
18 group. Of note all the other bioassays for TCE, TCA, DCA, and CH were in ad libitum fed mice.
19 Therefore, it is difficult to compare the dose-response curves for CH-treated mice on dietary
20 restriction to those fed ad libitum. However, the rationale for dietary restriction in the B6C3F1
21 mouse is to prevent the types of weight gain and corresponding high background tumor levels
22 observed in DeAngelo et al. (2008) and George et al. (2000). As stated previously, most other
23 studies of TCA, DCA, and TCE had background levels that, while varied, were lower than the ad
24 libitum fed mice studied in Leakey et al. (2003a).
25 Of note is that incidences of adenomas and carcinomas combined do not show
26 differences in tumor progression as carcinomas may increase and adenomas may regress. Liver
27 weight increases at 15-months did not correlate with 2-year tumor incidences in the ad libitum
28 group, but a consistent dose-response shape between these two measures is evident in the dietary
29 controlled group. However, of note is the reporting of liver weight at 15 months is for a time
30 period in which foci and liver tumors have been reported to have already occurred in other
31 studies, so hepatomegaly in the absence of these changes is hard to detect.
32 In terms of other noncancer effects that may be associated with tumor induction, it is
33 notable that while dietary restriction reduced the overall level of CH-mediated tumor induction,
34 it led to greater CH-mediated induction of peroxisome proliferation-associated enzymes.
35 Moreover, between control groups, dietary restricted mice appeared to have higher levels of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-295 DRAFT—DO NOT CITE OR QUOTE
-------
60%
0%
20
40 60
CH mg/kg-d
80
100
ad libitum -B- dietary control
2 Figure 4-14. Effects of dietary control on the dose-response curves for
3 changes in liver tumor incidences induced by CH in diet (Leakey et al.,
4 2003a).
5
6
7 lauric acid co-hydrolase activity than ad libitum-fed mice. Seng et al. (2003) report that lauric
8 acid p-hydroxylase and PCO were induced only at exposure levels >100 mg/kg CH, again with
9 dietary restricted groups showing the greatest induction. Such data argue against the role of
10 peroxisome proliferation in CH-liver tumor induction in mice.
11 Leakey et al. (2003a) gave no descriptions of liver pathology were given other than
12 incidence of mice with fatty liver changes. Hepatic malondialdehyde concentration in ad libitum
13 fed and dietary controlled mice did not change with CH exposure at 15 months but the dietary
14 controlled groups were all approximately half that of the ad libitum-fed mice. Thus, while
15 overall increased tumors observed in the ad libitum diet correlated with increased
16 malondialdehyde concentration, there was no association between CH dose and malondialdehyde
17 induction for either diet.
18 Overall, from the CH studies in mice, there is an apparent increase in liver adenomas and
19 carcinomas induced by CH treatment by either drinking water or gavage with all available
20 studies performed in male B6C3F1 mice. However, the background levels of hepatocellular
21 adenomas and carcinomas in these mice in George et al. (2000) and body-weight data from this
22 study are high, consistent with the association between large body weight and background tumor
23 susceptibility shown with dietary control (Leakey et al., 2003a). With dietary control, Leakey et
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-296 DRAFT—DO NOT CITE OR QUOTE
-------
1 al. (2003a) report a dose-response relationship between exposure and tumor incidence that is
2 proportional to dose.
O
4 4.5.6.3.2.5. Degree of concordance among trichloroethylene (TCE), trichloroacetic acid
5 (TCA), dichloroacetic acid (DCA), and chloral hydrate (CH) dose-response relationships.
6 Comparison of the dose-response for TCE hepatocarcinogenicity with that for TCA and DCA is
7 weakly suggestive a better concordance in dose-response shape between TCE and DCA or TCE
8 and CH than between TCE and TCA. However, differences across the databases of these
9 compounds, especially with respect to the comparability of study durations and control tumor
10 incidences, preclude a definitive conclusion from these data.
11
12 4.5.6.3.3. Inferences from liver tumor phenotype and genotype. A number of studies have
13 investigation tumor phenotypes, such as c-Jun staining, tincture, and dysplacity, or genotypes,
14 such as H-ras mutations, to inform both the identification of the active agents of TCE liver tumor
15 induction as well as what MOA(s) may be involved.
16
17 4.5.6.3.3.1. Tumor phenotype—staining and appearance. The descriptions of tumors in mice
18 reported by the NCI, NTP, and Maltoni et al studies are also consistent with phenotypic
19 heterogeneity as well as spontaneous tumor morphology (see Section E.3.4.1.5). As noted in
20 Section E.3.1, hepatocellular carcinomas observed in humans are also heterogeneous. For mice,
21 Maltoni et al. (1986) described malignant tumors of hepatic cells to be of different subhistotypes,
22 and of various degrees of malignancy and were reported to be unique or multiple, and have
23 different sizes (usually detected grossly at necropsy) from TCE exposure. In regard to
24 phenotype, tumors were described as usual type observed in Swiss and B6C3F1 mice, as well as
25 in other mouse strains, either untreated or treated with hepatocarcinogens and to frequently have
26 medullary (solid), trabecular, and pleomorphic (usually anaplastic) patterns. For the NC I (1976)
27 study, the mouse liver tumors were described in detail and to be heterogeneous "as described in
28 the literature" and similar in appearance to tumors generated by carbon tetrachloride. The
29 description of liver tumors in this study and tendency to metastasize to the lung are similar to
30 descriptions provided by Maltoni et al. (1986) for TCE-induced liver tumors in mice via
31 inhalation exposure. The NTP (1990) study reported TCE exposure to be associated with
32 increased incidence of hepatocellular carcinoma (tumors with markedly abnormal cytology and
33 architecture) in male and female mice. Hepatocellular adenomas were described as
34 circumscribed areas of distinctive hepatic parenchymal cells with a perimeter of normal
35 appearing parenchyma in which there were areas that appeared to be undergoing compression
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-297 DRAFT—DO NOT CITE OR QUOTE
-------
1 from expansion of the tumor. Mitotic figures were sparse or absent but the tumors lacked typical
2 lobular organization. Hepatocellular carcinomas were reported to have markedly abnormal
3 cytology and architecture with abnormalities in cytology cited as including increased cell size,
4 decreased cell size, cytoplasmic eosinophilia, cytoplasmic basophilia, cytoplasmic vacuolization,
5 cytoplasmic hyaline bodies and variations in nuclear appearance. Furthermore, in many instance
6 several or all of the abnormalities were reported to be present in different areas of the tumor and
7 variations in architecture with some of the hepatocellular carcinomas having areas of trabecular
8 organization. Mitosis was variable in amount and location. Therefore, the phenotype of tumors
9 reported from TCE exposure was heterogeneous in appearance between and within tumors from
10 all 3 of these studies.
11 Caldwell and Keshava (2006) report "that Bannasch (2001) and Bannasch et al. (2001)
12 describe the early phenotypes of preneoplastic foci induced by many oncogenic agents (DNA-
13 reactive chemicals, radiation, viruses, transgenic oncogenes and local hyperinsulinism) as
14 insulinomimetic. These foci and tumors have been described by tincture as eosinophilic and
15 basophilic and to be heterogeneous. The tumors derived from them after TCE exposure are
16 consistent with the description for the main tumor lines of development described by Bannasch
17 etal. (2001) (see Section 3.4.1.5). Thus, the response of liver to DCA (glycogenesis with
18 emergence of glycogen poor tumors) is similar to the progression of preneoplastic foci to tumors
19 induced from a variety of agents and conditions associated with increased cancer risk."
20 Furthermore Caldwell and Keshava (2006) note that Bull et al. (2002) report expression of
21 insulin receptor to be elevated in tumors of control mice or mice treated with TCE, TCA and
22 DCA but not in nontumor areas suggesting that this effect is not specific to DCA.
23 There is a body of literature that has focused on the effects of TCE and its metabolites
24 after rats or mice have been exposed to "mutagenic" agents to "initiate" hepatocarcinogenesis
25 and this is discussed in Section E.4.2. TCE and its metabolites were reported to affect tumor
26 incidence, multiplicity, and phenotype when given to mice as a coexposure with a variety of
27 "initiating" agents and with other carcinogens. Pereira and Phelps (1996) reported that
28 methylnitrosourea (MNU) alone induced basophilic foci and adenomas. MNU and low
29 concentrations of DCA or TCA in female mice were reported to induce heterogeneous for foci
30 and tumor with a higher concentration of DCA inducing more eosinophilic and a higher
31 concentration of TCA inducing more tumors that were basophilic. Pereira et al. (2001) reported
32 that not only dose, but gender also affected phenotype in mice that had already been exposed to
33 MNU and were then exposed to DCA. As for other phenotypic markers, Lantendresse and
34 Pereira (1997) reported that exposure to MNU and TCA or DCA induced tumors that had some
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-298 DRAFT—DO NOT CITE OR QUOTE
-------
1 commonalities, were heterogeneous, but for female mice were overall different between DCA
2 and TCA as coexposures with MNU.
3 With regard to the phenotype of TCA and DCA-induced tumors, Stauber and Bull (1997)
4 reported the for male B6C3F1 mice, DCA-induced "lesions" contained a number of smaller
5 lesions that were heterogeneous and more eosinophilic with larger "lesions" tending to less
6 numerous and more basophilic. For TCA results using this paradigm, the "lesions" were
7 reported to be less numerous, more basophilic, and larger than those induced by DCA. Carter et
8 al. (2003) used tissues from the DeAngelo et al. (1999) and examined the heterogeneity of the
9 DCA-induced lesions and the type and phenotype of preneoplastic and neoplastic lesions pooled
10 across all time points. Carter et al. (2003) examined the phenotype of liver tumors induced by
11 DCA in male B6C3 Fl mice and the shape of the dose-response curve for insight into its MOA.
12 They reported a dose-response of histopathologic changes (all classes of premalignant lesions
13 and carcinomas) occurring in the livers of mice from 0.05-3.5 g/L DCA for 26-100 weeks and
14 suggest foci and adenomas demonstrated neoplastic progression with time at lower doses than
15 observed DCA genotoxicity. Preneoplastic lesions were identified as eosinophilic, basophilic
16 and/or clear cell (grouped with clear cell and mixed cell) and dysplastic. Altered foci were
17 50% eosinophilic with about 30% basophilic. As foci became larger and evolved into
18 carcinomas they became increasingly basophilic. The pattern held true through out the exposure
19 range. There was also a dose and length of exposure related increase in atypical nuclei in
20 "noninvolved" liver. Glycogen deposition was also reported to be dose-dependent with
21 periportal accumulation at the 0.5 g/L exposure level. Carter et al. (2003) suggested that size and
22 evolution into a more malignant state are associated with increasing basophilia, a conclusion
23 consistent with those of Bannasch (1996) and that there a greater periportal location of lesions
24 suggestive as the location from which they arose. Consistent with the results of DeAngelo et al.
25 (1999), Carter et al. (2003) reported that DCA (0.05-3.5 g/L) increased the number of lesions
26 per animal relative to animals receiving distilled water, shortened the time to development of all
27 classes of hepatic lesions, and that the phenotype of the lesions were similar to those
28 spontaneously arising in controls. Along with basophilic and eosinophilic lesions or foci,
29 Carter et al. (2003) concluded that DCA-induced tumors also arose from isolated, highly
30 dysplastic hepatocytes in male B6C3F1 mice chronically exposed to DCA suggesting another
31 direct neoplastic conversion pathway other than through eosinophilic or basophilic foci.
32 Rather than male B6C3F1 mice, Pereira (1996) studied the dose-response relationship for
33 the carcinogenic activity of DCA and TCA and characterized their lesions (foci, adenomas and
34 carcinomas) by tincture in females (the generally less sensitive gender). Like the studies of TCE
35 by Maltoni et al. (1986), female mice were also reported to have increased liver tumors after
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-299 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCA and DC A exposures. Pereira (1996) pool lesions were pooled for phenotype analysis so the
2 affect of duration of exposure could not be determined nor adenomas separated from carcinomas
3 for "tumors." However, as the concentration of DCA was decreased the number of foci was
4 reported by Pereira (1996) to be decreased but the phenotype of the foci to go from primarily
5 eosinophilic foci (i.e., -95% eosinophilic at 2.58 g/L DCA) to basophilic foci (-57%
6 eosinophilic at 0.26 g/L). For TCA the number of foci was reported to -40 basophilic and
7 -60 eosinophilic regardless of dose. Spontaneously occurring foci were more basophilic by a
8 ratio of'7/3. Pereira (1996) described the foci of altered hepatocytes and tumors induced by
9 DCA in female B6C3F1 mice to be eosinophilic at higher exposure levels but at lower or
10 intermittent exposures to be half eosinophilic and half basophilic. Regardless of exposure level,
11 half of the TCA-induced foci were reported to be half eosinophilic and half basophilic with
12 tumors 75% basophilic. In control female mice, the limited numbers of lesions were mostly
13 basophilic, with most of the rest being eosinophilic with the exception of a few mixed tumors.
14 The limitations of descriptions tincture and especially for inferences regarding peroxisome
15 proliferator from the description of "basophilia" is discussed in Section E.3.4.1.5.
16 Thus, the results appear to differ between male and female B6C3F1 mice in regard to
17 tincture for DCA and TCA at differing doses. What is apparent is that the tincture of the lesions
18 is dependent on the stage of tumor progression, agent (DCA or TCA), gender, and dose. Also
19 what is apparent from these studies is the both DCA and TCA are heterogeneous in their tinctoral
20 characteristics.
21 Overall, tumors induced by TCA, DCA, CH, and TCE are all heterogeneous in their
22 physical and tinctural characteristics in a manner this not markedly distinguishable from
23 spontaneous lesions or those induced by a wide variety of chemical carcinogens. For instance,
24 Daniel et al. (1992), which studies DCA and CH carcinogenicity (discussed above) noted that
25 morphologically, there did not appear to be any discernable differences in the visual appearance
26 of the DCA- and CH-induced tumors. Therefore, these data do not provide strong insights into
27 elucidating the active agent(s) for TCE hepatocarcinogenicity or their MOA(s).
28
29 4.5.6.3.3.2. C-Jun staining. Stauber and Bull (1997) reported that in male B6C3F1 mice, the
30 oncoproteins c-Jun and c-Fos were expressed in liver tumors induced by DCA but not those
31 induced by TCA. Although Bull et al. (2004) have suggested that the negative expression of
32 c-Jun in TCA-induced tumors may be consistent with a characteristic phenotype shown in
33 general by peroxisome proliferators as a class, as pointed out by Caldwell and Keshava (2006),
34 there is no supporting evidence of this. Nonetheless, the observation that TCA and DCA have
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-300 DRAFT—DO NOT CITE OR QUOTE
-------
1 different levels of oncogene expression led to a number of follow-up studies by this group. No
2 data on oncoprotein immunostaining are available for CH.
3 Stauber et al. (1998) studied induction of "transformed" hepatocytes by DC A and TCE
4 treatment in vitro, including an examination of c-Jun staining. Stauber et al. (1998) isolated
5 primary hepatocytes from 5-8 week old male B6C3F1 mice (n = 3) and subsequently cultured
6 them in the presence of DCA or TCA. In a separate experiment 0.5 g/L DCA was given to mice
7 as pretreatment for 2 weeks prior to isolation. The authors assumed that the anchorage-
8 independent growth of these hepatocytes was an indication of an "initiated cell." After 10 days
9 in culture with DCA or TCA (0, 0.2, 0.5 and 2.0 mM), concentrations of 0.5 mM or more DCA
10 and TCA both induced an increase in the number of colonies that was statistically significant,
11 with DCA showing dose-dependence as well as slightly greater overall increases than TCA. In a
12 time course experiment the number of colonies from DCA treatment in vitro peaked by 10 days
13 and did not change through Days 15-25 at the highest dose and, at lower concentrations of DCA,
14 increased time in culture induced similar peak levels of colony formation by Days 20-25 as that
15 reached by 10 days at the higher dose. Therefore, the number of colonies formed was
16 independent of dose if the cells were treated long enough in vitro. However, not only did
17 treatment with DCA or TCA induce anchorage independent growth but untreated hepatocytes
18 also formed larger numbers of colonies with time, although at a lower rate than those treated
19 with DCA. The level reached by untreated cells in tissue culture at 20 days was similar to the
20 level induced by 10 days of exposure to 0.5 mM DCA. The time course of TCA exposure was
21 not tested to see if it had a similar effect with time as did DCA. The colonies observed at
22 10 days were tested for c-Jun expression with the authors noting that "colonies promoted by
23 DCA were primarily c-Jun positive in contrast to TCA promoted colonies that were
24 predominantly c-Jun negative." Of the colonies that arose spontaneously from tissue culture
25 conditions, 10/13 (76.9%) were reported to be c-Jun +, those treated with DCA 28/34 (82.3%)
26 were c-Jun +, and those treated with TCA 5/22 (22.7%) were c-Jun +. Thus, these data show
27 heterogeneity in cell in colonies but with more that were c-Jun + colonies occurring by tissue
28 culture conditions alone than in the presence of DCA, rather than in the presence of TCA.
29 Bull et al. (2002) administered TCE, TCA, DCA, and combinations of TCA and DCA to
30 male B6C3F1 mice by daily gavage (TCE) or drinking water (TCA, DCA, and TCA+DCA) for
31 52-79 weeks, in order to compare a number of tumor characteristics, including c-Jun expression,
32 across these different exposures. Bull et al. (2002) reported lesion reactivity to c-Jun antibody to
33 be dependent on the proportion of the DCA and TCA administered after 52 weeks of exposure.
34 Given alone, DCA was reported to produce lesions in mouse liver for which approximately half
35 displayed a diffuse immunoreactivity to a c-Jun antibody, half did not, and none exhibited a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-301 DRAFT—DO NOT CITE OR QUOTE
-------
1 mixture of the two. After TCA exposure alone, no lesions were reported to be stained with this
2 antibody. When given in various combinations, DCA and TCA coexposure induced a few
3 lesions that were only c-Jun+, many that were only c-Jun-, and a number with a mixed phenotype
4 whose frequency increased with the dose of DCA. For TCE exposure of 79 weeks, TCE-induced
5 lesions were reported to also have a mixture of phenotypes (42% c-Jun+, 34% c-Jun-, and
6 24% mixed) and to be most consistent with those resulting from DCA and TCA coexposure but
7 not either metabolite alone.
8 A number of the limitations of the experiment are discussed in Caldwell et al. (2008)
9 Specifically, for the DCA and TCA exposed animals, the experiment was limited by low
10 statistical power, a relatively short duration of exposure, and uncertainty in reports of lesion
11 prevalence and multiplicity due to inappropriate lesions grouping (i.e., grouping of hyperplastic
12 nodules, adenomas, and carcinomas together as "tumors"), and incomplete histopathology
13 determinations (i.e., random selection of gross lesions for histopathology examination). For
14 determinations of immunoreactivity to c-Jun, Bull et al. (2002) combined hyperplastic nodules,
15 adenomas, and carcinomas in most of their treatment groups, so differences in c-Jun expression
16 across differing types of lesions were not discernable.
17 Nonetheless, these data collectively strongly suggest that TCA is not the sole agent of
18 TCE-induced mouse liver tumors. In particular, TCE-induced tumors that were, in order of
19 frequency, c-Jun+, c-Jun-, and of mixed phenotype, while c-Jun+ tumors have never been
20 observed with TCA treatment. Nor do these data support DCA as the sole contributor, since
21 mixed phenotypes were not observed with DCA treatment.
22
23 4.5.6.3.3.3. Tumor genotype: H-ras mutation frequency and spectrum. An approach to
24 determine the potential MO As of DCA and TCA through examination of the types of tumors
25 each "induced" or "selected" was to examine H-ras activation (Ferreira-Gonzalez et al., 1995;
26 Anna et al., 1994; Bull et al., 2002; Nelson et al., 1990). No data of this type were available for
27 CH. This approach has also been used to try to establish an H-ras activation pattern for
28 "genotoxic" and "nongenotoxic" liver carcinogens compounds and to make inferences
29 concerning peroxisome proliferator-induced liver tumors. However, as noted by Stanley et al.
30 (1994), the genetic background of the mice used and the dose of carcinogen may affect the
31 number of activated H-ras containing tumors which develop. In addition, the stage of
32 progression of "lesions" (i.e., foci vs. adenomas vs. carcinomas) also has been linked the
33 observance of H-ras mutations. Fox et al. (1990) note that tumors induced by phenobarbital
34 (0.05% drinking water [H2O], 1 year), chloroform (200 mg/kg corn oil gavage, 2 times weekly
35 for 1 year) or ciprofibrate (0.0125% diet, 2 years) had a much lower frequency of H-ras gene
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-302 DRAFT—DO NOT CITE OR QUOTE
-------
1 activation than those that arose spontaneously (2-year bioassays of control animals) or induced
2 with the "genotoxic" carcinogen benzidine-2 hydrochloric acid (HC1) (120 ppm, drinking IH^O,
3 1 year) in mice. In that study, the term "tumor" was not specifically defined but a correlation
4 between the incidence of H-ras gene activation and development of either a hepatocellular
5 adenoma or hepatocellular carcinoma was reported to be made with no statistically significant
6 difference between the frequency of H-ras gene activation in the hepatocellular adenomas and
7 carcinomas. Histopathological examination of the spontaneous tumors, tumors induced with
8 benzidine-2 HC1, Phenobarbital, and chloroform was not reported to reveal any significant
9 changes in morphology or staining characteristics. Spontaneous tumors were reported to have
10 64% point mutation in codon 61 (n = 50 tumors examined) with a similar response for Benzidine
11 of 59% (n = 22 tumors examined), whereas for Phenobarbital the mutation rate was 7%
12 (n = 15 tumors examined), chloroform 21% (n = 24 tumors examined) and ciprofibrate 21%
13 (n = 39 tumors examined). The ciprofibrate-induced tumors were reported to be more
14 eosinophilic as were the surrounding normal hepatocytes.
15 Hegi et al. (1993) tested ciprofibrate-induced tumors in the NIH3T3 cotransfection-nude
16 mouse turnorigenicity assay, which the authors state is capable of detecting a variety of activated
17 protooncogenes. The tumors examined (ciprofibrate-induced or spontaneously arising) were
18 taken from the Fox et al. study (1990), screened previously, and found to be negative for H-ras
19 activation. With the limited number of samples examined, Hegi et al concluded that ras
20 protooncogene activation or activation of other protooncogenes using the nude mouse assay were
21 not frequent events in ciprofibrate-induced tumors and that spontaneous tumors were not
22 promoted with it. Using the more sensitive methods, the H-ras activation rate was reported to be
23 raised from 21 to 31% for ciprofibrate-induced tumors and from 64 to 66% for spontaneous
24 tumors. Stanley et al. (1994) studied the effect of methylclofenapate (MCP) (25 mg/kg for up to
25 2 years), a peroxisome proliferator, in B6C3F1 (relatively sensitive) and C57BL/10J (relatively
26 resistant) mice for H-ras codon 61 point mutations in MCP-induced liver tumors (hepatocellular
27 adenomas and carcinomas). In the B6C3F1 mice the number of tumors with codon 61 mutations
28 was 11/46 and for C57BL/10J mice 4/31. Unlike the findings of Fox et al. (1990), Stanley et al.
29 (1994) reported an increase in the frequency of mutation in carcinomas, which was reported to be
30 twice that of adenomas in both strains of mice, indicating that stage of progression was related to
31 the number of mutations in those tumors, although most tumors induced by MCP did not have
32 this mutation.
33 Anna et al. (1994) reported that the H-ras codon 61 mutation frequency was not
34 statistically different in liver tumors from DCA and TCE-treated mice from a highly variable
35 number of tumors examined. From their concurrent controls, they reported that H-ras codon 61
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-303 DRAFT—DO NOT CITE OR QUOTE
-------
1 mutations in 17% (n = 6) of adenomas and 100% (n = 5) of carcinomas. For historical controls
2 (published and unpublished), they reported mutations in 73% (n = 33) of adenomas and
3 mutations in 70% (n = 30) of carcinomas. For tumors from TCE-treated animals, they reported
4 mutations in 35% (n = 40) of adenomas and 69% (n = 36) of carcinomas, while for DCA-treated
5 animals, they reported mutations in 54% (n = 24) of adenomas and in 68% (n = 40) of
6 carcinomas. Anna et al. (1994) reported more mutations in TCE-induced carcinomas than
7 adenomas. In regard to mutation spectra in H-ras oncogenes in control or spontaneous tumors,
8 the patterns were slightly different but those from TCE treatment were mostly similar to that of
9 DCA-induced tumors (0.5% in drinking water).
10 The study of Ferreira-Gonzalez (1995) in male B6C3 Fl mice has the advantage of
11 comparison of tumor phenotype at the same stage of progression (hepatocellular carcinoma), for
12 allowance of the full expression of a tumor response (i.e., 104 weeks), and an adequate number
13 of spontaneous control lesions for comparison with DCA or TCA treatments. However, tumor
14 phenotype at an end stage of tumor progression may not be indicative of earlier stages of the
15 disease process. In spontaneous liver carcinomas, 58% were reported to show mutations in H-61
16 as compared with 50% of tumor from 3.5 g/L DCA-treated mice and 45% of tumors from
17 4.5 g/L TCA-treated mice. A number of peroxisome proliferators have been reported to have a
18 much smaller mutation frequency that spontaneous tumors (e.g., 13-24% H-ras codon 61
19 mutations after methylclofenopate depending on mouser strain, Stanely et al. [1994]: 21 to 31%
20 for ciprofibrate-induced tumors and from 64 to 66% for spontaneous tumors, Fox et al. [1990]
21 and Hegi et al [1993]). Thus, there was a heterogeneous response for this phenotypic marker for
22 the spontaneous, DCA-, and TCA- treatment induced hepatocellular carcinomas had similar
23 patterns H-ras mutations that differed from the reduced H-ras mutation frequencies reported for a
24 number of peroxisome proliferators.
25 In his review, Bull (2000) suggested "the report by Anna et al. (1994) indicated that
26 TCE-induced tumors possessed a different mutation spectra in codon 61 of the H-ras oncogene
27 than those observed in spontaneous tumors of control mice." Bull (2000) stated that "results of
28 this type have been interpreted as suggesting that a chemical is acting by a mutagenic
29 mechanism" but went on to suggest that it is not possible to a priori rule out a role for selection
30 in this process and that differences in mutation frequency and spectra in this gene provide some
31 insight into the relative contribution of different metabolites to TCE-induced liver tumors. Bull
32 (2000) noted that data from Anna et al. (1994), Ferreira-Gonzalez et al. (1995), and Maronpot et
33 al. (1995) indicated that mutation frequency in DCA-induced tumors did not differ significantly
34 from that observed in spontaneous tumors. Bull (2000) also noted that the mutation spectra
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-304 DRAFT—DO NOT CITE OR QUOTE
-------
1 found in DCA-induced tumors has a striking similarity to that observed in TCE-induced tumors,
2 and DCA-induced tumors were significantly different than that of TCA-induced liver tumors.
3 Bull et al. (2002) reported that mutation frequency spectra for the H-ras codon 61 in
4 mouse liver "tumors" induced by TCE (« = 37 tumors examined) were reported to be
5 significantly different than that for TCA (n = 41 tumors examined), with DCA-treated mice
6 tumors giving an intermediate result (n = 64 tumors examined). In this experiment,
7 TC A-induced "tumors" were reported to have more mutations in codon 61 (44%) than those
8 from TCE (21%) and DCA (33%). This frequency of mutation in the H-ras codon 61 for TCA is
9 the opposite pattern as that observed for a number of peroxisome proliferators in which the
10 number of mutations at H-ras codon 61 in tumors has been reported to be much lower than
11 spontaneously arising tumors (see above). Bull et al. (2002) noted that the mutation frequency
12 for all TCE, TCA or DCA tumors was lower in this experiment than for spontaneous tumors
13 reported in other studies (they had too few spontaneous tumors to analyze in this study), but that
14 this study utilized lower doses and was of shorter duration than that of Ferreira-Gonzalez (1995).
15 Furthermore, the disparities from previous studies may also be impacted by lesion grouping,
16 mentioned above, in which lower stages of progression are grouped with more advanced stages.
17 Overall, in terms of H-ras mutation, TCE-induced tumors appears to be more like
18 DCA-induced tumors (which are consistent with spontaneous tumors), or those resulting from a
19 coexposure to both DCA and TCA (Bull et al., 2002), than from those induced by TCA. As
20 noted above, Bull et al. (2002) reported the mutation frequency spectra for the H-ras codon 61 in
21 mouse liver tumors induced by TCE to be significantly different than that for TCA, with
22 DCA-treated mice tumors giving an intermediate result and for TCA-induced tumors to have a
23 H-ras profile that is the opposite than those of a number of other peroxisome proliferators. More
24 importantly, however, these data, along with the measures discussed above, show that mouse
25 liver tumors induced by TCE are heterogeneous in phenotype and genotype in a manner similar
26 to that observed in spontaneous tumors.
27
28 4.5.6.3.4. "Stop" experiments. Several stop experiments, in which treatment is terminated
29 early in some dose groups, have attempted to ascertain the whether progression differences exist
30 between TCA and DCA. After 37 weeks of treatment and then a cessation of exposure for
31 15 weeks, Bull et al. (1990) reported that after combined 52 week period, liver weight and
32 percent liver/body weight were reported to still be statistically significantly elevated after DCA
33 or TCA treatment. The authors partially attribute the remaining increases in liver weight to the
34 continued presence of hyperplastic nodules in the liver. In terms of liver tumor induction, the
35 authors stated that "statistical analysis of tumor incidence employed a general linear model
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-305 DRAFT—DO NOT CITE OR QUOTE
-------
1 ANOVA with contrasts for linearity and deviations from linearity to determine if results from
2 groups in which treatments were discontinued after 37 weeks were lower than would have been
3 predicted by the total dose consumed." The multiplicity of tumors (incidence was not used)
4 observed in male mice exposed to DCA or TCA at 37 weeks and then sacrificed at 52 weeks
5 were compared with those exposed for a full 52 weeks. The response in animals that received
6 the shorter duration of DCA exposure was very close to that which would be predicted from the
7 total dose consumed by these animals. By contrast, the response to TCA exposure for the shorter
8 duration was reported by the authors to deviate significantly (p = 0.022) from the linear model
9 predicted by the total dose consumed. However, in the prediction of "dose-response," foci,
10 adenomas, and carcinomas were combined into one measure. Therefore, foci, a certain
11 percentage of which have been commonly shown to spontaneously regress with time, were
12 included in the calculation of total "lesions." Moreover, only a sample of lesions were selected
13 for histological examination, and as is evident in the sample, some lesions appeared "normal"
14 upon microscopic examination (see below). Therefore, while suggesting that cessation of
15 exposure diminished the number of "lesions," methodological limitations temper any
16 conclusions regarding the identity and progression of lesion with continuous vs. noncontinuous
17 DCA and TCA treatment.
18 Additionally, Bull et al. (1990) noted that after stopping treatment, DCA lesions appeared
19 to arrest their progression in contrast to TCA lesions, which appeared to progress. In particular,
20 among those in the stop treatment group (at 2 g/L) with 0/19 lesions examined histologically
21 were carcinomas, while in the continuous treatment groups, a significant fraction of lesions
22 examined were carcinomas at the higher exposure (6/23 at 2 g/L). By contrast, at terminal
23 sacrifice, TCA lesions a larger fraction of the lesions examined were carcinomas in the stop
24 treatment group (3/5 at 2 g/L) than in the continuous treatment group (2/7 and 4/16 at 1 g/L and
25 2 g/L, respectively).
26 However, as mentioned above, these inferences are based on examination of only a
27 subset of lesions. Specifically, for TCA treatment the number of animals examined for
28 determination of which "lesions" were foci, adenomas, and carcinomas was 11 out of the
29 19 mice with "lesions" at 52 weeks while all 4 mice with lesions after 37 weeks of exposure and
30 15 weeks of cessation were examined. For DCA treatment the number of animals examined was
31 only 10 out of 23 mice with "lesions" at 52 weeks while all 7 mice with lesions after 37 weeks of
32 exposure and 15 weeks of cessation were examined. Most importantly, when lesions were
33 examined microscopically, some did not all turn out to be preneoplastic or neoplastic—for
34 example, two lesions appeared "to be histologically normal" and one necrotic.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-306 DRAFT—DO NOT CITE OR QUOTE
-------
1 While limited, the conclusions of Bull et al. (1990) are consistent with later experiments
2 performed by Pereira and Phelps (1996). They noted that in MNU-treated mice that were then
3 treated with DCA, the yield of altered hepatocytes decreases as the tumor yields increase
4 between 31 and 51 weeks of exposure suggesting progression of foci to adenomas, but that
5 adenomas did not appear to progress to carcinomas. For TCA, Pereira and Phelps (1996)
6 reported that "MNU-initiated" adenomas promoted with TCA continued to progress. However,
7 the use of MNU initiation complicates direct comparisons with treatment with TCA or DCA
8 alone.
9 No similar data comparing stop and continued treatment of TCE are available to assess
10 the consistency or lack-thereof with TCA or DCA. Moreover, the informative of such a
11 comparison would be limited by designs of the available TCA and DCA studies, which have
12 used higher concentrations in conjunction with the much lower durations of exposure. While
13 higher doses allow for responses to be more easily detected, it introduces uncertainty as to the
14 effects of the higher doses alone. In addition, because the overall duration of the experiments is
15 also generally much less than 104 weeks, it is not possible to discern whether the differences in
16 results between those animals in which treatment was suspended in comparison to those in which
17 had not had been conducted would persist with longer durations.
18
19 4.5.6.4. Conclusions Regarding the Role of Trichloroacetic Acid (TCA), Dichloroacetic Acid
20 (DCA), and Chloral Hydrate (CH) in Trichloroethylene (TCE)-Induced Effects in
21 the Liver
22 In summary, it is likely that oxidative metabolism is necessary for TCE-induced effects in
23 the liver. However, the specific metabolite or metabolites responsible for both noncancer and
24 cancer effects is less clear. TCE, TCA, and DCA exposures have all been associated with
25 induction of peroxisomal enzymes but are all weak PPARa agonists. The available data strongly
26 support TCA not being the sole or predominant active moiety for TCE-induced liver effects.
27 With respect to hepatomegaly, TCE and TCA dose-response relationships are quantitatively
28 inconsistent, for TCE leads to greater increases in liver/body weight ratios that expected from
29 predicted rates of TCA production. In fact, above a certain dose of TCE, liver/body weight
30 ratios are greater than that observed under any conditions studied so far for TCA. Histological
31 changes and effects on DNA synthesis are generally consistent with contributions from either
32 TCA or DCA, with a degree of polyploidization, rather than cell proliferation, likely to be
33 significant for TCE, TCA, and DCA. With respect to liver tumor induction, TCE leads to a
34 heterogeneous population of tumors, not unlike those that occur spontaneously or that are
35 observed following TCA-, DCA-, or CH-treatment. Moreover, some liver phenotype
36 experiments, particularly those utilizing immunostaining for c-Jun, support a role for both DCA
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-307 DRAFT—DO NOT CITE OR QUOTE
-------
1 and TCA in TCE-induced tumors, with strong evidence that TCA cannot solely account for the
2 characteristics of TCE-induced tumors. In addition, H-ras mutation frequency and spectrum of
3 TCE-induced tumors more closely resembles that of spontaneous tumors or of those induced by
4 DC A, and were less similar in comparison to that of TCA-induced tumors. The heterogeneity of
5 TCE-induced tumors is similar to that observed to be induced by a broad category of
6 carcinogens, and to that observed in human liver cancer. Overall, then, it is likely that multiple
7 TCE metabolites, and therefore, multiple pathways, contribute to TCE-induced liver tumors.
8
9 4.5.7. Mode of Action (MOA) for Trichloroethylene (TCE) Liver Carcinogenicity
10 This section will discuss the evidentiary support for several hypothesized modes of action
11 for liver carcinogenicity (including mutagenicity and peroxisome proliferation, as well as several
12 additional proposed hypotheses and key events with limited evidence or inadequate experimental
13 support), following the framework outlined in the Cancer Guidelines (U.S. EPA, 2005a, b).6
14
15 4.5.7.1. Mutagenicity
16 The hypothesis is that TCE acts by a mutagenic mode of action in TCE-induced
17 hepatocarcinogenesis. According to this hypothesis, the key events leading to TCE-induced liver
18 tumor formation constitute the following: TCE oxidative metabolite CH, after being produced in
19 the liver, cause direct alterations to DNA (e.g., mutation, DNA damage, and/or micronuclei
20 induction). Mutagenicity is a well established cause of carcinogenicity.
21
22 Experimental support for the hypothesized mode of action. The genotoxicity, as described by
23 the ability of TCE, CH, TCA, and DCA to induce mutations, was discussed previously in
24 Section 4.2. The strongest data for mutagenic potential are for CH, thought to be a relatively
25 short-lived intermediate in the metabolism of TCE that is rapidly converted to TCA and TCOH
26 in the liver (see Section 3.3). CH causes a variety of genotoxic effects in available in vitro and in
27 vivo assays, with particularly strong data as to its ability to induce aneuploidy. It has been
28 argued that CH mutagenicity is unlikely to be the cause of TCE carcinogenicity because the
29 concentrations required to elicit these responses are generally quite high, several orders of
6 As recently reviewed (Guyton et al, 2008) the approach to evaluating mode of action information described in US
EPA's Cancer Guidelines (2005a, b) considers the issue of human relevance of a hypothesized mode of action in the
context of hazard evaluation. This excludes, for example, consideration of toxicokinetic differences across species;
specifically, the Cancer Guidelines state, "the toxicokinetic processes that lead to formation or distribution of the
active agent to the target tissue are considered in estimating dose but are not part of the mode of action." In
addition, information suggesting quantitative differences in the occurrence of a key event between test species and
humans are noted for consideration in the dose-response assessment, but is not considered in human relevance
determination.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-308 DRAFT—DO NOT CITE OR QUOTE
-------
1 magnitude higher that achieved in vivo (Moore and Harrington-Brock, 2000). For example, peak
2 concentrations of CH in the liver of around 2-3 mg/kg have been reported after TCE
3 administration at doses that are hepatocarcinogenic in chronic bioassays (Abbas and Fisher,
4 1997; Greenberg et al., 1999). Assuming a liver density of about 1 kg/L, these concentrations
5 are orders of magnitude less than the minimum concentrations reported to elicit genotoxic
6 responses in the Ames test and various in vitro measures of micronucleus, aneuploidy, and
7 chromosome aberrations, which are in the 100-1,000 mg/L range. However, it is not clear how
8 much of a correspondence is to be expected from concentrations in genotoxicity assays in vitro
9 and concentrations in vivo, as reported in vivo CH concentrations are in whole-liver homogenate
10 while in vitro concentrations are in culture media. In addition, a few in vitro studies have
11 reported positive results at concentrations as low as 1 or 10 mg/L, including Furnus et al. (1990)
12 for aneuploidy in Chinese hamster CHED cells (10 mg/L), Eichenlaub-Ritter et al. (1996) for
13 bivalent chromosomes in meiosis I in MF1 mouse oocytes (10 mg/L), and Gibson et al. (1995)
14 for cell transformation in Syrian hamster embryo cells after 7 day treatment. Moreover, some in
15 vivo genotoxicity assays of CH reported positive results at doses similar to those eliciting a
16 carcinogenic response in chronic bioassays. For example, Nelson and Bull (1988) reported
17 increased DNA single strand breaks at 100 CH mg/kg (oral) in male B6C3F1 mice, although the
18 result was not replicated by Chang et al. (1992). In another example, four of six in vivo mouse
19 genotoxicity studies reported that CH induced micronuclei in mouse bone-marrow erythrocytes,
20 with the lowest effective doses in positive studies ranging from 83 to 500 mg/kg (positive: Russo
21 and Levis [1992], Russo et al. [1992], Marrazini et al. [1994], Beland et al. [1999]; negative:
22 Leuschner and Leuschner [1991], Leopardi et al. [1993]). However, the use of i.p.
23 administration in these and many other in vivo genotoxicity assays complicates the comparison
24 with carcinogenicity data. Also, it is difficult with the available data to assess the contributions
25 from the genotoxic effects of CH along with those from the genotoxic and nongenotoxic effects
26 of other oxidative metabolites (discussed below in Sections 4.5.5.2 and 4.5.5.3).
27 Furthermore, altered DNA methylation, another heritable mechanism by which gene
28 expression may be altered, is discussed below in the in Section 4.5.1.3.2.6. As discussed
29 previously, the differential patterns of H-ras mutations observed in liver tumors induced by TCE,
30 TCA, and DC A may be more indicative of tumor selection and tumor progression resulting from
31 exposure to these agents rather than a particular mechanism of tumor induction. The state of the
32 science of cancer and the role of epigenetic changes, in addition to genetic changes, in the
33 initiation and progression of cancer and specifically liver cancer, are discussed in Section E.3.1.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-309 DRAFT—DO NOT CITE OR QUOTE
-------
1 Therefore, while data are insufficient to conclude that a mutagenic MOA mediated by CH
2 is operant, a mutagenic MOA, mediated either by CH or by some other oxidative metabolite of
3 TCE, cannot be ruled out.
4
5 4.5.7.2. Peroxisome Proliferator Activated Receptor Alpha (PPARa) Receptor Activation
6 The hypothesis is that TCE acts by a PPARa agonism MOA in TCE-induced
7 hepatocarcinogenesis. According to this hypothesis, the key events leading to TCE-induced liver
8 tumor formation constitute the following: the TCE oxidative metabolite TCA, after being
9 produced in the liver, activates the PPARa receptor, which then causes alterations in cell
10 proliferation and apoptosis and clonal expansion of initiated cells. This MOA is assumed to
11 apply only to the liver.
12
13 Experimental support for the hypothesized mode of action. Proliferation of peroxisomes and
14 increased activity of a number of related marker enzymes has been observed in rodents treated
15 with TCE, TCA, and DCA. The peroxisome-related effects of TCE are most likely mediated
16 primarily through TCA based on TCE metabolism producing more TCA than DCA and the
17 lower doses of TCA required to elicit a response relative to DCA. However, Bull (2004) and
18 Bull et al. (2004) have recently suggested that peroxisome proliferation occurs at higher
19 exposure levels than those that induce liver tumors for TCE and its metabolites. They report that
20 a direct comparison in the no-effect level or low-effect level for induction of liver tumors in the
21 mouse and several other endpoints shows that, for TCA, liver tumors occur at lower
22 concentrations than peroxisome proliferation in vivo but that PPARa activation occurs at a lower
23 dose than either tumor formation or peroxisome proliferation. A similar comparison for DCA
24 shows that liver tumor formation occurs at a much lower exposure level than peroxisome
25 proliferation or PPARa activation. In vitro transactivation studies have shown that human and
26 murine versions of PPARa are activated by TCA and DCA, while TCE itself is relatively
27 inactive in the in vitro system, at least with mouse PPARa (Maloney and Waxman, 1999; Zhou
28 and Waxman, 1998). In addition, Laughter et al. (2004) reported that the responses of AGO,
29 PCO, and CYP4A induction by TCE, TCA, and DCA were substantially diminished in
30 PPARa-null mice. Therefore, evidence suggests that TCE, through its metabolites TCA and
31 DCA, activate PPARa, and that at doses relevant to TCE-induced hepatocarcinogenesis, the role
32 of TCA in PPARa agonism is likely to predominate.
33 It has been suggested that PPARa receptor activation is both the MOA for TCA liver
34 tumor induction as well as the MOA for TCE liver tumor induction, as a result of the metabolism
35 of TCE to TCA (NRC, 2006; Corton, 2008). Section E.3.4 addressed the status of the PPARa
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-310 DRAFT—DO NOT CITE OR QUOTE
-------
1 MOA hypothesis for liver tumor induction and provides a more detailed discussion. However, as
2 discussed previously and in Section E.2.1.10, TCE-induced increases in liver weight have been
3 reported in male and female mice that do not have a functional PPARa receptor (Nakajima et al.,
4 2000). The dose-response for TCE-induced liver weight increases differs from that of TCA (see
5 Section E.2.4.2). The phenotype of the tumors induced by TCE have been described to differ
6 from those by TCA and to be more like those occurring spontaneously in mice, those induced by
7 DCA, or those resulting from a combination of exposures to both DC A and TCA (see
8 Section E.2.4.4). As to whether TCA induces tumors through activation of the PPARa receptor,
9 the tumor phenotype of TC A-induced mouse liver tumors has been reported to have a different
10 pattern of H-ras mutation frequency from other peroxisome proliferators (see Section E.2.4.4;
11 Bull et al., 2002; Stanely et al., 1994; Fox et al., 1990; Hegi et al., 1993). While TCE, DCA, and
12 TCA are weak peroxisome proliferators, liver weight induction from exposure to these agents
13 has not correlated with increases in peroxisomal enzyme activity (e.g., PCO activity) or changes
14 in peroxisomal number or volume. By contrast, as discussed above, liver weight induction from
15 subchronic exposures appears to be a more accurate predictor of carcinogenic response for DCA,
16 TCA and TCE in mice (see also Section E.2.4.4). The database for cancer induction in rats is
17 much more limited than that of mice for determination of a carcinogenic response to these
18 chemicals in the liver and the nature of such a response.
19 While many compounds known to cause rodent liver tumors with long-term treatment
20 also activate the nuclear receptor PPARa, the mechanisms by which PPARa activation
21 contributes to tumorigenesis are not completely known (Klaunig et al., 2003; NRC, 2006;
22 Yang et al., 2007). As reviewed by Keshava and Caldwell (2006), PPARa activation leads to a
23 highly pleiotropic response and may play a role in toxicity in multiple organs as well as in
24 multiple chronic conditions besides cancer (obesity, atherosclerosis, diabetes, inflammation).
25 Klaunig et al. (2003) and NRC (2006) proposed that the key causal events for PPARa agonist-
26 induced liver carcinogenesis, after PPARa activation, are perturbation of cell proliferation and/or
27 apoptosis, mediated by gene expression changes, and selective clonal expansion. It has also been
28 proposed that sufficient evidence for this MOA consists of evidence of PPARa agonism (i.e., in
29 a receptor assay) in combination with either light- or electron-microscopic evidence for
30 peroxisome proliferation or both increased liver weight and one more of the in vivo markers of
31 peroxisome proliferation (Klaunig et al., 2003). However, it should be noted that peroxisome
32 proliferation and in vivo markers such as PCO are not considered causal events (Klaunig et al.,
33 2003; NRC, 2006), and that their correlation with carcinogenic potency is poor (Marsman et al.,
34 1988). Therefore, for the purposes of this discussion, peroxisome proliferation and its markers
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-311 DRAFT—DO NOT CITE OR QUOTE
-------
1 are considered indicators of PPARa activation, as it is well established that these highly specific
2 effects are mediated through PPARa (Klaunig et al., 2003; Peters et al., 1997).
3 As recently reviewed by Guyton et al. (2009), recent data suggest that PPARa activation
4 along with these hypothesized causal events may not be sufficient for carcinogenesis. In
5 particular, Yang et al. (2007) reported comparisons between mice treated with Wy-14643 and
6 transgenic mice in which PPARa was constitutively activated in hepatocytes without the
7 presence of ligand. Yang et al. (2007) reported that, in contrast to Wy-14643-treatment, the
8 transgene did not induce liver tumors at 11 months, despite inducing PPARa-mediated effects of
9 a similar type and magnitude seen in response to tumorigenic doses of Wy-14643 in wild-type
10 mice (decreased serum fatty acids, induction of PPARa target genes, altered expression of cell-
11 cycle control genes, and a sustained increase in cellular proliferation). Nonetheless, it is
12 important to discuss the extent to which PPARa activation mediates the effects proposed by
13 Klaunig et al. (2003) and NRC (2006), even if the hypothesized sequence of key events may not
14 be sufficient for carcinogenesis. Investigation continues into additional events that may also
15 contribute, such as nonparenchymal cell activation and micro-RNA-based regulation of
16 protooncogenes (Yang et al., 2007; Shah et al., 2007). Specifically addressed below are gene
17 expression changes, proliferation, clonal expansion, and mutation frequency or spectrum.
18 With respect to gene expression changes due to TCE, Laughter et al. (2004) evaluated
19 transcript profiles induced by TCE in wild-type and PPARa-null mice. As noted in
20 Sections E.3.4.1.3 and E.3.1.2, there are limitations to the interpretation of such studies, some of
21 which are discussed below. Also noted in Appendix E are discussions of how studies of
22 peroxisome proliferators, indicate of the need for phenotypic anchoring, especially since gene
23 expression is highly variable between studies and within studies using the same experimental
24 paradigm. Section E.3.4 in also provides detailed discussions of the status of the PPARa
25 hypothesis. Of note, all null mice at the highest TCE dose (1,500 mg/kg/d) were moribund prior
26 to the end of the planned 3-week experiment(Laughter et al., 2004), and it was proposed that this
27 may reflect a greater sensitivity in PPARa-null mice to hepatotoxins due to defects in tissue
28 repair abilities. Laughter et al. (2004) also noted that four genes known to be regulated by other
29 peroxisome proliferators also had altered expression with TCE treatment in wild-type, but not
30 null mice. However, in a comparative analysis, Bartosiewicz et al. (2001) concluded that TCE
31 induced a different pattern of transcription than two other peroxisome proliferators,
32 di(2-ethylhexyl) phthalate (DEHP) and clofibrate. In addition, Keshava and Caldwell (2006)
33 compared gene expression data from Wy-14643, dibutyl phthalate (DBF), GEM, and DEHP, and
34 noted a lack of consistent results across PPARa agonists. Thus, available data are insufficient to
35 conclude that TCE gene expression changes are similar to other PPAR agonists, or even that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-312 DRAFT—DO NOT CITE OR QUOTE
-------
1 there are consistent changes (beyond the in vivo markers of peroxisome proliferation, such as
2 AGO, PCO, CYP4A, etc.) among different agonists. It should also be noted that Laughter et al.
3 (2004) did not compare baseline (i.e., control levels of) gene expression between null and wild-
4 type control mice, hindering interpretation of these results (Keshava and Caldwell, 2006). The
5 possible relationship between PPARa activation and hypomethylation are discussed below in
6 Section 4.5.7.1.9.
7 In terms of proliferation, mitosis itself has not been examined in PPARa-null mice, but
8 BrdU incorporation, a measure of DNA synthesis that may reflect cell division, polyploidization,
9 or DNA repair, was observed to be diminished in null mice as compared to wild-type mice at 500
10 and 1,000 mg/kg/d TCE (Laughter et al., 2004). However, BrdU incorporation in null mice was
11 still about 3-fold higher than controls, although it was not statistically significantly different due
12 to the small number of animals, high variability, and the 2- to 3-fold higher baseline levels of
13 BrdU incorporation in control null mice as compared to control wild-type mice. Therefore,
14 while PPARa appears to contribute to the short-term increase in DNA synthesis observed with
15 TCE treatment, these results cannot rule out other contributing mechanisms. However, since it is
16 likely that both cellular proliferation and increased ploidy contribute to the observed TCE-
17 induced increases in DNA synthesis, it is not clear to whether the observed decrease in BrdU
18 incorporation is due to reduced proliferation, reduced polyploidization, or both.
19 With respect to clonal expansion, it has been suggested that tumor characteristics such as
20 tincture (i.e., the staining characteristics light microscopy sections of tumor using H&E stains)
21 and oncogene mutation status can be used to associate chemical carcinogens with a particular
22 MOA such as PPARa agonism (Klaunig et al., 2003; NRC, 2006). This approach is problematic
23 primarily because of the lack of specificity of these measures. For example, with respect to
24 tincture, it has been suggested that TCA-induced foci and tumors resemble those of other
25 peroxisome proliferators in basophilia and lack of expression of GOT and GST-pi. However, as
26 discussed in Caldwell and Keshava (2006), the term "basophilic" in describing foci and tumors
27 can be misleading, because, for example, multiple lineages of foci and tumors exhibit basophilia,
28 including those not associated with peroxisome proliferators (Bannasch, 1996; Bannasch et al.,
29 2001; Carter et al., 2003). Moreover, a number of studies indicate that foci and tumors induced
30 by other "classic" peroxisome proliferators may have different phenotypic characteristics from
31 that attributed to the class through studies of WY-14643, including DEHP (Voss et al., 2005) and
32 clofibric acid (Michel et al., 2007). Furthermore, even the combination of GGT and GST-pi
33 negative, basophilic foci are nonspecific to peroxisome proliferators, as they have been observed
34 in rats treated with AfBl and AfBl plus PB, none of which are peroxisome proliferators
35 (Kraupp-Grasl et al., 1998; Grasl-Kraupp et al., 1993). Finally, while Bull et al. (2004)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-313 DRAFT—DO NOT CITE OR QUOTE
-------
1 suggested that negative expression ofc-jun in TCA-induced tumors may be consistent with a
2 characteristic phenotype of peroxisome proliferators, no data could be located to support this
3 statement. Therefore, of phenotypic information does not appear to be reliable for associating a
4 chemical with a PPARa agonism MOA.
5 Mutation frequency or spectrum in oncogenes has also been suggested to be an indicator
6 of a PPARa agonism MOA being active (NRC, 2006), with the idea being that specific
7 genotypes are being promoted by PPARa agonists. Although not a highly specific marker, H-ras
8 codon 61 mutation frequency and spectra data do not support a similarity between mutations in
9 TCE-induced, TCA-, or DCA- tumors and those due to other peroxisome proliferators. For
10 example, while ciprofibrate and methylclofenopate had lower mutation frequencies than
11 historical controls (Hegi et al., 1993; Stanley et al., 1994), TCA-induced tumors had mutation
12 frequencies similar to or higher than historical controls (Ferreira-Gonzalez et al., 1995; Bull et
13 al., 2002). Anna et al. (1994) and Ferreira-Gonzalez et al. (1995) also reported TCE and DCA-
14 induced tumors to have mutation frequencies similar to historical controls, although Bull et al.
15 (2002) reported lower frequencies for these chemicals. However, the data reported by Bull et al.
16 (2002) consist of mixed lesions at different stages of progression, and such differing stages, in
17 addition to differences in genetic background and dose, can influence the frequency of H-ras
18 mutations (Stanley et al., 1994). In addition, a greater frequency of mutations was reported in
19 carcinomas than adenomas, and Bull et al. (2002) stated that this suggested that H-ras mutations
20 were a late event. Moreover, Fox et al. (1990) noted that tumors induced by phenobarbital,
21 chloroform, and ciprofibrate all had a much lower frequency of H-ras gene activation than those
22 that arose spontaneously, so this marker does not have good specificity. Mutation spectrum is
23 similarly of low utility for supporting a PPARa agonism MOA. First, because many peroxisome
24 proliferators been reported to have low frequency of mutations, the comparison of mutation
25 spectrum would be limited to a small fraction tumors. In addition to the low power due to small
26 numbers, the mutation spectrum is relatively nonspecific, as Fox et al. (1990) reported that of the
27 tumors with mutations, the spectra of the peroxisome proliferator ciprofibrate, historical controls,
28 and the genotoxic carcinogen benzidine-2 HC1 were similar.
29 In summary, TCE clearly activates PPARa, and some of the effects contributing to
30 tumorigenesis that Klaunig et al. (2003) and NRC (2006) propose to be the result of PPARa
31 agonism are observed with TCE, TCA, or DCA treatment. While this consistency is supportive a
32 role for PPARa, all of the proposed key causal effects with the exception of PPARa agonism
33 itself are nonspecific, and may be caused by multiple mechanisms. There is more direct
34 evidence that several of these effects, including alterations in gene expression and changes in
35 DNA synthesis, are mediated by multiple mechanisms in the case of TCE, and a causal linkage
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-314 DRAFT—DO NOT CITE OR QUOTE
-------
1 to PPARa specifically is lacking. Therefore, because, as discussed further in the MOA
2 discussion below, there are multiple lines of evidence supporting the role of multiple pathways
3 of TCE-induced tumorigenesis, the hypothesis that PPARa agonism and the key causal events
4 proposed by Klaunig et al. (2003) and NRC (2006) constitute the sole or predominant MOA for
5 TCE-induced carcinogenesis is considered unlikely.
6 Furthermore, as reviewed by Guyton et al. (2009), recent data strongly suggest that
7 PPARa and key events hypothesized by Klaunig et al. (2003) are not sufficient for
8 carcinogenesis induced by the purported prototypical agonist Wy-14643. Therefore, the
9 proposed PPARa MOA is likely "incomplete" in the sense that the sequence of key events7
10 necessary for cancer induction has not been identified. A recent 2-year bioassay of the
11 peroxisome proliferator DEHP showed that it can induce a liver tumor response in mice lacking
12 PPARa similar to that in wild-type mice (Ito et al., 2007). Klaunig et al. (2003) previously
13 concluded that PPARa agonism was the sole MOA for DEHP-induced liver tumorigenesis based
14 on the lack of tumors in PPARa-null mice after 11 months treatment with Wy-14643 (Peters et
15 al., 1997). They also assumed that due to the lack of markers of PPARa agonism in PPARa-null
16 mice after short-term treatment with DEHP (Ward et al., 1998), a long-term study of DEHP in
17 PPARa-null mice would yield the same results as for Wy-14643. However, due the finding by
18 Ito et al. (2007) that PPARa-null mice exposed to DEHP do develop liver tumors, they
19 concluded that DEHP can induce liver tumors by multiple mechanisms (Ito et al., 2007;
20 Takashima et al., 2008). Hence, since there is no 2-year bioassay in PPARa-null mice exposed
21 to TCE or its metabolites, it is not justifiable to use a similar argument based on Peters et al.
22 (1997) and short-term experiments to suggest that the PPARa MOA is operative. Therefore, the
23 conclusion is supported that the hypothesized PPARa MOA is inadequately specified because
24 the data do not adequately show the proposed key events individually being required for
25 hepatocarcinogenesis, nor do they show the sequence of key events collectively to be sufficient
26 for hepatocarcinogenesis.
27
7 As defined by the U.S. EPA Cancer Guidelines (2005a, b) a "key event" is "an empirically observable precursor
step that is itself a necessary element of the mode of action or is a biologically based marker for such an element,"
and the term "mode of action" (MOA) is defined as "a sequence of key events and processes, starting with
interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer
formation." Therefore, a single key event alone is necessary, but not necessarily sufficient for carcinogenesis;
however, the sequence of key events constituting a MOA needs to be sufficient for carcinogenesis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-315 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.5.7.3. Additional Proposed Hypotheses and Key Events with Limited Evidence or
2 Inadequate Experimental Support
3 Several effects that been hypothesized to be associated with liver cancer induction are
4 discussed in more detail below, including increased liver weight, DNA hypomethylation, and
5 pathways involved in glycogen accumulation such as insulin signaling proteins. As discussed
6 above, TCE and its metabolites reportedly increase nuclear size and ploidy in hepatocytes, and
7 these effects likely account for much of the increases in labeling index and DNA synthesis
8 caused by TCE. Importantly, these changes appear to persist with cessation of treatment, with
9 liver weights, but not nuclear sizes, returning to control levels(Kjellstrand et al., 1983a). In
10 addition, glycogen deposition, DNA synthesis, increases in mitosis, or peroxisomal enzyme
11 activity do not appear correlated with TCE-induced liver weight changes.
12
13 4.5.7.3.1. Increased liver weight. Increased liver weight or liver/body weight ratios
14 (hepatomegaly) is associated with increased risk of liver tumors in rodents, but it is relatively
15 nonspecific (Allen et al., 2004). The evidence presented above for TCE and its metabolites
16 suggest a similarity in dose-response between liver weight increases at short-term durations of
17 exposure and liver tumor induction observed from chronic exposure. Liver weight increases may
18 results from several concurrent processes that have been associated with increase cancer risk
19 (e.g., hyperplasia, increased ploidy, and glycogen accumulation) and when observed after
20 chronic exposure may result from the increased presence of foci and tumors themselves.
21 Therefore, there are inadequate data to adequately define a MOA hypothesis for
22 hepatocarcinogenesis based on liver weight increases.
23
24 4.5.7.3.2. "Negative selection." As discussed above, TCE, TCA, and DCA all cause transient
25 increases in DNA synthesis. This DNA synthesis has been assumed to result from proliferation
26 of hepatocytes. However, the dose-related TCA- and DCA-induced increases in liver weight not
27 correlate with patterns of DNA synthesis; moreover, there have been reports that DNA synthesis
28 in individual hepatocytes does not correlate with whole liver DNA synthesis measures
29 (Sanchez and Bull, 1990; Carter et al., 1995). With continued treatment, decreases in DNA
30 synthesis have been reported for DCA (Carter et al., 1995). More importantly, several studies
31 show that transient DNA synthesis is confined to a very small population of cells in the liver in
32 mice exposed to TCE for 10 days or to DCA or TCA for up to 14 days of exposure. Therefore,
33 generalized mitogenic stimulation is not likely to play a role in TCE-induced liver
34 carcinogenesis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-316 DRAFT—DO NOT CITE OR QUOTE
-------
1 Bull has proposed that the TCE metabolites TCA and DCA may contribute to liver tumor
2 induction through so-called "negative selection" by way of several possible processes
3 (Bull, 2000). First, it is hypothesized that the mitogenic stimulation by continued TCA and DCA
4 exposure is down-regulated in normal hepatocytes, conferring a growth advantage to initiated
5 cells that either do not exhibit the down-regulation of response or are resistant to the down-
6 regulating signals. This is implausible as both the normal rates of cell division in the liver and
7 the TCE-stimulated increases are very low. Polyploidization has been reported to decrease the
8 normal rates of cell division even further. That the transient and relatively low level of DNA
9 synthesis reported for TCE, DCA, and TCA is reflective of proliferation rather than
10 polyploidization is not supported by data on mitosis. A mechanism for such "down-regulation"
11 has not been identified experimentally.
12 A second proposed contributor to "negative-selection" is direct enhancement by TCA and
13 DCA in the growth of certain populations of initiated cells. While differences in phenotype of
14 end stage tumors have been reported between DCA and TCA, the role of selection and
15 emergence of potentially different foci has not been elucidated. Neither have pathway
16 perturbations been identified that are common to liver cancer in human and rodent for TCE,
17 DCA, and TCA. The selective growth of clones of hepatocytes that may progress fully to cancer
18 is a general feature of cancer and not specific to at TCE, TCA, or DCA MOA.
19 A third proposed mechanism by which TCE may enhance liver carcinogenesis within this
20 "negative selection" paradigm is through changing apoptosis. However, as stated above, TCE
21 has been reported to either not change apoptosis or to cause a slight increase at high doses.
22 Rather than increases in apoptosis, peroxisome proliferators have been suggested to inhibit
23 apoptosis as part of their carcinogenic MOA. However, the age and species studied appear to
24 greatly affect background rates of apoptosis (Snyder et al., 1995) with the rat having a greater
25 rate of apoptosis than the mouse. DCA has been reported to induce decreases in apoptosis in the
26 mouse (Carter et al., 1995; Snyder et al., 1995). However, the significance of the DCA-induced
27 reduction in apoptosis, from a level that is already inherently low in the mouse, for the MOA for
28 induction of DCA-induce liver cancer is difficult to discern.
29 Therefore, for a MOA for hepatocarcinogenesis based on "negative selection," there are
30 inadequate data to adequately define the MOA hypothesis, or the available data do not support
31 such a MOA being operative.
32
33 4.5.7.3.3. Polyploidization. Polyploidization may be an important key event in tumor
34 induction. For example, in addition to TCE, partial hepatectomy, nafenopin, methylclofenopate,
35 DEHP, diethylnitrosamine, jV-nitrosomorpholine, and various other exposures that contribute to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-317 DRAFT—DO NOT CITE OR QUOTE
-------
1 liver tumor induction also shift the hepatocyte ploidy distribution to be increasingly diploid or
2 polypoid (Hasmal and Roberts, 2000; Styles et al., 1988; Melchiorri et al., 1993; Miller et al.,
3 1996; Vickers et al., 1996). As discussed by Gupta (2000), "[w]orking models indicate that
4 extensive polyploidy could lead to organ failure, as well as to oncogenesis with activation of
5 precancerous cell clones." However, the mechanism(s) by which increased polypoidy enhances
6 carcinogenesis is not currently understood. Due to increased DNA content, polypoid cells will
7 generally have increased gene expression. However, polyploid cells are considered more highly
8 differentiated and generally divide more slowly and are more likely to undergo apoptosis,
9 perhaps thereby indirectly conferring a growth advantage to initiated cells (see Section E. 1). Of
10 note is that changes in ploidy have been observed in transgenic mouse models that are also prone
11 to develop liver cancer (see Section E.3.3.1). It is likely that polyploidization occurs with TCE
12 exposure and it is biologically plausible that polyploidization can contribute to liver
13 carcinogenesis, although the mechanism(s) is (are) not known. However, whether
14 polyploidization is necessary for TCE-induced carcinogenesis is not known, as no experiment in
15 which polyploidization specifically is blocked or diminished has been performed and the extent
16 of polyploidization has not been quantified. Therefore, there are inadequate data to adequately
17 define a MOA hypothesis for hepatocarcinogenesis based on polyploidization.
18
19 4.5.7.3.4. Glycogen storage. As discussed above, several studies have reported that DCA
20 causes accumulation of glycogen in mouse hepatocytes. Such glycogen accumulation has been
21 suggested to be pathogenic, as it is resistant to mobilization by fasting (Kato-Weinstein et al.,
22 1998). In humans, glycogenesis due to glycogen storage disease or poorly controlled diabetes
23 has been associated with increased risk of liver cancer (LaVecchia et al., 1994; Adami et al.,
24 1996; Wideroff et al., 1997; Rake et al., 2002). Glycogen accumulation has also been reported to
25 occur in rats exposed to DCA.
26 For TCE exposure in mice or rats, glycogen content of hepatocytes has been reported to
27 be somewhat less than or the same as controls, or not remarked upon in the studies. TCA
28 exposure has been reported to decrease glycogen content in rodent hepatocytes while DCA has
29 been reported to increase it (Kato-Weinstein et al., 2001). There is also evidence that DCA-
30 induced increases in glycogen accumulation are not proportional to liver weight increases and
31 only account for a relatively small portion of increases in liver mass. DCA-induced increases in
32 liver weight are not a function of cellular proliferation but probably include hypertrophy
33 associated with polyploidization, increased glycogen deposition and other factors.
34 While not accounting for increases in liver weight, excess glycogen can still be not only
35 be pathogenic but a predisposing condition for hepatocarcinogenesis. Some hypotheses
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-318 DRAFT—DO NOT CITE OR QUOTE
-------
1 regarding the possible relationship between glycogenesis and carcinogenesis have been posed
2 that lend them biological plausibility. Evert et al. (2003), using an animal model of hepatocyte
3 exposure to a local hyperinsulinemia from transplanted islets of Langerhans with remaining
4 tissue is hypoinsulinemic, reported that insulin induces alterations resembling preneoplastic foci
5 of altered hepatocytes that develop into hepatocellular tumors in later stages of carcinogenesis.
6 Lingohr et al. (2001) suggest that normal hepatocytes down-regulate insulin-signaling proteins in
7 response to the accumulation of liver glycogen caused by DCA and that the initiated cell
8 population, which does not accumulate glycogen and is promoted by DCA treatment, responds
9 differently from normal hepatocytes to the insulin-like effects of DCA. Bull et al. (Bull et al.,
10 2002) reported increased insulin receptor protein expression in tumor tissues regardless of
11 whether they were induced by TCE, TCA, or DCA. Given the greater activity of DCA relative
12 to TCA on carbohydrate metabolism, it is unclear whether changes in these pathways are causes
13 or simply reflect the effects of tumor progression. Therefore, it is biologically plausible that
14 changes in glycogen status may occur from the opposing actions of TCE metabolites, but
15 changes in glycogen content due to TCE exposure has not been quantitatively studied. The
16 possible contribution of these effects to TCE-induced hepatocarcinogenesis is unclear.
17 Therefore, there are inadequate data to adequately define a MO A hypothesis for TCE-induced
18 hepatocarcinogenesis based on changes in glycogen storage or even data to support increased
19 glycogen storage to result from TCE exposure.
20
21 4.5.7.3.5. Inactivation of GST-zeta. DCA has been shown to inhibit its own metabolism in that
22 pretreatment in rodents prior to a subsequent challenge dose leads to a longer biological half-life
23 (Schultz et al., 2002). This self-inhibition is hypothesized to occur through inactivation of
24 GST-zeta (Schultz et al., 2002). In addition, TCE has been shown to cause the same
25 prolongation of DCA half-life in rodents, suggesting that TCE inhibits GST-zeta, probably
26 through the formation of DCA (Schultz et al., 2002). DCA-induced inhibition of GST-zeta has
27 also been reported in humans, with GST-zeta polymorphisms reported to influence the degree of
28 inactivation (Blackburn et al., 2000; Blackburn et al., 2001; Tzeng et al., 2000). Board et al.
29 (2001) report one variant to have significantly higher activity with DCA as a substrate than other
30 GST-zeta isoforms, which could affect DCA susceptibility.
31 GST-zeta, which is identical to maleylacetoacetate isomerase, is part of the tyrosine
32 catabolism pathway which is disrupted in Type 1 hereditary tyrosinemia, a disease associated
33 with the development of hepatocellular carcinoma at a young age (Tanguay et al., 1996). In
34 particular, GST-zeta metabolizes maleylacetoacetate (MAA) to fumarylacetoacetate (FAA) and
35 maleylacetone (MA) to fumarylacetone(Cornett et al., 1999; Tanguay et al., 1996). It has been
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-319 DRAFT—DO NOT CITE OR QUOTE
-------
1 suggested that the increased cancer risk with this disease, as well as through DC A exposure,
2 results from accumulation of MAA and MA, both alkylating agents, or FAA, which displays
3 apoptogenic, mutagenic, aneugenic, and mitogenic activities (Bergeron et al., 2003; Cornett et
4 al., 1999; Jorquera and Tanguay, 2001; Kim et al., 2000; Tanguay et al., 1996). However, the
5 possible effects of DC A through this pathway will depend on whether MAA, MA, or FAA is the
6 greater risk factor, since inhibition of GST-zeta will lead to greater concentrations of MAA and
7 MA and lower concentrations of FAA. Therefore, if MAA is the more active agent, DCA may
8 increase carcinogenic risk, while if FAA is the more active, DCA may decrease carcinogenic
9 risk. Tzeng et al. (2000) propose the later based on the greater genotoxicity of FAA, and in fact
10 suggest that DCA may "merit consideration for trial in the clinical management of hereditary
11 tyrosinemia type 1."
12 Therefore, TCE-induced inactivation GST-zeta, probably through formation of DCA,
13 may play a role in TCE-induced hepatocarcinogenesis. However, this mode of action is not
14 sufficiently delineated at this point for further evaluation, as even the question of whether its
15 actions through this pathway may increase or decrease cancer risk has yet to be experimentally
16 tested.
17
18 4.5.7.3.6. Oxidative stress. Several studies have attempted to study the possible effects of
19 "oxidative stress" and DNA damage resulting from TCE exposures. The effects of induction of
20 metabolism by TCE, as well as through coexposure to ethanol, have been hypothesized to in
21 itself increase levels of "oxidative stress" as a common effect for both exposures (see
22 Section E.4.2.4). In terms of contributing to a carcinogenic MO A, the term "oxidative stress" is
23 a somewhat nonspecific term, as it is implicated as part of the pathophysiologic events in a
24 multitude of disease processes and is part of the normal physiologic function of the cell and cell
25 signaling. Commonly, it appears to refer to the formation of reactive oxygen species leading to
26 cellular or DNA damage. As discussed above, however, measures of oxidative stress induced by
27 TCE, TCA, and DCA appear to be either not apparent, or at the very most transient and
28 nonpersistent with continued treatment (Larson and Bull, 1992; Channel et al., 1998; Toraason et
29 al., 1999; Parrish et al., 1996). Therefore, while the available data are limited, there is
30 insufficient evidence to support a role for such effects in TCE-induced liver carcinogenesis.
31 Oxidative stress has been hypothesized to be part of the MOA for peroxisome
32 proliferators, but has been found to neither be correlated with cell proliferation nor carcinogenic
33 potency of peroxisome proliferators (see Section E.3.4.1.1). For instance, Parrish et al. (1996)
34 reported that increases in PCO activity noted for DCA and TCA were not associated with
35 SOHdG levels (which were unchanged) and also not with changes laurate hydrolase activity
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-320 DRAFT—DO NOT CITE OR QUOTE
-------
1 observed after either DCA or TCA exposure. The authors concluded that their data do not
2 support an increase in steady state oxidative damage to be associated with TCA initiation of
3 cancer and that extension of treatment to time periods sufficient to insure peroxisome
4 proliferation failed to elevate SOHdG in hepatic DNA. The authors thus, suggested that
5 peroxisome proliferative properties of TCA were not linked to oxidative stress or carcinogenic
6 response.
7
8 4.5.7.3.7. Changes in gene expression (e.g., hypomethylation). Studies of gene expression as
9 well as considerations for interpretation of studies of using the emerging technologies of DNA,
10 siRNA, and miRNA microarrays for MOA analyses are included in Sections E.3.1.2 and
11 E.3.4.2.2. Caldwell and Keshava (2006) and Keshava and Caldwell (2006) report on both
12 genetic expression studies and studies of changes in methylation status induced by TCE and its
13 metabolites as well as differences and difficulties in the patterns of gene expression between
14 differing PPARa agonists. In particular are concerns for the interpretation of studies which
15 employ pooling of data as well as interpretation of "snapshots in time of multiple gene changes."
16 For instance, in the Laughter et al. (2004) study, it is not clear whether transcription arrays were
17 performed on pooled data as well as the issue of phenotypic anchoring as data on percent
18 liver/body weight indicates significant variability within TCE treatment groups, especially in
19 PPARa-null mice. For studies of gene expression using microarrays Bartosiewicz et al. (2001)
20 used a screening analysis of 148 genes for xenobiotic-metabolizing enzymes, DNA repair
21 enzymes, heat shock proteins, cytokines, and housekeeping gene expression patterns in the liver
22 in response TCE. The TCE-induced gene induction was reported to be highly selective; only
23 Hsp 25 and 86 and Cyp2a were up-regulated at the highest dose tested. Collier et al. (2003)
24 reported differentially expressed mRNA transcripts in embryonic hearts from Sprague-Dawley
25 rats exposed to TCE with sequences down-regulated with TCE exposure appearing to be those
26 associated with cellular housekeeping, cell adhesion, and developmental processes. TCE was
27 reported to induce up-regulated expression of numerous stress-response and homeostatic genes.
28 For the Laughter et al. (2004) study, transcription profiles using macroarrays containing
29 approximately 1,200 genes were reported in response to TCE exposure with 43 genes reported to
30 be significantly altered in the TCE-treated wild-type mice and 67 genes significantly altered in
31 the TCE-treated PPARa knockout mice. However, the interpretation of this information is
32 difficult because in general, PPARa knockout mice have been reported to be more sensitive to a
33 number of hepatotoxins partly because of defects in the ability to effectively repair tissue damage
34 in the liver (Shankar et al., 2003; Mehendale, 2000) and because a comparison of gene
35 expression profiles between controls (wild-type and PPARa knockout) were not reported. As
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-321 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported by Voss et al. (2006), dose-, time course-, species-, and strain-related differences should
2 be considered in interpreting gene array data. The comparison of differing PPARa agonists
3 presented in Keshava and Caldwell (2006) illustrate the pleiotropic and varying liver responses
4 of the PPARa receptor to various agonists, but did not imply that these responses were
5 responsible for carcinogenesis.
6 As discussed above in Section E.3.3.5, Aberrant DNA methylation is a common hallmark
7 of all types of cancers, with hypermethylation of the promoter region of specific tumor
8 suppressor genes and DNA repair genes leading to their silencing (an effect similar to their
9 mutation) and genome-wide hypomethylation (Ballestar and Esteller, 2002; Berger and
10 Daxenbichler, 2002; Herman et al., 1998; Pereira et al., 2004; Rhee et al., 2002). Whether DNA
11 methylation is a consequence or cause of cancer is a long-standing issue (Ballestar and Esteller,
12 2002). Fraga et al. (2004, 2005) reported global loss of monoacetylation and trimethylation of
13 histone H4 as a common hallmark of human tumor cells; they suggested, however, that
14 genomewide loss of 5-methylcytosine (associated with the acquisition of a transformed
15 phenotype) exists not as a static predefined value throughout the process of carcinogenesis but
16 rather as a dynamic parameter (i.e., decreases are seen early and become more marked in later
17 stages).
18 DNA methylation is a naturally occurring epigenetic mechanism for modulating gene
19 expression, and disruption of this mechanism is known to be relevant to human carcinogenesis.
20 As reviewed by Calvisi et al. (2007),
21 [a]berrant DNA methylation occurs commonly in human cancers in the forms of
22 genome-wide hypomethylation and regional hypermethylation. Global DNA
23 hypomethylation (also known as demethylation) is associated with activation of
24 protooncogenes, such as c-Jun, c-Myc, and c-HA-Ras, and generation of genomic
25 instability. Hypermethylation on CpG islands located in the promoter regions of
26 tumor suppressor genes results in transcriptional silencing and genomic
27 instability.
28
29 While clearly associated with cancer, it has not been conclusively established whether these
30 epigenetic changes play a causative role or are merely a consequence of transformation
31 (Tryndyak et al., 2006). However, as Calvisi et al. (2007) note, "Current evidence suggests that
32 hypomethylation might promote malignant transformation via multiple mechanisms, including
33 chromosome instability, activation of protooncogenes, reactivation of transposable elements, and
34 loss of imprinting."
35 Although little is known about how it occurs, a hypothesis has also been proposed that
36 that the toxicity of TCE and its metabolites may arise from its effects on DNA methylation
37 status. In regard to methylation studies, many are coexposure studies as they have been
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-322 DRAFT—DO NOT CITE OR QUOTE
-------
1 conducted in initiated animals with some studies being very limited in their reporting and
2 conduct. Cal dwell and Keshava (2006) review the body of work regarding TCE, DC A, and
3 TCA. Methionine status has been noted to affect the emergence of liver tumors (Counts et al.,
4 1996). Tao et al. (2000) and Pereira et al. (2004) have studied the effects of excess methionine
5 in the diet to see if it has the opposite effects as a deficiency (i.e., and reduction in a carcinogenic
6 response rather than enhancement). However, Tao et al. (2000) report that the administration of
7 excess methionine in the diet is not without effect and can result in percent liver/body weight
8 ratios. Pereira et al. (2004) report that methionine treatment alone at the 8 g/kg level was
9 reported to increase liver weight, decrease lauryl-CoA activity and to increase DNA methylation.
10 Pereira et al. (2004) reported that very high level of methionine supplementation to an
11 AIN-760A diet, affected the number of foci and adenomas after 44 weeks of coexposure to
12 3.2 g/L DCA. However, while the highest concentration of methionine (8.0 g/kg) was reported
13 to decrease both the number of DCA-induce foci and adenomas, the lower level of methionine
14 coexposure (4.0 g/kg) increased the incidence of foci. Coexposure of methionine (4.0 or
15 8.0 g/kg) with 3.2 g/L DCA was reported to decrease by -25% DCA-induced glycogen
16 accumulation, increase mortality, but not to have much of an effect on peroxisome enzyme
17 activity (which was not elevated by more than 33% over control for DCA exposure alone). The
18 authors suggested that their data indicate that methioninine treatment slowed the progression of
19 foci to tumors. Given that increasing hypomethylation is associated with tumor progression,
20 decreased hypomethylation from large doses of methionine are consistent with a slowing of
21 progression. Whether, these results would be similar for lower concentrations of DCA and lower
22 concentrations of methionine that were administered to mice for longer durations of exposure,
23 cannot be ascertained from these data. It is possible that in a longer-term study, the number of
24 tumors would be similar. Finally, a decrease in tumor progression by methionine
25 supplementation is not shown to be a specific event for the MOA for DCA-induced liver
26 carcinogenicity.
27 Tao et al. (2000) reported that 7 days of gavage dosing of TCE (1,000 mg/kg in corn oil),
28 TCA (500 mg/kg, neutralized aqueous solution), and DCA (500 mg/kg, neutralized aqueous
29 solution) in 8-week old female B6C3F1 mice resulted in not only increased liver weight but also
30 increased hypomethylation of the promoter regions of c-jun and c-myc genes in whole liver
31 DNA. However, data were shown for 1-2 mice per treatment. Treatment with methionine was
32 reported to abrogate this response only at a 300 mg/kg i.p dose with 0-100 mg/kg doses of
33 methionine having no effect. Ge et al. (2001) reported DCA- and TCA-induced DNA
34 hypomethylation and cell proliferation in the liver of female mice at 500 mg/kg and decreased
35 methylation of the c-myc promoter region in liver, kidney and urinary bladder. However,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-323 DRAFT—DO NOT CITE OR QUOTE
-------
1 increased cell proliferation preceded hypomethylation. Ge et al. (2002) also reported
2 hypomethylation of the c-myc gene in the liver after exposure to the peroxisome proliferators
3 2,4-dichlorophenoxyacetic acid (1,680 ppm), DBF (20,000 ppm), gemfibrozil (8,000 ppm), and
4 Wy-14,643 (50-500 ppm, with no effect at 5 or 10 ppm) after 6 days in the diet. Caldwell and
5 Keshava (2006) concluded that hypomethylation did not appear to be a chemical-specific effect
6 at these concentrations. As noted Section E.3.3.5, chemical exposure to a number of differing
7 carcinogens have been reported to lead to progressive loss of DNA methylation..
8 After initiation by TV-methyl-7V-nitrosourea (25 mg/kg) and exposure to 20 mmol/L DCA
9 or TCA (46 weeks), Tao et al. (2004) report similar hypomethylation of total mouse liver DNA
10 by DCA and TCA with tumor DNA showing greater hypomethylation. A similar effect was
11 noted for the differentially methylated region-2 of the insulin-like growth factor-II (IGF-II) gene.
12 The authors suggest that hypomethylation of total liver DNA and the IGF-II gene found in
13 nontumorous liver tissue would appear to be the result of a more prolonged activity and not cell
14 proliferation, while hypomethylation of tumors could be an intrinsic property of the tumors. As
15 pointed out by Caldwell and Keshava (2006) over expression of IGF-II gene in liver tumors and
16 preneoplastic foci has been shown in both animal models of hepatocarcinogenesis and humans,
17 and may enhance tumor growth, acting via the over-expressed IGF-I receptor (Scharf et al.,
18 2001; Werner and Le Roith, 2000).
19 Diminished hypomethylation was observed in Wy-14643-treated PPARa-null mice as
20 compared to wild-type mice, suggestive of involvement of PPARa in mediating hypomethylation
21 (Pogribny et al., 2007), but it is unclear how relevant these results are to TCE and its metabolites.
22 First, the doses of Wy-14643 administered are associated with substantial liver necrosis and
23 mortality with long-term treatment (Woods et al., 2007), adding confounding factors the
24 interpretation of their results. Hypomethylation by Wy-14643 progressively increased with time
25 up to 5 months (Pogribny et al., 2007), consistent with the sustained DNA synthesis caused by
26 Wy-14643 and a role for proliferation in causing hypomethylation. Regardless, as discussed
27 above, it is unlikely that PPARa is the mediator of the observed transient increase in DNA
28 synthesis by DCA, so even if it is important for hypomethylation by TCA, there may be more
29 than one pathway for this effect.
30 To summarize, aberrant DNA methylation status, including hypomethylation, is clearly
31 associated with both human and rodent carcinogenesis. Hypomethylation itself appears to be
32 sufficient for carcinogenesis, as diets deficient in choline and methionine that induce
33 hypomethylation have been shown to cause liver tumors in both rats and mice (Ghoshal and
34 Farber, 1984; Mikol et al., 1983; Henning and Swendseid, 1996; Wainfan and Poirier, 1992).
35 However, it is not known to what extent hypomethylation is necessary for TCE-induced
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-324 DRAFT—DO NOT CITE OR QUOTE
-------
1 carcinogenesis. However, as noted by Bull (2004) and Bull et al. (2004), the doses of TCA and
2 DCA that have been tested for induction of hypomethylation are quite high compared to doses at
3 which tumor induction occurs—at least 500 mg/kg/d. Whether these effects are still manifest at
4 lower doses relevant to TCE carcinogenicity, particularly with respect to DCA, has not been
5 investigated. Finally, the role of PPARa in modulating hypomethylation, possibly through
6 increased DNA synthesis as suggested by experiments with Wy-14643, are unknown for TCE
7 and its metabolites.
8
9 4.5.7.3.8. Cytotoxicity. Cytotoxicity and subsequent induction of reparative hyperplasia have
10 been proposed as key events for a number of chlorinated solvents, such as chloroform and carbon
11 tetrachloride.. However, as discussed above and discussed by Bull (2004) and Bull et al. (2004),
12 TCE treatment at doses relevant to liver carcinogenicity results in relatively low cytotoxicity.
13 While a number of histological changes with TCE exposure are observed, in most cases necrosis
14 is minimal or mild, associated with vehicle effects, and with relatively low prevalence. This is
15 consistent with the low prevalence of necrosis observed with TCA and DCA treatment at doses
16 relevant to TCE exposure. Therefore, it is unlikely that cytotoxicity and reparative hyperplasia
17 play a significant role in TCE carcinogenicity
18
19 4.5.7.4. Mode of Action (MOA) Conclusions
20 Overall, although a role for many of the proposed key events discussed above cannot be
21 ruled out, there are inadequate data to support the conclusion that any of the particular MOA
22 hypotheses reviewed above are operant. Thus, the MOA of liver tumors induced by TCE is
23 considered unknown at this time, and the answer to the first key question "1. Is the hypothesized
24 mode of action sufficiently supported in the test animals?" is "no" at this time. Consequently,
25 the other key questions of "2. Is the hypothesized mode of action relevant to humans?" and
26 "J. Which populations or lifestages can be particularly susceptible to the hypothesized mode
27 of action?" will not be discussed in a MOA-specific manner. Rather, they are discussed below
28 in more general terms, first qualitatively and then quantitatively, using available relevant data.
29
30 4.5.7.4.1. Qualitative human relevance and susceptibility. No data exist that suggests that
31 TCE-induced liver tumorigenesis is caused by processes that irrelevant in humans. In addition,
32 as discussed above, several of the other effects such as polyploidization, changes in glycogen
33 storage, and inhibition of GST-zeta—are either clearly related to human carcinogenesis or areas
34 of active research as to their potential roles. For example, the effects of DCA on glycogen
35 storage parallel the observation that individuals with conditions that lead to glycogenesis appear
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-325 DRAFT—DO NOT CITE OR QUOTE
-------
1 to be at an increased risk of liver cancer (LaVecchia et al., 1994; Adami et al., 1996;
2 Wideroff et al., 1997; Rake et al., 2002). In addition, there may be some relationship between
3 the effects of DCA and the mechanism of increased liver tumor risk in childhood in those with
4 Type 1 hereditary tyrosinemia, though the hypotheses needs to be tested experimentally.
5 Similarly, with respect to PPARa activation and downstream events hypothesized to be causally
6 related to liver carcinogenesis, it is generally acknowledged that "a point in the rat/mouse key
7 events cascade where the pathway is biologically precluded in humans cannot be identified, in
8 principle" (Klaunig et al, 2003; NRC, 2006).
9 In terms of human relevance and susceptibility, it is also useful to briefly review what is
10 known about human HCC. A number of risk factors have been identified for human
11 hepatocellular carcinoma, including ethanol consumption, hepatitis B and C virus infection,
12 aflatoxin Bl exposure, and, more recently, diabetes and perhaps obesity (El-Serag and Rudolph,
13 2007). However, it is also estimated that a substantial minority of HCC patients, perhaps 15 to
14 50%, have no established risk factors (El-Serag and Rudolph, 2007). In addition, cirrhosis is
15 present in a large proportion of HCC patients, but the prevalence of HCC without underlying
16 cirrhosis, while not precisely known, is still significant, with estimates based on relatively small
17 samples ranging from 7 to 54% (Fattovisch, 2004).
18 However, despite the identification of numerous factors that appear to play a role in the
19 human risk of HCC, the mechanisms are still largely unclear (Yeh et al., 2007). Interestingly,
20 the observation by Leakey et al. (2003a, b) that body weight significantly and strongly impacts
21 background liver tumor rates in B6C3F1 mice parallels the observed epidemiologic associations
22 between liver cancer and obesity (review in El-Serag and Rudolph [2007]). This concordance
23 suggests that similar pathways may be involved in spontaneous liver tumor induction between
24 mice and humans. The extent to which TCE exposure may interact with known risk factors for
25 HCC cannot be determined at this point, but several hypotheses can be posed based on existing
26 data. If TCE affects some of the same pathways involved in human HCC, as suggested in the
27 discussion of several TCE-induced effects above, then TCE exposure may lead a risk that is
28 additive to background.
29 As discussed above, there are several parallels between the possible key events in TCE-
30 induced liver tumors in mice and what is known about mechanisms of human HCC, though none
31 have been experimentally tested. Altered ploidy distribution and DNA hypomethylation are
32 commonly observed in human HCC (Zeppa et al., 1998; Lin et al., 2003; Calvisi et al., 2007).
33 Interestingly, El-Serag and Rudolph (2007) have been suggested that the risk of HCC increases
34 with cirrhosis in part because the liver parenchymal cells have decreased proliferative capacity,
35 resulting in an altered milieu that promotes tumor cell proliferation. This description suggests a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-326 DRAFT—DO NOT CITE OR QUOTE
-------
1 similarity in mode of action, though via different mechanisms, with the "negative selection"
2 hypothesis proposed by Bull (2000) for TCE and its metabolites although for TCE changes in
3 apoptosis and cell proliferation have not been noted or examined to such an extent to provide
4 evidence of a similar environment. Increased ploidy decreases proliferative capacity, so that
5 may be another mechanism through which the effects of TCE mimic the conditions thought to
6 facilitate the induction of human HCC.
7 In sum, from the perspective of hazard characterization, the available data support the
8 conclusion that the mode of action for TCE-induced mouse liver tumors is relevant to humans.
9 No data suggest that any of the key events are biologically precluded in humans, and a number of
10 qualitative parallels exist between hypotheses for the mode of action in mice and what is known
11 about the etiology and induction of human HCC. A number of risk factors have been identified
12 that appear to modulate the risk of human HCC, and these may also modulate the susceptibility
13 to the effects from TCE exposure. As noted in Section E.4, TCE exposure in the human
14 population is accompanied not only by external exposures to its metabolites, but brominated
15 analogues of those metabolites that are also rodent carcinogens, a number of chlorinate solvents
16 that are hepatocarcinogenic and alcohol consumption. The types of tumors and the heterogeneity
17 of tumors induced by TCE in rodents parallel those observed in humans (see Section E.3.1.8).
18 The pathways identified for induction of cancer in humans for cancer are similar to those for the
19 induction of liver cancer (see Section E.3.2.1). However, while risk factors have been identified
20 for human liver cancer that have similarities to TCE-induced effects and those of its metabolites,
21 both the mechanism for human liver cancer induction and that for TCE-induced liver
22 carcinogenesis in rodents are not known.
23
24 4.5.7.4.2. Quantitative species differences. As a precursor to the discussion of quantitative
25 differences between humans and rodents and among humans, it should be noted that an adequate
26 explanation for the difference in response for TCE-liver cancer induction between rats and mice
27 has yet to be established or for that difference to be adequately described given the limitations in
28 the rat database. For TCA, there is only one available long-term study in rats that, while
29 suggestive that TCA is less potent in rats than mice, is insufficient to determine if there was a
30 TCA-induced effect or what its magnitude may be. While some have proposed that the lower
31 rate of TCA formation in rats relative to mice would explain the species difference, PBPK
32 modeling suggests that the differences (3-5-fold) may be inadequate to fully explain the
33 differences in carcinogenic potency. Moreover, inferences from comparing the effects of TCE
34 and TCA on liver weight, using PBPK model-based estimates of TCA internal dose metrics as a
35 result of TCE or TCA administration, indicate that TCA is not likely to play a predominant role
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-327 DRAFT—DO NOT CITE OR QUOTE
-------
1 in hepatomegaly. Combined with the qualitative correlation between rodent hepatomegaly and
2 hepatocarcinogenesis observed across many chemicals, this suggests that TCA similarly is not a
3 predominant factor in TCE-induced hepatocarcinogenesis. Indeed, there are multiple lines of
4 evidence that TCA is insufficient to account for TCE-induced tumors, including data on tumor
5 phenotype (e.g., c-Jun immunostaining) and genotype (e.g., H-ras mutation frequency and
6 spectrum). For DCA, only a single experiment in rats is available (reported in two publications),
7 and although it suggests lower hepatocarcinogenic potency in rats relative to mice, its relatively
8 low power limits the inferences that can be made as to species differences.
9 As TCA induces peroxisome proliferation in the mouse and the rat, some have suggested
10 that difference in peroxisomal enzyme induction is responsible for the difference in susceptibility
11 to TCA liver carcinogenesis. The study of DeAngelo et al. (1989) has been cited in the literature
12 as providing evidence of differences between rats and mice for peroxisomal response to TCA.
13 However, data from the most resistant strain of rat (Sprague-Dawley) have been cited in
14 comparisons of peroxisomal enzyme effects but the Osborne-Mendel and F344 rat were not
15 refractory and showed increased PCO activity so it is not correct to state that the rat is refractory
16 to TCA-induction of peroxisome activity (see Section E.2.3.1.5). In addition, as discussed
17 above, inferences based on PCO activity are limited by its high variability, even in control
18 animals, as well as its not necessarily being predictive of the peroxisome number or cytoplasmic
19 volume.
20 The same assumption of lower species sensitivity by measuring peroxisome proliferation
21 has been applied to humans, as peroxisome proliferation caused by therapeutic PPARa agonists
22 such as fibrates in humans is generally lower (<2-fold induction) than that observed in rodents
23 (20- to 50-fold induction). However, as mentioned above, it is known that peroxisome
24 proliferation is not a good predictor of potency (Marsman et al., 1988).
25 Limited data exist on the relative sensitivity of the occurrence of key events for liver
26 tumor induction between mice and humans and among humans. Pharmacokinetic differences are
27 addressed with PBPK modeling to the extent that data allow, so the discussion here will
28 concentrate on pharmacodynamic differences. Most striking is the difference in "background"
29 rates of liver tumors. Data from NTP indicates that control B6C3F1 mice in 2-year bioassays
30 have a background incidence of hepatocellular carcinomas of 26% in males and 10% in females,
31 with higher incidences for combined hepatocellular adenomas and carcinomas (Maronpot, 2007).
32 However, as discussed above, Leakey et al. (2003a, b) report that the background incidence rates
33 are very dependent on the weight of the mice. By contrast, the estimated lifetime risk of liver
34 and biliary tract cancer in the United States (about 75% of which are hepatocellular carcinomas)
35 is 0.97% for men and 0.43% for women (Ries et al., 2008). However, regions of the world
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-328 DRAFT—DO NOT CITE OR QUOTE
-------
1 where additional risk factors (hepatitis infection, alflatoxin exposure) have high prevalence have
2 liver cancer incidences up to more than 6-fold greater than the United States (Ferlay et al., 2004).
3 Therefore, one possible quantitative difference that can be flagged for use in dose-response
4 assessment is the background rate of liver tumors between species. Biologically-based dose-
5 response modeling by Chen (2000) suggested that the data were consistent with a purely
6 promotional model in which potency would be proportional to background tumor incidence.
7 However, it is notable that male Swiss mice, which have lower background liver tumor rates than
8 the B6C3F1 strain, were also positive in one long-term bioassay (Maltoni et al., 1986).
9 Similarly, in terms of intraspecies susceptibility, to the extent that TCE may
10 independently promote pre-existing initiated cells, it can be hypothesized that those with greater
11 risk for developing HCC due to one more of the known risk factors would have a proportional
12 increase in the any contributions from TCE exposure. In addition, in both humans and mice,
13 males appear to be at increased risk of liver cancer, possibly due to sexually dimorphism in
14 inflammatory responses (Lawrence et al., 2007; Naugler et al., 2007; Rakoff-Nahoun and
15 Medzhitov, 2007), suggesting that men may also be more susceptible to TCE-induced liver
16 tumorigenesis than women. It has been observed that human HCC is highly heterogeneous
17 histologically, but within patients and between patients, studies are only beginning to distinguish
18 the different pathways that may be responsible for this heterogeneity (Feitelson et al., 2002;
19 Chen et al., 2002; Yeh et al., 2007).
20 Appropriate quantitative data are generally lacking on interspecies differences in the
21 occurrence of most other proposed key events, although many have argued that there are
22 significant quantitative differences between rodents and humans related to PPARa activation
23 (Klaunig et al., 2003; NRC, 2006). For instance, it has been suggested that lower levels of
24 PPARa receptor in human hepatocytes relative to rodent hepatocytes contributes to lower human
25 sensitivity (Tugwood et al., 1996; Palmer et al., 1998; Klaunig et al., 2003). However, out of a
26 small sample of human livers (n = 6) show similar protein levels to mice (Walgren et al., 2000a).
27 Another proposed species difference has been ligand affinity, but while transactivation assays
28 showed greater affinity of Wy-14643 and perfluorooctanoic acid for rodent relative to human
29 PPARa, they showed TCA and DCA had a similar affinities between species (Maloney and
30 Waxman, 1999). Furthermore, it is not clear that receptor-ligand kinetics (capacity and affinity)
31 are rate-limiting for eliciting hepatocarcinogenic effects, as it is known that maximal receptor
32 occupation is not necessary for a maximal receptor mediated response (Stephenson, 1956, see
33 also review by Danhof et al., 2007).
34 There is also limited in vivo and in vitro data suggesting that increases in cell
35 proliferation mediated by PPARa agonists are diminished in humans and other primates relative
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-329 DRAFT—DO NOT CITE OR QUOTE
-------
1 to rodents (Klaunig et al., 2003; NRC, 2006; Hoivik et al., 2004). However, Walgren et al.
2 (2000b) reported that TCA and DCA were not mitogenic in either human or rodent hepatocytes
3 in vitro. Furthermore, TCE, TCA, and DCA all induce only transient increases in cell
4 proliferation, so the relevance to TCE of interspecies differences from PPARa agonists that to
5 produce sustained proliferation, such as Wy-14643, is not clear. In addition, comparisons
6 between primate and rodent models should take into account the differences in the ability to
7 respond to any mitogenic stimulation (see Section E.3.2). Primate and human liver respond
8 differently (and much more slowly) to a stimulus such as partial hepatectomy.
9 Recent studies in "humanized" mice (PPARa-null mice in which a human PPARa gene
10 was subsequently inserted and expressed in the liver) reported that treatment with a PPARa
11 agonist lead to greatly lower incidence of liver tumors as compared to wild-type mice
12 (Morimura et al., 2006). However, these experiments were performed with WY-14643 at a dose
13 causing systemic toxicity (reduced growth and survival), had a duration of less than 1 year, and
14 involved a limited number of animals. In addition, because liver tumors in mice at less than
15 1 year are extremely rare, the finding a one adenoma in WY-14643-treated humanized mice
16 suggests carcinogenic potential that could be further realized with continued treatment
17 (Keshava and Caldwell, 2006). In addition, Yang et al. (2007) recently noted that let-7C, a
18 microRNA involved in cell growth and thought to be a regulatory target of PPARa (Shah, 2008),
19 was inhibited by Wy-14643 in wild-type mice, but not in "humanized mice" in which had human
20 PPARa was expressed throughout the body on a PPARa-null background. However, these
21 humanized mice had about a 20-fold higher baseline expression of let-7C, as reported in control
22 mice, potentially masking any treatment effects. More generally, it is not known to what extent
23 PPARa-related events are rate-limiting in TCE-induced liver tumorigenesis, for which multiple
24 pathways appear to be operative. So even if quantitative differences mediated by PPARa were
25 well estimated, they would not be directly usable for dose-response assessment in the absence of
26 way to integrate the contributions from the different pathways.
27 In sum, the only quantitative data and inter- and intraspecies susceptibility suitable for
28 consideration in dose-response assessment are differences background liver tumor risk. These
29 may modulate the effects of TCE if relative risk, rather than additional risk, is the appropriate
30 common inter- and intraspecies metric. However, the extent to which relative risk would provide
31 a more accurate estimate of human risk is unknown.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-330 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.6. IMMUNOTOXICITY AND CANCERS OF THE IMMUNE SYSTEM
2 Chemical exposures may result in a variety of adverse immune-related effects, including
3 immunosuppression (decreased host resistance), autoimmunity, and allergy-hypersensitivity, and
4 may result in specific diseases such as infections, systemic or organ-specific autoimmune
5 diseases, or asthma. Measures of immune function (e.g., T-cell counts, immunoglobulin (Ig) E
6 levels, specific autoantibodies, cytokine levels) may provide evidence of an altered immune
7 response that precedes the development of clinically expressed diseases. The first section of this
8 chapter discusses effects relating to immunotoxicity, including risk of autoimmune diseases,
9 allergy and hypersensitivity, measures of altered immune response, and lymphoid cancers.
10 Studies pertaining to effects in humans are presented first, followed by a section discussing
11 relevant studies in animals. The second section of this chapter discusses evidence pertaining to
12 trichloroethylene in relation to lymphoid tissue cancers, including childhood leukemia.
13
14 4.6.1. Human Studies
15 4.6.1.1. Noncancer Immune-Related Effects
16 4.6.1.1.1. Immunosuppression, asthma, and allergies. In 1982, Lagakos et al. conducted a
17 telephone survey of residents of Woburn, Massachusetts, collecting information on residential
18 history and history of 14 types of medically diagnosed conditions (Lagakos, 1986). The survey
19 included 4,978 children born since 1960 who lived in Woburn before age 19. Completed
20 surveys were obtained from approximately 57% of the town residences with listed phone
21 numbers. Two of the wells providing the town's water supply from 1964 to 1979 had been
22 found to be contaminated with a number of solvents, including tetrachloroethylene (21 ppb) and
23 trichloroethylene (267 ppb) (as cited in [Lagakos, 1986]). Lagakos et al. used information from
24 a study by the Massachusetts Department of Environmental Quality and Engineering to estimate
25 the contribution of water from the two contaminated wells to the residence of each participant,
26 based on zones within the town receiving different mixtures of water from various wells, for the
27 period in which the contaminated wells were operating. This exposure information was used to
28 estimate a cumulative exposure based on each child's length of residence in Woburn. A higher
29 cumulative exposure measure was associated with conditions indicative of immunosuppression
30 (e.g., bacterial or viral infections) or hypersensitivity (e.g., asthma). In contrast, a recent study
31 using the National Health and Nutrition Examination Survey data collected from 1999-2000 in a
32 representative sample of the United States population (n = 550) did not find an association
33 between TCE exposure and self-report of a history of physician-diagnosed asthma (OR: 0.94,
34 95% CI: 0.77, 1.14) (Arif and Shah, 2007). TCE exposure, as well as exposure to 9 other
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-331 DRAFT—DO NOT CITE OR QUOTE
-------
1 volatile organic compounds, was determined through a passive monitor covering a period of
2 48-72 hours. No clear trend was seen with self-reported wheeze episodes (OR: 1.29, 95% CI:
3 [0.98, 1.68] for one to two episodes; OR: 0.21, 95% CI: [0.04, 10.05] for three or more episodes
4 in the past 12 months).
5 Allergy and hypersensitivity, as assessed with measures of immune system parameters or
6 immune function tests (e.g., atopy) in humans, have not been extensively studied with respect to
7 the effects of trichloroethylene (see Table 4-58). Lehmann et al. reported data pertaining to IgE
8 levels and response to specific antigens in relation to indoor levels of volatile organic compounds
9 among children (age 36 months) selected from a birth cohort study in Leipzig, Germany
10 (Lehmann et al., 2001). Enrollment into the birth cohort occurred between 1995 and 1996. The
11 children in this allergy study represent a higher-risk group for development of allergic disease,
12 with eligibility criteria that were based on low birth weight (between 1,500 and 2,500 g), or cord
13 blood IgE greater than 0.9 kU/L with double positive family history of atopy. These eligibility
14 criteria were met by 429 children; 200 of these children participated in the allergy study
15 described below, but complete data (IgE and volatile organic compound measurements) were
16 available for only 121 of the study participants. Lehmann et al. measured 26 volatile organic
17 compounds via passive indoor sampling in the child's bedroom for a period of 4 weeks around
18 the age of 36 months. The median exposure of trichloroethylene was 0.42 ug/m3 (0.17 ug/m3
19 and 0.87 ug/m3 for the 25th and 75th percentiles, respectively). Blood samples were taken at the
20 36-month-study examination and were used to measure the total IgE and specific IgE antibodies
21 directed to egg white, milk, indoor allergens (house dust mites, cat, molds), and outdoor
22 allergens (timothy-perennial grass, birch- tree). There was no association between
23 trichloroethylene exposure and any of the allergens tested in this study, although some of the
24 other volatile organic compounds (e.g., toluene, 4-ethyltoluene) were associated with elevated
25 total IgE levels and with sensitization to milk or eggs.
26
27 4.6.1.1.2. Generalized hypersensitivity skin diseases, with or without hepatitis. Occupational
28 exposure to trichloroethylene has been associated with a severe, generalized skin disorder that is
29 distinct from contact dermatitis in the clinical presentation of the skin disease (which often
30 involves mucosal lesions), and in the accompanying systemic effects that can include
31 lymphadenopathy, hepatitis, and other organ involvement. Kamijima et al. recently reviewed
32 case reports describing 260 patients with trichloroethylene-related generalized skin disorders
33 (Kamijima et al., 2007). Six of the patients were from the United States or Europe, with the
34 remainder occurring in China, Singapore, Philippines, and other Asian countries. One study in
35 Guangdong province, in southeastern China, included more than 100 of these cases in a single
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-332 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-58. Studies of immune parameters (IgE antibodies and cytokines) and trichloroethylene in humans
Parameter,
source of data
Results
Reference, location, diagnosis
period, sample size, age
IgE antibodies
blood sample, indoor air
sampling of 28 volatile organic
chemicals in child's bedroom
Trichloroethylene exposure not associated with sensitization to
indoor or outdoor allergens
Lehmann et al., 2001
Germany 1997-1999. n = 121
36-month old children
Cytokine secreting CD3+
T-cell populations
cord blood, indoor air sampling
of 28 volatile organic chemicals
in child's bedroom 4 wks after
birth
In CD3+ cord blood cells, some evidence of association between
increasing trichloroethylene levels and
decreased IL-4 >75th percentile OR: 0.6 (95% CI: 0.2, 2.1),
<25th percentile OR 4.4 (95% CI: 1.1, 17.8)
increased IFN-y >75th percentile OR: 3.6 (95% CI: 0.9, 14.9)
<25th percentile OR: 0.7 (95% CI: 0.2, 2.2)
Similar trends not seen with tumor necrosis factor-a or IL-2
Lehmann et al., 2002
Germany. 1995-1996. n = 85
newborns
Cytokine secreting CD3+ and
CD8+ T-cell populations
blood sample, indoor air
sampling of 28 volatile organic
chemicals in child's bedroom
Trichloroethylene exposure not associated with percentages of IL-4
CD3+ or IFN-y CD8+ T-cells
Lehmann et al., 2001
Germany. 1995-1999. n = 200
36-month old children.
Cytokine concentration—
serum
urine sample (trichloroacetic
acid concentration), blood
sample, questionnaire (smoking
history, age, residence),
workplace TCE measures
(personal samples, 4 exposed
and 4 nonexposed workers)
Nonexposed workers similar to office controls for all cytokine
measures. Compared to nonexposed workers, the trichloroethylene
exposed workers had
decreased IL-4 (mean 3.9 vs. 8.1 pg/mL)
increased IL-2 (mean 798 vs. 706 pg/mL)
increased IFN-y (mean 37.1 vs. 22.9 pg/mL)
lavicoli et al., 2005
Italy, n = 35 printers using TCE, 30
nonexposed workers (in same
factory, did not use or were not
near TCE), 40 office worker
controls. All men. Mean age
~33 yrs.
-------
1 year (Huang et al., 2002). Kamijima et al. categorized the case descriptions as indicative of
2 hypersensitivity syndrome (n = 124) or a variation of erythema multiforme, Stevens-Johnson
3 syndrome, and toxic epiderma necrolysis (n = 115), with 21 other cases unclassified in either
4 category. The fatality rate, approximately 10%, was similar in the two groups, but the
5 prevalence of fever and lymphadenopathy was higher in the hypersensitivity syndrome patients.
6 Hepatitis was seen in 92-94% of the multiforme, Stevens-Johnson syndrome, and toxic
7 epiderma necrolysis patients, but the estimates within the hypersensitivity syndrome group were
8 more variable (46-94%) (Kamijima et al., 2007).
9 Some of the case reports reviewed by Kamijima et al. provided information on the total
10 number of exposed workers, working conditions, and measures of exposure levels. From the
11 available data, generalized skin disease within a worksite occurred in 0.25 to 13% of workers in
12 the same location, doing the same type of work (Kamijima et al., 2007). The measured
13 concentration of trichloroethylene ranged from <50 mg/m3 to more than 4,000 mg/m3, and
14 exposure scenarios included inhalation only and inhalation with dermal exposures. Disease
15 manifestation generally occurred within 2-5 weeks of initial exposure, with some intervals up to
16 3 months. Most of the reports were published since 1995, and the geographical distribution of
17 cases reflects the newly industrializing areas within Asia.
18 Kamijima and colleagues recently conducted an analysis of urinary measures of
19 trichloroethylene metabolites (trichloroacetic acid and trichloroethanol) in 25 workers
20 hospitalized for hypersensitivity skin disease in 2002 (Kamijima et al., 2008). Samples taken
21 within 15 days of the last exposure to trichloroethylene exposure were available for 19 of the
22 25-patients, with a mean time of 8.4 days. Samples from the other patients were not used in the
23 analysis because the half life of urinary trichloroacetic acid is 50-100 hours. In addition,
24 3-6 healthy workers doing the same type of work in the factories of the affected worker, and
25 2 control workers in other factories not exposed to trichloroethylene were recruited in
26 2002-2003 for a study of breathing zone concentration of volatile organochlorines and urinary
27 measures of trichloroethylene metabolites. Worksite measures of trichloroethylene concentration
28 were also obtained. Adjusting for time between exposure and sample collection, mean urinary
29 concentration at the time of last exposure among the 19 patients was 206 mg/mL for
30 trichloroacetic acid. Estimates for trichloroethanol were not presented because of the shorter
31 half-life for this compound. Urinary trichloroacetic acid levels in the healthy exposed workers
32 varied among the 4 factories, with means (istandard deviation [SD]) of 41.6 (±18.0),
33 131 (±90.2), 180 (±92), and 395 (±684). The lower values were found in a factory in which the
34 degreasing machine had been partitioned from the workers after the illnesses had occurred.
35 Trichloroethylene concentrations (personal time-weighted averages) at the factories of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-334 DRAFT—DO NOT CITE OR QUOTE
-------
1 affected workers ranged from 164-2,330 mg/m3 (30-431 ppm). At the two factories with no
2 affected workers in the past 3 years, the mean personal time-weighted average trichloroethylene
3 concentrations were 44.9 mg/m3 (14 ppm) and 1,803 mg/m3 (334 ppm). There was no
4 commonality of additives or impurities detected among the affected factories that could explain
5 the occurrence of the hypersensitivity disorder.
6 To examine genetic influences on disease risk, Dai et al. conducted a case-control study
7 of 111 patients with trichloroethylene-related severe generalized dermatitis and
8 152 trichloroethylene-exposed workers who did not develop this disease (Dai et al., 2004).
9 Patients were recruited from May 1999 to November 2003 in Guangdong Province, and were
10 employed in approximately 80 electronic and metal-plating manufacturing plants. Initial
11 symptoms occurred within 3 months of exposure. The comparison group was drawn from the
12 same plants as the cases, and had worked for more than 3 months without development of skin or
13 other symptoms. Mean age in both groups was approximately 23 years. A blood sample was
14 obtained from study participants for genotyping of tumor necrosis factor (TNF)-a, TNF-P, and
15 interleukin (IL)-4 genotypes. The genes were selected based on the role of TNF and of
16 interleukin-4 in hypersensitivity and inflammatory responses. The specific analyses included
17 two polymorphisms in the promoter region of TNF-a (G —»• A substitution at position -308)
18 designated as TNF All, with wild-type designated TNFAI; and a G —> A substitution at position -
19 238), a polymorphism at the first intron on TNF-P, and a polymorphism in the promoter region
20 of IL-4 (C —> T substitution at -590). There was no difference in the frequency of the TNF-a"238,
21 TNF-P, or IL-4 polymorphisms between cases and controls, but the wild-type TNF-a"308
22 genotype was somewhat more common among cases (TNF A I/I genotype 94% in cases and 86%
23 in controls).
24 Kamijima et al. note the similarities, particular with respect to specific skin
25 manifestations, of the case presentations of trichloroethylene-related generalized skin diseases to
26 conditions that have been linked to specific medications (e.g., carbamezepine, allupurinol,
27 antibacterial sulfonamides), possibly in conjunction with reactivation of specific latent herpes
28 viruses (Kamijima et al., 2007). A previous review by these investigators discusses insights with
29 respect to drug metabolism that may be useful in developing hypotheses regarding susceptibility
30 to trichloroethylene-related generalized skin disorders (Nakajima et al., 2003). Based on
31 consideration of metabolic pathways and intermediaries, variability in CYP2E1,
32 UDP-glucoronyltransferase, glutathione-S transferase, and N-acetyl transferase (NAT) activities
33 could be hypothesized to affect the toxicity of trichloroethylene. NAT2 is most highly expressed
34 in liver, and the "slow" acetylation phenotype (which arises from a specific mutation) has been
35 associated with adverse effects of medications, including drug-induced lupus (Lemke and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-335 DRAFT—DO NOT CITE OR QUOTE
-------
1 McQueen, 1995) and hypersensitivity reactions (Spielberg, 1996). There are limited data
2 pertaining to genetic or other sources of variability in these enzymes on risk of trichloroethylene-
3 related generalized skin diseases, however. In a study in Guangdong province, CYP1A1,
4 GSTM1, GSTP1, GSTT1, and NAT2 genotypes in 43 cases of trichloroethylene-related
5 generalized skin disease were compared to 43 healthy trichloroethylene-exposed workers (Huang
6 et al., 2002). The authors reported that the NAT2 slow acetylation genotype was associated with
7 disease, but the data pertaining to this finding were not presented.
8
9 4.6.1.1.3. Cytokine profiles. Cytokines are produced by many of the immune regulatory cells
10 (e.g., macrophages, dendritic cells), and have many different effects on the immune system. The
11 T-helper Type 1 (Thl) cytokines, are characterized as "pro-inflammatory" cytokines, and include
12 TNF-a and interferon (IFN)-y. Although this is a necessary and important part of the innate
13 immune response to foreign antigens, an aberrant pro-inflammatory response may result in a
14 chronic inflammatory condition and contribute to development of scarring or fibrotic tissue, as
15 well as to autoimmune diseases. Th2 cytokines are important regulators of humoral (antibody -
16 related) immunity. IL-4 stimulates production of IgE and thus influences IgE-mediated effects
17 such as allergy, atopy, and asthma. Th2 cytokines can also act as "brakes" on the inflammatory
18 response, so the balance between different types of cytokine production is also important with
19 respect to risk of conditions resulting from chronic inflammation. Several studies have examined
20 cytokine profiles in relation to occupational or environmental TCE exposure (see Table 4-58).
21 The 2001 Lehmann et al. study of 36-month old children (described above) also included
22 a blood sample taken at the 3-year study visit, which was used to determine the percentages of
23 specific cytokine producing T-cells in relation to the indoor volatile organic compounds
24 exposures measured at birth. There was no association between trichloroethylene exposure and
25 either IL-4 CD3+ or IFN-y CD8+ T-cells (Lehmann et al., 2001).
26 Another study by Lehmann et al. examined the relationship between indoor exposures to
27 volatile organic compounds and T-cell subpopulations measured in cord blood of newborns
28 (Lehmann et al., 2002). The study authors randomly selected 85 newborns (43 boys and
29 42 girls) from a larger cohort study of 997 healthy, full-term babies, recruited between 1997 and
30 1999 in Germany. Exclusion criteria included a history in the mother of an autoimmune disease
31 or infectious disease during the pregnancy. Twenty-eight volatile organic compounds were
32 measured via passive indoor sampling in the child's bedroom for a period of 4 weeks after the
33 birth (a period which is likely to reflect the exposures during the prenatal period close to the time
34 of delivery). The levels were generally similar or slightly higher than the levels seen in the
35 previous study using samples from the bedrooms of the 36-month-old children. The highest
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-336 DRAFT—DO NOT CITE OR QUOTE
-------
1 levels of exposure were seen for limonene (median 24.3 ug/m3), a-pinene (median 19.3 ug/m3)
2 and toluene (median 18.3 ug/m3), and the median exposure of trichloroethylene was 0.6 ug/m3
3 (0.2 ug/m3 and 1.0 ug/m3 for the 25th and 75th percentiles, respectively). Flow cytometry was
4 used to measure the presence of CD3 T-cells obtained from the cord blood labeled with
5 antibodies against IFN-y, tumor necrosis factor-a, IL-2, and IL-4. There was some evidence of a
6 decreased level of IL-2 with higher trichloroethylene exposure in the univariate analysis, with
7 median percentage of IL-2 cells of 46.1 and 33.0% in the groups that were below the 75th
8 percentile and above the 75th percentile of trichloroethylene exposure, respectively. In analyses
9 adjusting for family history of atopy, gender and smoking history of the mother during
10 pregnancy, there was little evidence of an association with either IL-2 or IFN-y, but there was a
11 trend of increasing trichloroethylene levels associated with decreased IL-4 and increased IFN-y.
12 lavicoli et al. examined cytokine levels in 35 trichloroethylene-exposed workers (Group
13 A) from a printing area of a factory in Italy. Their work involved use of trichloroethylene in
14 degreasing (lavicoli et al., 2005). Two comparison groups were included. Group B consisted of
15 30 other factory workers who were not involved in degreasing activities and did not work near
16 this location, and Group C consisted of 40 office workers at the factory. All study participants
17 were male and had worked at their present position for at least 3 years, and all were considered
18 healthy. Personal breathing zone air samples from four workers in Group A and four workers in
19 Group B were obtained in three consecutive shifts (24 total samples) to determine air
20 concentration of trichloroethylene. A urine sample was obtained from each Group A and Group
21 B worker (end of shift at end of work week) for determination of trichloroacetic acid
22 concentrations (corrected for creatinine), and blood samples were collected for assessment of
23 IL-2, IL-4, and IFN-y concentrations in serum using enzyme-linked immunosorbent assays.
24 Among exposed workers, the mean trichloroethylene concentration was approximately 35 mg/m3
25 (30.75 ± SD 9.9, 37.75 ± 23.0, and 36.5 ± 8.2 mg/m3 in the morning, evening, and night shifts,
26 respectively). The urinary trichloroacetic acid concentrations were much higher in exposed
27 workers compared with nonexposed workers (mean ± SD, Group A 13.3 ± 5.9 mg/g creatinine;
28 Group B 0.02 ± 0.02 mg/g creatinine). There was no difference in cytokine levels between the
29 two control groups, but the exposed workers differed significantly (all ^-values <0.01 using
30 Dunnett's test for multiple comparisons) from each of the two comparison groups. The observed
31 differences were a decrease in IL-4 levels (mean 3.9, 8.1, and 8.1 pg/mL for groups A, B, and C,
32 respectively), and an increase in IL-2 levels (mean 798, 706, and 730 pg/mL for groups A, B,
33 and C, respectively) and in IFN-y levels (mean 37.1, 22.9, and 22.8 pg/mL for groups A, B, and
34 C, respectively).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-337 DRAFT—DO NOT CITE OR QUOTE
-------
1 The available data from these studies (Lehmann et al., 2001, 2002; lavicoli et al., 2005)
2 provide some evidence of an association between increased trichloroethylene exposure and
3 modulation of immune response involving an increase in pro-inflammatory cytokines (TL-2,
4 IFN-y) and a decrease in Th2 (allergy-related) cytokines (e.g., IL-4). These observations add
5 support to the influence of trichloroethylene in immune-related conditions affected by chronic
6 inflammation.
7
8 4.6.1.1.4. Autoimmune disease
9 4.6.1.1.4.1. Disease clusters and geographic-based studies. Reported clusters of diseases have
10 stimulated interest in environmental influences on systemic autoimmune diseases. These
11 descriptions include investigations into reported clusters of systemic lupus erythematosus (Balluz
12 et al., 2001; Dahlgren et al., 2007) and Wegener granulomatosis (Albert et al., 2005). Wegener
13 granulomatosis, an autoimmune disease involving small vessel vasculitis, usually with lung or
14 kidney involvement, is a very rare condition, with an incidence rate of 3-14 per million per year
15 (Mahr et al., 2006). Trichloroethylene was one of several ground water contaminants identified
16 in a recent study investigating a cluster of seven cases of Wegener granulomatosis around
17 Dublin, Pennsylvania. Because of the multiple contaminants, it is difficult to attribute the
18 apparent disease cluster to any one exposure.
19 In addition to the study of asthma and infectious disease history among residents of
20 Woburn, Massachusetts (Lagakos, 1986) (see Section 4.6.1.1.1), Byers et al. provide data
21 pertaining to immune function from 23 family members of leukemia patients in Woburn,
22 Massachusetts (Byers et al., 1988). Serum samples were collected in May and June of 1984 and
23 in November of 1985 (several years after 1979, when the contaminated wells had been closed).
24 Total lymphocyte counts and lymphocyte subpopulations (CD3, CD4, and CDS) and the
25 CD4/CD8 ratio were determined in these samples, and in samples from a combined control
26 group of 30 laboratory workers and 40 residents of Boston selected through a randomized
27 probability area sampling process. The study authors also assessed the presence of antinuclear
28 antibodies (ANA) or other autoantibodies (antismooth muscle, antiovarian, antithyroglobulin,
29 and antimicrosomal antibodies) in the family member samples and compared the results with
30 laboratory reference values. The age distribution of the control group, and stratified analyses by
31 age, are not provided. The lymphocyte subpopulations were higher and the CD4/CD8 ratio was
32 lower in the Woburn family members compared to the controls in both of the samples taken in
33 1984. In the 1985 samples, however, the subpopulation levels had decreased and the CD4/CD8
34 ratio had increased; the values were no longer statistically different from the controls. None of
35 the family member serum samples had antithyroglobulin or antimicrosomal antibodies, but
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-338 DRAFT—DO NOT CITE OR QUOTE
-------
1 10 family-member serum samples (43%) had ANA (compared to <5% expected based on the
2 reference value). Because the initial blood sample was taken in 1984, it is not possible to
3 determine the patterns at a time nearer to the time of the exposure. The coexposures that
4 occurred also make it difficult to infer the exact role of trichloroethylene in any alterations of the
5 immunologic parameters.
6 Kilburn and Warshaw reported data from a study of contamination by metal-cleaning
7 solvents (primarily trichloroethylene) and heavy metals (e.g., chromium) of the aquifer of the
8 Santa Cruz River in Tucson, Arizona (Kilburn and Warshaw, 1992). Exposure concentrations
9 above 5 ppb (6-500 ppb) had been documented in some of the wells in this area. A study of
10 neurological effects was undertaken between 1986 and 1989 (Kilburn and Warshaw, 1993), and
11 two of the groups within this larger study were also included in a study of symptoms relating to
12 systemic lupus erythematosus. Residents of Tucson (n = 362) were compared to residents of
13 southwest Arizona (n = 158) recruited through a Catholic parish. The Tucson residents were
14 selected from the neighborhoods with documented water contamination (>5 ppb
15 trichloroethylene for at least one year between 1957 and 1981). Details of the recruitment
16 strategy are not clearly described, but the process included recruitment of patients with lupus or
17 other rheumatic diseases (Kilburn and Warsaw, 1993, 1992). The prevalence of some self-
18 reported symptoms (malar rash, arthritis/arthralgias, Raynaud syndrome, skin lesions, and
19 seizure or convulsion was significantly higher in Tucson, but there was little difference between
20 the groups in the prevalence of oral ulcers, anemia, low white blood count or low platelet count,
21 pleurisy, alopecia, or proteinuria. The total number of symptoms reported was higher in Tucson
22 than in the other southwest Arizona residents (14.3 vs. 6.4% reported four or more symptoms,
23 respectively). Low-titer (1:80) ANA were seen in 10.6 and 4.7% of the Tucson and other
24 Arizona residents, respectively (p = 0.013). However, since part of the Tucson study group was
25 specifically recruited based on the presence of rheumatic diseases, it is difficult to interpret these
26 results.
27
28 4.6.1.1.4.2. Case-control studies. Interest in the role of organic solvents, including
29 trichloroethylene, in autoimmune diseases was spurred by the observation of a scleroderma-like
30 disease characterized by skin thickening, Raynaud's phenomenon, and acroosteolysis and
31 pulmonary involvement in workers exposed to vinyl chloride (Gama and Meira, 1978). A case
32 report in 1987 described the occurrence of a severe and rapidly progressive case of systemic
33 sclerosis in a 47-year-old woman who had cleaned X-ray tubes in a tank of trichloroethylene for
34 approximately 2.5 hours (Lockey et al., 1987).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-339 DRAFT—DO NOT CITE OR QUOTE
-------
1 One of the major impediments to autoimmune disease research is the lack of disease
2 registries, which make it difficult to identify incident cases of specific diseases (NIAMS, 2007).
3 There are no cohort studies of the incidence of autoimmune diseases in workers exposed to
4 trichloroethylene. Most of the epidemiologic studies of solvents and autoimmune disease rely on
5 general measures of occupational exposures to solvents, organic solvents, or chlorinated solvents
6 exposures. A 2- to 3-fold increased risk of systemic sclerosis (scleroderma) (Aryal et al., 2001;
7 Garabrant et al., 2003; Maitre et al., 2004), rheumatoid arthritis (Lundberg et al., 1994; Sverdrup
8 et al., 2005), undifferentiated connective tissue disease (Lacey et al., 1999), and antineutrophil-
9 cytoplasmic antibody (ANCA)-related vasculitis (Beaudreuil et al., 2005; Lane et al., 2003) has
10 generally been seen in these studies, but there was little evidence of an association between
11 solvent exposure and systemic lupus erythematosus in two recent case-control studies
12 (Cooper et al., 2004; Finckh et al., 2006).
13 Two case-control studies of scleroderma (Bovenzi et al., 2004; Maitre et al., 2004) and
14 two of rheumatoid arthritis (Olsson et al., 2004, 2000) provide data concerning solvent exposure
15 that occurred among metal workers or in jobs that involved cleaning metal (i.e., types of jobs
16 which were likely to use trichloroethylene as a solvent). There was a 2-fold increased risk
17 among male workers in the two studies of rheumatoid arthritis from Sweden (Olsson et al., 2004,
18 2000). The results from the smaller studies of scleroderma were more variable, with no exposed
19 cases seen in one study with 93 cases and 206 controls (Maitre et al., 2004), and an odds ratio of
20 5.2 (95% CI: 0.7, 37) seen in a study with 56 cases and 171 controls (Bovenzi et al., 2004).
21 Five other case-control studies provide data specifically about trichloroethylene exposure,
22 based on industrial hygienist review of job history data (see Table 4-59). Three of these studies
23 are of scleroderma (Diot et al., 2002; Garabrant et al., 2003; Nietert et al., 1998), one is of
24 undifferentiated connective tissue disease (Lacey et al., 1999), and one is of small vessel
25 vasculitidies involving ANCAs (Beaudreuil et al., 2005).
26 These studies included some kind of expert review of job histories, but only two studies
27 included a quantification of exposure (e.g., a cumulative exposure metric, or a "high" exposure
28 group) (Diot et al., 2002; Nietert et al., 1998). Most of the studies present data stratified by sex,
29 and as expected, the prevalence of exposure (either based on type of job or on industrial
30 hygienist assessment) is considerably lower in women compared with men. In men the studies
31 generally reported odds ratios between 2.0 and 8.0, and in women, the odds ratios were between
32 1.0 and 2.0. The incidence rate of scleroderma in the general population is approximately
33 5-10 times higher in women compared with men, which may make it easier to detect large
34 relative risks in men.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-340 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-59. Case-control studies of autoimmune diseases with measures of trichloroethylene exposure
to
vo £3'
I
I
§
***.
£3'
1
TO'
I
Disease, source of data
Results:
exposure prevalence, OR, 95% CI
Reference, location, sample
size, age
Scleroderma
Structured interview (specific jobs
and materials; jobs held 1 or more
years). Exposure classified by
self-report and by expert review
(job exposure matrix).
Structured interview (specific jobs
and materials; jobs held 6 or more
months). Exposure classified by
expert review.
Structured interview (specific jobs
and materials; jobs held 3 or more
months). Exposure classified by
self-report and by expert review.
Men
Maximum intensity 30% cases, 10% controls OR: 3.3 (1.0, 10.3)
Cumulative intensity 32% cases, 21% controls OR: 2.0 (0.7, 5.3)
Maximum probability 16% cases, 3% controls OR: 5 . 1 (not calculated)
Women:
Maximum intensity 6% cases, 7% controls OR: 0.9 (0.3, 2.3)
Cumulative intensity 10% cases, 9% controls OR: 1.2 (0.5, 2.6)
Maximum probability 4% cases, 5% controls OR: 0.7 (0.2, 2.2)
Men and women
any exposure: cases 16%, controls 8% OR: 2.4 (95% Cl: 1.0, 5.4)
high exposure:3 cases 9%, controls 1% OR: 7.6 (95% Cl: 1.5, 37.4)
Men
any exposure: cases 64%, controls 27% OR: 4.7 (95% Cl: 0.99, 22.0)
Women
any exposure: cases 9%, controls 4% OR: 2. 1 (95% Cl: 0.65, 6.8)
Women
Self report: cases 1.3%, controls 0.7% OR: 2.0 (95% Cl: 0.8, 4.8)
Expert review: cases 0.7%, controls 0.4% OR: 1.9 (95% Cl: 0.6, 6.6)
Nietertetal., 1998
South Carolina. Prevalent cases,
178 cases (141 women, 37
men), 200 hospital-based
controls. Mean age at onset
45.2yrs.
Diot et al, 2002
France. Prevalent cases, 80 cases
(69 women, 11 men), 160
hospital controls. Mean age at
diagnosis 48 yrs.
Garabrant et al., 2003
Michigan and Ohio. Prevalent
cases, 660 cases (all women),
2,227 population controls. b
Ages 18 and older.
Undifferentiated connective tissue disease
Structured interview (specific jobs
and materials; jobs held 3 or
more months). Exposure
classified by self-report and by
expert review.
Women
Self report: cases 0.5%, controls 0.7% OR: 0.88 (95% Cl: 0.11, 6.95)
Expert review: cases 0.5%, controls 0.4% OR: 1.67 (95% Cl: 0.19,
14.9)
Laceyetal., 1999
Michigan and Ohio. Prevalent
cases, 205 cases (all women),
2,095 population controls.
Ages 18 and older.
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
Table 4-59. Case-control studies of autoimmune diseases with measures of trichloroethylene exposure (continued)
Disease, source of data
Results:
exposure prevalence, OR, 95% CI
Reference, location, sample
size, age
ANCA-related diseases0
Structured interview (specific jobs
and materials; jobs held 6 or
more months). Exposure
classified by expert review.
Men and women (data not presented separately by sex)
cases 18.3%, controls 17.5% OR: 1.1 (0.5, 2.4)
Beaudreuil et al., 2005
France. Incident cases, 60
cases (-50% women), 120
hospital controls. Mean age 61
yrs.
TO'
""Cumulative exposure defined as product of probability x intensity x frequency x duration scores, summed across all jobs; scores of >1 classified as "high."
bTotal n; n with TCE data: self -report 606 cases, 2,138 control; expert review 606 cases, 2,137 controls.
Diseases included Wegener glomerulonephritis (n = 20), microscopic polyangiitis (n = 8), pauci-immune glomerulonephritis (n = 10), uveitis (n = 6),
Churg-Strauss syndrome (n = 4), stroke (n = 4) and other diseases (no more than 2 each).
I
TO
Co
H I
O
H
W
-------
1 The U.S. EPA conducted a meta-analysis of the three scleroderma studies with specific
2 measures of trichloroethylene (Diot et al., 2002; Garabrant et al., 2003; Nietert et al., 1998),
3 examining separate estimates for males and for females. The resulting combined estimate for
4 "any" exposure, using a random effects model to include the possibility of nonrandom error
5 between studies (DerSimonian and Laird, 1986), was OR: 2.5 (95% CI: 1.1, 5.4) for men and
6 OR: 1.2 (95% CI: 0.58, 2.6) in women. (Because the "any" exposure variable was not included
7 in the published report, Dr. Paul Nietert provided the U.S. EPA with a new analysis with these
8 results, e-mail communication from Paul Nietert to Glinda Cooper, November 28, 2007.)
9 Specific genes may influence the risk of developing autoimmune diseases, and genes
10 involving immune response (e.g., cytokines, major histocompatibility complex, B- and T-cell
11 activation) have been the focus of research pertaining to the etiology of specific diseases. The
12 metabolism of specific chemical exposures may also be involved (Cooper et al., 1999).
13 Povey et al. (2001) examined polymorphisms of two cytochrome CYP genes, CYP2E1 and
14 CYP2C19, in relation to solvent exposure and risk of developing scleroderma. These specific
15 genes were examined because of their hypothesized role in metabolism of many solvents,
16 including trichloroethylene. Seven scleroderma patients who reported a history of solvent
17 exposure were compared to 71 scleroderma patients with no history of solvent exposure and to
18 106 population-based controls. The CYP2E1*3 allele and the CYP2E1*4 allele were more
19 common in the 7 solvent-exposed patients (each seen in 2 of the 7 patients; 29%) than in either
20 of the comparison groups (approximately 5% for CYP2E1 *3 and 14% for CYP2E1 *4). The
21 authors present these results as observations that require a larger study for corroboration and
22 further elucidation of specific interactions.
23
24 4.6.1.2. Cancers of the Immune System, Including Childhood Leukemia
25 4.6.1.2.1. Description of studies. Human studies have reported cancers of the immune system
26 resulting from TCE exposure. Lymphoid tissue neoplasms arise in the immune system and result
27 from events that occur within immature lymphoid cells in the bone marrow or peripheral blood
28 (leukemias), or more mature cells in the peripheral organs (non-Hodgkin's lymphoma, NHL).
29 As such, the distinction between lymphoid leukemia and NHL is largely distributional with
30 overlapping entities, such that a particular lymphoid neoplasm may manifest both lymphomatous
31 and leukemic features during the course of the disease (Weisenberger, 1992). Lymphomas are
32 grouped according to the World Health Organization (WHO) classification as B-cell neoplasms,
33 T-cell/ natural killer (NK)-cell neoplasms, and Hodgkin's lymphoma, formerly known as
34 Hodgkin's disease (Harris et al., 2000).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-343 DRAFT—DO NOT CITE OR QUOTE
-------
1 Numerous studies are found in the published literature on lymphoma and either broad
2 exposure categories or occupational title. Most of these studies evaluate NHL, specifically. The
3 NHL studies generally report positive associations with organic solvents or job title as aircraft
4 mechanic, metal cleaner or machine tool operator, and printers, although associations are not
5 observed consistently across all studies, specific solvents are not identified, and different
6 lymphoma classifications are adopted (Alexander et al., 2007; Blair et al., 1993; Boffetta and de
7 Vocht, 2007; Chiu and Weisenburger, 2003; Dryver et al., 2004; Figgs et al., 1995;
8 Karunanayake et al., 2008; Lynge et al., 1997; Richardson et al., 2008; Seidler et al., 2007;
9 Mannetje et al., 2008; Tatham et al., 1997; Vineis et al., 2007; Schenk et al., 2009; Wang et al.,
10 2009). A major use of TCE is the degreasing as vapor or cold state solvent of metal and other
11 products with potential exposure in jobs in the metal industry, printing industry and aircraft
12 maintenance or manufacturing industry (Bakke et al., 2007). The recent NHL case-control study
13 of Purdue et al. (2009) examined degreasing tasks, specifically, and reported an increasing
14 positive trend between NHL risk in males and three degreasing exposure surrogates: average
15 frequency (hours/year) (p = 0.02), maximal frequency (hours/year), (p = 0.06), or cumulative
16 number of hours(p = 0.04).
17 As described in Appendix B, the U.S. EPA conducted a thorough and systematic search
18 of published epidemiological studies of cancer risk and trichloroethylene exposure using the
19 PubMed, ToxNet, and EMBASE bibliographic database. The U.S. EPA also requested
20 unpublished data pertaining to trichloroethylene from studies that may have collected these data
21 but did not include it in their published reports. ATSDR and state health department peer-
22 reviewed studies were also reviewed. Information from each of these studies relating to
23 specified design and analysis criteria was abstracted. These criteria included aspects of study
24 design, representativeness of study subjects, participation rate/loss to follow-up, latency
25 considerations, potential for biases related to exposure misclassification, disease
26 misclassification, and surrogate information, consideration of possible confounding, and
27 approach to statistical analysis. All studies are considered for hazard identification but those
28 studies more fully meeting the objective criteria provided the greater weight for identifying a
29 cancer hazard.
30 The body of evidence on lymphoma and trichloroethylene is comprised of occupational
31 cohort studies, population-based case-control studies and geographic studies. Four case-control
32 studies and four geographic studies also examine childhood leukemia and trichloroethylene.
33 Most studies report observed risk estimates and associated confidence intervals for lymphoma
34 and overall TCE exposure. The studies included a broad but sometimes slightly different group
35 of lymphosarcoma, reticulum-cell sarcoma, and other lymphoid tissue neoplasms, with the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-344 DRAFT—DO NOT CITE OR QUOTE
-------
1 exception of the Nordstrom et al. (1998) case-control study, which examined hairy cell leukemia,
2 now considered a lymphoma, and the Zhao et al. (2005) cohort study, which reported only results
3 for all lymphohematopoietic cancers, including nonlymphoid types. Persson and Fredrikson
4 (1999) do not identify the classification system for defining NHL, and Hardell et al. (1999)
5 define NHL using the Rappaport classification system. Miligi et al. (2006) used an NCI
6 classification system and considered chronic lymphocytic leukemias and NHLs together as
7 lymphomas, while Seidler et al. (2007) used the REAL classification system, which reclassifies
8 lymphocytic leukemias and NHLs as lymphomas of B-cell or T-cell origin. The cohort studies
9 (except for Zhao et al., 2005) and the case-control study of Siemiatycki (1991) have some
10 consistency in coding NHL, with NHL defined as lymphosarcoma and reticulum-cell sarcoma
11 (ICD code 200) and other lymphoid tissue neoplasms (ICD 202) using the ICD Revisions 7, 8, or
12 9. Revisions 7 and 8 are essentially the same with respect to NHL; under Revision 9, the
13 definition of NHL was broadened to include some neoplasms previously classified as Hodgkin's
14 lymphomas (Banks, 1992). Wang et al. (2009) refer to their cases as "NHL" cases; however,
15 according to the ICD-O classification system that they used, their cases are more specifically
16 various particular subtypes of malignant lymphoma (9590-9642, 9690-9701) and mast cell
17 tumors (9740-9750) (Morton et al., 2003). Fewer studies presented in published papers this
18 information for cell-specific lymphomas, leukemia, leukemia cell type, or multiple myeloma.
19 The seven cohort studies with data on the incidence of lymphopoietic and hematopoietic
20 cancer in relation to trichloroethylene exposure range in size (803 [Hansen et al., 2001] to 86,868
21 [Chang et al., 2005]), and were conducted in Denmark, Sweden, Finland, Taiwan and the United
22 States (see Table 4-60; for additional study descriptions, see Appendix B). Some subjects in the
23 Hansen et al. study are also included in a study reported by Raaschou-Nielsen et al. (2003);
24 however, any contribution from the former to the latter are minimal given the large differences in
25 cohort sizes of these studies (Hansen et al., 2001; Raaschou-Nielsen et al., 2003). The exposure
26 assessment techniques used in all studies except Chang et al. (2005) and Sung et al. (2007)
27 included a detailed job exposure matrix (Zhao et al., 2005; Blair et al., 1998), biomonitoring data
28 (Anttila et al., 1995; Axelson et al., 1994; Hansen et al., 2001), or reference to industrial hygiene
29 records on TCE exposure patterns and factors that affected exposure, indicating a high
30 probability of TCE exposure potential (Raaschou-Nielsen et al, 2003) with high probability of
31 TCE exposure to individual subjects. Subjects in Chang et al. (2005) and Sung et al. (2007), two
32 studies with overlapping subjects employed at an electronics plant in Taiwan, have potential
33 exposure to several solvents including TCE; all subjects are presumed as "exposed" because of
34 employment in the plant although individual subjects would be expected to have differing
35 exposure potentials. The lack of attribution of exposure intensity to individual subjects yields a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-345 DRAFT—DO NOT CITE OR QUOTE
-------
1 greater likelihood for exposure misclassification compared to the six other studies with exposure
2 assessment approaches supported by information on job titles, tasks, and industrial hygiene
3 monitoring data. Incidence ascertainment in two cohorts began 21 (Blair et al., 1998) and
4 38 years (Zhao et al., 2005) after the inception of the cohort. Specifically, Zhao et al. (2005)
5 note "results may not accurately reflect the effects of carcinogenic exposure that resulted in
6 nonfatal cancers before 1988." Because of the issues concerning case ascertainment raised by
7 this incomplete coverage, observations must be interpreted in light of possible bias reflecting
8 incomplete ascertainment of incident cases.
9 Eighteen cohort or PMR studies describing mortality risks from lymphopoietic and
10 hematopoietic cancer are summarized in Table 4-61 (for additional study descriptions, see
11 Appendix B). Two studies examined cancer incidence and are identified above (Blair et al.,
12 1998; Zhao et al., 2005). In 10 of the 18 studies presenting mortality risks (Blair et al., 1989;
13 Chang et al., 2003; Costa et al., 1989; Garabrant et al., 1988; Henschler et al., 1995; Sinks et al.,
14 1992; Sung et al., 2007; Wilcosky et al., 1984; ATSDR, 2004; Clapp and Hoffman, 2008), a
15 relatively limited exposure assessment methodology was used, study participants may not
16 represent the underlying population, or there was a low exposure prevalence of TCE exposure.
17 For reasons identified in the systematic review, these studies are given less weight in the overall
18 evaluation of the literature than the eight other cohort studies that better met the ideals of
19 evaluation criteria (Blair et al., 1998 and extended follow-up by Radican et al., 2008; Boice et
20 al., 1999, 2006; Greenland et al., 1994; Morgan et al., 1998; Ritz, 1999; Zhao et al., 2005).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-346 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
p
"I
I
§
^.
£3'
a
1
^t>
>
i
TO'
^5
I
f^l
w £
*i
•?•
I
g&
r* TO
C ^
^rl «s
H O
O §
og
^ a-
•^H SL"
g|
O^
HH Oq
H TO
SI
*^
O ^
S^
H
W
Table 4-60. Incidence cohort studies of TCE exposure and lymphopoietic and hematopoietic cancer risk
Population
exposure group
Lymphopoietic
cancer
Relative risk
(95% CI)a
«a
non-Hodgkin
lymphoma
Relative risk
(95% CI)a
«a
Leukemia
Relative risk
(95% CI)a
«a
Aerospace workers (Rocketdyne), CA
Any TCE exposure
Low cumulative TCE score
Medium cumulative TCE score
High cumulative TCE score
(p for trend)
Not reported
Not reported
1.0 (referent)
0.88 (0.47, 1.65)
0.20 (0.03, 1.46)
(0.097)
28
16
1
Electronic workers, Taiwan
All employees
Males
Females
Females
0.67 (0.42, 1.01)
0.73 (0.27, 1.60)
0.65 (0.37, 1.05)
22
6
16
Not reported
Not reported
Not reported
Not reported
0.78(0.49, 1.17)
23
Blue-collar workers, Denmark
Any exposure
Subcohort w/higher exposure"1
Employment duration
1-4.9 yrs
>5yrs
1.1 (1.0, 1.6)
Not reported
229
1.2(1.0, 1.5)
1.5(1.2,2.0)
1.5(1.1,2.1)
1.6(1.1,2.2)
96
65
35
30
1.2 (0.9, 1.4)
Not reported
82
Reference(s) and study description13
Zhao etal., 2005
n = 5,049 (2,689 with high cumulative
TCE exposure), began work before 1980,
worked at least 2 yrs, alive with no
cancer diagnosis in 1988, follow-up from
1988-2000, job exposure matrix
(intensity), internal referents (workers
with no TCE exposure). Leukemia
observations included in non-Hodgkin
lymphoma category
Chang et al., 2005; Sung et al., 2007
n = 88,868 (n = 70,735 female), follow-
up 1979-1997, does not identify TCE
exposure to individual subjects (Chang et
al., 2005)
n = 63,982 females, follow-up
1979-2001, dose not identify TCE
exposure to individual subjects (Sung et
al., 2007)
Raaschou-Nielsen et al., 2003
n = 40,049 (14,360 with presumed higher
level exposure to TCE), worked for at
least 3 months, follow-up from
1968-1997, documented TCE usec. U.S.
EPA based the lymphopoietic cancer
category on summation of ICD codes
200-204.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-60. Incidence cohort studies of TCE exposure and lymphopoietic and hematopoietic cancer risk
(continued)
Population
exposure group
Lymphopoietic
cancer
Relative risk
(95% CI)a
«a
non-Hodgkin
lymphoma
Relative risk
(95% CI)a
«a
Leukemia
Relative risk
(95% CI)a
«a
Biologically-monitored workers, Denmark
Any TCE exposure
Cumulative exposure (Ikeda), males
<17 ppm-yr
>17 ppm-yr
Mean concentration (Ikeda), males
<4ppm
4+ppm
Employment duration, males
<6.25 yr
>6.25 yr
2.0(1.1,3.3)
Not reported
Not reported
Not reported
15
3.1(1.3,6.1)
3.9(0.8,11)
3.1(0.6,9.1)
3.9(1.1,10)
3.2(1.1, 10)
2.5 (0.3, 9.2)
4.2(1.1,11)
8
3
3
4
4
2
4
2.0 (0.7, 4.4)
Not reported
Not reported
Not reported
6
Aircraft maintenance workers, Hill Air Force Base, UT
TCE Subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Not reported
1.0 (referent)
0.8 (0.4, 1.7)
0.7 (0.3, 1.8)
1.4 (0.6, 2.9)
36
12
7
17
Not reported
1.0 (referent)
0.9 (0.3, 2.6)
0.7 (0.2, 2.6)
1.0 (0.4, 2.9)
19
8
4
7
Not reported
1.0 (referent)
0.4(0.1,2.0)
0.9 (0.2, 3.7)
7
2
0
4
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0 (referent)
1.2(0.3,4.4)
1.9 (0.4, 8.8)
0.9(9.2,3.3)
3
2
3
1.0 (referent)
0.6(0.1,5.0)
0.9 (0.2, 4.5)
1
0
2
1.0 (referent)
2.4(0.3,21.8)
0
1
0
Reference(s) and study description13
Hansenetal.,2001
n = 803, U-TCA or air TCE samples,
follow-up 1968-1996 (subset of
Raaschlou-Nielsen et al. [2003] cohort).
U.S. EPA based the lymphopoietic cancer
category on summation of ICD codes
200-204
Blair etal., 1998
n = 10,461 men and 3,605 women (total
n = 14,066, n = 7,204 with TCE
exposure), employed at least 1 yr from
1952 to 1956, follow-up 1973-1990, job
exposure matrix (intensity), internal
referent (workers with no chemical
exposures)
00
I
o a.
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
2
•••2
I
o §
H I
Table 4-60. Incidence cohort studies of TCE exposure and lymphopoietic and hematopoietic cancer risk
(continued)
Population
exposure group
Biologically-monitored workers, Finland
Lymphopoietic
cancer
Relative risk
(95% CI)a
1.51(0.92,2.33)
«a
20
non-Hodgkin
lymphoma
Relative risk
(95% CI)a
1.81 (0.78, 3.56)
«a
8
Leukemia
Relative risk
(95% CI)a
1.08(0.35,2.53)
«a
5
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
1.36 (0.65, 2.49)
2.08 (0.95, 3.95)
10
9
2.01 (0.65, 4.69)
1.40(0.17,5.04)
5
2
0.39(0.01,2.19)
2.65 (0.72, 6.78)
1
4
Biologically-monitored workers, Sweden
Males, 2+ yrs exposure duration
0-17 ppm (Ikeda extrapolation)
18-35 ppm (Ikeda extrapolation)
>36 ppm (Ikeda extrapolation)
Females
1.17(0.47,2.40)
Not reported
Not reported
7
1.56 (0.51, 3.64)
1.44 (0.30, 4.20)
(0, 8.58)
6.25(0.16,34.8)
Not reported
5
o
J
0
1
Not reported
Not reported
Not reported
Reference(s) and study description13
Anttilaetal., 1995
n = 3,089 men and women, U-TCA
samples, follow-up 1967-1992
Axelsonetal., 1994
n = 1,421 men and 249 women (total
1,670), U-TCA samples, follow-up
1958 1987. U.o. EPA based the
lymphopoietic cancer category includes
ICD-7 200-203.
an = number of observed cases.
bStandardized incidence ratios using an external population referent group unless otherwise noted.
0 Exposure assessment based on industrial hygiene data on TCE exposure patterns and factors that affect such exposure (Raaschou-Nielsen et al., (2002), with
high probability of TCE exposure potential to individual subjects. Companies included iron and metal (48%), electronics (11%), painting (11%), printing (eo/
chemical (5%), dry cleaning (5%), and other industries.
dDefined as at least 1 year duration and first employed before 1980.
O
H
W
-------
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic
cancer risk
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% CI)
«a
non-Hodgkin lymphoma
Relative risk
(95% CI)
«a
Leukemia
Relative risk
(95% CI)
«a
Computer manufacturing workers (IBM), NY
Males
Females
2.24(1.01,4.19)
9
0
Aerospace workers (Rocketdyne), CA
Any TCE (utility /eng flush)
Any TCE exposure
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
(p for trend)
0.74 (0.34, 1.40)
Not reported
Not reported
9
0.21(0.01, 1.18)
Not reported
1.0 (referent)
1.49(0.86,2.57)
1.30(0.52,3.23)
(0.370)
1
60
27
27
6
1.08(0.35,2.53)
Not reported
5
Reference(s) and study description13
Clapp and Hoffman, 2008
n = 1 15 cancer deaths from
1969-2001, proportional cancer
mortality ratio, does not identify TCE
exposure to individual subjects. U.S.
EPA based the lymphopoietic cancer
category on "all lymphatic cancers."
Boiceetal.,2006
n = 41,351 (1,111 Santa Susana
workers with any TCE exposure),
employed on or after 1948-1999,
worked >6 months, follow-up to
1999, job exposure matrix without
quantitative estimate of TCE
intensity.
Zhao etal., 2005
n = 6,044 (n = 2,689 with high
cumulative level exposure to TCE),
began work and worked at least 2 yrs
in 1950 or later - 1993, follow-up to
2001, job exposure matrix (intensity),
internal referents (workers with no
TCE exposure). Leukemia
observations included in non-Hodgkin
lymphoma category.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% Cl)
«a
non-Hodgkin lymphoma
Relative risk
(95% Cl)
«a
Leukemia
Relative risk
(95% Cl)
«a
View-Master employees, OR
Males
Females
0.58(0.11, 1.69)
0.64(0.28, 1.26)
o
3
8
0.69 (0.08, 2.49)
0.52(0.14, 1.33)
2
4
0.50 (0.01, 2.79)
0.67(0.14, 1.96)
1
3
Electronic workers, Taiwan
All employees
Males
Females
Not reported
Not reported
1.27 (0.41, 2.97)
1.14(0.55,2.10)
5
10
0.44 (0.05, 1.59)
0.54 (0.23, 1.07)
2
8
Aerospace workers (Lockheed), CA
Routine TCE, any exposure
1.5 (0.81, 1.60)
36
1.19(0.65, 1.99)
14
1.05 (0.54, 1.84)
12
Routine-intermittent
Any TCE exposure
Duration of exposure
Oyrs
5yrs
p for trend
Not reported
Not reported
Not reported
1.0 (referent)
0.74 (0.32, 1.72)
1.33 (0.64, 2.78)
1.62 (0.82, 3.22)
0.20
32
7
10
14
Not reported
Not reported
Reference(s) and study description13
ATSDR, 2004
n = 616 deaths from 1989-2001,
proportional mortality ratio, does not
identify TCE exposure to individual
subjects. U.S. EPA based the non-
Hodgkin lymphoma cancer category
on "other lymphopoietic tissue."
Chang et al., 2003
n = 88,868 (n = 70,735 female), began
work 1978-1997, follow-up
1985 1997, does not identify ICE
exposure to individual subjects.
Boiceetal., 1999
n = 77,965 (n = 2,267 with routine
TCE exposure and n = 3.016 with
intermittent-routine TCE exposure),
began work >1960, worked at least 1
yr, follow-up from 1960-1996, job
exposure matrix without quantitative
estimate of TCE intensity.
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
o §
H I
O >
HH Oq
H ^
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% Cl)
na
non-Hodgkin lymphoma
Relative risk
(95% Cl)
«a
Leukemia
Relative risk
(95% Cl)
na
Uranium-processing workers (Fernald), OH
Any TCE exposure
No TCE exposure
Light TCE exposure, >2 yrs
Moderate TCE exposure, >2 yrs
Not reported
1.0 (referent)
1.45 (0.68, 3.06)c
1.17(0.15, 9.00)c
18
1
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Aerospace workers (Hughes), CA
TCE subcohort
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
0.99 (0.64, 1.47)
1.07(0.51, 1.96)
0.95 (0.53, 1.57)
25
10
15
0.96 (0.20, 2.81)d
1.01 (0.46, 1.92)e
1.79(0.22, 6.46)d
0.50 (0.01, 2.79)d
o
J
9
2
1
1.05 (0.50, 1.93)
0.85(0.17,2.47)
1.17(0.47,2.41)
10
o
J
1
TCE subcohort (Cox Analysis)
Never exposed
Ever exposed
1.0 (referent)
1.05(0.67, 1.65) f
82
25
1.0 (referent)
1.36(0.35, 5.22) d'f
8
o
J
1.0 (referent)
0.99(0.48, 2.03) f
32
10
Peak
No/Low
Medium/High
1.0 (referent)
1.08 (0.64, 1.82)
90
17
1.0 (referent)
1.31(0.28, 6.08)d
9
2
1.0 (referent)
1.10(0.49,2.49)
35
7
Cumulative
Referent
Low
High
1.0 (referent)
1.09(0.56,2.14)
1.03 (0.59, 1.79)
82
10
15
1.0 (referent)
2.25(0.46, ll.l)d
0.81(0.10, 6.49)d
8
2
1
1.0 (referent)
0.69 (0.21, 2.32)
1.14(0.5,2.60)
32
3
7
Reference(s) and study description13
Ritz, 1999
n = 3,814 (n = 2,971 with TCE),
began work 1951-1972, worked >3
months, follow-up to 1989, internal
referents (workers with no TCE
exposure).
Morgan etal., 1998
n = 20,508 (4,733 with TCE
exposure), worked >6 months
1950 1985, lollow-up to 1993,
external and internal (all non-TCE
exposed workers) workers referent,
job exposure matrix (intensity).
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
o §
H I
O >
HH Oq
H ^
O
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% Cl)
na
non-Hodgkin lymphoma
Relative risk
(95% Cl)
«a
Leukemia
Relative risk
(95% Cl)
na
Aircraft maintenance workers, Hill Air Force Base, UT
TCE subcohort
1.1(0.7, 1.8)g
66
2.0(0.9, 4.6) g
28
0.6(0.3, 1.2)g
16
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0 (referent)
1.1(0.6,2.1)
1.0(0.4,2.1)
1.3 (0.7, 2.5)
21
11
21
1.0 (referent)
1.8 (0.6,5.4)
1.9 (0.6,6.3)
1.1 (0.3,3.8)
10
6
5
1.0 (referent)
1.0(0.3,3.2)
1.2 (0.4, 3.6)
7
0
7
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0 (referent)
1.5 (0.6, 4.0)
0.7(0.1,4.9)
1.1(0.4,3.0)
1.06(0.75, 1.51)h
1.12(0.72, 1.73)
1.0 (referent)
1.04 (0.63, 1.74)
1.06 (0.49, 1.88)
1.25(0.75,2.09)
1.00(0.55, 1.83)
1.0 (referent)
1.10(0.48,2.54)
0.38 (0.05, 2.79)
1.11(0.53,2.31)
6
1
6
106
88
34
21
33
18
7
1
10
3.8(0.8,18.9)
3.6 (0.8, 16.2)
1.36(0.77, 2.39) h
1.56(0.79,4.21)
1.0 (referent)
1.83 (0.79, 4.21)
1.17(0.42,3.24)
1.50 (0.61, 3.69)
1.18(0.49,2.85)
1.0 (referent)
1.48 (0.47, 4.66)
1.30 (0.45, 3.77)
3
0
4
46
37
18
7
12
9
4
0
5
1.0 (referent)
0.4(0.1,3.2)
0.3(0.1,2.4)
0.64(0.35, 1.18)h
0.77 (0.37, 1.62)
1.0 (referent)
0.86 (0.36, 2.02)
0.51 (0.16, 1.63)
0.87(0.35,2.14)
0.36(0.10, 1.32)
1.0 (referent)
0.35 (0.05, 2.72)
0.48(0.10,2.19)
1
0
1
27
24
11
4
9
3
1
0
2
Reference(s) and study description13
Blair et al., 1998; Radican et al., 2008
n = 14,066 (n = 7,204 ever exposed to
TCE), employed at least 1 yr from
1952 to 1956, follow-up to 1990
(Blair etal., 1998) or to 2000
(Radican et al., 2008), job exposure
matrix, internal referent (workers with
no chemical exposures).
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% Cl)
na
non-Hodgkin lymphoma
Relative risk
(95% Cl)
«a
Leukemia
Relative risk
(95% Cl)
na
Cardboard manufacturing workers, Arnsburg, Germany
TCE-exposed subjects
Unexposed subjects from same
factory
1.10(0.03,6.12)
1.11 (0.03,6.19)
1
1
General Electric plant, Pittsfield, MA
0.76 (0.24, 2.42)''J
15
1.1(0.46,2.66)'
22
Cardboard manufacturing workers, Atlanta, GA
0.3 (0.0, 1.6)
1
Not reported
Not reported
U. S, Coast Guard employees
Marine inspectors
Noninspectors
1.57(0.91,2.51)
0.60 (0.24, 1.26)
17
7
1.75(0.48,4.49)
0.41(0.01,2.30)
4
1
1.55(0.62,3.19)
0.66(0.14, 1.94)
7
3
Reference(s) and study description13
Henschler et al., 1995
n = 169 TCE exposed and n = 190
unexposed men, employed >1 yr from
1956-1975, follow-up to 1992, local
population referent, qualitative
exposure assessment.
Greenland etal., 1994
Nested case-control study, n = 512
cancer (cases) and 1,202 noncancer
(controls) male deaths reported to
pension fund between 1969-1984
among workers employed <1984 and
with job history record, job exposure
matrix-ever held job with TCE
exposure.
Sinks etal., 1999
n = 2,050, employed on or before
1957-1988, follow-up to 1988,
Material Data Safety Sheets used to
identify chemicals used in work areas.
Blair etal., 1988
n=3,781 males (1,767 marine
inspectors), employed 1942-1970,
follow-up to 1980. TCE and nine
other chemicals identified as potential
exposures; no exposure assessment to
individual subjects.
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Lymphopoietic cancer
Relative risk
(95% Cl)
na
non-Hodgkin lymphoma
Relative risk
(95% CT)
n*
Leukemia
Relative risk
(95% Cl)
na
Aircraft manufacturing employees, Italy
All male subjects
0.80 (0.41, 1.40)
12
Not reported
Not reported
Aircraft manufacturing, San Diego, CA
All employees
0.82(0.56, 1.15)
o ^>
32
0.82 (0.44, 1.41)d
0.65(0.21, 1.52)k
13
5
0.82 (0.47, 1.32)
10
Reference(s) and study description13
Costa etal., 1989
n = 7,676 males, employed on or
before 1954-1981, followed to 1981,
job titles of white- and blue-collar
workers, technical staff, and admin.
clerks, does not identify TCE
exposure to individual subjects.
Garabrantetal., 1988
n = 14,067, employed at least 4 yrs
with company and >1 d at San Diego
plant from 1958-1982, followed to
1982, does not identify TCE exposure
to individual subjects.
I
H I
O >
HH Oq
H ^
O
H
W
-------
Table 4-61. Mortality cohort and PMR studies of TCE exposure and lymphopoietic and hematopoietic cancer
risk (continued)
Population,
exposure group
Solvent-exposed rubber workers
Lymphopoietic cancer
Relative risk
(95% CI)
2.4 '
na
o
J
non-Hodgkin lymphoma
Relative risk
(95% CI)
0.81
«a
o
J
Leukemia
Relative risk
(95% CI)
na
Reference(s) and study description13
Wilcosky et al., 1984
Nested case-control study, n = 9
lymphosarcoma and 10 leukemia
(cases) and 20% random sample of all
other deaths (controls) between
1964-1973 in cohort of n = 6,678,
exposure assessment by company
record for use in work area.
aw = number of observed cases.
bUnless otherwise noted, all studies reported standardized mortality ratios using an external population referent group.
'Logistic regression analysis with 15 lag for TCE exposure (Ritz, 1999).
dln Morgan et al. (1998) and Garabrant et al. (1988), this category was based on lymphosarcoma and reticulosarcoma.
eAs presented in Mandel et al. (2006), this category defined as ICD -7, ICDA-8, and ICD-9 codes of 200 and 202.
fRisk ratio from Cox Proportional Hazard Analysis, stratified by age and sex, from Environmental Health Strategies (1997) Final Report to Hughes Corporation
(Communication from Paul A. Cammer, President, Trichloroethylene Issues Group to Cheryl Siegel Scott, U.S. EPA, December 22, 1997).
8 Estimated relative risks from Blair et al. (1998) from Poisson regression models adjusted for date of hire, calendar year of death and sex.
h Estimated relative risks from Radican et al. (2008) from Cox proportional hazard models adjusted for age and sex.
1 Odds ratio from nested case-control analysis.
JLymphomas, lymphosarcomas, and reticulosarcomas (ICDA8 200-202) in Greenland et al. (1994).
kOther lymphatic and hematopoietic tissue neoplasms (Garabrant et al., 1988).
-------
1 Case-control studies of lymphoma or hairy cell leukemia (a lymphoma according to the
2 WHO's lymphoma classification system [Morton et al., 2007, 2006]) from United States
3 (Connecticut), Germany, Italy, Sweden, and Canada were identified, and are summarized in
4 Table 4-62 (for additional study descriptions, see Appendix B). These studies identified cases
5 from hospital records (Costantini et al., 2008; Hardell et al., 1994; Mester et al., 2006; Miligi et
6 al., 2006; Persson and Fredrikson, 1999; Seidler et al., 2007; Siemiatycki et al., 1991); the
7 Connecticut Tumor Registry (Wang et al., 2009); or the Swedish Cancer Registry (Nordstrom et
8 al., 1998), and population controls. These studies assign potential occupational TCE exposure to
9 cases and controls using self-reported information obtained from a mailed questionnaire (Hardell
10 et al., 1994; Nordstrom et al., 1998; Persson and Fredrikson, 1999) or from direct interview with
11 study subjects, with industrial hygienist ratings of exposure potential and a job exposure matrix
12 (Siemiatycki et al., 1991; Miligi et al., 2006; Seidler et al., 2007; Costantini et al., 2008; Wang et
13 al., 2009). Additionally, three of these large multiple center lymphoma case-control studies
14 examine specific types of NHL (Miligi et al., 2006; Seidler et al., 2007; Wang et al., 2009) or
15 leukemia (Costantini et al., 2008).
16 Four geographic based studies on lymphoma in adults are summarized in Table 4-63 (for
17 additional study descriptions, see Appendix B) and subjects in three studies are identified based
18 upon their residence in a community where TCE was detected in water serving the community
19 (Vartianen et al., 1993; Cohn et al., 1994; ATSDR, 2006). Both Cohn et al. (1994) and ATSDR
20 (2006) also present estimates for childhood leukemia and these observations are discussed below
21 with other studies reporting on childhood leukemia. A subject is assumed to have a probability
22 of exposure due to residence likely receiving water containing TCE. Most studies do not include
23 statistical models of water distribution networks, which may influence TCE concentrations
24 delivered to a home, nor a subject's ingestion rate to estimate TCE exposure to individual study
25 subjects. ATSDR (2004, 2006) adopts exposure modeling of soil vapor contamination to define
26 study area boundaries and to identify census tracts with a higher probability of exposure to
27 volatile organic solvents without identifying exposure concentrations to TCE and other solvents.
28 In these studies, one level of exposure to all subjects in a geographic area is assigned, although
29 there is some inherent measurement error and misclassification bias because not all subjects are
30 exposed uniformly.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-357 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-62. Case-control studies of TCE exposure and lymphopoietic cancer or leukemia
to
VO Lo'
I
I
§
***.
£3'
1
TO'
2
•••2
I
TO
2.
H I
O >
HH Oq
H TO
Population
Women aged 21-84 in CT, USA
Population in 6 German regions
Cancer type and exposure group
Odds ratio
(95% CI)
n exposed
cases
Non-Hodgkin lymphoma
Any TCE exposure
Low intensity TCE exposure
Medium-high intensity TCE exposure
(p for linear trend)
Low probability TCE exposure
Medium-high probability TCE exposure
(p for linear trend)
Low intensity TCE exposure/low probability
Low intensity /medium-high probability
Medium-high intensity/low probability
Medium-high intensity/medium-high
probability
1.2 (0.9, 1.8)
1.1(0.8, 1.6)
2.2 (0.9, 5.4)
0.06
1.1 (0.7, 1.8)
1.4 (0.9, 2.4)
0.37
0.9 (0.6, 1.5)
1.4 (0.9, 2.4)
2.2 (0.9, 5.4)
77
64
13
43
34
30
34
13
0
Non-Hodgkin lymphoma
Any TCE exposure
Cumulative TCE
0 ppm-yrs
>0-<4 ppm-yrs
4.4-<35 ppm-yrs
High exposure, >35 ppm-yrs
(p for linear trend)
>35 ppm-yrs, 10 yr lag
Not reported
1.0
0.7(0.4,1.1)
0.7 (0.5, 1.2)
2.1(1.0,4.8)
0.14
2.2(1.0,4.9)
610
40
32
21
Reference(s)
Wang etal., 2009
Seidler et al., 2007; Mester et al., 2006
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
2
•••2
I
Table 4-62. Case-control studies of TCE exposure and lymphopoietic cancer or leukemia (continued)
Population
Population in 6 German regions
(continued)
Cancer type and exposure group
Odds ratio
(95% CI)
n exposed
cases
B-cell NHL
Cumulative TCE
0 ppm-yrs
>0-<4 ppm-yrs
4.4-<35 ppm-yrs
High exposure, >35 ppm-yrs
(p for linear trend)
1.0
0.7 (0.5, 1.2)
0.8 (0.5, 1.3)
2.3 (1.0,5.3)
0.08
47
32
27
17
Diffuse B-cell NHL
Cumulative TCE
0 ppm-yrs
>0-<4 ppm-yrs
4.4-<35 ppm-yrs
High exposure, >35 ppm-yrs
(p for linear trend)
1.0
0.5 (0.2, 1.2)
0.8 (0.3, 1.8)
2.6 (0.7, 3.0)
0.03
139
6
7
4
Chronic lymphocytic Leukemia
Cumulative TCE
0 ppm-yrs
>0-<4 ppm-yrs
4.4-<35 ppm-yrs
High exposure, >35 ppm-yrs
(p for linear trend)
1.0
1.1(0.5,2.4)
0.7 (0.3, 1.7)
0.9(0.2,4.5)
0.46
610
10
6
2
Reference(s)
H I
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
Table 4-62. Case-control studies of TCE exposure and lymphopoietic cancer or leukemia (continued)
Population
Population in 8 Italian regions
Cancer type and exposure group
Odds ratio
(95% CI)
n exposed
cases
Non-Hodgkin lymphoma
Any TCE exposure
TCE exposure intensity
Very low/low
Medium/high
(p for linear trend)
Duration exposure, med/high TCE intensity
<15yr
>15yr
(p for linear trend)
Not reported
0.8 (0.5, 1.3)
1.2(0.7,2.0)
0.8
1.1(0.6,2.1)
1.0 (0.5, 2.6)
0.72
35
35
22
12
Other non-Hodgkin lymphoma
TCE exposure intensity, medium/high
Small lymphocytic NHL
FollicularNHL
Diffuse NHL
Other NHL
0.9(0.4,2.1)
Not presented
1.9(0.9,3.7)
1.2(0.6,2.4)
7
3
13
11
Leukemia
Any TCE exposure
TCE exposure intensity
Very low/low
Medium/high
Not reported
1.0 (0.5, 1.8)
0.7 (0.4, 1.5)
17
11
Acute myeloid leukemia
Any TCE exposure
TCE exposure intensity
Very low/low
Medium/high
Not reported
1.0(0.4,2.5)
1.1(0.5,2.9)
6
6
Reference(s)
Miligi et al., 2006
Costantini et al., 2008
H I
O >
HH Oq
H ^
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
Table 4-62. Case-control studies of TCE exposure and lymphopoietic cancer or leukemia (continued)
Population
Population in 8 Italian regions
(continued)
Population of Orebro and
Linkoping, Sweden
Population of Sweden
Population of Umea, Sweden
Population of Montreal, Canada
Cancer type and exposure group
Odds ratio
(95% CI)
n exposed
cases
Chronic lymphocytic leukemia
Any TCE exposure
TCE exposure intensity
Very low/low
Medium/high
Not reported
1.2(0.5,2.7)
0.9 (0.3, 2.6)
8
4
B-cell non-Hodgkin lymphoma
Any TCE exposure
1.2(0.5,2.4)
16
Hairy cell lymphoma
Any TCE exposure
1.5(0.7,3.3
9
Non-Hodgkin lymphoma
Any exposure to TCE
7.2(1.3,42)
4
Non-Hodgkin lymphoma
Any TCE exposure
Substantial TCE exposure
1.1(0.6,2.3)*
0.8 (0.2, 2.5)*
6
2
Reference(s)
Persson and Fredrikson, 1999
Nordstrom etal., 1998
Hardell et al., 1994
Siemiatycki et al., 1991
*90% confidence interval.
H I
O >
HH Oq
H ^
O
H
W
-------
Table 4-63. Geographic-based studies of TCE and non-Hodgkin lymphoma or leukemia in adults
to
vo £3'
I
I
§
***.
£3'
1
TO'
Population
Exposure group
Two study areas in Endicott, NY
Residents of 13 census tracts inRedland, CA
Population in New
Jersey
Population in Finland
non-Hodgkin lymphoma
Relative risk
(95% CI)
0.54(0.22, 1.12)
1.09 (0.84, 1.38)
n exposed
cases
7
111
Leukemia
Relative risk
(95% CI)
0.79 (0.34, 1.55)
1.02 (0.74, 1.35)
n exposed
cases
8
77
Males, maximum estimated TCE concentration (ppb) in municipal drinking water
<0.1
0.1-0.5
>5.0
1.00
1.28(1.10, 1.48)
1.20 (0.94, 1.52)
493
272
78
1.00
0.85 (0.71, 1.02)
1.10(0.84, 1.90)
438
162
63
Females, maximum estimated TCE concentration (ppb) in municipal drinking water
0.1
0.1-0.5
>5.0
Residents of Hausjarvi
Residents of Huttula
1.00
1.02 (0.87, 1.2)
1.36 (1.08, 1.70)
0.6(0.3, 1.1)
1.4(1.0,2.0)
504
26
87
14
13
1.00; 315
1.13 (0.93, 1.37)
1.43 (1.43, 1.90)
1.2 (0.8, 1.7)
0.7(0.4,1.1)
156
56
33
19
Reference
ATSDR, 2006
Morgan and Cassady, 2002
Conn et al., 1994
Vartiainen et al., 1993
I
o §
H I
O >
HH Oq
H ^
O
H
W
-------
1 NHL incidence is statistically significantly elevated in three high-quality studies (3.1,
2 95% CI: 1.3, 6.1 [Hansen et al., 2001]; 1.5, 95% CI: 1.2, 2.0, sub cohort with higher exposure
3 [Raaschou-Nielsen et al., 2003], 2.1, 95% CI: 1.0, 4.8, >35-ppm years cumulative TCE exposure
4 [Seidler et al., 2007]). Two of these incidence studies report statistically significantly
5 associations for all lymphopoietic and hematopoietic cancer, specifically NHL, for subjects with
6 longer employment duration as a surrogate of TCE exposure (>6.25 year, 4.2, 95% CI: 1.1, 11
7 [Hansen et al., 2001]; >5 year, 1.6, 95% CI: 1.1, 2.2, [Raaschou-Nielsen et al., 2003]) and
8 Seidler et al. (2007) report a positive trend with diffuse B-cell NHL and cumulative TCE
9 exposure (p = 0.03). Hansen et al. (2001) also examined two other exposure surrogates,
10 cumulative exposure and exposure intensity, with estimated risk larger in low exposure groups
11 than for high exposure groups. A fourth study from Sweden reports a large and imprecise risk
12 with TCE (7.2, 95% CI: 1.3, 42 [Hardell et al., 1994]) based on four exposed cases. High-quality
13 cohort mortality studies and other case-control studies observed a 10 to 50% increased risk
14 between NHL and any TCE exposure (1.2, 95% CI: 0.65, 1.99 [Boice et al., 1999]; 1.36, 95%
15 CI: 0.28, 6.08 [Morgan et al., 1998]; 1.5, 95% CI: 0.7, 3.3 [Nordstrom et al., 1998]; 1.2, 95% CI:
16 0.5, 2.4 [Persson and Fredrikson, 1999]; 1.36, 95% CI: 0.77, 2.39 [Radican et al., 2008]; 1.1,
17 95% CI: 0.6, 2.3 [Siemiatycki, 1991]; 1.2, 95% CI: 0.9, 1.8 [Wang et al., 2009]).
18 Odds ratios are higher for diffuse NHL, primarily a B-cell lymphoma, than for all
19 non-Hodgkin lymphomas in both studies which examine forms of lymphoma (Miligi et al., 2006;
20 Seidler et al., 2007) (see Table 4-63). Observations in the two other studies of B-cell lymphomas
21 (Persson and Fredrikson, 1999; Wang et al., 2009) appear consistent with Miligi et al. (2006) and
22 Seidler et al. (2007). Together, these observations suggest that the associations between
23 trichloroethylene and diffuse NHL are stronger than the associations seen with other forms of
24 lymphoma, and that disease misclassification may be introduced in studies examining
25 trichloroethylene and NHL as a broader category. Mortality observations in other occupational
26 cohorts (Wilcosky et al., 1984; Garabrant et al., 1988; Costa et al., 1989; Greenland et al., 1994;
27 Ritz, 1999; Henschler et al., 1995; Chang et al., 2003; ATSDR, 2004, Boice et al., 2006;
28 Sung et al., 2007) included a risk estimate of 1.0 in 95% confidence intervals; these studies
29 neither add to nor detract from the overall weight of evidence given their lower likelihood for
30 TCE exposure due to inferior exposure assessment approaches, lower prevalence of exposure,
31 lower statistical power, and fewer exposed deaths.
32 Seven studies presented estimated risks for leukemia and overall TCE exposure
33 (Anttila et al., 1995; Blair et al., 1998 and its update by Radican et al., 2008; Morgan et al., 1998;
34 Boice et al., 1999, 2006; Hansen et al., 2001; Raachou-Nielsen et al., 2003); only three studies
35 also presented estimated risks for a high exposure category (Anttila et al., 1995; Morgan et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-363 DRAFT—DO NOT CITE OR QUOTE
-------
1 1998; Blair et al., 1998). Two case-control studies presented estimated risk for leukemia
2 categories and low or high TCE exposure category (Seidler et al., 2007; Costantini et al., 2008);
3 however, neither study presented estimated risk for overall TCE exposure. Risk estimates in
4 high-quality cohort studies ranged from 0.64 (95% CI: 0.35, 1.18) (Radican et al., 2008) to 2.0
5 (95% CI: 0.7, 4.44) (Hansen et al, 2001). The largest study, with 82 observed incident leukemia
6 cases, reported a relative risk estimate of 1.2 (95% CI: 0.9, 1.4) (Raaschou-Nielsen et al., 2003).
7 Both case-control studies which examined leukemia risk and TCE exposure are quite limited in
8 statistical power, Costantini et al. (2008) was the largest with 11 exposed cases, and did not
9 provide evidence for an association.
10 The number of studies of childhood lymphoma including acute lymphatic leukemia and
11 trichloroethylene is much smaller than the number of studies of trichloroethylene and adult
12 lymphomas, and consists of four case-control studies (Costas et al., 2002; Lowengart et al., 1987;
13 McKinney et al., 1991; Shu et al., 1999) and four geographic based studies (Aickin et al., 1992;
14 AZ DHS, 1990, 1995; ATSDR, 2006, 2008; Cohn et al., 1994) (see Table 4-64). An additional
15 publication, focusing on ras mutations, based on one of the case-control studies is also available
16 (Shu et al., 2004). All four case-control studies evaluate maternal exposure, and three studies
17 also examine paternal occupational exposure (Lowengart et al., 1987; McKinney et al., 1991;
18 Shu et al., 2004, 1999). There are relatively few cases with maternal exposure (range 0 to 16) in
19 these case-control studies, and only Shu et al. have a large number (n = 136) of cases with
20 paternal exposure (Shu et al., 2004, 1999). The small numbers of exposed case parents limit
21 examination of possible susceptibility time windows. Overall, evidence for association between
22 parental trichloroethylene exposure and childhood leukemia is not robust or conclusive.
23 The results from the studies of Costas et al. (2002) and Shu et al. (1999, 2002) suggest a
24 fetal susceptibility to maternal exposure during pregnancy, with relative risks observed for this
25 time period equal or higher than the relative risks observed for periods before conception or after
26 birth (see Table 4-64). The studies by Lowengart et al. (1987) and McKinney et al. (1991) do
27 not provide informative data pertaining to this issue due to the small number (n = <3) of exposed
28 case mothers. A recent update of a cohort study of electronics workers at a plant in Taiwan
29 (Chang et al., 2003, 2005) reported a 4-fold increased risk (3.83; 95% CI: 1.17, 12.55
30 [Sung et al., 2008]) for childhood leukemia risk among the offspring of female workers
31 employed during the three months before to three months after conception. Exposures at this
32 factory included trichloroethylene, perchloroethylene, and other organic solvents (Sung et al.,
33 2008). The lack of TCE assignment to individual subjects in this study decrease its weight in the
34 overall analysis.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-364 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-64. Selected results from epidemiologic studies of TCE exposure and
childhood leukemia
Relative risk
(95% CI)
n observed
events
Reference(s)
Cohort studies (solvents)
Childhood leukemia among offspring of electronic workers
Nonexposed
Exposed pregnancy to organic solvents
1.0a
3.83(1.17, 12.55)
9
6
Sung et al., 2008
Case-control studies
Children's Cancer Group Study (children < 15 yrs age)
Acute lymphocytic leukemia
Maternal occupational exposure to TCE
Anytime
Preconception
During pregnancy
Postnatal
1.8(0.8,4.1)
1.8 (0.8, 5.2)
1.8 (0.5, 6.4)
1.4(0.5,4.1)
15
9
6
9
Paternal occupational exposure to TCE
Anytime
Preconception
During pregnancy
Postnatal
1.1 (0.8, 1.5)
1.1 (0.8, 1.5)
0.9 (0.6, 1.4)
1.0(0.7, 1.3)
136
100
56
77
K-ras + acute lymphocytic leukemia
Maternal occupational exposure to TCE
Anytime
Preconception
During pregnancy
Postnatal
1.8 (0.6, 4.8)
2.0 (0.7, 6.3)
3.1(1.0,9.7)
5
4
4
0
Paternal occupational exposure to TCE
Anytime
Preconception
During pregnancy
Postnatal
0.6 (0.3, 1.4)
0.6(0.3, 1.5)
0.3(0.1, 1.2)
0.4(0.1, 1.4)
9
8
2
3
Shuetal., 1999
Shu et al., 2004
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-365 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-64. Selected results from epidemiologic studies of TCE exposure and
childhood leukemia (continued)
Relative risk
(95% CI)
n observed
events
Residents of ages <19 in Woburn, MA
Maternal exposure 2 yrs before conception to diagnosis
Never
Least
Most
(p for linear trend)
1.00
5.00 (0.75, 33.5)
3.56(0.51,24.8)
>0.05
3
9
7
Maternal exposure 2 yrs before conception
Never
Least
Most
(p for linear trend)
1.00
2.48 (0.42, 15.2)
2.82 (0.30, 26.4)
>0.05
11
4
4
Birth to diagnosis
Never
Least
Most
(p for linear trend)
1.00
1.82(0.31, 10.8)
0.90(0.18,4.56)
>0.05
7
7
5
Maternal exposure during pregnancy
Never
Least
Most
(p for linear trend)
1.00
3.53(0.22,58.1)
14.3 (0.92, 224)
0.05
9
o
6
7
Population <14 yrs of age in 3 areas north England, United Kingdom
Acute lymphocytic leukemia and NHL
Maternal occupation exposure to TCE
Preconception
Paternal occupational exposure to TCE
Preconception
Periconception and gestation
Postnatal
1.16(0.13,7.91)
2.27(0.84,6.16)
4.49(1,15,21)
2.66(0.82,9.19)
2
9
7
7
Los Angeles Cancer Surveillance Program
Acute lymphocytic and nonlymphocytic leukemia, <10 yrs of age
Maternal occupational exposure to TCE
Paternal occupational exposure to TCE
One year before pregnancy
During pregnancy
After delivery
2.0 (p = 0.16)
2.0 (p =0.16)
2.7 (0.64, 15.6)
0
6/3b
6/3b
8/3b
Reference(s)
Costas et al., 2002
McKinneyetal., 1991
Lowengartetal., 1987
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-366 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-64. Selected results from epidemiologic studies of TCE exposure and
childhood leukemia (continued)
Relative risk
(95% CI)
n observed
events
Reference(s)
Geographic based studies
Two study areas in Endicott, NY
Leukemia, <19 yrs of age
Population in New Jersey
Not reported
<6
Acute lymphocytic leukemia
Maximum estimated TCE concentration (ppb) in municipal drinking water
Males
<0.1
0.1-0.5
>5.0
1.00
0.91(0.53, 1.57)
0.54(0.17, 17.7)
45
16
o
J
Females
<0.1
0.1-0.5
>5.0
1.00
1.85(1.03,3.70)
2.36(1.03,5.45)
25
22
7
Resident of Tucson Airport Area, AZ
Leukemia, <19 yrs of age
1970-1986
1987-1991
1.48(0.74,2.65)
0.80(0.31,2.05)
11
3
Resident of West Central Phoenix, AZ
Leukemia, <19 yrs of age
1.95(1.43,2.63)
38
ATSDR, 2006
Cohnetal., 1994
AZDHS, 1990, 1995
Aickinetal., 1992
"Internal referents, live born children among female workers not exposed to organic solvents.
bDiscordant pairs.
The evidence for an association between childhood leukemia and paternal exposure to
solvents is quite strong (Colt and Blair, 1998); however, for studies of TCE exposure, the small
numbers of exposed case fathers in two studies (McKinney et al., 1991; Lowengart et al., 1987)
and, for all three studies, likelihood of misclassification resulting from a high percentage of
paternal occupation information obtained from proxy interviews, limits observation
interpretations. Both Lowengart et al. (1987) and McKinney et al. (1991) provide some evidence
for a 2- to 4-fold increase of childhood leukemia risk and paternal occupational exposure
although the population study of Shu et al. (1999, 2002), with 13% of case father's occupation
reported by proxy respondents, does not appear to support the earlier and smaller studies.
The geographic based studies for adult lymphopoietic (see Table 4-63) or childhood
leukemias (see Table 4-64) do not greatly contribute to the overall weight of evidence. While
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-367 DRAFT—DO NOT CITE OR QUOTE
-------
some studies observed statistically significantly elevated risks for NHL or childhood cancer,
these studies generally fulfilled only the minimal of evaluation criteria with questions raised
about subject selection (Morgan and Cassady, 2002), their use of less sophisticated exposure
assessment approaches and associated assumption of an average exposure to all subjects (all
studies), and few cases with high level parental exposure (all studies).
4.6.1.2.2. Meta-analysis oflymphoma risk. Meta-analysis is adopted as a tool for examining
the body of epidemiologic evidence on lymphoma and TCE exposure and to identify possible
sources of heterogeneity. The meta-analysis oflymphoma examines 16 cohort and case-control
studies identified through a systematic review and evaluation of the epidemiologic literature on
TCE exposure (Siemiatycki et al., 1991; Axelson et al., 1994; Hardell et al., 1994; Anttila et al.,
1995; Greenland et al., 1994; Morgan et al., 1998; Nordstrom et al., 1998; Boice et al., 1999;
Persson and Fredrikson, 1999; Hansen et al., 2001; Raaschou-Nielsen et al., 2003; Zhao et al.,
2005; Miligi et al., 2006; Seidler et al., 2007; Radican et al., 2008; Wang et al., 2009) and two
studies as alternatives (Blair et al., 1998; Boice et al., 2006). These 18 studies oflymphoma and
TCE had high likelihood of exposure, were judged to have met, to a sufficient degree, the criteria
of epidemiologic design and analysis, and reported estimated risks for overall TCE exposure; 12
of these studies, also, presented estimated lymphoma risk with high level TCE exposure
(Siemiatycki et al., 1991; Axelson et al., 1994; Anttila et al., 1995; Blair et al., 1998; Morgan et
al., 1998; Boice et al., 1999; Hansen et al., 2001; Raaschou-Nielsen et al., 2003; Zhao et al.,
2005; Miligi et al., 2006; Seidler et al., 2007; Radican et al., 2008; Wang et al., 2009). Full
details of the systematic review, criteria to identify studies for including in the meta-analysis,
and meta-analysis methodology and findings are discussed in Appendices B and C.
The meta-analyses of the overall effect of TCE exposure on lymphoma suggest a small,
robust, and statistically significant increase in NHL risk. The pooled estimate from the primary
random effect meta-analysis (RRp) was 1.23 (95% CI: 1.04, 1.44) (Figure 4-15). This result and
its statistical significance were not influenced by individual studies. The result is similarly not
sensitive to individual risk ratio estimate selections except that substituting the Zhao et al. (2005)
mortality results with the study's incidence results leads to an RRp that is no longer statistically
significant of 1.19(95% CI: 1.00, 1.41).
Meta-analysis of the highest exposure groups, either duration, intensity, or their product,
cumulative exposure, results in an RRp of 1.57 (95% CI: 1.27, 1.94), which is greater than the
RRp from the overall exposure analysis, and provides additional support for an association
between NHL and TCE (Figure 4-16). The highest exposure category groups have a reduced
likelihood for exposure misclassification because they are believed to represent a greater
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-368 DRAFT—DO NOT CITE OR QUOTE
-------
differential TCE exposure compared to people identified with overall TCE exposure.
Observation of greater risk associated with higher exposure category compared to overall
(typically any versus none) exposure comparison additionally suggests an exposure-response
gradient between lymphoma and TCE, although estimation of a level of exposure associated with
the pooled or meta-relative risk is not possible.
Low-to-moderate heterogeneity in RRp is observed across the results of the 16 studies in
the meta-analysis of the overall effect of TCE, but it was not statistically significant (p = 0.10),
and no heterogeneity was observed in the meta-analysis of the highest exposure groups. In the
overall analysis, difference between cohort and case-control studies could explain much of the
observed heterogeneity. In the subgroup analysis, increased risk of lymphoma was strengthened
in analysis limited to cohort studies and virtually eliminated in the case-control study analysis.
Examination of heterogeneity in cohort and case-control studies separately was not statistically
significant in either case although some may be present given that statistical tests of
heterogeneity are generally insensitive in cases of minor heterogeneity. In general, sources of
heterogeneity are uncertain and may reflect several features known to influence epidemiologic
studies. One reason may be differences in exposure assessment and in overall TCE exposure
concentration between cohort and case-control studies. Several cohort studies (Anttila et al.,
1995; Axelson et al., 1994; Blair et al., 1998; Hansen et al., 2001; Raaschou-Nielsen et al., 2003)
adopt exposure assessment approaches that are expected to reduce potential for bias (NRC,
2006). Exposure misclassification bias due to random or measurement error and recall bias a
more likely in three case-control studies (Hardell et al., 1994; Nordstrom et al., 1998; Persson
and Fredrikson, 1999) with self-reported TCE exposure compared to Siemiatycki (1991), Miligi
et al. (2006), Seidler et al. (2007). No heterogeneity was observed in the meta-analysis of the
highest exposure groups, providing some evidence of exposure misclassification as a source of
heterogeneity in the overall analysis. In addition, a low overall TCE exposure prevalence is
anticipated in population case-control studies which would typically assess a large number of
workplaces and operations, where exposures are less well defined, and where case and control
subjects identified as exposed to TCE probably have minimal contact (NRC, 2006). Observed
higher risk ratios with higher exposure categories in NHL case-control studies support exposure
differences as a source of heterogeneity.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-369 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Lymphoma
Study name Statistics for each study
Rate Lower Upper
ratio limit limit p-Value
Ant+iio 1 QQC; -i s-in n ant; ^ K-IQ n na^
Avolcnn 1 QQA 1 t;9n (1 K77 7 K^9 ("I 74Q
Boice1999 1.190 0.705 2.009 0.515
f^roonlonrl 1 QCM n 7fif> f> 97Q 9 A17 f> fiA9
Honcon 9OO1 7 1 0O 1 ^^O fi 1 QQ O OO1
IWIrn-non 1 QQA 1 O1 0 O ^Ofi 1 Q41 n Q7R
Raaschou-Nielsen 2003 1.240 1.011 1.521 0.039
Radican2008 1.360 0.772 2.396 0.287
Zhao 2005 mort 1.437 0.899 2.297 0.130
Morrloll 1QCM 7 9OO 1 9fi7 Aft Q97 n O9fi
Miligi 2006 0.933 0.671 1.298 0.682
Mr»rrlotrr»rv» 1 QQA 1 ^OO O RQ1 7 O^7 O 7O^
Dciroor»nJ? CrorlriL'orn-i 1 QQQ1 OOO O ^4A O ROQ O R4Q
Seidler2007 0.800 0.566 1.131 0.207
Qiomiotw^U 1 QCM 1 1 nn n A7Q O ^9^ n R99
Wang 2008 1.200 0.849 1.697 0.302
1.228 1.044 1.444 0.013
0
1 0
Rate i
2 0
•atio
^^m
-
-
-
1
5 1
and £
•—
i
fe-
2
15% Cl
•
•
i ;
5 1
0
random effects model
Figure 4-15. Meta-analysis of lymphoma and overall TCE exposure. The pooled estimate is in the bottom row
Symbol sizes reflect relative weights of the studies. The horizontal midpoint of the bottom diamond represents the
RRp estimate and the horizontal extremes depict the 95% CI limits.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TCE and Lymphoma - highest exposure groups
Study name
Statistics for each study
Rate ratio and 95% C\
Rate Lower Upper
ratio limit limit p-Value
Anttila1995 1.400 0.350 5.598 0.634
Axelson 1994 6.250 0.880 44.369 0.067
Boice1999 1.620 0.818 3.210 0.167
Hansen 2001 cum exp 2.700 0.871 8.372 0.085
Morgan 1998 0.810 0.101 6.525 0.843
Raaschou-Nielsen 2003 1.600 1.119 2.288 0.010
Radican 2008 mort 1.400 0.705 2.780 0.336
Zhao 2005 mort 1.300 0.522 3.240 0.573
Miligi2006 1.200 0.709 2.028 0.497
Seidler2007 2.300 1.008 5.250 0.048
Siemiatycki 1991 0.800 0.195 3.275 0.756
Wang 2009 2.199 0.898 5.385 0.085
1.569 1.267 1.942 0.000
^^m
f
•—
•
•^
0.1 0.2 0.5 1
5 10
H I
O ^
HH Oq
H ^
O
random effects model; same for fixed
Figure 4-16. Meta-analysis of lymphoma and TCE exposure—highest exposure groups. The pooled estimate is in
the bottom row. Symbol sizes reflect relative weights of the studies. The horizontal midpoint of the bottom diamond
represents the RRp estimate and the horizontal extremes depict the 95% Cl limits.
H
W
-------
Diagnostic inaccuracies are likely another source of heterogeneity in the meta-analysis
through study differences in lymphoma groupings and in lymphoma classification schemes. All
studies include a broad but slightly different group of lymphosarcoma, reticulum-cell sarcoma,
and other lymphoid tissue neoplasms (Codes 200 and 202), except Nordstrom et al. (1998)
whose case-control study examined hairy cell leukemia, now considered a lymphoma. Cohort
studies have some consistency in coding NHL, with NHL defined as lymphosarcoma and
reticulum-cell sarcoma (200) and other lymphoid tissue neoplasms (202) using the ICD,
Revision 7, 200 and 202—four studies (Axelson et al., 1994; Anttila et al., 1995; Hansen et al.,
2001; Raaschou-Nielsen et al., 2003), ICD-Adapted, Revision 8 (Blair et al., 1998), and ICD-7,
8, 9, and 10, per the version in use at the time of death (Morgen et al., 1997, as presented in
Mandel et al., 2006; Boice et al., 1999; Radican et al., 2008), as does the case-control study of
Siemiatycki (1991) whose coding scheme for NHL is consistent with ICD 9, 200 and 202. Case-
control studies, on the other hand, have adopted other classification systems for defining NHL
including the NCI Working Formulation (Miligi et al., 2006), WHO (Seidler et al., 2007),
Rappaport (Hardell et al., 1994), or else do not identify the classification system for defining
NHL (Persson and Fredrikson, 1999).
There is some evidence of potential publication bias in this data set; however, it is
uncertain that this is actually publication bias rather than an association between standard error
and effect size resulting for some other reason, e.g., a difference in study populations or
protocols in the smaller studies. Furthermore, if there is publication bias in this data set, it does
not appear to account completely for the finding of an increased lymphoma risk.
NRC (2006) deliberations on trichloroethylene commented on two prominent evaluations
of the then-current epidemiologic literature using meta-analysis techniques. These studies were
by Wartenberg et al. (2000), and by Kelsh et al. (2005), submitted by Exponent-Health Sciences
to NRC during their deliberations and subsequently published in a paper on NHL (Mandel et al.,
2006) and a paper on multiple myeloma and leukemia (Alexander et al., 2006). The NRC found
weaknesses in the techniques used in each of these studies, and suggested that U.S. EPA conduct
a new meta-analysis of the epidemiologic data on trichloroethylene using objective and
transparent criteria so as to improve on the past analyses. U.S. EPA staff conducted their
analysis according to NRC (2006) suggestions for transparency, systematic review criteria, and
examination of both cohort and case-control studies. The U.S. EPA analysis of NHL analysis
considered a larger number of studies than in the previous analyses (Mandel et al., 2006;
Wartenberg et al., 2000), and includes recently published studies (Boice et al., 2006; Miligi et al.,
2006; Seidler et al., 2007; Zhao et al., 2005). Despite the weaknesses in Wartenberg et al.
(2000), Kelsh (2005) and Mandel et al. (2006), pooled NHL risk for overall TCE exposure in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-372 DRAFT—DO NOT CITE OR QUOTE
-------
these analyses is of a similar magnitude as that observed in U.S. EPA's updated analysis
(1.5, 95% CI: 0.9, 2.3, Tier 1 incidence; 1.2, 95% CI: 0.9, 1.7, Tier 1 mortality [Wartenberg et
al., 2000]; 1.59, 95% CI: 1.21, 2.08, Group I, TCE Subcohorts; 1.39, 95% CI: 0.62, 3.10, case-
control studies [Kelsh, 2005; Mandel et al, 2006]).
4.6.2. Animal Studies
The immunosuppressive and immunomodulating potential of TCE has not been fully
evaluated in animal models across various exposure routes, over various relevant durations of
exposure, across representative life stages, and/or across a wide variety of endpoints.
Nevertheless, the studies that have been conducted indicate a potential for TCE-induced
immunotoxicity, both following exposures in adult animals and during immune system
development (i.e., in utero and preweaning exposures).
4.6.2.1. Immunosuppression
A number of animal studies have indicated that moderate to high concentrations of TCE
over long periods have the potential to result in immunosuppression in animal models, dependant
on species and gender. These studies are described in detail below and summarized in
Table 4-65.
4.6.2.1.1. Inhalation exposures. Mature cross-bred dogs (5/group) were exposed to 0-, 200-,
500-, 700-, 1,000-, 1,500-, or 2,000-ppm TCE for 1-hour or to 700 ppm TCE for 4 hours, by
tracheal intubation under intravenous sodium pentobarbital anesthesia. An additional group of
dogs was exposed by venous injection of 50 mg/kg TCE administered at a rate of 1 mL/minute
(Hobara et al., 1984). Blood was sampled pre- and postexposure for erythrocyte and leukocyte
counts. Marked, transient decreases in leukocyte counts were observed at all exposure levels
30 minutes after initiation of exposure. At the end of the exposure period, all types of leukocytes
were decreased (by 85%); neutrophils were decreased 33%, and lymphocytes were increased
40%. There were no treatment-related changes in erythrocyte counts, hematocrit values, or
thrombocyte counts.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-373 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
-§
I
o §
H I
O >
HH Oq
H ^
O
Table 4-65. Summary of TCE immunosuppression studies
Exposure route/vehicle,
duration, dose
NOAEL; LOAEL3
Results
Reference, species/strain
sex/number
Inhalation Exposure Studies
Single 1-h exposure to all dose
groups; plus single 4-h exposure
at700ppmb
0, 200, 500, 700, 1,000, 1,500, or
2,000 ppm
Single 3-h exposure. Also, 3 h/d
on 5 d at lowest dose
0,2.6, 5.2, 10.6, 25.6, or 48 ppm
Single 3-h exposure,
50-200 ppnf
4-wk, 6 h/d, 5 d/wk
0, 100, 300, or 1,000 ppm
LOAEL: 200 ppm
NOAEL: 2.6 ppm
LOAEL: 5.2 ppm
NOAEL: 300 ppm
LOAEL: 1,000 ppm
Marked transient -I leukocyte counts at all exposure levels
30 min after initiating exposure. At end of exposure, 85% -i-
leukocyte counts (33% -I neutrophils, 40% -I lymphocytes).
Challenged with Streptococcus zooepidemicus to assess
susceptibility to infection and Klebsiella pneumoniae to
assess bacterial clearance. For single exposure: dose-related
sig. t mortality at >5.2 ppm over 14 d. Sig. J, in bactericidal
activity at 10.6 ppm.
Challenged with Streptococcus zooepidemicus. Dose-related
t mortality, bacterial antiphagocytic capsule formation, and
bacterial survival. Dose-related impairment of alveolar
macrophages; increased neutrophils in bronchoalveolar fluid
at 3 d postinfection.
At 1,000 ppm, 64% -i- plaque-forming cell assay response.
Hobaraetal., 1984
Dog, cross-bred, both sexes,
5/group
Aranyietal., 1986
Mouse, CD-I females, 4-5 wk old,
approx. 30 mice/group, 5-10
replications; for pulmonary
bactericidal activity assay, 17-24
mice/group
Park et al., 1993 (abstract)
Mouse, CD-I, (sex and #/group not
specified)
Woolhiseretal, 2006
Rat, Sprague-Dawley, female,
16/group
Oral Exposure Studies
Gavage in 10% emulphor, 14 d,
daily, 0, 24, or 240 mg/kg/d
Drinking water with 1%
emulphor, 4-6 months
0,0.1, 1.0, 2.5, or5.0mg/mL
Gavage, 14 d, 0, 14.4, or 144
mg/kg/d chloral hydrate
LOAEL: 24 mg/kg/d
LOAEL: 0.1 mg/kg/d
NOAEL: 144 mg/kg/d
Sig. -i- cell-mediated immune response to SRBC at both dose
levels.
In females, humoral immunity -I at 2.5 and 5 mg/mL TCE,
whereas cell-mediated immunity -i- and bone marrow stem
cell colonization -I at all four concentrations. The males
were relatively unaffected after both 4 and 6 months.
No treatment-related effects.
Sanders et al., 1982
Mouse, CD-I, male, 9-12/group
Sanders et al., 1982
Mouse, CD-I, male and female,
7-25/group
Kauffmann et al., 1982
Mouse, CD-I, male, 12/group
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table 4-65. Summary of TCE immunosuppression studies (continued)
Exposure route/vehicle,
duration, dose
NOAEL; LOAEL3
Results
Reference, species/strain
sex/number
Drinking water, 90 d, 0, 0.07, or
0.7 mg/mL chloral hydrate. (M:
0, 16, or 160 mg/kg/d; F: 0, 18, or
173 mg/kg/d)
NOAEL: 0.07 mg/mL
LOAEL: 0.7 mg/mL
Sig. -I cell-mediated immune response (plasma
hemagglutination liters and spleen antibody-producing cells
of mice sensitized to SRBC) in females at 0.7 mg/mL.
Kauffmann et al., 1982
Mouse, CD-I, male and female,
15-20/group
Drinking water, From mating to
PND 21 orPND 56, (emulphor
cone, not provided)
0 (emulphor), 1, or 10 ppm
LOAEL: 1 ppm
At 10 ppm, I body weight and length at PND 21. IgM
antibody response to SRBC challenge suppressed in both (51
and ? pups at 10 ppm, and (51 pups at 1 ppm, J, in splenic
CD4+CD8-T-cells. At 56 PND, striking ^ in natural killer
cell activity seen at both doses.
Adams et al., 2003 (abstract)
Mouse, B6C3F1, both sexes,
numbers of pups not stated
i
o a.
Drinking water, from GD 0 to 3
orSwksofage, 0, 1,400, or
14,000 ppb
LOAEL: 1,400 ppb
Suppressed PFC responses in both sexes and ages at 14,000
ppb, in males at both ages at 1,400 ppb, and in females at 8
wks at 1,400 ppb. Numbers of spleen B220+ cells ^ at
3-wks at 14,000 ppb. Pronounced t thymus T-cell
populations at 8 wks.
Peden-Adams et al., 2006
Mouse, B6C3F1, dams and both
sexes offspring, 5 litters/group;
5-7 pups/group at 3 wks;
4-5 pups/sex/group at 8 wks
Drinking water, from GD 0 to
7-8 wks of age; 0, 0.5, or
2.5 mg/mL
LOAEL: 0.5 mg/mL
At 0.5 mg/mL: Sig J, postweaning weight; sig.f IFNy
produced by splenic CD4+ cells at 5-6 wks; sig J, splenic
CD8+ and B220+ lymphocytes; sig.f IgG2a and histone; sig.
altered CD4-/CD8- and CD4+/CD8+ thymocyte profile
At 2.5 mg/mL: Sig J, postweaning weight; sig.f IFNy
produced by splenic CD4+ and CD8+ cells at 4-5 and 5-6
wks; sig I splenic CD4+, CD8+, and B220+ lymphocytes;
sig. altered CD4+/CD8+ thymocyte profile.
Blossom and Doss, 2007
Mouse, MRL +/+, dams and both
sexes offspring, 3 litters/group;
8-12 pups/group
H I
O >
HH Oq
H ^
O
H
W
Drinking water, from GD 0 to
PND 42; 0 or 0.1 mg/mL;
maternal dose = 25.7 mg/kg/d;
offspring PND 24-42 dose =
31.0 mg/kg/d
LOAEL: 0.1 mg/mL
At 0.1 mg/mL: at PND 20, sig. t thymocyte cellularity and
distribution, associated with sig. t in thymocyte subset
distribution; sig. t reactive oxygen species generation in
total thymocytes; sig. t in splenic CD4+ T-cell production of
IFN-y and IL-2 in females and TNF-a in males at PND 42.
Blossom et al., 2008
Mouse, MRL +/+, dams and both
sexes offspring, 8 litters/group; 3-8
pups/group
Drinking water, from GD 0 to 12
months of age; 0 (1% emulphor),
1,400, or 14,000 ppb
LOAEL: 1,400 ppb
At 1,400 ppb: splenic CD4-/CD8- cells sig.f in females;
thymic CD4+/CD8+ cells sig. i in males; 18% ^ in male
kidney weight.
At 14,000 ppb: thymic T-cell subpopulations (CD8+,
CD4/CD8-, CD4+) sig. ^ in males.
Peden-Adams et al., 2008 (in press)
Mouse, MRL +/+, dams and both
sexes offspring, unknown #
litters/group, 6-10
off spring/sex/group
-------
Table 4-65. Summary of TCE immunosuppression studies (continued)
Exposure route/vehicle,
duration, dose
NOAEL; LOAEL3
Results
Reference, species/strain
sex/number
Intraperitoneal Injection Exposure Studies
3 d, single daily injection, 0, 0.05,
0.5, or 5 mmol/kg/day
3 d, single daily injection, 0 or 10
mmol/kg/day
NOAEL: 0.05
mmol/kg/day
LOAEL: 0.5
mmol/kg/day
LOAEL: 10
mmol/kg/day
•I natural killer cell activity at 0.5 and 5 mmol/kg/day. -i-
splenocyte counts at 5 mmol/kg/day.
•I natural killer cell activity and -I spleen weights at 10
mmol/kg/day.
Wright etal., 1991
Rat, Sprague-Dawley
Wright etal., 1991
Mouse, B6C3F1
aNOAEL and LOAEL are based upon reported study findings.
blnhalation, tracheal intubation under anesthesia.
°Exact dose levels not specified.
1,1 = decreased, increased; sig. = statistically significant.
-------
1 In a study that examined the effects of a series of inhaled organic chemical air
2 contaminants on murine lung host defenses, Aranyi et al. exposed female CD-I mice to single
3 3-hour exposures of TCE at time-weighted concentrations of 0, 2.6, 5.2, 10.6, 25.6, or 48 ppm
4 (Aranyi et al., 1986). Additionally, at the dose at which no adverse treatment-related effect
5 occurred with a single exposure (i.e., 2.6 ppm), a multiple exposure test (5 days, 3 hours/day)
6 was conducted. Susceptibility to Streptococcus zooepidemicus aerosol infection and pulmonary
7 bactericidal activity to inhaled Klebsiella pneumoniae were evaluated. There was a significant
8 (p < 0.0001) treatment by concentration interaction for mortality, with the magnitude of the
9 effect increasing with concentration. A significant (p < 0.0001) treatment by concentration
10 interaction was also found for bactericidal activity. Single 3-hour exposures at 10.6, 25.6, and
11 48 ppm resulted in significant increases in mortality, although increases observed after single
12 exposures at 5.2 or 2.6 ppm or five exposures at 2.6 ppm were not significant. Pulmonary
13 bactericidal activity was significantly decreased after a single exposure at 10.6 ppm, but single
14 exposures to 2.6 or 5.2 ppm resulted in significant increases.
15 In a host-resistance assay, CD-I mice (sex and number/group not specified) exposed to
16 TCE by inhalation for 3 hours at 50-200 ppm were found to be more susceptible to increased
17 infection following challenge with Streptococcus zooepidemicus administered via aerosol
18 (Parketal., 1993). Dose-related increases in mortality, bacterial antiphagocytic capsule
19 formation, and bacterial survival were observed. Alveolar macrophage phagocytosis was
20 impaired in a dose-responsive manner, and an increase in neutrophils in bronchoalveolar lavage
21 fluid was observed in exposed mice 3 days post infection.
22 A guideline (OPPTS 870.3800) 4-week inhalation immunotoxicity study was conducted
23 in female Sprague-Dawley rats (Woolhiser et al., 2006). The animals (16/group) were exposed
24 to TCE at nominal levels of 0, 100, 300, or 1,000 ppm for 6 hours/day, 5 days/week. Effects on
25 the immune system were assessed using an antigen response assay, relevant organs weights,
26 histopathology of immune organs, and hematology parameters. Four days prior to study
27 termination, the rats were immunized with sheep red blood cells (SRBC), and within 24 hours
28 following the last exposure to TCE, a plaque forming cell assay was conducted to determine
29 effects on splenic anti-SRBC IgM response. Minor, transient effects on body weight and food
30 consumption were noted in treated rats for the first 2 weeks of exposure. Mean relative liver and
31 kidney weights were significantly (p = 0.05) increased at 1,000 ppm as compared to control,
32 while lung, spleen, and thymus weights were similar to control. No treatment-related effects
33 were observed for hematology, WBC differential counts, or histopathological evaluations
34 (including spleen, thymus, and lung-associated lymph nodes). At 1,000 ppm, rats demonstrated
35 a 64% decrease in plaque forming cell assay response. Lactate dehydrogenase, total protein
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-377 DRAFT—DO NOT CITE OR QUOTE
-------
1 levels, and cellular differentiation counts evaluated from bronchoalveolar lavage (BAL) samples
2 were similar between control and treated groups. A phagocytic assay using BAL cells showed
3 no alteration in phagocytosis, although these data were not considered fully reliable since (1) the
4 number of retrieved macrophage cells was lower than expected and pooling of samples was
5 conducted and (2) samples appear to have been collected at 24 hours after the last exposure
6 (rather than within approximately 2 hours of the last exposure), thereby allowing for possible
7 macrophage recovery. The NOAEL for this study was considered by the study authors to be
8 300 ppm, and the LOAEL was 1,000 ppm; however, the effect level may have actually been
9 lower. It is noted that the outcome of this study does not agree with the studies by Aranyi et al.
10 (1986) and Park et al. (1993), both of which identified impairment of macrophage phagocytic
11 activity in BAL following inhalation TCE exposures.
12
13 4.6.2.1.2. Oral exposures. In a study by Sanders et al., TCE was administered to male and
14 female CD-I mice for 4 or 6 months in drinking water at concentrations of 0, 0.1, 1, 2.5, or
15 5 mg/mL (Sanders et al., 1982). In females, humoral immunity was suppressed at 2.5 and
16 5 mg/mL, while cell-mediated immunity and bone marrow stem cell activity were inhibited at all
17 dose levels. Male mice were relatively unaffected either at 4 or 6 months, even though a
18 preliminary study in male CD-I mice (exposed to TCE for 14 days by gavage at 0, 24, or
19 240 mg/kg/d) had demonstrated a decrease in cell-mediated immune response to SRBC in male
20 mice at both treatment levels.
21 A significant decrease in humoral immunity (as measured by plasma hemagglutination
22 liters and the number of spleen antibody producing cells of mice sensitized to sheep
23 erythrocytes) was observed by Kaufmann et al. (1982) in female CD-I mice (15-20/group)
24 following a 90-day drinking water exposure to 0, 0.07, or 0.7 mg/mL (equivalent to 0, 18, or
25 173 mg/kg) chloral hydrate, a metabolite of TCE. Similar responses were not observed in male
26 CD-I mice exposed for 90 days in drinking water (at doses of 0, 16, or 160 mg/kg/d), or when
27 administered chloral hydrate by gavage to 12/group for 14 days at 14.4 or 144 mg/kg/d.
28 The potential for developmental immunotoxicity was assessed in B6C3F1 mice
29 administered TCE in drinking water at dose levels of 0, 1,400 or 14,000 ppb from gestation day
30 (GD) 0 to either 3 or 8 weeks of age (Adams et al., 2003 [preliminary data]; Peden-Adams et al.,
31 2006). At 3 and 8 weeks of age, offspring lymphocyte proliferation, NK cell activity, SRBC-
32 specific IgM production (PFC response), splenic B220+ cells, and thymus and spleen T-cell
33 immunophenotypes were assessed. Delayed-typed hypersensitivity and autoantibodies to
34 ds-DNA were evaluated in offspring at 8 weeks of age. Observed positive responses consisted of
35 suppressed PFC responses in males at both ages and both TCE treatment levels, and in females at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-378 DRAFT—DO NOT CITE OR QUOTE
-------
1 both ages at 14,000 ppb and at 8 weeks of age at 1,400 ppb. Spleen numbers of B220+ cells
2 were decreased in 3-week old pups at 14,000 ppb. Pronounced increases in all thymus T-cell
3 subpopulations (CD4+, CD8+, CD4+/CD8+, and CD4-/CD8-) were observed at 8 weeks of age.
4 Delayed hypersensitivity response was increased in 8-week old females at both treatment levels
5 and in males at 14,000 ppb only. No treatment-related increase in serum anti-ds-DNA antibody
6 levels was found in the offspring at 8 weeks of age.
7 In a study designed to examine potential susceptibility of the young (Blossom and Doss,
8 2007), TCE was administered to groups of pregnant MRL +/+ mice in drinking water at
9 occupationally-relevant levels of 0, 0.5, or 2.5 mg/mL. A total of 3 litters per treatment group
10 were maintained following delivery (i.e., a total of 11 pups at 0 mg/mL TCE, 8 pups at
11 0.5 mg/mL TCE, and 12 pups at 2.5 mg/mL TCE), and TCE was continuously administered to
12 the offspring until young adulthood (i.e., 7-8 weeks of age). Although there were no effects on
13 reproduction, offspring postweaning body weights were significantly decreased in both treated
14 groups. Additionally, TCE exposure was found to modulate the immune system following
15 developmental and early life exposures. Decreased spleen cellularity and reduced numbers of
16 CD4+, CD8+, and B220+ lymphocyte subpopulations were observed in the postweaning
17 offspring. Thymocyte development was altered by TCE exposures, as evidenced by significant
18 alterations in the proportions of double-negative subpopulations and inhibition of in vitro
19 apoptosis in immature thymocytes. TCE was also shown to induce a dose-dependent increase in
20 CD4+ and CD8+ T-lymphocyte IFNy in peripheral blood by 4-5 weeks of age, although these
21 effects were no longer observed at 7-8 weeks of age. Serum anti-histone autoantibodies and
22 total IgG2a were significantly increased in treated offspring; however, no histopathological signs
23 of autoimmunity were observed in the liver and kidneys at sacrifice.
24 This increase in T-cell hyperactivity was further explored in a study by Blossom et al.
25 (2008). In this study, MRL +/+ mice were treated in the drinking water with 0 or 0.1 mg/mL
26 TCE. Based on drinking water consumption data, average maternal doses of TCE were
27 25.7 mg/kg/d, and average offspring (PND 24-42) doses of TCE were 31.0 mg/kg/d. Treatment
28 was initiated at the time of mating, and continued in the females (8/group) throughout gestation
29 and lactation. Pups were weaned at PND 24, and the offspring were continued on drinking water
30 treatment in a group-housed environment until study termination (PND 42). Subsets of offspring
31 were sacrificed at PND 10 and 20, at which time developmental and functional endpoints in the
32 thymus were evaluated (i.e., total cellularity, CD4+/CD8+ ratios, CD24 differentiation markers,
33 and double-negative subpopulation counts). Indicators of oxidative stress were measured in the
34 thymus at PND 10 and 20, and in the brain at PND 42. Mitogen-induced intracellular cytokine
35 production by splenic CD4+ and CD8+ T-cells was evaluated in juvenile mice and brain tissue
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-379 DRAFT—DO NOT CITE OR QUOTE
-------
1 was examined at PND 42 for evidence of inflammation. Behavioral testing was also conducted;
2 these methods and results are described in Section 4.3. TCE treatment did not affect
3 reproductive capacity, parturition, or ability of dams to maintain litters. The mean body weight
4 of offspring was not different between the control and treated groups. Evaluation of the thymus
5 identified a significant treatment-related increase in cellularity, accompanied by alterations in
6 thymocyte subset distribution, at PND 20 (sexes combined). TCE treatment also appeared to
7 promote T-cell differentiation and maturation at PND 42, and ex vivo evaluation of cultured
8 thymocytes indicated increased reactive oxygen species (ROS) generation. Evaluation of
9 peripheral blood indicated that splenic CD4+ T-cells from TCE-exposed PND 42 mice produced
10 significantly greater levels of IFN-y and IL-2 in males and TNF-a in both sexes. There was no
11 effect on cytokine production on PND 10 or 20. The dose of TCE that resulted in adverse
12 offspring outcomes in this study (i.e., 0.1 mg/mL, equivalent to 25.7-31.0 mg/kg/d) is
13 comparable to that which has been previously demonstrated to result in immune system
14 alterations and autoimmunity in adult MRL +/+ mice (i.e., 0.1 mg/mL, equivalent to 21 mg/kg/d;
15 Griffin et al., 2000b).
16 Another study that examined the effects of developmental exposure to TCE on the
17 MRL+/+ mouse was conducted by Peden-Adams et al. (2008). In this study, MRL/MpJ (i.e.,
18 MRL +/+) mice (unspecified number of dams/group) were exposed to TCE (solubilized with 1%
19 emulphor) in drinking water at levels of 0, 1,400, or 14,000 ppb from GD 0 and continuing until
20 the offspring were 12 months of age. TCE concentrations in the drinking water were reported to
21 be analytically confirmed. Endpoints evaluated in offspring at 12 months of age included final
22 body weight; spleen, thymus, and kidney weights; spleen and thymus lymphocyte
23 immunophenotyping (CD4 or CDS); splenic B-cell counts; mitogen-induced splenic lymphocyte
24 proliferation; serum levels of autoantibodies to dsDNA and glomerular antigen (GA),
25 periodically measured from 4 to 12 months of age; and urinary protein measures. Reported
26 sample sizes for the offspring measurements varied from 6 to 10 per sex per group; the number
27 of source litters represented within each sample was not specified. The only organ weight
28 alteration was an 18% increase in kidney weight in the 1,400 ppb males. Splenic CD4-/CD8-
29 cells were altered in female mice (but not males) at 1,400 ppm only. Splenic T-cell populations,
30 numbers of B220+ cells, and lymphocyte proliferation were not affected by treatment.
31 Populations of thymic T-cell subpopulations (CD8+, CD4-/CD8-, and CD4+) were significantly
32 decreased in male but not female mice following exposure to 14,000-ppb TCE, and CD4+/CD8+
33 cells were significantly reduced in males by treatment with both TCE concentrations.
34 Autoantibody levels (anti-dsDNA and anti-GA) were not increased in the offspring over the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-3 80 DRAFT—DO NOT CITE OR QUOTE
-------
1 course of the study, indicating that TCE did not contribute to the development of autoimmune
2 disease markers following developmental exposures that continued into adult life.
3 Overall, the studies by Peden-Adams et al. (2006, 2008 in press), Blossom and Doss
4 (2007), and Blossom et al. (2008), which examined various immunotoxicity endpoints following
5 exposures that spanned the critical periods of immune system development in the rodent, were
6 generally not designed to assess issues such as posttreatment recovery, latent outcomes, or
7 differences in severity of response that might be attributed to the early life exposures.
8
9 4.6.2.1.3. Intraperitoneal administration. Wright et al. reported that following 3 days of
10 single intraperitoneal injections of TCE in Sprague-Dawley rats at 0, 0.05, 0.5, or 5 mmol/kg/day
11 and B6C3F1 mice at 0 or 10 mmol/kg/day, NK cell activity was depressed in the rats at the mid-
12 and high-dose levels, and in the mice at the high dose level (Wright et al., 1991). Also at the
13 highest dose levels tested, decreased splenocyte counts and relative spleen weight were observed
14 in the rats and mice, respectively. In vitro assays demonstrated treatment-related decreases in
15 splenocyte viability, inhibition of lipopolysaccharide-stimulated lymphocyte mitogenesis, and
16 inhibited NK cell activity suggesting the possibility that compromised immune function may
17 play a role in carcinogenic responses of experimental animals treated with TCE.
18
19 4.6.2.2. Hypersensitivity
20 Evidence of a treatment-related increase in delayed hypersensitivity response has been
21 observed in guinea pigs following dermal exposures with TCE and in mice following exposures
22 that occurred both during development and postnatally (see Table 4-66).
23 In a modified guinea pig maximization test, Tang et al. evaluated the contact allergenicity
24 potential of TCE and three metabolites (trichloroacetic acid, trichloroethanol, and chloral
25 hydrate) in 4 animals (FMMU strain, sex not specified) per group (Tang et al., 2002). Edema
26 and erythema indicative of skin sensitization (and confirmed by histopathology) were observed.
27 Sensitization rates were reported to be 71.4% for TCE and 58.3% for trichloroacetic acid, as
28 compared to a reference positive control response rate (i.e., 100% for 2,4-dinitrochlorobenzene).
29 In this study, the mean response scores for TCE, trichloroacetic acid, and
30 2,4-dinitrochlorobenzene were 2.3, 1.1, and 6.0, respectively. TCE was judged to be a strong
31 allergen and TCA was a moderate allergen, according to the criteria of Magnusson and Kligman
32 (Magnusson and Kligman, 1969). Trichloroethanol and chloral hydrate were not found to elicit a
33 dermal hypersensitivity response.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-3 81 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-66. Summary of TCE hypersensitivity studies
to
vo £3'
I
I
§
***.
£3'
1
TO'
I
Exposure route/vehicle, duration,
dose
NOAEL; LOAEL8
Results
Reference, species/strain
sex/number
Induction by single intradermal
injection, then challenge by dermal
application at 21 d
0 or 0.1 mL induction; 0 or 0.2 mL
challenge
TCE, TCA, TCOH, and chloral
hydrate
Edema and erythema (confirmed by
histopathology) indicative of skin
sensitization for TCE (strong sensitizer) and
TCA (moderate sensitizer)
Tang et al, 2002
Guinea pig, FMMU strain, sex
not specified, 4/group
Intradermal injection, 0, 167, 500,
1,500, or 4,500 mg/kg
Dermal patch, 0 or 900 mg/kg
Hypersensitivity: total dose from
induction through challenge <340
mg/kg
Intradermal NOAEL:
500 mg/kg
Intradermal LOAEL:
1,500 mg/kg
Dermal patch NOAEL:
900 mg/kg
Intradermal injection: At 1,500 mg/kg: Sig.
t AST; at 4,500 mg/kg, sig. ^ ALT and
AST, sig. I total protein and globulin; fatty
degeneration of liver
Dermal patch: no effects of treatment
Hypersensitivity: sensitization rate of 66%
(strong sensitizer), with edema and
erythema; sig. 1 ALT, AST, and lactate
dehydrogenase; sig. 1 relative liver weight;
sig. I albumin, IgA, and GOT; hepatic
lesions (ballooning changes)
Tang et al., 2008
Guinea pig, FMMU strain,
female, 5-6/group for
intradermal/dermal patch study,
10/group for hypersensitivity
study, female
H I
O ^
HH Oq
H ^
O
Drinking water, from GD 0 to 8
wks of age
0, 1,400, or 14,000 ppb
LOAEL: 1,400 ppb
Sig. t swelling of foot pad in females at
1,400 and in both sexes at 14,000 ppb
Peden-Adams et al., 2006
Mouse, B6C3F1, both sexes, 5
litters/group; 4-5 pups/sex/group
at 8 wksb
"NOAEL and LOAEL are based upon reported study findings.
bSubset of immunosuppression study.
1,1 = decreased, increased, sig. = statistically significant.
H
W
-------
Immune-mediated hepatitis associated with dermal hypersensitivity reactions in the
guinea pig following TCE exposures was characterized by Tang et al. (2008). In this study,
FMMU strain female guinea pigs (5-6/group) were treated with intradermal injection of 0, 167,
500, 1,500, or 4,500 mg/kg TCE or with a dermal patch containing 0 or 900 mg/kg TCE and
sacrificed at 48 hours posttreatment. At the intradermal dose of 1,500 mg/kg, a significant
increase (p < 0.05) in serum AST level was observed. At 4,500 mg/kg, significantly (p < 0.01)
increased ALT and AST levels were reported, and total protein and globulin decreased
significantly (p < 0.05). Histopathological examination of the liver revealed fatty degeneration,
hepatic sinusoid dilation, and inflammatory cell infiltration. No changes were observed at the
intradermal doses of 500 mg/kg or below, or the dermal patch dose of 900 mg/kg. A Guinea Pig
Maximization Test was also conducted according to the procedures of Magnusson and Kligman
on 10 FMMU females/group, in which the total TCE dosage from induction through challenge
phases was below 340 mg/kg. TCE treatment resulted in dermal erythema and edema, and the
sensitization rate was 66% (i.e., classified as a strong sensitizer). Significant increases (p < 0.05)
in ALT, AST, lactate dehydrogenase, and relative liver weight, and significant decreases
(p < 0.05) in albumin, IgA, and y-glutamyl transpeptidase (GGT) were observed. Additionally,
hepatic lesions (diffuse ballooning changes without lymphocyte infiltration and necrotic
hepatocytes) were noted. It was concluded that TCE exposure to guinea pigs resulted in delayed
type hypersensitivity reactions with hepatic injury that was similar to occupational
medicamentosa-like dermatitis disorders observed in human occupational studies.
Also, as indicated in Section 4.6.2.1.2 above, in a developmental immunotoxicity-type
study in B6C3F1 mice, administration of TCE in drinking water at dose levels of 0, 1,400, or
14,000 ppb from gestation Day 0 through to 8 weeks of age resulted in an increased delayed
hypersensitivity response in 8-week old female offspring at both treatment levels and in males at
the high dose of 14,000 ppb (Peden-Adams et al., 2006).
In an in vitro study that evaluated a number of chlorinated organic solvents, nonpurified
rat peritoneal mast cells (NPMC) and rat basophilic leukemia (RBL-2H3) cells were sensitized
with anti-dinitrophenol (DNP) monoclonal IgE antibody and then stimulated with DNP-
conjugated bovine serum albumin plus TCE (Seo et al., 2008). TCE enhanced antigen-induced
histamine release from NPMC and RBL-2H3 cells in a dose-related manner, and increased IL-4
and TNF-a production from the RBL-2H3 cells. In an in vivo study, i.p.-injected TCE was found
to markedly enhance passive cutaneous anaphylaxis reaction in antigen-challenged rats. These
results suggest that TCE increases histamine release and inflammatory mediator production from
antigen-stimulated mast cells via the modulation of immune responses; TCE exposure may lead
to the enhancement of allergic disease through this response.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-3 83 DRAFT—DO NOT CITE OR QUOTE
-------
4.6.2.3. Autoimmunity
A number of studies have been conducted to examine the effects of TCE exposure in
mouse strains (i.e., MRL +/+, MRL -Ipr, or NZB x NZW) which are all known to be genetically
susceptible to autoimmune disease. The studies have demonstrated the potential for TCE to
induce autoimmune disease (as demonstrated in Table 4-67 which summarizes those studies
which assessed serology, ex vivo assays of cultured splenocytes, and/or clinical or
histopathology). These and other studies conducted in susceptible mouse strains have proven to
be useful tools in exploring various aspects of the mode of action for this response.
Khan et al. used the MRL +/+ mouse model to evaluate the potential for TCE and one of
its metabolites, dichloroacetyl chloride (DCAC) to elicit an autoimmune response (Khan et al.,
1995). Female mice (4-5/group) were dosed by intraperitoneal injection with 10 mmol/kg TCE
or 0.2 mmol/kg DCAC every 4th day for 6 weeks and then sacrificed. Spleen weights and IgG
were increased. ANA and anti-ssDNA antibodies were detected in the serum of TCE- and
DCAC-treated mice; anti-cardiolipin antibodies were detected in the serum of DCAC-treated
mice. A greater magnitude of response observed with DCAC treatment suggested that the
metabolite may be important to the mechanism of TCE-induced autoimmunity.
Other studies in female MRL +/+ mice (8/group) examined exposure via drinking water.
In one of these studies, mice were treated with 2.5 or 5.0 mg/mL (455 or 734 mg/kg/d) TCE in
drinking water for up to 22 weeks (Gilbert et al., 1999; Griffin et al., 2000a). Serial sacrifices
were conducted at Weeks 4, 8, and 22. Significant increases in ANA and total serum
immunoglobulin were found at 4 weeks of TCE treatment (indicating an autoimmune response),
but not at 32 weeks. Increased expression of the activation marker C44 on splenic CD4+ cells
was observed at 32 weeks. In addition, at 4 and 32 weeks, splenic T-cells from treated mice
secreted more LFN-y than control T-cells (significant at 0.5 and 2.5 mg/mL), consistent with a
Thl immune or inflammatory response. By 22 weeks of TCE treatment, a specific immune
serum antibody response directed against dichloroacetylated proteins was activated in hepatic
tissues, indicating the presence of protein adducts. There was a slight but statistically significant
increase in serum alanine aminotransferase levels at 32 weeks at 0.5 mg/mL. Histopathological
evaluation at 32 weeks revealed extensive hepatic lymphocytic cell infiltration at 0.5 and
2.5 mg/mL; all treated groups contained significantly more hepatocyte reactive changes (i.e.,
presence of multinucleated hepatocytes, variations in hepatocyte morphology, and hepatocytes in
mitosis) than controls.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-3 84 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
Table 4-67. Summary of autoimmune-related studies of TCE and metabolites in mice and rats (by sex, strain,
and route of exposure)8
Nunber/group, vehicle, dose,
duration
NOAEL;
LOAELb
Results
Serology
Ex vivo assays of cultured
splenocytes
Clinical and
histopathology
Reference
Autoimmune-prone: Female MRL +/+ Mice, Drinking Water
8 per group, 0, 2.5, or
5 mg/mL TCE (average 0,
455, or 734 mg/kg/d), 4, 8, or
22wks
8 per group, 0,0.1,0.5, or 2.5
mg/mL TCE (0,21, 100, or
400 mg/kg/d), 4 or 32 wks
6-8 per group, 0,0.1, or 0.9
mg/mL trichloroacetaldehyde
hydrate (0, 24, or 220
mg/kg/d) or trichloroacetic
acid (0, 27, or 205 mg/kg/d),
4 wks
8 per group, 0, 0.1, 0.3, or
0.9 mg/mL
trichloroacetaldehyde hydrate
( 0, 13, 46, or 143 mg/kg/d),
40 wks
LOAEL:
2.5 mg/mL
LOAEL:
0. 1 mg/mL
LOAEL:
0. 1 mg/mL
LOAEL:
0.9 mg/mL
Increased ANA at 4 and
8 wks, no difference
between groups at 22 wks
Increased ANA in all
treated groups at 4 wks, but
not at 32 wks
Increased ANA and anti-
histone antibodies at
0.9 mg/mL
trichloroacetaldehyde
hydrate"
Slightly suppressed anti-
ssDNA, anti-dsDNA, and
anti-histone antibody
expression; differences not
statistically significant
Increased activated CD4+ T-
cells and IFN-y secretion
across doses at 4 wks, these
effects were reversed at 22
wks; decreased IL-4 secretion
(4 and 22 wks)
Increased activated CD4+ T-
cells (32 wks), IFN-y
secretion (4 and 32 wks), no
effect on IL-4 secretion
Increased activated CD4+ T-
cellsatO.l and0.9g/mL
doses of both metabolites.
At 0.9 mg/mL, increased
IFN-y secretion, no effect on
IL-4 secretion
Increased activated CD4+ T-
cells and increased INF-y
secretion, no effect on IL-4
secretion
No evidence of liver or renal
damage, based on serum
alanine aminotransferase,
sorbitol dehydrogenase, and
blood urea nitrogen.
Extensive hepatic
mononuclear cellular infiltrate
in 0.5 and 2.5 mg/mL groups,
and hepatocyte reactive
changes in all treated groups
at 32 wks.
No evidence of liver of kidney
damage, based on serum
alanine aminotransferase,
liver and kidney histology..
Diffuse alopecia, skin
inflammation and ulceration,
mononuclear cell infiltration,
mast cell hyperplasia, dermal
fibrosis. Statistically
significant increase at 0.9
mg/mL dose group, but also
increased at lower doses. No
liver or kidney histopathology
effects seen.
Griffin et al.
(2000a)
Griffin et al.
(2000b)
Blossom et al.
(2004)
Blossom et al.
(2007)
H I
O >
HH Oq
H ^
O
H
W
-------
Table 4-67. Summary of autoimmune-related studies of TCE and metabolites (by sex, strain, and route of
exposure) (continued)
Number/group, vehicle,
dose, duration
NOAEL;
LOAELb
Results
Serology
Ex vivo assays of cultured
splenocytes
Clinical and
histopathology
Reference
5 per group, 0 or 0.5 mg/mL
TCE (mean 60 ug/g-d),
48wks
LOAEL:
0.5 mg/mL
Increased ANA after
24 wks but not statistically
significant
Increased INF-y secretion
after 36 wks but not
statistically significant
Hepatic necrosis; hepatocyte
proliferation; leukocyte
infiltrate in the liver, lungs,
and kidneys; no difference in
serum aminotransferase liver
enzymes
Cai et al.
(2008)
Autoimmune-prone: male and female offspring MRL +/+ mice, drinking water
3 litters/group,
8-12 offspring/group; 0, 0.5,
or 2.5 mg/mL, GD 0 to
7-8 wks of age
LOAEL:
0.5 mg/mL
Increased anti-histone
antibodies and total IgG2a
in treated groups
Dose-dependant increase in
IFN-y secretion at 4-5 wks of
age but not 7-8 wks of age
No histopathological effects
in liver or kidneys
Blossom and
Doss (2007)
8 litters/group,
8-12 offspring/group; 0 or
0.1 mg/mL; maternal dose =
25.7 mg/kg/d; offspring PND
24-42 dose = 31.0 mg/kg/d;
GD 0 to PND 42
LOAEL:
0.1 mg/mL
Not evaluated
Increased IFN-y and IL-2 in
females, increased TNF- a in
both sexes
Not evaluated
Blossom et al.
(2008)
Unknown # litters/group,
6-10 offspring/sex/group; 0
(l%emulphor), 1400, or
14,000 ppb; GD 0 to 12
months of age
NOAEL:
1,400 ppb
No increase in
autoantibody levels
Not evaluated
Not evaluated
Peden-Adams
et al. (2008)
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table 4-67. Summary of autoimmune-related studies of TCE and metabolites (by sex, strain, and route of
exposure) (continued)
Number/group, vehicle,
dose, duration
NOAEL;
LOAELb
Results
Serology
Ex vivo assays of cultured
splenocytes
Clinical and
histopathology
Reference
Autoimmune-prone: Female MRL +/+ Mice, Intraperitoneal Injection
4-5 per group, 0 (corn oil),
10 mmol/kg TCE, or
0.2 mmol/kg dichloroacetyl
chloride, every 4th day for
6 wks
6 per group, 0 (corn oil),
0.2 mmol/kg dichloroacetyl
chloride, or 0.2 mmol/kg
dichloroacetic anhydride,
2 times per wk for 6 wks
LOAEL:
10 mmol/kg TCE,
0.2 mmol/kg
dichloroacetyl
chloride
LOAEL:
0.2 mmol/kg TCE,
0.2 mmol/kg
dichloroacetic
anhydride
In both groups, increased
ANA and anti-ssDNA
antibodies. In
dichloroacetyl chloride
group, anti-cardiolipin
antibodies. No difference
in anti-histone, -Sm, or -
DNA antibodies
In both treated groups,
increased ANA
Not evaluated
In both treated groups,
increased IL-lo, IL-1B, IL-3,
IL-6, IFN-y, G-CSF and
keratinocyte-derived
chemokine (KC) secretion;
decreased IL-5. In
dichloroacetyl chloride group,
increased IL-17 and INF-ad
Not evaluated
In both treated groups,
increased lymphocytes in
spleen, thickening of alveolar
septa with lymphocytic
interstitial infiltration
Khan et al.
(1995)
Cai et al.
(2006)
Autoimmune-prone: Female NZB x NZW Mice, Drinking Water
6 per group, 0, 1400, or
14,000 ppb TCE e'f, 27 wks
exposure
10 per group, 0, 1400, or
14,000 ppb TCE f, 27 wks
exposure
LOAEL: 1,400 ppb
LOAEL: 1,400 ppb
Increased anti-dsDNA
antibodies at 19 wks and at
32-32 wks in the 1,400 ppb
group
Increased anti-dsDNA
antibodies at 19 wks and at
32-32 wks in the 1,400 ppb
group
Not evaluated
No effect on splenocyte NK
activity
At 14,000 ppb, proteinuria
increased beginning at
20 wks; renal pathology
scores increased, no evidence
of liver disease
No effect on renal pathology
score; liver disease not
examined
Gilkeson et al.
(2004)
Kiel et al.
(2009)
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
o §
H I
Table 4-67. Summary of autoimmune-related studies of TCE and metabolites (by sex, strain, and route of
exposure) (continued)
Number/group, vehicle,
dose, duration
NOAEL;
LOAELb
Results
Serology
Ex vivo assays of cultured
splenocytes
Clinical and
histopathology
Reference
Autoimmune-prone: Male MRL — Ipr/lpr Mice, Inhalation
5 per group, 0, 500, 1000, or
2,OOOppmTCE,4h/d,
6 d/wk, 8 wks
LOAEL: 500 ppm
At >500 ppm, dose-related
liver inflammation,
splenomegaly and hyperplasia
of lymphatic follicles; at
1,000 ppm, immunoblastic
cell formation in lymphatic
follicles, no changes in
thymus
Kaneko et al.
(2000)
Autoimmune-inducible: Female Brown Norway Rat, Gavage
6-8 per group, 0, 100, 200,
400 mg/kg, 5 d/wk, 6 wks
followed by 1 mg/kg HgCl2
challenge
NOAEL 500
mg/kg
Not reported8
Not evaluated
Not evaluated
White et al.
(2000)
Nonautoimmune-prone: Female B6C3F1 Mice, Drinking Water
6 per group, 0, 1400, or
14,000 ppb TCE,e'f 30 wks
exposure
LOAEL: 1,400 ppb
Anti-dsDNA increased in
1,400 ppb group beginning
at age 32 wks and in the
14,000 ppb group
beginning at age 26 wks
No effect on splenocyte NK
activity
No renal disease observed
Gilkeson et al.
(2004)
O
H
W
-------
Table 4-67. Summary of autoimmune-related studies of TCE and metabolites (by sex, strain, and route of
exposure) (continued)
Number/group, vehicle,
dose, duration
10 per group, 0, 1400, or
14,000 ppb TCE,f 30 wks
exposure
NOAEL;
LOAELb
LOAEL: 1,400 ppb
Results
Serology
Anti-dsDNA increased
beginning at 26 wks in the
14,000 ppb group and at
32 wks of age in the
1,400 ppb group; increases
in anti-ssDNA antibodies
seen in both groups at
32 wks. Anti-GA were not
affected
Ex vivo assays of cultured
splenocytes
No effect on splenocyte NK
activity
Clinical and
histopathology
Increased renal pathology
scores in 1,400 ppb group;
Significant decrease in
thymus weight in both groups
Reference
Kiel et al.
(2009)
Selected endpoints, based on those reported across the majority of studies. Lupus-prone mouse strains develop lupus-like condition spontaneously, with virtually
complete penetrance. The autoimmune-inducible (Brown Norway) rat has been used as a model of mercuric chloride induced glomerulonephritis and
experimental autoimmune myasthenia gravis.
bNOAEL and LOAEL are based upon reported study findings.
°No difference reported in anti- ds-DNA, -ss-DNA, -ribonucleosome, -SSA, -SSB, -Sm, -Jo-1, or -Scl-70 antibodies.
dNo difference reported in secretion of other cytokines measured: IL-2, IL-4, IL-10, IL-12, TNF-a, granulocyte monocyte colony stimulating factor, macrophage
inflammatory protein-la, and RANTES (CCL-5).
Dose levels cited in the report (Gilkeson et al., 2004) were incorrect; corrections provided by personal communication from Margie Peden-Adams (Medical
University of South Carolina) to Glinda Cooper (U.S. EPA) on 13 August 2008; dose levels in this table are correctly report.
fDose in mg/kg/d not given.
gAnti-dsDNA tests were described in the methods section; no effect of TCE on serum IgE levels was seen, and it is not clear if the additional serological tests
were conducted in the TCE portion of this study or if they were conducted but not reported because no effect was seen.
-------
In a subsequent study which assessed occupationally relevant concentrations, TCE was
administered to female MRL +/+ mice (8/group) in drinking water at treatment levels of 0.1, 0.5,
or 2.5 mg/mL (21, 100, or 400 mg/kg/d) for 4 and 32 weeks (Griffin et al., 2000b). At 4 weeks,
significant increases in serum antinuclear antibody levels were observed at 0.1 and 0.5
mg/kg/d;at 32 weeks, the effects were observed at all three treatment levels. A dose-related
increase in the percentage of activated CD4+ T-cells in spleens and lymph nodes of treated mice
was observed at 32 weeks, and the CD4+ T-cells were found to secrete Thl-type cytokines at 4
and 32 weeks.
A similar response was observed by Cai et al. following chronic (48 weeks) exposure of
TCE to female MRL +/+ mice (5/group) in drinking water at 0 or 0.5 mg/mL (approximately
60 ug/g/day) (Cai et al., 2008). After 11 weeks of treatment, a statistically significant decrease
in body weight gain was observed. After 24 weeks of exposure, serum ANA were consistently
elevated in treated mice as compared to control, although statistical significance was not
achieved. Apparent treatment-related effects on serum cytokines included decreased IL-6 after
36 and 48 weeks, decreased TNF-a after 48 weeks, and increased granulocyte colony stimulating
factor (G-CSF) after 36 weeks of treatment. After 36 weeks of treatment, ex vivo cultured
splenocytes secreted higher levels of IFN-y than control splenocytes. Although there were no
observed effects on serum aminotransferase liver enzymes at termination, statistically significant
incidences of hepatocytic necrosis and leukocyte infiltration (including CD3+ T lymphocytes)
into liver lobules were observed in treated mice after 48 weeks of exposure. Hepatocyte
proliferation was also increased. TCE treatment for 48 weeks also induced necrosis and
extensive infiltration of leukocytes in the pancreas, infiltration of leukocytes into the perivascular
and peribronchial regions of the lungs, and thickening of the alveolar septa in the lungs. At 36
and 48 weeks of exposure, massive perivascular infiltration of leukocytes (including CD3+ T
lymphocytes) was observed in the kidneys, and immunoglobulin deposits were found in the
glomeruli.
To examine the role of metabolic activation in the autoimmune response, Griffin et al.
(2000c) treated MRL +/+ mice with 2.5 mg/mL (300 mg/kg/d) TCE in drinking water for
4 weeks (Griffin et al., 2000c). Immune responses were examined in the presence or absence of
subcutaneous doses of 200 mg/kg/d diallyl sulfide, a specific inhibitor of CYP2E1 which is
known to be a primary CYP cytochrome that is active in TCE metabolism. With diallyl sulfide
cotreatment that resulted in a decreased level of CYP2E1 apoprotein in liver microsomes, the
enhanced mitogen-induced proliferative capacity of T-cells was inhibited and the reduction in
IL-4 levels secreted by CD4+ T-cells was reversed for TCE-treated MRL +/+ mice. This study
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-390 DRAFT—DO NOT CITE OR QUOTE
-------
suggests that metabolism of TCE by CYP2E1 is responsible, at least in part, for the treatment-
related CD4+ T-cell alterations.
The TCE metabolite, trichloroacetaldehyde (TCAA) or trichloroacetaldehyde hydrate
(TCAH), was also evaluated in MRL +/+ mice (Blossom et al., 2007; Blossom and Gilbert,
2006; Gilbert et al., 2004) in order to determine if outcomes similar to the immunoregulatory
effects of TCE would be observed, and to attempt to further characterize the role of metabolism
in the mode of action for TCE. At concentrations ranging from 0.04 to 1 mM, TCAA stimulated
proliferation of murine Thl cells treated with anti-CD3 antibody or antigen in vitro. At similar
concentrations, TCAA induced phenotypic alterations consistent with upregulation of CD28 and
downregulation of CD62L in cloned memory Thl cells and DC4+ T-cells from untreated MRL
+/+ mice. Phosphorylation of activating transcription factor 2 (ATF-2) and c-Jun (two
components of the activator protein-a transcription factor) was also observed with TCAA-
induced Thl cell activation. Higher concentrations of TCAA formed a Schiff base on T-cells,
which suppressed the ability of TCAA to phosphorylate ATF-2. These findings suggested that
TCAA may promote T-cell activation by stimulating the mitogen-activated protein kinase
pathway in association with Schiff base formation on T-cell surface proteins (Gilbert et al.,
2004).
In order to determine whether metabolites of TCE could mediate the immunoregulatory
effects previously observed with TCE treatment (i.e., the generation of lupus and autoimmune
hepatitis, associated with activation of IFN-y-producing CD4+ T-cells), Blossom et al. (2004)
administered TCE metabolites, TCAH and trichloroacetic acid (TCA), to MRL +/+ mice
(6-8/group) in drinking water for 4 weeks. Drinking water concentrations were 0, 0.1, or
0.9 mg/mL; average daily doses were calculated as 0, 24, or 220 mg/kg/d for TCAH and 0, 27, or
205 mg/kg/d for TCA. These treatment levels were considered to be physiologically relevant
and to reflect occupational exposure. A phenotypic analysis of splenic and lymph node cells,
cytokine profile analysis, evaluation of apoptosis in CD4+ T-cells, and examination of serum
markers of autoimmunity (anti-ssDNA, anti-histone, or ANA) were conducted. Exposure to
TCAH or TCA at both treatment levels was found to promote CD4+ T-cell activation, as shown
by significant (p < 0.05) increases in the percentage of CD62L10 CD4+ T-cells in the spleens and
lymph nodes of the MRL +/+ mice. Increased levels of LFN-y were secreted by CD4+ T-cells
from mice treated by TCAH and TCA. No significant changes in body weight were observed;
spleen weights were similar between control and treated mice with the exception of a significant
decrease in spleen weight from mice treated with 0.9 mg/mL TCA. Liver and kidney histology
were not affected, and serum alanine aminotransferase levels were similar for control and treated
mice. A generalized trend towards an increase in serum autoantibodies (anti-ssDNA) was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-391 DRAFT—DO NOT CITE OR QUOTE
-------
observed in TCAH-treated mice, and slight but significant increases in anti-histone and anti-
nuclear antibody production were observed in mice treated with 0.9 mg/mL-day TCAH.
The autoimmune response of female MRL +/+ mice to DCAC, a metabolite of TCE, and
to dichloroacetic anhydride (DCAA) a similar acylating agent, was evaluated by Cai et al.
(2006). Six mice/group were injected intraperitoneally, twice weekly for 6 weeks, with
0.2 mmol/kg DCAC or DCAA in corn oil. Body weight gain was significantly decreased after 5
or 6 weeks treatment with DCAC and DCAA. DCAC treatment resulted in significant increases
in total serum IgG (77% increase over control) and IgGl (172% increase over control), as well as
the induction of DCAC-specific IgG and IgGl. Serum IgM levels were significantly decreased
by 25 and 18% in DCAC and DCAA-treated mice, respectively. IgE levels were increased
100% over controls in DCAC-treated mice. Of eight Thl/Th2 cytokines measured, only IL-5
was decreased in DCAC- and DCAA-treated mice. Serum ANA were detected in both DCAC-
and DCAA-treated mice. Treatment-related increases in cytokine and chemokine secretion in
cultured splenocytes were observed for DCAC and DCAA (IL-1, G-CSF, KC, IL-3, and IL-6).
DCAC-treated splenocytes also secreted more IL-17 and IFN-a than controls. Histopathological
changes were observed in the spleens of DCAC and DCAA-treated mice (lymphocyte population
increases in the red pulp). With both DCAC and DCAA treatment, the alveolar septa were
thickened in the lungs, moderate levels of lymphocytic interstitial infiltrates were present in
tissues, and alveolar capillaries were clogged with erythrocytes. These findings were attributed
both to the predisposition of the MRL +/+ mice towards autoimmune disease, and to the
treatment-related induction of autoimmune responses.
Fas-dependant activation-induced cell death leading to autoimmune disease has been
shown to be related to impaired Fas or FasL ligand expression in humans and mice, and defects
in the Fas-signaling pathways have been described in autoimmune disease models. The study by
Blossom and Gilbert examined the effects of TCAH on Fas-dependent autoimmune cell death
(Blossom and Gilbert, 2006). In this study, TCAH (1) inhibited apoptosis of antigen-activated
cells, (2) did not protect CD4+ T-cells from Fas-independent apoptosis, (3) did not inhibit
autoimmune cell death induced by direct engagement of the Fas receptor, (4) inhibited the
expression of FasL but not Fas on the surface of activated CD4+ T-cell, (5) increased release of
FasL from CD4+ cells in a metalloprotein-dependent manner, and (6) increased metalloprotein
MMP-7 expression.
Gilbert et al. (2006) studied the effect of treatment on apoptosis in CD4+ T-lymphocytes
isolated from MRL +/+ female mice that had been exposed to TCE (0, 0.1, 0.5, or 2.5 mg/mL) in
the drinking water for 4 or 32 weeks or to TCAH (0.1, 0.3, or 0.9 mg/mL) in drinking water for 4
or 40 weeks. After only 4 weeks, decreased activation-induced apoptosis was associated with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-392 DRAFT—DO NOT CITE OR QUOTE
-------
decreased FasL expression in the CD4+ T-cells, suggesting that TCE- and TCAH-induced
autoimmune disease was promoted through suppression of the process that would otherwise
delete activated self-reactive T-lymphocytes. By 32 weeks of treatment, TCE had induced
autoimmune hepatitis, which was associated with the promotion of oxidative stress, the
formation of liver protein adducts, and the stimulated production of antibodies to those adducts.
TCAH-treated mice did not exhibit autoimmune hepatitis by 40 weeks, but developed a dose-
dependant alopecia and skin inflammation (Blossom et al., 2007). TCAH appeared to modulate
the CD4+ T-cell subset by promoting the expression of an activated/effector phenotype with an
increased capacity to secrete the proinflammatory cytokine IFN-y. A 4-week exposure to TCAH
attenuated activation-induced cell death and the expression of the death receptor Fas in CD4+
cells; these effects were not seen after a 40-week exposure period. Differences in response were
tentatively attributed to higher levels of metalloproteinases (specifically MMP-7) at 4-weeks of
treatment, suggesting a possible mechanism for the promotion of skin pathology by TCAH.
The role of protein adduct formation in autoimmune response has been pursued by
various researchers. Halmes et al. administered a single i.p. dose of TCE in corn oil to male
Sprague-Dawley rats (2/group) at 0 or 1,000 mg/kg (Halmes et al., 1997). Using antiserum that
recognizes TCE covalently bound to protein, a single 50 kDa microsomal adduct was detected by
Western blot in livers of treated rats. Using affinity chromatography, a 50 kDa dichloroacetyl
protein was also isolated from rat plasma. The protein was reactive immunochemically with
anti-CYP2El antibodies. The data suggest that the protein adduct may be CYP2E1 that has been
released from TCE-damaged hepatocytes.
Cai et al. examined the role of protein haptenization in the induction of immune
responses (Cai et al., 2007). In this study, MRL +/+ mice were immunized with albumin adducts
of various TCE reactive intermediates of oxidative metabolism. Serum immunoglobulins and
cytokine levels were measured to evaluate immune responses against the haptenized albumin.
Antigen-specific IgG responses (subtypes: IgGl, IgG2a, and IgG2b) were found. Serum levels
of G-CSF were increased in immunized mice, suggesting macrophage activation. Following
immunization with formyl-albumin, lymphocyte infiltration in the hepatic lobule and portal area
was increased. This study suggests that proteins that are haptenized by metabolites of TCE may
act as antigens to induce humoral immune responses and T-cell-mediated hepatitis.
A possible role for oxidative stress in inflammatory autoimmune disease was proposed by
Khan et al. (2001). A study was performed in which female MRL +/+ mice were treated with
10 mmol/kg TCE or 0.2 mmol/kg DCAC via intraperitoneal injection every 4th day for 2, 4, 6, or
8 weeks. Anti-malondialdehyde serum antibodies, a marker of lipid peroxidation and oxidative
stress, were measured and were found to increase by 4 weeks of treatment, marginally for TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-393 DRAFT—DO NOT CITE OR QUOTE
-------
and significantly for DCAC. It was reported that anti-malondialdehyde antibodies has also been
found to be present in the serum of systemic lupus erythematosus-prone MRL-lpr/lpr mice.
In another study that addressed the association of oxidative and nitrosative stress, and the
role of lipid peroxidation and protein nitration, in TCE-mediated autoimmune response,
Wang et al. treated female MRL +/+ mice with 0.5 mg/mL TCE in drinking water for 48 weeks
(Wang et al., 2007b). The formation of antibodies in the serum to lipid peroxidation-derived
aldehyde protein adducts was evaluated. With TCE treatment, the serum levels of anti-
malondialdehyde and anti-4-hydroxynonenal protein adduct antibodies, inducible nitric oxide
synthase, and nitrotyrosine were increased. These were associated with increases in anti-
nuclear-, anti-ssDNA- and anti-dsDNA antibodies. The involvement of lipid
peroxidation-derived aldehyde protein adducts in TCE autoimmunity was further explored, using
female MRL +/+ mice that were administered by i.p. injections of TCE at 10 mmol/kg, either
every 4th day for 6 or 12 weeks (Wang et al., 2007a) or once per week for 4 weeks (Wang et al.,
2008). Significant increases in malondialdehyde and 4-hydroxynonenal protein adducts, as well
as significant induction of specific antibodies directed against these antigens were observed in
both studies. Wang et al. also demonstrated a significant proliferation of CD4+ T-cells in TCE-
treated mice, and splenic lymphocytes from TCE-treated mice released more IL-2 and IFN-y
when stimulated with MDA- or 4-hydroxynonenal-adducted mouse serum albumin (Wang et al.,
2008). Overall, the result of these studies suggest a role for lipid peroxidation aldehydes in the
induction and/or exacerbation of autoimmune response in the MRL +/+ animal model, and the
involvement of Thl cell activation.
In studies conducted in other rodent strains, less consistent outcomes have been observed.
Inhalation exposure of an autoimmune-prone strain of male mice (MRL-lpr/lpr) to 0-, 500-,
1,000-, or 2,000-ppm TCE for 4 hours/day, 6 days/week, for 8 weeks resulted in depressed serum
IgG levels and increased numbers of lymphoblastoid cells (Kaneko et al., 2000). Also at
2,000 ppm, changes in T-cell helper to suppressor cell ratios were observed. At
histopathological evaluation, dose-dependent inflammation and associated changes were noted in
the liver at >500 ppm, hyperplasia of the lymphatic follicles of the spleen and splenomegaly
were observed at >500 ppm, and the spleen exhibited the development of an immunoblastic-cell-
like structure at 1,000 ppm.
A 26-week drinking water study of TCE in NZB x NZW (NZBWF1) autoimmune-prone
mice demonstrated an increase in anti-dsDNA antibodies at 19 weeks and at 32 and 34 weeks in
the 1,400 ppb group, and increased kidney disease at 14,000 ppb (i.e., increased proteinuria at
20 weeks; increased renal pathology scores at termination, based upon glomerular proliferation,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-394 DRAFT—DO NOT CITE OR QUOTE
-------
inflammation, and necrosis) (Gilkeson et al., 2004).l Also in that study, a small increase in anti-
dsDNA antibody production, without kidney disease, was observed in B6C3F1 mice, with
statistically significant (p < 0.05) or borderline (p = 0.07) effects seen in the 1,400-ppb group at
observations between 32 and 39 weeks of age, and in the 14,000 ppb group at observations
between 26 and 39 weeks of age.
Keil et al. (2009) also assessed the effects of TCE exposure on NZWBF1 mice,
comparing the responses to those of TCE-exposed B6C3F1 mice, which are not autoimmune
prone (Keil et al. 2009). In this study, groups of NZWBF1 and B6C3F1 female mice (10/dose
level) were administered 0, 1400, or 14,000 ppb TCE in the drinking water. Treatment was
initiated at 9 weeks of age and continued until 36 weeks of age for the NZBWF1 and until
39 weeks of age for the B6C3F1 mice. Body weight; spleen, thymus, liver, and kidney weight;
spleen and thymus cellularity; and renal pathology were assessed. Splenic lymphocyte
proliferation, autoantiboidy production (anti-dsDNA, anti-ssDNA, and anti-glomerular), total
serum IgG, NK cell activity, and mitogen-induced lymphocyte proliferation were conducted.
Administration of TCE did not result in alterations in NK cell activity or T- or B-cell
proliferation in either strain of mice. In the NZBWF1 mice, there was little evidence of an
increase or of an acceleration in ss-DNA antibody production with TCE exposure, but as was
seen in the earlier study by these investigators (Gilkeson et al., 2004), ds-DNA antibodies were
increased at 19 weeks and at 32-34 weeks in the 1,400 ppb group. However, anti-glomerular
antibody levels were increased in NZBWF1 mice early in the study, returning to control levels
by 23 weeks of age. In the B6C3F1 mice the number of activated T-cells (CD4++/CD44+) was
increased (significantly at 14,000 ppm;/? < 0.05) and thymus weights were significantly
decreased (p < 0.05) in a dose-responsive manner. Renal pathology (as indicated by renal score
based on assessment of glomerular inflammation, proliferation, crescent formation and necrosis)
was significantly increased (p < 0.05) at 1,400 ppm. Also in the B6C3F1 mice, autoantibodies to
dsDNA were increased relative to controls beginning at 26 weeks in the 14,000-ppb group and at
32-weeks of age in the 1,400 ppb group; increases in anti-ssDNA antibodies were seen in both
groups at 32 weeks. Anti-glomerular antibodies were not affected in B6C3F1 mice. In summary,
the authors concluded that this study showed that 27-30 weeks of TCE drinking water
administration to NZBWF1 (autoimmune-prone) mice did not contribute to the progression of
autoimmune disease, while similar administration to B6C3F1 (nonautoimmune-prone) mice
increased the expression of a number of markers that are associated with autoimmune disease.
lrrhe study was reported in symposium proceedings. Dose levels cited in the proceedings were incorrect; however,
corrections were provided by personal communication from Margie Peden-Adams (Medical University of South
Carolina) to Glinda Cooper (U.S. EPA) on 13 August 2008, and dose levels are correctly reported here.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-395 DRAFT—DO NOT CITE OR QUOTE
-------
This study is important in that it demonstrates that autoimmune responses to TCE exposure in
animal models are not solely dependant upon a genetic predisposition to autoimmune disease.
White et al. conducted a study in female Brown Norway rats, which have been shown to
be susceptible to development of chemically-induced IgE mediated glomerulonephritis that is
similar to the nephritic damage seen in systemic lupus erythematosus (White et al., 2000). TCE
administered by gavage 5 days/week at 100, 200, or 400 mg/kg did not increase in IgE levels
after 6 weeks exposure, or after an additional challenge with 1 mg/kg mercuric chloride (HgCl2).
Several studies have examined the potential for autoimmune response following oral
exposures during pre- and postnatal immune system development, as described in
Section 4.6.2.1.2 above. Peden-Adams et al. conducted two such studies. In the first study,
B6C3F1 mice were treated with either 1,400 or 14,000 ppb TCE in drinking water from gestation
Day 0 to postnatal Week 8 (Peden-Adams et al., 2006). No treatment-related increases in serum
anti-ds-DNA antibody levels were observed in the 8-week old offspring, although it is noted that
the mouse strain used in the experiment is not an autoimmune-prone animal model. A more
recent study (Peden-Adams et al., 2008) exposed pregnant MRL +/+ mice to TCE in drinking
water at levels of 0, 1,400, or 14,000 ppb from GD 0 and continued the exposures until the
offspring were 12 months of age. Consistent with the findings of the 2006 publication,
autoantibody levels (anti-dsDNA and anti-glomerular) were not increased in the offspring over
the course of the study. Contrasting with these negative studies, the lupus-prone MRL +/+
mouse model was utilized in two additional drinking water studies with developmental exposures
in which there was some indication of a positive association between developmental exposures
to TCE and the initiation of autoimmune disease. Blossom and Doss (2007) administered TCE
to pregnant MRL +/+ mice in drinking water at levels of 0, 0.5, or 2.5 mg/mL and continued
administration to the offspring until approximately 7-8 weeks of age. TCE exposure induced a
dose-dependent increase in T-lymphocyte IFN-y in peripheral blood at 4-5 weeks of age, but this
effect was not observed in splenic T-lymphocytes at 7-8 weeks of age. Serum anti-histone
autoantibodies and total IgG2a were significantly increased in the TCE-treated offspring;
however, histopathological evaluation of the liver and kidneys did not reveal any treatment-
related signs of autoimmunity. In a study by Blossom et al. (2008), pregnant MRL +/+ mice
were administered TCE in the drinking water at levels of 0 or 0.1 mg/mL from GD 0 through
lactation, and continuing postweaning in the offspring until postnatal Day 42. Significant
treatment-related increases in pro-inflammatory cytokines (IFN-y and 11-2 in males and TNF-a in
both sexes) produced by splenic CD4+ T-cells were observed in PND 42 offspring.
In summary, TCE treatment induces and exacerbates autoimmune disease in genetically
susceptible strains of mice, and has also been shown to induce signs of autoimmune disease in a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-396 DRAFT—DO NOT CITE OR QUOTE
-------
nongenetically predisposed strain. Although the mechanism for this response is not fully
understood, a number of studies have been conducted to examine this issue. The primary
conclusion to date is that metabolism of the TCE to its chloral or dichloroacetic acid metabolites
is at least partially responsible for activating T-cells or altering T-cell regulation and survival
associated with polyclonal disease in susceptible mice strains.
4.6.2.4. Cancers of the Immune System
Cancers of the immune system that have been observed in animal studies and are
associated with TCE exposure are summarized in Tables 4-68 and 4-69. The specific tumor
types observed are malignant lymphomas, lymphosarcomas, and reticulum cell sarcomas in mice
and leukemias in rats.
In the NCI (1976) study, the results for Osborne-Mendel rats were considered
inconclusive due to significant early mortality, but exposure to B6C3F1 mice were also
analyzed. Limited increases in lymphomas over controls were observed in both sexes of mice
exposed (see Table 4-68). The NCI study (1976) used technical grade TCE which contained two
known carcinogenic compounds as stabilizers (epichlorohydrin and 1,2-epoxybutane). A later
study (Henschler et al., 1984) in which mice were given TCE that was pure, industrial, and
stabilized with one or both of these stabilizers did not find significant increases in lymphomas
over historical controls. A gavage study by NTP (1988), which used TCE stabilized with
diisopropylamine, did not see an increase in lymphomas in all four strains of rats (ACI, August,
Marshall, and Osborne-Mendel). The final NTP study (1990) in male and female F344 rats and
B6C3F1 mice, using epichlorohydrin-free TCE, again experienced early mortality in male rats.
This study did not observe significant increase in lymphomas over that of controls. Henschler et
al. (1980) tested NMRI mice, WIST rats and Syrian hamsters of both sexes, and observed a
variety of tumors in both sexes (Henschler et al., 1980), consistent with the spontaneous tumor
incidence in this strain (Deerberg and Muller-Peddinghaus, 1970; Deerberg et al., 1974).
Henschler et al. did not show an increase in lymphomas in rats or hamsters of either sex
(Henschler et al., 1980). Background levels of lymphomas in this mouse strain are high, making
it difficult to determine if the increased lymphomas in female mice is a treatment effect. In a
follow-up study, Henschler et al. (1984) examined the role of stabilizers of TCE in the
lymphomas demonstrated in female mice in the 1980 paper. Each exposure group had
-50 SPF-bred ICR/HA-Swiss mice and exposure was for 18 months. Background incidence of
tumors was high in all groups. Focusing just on malignant lymphomas (see Table 4-68), the high
background incidence in unexposed animals again makes it difficult to determine if there is TCE
and/or stabilizer-related incidence of lymphomas. There are no data at any other timepoint than
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-397 DRAFT—DO NOT CITE OR QUOTE
-------
18 months. A high mortality rate in all animals as well as the increased incidence of
'background' lymphomas in that report was also a problem and may have been related to the
shorter time frame.
Table 4-68. Malignant lymphomas incidence in mice exposed to TCE in
gavage and inhalation exposure studies
Cancer type, species, and sex
Exposure groups
Reference
Gavage exposure
Malignant lymphomas
Prevalence in: (n affected/total)
B6C3F1 mice, male
B6C3F1 mice, female
Vehicle control
11/50(22%)
7/48 (15%)
1,000 mg/kg/d
13/50 (26%)
13/49 (27%)
Lymphosarcomas and reticulum cell sarcomas
Prevalence in: (n affected/total)
B6C3F1 mice, male
B6C3F1 mice, female
Vehicle control
1/20 (5%)
1/20 (
5%)
Low dose
4/50
(8%)
5/50 (10%)
High dose
2/48 (4%)
5/47(11%)
Malignant lymphomas
Prevalence in: (n affected/total)
Swiss (ICR/HA) mice, male
Swiss (ICR/HA) mice, female
Control
19/50
(38%)
28/50
(56%)
Inhalation exposure
Malignant lymphomas
TCE-
pure
16/50
(32%)
21/50
(42%)
TCE-
indust
17/49
(35%)
19/50
(38%)
Control
Prevalence in:
Han:NMRI mice, male
Han:NMRI mice, female6
TCE-
EPC
11/49
(22%)
20/50
(40%)
TCE-
BO
11/49
(22%)
23/48
(48%)
96
(n affected/total)
7/30 (23%)
9/29(31%)
TCE-
EPC-BO
12/49
(24%)
18/50
(36%)
NTP, 1990a
NCI, 1976b
Henschler et al.,
1984C
480
7/29 (24%)
17/30 (57%)
6/30 (20%)
18/28 (64%)
Henschler et al.,
1980d
aAfter 103 weeks gavage exposure, beginning at 8 weeks of age.
bAfter 90 weeks gavage exposure, beginning at 5 weeks of age. Low dose is 1,200 mg/kg/d for male mice,
900 mg/kg/d for female mice (5 days/week). High dose is 2,400 mg/kg/d for male mice, 1,800 mg/kg/d for female
mice (5 days/week).
0 After 72 weeks gavage exposure (corn oil), beginning at 5 weeks of age. Male mice received 2,400 mg/kg/d,
female mice received 1,800 mg/kg/d. Stabilizers were added in the percent w/w: TCE-EPC, 0.8%, TCE-BO,
0.8%, TCE-EPC-BO, 0.25 and 0.25%.
dAfter 78 weeks inhalation exposure. Administered daily concentration: low dose is 96 (mg/m3) and high dose is
480 (mg/m3), equivalent to 100 and 500 ppm (100 ppm = 540 mg/m3), adjusted for 6 hours/day, 5 days/week
exposure.
Statistically significant by Cochran-Armitage trend test (p < 0.05).
Sources: NTP (1990) Tables 8, 9; NCI (1976) Table XXXa; Henschler et al. (1980) Table 3a.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-398 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-69. Leukemia incidence in rats exposed to TCE in gavage and
inhalation exposure studies
Species and sex
Exposure groups
Reference
Gavage exposure
Prevalence in (n affected/total)
Sprague-Dawley rats, male
Sprague-Dawley rats, female
August rats, female
Control
0/30
(0%)
1/30
(3.3%)
Control
0/50
(0%)
50 mg/kg
2/30 (6.7%)
0/30
(0%)
500 mg/kg
1/50
(2%)
250 mg/kg
3/30
(10.0%)
0/30
(0%)
1,000 mg/kg
5/50
(10%)
Maltoni et al., 1986 a
NTP, 1988b
Inhalation exposure
Prevalence in (n affected/total)
Sprague-Dawley rats, male
Sprague-Dawley rats, female
Control
9/135
(6.7)
7/145
(4.8)
100 ppm
13/130
(10.0)
9/130
(6.9)
300 ppm
14/130
(10.8)
2/130
(1.5)
600 ppm
15/130
(11.5)
11/130
(8.5)
Maltoni et al., 1988 c
aAfter 52 weeks gavage exposure, beginning at 13 weeks of age, olive oil vehicle. Percent affected and starting n
given in reported; U.S. EPA calculated n affected.
bAfter 104 weeks gavage exposure, beginning at 6.5-8 weeks of age, corn oil vehicle.
0 After 104 weeks inhalation exposure, BT304 and BT304bis. Percent affected and starting n given in reported; U.S.
EPA calculated n affected.
Maltoni et al reported a nonsignificant increase in leukemias in male rats exposed via
inhalation (Matoni et al., 1988, 1986). Maltoni et al. (1986) demonstrates a borderline higher
frequency of leukemias in male Sprague-Dawley rats following exposure by ingestion for
52 weeks, believed by the authors to be related to an increase in lymphoblastic lymphosarcomas
(see Table 4-69). The gavage study by NTP (1988), which used TCE stabilized with
diisopropylamine, observed leukemia in female August rats with a positive trend, but was not
significantly greater than the vehicle controls.
In summary, overall there is limited available data in animals on the role of TCE in
lymphomas and leukemias. There are few studies that analyze for lymphomas and/or leukemias.
Lymphomas were described in four studies (NTP, 1990; NCI, 1976; Henschler et al., 1980,
1984) but study limitations (high background rate) in most studies make it difficult to determine
if these are TCE-induced. Three studies found positive trends in leukemia in specific strains
and/or gender (Maltoni et al., 1986, 1988; NTP, 1988). Due to study limitations, these trends can
not be determined to be TCE-induced.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-399 DRAFT—DO NOT CITE OR QUOTE
-------
4.6.3. Summary
4.6.3.1. Noncancer Effects
The human and animal studies of TCE and immune-related effects provide strong
evidence for a role of TCE in autoimmune disease and in a specific type of generalized
hypersensitivity syndrome. The data pertaining to immunosuppressive effects is weaker.
The relation between systemic autoimmune diseases, such as scleroderma, and
occupational exposure to TCE has been reported in several recent studies. A meta-analysis of
scleroderma studies (Diot et al., 2002; Garabrant et al., 2003; Nietert et al., 1998) conducted by
the U.S. EPA resulted in a statistically significant combined odds ratio for any exposure in men
(OR: 2.5, 95% CI: 1.1, 5.4), with a lower relative risk seen in women in women (OR: 1.2,
95% CI: 0.58, 2.6). The incidence of systemic sclerosis among men is very low (approximately
1 per 100,000 per year), and is approximately 10 times lower than the rate seen in women
(Cooper and Stroehla, 2003). Thus, the human data at this time do not allow us to determine if
the difference in effect estimates between men and women reflects the relatively low background
risk of scleroderma in men, gender-related differences in exposure prevalence or in the reliability
of exposure assessment (Messing et al., 2003), a gender-related difference in susceptibility to the
effects of TCE, or chance. Changes in levels of inflammatory cytokines were reported in an
occupational study of degreasers exposed to TCE (lavicoli et al., 2005) and a study of infants
exposed to TCE via indoor air (Lehmann et al., 2001, 2002). Experimental studies support the
biological plausibility of these effects. Numerous studies have demonstrated accelerated
autoimmune responses in autoimmune-prone mice (Cai et al., 2008; Blossom et al., 2007, 2004;
Griffin et al., 2000a, b). With shorter exposure periods, effects include changes in cytokine
levels similar to those reported in human studies. More severe effects, including autoimmune
hepatitis, inflammatory skin lesions, and alopecia, were manifest at longer exposure periods, and
interestingly, these effects differ somewhat from the "normal" expression in these mice.
Immunotoxic effects, including increases in anti-ds DNA antibodies in adult animals and
decreased plaque forming cell response with prenatal and neonatal exposure, have been also
reported in B6C3F1 mice, which do not have a known particular susceptibility to autoimmune
disease (Gilkeson et al., 2004, Peden-Adams et al., 2006). Recent mechanistic studies have
focused on the roles of various measures of oxidative stress in the induction of these effects by
TCE (Wang et al., 2008, 2007b).
There have been a large number of case reports of a severe hypersensitivity skin disorder,
distinct from contact dermatitis and often accompanied by hepatitis, associated with occupational
exposure to TCE, with prevalences as high as 13% of workers in the same location (Kamijima et
al., 2008, 2007). Evidence of a treatment-related increase in delayed hypersensitivity response
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-400 DRAFT—DO NOT CITE OR QUOTE
-------
accompanied by hepatic damage has been observed in guinea pigs following intradermal
injection (Tang et al., 2008, 2006), and hypersensitivity response was also seen in mice exposed
via drinking water pre- and postnatally (gestation Day 0 through to 8 weeks of age)
(Peden-Adams et al., 2006).
Human data pertaining to TCE-related immunosuppression resulting in an increased risk
of infectious diseases is limited to the report of an association between reported history of
bacteria of viral infections in Woburn, Massachusetts (Lagakos, 1986). Evidence of localized
immunosuppression, as measured by pulmonary response to bacterial challenge (i.e., risk of
Streptococcal pneumonia-related mortality and clearance of Klebsiella bacteria) was seen in an
acute exposure study in CD-I mice (Aranyi et al., 1986). A 4-week inhalation exposure in
Sprague-Dawley rats reported a decrease in plaque forming cell response at exposures of
1,000 ppm (Woolhiser et al., 2006).
4.6.3.2. Cancer
Associations observed in epidemiologic studies of lymphoma and TCE exposure suggest
a causal relation between trichloroethylene exposure and lymphoma. Issues of study
heterogeneity, potential publication bias, and weaker exposure-response results contribute
uncertainty to the evaluation of the available data.
In a review of the lymphoma studies, 17 studies in which there is a high likelihood of
TCE exposure in individual study subjects (e.g., based on job-exposure matrices, biomarker
monitoring, or industrial hygiene data on TCE exposure patterns and factors that affect such
exposure) and which met, to a sufficient degree, the standards of epidemiologic design and
analysis were identified. These studies generally reported excess relative risk estimates for
lymphoma between 0.8 and 3.1 for overall TCE exposure. Statistically significant elevated
relative risk estimates with lymphoma and overall TCE exposure were observed in two cohort
(Hansen et al., 2001; Raaschou-Nielsen et al., 2003) and one case-control (Hardell et al., 1994)
study. Both cohort studies reported statistically significant associations with lymphoma for
subjects with longer employment duration as a surrogate of TCE exposure. Hardell et al. (1994)
reported a strong but imprecise association, in part reflecting possible bias from subject-reported
exposure history and few exposed cases. Other high-quality studies reported a 10 to 50%
elevated relative risk estimate with overall TCE exposure that were not statistically significant,
except for two population case-control studies of lymphoma, which did not report relative risk
estimates with overall TCE exposure but did for medium-high intensity or cumulative TCE
exposure (Miligi et al., 2006; Seidler et al., 2007). Fifteen additional studies were given less
weight because of their lesser likelihood of TCE exposure and other design limitations that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-401 DRAFT—DO NOT CITE OR QUOTE
-------
would decrease study power and sensitivity. The observed lack of association with lymphoma in
these studies likely reflects study design and exposure assessment limitations and is not
considered inconsistent with the overall evidence on TCE and lymphoma.
Consistency of the association between TCE exposure and lymphoma is further
supported by the results of meta-analyses of 16 high-quality studies reporting risk estimates for
overall TCE exposure. These meta-analyses found a statistically significant increased pooled
relative risk estimate for lymphoma of 1.23 (95% CI: 1.04, 1.44) for overall TCE exposure. The
analysis of lymphoma was robust to the removal of individual studies and the use of alternate
relative risk estimates from individual studies, and in only one cases was the resulting pooled
relative risk no longer statistically significant (lower 95% confidence bounds of 1.00). Some
evidence heterogeneity was observed, particularly between cohort and case-control studies, but it
was not statistically significant (p = 0.10); and, in addition, there was some evidence of potential
publication bias. Analyzing the cohort and case-control studies separately resolved most of the
heterogeneity, but the result for the pooled case-control studies was only a 7% increased relative
risk estimate and was not statistically significant. The sources of heterogeneity are uncertain but
may be the result of some bias associated with exposure assessment and/or disease classification,
or from differences between cohort and case-control studies in average TCE exposure.
Exposure-response relationships are examined in the TCE epidemiologic studies only to a
limited extent. Many studies examined only overall "exposed" versus "unexposed" groups and
did not provide exposure information by level of exposure. Others do not have adequate
exposure assessments to confidently distinguish between levels of exposure. The lymphoma
case-control study of Seidler et al. (2007) reported a statistically significant trend with TCE
exposure (p = 0.03 for Diffuse B-cell lymphoma trend with cumulative TCE exposure), and
lymphoma risk in Boice et al. (1999) appeared to increase with increasing exposure duration
(p = 0.20 for routine-intermittent exposed subjects). The borderline statistically significant trend
with TCE intensity in the case-control study of Wang et al. (2009 \p = 0.06]) is consistent with
Seidler et al. (2007). Further support was provided by meta-analyses using only the highest
exposure groups, which yielded a higher pooled relative risk estimate (1.57 [95% CI: 1.27, 1.94])
than for overall TCE exposure (1.27 [95% CI: 1.04, 1.44]).
Few risk factors are recognized for lymphoma, with the exception of viruses,
immunosuppression or smoking, which are associated with specific lymphoma subtypes.
Associations between lymphoma and TCE exposure are based on groupings of several
lymphoma subtypes. Three of the six lymphoma case-control studies adjusted for age, sex and
smoking in statistical analyses (Miligi et al., 2006; Seidler et al., 2007; Wang et al., 2009), the
other three case-control studies presented only unadjusted estimates of the odds ratio.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-402 DRAFT—DO NOT CITE OR QUOTE
-------
Animal studies describing rates of lymphomas and/or leukemias in relation to TCE
exposure (NTP, 1990, 1988; NCI, 1976; Henschler et al., 1980, 1984; Maltoni et al., 1986, 1988)
are available. Henschler et al. (1980) reported statistically significant increases in lymphomas in
female Han:NMRI mice treated via inhalation. While Henschler et al. (1980) suggested these
lymphomas were of viral origin specific to this strain, subsequent studies reported increased
lymphomas in female B6C3F1 mice treated via corn oil gavage (NTP, 1990) and leukemias in
male Sprague-Dawley and female August rats (Maltoni et al., 1986; NTP, 1988). However,
these tumors had relatively modest increases in incidence with treatment, and were not reported
to be increased in other studies.
4.7. RESPIRATORY TRACT TOXICITY AND CANCER
4.7.1. Epidemiologic Evidence
4.7.1.1. Chronic Effects: Inhalation
Two reports of a study of 1,091 gun-manufacturing workers are found on noncancer
pulmonary toxicity (Cakmak et al., 2004; Saygun et al., 2007). A subset of these workers
(n = 411) had potential exposure to multiple organic solvents including toluene, acetone, butanol,
xylene, benzene and TCE used to clean gun parts; however, both papers lacked information on
exposure concentration. Mean exposure duration in Cakmak et al. (2004) was 17 years
(SD = 7.9) for nonsmokers and 16 years (SD = 7.1) for smokers. Cakmak et al. (2004) indicated
effects of smoking and exposure to solvents, with smoking having the most important effect on
asthma-related symptoms (smoking, OR = 2.8, 95% CI: 2.0, 3.8; solvent exposure, OR = 1.4,
95% CI: 1.1, 1.9). Similarly, smoking, but not solvent exposure, was shown as a statistically
significantly predictor of lung function decrements. Saygun et al. (2007) reported on a five year
follow-up of 393 of the original 1,091 subjects, 214 of who were exposed to solvents. Of the
393 original subjects, the prevalence of definitive asthma symptoms, a more rigorous definition
than used by Cakmak et al. (2004), was 3.3% among exposed and 1.1% among nonexposed
subjects,p> 0.05. Saygun et al. (2007) presents observations on lung function tests for 697
current workers, a group which includes the 393 original study subjects. Smoking, but not
solvent exposure, was a predictor of mean annual forced expiratory volume (FEVi) decrease.
4.7.1.2. Cancer
Cancers of the respiratory tract including the lung, bronchus, and trachea are examined in
25 cohort, community studies and case-control studies of TCE. Twelve of the 25 studies
approached standards of epidemiologic design and analysis identified in the review of the
epidemiologic body of literature on TCE and cancer (see Appendix B; Siemiatycki, 1991;
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-403 DRAFT—DO NOT CITE OR QUOTE
-------
Axelson et al., 1994; Greenland et al., 1994; Anttila et al., 1995; Blair et al., 1998; Morgan et al.,
1998; Boice et al., 1999, 2006; Hansen et al., 2001; Raaschou-Nielsen et al., 2003; Zhao et al.,
2005; Radican et al., 2008). Cancers at other sites besides lung, bronchus, and trachea in the
respiratory system are more limitedly reported in these studies. Some information is available on
laryngeal cancer; however, only 9 of the 16 occupational cohort studies providing information on
lung cancer also reported findings for this site. Case-control studies of lung or laryngeal cancers
and occupational title or organic solvent exposure were found in the literature. Two case-control
studies of lung cancer, one population-based and the other nested within a cohort, were of TCE
exposure specifically. Lung and laryngeal cancer risk ratios reported in cohort, community and
case-control studies are found in Table 4-70.
Table 4-70. Selected results from epidemiologic studies of TCE exposure and
lung cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
£ for trend
All employees at electronics factory (Taiwan)
Not reported
1.00a
1.36(0.86,2.14)
1.11(0.60,2.06)
0.60
1.07 (0.72, 1.52)
43
35
14
30
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Any exposure, males
Any exposure, females
1.4(1.32, 1.55)
1.4(1.28, 1.51)
1.9(1.48,2.35)
632
559
73
Employment duration
5yrs
1.7(1.46, 1.93)
1.3(1.16, 1.52)
1.4(1.23, 1.63)
209
218
205
Zhao et al., 2005
Chang et al., 2005
Raaschou-Nielsen et al., 2003
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-404 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-70. Selected results from epidemiologic studies of TCE exposure and
lung cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Biologically -monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-yr
Mean concentration (Ikeda)
<4ppm
4+ppm
Employment duration
<6.25 yr
>6.25 yr
0.8(0.5, 1.3)
0.7(0.01,3.8)
Not reported
Not reported
Not reported
16
1
Aircraft maintenance workers (Hill Air Force Base, UT)
TCE subcohort
Not reported
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0"
1.0 (0.6, 2.0)
0.8 (0.4, 1.6)
0.8 (0.4, 1.7)
24
11
15
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0"
1
1
1
Biologically -monitored Finnish workers
All subjects
0.92 (0.59, 1.35)
25
Mean air-TCE (Ikeda extrapolation)
<6 ppm
6+ppm
1.02 (0.58, 1.66)
0.83 (0.33, 1.71)
16
7
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
0.69(0.31,1.30)
Not reported
9
Reference
Hansenetal., 2001
Blair etal., 1998
Anttila et al., 1995
Axelson et al., 1994
Cohort and PMR -mortality
Computer manufacturing workers (IBM), NY
Males
Females
1.03 (0.71, 1.42)
0.95 (0.20, 2.77)
35
3
Clapp and Hoffman 2008
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-405 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-70. Selected results from epidemiologic studies of TCE exposure and
lung cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Aerospace workers (Rocketdyne)
Any TCE (utility or engine flush workers)
1.24 (0.92, 1.63)
51
Engine flush — duration of exposure
Referent
0 yr (utility workers with TCE exposure)
<4yrs
>4yrs
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
1.0a
0.5 (0.22, 1.00)
0.8 (0.50, 1.26)
0.8 (0.46, 1.41)
Not reported
1.00a
1.05 (0.76, 1.44)
1.02 (0.68, 1.53)
0.91
472
7
27
24
99
62
33
View-Master employees
Males
Females
0.81 (0.42, 1.42)b
0.99(0.71, 1.35)b
12
41
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration0
Moderate TCE exposure, >2 yrs duration0
Not reported
Not reported
Not reported
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent exposure3
0.76 (0.60, 0.95)
Not reported
78
173
Duration of exposure
Oyrs
<1 yr
1-4 yrs
>5yrs
Trend test
1.0
0.85(0.65, 1.13)
0.98 (0.74, 1.30)
0.64 (0.46, 0.89)
^<0.05
288
66
63
44
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
1.10(0.89, 1.34)
1.49(1.09, 1.99)
0.90 (0.67, 1.20)
97
45
52
TCE subcohort (Cox Analysis) b
Never exposed
Ever exposed
1.00a
1.14(0.90, 1.44)
291
97
Peak
No/Low
1.00a
324
Reference
Boice et al., 2006
Zhao et al., 2005
ATSDR, 2004
Ritz, 1999
Boice etal., 1999
Morgan etal., 1998
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-406 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-70. Selected results from epidemiologic studies of TCE exposure and
lung cancer (continued)
Exposure group
Medium/High
Relative risk
(95% CI)
1.07 (0.82, 1.40)
No. obs.
events
64
Cumulative
Referent
Low
High
1.00a
1.47(1.07,2.03)
0.96 (0.72, 1.29)
291
45
52
Aircraft maintenance workers (Hill Air Force Base, Utah)
TCE subcohort
Any TCE exposure
0.9 (0.6, 1.3)a
109
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.0 (0.7, 1.6)
0.9 (0.5, 1.6)
1.1 (0.7, 1.8)
51
43
23
38
Females, Cumulative exp
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
0.6(0.1,2.4)
0.6(0.1,4.7)
0.4(0.1, 1.8)
2
2
11
2
TCE subcohort
Any TCE exposure
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
0.83 (0.63, 1.08)
0.91 (0.67, 1.24)
1.0a
0.96 (0.67, 1.37)
0.71(0.46, 1.11)
1.00 (0.69, 1.45)
0.53 (0.27, 1.07)
1.0"
0.69 (0.27, 1.77)
0.65(0.16,2.73)
0.39(0.14, 1.11)
166
155
66
31
58
11
5
2
4
Cardboard manufacturing workers in Arnsburg, Germany
TCE-exposed workers
Unexposed workers
Deaths reported to GE pension fund (Pittsfield, MA)
1.38(0.55,2.86)
1.06 (0.34, 2.47)
1.01 (0.69, 1.47)d
7
5
139
U.S. Coast Guard employees
Marine inspectors
Noninspectors
0.52(0.31,0.82)
0.81(0.55,1.16)
18
30
Reference
Blair etal., 1998
Radican et al., 2008
Henschler et al., 1995
Greenland etal., 1994
Blair etal., 1989
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-407 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-70. Selected results from epidemiologic studies of TCE exposure and
lung cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Aircraft manufacturing employees (Italy)
All employees
0.99 (0.73, 1.32)
99
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
Lamp manufacturing workers (GE)
Rubber industry workers (Ohio)
0.80 (0.68, 0.95)
0.58 (0.27, 1.27)
0.64 (p> 0.05) c
138
6
11
Reference
Costa etal., 1989
Garabrant et al., 1988
Shannon et al., 1988
Wilcosky et al., 1984
Case-control studies
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
0.9 (0.6, 1.5)e
0.6 (0.3, 1.2)e
21
9
Siemiatyckietal., 1991
Geographic based studies
Two study areas in Endicott, NY
1.28 (0.99, 1.62)
68
Residents of 13 census tracts
InRedland, CA
0.71 (0.61, 0.81)f
356
Iowa residents with TCE in water supply
Males
<0.15 ug/L
>0.15 ug/L
343.1s
345.7g
1,181
299
Females
<0.15 ug/L
>0.15 ug/L
58.7s
47.8g
289
59
ATSDR, 2006
Morgan and Cassidy, 2002
Isacson et al., 1985
"Internal referents, workers not exposed to TCE.
bRisk ratio from Cox Proportional Hazard Analysis, stratified by age, sex, and decade (Environmental Health
Strategies, 1997).
0 Odds ratio from nested case-control study.
dOdds ratio from nested case-control analysis.
e90% confidence interval.
f99% confidence interval.
8Average annual age-adjusted incidence (per 100,000).
Lung cancer relative risks were reported in 11 of 12 cohort studies of aircraft
manufacturing, aircraft maintenance, aerospace, and metal workers, with potential exposure to
TCE as a degreasing agent, and in occupational cohort studies employing biological markers of
TCE exposures. All 11 studies had a high likelihood of TCE exposure in individual study
subjects and were judged to have met, to a sufficient degree, the standards of epidemiologic
design and analysis (Axelson et al., 1994; Greenland et al., 1994; Anttila et al., 1995; Blair et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-408 DRAFT—DO NOT CITE OR QUOTE
-------
1998; Morgan et al., 1998; Boice et al., 1999, 2006; Hansen et al., 2001; Raaschou-Nielsen et al.,
2003; Zhao et al., 2005; Radican et al., 2008). Lung cancer risks were not reported for Fernald
uranium processing workers with potential TCE exposure (Ritz, 1999), a study of less weight
than the other 11 studies.. The incidence study of Raaschou-Nielsen et al. (2003) was the largest
cohort, with 40,049 subjects identified as potentially exposed to TCE in several industries
(primarily, in the iron/metal and electronic industries), including 14,360 of whom had
presumably higher level exposures to TCE. The study included 632 lung cancer cases and
reported a 40% elevated incidence in TCE exposed males and females combined (95% CI: 1.32,
1.55), with no exposure duration gradient. The 95% confidence intervals in other studies of lung
cancer incidence included a risk ratio of 1.0 (Axelson et al., 1994; Anttila et al., 1995; Blair et
al., 1998; Hansen et al., 2001; Zhao et al., 2005). Lung cancer mortality risks in studies of TCE
exposure to aircraft manufacturing, aircraft maintenance, and aerospace workers included a
relative risk of 1.0 in their 95% confidence intervals (Boice et al., 2006; Zhao et al., 2005;
Morgan et al., 1998; Blair et al., 1998). Boice et al. (1999) observed a 24% decrement
(95% CI: 0.60, 0.95) for subjects with routine TCE exposure. Exposure-response analyses using
internal controls (unexposed subjects at the same company) showed a statistically significant
decreasing trend between lung cancer risk and routine or intermittent TCE exposure duration.
The routine or intermittent category is broader and includes more subjects with potential TCE
exposure.
The population studied by Garabrant et al. (1998), ATSDR (2004) and Chang et al.
(2005) are all employees (white- and blue-collar) at a manufacturing facility or plant with
potential TCE exposures. Garabrant et al. (1988) observed a 20% deficit in lung cancer
mortality (95% CI: 0.68, 0.95) in their study of all employees working for 4 or more years at an
aircraft manufacturing company. Blair et al. (1989), a study of Coast Guard marine inspectors
with potential for TCE exposure but lacking assessment to individual subjects, observed a 48%
deficit in lung cancer mortality (95% CI: 0.31, 0.82). Confidence intervals (95% CI) in Costa et
al. (1989), Chang et al. (2005) and ATSDR (2004) included a risk of 1.0. TCE exposure was not
known for individual subjects in these studies. A wide potential for TCE exposure is likely
ranging from subjects with little to no TCE exposure potential to those with some TCE exposure
potential. Exposure misclassification bias, typically considered as a negative bias, is likely
greater in these studies compared to studies adopting more sophisticated exposure assessment
approaches, which are able to assign quantitative exposure metrics to individual study subjects.
All three studies were of lower likelihood for TCE exposure, in addition to limited statistical
power and other design limitations, and these aspects, in addition to potential exposure
misclassification bias were alternative explanations of observed findings.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-409 DRAFT—DO NOT CITE OR QUOTE
-------
One population case-control study examined the relationship between lung cancer and
TCE exposure (Siemiatycki et al., 1991) with risk ratios of 0.9 (95% CI: 0.6, 1.5) for any TCE
exposure and 0.6 (95% CI: 0.3, 1.2) for substantial TCE exposure after adjustment for cigarette
smoking. TCE exposure prevalence in cases in this study was 2.5% for any exposure. Only 1%
had "substantial" (author's term) exposure, limiting the sensitivity of this study. Relative risks
above 2.0 could only be detected with sufficient (80%) statistical power. The finding of no
association of lung cancer with TCE exposure, therefore, is not surprising. One nested case-
control study of rubber workers observed a smoking unadjusted risk of 0.64 (95% CI: not
presented in paper) in those who had >1 year cumulative exposure to TCE (Wilcosky et al.,
1984).
Three geographic based studies reported lung cancer incidence or mortality risks for
drinking water contamination with TCE (Isacson et al., 1985; Morgan and Cassidy, 2002;
ATSDR, 2006). Morgan and Cassidy (2002) observed a relative risk of 0.71 (99% CI: 0.61,
0.81) for lung cancer among residents of Redlands County, CA, whose drinking water was
contaminated with TCE and perchlorate. However, ATSDR (2006) reported a 28% increase
(95% CI: 0.99, 1.62) in lung cancer incidence among residents living in a area in Endicott, NY,
whose drinking water was contaminated with TCE and other solvents. No information on
smoking patterns is available for individual lung cancer cases as identified by the New York
State Department of Health (NYS DOH) for other cancer cases in this study (ATSDR, 2008).
Isacson et al. (1985) presented lung cancer age-adjusted incidence rates for Iowa residents by
TCE level in drinking water supplies and did not observe an exposure-response gradient.
Exposure information is inadequate in all three of these studies, with monitoring data, if
available, based on few samples and for current periods only, and no information on water
distribution, consumption patterns, or temporal changes. Thus, TCE exposure potential to
individual subjects was not known with any precision, introducing misclassification bias, and
greatly limiting their ability to inform evaluation of TCE and lung cancer.
Laryngeal cancer risks are presented in a limited number of cohort studies involving TCE
exposure. No case-control or geographic based studies of TCE exposure were found in the
published literature. All but one of the cohort studies providing information on laryngeal cancer
observed less that 5 incident cases or deaths. Accordingly, these studies are limited for
examining the relationship between TCE exposure and laryngeal cancer. Risk ratios for
laryngeal cancer are found in Table 4-71.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-410 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-71. Selected results from epidemiologic studies of TCE exposure and
laryngeal cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers with TCE exposure
Not reported
Danish blue-collar worker w/TCE exposure
Any exposure, males
Any exposure, females
Employment duration
5yrs
1.2 (0.87, 1.52)
1.7(0.33,4.82)
Not reported
53
3
Biologically -monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-yr
Mean concentration (Ikeda)
<4 ppm
4+ppm
Employment duration
<6.25 yr
>6.25 yr
1.1(0.1,3.9)
Not reported
Not reported
Not reported
2
0
(0.1 exp)
Aircraft maintenance workers (Hill Air Force Base, Utah)
TCE subcohort
Any exposure
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Not reported
Not reported
Not reported
Zhao etal., 2005
Raaschou-Nielsen et al.,
2003
Hansen etal., 2001
Blair etal., 1998
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-411 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-71. Selected results from epidemiologic studies of TCE exposure and
laryngeal cancer (continued)
Exposure group
Biologically -monitored Finnish workers
Mean air-TCE (Ikeda extrapolation from
U-TCA)
<6 ppm
6+ppm
Relative risk
(95% CI)
Not reported
Not reported
No. obs.
events
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
1.39(0.17,5.00)
Not reported
2
Reference
Anttilaetal., 1995
Axelsonetal., 1994
Cohort and PMR -Mortality
Computer manufacturing workers (IBM), NY
Not reported
Aerospace workers (Rocketdyne)
Any TCE (utility or engine flush workers)
Engine flush — duration of exposure
Referent
0 yr (utility workers with TCE exposure)
<4yrs
>4yrs
Any exposure to TCE
View-Master employees
Males
Females
1.45(0.18,5.25)
Not reported
Not reported
Not reported
2
All employees at electronic factory (Taiwan)
Males
Females
0
0
(0.90 exp)
0
(0.23 exp)
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration4
Moderate TCE exposure, >2 yrs
duration4
Not reported
Not reported
Not reported
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent exposure
1.10(0.30,2.82)
Not reported
4
Clapp and Hoffman (2008)
Boice et al., 2006
Zhao etal., 2005
ATSDR, 2004
Chang et al., 2003
Ritz, 1999
Boice etal., 1999
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-412 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-71. Selected results from epidemiologic studies of TCE exposure and
laryngeal cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
Peak
No/low
Medium/high
Cumulative
Referent
Low
High
Not reported
Not reported
Not reported
Aircraft maintenance workers (Hill Air Force Base, Utah)
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Cardboard manufacturing workers in Arnsburg,
Germany
Deaths reported to GE pension fund (Pittsfield, MA)
Not reported
Not reported
Not reported
Not reported
Not examined
U. S. Coast Guard employees
Marine inspectors
Noninspectors
0.57(0.01,3.17)
0.58 (0.01, 3.20)
1
1
Aircraft manufacturing employees (Italy)
All employees
0.27 (0.03, 0.98)
2
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
0
(7.41 exp)
Reference
Morgan etal., 1998
Blair etal., 1998
Henschleretal., 1995
Greenland etal., 1994
Blair etal., 1989
Costa etal., 1989
Garabrantetal., 1988
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-413 DRAFT—DO NOT CITE OR QUOTE
-------
1 In summary, studies in humans examining lung and laryngeal cancer and TCE exposure
2 are inconclusive and do not support either a positive or a negative association between TCE
3 exposure and lung cancer or laryngeal cancer. Raaschou-Nielsen et al. (2003), with the largest
4 numbers of lung cancer cases of all studies, was the only one to observe a statistically
5 significantly elevated lung cancer risk with TCE exposure. Raaschou-Nielsen et al. (2003) also
6 noted several factors that may have confounded or biased their results in either a positive or
7 negative direction. This study and other cohort studies, as with almost any occupational study,
8 were not able to control confounding by exposure to chemicals other than TCE (although no
9 such chemical was apparent in the reports). Information available for factors related to
10 socioeconomic status (e.g., diet, smoking, alcohol consumption) was also not available. Such
11 information may positively confound smoking-related cancers such as lung cancer, particularly
12 in those studies, which adopted national rates to derive expected numbers of site-specific cancer,
13 if greater smoking rates were over-represented in blue-collar workers or residents of lower socio-
14 economic status. The finding of a larger risk among subjects with shortest exposure also argues
15 against a causal interpretation for the observed association for all subjects (NRC, 2006).
16 Four studies reported a statistically significant deficit in lung cancer incidence (Blair et
17 al., 1989; Garabrant et al., 1988; Boice et al., 1999; Morgan and Cassidy, 2002). Absence of
18 smoking information in these studies would introduce a negative bias if the studied population
19 smoked less than the referent population and may partially explain the lung cancer decrements
20 observed in these studies. Morgan and Cassidy (2002) noted the relatively high education high
21 income levels, and high access to health care of subjects in this study compared to the averages
22 for the county as a whole, likely leading to a lower smoking rate compared to their referent
23 population. Garabrant et al. (1988) similarly attributed their observations to negative selection
24 bias introduced when comparison is made to national mortality rates, also known as a "healthy
25 worker effect." The statistically significant decreasing trend in Boice et al. (1999) with exposure
26 duration to intermittent or routine exposure may reflect a protective effect between TCE and lung
27 cancer. The use of internal controls in this analysis reduces bias associated with use of an
28 external population who may have different smoking patterns than an employed population.
29 However, the exposure assessment approach in this study is limited due to inclusion of subjects
30 identified with intermittent TCE exposure (i.e., workers who would be exposed only during
31 particular shop runs or when assisting other workers during busy periods) (Boice et al., 1999).
32 The Boice et al. (1999) analysis is based on twice as many lung cancer deaths (i.e., 173 lung
33 cancer deaths) among subjects with routine or intermittent TCE exposure compared to only
34 routinely exposed subjects (78 deaths). Subjects identified as intermittently exposed are
35 considered as having a lower exposure potential than routinely exposed subject and their
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-414 DRAFT—DO NOT CITE OR QUOTE
-------
1 inclusion in exposure-response analyses may introduce exposure misclassification bias. Such
2 bias is a possible explanation for the decreasing trend observation, particularly if workers with
3 lower potential for TCE exposure have longer exposure (employment) durations.
4 Thus, a qualitative assessment suggests the epidemiological literature on respiratory
5 cancer and TCE is quite limited and has sufficient power to detect only large relative risks.
6 These studies can only rule out risks of a magnitude of 2.0 or greater for lung cancer and relative
7 risks greater than 3.0 or 4.0 for laryngeal cancer for exposures to studied populations. Therefore,
8 the database is limited in its ability to detect lung cancer associated with TCE exposure,
9 especially if the magnitude of response is similar to those observed for other endpoints.
10
11 4.7.2. Laboratory Animal Studies
12 4.7.2.1. Respiratory Tract Animal Toxicity
13 Limited studies are available to determine the effects of TCE exposure on the respiratory
14 tract (summarized in Table 4-72). Many of these studies in mice have examined acute effects
15 following intraperitoneal administration at relatively high TCE doses. However, effects on the
16 bronchial epithelium have been noted in mice and rats with TCE administered via gavage, with
17 doses 1,000 mg/kg/d and higher reported to cause rales and dyspnea (Narotsky et al., 1995) and
18 pulmonary vasculitis (NTP, 1990) in rats. Mice appear to be more sensitive than rats to
19 histopathological changes in the lung via inhalation; pulmonary effects are also seen in rats with
20 gavage exposure. It is difficult to compare intraperitoneal to oral and inhalation routes of
21 exposure given the risk of peritonitis and paralytic ileus. Any inflammatory response from this
22 route of administration can also affect the pulmonary targets of TCE exposure such as the Clara
23 cells.
24 This section reviews the existing literature on TCE, and the role of the various TCE
25 metabolites in TCE-induced lung effects. The most prominent toxic effect reported is damage to
26 Clara cells in mouse lung. The nonciliated, columnar Clara cells comprise the majority of the
27 bronchiolar and terminal bronchiolar epithelium in mice, and alveolar Type I and Type II cells
28 constitute the alveolar epithelium. These cells have been proposed as a progenitor of lung
29 adenocarcinomas in both humans and mice (Kim et al., 2005). Long-term studies have not
30 focused on the detection of pulmonary adenoma carcinomas but have shown a consistently
31 positive response in mice but not rats. However, chronic toxicity data on noncancer effects is
32 very limited.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-415 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-72. Animal toxicity studies of trichloroethylene
Reference
Green et al.,
1997
Forkert and
Forkert, 1994
Villaschi et al.,
1991
Odum et al.,
1992
Kurasawa, 1988
(translation)
Animals (sex)
CD-1 mice (F)
CD-1 mice (M)
BC3F1 mice
(M)
CD-1 mice (F)
Alpk APfSD
rats (F)
Ethanol-treated
(130) and
nontreated
(110)Wistar
rats (M)
Exposure
route
Inhalation
Intraperitoneal
injection
Single
inhalation
Inhalation
Inhalation
Inhalation
Dose/exposure concentration
450-ppm, 6 h/d, 5 d with 2 d
break then 5 more days;
sacrificed 1 8 h after 1 , 5, 6, or 1 0
exposures
2,000 mg/kg in corn oil (0.01
mL/g BW); sacrificed 15, 30, 60
and 90 d after single exposure
30 min 500, 1,000, 2,000, 3,500,
and 7,000 ppm; sacrificed 2 h,
24 h, 2, 5, or 7 d post exposure
6 h/d; separate repeated study in
mice: 450 ppm for 6 h/d, 5 d/wk
for 2 wks; sacrificed 24 h after
exposure; repeat study sacrificed
at 2, 5, 6, 8, 9, 12, or 13d; mice:
20, 100,200,450, 1 ,000, or2,000
ppm
6 h/d; repeat study sacrificed at 2,
5,6,8,9, 12, or 13d; rats: 500,
or 1,000 ppm
500, 1,000, 2,000, 4,000, and
8,000 ppm for 2 h; sacrificed 22 h
after exposure
Exposed
5/group
10/group
3/group
4/group
4/group
10/group
Results
Increased vacuolation and proliferation of
Clara cells caused by accumulation of
chloral.
Increased fibrotic lesions, with early signs
visible at 15 d postexposure.
Increased vacuolation and proliferation of
nonciliated bronchial cells. Injury was
maximal at 24 h with some repair occurring
between 24 h and 48 h.
Dose-dependent increase in Clara cell
vacuolation in mice after a single exposure,
resolved after 5 d repeated exposures but
recurred following a 2-d break from
exposure. Changes accompanied by
decrease in CYP activity in mice. Exposure
to chloral alone demonstrated similar
response as TCE exposure in mice. No
changes were seen in rats.
TCE exposure resulted in highly selective
damage to Clara cells that occurred
between 8 and 22 h after the highest
exposure with repair by 4 wks post
exposure.
-------
to
o
vo
Table 4-72. Animal toxicity studies of trichloroethylene (continued)
'
I
I
§
Co
1
TO'
•3
I
i
I
TO
Reference
Forkert et a\.,
2006
Forkert et a\.,
1985
Forkert and
Birch, 1989
Stewart et a\.,
1979;
Le Mesurier et
a\., 1980
Lewis, 1984
Scott et al.,
1988
Animals (sex)
CD-1 mice (M);
wild-type
(mixed 1 29/Sv
and C57BL)
andCYP2E1-
null mice (M)
CD-1 mice (M)
CD-1 mice (M)
Wistar Rats (F)
Mice
CD-1 mice (M)
Exposure
route
Intraperitoneal
injection
Intraperitoneal
injection
Intraperitoneal
injection
Inhalation
(whole body
chamber)
Inhalation
(Pyrex bell jars)
Intraperitoneal
injection
Dose/exposure concentration
500, 750, and 1 ,000 mg/kg in
corn oil; for inhibition studies mice
pretreated with 100 mg/kg diallyl
sulfone; for immunoblotting, 250,
500, 750, and 1 ,000 mg/kg; for
PNP hydroxylation, 50, 100, 250,
500, 750, and 1 ,000 mg/kg;
sacrificed 4 h after exposure
2,000, 2,500 or 3, 000 mg/kg in
mineral oil; sacrificed 24 h
postexposure for dose response;
time course sacrificed 1,2, 12,
and 24 h postexposure
2,000 mg/kg in corn oil; sacrificed
1,2,4,8, 12, and 24 h
postexposure
30 min, 48.5 g/m3 (9,030 ppm);
sacrificed at 5 and 1 5 d
postexposure
1 0,000 ppm, 1 -4 h daily for 5
consecutive days; sacrificed 24 h
after last exposure
single injection of 2,500-3,000
mg/kg, sacrificed 24 h
postexposure
Exposed
4/group
10/group
10/group
5/group
~28/group
4/group
Results
TCE bioactivation by CYP2E1 and/or 2F2
correlated with bronchiolar cytotoxicity in
mice.
Clara cell injury was increased following
exposure at all doses tested; time course
demonstrated a rapid and marked reduction
in pulmonary microsomal cytochrome P450
content and aryl hydrocarbon hydroxylase
activity. Alveolar Type II cells were also
affected.
Necrotic changes seen in Clara cells as
soon as 1 h postexposure; increased
vacuolation was seen by 4 h postexposure;
covalent binding of TCE to lung
macromolecules peaked at 4 h and reached
a plateau at 12 and 24 h post exposure.
Decreased recovery of pulmonary
surfactant (dose-dependent).
Increased vacuolation and reduced activity
of pulmonary mixed function oxidases.
Clara cells were damaged and exfoliated
from the epithelium of the lung.
H I
O >
HH Oq
H TO
O
H
W
-------
Table 4-72. Animal toxicity studies of trichloroethylene (continued)
Reference
NTP, 1990
Prendergast et
al., 1967
Narotsky et al.,
1995
Animals (sex)
F344 rats (M,F)
B6C3F1 mice
(M,F)
Sprague-
Dawley or
Long-Evans
rats; Hartley
Guinea pigs;
New Zealand
albino rabbits;
beagle dogs;
squirrel
monkeys (sex
not given for
any species)
F344 rats (F)
Exposure
route
Gavage
Inhalation
Gavage
Dose/exposure concentration
Male rats: 0, 125,250,500,
1 ,000, and 2,000 mg/kg BW (corn
oil); female rats: 0, 62.5, 125,
250, 500 or 1 ,000 mg/kg BW
(corn oil); Mice: 0, 375, 750,
1,500, 3,000, and 6,000 mg/kg
BW (corn oil); dosed 5d/w for
13wks
730 ppm for 8 h/d, 5 d/w, 6 wks or
35 ppm for 90 d constant
0, 1,125, 1,500mg/kg/d
Exposed
10/group
Rats (15);
guinea
pigs (15);
rabbit (3);
dog (2);
monkey
(3)
21, 16, or
17 per
group
Results
Increased pulmonary vasculitis in the
high-dose groups of male and female rats
(6/10 group as compared to 1/10 in
controls). No pulmonary effects described
in mice at this time point.
No histopathological changes observed,
although rats were described to show a
nasal discharge in the 6 wk study. No
quantification was given.
Rales and dyspnea were observed in the
TCE high-dose group; two females with
dyspnea subsequently died.
-------
1 4.7.2.1.1. Acute and short-term effects: inhalation. Relatively high-dose single and multiple
2 inhalation exposures to TCE result in dilation of endoplasmic reticulum and vacuolation of
3 nonciliated (Clara) cells throughout the bronchial tree in mice. A single study in rats reported
4 similar findings. In mice, single exposure experiments show vacuolation at all dose levels tested
5 with the extent of damage increasing with dose. Villaschi et al. (1991) reported similar degrees
6 of vacuolation in B6C3F1 mice (3/group) at 24 hours after the start of exposure across all tested
7 doses (500, 1,000, 2,000, 3,500, and 7,000 ppm, 30 minutes), with the percentage of the
8 nonciliated cells remaining vacuolated at 48 hours increasing with dose. Clara cell vacuolation
9 was reported to be resolved 7 days after single 30 minute exposure to TCE. Odum et al. (1992)
10 reported that, when observed 24 hours after the start of 6 hours exposure, the majority of Clara
11 cells in mice were unaffected at the lowest dose of 20 ppm exposures, while marked vacuolation
12 was observed at 200 ppm (no quantitative measures of damage given and only 3 animals per
13 group were examined).
14 In rats, Odum et al. (1992) reported no morphological changes in the female Alpk APfSD
15 rat epithelium after 6 hours exposure (500 or 1,000 ppm) when observed 24 hours after the start
16 of exposure (n = 3/group). However, Kurasawa reported pronounced dose-related morphological
17 changes in Clara cells at the highest dose (8,000 ppm) for 2 hours in Wistar rats (n = 10 per
18 group). At 500 and 1,000 ppm, slight dilation of the apical surface was reported, but
19 morphological measurements (the ratio of the lengths of the apical surface to that of the base line
20 of apical cytoplasm) were not statistically-significantly different from controls. From 2,000 to
21 8,000 ppm, a progressively increasing flattening of the apical surface was observed. In addition,
22 at 2,000 ppm, slight dilation of the smooth endoplasmic reticulum was also observed, with
23 marked dilation and possible necrosis at 8,000 ppm. Kurasawa (1988) also examined the time-
24 course of Clara cell changes following a single 8,000-ppm exposure, reporting the greatest
25 effects at 1 day to 1 week, repair at 2 weeks, and nearly normal morphology at 4 weeks. The
26 only other respiratory effect that has been reported from one study in rats exposed via inhalation
27 is a reduction in pulmonary surfactant yield following 30 minute exposures at 9,030 ppm for 5 or
28 15 days (Stewart et al., 1979). Therefore, single inhalation experiments (Villaschi et al., 1991;
29 Odum et al., 1992; Kurasawa, 1988) suggest that the Clara cell is the target for TCE exposure in
30 both rats and mice and that mice are more susceptible to these effects. However, the database is
31 limited in its ability to discern quantitative differences in susceptibility or the nature of the dose-
32 response after a single dose of TCE.
33 Other experiments examined the effects of several days of TCE inhalation exposure in
34 mice and potential recovery. While single exposures require 1 to 4 weeks for complete recovery,
35 after short-term repeated exposure, the bronchial epithelium in mice appears to either adapt to or
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-419 DRAFT—DO NOT CITE OR QUOTE
-------
1 become resistant to damage Odum et al. (1992) and Green et al. (1997) observed Clara cells in
2 mice to be morphologically normal at the end of exposures 6 hours/day for 4 or 5 days. As with
3 single dose experiments, the extent of recovery in multidose exposures may be dose-dependent.
4 Using a very high dose, Lewis et al. (1984) report vacuolation of bronchial epithelial cells after
5 4 hours/day, but not 1 hours/day, (10,000 ppm) for 5 days in mice. In addition, Odum et al.
6 (1992) reported that the damage to Clara cells that resolved after repeated exposures of 5 days, a
7 sign of adaptation to TCE exposure, returned when exposure was resumed after 2 days.
8 In rats, only one inhalation study reported in two published articles (Stewart et al., 1979;
9 Le Mesurier et al., 1979) using repeated exposures examined pulmonary histopathology.
10 Interestingly, this study reported vacuolation in Type 1 alveolar cells, but not in Clara cells, after
11 5 days of exposure to approximately 9,030 ppm for 30 minutes/day (only dose tested). In
12 addition, abnormalities were observed in the endothelium (bulging of thin endothelial segments
13 into the microcirculatory lumen) and minor morphological changes in Type 2 alveolar cells.
14 Although exposures were carried out for 5 consecutive days, histopathology was recorded up to
15 15 days post exposure, giving cell populations time to recover. Because earlier time points were
16 not examined, it is not possible to discern whether the lack of reported Clara cell damage in rats
17 following repeated exposure is due to recovery or lack of toxicity in this particular experiment.
18 Although recovery of individual damaged cells may occur, cell proliferation, presumed
19 from labeling index data suggestive of increased DNA synthesis, contributes, at least in part, to
20 the recovery of the bronchial epithelium in mice. Villaschi et al. (1991) observed a dose-
21 dependent increase in labeling index as compared to controls in the mouse lung at 48 hours after
22 a single TCE exposure (30 minutes; 500, 1,000, 2,000, 3,500, 7,000 ppm), which decreased to
23 baseline values at 7 days postexposure. Morphological analysis of cells was not performed,
24 although the authors stated the dividing cells had the appearance of Clara cells. Interestingly,
25 Green et al. (1997) reported no increase in BrdU labeling 24 hours after a single exposure
26 (6 hours 450 ppm), but did see increased BrdU labeling at the end of multiple exposures
27 (I/day, 5 days) while Villaschi et al. (1991) reported increased [3H]Thymidine labeling 2, 5, and
28 7 days after single 30 minute exposures to 500-7,000 ppm. Therefore, the data for single
29 exposures at 450-500 ppm may be consistent if increased cell proliferation occurred only for a
30 short period of time around 48 hours postexposure, and was thereby effectively washed-out by
31 the longer "averaging time" in the experiments by Green et al. (1997). Also, these contradictory
32 results may be due to differences in methodology. Green et al. (1997) and Villaschi et al. (1991)
33 reported very different control labeling indices (6 and 0%, respectively) while reporting similar
34 absolute labeling indices at 450-500 ppm (6.5 and 5.2%, respectively). The different control
35 values may be a result of substantially-different times over which the label was incorporated: the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-420 DRAFT—DO NOT CITE OR QUOTE
-------
1 mice in Green et al. (1997) were given BrdU via a surgically-implanted osmotic pump over
2 4 days prior to sacrifice, while the mice in Villaschi et al. (1991) were given a single
3 intraperitoneal dose of [3H]Thymidine 1 hour prior to sacrifice. Stewart et al. (1979) observed
4 no stimulation of thymidine incorporation after daily exposure to TCE (9,000 ppm) for up to
5 15 days. This study did, however, report a nonstatistically significant reduction in orotate
6 incorporation, an indicator of RNA synthesis, after 15 days, although the data was not shown.
7 At the biochemical level, changes in pulmonary metabolism, particularly with respect to
8 CYP activity, have been reported following TCE exposure via inhalation or intraperitoneal
9 administration in mice. Odum et al. (1992) reported reduced enzyme activity in Clara cell
10 sonicates of ethoxycoumarin O-deethylase, aldrin epoxidation, and nicotinamide adenine
11 dinucleotide phosphate-oxidase (NADPH) cytochrome c reductase after 6 hour exposures to
12 20-2,000 ppm TCE, although the reduction at 20 ppm was not statistically significant. No
13 reduction of GST activity as determined by chlorodinitrobenzene as a substrate was detected.
14 With repeated exposure at 450 ppm, the results were substrate-dependent, with ethoxycoumarin
15 O-deethylase activity remaining reduced, while aldrin epoxidation and NADPH cytochrome c
16 reductase activity showing some eventual recovery by 2 weeks. The results reported by Odum et
17 al. (1992) for NADPH cytochrome c reductase were consistent with those of Lewis et al. (1984),
18 who reported similarly reduced NADPH cytochrome c reductase activity following a much
19 larger dose of 10,000 ppm for 1 and 4 hours/day for 5 days in mice (strain not specified). TCE
20 exposure has also been associated with a decrease in pulmonary surfactant. Repeated exposure
21 of female Wistar rats to TCE (9,000 ppm, 30 minutes/day) for 5 or 15 days resulted in a
22 significant decrease in pulmonary surfactant as compared to unexposed controls
23 (Le Mesurier et al., 1980).
24
25 4.7.2.1.1.1. Acute and short-term effects: intraperitoneal injection and savage exposure. As
26 stated above the intraperitoneal route of administration is not a relevant paradigm for human
27 exposure. A number of studies have used this route of exposure to study the effects of acute
28 TCE exposure in mice. In general, similar lung targets are seen following inhalation or
29 intraperitoneal treatment in mice (Forkert et al., 2006, 1985; Forkert and Birch, 1989; Scott et al.,
30 1988). Inhalation studies generally reported the Clara cell as the target in mice. No lung
31 histopathology from intraperitoneal injection studies in rats is available. Forkert et al. (1985) and
32 Forkert and Birch (1989) reported vacuolation of Clara cells as soon as 1 hour following
33 intraperitoneal administration of a single dose of 2,000 mg/kg in mice. At 2,500 mg/kg, both
34 Forkert et al. (1985) and Scott et al. (1988) reported exfoliation of Clara cells and parenchymal
35 changes, with morphological distortion in alveolar Type II cells and inconsistently observed
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-421 DRAFT—DO NOT CITE OR QUOTE
-------
1 minor swelling in Type I cells at 24 hours postexposure. Furthermore, at 3,000 mg/kg,
2 Scott et al. (1988) also reported a significant (85%) decrease in intracellularly stored surfactant
3 phospholipids at 24 hours postexposure. These data indicate that both Clara cells and alveolar
4 Type I and II cells are targets of TCE toxicity at these doses and using this route of
5 administration. Recently, Forkert et al. (2006) reported Clara cell toxicity that showed increased
6 severity with increased dose (pyknotic nuclei, exfoliation) at 500-1,000 mg/kg intraperitoneal
7 doses as soon as 4 hours postexposure in mice. Even at 500 mg/kg, a few Clara cells were
8 reported with pyknotic nuclei that were in the process of exfoliation. Damage to alveolar Type II
9 cells was not observed in this dose range. The study by Scott et al. (1988) examined surfactant
10 phospholipids and phospholipase A2 activity in male CD-I mice exposed by intraperitoneal
11 injection of TCE (2,500 or 3,000 mg/kg, 24 hours). The lower concentration led to damage to
12 and exfoliation of Clara cells from the epithelial lining into the airway lumen, while only the
13 higher concentration led to changes in surfactant phospholipids. This study demonstrated an
14 increase in total phospholipid content in the lamellar body fractions in the mouse lung.
15 The study by Narotsky et al. (1995) exposed F344 timed-pregnant rats to TCE (0, 1,125,
16 and 1,500 mg/kg BW) by gavage and examined both systemic toxicity and developmental effects
17 at 14 days postexposure. Rales and dyspnea in the dams were observed in the high-dose group,
18 with two of the animals with dyspnea subsequently dying. The developmental effects observed
19 in this study are discussed in more detail in Section 4.8.
20
21 4.7.2.1.1.2. Subchronic and chronic effects. There are a few reports of the subchronic and
22 chronic noncancer effects of TCE on the respiratory system from intraperitoneal exposure in
23 mice and from gavage exposure in rats. Forkert and Forkert (1994) reported pulmonary fibrosis
24 in mice 90 days after intraperitoneal administration of a single 2,000 mg/kg dose of TCE. The
25 effects were in the lung parenchyma, not the bronchioles where Clara cell damage has been
26 observed after acute exposure. It is possible that fibrotic responses in the alveolar region occur
27 irrespective of where acute injury occurs. Effects upon Clara cells can also impact other areas of
28 the lung via cytokine regulation (Elizur et al., 2008). Alternatively, the alveolar and/or capillary
29 components of the lung may have been affected by TCE in a manner that was not
30 morphologically apparent in short-term experiments. In addition effects from a single or a few
31 short-term exposures may take longer to manifest. The latter hypothesis is supported by the
32 alveolar damage reported by Odum et al. (1992) after chloral administration by inhalation, and
33 by the adducts reported in alveolar Type II cells by Forkert et al. (2006) after 500-1,000 mg/kg
34 TCE intraperitoneal administration.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-422 DRAFT—DO NOT CITE OR QUOTE
-------
1 As noted previously, rats have responded to short-term inhalation exposures of TCE with
2 Clara cell and alveolar Type I and II effects. After repeated inhalation exposures over 6 weeks
3 (8 hours/day, 5 days/week, 730 ppm) and continuous exposures over 90 days (35 ppm),
4 Prendergast et al. (1967) noted no histopathologic changes in rats, guinea pigs, rabbits, dogs, or
5 monkeys after TCE exposure, but did describe qualitatively observing some nasal discharge in
6 the rats exposed for 6 weeks. The study details in Prendergast et al. (1967) are somewhat
7 limited. Exposed animals are described as "typically" 15 Long-Evans or Sprague-Dawley rats,
8 15 Hartley guinea pigs, 3 squirrel monkeys, 3 New Zealand albino rabbits, and 2 beagle dogs.
9 Controls were grouped between studies. In a 13-week NTP study in F344/N rats (n = 10/group)
10 exposed to TCE (0-2,000 mg/kg/d 5 days/week) by gavage, pulmonary vasculitis was observed
11 in 6/10 animals of each sex of the highest dose group (2,000 mg/kg/d), in contrast tol/10 in
12 control s of each sex (NTP, 1990).
13
14 4.7.2.2. Respiratory Tract Cancer
15 Limited studies have been performed examining lung cancer following TCE exposure
16 (summarized in Table 4-73). TCE inhalation exposure was reported to cause statistically
17 significant increase in pulmonary tumors (i.e., pulmonary adenocarcinomas) in some studies in
18 mice, but not in studies in rats and hamsters. Oral administration of TCE frequently resulted in
19 elevated lung tumor incidences in mice, but not in any tested species was there a statistically
20 significant increase. This section will describe the data regarding TCE induction of pulmonary
21 tumors in rodent models. The next sections will consider the role of metabolism and potential
22 MO As for inhalation carcinogenicity, primarily in mice.
23
24 4.7.2.2.1. Inhalation. There are three published inhalation studies examining the
25 carcinogenicity of TCE at exposures from 0-600 ppm, two of which reported statistically
26 significantly increased lung tumor incidence in mice at the higher concentrations (Fukuda et al.,
27 1983; Maltoni et al., 1986, 1988; Henschler et al., 1980). Rats and hamsters did not show an
28 increase in lung tumors following exposure.
29
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-423 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-73. Animal carcinogenicity studies of trichloroethylene
to
o
vo
Reference
Fukuda et a\.,
1983
Maltoni et al.,
1986, 1988
Henschleret
al., 1980
Henschleret
al., 1984
Van Duuren et
al., 1979
NCI, 1976
Animals
(sex)
ICR mice (F)
S-D rats (F)
S-D rats (M,
F) Swiss
mice (M, F)
B6C3F1 mice
(M, F)
Wistar rats
(M, F)
Syrian
hamsters (M,
F)
NMRI mice
Swiss mice
(M,F)
Swiss mice
(M,F)
Osborne-
Mendel rats
(M,F)
B6C3f1 mice
(M,F)
Exposure route
Inhalation, 7 h/d, 5
d/wk, 104 wk, hold
until 107 wk
Inhalation, 7 h/d, 5
d/wk, 104 wk, hold
until death
Inhalation, 6 h/d, 5
d/wk, 78 wks, hold
until 130 wk (mice
and hamsters) or
156wk(rats)
Gavage, 5/wk, 72
wkhold 104 wk
Gavage, 1/wk, 89
wk
Gavage, 5/wk, 78
wk, hold until 110
wk (rats) or 90 wk
(mice)
Dose/exp cone
(stabilizers, if any)
0, 50, 150, or 450 ppm
(epichlorohydrin)
0, 100, 300, or 600 ppm
0, 100, or 500 ppm
(triethanolamine)
2.4g/kgBW(M), 1.8
g/kg BW (F) all
treatments; (control,
triethanolamine,
industrial,
epichlorohydrin,
1 ,2-epoxybutane, both)
0 or 0.5 mg (unknown)
Rats: TWA: 0, 549, or
1 ,097 mg/kg
Mice: TWA: M: 0, 1,169,
or 2, 339 mg/kg; F: 0,
869, or 1 ,739 mg/kg
(epoxybutane,
epichlorohydrin)
Pulmonary tumor incidences
Benign+malignant
Mice: 6/49, 5/50, 13/50, 11/46;
Rats: 0/50, 0/50, 1/47, 1/51
Rats: 0/280, 0/260, 0/260,
0/260;
Swiss Mice: M: 10/90, 11/90,
23/90*, 27/90**; F: 15/90,
15/90, 13/90,20/90;
B6C3F1 Mice: M: 2/90, 2/90,
3/90,1/90; F: 4/90, 6/90,
7/90, 15/90*;
Rats: M: 1/29, 1/30, 1/30; F:
0/28; 1/30; 0/30;
Hamsters: 0/60, 0/59, 0/60;
Mice:M: 1/30,3/29, 1/30; F:
3/29, 0/30,1/28
Male: 18/50, 17/50, 14/50,
21/50, 15/50, 18/50;
Female: 12/50,20/50,21/50,
17/50, 18/50, 18/50
0/30 for all groups
Rats: M: 1/20, 0/50, 0/50; F:
0/20, 1/47, 0/50
Mice: M: 0/20, 5/50, 2/48; F:
1/20, 4/50, 7/47
Malignant only
Mice: 1/49; 3/50; 8/50*;
7/46*; Rats: none
Rats: 0/280, 0/260, 0/260,
0/260;
Swiss Mice: M: 0/90, 0/90,
0/90, 1/90; F: 2/90, 0/90,
0/90, 2/90;
B6C3F1 Mice M: 0/90, 0/90,
0/90, 0/90; F: 0/90, 1/90,
0/90, 0/90;
Rats: M: 1/29, 1/30, 1/30; F:
0/28; 1/30; 0/30;
Hamsters: 0/60, 0/59, 0/60;
Mice: M: 5/30, 3/29, 1/30; F:
1/29,3/30,0/28
Male: 8/50, 6/50, 7/50, 5/50,
7/50, 7/50;
Female: 5/50, 11/50,8/50,
3/50, 7/50, 7/50
0/30 for all groups
Rats: M: 0/20, 0/50, 0/50; F:
0/20, 1/47, 0/50
Mice: M: 0/20, 0/50, 1/48; F:
0/20, 2/50, 2/47
'
I
I
§
Co
1
TO'
•3
tl
"4
I
H I
O
Oq
O
H
W
-------
Table 4-73. Animal carcinogenicity studies of trichloroethylene (continued)
Reference
NTP, 1988
NTP, 1990
Maltoni et al.,
1986
Animals
(sex)
ACI, August,
Marshall,
Osborne-
Mendel rats
F344 rats (M,
F) B6C3F1
mice (M, F)
S-D rats (M,
F)
Exposure route
Gavage, 1/d, 5
d/wk, 103wk
Gavage, 1/day, 5
days/wk, 1 03 wk
Gavage, 1/d, 4-5
d/wk, 56 wk; hold
until death
Dose/exp cone
(stabilizers, if any)
0, 500, or 1 ,000 mg/kg
(diisopropylamine)
Mice: 0 or 1 ,000 mg/kg
Rats: 0,500, 1,000
mg/kg
0, 50 or 250 mg/kg
Pulmonary tumor incidences
Benign+malignant
ACI M: 1/50, 4/47, 0/46; F:
0/49, 2/47, 2/42
August M: 1/50, 1/50,0/49; F:
1/50, 1/50, 0/50
Marshall M: 3/49, 2/50, 2/47;
F: 3/49, 3/49, 1/46
Osborne-Mendel M: 2/50,
1/50, 1/50; F: 0/50, 3/50, 2/50
Mice: M: 7/49, 6/50; F: 1/48,
4/49
Rats: M: 4/50, 2/50, 3/49; F:
1/50, 1/49,4/50
M: 0/30, 0/30, 0/30; F: 0/30,
0/30, 0/30
Malignant only
ACI M: 1/50,2/47,0/46; F:
0/49, 1/47, 2/42
August M: 0/50, 1/50,0/49;
F: 1/50, 0/50, 0/50
Marshall M: 3/49, 2/50,
2/47; F: 3/49, 3/49, 1/46
Osborne-Mendel M: 1/50,
1/50, 0/50; F: 0/50, 3/50,
1/50
Mice: M: 3/49, 1/50; F: 1/48,
0/49
Rats: M: 3/50, 2/50, 3/49; F:
0/50, 0/49, 2/50
M: 0/30, 0/30, 0/30; F: 0/30,
0/30, 0/30
*Statistically-significantly different from controls by Fisher's exact test (p < 0.05).
**Statistically-significantry different from controls by Fisher's exact test (p < 0.01).
-------
1 The inhalation studies by Fukuda et al. (1983), which involved female ICR mice and
2 Sprague-Dawley rats, observed a 3-fold increase in lung tumors per mouse in those exposed to
3 the two higher concentrations (150-450 ppm) but reported no increase in lung tumors in the rats.
4 Maltoni et al. (1986, 1988) reported statistically-significantly increased pulmonary tumors in
5 male Swiss and female B6C3F1 mice at the highest dose of 600 ppm, but no significant increases
6 in any of the other species/strains/sexes tested. Henschler et al. (1980) tested NMRI mice,
7 Wistar rats and Syrian hamsters of both sexes, and reported no observed increase in pulmonary
8 tumors any of the species tested (see Section 4.4 and Appendix E for details of the conduct of
9 these studies).
10
11 4.7.2.2.2. Gavage. None of the six chronic gavage studies (Van Duuren et al., 1979; NCI,
12 1976; Henschler et al., 1984; NTP, 1988, 1990; Maltoni et al., 1986), which exposed multiple
13 strains of rats and mice to 0-3,000 mg/kg TCE for at least 56 weeks, reported a statistically-
14 significant excess in lung tumors, although nonstatistically-significant increases were frequently
15 observed in mice.
16 The study by Van Duuren et al. (1979) examined TCE along with 14 other halogenated
17 compounds for carcinogenicity in both sexes of Swiss mice. While no excess tumors were
18 observed, the dose rate of 0.5 mg once per week is equivalent to an average dose rate of
19 approximately 2.4 mg/kg/d for a mouse weighing 30 g, which is about 400-fold smaller than that
20 in the other gavage studies. In the NCI (1976) study, the results for Osborne-Mendel rats were
21 considered inconclusive due to significant early mortality, but female B6C3F1 mice (though not
22 males) exhibited a nonstatistically-significant elevation in pulmonary tumor incidence. The NCI
23 study (1976) used technical grade TCE which contained two known carcinogenic compounds as
24 stabilizers (epichlorohydrin and 1,2-epoxybutane), but a later study by Henschler et al. (1984) in
25 which mice were given TCE that was either pure, industrial, and stabilized with one or both of
26 these stabilizers found similar pulmonary tumors regardless of the presence of stabilizers. In this
27 study, female mice (n = 50) had elevated, but again not statistically-significant, increases in
28 pulmonary tumors. A later gavage study by NTP (1988), which used TCE stabilized with
29 diisopropylamine, observed no pulmonary tumors, but chemical toxicity and early mortality
30 rendered this study inadequate for determining carcinogenicity. The final NTP study (1990) in
31 male and female F344 rats and B6C3F1 mice, using epichlorohydrin-free TCE, again showed
32 early mortality in male rats. Similar to the other gavage studies, a nonstatistically significant
33 elevation in (malignant) pulmonary tumors was observed in mice, in this case in both sexes.
34 These animal studies show that while there is a limited increase in lung tumors following gavage
35 exposure to TCE in mice, the only statistically significant increase in lung tumors occurs
36 following inhalation exposure in mice.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-426 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.7.3. Role of Metabolism in Pulmonary Toxicity
2 TCE oxidative metabolism has been demonstrated to play a main role in TCE pulmonary
3 toxicity in mice. However, data are not available on the role of specific oxidative metabolites in
4 the lung. The Clara cell is thought to be the cell type responsible for much of the CYP
5 metabolism in the lung. Therefore, damage to this cell type would be expected to also affect
6 metabolism. More direct measures of CYP and isozyme-specific depression following TCE
7 exposure have been reported following intraperitoneal administration in mice. Forkert et al.
8 (1985) reported significant reduction in microsomal aryl hydrocarbon hydroxylase activity as
9 well as CYP content between 1 and 24 hours after exposure (2,000-3,000 mg/kg i.p. TCE).
10 Maximal depression occurred between 2 and 12 hours, with aryl hydrocarbon hydroxylase
11 activity (a function of CYP) less than 50% of controls and CYP content less than 20% of
12 controls. While there was a trend towards recovery from 12 to 24 hours, depression was still
13 significant at 24 hours. Forkert et al. (2005) reported decreases in immunoreactive CYP2E1,
14 CYP2F2, and CYP2B1 in the 4 hours after TCE treatment with 750 mg/kg intraperitoneal
15 injection in mice. The amount and time of maximal reduction was isozyme dependent
16 (CYP2E1: 30% of controls at 2 hours; CYP2F2: abolished at 30 minutes; CYP2B1: 43% of
17 controls at 4 hours). Catalytic markers for CYP2E1, CYP2F2, and CYP2B enzymes showed
18 rapid onset (15 minutes or less after TCE administration) of decreased activity, and continued
19 depression through 4 hours. Decrease in CYP2E1 and CYP2F2 activity (measured by PNP
20 hydroxylase activity) was greater than that of CYP2B (measured by pentoxyresorufin
21 O-dealkylase activity). Forkert et al. (2006) reported similar results in which 4 hours after
22 treatment, immunodetectable CYP2E1 protein was virtually abolished at doses 250-1,000 mg/kg
23 and immunodetectable CYP2F2 protein, while still detectable, was reduced. PNP hydroxylase
24 activity was also reduced 4 hours after treatment to 37% of controls at the lowest dose tested of
25 50 mg/kg, with further decreases to around 8% of control levels at doses of 500 mg/kg and
26 higher. These results correlate with previously described increases in Clara cell cytotoxicity, as
27 well as dichloroacetyl lysine (DAL) protein adduct formation. DAL adducts were observed in
28 the bronchiolar epithelium of CD-I mice and most prominent in the cellular apices of Clara cells
29 (Forkert et al., 2006). This study also examined the effect of TCE in vitro exposure on the
30 formation of chloral hydrate in lung microsomes from male CD-I mice and CYP2E1 knock-out
31 mice. The rates of CH formation were the same for lysosomes from both CD-I and CYP2E1
32 knockout mice from 0.25 mM to 0.75 mM, but the CH formation peaked earlier for in the wild-
33 type lysosomes (0.75 mM) as compared to CYP2El-null lysosomes (1 mM).
34 The strongest evidence for the necessary role of TCE oxidation is that pretreatment of
35 mice with diallyl sulfone (DASO2), an inhibitor of CYP2E1 and CYP2F2, protected against
36 TCE-induced pulmonary toxicity. In particular, following an intraperitoneal TCE dose of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-427 DRAFT—DO NOT CITE OR QUOTE
-------
1 750 mg/kg, Clara cells and the bronchiolar epithelium in mice pretreated with the
2 CYP2E1/CYP2F2 inhibitor appeared normal. In naive mice given the same dose, the epithelium
3 was attenuated due to exfoliation and there was clear morphological distortion of Clara cells
4 (Forkert et al., 2005). In addition, the greater susceptibility of mouse lungs relative to rat lungs
5 is consistent with their larger capacity to oxidize TCE, as measured in vitro in lung microsomal
6 preparations (Green et al., 1997). Analysis by immunolocalization also found considerably
7 higher levels of CYP2E1 in the mouse lung, heavily localized in Clara cells, as compared to rat
8 lungs, with no detectable CYP2E1 in human lung samples (Green et al., 1997). In addition, both
9 Green et al. (1997) and Forkert et al. (2006) report substantially lower metabolism of TCE in
10 human lung microsomal preparations than either rats or mice. It is clear that CYP2E1 is not the
11 only CYP enzyme involved in pulmonary metabolism, as lung microsomes from CYP2El-null
12 mice showed greater or similar rates of CH formation compared to those from wild-type mice.
13 Recent studies have suggested a role for CYP2F2 in TCE oxidative metabolism, although more
14 work is needed to make definitive conclusions. In addition, there may be substantial variability
15 in human lung oxidative metabolism, as Forkert et al. (2006) reported that in microsomal
16 samples from eight individuals, five exhibited no detectable TCE oxidation (<0.05 pmol/mg
17 protein/20 minutes), while others exhibited levels well above the limit of detection
18 (0.4-0.6 pmol/mg protein/minute).
19 In terms of direct pulmonary effects of TCE metabolites, Odum et al. (1992) reported that
20 mice exposed to 100 ppm via inhalation of chloral for 6 hours resulted in bronchiolar lesions
21 similar to those seen with TCE, although with a severity equivalent to 1,000 ppm TCE
22 exposures. In addition, some alveolar necrosis, alveolar oedema, and desquamation of the
23 epithelium were evident. In the same study, TCOH (100 and 500 ppm) also produced Clara cell
24 damage, but with lower incidence than TCE, and without alveolar lesions, while TCA treatment
25 produced no observable pulmonary effects. Therefore, it has been proposed that chloral is the
26 active metabolite responsible for TCE pulmonary toxicity, and the localization of damage to
27 Clara cells (rather than to other cell types, as seen with direct exposure to chloral) is due to the
28 localization of oxidative metabolism in that cell type (Odum et al., 1992; Green et al., 1997;
29 Green, 2000). However, the recent identification by Forkert et al. (2006) of DAL adducts, also
30 localized with Clara cell, suggests that TCE oxidation to dichloroacetyl chloride, which is not
31 believed to be derived from chloral, may also contribute to adverse health effects.
32 Due to the histological similarities between TCE- and chloral-induced pulmonary
33 toxicity, consistent with chloral being the active moiety, it has been proposed that the limited or
34 absent capacity for reduction of chloral (rapidly converted to CH in the presence of water) to
35 TCOH and glucuronidation of TCOH to TCOG in mouse lungs leads to "accumulation" of
36 chloral in Clara cells. However, the lack of TCOH glucuronidation capacity of Clara cells
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-428 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported by Odum et al. (1992), while possibly an important determinant of TCOH
2 concentrations, should have no bearing on CH concentrations, which depend on the production
3 and clearance of CH only. While isolated mouse Clara cells form smaller amounts of TCOH
4 relative to CH (Odum et al., 1992), the cell-type distribution of the enzymes metabolizing CH is
5 not clear. Indeed, cytosolic fractions of mouse, rat and human whole lungs show significant
6 activity for CH conversion to TCOH (Green et al., 1997). In particular, in mouse lung
7 subcellular fractions, 1 micromole of TCE in a 1.3 mL reactivial was converted to CH at a rate of
8 1 nmol/minute/mg microsomal protein, while 10 nmol CH in a 1.3 mL reactivial was converted
9 to TCOH at a rate of 0.24 nmol/minute/mg cytosolic protein (Green et al., 1997). How this
10 4-fold difference in activity would translate in vivo is uncertain given the 100-fold difference in
11 substrate concentrations, lack of information as to the concentration-dependence of activity, and
12 uncertain differences between cytosolic and microsomal protein content in the lung. It is unclear
13 whether local pulmonary metabolism of chloral is the primary clearance process in vivo, as in the
14 presence of water, chloral rapidly converts to chloral hydrate, which is soluble in water and
15 hence can rapidly diffuse to surrounding tissue and to the blood, which also has the capacity to
16 metabolize chloral hydrate (Lipscomb et al., 1996). Nonetheless, experiments with isolated
17 perfused lungs of rats and guinea pigs found rapid appearance of TCOH in blood following TCE
18 inhalation exposure, with no detectable chloral hydrate or TCOG (Dalbey and Bingham, 1978).
19 Therefore, it appears likely that chloral in the lung either is rapidly metabolized to TCOH, which
20 then diffuses to blood, or diffuses to blood as CH and is rapidly metabolized to TCOH by
21 erythrocytes (Lipscomb et al., 1996).
22 This hypothesis is further supported by in vivo data. No in vivo data in rats on CH after
23 TCE administration were located, and Fisher et al. (1998) reported CH in blood of human
24 volunteers exposed to TCE via inhalation were below detection limits. In mice, however, after
25 both inhalation and oral gavage exposure to TCE, CH has been reported in whole lung tissue at
26 concentrations similar to or somewhat greater than that in blood (Abbas and Fisher, 1997;
27 Greenberg et al., 1999). A peak concentration (1.3 |ig/g) of pulmonary CH was reported after
28 inhalation exposure to 600 ppm—at or above exposures where Clara cell toxicity was reported in
29 acute studies (Odum et al., 1992; Green et al., 1997). However, this was 5-fold less than the
30 reported pulmonary CH concentration (6.65 |ig/g) after gavage exposures of 1,200 mg/kg.
31 Specifically, a 600-ppm exposure or 450-ppm exposure reported in the Maltoni et al. and Fukuda
32 et al. studies results in a greater incidence in lung tumors than the 1,000-1,200 mg/kg/d
33 exposures in the NTP (1990) and NCI (1976) bioassays. However, the peak CH levels measured
34 in whole lung tissues after inhalation exposure to TCE at 600 ppm were reported to be about
35 5-fold lower than that at 1,200 mg/kg by gavage, therefore, showing the opposite pattern
36 (Greenberg et al., 1999; Abbas and Fisher, 1997). No studies of Clara cell toxicity after gavage
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-429 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposures were located, but several studies in mice administered TCE via intraperitoneal
2 injection did show Clara cell toxicity at around a dose of 750 mg/kg (Forkert et al., 2006) or
3 above (e.g., Forkert and Forkert, 1994; Forkert and Birch, 1989). However, as noted previously,
4 i.p. exposures are subject to an inflammatory response, confounding direct comparisons of dose
5 via other routes of administration.
6 Although, whole lung CH concentrations may not precisely reflect the concentrations
7 within specific cell types, as discussed above, the water solubility of CH suggests rapid
8 equilibrium between cell types and between tissues and blood. Both Abbas and Fisher (1997)
9 and Greenberg et al. (1999) were able to fit CH blood and lung levels using a PBPK model that
10 did not include pulmonary metabolism, suggesting that lung CH levels may be derived largely by
11 systemic delivery, i.e., from CH formed in the liver. However, a more detailed PBPK model-
12 based analysis of this hypothesis has not been performed, as CH is not included in the PBPK
13 model developed by Hack et al. (2006) that was updated in Section 3.5.
14 Two studies have reported formation of reactive metabolites in pulmonary tissues as
15 assessed by macromolecular binding after TCE intraperitoneal administration. Forkert and Birch
16 (1989) reported temporal correlations between the severity of Clara cell necrosis with increased
17 levels of covalent binding macromolecules in the lung of TCE or metabolites with a single
18 2,000 mg/kg dose of [14C] TCE. The amount of bound TCE or metabolites per gram of lung
19 tissue, DNA, or protein peaked at 4 hours and decreased progressively at 8, 12, and 24 hours.
20 The fraction of radioactivity in lung tissue macromolecules that was covalently bound reached a
21 plateau of about 20% from 4-24 hours, suggesting that clearance of total and covalently bound
22 TCE or metabolites was similar. The amount of covalent binding in the liver was 3- to 10-fold
23 higher than in the lung, although hepatic cytotoxicity was not apparent. This tissue difference
24 could either be due to greater localization of metabolism in the lung, so that concentrations
25 reactive metabolites in individual Clara cells are greater than both the lung as a whole and
26 hepatocytes, or because of greater sensitivity of Clara cells as compared to hepatocytes to
27 reactive metabolites. More recently, Forkert et al. (2006) examined DAL adducts resulting from
28 metabolism of TCE to dichloroacetyl chloride as an in vivo marker of production of reactive
29 metabolites. Following intraperitoneal administration of 500-1,000 mg/kg TCE in CD-I mice,
30 they found localization of DAL adducts believed to be from oxidative metabolism within Clara
31 cell apices, with dose-dependent increase in labeling with a polyclonal anti-DAL antibody that
32 correlated with increased Clara cell damage. Dose-dependent DAL adducts were also found in
33 alveolar Type II cells, although no morphologic changes in those cells were observed Both Clara
34 cell damage (as discussed above) and DAL labeling were abolished in mice pretreated with
35 DASC-2, an inhibitor of CYP2E1 and CYP2F2. However, Clara cell damage in treated CYP2E1-
36 null mice was more severe than in CD-I mice. Although DAL labeling was less pronounced in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-430 DRAFT—DO NOT CITE OR QUOTE
-------
1 CYP2El-null mice as compared to CD-I mice, this was due in part to the greater histopathologic
2 damage leading to attenuation of the epithelium and loss of Clara cells in the null mice. In
3 addition, protein immunoblotting with anti-DAL, anti-CYP2El and anti-CYP2F2 antibodies
4 suggested that a reactive TCE metabolite including dichloroacetyl chloride was formed that is
5 capable of binding to CYP2E1 and CYP2F2 and changing their protein structures. Follow-up
6 studies are needed in the lung and other target tissues to determine the potential role of the DAL
7 adducts in TCE-induced toxicity.
8 Finally, although Green (2000) and others have attributed species differences in
9 pulmonary toxicity to differences in the capacity for oxidative metabolism in the lung, it should
10 be noted that the concentration of the active metabolite is determined by both its production and
11 clearance (Clewell et al., 2000). Therefore, while the maximal pulmonary capacity to produce
12 oxidative metabolites is clearly greater in the mouse than in rats or humans, there is little
13 quantitative information as to species differences in clearance, whether by local chemical
14 transformation/metabolism or by diffusion to blood and subsequent systemic clearance. In
15 addition, existing in vitro data on pulmonary metabolism are at millimolar TCE concentrations
16 where metabolism is likely to be approaching saturation, so the relative species differences at
17 lower doses has not been characterized. Studies with recombinant CYP enzymes examined
18 species differences in the catalytic efficiencies of CYP2E1, CYP2F, and CYP2B1, but the
19 relative contributions of each isoform to pulmonary oxidation of TCE in vivo remains unknown
20 (Forkert et al., 2005). Furthermore, systemic delivery of oxidative metabolites to the lung may
21 contribute, as evidenced by respiratory toxicity reported with i.p. administration. Therefore,
22 while the differences between mice and rats in metabolic capacity are correlated with their
23 pulmonary sensitivity, it is not clear that differences in capacity alone are accurate quantitative
24 predictors of toxic potency. Thus, while it is likely that the human lung is exposed to lower
25 concentrations of oxidative metabolites, quantitative estimates for differential sensitivity made
26 with currently available data and dosimetry models are highly uncertain.
27 In summary, it appears likely that pulmonary toxicity is dependent on in situ oxidative
28 metabolism, however, the active agent has not been confidently identified. The similarities in
29 histopathologic changes in Clara cells between TCE and chloral inhalation exposure, combined
30 with the wider range of cell types affected by direct chloral administration relative to TCE, led
31 some to hypothesize that chloral is the toxic moiety in both cases, but with that generated in situ
32 from TCE in Clara cells "accumulating" in those cells (Green, 2000). However, chemical and
33 toxicokinetic data suggest that such "accumulation" is unlikely for several reasons. These
34 include the rapid conversion of chloral to chloral hydrate in the presence of water, the water
35 solubility of CH leading to rapid diffusion to other cell types and blood, the likely rapid
36 metabolism of chloral hydrate to TCOH either in pulmonary tissue or in blood erythrocytes, and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-431 DRAFT—DO NOT CITE OR QUOTE
-------
1 in vivo data showing lack of correlation across routes of exposure between whole-lung CH
2 concentrations and pulmonary carcinogenicity and toxicity. However, additional possibilities for
3 the active moiety exist, such as dichloroacetyl chloride, which is derived through a TCE
4 oxidation pathway independent of chloral and which appears to result in adducts with lysine
5 localized in Clara cells.
6
7 4.7.4. Mode of Action for Pulmonary Carcinogenicity
8 A number of effects have been hypothesized to be key events in the pulmonary
9 carcinogenicity of TCE, including cytotoxicity leading to increased cell proliferation, formation
10 of DAL protein adducts, and mutagenicity. As stated previously, the target cell for pulmonary
11 adenocarcinoma formation has not been established. Much of the hazard and MO A information
12 has focused on Clara cell effects from TCE which is a target in both susceptible and
13 nonsusceptible rodent species for lung tumors. However, the role of Clara cell susceptibility to
14 TCE-induced lung toxicity or to other potential targets such as lung stem cells that are activated
15 to repopulate both Clara and Type II alveolar cells after injury, has not been determined for
16 pulmonary carcinogenicity. While all of the events described above may be plausibly involved
17 in the MO A for TCE pulmonary carcinogenicity, none have been directly shown to be necessary
18 for carcinogenesis.
19
20 4.7 A.I. Mutagenicity via OxidativeMetabolism
21 The hypothesis is that TCE acts by a mutagenic MOA in TCE- induced lung tumors.
22 According to this hypothesis, the key events leading to TCE-induced lung tumor formation
23 constitute the following: the oxidative metabolism of TCE producing chloral/chloral hydrate
24 delivered to pulmonary tissues, causes direct alterations to DNA (e.g., mutation, DNA damage,
25 and/or micronuclei induction). Mutagenicity is a well-established cause of carcinogenicity.
26
27 4.7.4.1.1. Experimental support for the hypothesized mode of action. Pulmonary toxicity has
28 been proposed to be dependent on in situ oxidative metabolism, however, the active agent has
29 not been confidently identified. The similarities in histopathologic changes in Clara cells
30 between TCE and chloral inhalation exposure, combined with the wider range of cell types
31 affected by direct chloral administration relative to TCE, led some to hypothesize that chloral is
32 the toxic moiety. Chloral that is formed from the metabolism of TCE is quickly converted to CH
33 upon hydration under physiological conditions. As discussed in Section 4.2.4, CH clearly
34 induces aneuploidy in multiple test systems, including bacterial and fungal assays in vitro (Kafer,
35 1986; Kappas, 1989; Crebelli et al., 1991), mammalian cells in vitro (Vagnarelli et al., 1990;
36 Sbrana et al., 1993), and mammalian germ-line cells in vivo (Russo et al., 1984; Miller and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-432 DRAFT—DO NOT CITE OR QUOTE
-------
1 Adler, 1992). Conflicting results were observed in in vitro and in vivo mammalian studies of
2 micronuclei formation (Degrassi and Tanzarella, 1988; Nesslany and Marzin, 1999; Russo and
3 Levis, 1992a, b; Giller et al., 1995; Beland, 1999), with positive results in germ-line cells
4 (Nutley et al., 1996; Allen et al., 1994). In addition, it is mutagenic in the Ames bacterial
5 mutation assay for some strains (Haworth et al., 1983; Ni et al., 1994; Beland, 1999; Giller et al.,
6 1995). Structurally related chlorinated aldehydes 2-chloroacetyaldehyde and
7 2,2-dichloroacetaldehyde are both alkylating agents, are both positive in a genotoxic assay
8 (Bignami et al., 1980), and both interact covalently with cellular macromolecules
9 (Guengerich et al., 1979).
10 As discussed in the section describing the experimental support for the mutagenic MOA
11 for liver carcinogenesis (see Section 4.5.7.1), it has been argued that CH mutagenicity is unlikely
12 to be the cause of TCE carcinogenicity because the concentrations required to elicit these
13 responses are several orders of magnitude higher that achieved in vivo (Moore and Harrington-
14 Brock, 2000). Similar to the case of the liver, it is not clear how much of a correspondence is to
15 be expected from concentrations in genotoxicity assays in vitro and concentrations in vivo, as
16 reported in vivo CH concentrations are in whole lung homogenate while in vitro concentrations
17 are in culture media. None of the available in vivo genotoxicity assays used the inhalation route
18 that elicited the greatest lung tumor response under chronic exposure conditions, so direct in vivo
19 comparisons are not possible. Finally, as discussed in Section 4.5.7.1, the use of i.p.
20 administration in many other in vivo genotoxicity assays complicates the comparison with
21 carcinogenicity data.
22 As discussed above (see Section 4.7.3), chemical and toxicokinetic data are not
23 supportive of CH being the active agent of TCE-induced pulmonary toxicity, and directly
24 contradict the hypothesis of chloral "accumulation." Nonetheless, CH has been measured in the
25 mouse lung following inhalation and gavage exposures to TCE (Abbas and Fisher, 1997;
26 Greenberg et al., 1999), possibly the result of both in situ production and systemic delivery.
27 Therefore, in principle, CH could cause direct alterations in DNA in pulmonary tissue.
28 However, as discussed above, the relative amounts of CH measured in whole lung tissue from
29 inhalation and oral exposures do not appear to correlate with sensitivity to TCE lung tumor
30 induction across exposure routes. While these data cannot rule out a role for mutagenicity
31 mediated by CH due to various uncertainties, such as whether whole lung CH concentrations
32 accurately reflect cell-type specific concentrations and possible confounding due to strain
33 differences between inhalation and oral chronic bioassays, they do not provide support for this
34 MOA.
35 Additional possibilities for the active moiety exist, such as dichloroacetyl chloride, which
36 is derived through a TCE oxidation pathway independent of chloral and which appears to result
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-433 DRAFT—DO NOT CITE OR QUOTE
-------
1 in adducts with lysine localized in Clara cells (Forkert et al., 2006). DCA, which has some
2 genotoxic activity, is, also, presumed to be formed through this pathway (see Section 3.3).
3 Currently, however, there are insufficient data to support a role for these oxidative metabolites in
4 a mutagenic MOA.
5
6 4.7.4.2. Cytotoxicity Leading to Increased Cell Proliferation
1 The hypothesis is that TCE acts by a cytotoxicity MOA in TCE-induced pulmonary
8 carcinogenesis. According to this hypothesis, the key events leading to TCE-induced lung tumor
9 formation constitute the following: TCE oxidative metabolism in situ leads to currently unknown
10 reactive metabolites that cause cytotoxicity, leading to compensatory cellular proliferation and
11 subsequently increased mutations and clonal expansion of initiated cells.
12
13 4.7.4.2.1. Experimental support for the hypothesized mode of action. Evidence for the
14 hypothesized MOA consists primarily of (1) the demonstration of acute cytotoxicity and
15 transient cell proliferation following TCE exposure in laboratory mouse studies; (2) toxicokinetic
16 data supporting oxidative metabolism being necessary for TCE pulmonary toxicity; (3) the
17 association of lower pulmonary oxidative metabolism and lower potency for TCE-induced
18 cytotoxicity with the lack of observed pulmonary carcinogenicity in laboratory rats. However,
19 there is a lack of experimental support linking TCE acute pulmonary cytotoxicity to sustained
20 cellular proliferation of chronic exposures or clonal expansion of initiated cells.
21 As discussed above, a number of acute studies have shown that TCE is particularly
22 cytotoxic to Clara cells in mice, which has been suggested to be involved in the development of
23 mouse lung tumors (Buckpitt et al., 1995; Forkert and Forkert, 1994, Kim et al., 2005). In
24 addition, studies examining cell labeling by either BrdU (Green et al., 1997) or 3H-thymidine
25 incorporation (Villaschi et al., 1991) suggest increased cellular proliferation in mouse Clara cells
26 following acute inhalation exposures to TCE. Moreover, in short-term studies, Clara cells appear
27 to become resistant to cytotoxicity with repeated exposure, but regain their susceptibility after
28 2 days without exposure. This observation led to the hypothesis that the 5 day/week inhalation
29 dosing regime (Fukuda et al., 1983; Maltoni et al., 1986, 1988; Henschler et al., 1980) in the
30 chronic mouse studies leads to periodic cytotoxicity in the mouse lung at the beginning of each
31 week followed by cellular regeneration, and that the increased rate of cell division leads to
32 increased incidence of tumors by increasing the overall mutation rate and by increasing the
33 division rate of already initiated cells (Green, 2000). However, longer-term studies to test this
34 hypothesis have not been carried out.
35 As discussed above (see Section 4.7.3), there is substantial evidence that pulmonary
36 oxidative metabolism is necessary for TCE-induced pulmonary toxicity, although the active
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-434 DRAFT—DO NOT CITE OR QUOTE
-------
1 moiety remains unknown. In addition, the lower capacity for pulmonary oxidative metabolism
2 in rats as compared to mice is consistent with studies in rats not reporting pulmonary cytoxicity
3 until exposures higher than those in the bioassays, and the lack of reported pulmonary
4 carcinogenicity in rats at similar doses to mice. However, rats also have a lower background rate
5 of lung tumors (Green, 2000), and so would be less sensitive to carcinogenic effects in that tissue
6 to the extent that relative risks is the important metric across species. In addition, this MOA
7 hypothesis requires a number of additional key assumptions for which there are currently no
8 direct evidence. First, the cycle of cytotoxicity, repair, resistance to toxicity, and loss of
9 resistance after exposure interruption, has not been documented and under the proposed MOA
10 should continue under chronic exposure conditions. This cycle has thus, far only been observed
11 in short term (up to 13-day) studies. In addition, although Clara cells have been identified as the
12 target of toxicity whether they or endogenous stem cells in the lung are the cells responsible for
13 mouse lung tumors has not been established. There is currently no data as to the cell type of
14 origin for TCE-induced lung tumors.
15
16 4.7.4.3. Additional Hypothesized Modes of Action with Limited Evidence or Inadequate
17 Experimental Support
18 4.7.4.3.1. Role of formation of DAL protein adducts. As discussed above, Forkert et al.
19 (2006) recently observed dose-dependent formation of DAL protein adducts in the Clara cells of
20 mice exposed to TCE via intraperitoneal injection. While adducts were highly localized in Clara
21 cells, they were also found in alveolar Type II cells, though these cells did not show signs of
22 cytotoxicity in this particular experimental paradigm. In terms of the MOA for TCE-induced
23 pulmonary carcinogenicity, these adducts may either be causally important in and of themselves,
24 or they may be markers of a different causal effect. For instance, it is possible that these adducts
25 are a cause for the observed Clara cell toxicity, and Forkert et al. (2006) suggested that the lack
26 of toxicity in alveolar Type II cells may indicate that "there may be a threshold in adduct
27 formation and hence bioactivation at which toxicity is manifested." In this case, they are an
28 additional precursor event in the same causal pathway proposed above. Alternatively, these
29 adducts may be indicative of effects related to carcinogenesis but unrelated to cytotoxicity. In
30 this case, the Clara cell need not be the cell type of origin for mouse lung tumors.
31 Because of their recent discovery, there is little additional data supporting, refuting, or
32 clarifying the potential role for DAL protein adducts in the MOA for TCE-induced pulmonary
33 carcinogenesis. For instance, the presence and localization of such adducts in rats has not been
34 investigated, and could indicate the extent to which the level of adduct formation is correlated
35 with existing data on species differences in metabolism, cytotoxicity, and carcinogenicity. In
36 addition, the formation of these adducts has only been investigated in a single dose study using
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-43 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 i.p. injection. As stated above, i.p. injection may involve the initiation of a systemic
2 inflammatory response that can activate lung macrophages or affect Clara cells. Experiments
3 with repeated exposures over chronic durations and by inhalation or oral of administration would
4 be highly informative. Finally, the biological effects of these adducts, whether cytotoxicity or
5 something else, have not been investigated.
6
7 4.7.4.4. Conclusions About the Hypothesized Modes of Action
8 4.7.4.4.1. (1) Is the hypothesized mode of action sufficiently supported in the test animals?
9 4.7.4.4.1.1. Mutasenicity. Chloral hydrate is clearly genotoxic, as there are substantial data
10 from multiple in vitro and in vivo assays supporting its ability induce aneuploidy, with more
11 limited data as to other genotoxic effects, such as point mutations. Chloral hydrate is also clearly
12 present in pulmonary tissues of mice following TCE exposures similar to those inducing lung
13 tumors in chronic bioassays. However, chemical and toxicokinetic data are not supportive of CH
14 being the predominant metabolite for TCE carcinogenicity. Such data include the water
15 solubility of CH leading to rapid diffusion to other cell types and blood, its likely rapid
16 metabolism to TCOH either in pulmonary tissue or in blood erythrocytes, and in vivo data
17 showing lack of correlation across routes of exposure between whole lung CH concentrations
18 and pulmonary carcinogenicity. Therefore, while a role for mutagenicity via CH in the MOA of
19 TCE-induced lung tumors cannot be ruled about, available evidence is inadequate to support the
20 conclusion that direct alterations in DNA caused by CH produced in or delivered to the lung after
21 TCE exposure constitute a MOA for TCE-induced lung tumors.
22
23 4.7.4.4.1.2. Cytotoxicity. The MOA hypothesis for TCE-induced lung tumors involving
24 cytotoxicity is supported by relatively consistent and specific evidence for cytotoxicity at
25 tumorigenic doses in mice. However, the majority of cytotoxicity-related key events have been
26 investigated in studies less than 13 days, and none has been shown to be causally related to TCE-
27 induced lung tumors. In addition, the cell type (or types) of origin for the observed lung tumors
28 in mice has not been determined, so the contribution to carcinogenicity of Clara cell toxicity and
29 subsequent regenerative cell division is not known. Similarly, the relative contribution from
30 recently discovered dichloroacetyl-lysine protein adducts to the tumor response has not been
31 investigated and has currently only been studied in i.p. exposure paradigms of short duration. In
32 summary, while there are no data directly challenging the hypothesized MOA described above,
33 the existing support for their playing a causal role in TCE-induced lung tumors is largely
34 associative, and based on acute or short term studies. Therefore, there are inadequate data to
35 support a cytotoxic MOA based on the TCE-induced cytotoxicity in Clara cells in the lungs of
36 test animals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-436 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.7.4.4.1.3. Additional hypothesis. Inadequate data are available to develop a MOA hypothesis
2 based on recently discovered DAL adducts induced by TCE inhalation and i.p. exposures. It will
3 therefore, not be considered further in the conclusions below.
4 Overall, therefore, the MOA for TCE-induced lung tumors is considered unknown at this
5 time.
6
7 4.7.4.4.2. (2) Is the hypothesized mode of action relevant to humans?
8 4.7.4.4.2.1. Mutasenicity. The evidence discussed above demonstrates that CH is mutagenic in
9 microbial as well as test animal species. There is therefore, the presumption that they would be
10 mutagenic in humans. Therefore, this MOA is considered relevant to humans.
11
12 4.7.4.4.2.2. Cytotoxicity. No data from human studies are available on the cytotoxicity of TCE
13 and its metabolites in the lung, and no causal link between cytotoxi city and pulmonary
14 carcinogenicity has been demonstrated in animal or human studies. Nonetheless, in terms of
15 human relevance, no data suggest that the proposed key events are not biologically plausible in
16 humans, therefore, qualitatively, TCE-induced lung tumors are considered relevant to humans.
17 Information about the relative pharmacodynamic sensitivity between rodents and humans is
18 absent, but information on pharmacokinetic differences in lung oxidative metabolism does exist
19 and will be considered in dose-response assessment when extrapolating between species (see
20 Section 5.2.1.2).
21
22 4.7.4.4.3. (3) Which populations or lifestages can be particularly susceptible to the
23 hypothesized mode of action ?
24 4.7.4.4.3.1. Mutasenicity. The mutagenic MOA is considered relevant to all populations and
25 lifestages. According to U.S. EPA's Cancer Guidelines (U.S. EPA, 2005a) and Supplemental
26 Guidance (U.S. EPA, 2005b), there may be increased susceptibility to early-life exposures for
27 carcinogens with a mutagenic mode of action. However, because the weight of evidence is
28 inadequate to support a mutagenic MOA for TCE pulmonary carcinogenicity, and in the absence
29 of chemical-specific data to evaluate differences in susceptibility, the ADAFs should not be
30 applied, in accordance with the Supplemental Guidance.
31
32 4.7.4.4.3.2. Cytotoxicity. No information based is available as to which populations or
33 lifestages may be particularly susceptible to TCE-induced lung tumors. However,
34 pharmacokinetic differences in lung oxidative metabolism among humans do exist, and because
35 of the association between lung oxidative metabolism and toxicity, will be considered in dose-
36 response assessment when extrapolating within species.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-437 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.7.5. Summary and Conclusions
2 The studies described here show pulmonary toxicity found mainly in Clara cells in mice
3 (Green et al., 1997; Villaschi et al., 1991; Odum et al., 1992; Forkert et al., 1985; Forkert and
4 Birch, 1989) and rats (Kurasawa, 1988). The most convincing albeit limited data regarding this
5 type of toxicity was demonstrated predominantly in mice exposed via inhalation, although some
6 toxicity was shown in intraperitoneal injection studies. Increased vacuolation of Clara cells was
7 often seen within the first 24-hours-of-exposure, depending on dose, but with cellular repair
8 occurring within days or weeks of exposure. Continued exposure led to resistance to TCE-
9 induced Clara cell toxicity, but damage recurred if exposure was stopped after 5 days and then
10 resumed after 2 days without exposure. However, Clara cell toxicity has only been observed in
11 acute and short-term studies, and it is unclear whether they persist with subchronic or chronic
12 exposure, particularly in mice, which are the more sensitive species. With respect to pulmonary
13 carcinogenicity, statistically-significantly increased incidence of lung tumors from chronic
14 inhalation exposures to TCE was observed female ICR mice (Fukuda et al., 1983), male Swiss
15 mice, and female B6C3F1 mice (Maltoni et al., 1986), though not in other sex/strain
16 combinations, nor in rats (Henschler et al., 1980; Maltoni et al., 1986). However, lung toxicity
17 and Clara cell effects have also been observed in rats. Overall, the limited carcinogenesis studies
18 described above are consistent with TCE causing mild increases in pulmonary tumor incidence
19 in mice, but not in other species tested such as rats and hamsters.
20 The epidemiologic studies are quite limited for examining the role of TCE in cancers of
21 the respiratory system, with no studies found on TCE exposure specifically examining toxicity of
22 the respiratory tract. The two studies found on organic solvent exposure which included TCE
23 suggested smoking as a primary factor for observed lung function decreases among exposed
24 workers. Animal studies have demonstrated toxicity in the respiratory tract, particularly damage
25 to the Clara cells (nonciliated bronchial epithelial cells), as well as decreases in pulmonary
26 surfactant following both inhalation and intraperitoneal exposures, especially in mice. Dose-
27 related increases in vacuolation of Clara cells have been observed in mice and rats as early as
28 24 hours postexposure (Odum et al., 1992; Kurasawa, 1988; Forkert et al., 1985, 2006; Forkert
29 and Birch, 1989; Scott et al., 1988). Mice appear to be more sensitive to these changes, but both
30 species show a return to normal cellular morphology at four weeks postexposure (Odum et al.,
31 1992). Studies in mice have also shown an adaptation or resistance to this damage after only 4 to
32 5 days of repeated exposures (Odum et al., 1992; Green et al., 1997). The limited
33 epidemiological literature on lung and laryngeal cancer in TCE-exposed groups is inconclusive
34 due to study limitations (low power, null associations, confidence intervals on relative risks that
35 include 1.0). These studies can only rule out risks of a magnitude of 2.0 or greater for lung
36 cancer and relative risks greater than 3.0 or 4.0 for laryngeal cancer for exposures to studied
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-43 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 populations and thus, may not detect a level of response consistent with other endpoints. Animal
2 studies demonstrated a statistically significant increase in pulmonary tumors in mice following
3 chronic inhalation exposure to TCE (Fukuda et al., 1983; Maltoni et al., 1988, 1986). These
4 results were not seen in other species tested (rats, hamsters; Maltoni et al., 1986, 1988; Fukuda et
5 al., 1983; Henschler et al., 1980). By gavage, elevated, but not statistically significant,
6 incidences of benign and/or malignant pulmonary tumors have been reported in B6C3F1 mice
7 (NCI, 1976; Henschler et al., 1984; NTP, 1990). No increased pulmonary tumor incidences have
8 been reported in rats exposed to TCE by gavage (NCI, 1976; NTP, 1988, 1990), although all the
9 studies suffered from early mortality in at least one sex of rat.
10 Although no epidemiologic studies on the role of metabolism of TCE in adverse
11 pulmonary health effects have been published, animal studies have demonstrated the importance
12 of the oxidative metabolism of TCE by CYP2E1 and/or CYP2F2 in pulmonary toxicity.
13 Exposure to diallyl sulfone (DASC^), an inhibitor of both enzymes protects against pulmonary
14 toxicity in mice following exposure to TCE (Forkert et al., 2005). The increased susceptibility in
15 mice correlates with the greater capacity to oxidize TCE based on increased levels of CYP2E1 in
16 mouse lungs relative to lungs of rats and humans (Green et al., 1997; Forkert et al., 2006), but it
17 is not clear that these differences in capacity alone are accurate quantitative predictors of
18 sensitivity to toxicity. In addition, available evidence argues against the previously proposed
19 hypothesis (e.g., Green, 2000) that "accumulation" of chloral in Clara cells is responsible for
20 pulmonary toxicity, since chloral is first converted the water-soluble compounds chloral hydrate
21 and TCOH that can rapidly diffuse to surrounding tissue and blood. Furthermore, the
22 observation of DAL protein adducts, likely derived dichloroacetyl chloride and not from chloral,
23 that were localized in Clara cells suggests an alternative to chloral as the active moiety. While
24 chloral hydrate has shown substantial genotoxic activity, chemical and toxicokinetic data on CH
25 as well as the lack of correlation across routes of exposure between in vivo measurements of CH
26 in lung tissues and reported pulmonary carcinogenicity suggest that evidence is inadequate to
27 conclude that a mutagenic MOA mediated by CH is operative for TCE-induced lung tumors.
28 Another MOA for TCE-induced lung tumors has been plausibly hypothesized to involve
29 cytotoxicity leading to increased cell proliferation, but the available evidence is largely
30 associative and based on short-term studies, so a determination of whether this MOA is operative
31 cannot be made. The recently discovered formation of DAL protein adducts in pulmonary
32 tissues may also play a role in the MOA of TCE-induced lung tumors, but an adequately defined
33 hypothesis has yet to be developed. Therefore, the MOA for TCE-induced lung tumors is
34 currently considered unknown, and this endpoint is thus, considered relevant to humans.
35 Moreover, none of the available data suggest that any of the currently hypothesized mechanisms
36 would be biologically precluded in humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-439 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8. REPRODUCTIVE AND DEVELOPMENTAL TOXICITY
2 4.8.1. Reproductive Toxicity
3 An assessment of the human and experimental animal data, taking into consideration the
4 overall weight of the evidence, demonstrates a concordance of adverse reproductive outcomes
5 associated with TCE exposures. Effects on male reproductive system integrity and function are
6 particularly notable and are discussed below. Cancers of the reproductive system in both males
7 and females have also been identified and are discussed below.
8
9 4.8.1.1. Human Reproductive Outcome Data
10 A number of human studies have been conducted that examined the effects of TCE on
11 male and female reproduction following occupational and community exposures. These are
12 described below and summarized in Table 4-74. Epidemiological studies of female human
13 reproduction examined infertility and menstrual cycle disturbances related to TCE exposure.
14 Other studies of exposure to pregnant women are discussed in the section on human
15 developmental studies (see Section 4.8.2.1). Epidemiological studies of male human
16 reproduction examined reproductive behavior, altered sperm morphology, altered endocrine
17 function, and infertility related to TCE exposure.
18
19 4.8.1.1.1. Female and male combined human reproductive effects.
20 Reproductive behavior. A residential study of individuals living near the Rocky Mountain
21 Arsenal in Colorado examined the reproductive outcomes in 75 men and 71 women exposed to
22 TCE in drinking water (ATSDR, 2001). TCE exposure was classified as high (>10.0 ppb),
23 medium (>5.0 to <10.0 ppb), and low (<5.0 ppb). Altered libido for men and women combined
24 was observed in a dose-response fashion, although the results were nonsignificant. The results
25 were not stratified by gender.
26
27 4.8.1.1.2. Female human reproductive effects.
28 4.8.1.1.2.1. Infertility. Sallmen et al. (1995) examined maternal occupational exposure to
29 organic solvents and time-to-pregnancy. Cases of spontaneous abortion and controls from a
30 prior study of maternal occupational exposure to organic solvents in Finland during 1973-1983
31 and pregnancy outcome (Lindbohm et al., 1990) were used to study time-to-pregnancy of
32 197 couples. Exposure was assessed by questionnaire during the first trimester and confirmed
33 with employment records. Biological measurements of TCA in urine in 64 women who held the
34 same job during pregnancy and measurement (time of measurement not stated) had a median
35 value of 48.1 |imol/L (mean: 96.2 ± 19.2 |imol/L) (Lindbohm et al., 1990). Nineteen women had
36 low exposure to TCE (used <1 or 1-4 times/week), and 9 had high exposure to TCE (daily use).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-440 DRAFT—DO NOT CITE OR QUOTE
-------
1 In this follow-up study, an additional questionnaire on time-to-pregnancy was answered by the
2 mothers (Sallmen et al., 1995). The incidence density ratio (IDR) was used in this study to
3 estimate the ratio of average incidence rate of pregnancies for exposed women compared to
4 nonexposed women; therefore, a lower IDR indicates infertility. For TCE, a reduced incidence
5 of fecundability was observed in the high exposure group (IDR: 0.61, 95% CI: 0.28-1.33) but
6 not in the low exposure group (IDR: 1.21, 95% CI: 0.73-2.00). A similar study of paternal
7 occupational exposure (Sallmen et al., 1998) is discussed in Section 4.2.1.2.
8 The residential study in Colorado discussed above did not observe an effect on lifetime
9 infertility infertility in the medium (ORadj: 0.45; 95% CI: 0.02-8.92) or high exposure groups
10 (ORadj: 0.88; 95% CI: 0.13-6.22) (ATSDR, 2001). Curiously, exposed women had more
11 pregnancies and live births than controls.
12
13 4.8.1.1.2.2. Menstrual cycle disturbance. The ATSDR (2001) study discussed above also
14 examined effects on the menstrual cycle (ATSDR, 2001). Nonsignificant associations without a
15 dose-response were seen for abnormal menstrual cycle in women (ORadj: 2.23,
16 95% CI: 0.45-11.18).
17 Other studies have examined the effect of TCE exposure on the menstrual cycle. One
18 study examined women working in a factory assembling small electrical parts (Zielinski, 1973,
19 translated). The mean concentration of TCE in indoor air was reported to be 200 mg/m3.
20 Eighteen percent of the 140 exposed women suffered from amenorrhea, compared to only 2% of
21 the 44 nonexposed workers. The other study examined 75 men and women working in dry
22 cleaning or metal degreasing (Bardodej and Vyskocil, 1956). Exposures ranged from
23 0.28-3.4 mg/L, and length of exposure ranged from 0.5 to 25 years. This study reported that
24 many women experienced menstrual cycle disturbances, with a trend for increasing air
25 concentrations and increasing duration of exposure.
26 An additional case study of a 20-year-old woman was occupationally exposed to TCE via
27 inhalation. The exposure was estimated to be as high as 10 mg/mL or several thousand ppm,
28 based on urine samples 21-25 days after exposure of 3.2 ng/mL of total trichloro-compounds.
29 The primary effect was neurological, although she also experienced amenorrhea, followed by
30 irregular menstruation and lack of ovulation as measured by basal body temperature curves
31 (Sagawa et al., 1973).
32
33 4.8.1.1.3. Male human reproductive effects.
34 4.8.1.1.3.1. Reproductive behavior. One study reported on the effect of TCE exposure on the
35 male reproductive behavior in 75 men working in dry cleaning or metal degreasing (Bardodej
36 and Vyskocil, 1956). Exposures ranged from 0.28-3.4 mg/L, and length of exposure ranged
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-441 DRAFT—DO NOT CITE OR QUOTE
-------
1 from 0.5 to 25 years. This study found that men experienced decreased potency or sexual
2 disturbances; the authors speculated that the effects on men could be due to the CNS effects of
3 TCE exposure. This study also measured serial neutral 17-ketosteroid determinations but they
4 were found to be not statistically significant (Bardodej and Vyskocil, 1956).
5 An occupational study of 30 men working in a money printing shop were exposed to
6 TCE for <1 year to 5 years (El Ghawabi et al., 1973). Depending on the job description, the
7 exposures ranged from 38-172-ppm TCE. Ten (33%) men suffered from decreased libido,
8 compared to three (10%) of unexposed controls. However, these results were not stratified by
9 exposure level or duration. The authors speculate that decreased libido was likely due to the
10 common symptoms of fatigue and sleepiness.
11 A case study described a 42 year-old man exposed to TCE who worked as an aircraft
12 mechanic for approximately 25 years (Saihan et al., 1978). He suffered from a number of health
13 complaints including gynaecomastia and impotence, along with neurotoxicity and
14 immunotoxicity. In addition, he drank alcohol daily which could have increased his response to
15 TCE.
16
17 4.8.1.1.3.2. Altered sperm duality. Genotoxic effects on male reproductive function were
18 examined in a study evaluating occupational TCE exposure in 15 male metal degreasers
19 (Rasmussen et al., 1988). No measurement of TCE exposure was reported. Sperm count,
20 morphology, and spermatozoa Y-chromosomal nondisjunction during spermatogenesis were
21 examined, along with chromosomal aberrations in cultured lymphocytes. A nonsignificant
22 increase in percentage of two fluorescent Y-bodies (YFF) in spermatozoa were seen in the
23 exposed group (p > 0.10), and no difference was seen in sperm count or morphology compared
24 to controls.
25 An occupational study of men using TCE for electronics degreasing (Chia et al., 1996,
26 1997; Goh et al., 1998) examined subjects (n = 85) who were offered a free medical exam if they
27 had no prior history related to endocrine function, no clinical abnormalities, and normal liver
28 function tests; no controls were used. These participants provided urine, blood, and sperm
29 samples. The mean urine TCA level was 22.4 mg/g creatinine (range: 0.8-136.4 mg/g
30 creatinine). In addition, 12 participants provided personal 8-hour air samples, which resulted in
31 a mean TCE exposure of 29.6 ppm (range: 9-131 ppm). Sperm samples were divided into two
32 exposure groups; low for urine TCE less than 25 mg/g creatinine, and high for urine TCA greater
33 than or equal to 25 mg/g creatinine. A decreased percentage of normal sperm morphology was
34 observed in the sperm samples in the high exposure group (n = 48) compared to the low
35 exposure group (n = 37). However, TCE exposure had no effect on semen volume, sperm
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-442 DRAFT—DO NOT CITE OR QUOTE
-------
1 density, or motility. There was also an increased prevalence of hyperzoospermia (sperm density
2 of >120 million sperm per mL ejaculate) with increasing urine TCA levels (Chia et al., 1996).
O
4 4.8.1.1.3.3. Altered endocrine function. Two studies followed up on the study by Chia et al.
5 (1996) to examine endocrine function (Chia et al., 1997; Goh et al., 1998). The first examined
6 serum testosterone, follicle-stimulating hormone (FSH), dehydroepiandrosterone sulphate
7 (DHEAS), and sex-hormone binding globulin (SHBG) (Chia et al., 1997). With increased years
8 of exposure to TCE, an increase in DHEAS levels were seen, from 255 ng/mL for <3 years to
9 717.8 ng/mL >7 years exposure. Also with increased years of exposure to TCE, decreased FSH,
10 SHBG and testosterone levels were seen. The authors speculated these effects could be due to
11 decreased liver function related to TCE exposure (Chia et al., 1997).
12 The second follow-up study of this cohort studied the hormonal effects of chronic low-
13 dose TCE exposure in these men (Goh et al., 1998). Because urine TCE measures only indicate
14 short-term exposure, long-term exposure was indicated by years of exposure. Hormone levels
15 examined include androstenedione, cortisol, testosterone, aldosterone, SHBG, and insulin.
16 Results show that a decrease in serum levels of testosterone and SHBG were significantly
17 correlated with years of exposure to TCE, and an increase in insulin levels were seen in those
18 exposed for less than 2 years. Androstenedione, cortisol, and aldosterone were in normal ranges
19 and did not change with years of exposure to TCE.
20
21 4.8.1.1.3.4. Infertility. Sallmen et al. (1998) examined paternal occupational exposure and
22 time-to-pregnancy among their wives. Cases of spontaneous abortion and controls from a prior
23 study of pregnancy outcome (Taskinen et al., 1989) were used to study time-to-pregnancy of
24 282 couples. Exposure was determined by biological measurements of the father who held the
25 same job during pregnancy and measurement (time of measurement not stated) and
26 questionnaires answered by both the mother and father. An additional questionnaire on time-to-
27 pregnancy was answered by the mother for this study six years after the original study
28 (Sallmen et al., 1998). The level of exposure was determined by questionnaire and classified as
29 "low/intermediate" if the chemical was used <1 or 1-4 days/week and biological measures
30 indicated high exposure (defined as above the reference value for the general population), and
31 "high" if used daily or if biological measures indicated high exposure. For 13 men highly
32 exposed, mean levels of urine TCA were 45 |imol/L (SD 42 |imol/L; median 31 |imol/L); for
33 22 men low/intermediately exposed, mean levels of urine TCA were 41 |imol/L (SD 88 |imol/L;
34 median 15 |imol/L). The terminology IDR was replaced by fecundability density ratio (FDR) in
35 order to reflect that pregnancy is a desired outcome; therefore, a high FDR indicates infertility.
36 No effect was seen on fertility in the low exposure group (FDR: 0.99, 95% CI: 0.63-1.56) or in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-443 DRAFT—DO NOT CITE OR QUOTE
-------
1 the intermediate/high exposure group (FDR: 1.03, 95% CI: 0.60-1.76). However, the exposure
2 categories were grouped by low/intermediate versus high, whereas the outcome categories were
3 grouped by low versus intermediate/high, making a dose-response association difficult.
4 A small occupational study reported on eight male mechanics exposed to TCE for at least
5 two years who sought medical treatment for infertility (Forkert et al., 2003). The wives were
6 determined to have normal fertility. Samples of urine from two of the eight male mechanics
7 contained TCA and/or TCOH, demonstrating the rapid metabolism in the body. However,
8 samples of seminal fluid taken from all eight individuals detected TCE and the metabolites
9 chloral hydrate and TCOH, with two samples detecting DCA and one sample detecting TCA.
10 Five unexposed controls also diagnosed with infertility did not have any TCE or metabolites in
11 samples of seminal fluid. There was no control group that did not experience infertility.
12 Increased levels of TCE and its metabolites in the seminal fluid of exposed workers compared to
13 lower levels found in their urine samples was explained by cumulative exposure and
14 mobilization of TCE from adipose tissue, particularly that surrounding the epididymis. In
15 addition, CYP2E1 was detected in the epididymis, demonstrating that metabolism of TCE can
16 occur in the male reproductive tract. However, this study could not directly link TCE to the
17 infertility, as both the exposed and control populations were selected due to their infertility.
18 The ATSDR (2001) study discussed above on the reproductive effects from TCE in
19 drinking water of individuals living near the Rocky Mountain Arsenal in Colorado did not
20 observe infertility or other adverse reproductive effects for the high exposure group compared to
21 the low exposure group (ORadj: 0.83; 95% CI: 0.11-6.37). Curiously, exposed men had more
22 pregnancies and live births than controls.
23
24 4.8.1.1.4. Summary of human reproductive toxicity. Following exposure to TCE, adverse
25 effects on the female reproductive system observed include reduced incidence of fecundability
26 (as measured by time-to-pregnancy) and menstrual cycle disturbances. Adverse effects on the
27 male reproductive system observed include altered sperm morphology, hyperzoospermia, altered
28 endocrine function, decreased sexual drive and function, and altered fertility. These are
29 summarized in Table 4-74.
30
31 4.8.1.2. Animal Reproductive Toxicity Studies
32 A number of animal studies have been conducted that examined the effects of TCE on
33 reproductive organs and function following either inhalation or oral exposures. These are
34 described below and summarized in Tables 4-75 and 4-76. Other animal studies of offspring
35 exposed during fetal development are discussed in the section on animal developmental studies
36 (see Section 4.8.2.2).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-444 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-74. Human reproductive effects
Subjects
Exposure
Effect
Reference
Female and male combined effects
Reproductive behavior
75 men and 71 women
living near Rocky
Mountain Arsenal,
Colorado
Low: <5.0 ppb
Medium: >5.0-<10.0 ppb
High: <10.0 ppb
Highest: <15 ppb
Altered libido a
Low: referent
Med: ORadj: 0.67 (95% CI: 0.18-2.49)
High: ORad]: 1.65 (95% CI: 0.54-5.01)
Highest: ORadj: 2.46 (95%
CI: 0.59-10.28)
ATSDR,
2001
Female effects
Infertility
197 women
occupationally exposed to
solvents in Finland
1973-1983
71 women living near
Rocky Mountain Arsenal,
Colorado
U-TCA (umol/L) b
Median: 48.1
Mean: 96.2 ± 19.2
Low: <5.0 ppb
Med: >5.0 to <10.0 ppb
High: <10.0 ppb
Reduced incidence of fecundability in the
high exposure group ° as measured by time to
pregnancy
Low: IDR= 1.21 (95%CI: 0.73-2.00)
High: IDR = 0.61 (95%CI: 0.28-1.33)
No effect on lifetime infertility a
Low: referent
Med: ORadj: 0.45 (95% CI: 0.02-8.92)
High: ORadj: 0.88 (95% CI: 0.13-6.22)
Menstrual cycle disturbance
71 women living near
Rocky Mountain Arsenal,
Colorado
184 women working in a
factory assembling small
electrical parts in Poland
32 women working in dry
cleaning or metal
degreasing in
Czechoslovakia"1
20-yr-old woman was
occupationally exposed to
TCE via inhalation
Low: <5.0 ppb
Med: >5.0 to <10.0 ppb
High: <10.0 ppb
Mean indoor air TCE: 200
mg/m3
0.28-3. 4 mg/L TCE for
0.5-25 yrs
Urine total trichloro-
compounds 3.2 ng/mL
(21-25 days after exposure)
Increase in abnormal menstrual cycle
(defined as <26 days or >30 days)
Low: referent
Med: ORadj: 4.17 (95% CI: 0.31-56.65)
High: ORadj: 2.39 (95% CI: 0.41-13.97)
18% reporting increase in amenorrhea in
exposed group (n = 140), compared to 2%
increase in unexposed group (n = 44)
3 1% reporting increase in menstrual
disturbances a
Amenorrhea, followed by irregular
menstruation and lack of ovulation
Sallmen et
al., 1995
ATSDR,
2001
ATSDR,
2001
Zielinski,
1973
Bardodej
and
Vyskocil,
1956
Sagawa et
al., 1973
Male effects
Reproductive behavior
43 men working in dry
cleaning or metal
degreasing in
Czechoslovakia
30 male workers in a
money printing shop in
Egypt
42 yr-old male aircraft
mechanic in UK
0.28-3. 4 mg/L TCE for
0.5-25 yrs
38-172 ppm TCE
TCE exposure reported but
not measured; exposure for
25 yrs
30% reporting decreased potency a
Decreased libido reported in 10 men (33%),
compared to 3 men in the control group
(10%)
Gynaecomastia, impotence
Bardodej
and
Vyskocil,
1956
El Ghawabi
etal., 1973
Saihanetal.,
1978
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-445 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-74. Human reproductive effects (continued)
Subjects
Exposure
Effect
Altered sperm quality
15 men working as metal
degreasers in Denmark
85 men of Chinese
descent working in an
electronics factory
TCE exposure reported but
not measured
Mean personal air TCE:
29.6 ppm; Mean U-TCA:
22.4 mg/g creatinine
Nonsignificant increase in percentage of two
YFF in spermatozoa; no effect on sperm
count or morphology
Decreased normal sperm morphology and
hyperzoospermia
Altered endocrine function
85 men of Chinese
descent working in an
electronics factory
85 men of Chinese
descent working in an
electronics factory
Mean personal air TCE:
29.6 ppm; Mean U-TCA:
22.4 mg/g creatinine
Mean personal air TCE:
29.6 ppm; Mean U-TCA:
22.4 mg/g creatinine
Increased DHEAS and decreased FSH,
SHBG and testosterone levels; dose-response
observed
Decreased serum levels of testosterone and
SHBG were significantly correlated with
years of exposure to TCE; increased insulin
levels for exposure <2 yrs
Infertility
282 men occupationally
exposed to solvents in
Finland 1973-1983
8 male mechanics seeking
treatment for infertility in
Canada
75 men living near Rocky
Mountain Arsenal,
Colorado
U-TCA (umol/L):
High exposure: °
Mean: 45 (SD 42)
Median 3 1
Low exposure: °
Mean: 41 (SD 88)
Median: 15
Urine (umol/):
TCA: 0.30-4.22
TCOH: 0.60-0.89
Seminal fluid (pg/extract):
TCE: 20.4-5,419.0
Chloral: 61.2-1,739.0
TCOH 2.7-25.5
TCA: <100-5,504
DCA: <100-13,342
Low: <5.0 ppb
Med: >5.0 to <10.0 ppb
High: <10.0 ppb
No effect on fecundability ° (as measured by
time to pregnancy)
Low: FDR: 0.99 (95% CI: 0.63-1.56)
Intermediate/High: FDR:C 1.03 (95%
CI: 0.60-1.76)
Infertility could not be associated with TCE
as controls were 5 men also in treatment for
infertility
No effect on lifetime infertility (not defined)
Low: referent
Med: n/a
High: ORadj: 0.83 (95% CI: 0.11-6.37)
Reference
Rasmussen
etal., 1988
Chia et al.,
1996
Chia et al.,
1997
Goh et al.,
1998
Sallmen et
al., 1998
Forkert et
al., 2003
ATSDR,
2001
2
3
4
5
6
7
aNot defined by the authors.
bAs reported in Lindbohm et al. (1990).
'Low/intermediate exposure indicated use of TCE <1 or 1-4 days/week, and biological measures indicated high
exposure. High exposure indicated daily use of TCE, or if biological measures indicated high exposure.
dNumber inferred from data provided in Tables 2 and 3 in Bardodej and Vyskocil (1956).
9 UK = United Kingdom.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-446 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-75. Summary of mammalian in vivo reproductive toxicity studies—
inhalation exposures
Reference
Forkert et al.,
2002
Kan et al.,
2007
Kumar et al.,
2000a
Kumar et al.,
2000b
Kumar et al.,
2001
Land et al.,
1981
Species/strain/
sex/number
Mouse, CD-I,
male, 6/group
Mouse, CD-I,
male, 4/group
Rat, Wistar,
male,
12-13/group
Rat, Wistar,
males,
12-13/group
Rat, Wistar,
male, 6/group
Mouse,
C57BlxC3H
(Fl), male, 5
or 10/group
Exposure
level/duration
0 or 1,000 ppm
(5,374 mg/m3)b
6 h/d, 5 d/wk,
19 dover
4 wks
0 or 1,000 ppm
6 h/d,5 d/wk,
1 to 4 wks
0 or 376 ppm
4 h/d, 5 d/wk,
2 to 10 wks
exposure, 2 to
8 wks rest
period
0 or 376 ppm
4 h/d, 5 d/wk,
12 and 24 wks
0 or 376 ppm
4 h/d, 5 d/wk,
12 and 24 wks
0, 0.02%, or
0.2%
4 h/d, 5 d, 23 d
rest
NOAEL;
LOAEL a
LOAEL: 1,000
ppm
LOAEL: 1,000
ppm
LOAEL: 376
ppm
LOAEL: 376
ppm
LOAEL: 376
ppm
NOAEL:
0.02%
LOAEL: 0.2%
Effects
U-TCA and U-TCOH increased by 2nd and 3rd
wk, respectively. Cytochrome P450 2E1 andp-
nitrophenol hydroxylation in epididymal
epithelium > testicular Leydig cells. Choral
also generated from TCE in epididymis > testis.
Sloughing of epididymal epithelial cells after
4 wk exposure.
Light microscopy findings: degeneration and
sloughing of epididymal epithelial cells as early
as 1 wk into exposure; more severe by 4 wks.
Ultrastructional findings: vesiculation in
cytoplasm, disintegration of basolateral cell
membranes, sloughing of epithelial cells.
Sperm found in situ in cytoplasm of
degenerated epididymal cells. Abnormalities of
the head and tail in sperm located in the
epididymal lumen.
Alterations in testes histopathology (smaller,
necrotic spermatogenic tubules), t sperm
abnormalities, and sig. t pre- and/or
postimplantation loss in litters observed in the
groups with 2 or 10 wks of exposure, or 5 wks
of exposure with 2 wks rest.
Sig. -I in total epididymal sperm count and
sperm motility, with sig. -i- in serum
testosterone, sig. t in testes cholesterol, sig. -i-
of glucose 6-phosphate dehydrogenase and
17-p-hydroxy steroid dehydrogenase at 12 and
24 wks exposure.
BW gain sig. |. Testis weight, sperm count and
motility sig. |, effect stronger with exposure
time. After 12 wk, numbers of spermatogenic
cells and spermatids |, some of the
spermatogenic cells appeared necrotic. After
24 wk testes were atrophied, tubules were
smaller, had Sertoli cells and were almost
devoid of spermatocytes and spermatids.
Leydig cells were hyperplastic. SDH, G6PDH
sig. |, GOT and p-glucuronidase sig. |; effects
stronger with exposure time.
Sig. t percent morphologically abnormal
epididymal sperm.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-447 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-75. Summary of mammalian in vivo reproductive toxicity studies—
inhalation exposures (continued)
Reference
Xuetal.,
2004
Species/strain/
sex/number
Mouse, CD-I,
male, 4 to
27/group
Exposure
level/duration
0 or 1,000 ppm
(5.37 mg/L)b
6 h/d, 5 d/wk,
1-6 wks
NOAEL;
LOAEL a
LOAEL: 1,000
ppm
Effects
Sig. I in vitro sperm-oocyte binding and in vivo
fertilization
3 aNOAEL and LOAEL are based upon reported study findings.
4 bDose conversion calculations by study author(s).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-448 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-76. Summary of mammalian in vivo reproductive toxicity studies—
oral exposures
Reference
Species/strain/
sex/number
Dose
level/exposure
duration
Route/vehicle
NOAEL;
LOAEL a
Effects
Studies assessing male reproductive outcomes
DuTeaux et
al, 2003
DuTeaux et
al., 2004b
Veeramachan
enietal., 2001
Zenicketal.,
1984
Rat, Sprague-
Dawley, male,
3/group
Rat, Sprague-
Dawley, male,
3/group, or
Simonson
albino
(UC-Davis),
male, 3/group
Rabbit, Dutch
belted, females
and offspring;
7-9
offspring/group
Rat, Long-
Evans, male,
10/group
0, 0.2, or 0.4%
(0,143, or 270
mg/kg/d)
0,0.2, or 0.4%
(0, 143, or 270
mg/kg/d)
14 d
9.5-or28.5-ppm
TCEd
GD 20 thru
lactation, then to
offspring thru
postnatal wk 15
0, 10, 100, or
1,000 mg/kg/d
6 wk, 5 d/wk;
4 wks recovery
Drinking
water; 3%
ethoxylated
castor oil
vehicle
Drinking
water, 3%
ethoxylated
castor oil
vehicle
Drinking
water
Gavage, corn
oil vehicle
LOEL: 0.2%
LOAEL: 0.2%
LOAEL:
9.5 ppm
NOAEL: 100
mg/kg/d
LOAEL:
1,000 mg/kg/d
TCE metabolite-protein
adducts formed by a
cytochrome P450-mediated
pathway were detected by
fluorescence
imunohistochemistry in the
epithelia of corpus
epididymis and in efferent
ducts.
Dose-dependent -i- in ability
of sperm to fertilize oocytes
collected from untreated $s.
Oxidative damage to sperm
membrane in head and mid-
piece was indicated by dose-
related t in oxidized proteins
and lipid peroxidation.
Decreased copulatory
behavior; acrosomal
dysgenesis, nuclear
malformations; sig. -i- LH and
testosterone.
At 1,000 mg/kg, BW ^
liver/BW ratios t, and
impaired copulatory behavior.
Copulatory performance
returned to normal by 5th wk
of exposure. At wk 6, TCE
and metabolites concentrated
to a significant extent in male
reproductive organs.
Studies assessing female reproductive outcomes
Berger and
Horner, 2003
Cosby and
Dukelow,
1992
Rat, Simonson
(S-D derived),
female,
(5-6) x 3/group
Mouse,
B6D2F1,
female,
7-12/group
0 or 0.45%
2 wks
0,24, or 240
mg/kg/d
GD 1-5, 6-10, or
11-15
Drinking
water, 3%
Tween
vehicle
Gavage, corn
oil vehicle
LOAEL:
0.45%
NOAEL: 240
mg/kg/d
In vitro fertilization and
sperm penetration of oocytes
sig. -I with sperm harvested
from untreated males.
No treatment-related effects
on in vitro fertilization in
dams or offspring.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-449 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-76. Summary of mammalian in vivo reproductive toxicity studies—
oral exposures (continued)
Reference
Manson et al.,
1984
Wuand
Berger, 2007
Wuand
Berger, 2008
Species/strain/
sex/number
Rat, Long-
Evans, female,
23-25/group
Rat, Simonson
(S-D derived),
female,
(no. /group not
reported)
Rat, Simonson
(S-D derived),
female,
(no. /group not
reported)
Dose
level/exposure
duration
0, 10, 100, or
1,000 mg/kg/d
6 wks: 2 wk
premating, 1 wk
mating period,
GD 1-21
0 or 0.45%
(0.66g/kg-d)b
Preovulation days
1-5, 6-10,
11-14, or 1-14
0 or 0.45%
(0.66g/kg-d)b
Ior5d
Route/vehicle
Gavage, corn
oil vehicle
Drinking
water, 3%
Tween
vehicle
Drinking
water, 3%
Tween
vehicle
NOAEL;
LOAEL a
NOAEL: 100
mg/kg/d
LOAEL: 1,000
mg/kg/d
LOAEL:
0.45%
NOEL: 0.45%
Effects
Female fertility and mating
success was not affected. At
1,000 mg/kg/d group, 5/23
females died, gestation B W
gain was sig. -I: After
subchronic oral TCE
exposure, TCE was detected
in fat, adrenals, and ovaries;
TCA levels in uterine tissue
were high.
At 1,000 mg/kg/d, neonatal
deaths (female pups) were t
onPNDs 1, 10, and 14.
Dose-related t seen in TCA
in blood, liver and milk in
stomach of ? pups, not <$s.
In vitro fertilization and
sperm penetration of oocytes
sig. •I' with sperm harvested
from untreated males.
Ovarian mRNA expression
for ALCAM and Cudzl
protein were not altered.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-450 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-76. Summary of mammalian in vivo reproductive toxicity studies—
oral exposures (continued)
Reference
Species/strain/
sex/number
Dose
level/exposure
duration
Route/vehicle
NOAEL;
LOAEL a
Effects
Studies assessing fertility and reproductive outcome in both sexes
George et al.,
1985
George et al.,
1986
Mouse, CD-I,
male and
female, 20
pairs/treatment
group; 40
controls/sex
Rat, F334,
males and
female,
20 pairs/
treatment
group,
40 controls/sex
0,0.15, 0.30, or
0.60% c micro-
encapsulated TCE
(TWA dose
estimates: 0, 173,
362, or 737
mg/kg/d) b
Breeders exposed
1 wk premating,
then for 13 wk;
pregnant females
throughout
gestation (i.e.,
ISwktotal)
0,0.15, 0.30 or
0.60% c micro-
encapsulated TCE
Breeders exposed
1 wk premating,
then for 13 wk;
pregnant females
throughout
gestation (i.e.,
ISwktotal)
Dietary
Dietary
Parental
systemic
toxicity:
NOAEL:
0.30%
LOAEL:
0.60%
Parental
reproductive
function:
LOAEL:
0.60% c
Offspring
toxicity:
NOAEL:
0.30%
LOAEL:
0.60%
Parental
systemic
toxicity:
LOAEL:
0.15%
At 0.60%, in F0:sig.t liver
weights in both sexes; sig. -i-
testis and seminal vesicle
weight; histopathology of
liver and kidney in both
sexes.
At 0.60%, inFl: sig. ^ BW
on PND 74, and in
postpartum Fl dams; sig. t
liver, testis, and epididymis
weights in males, sig. t
kidney weights in both sexes;
sig. -i- testis and seminal
vesicle weight;
histopathology of liver and
kidney in both sexes.
At0.60%,inFOandFl
males: sig. -i- sperm motility.
At 0.60%, inFl pups: sig. -i-
live birth weights, sig. -i- PND
4 pup BW; perinatal mortality
t (PND 0-21).
At 0.60%, in FO: sig. i
postpartum dam B W; sig. -i-
term. B W in both sexes; sig.
t liver, and kidney /adrenal
weights in both sexes; sig. t
testis/epididymis weights; in
Fl: sig. -i- testis weight.
At all doses in Fl : sig. -i-
postpartum dam BW; sig.^
term. B W in both sexes, sig.
t liver wt. in both sexes.
At 0.30 and 0.60%, inFl: sig.
T liver wt. in females.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-451 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-76. Summary of mammalian in vivo reproductive toxicity studies—
oral exposures (continued)
Reference
George et al.,
1986
(continued)
Species/strain/
sex/number
Dose
level/exposure
duration
Route/vehicle
NOAEL;
LOAEL a
Parental
reproductive
function:
LOAEL:
0.60% c
Offspring
toxicity:
LOAEL:
0.15%
Effects
At 0.60%, sig ^ mating in FO
males and females (in cross-
over mating trials).
At 0.60%, sig. ^ Fl BW on
PND 4 and 14.
At all doses, sig. ^ Fl BW on
PND 2 land 80.
At 0.3 and 0.60%, sig. i live
Fl pups/litter.
At 0.15 and 0.60%, trend
toward ^ Fl survival from
PND 21 to PND 80.
2 aNOAEL, LOAEL, NOEL, and LOEL (lowest-observed-effect level) are based upon reported study findings.
3 bDose conversion calculations by study author(s).
4 fertility and reproduction assessment of last litter from continuous breeding phase and cross-over mating
5 assessment (rats only) were conducted for 0 or 0.60% dose groups only.
6 dConcurrent exposure to several ground water contaminants; values given are for TCE levels in the mixture.
7
8
9 4.8.1.2.1. Inhalation exposures. Studies in rodents exposed to TCE via inhalation are
10 described below and summarized in Table 4-75. These studies focused on various aspects of
11 male reproductive organ integrity, spermatogenesis, or sperm function in rats or mice. In the
12 studies published after the year 2000, the effects of either 376 or 1,000-ppm TCE were studied
13 following exposure durations ranging from 1 to 24 weeks, and adverse effects on male
14 reproductive endpoints were observed.
15 Kumar et al. (2000a) exposed male Wistar rats in whole body inhalation chambers to
16 376-ppm TCE for 4 hours/day, 5 days/week over several duration scenarios. These were
17 2-weeks (to observe the effect on the epididymal sperm maturation phase), 10 weeks (to observe
18 the effect on the entire spermatogenic cycle), 5 weeks with 2 weeks rest (to observe the effect on
19 primary spermatocytes differentiation to sperm), 8 weeks with 5 weeks rest (to observe effects
20 on an intermediate stage of spermatogenesis), and 10 weeks with 8 weeks rest (to observe the
21 effect on spermatogonial differentiation to sperm). Control rats were exposed to ambient air.
22 Weekly mating with untreated females was conducted. At the end of the treatment/rest periods,
23 the animals were sacrificed; testes and cauda epididymes tissues were collected. Alterations in
24 testes histopathology (smaller, necrotic spermatogenic tubules), increased sperm abnormalities,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-452 DRAFT—DO NOT CITE OR QUOTE
-------
1 and significantly increased pre- and/or postimplantation loss in litters were observed in the
2 groups with 2 or 10 weeks of exposure, or 5 weeks of exposure with 2 weeks rest. It was
3 hypothesized that postmeiotic cells of spermatogenesis and epididymal sperm were affected by
4 TCE exposure, leading to reproductive impairment.
5 To test the hypothesis that TCE exposure adversely affects sperm function and
6 fertilization, Xu et al. (2004) conducted a study in which male CD-I mice were exposed by
7 inhalation to atmospheres containing 1,000 ppm (5.37 mg/L) TCE for 1 to 6 weeks (6 hours/day,
8 5 days/week). After each TCE exposure, body weights were recorded. Following termination,
9 the right testis and epididymis of each treated male were weighed, and sperm was collected from
10 the left epididymis and vas deferens for assessment of the number of total sperm and motile
11 sperm. Sperm function was evaluated in the following experiments: (1) suspensions of
12 capacitated vas deferens/cauda epididymal sperm were examined for spontaneous acrosome
13 reaction, (2) in vitro binding of capacitated sperm to mature eggs from female CF-1 mice
14 (expressed as the number of sperm bound per egg) was assessed, and (3) in vivo fertilization was
15 evaluated via mating of male mice to superovulated female CF-1 mice immediately following
16 inhalation exposure; cumulus masses containing mature eggs were collected from the oviducts of
17 the females, and the percentage of eggs fertilized was examined. Inhalation exposure to TCE did
18 not result in altered body weight, testis and epididymis weights, sperm count, or sperm
19 morphology or motility. Percentages of acrosome-intact sperm populations were similar
20 between treated and control animals. Nevertheless, for males treated with TCE for 2 or more
21 weeks decreases were observed in the number of sperm bound to the oocytes in vitro (significant
22 at 2 and 6 weeks, p < 0.001). In a follow-up assessment, control sperm were incubated for
23 30-minutes in buffered solutions of TCE or metabolites (chloral hydrate or trichloroethanol);
24 while TCE-incubation had no effect on sperm-oocyte binding, decreased binding capacity was
25 noted for the metabolite-incubated sperm. The ability for sperm from TCE-exposed males to
26 bind to and fertilize oocytes in vivo was also found to be significantly impaired (p < 0.05).
27 A study designed to investigate the role of testosterone, and of cholesterol and ascorbic
28 acid (which are primary precursors of testosterone) in TCE-exposed rats with compromised
29 reproductive function was conducted by Kumar et al. (2000b). Male Wistar rats (12-13/group)
30 were exposed (whole body) to 376 ppm TCE by inhalation for 4 hours/day, 5 days/week, for
31 either 12 or 24 weeks and then terminated. Separate ambient-air control groups were conducted
32 for the 12- and 24-week exposure studies. Epididymal sperm count and motility were evaluated,
33 and measures of 17-p-hydroxy steroid dehydrogenase (17-P-HSD), testicular total cholesterol
34 and ascorbic acid, serum testosterone, and glucose 6-p dehydrogenase (G6PDH) in testicular
35 homogenate were assayed. In rats exposed to TCE for either 12 or 24 weeks, total epididymal
36 sperm count and motility, serum testosterone concentration, and specific activities of both
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-453 DRAFT—DO NOT CITE OR QUOTE
-------
1 17-P-HSD and G6PDH were significantly decreased (p < 0.05), while total cholesterol content
2 was significantly (p < 0.05) increased. Ascorbic acid levels were not affected.
3 In another study, Kumar et al. (2001) utilized the same exposure paradigm to examine
4 cauda epididymal sperm count and motility, testicular histopathology, and testicular marker
5 enzymes: sorbitol dehydrogenase (SDH), G6PDH, glutamyl transferase (GT), and glucuronidase,
6 in Wistar rats (6/group). After 24 weeks of exposure, testes weights and epididymal sperm count
7 and motility were significantly decreased (p < 0.05). After 12 weeks of TCE exposure,
8 histopathological examination of the testes revealed a reduced number of spermatogenic cells in
9 the seminiferous tubules, fewer spermatids as compared to controls, and the presence of necrotic
10 spermatogenic cells. Testicular atrophy, smaller tubules, hyperplastic Ley dig cells, and a lack of
11 spermatocytes and spermatids in the tubules were observed after 24 weeks of TCE exposure.
12 After both 12 and 24 weeks of exposure, SDH and G6PDH were significantly (p < 0.05) reduced
13 while GT and p-glucuronidase were significantly (p < 0.05) increased.
14 In a study by Land et al. (1981), 8-10 week old male mice (C57BlxC3H)Fl (5 or
15 10/group) were exposed (whole body) by inhalation to a number of anesthetic agents for
16 5 consecutive days at 4 hours/day and sacrificed 28 days after the first day of exposure.
17 Chamber concentration levels for the TCE groups were 0.02 and 0.2%. The control group
18 received ambient air. Epididymal sperm were evaluated for morphological abnormalities. At
19 0.2% TCE, the percent abnormal sperm in a sample of 1,000 was significantly (p < 0.01)
20 increased as compared to control mice; no treatment-related effect on sperm morphology was
21 observed at 0.02% TCE.
22 Forkert et al. (2002) exposed male CD-I mice by inhalation to 1,000-ppm TCE
23 (6 hours/day, 5 day/week) for 4 consecutive weeks and observed sloughing of portions of the
24 epithelium upon histopathological evaluation of testicular and epididymal tissues.
25 Kan et al. (2007) also demonstrated that damage to the epididymal epithelium and sperm
26 of CD-I mice (4/group) resulted from exposure to 0 or 1,000-ppm TCE by inhalation for
27 6 hours/day, 5 days/week, for 1 to 4 weeks. Segments of the epididymis (caput, corpus, and
28 cauda) were examined by light and electron microscope. As early as 1 week after TCE exposure,
29 degeneration and sloughing of epithelial cells from all three epididymal areas were observed by
30 light microscopy; these findings became more pronounced by 4 weeks of exposure. Vesiculation
31 in the cytoplasm, disintegration of basolateral cell membranes, and epithelial cell sloughing were
32 observed with electron microscopy. Sperm were found in situ in the cytoplasm of degenerated
33 epididymal cells. A large number of sperm in the lumen of the epididymis were abnormal,
34 including head and tail abnormalities.
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-454 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.1.2.2. Oral exposures. A variety of studies were conducted to assess various aspects of
2 male and/or female reproductive capacity in laboratory animal species following oral exposures
3 to TCE. These are described below and summarized in Table 4-76. They include studies that
4 focused on male reproductive outcomes in rats or rabbits following gavage or drinking water
5 exposures (Zenick et al., 1984; DuTeaux et al., 2003, 2004b; Veeramachaneni et al., 2001),
6 studies that focused on female reproductive outcomes in rats following gavage or drinking water
7 exposures (Berger and Horner, 2003; Cosby and Dukelow, 1992; Manson et al., 1984; Wu and
8 Berger, 2007, 2008), and studies assessed fertility and reproductive outcome in both sexes
9 following dietary exposures to CD-I mice or F344 rats (George et al., 1985, 1986).
10
11 4.8.1.2.2.1. Studies assessing male reproductive outcomes. Zenick et al. (1984) conducted a
12 study in which sexually experienced Long-Evans hooded male rats were administered 0, 10, 100,
13 or 1,000 mg/kg/d TCE by gavage in corn oil for 6 weeks. A 4-week recovery phase was also
14 incorporated into the study design. Endpoints assessed on Weeks 1 and 5 of treatment included
15 copulatory behavior, ejaculatory plug weights, and ejaculated or epididymal sperm measures
16 (count, motility, and morphology). Sperm measures and plug weights were not affected by
17 treatment, nor were Week 6 plasma testosterone levels found to be altered. TCE effects on
18 copulatory behavior (ej aculation latency, number of mounts, and number of intromissions) were
19 observed at 1,000 mg/kg/d; these effects were recovered by 1-4 weeks post-treatment. Although
20 the effects on male sexual behavior in this study were believed to be unrelated to narcotic effects
21 of TCE, a later study by Nelson and Zenick (1986) showed that naltrexone (an opioid receptor
22 antagonist, 2.0 mg/kg, i.p., administered 15 minutes prior to testing) could block the effect.
23 Thus, it was hypothesized that the adverse effects of TCE on male copulatory behavior in the rat
24 at 1,000 ppm may in fact be mediated by the endogenous opioid system at the CNS level.
25 In a series of experiments by DuTeaux et al. (2003, 2004b), adult male rats were
26 administered 0, 0.2, or 0.4% TCE (v/v) (equivalent to 0, 2.73 mg/L, or 5.46 mg/L) in a solution
27 of 3% ethoxylated castor oil in drinking water for 14 days. These concentrations were within the
28 range of measurements obtained in formerly contaminated drinking water wells, as reported by
29 ATSDR (1997). The average ingested doses of TCE (based upon animal body weight and
30 average daily water consumption of 28 mL) were calculated to be 143 or 270 mg/kg/d for the
31 low and high-dose groups, respectively (DuTeaux et al., 2008). Cauda epididymal and vas
32 deferens sperm from treated males were incubated in culture medium with oviductal cumulus
33 masses from untreated females to assess in vitro fertilization capability. Treatment with TCE
34 resulted in a dose-dependent decrease in the ability of sperm to fertilize oocytes. Terminal body
35 weights and testis/epididymal weights were similar between control and treated groups.
36 Evaluation of sperm concentration or motility parameters did not reveal any treatment-related
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-455 DRAFT—DO NOT CITE OR QUOTE
-------
1 alterations; acrosomal stability and mitochondrial membrane potential were not affected by
2 treatment. Although no histopathological changes were observed in the testis or in the caput,
3 corpus, or cauda epididymis, exposure to 0.2 and 0.4% TCE resulted in slight cellular alterations
4 in the efferent ductule epithelium.
5 Veeramachaneni et al. (2001) evaluated the effects of drinking water containing
6 chemicals typical of ground water near hazardous waste sites (including 9.5- or 28.5-ppm TCE)
7 on male reproduction. In this study, pregnant Dutch-belted rabbits were administered treated
8 drinking water from gestation Day 20; treatment continued through the lactation period and to
9 weaned offspring (7-9/group) through postnatal Week 15. Deionized water was administered
10 from postnatal weeks 16-61, at which time the animals were terminated. At 57-61 weeks of
11 age, ejaculatory capability, and seminal, testicular, epididymal, and endocrine characteristics
12 were evaluated. In both treated groups, long-term effects consisted of decreased copulatory
13 behavior (interest, erection, and/or ejaculation), significant increases in acrosomal dysgenesis
14 and nuclear malformations (p < 0.03), and significant decreases in serum concentration of
15 luteinizing hormone (p < 0.05) and testosterone secretion after human chorionic gonadotropin
16 administration (p < 0.04). There were no effects on total spermatozoa per ejaculate or on daily
17 sperm production. The contribution of individual drinking water contaminants to adverse male
18 reproductive outcome could not be discerned in this study. Additionally, it was not designed to
19 distinguish between adverse effects that may have resulted from exposures in late gestation (i.e.,
20 during critical period of male reproductive system development) versus postnatal life.
21
22 4.8.1.2.2.2. Studies assessing female reproductive outcomes. In a study that evaluated
23 postnatal growth following gestational exposures, female B6D2F1 mice (7-12/group) were
24 administered TCE at doses of 0, 1% LD50 (24 mg/kg/d), and 10% LD50 (240 mg/kg/d) by gavage
25 in corn oil from gestation days 1-5, 6-10, or 11-15 (day of mating was defined as gestation
26 Day 1) (Cosby and Dukelow, 1992). Litters were examined for pup count, sex, weight, and
27 crown-rump measurement until postnatal Day 21. Some offspring were retained to 6 weeks of
28 age, at which time they were killed and the gonads were removed, weighed and preserved. No
29 treatment-related effects were observed in the dams or offspring. In a second series of studies
30 conducted by Cosby and Dukelow and reported in the same paper, TCE and its metabolites
31 DCA, TCA, and TCOH were added to culture media with capacitated sperm and cumulus masses
32 from B6D2F1 mice to assess effects on in vitro fertilization. Dose-related decreases in
33 fertilization were observed for DCA, TCA, and TCOH at 100 and 1,000 ppm, but not with TCE.
34 Synergystic effects were not observed with TCA and TCOH.
35 A study was conducted by Manson et al. (1984) to determine if subchronic oral exposure
36 to TCE affected female reproductive performance, and if TCE or its metabolites trichloroacetic
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-456 DRAFT—DO NOT CITE OR QUOTE
-------
1 acid or trichloroethanol accumulated in female reproductive organs or neonatal tissues. Female
2 Long-Evans hooded rats (22-23/group) were administered 0 (corn oil vehicle), 10, 100, or
3 1,000 mg/kg/d of TCE by gavage for 2 weeks prior to mating, throughout mating, and to
4 gestation Day 21. Delivered pups were examined for gross anomalies, and body weight and
5 survival were monitored for 31 days. Three maternal animals per group and 8-10 neonates per
6 group (killed on postnatal Days 3 and 31) were analyzed for TCE and metabolite levels in
7 tissues. TCE exposure resulted in 5 deaths and decreased maternal body weight gain at
8 1,000 mg/kg/d, but did not affect estrous cycle length or female fertility at any dose level. There
9 were no evident developmental anomalies observed at any treatment level; however, at
10 1,000 mg/kg/d there was a significant increase in the number of pups (mostly female) born dead,
11 and the cumulative neonatal survival count through PND 18 was significantly decreased as
12 compared to control. TCE levels were uniformly high in fat, adrenal glands, and ovaries across
13 treatment groups, and TCA levels were high in uterine tissue. TCE levels in the blood, liver, and
14 milk contents of the stomach increased in female PND-3 neonates across treatment groups.
15 These findings suggest that increased metabolite levels did not influence fertility, mating
16 success, or pregnancy outcome.
17 In another study that examined the potential effect of TCE on female reproductive
18 function, Berger and Horner (2003) conducted 2-week exposures of Sprague-Dawley derived
19 female Simonson rats to tetrachloroethylene, trichloroethylene, several ethers, and
20 4-vinylcyclohexene diepoxide in separate groups. The TCE-treated group received 0.45% TCE
21 in drinking water containing 3% Tween vehicle; control groups were administered either
22 untreated water, or water containing the 3% Tween vehicle. There were 5-6 females/group, and
23 three replicates were conducted for each group. At the end of exposure, ovulation was induced,
24 the rats were killed, and the ovaries were removed. The zona pellucida was removed from
25 dissected oocytes, which were then placed into culture medium and inseminated with sperm from
26 untreated males. TCE treatment did not affect female body weight gain, the percentage of
27 females ovulating, or the number of oocytes per ovulating female. Fertilizability of the oocytes
28 from treated females was reduced significantly (46% for TCE-treated females versus 56% for
29 vehicle controls). Oocytes from TCE-treated females had reduced ability to bind sperm plasma
30 membrane proteins compared with vehicle controls.
31 In subsequent studies, Wu and Berger (2007, 2008) examined the effect of TCE on
32 oocyte fertilizibility and ovarian gene expression. TCE was administered to female Simonson
33 rats (number of subjects not reported) in the drinking water at 0 or 0.45% (in 3% Tween vehicle);
34 daily doses were estimated to be 0.66 g TCE/kg body weight/day. In the oocyte fertilizibility
35 study (Wu and Berger, 2007), the female rats were treated on Days 1-5, 6-10, 11-14, or 1-14 of
36 the 2-week period preceding ovulation (on Day 15). Oocytes were extracted and fertilized in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-457 DRAFT—DO NOT CITE OR QUOTE
-------
1 vitro with sperm from a single male donor rat. With any duration of TCE exposure, fertilization
2 (as assessed by the presence of decondensed sperm heads) was significantly (p < 0.05) decreased
3 as compared to controls. After exposure on Days 6-10, 11-14, or 1-14, the oocytes from TCE-
4 treated females had a significantly decreased ability to bind sperm (p < 0.05) in comparison to
5 oocytes from vehicle controls. Increased protein carbonyls (an indicator of oxidatively modified
6 proteins) were detected in the granulosa cells of ovaries from females exposed to TCE for
7 2 weeks. The presence of oxidized protein was confirmed by Western blot analysis.
8 Microsomal preparations demonstrated the localization of cytochrome P450 2E1 and glutathione
9 s-transferase (TCE-metabolizing enzymes) in the ovary. Ovarian mRNA transcription for
10 ALCAM and Cuzdl protein was not found to be altered after 1 or 5 days of exposure (Wu and
11 Berger, 2008), suggesting that the post-translational modification of proteins within the ovary
12 may partially explain the observed reductions in oocyte fertilization.
13
14 4.8.1.2.2.3. Studies assessing fertility and reproductive outcomes in both sexes. Assessments
15 of reproduction and fertility with continuous breeding were conducted in NTP studies in CD-I
16 mice (George et al., 1985) and Fischer 344 rats (George et al., 1986). TCE was administered to
17 the mice and rats at dietary levels of 0, 0.15, 0.30, or 0.60%, based upon the results of
18 preliminary 14-day dose-range finding toxicity studies. Actual daily intake levels for the study
19 in mice were calculated from the results of dietary formulation analyses and body weight/food
20 consumption data at several time points during study conduct; the most conservative were from
21 the second week of the continuous breeding study: 0, 52.5, 266.3, and 615.0 mg/kg/d. No intake
22 calculations were presented for the rat study. In these studies, which were designed as described
23 by Chapin and Sloane (1996), the continuous breeding phase in FO adults consisted of a 7-day
24 premating exposure, 98-day cohabitation period, and 28-day segregation period. In rats, a
25 crossover mating trial (i.e., control males x control females; 0.60% TCE males x control
26 females; control males x 0.60% TCE females) was conducted to further elucidate treatment-
27 related adverse reproductive trends observed in the continuous breeding phase. The last litter of
28 the continuous breeding phase was raised to sexual maturity for an assessment of fertility and
29 reproduction in control and high-dose groups; for the rats, this included an open field behavioral
30 assessment of Fl pups. The study protocol included terminal studies in both generations,
31 including sperm evaluation (count morphology, and motility) in 10 selected males per dose level,
32 macroscopic pathology, organ weights, and histopathology of selected organs.
33 In the continuous breeding phase of the CD-I mouse study (George et al., 1985), no
34 clinical signs of toxicity were observed in the parental (FO) animals, and there were no treatment-
35 related effects on the proportion of breeding pairs able to produce a litter, the number of live
36 pups per litter, the percent born live, the proportion of pups born live, the sex of pups born live,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-458 DRAFT—DO NOT CITE OR QUOTE
-------
1 absolute live pup weights, or adjusted female pup weights. At the high dose level of 0.60%, a
2 number of adverse outcomes were observed. In the parental animals, absolute and body-weight-
3 adjusted male and female liver weight values were significantly increased (p < 0.01), and right
4 testis and seminal vesicle weights were decreased (p < 0.05), but kidney/adrenal weights were
5 not affected. Sperm motility was significantly (p < 0.01) decreased by 45% in treated males as
6 compared to controls. Histopathology examination revealed lesions in the liver (hypertrophy of
7 the centrilobular liver cells) and kidneys (tubular degeneration and karyomegaly of the
8 corticomedullary renal tubular epithelium) of FO males and females. In the pups at 0.60%,
9 adjusted live birth weights for males and both sexes combined were significantly decreased
10 (p < 0.01) as compared to control. The last control and high-dose litters of the continuous
11 breeding assessment were raised to the age of sexual maturity for a further assessment of
12 reproductive performance. In these Fl pups, body weights (both sexes) were significantly
13 decreased at PND 4, and male offspring body weights were significantly (p < 0.05) less than
14 controls at PND 74 (±10). It was reported that perinatal mortality (PND 0-21) was increased,
15 with a 61.3% mortality rate for TCE-treated pups versus a 28.3% mortality rate for control pups.
16 Reproductive performance was not affected by treatment, and postmortem evaluations of the Fl
17 adult mice revealed significant findings at 0.60% TCE that were consistent with those seen in the
18 FO adults and additionally demonstrated renal toxicity, i.e., elevated liver and kidney/adrenal
19 weights and hepatic and renal histopathological lesions in both sexes, elevated testis and
20 epididymis weights in males, and decreased sperm motility (18% less than control).
21 The F344 rat study continuous breeding phase demonstrated no evidence of treatment-
22 related effects on the proportion of breeding pairs able to produce a litter, percent of pups born
23 alive, the sex of pups born alive, or absolute or adjusted pup weights (George et al., 1986).
24 However, the number of live pups per litter was significantly (p < 0.05) decreased at 0.30 and
25 0.60% TCE, and a significant (p < 0.01) trend toward a dose-related decrease in the number of
26 live litters per pair was observed; individual data were reported to indicate a progressive decrease
27 in the number of breeding pairs in each treatment group producing third, fourth, and fifth litters.
28 The crossover mating trial conducted in order to pursue this outcome demonstrated that the
29 proportion of detected matings was significantly depressed (p < 0.05) in the mating pairs with
30 TCE-treated partners compared to the control pairs. In the FO adults at 0.60% TCE, postpartum
31 dam body weights were significantly decreased (p < 0.01 or 0.05) in the continuous breeding
32 phase and the crossover mating trials, and terminal body weights were significantly decreased
33 (p < 0.01) for both male and female rats. Postmortem findings for FO adults in the high-dose
34 group included significantly increased absolute and body-weight-adjusted liver and
35 kidney/adrenal weights in males, increased adjusted liver and kidney/adrenal weights in females,
36 and significantly increased adjusted left testis/epididymal weights. Sperm assessment did not
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-459 DRAFT—DO NOT CITE OR QUOTE
-------
1 identify any effects on motility, concentration or morphology, and histopathological examination
2 was negative. The last control and high-dose litters of the continuous breeding assessment were
3 raised to the age of sexual maturity for assessment of open field behavior and reproductive
4 performance. In these Fl pups at 0.60% TCE, body weights of male and females were
5 significantly (p < 0.05 or 0.01, respectively) decreased at PND 4 and 14. By PND 21, pup
6 weights in both sexes were significantly reduced in all treated groups, and this continued until
7 termination (approximately PND 80). A tendency toward decreased postweaning survival (i.e.,
8 from PND 21 to PND 81 ± 10) was reported for Fl pups at the 0.15 and 0.60% levels. Open
9 field testing revealed a significant (p < 0.05) dose-related trend toward an increase in the time
10 required for male and female Fl weanling pups to cross the first grid in the testing device,
11 suggesting an effect on the ability to react to a novel environment. Reproductive performance
12 assessments conducted in this study phase were not affected by treatment. Postpartum Fl dam
13 body weights were significantly decreased (p < 0.05 or 0.01) in all of the TCE-treated groups as
14 compared to controls, as were terminal body weights for both adult Fl males and females.
15 Postmortem evaluations of the Fl adult rats revealed significantly (p < 0.01) decreased left
16 testis/epididymis weight at 0.60% TCE, and significantly (p < 0.05 or 0.01) increased adjusted
17 mean liver weight in all treated groups for males and at 0.30 and 0.60% for females. Sperm
18 assessments for Fl males revealed a significant increase (p < 0.05) in the percent abnormal
19 sperm in the 0.30% TCE group, but no other adverse effects on sperm motility, concentration, or
20 morphology were observed. As with the FO adults, there were no adverse treatment-related
21 findings revealed at histopathological assessment. The study authors concluded that the
22 observed effects to TCE exposure in this study were primarily due to generalized toxicity and not
23 to a specific effect on the reproductive system; however, based upon the overall toxicological
24 profile for TCE, which demonstrates that the male reproductive system is a target for TCE
25 exposures, this conclusion is not supported.
26
27 4.8.1.3. Discussion/Synthesis ofnoncancer reproductive toxicity findings
28 The human epidemiological findings and animal study evidence consistently indicate that
29 TCE exposures can result in adverse reproductive outcomes. Although the epidemiological data
30 may not always be robust or unequivocal, they demonstrate the potential for a wide range of
31 exposure-related adverse outcomes on female and male reproduction. In animal studies, there is
32 some evidence for female-specific reproductive toxicity; but there is strong and compelling
33 evidence for adverse effects of TCE exposure on male reproductive system and function.
34
35 4.8.1.3.1. Female reproductive toxicity. Although few epidemiological studies have examined
36 TCE exposure in relation to female reproductive function (see Table 4-77), the available studies
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-460 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
provide evidence of decreased fertility, as measured by time to pregnancy (Sallmen et al., 1995),
and effects on menstrual cycle patterns, including abnormal cycle length (ATSDR, 2001),
amenorrhea (Sagawa et al., 1973; Zielinski, 1973), and menstrual "disturbance" (Bardodej and
Vyskocil, 1956). In experimental animals, the effects on female reproduction include evidence
of reduced in vitro oocyte fertilizability in rats (Berger and Horner, 2003; Wu and Berger, 2007).
However, in other studies that assessed reproductive outcome in female rodents (Cosby and
Dukelow, 1992; George et al., 1985, 1986; Manson et al., 1984), there was no evidence of
adverse effects of TCE exposure on female reproductive function. Overall, although the data are
suggestive, there are inadequate data to make conclusions as to whether adverse effects on
human female reproduction are caused by TCE.
Table 4-77. Summary of adverse female reproductive outcomes associated
with TCE exposures
Finding
Menstrual cycle disturbance
Reduced fertility
Species
Human
Humana
Ratb
Citation
ATSDR, 2001a
Bardodej and Vyskocil, 1956
Sagawa et al., 1973
Zielinski, 1973
Sallmen et al., 1995
Berger and Horner, 2003
Wu and Berger, 2007
aNot significant.
bln vitro oocyte fertilizability.
4.8.1.3.2. Male reproductive toxicity. Notably, the results of a number of studies in both
humans and experimental animals have suggested that exposure to TCE can result in targeted
male reproductive toxicity (see Table 4-78). The adverse effects that have been observed in both
male humans and male animal models include altered sperm count, morphology, or motility
(Chia et al., 1996; George et al., 1985; Kumar et al, 2000a, b, 2001; Land et al., 1981;
Rasmussen et al., 1988; Veeramachaneni et al., 2001); decreased libido or copulatory behavior
(Bardodej and Vyskocil, 1956; El Ghawabi et al., 1973; George et al., 1986; Saihan et al., 1978;
Veeramachaneni et al., 2001; Zenick et al., 1984); alterations in serum hormone levels
(Chia et al., 1997; Goh et al., 1998; Kumar et al., 2000b; Veeramachaneni et al., 2001); and
reduced fertility (George et al., 1986). However, other studies in humans did not see evidence of
altered sperm count or morphology (Rasmussen et al., 1988) or reduced fertility (Forkert et al.,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-461 DRAFT—DO NOT CITE OR QUOTE
-------
1 2003; Sallmen et al., 1998), and some animal studies also did not identify altered sperm
2 measures (Cosby and Dukelow, 1992; Xu et al., 2004; Zenick et al., 1984; George et al, 1986).
3 Additional adverse effects observed in animals include histopathological lesions of the testes
4 (George et al., 1986; Kumar et al., 2000a, 2001) or epidiymides (Forkert et al., 2002; Kan et al.,
5 2007) and altered in vitro sperm-oocyte binding and/or in vivo fertilization for TCE and/or its
6 metabolites (Xu et al., 2004; DuTeaux et al., 2004b).
7 In spite of the preponderance of studies demonstrating effects on sperm parameters, there
8 is an absence of overwhelming evidence in the database of adverse effects of TCE on overall
9 fertility in the rodent studies. That is not surprising, however, given the redundancy and
10 efficiency of rodent reproductive capabilities. Nevertheless, the continuous breeding
11 reproductive toxicity study in rats (George et al., 1986) did demonstrate a trend towards
12 reproductive compromise (i.e., a progressive decrease in the number of breeding pairs producing
13 third, fourth, and fifth litters).
14 It is noted that in the studies by George et al. (1985, 1986), adverse reproductive
15 outcomes in male rats and mice were observed at the highest dose level tested (0.060% TCE in
16 diet) which was also systemically toxic (i.e., demonstrating kidney toxicity and liver enzyme
17 induction and toxicity, sometimes in conjunction with body weight deficits). Because of this, the
18 study authors concluded that the observed reproductive toxicity was a secondary effect of
19 generalized systemic toxicity; however, this conclusion is not supported by the overall
20 toxicological profile of TCE which provides significant evidence indicating that TCE is a
21 reproductive toxicant.
22
23 4.8.1.3.2.1. The role of metabolism in male reproductive toxicity. There has been particular
24 focus on evidence of exposure to male reproductive organs by TCE and/or its metabolites, as
25 well as the role of TCE metabolites in the observed toxic effects.
26 In humans, a few studies demonstrating male reproductive toxicity have measured levels
27 of TCE in the body. U-TCA was measured in men employed in an electronics factory, and
28 adverse effects observed included abnormal sperm morphology and hyperzoospermia and altered
29 serum hormone levels (Chia et al., 1996, 1997; Goh et al., 1998). U-TCA was also measured as
30 a marker of exposure to TCE in men occupationally exposed to solvents, although this study did
31 not report any adverse effects on fertility (Sallmen et al., 1998).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-462 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-78. Summary of adverse male reproductive outcomes associated
with TCE exposures
Finding
Testicular toxicity /pathology
Epididymal toxicity/pathology
Decreased sperm quantity /quality
Altered in vitro sperm-oocyte binding or in vivo
fertilization
Altered sexual drive or function
Altered serum testosterone levels
Reduced fertility
Gynaecomastia
Species
Rat
Mouse
Mouse
Human
Rat
Mouse
Rabbit
Rat
Mouse
Human
Rat
Rabbit
Human
Rat
Rabbit
Rat
Human
Citation
George et al., 1986
Kumar et al., 2000a
Kumar etal., 2001
Kan et al., 2007
Forkert et al., 2002
Chiaetal., 1996
Rasmussen et al., 1988a
Kumar et al., 2000a, b, 2001
George etal., 1985
Land etal., 1981
Veeramachaneni et al., 2001
DuTeaux et al., 2004b
Cosby and Dukelow, 1992b
Xu et al., 2004b
El Ghawabi et al., 1973
Saihanetal., 1978C
Bardodej and Vyskocil, 1956
George et al., 1986
Zenicketal., 1984
Veeramachaneni et al., 2001
Chiaetal., 1997d
Gohetal., 1998e
Kumar et al., 2000b
Veeramachaneni et al., 2001
George et al., 1986
Saihanetal., 1978C
2
3
4
5
6
7
a Nonsignificant increase in percentage of two YFF in spermatozoa; no effect on sperm count or morphology.
b Observed with metabolite(s) of TCE only.
0 Case study of one individual.
d Also observed altered levels of DHEAS, FSH, and SHBG.
e Also observed altered levels of SHBG.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-463 DRAFT—DO NOT CITE OR QUOTE
-------
1 In the study in Long-Evans male rats by Zenick et al. (1984), blood and tissue levels of
2 TCE, TCA, and TCOH were measured in three rats/group following 6 weeks of gavage treatment
3 at 0, 10, 100, and 1,000 mg/kg/d. Additionally the levels of TCE and metabolites were measured
4 in seminal plugs recovered following copulation at Week 5. Marked increases in TCE levels
5 were observed only at 1,000 mg/kg/d, in blood, muscle, adrenals, and seminal plugs. It was
6 reported that dose-related increases in TCA and TCOH concentrations were observed in the
7 organs evaluated, notably including the reproductive organs (epididymis, vas deferens, testis,
8 prostate, and seminal vesicle), thus, creating a potential for interference with reproductive
9 function.
10 This potential was explored further in a study by Forkert et al. (2002), in which male
11 CD-I mice were exposed by inhalation to 1,000-ppm TCE (6 hours/day, 5 day/week) for
12 4 consecutive weeks. Urine was obtained on Days 4, 9, 14, and 19 of exposure and analyzed for
13 concentrations of TCE and TCOH. Microsomal preparations from the liver, testis and
14 epididymis were used for immunoblotting, determining/>-nitrophenol hydroxylase and CYP2E1
15 activities, and evaluating the microsomal metabolism of TCE.
16 Subsequent studies conducted by the same laboratory (Forkert et al., 2003) evaluated the
17 potential of the male reproductive tract to accumulate TCE and its metabolites including chloral,
18 TCOH, TCA, and DCA. Human seminal fluid and urine samples from eight mechanics
19 diagnosed with clinical infertility and exposed to TCE occupationally were analyzed. Urine
20 samples from two of the eight subjects contained TCA and/or TCOH, suggesting that TCE
21 exposure and/or metabolism was low during the time just prior to sample collection. TCE,
22 chloral, and TCOH were detected in seminal fluid samples from all eight subjects, while TCA
23 was found in one subject, and DCA was found in two subjects. Additionally, TCE and its
24 metabolites were assessed in the epididymis and testis of CD-I mice (4/group) exposed by
25 inhalation (6 hours/day, 5 days/week) to 1,000 ppm TCE for 1, 2, and 4 weeks. TCE, chloral,
26 and TCOH were found in the epididymis at all timepoints, although TCOH levels were increased
27 significantly (tripled) at four weeks of exposure. This study showed that the metabolic
28 disposition of TCE in humans is similar to that in mice, indicating that the murine model is
29 appropriate for investigating the effects of TCE-induced toxicity in the male reproductive
30 system. These studies provide support for the premise that TCE is metabolized in the human
31 reproductive tract, mainly in the epididymis, resulting in the production of metabolites that cause
32 damage to the epididymal epithelium and affect the normal development of sperm.
33 Immunohistochemical experiments (Forkert et al., 2002) confirmed the presence of
34 CYP2E1 in the epididymis and testis of mice; it was found to be localized in the testicular
35 Ley dig cells and the epididymal epithelium. Similar results were obtained with the
36 immunohistochemical evaluation of human and primate tissue samples. CYP2E1 has been
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-464 DRAFT—DO NOT CITE OR QUOTE
-------
1 previously shown by Lipscomb et al. (1998) to be the predominant CYP enzyme catalyzing the
2 hepatic metabolism of TCE in both animals and rodents. These findings support the role of
3 CYP2E1 in TCE metabolism in the male reproductive tract of humans, primates, and mice.
4
5 4.8.1.3.2.2. Mode of action for male reproductive toxicity. A number of studies have been
6 conducted to attempt to characterize various aspects of the mode of action for observed male
7 reproductive outcomes.
8 Studies by Kumar et al. (2000b, 2001) suggest that perturbation of testosterone
9 biosynthesis may have some role in testicular toxicity and altered sperm measures. Significant
10 decreases in the activity of G6PDH and accumulation of cholesterol are suggestive of an
11 alteration in testicular steroid biosynthesis. Increased testicular lipids, including cholesterol,
12 have been noted for other testicular toxicants such as lead (Saxena et al., 1987),
13 triethylenemelamine (Johnson et al., 1967), and quinalphos (Ray et al., 1987), in association with
14 testicular degeneration and impaired spermatogenesis. Since testosterone has been shown to be
15 essential for the progression of spermatogenesis (O'Donnell et al., 1994), alterations in
16 testosterone production could be a key event in male reproductive dysfunction following TCE
17 exposure. Additionally, the observed TCE-related reduction of 17-P-HSD, which is involved in
18 the conversion of androstenedione to testosterone, has also been associated with male
19 reproductive insufficiency following exposure to phthalate esters (Srivastava and Srivastava,
20 1991), quinalphos (Ray et al., 1987), and lead (Saxena et al., 1987). Reductions in SDH, which
21 are primarily associated with the pachytene spermatocyte maturation of germinal epithelium,
22 have been shown to be associated with depletion of germ cells (Mills and Means, 1970;
23 Chapin et al., 1982), and the activity of G6PDH is greatest in premeiotic germ cells and Leydig
24 cells of the interstitium (Blackshaw et al., 1970). The increased GT and glucuronidase observed
25 following TCE exposures appear to be indicative of impaired Sertoli cell function (Hodgen and
26 Sherins, 1973; Sherins and Hodgen, 1976). Based upon the conclusions of these studies,
27 Kumar et al. (2001) hypothesized that the reduced activity of G6PDH and SDH in testes of
28 TCE-exposed male rats is indicative of the depletion of germ cells, spermatogenic arrest, and
29 impaired function of the Sertoli cells and Leydig cells of the interstitium.
30 In the series of experiments by DuTeaux et al. (2003, 2004b), protein dichloroacetyl
31 adducts were found in the corpus epididymis and in the efferent ducts of rats administered TCE;
32 this effect was also demonstrated following in vitro exposure of reproductive tissues to TCE.
33 Oxidized proteins were detected on the surface of spermatozoa from TCE-treated rats in a
34 dose-response pattern; this was confirmed using a Western blotting technique. Soluble (but not
35 mitochondrial) cysteine-conjugate p-lyase was detected in the epididymis and efferent ducts of
36 treated rats. Following a single intraperitoneal injection of DCVC, no dichloroacetylated protein
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-465 DRAFT—DO NOT CITE OR QUOTE
-------
1 adducts were detected in the epididymis and efferent ducts. The presence of CYP2E1 was found
2 in epididymis and efferent ducts, suggesting a role of cytochrome P450-dependent metabolism
3 in adduct formation. An in vitro assay was used to demonstrate that epididymal and efferent
4 duct microsomes are capable of metabolizing TCE; TCE metabolism in the efferent ducts was
5 found to be inhibited by anti-CYP2El antibody. Lipid peroxidation in sperm, presumably
6 initiated by free radicals, was increased in a significant (p < 0.005) dose-dependent manner after
7 TCE-exposure.
8 Overall, it has been suggested (DuTeaux et al., 2004b) that reproductive organ toxicities
9 observed following TCE exposure are initiated by metabolic bioactivation, leading to subsequent
10 protein adduct formation. It has been hypothesized that epoxide hydrolases in the rat epididymis
11 may play a role in the biological activation of metabolites (DuTeaux et al., 2004a).
12 4.8.1.3.3. Summary of noncancer reproductive toxicity. The toxi col ogical database for TCE
13 includes a number of studies that demonstrate adverse effects on the integrity and function of the
14 reproductive system in females and males. Both the epidemiological and animal toxicology
15 databases provide suggestive, but limited, evidence of adverse outcomes to female reproductive
16 outcomes. However, much more extensive evidence exists in support of an association between
17 TCE exposures and male reproductive toxicity. The available epidemiol ogical data and case
18 reports that associate TCE with adverse effects on male reproductive function are limited in size
19 and provide little quantitative dose data (Lamb and Hentz, 2006). However, the animal data
20 provide extensive evidence of TCE-related male reproductive toxicity. Strengths of the database
21 include the presence of both functional and structural outcomes, similarities in adverse
22 treatment-related effects observed in multiple species, and evidence that metabolism of TCE in
23 male reproductive tract tissues is associated with adverse effects on sperm measures in both
24 humans and animals (suggesting that the murine model is appropriate for extrapolation to human
25 health risk assessment). Additionally some aspects of a putative MOA (e.g., perturbations in
26 testosterone biosynthesis) appear to have some commonalities between humans and animals.
27
28 4.8.2. Cancers of the Reproductive System
29 The effects of TCE on cancers of the reproductive system have been examined for males
30 and females in both epidemiol ogical and experimental animal studies. The epidemiol ogical
31 literature includes data on prostate in males and cancers of the breast and cervix in females. The
32 experimental animal literature includes data on prostate and testes in male rodents; and uterus,
33 ovary, mammary gland, vulva, and genital tract in female rodents. The evidence for these
34 cancers is generally not robust.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-466 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.2.1. Human Data
2 The epidemiologic evidence on TCE and cancer of the prostate, breast, and cervix is from
3 cohort and geographic based studies. Two additional case-control studies of prostate cancer in
4 males are nested within cohorts (Greenland et al., 1994; Krishnadasan et al., 2007). The nested
5 case-control studies are identified in Tables 4-79-4-81 with cohort studies given their source
6 population for case and control identification. One population-based case-control study
7 examined on TCE exposure and prostate (Siemiatycki, 1991); however, no population case-
8 control studies on breast or cervical cancers and TCE exposure were found in the peer-reviewed
9 literature.
10
11 4.8.2.1.1. Prostate cancer. Sixteen cohort or PMR studies, two nested case-control, one
12 population case-control, and two geographic-based studies present relative risk estimates for
13 prostate cancer (Wilcosky et al., 1984; Garabrant et al., 1988; Shannon et al., 1988; Blair et al.,
14 1989; Axelson et al., 1994; Siemiatycki, 1991; Greenland et al., 1994; Anttila et al., 1995; Blair
15 et al., 1998; Morgan et al., 1998; Boice et al., 1999, 2006; Ritz, 1999; Hansen et al., 2001;
16 Morgan and Cassady, 2002; Raaschou-Nielsen et al., 2003; Chang et al., 2003, 2005; ATSDR,
17 2004, 2006; Krishnadasan et al., 2007; Radican et al., 2008). Three small cohort studies (Costa
18 et al., 1989; Sinks et al., 1992; Henschler et al., 1995), one multiple-site population case-control
19 (Siemiatycki, 1991) and one geographic based study (Vartiainen et al., 1993) do not report
20 estimates for prostate cancer in their published papers. Twelve of the 19 studies with prostate
21 cancer relative risk estimates had high likelihood of TCE exposure in individual study subjects
22 and were judged to have met, to a sufficient degree, the standards of epidemiologic design and
23 analysis (Siemiatycki, 1991; Axelson et al., 1994; Anttila et al., 1994; Greenland et al., 1994,
24 Blair et al., 1998; Morgan et al., 1998, 2000; Boice et al., 1999, 2006; Hansen et al., 2001;
25 Raaschou-Nielsen et al., 2003; Krishnadasan et al., 2007; Radican et al., 2008). Krishnadasan et
26 al. (2007) in their nested case-control study of prostate cancer observed a 2-fold odds ratio
27 estimate with high cumulative TCE exposure score (2.4, 95% CI: 1.3, 4.4, 20 year lagged
28 exposure) and an increasing positive relationship between prostate cancer incidence and TCE
29 cumulative exposure score (p = 0.02). TCE exposure was positively correlated with several
30 other occupational exposures, and Krishnadasan et al. (2007) adjusted for possible confounding
31 from all other chemical exposures as well as age at diagnosis, occupational physical activity, and
32 socio-economic status in statistical analyses. Relative risk estimates in studies other than
33 Krishnadasan et al. (2007) were above 1.0 for overall TCE exposure (1.8, 95% CI: 0.8, 4.0
34 [Siemiatycki, 1991]; 1.1, 95% CI: 0.6, 1.8 [Blair et al., 1998] and 1.20, 95% CI: 0.92, 1.76, with
35 an additional 10-year follow-up [Radican et al., 2008]; 1.58, 95% CI: 0.96, 2.62 [Morgan et al.,
36 1998, 2000; Environmental Health Strategies, 1997]; 1.3, 95% CI: 0.52, 2.69 [Boice et al.,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-467 DRAFT—DO NOT CITE OR QUOTE
-------
1 1999]; 1.38, 95% CI: 0.73, 2.35 [Anttila et al., 1995]) and prostate cancer risks did not appear to
2 increase with increasing exposure. Four studies observed relative risk estimates below 1.0 for
3 overall TCE exposure (0.93, 95% CI: 0.60, 1.37 [Garabrant et al., 1988]; 0.6, 95% CI: 0.2, 1.30
4 [Hansen et al., 2001]; 0.9, 95% CI: 0.79, 1.08 [Raaschou-Nielsen et al., 2003]; 0.82, 95% CI:
5 0.36, 1.62 [Boice et al., 2006]), and are not considered inconsistent because alternative
6 explanations are possible and included observations are based on few subjects, lowering
7 statistical power, or to poorer exposure assessment approaches that may result in a higher
8 likelihood of exposure misclassification.
9 Seven other cohort, PMR, and geographic based studies were given less weight in the
10 analysis because of their lesser likelihood of TCE exposure and other study design limitations
11 that would decrease statistical power and study sensitivity (Wilcosky et al., 1984; Shannon et al.,
12 1988; Blair et al., 1989; Morgan and Cassady, 2002; ATSDR, 2004, 2006; Chang et al., 2005).
13 Chang et al. (2005) observed a statistically significant deficit in prostate cancer risk, based on
14 one case, and an insensitive exposure assessment (0.14, 95% CI: 0.00, 0.76). Relative risks in
15 the other five studies ranged from 0.62 (CI not presented in paper) (Wilcosky et al., 1984) to
16 1.11 (95% CI: 0.98, 1.25) (Morgan and Cassady, 2002).
17 Risk factors for prostate cancer include age, family history of prostate cancer, and
18 ethnicity as causal with inadequate evidence for a relationship with smoking or alcohol
19 (Wigle et al., 2008). All studies except Krishnadasan et al. (2007) were not able to adjust for
20 possible confounding from other chemical exposures in the work environment. None of the
21 studies including Krishnadasan et al. (2007) accounted for other well-established
22 nonoccupational risk factors for prostate cancer such as race, prostate cancer screening and
23 family history. There is limited evidence that physical activity may provide a protective effect
24 for prostate cancer (Wigle et al., 2008). Krishnadasan et al. (2008) examined the effect of
25 physical activity in the Rocketdyne aerospace cohort (Zhao et al., 2005; Krishnadasan et al.,
26 2007). Their finding of a protective effect with high physical activity (0.55, 95% CI: 0.32, 0.95,
27 p trend = 0.04) after control for TCE exposure provides additional evidence (Krishnadasan et al.,
28 2008) and suggests underlying risk may be obscured in studies lacking adjustment for physical
29 activity.
30
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-468 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-79. Summary of human studies on TCE exposure and prostate
cancer
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
Events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Low/moderate TCE score
High TCE score
p for trend
Low/moderate TCE score
High TCE score
p for trend
All employees at electronics factory (Taiwan)
1.3 (0.81, 2.1fb
2.1(1.2, 3.9fb
0.02
1.3 (0.81, 2.1)a'c
2.4(1.3,4.4)a'c
0.01
0.14(0.00, 0.76)d
90
45
1
Danish blue-collar worker with TCE exposure
Any exposure
0.9 (0.79, 1.08)
163
Biologically -monitored Danish workers
Any TCE exposure, females
0.6 (0,2, 1.3)
6
Aircraft maintenance workers (Hill Air Force Base, UT)
TCE subcohort
Cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Biologically -monitored Finnish workers
Mean air-TCE (Ikeda extrapolation
<6ppm
6+ppm
Not reported
1.0e
1.1(0.7, 1.6)
1.0 (0.6, 1.6)
1.2(0.8, 1.8)
1.2 (0.92, 1.76)
1.0"
1.03 (0.65, 1.62)
1.33(0.82,2.15)
1.31(0.84,2.06)
1.38(0.73,2.35)
1.43 (0.62, 2.82)
0.68 (0.08, 2.44)
158
64
38
56
116
41
42
43
13
8
2
Cardboard manufacturing workers in Arnsburg, Germany
Exposed workers
Biologically -monitored Swedish workers
Cardboard manufacturing workers, Atlanta area, GA
Not reported
1.25 (0.84, 1.84)
Not reported
26
Krishnadasan et al., 2007
Chang et al., 2005
Raaschou-Nielsen et al.,
2003
Hansenetal., 2001
Blair etal., 1998
Radican et al. 2008
Anttila et al., 1995
Henschleretal., 1995
Axelson et al., 1994
Sinks etal., 1992
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-469 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-79. Summary of human studies on TCE exposure and prostate
cancer (continued)
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
Events
Reference
Cohort and PMR-mortality
Aerospace workers (Rocketdyne)
Any TCE (utility/eng flush)
View-Master employees
All employees at electronics factory (Taiwan)
0.82 (0.36, 1.62)
1.69 (0.68, 3.48)f
Not reported
8
8
0
Fernald workers
Any TCE exposure
Light TCE exposure, >2 yrs duration
Moderate TCE exposure, >2 yrs duration
Not reported
0.91(0.38, 2.18)e'g
1.44(0.19, 11.4)e'g
10
1
Aerospace workers (Lockheed)
Routine exposure to TCE
Routine-intermittent
1.31 (0.52,2.69)
Not reported
7
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
1.18(0.73, 1.80)
1.03 (0.51, 1.84)
0.47(0.15, 1.11)
21
7
14
TCE subcohort (Cox Analysis)
Never exposed
Ever exposed
1.00e
1.58 (0.96, 2.62)h
Peak
No/Low
Medium/high
1.00e
1.39 (0.80, 2.41)h
Cumulative
Referent
Low
High
1.00e
1.72 (0.78, 3.80)h
1.53 (0.85, 2.75)h
Aircraft maintenance workers (Hill Air Force Base, UT)
TCE subcohort
1.1 (0.6, 1.8)
54
Cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0"
0.9(0.5, 1.8)
1.0(0.5,2.1)
1.3 (0.7, 2.4)
19
13
22
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Deaths reported to GE pension fund (Pittsfield, MA)
Cardboard manufacturing workers, Atlanta area, GA
Not reported
0.82 (0.46, 1.46)a
Not reported
58
0
Boiceetal.,2006
ATSDR, 2004
Chang etal., 2003
Ritz, 1999
Boiceetal., 1999
Morgan et al., 1998, 2000
Blair etal., 1998
Henschleretal., 1995
Greenland etal., 1994
Sinks etal., 1992
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-470 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-79. Summary of human studies on TCE exposure and prostate
cancer (continued)
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
Events
U. S. Coast Guard employee
Marine inspectors
Noninspectors
1.06(0.51, 1.95)
0.57(0.15, 1.45)
10
7
Aircraft manufacturing plant employees (Italy)
Aircraft manufacturing plant employees (San Diego,
CA)
Lamp manufacturing workers (GE)
0.93 (0.60, 1.37)
1.56 (0.63, 3.22)
25
7
Rubber workers
Any TCE exposure
0.62 (not reported)
3
Reference
Blair etal., 1989
Costa etal., 1989
Garabrantetal., 1988
Shannon et al., 1988
Wilcoskyetal., 1984
Case-control studies
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
1.1(0.6,2.1)'
1.8 (0.8, 4.0)1
11
7
Siemiatycki, 1991
Geographic based studies
Residents in two study areas in Endicott, NY
Residents of 13 census tracts inRedlands, CA
1.05 (0.75, 1.43)
1.11(0.98, 1.25)1
40
483
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
ATSDR, 2006
Morgan and Cassady, 2002
Vartiainenetal., 1993
1
2
O
4
5
6
7
8
9
10
11
12
13
"Odds ratio from nested case-control study.
bOdds ratio, zero lag.
°Odds ratio, 20 year lag.
dChang et al. (2005) presents SIRs for a category site of all cancers of male genital organs.
Internal referents, workers without TCE exposure.
Proportional mortality ratio.
8Analysis for >2 years exposure duration and a lagged TCE exposure period of 15 years.
hRisk ratio from Cox Proportional Hazard Analysis, stratified by age and sex, from Environmental Health Strategies
(1997) Final Report to Hughes Corporation (Communication from Paul A. Cammer, President, Trichloroethylene
Issues Group to Cheryl Siegel Scott, U.S. EPA, December 22, 1997).
'90% confidence interval.
J99% confidence interval.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-471 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-80. Summary of human studies on TCE exposure and breast cancer
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any TCE exposure
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
All employees at electronics factory (Taiwan)
Females
Females
Not reported
1.09 (0.96, 1.22)a
1.19(1.03, 1.36)
286
215
Danish blue-collar worker with TCE exposure
Any exposure, males
Any exposure, females
0.5 (0.06, 1.90)
1.1 (0.89, 1.24)
2
145
Biologically -monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
0.9 (0.2, 2.3)
0
(0.2 exp)
4
Aircraft maintenance workers (Hill Air Force Base, UT)
TCE subcohort
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Biologically -monitored Finnish workers
Not reported
1.0b
0.3 (0.1, 1.4)
0.4 (0.1,2.9)
0.4 (0.4, 1.2)
Not reported
34
20
11
3
Cardboard manufacturing workers in Arnsburg, Germany
Exposed workers
Biologically -monitored Swedish workers
Cardboard manufacturing workers, Atlanta area, GA
Not reported
Not reported
Not reported
Zhao et al., 2005
Sung etal., 2007
Chang et al., 2005
Raaschou-Nielsen et al.,
2003
Hansen etal., 2001
Blair etal., 1998
Anttila et al., 1995
Henschleretal., 1995
Axelson et al., 1994
Sinks etal., 1992
Cohort and PMR-mortality
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
Not reported
Not reported
Not reported
Not reported
Not reported
View-Master employees
Males
Females
1.02 (0.67, 1.49)c
0
(0.05 exp)
27
Boice etal., 2006
Zhao etal., 2005
ATSDR, 2004
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-472 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-80. Summary of human studies on TCE exposure and breast cancer
(continued)
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
events
Fernald workers
Any TCE exposure
Light TCE exposure, >2 yrs duration
Moderate TCE exposure, >2 yrs
duration
Not reported
Not reported
Not reported
Aerospace workers (Lockheed)
Routine exposure to TCE
Routine-intermittent3
1.31(0.52, 2.69)d
Not reported
7
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
0.75 (0.43, 1.22)d
1.03 (0.51, 1.84)d
0.47(0.15, l.ll)d
16
11
5
TCE subcohort (Cox Analysis)
Never exposed
Ever exposed
1.00d
0.94(0.51, 1.75) d'e
NR
NR
Peak
No/Low
Medium/high
1.00d
1.14(0.48, 2.70)^
NR
Cumulative
Referent
Low
High
1.00b
1.20 (0.60, 2.40)d'e
0.65 (0.25, 1.69)^
NR
NR
Aircraft maintenance workers (Hill Air Force Base, UT)
TCE subcohort (females)
2.0 (0.9, 4.6)
20
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Low level intermittent exposure
Low level continuous exposure
Frequent peaks
TCE subcohort (females)
1.0b
2.4(1.1,5.2)
1.2 (0.3, 5.4)
1.4 (0.6, 3.2)
3.1(1.5,6.2)
3.4(1.4,8.0)
1.4 (0.7, 3.2)
1.23 (0.73, 2.06)
10
21
8
15
8
10
26
Reference
Ritz, 1999
Boiceetal., 1999
Morgan et al., 1998
Blair etal., 1998
Radicanetal.,2008
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-473 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-80. Summary of human studies on TCE exposure and breast cancer
(continued)
Studies
Exposure group
Relative risk
(95% CI)
No. obs.
events
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Low level intermittent exposure
Low level continuous exposure
Frequent peaks
1.0b
1.57 (0.81, 3.04)
1.01(0.31,3.30)
1.05 (0.53, 2.07)
1.92 (1.08, 3.43)
1.71 (0.79, 3.71)
1.08 (0.57, 2.02)
12
3
11
18
8
14
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Deaths reported to GE pension fund (Pittsfield, MA)
Cardboard manufacturing workers, Atlanta area, GA
Not examined
Not reported
Not reported
0
U. S. Coast Guard employees
Marine inspectors
Noninspectors
Aircraft manufacturing plant employees (Italy)
Not reported
Not reported
Not reportedf
Aircraft manufacturing plant employees (San Diego, CA)
All subjects, females
Lamp manufacturing workers (GE)
Coil/wire drawing
Other areas
0.81 (0.52, 1.48)d
2.04 (0.88, 4.02)
0.97 (0.57, 1.66)
16
8
13
Reference
Henschleretal., 1995
Greenland etal., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrantetal., 1988
Shannon et al., 1988
Case-control Studies
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
Not reported
Not reported
Siemiatycki, 1991
Geographic Based Studies
Residents in two study areas in Endicott, NY
Residents of 13 census tracts inRedlands, CA
0.88(0.65, 1.18)
1.09 (0.97, 1.21)
46
536
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
ATSDR, 2006
Morgan and Cassady, 2002
Vartiainenetal., 1993
1
2
3
4
5
6
7
8
9
10
11
12
al5 year lag.
blnternal referents, workers not exposed to TCE.
0 Proportional mortality ratio.
dln Garabramt et al. (1998), Morgan et al. (1998) and Boice et al. (1999), breast cancer risk is for males and females
combined (ICD-9, 174, 175).
eRisk ratio from Cox Proportional Hazard Analysis, stratified by age and sex, from Environmental Health Strategies
(1997) Final Report to Hughes Corporation (Communication from Paul A. Cammer, President, Trichloroethylene
Issues Group to Cheryl Siegel Scott, U.S. EPA, December 22, 1997).
fThe cohort of Blair et al. (1989) and Costa et al. (1989) are composed of males only.
NR = not reported
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-474 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-81. Summary of human studies on TCE exposure and cervical
cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
All employees at electronics factory (Taiwan)
Not reported
Not reported
0.96 (0.86, 1.22)a
337
Danish blue-collar worker w/TCE exposure
Any exposure
1.9 (1.42,2.37)
62
Exposure lag time
20yrs
1.5 (0.7, 2.9)
9
Employment duration
5yrs
2.5(1.7,3.5)
1.6(1.0,2.4)
1.3 (0.6, 2.4)
30
22
10
Biologically -monitored Danish workers
Any TCE exposure
3.8(1.0,9.8)
4
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-yr
2.9 (0.04, 16)
2.6 (0.03, 14)
1
1
Mean concentration (Ikeda)
<4ppm
4+ppm
3.4 (0.4, 12)
4.3 (0.5, 16)
2
2
Employment duration
<6.25 yr
>6.25 yr
3.8(0.1,21)
2.1(0.03, 12)
1
1
Aircraft maintenance workers from Hill Air Force Base, UT
TCE subcohort
Cumulative exposure
Not reported
Not reported
Biologically -monitored Finnish workers
All subjects
2.42 (1.05, 4.77)
8
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
1.86 (0.38, 5.45)
4.35(1.41, 10.1)
o
J
5
Cardboard manufacturing workers in Arnsburg, Germany
Exposed workers
Not reported
Zhao et al., 2005
Sung etal., 2007
Raaschou-Nielsen et al.,
2003
Hansen etal., 2001
Blair etal., 1998
Anttila et al., 1995
Henschleretal., 1995
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-475 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-81. Summary of human studies on TCE exposure and cervical
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Biologically -monitored Swedish workers
Any TCE exposure
Not reported
Cardboard manufacturing workers, Atlanta area, GA
All subjects
Not reported
Reference
Axelson et al., 1994
Sinks etal., 1992
Cohort studies-mortality
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Any exposure to TCE
Not reported
Not reported
View-Master employees
Females
1.77 (0.57, 4.12)b
5
United States uranium-processing workers (Fernald, OH)
Any TCE exposure
Light TCE exposure, >2 yrs duration
Moderate TCE exposure, >2 yrs duration
Not reported
Not reported
Not reported
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent
- (0.00, 5.47)
Not reported
0
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
(0.00, 1.07)
0
(3.5 exp)
0
(1.91 exp)
0
(1.54 exp)
Aircraft maintenance workers (Hill AFB, UT)
TCE subcohort
1.8 (0.5, 6.5)c
5
Cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE sucohort
1.0°
0.9(0.1,8.3)
3.0(0.8, 11.7)
1.67 (0.54, 5.22)
1
0
4
6
Cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0C
0.76 (0.09, 6.35)
2.83 (0.86, 9.33)
1
0
5
Boiceetal.,2006
Zhao etal., 2005
ATSDR, 2004
Ritz, 1999
Boiceetal., 1999
Morgan etal., 1998
Blair etal., 1998
Radican etal., 2008
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-476 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-81. Summary of human studies on TCE exposure and cervical
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Unexposed workers
Deaths reported to GE pension fund (Pittsfield, MA)
Cardboard manufacturing workers, Atlanta area, GA
U. S. Coast Guard employees
Aircraft manufacturing plant employees (Italy)
Not reported
Not reported
Not examinedd
Not reported
Not reported6
Not reported6
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
0.61 (0.25, 1.26)f
7
Lamp manufacturing workers (GE)
Coil/wire drawing
Other areas
1.05 (0.03, 5.86)
1.16(0.32,2.97)
1
4
Reference
Henschleretal., 1995
Greenland etal., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrantetal., 1988
Shannon et al., 1988
Case-control studies
Geographic based studies
Residents in two study areas in Endicott, NY
Residents in Texas
Counties reporting any air TCE release
Countires not reporting any air TCE
relesease
Residents of 13 census tracts in Redlands, CA
1.06 (0.29, 2.71)
66.4g
60.8s
0.65 (0.38, 1.02)
<6
29
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
ATSDR, 2006
Coyle et al, 2005
Morgan and Cassady, 2002
Vartiainenetal., 1993
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
"Standardized incidence ratio for females in Sung et al. (2007) reflects a 15-year lag period.
bProportional mortality ratio.
Internal referents, workers not exposed to TCE.
dNested case-control analysis.
eMales only in cohort.
fSMR is for cancer of the genital organs (cervix, uterus, endometrium, etc.).
8 Median annual age-adjusted breast cancer rate (1995-2000).
4.8.2.1.2. Breast cancer. Fifteen studies of TCE exposure reported findings on breast cancer
in males and females combined (Garabrant et al., 1988; Greenland et al., 1994; Boice et al.,
1999), in males and females, separately (Hansen et al., 2001; Raaschou-Nielsen et al., 2003;
ATSDR, 2004; Clapp and Hoffman, 2008), or in females only (Shannon et al., 1988; Blair et al.,
1998; Morgan et al., 1998; Coyle et al., 2005; ATSDR, 2006; Chang et al., 2005; Sung et al.,
2007; Radican et al., 2008). Six studies have high likelihood of TCE exposure in individual
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-477 DRAFT—DO NOT CITE OR QUOTE
-------
1 study subjects and met, to a sufficient degree, the standards of epidemiologic design and analysis
2 (Blair et al., 1998; Morgan et al., 1998; Boice et al., 1999; Hansen et al., 2001; Raaschou-Nielsen
3 et al., 2003; Radican et al. 2008). Four other high-quality studies with risk estimates for other
4 cancer sites do not report risk estimates for breast cancer (Siemiatycki, 1991; Axelson et al.,
5 1994; Anttila et al., 1995; Boice et al., 2006). No case-control studies were found on TCE
6 exposure, although several studies examine occupational title or organic solvent as a class
7 (Weiderpass et al., 1999; Band et al., 2000; Rennix et al., 2005; Ji et al., 2008). While
8 association is seen with occupational title or industry and breast cancer (employment in aircraft
9 and aircraft part industry, 2.48, 95% CI: 1.14, 5.39 [Band et al., 2000]; solvent user: 1.48,
10 95% CI: 1.03, 2.12 [Rennix et al., 2005]), TCE exposure is not uniquely identified. The two
11 studies suggest association between organic solvents and female breast cancer needs further
12 investigation of possible risk factors.
13 Relative risk estimates in the five high-quality studies ranged from 0.75 (0.43, 1.22)
14 (females and males; Morgan et al., 1998) to 2.0 (0.9, 4.6) (mortality in females; Blair et al.,
15 1998). Blair et al. (1998), additionally, observed stronger risk estimates for breast cancer
16 mortality among females with low level intermittent (3.1, 95% CI: 1.5, 6.2) and low level
17 continuous (3.4, 95% CI: 1.4, 8.0) TCE exposures, but not with frequent peaks (1.4, 95% CI: 0.7,
18 3.2). A similar pattern of risks was also observed by Radican et al. (2008) who studied mortality
19 in this cohort and adding 10 years of follow-up, although the magnitude of breast cancer risk in
20 females was lower than that observed in Blair et al. (1998). Risk estimates did not appear to
21 increase with increasing cumulative exposure in the two studies that included exposure-response
22 analyses (Blair et al., 1998; Morgan et al., 1998). None of the five high quality studies reported
23 a statistically significant deficit in breast cancer and confidence intervals on relative risks
24 estimates included 1.0 (no risk). Few female subjects in these studies appear to have high TCE
25 exposure. For example, Blair et al. (1998) identified 8 of the 28 breast cancer deaths and 3 of the
26 34 breast cancer cases with high cumulative exposure.
27 Relative risk estimates in six studies of lower likelihood TCE exposure and other design
28 deficiencies ranged from 0.81 (95% CI: 0.52, 1.48) (Garabrant et al., 1988) to 1.19 (1.03, 1.36)
29 (Chang et al., 2005). These studies lack a quantitative surrogate for TCE exposure to individual
30 subjects and instead classify all subjects as "potentially exposed", with resulting large dilution of
31 actual risk and decreased sensitivity (Garabrant et al., 1988; Shannon et al., 1988; Morgan and
32 Cassady, 2002; Chang et al., 2005; ATSDR, 2006; NRC, 2006; Sung et al., 2007).
33 Four studies reported on male breast cancer separately (Hansen et al., 2001; Raaschou-
34 Nielsen et al., 2003; ATSDR, 2004; Clapp and Hoffman, 2008) and a total of three cases were
35 observed. Breast cancer in men is a rare disease and is best studied using a case-control
36 approach (Weiss et al., 2005). Reports exist of male breast cancer among former residents of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-478 DRAFT—DO NOT CITE OR QUOTE
-------
1 Camp Lejuene (U. S. EPA, 2009). Further assessment of TCE exposure and male breast cancer is
2 warranted.
3 Overall, the epidemiologic studies on TCE exposure and breast cancer are quite limited in
4 statistical power; observations are based on few breast cancer cases in high-quality studies or on
5 inferior TCE exposure assessment in studies with large numbers of observed cases.
6 Additionally, adjustment for nonoccupational breast cancer risk factors is less likely in cohort
7 and geographic based studies given their use of employment and public records. Breast cancer
8 mortality observations in Blair et al. (1998) and further follow-up of this cohort by Radican et al.
9 (2008) of an elevated risk with overall TCE exposure, particulalry low level intermittent and
10 continuous TCE exposure, provide evidence of an association with TCE. No other high-quality
11 study reported a statistically significant association with breast cancer, although few observed
12 cases leading to lower statistical power or examination of risk for males and females combined
13 are alternative explanations for the null observations in these studies. Both Chang et al. (2005)
14 and Sung et al. (2007), two overlapping studies of female electronics workers exposed to TCE,
15 perchloroethylene, and mixed solvents, reported association with breast cancer incidence, with
16 breast cancer risk in Chang et al. (2005) appearing to increase with employment duration. Both
17 studies, in addition to association provided by studies of exposure to broader category of organic
18 solvents (Band et al., 2000; Rennix et al., 2005), support Blair et al. (1998) and Radican et al.
19 (2008), although the lack of exposure assessment is an uncertainty. The epidemiologic evidence
20 is limited for examining TCE and breast cancer, and while these studies do not provide any
21 strong evidence for association with TCE exposure they in turn do not provide evidence of an
22 absence of association.
23
24 4.8.2.1.3. Cervical cancer. Eleven cohort or PMR studies and 2 geographic based studies
25 present relative risk estimates (Garabrant et al., 1988; Shannon et al., 1988; Anttila et al., 1995;
26 Blair et al., 1998; Morgan et al., 1998; Boice et al., 1999; Hansen et al., 2001; Morgan and
27 Cassady, 2002; Raaschou-Nielsen et al., 2003; ATSDR, 2004, 2006; Sung et al., 2007; Radican
28 et al., 2008). Seven of these studies had high likelihood of TCE exposure in individual study
29 subjects and were judged to have met, to a sufficient degree, the standards of epidemiologic
30 design and analysis (Anttila et al., 1995; Blair et al., 1998; Morgan et al., 1998; Boice et al.,
31 1999; Hansen et al., 2001; Raaschou-Nielsen et al., 2003; Radican et al., 2008). Three small
32 cohort studies (Costa et al., 1989; Sinks et al., 1992; Henschler et al., 1995) as well as three high-
33 quality studies (Axelson et al., 1994; Zhao et al., 2005; Boice et al., 2006) did not present
34 relative risk estimates for cervical cancer. Additionally, one population case-control and one
35 geographic study of several site-specific cancers do not present information on cervical cancer
36 (Siemiatycki, 1991; Vartiainen et al., 1993).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-479 DRAFT—DO NOT CITE OR QUOTE
-------
1 Five high-quality studies observed elevated risk for cervical cancer and overall TCE
2 exposure (2.42, 95% CI: 1.05, 4.77 [Anttila et al., 1995]; 1.8, 95% CI: 0.5, 6.5 [Blair et al., 1998]
3 that changed little with an additional 10 years follow-up, 1.67, 95% CI: 0.54, 5.22
4 [Radican et al., 2008]; 3.8, 95% CI: 1.42, 2.37 [Hansen et al., 2001]; 1.9, 95% CI: 1.42, 2.37
5 [Raaschou-Nielsen et al., 2003]). The observations of a 3- to 4-fold elevated cervical cancer risk
6 with high mean TCE exposure compared to subjects in the low exposure category (6+ ppm: 4.35,
7 95% CI: 1.41, 10.1 [Anttila et al., 1995]; 4+ ppm: 4.3, 95% CI: 0.5, 16 [Hansen et al., 2001]) or
8 with high cumulative TCE exposure (0.25-ppm year: 3.0, 95% CI: 0.8, 11.7 [Blair et al., 1998],
9 2.83, 95% CI: 0.86, 9.33 [Radican et al., 2008]) provides additional support for association with
10 TCE. Cervical cancer risk was lowest for subjects in the high exposure duration category
11 (Hansen et al., 2001; Raaschou-Nielsen et al., 2003); however, duration of employment is a poor
12 exposure metric given subjects may have differing exposure intensity with similar exposure
13 duration (NRC, 2006). No deaths due to cervical cancer were observed in two other high-quality
14 studies (Morgan et al., 1998; Boice et al., 1999), less than 4 deaths were expected, suggesting
15 these cohorts contained few female subjects with TCE exposure.
16 Human papilloma virus and low socioeconomic status are known risk factors for cervical
17 cancer (ACS, 2008). Subjects in Raaschou-Nielsen et al. (2003) are blue-collar workers and low
18 socioeconomic status likely explains observed associations in this and the other high-quality
19 studies. The use of internal controls in Blair et al. (1998) who are similar in socioeconomic
20 status as TCE subjects is believed to partly account for possible confounder related to socio-
21 economic status; however, direct information on individual subjects is lacking.
22 Six other cohort, PMR, and geographic based studies were given less weight in the
23 analysis because of their lesser likelihood of TCE exposure and other study design limitations
24 that would decrease statistical power and study sensitivity (Garabrant et al., 1988; Shannon et al.,
25 1988; Morgan and Cassady, 2002; ATSDR, 2004, 2006; Sung et al., 2007). Cervical cancer risk
26 estimates in these studies ranged between 0.65 (95% CI: 0.38, 1.02) (Morgan and Cassady,
27 2002) to 1.77 (proportional mortality ratio; 95% CI: 0.57, 4.12; ATSDR, 2004). No study
28 reported a statistically significant deficit in cervical cancer risk.
29
30 4.8.2.2. Animal studies
31 Histopathology findings have been noted in reproductive organs in various cancer
32 bioassay studies conducted with TCE. A number of these findings (summarized in Table 4-82)
33 do not demonstrate a treatment-related profile.
34
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-480 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-82. Histopathology findings in reproductive organs
Tumor incidence in mice after 18 months inhalation exposure a
Males
Females
Tissue
Finding
No. examined:
Prostate
Testis
Myoma
Carcinoma
Cyst
No. examined:
Uterus
Ovary
Adenocarcinoma
Adenocarcinoma
Adenoma
Carcinoma
Granulosa cell tumor
Control
30
1
0
0
29
1
1
3
0
4
100 ppm
29
0
0
0
30
0
0
1
2
0
500 ppm
30
0
1
1
28
0
0
3
2
2
Tumor incidence in rats after 18 months inhalation exposure a
Males
Females
Tissue
Finding
No. examined:
Testis
Interstitial cell tumors
No. examined:
Mammary
Uterus
Ovary
Genital tract
Fibroadenoma
Adenocarcinoma
Adenocarcinoma
Carcinoma
Granulosa cell tumor
Squamous cell carcinoma
Control
29
4
28
2
3
3
4
1
0
100 ppm
30
0
30
0
2
1
0
0
2
500 ppm
30
3
30
0
2
4
1
0
0
Tumor incidence in hamsters after 18 months inhalation exposure a
Females
Tissue
Finding
No. examined:
Ovary
Cystadenoma
Control
30
1
100 ppm
29
0
500 ppm
30
0
Tumor incidence in mice after 18 months gavage administration b
Females
Tissue
Finding
No. examined:
Mammary
Ovary
Vulva
Carcinoma
Granulosa cell tumor
Squamous cell carcinoma
Con- TCE TCE
trol Pure Industrial
50 50 50
1 2 0
0 1 0
00 0
TCE+
EPC
50
0
0
0
TCE TCE +EPC
+BO +BO
48 50
0 0
0 0
1 1
2
3
4
"Henschleretal. (1980).
bHenschler et al. (1984); EPC = epichlorohydrin; BO = 1,2-epoxybutane.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-481 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
Cancers of the reproductive system that are associated with TCE exposure and observed
in animal studies are comprised of testicular tumors (interstitial cell and Leydig cell) (U.S. EPA,
2001). A summary of the incidences of testicular tumors observed in male rats is presented in
Table 4-83.
Table 4-83. Testicular tumors in male rats exposed to TCE, adjusted for
reduced survival3
Interstitial cell tumors after 103 wks gavage exposure, beginning at 6.5-8 wks of age (NTP,
1988, 1990)
Administered dose (mg/kg/d)
Male ACT rats
Male August rats
Male Marshall ratsb
Male Osborne-Mendel rats
Male F344/N rats
Untreated
control
38/45 (84%)
36/46 (78%)
16/46 (35%)
1/30 (3%)
44/47 (94%)
Vehicle
control
36/44 (82%)
34/46 (74%)
17/46 (37%)
0/28 (0%)
47/48 (98%)
500
23/26 (88%)
30/34 (88%)
21/33 (64%)
0/25 (0%)
47/48 (98%)
1,000
17/19 (89%)
26/30 (87%)
32/39 (82%)
1/19 (5%)
32/44 (73%)
Leydig cell tumors after 104 wks inhalation exposure, beginning at 12 wks of age (Maltoni
et al., 1986)
Administered daily
concentration (mg/m3)c
Male Sprague-Dawley ratsb
Control
6/114(5%)
112.5
16/105 (15%)
337.5
30/107 (28%)
675
31/113 (27%)
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
aACI rats alive at Week 70, August rats at Week 65, Marshall rats at Week 32, Osborne-Mendel rats at Week 97,
F344/N rats at Week 32, Sprague-Dawley rats at Week 81 (except BT304) or Week 62 (except BT304 bis).
Equivalent to 100, 300, or 600 ppm (100 ppm = 540 mg/m3), adjusted for 7 hours/day, 5 days/week exposure.
Statistically significant by Cochran-Armitage trend test (p < 0.05).
Sources: NTP (1988) Tables A2, C2, E2, G2; NTP (1990) Table A3; Maltoni et al. (1986) IV/IV Table 21, IV/V
Table 21.
4.8.2.3. Mode of Action for Testicular Tumors
The database for TCE does not include an extensive characterization of the mode of
action for Leydig cell tumorigenesis in the rat, although data exist that are suggestive of
hormonal disruption in male rats. A study by Kumar et al. (2000b) found significant decreases in
serum testosterone concentration and in 17-P-HSD, G6PDH, and total cholesterol and ascorbic
acid levels in testicular homogenate from male rats that had been exposed via inhalation to
376 ppm TCE for 12 or 24 weeks. In a follow-up study, Kumar et al. (2001) also identified
decreases in SDH in the testes of TCE-treated rats. These changes are markers of disruption to
testosterone biosynthesis. Evidence of testicular atrophy, observed in the 2001 study by
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-482 DRAFT—DO NOT CITE OR QUOTE
-------
1 Kumar et al., as well as the multiple in vivo and in vitro studies that observed alterations in
2 spermatogenesis and/or sperm function, could also be consistent with alterations in testosterone
3 levels. Therefore, while the available data are suggestive of a MOA involving hormonal
4 disruption for TCE-induced testicular tumors, the evidence is inadequate to specify and test a
5 hypothesized sequence of key events.
6 Leydig cell tumors can be chemically induced by alterations of steroid hormone levels,
7 through mechanisms such as agonism of estrogen, gonadotropin releasing hormone, or dopamine
8 receptors; antagonism of androgen receptors; and inhibition of 5a-reductase, testosterone
9 biosynthesis, or aromatase (Cook et al., 1999). For those plausible mechanisms that involve
10 disruption of the hypothalamic-pituitary-testis (HPT) axis, decreased testosterone or estradiol
11 levels or recognition is involved, and increased luteinizing hormone (LH) levels are commonly
12 observed. Although there is evidence to suggest that humans are quantitatively less sensitive
13 than rats in their proliferative response to LH, evidence of treatment-related Leydig cell tumors
14 in rats that are induced via HPT disruption is considered to represent a potential risk to humans
15 (with the possible exception of GnRh or dopamine agonists), since the pathways for regulation of
16 the HPT axis are similar in rats and humans (Clegg et al., 1997).
17
18 4.8.3. Developmental Toxicity
19 An evaluation of the human and experimental animal data for developmental toxicity,
20 considering the overall weight and strength of the evidence, suggests a potential for adverse
21 outcomes associated with pre- and/or postnatal TCE exposures.
22
23 4.8.3.1. Human Developmental Data
24 Epidemiological developmental studies (summarized in Table 4-84) examined the
25 relationship between TCE exposure and prenatal developmental outcomes including spontaneous
26 abortion and perinatal death; decreased birth weight, small for gestational age, and postnatal
27 growth; congenital malformations; and other adverse birth outcomes. Postnatal developmental
28 outcomes examined include developmental neurotoxicity, developmental immunotoxicity, other
29 developmental outcomes, and childhood cancer.
30
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-483 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-84. Developmental studies in humans
to
VO Lo'
I
I
§
***.
£3'
1
TO'
tl
I
TO
Subjects
Exposure
Effect
Reference
Adverse fetal/birth outcomes
Spontaneous abortion and perinatal death
371 men occupationally exposed to
solvents in Finland 1973-1983
535 women occupationally exposed to
solvents in Finland 1973-1986
3,265 women occupationally exposed
to organic solvents in Finland
1973-1983
361 women occupationally and
residentially exposed to solvents in
Santa Clara County, CA June
1986-February 1987 (735 controls)
4,396 pregnancies among residents of
Woburn, MA 1960-1982
707 parents of children with congenital
heart disease in Tucson Valley, AZ
1969-1987
75 men and 71 women living near
Rocky Mountain Arsenal, CO
1981-1986
1,440 pregnancies among residents of
Endicott, NY 1978-2002
Questionnaire
Low/rare used <1 d/wk;
Intermediate used 1-4 d/wk or
intermediate/low TCA urine
levels;
High/frequent used daily or high
TCA urine levels
Questionnaire
Rare used 1-2 d/wk;
Frequent used >3 d/wk
Questionnaire
U-TCA: median: 48.1 umol/L; mean
96.2 ± 19.2 umol/L
Questionnaire
TCE: 267 ug/L
Tetrachloroethylene: 21 ug/L
Chloroform: 12 ug/L
6-239 ppb TCE, along with DCA and
chromium
Low: <5.0 ppb
Medium: >5.0 to <10.0 ppb
High: <10.0 ppb
indoor air from soil vapor: 0. 18-140
mg/m3
No risk of spontaneous abortion after paternal exposure,
based on 17 cases and 35 controls exposed to TCE (OR: 1.0,
95% Cl: 0.6-2.0)
Increased risk of spontaneous abortion among frequently-
exposed women, based on 7 cases and 9 controls exposed to
TCE (OR: 1.6, 95% Cl: 0.5-4.8)
No increased risk of spontaneous abortion based on 3 cases
and 13 controls exposed to TCE
OR: 0.6, 95% Cl: 0.2-2.3
Increased risk of spontaneous abortion based on 6 cases and
4 controls exposed to TCEa
OR: 3.1, 95% Cl: 0.92-10.4
Increased risk of perinatal death (n = 67) after 1970
(p = 0.55) but not before 1970 (OR: 10, p = 0.003)
No increased risk of spontaneous abortion (n = 520;
p = 0.66)
No increased risk of fetal death (not quantified) based on 246
exposed and 461 unexposed cases
Increased risk of miscarriage
ORadj: 4.44, 95% Cl: 0.76-26.12
Increased risk of no live birth
ORadj: 2.46, 95% Cl: 0.24-24.95
No increase in spontaneous fetal death
SIR: 0.66, 95% Cl: 0.22-1. 55
Taskinen et al.,
1989
Taskinen et al.,
1994
Lindbohm et al.,
1990
Windham et al.,
1991
Lagakos etal.,
1986
Goldberg et al.,
1990
ATSDR,2001
ATSDR, 2006,
2008
H I
O >
HH Oq
H TO
O
H
W
-------
Table 4-84. Developmental studies in humans (continued)
Subjects
81,532 pregnancies among residents of
75 New Jersey towns 1985-1988 (3
control groups)
Exposure
55 ppb TCE, along with many other
compounds
Effect
No increased risk of fetal death for > 10 ppb
OR: 1.12
Reference
Bove, 1996; Bove
etal., 1995
Decreased birth weight, small for gestational age, and postnatal growth
361 women occupationally and
residentially exposed to solvents in
Santa Clara County, CA June
1986-February 1987 (735 controls)
3,462 births in Woburn, MA
1960-1982
1,099 singleton birthsb to residents of 3
census tracts near Tucson International
Airport 1979-1981 (877 controls)
1,440 births0 to residents of Endicott,
NY 1978-2002
6,289 pregnancies among women
residing at Camp Lejeune, NC
1968-1985 (141 short-term and 3 1
long-term TCE-exposed, 5,681
unexposed controls)"1
Questionnaire
267 ug/L TCE in drinking water,
along with tetrachloroethylene and
chloroform
<5-107 ug/L
Indoor air from soil vapor:
0. 18-140 mg/m3
Tarrawa Terrace:
TCE: 8 ppb;
1,2-DCE: 12 ppb
PCE:215ppb
Hadnot Point:
TCE: 1,400 ppb
1,2-DCE: 407 ppb
Increased risk of IUGR based on one case exposed to both
TCE and tetrachloroethylene
OR: 12.5
No increase in low birth weight (p = 0.77)
No increase in full-term low birth weight (OR: 0.81)
No increase in low birth weight (OR: 0.9)
Increase in very low birth weight
OR: 3.3, 95% CI: 0.53-20.6
Small increase in low birth weight
OR: 1.26, 95% CI: 1.00-1.59
Small increase in small for gestational age
OR: 1.22, 95% CI: 1.02-1.45
Increase in full-term low birth weight
OR: 1.41, 95% CI: 1.01-1.95
Change in mean birth weight
Long-term total: -139 g, 90% CI: -277, -1
Long-term males: -312 g, 90% CI: -540, -85
Short term total: +70g, 90% CI: -6, 146
Increase in SGA
Long-term total: OR: 1.5, 90% CI: 0.5, 3.8
Long-term males: OR: 3.9, 90% CI: 1.1-11.9
Short term total: OR: 1.1, 90% CI: 0.2-1.1
Windhametal.,
1991
Lagakos etal.,
1986
Rodenbeck et al.,
2000
ATSDR, 2006,
2008
ATSDR, 1998
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
I
TO
Table 4-84. Developmental studies in humans (continued)
Subjects
81,532 pregnancieseamong residents of
75 New Jersey towns 1985-1988
Exposure
55 ppb TCE, along with many other
compounds
Effect
Decreased birth weight at >5 ppb by 17.9g
No increase in prematurity at >10 ppb: OR: 1.02
Increase in low birth weight, term
>10 ppb: OR: 1.23, 50% Cl: 1.09-1.39
No risk for very low birth weight
Reference
Bove, 1996; Bove
etal., 1995
Congenital malformations
1,148 men and 969 women
occupationally exposed to TCE in
Finland 1963-1976
371 men occupationally exposed to
solvents in Finland 1973-1983
100 babies with oral cleft defects born
to women occupationally exposed in
Europe 1989-1992
4,396 pregnancies among residents of
Woburn, MA 1960-1982
707 children with congenital heart
disease in Tucson Valley, AZ
1969-1987 (246 exposed, 461
unexposed)
U-TCA:
<10 to >500 mg/L
Low/rare used <1 d/wk;
Intermediate used 1-4 d/wk or if
biological measures indicated high
exposure;
High/frequent used daily or if
biological measures indicated high
exposure
Questionnaire
TCE: 267 ug/L
Tetrachloroethylene: 21 ug/L
Chloroform: 12 ug/L
Wells contaminated with TCE (range:
6-239 ppb), along with DCA and
chromium
No congenital malformations reported
No increase in congenital malformations based on 17 cases
and 35 controls exposed to TCE
OR: 0.6, 95% Cl: 0.2-2.0
Increase in cleft lip based on 2 of 4 TCE-exposed women
ORadj: 3.21, 95% Cl: 0.49-20.9
Increase in cleft palate based on 2 of 4 TCE-exposed women
ORadj: 4.47, 95% Cl: 1.02-40.9
Increase in eye/ear birth anomalies: OR: 14.9, p< 0.0001
Increase in CNS/chromosomal/oral cleft anomalies:
OR: 4.5, p = 0.01
Increase in kidney /urinary tract disorders:
OR: 1.35, p = 0.02
Small increase in lung/respiratory tract disorders:
OR: 1.16, p = 0.05
No increase in cardiovascular anomalies (n = 5): p = 0.91
Increase in congenital heart disease
<1981: OR: ~3(p< 0.005)
>1981:OR:~1
Increased prevalence after maternal exposure during first
trimester (p < 0.001, 95% Cl: 1.14-4.14)
Tolaetal., 1980
Taskinen et al.,
1989
Lorente et al.,
2000
Lagakos etal.,
1986
Goldberg et al.,
1990
H I
O >
HH Oq
H TO
O
H
W
-------
to
Table 4-84. Developmental studies in humans (continued)
'
I
I
§
Co
1
TO'
•3
I
1 >;
I
TO
Subjects
Exposure
Effect
Reference
75 men, 71 women living near Rocky
Mountain Arsenal, CO 1981-1986
Low: <5.0 ppb
Medium: >5.0 to <10.0 ppb
High: <10.0 ppb
Increase in total birth defects (n = 9)
OR: 5.87, 95% CI: 0.59-58.81
ATSDR, 2001
Births to residents of Endicott, NY
1983-20001
Indoor air from soil vapor: 0.18-140
mg/m3
No increase in total birth defects:
RR: 1.08, 95% CI: 0.82-1.42
Increase in total cardiac defects:
RR: 1.94, 95% CI: 1.21-3.12
Increase in major cardiac defects:
RR: 2.52, 95% CI: 1.2-5.29
Increase in conotruncal heart defects:
RR: 4.83, 95% CI: 1.81-12.89
ATSDR, 2006,
2008
81,532 pregnancies among residents of
75 New Jersey towns 1985-1988
55 ppb TCE, along with many other
compounds
No increase in total birth defects: >10 ppb: OR: 1.12
Increase in total CNS defects at high dose
>l-5 ppb: OR: 0.93, 90% CI: 0.47-1.77
>10 ppb: OR: 1.68, 90% CI: 0.76-3.52
Increase in neural tube defects
>l-5 ppb: OR: 1.58, 90% CI: 0.69-3.40
>10 ppb: OR: 2.53, 90% CI: 0.91-6.37
Increase in oral clefts:
>5 ppb: OR: 2.24, 95% CI: 1.16-4.20
Increase in major cardiac defects:
>10 ppb: OR: 1.24, 50% CI: 0.75-1.94
Increase in ventrical septal defects
>5ppb: OR: 1.30, 95% CI: 0.88-1.87
Bove, 1996; Bove
etal., 1995
H I
O ^
HH Oq
H TO
O
H
W
1,623 children <20 yrs old dying from
congenital anomalies in Maricopa
County, AZ 1966-1986
8.9 and 29 ppb TCE in drinking water
Increase in deaths due to congenital anomalies in East
Central Phoenix
1966-1969: RR: 1.4, 95% CI: 1.1-1.7
1970-1981: RR: 1.5, 95% CI: 1.3-1.7
1982-1986: RR: 2.0, 95% CI: 1.5-2.5
AZ DHS, 1988
4,025 infants born with congenital
heart defects in Milwaukee, WI
1997-1999
Maternal residence within 1.32 miles
from at least one TCE emissions
source
Increase in congenital heart defects for mothers >38 yrs old
Exposed: OR: 6.2, 95% CI: 2.6-14.5
Unexposed: OR: 1.9, 95% CI: 1.1-3.5
No increase in congenital heart defects for exposed mothers
<38 yrs old: OR: 0.9, 95% CI: 0.6-1.2
Yauck et al., 2004
-------
to
VO Lo'
I
I
§
***.
£3'
1
TO'
•3
'I
• >;
I
TO
Table 4-84. Developmental studies in humans (continued)
Subjects
12 children exposed to TCE in well
water in Michigan
Exposure
5-10 yrs to 8-14 ppm
Effect
1 born with multiple birth defects
Reference
Bernad et al.,
1987, abstract
Other adverse birth outcomes
34 live births for which inhalation of
TCE for anesthesia was used in Japan
1962-1697
5 1 UK women whose fetus was
considered to be at risk for hypoxia
during labor administered TCE as an
analgesic (50 controls)
2-8 mL (mean 4.3 mL) for 2-98 min
(mean: 34.7 min)
Amount and route of exposure not
reported
1 case of asphyxia; 3 "sleepy babies" with Apgar scores of
5-9. Delayed appearance of newborn reflexes
TCE caused fetal pH to fall more, base deficit increased
more, and PO2 fell more than the control group by 4-fold or
more compared to other analgesics used
Beppu, 1968
Phillips and
Macdonald, 1971
Postnatal developmental outcomes
Developmental neurotoxicity
54 individuals from 3 residential
cohorts in the United States exposed to
TCE in drinking water
284 cases of ASD diagnosed <9 yrs old
and 657 controls born in the San
Francisco Bay Area 1994
948 children (<18 yrs) in the
trichloroethylene Subregistry
Woburn, MA
63-400 ppb for <1-12 yrs
Alpha, OH
3.3-330 ppb for 5-17 yrs
Twin Cities, MN
261-2,440 ppb for 0.25-25 yrs
Births geocoded to census tracts, and
linked to HAPs data
0.4 to >5,000 ppb TCE
Woburn, MA
Verbal naming/language impairment in 6/13 children (46%)
Alpha, OH
Verbal naming/language impairment in 1/2 children (50%)
Twin Cities, MN
Verbal naming/language impairment in 4/4 children (100%)
Memory impairment in 4/4 children (100%)
Academic impairment in 4/4 children (100%)
Moderate encephalopathy in 4/4 children (100%)
Poor performance on reading/spelling test in 3/4 children
(75%)
Poor performance on information test in 3/4 children (75%)
Increase in ASD
upper 3rd quartile: OR: 1.37, 95% Cl: 0.96-1.95
upper 4th quartile: OR: 1.47, 95% Cl: 1.03-2.08
Increase in speech impairment:
0-9 yrs old: RR: 2.45, 99% Cl: 1.31-4.58
10-17 yrs old: RR: 1.14, 99% Cl: 0.46-2.85
Increase in hearing impairment:
0-9 yrs old: RR: 2.13, 99% Cl: 1.12-4.07
10-17 yrs old: RR: 1.12, 99% Cl: 0.52-2.24
White etal., 1997
Windham et al.,
2006
ATSDR, 2003a;
Burg etal., 1995;
Burg and Gist,
1999
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
I
TO
Table 4-84. Developmental studies in humans (continued)
Subjects
12 children exposed to TCE in well
water in Michigan
Exposure
5-10 yrs to 8-14 ppm
Effect
9 of 12 children (75%) had poor learning ability, aggressive
behavior, and low attention span
Reference
Bernad et al.,
1987, abstract
Developmental immunotoxicity
200 children aged 36 months old born
prematurely8 and at risk of atopyh in
Lepzig, Germany 1995-1996
85 healthy1 full-term neonates born in
Lepzig, Germany 1997-1999
Median air level in child's bedroom:
0.42 ug/m3
Median air level in child's bedroom
3-4 wks afterbirth: 0.6 ug/m3
No association with allergic sensitization to egg white and
milk, or to cytokine producing peripheral T-cells
Significant reduction of Thl IL-2 producing T-cells
Lehmann et al.,
2001
Lehmann et al.,
2002
Other developmental outcomes
55 children (6 months to 10 yrs old)
were anesthetized for operations to
repair developmental defects of the jaw
and face in Poland 1964
>10 mL TCE
Reports of bradycardia, accelerated heart rate, and
respiratory acceleration observed; no arrhythmia was
observed
Jasinka, 1965,
translation
Childhood cancer
98 children (<10 yrs old) diagnosed
with brain tumors in Los Angeles
County 1972-1977
22 children (<19 yrs old) diagnosed
with neuroblastoma in United States
and Canada 1992-1994 (12 controls)
61 boys and 62 girls (<10 yrs old)
diagnosed with leukemia and 123
controls in Los Angeles County
1980-1984
1,842 children (<15 yrs old) diagnosed
with ALL in United States and Canada
1989-1993 (1986 controls)
Questionnaire of parental
occupational exposures
Questionnaire of parental
occupational exposures
Questionnaire of parents for
occupational exposure
Questionnaire of parents for
occupational exposure
Two cases were reported for TCE exposure, one with methyl
ethyl ketone
Increase in neuroblastoma after paternal exposure
OR: 1.4, 95% CI: 0.7-2.9
Maternal exposure not reported
Increase in leukemia after paternal exposure
Preconception (1 yr): OR: 2.0, p = 0.16
Prenatal: OR: 2.0, p = 0.16
Postnatal: OR: 2.7, /? = 0.7
Maternal exposure not reported
Increase in ALL after maternal exposure
Preconception: OR: 1.8, 95% CI: 0.6-5.2
Pregnancy: OR: 1.8, 95% CI: 0.5-6.4
Postnatal: OR: 1.4, 95% CI: 0.5-4.1
Anytime: OR: 1.8, 95% CI: 0.8-4.1
No increase in ALL after paternal exposure
Anytime: OR: 1.1, 95% CI: 0.8-1.5
Peters etal., 1981
De Roos et al.,
2001
Lowengartetal.,
1987
Shu etal., 1999
H I
O >
HH Oq
H TO
O
H
W
-------
Table 4-84. Developmental studies in humans (continued)
Subjects
109 children (<15 yrs old) born in UK
1974-1988 (218 controls)
22 children (<15 yrs old) diagnosed
with childhood cancer in California
1988-1998
1,190 children (<20 yrs old) diagnosed
with leukemia in 4 counties in New
Jersey 1979-1987
24 children (<15 yrs old) diagnosed
with leukemia in Woburn, MA
1969-1997
347 children (<20 yrs old) diagnosed
with cancer in Endicott, NY
1980-2001
189 children (<20 yrs old) diagnosed
with cancer in Maricopa County, AZ
1965-1990
16 children (<20 yrs old) diagnosed
with cancer in East Phoenix, AZ
1965-1986
Exposure
Questionnaire of parents for
occupational exposure
0.09-97 ppb TCE in drinking water
0-67 ppb TCE in drinking water
267 ug/L TCE in drinking water,
along with tetrachloroethylene,
arsenic, and chloroform
indoor air from soil vapor: 0. 18-140
mg/m3
8.9 and 29 ppb TCE in drinking water
TCE, TCA, and other contaminants in
drinking water
Effect
Increase in leukemia and NHL after paternal exposure
Preconception: OR: 2.27, 95% CI: 0.84-6.16
Prenatal: OR: 4.40, 95% CI: 1.15-21.01
Postnatal: OR: 2.66, 95% CI: 0.82-9.19
No increase in leukemia and NHL after maternal exposure
Preconception: OR: 1.16, 95% CI: 0.13-7.91
No increase in total cancer: SIR: 0.83, 99% CI: 0.44-1.40
No increase in CNS cancer: SIR: 1.05, 99% CI: 0.24-2.70
No increase in leukemia: SIR: 1.09, 99% CI: 0.38-2.31
Increase in ALL in girls with >5 ppb exposure
<20 yrs old: RR: 3.36, 95% CI: 1.29-8.28
<5 yrs old: RR: 4.54, 95% CI: 1.47-10.6
Increase in childhood leukemia
Preconception: ORadj: 2.61, 95% CI: 0.47-14.97
Pregnancy: OR^: 8.33, 95% CI: 0.73-94.67
Postnatal: ORadj: 1.18, 95% CI: 0.28-5.05
Ever: ORadj: 2.39, 95% CI: 0.54-10.59
No increase in cancer (<6 cases, similar to expected)
Increase in leukemia:
1965-1986: SIR: 1.67, 95% CI: 1.20-2.27
1982-1986: SIR: 1.91, 95% CI: 1.11-3.12
No increase in total childhood cancers, lymphoma,
brain/CNS, or other cancers
No increase in leukemia: SIR: 0.85, 95% CI: 0.50-1.35
Reference
McKinney etal.,
1991
Morgan and
Cassady, 2002
Cohnetal., 1994
Costas etal., 2002;
Cutler etal., 1986;
Lagakos etal.,
1986;MADPH,
1997J
ATSDR, 2006,
2008
AZ DHS, 1988,
1990a, 1997k
AZ DHS, 1990b
-------
Table 4-84. Developmental studies in humans (continued)
Subjects
Exposure
Effect
Reference
37 children (<20 yrs old) diagnosed
with cancer in Pima County, AZ
1970-1986
1.1-239 ppb TCE, along with 1,1-
DCE, chloroform and chromium in
drinking water
Increase in leukemia (n= 11):
SIR: 1.50, 95% CI: 0.76-2.70
No increase in testicular cancer (n = 6):
SIR: 0.78, 95% CI: 0.32-1.59
No increase in lymphoma (n = 2):
SIR: 0.63, 95% CI: 0.13-1.80
No increase in CNS/brain cancer (n = 3):
SIR: 0.84, 95% CI: 0.23-2.16
Increase in other cancer (n = 15):
SIR: 1.40, 95% CI: 0.79-2.30
AZ DHS, 1990c
aOf those exposed to TCE, four were also exposed to tetrachloroethylene and one was also exposed to paint strippers and thinners.
bFull term defined as between 35 and 46 weeks gestation, low birth weight as <2501 g, and very low birth weight as <1,501 g.
°Low birth weight defined as <2,500, moderately low birth weight (1,500-<2,500 g), term low birth weight (>37 weeks gestation and <25,000 g).
dUnexposed residents resided at locations not classified for long-term or short-term TCE exposure. Long-term TCE exposed mothers resided at Hospital Point
during 1968-1985 for at least one week prior to birth. Short-term TCE exposed mothers resided at Berkeley Manor, Midway Park, Paradise Point, and Wakins
Village at the time of birth and at least 1 week during January 27 to February 7, 1985. In addition, the mother's last menstrual period occurred on or before
January 31, 1985 and the birth occurred after February 2, 1985.
eLow birth weight defined as <2,500 g, very low birth weight as <1,500 g.
fl,440 births reported for years 1978-2002, but number not reported for years 1983-2000.
gPremature defined as 1,500-2,500 g at birth.
hRisk of atopy defined as cord blood IgE >0.9 kU/L; double positive family atopy history.
'Healthy birth defined as >2,500 g and >37 weeks gestation.
JOnly results from Costas et al. (2002) are reported in the table.
kOnly results from AZ DHS (1990a) are reported in the table.
PCE = perchloroethylene, UK = United Kingdom.
-------
1 4.8.3.1.1. Adverse fetal/birth outcomes.
2 4.8.3.1.1.1. Spontaneous abortion and perinatal death. Spontaneous abortion or miscarriage
3 is defined as nonmedically induced premature delivery of a fetus prior to 20 weeks gestation.
4 Perinatal death is defined as stillbirths and deaths before 7 days after birth. Available data comes
5 from several studies of occupational exposures in Finland and Santa Clara, California, and by
6 geographic-based studies in areas with known contamination of water supplies in Woburn, MA;
7 Tucson Valley, AZ; Rocky Mountain Arsenal, CO; Endicott, NY; and New Jersey.
8
9 4.8.3.1.1.1.1. Occupational studies. The risks of spontaneous abortion and congenial
10 malformations among offspring of men occupationally exposed to TCE and other organic
11 solvents were examined by Taskinen et al. (1989). This nested case-control study was conducted
12 in Finland from 1973-1983. Exposure was determined by biological measurements of the father
13 and questionnaires answered by both the mother and father. The level of exposure was classified
14 as "low/rare" if the chemical was used <1 days/week, "intermediate" if used 1-4 days/week or if
15 TCA urine measurements indicated intermediate/low exposure, and "high/frequent" if used daily
16 or if TCA urine measurements indicated clear occupational exposure (defined as above the RfV
17 for the general population). There was no risk of spontaneous abortion from paternal TCE
18 exposure (OR: 1.0, 95% CI: 0.6-2.0), although there was a significant increase for paternal
19 organic solvent exposure (OR: 2.7, 95% CI: 1.3-5.6) and a nonsignificant increase for maternal
20 organic solvent exposure (OR: 1.4, 95% CI: 0.6-3.0). (Also see section below for results from
21 this study for congenital malformations).
22 Another case-control study in Finland examined pregnancy outcomes in 1973-1986
23 among female laboratory technicians aged 20-34 years (Taskinen et al., 1994). Exposure was
24 reported via questionnaire, and was classified as "rare" if the chemical was used 1-2 days/week,
25 and "frequent" if used at least 3 days/week. Cases of spontaneous abortion (n = 206) were
26 compared with controls who had delivered a baby and did not report prior spontaneous abortions
27 (n = 329). A nonstatistically significant increased risk was seen between spontaneous abortion
28 and TCE use at least 3-days-a-week (OR: 1.6, 95% CI: 0.5-4.8).
29 The association between maternal exposure to organic solvents and spontaneous abortion
30 was examined in Finland for births 1973-1983 (Lindbohm et al., 1990). Exposure was assessed
31 by questionnaire and confirmed with employment records, and the level of exposure was either
32 high, low or none based on the frequency of use and known information about typical levels of
33 exposure for job type. Biological measurements of trichloroacetic acid in urine were also taken
34 on 64 women, with a median value of 48.1 |imol/L (mean: 96.2 ± 19.2 |imol/L). Three cases and
35 13 controls were exposed to TCE, with no increased risk seen for spontaneous abortion (OR: 0.6,
36 95% CI: 0.2-2.3, p. 0.45).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-492 DRAFT—DO NOT CITE OR QUOTE
-------
1 A case-control study in Santa Clara County, California, examined the association
2 between solvents and adverse pregnancy outcomes in women >18 years old (Windham et al.,
3 1991). For pregnancies occurring between June 1986 and February 1987, 361 cases of
4 spontaneous abortion were compared to 735 women who had a live birth during this time period.
5 Telephone interviews included detailed questions on occupational solvent exposure, as well as
6 additional questions on residential solvent use. For TCE exposure, six cases of spontaneous
7 abortion were compared to four controls of live births; of these ten TCE-exposed individuals,
8 four reported exposure to tetrachloroethylene, and one reported exposure to paint strippers and
9 thinners. An increased risk of spontaneous abortions was seen with TCE exposure (OR: 3.1,
10 95% CI: 0.92-10.4), with a statistically significant increased risk for those exposed
11 >0.5 hours/week (OR: 7.7, 95% CI: 1.3-47.4). An increased risk for spontaneous abortion was
12 also seen for those reporting a more "intense" exposure based primarily on odor, as well as skin
13 contact or other symptoms (OR: 3.9,p = 0.04). (Also see section below from this study on low
14 birth weight.)
15
16 4.8.3.1.1.1.2. Geographic-based studies. A community in Woburn, MA with contaminated
17 well water experienced an increased incidence of adverse birth outcomes and childhood
18 leukemia (Lagakos et al., 1986). In 1979, the wells supplying drinking water were found to be
19 contaminated with 267 ppb TCE, 21 ppb tetrachloroethylene, 11.8 ppb, and 12 ppb chloroform,
20 and were subsequently closed. Pregnancy and childhood outcomes were examined from
21 4,396 pregnancies among residents (Lagakos et al., 1986). No association between water access
22 and incidence of spontaneous abortion (n = 520) was observed (p = 0.66). The town's water
23 distribution system was divided into five zones, which was reorganized in 1970. Prior to 1970,
24 no association was observed between water access and incidence of perinatal deaths (n = 46 still
25 births and 21 deaths before 7 days) (p = 0.55). However, after 1970, a statistically significant
26 positive association between access to contaminated water and perinatal deaths was observed
27 (OR: 10.0, p = 0.003). The authors could not explain why this discrepancy was observed, but
28 speculated that contaminants were either not present prior to 1970, or were increased after 1970.
29 (Also see sections below on decreased birth weight, congenital malformations, and childhood
30 cancer for additional results from this cohort.)
31 A community in Tucson Valley, Arizona with contaminated well water had a number of
32 reported cases of congenital heart disease. The wells were found to be contaminated with TCE
33 (range = 6-239 ppb), along with dichloroethylene and chromium (Goldberg et al., 1990). This
34 study identified 707 children born with congenital heart disease during the years 1969-1987. Of
35 the study participants, 246 families had parental residential and occupational exposure during
36 one month prior to conception and during the first trimester of pregnancy, and 461 families had
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-493 DRAFT—DO NOT CITE OR QUOTE
-------
1 no exposure before the end of the first trimester. In addition to this control group, two others
2 were used: (1) those that had contact with the contaminated water area, and (2) those that had
3 contact with the contaminated water area and matched with cases for education, ethnicity, and
4 occupation. Among these cases of congenital heart disease, no significant difference was seen
5 for fetal death (not quantified) for exposed cases compared to unexposed cases. (Also see
6 section below on congenital malformations for additional results from this cohort.)
7 A residential study of individuals living near the Rocky Mountain Arsenal in Colorado
8 examined the outcomes in offspring of 75 men and 71 women exposed to TCE in drinking water
9 (ATSDR, 2001). TCE exposure was stratified by high (>10.0 ppb), medium (>5.0 ppm to
10 <10.0 ppb), and low (<5.0 ppb). Among women with >5 ppb exposure experiencing miscarriage
11 (n = 22/57) compared to unexposed women experiencing miscarriage (n = 2/13) an elevated
12 nonsignificant association was observed (ORadj: 4.44, 95% CI: 0.76-26.12). For lifetime number
13 of miscarriages reported by men and women, results were increased but without dose-response
14 for women (medium: ORadj: 8.56, 95% CI: 0.69-105.99; high: ORadj: 4.16, 95% CI: 0.61-25.99),
15 but less for men (medium: ORadj: 1.68, 95% CI: 0.26-10.77; high: ORadj: 0.65,
16 95% CI: 0.12-3.48). Among women with >5 ppb exposure experiencing no live birth (n = 9/57)
17 compared to unexposed women experiencing no live birth (n = 1/13) an elevated nonsignificant
18 association was observed (ORadj: 2.46, 95% CI: 0.24-24.95). (Also see below for results from
19 this study on birth defects.)
20 NYS DOH and ATSDR conducted a study in Endicott, NY to examine childhood cancer
21 and birth outcomes in an area contaminated by a number of volatile organic compounds (VOCs),
22 including "thousands of gallons" of TCE (ATSDR, 2006). Soil vapor levels tested ranged from
23 0.18-140 mg/m3 in indoor air. A follow-up study by ATSDR (2008) reported that during the
24 years 1978-1993 only five spontaneous fetal deaths occurring >20 weeks gestation were
25 reported when 7.5 were expected (SIR: 0.66, 95% CI: 0.22-1.55). (See sections on low birth
26 weight, congenital malformations, and childhood cancer for additional results from this cohort.)
27 Women were exposed to contaminated drinking water while pregnant and living in 75
28 New Jersey towns during the years 1985-1988 (Bove, 1996; Bove et al., 1995). The water
29 contained multiple trihalomethanes, including an average of 55 ppb TCE, along with
30 tetrachloroethylene, 1,1,1-trichloroethane, carbon tetrachloride, 1,2-dichloroethane, and benzene.
31 A number of birth outcomes were examined for 81,532 pregnancies, which resulted in
32 80,938 live births and 594 fetal deaths. No association was seen for exposure to >10 ppb TCE
33 and fetal death (ORadj: 1.12). (See below for results from this study on decreased birth weight
34 and congenital malformations.)
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-494 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.3.1.1.2. Decreased birth weight, small for gestational age, and postnatal growth.
2 Available data pertaining to birth weight and other growth-related outcomes come from the case-
3 control study in Santa Clara, CA (discussed above), and by geographic-based studies as well as
4 geographic areas with known contamination of water supplies areas in Woburn, MA; Tucson,
5 AZ, Endicott, NY; Camp Lejeune, NC; and New Jersey.
6
7 4.8.3.1.1.2.1. Occupational studies. The case-control study of the relationship between solvents
8 and adverse pregnancy outcomes discussed above (Windham et al., 1991) also examined
9 intrauterine growth restriction (IUGR). Telephone interviews included detailed questions on
10 occupational solvent exposure, as well as additional questions on residential solvent use. An
11 increased risk of IUGR was observed (OR: 12.5), although this was based only on one case that
12 was exposed to both TCE and tetrachloroethylene (also see section above on spontaneous
13 abortion).
14
15 4.8.3.1.1.2.2. Geographic-based studies. The study of Woburn, MA with contaminated well
16 water discussed above (Lagakos et al., 1986) examined birth weight. Of 3,462 live births
17 surviving to 7 days, 220 were less than 6 pounds at birth (6.4%). No association was observed
18 between water access and low birth weight (p = 0.77). (See section on spontaneous abortion for
19 study details, and see sections on spontaneous abortion, congenital malformations, and childhood
20 cancer for additional results from this cohort.)
21 An ecological analysis of well water contaminated with TCE in Tucson and birth-weight
22 was conducted by Rodenbeck et al. (2000). The source of the exposure was a U.S. Air Force
23 plant and the Tucson International Airport. The wells were taken out of service in 1981 after
24 concentrations of TCE were measured in the range of <5 |ig/L to 107 |ig/L. The study
25 population consisted of 1,099 babies born within census tracts between 1979 and 1981, and the
26 comparison population consisted of 877 babies from nearby unexposed census tracts. There was
27 a nonsignificant increased risk for maternal exposure to TCE in drinking water and very-low-
28 birth-weight (<1,501 g) (OR: 3.3, 95% CI: 0.53-20.6). No increases were observed in the low-
29 birth-weight (<2,501 g) (OR: 0.9) or full-term (>35-week and <46-week gestation) low-birth-
30 weight (OR: 0.81).
31 The study of VOC exposure in Endicott, NY reported data on low birth weight and small
32 for gestational age (ATSDR, 2006, see section on spontaneous abortion for study details). For
33 births occurring during the years 1978-2002, low birth weight was slightly but statistically
34 elevated (OR: 1.26, 95% CI: 1.00-1.59), as was small for gestational age (SGA; OR: 1.22,
35 95% CI: 1.02-1.45), and full-term low birth weight (OR: 1.41, 95% CI: 1.01-1.95). (Also see
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-495 DRAFT—DO NOT CITE OR QUOTE
-------
1 sections on spontaneous abortion, congenital malformations, and childhood cancer for additional
2 results from this cohort.)
3 Well water at the U.S. Marine Corps Base in Camp Lejeune, NC was identified to be
4 contaminated with TCE, tetrachloroethylene, and 1,2-dichloroethane in April, 1982 and the wells
5 were closed in December, 1984. AT SDR examined pregnancy outcomes among women living
6 on the base during the years 1968-1985 (ATSDR, 1998). Compared to unexposed residents2
7 (n = 5,681), babies exposed to TCE long-term3 (n = 31) had a lower mean birth weight after
8 adjustment for gestational age (-139 g, 90% CL = -277, -1), and babies exposed short-term4
9 (n = 141) had a slightly higher mean birth weight (+70g, 90% CL = -6, 146). For the long-term
10 group, no effect was seen for very low birth weight (<1,500 grams) or prematurity (>5 ppb,
11 OR: 1.05). No preterm births were reported in the long-term group and those (n = 8) in the
12 short-term group did not have an increased risk (OR: 0.7, 90% CI: 0.3-1.2). A higher
13 prevalence of SGA5 was seen in the long-term exposed group (n = 3; OR 1.5, 90% CL: 0.5, 3.8)
14 compared to the short-term exposed group (OR: 1.1, 90% CI: 0.2-1.1). When the long-term
15 group was stratified by gender, male offspring were at more risk for both reduced birth weight
16 (-312 g, 90% CL = -632, -102) and SGA (OR: 3.9, 90% CL: 1.1-11.8). This study is limited
17 due the mixture of chemicals in the water, as well as it small sample size. ATSDR is currently
18 reanalyzing the findings because of an error in the exposure assessment related to the start-up
19 date of a water treatment plant (ATSDR, 2007, 2009; GAO, 2007a, b).
20 Pregnancy outcomes among women were exposed to contaminated drinking water while
21 pregnant and living in 75 New Jersey towns during the years 1985-1988 was examined by
22 Bove et al. (Bove, 1996; Bove et al., 1995). The water contained multiple trihalomethanes,
23 including an average of 55 ppb TCE, along with tetrachloroethylene, 1,1,1-trichloroethane,
24 carbon tetrachloride, 1,2-dichloroethane, and benzene. A number of birth outcomes were
25 examined for 81,532 pregnancies, which resulted in 80,938 live births and 594 fetal deaths. A
26 slight decrease of 17.9 grams in birth weight was seen for exposure >5 ppb, with a slight increase
27 in risk for exposure >10 ppb (OR: 1.23), but no effect was seen for very low birth weight or
28 SGA/prematurity (>5 ppb, OR: 1.05). However, due to the multiple contaminants in the water, it
29 is difficult to attribute the results solely to TCE exposure. (See below for results from this study
30 on congenital malformations.)
31
Unexposed residents resided at locations not classified for long-term or short-term TCE exposure.
3Long-term TCE exposed mothers resided at Hospital Point during 1968-1985 for at least one week prior to birth.
4Short-term TCE exposed mothers resided at Berkeley Manor, Midway Park, Paradise Point, and Wakins Village at
the time of birth and at least 1 week during January 27 to February 7, 1985. In addition, the mother's last menstrual
period occurred on or before January 31, 1985 and the birth occurred after February 2, 1985.
5The criteria for SGA being singleton births less than the 10th percentile of published sex-specific growth curves.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-496 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.3.1.1.3. Congenital malformations. Three studies focusing on occupational solvent
2 exposure and congenital malformations from Europe provide data pertaining to TCE. Analyses
3 of risk of congenital malformations were also included in the studies in the four geographic areas
4 described above (Woburn, MA; Tucson, AZ, Rocky Mountain Arsenal, CO; Endicott, NY; and
5 New Jersey), as well as additional sites in Phoenix, AZ; and Milwaukee, WI. Specific categories
6 of malformations examined include cardiac defects, as well as cleft lip or cleft palate.
7
8 4.8.3.1.1.3.1. Occupational studies. A study of 1,148 men and 969 women occupationally
9 exposed to TCE in Finland from 1963-1976 to examined congenital malformations of offspring
10 (Tola et al., 1980). Urinary trichloroacetic acid measurements available for 2,004 employees
11 ranged from <10 to >500 mg/L, although 91% of the samples were below 100 mg/L. No
12 congenital malformations were seen in the offspring of women between the ages of 15-49 years,
13 although 3 were expected based on the national incidence. Expected number of cases for the
14 cohort could not be estimated because the number of pregnancies was unknown.
15 Men from Finland occupationally exposed to organic solvents including TCE did not
16 observe a risk of congenital malformations from paternal organic solvent exposure based on
17 17 cases and 35 controls exposed to TCE (OR: 0.6, 95% CI: 0.2-2.0) (Taskinen et al., 1989).
18 (Also see section above on spontaneous abortion for study details and additional results from this
19 cohort.)
20 An occupational study of 100 women who gave birth to babies born with oral cleft
21 defects and 751 control women with normal births were examined for exposure to a number of
22 agents including TCE during the first trimester of pregnancy (Lorente et al., 2000). All women
23 were participants in a multicenter European case-referent study whose children were born
24 between 1989 and 1992. Four women were exposed to TCE, resulting in two cases of cleft lip
25 (ORadj: 3.21, 95% CI: 0.49-20.9), and two cases of cleft palate (ORadj: 4.47,
26 95% CI: 1.02-40.9). Using logistic regression, the increased risk of cleft palate remained high
27 (OR: 6.7, 95% CI: 0.9-49.7), even when controlling for tobacco and alcohol consumption
28 (OR: 7.8, 95% CI: 0.8-71.8). However, the number of cases was small, and exposure levels
29 were not known.
30
31 4.8.3.1.1.3.2. Geographic-based studies. A community in Woburn, MA with contaminated
32 well water experienced an increased incidence of adverse birth outcomes and childhood
33 leukemia (Lagakos et al., 1986, see section on spontaneous abortion for study details).
34 Statistically significant positive association between access to contaminated water and eye/ear
35 birth anomalies (OR: 14.9,p< 0.0001), CNS/chromosomal/oral cleft anomalies (OR: 4.5,
36 p = 0.01), kidney/urinary tract disorders (OR: 1.35,p = 0.02) and lung/respiratory tract disorders
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-497 DRAFT—DO NOT CITE OR QUOTE
-------
1 (OR: 1.16,p = 0.05) were observed. There were also five cases of cardiovascular anomalies, but
2 there was not a significant association with TCE (p = 0.91). However, since organogenesis
3 occurs during gestational weeks 3-5 in humans, some of these effects could have been missed if
4 fetal loss occurred. (Also see sections on spontaneous abortion, perinatal death, decreased birth
5 weight, and childhood cancer for additional results from this cohort.)
6 A high prevalence of congenital heart disease was found within an area of Tucson Valley,
7 AZ (Goldberg et al., 1990, see section on spontaneous abortion for study details and additional
8 results). Of the total 707 case families included, 246 (35%) were exposed to wells providing
9 drinking water found to be contaminated with TCE (range = 6-239 ppb), along with
10 dichloroethylene and chromium. Before the wells were closed after the contamination was
11 discovered in 1981, the OR of congenital heart disease was 3 times higher for those exposed to
12 contaminated drinking water compared to those not exposed; after the wells were closed, there
13 was no difference seen. This study observed 18 exposed cases of congenital heart disease when
14 16.4 would be expected (RR: 1.1). Prevalence of congenital heart disease in offspring after
15 maternal exposure during the first trimester (6.8 in 1,000 live births) was significantly increased
16 compared to nonexposed families (2.64 in 1,000 live births) (p < 0.001, 95% CI: 1.14-4.14). No
17 difference in prevalence was seen if paternal data was included, and there was no difference in
18 prevalence by ethnicity. In addition, no significant difference was seen for cardiac lesions.
19 A residential study of individuals living near the Rocky Mountain Arsenal in Colorado
20 examined the outcomes in offspring of 75 men and 71 women exposed to TCE in drinking water
21 (ATSDR, 2001). The risk was elevated for the nine birth defects observed (OR: 5.87,
22 95% CI: 0.59-58.81), including one nervous system defect, one heart defect, and one incidence
23 of cerebral palsy. The remaining cases were classified as "other," and the authors speculate
24 these may be based on inaccurate reports. (See above for study details and results on
25 spontaneous abortion.)
26 The study of VOC exposure in Endicott, NY examined a number of birth defects during
27 the years 1983-2000 (ATSDR, 2006, see section on spontaneous for study details). These
28 include total reportable birth defects, structural birth defects, surveillance birth defects, total
29 cardiac defects, major cardiac defects, cleft lip/cleft palate, neural tube defects, and choanal
30 atresia (blocked nasal cavities). There were 56 expected cases of all birth defects and 61 were
31 observed resulting in no elevation of risk (rate ratio, RR: 1.08, 95% CI: 0.82-1.42). There were
32 no cases of cleft lip/cleft palate, neural tube defects, or choanal atresia. Both total cardiac
33 defects (n = 15; RR: 1.94, 95% CI: 1.21-3.12) and major cardiac defects (n = 6; RR: 2.52,
34 95% CI: 1.2-5.29) were statistically increased. A follow-up study by ATSDR (2008) reported
35 that conotruncal heart malformations were particularly elevated (n = 4; RR: 4.83, 95% CI:
36 1.81-12.89). The results remained significantly elevated (aRR: 3.74; 95% CI: 1.21-11.62)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-498 DRAFT—DO NOT CITE OR QUOTE
-------
1 when infants with Down syndrome were excluded from the analysis. (Also see sections on
2 spontaneous abortion, decreased birth weight, and childhood cancer for additional results from
3 this cohort.)
4 In the New Jersey study described previously, the prevalence of birth defects reported by
5 surveillance systems was examined among the women exposed to TCE and other contaminants
6 in water while pregnant between 1985-1988 (Bove, 1996; Bove et al., 1995). For exposure
7 >10 ppb (n = 1,372), an increased risk, with relatively wide confidence intervals, was seen for all
8 birth defects (OR: 2.53, 95% CI: 0.77-7.34). An increased risk was also seen for CNS defects
9 (>10 ppb: OR: 1.68), specifically 56 cases of neural tube defects (10 ppb: OR: 2.53, 95% CI: 0.77-7.34). A slight increase was seen in
11 major cardiac defects (>10 ppb: OR: 1.24, 50% CI: 0.75-1.94), including ventrical septal defects
12 (>5 ppb: OR: 1.30, 95% CI: 0.88-1.87). An elevated risk was seen for 9 cases of oral clefts
13 (<5 ppb: OR: 2.24, 95% CI: 1.04-4.66), although no dose-response was seen (>10 ppb,
14 OR: 1.30). However, due to the multiple contaminants in the water, it is difficult to attribute the
15 results solely to TCE exposure. (See above for results from this study on fetal death and
16 decreased birth weight.)
17 Arizona Department of Heath Services (AZ DHS) conducted studies of contaminated
18 drinking water and congenital malformations (<20 years old) in Maricopa County, which
19 encompasses Phoenix and the surrounding area (AZ DHS, 1988). TCE contamination was
20 associated with elevated levels of deaths in children less than 20 years old due to total congenital
21 anomalies in East Central Phoenix from 1966-1969 (RR: 1.4, 95% CI: 1.1-1.7), from
22 1970-1981 (RR: 1.5, 95% CI: 1.3-1.7), and from 1982-1986 (RR: 2.0, 95% CI: 1.5-2.5), as
23 well as in other areas of the county. (See below for results from this study on childhood
24 leukemia.)
25 A study was conducted of children born 1997-1999 with congenital heart defects in
26 Milwaukee, WI (Yauck et al., 2004). TCE emissions data were ascertained from state and U.S.
27 EPA databases, and distance between maternal residence and the emission source was
28 determined using a GIS. Exposure was defined as those within 1.32 miles from at least one site.
29 Results showed that an increased risk of congenital heart defects was seen for the offspring of
30 exposed mothers 38 years old or older (OR: 6.2, 95% CI: 2.6-14.5), although an increased risk
31 was also seen for offspring of unexposed mothers 38 years old or older (OR: 1.9,
32 95% CI: 1.1-3.5), and no risk was seen for offspring of exposed mothers younger than 38 years
33 (OR: 0.9, 95% CI: 0.6-1.2). The authors speculate that studies that did not find a risk only
34 examined younger mothers. The authors also note that statistically-significant increased risk was
35 seen for mothers with preexisting diabetes, chronic hypertension, or alcohol use during
36 pregnancy.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-499 DRAFT—DO NOT CITE OR QUOTE
-------
1 An abstract reported that twenty-eight people living in a Michigan town were exposed for
2 5-10 years to 8-14 ppm TCE in well water (Bernad et al., 1987, abstract). One child was born
3 with multiple birth defects, with no further details.
4 4.8.3.1.1.4. Other adverse birth outcomes. TCE was previously used as a general anesthetic
5 during pregnancy. One study measured the levels of TCE in maternal and newborn blood after
6 use during 34 vaginal childbirths (Beppu, 1968). TCE was administered through a vaporizer
7 from two to 98 minutes (mean 34.7 minutes) at volumes from 2 to 8 mL (mean 4.3 mL). Mean
8 blood TCE concentrations were 2.80 ± 1.14 mg/dL in maternal femoral arteries; 2.36 ±1.17
9 mg/dL in maternal cubital veins; 1.83 ± 1.08 mg/dL in umbilical vein; and 1.91 ± 0.95 mg/dL in
10 the umbilical arteries. A significant correlation was seen for maternal arterial blood and infants'
11 venous blood, and the concentration of the fetal blood was lower than that of the mother. Of
12 these newborns, one had asphyxia and three "sleepy babies" had Apgar scores of 5 to 9;
13 however, these results could not be correlated to length of inhalation and there was no difference
14 in the TCE levels in the mother or newborn blood compared to those without adverse effects.
15 Discussion included delayed newborn reflexes (raising the head and buttocks, bending the spine,
16 and sound reflex), blood pressure, jaundice, and body weight gain; however, the results were
17 compared to newborns exposed to other compounds, not to an unexposed population. This study
18 also examined the concentration of TCE in one mother at 22-weeks gestation exposed for four
19 minutes, after which the fetus was "artificially delivered." Maternal blood concentration was
20 3.0 mg/dL, and 0.9 mg/dL of TCE was found in the fetal heart, but not in other organs.
21 Another study of TCE administered during childbirth to the mother as an analgesic
22 examined perinatal measures, including fetal pH, fetal partial pressure carbon dioxide (PCO2J
23 fetal base deficit, fetal partial pressure oxygen (PO2), Apgar scores, and neonatal capillary blood
24 (Phillips and Macdonald, 1971). The study consisted of 152 women whose fetus was considered
25 to be at risk for hypoxia during labor. Out of this group, 51 received TCE (amount and route of
26 exposure not reported). TCE caused fetal pH to fall more, base deficit increased more, and PO2
27 fell more than the control group by 4-fold or more compared to other analgesics used.
28
29 4.8.3.1.2. Postnatal developmental outcomes.
30 4.8.3.1.2.1. Developmental neurotoxicity. The studies examining neurotoxic effects from TCE
31 exposure are discussed in Section 4.3, and the human developmental neurotoxic effects are
32 reiterated here.
33
34 4.8.3.1.2.1.1. Occupational studies. An occupational study examined the neurodevelopment of
35 the offspring of 32 women exposed to various organic solvents during pregnancy (Laslo-Baker et
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-500 DRAFT—DO NOT CITE OR QUOTE
-------
1 al., 2004; Till et al., 2001). Three of these women were exposed to TCE; however, no levels
2 were measured and the results for examined outcomes are for total organic solvent exposure, and
3 are not specific to TCE.
4
5 4.8.3.1.2.1.2. Geographic-based studies. A study of three residential cohorts (Woburn, MA,
6 Alpha, OH, and Twin Cities, MN) examined the neurological effects of TCE exposure in
7 drinking water (White et al., 1997). For Woburn, MA, 28 individuals ranging from 9-55 years
8 old were assessed, with exposure from a tanning factor and chemical plant at levels 63-400 ppb
9 for <1 to 12 years; the time between exposure and neurological examination was about 5 years.
10 In this cohort, six of thirteen children (46%) had impairments in the verbal naming/language
11 domain. For Alpha, OH, 12 individuals ranging from 12-68 years old were assessed, with
12 exposure from degreasing used at a manufacturing operation at levels 3.3-330 ppb for
13 5-17 years; the time between exposure and neurological examination was 5-17 years. In this
14 cohort, one of two children (50%) had impairments in the verbal naming/language domain. For
15 Twin Cities, MN, 14 individuals ranging from 8-62 years old were assessed, with exposure from
16 an army ammunition plant at levels 261-2,440 ppb for 0.25-25 years; the time between
17 exposure and neurological examination was 4-22 years. In this cohort, four of four children
18 (100%) had impairments in the verbal naming/language, memory, and academic domains and
19 were diagnosed with moderate encephalopathy; and three of four children (75%) performed
20 poorly on the WRAT-R Reading and Spelling and WAIS-R Information tests.
21 A case-control study was conducted to examine the relationship between multiple
22 environmental agents and autism spectrum disorder (ASD) (Windham et al., 2006). Cases
23 (n = 284) and controls (n = 657) were born in 1994 in the San Francisco Bay Area. Cases were
24 diagnosed before age nine. Exposure was determined by geocoding births to census tracts, and
25 linking to hazardous air pollutants (HAPs) data. An elevated risk was seen for TCE in the upper
26 3rd quartile (OR: 1.37, 95% CI: 0.96-1.95), and a statistically significant elevated risk was seen
27 for the upper 4th quartile (OR: 1.47, 95% CI: 1.03-2.08).
28 The Trichloroethylene Subregistry (Burg et al., 1995; Burg and Gist, 1999), including
29 948 children <18 years old from 13 sites located in 3 states, was examined for any association of
30 ingestion of drinking water contaminated with TCE and various health effects (Burg et al., 1995;
31 Burg and Gist, 1999; ATSDR, 2003a). Exposure groups included (1) maximum TCE exposure,
32 (2) cumulative TCE exposure, (3) cumulative chemical exposure, and (4) duration of exposure.
33 Exposed children 0-9 years old had statistically increased hearing impairment compared to
34 controls (RR: 2.13, 99% CI: 1.12-4.07), with children <5 having a 5.2-fold increase over
35 controls. Exposed children 0-9 years old also had statistically increased speech impairment
36 (RR: 2.45, 99% CI: 1.31-4.58). In addition, anemia and other blood disorders were statistically
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-501 DRAFT—DO NOT CITE OR QUOTE
-------
1 higher for males 0-9 years old. The authors noted that exposure could have occurred prenatally
2 or postnatally. There was further analysis on the 116 exposed children and 182 controls who
3 were under 10 years old at the time that the baseline study was conducted by ATSDR. This
4 analysis did not find a continued association with speech and hearing impairment in these
5 children; however, the absence of acoustic reflexes (contraction of the middle ear muscles in
6 response to sound) remained significant (ATSDR, 2003a). No differences were seen when
7 stratified by prenatal and postnatal exposure.
8 Twenty-eight people living in a Michigan town were exposed for 5-10 years to
9 8-14 ppm TCE in well water (Bernad et al., 1987). Ten adults and 12 children completed a
10 questionnaire on neurotoxic endpoints. Nine of the 12 children had poor learning ability,
11 aggressive behavior, and low attention span.
12
13 4.8.3.1.2.2. Developmental immunotoxicity. The studies examining human immunotoxic
14 effects from TCE exposure are discussed in Section 4.6.1. The studies reporting developmental
15 effects are reiterated briefly here.
16 Two studies focused on immunological development in children after maternal exposure
17 to VOCs (Lehmann et al., 2001, 2002). The first examined premature neonates (1,500-2,500 g)
18 and neonates at risk of atopy (cord blood IgE >0.9 kU/L; double positive family atopy history) at
19 36 months of age (Lehmann et al., 2001). Median air level in child's bedroom measured
20 0.42 |ig/m3. There was no association with allergic sensitization to egg white and milk, or to
21 cytokine producing peripheral T-cells. The second examined healthy, full-term neonates
22 (>2,500 g; >37 weeks gestation) born in Lepzig, Germany (Lehmann et al., 2002). Median air
23 level in the child's bedroom 3-4 weeks after birth measured 0.6 |ig/m3. A significant reduction
24 of Thl IL-2 producing T-cells was observed.
25 Byers et al. (1988) observed altered immune response in family members of children
26 diagnosed with leukemia in Woburn, MA (Lagakos et al., 1986, see below for results of this
27 study). The family members included 13 siblings under 19 years old at the time of exposure;
28 however, an analysis looking at only these children was not done. This study is discussed in
29 further detail in Section 4.6.1.
30
31 4.8.3.1.2.3. Other developmental outcomes. A study demonstrated the adverse effects of TCE
32 used as an anesthetic in children during operations during 1964 in Poland to repair
33 developmental defects of the jaw and face (Jasinka, 1965, translation). Fifty-five children
34 ranging from 6 months to 10 years old were anesthetized with at least 10 mL TCE placed into an
35 evaporator. Bradycardia occurred in 2 children, an accelerated heart rate of 20-25 beats per
36 minute occurred in 7 children, no arrhythmia was observed, and arterial blood pressure remained
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-502 DRAFT—DO NOT CITE OR QUOTE
-------
1 steady or dropped by 10 mmHG only. Respiratory acceleration was observed in 25 of the
2 children, and was seen more in infants and younger children.
3
4 4.8.3.1.2.4. Childhood cancer. Several studies of parental occupational exposure were
5 conducted in North America and the United Kingdom to determine an association with
6 childhood cancer. A number of geographic-based studies were conducted in California; New
7 Jersey; Woburn, MA; Endicott, NY; Phoenix, AZ; and Tucson, AZ. Specific categories of
8 childhood cancers examined include leukemia, non-Hodgkin's lymphoma, and CNS tumors.
9
10 4.8.3.1.2.4.1. Occupational studies. Brain tumors in 98 children less than 10 years old at
11 diagnosis from 1972-1977 in Los Angeles County have been observed in the offspring of fathers
12 (Peters etal., 1981, 1985). Exposure was determined by questionnaire. Two cases with TCE
13 exposure were reported: one case of oligodendroglioma in an 8-year-old whose father was a
14 machinist, and astrocytoma in a 7-year-old whose father was an inspector for production
15 scheduling and parts also exposed to methyl ethyl ketone (Peters et al., 1981). Peters et al.
16 (1985) also briefly mentioned 5 cases and no controls of paternal exposure to TCE and brain
17 tumors in the offspring (resulting in an inability to calculate an odds ratio), but without providing
18 any additional data.
19 A case-control study was conducted to assess an association between parental
20 occupational exposure and neuroblastoma diagnosed in offspring <19 years old in the United
21 States and Canada from May 1992 to April 1994 (De Roos et al., 2001). Paternal self-reported
22 exposure to TCE was reported in 22 cases and 12 controls, resulting in an elevated risk of
23 neuroblastoma in the offspring (OR: 1.4, 95%CI: 0.7-2.9). Maternal exposure to TCE was not
24 reported.
25 A case-control study of parental occupational exposure and childhood leukemia was
26 conducted in Los Angeles County (Lowengart et al., 1987). Children (61 boys and 62 girls)
27 diagnosed less than 10 years old (mean age 4 years) from 1980 to 1984 were included in the
28 analysis. Paternal occupation exposure to TCE was elevated for one year preconception
29 (OR: 2.0, p = 0.16), prenatal (OR: 2.0, p = 0.16), and postnatal (OR: 2.7, p = 0.7). Maternal
30 exposure to TCE was not reported.
31 A case-control study children diagnosed with acute lymphoblastic leukemia (ALL)
32 examined parental occupational exposure to hydrocarbons in the United States and Canada
33 (Shu et al., 1999). Children were under the age of 15 years at diagnosis during the years 1989 to
34 1993. Cases were confirmed with a bone marrow sample. 1,842 case-control pairs were given
35 questionnaires on maternal and paternal exposures, resulting in 15 cases and 9 controls
36 maternally exposed and 136 cases and 104 controls paternally exposed to TCE. There was an
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-503 DRAFT—DO NOT CITE OR QUOTE
-------
1 increased but nonsignificant risk for maternal exposure to TCE during preconception (OR: 1.8,
2 95% CI: 0.6-5.2), pregnancy (OR: 1.8, 95% CI: 0.5-6.4), postnatally (OR: 1.4,
3 95% CI: 0.5-4.1), or any of these periods (OR: 1.8, 95% CI: 0.8-4.1). However, there was no
4 increased risk for paternal exposure to TCE.
5 Occupational exposure in communities in the United Kingdom was examined to
6 determine an association with leukemia and non-Hodgkin's lymphoma diagnosed in the
7 offspring (McKinney et al., 1991). Paternal occupational exposure was elevated for exposure
8 occurring during preconception (OR: 2.27, 95% CI: 0.84-6.16), prenatal (OR: 4.40,
9 95% CI: 1.15-21.01), and postnatal (OR: 2.66, 95% CI: 0.82-9.19). Risk from maternal
10 preconception exposure was not elevated (OR: 1.16, 95% CI: 0.13-7.91). However, the number
11 of cases examined in this study was low, particularly for maternal exposure.
12
13 4.8.3.1.2.4.2. Geographic-based studies. A California community exposed to TCE
14 (0.09-97 ppb) in drinking water from contaminated wells was examined for cancer (Morgan and
15 Cassady, 2002). A specific emphasis was placed on the examination of 22 cases of childhood
16 cancer diagnosed before 15 years old. However, the incidence did not exceed those expected for
17 the community for total cancer (SIR: 0.83, 99% CI: 0.44-1.40), CNS cancer (SIR: 1.05,
18 99% CI: 0.24-2.70), and leukemia (SIR: 1.09, 99% CI: 0.38-2.31).
19 An examination of drinking water was conducted in four New Jersey counties to
20 determine an association with leukemia and non-Hodgkin's lymphoma (Cohn et al., 1994). A
21 number of contaminants were reported, including VOCs and trihalomethanes. TCE was found as
22 high as 67 ppb, and exposure categories were assigned to be >0.1, 0.1-5 and >5 ppb. A
23 significantly elevated dose-response risk for ALL was observed for girls diagnosed before
24 20 years old (RR: 3.36, 95% CI: 1.29-8.28), which was increased among girls diagnosed before
25 5 years old (RR:4.54, 95% CI: 1.47-10.6). A significantly elevated dose-response risk for girls
26 was also observed for total leukemia (RR: 1.43, 95% CI: 1.07-1.98).
27 The Woburn, MA community with contaminated well water experienced an increase in
28 the incidence of childhood leukemia (Costas et al., 2002; Cutler et al., 1986; Lagakos et al.,
29 1986; MA DPH, 1997). An initial study examined twelve cases of childhood leukemia
30 diagnosed in children less than 15 years old between 1969-1979, when 5.2 cases were expected,
31 and a higher risk was observed in boys compared to girls; however, no factors were observed to
32 account for this increase (Cutler et al., 1986). Another study observed statistically significant
33 positive association between access to contaminated water and 20 cases of childhood cancer
34 were observed for both cumulative exposure metric (OR: 1.39, p = 0.03), and none versus some
35 exposure metric (OR: 3.03,p = 0.02) (Lagakos et al., 1986). Massachusetts Department of
36 Public Health (MA DPH, 1997) conducted a case-control study of children less than 20 years old
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-504 DRAFT—DO NOT CITE OR QUOTE
-------
1 living in Woburn and diagnosed with leukemia between 1969 and 1989 (n = 21) and observed
2 that consumption of drinking water increased the risk of leukemia (OR: 3.03, 95%
3 CI: 0.82-11.28), with the highest risk from exposure during fetal development (OR: 8.33,
4 95% CI: 0.73-94.67). This study found that paternal occupational exposure to TCE was not
5 related to leukemia in the offspring (MA DPH, 1997). In the most recent update, Costas et al.
6 (2002) reported that between the years 1969 and 1997, 24 cases of childhood leukemia were
7 observed when 11 were expected. Risk was calculated for cumulative exposure to contaminated
8 drinking water two years prior to conception (ORadj: 2.61, 95% CI: 0.47-14.97), during
9 pregnancy (ORadj: 8.33, 95% CI: 0.73-94.67), postnatal (ORadj: 1.18, 95% CI: 0.28-5.05), and
10 any of these time periods (ORadj: 2.39, 95% CI: 0.54-10.59). A dose response was observed
11 during pregnancy only. Cases were more likely to be male (76%), <9 years old at diagnosis
12 (62%), breast-fed (OR: 10.17, 95% CI: 1.22-84.50), and exposed during pregnancy (adjusted
13 OR: 8.33, 95% CI: 0.73-94.67). A dose-response was seen during the pregnancy exposure
14 period, with the most exposed having an adjusted OR of 14.30 (95% CI: 0.92-224.52). Other
15 elevated risks observed included maternal alcohol intake during pregnancy (OR: 1.50,
16 95% CI: 0.54-4.20), having a paternal grandfather diagnosed with cancer (OR: 2.01,
17 95% CI: 0.73-5.58), father employed in a high risk industry (OR: 2.55, 95% CI: 0.78-8.30), and
18 public water being the subject's primary beverage (OR: 3.03, 95% CI: 0.82-11.28). (Also see
19 sections on spontaneous abortion, perinatal death, decreased birth weight, and congenital
20 malformations for additional results from this cohort.)
21 The study of VOC exposure in Endicott, NY discussed above observed fewer than six
22 cases of cancer that were diagnosed between 1980 and 2001 in children less than 20 years old,
23 and did not exceed expected cases or types (ATSDR, 2006). (See section on spontaneous
24 abortion for study details, and sections on spontaneous abortion, decreased birth weight, and
25 congenital malformations for additional results from this cohort.)
26 The AZ DHS conducted a number of studies of contaminated drinking water and 189
27 cases of childhood cancer (<20 years old) (AZ DHS, 1988, 1990a, b, c, 1997). In Maricopa
28 County, which encompasses Phoenix and the surrounding area, TCE contamination (8.9 and
29 29 ppb in two wells) was associated with elevated levels of childhood leukemia (n = 67) in west
30 central Phoenix during 1965-1986 (SIR: 1.67, 95% CI: 1.20-2.27) and 1982-1986 (SIR: 1.91,
31 95% CI: 1.11-3.12), but did not observe a significant increase in total childhood cancers,
32 lymphoma, brain/CNS, or other cancers during these time periods (AZ DHS, 1990a). (See above
33 for results from this study on congenital anomalies.) A follow-up study retrospectively asked
34 parents about exposures and found that residence within 2 miles of wells contaminated with TCE
35 was not a risk factor for childhood leukemia, but identified a number of other risk factors
36 (AZ DHS, 1997). A further study of East Phoenix, reported on TCE contamination found along
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-505 DRAFT—DO NOT CITE OR QUOTE
-------
1 with 1,1,1-trichloroethane and 25 other contaminants in well water (levels not reported) and
2 found no increase in incidence of childhood leukemia (SIR: 0.85, 95% CI: 0.50-1.35) based on
3 16 cases (AZ DHS, 1990b). There were also 16 cases of other types of childhood cancer, but
4 were too few to be analyzed separately. In Pima County, which encompasses Tucson and the
5 surrounding area, TCE was found in drinking wells (1.1-239 ppb), along with
6 1,1-dichloroethylene (DCE), chloroform and chromium and found a nonstatistically elevated risk
7 of leukemia was observed (SIR: 1.50, 95% CI: 0.76-2.70), but no risk was observed for
8 testicular cancer, lymphoma, or CNS/brain cancer (AZ DHS, 1990c).
9
10 4.8.3.1.3. Summary of human developmental toxicity. Epidemiological developmental
11 studies examined the association between TCE exposure and a number of prenatal and postnatal
12 developmental outcomes. Prenatal developmental outcomes examined include spontaneous
13 abortion and perinatal death; decreased birth weight, small for gestational age, and postnatal
14 growth; congenital malformations; and other adverse birth outcomes. Postnatal developmental
15 outcomes examined include developmental neurotoxicity, developmental immunotoxicity, other
16 developmental outcomes, and childhood cancer related to TCE exposure.
17 More information on developmental outcomes is expected. A follow-up study of the
18 Camp Lejeune cohort (ATSDR, 1998) for birth defects and childhood cancers was initiated in
19 1999 (ATSDR, 2003b) and expected to be completed soon (GAO, 2007a, b; ATSDR, 2009).
20 Out of a total of 106 potential cases of either birth defects or childhood cancer, 57 have been
21 confirmed and will constitute the cases. These will be compared 548 control offspring of
22 mothers who also lived at Camp Lejeune during their pregnancy from 1968-1985. As part of
23 this study, a drinking water model was developed to determine a more accurate level and
24 duration of exposure to these pregnant women (ATSDR, 2007). Additional health studies have
25 been suggested, including adverse neurological or behavioral effects or pregnancy loss.
26
27 4.8.3.2. Animal Developmental Toxicology Studies
28 A number of animal studies have been conducted to assess the potential for
29 developmental toxicity of TCE. These include studies conducted in rodents by prenatal
30 inhalation or oral exposures (summarized in Tables 4-85 and 4-86), as well as assessments in
31 nonmammalian species (e.g., avian, amphibian, and invertebrate species) exposed to TCE during
32 development. Studies have been conducted that provide information on the potential for effects
33 on specific organ systems, including the developing nervous, immune, and pulmonary systems.
34 Additionally, a number of research efforts have focused on further characterization of the mode
35 of action for cardiac malformations that have been reported to be associated with TCE exposure.
36
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-506 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-85. Summary of mammalian in vivo developmental toxicity
studies—inhalation exposures
Reference
Carney et
al, 2006
Dorfmueller
etal., 1979
Hardin et
al., 1981
Healy et al.,
1982
Schwetz et
al., 1975
Westergren
etal., 1984
Species/strain/
sex/number
Rat, Sprague-
Dawley,
females, 27
dams/group
Rat, Long-
Evans,
females, 30
dams/group
Rat, Sprague-
Dawley,
female,
nominal
30/group
Rabbit, New
Zealand white,
female,
nominal
20/group
Rat, Wistar,
females, 31-32
dams/group
Rat, Sprague-
Dawley,
female,
20-35/group
Mouse, Swiss-
Webster,
females, 30-40
dams/group
Mouse, NMRI,
male and
female, 6-12
offspring/group
Exposure level/
duration
0, 50, 150, or
600 ppm
(600 ppm =
3.2 mg/L)b
6h/d;
GD 6-20
0 or 1,800 +200
ppm
(9,674 + 1,075
mg/m3)b
2 wks, 6 h/d,
5 d/wk; prior to
mating and/or
on GD 0-20
0 or 500 ppm
6-7 h/d;
GD 1-19
0 or 500 ppm
6-7 h/d;
GD 1-24
0 or 100 ppm
4 h/d;
GD 8-21
0 or 300 ppm
7 h/d;
GD 6-15
0 or 150 ppm
24 h/d;
30 d (during 7 d
of mating and
until GD 22)
NOAEL; LOAEL a
Mat. NOAEL: 150 ppm
Mat. LOAEL: 600 ppm
Dev. NOAEL: 600 ppm
Mat. NOAEL: 1,800 +
200 ppm
Dev. LOAEL: 1,800 +
200 ppm
Mat. NOAEL: 500 ppm
Dev. NOAEL: 500 ppm
Mat. NOAEL: 500 ppm
Dev. LOAEL: 500 ppm
Mat. NOAEL: 100 ppm
Dev. LOAEL: 100 ppm
Mat. LOAEL: 300 ppm
Dev. NOAEL: 300 ppm
Dev. LOAEL:
150 ppm°
Effects
^ BW gain (22% less than control) on
GD 6-9 at 600 ppm.
No evidence of developmental toxicity,
including heart defects.
No maternal abnormalities.
Sig. t skeletal and soft tissue anomalies
in fetuses from dams exposed during
pregnancy only. No sig. treatment
effects on behavior of offspring 10, 20,
or 100 d postpartum. BW gains sig. -i-
in pups from dams with pregestational
exposure.
No maternal toxicity
No embryonic or fetal toxicity.
No maternal toxicity.
Hydrocephaly observed in 2 fetuses of
2 litters, considered equivocal evidence
of teratogenic potential.
No maternal abnormalities.
Litters with total resorptions sig. t.
Sig. -i- fetal weight, and t bipartite or
absent skeletal ossification centers.
4-5% ^ maternal BW
No embryonic or fetal toxicity; not
teratogenic.
Specific gravity of brains sig. -i- atPND
0, 10, and 20-22. Similar effects at
PND 20-22 in occipital cortex and
cerebellum. No effects at 1 month of
age.
2 "NOAEL and LOAEL are based upon reported study findings. Mat. = maternal; Dev. = developmental.
3 bDose conversions provided by study author(s).
4 'Parental observations not reported.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-507 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-86. Ocular defects observed (Narotsky et al., 1995)
2
3
4
5
6
7
8
9
10
11
Dose TCE (mg/kg/d)
0
10.1
32
101
320
475
633
844
1,125
Incidence
(no. affected
pups/total no. pups)*
1/197
0/71
0/85
3/68
3/82
6/100
6/100
7/58
12/44
Percent pups
with eye
defects
0.51
0.00
0.00
4.41
3.66
6.00
6.00
12.07
27.27
*Reported in Barton and Das (1996).
4.8.3.2.1. Mammalian studies
Studies that have examined the effects of TCE on mammalian development following
either inhalation or oral exposures are described below and summarized in Tables 4-85 and 4-87,
respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-508 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-87. Summary of mammalian in vivo developmental toxicity
studies—oral exposures
Reference
Blossom
and Doss,
2007
Blossom et
al., 2008
Species/strain/
sex/number
Mouse, MRL +/+,
dams and both sexes
offspring, 3
litters/group, 8-12
offspring/group
Mouse, MRL +/+,
dams and both sexes
offspring, 8
litters/group, 3-8
offspring/group
Dose level/
exposure
duration
0,0.5, or 2.5
mg/mL
Parental mice
and/or offspring
exposed from GD
0 to 7-8 months
of age
0 or 0.1 mg/mL
(maternal dose =
25.7 mg/kg/d;
offspring PND
24-42 dose— 31.0
mg/kg/d
Parental mice
and/or offspring
exposed from GD
0 to PND 42
Route/
vehicle
Drinking
water
Drinking
water
NOAEL;
LOAEL3
Dev. LOAEL
= 0.5 mg/mLb
Dev. LOAEL
= 1,400 ppbb
Effects
At 0.5 mg/mL: Sig |
postweaning weight; sig.f
IFNy produced by splenic
CD4+ cells at 5-6 wks; sig |
splenic CD8+and B220+
lymphocytes; sig.f IgG2aand
histone; sig. altered CD4-
/CD8- and CD4+/CD8+
thymocyte profile.
At 2. 5 mg/mL: Sig |
postweaning weight; sig.f
IFNy produced by splenic
CD4+ and CD8+ cells at 4-5
and 5-6 wks; sig J, splenic
CD4+, CD8+, and B220+
lymphocytes; sig. altered
CD4+/CD8+ thymocyte
profile.
At 0.1 mg/mL: at PND 20,
sig. t thymocyte cellularity
and distribution, associated
with sig. t in thymocyte
subset distribution; sig. t
reactive oxygen species
generation in total
thymocytes; sig. t in splenic
CD4+ T-cell production of
IFN-y and IL-2 in females
and TNF-a in males at PND
42.
Significantly impaired nest-
building behaviors at PND
35. Increased aggressive
activities, and increased
oxidative stress and impaired
thiol status in the cerebellar
tissue of male offspring at
PND 40.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-509 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-87. Summary of mammalian in vivo developmental toxicity
studies—oral exposures (continued)
Reference
Collier et
al, 2003
Cosby and
Dukelow,
1992
Dawson, et
al., 1993
Fisher et
al., 2001;
Warren et
al., 2006
Species/strain/
sex/number
Rat, Sprague-
Dawley, female, no.
dams/group not
reported
Mouse, B6D2F1,
female, 28-62
dams/group
Rat, Sprague-
Dawley, 116 females
allocated to 1 1
groups
Rat, Sprague-
Dawley, female,
20-25 dams/group
Dose level/
exposure
duration
0,0.11, or 1.1
mg/mL
(0,830, or 8,300
ugM)°
GDO-11
0,24, or 240
mg/kg/d
GD 1-5, 6-10, or
11-15
0, 1.5, or 1,100
ppm
2 mo before
mating and/or
during gestation
0 or 500 mg/kg/d
GD 6-15
Route/
vehicle
Drinking
water
Gavage in
corn oil
Drinking
water
Gavage in
soybean oil
NOAEL;
LOAEL"
Dev. LOEL:
0.11 mg/mL
Mat. NOAEL:
240 mg/kg/d
Dev. NOAEL:
240 mg/kg/d
Mat. NOAEL:
1,100 ppm
Dev. LOAEL:
1.5 ppm
Mat. NOAEL:
500 mg/kg/d
Dev. NOAEL:
500 mg/kg/d
Effects
Embryos collected between
GD 10.5 and 11. Gene
expression at 1.1 mg/mL
TCE: 8 housekeeping genes
t, and one gene |; 3 stress
response genes t, IL-10 J,; 2
cyto-skeletal/cell
adhesion/blood related genes
t, 3 genes ^; 2 heart-specific
genes t . Effects at 0. 1 1
mg/mL reduced considerably.
Two possible markers for
fetal TCE exposure identified
as Serca-2 Ca+2 ATPase and
GPI-pl37.
No maternal toxicity.
No effects on embryonic or
fetal development.
No maternal toxicity.
Sig. t in heart defects,
primarily atrial septal defects,
found at both dose levels in
groups exposed prior to
pregnancy and during
pregnancy, as well as in
group exposed to 1,100 ppm
dose during pregnancy only.
No sig. t in congenital heart
defects in groups exposed
prior to pregnancy only.
No maternal toxicity.
No developmental toxicity.
The incidence of heart
malformations for fetuses
from TCE-treated dams
(3-5%) did not differ from
neg. controls. No eye defects
observed.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-510 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-87. Summary of mammalian in vivo developmental toxicity
studies—oral exposures (continued)
Reference
Fredriksson
etal., 1993
George et
al, 1986
Isaacson
and Taylor,
1989
Johnson et
al., 2003
Narotsky et
al., 1995
Species/strain/
sex/number
Mouse, NMRI, male
pups, 12 pups from
3-4 different
litters/group
Rat, F334, male and
female, 20
pairs/treatment
group,
40 controls/sex
Rat, Sprague-
Dawley, females, 6
dams/group
Rat, Sprague-
Dawley, female,
9-13/group, 55 in
control group
Rat, Fischer 344,
females, 8-12
dams/group
Dose level/
exposure
duration
0,50, or 290
mg/kg/d
PND 10-16
0,0.15, 0.30 or
0.60% micro-
encapsulated TCE
Breeders exposed
1 wk premating,
then for 13 wk;
pregnant ?s
throughout
pregnancy (i.e., 18
wk total)
0,312, or 625
mg/L.
(0,4.0, or 8.1
mg/d)c
Dams (and pups)
exposed from 14 d
prior to mating
until end of
lactation.
0, 2.5 ppb, 250
ppb, 1.5 ppm, or
1,100 ppm
(0, 0.00045,
0.048, 0.218, or
129 mg/kg/d)c
GD 0-22
0, 10.1,32, 101,
320, 475, 633,
844, or 1,125
mg/kg/d
GD 6-15
Route/
vehicle
Gavage in a
20% fat
emulsion
prepared
from egg
lecithin and
peanut oil
Dietary
Drinking
water
Drinking
water
Gavage in
corn oil
NOAEL;
LOAEL3
Dev. LOAEL:
50 mg/kg/d
LOAEL:
0.15%
Dev. LOAEL:
312mg/Lb
Dev. NOAEL:
2.5 ppb
Dev. LOAEL:
250 ppbb
Mat. LOAEL:
475 mg/kg/d
Effects
Rearing activity sig. -I at both
dose levels on PND 60.
Open field testing in pups: a
sig. dose-related trend toward
t time required for male and
female pups to cross the first
grid in the test devise.
Sig. -I myelinated fibers in
the stratum lacunosum-
moleculare of pups.
Reduction in myelin in the
hippocampus.
Sig. t in percentage of
abnormal hearts and the
percentage of litters with
abnormal hearts at >250 ppb.
Sig. dose-related -i- dam B W
gain at all dose levels on GD
6-8 and 6-20. Delayed
parturition at >475 mg/kg/d;
ataxia at >633 mg/kg/d;
mortality at 1,125 mg/kg/d.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-511 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-87. Summary of mammalian in vivo developmental toxicity
studies—oral exposures (continued)
Reference
Narotsky et
al., 1995
(continued)
Narotsky
and
Kavlock,
1995
Noland-
Gerbec et
al., 1986
Peden-
Adams et
al., 2006
Species/strain/
sex/number
Rat, Fischer 344,
females, 16-21
dams/group
Rat, Sprague-
Dawley, females,
9-11 dams/group
Mouse, B6C3F1,
dams and
both sexes offspring,
5 dams/group; 5-7
pups/group at 3 wks;
4-5 pups/sex/group
at 8 wks
Dose level/
exposure
duration
0, 1,125, or
1,500 mg/kg/d
GD 6-19
Oor312mg/L
(Avg. total intake
of dams: 825 mg
TCEover61d.)c
Dams (and pups)
exposed from 14 d
prior to mating
until end of
lactation.
0, 1,400, or
14,000 ppb
Parental mice
and/or offspring
exposed during
mating, and from
GD 0 thru 3 or 8
wks of age
Route/
vehicle
Gavage in
corn oil
Drinking
water
Drinking
water
NOAEL;
LOAEL3
Dev. NOAEL:
32 mg/kg/d
Dev. LOAEL:
101 mg/kg/d
Mat. LOAEL:
1,125 mg/kg/d
Dev. LOAEL:
1,125 mg/kg/d
Dev. LOEL:
312mg/Lb
Dev. LOAEL:
1,400 ppbb
Effects
t full litter resorption and
postnatal mortality at >425
mg/kg/d. Sig. prenatal loss at
1,125 mg/kg/d. PupBW^
(not sig.) on PND 1 and 6.
Sig. t in pups with eye
defects at 1,125 mg/kg/d.
Dose-related (not sig.) t in
pups with eye defects at > 101
mg/kg/d.
Ataxia, -I activity,
piloerection; dose-related -i-
BW gain.
Sig. t full litter resorptions, -i-
live pups/litter; sig. -i- pup BW
on PND 1; sig. t incidences of
microophthalmia and
anophthalmia.
Sig. ^ uptake of 3H-2-DG in
whole brains and cerebella
(no effect in hippocampus)
of exposed pups at 7, 11,
and 16 d, but returned to
control levels by 2 1 d.
At 1,400 ppb: Suppressed
plaque-forming cell (PFC)
responses in males at 3 and 8
wks of age and in females at
8 wks of age. Delayed
hypersensitivity response
increased at 8 wks of age in
females.
At 14,000 ppb: Suppressed
PFC responses in males and
females at 3 and 8 wks of
age. Splenic cell population
decreased in 3 wk old pups.
Increased thymic T-cells at 8
wks of age. Delayed
hypersensitivity response
increased at 8 wks of age in
males and females.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-512 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-87. Summary of mammalian in vivo developmental toxicity
studies—oral exposures (continued)
Reference
Peden-
Adams et
al, 2008
Taylor et
al., 1985
Species/strain/
sex/number
Mouse, MRL +/+,
dams and both sexes
offspring, unknown
no. litters/group,
6-10
offspring/sex/group
Rat, Sprague-
Dawley, females, no.
dams/group not
reported
Dose level/
exposure
duration
0, 1,400, or
14,000 ppb
(vehicle = 1%
emulphore)
Parental mice
and/or offspring
exposed from GD
0 to 12 months of
age
0,312, 625, or
1,250 mg/L
Dams (and pups)
exposed from 14 d
prior to mating
until end of
lactation
Route/
vehicle
Drinking
water
Drinking
water
NOAEL;
LOAEL3
Dev. LOAEL
= 1,400 ppbb
Dev. LOAEL:
312mg/Lb
Effects
At 1,400 ppb: splenic CD4-
/CD8- cells sig.f in females;
thymic CD4+/CD8+ cells sig.
•I in males; 18% f in male
kidney weight.
At 14,000 ppb: thymic T-cell
subpopulations (CD8+,
CD4/CD8-, CD4+) sig. ^ in
males.
Exploratory behavior sig. t in
60- and 90-d old male rats at
all treatment levels.
Locomotor activity was
higher in rats from dams
exposed to 1,250 ppm TCE.
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
aNOAEL, LOAEL, and LOEL (lowest-observed-effect level) are based upon reported study findings. Mat. =
Maternal; Dev. = Developmental.
bDose conversions provided by study author(s).
°Maternal observations not reported.
4.8.3.2.1.1. Inhalation exposures. Dorfmueller et al. (1979) conducted a study in which TCE
was administered by inhalation exposure to groups of approximately 30 female Long-Evans
hooded rats at a concentration of 1,800 ± 200 ppm before mating only, during gestation only, or
throughout the premating and gestation periods. Half of the dams were killed at the end of
gestation and half were allowed to deliver. There were no effects on body weight change or
relative liver weight in the dams. The number of corpora lutea, implantation sites, live fetuses,
fetal body weight, resorptions, and sex ratio were not affected by treatment. In the group
exposed only during gestation, a significant increase in four specific sternebral, vertebral, and rib
findings, and a significant increase in displaced right ovary were observed upon fetal skeletal and
soft tissue evaluation. Mixed function oxidase enzymes (ethoxycoumarin and ethoxyresorbin)
which are indicative of cytochrome P450 and P448 activities, respectively, were measured in the
livers of dams and fetuses, but no treatment-related findings were identified. Postnatal growth
was significantly (p < 0.05) decreased in the group with gestation-only exposures. Postnatal
behavioral studies, consisting of an automated assessment of ambulatory response in a novel
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-513 DRAFT—DO NOT CITE OR QUOTE
-------
1 environment on postnatal days 10, 20, and 100, did not identify any effect on general motor
2 activity of offspring following in utero exposure to TCE.
3 In a study by Schwetz et al. (1975), pregnant Sprague-Dawley rats and Swiss Webster
4 mice (30-40 dams/group) were exposed to TCE via inhalation at a concentration of 300 ppm for
5 7 hours/day on gestation days 6-15. The only adverse finding reported was a statistically
6 significant 4-5% decrease in maternal rat body weight. There were no treatment related effects
7 on pre- and postimplantation loss, litter size, fetal body weight, crown-rump length, or external,
8 soft tissue, or skeletal findings.
9 Hardin et al. (1981) summarized the results of inhalation developmental toxicology
10 studies conducted in pregnant Sprague-Dawley rats and New Zealand white rabbits for a number
11 of industrial chemicals, including TCE. Exposure concentrations of 0 or 500 ppm TCE were
12 administered for 6-7 hours/day, on gestations days 1-19 (rats) or 1-24 (rabbits), and cesarean
13 sections were conducted on gestation days 21 or 30, respectively. There were no adverse
14 findings in maternal animals. No statistically significant increase in the incidence of
15 malformations was reported for either species; however, the presence of hydrocephaly in two
16 fetuses of two TCE-treated rabbit litters was interpreted as a possible indicator of teratogenic
17 potential.
18 Healy et al. (1982) did not identify any treatment-related fetal malformations following
19 inhalation exposure of pregnant inbred Wistar rats to 0 or 100 ppm (535 mg/m3) on GD 8-21. In
20 this study, significant differences between control and treated litters were observed as an
21 increased incidence of total litter loss (p < 0.05), decreased mean fetal weight (p < 0.05), and
22 increased incidence of minor ossification variations (p = 0.003) (absent or bipartite centers of
23 ossification).
24 Carney et al. (2006) investigated the effects of whole-body inhalation exposures to
25 pregnant Sprague-Dawley rats at nominal (and actual) chamber concentrations of 0, 50, 150, or
26 600 ppm TCE for 6 hours/day, 7 days/week on gestation days 6-20. This study was conducted
27 under Good Laboratory Practice regulations according to current U.S. EPA and Organisation for
28 Economic Co-operation and Development (OECD) regulatory testing guidelines (i.e., OPPTS
29 870.3700 and OECD GD 414). Maternal toxicity consisted of a statistically significant decrease
30 (22%) in body weight gain during the first 3 days of exposure to 600-ppm TCE, establishing a
31 no-observed-effect concentration (NOEC) of 150 ppm for dams. No significant difference
32 between control and TCE-treated groups was noted for pregnancy rates, number of corpora lutea,
33 implantations, viable fetuses per litter, percent pre- and postimplantation loss, resorption rates,
34 fetal sex ratios, or gravid uterine weights. External, soft tissue, and skeletal evaluation of fetal
35 specimens did not identify any treatment-related effects. No cardiac malformations were
36 identified in treated fetuses. The fetal NOEC for this study was established at 600 ppm.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-514 DRAFT—DO NOT CITE OR QUOTE
-------
1 Westergren et al. (1984) examined brain specific gravity of NMRI mice pups following
2 developmental exposures to TCE. Male and female mice were separately exposed 24 hours/day
3 (except for limited periods of animal husbandry activities) to 0- or 150-ppm TCE for 30 days and
4 mated during exposure for 7 days. Exposure of the females was continued throughout gestation,
5 until the first litter was born. Offspring (6-12/group; litter origin not provided in report) were
6 terminated by decapitation on PND 1, 10, 21-22, or 30. The specific gravity of the brain frontal
7 cortex, cortex, occipital cortex, and cerebellum were measured. The cortex specific gravity was
8 significantly decreased at PND 1 (p < 0.001) and 10 (p < 0.01) in pups from exposed mice.
9 There were also significant differences (p < 0.05) in the occipital cortex and cerebellum at
10 PND 20-22. This was considered suggestive of delayed maturation. No significant differences
11 between control and treated pups were observed at one month of age.
12
13 4.8.3.2.1.2. Oral exposures. A screening study conducted by Narotsky and Kavlock (1995)
14 assessed the developmental toxicity potential of a number of pesticides and solvents, including
15 TCE. In this study, Fischer 344 rats were administered TCE by gavage at 0, 1,125, and
16 1,500 mg/kg/d on gestation days 6-19, and litters were examined on postnatal days 1, 3, and 6.
17 TCE-related increased incidences of full-litter resorptions, decreased litter sizes, and decreased
18 mean pup birth weights were observed at both treatment levels. Additionally, TCE treatment
19 was reported to be associated with increased incidences of eye abnormalities (microphthalmia or
20 anophthalmia). Increased incidences of fetal loss and percent pups with eye abnormalities were
21 confirmed by Narotsky et al. (1995) in a preliminary dose-setting study that treated Fischer 344
22 rats with TCE by gavage doses of 475, 633, 844, or 1,125 mg/kg/d on gestation days 6-15, and
23 then in a 5 x 5 x 5 mixtures study that used TCE doses of 0, 10.1, 32, 101, and 320 mg/kg/d on
24 GD 6-15. In both studies, dams were allowed to deliver, and pups were examined postnatally.
25 The incidence of ocular defects observed across all TCE treatment levels tested is presented in
26 Table 4-86.
27 Other developmental findings in this study included increased full litter resorption at 475,
28 844, and 1,125 mg/kg/d; increased postnatal mortality at 425 mg/kg/d. Pup body weights were
29 decreased (not significantly) on PND 1 and 6 at 1,125 mg/kg/d. In both the Narotsky and
30 Kavlock (1995) and Narotsky et al. (1995) studies, significantly decreased maternal body weight
31 gain was observed at the same treatment levels at which full litter resorption was noted.
32 Additionally, in Narotsky et al. (1995) maternal observations included delayed parturition at 475,
33 844, and 1,125 mg/kg/d, ataxia at 633 mg/kg/d, and mortality at 1,125 mg/kg/d.
34 Cosby and Dukelow (1992) administered TCE in corn oil by gavage to female B6D2F1
35 mice (28-62/group) on gestation days 1-5, 6-10, or 11-15 (where mating = GD 1). Dose levels
36 were 0, 1/100 and 1/10 of the oral LD50 (i.e., 0, 24.02, and 240.2 mg/kg body weight). Dams
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-515 DRAFT—DO NOT CITE OR QUOTE
-------
1 were allowed to deliver; litters were evaluated for pup count sex, weight, and crown-rump length
2 until weaning (PND 21). Some litters were retained until 6 weeks of age at which time gonads
3 (from a minimum of 2 litters/group) were removed, weighed, and examined. No treatment-
4 related reproductive or developmental abnormalities were observed.
5 A single dose of TCE was administered by gavage to pregnant CD-I mice (9-19/group)
6 at doses of 0, 0.1, or 1.0 ug/kg in distilled water, or 0, 48.3, or 483 mg/kg in olive oil, 24 hours
7 after premating human chorionic gonadotropin (hCG) injection (Coberly et al., 1992). At
8 53 hours after the hCG-injection, the dams were terminated, and the embryos were flushed from
9 excised oviducts. Chimera embryos were constructed, cultured, and examined. Calculated
10 proliferation ratios did not identify any differences between control and treated blastomeres. A
11 lack of treatment-related adverse outcome was also noted when the TCE was administered by i.p.
12 injection to pregnant mice (16-39/group) at 24 and 48 hours post-hCG at doses of 0, 0.01, 0.02,
13 or 10 ug/kg body weight.
14 In a study intended to confirm or refute the cardiac teratogenicity of TCE that had been
15 previously observed in chick embryos, Dawson et al. (1990) continuously infused the gravid
16 uterine horns of Sprague-Dawley rats with solutions of 0-, 15-, or 1,500-ppm TCE (or 1.5 or
17 150-ppm dichloroethylene) on gestation days 7-22. At terminal cesarean section on gestation
18 Day 22, the uterine contents were examined, and fetal hearts were removed and prepared for
19 further dissection and examination under a light microscope. Cardiac malformations were
20 observed in 3% of control fetuses, 9% of the 15-ppm TCE fetuses (p = 0.18), and 14% of the
21 1,500-ppm TCE fetuses, (p = 0.03). There was a >60% increase in the percent of defects with a
22 100-fold increase in dose. No individual malformation or combination of abnormalities was
23 found to be selectively induced by treatment.
24 To further examine these TCE-induced cardiac malformations in rats, Dawson et al.
25 (1993) administered 0, 1.5 or 1,100-ppm TCE in drinking water to female Sprague-Dawley rats.
26 Experimental treatment regimens were (1) a period of approximately 2 months prior to
27 pregnancy plus the full duration of pregnancy, (2) the full duration of pregnancy only, or (3) an
28 average of 3 months before pregnancy only. The average total daily doses of TCE consumed for
29 each exposure group at both dose levels were
30
31
Group 1
Group 2
Group 3
1.5 ppm
23.5 |oL
0.78 nL
3.97 nL
1,100 ppm
1,206 nL
261 nL
1,185 nL
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-516 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
The study also evaluated 0, 0.15, or 110-ppm dichloroethylene in drinking water, with treatment
administered (1) two months prior to pregnancy plus the full duration of pregnancy, or (2) an
average of 2 months before pregnancy only. At terminal cesarean section, uterine contents were
examined, fetuses were evaluated for external defects, and the heart of each fetus was removed
for gross histologic examination under a dissecting microscope, conducted without knowledge of
treatment group. There were no differences between TCE-treated and control group relative to
percentage of live births, implants, and resorptions. The percentage of cardiac defects in TCE-
treated groups ranged from 8.2 to 13.0%, and was statistically significant as compared to the
control incidence of 3%. The dose-response was relatively flat, even in spite of the extensive
difference between the treatment levels. There was a broad representation of various types of
cardiac abnormalities identified, notably including multiple transposition, great artery, septal,
and valve defects (see Table 4-88). No particular combination of defects or syndrome
predominated. Exposure before pregnancy did not appear to be a significant factor in the
incidence of cardiac defects.
Table 4-88. Types of congenital cardiac defects observed in TCE-exposed
fetuses (Dawson et al., 1993, Table 3)
Cardiac abnormalities
d-transposition (right chest)
1-transposition (left chest)
Great artery defects
Atrial septal defects
Mitral valve defects
Tricuspid valve defects
Control
2
1
TCE concentrations
Premating
1,100 ppm
7
1
1.5 ppm
3
Premating/gestation
1,100 ppm
1
19
5
1
1.5 ppm
2
2
5
8
2
Gestation only
1,100 ppm
7
1.5 ppm
1
1
4
Ventricular septal defects
Subaortic
Membranous
Muscular
Endocardial cushion defect
Pulmonary valve defects
Aortic valve defects
Situs inversus
Total abnormalities
Total abnormal hearts
1
2
1
7
7
1
9
9
1
3
1
8
8
4
2
4
2
2
1
41
40
1
1
2
23
23
1
4
1
2
15
11
2
1
1
10
9
18
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-517 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
In an attempt to determine a threshold for cardiac anomalies following TCE exposures,
Johnson et al. (2003, 2005) compiled and reanalyzed data from five studies conducted from
1989-1995. In these studies, TCE was administered in drinking water to Sprague-Dawley rats
throughout gestation (i.e., a total of 22 days) at levels of 2.5 ppb (0.0025 ppm), 250 ppb
(0.25-ppm), 1.5, or 1,100 ppm. The dams were terminated on the last day of pregnancy and
fetuses were evaluated for abnormalities of the heart and great vessels. The control data from the
five studies were combined prior to statistical comparison to the individual treated groups, which
were conducted separately. The study author reported that significant increases in the percentage
of abnormal hearts and the percentage of litters with abnormal hearts were observed in a
generally dose-responsive manner at 250 ppb and greater (see Table 4-89).
Table 4-89. Types of heart malformations per 100 fetuses (Johnson et al.,
2003, Table 2, p. 290)
Type of defect/100 fetuses
Abnormal looping
Coronary artery /sinus
Aortic hypoplasia
Pulmonary artery hypoplasia
Atrial septal defect
Mitral valve defect
Tricuspid valve defect
Ventricular septal defect
Perimembranous (subaortic)
Muscular
Atriventricular septal defect
Pulmonary valve defect
Aortic valve defects
Fetuses with abnormal hearts («)
Total fetuses («)
Litters with fetuses with abnormal hearts/litter (»)
Litter with fetuses with abnormal hearts/no, litters (%)
Control
0.33
1.16
0.17
0.33
0.33
0.17
13
606
9/55
16.4
TCE dose group
1,100 ppm
6.67
2.86
0.95
0.95
1.9
11
105
6/9
66.7
1.5 ppm
1
0.55
0.55
2.21
1.66
0.55
9
181
5/13
38.5
250 ppb
1.82
0.91
0.91
0.91
0.91
5
110
4/9
44.4
2.5 ppb
0
144
0/12
0.0
14
15
16
17
In a study by Fisher et al. (2001), pregnant Sprague-Dawley rats were administered daily
gavage doses on GD 6-15 of TCE (500 mg/kg/d), TCA (300 mg/kg/d), or DCA (300 mg/kg/d).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
1 0/20/09 4-5 1 8 DRAFT— DO NOT CITE OR QUOTE
-------
1 Cesarean delivery of fetuses was conducted on GD 21. Water and soybean oil negative control
2 groups, and a retinoic acid positive control group were also conducted simultaneously. Maternal
3 body weight gain was not significantly different from control for any of the treated groups. No
4 significant differences were observed for number of implantations, resorptions, or litter size.
5 Mean fetal body weight was reduced by treatment with TCA and DC A. The incidence of heart
6 malformations was not significantly increased in treated groups as compared to controls. The
7 fetal rate of cardiac malformations ranged from 3 to 5% across the TCE, TCA, and DCA dose
8 groups and from 6.5 to 2.9% for the soybean and water control dose groups, respectively. It was
9 suggested that the apparent differences between the results of this study and the Dawson et al.
10 (1993) study may be related to factors such as differences in purity of test substances or in the rat
11 strains, or differences in experimental design (e.g., oral gavage versus drinking water, exposure
12 only during the period of organogenesis versus during the entire gestation period, or the use of a
13 staining procedure). The rats from this study were also examined for eye malformations to
14 follow-up on the findings of Narotsky (1995). As reported in Warren et al. (2006), gross
15 evaluation of the fetuses as well as computerized morphometry conducted on preserved and
16 sectioned heads revealed no ocular anomalies in the groups treated with TCE. This technique
17 allowed for quantification of the lens area, global area, medial canthus, distance, and interlocular
18 distance. DCA treatment was associated with statistically significant reductions in the lens area,
19 globe area, and interlocular distance. All four measures were reduced in the TCA-treated group,
20 but not significantly. The sensitivity of the assay was demonstrated successfully with the use of
21 a positive control group that was dosed on GD 6-15 with a known ocular teratogen, retinoic acid
22 (15mg/kg/d).
23 Johnson et al. (1998a, b) conducted a series of studies to determine whether specific
24 metabolites of TCE or dichloroethylene were responsible for the cardiac malformations observed
25 in rats following administration during the period of organogenesis. Several metabolites of the
26 two chemicals were administered in drinking water to Sprague-Dawley rats from GD 1-22.
27 These included carboxy methylcystine, dichloroacetaldehyde, dichlorovinyl cystine,
28 monochloroacetic acid, trichloroacetic acid, trichloroacetaldehyde, and trichloroethanol.
29 Dichloroacetic acid, a primary common metabolite of TCE and dichloroethylene, was not
30 included in these studies. The level of each metabolite administered in the water was based upon
31 the dosage equivalent expected if 1,100 ppm (the limit of solubility) TCE broke down
32 completely into that metabolite. Cesarean sections were performed on GD 22, uterine contents
33 were examined, and fetuses were processed and evaluated for heart defects according to the
34 procedures used by Dawson et al. (1993). No treatment-related maternal toxicity was observed
35 for any metabolite group. Adverse fetal outcomes were limited to significantly increased
36 incidences of fetuses with abnormal hearts (see Table 4-90). Significant increases in fetuses with
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-519 DRAFT—DO NOT CITE OR QUOTE
-------
1 cardiac defects (on a per-fetus and per-litter basis) were observed for only one of the metabolites
2 evaluated, i.e., trichloroacetic acid (2,730 ppm, equivalent to a dose of 291 mg/kg/d). Notably,
3 significant increases in fetuses with cardiac malformations were also observed with 1.5 or
4 l,100-ppmTCE(0.218or 129 mg/kg/d), or with 0.15 or 110-ppm DCE (0.015 or
5 10.64 mg/kg/d), but in each case only with prepregnancy-plus-pregnancy treatment regimens.
6 The cardiac abnormalities observed were diverse and did not segregate to any particular anomaly
7 or grouping. Dose related increases in response were observed for the overall number of fetuses
8 with any cardiac malformation for both TCE and DCE; however, no dose-related increase
9 occurred for any specific cardiac anomaly (Johnson et al., 1998b).
10 The TCE metabolites TCA and DCA were also studied by Smith et al. (1989, 1992).
11 Doses of 0, 330, 800, 1,200, or 1,800 mg/kg TCA were administered daily by oral gavage to
12 Long-Evan hooded rats on gestation days 6-15. Similarly, DCA was administered daily by
13 gavage to Long-Evans rats on GD 6-15 in two separate studies, at 0, 900, 1,400, 1,900, or
14 2,400 mg/kg/d and 0, 14, 140, or 400 mg/kg/d. Embryo lethality and statistically or biologically
15 significant incidences of orbital anomalies (combined soft tissue and skeletal findings) were
16 observed for TCA at >800 mg/kg/d, and for DCA at >900 mg/kg/d. Fetal growth (body weight
17 and crown-rump length) was affected at >330 mg/kg/d for TCE and at >400 mg/kg/d for DCA.
18 For TCA, the most common cardiac malformations observed were levocardia at >330 mg/kg/d
19 and interventricular septal defect at >800 mg/kg/d. For DCA, levocardia was observed at
20 >900 mg/kg/d, interventricular septal defect was observed at >1,400 mg/kg/d, and a defect
21 between the ascending aorta and right ventricle was observed in all treated groups (i.e.,
22 >14 mg/kg/d, although the authors appeared to discount the single fetal finding at the lowest dose
23 tested). Thus, NOAELs were not definitively established for either metabolite, although it
24 appears that TCA was generally more potent than DCA in inducing cardiac abnormalities.
25
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-520 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-90. Congenital cardiac malformations (Johnson et al., 1998b, Table 2, p. 997)
to
O k^j
o J5"
vo £;•
I
I
§
^.
Co'
1
TO'
I
TO
Heart abnormalities
Abnormal looping
Aortic hypoplasia
Pulmonary artery hypoplasia
Atrial septal defects
Mitral valve defects, hypoplasia or ectasia
Tricuspid valve defects, hypoplasia or ectasia
Ventricular septal defects
Perimembranous a
Muscular
Atrioventricual septal defects
Pulmonary valve defects
Aortic valve defects
Situs inversus
Total
Abnormal hearts
Fetuses with abnormal hearts
Fetuses
Treatment group
Normal
water
2
-
-
7
1
-
2
2
1
-
-
-
15
13
605
TCE
p+p
1,100
ppm
-
1
-
19
5
1
6
4
-
2
2
1
41
40*
434
TCE
p+p
1.5
ppm
2
1
1
5
8
1
2
-
-
1
2
-
23
22*
255
TCE
P
1,100
ppm
-
-
-
7
-
-
1
4
1
-
2
-
15
11*
105
DCE
p+p
110
ppm
-
1
-
11
4
1
4
2
1
1
2
-
25
24*
184
DCE
p+p
0.15
ppm
-
-
-
7
o
J
-
1
1
-
-
3
-
15
14*
121
TCAA
P
2,730
ppm
-
1
2
3
1
-
4
1
-
1
-
-
13
12*
114
MCAA
P
1,570
ppm
-
-
1
3
-
-
-
-
-
3
-
-
7
6
132
TCEth
P
1,249
ppm
-
1
-
-
1
1
-
1
-
1
1
-
6
5
121
TCAld
P
1,232
ppm
-
-
-
2
2
-
o
J
-
-
1
-
-
8
8
248
DCAld
P
174
ppm
-
1
2
-
-
-
-
-
-
-
-
-
3
3
101
CMC
P
473
ppm
-
-
-
-
-
-
1
2
-
-
1
-
4
4
85
DCVC
P
50
ppm
-
1
-
1
1
-
-
2
-
-
-
-
5
5
140
O I "Subaortic.
^ ^ bPer-fetus statistical significance (Fisher exact test).
hHOq
H-] <^
W S p+p = pregnancy and prepregnancy, p = pregnancy.
O
H
W
-------
1 These findings were followed up by a series of studies on DC A reported by Epstein et al.
2 (1992), which were designed to determine the most sensitive period of development and further
3 characterize the heart defects. In these studies, Long-Evans hooded rats were dosed by oral
4 gavage with a single dose of 2,400 mg/kg/d on selected days of gestation (6-8, 9-11, or 12-15);
5 with a single dose of 2,400 mg/kg on Days 10, 11, 12, or 13; or with a single dose of
6 3,500 mg/kg on Days 9, 10, 11, 12, or 13. The heart defects observed in these studies were
7 diagnosed as high interventricular septal defects rather than membranous type interventricular
8 septal defects. The authors hypothesized that high intraventricular septal defects are a specific
9 type of defect produced by a failure of proliferating interventricular septal tissue to fuse with the
10 right tubercle of the atrioventricular cushion tissue. This study identified gestation days 9
11 through 12 as a particularly sensitive period for eliciting high interventricular septal defects. It
12 was postulated that DC A interferes with the closure of the tertiary interventricular foramen,
13 allowing the aorta to retain its embryonic connection with the right ventricle. Further, it was
14 suggested that the selectivity of DC A in inducing cardiac malformations may be due to the
15 disruption of a discrete cell population.
16 TCE and its metabolites DCE and TCAA were administered in drinking water to
17 pregnant Sprague-Dawley rats from gestation days 0-11 (Collier et al., 2003). Treatment levels
18 wereO, 110, or 1,100 ppm (i.e., 0, 830 or 8,300 ugM) TCE; 0, 11, or 110 ppm (i.e., 0, 110, or
19 1,100 ugM) DCE; 0, 2.75, or 27.3 mg/mL (i.e., 0, 10, or 100 mM) TCAA. Embryos (including
20 hearts) were harvested between embryonic days 10.5-11, since this is the stage at which the
21 developmental processes of myoblast differentiation, cardiac looping, atrioventricular valve
22 formation, and trabeculation would typically be occurring. A PCR based subtraction scheme
23 was used to identify genes that were differentially regulated with TCE or metabolite exposure.
24 Numerous differentially regulated gene sequences were identified. Up-regulated transcripts
25 included genes associated with stress response (Hsp 70) and homeostasis (several ribosomal
26 proteins). Down-regulated transcripts included extracellular matrix components (GPI-pl37 and
27 vimentin) and Ca2+ responsive proteins (Serca-2 Ca2+-ATPase and p-catenin). Serca-2 Ca2+ and
28 GPI-pl37 were identified as two possible markers for fetal TCE exposure. Differential
29 regulation of expression of these markers by TCE was confirmed by dot blot analysis and
30 semi quantitative real time PCR with decreased expression seen at levels of TCE exposure
31 between 100 and 250 ppb (0.76 and 1.9 uM).
32
33 4.8.3.2.1.2.1. Developmental neurotoxicity and developmental immunotoxicity. Several studies
34 were conducted that included assessments of the effects of TCE oral exposure on the developing
35 nervous system (Fredriksson et al., 1993; Isaacson and Taylor, 1989; Noland-Gerbec et al., 1986;
36 George et al., 1986; Dorfmueller et al., 1979; Blossom et al., 2008) or immune system (Peden-
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-522 DRAFT—DO NOT CITE OR QUOTE
-------
1 Adams et al., 2006, 2008; Blossom and Doss, 2007; Blossom et al., 2008). These studies,
2 summarized below, are addressed in additional detail in Section 4.3 (nervous system) and
3 Section 4.6.2.1.2 (immune system).
4
5 4.8.3.2.1.2.2. Developmentalneurotoxicity. Fredriksson et al. (1993) conducted a study in male
6 NMRI weanling mice (12/group, selected from 3-4 litters), which were exposed to
7 trichloroethylene by oral gavage at doses of 0 (vehicle), 50, or 290 mg/kg/d TCE in a fat
8 emulsion vehicle, on PNDs 10-16. Locomotor behavior (horizontal movement, rearing and total
9 activity) were assessed over three 20-minute time periods at postnatal days 17 and 60. There
10 were no effects of treatment in locomotor activity at PND 17. At PND 60, the mice treated with
11 50 and 290 mg/kg/d TCE showed a significant (p < 0.01) decrease in rearing behavior at the
12 0-20 and 20-40 minute time points, but not at the 40-60 minute time point. Mean rearing
13 counts were decreased by over 50% in treated groups as compared to control. Horizontal activity
14 and total activity were not affected by treatment.
15 Open field testing was conducted in control and high-dose Fl weanling Fischer 344 rat
16 pups in an NTP reproduction and fertility study with continuous breeding (George et al., 1986).
17 In this study, TCE was administered at dietary levels of 0, 0.15, 0.30, or 0.60%. The open field
18 testing revealed a significant (p < 0.05) dose-related trend toward an increase in the time required
19 for male and female pups to cross the first grid in the testing device, suggesting an effect on the
20 ability to react to a novel environment.
21 Taylor et al. (1985) administered TCE in drinking water (0, 312, 625, or 1,250 ppm) to
22 female Sprague-Dawley rats for 14 days prior to breeding, and from gestation Day 0 through
23 offspring postnatal Day 21. The number of litters/group was not reported, nor did the study state
24 how many pups per litter were evaluated for behavioral parameters. Exploratory behavior was
25 measured in the pups in an automated apparatus during a 15-minute sampling period on PND 28,
26 60, and 90. Additionally, wheel-running, feeding, and drinking behavior was monitored
27 24 hours/day on PND 55-60. The number of exploratory events was significantly increased by
28 approximately 25-50% in 60- and 90-day old male TCE-treated rats at all dose levels, with the
29 largest effect observed at the highest dose level tested, although there were no effects of
30 treatment on the number of infrared beam-breaks. No difference between control and treated rats
31 was noted for pups tested on PND 28. Wheel-running activity was increased approximately 40%
32 in 60-day old males exposed to 1,25-ppm TCE as compared to controls. It is notable that
33 adverse outcomes reported in the developmentally-exposed offspring on this study were
34 observed long after treatment ceased.
35 Using a similar treatment protocol, the effects of TCE on development of myelinated
36 axons in the hippocampus was evaluated by Isaacson and Taylor (1989) in Sprague-Dawley rats.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-523 DRAFT—DO NOT CITE OR QUOTE
-------
1 Female rats (6/group) were exposed in the drinking water from 14 days prior to breeding and
2 through the mating period; then the dams and their pups were exposed throughout the prenatal
3 period and until PND 21, when they were sacrificed. The dams received 0, 312 or 625 ppm (0,
4 4, or 8.1 mg/day TCE in the drinking water. Myelinated fibers were counted in the hippocampus
5 of 2-3 pups per treatment group at PND 21, revealing a decrease of approximately 40% in
6 myelinated fibers in the CA1 area of the hippocampus of pups from dams at both treatment
7 levels, with no dose-response relationship. There was no effect of TCE treatment on myelination
8 in several other brain regions including the internal capsule, optic tract or fornix.
9 A study by Noland-Gerbec et al. (1986) examined the effect of pre- and perinatal
10 exposure to TCE on 2-deoxyglucose (2-DG) uptake in the cerebellum, hippocampus and whole
11 brain of neonatal rats. Sprague-Dawley female rats (9-1 I/group) were exposed via drinking
12 water to 0 or 312 mg TCE/liter distilled water from 14 days prior to mating until their pups were
13 euthanized at postnatal Day 21. The total TCE dose received by the dams was 825 mg over the
14 61-day exposure period. Pairs of male neonates were euthanized on PND 7, 11, 16, and 21.
15 There was no significant impairment in neonatal weight or brain weight attributable to treatment,
16 nor were other overt effects observed. 2-DG uptake was significantly reduced from control
17 values in neonatal whole brain (9-11%) and cerebellum (8-16%) from treated rats at all ages
18 studied, and hippocampal 2-DG uptake was significantly reduced (7-21% from control) in
19 treated rats at all ages except at PND 21.
20 In a study by Blossom et al. (2008), MRL +/+ mice were treated in the drinking water
21 with 0 or 0.1 mg/mL TCE from maternal GD 0 through offspring PND 42. Based on drinking
22 water consumption data, average maternal doses of TCE were 25.7 mg/kg/d, and average
23 offspring (PND 24-42) doses of TCE were 31.0 mg/kg/d. In this study, a subset of offspring
24 (3 randomly selected neonates from each litter) was evaluated for righting reflex on PNDs 6, 8,
25 and 10; bar-holding ability on PNDs 15 and 17; and negative geotaxis on PNDs 15 and 17; none
26 of these were impaired by treatment. In an assessment of offspring nest building on PND 35,
27 there was a significant association between impaired nest quality and TCE exposure; however,
28 TCE exposure did not have an effect on the ability of the mice to detect social and nonsocial
29 odors on PND 29 using olfactory habituation and dishabituation methods. Resident intruder
30 testing conducted on PND 40 to evaluate social behaviors identified significantly more
31 aggressive activities (i.e., wrestling and biting) in TCE-exposed juvenile male mice as compared
32 to controls. Cerebellar tissue homogenates from the male TCE-treated mice had significantly
33 lower GSH levels and GSH:oxidized GSH (GSH:GSSG) ratios, indicating increased oxidative
34 stress and impaired thiol status; these have been previously reported to be associated with
35 aggressive behaviors (Franco et al., 2006). Qualitative histopathological examination of the
36 brain did not identify alterations indicative of neuronal damage or inflammation. Although the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-524 DRAFT—DO NOT CITE OR QUOTE
-------
1 study author attempted to link the treatment-related alterations in social behaviors to the potential
2 for developmental exposures to TCE to result in autism in humans, this association is not
3 supported by data and is considered speculative at this time.
4 As previously noted, postnatal behavioral studies conducted by Dorfmueller et al. (1979)
5 did not identify any changes in general motor activity measurements of rat offspring on PND 10,
6 20, and 100 following maternal gestational inhalation exposure to TCE at 1,800 ± 200 ppm.
7
8 4.8.3.2.1.2.3. Developmental immunotoxicity. Peden-Adams et al. (2006) assessed the potential
9 for developmental immunotoxicity following TCE exposures. In this study, B6C3F1 mice
10 (5/sex/group) were administered TCE via drinking water at dose levels of 0, 1,400 or 14,000 ppb
11 from maternal gestation Day 0 to either postnatal 3 or 8, when offspring lymphocyte
12 proliferation, NK cell activity, SRBC-specific IgM production (PFC response), splenic B220+
13 cells, and thymus and spleen T-cell immunophenotypes were assessed. (A total of 5-7 pups per
14 group were evaluated at Week 3, and the remainder were evaluated at Week 8.) Observed
15 positive responses consisted of suppressed PFC responses in males at both ages and both TCE
16 treatment levels, and in females at both ages at 14,000 ppb and at 8 weeks of age at 1,400 ppb.
17 Spleen numbers of B220+ cells were decreased in 3-week old pups at 14,000 ppb. Pronounced
18 increases in all thymus T-cell subpopulations (CD4+, CD8+, CD4+/CD8+, and CD4-/CD8-)
19 were observed at 8-weeks of age. Delayed hypersensitivity response, assessed in offspring at
20 8 weeks of age, was increased in females at both treatment levels and in males at 14,000 ppb
21 only. No treatment-related increase in serum anti-dsDNA antibody levels was found in the
22 offspring at 8 weeks of age.
23 In a study by Blossom and Doss (2007), TCE was administered to groups of pregnant
24 MRL +/+ mice in drinking water at levels of 0, 0.5 or 2.5 mg/mL. TCE was continuously
25 administered to the offspring until young adulthood (i.e., 7-8 weeks of age). Offspring
26 postweaning body weights were significantly decreased in both treated groups. Decreased
27 spleen cellularity and reduced numbers of CD4+, CD8+, and B220+ lymphocyte subpopulations
28 were observed in the postweaning offspring. Thymocyte development was altered by TCE
29 exposures (significant alterations in the proportions of double-negative subpopulations and
30 inhibition of in vitro apoptosis in immature thymocytes). A dose-dependent increase in CD4+
31 and CD8+ T-lymphocyte IFNy was observed in peripheral blood by 4-5 weeks of age, although
32 these effects were no longer observed at 7-8 weeks of age. Serum anti-histone autoantibodies
33 and total IgG2a were significantly increased in treated offspring; however, no histopathological
34 signs of autoimmunity were observed in the liver and kidneys at sacrifice.
35 Blossom et al. (2008) administered TCE to MRL +/+ mice (8 dams/group) in the drinking
36 water at levels of 0 or 0.1 mg/mL from GD 0 through offspring postnatal Day 42. Average
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-525 DRAFT—DO NOT CITE OR QUOTE
-------
1 maternal doses of TCE were 25.7 mg/kg/d, and average offspring (PND 24-42) doses of TCE
2 were 31.0 mg/kg/d. Subsets of offspring were sacrificed at PND 10 and 20, and thymus
3 endpoints (i.e., total cellularity, CD4+/CD8+ ratios, CD24 differentiation markers, and double-
4 negative subpopulation counts) were evaluated. Evaluation of the thymus identified a significant
5 treatment-related increase in cellularity, accompanied by alterations in thymocyte subset
6 distribution, at PND 20 (sexes combined). TCE treatment also appeared to promote T-cell
7 differentiation and maturation at PND 42. Indicators of oxidative stress were measured in the
8 thymus at PND 10 and 20, and in the brain at PND 42,.and ex vivo evaluation of cultured
9 thymocytes indicated increased ROS generation. Mitogen-induced intracellular cytokine
10 production by splenic CD4+ and CD8+ T-cells was evaluated in juvenile mice and brain tissue
11 was examined at PND 42 for evidence of inflammation. Evaluation of peripheral blood
12 indicated that splenic CD4+ T-cells from TCE-exposed PND 42 mice produced significantly
13 greater levels of IFN-y and IL-2 in males and TNF-a in both sexes. There was no effect on
14 cytokine production on PND 10 or 20.
15 Peden-Adams et al. (2008) administered TCE to MRL+/+ mice (unspecified number of
16 dams/group) in drinking water at levels of 0, 1,400, or 14,000 ppb from GD 0 and continuing
17 until the offspring were 12 months of age. At 12 months of age, final body weight; spleen,
18 thymus, and kidney weights; spleen and thymus lymphocyte immunophenotyping (CD4 or
19 CDS); splenic B-cell counts; mitogen-induced splenic lymphocyte proliferation; serum levels of
20 autoantibodies to dsDNA and GA, periodically measured from 4 to 12 months of age; and
21 urinary protein measures were recorded. Reported sample sizes for the offspring measurements
22 varied from 6 to 10 per sex per group; the number of source litters represented within each
23 sample was not specified. The only organ weight alteration was an 18% increase in kidney
24 weight in the 1,400 ppb males. Splenic CD4-/CD8- cells were altered in female mice (but not
25 males) at 1,400 ppm only. Splenic T-cell populations, numbers of B220+ cells, and lymphocyte
26 proliferation were not affected by treatment. Populations of thymic T-cell subpopulations
27 (CD8+, CD4-/CD8-, and CD4+) were significantly decreased in male but not female mice
28 following exposure to 14,000 ppb TCE, and CD4+/CD8+ cells were significantly reduced in
29 males by treatment with both TCE concentrations. Autoantibody levels (anti-dsDNA and anti-
30 GA) were not increased in the offspring over the course of the study.
31 Although all of the developmental immunotoxicity studies with TCE (Peden-Adams et
32 al., 2006, 2008; Blossom and Doss, 2007; Blossom et al., 2008) exposed the offspring during
33 critical periods of pre- and postnatal immune system development, they were not designed to
34 assess issues such as post-treatment recovery, latent outcomes, or differences in severity of
35 response that might be attributed to the early life exposures.
36
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-526 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.3.2.1.3. Intraperitoneal exposures. The effect of TCE on pulmonary development was
2 evaluated in a study by Das and Scott (1994). Pregnant Swiss-Webster mice (5/group) were
3 administered a single intraperitoneal injection of TCE in peanut oil at doses of 0 or 3,000 mg/kg
4 on gestation Day 17 (where mating = Day 1). Lungs from GD 18 and 19 fetuses and from
5 neonates on PND 1, 5, and 10 were evaluated for phospholipid content, DNA, and microscopic
6 pathology. Fetal and neonatal (PND 1) mortality was significantly increased (p < 0.01) in the
7 treated group. Pup body weight and absolute lung weight were significantly decreased (p < 0.05)
8 on PND 1, and mean absolute and relative (to body weight) lung weights were significantly
9 decreased on GD 18 and 19. Total DNA content (ug/mg lung) was similar between control and
10 treated mice, but lung phospholipid was significantly (p < 0.05) reduced on GD 19 and
11 significantly increased (p < 0.05) on PND 10 in the TCE-treated group. Microscopic
12 examination revealed delays in progressive lung morphological development in treated offspring,
13 first observed at GD 19 and continuing at least through PND 5.
14
15 4.8.3.2.2. Studies in nonmammalian species.
16 4.8.3.2.2.1. Avian. Injection of White Leghorn chick embryos with 1, 5, 10, or 25 umol TCE
17 per egg on Days 1 and 2 of embryogenesis demonstrated mortality, growth defects, and
18 morphological anomalies at evaluation on Day 14 (Bross et al., 1983). These findings were
19 consistent with a previous study that had been conducted by Elovaara et al. (1979). Up to 67%
20 mortality was observed in the treated groups, and most of the surviving embryos were
21 malformed (as compared to a complete absence of malformed chicks in the untreated and
22 mineral-oil-treated control groups). Reported anomalies included subcutaneous edema,
23 evisceration (gastroschisis), light dermal pigmentation, beak malformations, club foot, and
24 patchy feathering. Retarded growth was observed as significantly (p < 0.05) reduced crown-
25 rump, leg, wing, toe, and beak lengths as compared to untreated controls. This study did not
26 identify any liver damage or cardiac anomalies.
27 In a study by Loeber et al. (1988), 5, 10, 15, 20, or 25 umol TCE was injected into the air
28 space of White Longhorn eggs at embryonic stages 6, 12, 18, or 23. Embryo cardiac
29 development was examined in surviving chicks in a double-blinded manner at stages 29, 34, or
30 44. Cardiac malformations were found in 7.3% of TCE-treated hearts, compared to 2.3% of
31 saline controls and 1.5% of mineral oil controls. The observed defects included septal defects,
32 cor biloculare, conotruncal abnormalities, atrioventricular canal defects, and abnormal cardiac
33 muscle.
34 Drake et al. (2006a) injected embryonated White Leghorn chicken eggs (Babcock or
35 Bovan strains) with 0, 0.4, 8, or 400 ppb TCE per egg during the period of cardiac valvuloseptal
36 morphogenesis (i.e., 2-3.3 days incubation). The injections were administered in four aliquots at
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-527 DRAFT—DO NOT CITE OR QUOTE
-------
1 Hamberger and Hamilton (HH) stages 13, 15, 17, and 20, which spanned the major events of
2 cardiac cushion formation, from induction through mesenchyme transformation and migration.
3 Embryos were harvested 22 hours after the last injection (i.e., HH 24 or HH 30) and evaluated
4 for embryonic survival, apoptosis, cellularity and proliferation, or cardiac function. Survival was
5 significantly reduced for embryos at 8 and 400 ppb TCE at HH 30. Cellular morphology of
6 cushion mesenchyme, cardiomyocytes, and endocardiocytes was not affected by TCE treatment;
7 however, the proliferative index was significantly increased in the atrioventricular canal (AVC)
8 cushions at both treatment levels and in the outflow tract (OFT) cushions at 8 ppb. This resulted
9 in significant cushion hypercellularity for both the OFT and AVC of TCE-treated embryos.
10 Similar outcomes were observed in embryos when TCA or TCOH was administered, and the
11 effects of TCA were more severe than for TCE. Doppler ultrasound assessment of cardiac
12 hemodynamics revealed no effects of TCE exposure on cardiac cycle length or heart rate;
13 however, there was a reduction in dorsal aortic blood flow, which was attributed to a 30.5%
14 reduction in the active component of atrioventricular blood flow. Additionally the passive-to-
15 active atrioventricular blood flow was significantly increased in treated embryos, and there was a
16 trend toward lower stroke volume. The overall conclusion was that exposure to 8 ppb TCE
17 during cushion morphogenesis reduced the cardiac output of the embryos in this study. The
18 findings of cardiac malformations and/or mortality following in ovo exposure to chick embryos
19 with 8 ppb TCE during the period of valvuloseptal morphogenesis has also been confirmed by
20 Rufer et al. (2008).
21 In a follow-up study, Drake et al. (2006b) injected embryonated White Leghorn chicken
22 eggs with TCE or TCA during the critical window of avian heart development, beginning at HH
23 stage 3+ when the primary heart field is specified in the primitive streak and ending
24 approximately 50 hours later at HH stage 17, at the onset of chambering. Total dosages of 0, 0.2,
25 2, 4, 20, or 200 nmol (equivalent to 0, 0.4, 4, 8, 40, or 400 ppb) were injected in four aliquots
26 into each egg yolk during this window (i.e., at stages 3+, 6, 13, and 17: hours 16, 24, 46, and 68).
27 Embryos were harvested at 72 hours, 3.5 days, 4 days or 4.25 days (HH stages 18, 21, 23, or 24,
28 respectively) and evaluated for embryonic survival, cardiac function, or cellular parameters.
29 Doppler ultrasound technology was utilized to assess cardiovascular effects at HH 18, 21, and
30 23. In contrast with the results of Drake et al. (2006a), all of the functional parameters assessed
31 (i.e., cardiac cycle length, heart rate, stroke volume, and dorsal aortic and atrioventricular blood
32 flow) were similar between control and TCE- or TCA-treated embryos. The authors attributed
33 this difference in response between studies to dependence upon developmental stage at the time
34 of exposure. In this case, the chick embryo was relatively resistant to TCE when exposure
35 occurred during early cardiogenic stages, but was extremely vulnerable when TCE exposure
36 occurred during valvuloseptal morphogenesis. It was opined that this could explain why some
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-528 DRAFT—DO NOT CITE OR QUOTE
-------
1 researchers have observed no developmental cardiac effects after TCE exposure to mammalian
2 models, while others have reported positive associations.
O
4 4.8.3.2.2.2. Amphibian. The developmental toxicity of TCE was evaluated in the Frog Embryo
5 Teratogenesis Assay: Xenopus by Fort et al. (1991, 1993). Late Xenopus laevis blastulae were
6 exposed to TCE, with and without exogenous metabolic activation systems, or to TCE
7 metabolites (dichloroacetic acid, trichloroacetic acid, trichloroethanol, or oxalic acid), and
8 developmental toxicity ensued. Findings included alterations in embryo growth, and increased
9 types and severity of induced malformations. Findings included cardiac malformations that were
10 reportedly similar to those that had been observed in avian studies. It was suggested that a mixed
11 function oxidase-mediated reactive epoxide intermediate (i.e., TCE-oxide) may play a significant
12 role in observed developmental toxicity in in vitro tests.
13 Likewise, McDaniel et al. (2004) observed dose-dependent increases in developmental
14 abnormalities in embryos of four North American amphibian species (wood frogs, green frogs,
15 American toads, and spotted salamanders) following 96-hour exposures to TCE. Median
16 effective concentrations (ECso) for malformations was 40 mg/L for TCE in green frogs, while
17 American toads were less sensitive (with no ECso at the highest concentration tested—85 mg/L).
18 Although significant mortality was not observed, the types of malformations noted would be
19 expected to compromise survival in an environmental context.
20
21 4.8.3.2.2.3. Invertebrate. The response of the daphnid Ceriodaphnia dubia to six industrial
22 chemicals, including TCE, was evaluated by Niederlehner et al. (1998). Exposures were
23 conducted for 6-7 days, according to standard U.S. EPA testing guidelines. Lethality,
24 impairment of reproduction, and behavioral changes, such as narcosis and abnormal movement,
25 were observed with TCE exposures. The reproductive sublethal effect concentration value for
26 TCE was found to be 82 uM.
27
28 4.8.3.2.3. In vitro studies. Rat whole embryo cultures were used by Saillenfait et al. (1995) to
29 evaluate the embryotoxicity of TCE, tetrachloroethylene, and four metabolites (trichloroacetic
30 acid, dichloroacetic acid, chloral hydrate, and trichloroacetyl chloride). In this study, explanted
31 embryos of Sprague-Dawley rats were cultured in the presence of the test chemicals for 46 hours
32 and subsequently evaluated. Concentration-dependant decreases in growth and differentiation,
33 and increases in the incidence of morphologically abnormal embryos were observed for TCE at
34 >5 mM.
35 Whole embryo cultures were also utilized by Hunter et al. (1996) in evaluating the
36 embryotoxic potential of a number of disinfection by-products, including the TCE metabolites
37 DCA and TCA. CD-I mouse conceptuses (GD 9; 3-6 somites) were cultured for 24-26 hours in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-529 DRAFT—DO NOT CITE OR QUOTE
-------
1 treated medium. DCA levels assessed were 0, 734, 1,468, 4,403, 5,871, 7,339, 11,010, or
2 14,680 uM; TCA levels assessed were 0, 500, 1,000, 2,000, 3,000, 4,000, 5,000 uM. For DCA,
3 neural tube defects were observed at levels of >5,871 uM, heart defects were observed at
4 >7,339 uM, and eye defects were observed at levels of >11,010 uM. For TCA, neural tube
5 defects were observed at levels of >2,000 uM, heart and eye defects were observed at
6 >3,000 uM. The heart defects for TCA were reported to include incomplete looping, a reduction
7 in the length of the heart beyond the bulboventricular fold, and a marked reduction in the caliber
8 of the heart tube lumen. Overall benchmark concentrations (i.e., the lower limit of the 95%
9 confidence interval required to produce a 5% increase in the number of embryos with neural tube
10 defects) were 2,451.9 uM for DCA and 1,335.8 uM for TCA (Richard and Hunter, 1996).
11 Boyer et al. (2000) used an in vitro chick-atrioventricular (AV) canal culture to test the
12 hypothesis that TCE might cause cardiac valve and septal defects by specifically perturbing
13 epithelial-mesenchymal cell transformation of endothelial cells in the AV canal and outflow tract
14 areas of the heart. AV explants from Stage 16 White Leghorn chick embryos were placed in
15 hydrated collagen gels, with medium and TCE concentrations of 0, 50, 100, 150, 200, or
16 250 ppm. TCE was found to block the endothelial cell-cell separation process that is associated
17 with endothelial activation as well as to inhibit mesenchymal cell formation across all TCE
18 concentrations tested. TCE did not, however, have an effect on the cell migration rate of fully
19 formed mesenchymal cells. TCE-treatment was also found to inhibit the expression of
20 transformation factor Mox-1 and extracellular matrix protein fibrillin 2, two protein markers of
21 epithelial-mesenchyme cell transformation.
22 4.8.3.3. Discussion/Synthesis of 'Developmental Data
23 In summary, an overall review of the weight of evidence in humans and experimental
24 animals is suggestive of the potential for developmental toxicity with TCE exposure. A number
25 of developmental outcomes have been observed in the animal toxicity and the epidemiological
26 data, as discussed below. These include adverse fetal/birth outcomes including death
27 (spontaneous abortion, perinatal death, pre- or postimplantation loss, resorptions), decreased
28 growth (low birth weight, small for gestational age, intrauterine growth restriction, decreased
29 postnatal growth), and congenital malformations, in particular cardiac defects. Postnatal
30 developmental outcomes include developmental neurotoxicity, developmental immunotoxicity,
31 and childhood cancer.
32
33 4.8.3.3.1. Adverse fetal and early neonatal outcomes. Studies that demonstrate adverse fetal
34 or early neonatal outcomes are summarized in Table 4-91. In human studies of prenatal TCE
35 exposure, increased risk of spontaneous abortion was observed in some studies (ATSDR, 2001;
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-530 DRAFT—DO NOT CITE OR QUOTE
-------
1 Taskinen et al., 1994; Windham et al., 1991), but not in others (ATSDR, 2001, 2008;
2 Goldberg et al., 1990; Lagakos et al., 1986; Lindbohm et al., 1990; Taskinen et al., 1989). In
3 addition, perinatal deaths were observed after 1970, but not before 1970 (Lagakos et al., 1986).
4 In rodent studies that examined offspring viability and survival, there was an indication that TCE
5 exposure may have resulted in increased pre-and/or postimplantation loss (Kumar et al., 2000a;
6 Healy et al., 1982; Narotsky and Kavlock, 1995), and in reductions in live pups born as well as in
7 postnatal and postweaning survival (George et al., 1985, 1986).
Table 4-91. Summary of adverse fetal and early neonatal outcomes
associated with TCE exposures
10
Positive finding
Spontaneous abortion, miscarriage,
pre-and/or postimplantation loss
Perinatal death, reduction in live births
Postnatal and postweaning survival
Decreased birth weight, small for
gestational age, postnatal growth
Species
Human
Rat
Human
Mouse
Rat
Mouse
Rat
Human
Mouse
Rat
Citation
ATSDR, 200 la
Taskinen et al., 1994a
Windham et al., 1991
Kumar et al., 2000a
Healy etal., 1982
Narotsky and Kavlock, 1995
Narotsky et al., 1995
Lagakos et al., 1986b
George etal., 1985
George et al., 1986
George etal., 1985
George etal., 1986
ATSDR, 1998
ATSDR, 2006
Rodenbeck et al., 2000C
Windham et al., 1991
George etal., 1985
George etal., 1986
Healy etal., 1982
Narotsky and Kavlock, 1995
Narotsky et al., 1995
11
12
13
14
15
16
17
18
aNot significant.
bObserved for exposures after 1970, but not before.
Increased risk for very low birth weight but not low birth weight or full-term low birth weight.
Decreased birth weight and small for gestational age was observed (ATSDR, 1998, 2006;
Rodenbeck et al., 2000; Windham et al., 1991), however, no association was observed in other
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-531 DRAFT—DO NOT CITE OR QUOTE
-------
1 studies (Bove, 1996; Bove et al., 1995; Lagakos et al., 1986). While comprising both
2 occupational and environmental exposures, these human studies are overall not highly
3 informative due to their small numbers of cases and limited exposure characterization or to the
4 fact that exposures to mixed solvents were involved. However, decreased fetal weight, live birth
5 weights and postnatal growth were also observed in rodents (George et al., 1985, 1986; Healy et
6 al., 1982; Narotsky and Kavlock, 1995), adding to the weight of evidence for this endpoint. It is
7 noted that the rat studies reporting effects on fetal or neonatal viability and growth used Fischer
8 344 or Wistar rats, while several other studies, which used Sprague-Dawley rats, reported no
9 increased risk in these developmental measures (Carney et al., 2006; Hardin et al., 1981;
10 Schwetz et al., 1975).
11 Overall, based on weakly suggestive epidemiologic data and fairly consistent laboratory
12 animal data, it can be concluded that TCE exposure poses a potential hazard for prenatal losses
13 and decreased growth or birth weight of offspring.
14
15 4.8.3.3.2. Cardiac malformations. A discrete number of epidemiological studies and studies
16 in laboratory animal models have identified an association between TCE exposures and cardiac
17 defects in developing embryos and/or fetuses. These are listed in Table 4-92. Additionally, a
18 number of avian and rodent in vivo studies and in vitro assays have examined various aspects of
19 the induction of cardiac malformations.
20 In humans, an increased risk of cardiac defects has been observed after exposure to TCE
21 in studies reported by ATSDR (2006, 2008) and Yauck et al. (2004), although others saw no
22 significant effect (Bove et al., 1995; Bove, 1996; Goldberg et al., 1990; Lagakos et al., 1986),
23 possibly due to a small number of cases. In addition, altered heart rate was seen in one study
24 (Jasinka, 1965, translation). A cohort of water contamination in Santa Clara County, California
25 is often cited as a study of TCE exposure and cardiac defects; however, the chemical of exposure
26 is in fact trichloroethane, not TCE (Deane et al., 1989; Swan et al., 1989).
27 In laboratory animal models, avian studies were the first to identify adverse effects of
28 TCE exposure on cardiac development. As described in Section 4.8.2.2.1, cardiac malformations
29 have been reported in chick embryos exposed to TCE (Bross et al., 1983; Loeber et al., 1988;
30 Boyer et al., 2000; Drake et al., 2006a, b; Mishima et al., 2006; Rufer et al., 2008). Additionally,
31 a number of studies were conducted in rodents in which cardiac malformations were observed in
32 fetuses following the oral administration of TCE to maternal animals during gestation (Dawson
33 et al., 1990, 1993; Johnson et al., 2003, 2005; see Section 4.8.2.2.1.2). Cardiac defects were also
34 observed in rats following oral gestational treatment with metabolites of TCE (Johnson et al.,
35 1998a, b; Smith et al., 1989, 1992; Epstein et al., 1992).
36
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-532 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 4-92. Summary of studies that identified cardiac malformations
associated with TCE exposures
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Finding
Cardiac defects
Altered heart rate
Species
Human
Rat
Chicken
Human
Citations
ATSDR, 2006, 2008;
Yauck et al., 2004;
Dawson et al., 1990, 1993
Johnson et al., 2003, 2005
Johnson et al., 1998a, b*
Smith etal., 1989,* 1992*
Epstein et al., 1992*
Brossetal., 1983
Boyer et al., 2000
Loeberetal., 1988
Drake et al., 2006a, b
Mishima et al., 2006
Rufer et al., 2008
Jasinka, 1965, translation
*MetabolitesofTCE.
However, cardiac malformations were not observed in a number of other studies in
laboratory animals in which TCE was administered during the period of cardiac organogenesis
and fetal visceral findings were assessed. These included inhalation studies in rats (Dorfmueller
et al., 1979; Schwetz et al., 1975; Hardin et al., 1981; Healy et al., 1982; Carney et al., 2006) and
rabbits (Hardin et al., 1981), and oral gavage studies in rats (Narotsky et al., 1995; Narotsky and
Kavlock, 1995; Fisher et al., 2001) and mice (Cosby and Dukelow, 1992).
It is generally recognized that response variability among developmental bioassays
conducted with the same chemical agent may be related to factors such as the study design (e.g.,
the species and strain of laboratory animal model used, the day(s) or time of day of dose
administration in relation to critical developmental windows, the route of exposure, the vehicle
used, the day of study termination), or the study methodologies (e.g., how fetuses were
processed, fixed, and examined; what standard procedures were used in the evaluation of
morphological landmarks or anomalies, and whether there was consistency in the fetal
evaluations that were conducted). In the case of studies that addressed cardiac malformations,
there is additional concern as to whether detailed visceral observations were conducted, whether
or not cardiac evaluation was conducted using standardized dissection procedures (e.g., with the
use of a dissection microscope or including confirmation by histopathological evaluation, and
whether the examinations were conducted by technicians who were trained and familiar with
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-533 DRAFT—DO NOT CITE OR QUOTE
-------
1 fetal cardiac anatomy). Furthermore, interpretation of the findings can be influenced by the
2 analytical approaches applied to the data as well as by biological considerations such as the
3 historical incidence data for the species and strain of interest. These issues have been critically
4 examined in the case of the TCE developmental toxicity studies (Hardin et al., 2005;
5 Watson et al., 2006).
6 In the available animal developmental studies with TCE, differences were noted in the
7 procedures used to evaluate fetal cardiac morphology following TCE gestational exposures
8 across studies, and some of these differences may have resulted in inconsistent fetal outcomes
9 and/or the inability to detect cardiac malformations. Most of the studies that did not identify
10 cardiac anomalies used a traditional free-hand sectioning technique (as described in Wilson,
11 1965) on fixed fetal specimens (Dorfmueller et al., 1979; Schwetz et al., 1975; Hardin et al.,
12 1981; Healy et al., 1982). Detection of cardiac anomalies can be enhanced through the use of a
13 fresh dissection technique as described by Staples (1974) and Stuckhardt and Poppe (1984); a
14 significant increase in treatment-related cardiac heart defects was observed by Dawson et al.
15 (1990) when this technique was used. Further refinement of this fresh dissection technique was
16 employed by Dawson and colleagues at the University of Arizona (UA), resulting in several
17 additional studies that reported cardiac malformations (Dawson et al., 1993; Johnson et al., 2003,
18 2005). However, two studies conducted in an attempt to verify the teratogenic outcomes of the
19 UA laboratory studies used the same or similar enhanced fresh dissection techniques and were
20 unable to detect cardiac anomalies (Fisher et al., 2001; Carney et al., 2001). Although the
21 Carney et al. study was administered via inhalation (a route which has not previously been
22 shown to produce positive outcomes), the Fisher et al. study was administered orally and
23 included collaboration between industry and UA scientists. It was suggested that the apparent
24 differences between the results of the Fisher et al. study and the Dawson et al. (1993) and
25 Johnson et al. studies may be related to factors such as differences in purity of test substances or
26 in the rat strains, or differences in experimental design (e.g., oral gavage versus drinking water,
27 exposure only during the period of organogenesis versus during the entire gestation period, or the
28 use of a staining procedure).
29 It is notable that all studies that identified cardiac anomalies following gestational
30 exposure to TCE or its metabolites were (1) conducted in rats and (2) dosed by an oral route of
31 exposure (gavage or drinking water). Cross-species and route-specific differences in fetal
32 response may be due in part to toxicokinetic factors. Although a strong accumulation and
33 retention of TCA was found in the amniotic fluid of pregnant mice following inhalation
34 exposures to TCE (Ghantous et al., 1986), other toxicokinetic factors may be critical. The
35 consideration of toxicokinetics in determining the relevance of murine developmental data for
36 human risk assessment is briefly discussed by Watson et al. (2006). There are differences in the
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-534 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolism of TCE between rodent and humans in that TCE is metabolized more efficiently in
2 rats and mice than humans, and a greater proportion of TCE is metabolized to DC A in rodents
3 versus to TCA in humans. Studies that examined the induction of cardiac malformations with
4 gestational exposures of rodents to various metabolites of TCE identified TCA and DCA as
5 putative cardiac teratogens. Johnson et al. (1998a, b) and Smith et al. (1989) reported increased
6 incidences of cardiac defects with gestational TCA exposures, while Smith et al. (1992) and
7 Epstein et al. (1992) reported increased incidences following DCA exposures.
8 In all studies that observed increased cardiac defects, either TCE or its metabolites were
9 administered during critical windows of in utero cardiac development, primarily during the
10 entire duration of gestation, or during the period of major organogenesis (e.g., GD 6-15 in the
11 rat). The study by Epstein et al. (1992) used dosing with DCA on discrete days of gestation and
12 had identified gestation days 9 through 12 as a particularly sensitive period for eliciting high
13 interventricular septal defects associated with exposures to TCE or its metabolites.
14 In the oral studies that identified increased incidences of cardiac malformations following
15 gestational exposure to TCE, there was a broad range of administered doses at which effects
16 were observed. In drinking water studies, Dawson et al. (1993) observed cardiac anomalies at
17 1.5 and 1,100 ppm (with no NOAEL) and Johnson et al. (2003, 2005) reported effects at 250 ppb
18 (with a NOAEL of 2.5. ppb). One concern is the lack of a clear dose-response for the incidence
19 of any specific cardiac anomaly or combination of anomalies was not identified, a disparity for
20 which no reasonable explanation for this disparity has been put forth.
21 The analysis of the incidence data for cardiac defects observed in the Johnson et al.
22 (2003, 2005) studies has been critiqued (Watson et al., 2006). Issues of concern that have been
23 raised include the statistical analyses of findings on a per-fetus (rather than the more appropriate
24 per-litter) basis (Benson, 2004), and the use of nonconcurrent control data in the analysis
25 (Hardin et al., 2004). In response, the study author has further explained procedures used
26 (Johnson, 2004) and has provided individual litter incidence data to the U.S. EPA for
27 independent statistical analysis (P. Johnson, personal communication, 2008) (see Section 5.1.2.8,
28 dose-response). In sum, while the studies by Dawson et al. (1993) and Johnson et al. (2003,
29 2005) have significant limitations, there is insufficient reason to dismiss their findings.
30
31 4.8.3.3.2.1. Mode of action for cardiac malformations. A number of in vitro studies have
32 been conducted to further characterize the potential for alterations in cardiac development that
33 have been attributed to exposures with TCE and/or its metabolites. It was noted that many of the
34 cardiac defects observed in humans and laboratory species (primarily rats and chickens) involved
35 septal and valvular structures.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-535 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
During early cardiac morphogenesis, outflow tract and atrioventricular (A-V) endothelial
cells differentiate into mesenchymal cells. These mesenchymal cells have characteristics of
smooth muscle-like myofibroblasts and form endocardial cushion tissue, which is the primordia
of septa and valves in the adult heart. Events that take place in cardiac valve formation in
mammals and birds are summarized by NRC (2006) and reproduced in Table 4-93.
Table 4-93. Events in cardiac valve formation in mammals and birds3
Stage and event
Early cardiac development
Epithelial-mesenchymal
cell transformation
Mesenchymal cell
migration and proliferation
Development of septa and
valvular structures
Structural description b
The heart is a hollow, linear, tube-like structure with two cell layers. The outer
surface is a myocardial cell layer, and the inner luminal surface is an endothial layer.
Extracellular matrix is between the two cell layers.
A subpopulation of endothelial sells lining the atrioventricular canal detaches from
adjacent cells and invades the underlying extracellular matrix.
Three events occur
> Endothelial cell activation (avian stage 14)
> Mesenchymal cell formation (avian stage 16)
> Mesenchymal cell migration into the extracellular matrix (avian stages 17 and
18).
Endothelial-derived mesenchymal cells migrate toward the surrounding myocardium
and proliferate to populate the atrioventricular (A-V) canal extracellular matrix.
Cardiac mesenchyme provides cellular constituents for
> Septum intermedium
> Valvular leaflets of the mitral and tricuspid A-V valves.
The septum intermedium subsequently contributes to
> Lower portion of the interatrial septum
> Membranous portion of the interventricular septum.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
aAs summarized in NRC (2006)
bMarkwald et al. (1984, 1996), Boyer et al. (2000).
Methods have been developed to extract the chick stage 16 atrioventricular canal from
the embryo and culture it on a hydrated collagen gel for 24-48 hours, allowing evaluation of the
described stages of cardiac development and their response to chemical treatment. Factors that
have been shown to influence the induction of endocardial cushion tissue include molecular
components such as fibronectin, laminin, and galactosyltransferase (Mjaatvedt et al., 1987;
Loeber and Runyan, 1990), components of the extracellular matrix (Mjaatvedt et al., 1991), and
smooth muscle a-actin and transforming growth factor P3 (Nakajima et al., 1997; Ramsdell and
Markwald, 1997).
Boyer et al. (2000) utilized the in vitro chick A-V canal culture system to examine the
molecular mechanism of TCE effects on cardiac morphogenesis. A-V canal explants from stage
16 chick embryos (15/treatment level) were placed onto collagen gels and treated with 0, 50,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-536 DRAFT—DO NOT CITE OR QUOTE
-------
1 100, 150, 200, or 250-ppm TCE and incubated for a total of 54 hours. Epithelial-mesenchymal
2 transformation, endothelial cell density, cell migration, and immunohistochemistry were
3 evaluated. TCE treatment was found to inhibit endothelial cell activation and normal
4 mesenchymal cell transformation, endothelial cell-cell separation, and protein marker expression
5 (i.e., transcription factor Mox-1 and extracellular matrix protein fibrillin 2). Mesenchymal cell
6 migration was not affected, nor was the expression of smooth muscle a-actin. The study authors
7 proposed that TCE may cause cardiac valvular and septal malformations by inhibiting
8 endothelial separation and early events of mesenchymal cell formation. Hoffman et al. (2004)
9 has proposed alternatively that TCE may be affecting the adhesive properties of the endocardial
10 cells. No experimental data are currently available that address the levels of TCE in cardiac
11 tissue in vivo, resulting in some questions (Dugard, 2000) regarding the relevance of these
12 mechanistic findings to human health risk assessment.
13 In a study by Mishima et al. (2006), White Leghorn chick whole embryo cultures (stage
14 13 and 14) were used to assess the susceptibility of endocardial epithelial-mesenchymal
15 transformation in the early chick heart to TCE at analytically determined concentrations of 0, 10,
16 20, 40, or 80 ppm. This methodology maintained the anatomical relationships of developing
17 tissues and organs, while exposing precisely staged embryos to quantifiable levels of TCE and
18 facilitating direct monitoring of developmental morphology. Following 24 hours of incubation
19 the numbers of mesenchymal cells in the inferior and superior AV cushions were counted. TCE
20 treatment significantly reduced the number of mesenchymal cells in both the superior and
21 inferior AV cushions at 80 ppm.
22 Ou et al. (2003) examined the possible role of endothelial nitric oxide synthase (which
23 generates nitric oxide that has an important role in normal endothelial cell proliferation and
24 hence normal blood vessel growth and development) in TCE-mediated toxicity. Cultured
25 proliferating bovine coronary endothelial cells were treated with TCE at 0-100 uM and
26 stimulated with a calcium ionophore to determined changes in endothelial cells and the
27 generation of endothelial nitric oxide synthase, nitric oxide, and superoxide anion. TCE was
28 shown to alter heat shock protein interactions with endothelial nitric oxide synthase and induce
29 endothelial nitric oxide synthase to shift nitric oxide to superoxide-anion generation. These
30 findings provide insight into how TCE impairs endothelial proliferation.
31 Several studies have also identified a TCE-related perturbation of several proteins
32 involved in regulation of intracellular Ca2+. After 12 days of maternal exposure to TCE in
33 drinking water, Serca2a (sarcoendoplasmic reticulum Ca2+ ATPase) mRNA expression was
34 reduced in rat embryo cardiac tissues (Collier et al., 2003). Selmin et al. (2008) conducted a
35 microarray analysis of a P19 mouse stem cell line exposed to 1-ppm TCE in vitro, identifying
36 altered expression of Ryr (ryanodine receptor isoform 2). Caldwell et al. (2008) used real-time
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-537 DRAFT—DO NOT CITE OR QUOTE
-------
1 PCR and digital imaging microscopy to characterize the effects of various doses of TCE on gene
2 expression and Ca2+ response to vasopressin in rat cardiac myocytes (H9c2) Serca2a and Ryr2
3 expression were reduced at 12 and 48 hours following exposure to TCE. Additionally, Ca2+
4 response to vasopressin was altered following TCE treatment. Overall, these data suggest that
5 TCE may disrupt the ability to regulate cellular Ca2+ fluxes, leading to morphogenic
6 consequences in the developing heart. This remains an open area of research.
7 Thus, in summary, a number of studies have been conducted in an attempt to characterize
8 the MOA for TCE-induced cardiac defects. A major research focus has been on disruptions in
9 cardiac valve formation, using avian in ovo and in vitro studies. These studies demonstrated
10 treatment-related alterations in endothelial cushion development that could plausibly be
11 associated with defects involving septal and valvular morphogenesis in rodents and chickens.
12 However, a broad array of cardiac malformations has been observed in animal models following
13 TCE exposures (Dawson et al., 1993; Johnson et al., 2003, 2005), and other evidence of
14 molecular disruption of Ca2+ during cardiac development has been examined (Caldwell et al.,
15 2008; Collier et al., 2003; Selmin et al., 2008) suggesting the possible existence of multiple
16 MO As.
17
18 4.8.3.3.2.2. Association of peroxisome proliferator activated receptor alpha (PPAR) with
19 developmental outcomes. The PPARs are ligand activated receptors that belong to the nuclear
20 hormone receptor family. Three isotypes have been identified (PPARa, PPAR5 [also known as
21 PPARP], and PPARy). These receptors, upon binding to an activator, stimulate the expression of
22 target genes implicated in important metabolic pathways. In rodents, all three isotypes show
23 specific time and tissue-dependent patterns of expression during fetal development and in adult
24 animals. In development, they have been especially implicated in several aspects of tissue
25 differentiation, e.g., of the adipose tissue, brain, placenta and skin. Epidermal differentiation has
26 been linked strongly with PPARa and PPAR6 (Michalik et al., 2002). PPARa starts late in
27 development, with increasing levels in organs such as liver, kidney, intestine, and pancreas; it is
28 also transiently expressed in fetal epidermis and CNS (Braissant and Wahli, 1998) and has been
29 linked to phthalate-induced developmental and testicular toxicity (Gorton and Lapinskas, 2005).
30 Liver, kidney, and heart are the sites of highest PPARa expression (Toth et al., 2007). PPAR5
31 and PPARy have been linked to placental development and function, with PPARy found to be
32 crucial for vascularization of the chorioallantoic placenta in rodents (Wendling et al., 1999), and
33 placental anomalies mediated by PPARy have been linked to rodent cardiac defects (Barak et al.,
34 2008). While it might be hypothesized that there is some correlation between PPAR signaling,
35 fetal deaths, and/or cardiac defects observed following TCE exposures in rodents, no definitive
36 data have been generated that elucidate a possible PPAR-mediated MOA for these outcomes.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-538 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.8.3.3.2.3. Summary of the weight of evidence on cardiac malformations. The evidence for
2 an association between TCE exposures in the human population and the occurrence of congenital
3 cardiac defects is not particularly strong. Many of the epidemiological study designs were not
4 sufficiently robust to detect exposure-related birth defects with a high degree of confidence.
5 However, two well-conducted studies by ATSDR (2006, 2008) clearly demonstrated an
6 elevation in cardiac defects. It could be surmised that the identified cardiac defects were
7 detected because they were severe, and that additional cases with less severe cardiac anomalies
8 may have gone undetected.
9 The animal data provide strong, but not unequivocal, evidence of the potential for TCE-
10 induced cardiac malformations following oral exposures during gestation. Strengths of the
11 evidence are the duplication of the adverse response in several studies from the same laboratory
12 group, detection of treatment-related cardiac defects in both mammalian and avian species (i.e.,
13 rat and chicken), general cross-study consistency in the positive association of increased cardiac
14 malformations with test species (i.e., rat), route of administration (i.e., oral), and the
15 methodologies used in cardiac morphological evaluation (i.e., fresh dissection of fetal hearts).
16 Furthermore, when differences in response are observed across studies they can generally be
17 attributed to obvious methodological differences, and a number of in ovo and in vitro studies
18 demonstrate a consistent and biologically plausible MOA for one type of malformation observed.
19 Weaknesses in the evidence include lack of a clear dose-related response in the incidence of
20 cardiac defects, and the broad variety of cardiac defects observed, such that they cannot all be
21 grouped easily by type or etiology.
22 Taken together, the epidemiological and animal study evidence raise sufficient concern
23 regarding the potential for developmental toxicity (increased incidence of cardiac defects) with
24 in utero TCE exposures.
25 4.8.3.3.3. Other structural developmental outcomes. A summary of other structural
26 developmental outcomes that have been associated with TCE exposures is presented in
27 Table 4-94.
28 In humans, a variety of birth defects other than cardiac have been observed. These
29 include total birth defects (Bove, 1996; Bove et al., 1995; AZ DHS, 1988; ATSDR, 2001), CNS
30 birth defects (ATSDR, 2001; Bove, 1996; Bove et al., 1995; Lagakos et al., 1986), eye/ear birth
31 anomalies (Lagakos et al., 1986); oral cleft defects (Bove, 1996; Bove et al., 1995; Lagakos et
32 al., 1986; Lorente et al., 2000); kidney/urinary tract disorders (Lagakos et al., 1986);
33 musculoskeletal birth anomalies (Lagakos et al., 1986); anemia/blood disorders (Burg and Gist,
34 1999); and lung/respiratory tract disorders (Lagakos et al., 1986). While some of these results
35 were statistically significant, they have not been reported elsewhere. Occupational cohort
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-539 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
studies, while not reporting positive results, are generally limited by the small number of
observed or expected cases of birth defects (Lorente et al., 2000; Tola et al., 1980; Taskinen et
al., 1989).
Table 4-94. Summary of other structural developmental outcomes associated
with TCE exposures
Finding
Eye/ear birth anomalies
Oral cleft defects
Kidney/urinary tract
disorders
Musculoskeletal birth
anomalies
Anemia/blood disorders
Lung/respiratory tract
disorders
Skeletal
Other*
Species
Human
Rat
Human
Human
Human
Human
Human
Mouse
Rat
Human
Citations
Lagakos etal., 1986
Narotsky, 1995
Narotsky and Kavlock, 1995
Bove, 1996
Boveetal., 1995
Lagakos et al., 1986
Lorente et al., 2000
Lagakos et al., 1986
Lagakos etal., 1986
Burg and Gist, 1999
Lagakos et al., 1986
Das and Scott, 1994
Healyetal., 1982
ATSDR, 2001
* As reported by the authors.
In experimental animals, a statistically significant increase in the incidence of fetal eye
defects, primarily micropththalmia and anopththalmia, manifested as reduced or absent eye
bulge, was observed in rats following gavage administration of 1,125 mg/kg/d TCE during the
period of organogenesis (Narotsky et al., 1995; Narotsky and Kavlock, 1995). Dose-related
nonsignificant increases in the incidence of Fischer 344 rat pups with eye defects were also
observed at lower dose levels (101, 320, 475, 633, and 844 mg/kg/d) in the Narotsky et al. (1995)
study (also reported in Barton and Das [1996]). However, no other developmental or
reproductive toxicity studies identified abnormalities of eye development following TCE
exposures. For example, in a study reported by Warren et al. (2006), extensive computerized
morphometric ocular evaluation was conducted in Sprague-Dawley rat fetuses that had been
examined for cardiac defects by Fisher et al. (2001); the dams had been administered TCE
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-540 DRAFT—DO NOT CITE OR QUOTE
-------
1 (500 mg/kg/d), DCA (300 mg/kg/d), or TCA (300 mg/kg/d) during gestation days 6-15. No
2 ocular defects were found with TCE exposures; however, significant reductions in the lens area,
3 globe area, and interocular distance were observed with DCA exposures, and nonsignificant
4 decreases in these measures as well as the medial canthus distance were noted with TCA
5 exposures. Developmental toxicity studies conducted by Smith et al. (1989, 1992) also identified
6 orbital defects (combined soft tissue and skeletal abnormalities) in Long Evans rat fetuses
7 following GD 6-15 exposures with TCA and DCA (statistically or biologically significant at
8 >800 mg/kg/d and >900 mg/kg/d, respectively). Overall, the study evidence indicates that TCE
9 and its oxidative metabolites can disrupt ocular development in rats. In addition to the evidence
10 of alteration to the normal development of ocular structure, these findings may also be an
11 indicator of disruptions to nervous system development. It has been suggested by Warren et al.
12 (2006) and Williams and DeSesso (2008) that the effects of concern (defined as statistically
13 significant outcomes) are observed only at high dose levels and are not relevant to risk
14 assessment for environmental exposures. On the other hand, Barton and Das (1996) point out
15 that benchmark dose modeling of the quantal eye defect incidence data provides a reasonable
16 approach to the development of oral toxicity values for TCE human health risk assessment. It is
17 also noted that concerns may exist not only for risks related to low level environmental
18 exposures, but also for risks resulting from acute or short-term occupational or accidental
19 exposures, which may be associated with much higher inadvertent doses.
20 It was also notable that a study using a single intraperitoneal dose of 3,000 mg/kg TCE to
21 mice during late gestation (GD 17) identified apparent delays in lung development and increased
22 neonatal mortality (Das and Scott, 1994). No further evaluation of this outcome has been
23 identified in the literature.
24 Healy et al. (1982) did not identify any treatment-related fetal malformations following
25 inhalation exposure of pregnant inbred Wistar rats to 0 or 100 ppm (535 mg/m3) on GD 8-21. In
26 this study, significant differences between control and treated litters were observed as an
27 increased incidence of minor ossification variations (p = 0.003) (absent or bipartite centers of
28 ossification).
29
30 4.8.3.3.4. Developmental neurotoxicity. Studies that address effects of TCE on the developing
31 nervous system are discussed in detail in Section 4.3, addressed above in the sections on human
32 developmental toxicity (Section 4.8.3) and on mammalian studies (Section 4.8.3.2.1) by route of
33 exposure, and summarized in Table 4-95. The available data collectively suggest that the
34 developing brain is susceptible to TCE exposures.
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-541 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-95. Summary of developmental neurotoxicity associated with TCE
exposures
2
3
4
5
6
7
Positive findings
CNS defects, neural tube defects
Eye defects
Delayed newborn reflexes
Impaired learning or memory
Aggressive behavior
Hearing impairment
Speech impairment
Encephalopathy
Impaired executive function
Impaired motor function
Attention deficit
ASD
Delayed or altered biomarkers of
CNS development
Behavioral alterations
Species
Human
Rat
Human
Human
Human
Rat
Human
Human
Human
Human
Human
Human
Human
Rat
Mice
Rat
Citations
ATSDR, 2001
Bove, 1996; Bove et al., 1995
Lagakos et al., 1986
Narotsky, 1995;
Narotsky and Kavlock, 1995
Beppu, 1968
Bernad et al., 1987, abstract
White etal., 1997
Bernad et al., 1987, abstract
Blossom et al., 2008
ATSDR, 2003a;
Burg etal., 1995;
Burg and Gist, 1999
Beppu, 1968
ATSDR, 2003a;
Burg etal., 1995;
Burg and Gist, 1999
White etal., 1997
White etal., 1997
White etal., 1997
White etal., 1997
Bernad et al., 1987, abstract
Windham et al., 2006
Isaacson and Taylor, 1989
Noland-Gerbec et al., 1986
Westergren et al., 1984
Blossom et al., 2008
Fredriksson et al., 1993
George etal., 1986
Taylor etal., 1985
In humans, CNS birth defects were observed in a few studies (ATSDR, 2001; Bove,
1996; Bove et al., 1995; Lagakos et al., 1986). Postnatally, observed adverse effects in humans
include delayed newborn reflexes following use of TCE during childbirth (Beppu, 1968),
impaired learning or memory (Bernad et al., 1987, abstract; White et al., 1997); aggressive
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-542 DRAFT—DO NOT CITE OR QUOTE
-------
1 behavior (Bernad et al., 1987, abstract); hearing impairment (Beppu, 1968; Burg et al., 1995;
2 Burg and Gist, 1999; ATSDR, 2003a); speech impairment (Berg et al., 1995; Burg and Gist,
3 1999; White et al., 1997); encephalopathy (White et al., 1997); impaired executive and motor
4 function (White et al., 1997); attention deficit (Bernad et al., 1987, abstract; White et al., 1997),
5 and autism spectrum disorder (Windham et al., 2006). While there are broad developmental
6 neurotoxic effects that have been associated with TCE exposure, there are many limitations in
7 the studies.
8 More compelling evidence for the adverse effect of TCE exposure on the developing
9 nervous system is found in the animal study data, although a rigorous evaluation of potential
10 outcomes has not been conducted. For example, there has not been an assessment of cognitive
11 function (i.e., learning and memory) following developmental exposures to TCE, nor have most
12 of the available studies characterized the pre- or postnatal exposure of the offspring to TCE or its
13 metabolites. Nevertheless, there is evidence of treatment-related alterations in brain
14 development and in behavioral parameters (e.g., spontaneous motor activity and social
15 behaviors) associated with exposures during neurological development. The animal study
16 database includes the following information: Following inhalation exposures of 150 ppm to mice
17 during mating and gestation, the specific gravity of offspring brains were significantly decreased
18 at postnatal time points through the age of weaning; however, this effect did not persist to
19 1 month of age (Westergren et al., 1984). In studies reported by Taylor et al. (1985), Isaacson
20 and Taylor (1989), and Noland-Gerbec et al. (1986), 312 mg/L exposures in drinking water that
21 were initiated 2 weeks prior to mating and continued to the end of lactation resulted,
22 respectively, in (a) significant increases in exploratory behavior at postnatal days 60 and 90, (b)
23 reductions in myelination in the brains of offspring at weaning, and (c) significantly decreased
24 uptake of 2-deoxyglucose in the neonatal rat brain (suggesting decreased neuronal activity).
25 Ocular malformations in rats observed by Narotsky (1995) and Narotsky and Kavlock (1995)
26 following maternal gavage doses of 1,125 mg/kg/d during gestation may also be indicative of
27 alterations of nervous system development. Gestational exposures to mice (Fredriksson et al.,
28 1993) resulted in significantly decreased rearing activity on postnatal Day 60, and dietary
29 exposures during the course of a continuous breeding study in rats (George et al., 1986) found a
30 significant trend toward increased time to cross the first grid in open field testing. In a study by
31 Blossom et al. (2008), alterations in social behaviors (deficits in nest-building quality and
32 increased aggression in males) were observed in pubertal-age MRL +/+ mice that had been
33 exposed to 0.1 mg/mL TCE via drinking water during prenatal and postnatal development (until
34 PND 42). Dorfmueller et al. (1979) was the only study that assessed neurobehavioral endpoints
35 following in utero exposure (maternal inhalation exposures of 1,800 ± 200 ppm during gestation)
36 and found no adverse effects that could be attributed to TCE exposure. Specifically, an
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-543 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
automated assessment of ambulatory response in a novel environment on postnatal days 10, 20
and 100, did not identify any effect on general motor activity of offspring.
4.8.3.3.5. Developmental immunotoxicity. Studies that address the developmental
immunotoxic effects of TCE are discussed in detail in Section 4.6, addressed above in the
sections on human developmental toxicity (Section 4.8.3) and on mammalian studies
(Section 4.8.3.2.1) by route of exposure, and summarized in Table 4-96.
Table 4-96. Summary of developmental immunotoxicity associated with
TCE exposures
10
Finding
Significant reduction in Thl IL-2 producing cells
Altered immune response
Suppression of PFC responses, increased T-cell
subpopulations, decreased spleen cellularity, and
increased hypersensitivity response
Altered splenic and thymic T-cell subpopulations
Altered thymic T-cell subpopulations; transient
increased proinflammatory cytokine production by
T-cells; increased autoantibody levels and IgG
Increased proinflammatory cytokine production by
T-cells
Species (strain)
Human
Human
Mouse (B6C3F1)
Mouse (MRL +/+)
Mouse (MRL +/+)
Mouse (MRL +/+)
Citations
Lehmann et al., 2002
Byersetal., 1988
Peden- Adams et al.,
2006
Peden-Adams et al.,
2008
Blossom and Doss,
2007
Blossom et al., 2008
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Two epidemiological studies that addressed potential immunological perturbations in
children that were exposed to TCE were reported by Lehmann et al. (2001, 2002). In the 2001
study, no association was observed between TCE and allergic sensitization to egg white and
milk, or to cytokine producing peripheral T-cells, in premature neonates and 36-month-old
neonates that were at risk of atopy. In the 2002 study, there was a significant reduction in Thl
IL-2 producing cells. Another study observed altered immune response in family members of
those diagnosed with childhood leukemia, including 13 siblings under age 19 at the time of
exposure, but an analysis looking at only these children was not done (Byers et al., 1988).
Several studies were identified (Peden-Adams et al., 2006, 2008; Blossom and Doss,
2007; Blossom et al., 2008) which assessed the potential for developmental immunotoxicity in
mice following oral (drinking water) TCE exposures during critical pre- and postnatal stages of
immune system development. Peden-Adams et al. (2006) noted evidence of immune system
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-544 DRAFT—DO NOT CITE OR QUOTE
-------
1 perturbation (suppression of PFC responses, increased T-cell subpopulations, decreased spleen
2 cellularity, and increased hypersensitivity response) in B6C3F1 offspring following in utero and
3 8 weeks of postnatal exposures to TCE. Evidence of autoimmune response was not observed in
4 the offspring of this nonautoimmune-prone strain of mice. However, in a study by Peden-Adams
5 et al. (2008) MRL +/+ mice, which are autoimmune-prone, were exposed from conception until
6 12 months of age. Consistent with the Peden-Adams et al. (2006) study, no evidence of
7 increased autoantibody levels was observed in the offspring. In two other studies focused on
8 autoimmune responses following drinking water exposures of MRL +/+ mice to TCE during in
9 utero development and continuing until the time of sexual maturation, Blossom and Doss (2007)
10 and Blossom et al. (2008) reported some peripheral blood changes that were indicative of
11 treatment-related autoimmune responses in offspring. Positive response levels were 0.5 and
12 2.5 mg/mL for Blossom and Doss (2007) and 0.1 mg/mL for Blossom et al. (2008). None of
13 these studies were designed to extensively evaluate recovery, latent outcomes, or differences in
14 severity of response that might be attributed to the early life exposures. Consistency in response
15 in these animal studies was difficult to ascertain due to the variations in study design (e.g.,
16 animal strain used, duration of exposure, treatment levels evaluated, timing of assessments, and
17 endpoints evaluated). Likewise, the endpoints assessed in the few epidemiological studies that
18 evaluated immunological outcomes following developmental exposures to TCE were dissimilar
19 from those evaluated in the animal models, and so provided no clear cross-species correlation.
20 The most sensitive immune system response noted in the studies that exposed developing
21 animals were the decreased PFC and increased hypersensitivity observed by Peden-Adams et al.
22 (2006); treatment-related outcomes were noted in mice exposed in the drinking water at a
23 concentration of 1,400 ppb. None of the other studies that treated mice during immune system
24 development assessed these same endpoints; therefore, direct confirmation of these findings
25 across studies was not possible. It is noted, however, that similar responses were not observed in
26 studies in which adult animals were administered TCE (e.g., Woolhiser et al., 2006), suggesting
27 increased susceptibility in the young. Differential lifestage-related responses have been observed
28 with other diverse chemicals (e.g., diethylstilbestrol; diazepam; lead; 2,3,7,8-tetrachlorobenzo-
29 p dioxin; and tributyltin oxide) in which immune system perturbations were observed at lower
30 doses and/or with greater persistence when tested in developing animals as compared to adults
31 (Luebke et al., 2006). Thus, such an adverse response with TCE exposure is considered
32 biologically plausible and an issue of concern for human health risk assessment.
33
34 4.8.3.3.6. Childhood cancers. A summary of childhood cancers that have been associated with
35 TCE exposures discussed above is presented in Table 4-97. A summary of studies that observed
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-545 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
childhood leukemia is also discussed in detail in Section 4.6.1.3 and Section 4.8.3.1.2.4 contains
details of epidemiologic studies on childhood brain cancer.
Table 4-97. Summary of childhood cancers associated with TCE exposures
Finding
Leukemia
Neuroblastoma
Species
Human
Human
Citations
AZDHS, 1988, 1990a
AZDHS, 1990c
Cohnetal., 1994
Cutler et al., 1986; Costas et al., 2002; Lagakos et al., 1986;
MADPH, 1997
Lowengart et al., 1987
McKinney et al., 1991
Shuetal., 1999
DeRoosetal., 2001
Peters etal., 1981, 1985
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
A nonsignificant increased risk of leukemia diagnosed during childhood has been
observed in a number of studies examining TCE exposure (AZ DHS, 1998, 1990a, c; Cohn et al.,
1994; Costas et al., 2002; Lagakos et al., 1986; Lowengart et al., 1987; MA DPH, 1997;
McKinney et al., 1991; Shu et al., 1999). However, other studies did not observed an increased
risk for childhood leukemia after TCE exposure (AZ DHS, 1990b, 1997; Morgan and Cassady,
2002), possibly due to the limited number of cases or the analysis based on multiple solvents.
CNS cancers during childhood have been reported on in a few studies. Neuroblastomas were not
statistically elevated in one study observing parental exposure to multiple chemicals, including
TCE (De Roos et al., 2001). Brain tumors were observed in another study, but the odds ratio
could not be determined (Peters et al., 1981, 1985). CNS cancers were not elevated in other
studies (AZ DHS, 1990c; Morgan and Cassady, 2002). Other studies did not see an excess risk
of total childhood cancers (ATSDR, 2006; Morgan and Cassady, 2002).
A follow-up study of the Camp Lejeune cohort that will examine childhood cancers
(along with birth defects) was initiated in 1999 (ATSDR, 2003b), is expected to be completed
soon (GAO, 2007a, b; ATSDR, 2009), and may provide additional insight.
No studies of cancers in experimental animals in early lifestages have been identified.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-546 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.9. OTHER SITE-SPECIFIC CANCERS
2 4.9.1. Esophageal Cancer
3 Increasing esophageal cancer incidence has been observed in males, but not females in
4 the United States between 1975 and 2002, a result of increasing incidence of esophageal
5 adenocarcinoma (Ward et al., 2006). Males also have higher age-adjusted incidence and
6 mortality rates (incidence, 7.8 per 100,000; mortality, 7.8 per 100,000) than females (incidence,
7 2.0 per 100,000; mortality, 1.7 per 100,000) (Ries et al., 2008). Survival for esophageal cancer
8 remains poor and age-adjusted mortality rates are just slightly lower than incidence rates. Major
9 risk factors associated with esophageal cancer are smoking and alcohol for squamous cell
10 carcinoma, typically found in the upper third of the esophagus, and obesity, gastroesophageal
11 reflux, and Barrett's esophagus for adenocarcinoma that generally occurs in the lower esophagus
12 (Ward et al., 2006).
13 Seventeen epidemiologic studies on TCE exposure reported relative risks for esophageal
14 cancer (Garabrant et al., 1988; Blair et al., 1989; Costa et al., 1989; Siemiatycki, 1991;
15 Greenland et al., 1994; Blair et al., 1998; Boice et al., 1999, 2006; Ritz, 1999; Hansen et al.,
16 2001; Raaschou-Nielsen et al., 2003; ATSDR, 2004, 2006; Zhao et al., 2005; Sung et al., 2007;
17 Clapp and Hoffman, 2008; Radican et al., 2008). Ten studies had high likelihood of TCE
18 exposure in individual study subjects and were judged to have met, to a sufficient degree, the
19 standards of epidemiologic design and analysis (Siemiatycki, 1991; Greenland et al., 1994; Blair
20 et al., 1998; Boice et al., 1999, 2006; Ritz, 1999; Hansen et al., 2001; Raaschou-Nielsen et al.,
21 2003; Zhao et al., 2005; Radican et al., 2008). Four studies with high quality information
22 (Axelson et al., 1994; Anttila et al., 1995; Blair et al., 1998 [Incidence]; Morgan et al., 1998) do
23 not present relative risk estimates for esophageal cancer and TCE exposure nor do two other
24 studies which carry less weight in the analysis because of design limitations (Sinks et al., 1992;
25 Henschler et al., 1995). Only Raaschou-Nielsen et al. (2003) examines esophageal cancer
26 histologic type, an important consideration given differences between suspected risk factors for
27 adenocarcinoma and those for squamous cell carcinoma. Appendix B identifies these study's
28 design and exposure assessment characteristics.
29 Several population case-control studies (Yu et al., 1988; Gustavsson et al., 1998; Parent
30 et al., 2000; Weiderpass et al., 2003; Engel et al., 2002; Ramanakumar et al., 2008; Santibafiez et
31 al., 2008) examine esophageal cancer and organic solvents or occupational job titles with past
32 TCE use documented (Bakke et al., 2006). Relative risk estimates in case-control studies that
33 examine metal occupations or job titles, or solvent exposures are found in Table 4-98. The lack
34 of exposure assessment to TCE, low prevalence of exposure to chlorinated hydrocarbon solvents,
35 or few exposed cases and controls in those studies lowers their sensitivity for informing
36 evaluations of TCE and esophageal cancer.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-547 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-98. Selected observations from case-control studies of TCE exposure and esophageal cancer
to
VO Lo'
I
I
§
***.
£3'
1
TO'
Study
population
Exposure group
All esophageal cancers
Relative risk
(95% CY)
No. obs.
events
Squamous cell cancer
Relative risk
(95% CI)
No. obs.
events
Adenocarcinoma
Relative risk
(95% CI)
No. obs.
events
Population of regions in Eastern Spain
Metal molders, welders, etc.
Metal-processing plant
operators
0.94(0.14,6.16)
1.14(0.29,4.44)
o
J
5
0.40(0.05,3.18)
1.23(0.23,6.51)
2
4
3.55 (0.28, 44.70)
0.86 (0.08, 8.63)
1
1
Chlorinated hydrocarbon solvents
Low exposure
High exposure
1.05(0.15,7.17)
1.76 (0.40, 7.74)
2
6
2.18(0.41, 11.57)
0
5
4.92 (0.69, 34.66)
3.03(0.28,32.15)
2
1
Population of Montreal, Canada
Painter, Metal coatings
Any exposure
Substantial exposure
1.3(0.4,4.2)
4.2(1.1, 17.0)
6
4
Solvents
Any exposure
Nonsubstantial exposure
Substantial exposure
1.1 (0.7, 1.7)
1.0(0.5, 1.9)
1.1(0.6, 1.9)
39
16
39
1.4 (0.8, 2.5)
1.3 (0.6, 2.6)
1.4 (0.8, 2.5)
30
12
30
Population of Sweden
Organic solvents
No exposure
Moderate exposure
High exposure
Test for trend
No exposure
Moderate exposure
High exposure
Test for trend
1.0
0.7 (0.4, 1.5)
1.3(0.7,2.3)
p = 0.47
1.0
0.5(0.1,3.9)*
0.4(0.1, 1.8)*
p = 0.44
145
15
21
1
2
1.0
1.2(0.6,2.3)
1.4 (0.7, 2.5)
^ = 0.59
1.0
0.4(0.1, 1.5)*
0.9 (0.5, 1.6)*
^ = 0.36
128
14
18
2
12
Reference
Santibanez et al., 2008
Ramanakumar et al.,
2008; Parent et al., 2000
Janssen et al., 2006a, b
00
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
2
•••2
I
TO
Co
Table 4-98. Selected observations from case-control studies of TCE exposure and esophageal cancer (continued)
Study
population
Exposure group
All esophageal cancers
Relative risk
(95% CY)
No. obs.
events
Squamous cell cancer
Relative risk
(95% CT)
No. obs.
events
Adenocarcinoma
Relative risk
(95% CI)
No. obs.
events
Population of Finland (Females)
Chlorinated hydrocarbon solvents
Low level exposure
High level exposure
0.95 (0.54, 1.66)
0.62(0.34, 1.13)
Not
reported
Not
reported
Population of NJ, CT, WA State
Precision metal workers
Metal product manufacturing
Not reported
Not reported
0.7(0.3, 1.5)
0.8(0.3, 1.8)
12
15
1.4 (0.8, 2.3)
1.3(0.8,2.3)
25
26
Reference
Weiderpassetal.,2003
Engeletal.,2002
H I
* Jansson et al. (2006b) is a registry-based study of the Swedish Construction Worker Cohort. Relative risks are incidence rate ratios from Cox regression
analysis using calendar time and adjustment for attained age, calendar period at entry into the cohort, tobacco smoking status at entry into the cohort and
at entry into the cohort.
O
H
W
-------
1 Table 4-99 presents risk estimates for TCE exposure and esophageal cancer observed in
2 cohort, PMR, case-control, and geographic based studies. Ten studies in which there is a high
3 likelihood of TCE exposure in individual study subjects (e.g., based on job-exposure matrices or
4 biomarker monitoring) reported risk estimates for esophageal cancer (Siemtiatycki, 1991;
5 Greenland et al., 1994; Blair et al., 1998; Boice et al., 1999; Ritz et al., 1999; Hansen et al.,
6 2001; Raaschou-Nielsen et al., 2003; Zhao et al., 2005; Boice et al., 2006; Radican et al., 2008).
7 Some evidence for association with esophageal cancer and overall TCE exposure comes from
8 studies with high likelihood of TCE exposure (5.6, 95% CI: 0.7, 44.5 [Blair et al., 1998] and
9 1.88, 95% CI: 0.61, 5.79 [Radican et al., 2008, which was an update of Blair et al., 1998 with an
10 additional 10 years of follow-up]; 4.2, 95% CI: 1.5, 9.2, [Hansen et al., 2001]; 1.2, 95% CI: 0.84,
11 1.57 [Raaschou-Nielsen et al., 2003]). Two studies support an association with adenocarcinoma
12 histologic type of esophageal cancer and TCE exposure (five of the six observed esophageal
13 cancers were adenocarcinomas [less than 1 expected; Hansen et al., 2001]); 1.8, 95% CI: 1.2, 2.7
14 (Raaschou-Nielsen et al., 2003). Risk estimates in other high-quality studies are based on few
15 deaths, low statistical power to detect a doubling of esophageal cancer risk, and confidence
16 intervals which include a risk estimate of 1.0 (no increased risk).
17 Seven other studies (Garabrant et al., 1988; Blair et al., 1989; Costa et al., 1989; Sung et
18 al., 2007; ATSDR, 2004, 2006; Clapp and Hoffman, 2008) with lower likelihood for TCE
19 exposure, in addition to limited statistical power and other design limitations, observed relative
20 risk estimates between 0.21 (95% CI: 0.0.01, 1.17) (Costa et al., 1989) to 1.14 (95% CI: 0.62,
21 1.92) (Garabrant et al., 1988). For these reasons, esophageal cancer observations in these studies
22 are not inconsistent with Blair et al. (1998) and its update Radican et al. (2008), Hansen et al.,
23 (2001), or Raaschou-Nielsen et al. (2003). No study reported a statistically significant deficit in
24 the esophageal cancer risk estimate and overall of TCE exposure. Of those studies with
25 exposure-response analyses, a pattern of increasing esophageal cancer relative risk with
26 increasing exposure metric is not generally noted (Siemiatycki, 1991; Blair et al., 1998; Boice et
27 al., 1999; Zhao et al., 2005; Radican et al., 2008) except for Hansen et al. (2001) and Raaschou-
28 Nielsen et al. (2003). In these last two studies, esophageal cancer relative risk estimates
29 associated with long employment duration were slightly higher (SIR: 6.6, 95% CI: 1.8, 7.0.8, 3.7
30 [Hansen et al., 2001]; SIR: 1.9, 95% CO: 0.8, 3.7 [Raaschou-Nielsen et al., 2003]) than those for
31 short employment duration (SIR: 4.4, 95% CI: 0.5, 19 [Hansen et al., 2001]; SIR: 1.7, 95% CI:
32 0.6, 3.6 [Raaschou-Nielsen et al., 2003]). Hansen et al. (2001) also reports risk for two other
33 TCE exposure surrogates, average intensity and cumulative exposure, and in both cases observed
34 lower risk estimates with the higher exposure surrogate.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-550 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-99. Summary of human studies on TCE exposure and esophageal
cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Med cumulative TCE score
High TCE score
p for trend
Not reported
1.00a
1.66 (0.62, 4.41)b
0.82(0.17, 3.95)b
p = 0.974
9
8
2
All employees at electronics factory (Taiwan)
Males
Females
Not reported
1.16(0.0.14, 4.20)c
2
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Any exposure, males
Any exposure, females
Any exposure, males
Any exposure, females
1.2 (0.84, 1.57)
1.1(0.81, 1.53)
2.0(0.54,5.16)
1.8(1. 15, 2.73)d
44
40
4
23
0 (0.4 exp)d
Exposure lag time
20yrs
1.7 (0.8, 3.0)d
10
Employment duration
5yrs
1.7 (0.6, 3.6)d
1.9 (0.9, 3.6)d
1.9 (0.8, 3.7)d
6
9
8
Subcohort with higher exposure
Any TCE exposure
Employment duration
1-4.9 yrs
>5yrs
Biologically -monitored Danish workers
Any TCE exposure, males
Adenocarcinoma histologic type
Any TCE exposure, females
1.7 (0.9, 2.9)d
1.6 (0.6, 3.4)d
1.9 (0.8, 3.8)d
4.0 1.5, 8.72)
4.2(1.5,9.2)
3.6(1.2, 8.3)e
13
6
7
6
6
5
O(O.lexp)
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-vr
6.5(1.3, 19)
4.2(1.5,9.2)
3
3
Mean concentration (Ikeda)
<4ppm
4+ppm
8.0 (2.6, 19)
1.3 (0.02, 7.0)
5
1
Zhao et al., 2005
Sung et al., 2007
Raaschou-Nielsen et al.,
2003
Hansenetal.,2001
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-551 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-99. Summary of human studies on TCE exposure and esophageal
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Employment duration
<6.25 yr
>6.25 yr
4.4 (0.5, 16)
6.6(1.8, 17)
2
4
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Not reported
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
Not reported
Not reported
Not reported
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
Not reported
Not reported
Not reported
Biologically -monitored Finnish workers
All subjects
Not reported
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
Not reported
Not reported
Cardboard manufacturing workers in Arnsburg, Germany
Exposed workers
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
Not reported
Not reported
Not reported
Cardboard manufacturing workers, Atlanta area, GA
All subjects
Not reported
Reference
Blair etal., 1998
Anttila et al., 1995
Henschler et al., 1995
Axelsonetal., 1994
Sinks etal., 1992
Cohort and PMR studies-mortality
Computer manufacturing workers (IBM), NY
Males
1.12(0.30, 2.86)f
5.24 (0.13, 29.2)f
Clapp and Hoffman, 2008
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-552 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-99. Summary of human studies on TCE exposure and esophageal
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Aerospace workers (Rocketdyne)
Any TCE (utility/eng flush)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
0.88(0.18,2.58)
Not reported
1.00a
1.40 (0.70, 2.82)b
1.27(0.52, 3. 13)b
;? = 0.535
o
J
18
15
7
View-Master employees
Males
Females
0.62 (0.02, 3.45)f
1
0 (1.45 exp)f
All employees at electronics factory (Taiwan)
Males
Females
0(3.34exp)
0 (0.83 exp)
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration
Moderate TCE exposure, >2 yrs
duration
Not reported
2.61 (0.99, 6.88)g
12
0
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent3
Duration of exposure
Oyrs
5 vrs
p for trend
0.83 (0.34, 1.72)
Not presented
1.0a
0.23 (0.05, 0.99)
0.57 (0.20, 1.67)
0.91(0.38,2.22)
^>0.20
7
11
28
2
4
7
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
TCE subcohort (Cox Analysis)
Never exposed
Ever exposed
Peak
No/Low
Medium/high
Not reported
Not reported
Not reported
Reference
Boice et al., 2006
Zhao et al., 2005
ATSDR, 2004
Chang etal., 2003
Ritz, 1999
Boice et al., 1999
Morgan etal., 1998
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-553 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-99. Summary of human studies on TCE exposure and esophageal
cancer (continued)
Exposure group
Cumulative
Referent
Low
High
Relative risk
(95% CI)
Not reported
No. obs.
events
Aircraft maintenance workers (Hill AFB, UT)
TCE subcohort
5.6(0.7, 44.5) a
10
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
Not reported11
Not reported11
Not reported11
3
2
4
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
3.6 (0.2, 58)
1.88 (0.61, 5.79)
1.66 (0.48, 5.74)
1.0a
1.84(0.48,7.14)
1.33 (0.27, 6.59)
1.67(0.40,7.00)
2.81(0.25,31.10)
1.0a
3.99 (0.25, 63.94)
9,.59(0.60, 154.14)
1
0
0
17
15
7
3
5
2
1
1
0
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Unexposed workers
Deaths reported to among GE pension fund
(Pittsfield, MA)
Cardboard manufacturing workers, Atlanta area,
GA
Not reported
Not reported
0.95(0.1,3.17)'
Not reported
13
U. S. Coast Guard employees
Marine inspectors
Noninspectors
0.72 (0.09, 2.62) 2
0.74 (0.09, 2.68) 2
Aircraft manufacturing plant employees (Italy)
All subjects
0.21(0.01, 1.17)
1
Reference
Blair etal., 1998
Radican et al., 2008
Henschler et al., 1995
Greenland et al., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-554 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-99. Summary of human studies on TCE exposure and esophageal
cancer (continued)
Exposure group
Rubber Workers
Relative risk
(95% CI)
Not reported1
No. obs.
events
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
1.14(0.62, 1.92)
14
Reference
Wilcosky et al., 1984
Garabrant et al., 1988
Case-control studies
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
0.5(0.1, 2.5)1
0.8(0.1, 4.6)1
1
1
Siemiatyckietal., 1991;
Parent et al., 2000
Geographic based studies
Residents in two study areas in Endicott, NY
Residents of 13 census tracts inRedlands, CA
0.78 (0.29, 1.70)
Not reported
6
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
ATSDR, 2006
Morgan and Cassidy, 2002
Vartiainen et al., 1993
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
"Internal referents, workers not exposed to TCE.
bRitz (1999) and Zhao et al. (2005) reported relative risks for the combined site of esophagus and stomach.
cSung et al. (2007) and Chang et al. (2005)—SIR for females and reflects a 10-year lag period.
dSIR for adenocarcinoma of the esophagus.
eThe SIR for adenocarcinoma histologic type can not be calculated because Hansen et al. (2001) do not present
expected numbers for adenocarcinoma histologic type of esophageal cancer. An approximation of the SIR for
adenocarcinoma histologic type is presented using the expected number of total number of expected esophageal
cancers for males (n = 1.4). The expected numbers of esophageal adenocarcinomas in males will be lower;
Hansen et al. (2001) noted the proportion of adenocarcinomas among the comparable Danish male population
during the later period of the study (1990-1996) as 38%. A rough approximation of the expected number of
esophageal carcinomas would be 0.5 expected cases and an approximated SIR of 9.4 (3.1, 22).
Proportional mortality ratio.
8Adjusted relative risks for >2 year exposure duration and 15 year lag from 1st exposure.
hNo esophageal cancer deaths occurred in the referent population in Blair et al. (1998) and relative risk in could not
be calculated for this reason.
'Odds ratio from nested case-control analysis.
J90% confidence interval.
Meta-analysis is not adopted as a tool for examining the body of epidemiologic evidence
on esophageal cancer and TCE exposure given the absence of reported relative risk estimates in
several of the high-quality studies (Axelson et al., 1994; Anttila et al., 1995; Morgan et al.,
1998).
Overall, three high-quality cohort studies provide some evidence of association for
esophageal cancer and TCE exposure. The finding in two of these studies of esophageal risk
estimates among subjects with long employment duration were higher than those associated with
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-555 DRAFT—DO NOT CITE OR QUOTE
-------
1 low employment duration provides additional evidence (Hansen et al., 2001; Raaschou-Nielsen
2 et al., 2003). The cohort studies are unable to directly examine possible confounding due to
3 suspected risk factors for esophageal cancer such as smoking, obesity and alcohol. The use of an
4 internal referent group, similar in socioeconomic status as exposed subjects, is believed to
5 minimize but may not completely control for possible confounding related to smoking and health
6 status (Blair et al., 1998; its follow-up Radican et al., 2008; Zhao et al., 2005; Boice et al, 2006).
7 Observation of a higher risk for adenocarcinoma histologic type than for a combined category of
8 esophageal cancer in Raaschou-Nielsen et al. (2003) also suggests minimal confounding from
9 smoking. Smoking is not identified as a possible risk factor for the adenocarcinoma histologic
10 type of esophageal cancer but is believed a risk factor for squamous cell histologic type.
11 Furthermore, the magnitude of lung cancer risk in Raaschou-Nielsen et al. (2003) suggests a high
12 smoking rate is unlikely. The lack of association with overall TCE exposure and the absence of
13 exposure-response patterns in the other studies of TCE exposure may reflect limitations in
14 statistical power, the possibility of exposure misclassification, and differences in measurement
15 methods. These studies do not provide evidence against an association between TCE exposure
16 and esophageal cancer.
17
18 4.9.2. Bladder Cancer
19 Twenty-five epidemiologic studies present risk estimates for bladder cancer (Garabrant et
20 al., 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Mallin, 1990; Siemiatycki,
21 1991; Sinks et al., 1992; Axelson et al., 1994; Greenland et al., 1994; Anttila et al., 1995; Blair et
22 al., 1998; Morgan et al., 1998; Boice et al., 1999, 2006; Pesch et al., 2000b; Hansen et al., 2001;
23 Cassidy and Morgan, 2002; Chang et al., 2003, 2005; Raaschou-Nielsen et al., 2003; ATSDR,
24 2004, 2006; Zhao et al., 2005; Sung et al., 2007; Radican et al., 2008). Table 4-100 presents risk
25 estimates for TCE exposure and bladder cancer observed in cohort, case-control, and geographic
26 based studies. Thirteen studies, all either cohort or case-control studies, which there is a high
27 likelihood of TCE exposure in individual study subjects (e.g., based on job-exposure matrices or
28 biomarker monitoring) or which met, to a sufficient degree, the standards of epidemiologic
29 design and analysis in a systematic review, reported relative risk estimates for bladder or
30 urothelial cancer between 0.6 (Siemiatycki, 1991) and 1.7 (Boice et al., 2006) and overall TCE
31 exposure. Relative risk estimates were generally based on small numbers of cases or deaths,
32 except for one study (Raaschou-Nielsen et al., 2004), with the result of wide confidence intervals
33 on the estimates. Of high-quality studies, two reported statistically significant elevated bladder
34 or urothelial cancer risks with the highest cumulative TCE exposure category (2.71, 95% CI:
35 1.10, 6.65 [Morgan et al., 1998]; 1.8, 95% CI: 1.2, 2.7 [Pesch et al., 2000b]) and five presented
36 risk estimates and categories of increasing cumulative TCE exposure (Blair et al., 1998; Morgan
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-556 DRAFT—DO NOT CITE OR QUOTE
-------
1 et al., 1998; Pesch et al., 2000b; Zhao et al., 2005; Radican et al., 2008). Risk estimates in
2 Morgan et al. (1998), Pesch et al. (2000b), and Zhao et al. (2005) appeared to increase with
3 increasing cumulative TCE exposure with the/>-value for trend of 0.07 in Zhao et al. (2005), the
4 only study to present a formal statistical test for linear trend. Risk estimates did not appear to
5 either increase or decrease with increasing cumulative TCE exposure in Blair et al. (1998) or its
6 update Radican et al. (2008), which added another 10 years of follow-up. Twelve additional
7 studies were given less weight because of their lesser likelihood of TCE exposure and other
8 design limitations that would decrease statistical power and study sensitivity (Garabrant et al.,
9 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Mallin, 1990; Sinks et al., 1992;
10 Cassidy and Morgan, 2002; Chang et al., 2003, 2005; ATSDR, 2004, 2006; Sung et al., 2007).
11 Meta-analysis is not adopted as a tool for examining the body of epidemiologic evidence
12 on bladder cancer and TCE.
13 Overall, three high-quality cohort or case-control studies provide some evidence of
14 association for bladder or urothelial cancer and high cumulative TCE exposure (Morgan et al.,
15 1998; Pesch et al., 2000b; Zhao et al., 2005). The case-control study of Pesch et al. (2000b)
16 adjusted for age, study center, and cigarette smoking, with a finding of a statistically significant
17 risk estimate between urothelial cancer and the highest TCE exposure category. Cancer cases in
18 this study are of several sites, bladder, ureter, and renal pelvis, and grouping different site-
19 specific cancers with possible etiologic heterogeneity may introduce misclassification bias. The
20 cohort studies are unable to directly examine possible confounding due to suspected risk factors
21 for esophageal cancer such as smoking, obesity, and alcohol. The use of an internal referent
22 group, similar in socioeconomic status as exposed subjects, by Morgan et al. (1998) and Zhao et
23 al. (2005) is believed to minimize but may not completely control for possible confounding
24 related to smoking and health status. The lack of association with overall TCE exposure in other
25 studies and the absence of exposure-response patterns with TCE exposure in Blair et al. (1998)
26 and Radican et al. (2008) may reflect limitations in statistical power, the possibility of exposure
27 misclassification, and differences in measurement methods. These studies do not provide
28 evidence against an association between TCE exposure and bladder cancer.
29
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-557 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-100. Summary of human studies on TCE exposure and bladder
cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
Not reported
1.00a
1.54 (0.81, 2.92)b
1.98 (0.93, 4.22) b
p = 0.069
20
19
11
TCE, 20 yrs exposure lag
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
1.00a
1.76(0.61,5.10)°
3.68(0.87, 15.5)c
p = 0.064
20
20
10
All employees at electronics factory (Taiwan)
Males
Females
Males
Females
Not reported
0.34 (0.07, 1.00)
1.06(0.45, 2.08) d
1.09(0.56, 1.91) d
10
8
12
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Any exposure, males
Any exposure, females
Biologically -monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
1.1 (0.92, 1.21)
1.0(0.89, 1.18)
1.6 (0.93, 2.57)
1.0 (0.48, 1.86)
1.1 (0.50,2.0)
0.5 expected
220
203
17
10
10
0
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Not reported
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.7 (0.6, 4.4)
1.7 (0.6, 4.9)
1.4(0.5,4.1)
13
9
9
Zhao et al., 2005
Sung et al., 2007
Chang et al., 2005
Raaschou-Nielsen et al., 2003
Hansenetal., 2001
Blair etal., 1998
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-558 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-100. Summary of human studies on TCE exposure and bladder
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.1(0.1, 10.8)
1.0(0.1,9.1)
1
0
1
Biologically -monitored Finnish workers
All subjects
0.82 (0.27, 1.90)
5
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
1.02 (0.44, 2.00)
Not reported
8
Reference
Anttila et al., 1995
Axelsonetal., 1994
Cohort and PMR studies-mortality
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Any exposure to TCE
Low cumulative TCE score
Med cumulative TCE score
High TCE score
p for trend
1.66 (0.54, 3.87)
Not reported
1.00a
1.27(0.43, 3.73)b
1.15(0.29, 4.51)b
p = 0.809
5
8
6
3
TCE, 20 yrs exposure lag
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
1.00a
0.95(0.15, 6.02) c
1.85(0.12, 27.7)c
;? = 0.533
8
7
2
View-Master employees
Males
Females
1.22(0.15,4.40)
0.78 (0.09, 2.82)
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration
Moderate TCE exposure, >2 yrs duration
Not reported
Not reported
Not reported
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent3
0.55(0.18, 1.28)
Not reported
5
Boiceetal.,2006
Zhao et al., 2005
ATSDR, 2004
Ritz, 1999
Boiceetal., 1999
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-559 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-100. Summary of human studies on TCE exposure and bladder
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
1.36 (0.59, 2.68)
0.51 (0.01,2.83)
1.79 (0.72, 3.69)
8
1
7
TCE subcohort (Cox Analysis)
Never exposed
Ever exposed
1.0a
2.05(0.86, 4.85) e
8
Peak
No/low
Medium/high
1.0a
1.41(0.52,3.81)
5
Cumulative
Referent
Low
High
1.0a
0.69 (0.09, 5.36)
2.71(1.10,6.65)
1
7
Aircraft maintenance workers (Hill AFB, UT)
TCE subcohort
1.2(0.5, 2.9) a
17
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.8 (0.5, 6.2)
2.1(0.6,8.0)
1.0(0.2,5.1)
7
5
3
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
0.8(0.1,7.5)
0.80(0.41, 1.58)
1.05(0.47,2.35)
1.0a
0.96(0.37,2.51)
1.77 (0.70, 4.52)
0.67(0.15,2.95)
0
0
1
25
24
9
10
5
Reference
Morgan etal., 1998
Blair etal., 1998
Radican et al., 2008
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-560 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-100. Summary of human studies on TCE exposure and bladder
cancer (continued)
Exposure group
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Relative risk
(95% CI)
0.22 (0.03, 1.83)
1.0a
2.86 (0.27, 29.85)
No. obs.
events
1
0
1
0
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Unexposed workers
Deaths reported to GE pension fund (Pittsfield,
MA)
Not reported
Not reported
0.85 (0.32, 2.23)f
20
Cardboard manufacturing workers, Atlanta area, GA
0.3 (0.0, 1.6)
1
U. S. Coast Guard employees
Marine inspectors
Noninspectors
0.50 (0.06, 1.79)
0.90(0.18,2.62)
2
3
Aircraft manufacturing plant employees (Italy)
All subjects
0.74 (0.30, 1.53)
7
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
Lamp manufacturing workers (GE)
1.26 (0.74, 2.03)
0.93 (0.19,2.72)
17
3
Reference
Henschler et al., 1995
Greenland etal., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrant et al., 1988
Shannon et al., 1988
Case-control studies
Population of 5 regions in Germany
Any TCE exposure
Males
Females
Not reported
Not reported
Not reported
Males
Medium
High
Substantial
0.8(0.6, 1.2) 8
1.3(0.8, 1.7)8
1.8(1.2, 2.7)8
47
74
36
Population of Montreal, Canada
Any TCE exposure
Substantial TCE exposure
0.6 (0.3, 1.2)
0.7 (0.3, 1.6)
8
5
Peschetal.,2000b
Siemiatycki, 1991;
Siemiatyckietal., 1994
Geographic based studies
Residents in two study areas in Endicott, NY
0.71 (0.38, 1.21)
13
ATSDR, 2006
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-561 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-100. Summary of human studies on TCE exposure and bladder
cancer (continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Residents of 13 census tracts in Redlands, CA
0.98(0.71, 1.29)h
82
Finnish residents
Residents of Hausjarvi
Residents of Huttula
Not reported
Not reported
Residents of 9 county area in Northwestern Illinois
All zip codes in study area
Males
Females
1.4(1.1, 1.9)
1.8(1.2,2.7)
47
21
Cluster community
Males
Females
1.7(1.1,2.6)
2.6 (1.2, 4.7)
21
10
Adjacent community
Males
Females
1.2 (0.6, 2.0)
1.6(0.5,3.8)
12
5
Remainder of zip code areas
Males
Females
1.4 (0.8, 2.2)
1.4 (0.5, 3.0)
14
6
Reference
Morgan and Cassidy, 2002
Vartiainen et al., 1993
Mallin, 1990
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Internal referents, workers not exposed to TCE.
bRelative risk estimates for TCE exposure after adjustment for 1st employment, socioeconomic status, and age at
event.
'Relative risk estimates for TCE exposure after adjustment for 1st employment, socioeconomic status, age at event,
and all other carcinogen exposures, including hydrazine.
dChang et al. (2005) and Costa et al. (1989) report estimated risks for a combined site of all urinary organ cancers.
eRisk ratio from Cox Proportional Hazard Analysis, stratified by age, sex and decade (Environmental Health
Strategies, 1997).
fOdds ratio from nested case-control analysis.
8Odds ratio for urothelial cancer, a category of bladder, ureter, and renal pelvis cancers) and cumulative TCE
exposure, as assigned using a job-task-exposure matrix (ITEM) approach (Pesch et al., 2000b).
h99% confidence interval.
4.9.3. Central Nervous System and Brain Cancers
Brain cancer is examined in most cohort studies and in one case-control study (Garabrant
et al., 1988; Blair et al., 1989; Costa et al., 1989; Greenland et al., 1994; Heineman et al., 1994;
Anttila et al., 1995; Henschler et al., 1995; Blair et al., 1998; Morgan et al., 1998; Boice et al.,
1999, 2006; Ritz, 1999; Hansen et al., 2001; Chang et al., 2003, 2005; Raaschou-Nielsen et al.,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-562 DRAFT—DO NOT CITE OR QUOTE
-------
1 2003; Zhao et al., 2005; Sung et al., 2007; Clapp and Hoffman, 2008; Radican et al., 2008).
2 Overall, these epidemiologic studies do not provide strong evidence for or against association
3 between TCE and brain cancer in adults (see Table 4-101). Relative risk estimates in well
4 designed and conducted cohort studies, Axelson et al. (1994), Anttila et al. (1995), Blair et al.
5 (1998), its follow-up reported in Radican et al. (2008), Morgan et al. (1998), Boice et al. (1999),
6 Zhao et al. (2005), and Boice et al. (2006), are near a risk of 1.0 and imprecise, confidence
7 intervals all include a risk estimate of 1.0. All studies except Raaschou-Nielsen et al. (2003),
8 observations are based on few events and lowered statistical power. Bias resulting from
9 exposure misclassification is likely in these studies, although of a lower magnitude compared to
10 other cohort studies identified in Table 4-101, and may partly explain observations. Exposure
11 misclassification is also likely in the case-control study of occupational exposure of Heineman et
12 al. (1994) who do not report association with TCE exposure.
13 Three geographic-based studies and one case-control study examined childhood brain
14 cancer (AZ DHS, 1990, 1995; De Roos et al., 2001; Morgan and Cassidy, 2002; ATSDR, 2006).
15 The strongest study, De Roos et al. (2001), a population case-control study which examined
16 paternal exposure, used expert judgment to evaluate the probably of TCE exposure from self-
17 reported information in an attempt to reduce exposure misclassification bias. The odds ratio
18 estimate in this study was 0.9 (95% CI: 0.3, 2.5). Like many population case-control studies, a
19 low prevalence of TCE exposure was found, only 9 fathers were identified with probable TCE
20 exposure by the industrial hygiene review, and greatly impacted statistical power. There is some
21 concern for childhood brain cancer and organic solvent exposure based on Peters et al. (1981)
22 whose case-control study of childhood brain cancer reported to the Los Angeles County Cancer
23 Surveillance Program observed a high odds ratio estimate for paternal employment in the aircraft
24 industry (OR: co,p< 0.001). This study does not present an odds ratio for TCE exposure only
25 although it did identify two of the 14 case and control fathers with previous employment in the
26 aircraft industry reported exposure to TCE.
27
28 4.10. SUSCEPTIBLE LIFESTAGES AND POPULATIONS
29 Variation in response among segments of the population may be due to age, genetics, and
30 ethnicity, as well as to differences in lifestyle, nutrition, and disease status. These could be
31 potential risk factors that play an important role in determining an individual's susceptibility and
32 sensitivity to chemical exposures. Studies on TCE toxicity in relation to some of these risk
33 factors including lifestage, gender, genetics, race/ethnicity, pre-existing health status, and
34 lifestyle are discussed below. Others have also reviewed factors related to human variability and
35 their potential for susceptibility to TCE (Barton et al., 1996; Clewell et al., 2000; Davidson and
36 Beliles, 1991; NRC, 2006; Pastino et al., 2000).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-563 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-101. Summary of human studies on TCE exposure and brain cancer
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
Not reported
1.00a
0.46(0.09, 2.25) b
0.47(0.06, 3.95) b
p= 0.382
7
2
1
All employees at electronics factory (Taiwan)
Males
Females
Males
Females
Not reported
1.07(0.59, 1.80) c
0.40 (0.05, 1.46)
0.97 (0.54, 1.61)
2
15
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Any exposure, males
Any exposure, females
Biologically -monitored Danish workers
Any TCE exposure, males
Any TCE exposure, females
1.0 (0.84, 1.24)
1.0(0.76, 1.18)
1.1 (0.67, 1.74)
0.3 (0.01, 1.86)
0.4(0.01,2.1)
0.5 expected
104
85
19
1
1
0
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Not reported
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
2.0 (0.2, 19.7)
3.9 (0.4, 34.9)
0.8(0.1, 13.2)
3
4
1
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
0
0
0
Biologically -monitored Finnish workers
All subjects
1.09(0.50,2.07)
9
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
1.52(0.61,3.13)
0.76 (0.01, 2.74)
7
2
Biologically -monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
Not reported
Not reported
Zhao etal., 2005
Sung etal, 2007
Chang et al., 2005
Raaschou-Nielsen et al., 2003
Hansen etal., 2001
Blair etal., 1998
Anttilaetal., 1995
Axelsonetal., 1994
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-564 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-101. Summary of human studies on TCE exposure and brain cancer
(continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Reference
Cohort and PMR studies-mortality
Computer manufacturing workers (IBM), NY
Males
Females
1.90(0.52,4.85)
4
0
Aerospace workers (Rocketdyne)
Any TCE (utility /eng flush)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
p for trend
0.81(0.17,2.36)
Not reported
1.00a
0.42(0.12, 1.50)
0.83 (0.23, 3.08)
;? = 0.613
3
12
3
o
J
View-Master employees
Males
Females
Not reported
Not reported
All employees at electronics factory (Taiwan)
Males
Females
0.96(0.01,5.36)
0.96(0.01,5.33)
1
1
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration, 0 lag
Moderate TCE exposure, >2 yrs duration,
Olag
Light TCE exposure, >5 yrs duration, 15
yrlag
Moderate TCE exposure, >5 yrs duration,
15 yrlag
Not reported
1.81 (0.49, 6.71) d
3.26(0.37, 28.9) d
5.41(0.87, 33.9) d
14.4(1.24, 167) d
6
1
3
1
Aerospace workers (Lockheed)
Routine exposure
Routine-intermittent3
0.54(0.15,1.37)
Not presented
4
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)d
High intensity (>50 ppm) d
0.99 (0.64, 1.47)
0.73 (0.09, 2.64)
0.44 (0.05, 1.58)
4
2
2
Aircraft maintenance workers (Hill AFB, Utah)
TCE subcohort
0.8(0.2, 2.2) a
11
Clapp and Hoffman, 2008
Boice et al., 2006
Zhao etal., 2005
ATSDR, 2004
Chang etal., 2003
Ritz, 1999
Boice etal., 1999
Morgan etal., 1998
Blair etal., 1998
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-565 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-101. Summary of human studies on TCE exposure and brain cancer
(continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
0.7 (0.7, 3,3)
2.0 (0.5, 8.4)
0.9 (0.2, 4.4)
3
5
2
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
1.0a
1.02 (0.39, 2.67)
1.26 (0.43, 3.75)
1.0a
1.46 (0.44, 4.86)
1.74(0.49,6.16)
0.66(0.15,2.95)
0
0
0
17
17
8
6
3
0
Cardboard manufacturing workers in Arnsburg, Germany
TCE exposed workers
Unexposed workers
Deaths reported to GE pension fund (Pittsfield,
MA)
3.70 (0.09, 20.64)
9.38(1.93,27.27)
0.93 (0.32, 2.69)e
1
3
16
Cardboard manufacturing workers, Atlanta area, GA
Not reported
U. S. Coast Guard employees
Marine inspectors
Noninspectors
1.70 (0.55, 3.95)
1.36(0.44,3.17)
5
5
Aircraft manufacturing plant employees (Italy)
All subjects
0.79(0.16,2.31)
3
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
0.78 (0.42, 1.34)
16
Reference
Radican et al., 2008
Henschleretal., 1995
Greenland et al., 1994
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrantetal., 1988
Case-control studies
Children's Cancer Group/Pediatric Oncology Group
Any TCE exposure
1.64(0.95,2.84)
37
Neuroblastoma, <15 yrs age
DeRoos etal., 2001
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-566 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-101. Summary of human studies on TCE exposure and brain cancer
(continued)
Exposure group
Relative risk
(95% CI)
No. obs.
events
Paternal TCE exposure
Serf-reported exposure
IH assignment of probable exposure
1.4 (0.7, 2.9)
0.9 (0.3, 2.5)
22
9
Population of So. LA, NJ, Philadelphia PA
Any TCE exposure
Low exposure
Medium exposure
High exposure
p for trend
1.1 (0.8, 1.6)
1.1 (0.7, 1.7)
1.1 (0.6, 1.8)
1.1(0.5,2.8)
0.45
128
27
42
12
Reference
Heineman et al., 1994
Geographic based studies
Residents in two study areas in Endicott, NY
Brain/CNS, <19 yrs of age
Residents of 13 census tracts inRedlands, CA
Brain/CNS, <15 yrs of age
Not reported
1.05(0.24, 2.70) f
<6
6
Resident of Tucson Airport Area, AZ
Brain/CNS, <19 yrs of age
1970-1986
1987-1991
0.84(0.23,2.16)
0.78 (0.26, 2.39)
3
2
ATSDR, 2006
Morgan and Cassidy, 2002
AZ DHS, 1990, 1995
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
"Internal referents, workers not exposed to TCE.
bRelative risks for TCE exposure after adjustment for 1st employment, socioeconomic status, and age at event.
Standardized incidence ratio from analyses lagging exposure 10 years prior to end of follow-up or date of incident
cancer.
dRelative risks for TCE exposure after adjustment for time since 1st hired, external and internal radiation dose, and
same chemical at a different level.
eOdds ratio from nested case-control analysis.
f99% confidence interval.
4.10.1. Lifestages
Individuals of different lifestages are physiologically, anatomically, and biochemically
different. Early (infants and children) and later (the elderly) lifestages differ greatly from
adulthood in body composition, organ function, and many other physiological parameters that
can influence the toxicokinetics of chemicals and their metabolites in the body (ILSI, 1992). The
limited data on TCE exposure suggest that these segments of the population—particularly
individuals in early lifestages—may have greater susceptibility than does the general population.
This section presents and evaluates the pertinent published literature available to assess how
individuals of differing lifestages may respond differently to TCE.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-567 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.1.1. Early Lifestages
2 4.10.1.1.1. Early lifestage-specific exposures. Section 2.4 describes the various exposure
3 pathways of concern for TCE. For all postnatal lifestages, the primary exposure routes of
4 concern include inhalation and contaminated drinking water. In addition, there are exposure
5 pathways to TCE are unique to early lifestages. Fetal and infant exposure to TCE can occur
6 through placental transfer and breast milk consumption if the mother has been exposed, and
7 could potentially increase overall TCE exposure. Placental transfer of TCE has been
8 demonstrated in humans (Beppu, 1968; Laham, 1970), rats (Withey and Karpinski, 1985), mice
9 (Ghantous et al., 1986), rabbits (Beppu, 1968), and sheep and goats (Helliwell and Hutton,
10 1950). Similarly, TCE has been found in breast milk in humans (Fisher et al., 1997; Pellizzari et
11 al., 1982), goats (Hamada and Tanaka, 1995), and rats (Fisher et al., 1990). Pellizzari et al.
12 (1982) conducted a survey of environmental contaminants in human milk, using samples from
13 cities in the northeastern region of the United States and one in the southern region and detected
14 TCE in 8 milk samples taken from 42 lactating women. No details of times postpartum, milk
15 lipid content or TCE concentration in milk or blood were reported. Fisher et al. (1997) predicted
16 that a nursing infant would consume 0.496 mg TCE during a 24-hour period. In lactating rats
17 exposed to 600 ppm (3,225 mg/m3) TCE for 4 hours resulted in concentrations of TCE in milk of
18 110 |ig/mL immediately following the cessation of exposure (Fisher et al., 1990).
19 Direct childhood exposures to TCE from oral exposures may also occur. A
20 contamination of infant formula resulted in levels of 13 ppb (Fan, 1988). Children consume high
21 levels of dairy products, and TCE has been found in butter and cheese (Wu and Schaum, 2000).
22 In addition, TCE has been found in food and beverages containing fats such as margarine
23 (Wallace et al., 1984), grains and peanut butter (Wu and Schaum, 2000), all of which children
24 consume in high amounts. A number of studies have examined the potential adverse effects of
25 childhood exposure to drinking water contaminated with TCE (ATSDR, 1998, 2001;
26 Bernad et al., 1987; Bove, 1996; Bove et al., 1995; Burg and Gist, 1999; Goldberg et al., 1990;
27 Lagakos et al., 1986; Rodenbeck et al., 2000; Sonnenfeld et al., 2001; White et al., 1997; see
28 Section 4.10.2.1). TCE in residential water may also be a source of dermal or inhalation
29 exposure during bathing and showering (Fan, 1988; Franco et al., 2007; Giardino and Andelman,
30 1996; Lee et al., 2002; Weisel and Jo, 1996; Wu and Schaum, 2000); it has been estimated that
31 showering and bathing scenarios in water containing 3-ppm TCE, a child of 22 kg receives a
32 higher dose (about 1.5 times) on a mg/kg basis than a 70 kg adult (Fan, 1988).
33 Direct childhood inhalation exposure to TCE have been documented in both urban and
34 rural settings. A study of VOCs measured personal, indoor and outdoor TCE in 284 homes, with
35 72 children providing personal measures and time-activity diaries (Adgate et al., 2004a). The
36 intensive-phase of the study found a mean personal level of 0.8 |ig/m3 and mean indoor and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-568 DRAFT—DO NOT CITE OR QUOTE
-------
1 outdoor levels of 0.6 |ig/m3, with urban homes have significantly higher indoor levels of TCE
2 than nonurban homes (t = 2.3, p = 0.024) (Adgate et al., 2004a). A similar study of personal,
3 indoor and outdoor TCE was conducted in two inner-city elementary schools as well as in the
4 homes of 113 children along with time-activity diaries, and found a median a median personal
5 level of 0.3 |ig/m3, a median school indoor level of 0.2 |ig/m3, a median home indoor level of
6 0.3 |ig/m3, a median outdoor level of 0.3 |ig/m3 in the winter, with slightly lower levels in the
7 spring (Adgate et al., 2004b). Studies from Leipzig, Germany measured the median air level of
8 TCE in children's bedrooms to be 0.42 |ig/m3 (Lehmann et al., 2001) and 0.6 |ig/m3
9 (Lehmann et al., 2002). A study of VOCs in Hong Kong measured air levels in schools,
10 including an 8-hour average of 1.28 |ig/m3, which was associated with the lowest risk of cancer
11 in the study (Guo et al., 2004). Another found air TCE levels to be highest in school/work
12 settings, followed by outside, in home, in other, and in transit settings (Sexton et al., 2007).
13 Measured indoor air levels ranged from 0.18-140 ug/m3 for children exposed through vapor
14 intrusion from soil vapor (ATSDR, 2006). Contaminated soil may be a source of either dermal
15 or ingestion exposure of TCE for children (Wu and Schaum, 2000).
16 Additional TCE exposure has also been documented to have occurred during medical
17 procedures. TCE was used in the past as an anesthetic during childbirth (Beppu, 1968; Phillips
18 and Macdonald, 1971) and surgery during childhood (Jasinka, 1965). These studies are
19 discussed in more detail in Section 4.8.3.1.1. In addition, the TCE metabolite chloral hydrate has
20 been used as an anesthetic for children for CAT scans (Steinberg, 1993).
21 Dose received per body weight for 3-ppm TCE via oral, dermal, dermal plus inhalation,
22 and bathing scenarios was estimated for a 10-kg infant, a 22-kg child, and a 70-kg adult (Fan,
23 1988; see Table 4-102). For the oral route (drinking water), an infant would receive a higher
24 daily dose than a child, and the child more than the adult. For the dermal and dermal plus
25 inhalation route, the child would receive more than the adult. For the bathing scenario, the infant
26 and child would receive comparable amounts, more than the adult.
27
28 4.10.1.1.2. Early lifestage-specific toxicokinetics. Chapter 3 describes the toxicokinetics of
29 TCE. However, toxicokinetics in developmental lifestages are distinct from toxicokinetics in
30 adults (Benedetti et al., 2007; Ginsberg et al., 2002, 2004a, 2004b; Hattis et al., 2003) due to, for
31 example, altered ventilation rates, percent adipose tissue, and metabolic enzyme expression.
32 Early lifestage-specific information is described below for absorption, distribution, metabolism,
33 and excretion, followed by available early lifestage-specific PBPK models.
34
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-569 DRAFT—DO NOT CITE OR QUOTE
-------
Table 4-102. Estimated lifestage-specific daily doses for TCE in water*
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Drinking water
Showering — dermal
Showering — dermal
and inhalation
Bathing — 15 min
Bathing — 5 min
Body weight
Infant (10 kg)
0.3 mg/kg
-
-
-
0.08 mg/kg
Child (22 kg)
0.204 mg/kg
0.1 mg/kg
0.1 29 mg/kg
0.24 mg/kg
0.08 mg/kg
Adult (70 kg)
0.086 mg/kg
0.064 mg/kg
0.083 mg/kg
0.1 54 mg/kg
0.051 mg/kg
*Adapted from Fan (1988).
4.10.1.1.2.1. Absorption. As discussed in Section 3.1, exposure to TCE may occur via
inhalation, ingestion, and dermal absorption. In addition, prenatal exposure may result in
absorption via the transplacental route. Exposure via inhalation is proportional to the ventilation
rate, duration of exposure, and concentration of expired air, and children have increased
ventilation rates per kg body weight compared to adults, with an increased alveolar surface area
per kg body weight for the first two years (U.S. EPA, 2008). It is not clear to what extent dermal
absorption may be different for children compared to adults; however, infants have a 2-fold
increase in surface area compared to adults, although similar permeability (except for premature
babies) compared to adults (U.S. EPA, 2008).
4.10.1.1.2.2. Distribution. Both human and animal studies provide clear evidence that TCE
distributes widely to all tissues of the body (see Section 3.2). For lipophilic compounds such as
TCE, percentage adipose tissue, which varies with age, will affect absorption and retention of the
absorbed dose. Infants have a lower percentage of adipose tissue per body weight than adults,
resulting in a higher concentration of the lipophilic compound in the fat of the child (NRC,
1993).
During pregnancy of humans and experimental animals, TCE is distributed to the
placenta (Beppu, 1968; Ghantous et al., 1986; Helliwell and Hutton, 1950; Laham, 1970; Withey
and Karpinski, 1985). In humans, TCE has been found in newborn blood after exposure to TCE
during childbirth with ratios of concentrations in fetal:maternal blood ranging from
approximately 0.5 to approximately 2 (Laham, 1970). In childhood, blood levels concentrations
of TCE were found to range from 0.01-0.02 ng/mL (Sexton et al., 2005). Pregnant rats exposed
to TCE vapors on GD 17 resulted in concentrations of TCE in fetal blood approximately one-
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-570 DRAFT—DO NOT CITE OR QUOTE
-------
1 third the concentration in corresponding maternal blood, and was altered based upon the position
2 along the uterine horn (Withey and Karpinski, 1985). TCE has also been found in the organs of
3 prenatal rabbits including the brain, liver, kidneys and heart (Beppu, 1968). Rats prenatally
4 exposed to TCE had increased levels measured in the brain at PND10, compared to rats exposed
5 as adults (Rodriguez et al., 2007). TCE can cross the blood-brain barrier during both prenatal
6 and postnatal development, and may occur to a greater extent in younger children. It is also
7 important to note that it has been observed in mice that TCE can cycle from the fetus into the
8 amniotic fluid and back to the fetus (Ghantous et al., 1986).
9 Studies have examined the differential distribution by age to a mixture of six VOCs
10 including TCE to children aged 3-10 years and adults aged 20-82 years old (Mahle et al., 2007)
11 and in rats at PND10, 2 months (adult), and 2 years (aged) (Mahle et al., 2007; Rodriguez et al.,
12 2007). In humans, the blood:air partition coefficient for male or female children was
13 significantly lower compared to adult males (Mahle et al., 2007). In rats, the difference in
14 tissue:air partition coefficients increased with age (Mahle et al., 2007). Higher peak
15 concentrations of TCE in the blood were observed in the PND10 rat compared to the adult rat
16 after inhalation exposure, likely due to the lower metabolic capacity of the young rats
17 (Rodriguez et al., 2007).
18
19 4.10.1.1.2.3. Metabolism. Section 3.3 describes the enzymes involved in the metabolism of
20 TCE, including CYP and GST. Expression of these enzymes changes during various stages of
21 fetal development (Dome et al., 2005; Hakkola et al., 1996a, b, 1998a, b; Hines and McCarver,
22 2002; Shao et al., 2007; van Lieshout et al., 1998) and during postnatal development
23 (Blake et al., 2005; Dome et al., 2005; Tateishi et al., 1997), and may result in altered
24 susceptibility.
25 Expression of CYP enzymes have been shown to play a role in decreasing the
26 metabolism of TCE during pregnancy in rats, although metabolism increased in young rats
27 (3-week-old) compared to adult rats (18-week-old) (Nakajima et al., 1992a). For TCE, CYP2E1
28 is the main metabolic CYP enzyme, and expression of this enzyme has been observed in humans
29 in prenatal brain tissue at low levels beginning at 8-weeks gestation and increasing throughout
30 gestation (Brzezinski et al., 1999). Very low levels of CYP2E1 have been detected in some
31 samples fetal liver during the second trimester (37% of samples) and third trimester (80% of
32 samples) (Carpenter et al., 1996; Johnsrud et al., 2003), although hepatic expression surges
33 immediately after birth in most cases (Johnsrud et al., 2003; Vieira et al., 1996) and in most
34 infants reaches adult values by 3 months of age (Johnsrud et al., 2003; Vieira et al., 1996).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-571 DRAFT—DO NOT CITE OR QUOTE
-------
1 Although there is some uncertainty as to which GST isoforms mediate TCE conjugation,
2 it should be noted that their expression changes with fetal development (McCarver and Hines,
3 2002; Raijmakers et al., 2001; van Lieshout et al., 1998).
4
5 4.10.1.1.2.4. Excretion. The major processes of excretion of TCE and its metabolites are
6 discussed in Section 3.4, yet little is know about whether there are age-related differences in
7 excretion of TCE. The major pathway for elimination of TCE is via exhalation, and its
8 metabolites via urine and feces, and it is known that renal processes are not mature until about
9 6 months of age (NRC, 1993). Only one study was identified that measured TCE or its
10 metabolites in exhaled breath and urine in a 17-year old who ingested a large quantity of TCE
11 (Briining et al., 1998). TCE has also been measured in the breast milk in lactating women
12 (Fisher et al., 1997; Pellizzari et al., 1982), goats (Hamada and Tanaka, 1995), and rats (Fisher et
13 al., 1990).
14
15 4.10.1.1.2.5. Physiologically-based pharmacokinetic (PBPK) models. Early lifestage-specific
16 information regarding absorption, distribution, metabolism, and excretion needs to be considered
17 for a child-specific and chemical-specific PBPK model. To adequately address the risk to infants
18 and children, age-specific parameters for these values should be used in PBPK models that can
19 approximate the internal dose an infant or child receives based on a specific exposure level (see
20 Section 3.5).
21 Fisher et al. developed PBPK models to describe the toxicokinetics of TCE in the
22 pregnant rat (Fisher et al., 1989), lactating rat and nursing pup (Fisher et al., 1990). The prenatal
23 study demonstrates that approximately two-thirds of maternal exposure to both TCE and TCA
24 reached the fetus after maternal inhalation, gavage, or drinking water exposure (Fisher et al.,
25 1989). After birth, only 2% of maternal exposure to TCE reaches the pup; however, 15% and
26 30% of maternal TCA reaches the pup after maternal inhalation and drinking water exposure,
27 respectively (Fisher et al., 1990). One analysis of PBPK models examined the variability in
28 response to VOCs including TCE between adults and children, and concluded that the
29 intraspecies uncertainty factor for pharmacokinetics is sufficient to capture variability between
30 adults and children (Pelekis et al., 2001).
31
32 4.10.1.1.3. Early lifestage-specific effects. Although limited data exist on TCE toxicity as it
33 relates to early lifestages, there is enough information to discuss the qualitative differences. In
34 addition to the evidence described below, Section 4.8 contains information reproductive and
35 developmental toxicity. In addition, Sections 4.3 on neurotoxicity and Section 4.6 on
36 immunotoxicity characterize a wide array of postnatal developmental effects.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-572 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.1.1.3.1. Differential effects in early lifestages. There are a few adverse health outcomes, in
2 particular birth defects, which are observed only after early lifestage exposure to TCE.
O
4 Birth Defects. A summary of structural developmental outcomes that have been associated with
5 TCE exposures is presented in Sections 4.8.2.3. In particular, cardiac birth defects have been
6 observed after exposure to TCE in humans (ATSDR, 2006; Goldberg et al., 1990; Lagakos et al.,
7 1986; Yauck et al., 2004), rodents (Dawson et al., 1990, 1993; Johnson et al., 1998a, b, 2003,
8 2005; Smith et al., 1989, 1992), and chicks (Bross et al., 1983; Loeber et al., 1988; Boyer et al.,
9 2000; Drake et al., 2006a, b; Mishima et al., 2006; Rufer et al., 2008). However, it is notable
10 that cardiac malformations were not observed in a number of other studies in humans
11 (Lagakos et al., 1986; Taskinen et al., 1989; Tola et al., 1980), rodents (Carney et al., 2006;
12 Coberly et al., 1992; Cosby and Dukelow, 1992; Dorfmueller et al., 1979; Fisher et al., 2001;
13 Hardin et al., 1981; Healy et al., 1982; Narotsky and Kavlock, 1995; Narotsky et al., 1995;
14 Schwetz et al., 1975), and rabbits (Hardin et al., 1981). See Section 4.8.2.3.2 for further
15 discussion on cardiac malformations.
16 Structural CNS birth defects were observed in humans (ATSDR, 2001; Bove, 1996;
17 Bove et al., 1995; Lagakos et al., 1986). In addition, a number of postnatal nonstructural adverse
18 effects have been observed in humans and experimental animals following prenatal exposure to
19 TCE. See Sections 4.3.10 and 4.8.2.3.3 for further discussion on developmental neurotoxicity.
20 A variety of other birth defects have been observed—including eye/ear birth anomalies in
21 humans and rats (Lagakos et al., 1986; Narotsky et al., 1995; Narotsky and Kavlock, 1995);
22 lung/respiratory tract disorders in humans and mice (Das and Scott, 1994; Lagakos et al., 1986);
23 and oral cleft defects (Bove, 1996; Bove et al., 1995; Lagakos et al., 1986), kidney/urinary tract
24 disorders, musculoskeletal birth anomalies (Lagakos et al., 1986), and anemia/blood disorders
25 (Burg and Gist, 1999) in humans. See Section 4.8.2.3.5 for further discussion on other structural
26 developmental outcomes. A current follow-up study of the Camp Lejeune cohort will examine
27 birth defects and may provide additional insight (ATSDR, 2003b; GAO, 2007a, b; ATSDR,
28 2009).
29
30 4.10.1.1.3.2. Susceptibility to noncancer outcomes in early lifestages. There are a number of
31 adverse health outcomes observed after exposure to TCE that are observed in both children and
32 adults. Below is a discussion of differential exposure, incidence and/or severity in early
33 lifestages compared to adulthood.
34 Occupational TCE poisonings via inhalation exposure resulted in an elevated percent of
35 cases in the adolescents aged 15-19 years old (McCarthy and Jones, 1983). In addition, there is
36 concern for intentional exposure to TCE during adolescence, including a series of deaths
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-573 DRAFT—DO NOT CITE OR QUOTE
-------
1 involving inhaling typewriter correction fluid (King et al., 1985), a case of glue sniffing likely
2 associated with cerebral infarction in a 12-year-old boy with a 2-year history of exposure
3 (Parker et al., 1984), and a case of attempted suicide by ingestion of 70 mg TCE in a 17-year-old
4 boy (Briining et al., 1998).
5
6 4.10.1.1.3.2.1. Neurotoxicity. Adverse CNS effects observed after early lifestage exposure to
7 TCE in humans include delayed newborn reflexes (Beppu, 1968), impaired learning or memory
8 (Bernad et al., 1987, abstract; White et al., 1997); aggressive behavior (Bernad et al., 1987;
9 Blossom et al., 2008); hearing impairment (Burg and Gist, 1999); speech impairment (Burg and
10 Gist, 1995; White et al., 1997); encephalopathy (White et al., 1997); impaired executive and
11 motor function (White et al., 1997); attention deficit (Bernad et al., 1987; White et al., 1997), and
12 autism spectrum disorder (Windham et al., 2006). One analysis observed a trend for increased
13 adversity during development, with those exposed during childhood demonstrating more deficits
14 than those exposed during adulthood (White et al., 1997). In experimental animals, observations
15 include decreased specific gravity of newborn brains until weaning (Westergren et al., 1984),
16 reductions in myelination in the brains at weaning, significantly decreased uptake of
17 2-deoxyglucose in the neonatal rat brain, significant increase in exploratory behavior (Isaacson
18 and Taylor, 1989; Noland-Gerbec et al., 1986; Taylor et al., 1985), decreased rearing activity
19 (Fredriksson et al., 1993), and increased time to cross the first grid in open field testing
20 (George etal., 1986).
21 Two studies addressed whether or not children are more susceptible to CNS effects
22 (Burg et al., 1995; White et al., 1997). An analysis of three residential exposures of TCE
23 observed speech impairments in younger children and not at any other lifestage (White et al.,
24 1997). A national exposure registry also observed statistically significant speech impairment and
25 hearing impairment in 0-9 year olds and no other age group (Burg et al., 1995). However, a
26 follow-up study did not find a continued association with speech and hearing impairment in these
27 children, although the absence of acoustic reflexes remained significant (ATSDR, 2003a). See
28 Section 4.3 for further information on central nervous system toxicity, and Section 4.8.3.3.3 for
29 further information on developmental neurotoxicity.
30
31 4.10.1.1.3.2.2. Liver toxicity. No early lifestage-specific effects were observed after TCE
32 exposure. See Section 4.4 for further information on liver toxicity.
33
34 4.10.1.1.3.2.3. Kidney toxicity. Residents of Woburn, Massachusetts including 4,978 children
35 were surveyed on residential and medical history to examine an association with contaminated
36 wells; an association was observed for higher cumulative exposure measure and history of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-574 DRAFT—DO NOT CITE OR QUOTE
-------
1 kidney and urinary tract disorders (primarily kidney or urinary tract infections) and with lung and
2 respiratory disorders (asthma, chronic bronchitis, or pneumonia) (Lagakos et al., 1986). See
3 Section 4.5 for further information on kidney toxicity.
4
5 4.10.1.1.3.2.4. Immunotoxicity. Several studies in exposure to TCE in early lifestages of humans
6 and experimental animals were identified that assessed the potential for developmental
7 immunotoxicity (Adams et al., 2003; Blossom and Doss, 2007; Blossom et al., 2008;
8 Lehmann et al., 2001, 2002; Peden-Adams et al., 2006, 2008). All noted evidence of immune
9 system perturbation except one (Lehman et al., 2001). See Section 4.6 for further information on
10 immunotoxicity, and Section 4.8.2.3.4 for further discussion on developmental immunotoxicity.
11
12 4.10.1.1.3.2.5. Respiratory toxicity. Residents of Woburn, Massachusetts including
13 4,978 children were surveyed on residential and medical history to examine an association with
14 contaminated wells; an association was observed for lung and respiratory disorders (asthma,
15 chronic bronchitis, or pneumonia) (Lagakos et al., 1986). See Section 4.7 for further information
16 on respiratory tract toxicity.
17
18 4.10.1.1.3.3. Susceptibility to cancer outcomes in early lifestages. The epidemiologic and
19 experimental animal evidence is limited regarding susceptibility to cancer from exposure to TCE
20 during early life stages. The human epidemiological evidence is summarized above for cancer
21 diagnosed during childhood (see Sections 4.8.2.1 and 4.8.2.3.5), including a discussion of
22 childhood cancers of the nervous system including neuroblastoma and the immune system
23 including leukemia (see Section 4.6.1.3). A current follow-up study of the Camp Lejeune cohort
24 will examine childhood cancers and may provide additional insight (ATSDR, 2003b; GAO,
25 2007a, b; ATDSR, 2009). No studies of cancers in experimental animals in early lifestages have
26 been observed.
27
28 4.10.1.1.3.3.1. Total childhood cancer. Total childhood cancers have been examined in
29 relationship to TCE exposure (ATSDR, 2006; Morgan and Cassady, 2002). Two studies
30 examining total childhood cancer in relation to TCE in drinking water did not observe an
31 association. A study in Endicott, NY contaminated by a number of VOCs, including "thousands
32 of gallons" of TCE observed fewer than 6 cases of cancer diagnosed between 1980 and 2001 in
33 children aged 0-19 years, and did not exceed expected cases or types (ATSDR, 2006). A
34 California community exposed to TCE in drinking water from contaminated wells was examined
35 for cancer, with a specific emphasis on childhood cancer (<15 years old); however, the incidence
36 did not exceed those expected for the community (Morgan and Cassady, 2002). A third study of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-575 DRAFT—DO NOT CITE OR QUOTE
-------
1 childhood cancer in relation to TCE in drinking water in Camp Lejeune, North Carolina is
2 currently underway (GAO, 2007a, b).
O
4 4.10.1.1.3.3.2. Childhood leukemia. Childhood leukemia has been examined in relationship to
5 TCE exposure (Cohn et al., 1994; Lagakos et al., 1986; Lowengart et al., 1987; McKinney et al.,
6 1991; Costas et al., 2002; Shu et al., 1999). In a study examining drinking water exposure to
7 TCE in 75 New Jersey towns, childhood leukemia, (including ALL) was significantly increased
8 for girls (n = 6) diagnosed before age 20 years, but this was not observed for boys (Cohn et al.,
9 1994). A community in Woburn, MA with contaminated well water including TCE experienced
10 20 cases of childhood leukemia, significantly more than expected (Lagakos et al., 1986). Further
11 analysis by Costas et al. (2002) also observed a greater than 2-fold increase over expected cases
12 of childhood leukemia. Cases were more likely to be male (76%), <9 years old at diagnosis
13 (62%), breast-fed (OR: 10.17, 95% CI: 1.22-84.50), and exposed during pregnancy (adjusted
14 OR: 8.33, 95% CI: 0.73-94.67). The highest risk was observed for exposure during pregnancy
15 compared to preconception or postnatal exposure, and a dose-response was seen for exposure
16 during pregnancy (Costas et al., 2002). In addition, family members of those diagnosed with
17 childhood leukemia, including 13 siblings under age 19 at the time of exposure, had altered
18 immune response, but an analysis looking at only these children was not done (Byers et al.,
19 1988).
20 Case-control studies examined children diagnosed with ALL for parental occupational
21 exposures and found a nonsignificant 2- to 4-fold increase of childhood leukemia risk for
22 exposure to TCE during preconception, pregnancy, postnatally, or all developmental periods
23 combined (Lowengart et al., 1987; McKinney et al., 1991; Shu et al., 1999). Some studies
24 showed an elevated risk for maternal (Shu et al., 1999) or paternal exposure (Lowengart et al.,
25 1987; McKinney et al., 1991), while others did not show an elevated risk for maternal
26 (McKinney et al., 1991) or paternal exposure (Shu et al., 1999), possibly due to the small number
27 of cases. No variability was observed in the developmental stages in Shu et al. (1999), although
28 Lowengart et al. (1987) observed the highest risk to be paternal exposure to TCE after birth.
29
30 4.10.1.1.3.3.3. CNS tumors. In a case-control study of parental occupational exposures, paternal
31 self-reported exposure to TCE was not significantly associated with neuroblastoma in the
32 offspring (OR = 1.4, 95%CI: 0.7-2.9) (De Roos et al., 2001). Brain tumors have also been
33 observed in the offspring of fathers exposed to TCE, but the odds ratio could not be determined
34 (Peters etal., 1981, 1985).
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-576 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.1.1.3.3.4. Age-dependent adjustment factors (ADAFs). According to U.S. EPA's
2 Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
3 (U.S. EPA, 2005b) there may be increased susceptibility to early-life exposures for carcinogens
4 with a mutagenic MOA. Therefore, because the weight of evidence supports a mutagenic MOA
5 for TCE carcinogenicity in the kidney (see Section 4.4.7), and in the absence of chemical-
6 specific data to evaluate differences in susceptibility, early-life susceptibility should be assumed
7 and the ADAFs should be applied, in accordance with the Supplemental Guidance.
8
9 4.10.1.2. Later Lifestages
10 Few studies examine the differential effects of TCE exposure for elderly adults
11 (>65 years old). These limited studies suggest that older adults may experience increased
12 adverse effects than younger adults. However, there is no further evidence for elderly
13 individuals exposed to TCE beyond these studies.
14 Toxicokinetics in later lifestages are distinct from toxicokinetics in younger adults
15 (Benedetti et al., 2007; Ginsberg et al., 2005). Studies have examined the age differences in TK
16 after exposure to a mixture of six VOCs including TCE for humans (Mahle et al., 2007) and rats
17 (Mahle et al., 2007; Rodriguez et al., 2007). In humans, the blood:air partition coefficient for
18 adult males (20-82 years) was significantly (p < 0.05) higher (11.7 ± 1.9) compared to male
19 (11.2 ± 1.8) or female (11.0 ± 1.6) children (3-10 years) (Mahle et al., 2007); when the data was
20 stratified for adults above and below 55 years of age, there was no significant difference
21 observed between adults (20-55 years) and aged (56-82) (data not reported). In rats, the
22 difference in tissue:air partition coefficients also increased from PND10 to adult (2 months) to
23 aged (2 years) rat (Mahle et al., 2007). TCE has also been measured in the brain of rats, with an
24 increased level observed in older (2 year old) rats compared to adult (2 month old) rats
25 (Rodriguez et al., 2007). It was also observed that aged rats reached steady state slower with
26 higher concentrations compared to the adult rat; the authors suggest that the almost 2-fold greater
27 percentage of body fat in the elderly is responsible for this response (Rodriguez et al., 2007). An
28 age-related difference in CYP expression (Dome et al., 2005), in particular CYP2E1 activity
29 were observed in human liver (George et al., 1995), with the lowest activity in those >60 years
30 and the highest in those <20 years old (Parkinson et al., 2004). Also, GST expression has been
31 observed to decrease with age in human lymphocytes, with the lowest expression in those aged
32 60-80 years old (van Lieshout and Peters, 1998).
33 One cohort of TCE exposed metal degreasers found an increase in psychoorganic
34 syndrome and increased vibration threshold related to increasing age (Rasmussen et al., 1993a, b,
35 c), although the age groups were <29 years, 30-39 years, and 40+ years, but the age ranged only
36 from 18-68 years and did not examine >65 years as a separate category.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-577 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2. Other Susceptibility Factors
2 Aside from age, many other factors may affect susceptibility to TCE toxicity. A partial
3 list of these factors includes gender, genetic polymorphisms, pre-existing disease status,
4 nutritional status, diet, and previous or concurrent exposures to other chemicals. The toxicity
5 that results due to changes in multiple factors may be quite variable, depending on the exposed
6 population and the type of exposure. Qualitatively, the presence of multiple susceptibility
7 factors will increase the variability that is seen in a population response to TCE toxicity.
8
9 4.10.2.1. Gender
10 Individuals of different genders are physiologically, anatomically, and biochemically
11 different. Males and females can differ greatly in many physiological parameters such as body
12 composition, organ function, and ventilation rate, which can influence the toxicokinetics of
13 chemicals and their metabolites in the body (Gandhi et al., 2004; Gochfeld, 2007).
14
15 4.10.2.1.1. Gender-specific toxicokinetics. Chapter 3 describes the toxicokinetics of TCE.
16 Gender-specific information is described below for absorption, distribution, metabolism, and
17 excretion, followed by available gender-specific PBPK models.
18
19 4.10.2.1.1.1. Absorption. As discussed in Section 3.1, exposure to TCE may occur via
20 inhalation, ingestion, and skin absorption. Exposure via inhalation is proportional to the
21 ventilation rate, duration of exposure, and concentration of expired air, and women have
22 increased ventilation rates during exercise compared to men (Gochfeld, 2007). Percent body fat
23 varies with gender (Gochfeld, 2007), which for lipophilic compounds such as TCE will affect
24 absorption and retention of the absorbed dose. After experimental exposure to TCE, women
25 were found to absorb a lower dose due to lower alveolar intake rates compared to men (Sato,
26 1993; Sato etal., 1991b).
27
28 4.10.2.1.1.2. Distribution. Both human and animal studies provide clear evidence that TCE
29 distributes widely to all tissues of the body (see Section 3.2). The distribution of TCE to specific
30 organs will depend on organ blood flow and the lipid and water content of the organ, which may
31 vary between genders (Gochfeld, 2007). After experimental exposure to humans, higher
32 distribution of TCE into fat tissue was observed in women leading to a greater blood
33 concentration 16 hours after exposure compared to men (Sato, 1993; Sato et al., 1991b). In
34 experimental animals, male rats generally have higher levels of TCE in tissues compared to
35 female rats, likely due to gender differences in metabolism (Lash et al., 2006). In addition, TCE
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-578 DRAFT—DO NOT CITE OR QUOTE
-------
1 has been observed in the male reproductive organs (epididymis, vas deferens, testis, prostate, and
2 seminal vesicle) (Zenick et al., 1984).
O
4 4.10.2.1.1.3. Metabolism. Section 3.3 describes the metabolic processes involved in the
5 metabolism of TCE, including CYP and GST enzymes. In addition, the role of metabolism in
6 male reproductive toxicity is discussed in Section 4.8.1.3.2. In general, there is some indication
7 that TCE metabolism is different between males and females, with females more rapidly
8 metabolizing TCE after oral exposure to rats (Lash et al., 2006), intraperitoneal injections in rats
9 (Verma and Rana, 2003), and in mouse, rat and human liver microsomes (Elfarra et al., 1998).
10 CYP expression may differ between genders (Gandhi et al., 2004; Gochfeld, 2007; Lash
11 et al., 2006; Parkinson et al., 2004). CYP2E1 was detected in the epididymis and testes of mice
12 (Forkert et al., 2002), and CYP2E1 and GST-a has been detected in the ovaries of rats (Wu and
13 Berger, 2008), indicating that metabolism of TCE can occur in both the male and female
14 reproductive tracts. Unrelated to TCE exposure, there is no gender-related difference in
15 CYP2E1 activity observed in human liver microsomes (Parkinson et al., 2004). One study of
16 TCE exposure in mice observed induced CYP2E1 expression in the liver of males only
17 (Nakajima et al., 2000). Male rats have been shown to have higher levels of TCE metabolites in
18 the liver (Lash et al., 2006), and lower levels of TCE metabolites in the kidney (Lash et al.,
19 2006) compared to female rats. However, another study did not observe ant sex-related
20 differences in the metabolism of TCE in rats (Nakajima et al., 1992a).
21 Unlike CYP-mediated oxidation, quantitative differences in the polymorphic distribution
22 or activity levels of GST isoforms in humans are not presently known. However, the available
23 data (Lash et al., 1999a, b) do suggest that significant variation in GST-mediated conjugation of
24 TCE exists in humans. One study observed that GSH conjugation is higher in male rats
25 compared to female rats (Lash et al., 2000); however, it has also been speculated that any gender
26 difference may be due to a polymorphism in GSH conjugation of TCE rather than a true gender
27 difference (Lash et al., 1999a). Also, induction of PPARa expression in male mice was greater
28 than that in females (Nakajima et al., 2000).
29
30 4.10.2.1.1.4. Excretion. The major processes of excretion of TCE and its metabolites are
31 discussed in Section 3.4. Two human voluntary inhalation exposure studies observed the levels
32 of TCE and its metabolites in exhaled breath and urine (Kimmerle and Eben, 1973; Nomiyama
33 and Nomiyama, 1971). Increased levels of TCE in exhaled breath in males were observed in one
34 human voluntary inhalation exposure study of 250-380 ppm for 160 minutes (Nomiyama and
35 Nomiyama, 1971), but no difference was observed in another study of 40 ppm for 4 hours or
36 50 ppm for 4 hours for 5 days (Kimmerle and Eben, 1973).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-579 DRAFT—DO NOT CITE OR QUOTE
-------
1 After experimental exposure to TCE, women were generally found to excrete higher
2 levels of TCE and TCA compared to men (Kimmerle and Eben, 1973; Nomiyama and
3 Nomiyama, 1971). However, other studies observed an increase in TCE in the urine of males
4 (Inoue et al., 1989), an increase in TCA in the urine of males (Sato et al., 1991b), or no
5 statistically significant (p > 0.10) gender difference for TCA in the urine (Inoue et al., 1989).
6 Others found that the urinary elimination half-life of TCE metabolites is longer in women
7 compared to men (Ikeda, 1977; Ikeda and Imamura, 1973).
8 In addition to excretion pathways that occur in both genders, excretion occurs uniquely in
9 men and women. In both humans and experimental animals, it has been observed that females
10 can excrete TCE and metabolites in breast milk (Fisher et al., 1990, 1997; Hamada and Tanaka,
11 1995; Pellizzari et al., 1982), while males can excrete TCE and metabolites in seminal fluid
12 (Forkert et al., 2003; Zenick et al., 1984).
13
14 4.10.2.1.1.5. Physiologically-based pharmacokinetic fPBPK) models. Gender-specific
15 differences in uptake and metabolism of TCE were incorporated into a PBPK model using
16 human exposure data (Fisher et al., 1998). The chemical-specific parameters included cardiac
17 output at rest, ventilation rates, tissue volumes, blood flow, and fat volume. This model found
18 that gender differences for the toxicokinetics of TCE are minor.
19
20 4.10.2.1.2. Gender-specific effects.
21 4.10.2.1.2.1. Gender susceptibility to noncancer outcomes.
22 4.10.2.1.2.1.1. Liver toxicity. No gender susceptibility to noncancerous outcomes in the liver
23 was observed. A detailed discussion of the studies examining the effects of TCE on the liver can
24 be found in Section 4.4.
25
26 4.10.2.1.2.1.2. Kidney toxicity. A detailed discussion of the studies examining the noncancer
27 effects of TCE on the kidney can be found in Section 4.5. A residential study found that females
28 aged 55-64 years old had an elevated risk of kidney disease (RR = 4.57, 99% CI: 2.10-9.93),
29 although an elevated risk of urinary tract disorders was reported for both males and females
30 (Burg et al., 1995). Additionally, a higher rate of diabetes in females exposed to TCE was
31 reported in two studies (Burg et al., 1995; Davis et al., 2005). In rodents, however, and kidney
32 weights were increased more in male mice than in females (Kjellstrand et al., 1983a, b), and
33 male rats have exhibited increased renal toxicity to TCE (Lash et al., 1998, 2001).
34
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-580 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2.1.2.1.3. Immunotoxicity. A detailed discussion of the studies examining the immunotoxic
2 effects of TCE can be found in Section 4.6. Most of the immunotoxicity studies present data
3 stratified by sex. The prevalence of exposure to TCE is generally lower in women compared
4 with men. In men, the studies generally reported odds ratios between 2.0 and 8.0, and in women,
5 the odds ratios were between 1.0 and 2.0. Based on small numbers of cases, an occupational
6 study of TCE exposure found an increased risk for systemic sclerosis for men (OR: 4.75,
7 95% CI: 0.99-21.89) compared to women (OR: 2.10; 95% CI: 0.65-6.75) (Diot et al., 2002).
8 Another study found similar results, with an elevated risk for men with a maximum intensity,
9 cumulative intensity and maximum probability of exposure to TCE compared to women
10 (Nietert et al., 1998). These two studies, along with one focused exclusively on the risk of
11 scleroderma to women (Garabrant et al., 2003), were included in a meta-analysis conducted by
12 the U.S. EPA resulting in a combined estimate for "any" exposure, was OR = 2.5 (95% CI: 1.1,
13 5.4) for men and OR = 1.2 (95% CI: 0.58, 2.6) in women.
14
15 4.10.2.1.2.1.4. Respiratory toxicity. No gender susceptibility to noncancerous outcomes in the
16 respiratory tract was observed. A detailed discussion of the studies examining the respiratory
17 effects of TCE can be found in Section 4.7.
18
19 4.10.2.1.2.1.5. Reproductive toxicity. A detailed discussion of the studies examining the gender-
20 specific noncancer reproductive effects of TCE can be found in Section 4.8.1.
21 Studies examining males after exposure to TCE observed altered sperm morphology and
22 hyperzoospermia (Chia et al., 1996), altered endocrine function (Chia et al., 1997; Goh et al.,
23 1998), decreased sexual drive and function (Bardodej and Vyskocil, 1956; El Ghawabi et al.,
24 1973; Saihan et al., 1978), and altered fertility to TCE exposure. Infertility was not associated
25 with TCE exposure in other studies (Forkert et al., 2003; Sallmen et al., 1998), and sperm
26 abnormalities were not observed in another study (Rasmussen et al., 1988).
27 There is more limited evidence for reproductive toxicity in females. There are
28 epidemiological indicators of a possible effect of TCE exposure on female fertility
29 (Sallmen et al., 1995), increased rate of miscarriage (ATSDR, 2001), and menstrual cycle
30 disturbance (ATSDR, 2001; Bardodej and Vyskocil, 1956; Zielinski, 1973). In experimental
31 animals, the effects on female reproduction include evidence of reduced in vitro oocyte
32 fertilizability in rats (Berger and Horner, 2003; Wu and Berger, 2007, 2008). However, in other
33 studies that assessed reproductive outcome in female rodents (Cosby and Dukelow, 1992;
34 George et al., 1985, 1986; Manson et al., 1984), there was no evidence of adverse effects of TCE
35 exposure on female reproductive function.
36
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-581 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2.1.2.1.6. Developmental toxicity. A detailed discussion of the studies examining the
2 gender-specific noncancer developmental effects of TCE can be found in Section 4.8.3. Only
3 one study of contaminated drinking water exposure in Camp Lejeune, North Carolina observed a
4 higher risk of SGA in males (ATSDR, 1998; Sonnenfeld et al., 2001).
5
6 4.10.2.1.2.2. Gender susceptibility to cancer outcomes. A detailed discussion of the studies
7 examining the carcinogenic effects of TCE can be found on the liver in Section 4.4, on the
8 kidney in Section 4.5, in the immune system in Section 4.6.4, in the respiratory system in
9 Sections 4.7.1.2 and 4.7.3, and on the reproductive system in Section 4.8.2.
10
11 4.10.2.1.2.2.1. Liver cancer. An elevated risk of liver cancer was observed for females in both
12 human (Raaschou-Nielsen et al., 2003) and rodent (Elfarra et al., 1998) studies. In addition,
13 gallbladder cancer was significantly elevated for women (Raaschou-Nielsen et al., 2003). A
14 detailed discussion of the studies examining the gender-specific liver cancer effects of TCE can
15 be found in Section 4.4.
16
17 4.10.2.1.2.2.2. Kidney cancer. One study of occupational exposure to TCE observed an increase
18 in renal cell carcinoma for women compared to men (Dosemeci et al., 1999), but no gender
19 difference was observed in other studies (Pesch et al., 2000; Raaschou-Nielsen et al., 2003).
20 Blair et al. (1998) and Hansen et al. (2001) also present some results by sex, but both of these
21 studies have too few cases to be informative about a sex difference for kidney cancer. Exposure
22 differences between males and females in Dosemeci et al. (1999) may explain their finding.
23 These studies, however, provide little information to evaluate susceptibility between sexes
24 because of their lack of quantitative exposure assessment and lower statistical power. A detailed
25 discussion of the studies examining the gender-specific kidney cancer effects of TCE can be
26 found in Section 4.5.
27
28 4.10.2.1.2.2.3. Cancers of the immune system. Two drinking water studies suggest that there
29 may be an increase of leukemia (Cohn et al., 1994; Fagliano et al., 1990) and NHL (Cohn et al.,
30 1994) among females. An occupational study also observed an elevated risk of leukemia in
31 females (Raaschou-Nielsen et al., 2003), although study of contaminated drinking water in
32 Woburn, Massachusetts observed an increased risk of childhood leukemia in males (Costas et al.,
33 2002). A detailed discussion of the studies examining the gender-specific cancers of the immune
34 system following TCE exposure can be found in Section 4.6.4.
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-582 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2.1.2.2.4. Respiratory cancers. One study observed significantly elevated risk of lung
2 cancer following occupational TCE exposure for both men and women, although the risk was
3 found to be higher for women (Raaschou-Nielsen et al., 2003). This same study observed a
4 nonsignificant elevated risk in both men and women for laryngeal cancer, again with an
5 increased risk for women (Raaschou-Nielsen et al., 2003). Conversely, a study of Iowa residents
6 with TCE-contaminated drinking water observed a 7-fold increased annual age-adjusted
7 incidence for males compared to females (Isacson et al., 1985). However, other studies did not
8 observe a gender-related difference (ATSDR, 2003a; Blair et al., 1998; Hansen et al., 2001). A
9 detailed discussion of the studies examining the gender-specific respiratory cancers following
10 TCE exposure can be found in Sections 4.7.1.2 and 4.7.3.
11
12 4.10.2.1.2.2.5. Reproductive cancers. Breast cancer in females and prostate cancer in males was
13 reported after exposure to TCE in drinking water (Isacson et al., 1985). A statistically elevated
14 risk for cervical cancer, but not breast, ovarian or uterine cancer, was observed in women in
15 another study (Raaschou-Nielsen et al., 2003). This study also did not observe elevated prostate
16 or testicular cancer (Raaschou-Nielsen et al., 2003). A detailed discussion of the studies
17 examining the gender-specific reproductive cancers following TCE exposure can be found in
18 Section 4.8.2.
19
20 4.10.2.1.2.2.6. Other Cancers. Bladder and rectal cancer was increased in men compared to
21 women after exposure to TCE in drinking water, but no gender difference was observed for
22 colon cancer (Isacson et al., 1985). After occupational TCE exposure, bladder, stomach, colon,
23 and esophageal cancer was nonsignificantly elevated in women compared to men (Raaschou-
24 Nielsen et al., 2003).
25
26 4.10.2.2. Genetic Variability
27 Section 3.3 describes the metabolic processes involved in the metabolism of TCE.
28 Human variation in response to TCE exposure may be associated with genetic variation. TCE is
29 metabolized by both CYP and GST; therefore, it is likely that polymorphisms will alter the
30 response to exposure (Garte et al., 2001; Nakajima and Aoyama, 2000), as well as other
31 chemicals that may alter the metabolism of TCE (Lash et al., 2007). It is important to note that
32 even with a given genetic polymorphism, metabolic expression is not static, and depends on
33 lifestage (see Section 4.10.1.1.2), obesity (see Section 4.10.2.4.1), and alcohol intake (see
34 Section 4.10.2.5.1).
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-583 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2.2.1. CYPgenotypes. Variability in CYP expression occurs both within humans (Dome et
2 al., 2005) and across experimental animal species (Nakajima et al., 1993). In particular,
3 increased CYP2E1 activity may lead to increased susceptibility to TCE (Lipscomb et al., 1997).
4 The CYP2E1 *3 allele and the CYP2E1 *4 allele were more common among those who
5 developed scleroderma who were exposed to solvents including TCE (Povey et al., 2001). A
6 PBPK model of CYP2E1 expression after TCE exposure has been developed for rats and humans
7 (Yoon et al., 2007).
8 In experimental animals, toxicokinetics of TCE differed among CYP2E1 knockout and
9 wild-type mice (Kim and Ghanayem, 2006). This study found that exhalation was more
10 prevalent among the knockout mice, whereas urinary excretion was more prevalent among the
11 wild-type mice. In addition, the dose was found to be retained to a greater degree by the
12 knockout mice compared to the wild-type mice.
13 4.10.2.2.2. GST genotype. There is a possibility that GST polymorphisms could play a role in
14 variability in toxic response (Caldwell and Keshava, 2006), but this has not been sufficiently
15 tested (NRC, 2006). One study of renal cell cancer in workers exposed to TCE demonstrated a
16 significant increased for those with GSTM1+ and GSTT1+ polymorphisms, compared to a
17 negative risk for those with GSTM1- and GSTTl-polymorphisms (Briining et al., 1997).
18 However, another study did not confirm this hypothesis, observing no clear relationship between
19 GSTM1 and GSTT1 polymorphisms and renal cell carcinoma among TCE exposed individuals,
20 although they did see a possible association with the homozygous wild-type allele GSTP1*A
21 (Wiesenhiitter et al., 2007). A third study unrelated to TCE exposure found GSTT1- to be
22 associated with an increased risk of renal cell carcinoma, but no difference was seen for GSTM1
23 and GSTP1 alleles (Sweeney et al., 2000).
24
25 4.10.2.2.3. Other genotypes. Other genetic polymorphisms could play a role in variability in
26 toxic response, in particular TCE-related skin disorders. Studies have found that many TCE-
27 exposed patients diagnosed with skin conditions exhibited the slow-acetylator NAT2 genotype
28 (Huang et al., 2002; Nakajima et al., 2003); whereas there was no difference in NAT2 status for
29 those diagnosed with renal cell carcinoma (Wiesenhiitter et al., 2007). Other studies have found
30 that many TCE-exposed patients diagnosed with skin conditions expressed variant HLA alleles
31 (Li et al., 2007; Yue et al., 2007), in particular HLA-B*1301 which is more common in Asians
32 compared to whites (Cao et al., 2001; Williams et al., 2001); or TNF a-308 allele (Dai et al.,
33 2004). Also, an in vitro study of human lung adenocarcinoma cells exposed to TCE varied in
34 response based on their p53 status, with p53-wild-type cells resulting in severe cellular damage,
35 but not the p53-null cells (Chen et al., 2002).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-584 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.2.3. Race/Ethnicity
2 Different racial or ethnic groups may express metabolic enzymes in different ratios and
3 proportions due to genetic variability (Garte et al., 2001). In particular, ethnic variability in CYP
4 expression has been reported (Dome et al., 2005; McCarver et al., 1998; Parkinson et al., 2004;
5 Shimada et al., 1994; Stephens et al., 1994). It has been observed that the metabolic rate for
6 TCE may differ between the Japanese and Chinese (Inoue et al., 1989). Also, body size varies
7 among ethnic groups, and increased body size was related to increased absorption of TCE and
8 urinary excretion of TCE metabolites (Sato et al., 1991b).
9
10 4.10.2.4. Pre-Existing Health Status
11 It is known that kidney and liver diseases can affect the clearance of chemicals from the
12 body, and therefore, poor health may lead to increased half-lives for TCE and its metabolites.
13 There is some data indicating that obesity/metabolic syndrome, diabetes and hypertension may
14 increase susceptibility to TCE exposure through altered toxicokinetics. In addition, some of
15 these conditions lead to increased risk for adverse effects that have also been associated with
16 TCE exposure, though the possible interaction between TCE and known risk factors for these
17 effects is not understood.
18
19 4.10.2.4.1. Obesity and metabolic syndrome. TCE is lipophilic and stored in adipose tissue;
20 therefore, obese individuals may have an increased body burden of TCE (Clewell et al., 2000).
21 Immediately after exposure, blood concentrations are higher and urinary excretion of metabolites
22 are faster in thin men than obese men due to the storage of TCE in the fat. However, the release
23 of TCE from the fat tissue beginning three hours after exposure reverses this trend and obese
24 men have increased blood concentrations and urinary excretion of metabolites are compared to
25 thin men (Sato, 1993; Sato et al., 1991b). This study also reported that increased body size was
26 related to increased absorption and urinary excretion of TCE metabolites (Sato et al., 1991b).
27 After evaluating the relationship between mean daily uptake and mean minute volume, body
28 weight, lean body mass, and amount of adipose tissue, the variation in uptake was more closely
29 correlated with lean body mass, but not adipose tissue content (Monster et al., 1979). Thus,
30 adipose tissue may play an important role in postexposure distribution, but is not a primary
31 determinant of TCE uptake. Increased CYP2E1 expression has been observed in obese
32 individuals (McCarver et al., 1998). Accumulation into adipose tissue may prolong internal
33 exposures (Davidson and Beliles, 1991; Lash et al., 2000), as evidenced by increased durations
34 of elimination in subjects with larger body mass indices (Monster, 1979).
35 In addition, individuals with high BMI are at increased risk of some of the same health
36 effects associated with TCE exposure. For example, renal cell carcinoma, liver cancer, and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-585 DRAFT—DO NOT CITE OR QUOTE
-------
1 prostate cancer may be positively associated with BMI or obesity (Asal et al., 1988a, b;
2 Benichou et al., 1998; El-Serag and Rudolph, 2007; Wigle et al., 2008). However, whether and
3 how TCE interacts with known risk factors for such diseases is unknown, as existing
4 epidemiologic studies have only examined these factors as possible confounders for effects
5 associated with TCE, or vice versa (Charbotel et al., 2006; Krishnadasan et al., 2008).
6
7 4.10.2.4.2. Diabetes. A higher rate of diabetes in females exposed to TCE was reported in two
8 studies (Burg et al., 1995; Davis et al., 2005). Whether the TCE may have caused the diabetes or
9 the diabetes may have increased susceptibility to TCE is not clear. However, it has been
10 observed that CYP2E1 expression is increased in obese Type II diabetics (Wang et al., 2003),
11 and in poorly controlled Type I diabetics (Song et al., 1990), which may consequently alter the
12 metabolism of TCE.
13
14 4.10.2.4.3. Hypertension. One study found no difference in risk for renal cell carcinoma among
15 those diagnosed with hypertension among those living in an area with high TCE exposure;
16 however, a slightly elevated risk was seen for those being treated for hypertension (OR: 1.57,
17 95% CI: 0.90-2.72) (Charbotel et al., 2006). Unrelated to TCE exposure, hypertension has been
18 associated with increase risk of renal cell carcinoma in women (Benichou et al., 1998).
19
20 4.10.2.5. Lifestyle Factors and Nutrition Status
21 4.10.2.5.1. Alcohol intake. A number of studies have examined the interaction between TCE
22 and ethanol exposure in both humans (Bardodej and Vyskocil, 1956; Barret et al., 1984;
23 McCarver et al., 1998; Muller et al., 1975; Sato, 1993; Sato et al., 1981, 1991a; Stewart et al.,
24 1974) and experimental animals (Kaneko et al., 1994; Larson and Bull, 1989; Nakajima et al.,
25 1988, 1990, 1992b; Okino et al., 1991; Sato et al., 1980, 1983; Sato and Nakajima, 1985; White
26 and Carlson, 1981).
27 The coexposure causes metabolic inhibition of TCE in humans (Muller et al., 1975;
28 Windemuller and Ettema, 1978), male rats (Kaneko et al., 1994; Larson and Bull, 1989;
29 Nakajima et al., 1988, 1990; Nakanishi et al., 1978; Okino et al., 1991; Sato and Nakajima, 1985;
30 Sato et al., 1981), and rabbits (White and Carlson, 1981). Similarly, individuals exposed to TCE
31 reported an increase in alcohol intolerance (Bardodej and Vyskocil, 1956; Grandjean et al., 1955;
32 Rasmussen and Sabroe, 1986). Disulfiram, used to treat alcoholism, has also been found to
33 decrease the elimination of TCE and TCA (Bartonicek and Teisinger, 1962).
34 A "degreasers flush" has been described, reflecting a reddening of the face of those
35 working with TCE after drinking alcohol, and measured an elevated level of TCE in exhaled
36 breath compared to nondrinkers exposed to TCE (Stewart et al., 1974). This may be due to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-586 DRAFT—DO NOT CITE OR QUOTE
-------
1 increased CYP2E1 expression in those that consume alcohol (Caldwell et al., 2008;
2 Liangpunsakul et al., 2005; Lieber, 2004; McCarver et al., 1998; Parkinson et al., 2004;
3 Perrot et al., 1989), which has also been observed in male rats fed alcohol (Nakajima et al.,
4 1992b), although another study of male rats observed that ethanol did not decrease CYP activity
5 (Okino et al., 1991). It is important to note that there a further increased response of TCE and
6 ethanol has been reported when also combined with low fat diets or low carbohydrate diets in
7 male rats (Sato et al., 1983).
8 Since the liver is a target organ for both TCE and alcohol, decreased metabolism of TCE
9 could be related to cirrhosis of the liver as a result of alcohol abuse (McCarver et al., 1998), and
10 an in increase in clinical liver impairment along with degreasers flush has been observed
11 (Barret et al., 1984).
12 The central nervous system may also be impacted by the coexposure. Individuals
13 exposed to TCE and ethanol reported an increase in altered mood states (Reif et al., 2003),
14 decreased mental capacity as described as small increases in functional load (Windemuller and
15 Ettema, 1978), and those exposed to TCE and tetrachloroethylene who consumed alcohol had an
16 elevated color confusion index (Valic et al., 1997).
17
18 4.10.2.5.2. Tobacco smoking. Individuals who smoke tobacco may be at increased risk of the
19 health effects from TCE exposure. One study examining those living in an area with high TCE
20 exposure found an increasing trend of risk (p = 0.008) for renal cell carcinoma among smokers,
21 with the highest OR among those with >40 pack-years (OR = 3.27, 95% CI: 1.48-7.19)
22 (Charbotel et al., 2006). It has been shown that renal cell carcinoma is independently associated
23 with smoking in a dose-response manner (Yuan et al., 1998), particularly in men (Benichou et
24 al., 1998).
25 A number of factors correlated to smoking (e.g., socioeconomic status, diet, alcohol
26 consumption) may positively confound results if greater smoking rates were over-represented in
27 a cohort (Raaschou-Nielsen et al., 2003). Absence of smoking information, on the other hand,
28 could introduce a negative bias. Morgan and Cassidy (2002) noted the relatively high education
29 high income levels, and high access to health care of subjects in this study compared to the
30 averages for the county as a whole likely leads to a lower smoking rate. Garabrant et al. (1988)
31 similarly attributed their observations to negative selection bias introduced when comparison is
32 made to national mortality rats known as "the healthy worker effect."
33
34 4.10.2.5.3. Nutritional status. Malnutrition may also increase susceptibility to TCE.
35 Bioavailability of TCE after oral and intravenous exposure increased with fasting from
36 approximately 63% in nonfasted rats to greater than 90% in fasted rats, with blood levels in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-587 DRAFT—DO NOT CITE OR QUOTE
-------
1 fasted rats were elevated 2-3-fold, and increased half-life in the blood of fasted rats
2 (D'Souza et al., 1985). Food deprivation (Sato and Nakajima, 1985) and carbohydrate restriction
3 (Nakajima et al., 1982; Sato and Nakajima, 1985) enhanced metabolism of TCE in male rats, but
4 this was not observed for dietary changes in protein or fat levels (Nakajima et al., 1982).
5 Vitamin intake may also alter susceptibility to TCE. An in vitro study of cultured normal
6 human epidermal keratinocyte demonstrated an increased lipid peroxidation in a dose-dependant
7 manner after exposure to TCE, which were then attenuated by exposure to Vitamin E
8 (Ding et al., 2006).
9
10 4.10.2.5.4. Physical activity. Increased inhalation during physical activity leads increases TCE
11 concentrations in the alveoli when compared to inhalation in a resting state (Astrand, 1975).
12 Studies have examined the time course of inhaled TCE and metabolites in blood and urine in
13 individuals with different workloads (Astrand and Ovrum, 1976; Jakubowski and Wieczorek,
14 1988; Monster et al., 1976; Vesterberg et al., 1976; Vesterberg and Astrand, 1976). These
15 studies demonstrate that an increase in pulmonary ventilation increases the amount of TCE taken
16 up during exposure (Astrand and Ovrum, 1976; Jakubowski and Wieczorek, 1988;
17 Monster et al., 1976; Sato, 1993).
18 The Rocketdyne aerospace cohort exposed to TCE (and other chemicals) found a
19 protective effect with high physical activity, but only after controlling for TCE exposure and
20 socioeconomic status (OR = 0.55, 95% CI: 0.32-0.95, p trend = 0.04) (Krishnadasan et al.,
21 2008). In general, physical activity may provide a protective effect for prostate cancer
22 (Wigle et al., 2008) (see Section 4.8.3.1.1).
23
24 4.10.2.5.5. Socioeconomic status. Socioeconomic status (SES) can be an indicator for a number
25 of coexposures, such as increased tobacco smoking, poor diet, education, income, and health care
26 access, which may play a role in the results observed in the health effects of TCE exposure
27 (Morgan and Cassidy, 2002).
28 Children's exposure to TCE was measured in a low SES community, as characterized by
29 income, educational level, and receipt of free or reduced cost school meals (Sexton et al., 2005);
30 however, this study did not compare data to a higher SES community, nor examine health
31 effects.
32 An elevated risk of NHL and esophagus/adenocarcinoma after exposure to TCE was
33 observed for blue-collar workers compared to white collar and unknown SES
34 (Raaschou-Nielsen et al., 2003). Authors speculate that these results could be confounding due
35 to other related factors to SES such as smoking.
36
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-588 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.10.3. Uncertainty of Database for Susceptible Populations
2 There is some evidence that certain subpopulations may be more susceptible to exposure
3 to TCE. These subpopulations include early and later lifestages, gender, genetic polymorphisms,
4 race/ethnicity, pre-existing health status, and lifestyle factors and nutrition status. Although
5 there is more information on early life exposure to TCE than on other potentially susceptible
6 populations, there remain a number of uncertainties regarding children's susceptibility.
7 Improved PBPK modeling for using childhood parameters early lifestages as recommended by
8 the NRC (2006), and validation of these models, will aid in determining how variations in
9 metabolic enzymes affect TCE metabolism. In particular, the NRC states that it is prudent to
10 assume children need greater protection than adults—unless sufficient data are available to
11 justify otherwise (NRC, 2006).
12 More studies specifically designed to evaluate effects in early and later lifestages are
13 needed in order to more fully characterize potential life stage-related TCE toxicity. Because the
14 neurological effects of TCE constitute the most sensitive endpoints of concern for noncancer
15 effects, it is quite likely that the early lifestages may be more susceptible to these outcomes than
16 are adults. Lifestage-specific neurotoxic effects, particularly in the developing fetus, need
17 further evaluation. It is important to consider the use of age-appropriate testing for assessment of
18 these and other outcomes, both for cancer and noncancer outcomes. Data specific to the
19 carcinogenic effects of TCE exposure during the critical periods of development of experimental
20 animals and humans also are sparse.
21 There is a need to better characterize the implications of TCE exposures to susceptible
22 populations. There is suggestive evidence that there may be greater susceptibility for exposures
23 to the elderly. Gender and race/ethnic differences in susceptibility are likely due to variation in
24 physiology and exposure, and genetic variation likely has an effect on the toxicokinetics of TCE.
25 Diminished health status (e.g., impaired kidney liver or kidney), alcohol consumption, tobacco
26 smoking, and nutritional status will likely affect an individual's ability to metabolize TCE. In
27 addition, further evaluation of the effects due to coexposures to other compounds with similar or
28 different MO As need to be evaluated. Future research should better characterize possible
29 susceptibility for certain lifestages or populations.
30
31 4.11. HAZARD CHARACTERIZATION
32 4.11.1. Characterization of Noncancer Effects
33 4.11.1.1. Neurotoxicity
34 Both human and animal studies have associated TCE exposure with effects on several
35 neurological domains. The strongest neurological evidence of hazard in humans is for changes
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-589 DRAFT—DO NOT CITE OR QUOTE
-------
1 in trigeminal nerve function or morphology and impairment of vestibular function. Fewer and
2 more limited evidence exists in humans on delayed motor function, and changes in auditory,
3 visual, and cognitive function or performance. Acute and subchronic animal studies show
4 morphological changes in the trigeminal nerve, disruption of the peripheral auditory system
5 leading to permanent function impairments and histopathology, changes in visual evoked
6 responses to patterns or flash stimulus, and neurochemical and molecular changes. Additional
7 acute studies reported structural or functional changes in hippocampus, such as decreased
8 myelination or decreased excitability of hippocampal CA1 neurons, although the relationship of
9 these effects to overall cognitive function is not established. Some evidence exists for motor-
10 related changes in rats/mice exposed acutely/subchronically to TCE, but these effects have not
11 been reported consistently across all studies.
12 Epidemiologic evidence supports a relationship between TCE exposure and trigeminal
13 nerve function changes, with multiple studies in different populations reporting abnormalities in
14 trigeminal nerve function in association with TCE exposure (Barret et al., 1982, 1984, 1987;
15 Feldman et al., 1988, 1992; Kilburn and Warshaw, 1993; Ruitjen et al., 2001; Kilburn, 2002a;
16 Mhiri et al., 2004). Of these, two well conducted occupational cohort studies, each including
17 more than 100 TCE-exposed workers without apparent confounding from multiple solvent
18 exposures, additionally reported statistically significant dose-response trends based on ambient
19 TCE concentrations, duration of exposure, and/or urinary concentrations of the TCE metabolite
20 TCA (Barret et al., 1984; Barret et al., 1987). Limited additional support is provided by a
21 positive relationship between prevalence of abnormal trigeminal nerve or sensory function and
22 cumulative exposure to TCE (most subjects) or CFC-113 (<25% of subjects) (Rasmussen et al.,
23 1993c). Test for linear trend in this study was not statistically significant and may reflect
24 exposure misclassification since some subjects included in this study did not have TCE exposure.
25 The lack of association between TCE exposure and overall nerve function in three small studies
26 (trigeminal: El-Ghawabi et al., 1973; ulnar and medial: Triebig et al., 1982, 1983) does not
27 provide substantial evidence against a causal relationship between TCE exposure and trigeminal
28 nerve impairment because of limitations in statistical power, the possibility of exposure
29 misclassification, and differences in measurement methods. Laboratory animal studies have also
30 shown TCE-induced changes in the trigeminal nerve. Although one study reported no significant
31 changes in trigeminal somatosensory evoked potential in rats exposed to TCE for 13 weeks
32 (Albee et al., 2006), there is evidence of morphological changes in the trigeminal nerve
33 following short-term exposures in rats (Barret et al., 1991, 1992).
34 Human chamber, occupational, geographic based/drinking water, and laboratory animal
35 studies clearly established TCE exposure causes transient impairment of vestibular function.
36 Subjective symptoms such as headaches, dizziness, and nausea resulting from occupational
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-590 DRAFT—DO NOT CITE OR QUOTE
-------
1 (Granjean et al., 1955; Liu et al., 1988; Rasmussen and Sabroe, 1986; Smith et al., 1970),
2 environmental (Hirsch et al., 1996), or chamber exposures (Stewart et al., 1970; Smith et al.,
3 1970) have been reported extensively. A few laboratory animal studies have investigated
4 vestibular function, either by promoting nystagmus or by evaluating balance (Niklasson et al.,
5 1993; Tham et al., 1979; Tham et al., 1984; Umezu et al., 1997).
6 In addition, mood disturbances have been reported in a number of studies, although these
7 effects also tend to be subjective and difficult to quantify (Gash et al., 2007; Kilburn and
8 Warshaw, 1993; Kilburn, 2002a, 2002b; McCunney et al., 1988; Mitchell et al., 1969;
9 Rasmussen and Sabroe, 1986; Troster and Ruff, 1990), and a few studies have reported no
10 effects from TCE on mood (Reif et al., 2003; Triebig et al., 1976, 1977a). Few comparable
11 mood studies are available in laboratory animals, although both Moser et al. (2003) and Albee et
12 al. (2006) report increases in handling reactivity among rats exposed to TCE. Finally,
13 significantly increased number of sleep hours was reported by Arito et al. (1994) in rats exposed
14 via inhalation to 50-300-ppm TCE for 8 hours/day for 6 weeks.
15 Four epidemiologic studies of chronic exposure to TCE observed disruption of auditory
16 function. One large occupational cohort study showed a statistically significant difference in
17 auditory function with cumulative exposure to TCE or CFC-113 as compared to control groups
18 after adjustment for possible confounders, as well as a positive relationship between auditory
19 function and increasing cumulative exposure (Rasmussen et al., 1993b). Of the three studies
20 based on populations from ATSDR's TCE Subregistry from the National Exposure Registry,
21 more limited than Rasmussen et al. (1993b) due to inferior exposure assessment, Burg et al.
22 (1995) and Burg and Gist (1999) reported a higher prevalence of self-reported hearing
23 impairments. The third study reported that auditory screening revealed abnormal middle ear
24 function in children less than 10 years of age, although a dose-response relationship could not be
25 established and other tests did not reveal differences in auditory function (ATSDR, 2003a).
26 Further evidence for these effects is provided by numerous laboratory animal studies
27 demonstrating that high dose subacute and subchronic TCE exposure in rats disrupts the auditory
28 system leading to permanent functional impairments and histopathology.
29 Studies in humans exposed under a variety of conditions, both acutely and chronically,
30 report impaired visual functions such as color discrimination, visuospatial learning tasks, and
31 visual depth perception in subjects with TCE exposure. Abnormalities in visual depth perception
32 were observed with a high acute exposure to TCE under controlled conditions (Vernon and
33 Ferguson, 1969). Studies of lower TCE exposure concentrations also observed visuofunction
34 effects. One occupational study (Rasmussen et al., 1993b) reported a statistically significant
35 positive relationship between cumulative exposure to TCE or CFC-113 and visual gestalts
36 learning and retention among Danish degreasers. Two studies of populations living in a
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-591 DRAFT—DO NOT CITE OR QUOTE
-------
1 community with drinking water containing TCE and other solvents furthermore suggested
2 changes in visual function (Kilburn et al., 2002a; Reif et al., 2003). These studies used more
3 direct measures of visual function as compared to Rasmussen et al. (1993b), but their exposure
4 assessment is more limited because TCE exposure is not assigned to individual subjects
5 (Kilburn et al., 2002a), or because there are questions regarding control selection (Kilburn et al.,
6 2002a) and exposure to several solvents (Kilburn et al., 2002a; Reif et al., 2003).
7 Additional evidence of effects of TCE exposure on visual function is provided by a
8 number of laboratory animal studies demonstrating that acute or subchronic TCE exposure
9 causes changes in visual evoked responses to patterns or flash stimulus (Boyes et al., 2003, 2005;
10 Blain et al., 1994). Animal studies have also reported that the degree of some effects is
11 correlated with simultaneous brain TCE concentrations (Boyes et al., 2003, 2005) and that, after
12 a recovery period, visual effects return to control levels (Blain et al., 1994; Rebert et al., 1991).
13 Overall, the human and laboratory animal data together suggest that TCE exposure can cause
14 impairment of visual function, and some animal studies suggest that some of these effects may
15 be reversible with termination of exposure.
16 Studies of human subjects exposed to TCE either acutely in chamber studies or
17 chronically in occupational settings have observed deficits in cognition. Five chamber studies
18 reported statistically significant deficits in cognitive performance measures or outcome measures
19 suggestive of cognitive effects (Stewart et al., 1970; Gamberale et al., 1976; Triebig et al., 1976,
20 1977a; Gamberale et al., 1977). Danish degreasers with high cumulative exposure to TCE or
21 CFC-113 had a high risk (OR = 13.7, 95% CI: 2.0-92.0) for psychoorganic syndrome
22 characterized by cognitive impairment, personality changes, and reduced motivation, vigilance,
23 and initiative compared to workers with low cumulative exposure. Studies of populations living
24 in a community with contaminated groundwater also reported cognitive impairments (Kilburn
25 and Warshaw, 1993; Kilburn, 2002a), although these studies carry less weight in the analysis
26 because TCE exposure is not assigned to individual subjects and their methodological design is
27 weaker.
28 Laboratory studies provide some additional evidence for the potential for TCE to affect
29 cognition, although the predominant effect reported has been changes in the time needed to
30 complete a task, rather than impairment of actual learning and memory function (Kulig et al.,
31 1987; Kishi et al., 1993; Umezu et al., 1997). In addition, in laboratory animals, it can be
32 difficult to distinguish cognitive changes from motor-related changes. However, several studies
33 have reported structural or functional changes in the hippocampus, such as decreased
34 myelination (Issacson et al., 1990; Isaacson and Taylor, 1989) or decreased excitability of
35 hippocampal CA1 neurons (Ohta et al., 2001), although the relationship of these effects to
36 overall cognitive function is not established.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-592 DRAFT—DO NOT CITE OR QUOTE
-------
1 Two studies of TCE exposure, one chamber study of acute exposure duration and one
2 occupational study of chronic duration, reported changes in psychomotor responses. The
3 chamber study of Gamberale et al. (1976) reported a dose-related decrease in performance in a
4 choice reaction time test in healthy volunteers exposed to 100 and 200-ppm TCE for 70 minutes
5 as compared to the same subjects without exposure. Rasmussen et al. (1993c) reported a
6 statistically significant association with cumulative exposure to TCE or CFC-113 and
7 dyscoordination trend among Danish degreasers. Observations in a third study (Gun et al., 1978)
8 are difficult to judge given the author's lack of statistical treatment of data. In addition, Gash et
9 al. (2007) reported that 14 out of 30 TCE-exposed workers exhibited significantly slower fine
10 motor hand movements as measured through a movement analysis panel test. Studies of
11 population living in communities with TCE and other solvents detected in groundwater supplies
12 reported significant delays in simple and choice reaction times in individuals exposed to TCE in
13 contaminated groundwater as compared to referent groups (Kilburn, 2002a; Kilburn and
14 Warshaw, 1993; Kilburn and Thornton, 1996). Observations in these studies are more uncertain
15 given questions of the representativeness of the referent population, lack of exposure assessment
16 to individual study subjects, and inability to control for possible confounders including alcohol
17 consumption and motivation. Finally, in a presentation of 2 case reports, decrements in motor
18 skills as measured by the grooved pegboard and finger tapping tests were observed (Troster and
19 Ruff, 1990).
20 Laboratory animal studies of acute or subchronic exposure to TCE observed psychomotor
21 effects, such as loss of righting reflex (Umezu et al., 1997; Shih et al., 2001) and decrements in
22 activity, sensory-motor function, and neuromuscular function (Kishi et al., 1993; Moser et al.,
23 1995; Moser et al., 2003). However, two studies also noted an absence of significant changes in
24 some measures of psychomotor function (Kulig et al., 1987; Albee et al., 2006). In addition, less
25 consistent results have been reported with respect to locomotor activity in rodents. Some studies
26 have reported increased locomotor activity after an acute i.p. dosage (Wolff and Siegmund,
27 1978) or decreased activity after acute or short term oral gavage dosing (Moser et al., 1995,
28 2003). No change in activity was observed following exposure through drinking water
29 (Waseem et al., 2001), inhalation (Kulig et al., 1987) or orally during the neurodevelopment
30 period (Fredriksson et al., 1993).
31 Several neurochemical and molecular changes have been reported in laboratory
32 investigations of TCE toxicity. Kjellstrand et al. (1987) reported inhibition of sciatic nerve
33 regeneration in mice and rats exposed continuously to 150-ppm TCE via inhalation for 24 days.
34 Two studies have reported changes in GABAergic and glutamatergic neurons in terms of GABA
35 or glutamate uptake (Briving et al., 1986) or response to GABAergic antagonistic drugs
36 (Shih et al., 2001) as a result of TCE exposure, with the Briving et al. (1986) conducted at
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-593 DRAFT—DO NOT CITE OR QUOTE
-------
1 50 ppm for 12 months. Although the functional consequences of these changes is unclear,
2 Tham et al. (1979, 1984) described central vestibular system impairments as a result of TCE
3 exposure that may be related to altered GAB Aergic function. In addition, several in vitro studies
4 have demonstrated that TCE exposure alters the function of inhibitory ion channels such as
5 receptors for GABAA glycine, and serotonin (Krasowski and Harrison, 2000; Beckstead et al.,
6 2000; Lopreato et al., 2003) or of voltage-sensitive calcium channels (Shafer et al., 2005).
7
8 4.11.1.2. Kidney Toxicity
9 There are few human data pertaining to TCE-related noncancer kidney toxicity.
10 Observation of elevated excretion of urinary proteins in the available studies (Rasmussen et al.,
11 1993a; Briining et al., 1999a, b; Bolt et al., 2004; Green et al., 2004) indicates the occurrence of
12 a toxic insult among TCE-exposed subjects compared to unexposed controls. Two studies are of
13 subjects with previously diagnosed kidney cancer (Briining et al., 1999a; Bolt et al., 2004), while
14 subjects in the other studies are disease free. Urinary proteins are considered nonspecific
15 markers of nephrotoxicity and include al-microglobulin, albumin, and NAG (Price et al., 1996;
16 Lybarger et al., 1999; Price et al., 1999). Four studies measure al-microglobulin with elevated
17 excretion observed in the German studies (Briining et al., 1999a, b; Bolt et al., 2004) but not
18 Green et al. (2004). However, Rasmussen et al. (1993a) reported a positive relationship between
19 increasing urinary NAG, another nonspecific marker of tubular toxicity, and increasing exposure
20 duration; and Green et al. (2004) found statistically significant group mean differences in NAG.
21 Observations in Green et al. (2004) provide evidence of tubular damage among workers exposed
22 to trichloroethylene at current occupational levels. Elevated excretion of NAG has also been
23 observed with acute TCE poisoning (Carrieri et al., 2007). Some support for TCE nephrotoxicity
24 in humans is provided by a study of end-stage renal disease in a cohort of workers at Hill Air
25 Force Base (Radican et al., 2006), although subjects in this study were exposed to hydrocarbons,
26 JP-4 gasoline, and solvents in addition to TCE, including 1,1,1-trichloroethane.
27 Laboratory animal and in vitro data provide additional support for TCE nephrotoxicity.
28 Multiple studies with both gavage and inhalation exposure show that TCE causes renal toxicity
29 in the form of cytomegaly and karyomegaly of the renal tubules in male and female rats and
30 mice (summarized in Section 4.4.4). Further studies with TCE metabolites have demonstrated a
31 potential role for DCVC, TCOH, and TCA in TCE-induced nephrotoxicity. Of these, available
32 data suggest that DCVC induced renal effects most like those of TCE and is formed in sufficient
33 amounts following TCE exposure to account for these effects. TCE or DCVC have also been
34 shown to be cytotoxic to primary cultures of rat and human renal tubular cells (Cummings et al.,
35 2000a, b; Cummings and Lash, 2000).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-594 DRAFT—DO NOT CITE OR QUOTE
-------
1 Overall, multiple lines of evidence support the conclusion that TCE causes nephrotoxicity
2 in the form of tubular toxicity, mediated predominantly through the TCE GSH conjugation
3 product DCVC.
4
5 4.11.1.3. Liver Toxicity
6 Few studies on liver toxicity and TCE exposure are found in humans. Of these, three
7 studies reported significant changes in serum liver function tests, widely used in clinical settings
8 in part to identify patients with liver disease, in metal degreasers whose TCE exposure was
9 assessed using urinary trichloro-compounds as a biomarker (Nagaya et al., 1993; Rasmussen et
10 al., 1993; Xu et al., 2009). Two additional studies reported plasma or serum bile acid changes
11 (Neghab et al., 1997; Driscoll et al., 1992). One study of subjects from the TCE subregistry of
12 ATSDR' s National Exposure Registry is suggestive of liver disorders but limitations preclude
13 inferences whether TCE caused these conditions is not possible given the study's limitations
14 (Davis et al., 2005). Furthermore, a number of case reports exist of liver toxicity including
15 hepatitis accompanying immune-related generalized skin diseases described as a variation of
16 erythema multiforme, Stevens-Johnson syndrome, toxic epidermal necrolysis patients, and
17 hypersensitivity syndrome (Kamijima et al., 2007) in addition to jaundice, hepatomegaly,
18 hepatosplenomegaly, and liver failure TCE-exposed workers (Thiele, 1982; Huang et al., 2002).
19 Cohort studies have examined cirrhosis mortality and either TCE exposure (Blair et al., 1989;
20 Morgan et al., 1998; Boice et al., 1999, 2006; Garabrant et al., 1988; Blair et al., 1998; Ritz et al.,
21 1999; ATSDR, 2004; Radican et al., 2008) or solvent exposure (Leigh and Jiang, 1993), but are
22 greatly limited by their use of death certificates where there is a high degree (up to 50%) of
23 underreporting (Blake et al., 1988), so these null findings do not rule out an effect of TCE on
24 cirrhosis. Overall, while there some evidence exists of liver toxicity as assessed from liver
25 function tests, the data are inadequate for making conclusions regarding causality.
26 In laboratory animals, TCE exposure is associated with a wide array of hepatotoxic
27 endpoints. Like humans, laboratory animals exposed to TCE have been observed to have
28 increased serum bile acids (Bai et al., 1992b; Neghab et al., 1997), although the toxicologic
29 importance of this effect is unclear. Most other effects in laboratory animals have not been
30 studied in humans, but nonetheless provide evidence that TCE exposure leads to hepatotoxicity.
31 These effects include increased liver weight, small transient increases in DNA synthesis,
32 cytomegaly in the form of "swollen" or enlarged hepatocytes, increased nuclear size probably
33 reflecting polyploidization, and proliferation of peroxisomes. Liver weight increases
34 proportional to TCE dose are consistently reported across numerous studies and appear to be
35 accompanied by periportal hepatocellular hypertrophy (Nunes et al., 2001; Tao et al., 2000,
36 Tucker et al., 1982; Goldsworthy and Popp, 1987; Elcombe et al., 1985; Dees and Travis, 1993;
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-595 DRAFT—DO NOT CITE OR QUOTE
-------
1 Nakajima et al., 2000; Berman et al., 1995; Melnick et al., 1987; Laughter et al., 2004;
2 Merrick et al., 1989; Goel et al., 1992; Kjellstand et al., 1981, 1983a, b; Buben and O'Flaherty,
3 1985). There is also evidence of increased DNA synthesis in a small portion of hepatocytes at
4 around 10 days in vivo exposure (Mirsalis et al., 1989; Elcombe et al., 1985; Dees and Travis,
5 1993; Channel et al., 1998). The lack of correlation of hepatocellular mitotic figures with whole
6 liver DNA synthesis or DNA synthesis observed in individual hepatocytes (Elcombe et al., 1985;
7 Dees and Travis, 1993) supports the conclusions that cellular proliferation is not the predominant
8 cause of increased DNA synthesis and that nonparenchymal cells may also contribute to such
9 synthesis. Indeed, nonparenchymal cell activation or proliferation has been noted in several
10 studies (Kjellstrand et al., 1983b; Goel et al., 1992). Moreover, the histological descriptions of
11 TCE-exposed livers are consistent with and, in some cases, specifically note increased
12 polyploidy (Buben and O'Flaherty, 1985). Interestingly, changes in TCE-induced hepatocellular
13 ploidy, as indicated by histological changes in nuclei, have been noted to remain after the
14 cessation of exposure (Kjellstrand et al., 1983a). In regard to apoptosis, TCE has been reported
15 either to have no effect or to cause a slight increase at high doses (Dees and Travis, 1993;
16 Channel et al., 1998). Some studies have also noted effects from dosing vehicle alone (such as
17 corn oil, in particular) not only on liver pathology, but also on DNA synthesis (Merrick et al.,
18 1989; Channel et al., 1998). Available data also suggest that TCE does not induce substantial
19 cytotoxicity, necrosis, or regenerative hyperplasia, as only isolated, focal necroses and mild to
20 moderate changes in serum and liver enzyme toxicity markers having been reported
21 (Elcombe et al., 1985; Dees and Travis, 1993; Channel et al., 1998). Data on peroxisome
22 proliferation, along with increases in a number of associated biochemical markers, show effects
23 in both mice and rats (Elcombe et al., 1985; Channel et al., 1998; Goldsworthy and Popp, 1987).
24 These effects are consistently observed across rodent species and strains, although the degree of
25 response at a given mg/kg/d dose appears to be highly variability across strains, with mice on
26 average appearing to be more sensitive.
27 While it is likely that oxidative metabolism is necessary for TCE-induced effects in the
28 liver, the specific metabolite or metabolites responsible is less clear. TCE, TCA, and DCA
29 exposures have all been associated with induction of changes in liver weight, DNA synthesis,
30 and peroxisomal enzymes. The available data strongly support TCA not being the sole or
31 predominant active moiety for TCE-induced liver effects, particularly with respect to
32 hepatomegaly. In particular, TCE and TCA dose-response relationships are quantitatively
33 inconsistent, for TCE leads to greater increases in liver/body weight ratios that expected from
34 predicted rates of TCA production (see analysis in Section 4.5.6.2.1). In fact, above a certain
35 dose of TCE, liver/body weight ratios are greater than that observed under any conditions studied
36 so far for TCA. Histological changes and effects on DNA synthesis are generally consistent with
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-596 DRAFT—DO NOT CITE OR QUOTE
-------
1 contributions from either TCA or DC A, with a degree of polyploidization, rather than cell
2 proliferation, likely to be significant for TCE, TCA, and DCA.
3 Overall, TCE, likely through its oxidative metabolites, clearly leads to liver toxicity in
4 laboratory animals, with mice appearing to be more sensitive than other laboratory animal
5 species, but there is only limited epidemiologic evidence of hepatotoxicity being associated with
6 TCE exposure.
7
8 4.11.1.4. Immunotoxicity
9 Studies in humans provide evidence of associations between TCE exposure and a number
10 of immunotoxicological endpoints. The relation between systemic autoimmune diseases, such as
11 scleroderma, and occupational exposure to TCE has been reported in several recent studies. A
12 meta-analysis of scleroderma studies (Diot et al., 2002; Garabrant et al., 2003; Nietert et al.,
13 1998) conducted by the U.S. EPA resulted in a statistically significant combined odds ratio for
14 any exposure in men (OR: 2.5, 95% CI: 1.1, 5.4), with a lower relative risk seen in women (OR:
15 1.2, 95% CI: 0.58, 2.6). The incidence of systemic sclerosis among men is very low
16 (approximately 1 per 100,000 per year), and is approximately 10 times lower than the rate seen
17 in women (Cooper and Stroehla, 2003). Thus, the human data at this time do not allow
18 determination of whether the difference in effect estimates between men and women reflects the
19 relatively low background risk of scleroderma in men, gender-related differences in exposure
20 prevalence or in the reliability of exposure assessment (Messing et al., 2003), a gender-related
21 difference in susceptibility to the effects of TCE, or chance. Changes in levels of inflammatory
22 cytokines were reported in an occupational study of degreasers exposed to TCE (lavicoli et al.,
23 2005) and a study of infants exposed to TCE via indoor air (Lehmann et al., 2001, 2002).
24 Experimental studies provide additional support for these effects. Numerous studies have
25 demonstrated accelerated autoimmune responses in autoimmune-prone mice (Cai et al., 2008;
26 Blossom et al., 2007, 2004; Griffin et al., 2000a, b). With shorter exposure periods, effects
27 include changes in cytokine levels similar to those reported in human studies. More severe
28 effects, including autoimmune hepatitis, inflammatory skin lesions, and alopecia, were manifest
29 at longer exposure periods, and interestingly, these effects differ somewhat from the "normal"
30 expression in these mice. Immunotoxic effects, including increases in anti-ds DNA antibodies in
31 adult animals, decreased thymus weights, and decreased plaque forming cell response with
32 prenatal and neonatal exposure, have been also reported in B6C3F1 mice, which do not have a
33 known particular susceptibility to autoimmune disease (Gilkeson et al., 2004; Keil et al., 2009;
34 Peden-Adams et al., 2006). Recent mechanistic studies have focused on the roles of various
35 measures of oxidative stress in the induction of these effects by TCE (Wang et al., 2008, 2007b).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-597 DRAFT—DO NOT CITE OR QUOTE
-------
1 There have been a large number of case reports of a severe hypersensitivity skin disorder,
2 distinct from contact dermatitis and often accompanied by hepatitis, associated with occupational
3 exposure to TCE, with prevalences as high as 13% of workers in the same location
4 (Kamijima et al., 2008, 2007). Evidence of a treatment-related increase in delayed
5 hypersensitivity response accompanied by hepatic damage has been observed in guinea pigs
6 following intradermal injection (Tang et al., 2008, 2002), and hypersensitivity response was also
7 seen in mice exposed via drinking water pre- and postnatally (gestation Day 0 through to
8 8 weeks of age) (Peden-Adams et al., 2006).
9 Human data pertaining to TCE-related immunosuppression resulting in an increased risk
10 of infectious diseases is limited to the report of an association between reported history of
11 bacteria of viral infections in Woburn, Massachusetts (Lagakos, 1986). Evidence of localized
12 immunosuppression, as measured by pulmonary response to bacterial challenge (i.e., risk of
13 Streptococcal pneumonia-related mortality and clearance of Klebsiella bacteria) was seen in an
14 acute exposure study in CD-I mice (Aranyi et al., 1986). A 4-week inhalation exposure in
15 Sprague-Dawley rats reported a decrease in plaque forming cell response at exposures of
16 1,000 ppm (Woolhiser et al., 2006).
17 Overall, the human and animal studies of TCE and immune-related effects provide strong
18 evidence for a role of TCE in autoimmune disease and in a specific type of generalized
19 hypersensitivity syndrome, while there are less data pertaining to immunosuppressive effects.
20
21 4.11.1.5. Respiratory Tract Toxicity
22 There are very limited human data on pulmonary toxicity and TCE exposure. Two recent
23 reports of a study of gun manufacturing workers reported asthma-related symptoms and lung
24 function decrements associated with solvent exposure (Cakmak et al., 2004; Saygun et al., 2007),
25 but these studies are limited by multiple solvent exposures and the significant effect of smoking
26 on pulmonary function. Laboratory studies in mice and rats have shown toxicity in the bronchial
27 epithelium, primarily in Clara cells, following acute exposures to TCE by inhalation (see
28 Section 4.7.2.1.1). A few studies of longer duration have reported more generalized toxicity,
29 such as pulmonary fibrosis 90 days after a single 2,000 mg/kg i.p. dose in mice and pulmonary
30 vasculitis after 13-week oral gavage exposures to 2,000 mg/kg/d in rats (Forkert and Forkert,
31 1994; NTP, 1990). However, respiratory tract effects were not reported in other longer-term
32 studies. Acute pulmonary toxicity appears to be dependent on oxidative metabolism, although
33 the particular active moiety is not known. While earlier studies implicated chloral produced in
34 situ by CYP enzymes in respiratory tract tissue was responsible for toxicity (reviewed in Green,
35 2000), the evidence is inconsistent, and several other possibilities are viable. First, substantial
36 "accumulation" of chloral is unlikely, as it is likely either to be rapidly converted to TCOH in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-598 DRAFT—DO NOT CITE OR QUOTE
-------
1 respiratory tract tissue or to diffuse rapidly into blood and be converted to TCOH in erythrocytes
2 or the liver. Conversely, a role for systemically produced oxidative metabolites cannot be
3 discounted, as CH and TCOH in blood have both been reported following inhalation dosing in
4 mice. In addition, a recent study reported dichloroacetyl chloride protein adducts in the lungs of
5 mice to which TCE was administered by i.p. injection, suggesting dichloroacetyl chloride, which
6 is not believed to be derived from chloral, may also contribute to TCE respiratory toxicity.
7 Although humans appear to have lower overall capacity for enzymatic oxidation in the lung
8 relative to mice, CYP enzymes do reside in human respiratory tract tissue, suggesting that,
9 qualitatively, the respiratory tract toxicity observed in rodents is biologically plausible in
10 humans. However, quantitative estimates of differential sensitivity across species due to
11 respiratory metabolism are highly uncertain due to limited data. Therefore, overall, data are
12 suggestive of TCE causing respiratory tract toxicity, based primarily on short-term studies in
13 mice and rats, and no data suggest that such hazards would be biologically precluded in humans.
14
15 4.11.1.6. Reproductive Toxicity
16 Reproductive toxicity related to TCE exposure has been evaluated in human and
17 experimental animal studies for effects in males and females. Only a limited number of studies
18 have examined whether TCE causes female reproductive toxicity. Epidemiologic studies have
19 identified possible associations of TCE exposure with effects on female fertility (Sallmen et al.,
20 1995; ATSDR, 2001) and with menstrual cycle disturbances (ATSDR, 2001; Bardodej and
21 Vyskocil, 1956; Sagawa et al., 1973; Zielinski, 1973). Reduced in vitro oocyte fertilizability has
22 been reported as a result of TCE exposure in rats (Berger and Horner, 2003; Wu and Berger,
23 2007), but a number of other laboratory animal studies did not report adverse effects on female
24 reproductive function (Cosby and Dukelow, 1992; George et al., 1985, 1986; Manson et al.,
25 1984). Overall, there are inadequate data to conclude whether adverse effects on human female
26 reproduction are caused by TCE.
27 By contrast, a number of human and laboratory animal studies suggest that TCE exposure
28 has the potential for male reproductive toxicity. In particular, human studies have reported TCE
29 exposure to be associated, in several cases statistically-significantly, with increased sperm
30 density and decreased sperm quality (Chia et al., 1996; Rasmussen et al., 1988), altered sexual
31 drive or function (El Gawabi et al., 1973; Saihan et al., 1978; Bardodej and Vyskocil, 1956), or
32 altered serum endocrine levels (Chia et al., 1997; Goh et al., 1998). In addition, three studies
33 that reported measures of fertility did not or could not report changes associated with TCE
34 exposure (ATSDR, 2001; Forkert et al., 2003; Sallmen et al., 1998), although the statistical
35 power of these studies is quite limited. Further evidence of similar effects is provided by several
36 laboratory animal studies that reported effects on sperm (Kumar et al., 2000a, b, 2001;
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-599 DRAFT—DO NOT CITE OR QUOTE
-------
1 George et al., 1985; Land et al., 1981; Veeramachaneni et al., 2001), libido/copulatory behavior
2 (George et al., 1986; Zenick et al., 1984; Veeramachaneni et al., 2001), and serum hormone
3 levels (Kumar et al., 2000b; Veeramachaneni et al., 2001). As with the human database, some
4 studies that assessed sperm measures did not report treatment-related alterations (Cosby and
5 Dukelow, 1992; Xu et al., 2004; Zenick et al., 1984; George et al., 1986). Additional adverse
6 effects on male reproduction have also been reported, including histopathological lesions in the
7 testes or epididymides (George et al., 1986; Kumar et al., 2000a, 2001; Forkert et al., 2002;
8 Kan et al., 2007) and altered in vitro sperm-oocyte binding or in vivo fertilization due to TCE or
9 metabolites (Xu et al., 2004; DuTeaux et al., 2004b). While reduced fertility in rodents was only
10 observed in one study (George et al., 1986), this is not surprising given the redundancy and
11 efficiency of rodent reproductive capabilities. Furthermore, while George et al. (1986) proposed
12 that the adverse male reproductive outcomes observed in rats were due to systemic toxicity, the
13 database as a whole suggests that TCE does induce reproductive toxicity independent of
14 systemic effects. Therefore, overall, the human and laboratory animal data together support the
15 conclusion that TCE exposure poses a potential hazard to the male reproductive system.
16
17 4.11.1.7. Developmental Toxicity
18 The relationship between TCE exposure (direct or parental) and adverse developmental
19 outcomes has been investigated in a number of epidemiologic and laboratory animal studies.
20 Prenatal effects examined include death (spontaneous abortion, perinatal death, pre- or
21 postimplantation loss, resorptions), decreased growth (low birth weight, small for gestational
22 age, intrauterine growth restriction, decreased postnatal growth), and congenital malformations,
23 in particular eye and cardiac defects. Postnatal developmental outcomes examined include
24 growth and survival, developmental neurotoxicity, developmental immunotoxicity, and
25 childhood cancers.
26 A few epidemiological studies have reported associations between parental exposure to
27 TCE and spontaneous abortion or perinatal death (Taskinen et al., 1994; Windham et al., 1991;
28 ATSDR, 2001), although other studies reported mixed or null findings (ATSDR, 2006, 2008;
29 Bove, 1996; Bove et al., 1995; Goldberg et al., 1990; Lagakos et al., 1986; Lindbohm et al.,
30 1990; Taskinen et al., 1989). Studies examining associations between TCE exposure and
31 decreased birth weight or small for gestational age have reported small, often nonstatistically
32 significant, increases in risk for these effects (ATSDR, 1998, 2006, 2008; Windham et al., 1991).
33 However, other studies observed mixed or no association (Bove, 1996; Bove et al., 1995;
34 Lagakos et al., 1986; Rodenbeck et al., 2000). While comprising both occupational and
35 environmental exposures, these studies are overall not highly informative due to their small
36 numbers of cases and limited exposure characterization or to the fact that exposures to mixed
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-600 DRAFT—DO NOT CITE OR QUOTE
-------
1 solvents were involved. However, a number of laboratory animal studies show analogous effects
2 of TCE exposure in rodents. In particular, pre- or postimplantation losses, increased resorptions,
3 perinatal death, and decreased birth weight have been reported in multiple well-conducted
4 studies in rats and mice (Healy et al., 1982; Kumar et al., 2000a; George et al., 1985, 1986;
5 Narotsky et al., 1995; Narotsky and Kavlock, 1995). Interestingly, the rat studies reporting these
6 effects used Fischer 344 or Wistar rats, while several other studies, all of which used Sprague-
7 Dawley rats, reported no increased risk in these developmental measures (Carney et al., 2006;
8 Hardin et al., 1981; Schwetz et al., 1975). Overall, based on weakly suggestive epidemiologic
9 data and fairly consistent laboratory animal data, it can be concluded that TCE exposure poses a
10 potential hazard for prenatal losses and decreased growth or birth weight of offspring.
11 Epidemiologic data provide some support for the possible relationship between maternal
12 TCE exposure and birth defects in offspring, in particular cardiac defects. Other developmental
13 outcomes observed in epidemiology and experimental animal studies include an increase in total
14 birth defects (AZ DHS, 1988; ATSDR, 2001), CNS defects (ATSDR, 2001; Bove, 1996;
15 Bove et al., 1995; Lagakos et al., 1986), oral cleft defects (Bove, 1996; Bove et al., 1995;
16 Lagakos et al., 1986; Lorente et al., 2000), eye/ear defects (Lagakos et al., 1986; Narotsky et al.,
17 1995; Narotsky and Kavlock, 1995), kidney/urinary tract disorders (Lagakos et al., 1986),
18 musculoskeletal birth anomalies (Lagakos et al., 1986), lung/respiratory tract disorders
19 (Lagakos et al., 1986; Das and Scott, 1994), and skeletal defects (Healy et al., 1982).
20 Occupational cohort studies, while not consistently reporting positive results, are generally
21 limited by the small number of observed or expected cases of birth defects (Lorente et al., 2000;
22 Tola et al., 1980; Taskinen et al., 1989).
23 While only one of the epidemiological studies specifically reported observations of eye
24 anomalies (Lagakos et al., 1986), studies in rats have identified increases in the incidence of fetal
25 eye defects following oral exposures during the period of organogenesis with TCE
26 (Narotsky et al., 1995; Narotsky and Kavlock, 1995) or its oxidative metabolites DC A and TCA
27 (Smith et al., 1989, 1992; Warren et al., 2006). No other developmental or reproductive toxicity
28 studies identified abnormalities of eye development following TCE exposures, which may have
29 been related to the administered dose or other aspects of study design (e.g., level of detail applied
30 to fetal ocular evaluation). Overall, the study evidence suggests a potential for the disruption of
31 ocular development by exposure to TCE and its oxidative metabolites.
32 The epidemiological studies, while individually limited, as a whole show relatively
33 consistent elevations, some of which were statistically significant, in the incidence of cardiac
34 effects in TCE-exposed populations compared to reference groups (ATSDR, 2001, 2006, 2008;
35 Bove et al., 1995; Bove, 1996; Goldberg et al., 1990; Yauck et al., 2004). Interestingly,
36 Goldberg et al. (1990) noted that the odds ratio for congenital heart disease in offspring declined
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-601 DRAFT—DO NOT CITE OR QUOTE
-------
1 from 3-fold to no difference as compared to controls after TCE-contaminated drinking water
2 wells were closed, suggestive of a causal relationship. However, this study reported no
3 significant differences in cardiac lesions between exposed and nonexposed groups
4 (Goldberg et al., 1990). One additional community study reported that, among the 5 cases of
5 cardiovascular anomalies, there was no significant association with TCE (Lagakos et al., 1986),
6 but due to the small number of cases this does not support an absence of effect. In laboratory
7 animal models, avian studies were the first to identify adverse effects of TCE exposure on
8 cardiac development, and the initial findings have been confirmed multiple times (Bross et al.,
9 1983; Loeber et al., 1988; Boyer et al., 2000; Drake et al., 2006a, b; Mishima et al., 2006;
10 Rufer et al., 2008). Additionally, administration of TCE and TCE metabolites TCA and DCA in
11 maternal drinking water during gestation has been reported to induce cardiac malformations in
12 rat fetuses (Dawson et al., 1990, 1993; Johnson et al., 1998a, b, 2003, 2005; Smith et al., 1989,
13 1992; Epstein et al., 1992). However, it is notable that a number of other studies, several of
14 which were well conducted, did not report induction of cardiac defects in rats or rabbits from
15 TCE administered by inhalation (Dorfmueller et al., 1979; Schwetz et al., 1975; Hardin et al.,
16 1981; Healy et al., 1982; Carney et al., 2006) or in rats and mice by gavage (Cosby and
17 Dukelow, 1992; Narotsky et al., 1995; Narotsky and Kavlock, 1995; Fisher et al., 2001).
18 The potential importance of these effects warrants a more detailed discussion of possible
19 explanations for the apparent inconsistencies in the laboratory animal studies. Many of the
20 studies that did not identify cardiac anomalies used a traditional free-hand section technique on
21 fixed fetal specimens (Dorfmueller et al., 1979; Schwetz et al., 1975; Hardin et al., 1981;
22 Healy et al., 1982). Detection of such anomalies can be enhanced through the use of a fresh
23 dissection technique as described by Staples (1974) and Stuckhardt and Poppe (1984), and this
24 was the technique used in the study by Dawson et al. (1990), with further refinement of the
25 technique used in the positive studies by Dawson et al. (1993) and Johnson et al. (2003, 2005).
26 However, two studies that used the same or similar fresh dissection technique did not report
27 cardiac anomalies (Fisher et al., 2001; Carney et al., 2006), although it has been suggested that
28 differences in experimental design (e.g., inhalation versus gavage versus drinking water route of
29 administration, exposure during organogenesis versus the entire gestational period, or varied
30 dissection or evaluation procedures) may have been contributing factors to the differences in
31 observed response. A number of other limitations in the studies by Dawson et al. (1993) and
32 Johnson et al. (2003, 2005) have been suggested (Hardin et al., 2005; Watson et al., 2006). One
33 concern is the lack of clear dose-response relationship for the incidence of any specific cardiac
34 anomaly or combination of anomalies, a disparity for which no reasonable explanation has been
35 put forth. In addition, analyses on a fetal- rather than litter-basis and the pooling of data
36 collected over an extended period, including nonconcurrent controls, have been criticized. With
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-602 DRAFT—DO NOT CITE OR QUOTE
-------
1 respect to the first issue, the study authors provided individual litter incidence data to U.S. EPA
2 for analysis (see Chapter 5, dose-response), and, in response to the second issue, the study
3 authors provided further explanation as to their experimental procedures (Johnson et al., 2004).
4 In sum, while the studies by Dawson et al. (1993) and Johnson et al. (2003, 2005) have
5 significant limitations, there is insufficient reason to dismiss their findings.
6 Finally, mechanistic studies, particularly based on the avian studies mentioned above,
7 provide additional support for TCE-induced fetal cardiac malformation, particularly with respect
8 to defects involving septal and valvular morphogenesis. As summarized by NRC (2006), there is
9 substantial concordance in the stages and events of cardiac valve formation between mammals
10 and birds. While quantitative extrapolation of findings from avian studies to humans is not
11 possible without appropriate kinetic data for these experimental systems, the treatment-related
12 alterations in endothelial cushion development observed in avian in ovo and in vitro studies
13 (Boyer et al., 2000; Mishima et al., 2006; Ou et al., 2003) provide a plausible mechanistic basis
14 for defects in septal and valvular morphogenesis observed in rodents, and consequently support
15 the plausibility of cardiac defects induced by TCE in humans.
16 Postnatal developmental outcomes examined after TCE prenatal and/or postnatal
17 exposure in both humans and experimental animals include developmental neurotoxicity,
18 developmental immunotoxicity, and childhood cancer. Effects on the developing nervous
19 system included a broad array of structural and behavioral alterations in humans (White et al.,
20 1997; Windham et al., 2006; Burg et al., 1995; Burg and Gist, 1997; Bernad et al., 1987;
21 Laslo-Baker et al., 2004; Till et al., 2001; Beppu, 1968; ATSDR, 2003a) and animals
22 (Fredriksson et al., 1993; George et al., 1986; Isaacson and Taylor, 1989; Narotsky and Kavlock,
23 1995; Noland-Gerbec et al., 1986; Taylor et al., 1985; Westergren et al., 1984; Blossom et al.,
24 2008). Adverse immunological findings in humans following developmental exposures to TCE
25 were reported by Lehmann et al. (2002) and Byers et al. (1988). In mice, alterations in T-cell
26 subpopulations, spleen and/or thymic cellularity, cytokine production, autoantibody levels (in an
27 autoimmune-prone mouse strain), and/or hypersensitivity response were observed after
28 exposures during development (Blossom and Doss, 2007; Blossom et al., 2008; Peden-
29 Adams et al., 2006, 2008), Childhood cancers included leukemia and non-Hodgkin's lymphoma
30 (Morgan and Cassady, 2002; McKinney et al., 1991; Lowengart et al., 1987; Cohn et al., 1994;
31 Cutler et al., 1986; Lagakos et al., 1986; Costas et al., 2002; MA DPH, 1997; Shu et al., 1999;
32 AZ DHS, 1988, 1990a, b, c, 1997), CNS tumors (Morgan and Cassady, 2002; AZ DHS, 1998,
33 1990a, c, 1997; DeRoos et al., 2001; Peters and Preston-Martin, 1984; Peters et al., 1981, 1985),
34 and total cancers (Morgan and Cassady, 2002; ATSDR, 2006, 2008; AZ DHS, 1988, 1990a,
35 1997). These outcomes are discussed in the other relevant sections for neurotoxicity,
36 immunotoxicity, and carcinogenesis.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-603 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.11.2. Characterization of Carcinogenicity
2 In 1995, IARC concluded that trichloroethylene is "probably carcinogenic to humans"
3 (IARC, 1995). In 2000, National Toxicology Program (NTP) concluded that trichloroethylene is
4 "reasonably anticipated to be a human carcinogen" (NTP, 2000). In 2001, the draft U.S. EPA
5 health risk assessment of TCE concluded that TCE was "highly likely" to be carcinogenic in
6 humans. In 2006, a committee of the National Research Council stated that "findings of
7 experimental, mechanistic, and epidemiologic studies lead to the conclusion that
8 trichloroethylene can be considered a potential human carcinogen" (NRC, 2006).
9 Following U.S. EPA (2005a) Guidelines for Carcinogen Risk Assessment, based on the
10 available data as of 2009, TCE is characterized as "Carcinogenic to Humans" by all routes of
11 exposure. This conclusion is based on convincing evidence of a causal association between TCE
12 exposure in humans and kidney cancer. The human evidence of carcinogenicity from
13 epidemiologic studies of TCE exposure is compelling for lymphoma but less convincing than for
14 kidney cancer, and more limited for liver and biliary tract cancer. Additionally, there are several
15 lines of supporting evidence for TCE carcinogenicity in humans. First, TCE induces site-
16 specific tumors in rodents given TCE by oral gavage and inhalation. Second, toxicokinetic data
17 indicate that TCE absorption, distribution, metabolism, and excretion are qualitatively similar in
18 humans and rodents. Finally, with the exception of a mutagenic MOA for TCE-induced kidney
19 tumors, MO As have not been established for TCE-induced tumors in rodents, and no
20 mechanistic data indicate that any hypothesized key events are biologically precluded in humans.
21
22 4.11.2.1. Summary Evaluation of Epidemiologic Evidence of Trichloroethylene (TCE) and
23 Cancer
24 The available epidemiologic studies provide convincing evidence of a causal association
25 between TCE exposure and cancer. The strongest epidemiologic evidence consists of reported
26 increased risks of kidney cancer, with more limited evidence for lymphoma and liver cancer, in
27 several well-designed cohort and case-control studies (discussed below). The summary
28 evaluation below of the evidence for causality is based on guidelines adapted from Hill (1965)
29 by U.S. EPA (2005), and focuses on evidence related to kidney cancer, lymphoma, and liver
30 cancer.
31
32 4.11.2.1.1. (a) Consistency of observed association. Elevated risks for kidney cancer have been
33 observed across many independent studies. Eighteen studies in which there is a high likelihood
34 of TCE exposure in individual study subjects (e.g., based on job-exposure matrices or biomarker
35 monitoring) and which were judged to have met, to a sufficient degree, the standards of
36 epidemiologic design and analysis, were identified in a systematic review of the epidemiologic
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-604 DRAFT—DO NOT CITE OR QUOTE
-------
1 literature. Of the 14 of these studies reporting risks of kidney cancer, most estimated relative
2 risks between 1.1 and 1.9 for overall exposure to TCE. Five of these 14 studies reported
3 statistically significant increased risks either for overall exposure to TCE (Dosemeci et al., 1999;
4 Bruning et al., 2003; Raaschou-Nielsen et al., 2003) or for one of the highest TCE exposure
5 group (Raaschou-Nielsen et al., 2003; Zhao et al., 2005; Charbotel et al., 2006). Thirteen other
6 cohort, case-control, and geographic based studies were given less weight because of their lesser
7 likelihood of TCE exposure and other study design limitations that would decrease statistical
8 power and study sensitivity.
9 The consistency of association between TCE exposure and kidney cancer is further
10 supported by the results of the meta-analyses of the 14 cohort and case-control studies of
11 sufficient quality and with high probability TCE exposure potential to individual subjects. These
12 analyses observed a statistically significant increased pooled relative risk estimate (RRp) for
13 kidney cancer of 1.25 (95% CI: 1.11, 1.41) for overall TCE. The pooled relative risk were robust
14 and did not change appreciably with the removal of any individual study or with the use of
15 alternate relative risk estimates from individual studies. In addition, there was no evidence for
16 heterogeneity or publication bias.
17 The consistency of increased kidney cancer relative risk estimates across a large number
18 of independent studies of different designs and populations from different countries and
19 industries argues against chance, bias or confounding as the basis for observed associations.
20 This consistency, thus, provides substantial support for a causal effect between kidney cancer
21 and TCE exposure.
22 Some evidence of consistency is found between TCE exposure and lymphoma and liver
23 cancer. In a weight-of-evidence review of the lymphoma studies, 16 studies in which there is a
24 high likelihood of TCE exposure in individual study subjects (e.g., based on job-exposure
25 matrices or biomarker monitoring) and which met, to a sufficient degree, the standards of
26 epidemiologic design and analysis were identified. These studies generally reported excess
27 relative risk estimates for lymphoma between 0.8 and 3.1 for overall TCE exposure. Statistically
28 significant elevated relative risk estimates were observed in two cohort (Hansen et al., 2001;
29 Raaschou-Nielsen et al., 2003) and one case-control (Hardell et al., 1994) studies. The other 13
30 high-quality studies reported elevated relative risk estimates with overall TCE exposure that
31 were not statistically significant. Fifteen additional studies were given less weight because of
32 their lesser likelihood of TCE exposure and other design limitations that would decrease study
33 power and sensitivity. The observed lack of association with lymphoma in these studies likely
34 reflects study design and exposure assessment limitations and is not considered inconsistent with
35 the overall evidence on TCE and lymphoma.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-605 DRAFT—DO NOT CITE OR QUOTE
-------
1 Consistency of the association between TCE exposure and lymphoma is further
2 supported by the results of meta-analyses. These meta-analyses found a statistically significant
3 increased pooled relative risk estimate for lymphoma of 1.23 (95% CI: 1.04, 1.44) for overall
4 TCE exposure. This result and its statistical significance were not overly influenced by most
5 individual studies. In terms of the statistical significance of the RRp estimate, the only alternate
6 analysis (involving either a study removal or an alternate RR estimate) that did not yield a
7 statistically significant RRp was the analysis in which the Zhao et al. (2005) mortality RR
8 estimate was substituted with the incidence estimate, resulting in an RRp estimate of 1.19 (95%
9 CI: 1.00, 1.41]).. Some heterogeneity was observed across the 16 studies, though it was not
10 statistically significant (p = 0.10). Analyzing the cohort and case-control studies separately
11 resolved most of the heterogeneity, but the result for the pooled case-control studies was only
12 about a 7% increased relative risk estimate and was not statistically significant. The sources of
13 heterogeneity are uncertain but may be the result of some bias associated with exposure
14 assessment and/or disease classification, or from differences between cohort and case-control
15 studies in average TCE exposure. Notably, no heterogeneity was observed in the meta-analysis
16 of the highest exposure group, providing some evidence of exposure misclassification as a source
17 of heterogeneity in the overall analysis. In addition, there is some evidence of potential
18 publication bias in this data set; however, it is uncertain that this is actually publication bias
19 rather than an association between standard error and effect size resulting for some other reason,
20 e.g., a difference in study populations or protocols in the smaller studies. Furthermore, if there is
21 publication bias in this data set, it does not appear to account completely for the finding of an
22 increased lymphoma risk.
23 There are fewer studies on liver cancer than for kidney cancer and lymphoma. Of nine
24 studies, all of them cohort studies, in which there is a high likelihood of TCE exposure in
25 individual study subjects (e.g., based on job-exposure matrices or biomarker monitoring) and
26 which met, to a sufficient degree, the standards of epidemiologic design and analysis in a
27 systematic review, most reported relative risk estimates for liver and gallbladder cancer between
28 0.5 and 2.0 for overall exposure to TCE. Relative risk estimates were generally based on small
29 numbers of cases or deaths, with the result of wide confidence intervals on the estimates, except
30 for one study (Raaschou-Nielsen et al., 2003). This study has almost 6 times more cancer cases
31 than the next largest study and observed a statistically significant elevated liver and gallbladder
32 cancer risk with overall TCE exposure (RRp = 1.35 [95% CI: 1.03, 1.77]). Ten additional
33 studies were given less weight because of their lesser likelihood of TCE exposure and other
34 design limitations that would decrease statistical power and study sensitivity.
35 Consistency of the association between TCE exposure and liver cancer is further
36 supported by the results of meta-analyses. These meta-analyses found a statistically significant
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-606 DRAFT—DO NOT CITE OR QUOTE
-------
1 increased pooled relative risk estimate for liver and biliary tract cancer of 1.33 (95% CI: 1.09,
2 1.64) with overall TCE exposure. Although there was no evidence of heterogeneity or
3 publication bias and the pooled estimate was fairly insensitive to the use of alternative relative
4 risk estimates, the statistical significance of the pooled estimate depends heavily on the one large
5 study by Raaschou-Nielsen et al. (2003). However, there were fewer adequate studies available
6 for meta-analysis of liver cancer (9 versus 16 for lymphoma and 14 for kidney), leading to lower
7 statistical power, even with pooling. Moreover, liver cancer is comparatively rarer, with age-
8 adjusted incidences roughly half or less those for kidney cancer or lymphoma; thus, fewer liver
9 cancer cases are generally observed in individual cohort studies.
10
11 4.11.2.1.2. (b) Strength of the observed association. In general, the observed associations
12 between TCE exposure and cancer are modest, with relative risks or odds ratios for overall TCE
13 exposure generally less than 2.0, and higher relative risks or odds ratios for high exposure
14 categories. Among the highest statistically significant relative risks were those reported for
15 kidney cancer in the studies by Henschler et al. (1995) (7.97 [95% CI: 2.59, 8.59]) and
16 Vamvakas et al. (1998) (10.80 [95% CI: 3.36, 34.75]). As discussed in Section 4.5.3., risk
17 magnitude in both studies is highly uncertain due, in part, to possible selection biases, and
18 neither was included in the meta-analyses. However, the findings of these studies were
19 corroborated, though with lower reported relative risks, by later studies which overcame many of
20 their deficiencies, such as Briining et al. (2003) (2.47 [95% CI: 1.36, 4.49]) and Charbotel et al.
21 (2006, 2009) (2.16 [95% CI: 1.02, 4.60] for the high cumulative exposure group]. In addition,
22 the very high apparent exposure in the subjects of Henschler et al. (1995) and Vamvakas et al.
23 (1998) may have contributed to their reported relative risks being higher than those in other
24 studies. Exposures in most population case-control studies are of lower overall TCE intensity
25 compared to exposures in Briining et al. (2003) and Charbotel et al. (2006, 2009), and, as would
26 be expected, observed relative risk estimates are lower (1.24 [95% CI: 1.03, 1.49]), Pesch et al.,
27 2000a; 1.30 [95% CI: 0.9, 1.9], Dosemeci et al., 1999). A few high-quality cohort studies
28 reported statistically significant relative risks of approximately 2.0 with highest exposure,
29 including Zhao et al. (2005) (4.9 [95% CI: 1.23, 19.6] for high TCE score), Raaschou-Nielsen et
30 al. (2003) (1.7 [95% CI: 1.1, 2.4] for >5 year exposure duration, subcohort with higher
31 exposure]), and Charbotel et al. (2006) (2.16 [95% CI: 1.02, 4.60] for high cumulative exposure
32 and 2.73 [95% CI: 1.06, 7.07] for high cumulative exposure plus peaks).
33 Among the highest statistically significant relative risks reported for lymphoma were
34 those of Hansen et al. (2001) (3.1 [95% CI: 1.3, 6.1]) and Hardell et al. (1994) (7.2 [95% CI: 1.3,
35 42]), the latter a case-control study whose magnitude of risk is uncertain because of self-reported
36 occupational TCE exposure. However, these findings are corroborated in Seidler et al. (2007)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-607 DRAFT—DO NOT CITE OR QUOTE
-------
1 (2.1 [95% CI: 1.0, 4.88] for high cumulative exposure), a population case-control study with a
2 higher quality exposure assessment approach. Observed relative risk estimates for liver cancer
3 and overall TCE exposure are generally more modest.
4 Overall, the strength of association between TCE exposure and cancer is not large with
5 overall TCE exposure. Large relative risk estimates are considered strong evidence of causality;
6 however, a modest risk does not preclude a causal association and may reflect a lower level of
7 exposure, an agent of lower potency, or a common disease with a high background level (U.S.
8 EPA, 2005). Modest relative risk estimates have been observed with several well-established
9 human carcinogens such as benzene and secondhand smoke. Chance cannot explain the
10 observed association between TCE and cancer; statistically significant associations are found in a
11 number of the studies that contribute greater weight to the overall evidence, given their design
12 and statistical analysis approaches. In addition, other known or suspected risk factors can not
13 fully explain the observed elevations in kidney cancer relative risks. All kidney cancer case-
14 control studies included adjustment for possible confounding effects of smoking, and some
15 studies included body mass index and hypertension. The associations between kidney cancer
16 and TCE exposure remained in these studies after adjustment for possible known and suspected
17 confounders. Charbotel et al. (2009) observed a nonstatistically significantly kidney cancer risk
18 with exposure to only TCE with cutting fluids (1.11 [95% CI: 0.11, 10.71]) or to only cutting
19 fluids without TCE (1.24 [95% CI: 0.39, 3.93]); however, the finding of a 4-fold higher risk with
20 both cutting fluid and time-weight-average TCE exposure >50 ppm (3.74 [95% CI: 1.32, 10.57])
21 supports association with TCE. Although direct examination of smoking and other suspected
22 kidney cancer risk factors is usually not possible in cohort studies, confounding is less likely in
23 Zhao et al. (2005), given their use of an internal referent group and adjustment for
24 socioeconomic status, an indirect surrogate for smoking, and other occupational exposures. In
25 addition, the magnitude of the lung cancer risk in Raaschou-Nielsen et al. (2003) suggests a high
26 smoking rate is unlikely and cannot explain their finding on kidney cancer.
27 Few risk factors are recognized for lymphoma, with the exception of viruses and
28 suspected factors such as immunosuppression or smoking, which are associated with specific
29 lymphoma subtypes. Associations between lymphoma and TCE exposure are based on
30 groupings of several lymphoma subtypes. Three of the six lymphoma case-control studies
31 adjusted for age, sex and smoking in statistical analyses (Miligi et al., 2006; Seidler et al., 2007;
32 Wang et al., 2009), the other three case-control studies presented only unadjusted estimates of
33 the odds ratio. Like for kidney cancer, direct examination of possible confounding in cohort
34 studies is not possible. The use of internal controls in some of the higher quality cohort studies
35 is intended to reduce possible confounding related to lifestyle differences, including smoking
36 habits, between exposed and referent subjects.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-608 DRAFT—DO NOT CITE OR QUOTE
-------
1 Heavy alcohol use and viral hepatitis are established risk factors for liver cancer, with
2 severe obesity and diabetes characterized as a metabolic syndrome associated with liver cancer.
3 Only cohort studies for liver cancer are available, and they were not able to consider these
4 possible risk factors.
5
6 4.11.2.1.3. (c) Specificity of the observed association. Specificity is generally not as relevant as
7 other aspects for judging causality. As stated in the U.S. EPA Guidelines for Carcinogen Risk
8 Assessment (2005), based on our current understanding that many agents cause cancer at multiple
9 sites, and cancers have multiple causes, the absence of specificity does not detract from evidence
10 for a causal effect. Evidence for specificity could be provided by a biological marker in tumors
11 that was specific to TCE exposure. There is some evidence suggesting particular VHL mutations
12 in kidney tumors may be caused by TCE, but uncertainties in these data preclude a definitive
13 conclusion.
14
15 4.11.2.1.4. (d) Temporal relationship of the observed association. Each cohort study was
16 evaluated for the adequacy of the follow-up period to account for the latency of cancer
17 development. The studies with the greatest weight based on study design characteristics (e.g.,
18 those used in the meta-analysis) all had adequate follow-up to assess associations between TCE
19 exposure and cancer. Therefore, the findings of those studies are consistent with a temporal
20 relationship.
21
22 4.11.2.1.5. (e) Biological gradient (exposure-response relationship). Exposure-response
23 relationships are examined in the TCE epidemiologic studies only to a limited extent. Many
24 studies examined only overall "exposed" versus "unexposed" groups and did not provide
25 exposure information by level of exposure. Others do not have adequate exposure assessments
26 to confidently distinguish between levels of exposure. For example, many studies used duration
27 of employment as an exposure surrogate; however, this is a poor exposure metric given subjects
28 may have differing exposure intensity with similar exposure duration (NRC, 2006).
29 Two studies of kidney cancer reported a statistically significant trend of increasing risk
30 with increasing TCE exposure, Zhao et al. (2005) (p = 0.023 for trend with TCE score) and
31 Charbotel et al. (2005, 2007) (p = 0.04 for trend with cumulative TCE exposure). Charbotel et
32 al. (2007) was specifically designed to examine TCE exposure and had a high-quality exposure
33 assessment. Zhao et al. (2005) also had a relatively well-designed exposure assessment. A
34 positive trend was also observed in one other study (Raaschou-Nielsen et al., 2003, with
35 empl oyment durati on).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-609 DRAFT—DO NOT CITE OR QUOTE
-------
1 Biological gradient is further supported by meta-analyses for kidney cancer using only
2 the highest exposure groups and accounting for possible reporting bias, which yielded a higher
3 pooled relative risk estimate (1.53 [95% CI: 1.23, 1.91]) than for overall TCE exposure (1.25
4 [95% CI: 1.11, 1.41]). Although this analysis uses a subset of studies in the overall TCE
5 exposure analysis, the finding of higher risk in the highest exposure groups, where such groups
6 were available, is consistent with a trend of increased risk with increased exposure.
7 The lymphoma case-control study of Seidler et al. (2007) reported a statistically
8 significant trend with TCE exposure (p = 0.03 for Diffuse B-cell lymphoma trend with
9 cumulative TCE exposure), and lymphoma risk in Boice et al. (1999) appeared to increase with
10 increasing exposure duration (p = 0.20 for routine-intermittent exposed subjects). The borderline
11 trend with TCE intensity in the case-control study of Wang et al. (2009) (p = 0.06) is consistent
12 with Seidler et al. (2007). As with kidney cancer, further support was provided by meta-analyses
13 using only the highest exposure groups, which yielded a higher pooled relative risk estimate
14 (1.57 [95% CI: 1.27, 1.94]) than for overall TCE exposure (1.23 [95% CI: 1.04, 1.44]). For liver
15 cancer, the meta-analyses using only the highest exposure groups yielded a lower, and
16 nonstatistically significant, pooled estimate for primary liver cancer (1.25 [95% CI: 0.87, 1.79])
17 than overall TCE exposure (1.28 [95% CI: 0.93, 1.77]). There were no case-control studies on
18 liver cancer and TCE, and the cohort studies generally had few liver cancer cases, making it
19 more difficult to assess exposure-response relationships. The one large study (Raaschou-Nielsen
20 et al., 2003) used only duration of employment, which is an inferior exposure metric.
21
22 4.11.2.1.6. (f) Biological plausibility. TCE metabolism is similar in humans, rats, and mice and
23 results in reactive metabolites. TCE is metabolized in multiple organs and metabolites are
24 systemically distributed. Several oxidative metabolites produced primarily in the liver, including
25 CH, TCA and DCA, are rodent hepatocarcinogens. Two other metabolites, DCVC and DCVG,
26 which can be produced and cleared by the kidney, have shown genotoxic activity, suggesting the
27 potential for carcinogenicity. Kidney cancer, lymphomas, and liver cancer have all been
28 observed in rodent bioassays (see below). The laboratory animal data for liver and kidney cancer
29 are the most robust, corroborated in multiple studies, sexes, and strains, although each has only
30 been reported in a single species and the incidences of kidney cancer are quite low. Lymphomas
31 were only reported to be statistically significantly elevated in a single study in mice, but one
32 additional mouse study reported elevated lymphoma incidence and one rat study reported
33 elevated leukemia incidence. In addition, there is some evidence both in humans and laboratory
34 animals for kidney, liver and immune system noncancer toxicity from TCE exposure. Several
35 hypothesized modes of action have been presented for the rodent tumor findings, although there
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-610 DRAFT—DO NOT CITE OR QUOTE
-------
1 are insufficient data to support any one mode of action, and the available evidence does not
2 preclude the relevance of the hypothesized modes of action to humans.
3 4.11.2.1.7. (g) Coherence. Coherence is defined as consistency with the known biology. As
4 discussed under biological plausibility, the observance of kidney and liver cancer, and
5 lymphomas in humans is consistent with the biological processing and toxicity of TCE.
6
7 4.11.2.1.8. (h) Experimental evidence (from human populations). Few experimental data from
8 human populations are available on the relationship between TCE exposure and cancer. The only
9 study of a "natural experiment" (i.e., observations of a temporal change in cancer incidence in
10 relation to a specific event) notes that childhood leukemia cases appeared to be more evenly
11 distributed throughout Woburn, MA, after closure of the two wells contaminated with
12 trichloroethylene and other organic solvents (MA DPH, 1997).
13
14 4.11.2.1.9. (i) Analogy. Exposure to structurally related chlorinated solvents such as
15 tetrachloroethylene and dichloromethane have also been associated with kidney, lymphoid, and
16 liver tumors in human, although the evidence for TCE is considered stronger.
17
18 4.11.2.1.10.Conclusion. In conclusion, based on the weight-of-evidence analysis for kidney
19 cancer and in accordance with U.S. EPA guidelines, TCE is characterized as "Carcinogenic to
20 Humans." This hazard descriptor is used when there is convincing epidemiologic evidence of a
21 causal association between human exposure and cancer. Convincing evidence is found in the
22 consistency of the kidney cancer findings. The consistency of increased kidney cancer relative
23 risk estimates across a large number of independent studies of different designs and populations
24 from different countries and industries provides compelling evidence given the difficulty, a
25 priori, in detecting effects in epidemiologic studies when the relative risks are modest, the
26 cancers are relatively rare, and therefore, individual studies have limited statistical power. This
27 strong consistency argues against chance, bias, and confounding as explanations for the elevated
28 kidney cancer risks. In addition, statistically significant exposure-response trends are observed
29 in high-quality studies. These studies were designed to examine kidney cancer in populations
30 with high TCE exposure intensity. These studies addressed important potential confounders and
31 biases, further supporting the observed associations with kidney cancer as causal. In a meta-
32 analysis of 14 high-quality studies, a statistically significant pooled relative risk estimate was
33 observed for overall TCE exposure (RRp: 1.25 [95% CI: 1.11, 1.41]). The pooled relative risk
34 estimate was greater for the highest TCE exposure groups (RRp: 1.53 [95% CI: 1.23, 1.91]; n =
35 12 studies). Meta-analyses investigating the influence of individual studies and the sensitivity of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-611 DRAFT—DO NOT CITE OR QUOTE
-------
1 the results to alternate relative risk estimate selections found the pooled relative risk estimates to
2 be highly robust. Furthermore, there was no indication of publication bias or significant
3 heterogeneity. It would require a substantial amount of high-quality negative data to contradict
4 this observed association.
5 The evidence is less convincing for lymphoma and liver cancer. While the evidence is
6 strong for lymphoma, issues of (non-statistically significant) study heterogeneity, potential
7 publication bias, and weaker exposure-response results contribute greater uncertainty. The
8 evidence is more limited for liver cancer mainly because only cohort studies are available and
9 most of these studies have small numbers of cases.
10
11 4.11.2.2. Summary of Evidence for Trichloroethylene (TCE) Carcinogenicity in Rodents
12 Additional evidence of TCE carcinogenicity consists of increased incidences of tumors
13 reported in multiple chronic bioassays in rats and mice. In total, this database identifies some of
14 the same target tissues of TCE carcinogenicity also seen in epidemiological studies, including the
15 kidney, liver, and lymphoid tissues.
16 Of particular note is the site-concordant finding of TCE-induced kidney cancer in rats. In
17 particular, low, but biologically and sometimes statistically significant, increases in the incidence
18 of kidney tumors were observed in multiple strains of rats treated with TCE by either inhalation
19 or corn oil gavage (Maltoni et al., 1986; NTP, 1988, 1990). For instance, Maltoni et al. (1986)
20 reported that although only 4/130 renal adenocarcinomas in rats in the highest dose group, these
21 tumors had never been observed in over 50,000 Sprague-Dawley rats (untreated, vehicle-treated,
22 or treated with different chemicals) examined in previous experiments in the same laboratory In
23 addition, the gavage study by NCI (1976) and two inhalation studies by Henschler et al. (1980),
24 and Fukuda et al. (1983) each observed one renal adenoma or adenocarcinoma in some dose
25 groups and none in controls. The largest (but still small) incidences were observed in treated
26 male rats, only in the highest dose groups. However, given the small numbers, an effect in
27 females cannot be ruled out. Several studies in rats were limited by excessive toxicity,
28 accidental deaths, or deficiencies in reporting (NCI, 1976; NTP, 1988, 1990). Individually,
29 therefore, these studies provide only suggestive evidence of renal carcinogenicity. Overall,
30 given the rarity of these types of tumors in the rat strains tested and the repeated similar results
31 across experiments and strains, these studies taken together support the conclusion that TCE is a
32 kidney carcinogen in rats, with males being more sensitive than females. No other tested
33 laboratory species (i.e., mice and hamsters) have exhibited increased kidney tumors, although
34 high incidences of kidney toxicity have been reported in mice (NCI, 1976; Maltoni et al., 1986;
35 NTP, 1990). The GSH-conjugation-derived metabolites suspected of mediating TCE-induced
36 kidney carcinogenesis have not been tested in a standard 2-year bioassay, so their role cannot be
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-612 DRAFT—DO NOT CITE OR QUOTE
-------
1 confirmed definitively. However, it is clear that GSH conjugation of TCE occurs in humans and
2 that the human kidney contains the appropriate enzymes for bioactivation of GSH conjugates.
3 Therefore, the production of the active metabolites thought to be responsible for kidney tumor
4 induction in rats likely occurs in humans.
5 Statistically significant increases in TCE-induced liver tumors have been reported in
6 multiple inhalation and gavage studies with male Swiss mice and B6C3F1 mice of both sexes
7 (NCI, 1976; Maltoni et al., 1986; NTP, 1990; Anna et al., 1994; Herren-Freund et al., 1987;
8 Bull et al., 2002). In female Swiss mice, on the other hand, Fukuda et al. (1983), in CD-I (ICR,
9 Swiss-derived) mice, and Maltoni et al. (1986) both reported small, nonsignificant increases at
10 the highest dose by inhalation. Henschler et al. (1980, 1984) reported no increases in either sex
11 of Han:NMRI (also Swiss-derived) mice exposed by inhalation and ICR/HA (Swiss) mice
12 exposed by gavage. However, the inhalation study (Henschler et al., 1980) had only 30 mice per
13 dose group and the gavage study (Henschler et al., 1984) had dosing interrupted due to toxicity.
14 Studies in rats (NCI, 1976; Henschler et al., 1980; Maltoni et al., 1986; NTP, 1988, 1990) and
15 hamsters (Henschler et al., 1980) did not report statistically significant increases in liver tumor
16 induction with TCE treatment. However, several studies in rats were limited by excessive
17 toxicity or accidental deaths (NCI, 1976; NTP, 1988, 1990), and the study in hamsters only had
18 30 animals per dose group. These data are inadequate for concluding that TCE lacks
19 hepatocarcinogenicity in rats and hamsters, but are indicative of a lower potency in these species.
20 Moreover, it is notable that a few studies in rats reported low incidences (too few for statistical
21 significance) of very rare biliary- or endothelial-derived tumors in the livers of some treated
22 animals (Fukuda et al., 1983; Henschler et al., 1980; Maltoni et al.,1986). Further evidence for
23 the hepatocarcinogenicity of TCE is derived from chronic bioassays of the TCE oxidative
24 metabolites CH, TCA, and DCA in mice (e.g., George et al., 2000; Leakey et al., 2003a;
25 Bull et al., 1990; DeAngelo et al., 1996, 1999, 2008), all of which reported
26 hepatocarcinogenicity. Very limited testing of these TCE metabolites has been done in rats, with
27 a single experiment reported in both Richmond et al. (1995) and DeAngelo et al. (1996) finding
28 statistically significant DCA-induced hepatocarcinogenicity. With respect to TCA, DeAngelo et
29 al. (1997), often cited as demonstrating lack of hepatocarcinogenicity in rats, actually reported
30 elevated adenoma multiplicity and carcinoma incidence from TCA treatment. However,
31 statistically, the role of chance could not be confidently excluded because of the low number of
32 animals per dose group (20-24 per treatment group at final sacrifice). Overall, TCE and its
33 oxidative metabolites are clearly carcinogenic in mice, with males more sensitive than females
34 and the B6C3F1 strain appearing to be more sensitive than the Swiss strain. Such strain and sex
35 differences are not unexpected, as they appear to parallel, qualitatively, differences in
36 background tumor incidence. Data in other laboratory animal species are limited. Thus, except
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-613 DRAFT—DO NOT CITE OR QUOTE
-------
1 for DCA, which is carcinogenic in rats, inadequate evidence exists to evaluate the
2 hepatocarcinogenicity of these compounds in rats or hamsters. However, to the extent that there
3 is hepatocarcinogenic potential in rats, TCE is clearly less potent in the strains tested in this
4 species than in B6C3F1 and Swiss mice.
5 Additionally, there is more limited evidence for TCE-induced lymphatic cancers in rats
6 and mice, lung tumors in mice, and testicular tumors in rats. With respect to the lymphomas,
7 Henschler et al. (1980) reported statistically significant increases in lymphomas in female
8 Han:NMRI mice treated via inhalation. While Henschler et al. (1980) suggested these
9 lymphomas were of viral origin specific to this strain, subsequent studies reported increased
10 lymphomas in female B6C3F1 mice treated via corn oil gavage (NTP, 1990) and leukemias in
11 male Sprague-Dawley and female August rats (Maltoni et al., 1986; NTP, 1988). However,
12 these tumors had relatively modest increases in incidence with treatment, and were not reported
13 to be increased in other studies. With respect to lung tumors, rodent bioassays have
14 demonstrated a statistically significant increase in pulmonary tumors in mice following chronic
15 inhalation exposure to TCE (Fukuda et al., 1983; Maltoni et al., 1988, 1986). Pulmonary tumors
16 were not reported in other species tested (i.e., rats and hamsters; Maltoni et al., 1986, 1988;
17 Fukuda et al., 1983; Henschler et al., 1980). Chronic oral exposure to TCE led to a
18 nonstatistically significant increase in pulmonary tumors in mice but, again, not in rats or
19 hamsters (Henschler et al., 1984; Van Duuren et al., 1979; NCI, 1976; NTP, 1988, 1990; Maltoni
20 et al., 1986). A lower response via oral exposure would be consistent with a role of respiratory
21 metabolism in pulmonary carcinogenicity. Finally, increased testicular (interstitial cell and
22 Ley dig cell) tumors have been observed in rats exposed by inhalation and gavage (NTP, 1988,
23 1990; Maltoni et al., 1986). Statistically significant increases were reported in Sprague-Dawley
24 rats exposed via inhalation (Maltoni et al., 1986) and Marshall rats exposed via gavage (NTP,
25 1988). In three rat strains, ACI, August, and F344/N, a high (>75%) control rate of testicular
26 tumors was observed, limiting the ability to detect a treatment effect (NTP, 1988, 1990).
27 In summary, there is clear evidence for TCE carcinogenicity in rats and mice, with
28 multiple studies showing TCE to cause tumors at multiple sites. The apparent lack of site
29 concordance across laboratory animal species may be due to limitations in design or conduct in a
30 number of rat bioassays and/or genuine interspecies differences in sensitivity. Nonetheless, these
31 studies have shown carcinogenic effects across different strains, sexes, and routes of exposure,
32 and site-concordance is not necessarily expected for carcinogens.
33
34 4.11.2.3. Summary of Additional Evidence on Biological Plausibility
35 Additional evidence from toxicokinetic, toxicity, and mechanistic studies supports the
36 biological plausibility of TCE carcinogenicity in humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-614 DRAFT—DO NOT CITE OR QUOTE
-------
1
2 4.11.2.3.1. Toxicokinetics. As described in Chapter 3, there is no evidence of major qualitative
3 differences across species in TCE absorption, distribution, metabolism, and excretion. In
4 particular, available evidence is consistent with TCE being readily absorbed via oral, dermal, and
5 inhalation exposures, and rapidly distributed to tissues via systemic circulation. Extensive in
6 vivo and in vitro data show that mice, rats, and humans all metabolize TCE via two primary
7 pathways: oxidation by CYPs and conjugation with glutathione via GSTs. Several metabolites
8 and excretion products from both pathways, including TCA, DCA, TCOH, TCOG, NAcDCVC,
9 and DCVG, have been detected in blood and urine from exposed humans was well as from at
10 least one rodent species. In addition, the subsequent distribution, metabolism, and excretion of
11 TCE metabolites are qualitatively similar among species. Therefore, humans possess the
12 metabolic pathways that produce the TCE metabolites thought to be involved in the induction of
13 rat kidney and mouse liver tumors, and internal target tissues of both humans and rodents
14 experience a similar mix of TCE and metabolites.
15 As addressed in further detail elsewhere (see Chapters 3 and 5), examples of quantitative
16 interspecies differences in toxicokinetics include differences in partition coefficients, metabolic
17 capacity and affinity in various tissues, and plasma binding of the metabolite TCA. These and
18 other differences are addressed through PBPK modeling, which also incorporates physiological
19 differences among species (see Section 3.5), and are accounted for in the PBPK model-based
20 dose-response analyses (see Chapter 5). Importantly, these quantitative differences affect only
21 interspecies extrapolations of carcinogenic potency, and do not affect inferences as to the
22 carcinogenic hazard for TCE. In addition, available data on toxicokinetic differences do not
23 appear sufficient to explain interspecies differences in target sites of TCE carcinogenicity
24 (discussed further in Chapter 5: Dose-Response).
25
26 4.11.2.3.2. Toxicity and mode of action. Many different MO As have been proposed for TCE-
27 induced carcinogenesis. With respect to genotoxicity, although it appears unlikely that TCE, as a
28 pure compound, causes point mutations, there is evidence for TCE genotoxicity with respect to
29 other genetic endpoints, such as micronucleus formation (see Section 4.2.1.4.4). In addition, as
30 discussed further below, several TCE metabolites have tested positive in genotoxicity assays.
31 The MOA conclusions for specific target organs in laboratory animals are summarized below.
32 Only in the case of the kidney is it concluded that the data are sufficient to support a particular
33 MOA being operative. However, the available evidence do not indicate that qualitative
34 differences between humans and test animals would preclude any of the hypothesized key events
35 in rodents from occurring in humans.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-615 DRAFT—DO NOT CITE OR QUOTE
-------
1 For the kidney, the predominance of positive genotoxicity data in the database of
2 available studies of TCE metabolites derived from GSH conjugation (in particular DCVC, see
3 Section 4.2.5), together with toxicokinetic data consistent with their systemic delivery to and in
4 situ formation in the kidney, supports the conclusion that a mutagenic MOA is operative in TCE-
5 induced kidney tumors (see Section 4.4.7.1). Relevant data include demonstration of
6 genotoxicity in available in vitro assays of GSH conjugation metabolites and reported kidney-
7 specific genotoxicity after in vivo administration of TCE or DCVC. Mutagenicity is a well-
8 established cause of carcinogenicity. While supporting the biological plausibility of this
9 hypothesized MOA, available data on the VHL gene in humans or transgenic animals do not
10 conclusively elucidate the role of VHL mutation in TCE-induced renal carcinogenesis.
11 Cytotoxicity and compensatory cell proliferation, also presumed to be mediated through
12 metabolites formed after GSH-conjugation of TCE, have also been suggested to play a role in the
13 MOA for renal carcinogenesis, as high incidences of nephrotoxicity have been observed in
14 animals at doses that also induce kidney tumors. Human studies have reported markers for
15 nephrotoxicity at current occupational exposures, although data are lacking at lower exposures.
16 Toxicity is observed in both mice and rats, in some cases with nearly 100% incidence in all dose
17 groups, but kidney tumors are only observed at low incidences in rats at the highest tested doses.
18 Therefore, nephrotoxicity alone appears to be insufficient, or at least not rate-limiting, for rodent
19 renal carcinogenesis, since maximal levels of toxicity are reached before the onset of tumors. In
20 addition, nephrotoxicity has not been shown to be necessary for kidney tumor induction by TCE
21 in rodents. In particular, there is a lack of experimental support for causal links, such as
22 compensatory cellular proliferation or clonal expansion of initiated cells, between nephrotoxicity
23 and kidney tumors induced by TCE. Furthermore, it is not clear if nephrotoxicity is one of
24 several key events in a MOA, if it is a marker for an "upstream" key event (such as oxidative
25 stress) that may contribute independently to both nephrotoxicity and renal carcinogenesis, or if it
26 is incidental to kidney tumor induction. Moreover, while toxicokinetic differences in the GSH
27 conjugation pathway, along with their uncertainty, are addressed through PBPK modeling, no
28 data suggest that any of the proposed key events for TCE-induced kidney tumors rats are
29 precluded in humans. Therefore, TCE-induced rat kidney tumors provide additional support for
30 the convincing human evidence of TCE-induced kidney cancer, with mechanistic data supportive
31 of a mutagenic MOA.
32 The strongest data supporting the hypothesis of a mutagenic MOA in either the lung or
33 the liver are those demonstrating the genotoxicity of CH (see Section 4.2.4), which is produced
34 in these target organs as a result of oxidative metabolism of TCE. It has been suggested that CH
35 mutagenicity is unlikely to be the cause of TCE hepatocarcinogenicity because the
36 concentrations required to elicit these responses are several orders of magnitude higher that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-616 DRAFT—DO NOT CITE OR QUOTE
-------
1 achieved in vivo (Moore and Harrington-Brock, 2000). However, it is not clear how much of a
2 correspondence is to be expected from concentrations in genotoxicity assays in vitro and
3 concentrations in vivo, as reported in vivo CH concentrations are in whole liver homogenate
4 while in vitro concentrations are in culture media. The use of i.p. administration, which leads to
5 an inflammatory response, in many other in vivo genotoxicity assays in the liver and lung
6 complicates the comparison with carcinogenicity data. Also, it is difficult with the available data
7 to assess the contributions from genotoxic effects of CH along with those from the genotoxic and
8 nongenotoxic effects of other oxidative metabolites (e.g., DCA and TCA). Therefore, while data
9 are insufficient to conclude that a mutagenic MOA mediated by CH is operant, a mutagenic
10 MOA in the liver or lung, either mediated by CH or by some other oxidative metabolite of TCE,
11 cannot be ruled out.
12 A second MOA hypothesis for TCE-induced liver tumors involves activation of the
13 PPARa receptor. Clearly, in vivo administration of TCE leads to activation of PPARa in rodents
14 and likely does so in humans as well (based on in vitro data for TCE and its oxidative
15 metabolites). However, the evidence as a whole does not support the view that PPAR-a is the
16 sole operant MOA mediating TCE hepatocarcinogenesis. Although metabolites of TCE activate
17 PPARa, the data on the subsequent elements in the hypothesized MOA (e.g., gene regulation,
18 cell proliferation, apoptosis, and selective clonal expansion), while limited, indicate significant
19 differences between PPAR-a agonists such as Wy-14643 and TCE or its metabolites. For
20 example, compared with other agonists, TCE induces transient as opposed to persistent increases
21 in DNA synthesis; increases (or is without effect on), as opposed to decreases, apoptosis; and
22 induces a different H-ras mutation frequency or spectrum. These data support the view that
23 mechanisms other than PPARa activation may contribute to these effects; besides PPARa
24 activation, the other hypothesized key events are nonspecific, and available data (e.g., using
25 knockout mice) do not indicate that they are solely or predominantly dependent on PPARa. A
26 second consideration is whether certain TCE metabolites (e.g., TCA) that activate PPAR-a are
27 the sole contributors to its carcinogenicity. As summarized above (see Section 4.11.1.3), TCA is
28 not the only metabolite contributing to the observed noncancer effects of TCE in the liver. Other
29 data also suggest that multiple metabolites may also contribute to the hepatic carcinogenicity of
30 TCE. Liver phenotype experiments, particularly those utilizing immunostaining for c-Jun,
31 support a role for both DCA and TCA in TCE-induced tumors, with strong evidence that TCA
32 cannot solely account for the characteristics of TCE-induced tumors (e.g., Bull et al., 2002). In
33 addition, H-ras mutation frequency and spectrum of TCE-induced tumors more closely
34 resembles that of spontaneous tumors or of those induced by DCA, and were less similar in
35 comparison to that of TCA-induced tumors. The heterogeneity of TCE-induced tumors is similar
36 to that observed to be induced by a diversity carcinogens including those that do not activate
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-617 DRAFT—DO NOT CITE OR QUOTE
-------
1 PPAR-a, and to that observed in human liver cancer. Taken together, the available data indicate
2 that, rather than being solely dependent on a single metabolite (TCA) and/or molecular target
3 (PPAR-a) multiple TCE metabolites and multiple toxicity pathways contribute to TCE-induced
4 liver tumors.
5 Other considerations as well as new data published since the NRC (2006) review are also
6 pertinent to the liver tumor MOA conclusions. It is generally acknowledged that, qualitatively,
7 there are no data to support the conclusion that effects mediated by the PPAR-a receptor that
8 contribute to hepatocarcinogenesis would be biologically precluded in humans (Klaunig et al.,
9 2003; NRC, 2006). It has, on the other hand, been argued that due to quantitative toxicokinetic
10 and toxicodynamic differences, the hepatocarcinogenic effects of chemicals activating this
11 receptor are "unlikely" to occur in humans (Klaunig et al., 2003; NRC, 2006); however, several
12 lines of evidence strongly undermine the confidence in this assertion. With respect to
13 toxicokinetics, as discussed above, quantitative differences in oxidative metabolism are
14 accounted for in PBPK modeling of available in vivo data, and do not support interspecies
15 differences of a magnitude that would preclude hepatocarcinogenic effects based on
16 toxicokinetics alone. With respect to the MOA proposed by Klaunig et al. (2003), recent
17 experiments have demonstrated that PPAR-a activation and the sequence of key events in the
18 hypothesized MOA are not sufficient to induce hepatocarcinogenesis (Yang et al., 2007).
19 Moreover, the demonstration that the PPAR-a agonist DEHP induces tumors in PPAR-a-null
20 mice supports the view that the events comprising the hypothesized MOA are not necessary for
21 liver tumor induction in mice by this PPARa agonist (Ito et al., 2007). Therefore, several lines
22 of evidence, including experiments published since the NRC (2006) review, call into question
23 the scientific validity of using the PPAR-a MOA hypothesis as the basis for evaluating the
24 relevance to human carcinogenesis of rodent liver tumors (Guyton et al., 2009).
25 In summary, available data support the conclusion that the MOA for TCE-induced liver
26 tumors in laboratory animals is not known. However, a number of qualitative similarities exist
27 between observations in TCE-exposed mice and what is known about the etiology and induction
28 of human hepatocellular carcinomas. Polyploidization, changes in glycogen storage, inhibition
29 of GST-zeta, and aberrant DNA methylation status, which have been observed in studies of mice
30 exposed to TCE or its oxidative metabolites, are all either clearly related to human
31 carcinogenesis or are areas of active research as to their potential roles (PPARa activation is
32 discussed below). The mechanisms by which TCE exposure may interact with known risk
33 factors for human hepatocellular carcinomas are not known. However, available data do not
34 suggest that TCE exposure to mice results in liver tumors that are substantially different in terms
35 of their phenotypic characteristics either from human hepatocellular carcinomas or from rodent
36 liver tumors induced by other chemicals.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-618 DRAFT—DO NOT CITE OR QUOTE
-------
1 Comparing various other, albeit relatively nonspecific, tumor characteristics between
2 rodent species and humans provides additional support to the biologic plausibility of TCE
3 carcinogenicity. For example, in the kidney and the liver, the higher incidences of background
4 and TCE-induced tumors in male rats and mice, respectively, as compared to females parallels
5 the observed higher human incidences in males for these cancers (Ries et al., 2008). For the
6 liver, while there is a lower background incidence of liver tumors in humans than in rodents, in
7 the United States there is an increasing occurrence of liver cancer associated with several factors,
8 including viral hepatitis, higher survival rates for cirrhosis, and possibly diabetes (reviewed in
9 El-Serag, 2007). In addition, Leakey et al. (2003) reported that increased body weight in
10 B6C3F1 mice is strongly associated with increased background liver tumor incidences, although
11 the mechanistic basis for this risk factor in mice has not been established. Nonetheless, it is
12 interesting that recent epidemiologic studies have suggested obesity, in addition to associated
13 disorders such as diabetes and metabolic syndrome, as a risk factor for human liver cancer
14 (El-Serag, 2007; El-Serag and Rudolph, 2007). Furthermore, the phenotypic and morphologic
15 heterogeneity of tumors seen in the human liver is qualitatively similar to descriptions of mouse
16 liver tumors induced by TCE exposure, as well as those observed from exposure to a variety of
17 other chemical carcinogens. These parallels suggest similar pathways (e.g., for cell signaling) of
18 carcinogenesis may be active in mice and humans and support the qualitative relevance of mouse
19 models of liver to human liver cancer.
20 For mouse lung tumors, MOA hypotheses have centered on TCE metabolites produced
21 via oxidative metabolism in situ. As discussed above, the hypothesis that the mutagenicity of
22 reactive intermediates or metabolites (e.g., CH) generated during CYP metabolism contributes to
23 lung tumors cannot be ruled out, although available data are inadequate to conclusively support
24 this MOA. An alternative MOA has been posited involving other effects of such oxidative
25 metabolites, particularly CH, including cytotoxicity and regenerative cell proliferation.
26 Experimental support for this alternative hypothesis remains limited, with no data on proposed
27 key events in experiments of duration 2 weeks or longer. While the data are inadequate to
28 support this MOA hypothesis, the data also do not suggest that any proposed key events would
29 be biologically plausible in humans. Furthermore, the focus of the existing MOA hypothesis
30 involving cytotoxicity has been CH, and, as summarized above (see Section 4.11.1.5), other
31 metabolites may contribute to respiratory tract noncancer toxicity or carcinogenicity. In sum, the
32 MOA for mouse lung tumors induced by TCE is not known.
33 A MOA subsequent to in situ oxidative metabolism, whether involving mutagenicity,
34 cytotoxicity, or other key events, may also be relevant to other tissues where TCE would
35 undergo CYP metabolism. For instance, CYP2E1, oxidative metabolites, and protein adducts
36 have been reported in the testes of rats exposed to TCE, and, in some rat bioassays, TCE
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-619 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure increased the incidence of rat testicular tumors. However, inadequate data exist to
2 adequately define a MOA hypothesis for this tumor site.
3 4.11.3. Characterization of Factors Impacting Susceptibility
4 As discussed in more detail in Section 4.10, there is some evidence that certain
5 subpopulations may be more susceptible to exposure to TCE. Factors affecting susceptibility
6 examined include lifestage, gender, genetic polymorphisms, race/ethnicity, pre-existing health
7 status, and lifestyle factors and nutrition status.
8 Examination of early lifestages includes exposures such as transplacental transfer
9 (Beppu, 1968; Laham, 1970; Withey and Karpinski, 1985; Ghantous et al., 1986; Helliwell and
10 Hutton, 1950) and breast milk ingestion (Fisher et al., 1990, 1997; Pellizzari et al., 1982;
11 Hamada and Tanaka, 1995), early lifestage-specific toxicokinetics, PBPK models (Fisher et al.,
12 1989, 1990), and differential outcomes in early lifestages such as developmental cardiac defects.
13 Although there is more information on susceptibility to TCE during early lifestages than on
14 susceptibility during later lifestages or for other populations with potentially increased
15 susceptibility, there remain a number of uncertainties regarding children's susceptibility.
16 Improved PBPK modeling for using childhood parameters for early lifestages as recommended
17 by the NRC (2006), and validation of these models will aid in determining how variations in
18 metabolic enzymes affect TCE metabolism. In particular, the NRC states that it is prudent to
19 assume children need greater protection than adults, unless sufficient data are available to justify
20 otherwise (NRC, 2006). Because the weight of evidence supports a mutagenic MOA for TCE
21 carcinogenicity in the kidney (see Section 4.4.7), and there is an absence of chemical-specific
22 data to evaluate differences in carcinogenic susceptibility, early-life susceptibility should be
23 assumed and the ADAFs should be applied, in accordance with the Supplemental Guidance
24 (discussed further in Chapter 5).
25 Fewer data are available on later lifestages, although there is suggestive evidence to
26 indicate that older adults may experience increased adverse effects than younger adults (Mahle et
27 al., 2007; Rodriguez et al., 2007). In general, more studies specifically designed to evaluate
28 effects in early and later lifestages are needed in order to more fully characterize potential life
29 stage-related TCE toxicity.
30 Examination of gender-specific susceptibility includes toxicokinetics, PBPK models
31 (Fisher et al., 1998), and differential outcomes. Gender differences observed are likely due to
32 variation in physiology and exposure.
33 Genetic variation likely has an effect on the toxicokinetics of TCE. In particular,
34 differences in CYP2E1 activity may affect susceptibility of TCE due to effects on production of
35 toxic metabolites (Kim and Ghanayem, 2006; Lipscomb et al., 1997; Povey et al., 2001; Yoon et
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-620 DRAFT—DO NOT CITE OR QUOTE
-------
1 al., 2007). GST polymorphisms could also play a role in variability in toxic response (Briining et
2 al., 1997; Wiesenhiitter et al., 2007), as well as other genotypes, but these have not been
3 sufficiently tested. Differences in genetic polymorphisms related to the metabolism of TCE have
4 also been observed among various race/ethnic groups (Inoue et al., 1989; Sato et al., 1991b).
5 Pre-existing diminished health status may alter the response to TCE exposure.
6 Individuals with increased body mass may have an altered toxicokinetic response (Clewell et al.,
7 2000; Sato, 1993; Sato et al., 1991b; Monster et al., 1979; McCarver et al., 1998; Davidson and
8 Beliles, 1991; Lash et al., 2000) resulting in changes the internal concentrations of TCE or in the
9 production of toxic metabolites. Other conditions, including diabetes and hypertension, are risk
10 factors for some of the same health effects that have been associated with TCE exposure, such as
11 renal cell carcinoma. However, the interaction between TCE and known risk factors for human
12 diseases is not known, and further evaluation of the effects due to these factors is needed.
13 Lifestyle and nutrition factors examined include alcohol consumption, tobacco smoking,
14 nutritional status, physical activity, and socioeconomic status. In particular, alcohol intake has
15 been associated with metabolic inhibition (altered CYP2E1 expression) of TCE in both humans
16 and experimental animals (Bardodej and Vyskocil, 1956; Barret et al., 1984; McCarver et al.,
17 1998; Muller et al., 1975; Sato, 1993; Sato et al., 1980, 1981, 1983, 1991a; Stewart et al., 1974;
18 Kaneko et al., 1994; Larson and Bull, 1989; Nakajima et al., 1988, 1990, 1992b; Okino et al.,
19 1991; Sato and Nakajima, 1985; White and Carlson, 1981). In addition, such factors have been
20 associated with increased baseline risks for health effects associated with TCE, such as kidney
21 cancer (e.g., smoking) and liver cancer (e.g., alcohol consumption). However, the interaction
22 between TCE and known risk factors for human diseases is not known, and further evaluation of
23 the effects due to these factors is needed.
24 In sum, there is some evidence that certain subpopulations may be more susceptible to
25 exposure to TCE. Factors affecting susceptibility examined include lifestage, gender, genetic
26 polymorphisms, race/ethnicity, pre-existing health status, and lifestyle factors and nutrition
27 status. However, except in the case of toxicokinetic variability characterized using the PBPK
28 model described in Section 3.5, there are inadequate chemical-specific data to quantify the
29 degree of differential susceptibility due to such factors.
30
31
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 4-621 DRAFT—DO NOT CITE OR QUOTE
-------
1 5. DOSE-RESPONSE ASSESSMENT
2
3
4 5.1. DOSE-RESPONSE ANALYSES FOR NONCANCER ENDPOINTS
5 Because of the large number of noncancer health effects associated with trichloroethylene
6 (TCE) exposure and the large number of studies reporting on these effects, a screening process,
7 described below, was used to reduce the number of endpoints and studies to those that would
8 best inform the selection of the critical effects for the inhalation reference concentration (RfC)
9 and oral reference dose (RfD).l The screening process helped identify the more sensitive
10 endpoints for different types of effects within each health effect domain (e.g., different target
11 systems) and provided information on the exposure levels that could contribute to the most
12 sensitive effects, used for the RfC and RfD, as well as to additional noncancer effects as
13 exposure increases. These more sensitive endpoints were also used to investigate the impacts of
14 pharmacokinetic uncertainty and variability.
15 The general process used to derive the RfD and RfC was as follows (see Figure 5-1):
16
17 (1) Consider all studies described in Chapter 4 that report adverse noncancer health effects
18 and provide quantitative dose-response data.
19 (2) Consider for each study/endpoint possible points of departure (PODs) on the basis of
20 applied dose, with the order of preference being first a benchmark dose (BMD)2 derived
21 from empirical modeling of the dose-response data, then a no-observed-adverse-effect
22 level (NOAEL), and lastly a lowest-observed-adverse-effect level (LOAEL).
23 (3) Adjust each POD by endpoint/study-specific "uncertainty factors" (UFs), accounting for
24 uncertainties and adjustments in the extrapolation from the study conditions to conditions
25 of human exposure, to derive candidate RfCs (cRfCs) or RfDs (cRfDs) intended to be
26 protective for each endpoint (individually) on the basis of applied dose.
27 (4) Array the cRfCs and cRfDs across the following health effect domains: (1) neurotoxic
28 effects; (2) systemic (body weight) and organ toxicity (kidney, liver) effects; (3)
29 immunotoxic effects; (4) reproductive effects; and (5) developmental effects.
1 In U.S. EPA noncancer health assessments, the RfC (RfD) is an estimate (with uncertainty spanning perhaps an
order of magnitude) of a continuous inhalation (daily oral) exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. It can be derived
from a NOAEL, LOAEL, or benchmark concentration (dose), with uncertainty factors generally applied to reflect
limitations of the data used.
2More precisely, it is the BMDL, i.e., the (one-sided) 95% lower confidence bound on the dose corresponding to the
benchmark response for the effect that is used as the POD.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-1 DRAFT—DO NOT CITE OR QUOTE
-------
y AM
studies
Points of
Departure
(POPs)
Apply
ncertainty
Factors
(UFs;
Candidate RfCsV
(cRfCs) & V
candidate RfDs
(cRfDs)
[applied dose]
41 i \
j ^
•^ Lowest
values within
each domain
V
5)
T Candidate
v critical effects/
' studies,
cRfCs & cRfDs
_@_\
Apply PBPK \
model; ,
Update UFs /
V
PBPK-based
candidate RfCs
(p-cRfDs) &
candidate RfDs,
(P-cRfDs)
(g^Consider and evaluate most sensitive
estimates across domains and their
uncertainties
I
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
RfC and RfD for
noncancer effects
Figure 5-1. Flow-chart of the process used to derive the RfD and RfC for
noncancer effects.
(5) Select as candidate critical effects those endpoints with the lowest cRfCs or cRfDs,
within each of these effect domains, taking into account the confidence in each estimate.
When there are alternative estimates available for a particular endpoint, preference is
given to studies whose design characteristics (e.g., species, statistical power, exposure
level(s) and duration, endpoint measures) are better suited for determining the most
sensitive human health effects of chronic TCE exposure.
(6) For each candidate critical effect selected in step 5, use, to the extent possible, the
physiologically based pharmarcokinetic (PBPK) model developed in Section 3.5 to
calculate an internal dose POD (idPOD) for plausible internal dose metrics that were
selected on the basis of what is understood about the role of different TCE metabolites in
toxicity and the mode of action (MOA) for toxicity.
(7) For each idPOD for each candidate critical effect, use the PBPK model to estimate
interspecies and within-human pharmacokinetic variability (or just within-human
variability for human-based PODs). The results of this calculation are 99th percentile
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimates of the human equivalent concentration and human equivalent dose (HECgg and
2 HED99) for each candidate critical effect.
3 (8) Adjust each HECgg or HED99 by endpoint/study-specific UFs (which, due to the use of
4 the PBPK model, may differ from the UFs used in step 3) to derive a PBPK model-based
5 candidate RfCs (p-cRfC) and RfD (p-cRfD) for each candidate critical effect.
6 (9) Characterize the uncertainties in the cRfCs, cRfDs, p-cRfCs, and p-cRfDs, with the
7 inclusion of quantitative uncertainty analyses of pharmacokinetic uncertainty and
8 variability as derived from the Bayesian population analysis using the PBPK model.
9 (10) Evaluate the most sensitive cRfCs, p-cRfCs, cRfDs, and p-cRfDs, taking into account the
10 confidence in the estimates, to arrive at an RfC and RfD for TCE.
11
12 In contrast to the approach used in most assessments, in which the RfC and RfD are each based
13 on a single critical effect, the final RfC and RfD for TCE were based on multiple critical effects
14 that resulted in very similar candidate RfC and RfD values at the low end of the full range of
15 values. This approach was taken here because it provides robust estimates of the RfC and RfD
16 and because it highlights the multiple effects that are all yielding very similar candidate values.
17 The results of this process are summarized in the sections below, with technical details presented
18 in Appendix F.
19
20 5.1.1. Modeling Approaches and Uncertainty Factors for Developing Candidate
21 Reference Values Based on Applied Dose
22 This section summarizes the general methodology used with all the TCE studies and
23 endpoints for developing cRfCs and cRfDs on the basis of applied dose. A detailed discussion of
24 the application of these approaches to the studies and endpoints for each health effect domain
25 follows in the next section (see Section 5.1.2).
26 Standard adjustments3 were made to the applied doses to obtain continuous inhalation
27 exposures and daily average oral doses over the study exposure period (see Appendix F for
28 details), except for effects for which there was sufficient evidence that the effect was more
29 closely associated with administered exposure level (e.g., changes in visual function). The PODs
30 based on applied dose in the following sections and in Appendix F are presented in terms of the
31 adjusted doses (except where noted).
Discontinuous exposures (e.g., gavage exposures once a day, 5 days/week, or inhalation exposures for 5 days/week,
6 hours/day) were adjusted to the continuous exposure yielding the same cumulative exposure. For inhalation
studies, these adjustments are equivalent to those recommended by U.S. EPA (1994) for deriving a human
equivalent concentration for a Category 3 gas for which the blood: air partition coefficient in laboratory animals is
greater than that in humans (see Section 3.1 for discussion of the TCE blood:air partition coefficient).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 As described above, wherever possible,4 benchmark dose modeling was conducted to
2 obtain benchmark dose lower bounds (BMDLs) to serve as PODs for the cRfCs and cRfDs.
3 Note that not all quantitative dose-response data are amenable to benchmark dose modeling. For
4 example, while nonnumerical data (e.g., data presented in line or bar graphs rather than in tabular
5 form) were considered for developing LOAELs or NOAELs, they were not used for benchmark
6 dose modeling. In addition, sometimes the available models used do not provide an adequate fit
7 to the data. For the benchmark dose modeling for this assessment, the U.S. EPA's BenchMark
8 Dose Software (BMDS), which is freely available at www.epa.gov/ncea/bmds, was used. For
9 dichotomous responses, the Log-logistic, multistage, and Weibull models were fitted. This
10 subset of BMDS dichotomous models was used to reduce modeling demands, and these
11 particular models were selected because, as a group, they have been found to be capable of
12 describing the great majority of dose-response data sets, and specifically for some TCE data sets
13 (Filipsson and Victorin, 2003). For continuous responses, the distinct models available in
14 BMDS—the power, polynomial, and Hill models—were fitted. For some reproductive and
15 developmental data sets, two nested models (the nested logistic and the Rai and Van Ryzin
16 models in BMDS5) were fitted to examine and account for potential intralitter correlations.
17 Models with unconstrained power parameters <1 were considered when the dose-response
18 relationship appeared supralinear, but these models often yield very low BMDL estimates and
19 there was no situation in which an unconstrained model with a power parameter <1 was selected
20 for the data sets modeled here. In most cases, a constrained model or the Hill model provided an
21 adequate fit to such a dose-response relationship. In a few cases, the highest-dose group was
22 dropped to obtain an improved fit to the lower-dose groups. See Appendix F for further details
23 on model fitting and parameter constraints.
24 After the fitting these models to the data sets, the following procedure for model selection
25 was applied. First, models were rejected if the p-va\ue for goodness of fit was <0.10.6 Second,
26 models were rejected if they did not appear to adequately fit the low-dose region of the dose-
27 response relationship, based on an examination of graphical displays of the data and scaled
28 residuals. If the BMDL estimates from the remaining models were "sufficiently close" (with a
29 criterion of within 2-fold for "sufficiently close"), then the model with the lowest Akaike
4An exception was for the systemic effect of decreased body weight, which was observed in multiple chronic
studies. Dose-response data were available, but the resources were not invested into modeling these data because
the endpoint appeared a priori to be less sensitive than others and was not expected to be a critical effect.
5The National Center for lexicological Research model failed with the TCE datasets.
6In a few cases in which none of the models fit the data with/) > 0.10, linear models were selected on the basis of an
adequate visual fit overall.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 Information Criteria (AIC) was selected.7 If the BMDL estimates from the remaining models are
2 not sufficiently close, some model dependence is assumed. With no clear biological or statistical
3 basis to choose among them, the lowest BMDL was chosen as a reasonable conservative
4 estimate, unless the lowest BMDL appeared to be an outlier, in which case further judgments
5 were made. Additionally, for continuous models, constant variance models were used for model
6 parsimony unless the/>-value for the test of homogenous variance was <0.10, in which case the
7 modeled variance models were considered.
8 For benchmark response (BMR) selection, statistical and biological considerations were
9 taken into account. For dichotomous responses, our general approach was to use 10% extra risk
10 as the BMR for borderline or minimally adverse effects and either 5% or 1% extra risk for
11 adverse effects, with 1% reserved for the most severe effects. For continuous responses, the
12 preferred approach for defining the BMR is to use a pre-established cut-point for the minimal
13 level of change in the endpoint at which the effect is generally considered to become biologically
14 significant (e.g., there is substantial precedence for using a 10% change in weight for organ and
15 body weights and a 5% change in weight for fetal weight). In the absence of a well-established
16 cut-point, a BMR of 1 (control) standard deviation (SD) change from the control mean, or 0.5
17 SD for effects considered to be more serious, was generally selected. For one neurological effect
18 (traverse time), a doubling (i.e., 2-fold change) was selected because the control SD appeared
19 unusually small.
20 After the PODs were determined for each study/endpoint, UFs were applied to obtain the
21 cRfCs and cRfDs. Uncertainty factors are used to address differences between study conditions
22 and conditions of human environmental exposure (U.S. EPA, 2002). These include
23
24 (a) Extrapolating from laboratory animals to humans: If a POD is derived from
25 experimental animal data, it is divided by an UF to reflect pharmacokinetic and
26 pharmacodynamic differences that may make humans more sensitive than laboratory
27 animals. For oral exposures, the standard value for the interspecies UF is 10, which
28 breaks down (approximately) to a factor of three for pharmacokinetic differences and a
29 factor of three for pharmacodynamic differences. For inhalation exposures, ppm
30 equivalence across species is generally assumed, in which case pharmacokinetic
31 differences are considered to be negligible, and the standard value used for the
32 interspecies UF is 3, which is ascribed to pharmacodynamic differences8. These standard
7 Akaike Information Criteria—a measure of information loss from a dose-response model that can be used to
compare a set of models. Among a specified set of models, the model with the lowest AIC is considered the "best."
If two or more models share the lowest AIC, an average of the BMDLs could be used, but averaging was not used in
this assessment because for the one occasion in which models shared the lowest AIC, a selection was made based on
visual fit.
8Note that the full attribution of the scaling effect, under the assumption that response scales across species in
accordance with ppm equivalence, to pharmacokinetics is an oversimplification and is only one way to think about
how to interpret cross-species scaling. See Section 5.1.3.1 for further discussion of scaling issues.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-5 DRAFT—DO NOT CITE OR QUOTE
-------
1 values were used for all the cRfCs and cRfDs based on laboratory animal data in this
2 assessment.
3 (b) Human (intraspecies) variability: RfCs and RfDs apply to the human population,
4 including sensitive subgroups, but studies rarely examine sensitive humans. Sensitive
5 humans could be adversely affected at lower exposures than a general study population;
6 consequently, PODs from general-population studies are divided by an UF to address
7 sensitive humans. Similarly, the animals used in most laboratory animal studies are
8 considered to be "typical" or "average" responders, and the human (intraspecies)
9 variability UF is also applied to PODs from such studies to address sensitive subgroups.
10 The standard value for the human variability UF is 10, which breaks down
11 (approximately) to a factor of three for pharmacokinetic variability and a factor of three
12 for pharmacodynamic variability. This standard value was used for all the PODs in this
13 assessment with the exception of the PODs for a few immunological effects that were
14 based on data from a sensitive (autoimmune-prone) mouse strain; for those PODs, an UF
15 of 3 was used for human variability.
16 (c) Uncertainty in extrapolating from subchronic to chronic exposures: RfCs and RfDs
17 apply to lifetime exposure, but sometimes the best (or only) available data come from
18 less-than-lifetime studies. Lifetime exposure can induce effects that may not be apparent
19 or as large in magnitude in a shorter study; consequently, a dose that elicits a specific
20 level of response from a lifetime exposure may be less than the dose eliciting the same
21 level of response from a shorter exposure period. Thus, PODs based on subchronic
22 exposure data are generally divided by a subchronic-to-chronic UF, which has a standard
23 value of 10. If there is evidence suggesting that exposure for longer time periods does
24 not increase the magnitude of an effect, a lower value of three or one might be used. For
25 some reproductive and developmental effects, chronic exposure is that which covers a
26 specific window of exposure that is relevant for eliciting the effect, and subchronic
27 exposure would correspond to an exposure that is notably less than the full window of
28 exposure.
29 (d) Uncertainty in extrapolating from LOAELs to NOAELs: PODs are intended to be
30 estimates of exposure levels without appreciable risk under the study conditions so that,
31 after the application of appropriate UFs for interspecies extrapolation, human variability,
32 and/or duration extrapolation, the absence of appreciable risk is conveyed to the RfC or
33 RfD exposure level to address sensitive humans with lifetime exposure. Under the
34 NOAEL/LOAEL approach to determining a POD, however, adverse effects are
35 sometimes observed at all study doses. If the POD is a LOAEL, it is divided by an UF to
36 better estimate a NOAEL. The standard value for the LOAEL-to-NOAEL UF is 10,
37 although sometimes a value of three is used if the effect is considered minimally adverse
38 at the response level observed at the LOAEL or even one if the effect is an early marker
39 for an adverse effect. For one POD in this assessment, a value of 30 was used for the
40 LOAEL-to-NOAEL UF because the incidence rate for the adverse effect was >90% at the
41 LOAEL.
42 (e) Additional database uncertainties: Sometimes a database UF of 3 or 10 is used to reflect
43 other factors contributing uncertainties that are not explicitly treated by the UFs described
44 above. Such factors include lack of completeness of the overall database, minimal
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-6 DRAFT—DO NOT CITE OR QUOTE
-------
1 sample size, or poor exposure characterization. No database UF was used in this
2 assessment. See Section 5.1.4.1 for additional discussion of the uncertainties associated
3 with the overall database for TCE.
4
5 5.1.2. Candidate Critical Effects by Effect Domain
6 A large number of endpoints and studies were considered within each of the five health
7 effect domains. A comprehensive list of all endpoints/studies that were considered for
8 developing cRfCs and cRfDs is shown in Tables 5-1-5-5. These tables also summarize the
9 PODs for the various study endpoints, the UFs applied, and the resulting cRfCs or cRfDs.
10 Inhalation and oral studies are presented together so that the extent of the available data, as well
11 as concordance or lack thereof in the responses across routes of exposure, is evident. In addition,
12 the PBPK model developed in Section 3.5 will be applied to each candidate critical effect to
13 develop a POD based on internal dose (idPOD); and subsequent extrapolation of the idPOD to
14 pharmacokinetically sensitive humans is performed for both inhalation and oral human
15 exposures, regardless of the route of exposure in the original study.
16 The sections below discuss the cRfCs and cRfDs developed from the effects and studies
17 identified in the hazard characterization (see Chapter 4) that were suitable for the derivation of
18 reference values (i.e., that provided quantitative dose-response data). Because the general
19 approach for applying UFs was discussed above, the sections below only discuss the selection of
20 particular UFs when there are study characteristics that require additional judgment as to the
21 appropriate UF values and possible deviations from the standard values usually assigned.
22
23 5.1.2.1. Candidate Critical Neurological Effects on the Basis of Applied Dose
24 As summarized in Section 4.11.1.1, both human and experimental animal studies have
25 associated TCE exposure with effects on several neurological domains. The strongest
26 neurological evidence of hazard is for changes in trigeminal nerve function or morphology and
27 impairment of vestibular function. There is also evidence for effects on motor function; changes
28 in auditory, visual, and cognitive function or performance; structural or functional changes in the
29 brain; and neurochemical and molecular changes. Studies with numerical dose-response
30 information, with their corresponding cRfCs or cRfDs, are shown in Table 5-1. Because
31 impairment of vestibular function occurs at higher exposures, such changes were not considered
32 candidate critical effects; but the other neurological effect domains are represented.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-7 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-1. Neurological effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
to
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
'J'comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Trigeminal nerve effects
Mhiri et al., 2004
Ruitjen etal., 1991
Barret et al., 1992
Human
Human
Human
Rat
LOAEL
LOAEL
LOAEL
LOAEL
40
6
14
1,800
1
1
1
10
1
1
1
10
10
10
10
10
10
10
3
10
1
1
1
1
100
100
30
10,000C
0.40
0.06
0.47
0.18
Abnormal trigeminal somatosensory
evoked potentials; preferred POD based
on middle of reported range of
50-150 ppm.
Alternate POD based on U-TCA and
Ikedaetal. (1972).
Trigeminal nerve effects; POD based on
mean cumulative exposure and mean
duration, UFIoael = 3 due to early marker
effect and minimal degree of change.
Morphological changes; uncertain
adversity; some effects consistent with
demyelination.
Auditory effects
Rebertetal., 1991
Albee etal. ,2006
Crofton and Zhao,
1997
Rat
Rat
Rat
NOAEL
NOAEL
BMDL
800
140
274
10
10
10
3
3
3
10
10
10
1
1
1
1
1
1
300
300
300
2.7
0.47
0.91
Preferred, due to better dose-response
data, amenable to BMD modeling.
BMR = 10dB absolute change.
Psychomotor effects
Waseem et al.,
2001
Nunesetal., 2001
Moseretal., 1995
Rat
Rat
Rat
Rat
LOAEL
LOAEL
BMDL
NOAEL
45
2,000
248
500
1
10
3
3
3
10
10
10
10
10
10
10
3
3
1
1
1
1
1
1
3,000
300
300
0.45
0.67
0.83
1.7
Changes in locomotor activity; transient,
minimal degree of adversity; no effect
reported in same study for oral exposures
(2 10 mg/kg/d).
| Foot splaying; minimal adversity.
| # rears (standing on hindlimbs);
BMR = 1 SD change.
| Severity score for neuromuscular
changes.
Co
TO
Co
I
TO
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-1. Neurological effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
(continued)
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UFloael
UFdb
[1C b
"-"^comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Visual function effects
Blain etal., 1994
Rabbit
LOAEL
350
10
3
10
10
1
3,000
0.12
POD not adjusted to continuous exposure
because visual effects more closely
associated with administered exposure.
Cognitive effects
Kuligetal., 1987
Isaacson et al.,
1990
Rat
Rat
NOAEL
LOAEL
500
47
1
10
3
10
10
10
1
10
1
1
30
10,000C
17
0.0047
| time in 2-choice visual discrim. test; test
involves multiple systems but largely
visual so not adjusted to continuous
exposure.
Demyelination in hippocampus.
Mood and sleep disorders
Albee etal. ,2006
Aritoetal., 1994
Rat
Rat
NOAEL
LOAEL
140
12
10
3
3
3
10
10
1
10
1
1
300
1,000
0.47
0.012
Hyperactivity.
Changes in wakefulness.
Other neurological effects
Kjellstrand et al.,
1987
Gash et al., 2007
Rat
Mouse
Rat
LOAEL
LOAEL
LOAEL
300
150
710
10
10
10
3
3
10
10
10
10
10
10
10
1
1
1
3,000
3,000
10,000C
0.10
0.050
0.071
J, regeneration of sciatic nerve.
J, regeneration of sciatic nerve.
Degeneration of dopaminergic neurons.
Co
H £
""Adjusted to continuous exposure unless otherwise noted. For inhalation studies, adjustments yield a POD that is a human equivalent concentration as
recommended for a Category 3 gas in U.S. EPA (1994) in the absence of PBPK modeling. Same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors.
°U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with a composite UF of greater than 3,000;
however, composite UFs exceeding 3,000 are considered here because the derivation of the cRfCs and cRfDs is part of a screening process and the subsequent
application of the PBPK model for candidate critical effects will reduce the values of some of the individual UFs.
UFSC = subchronic -to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFloael = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded studies/endpoints were selected as candidate critical effects/studies.
-------
1 For trigeminal nerve effects, cRfC estimates based on two human studies are in a similar
2 range of 0.4-0.5 ppm (Mhiri et al., 2004; Ruitjen et al., 1991). There remains some uncertainty
3 as to the exposure characterization, as shown by the use of an alternative POD for Mhiri et al.
4 (2004) based on urinary trichloroacetic acid (TCA) resulting in a 5-fold smaller cRfC. However,
5 the overall confidence in these estimates is relatively high because they are based on humans
6 exposed under chronic or nearly chronic conditions. Other human studies (e.g., Barret et al.,
7 1984), while indicative of hazard, did not have adequate exposure information for quantitative
8 estimates of an inhalation POD. A cRfD of 0.2 mg/kg/d was developed from the only oral study
9 demonstrating trigeminal nerve changes, an acute study in rats (Barret et al., 1992). This
10 estimate required multiple extrapolations with a composite uncertainty factor of 10,000.9
11 For auditory effects, a high confidence cRfC of about 0.7 ppm was developed based on
12 BMD modeling of data from Crofton and Zhao (1997); and cRfCs developed from two other
13 auditory studies (Albee et al., 2006; Rebert et al., 1991) were within about 4-fold. No oral data
14 were available for auditory effects. For psychomotor effects, the available human studies (e.g.,
15 Rasmussen et al., 1983) did not have adequate exposure information for quantitative estimates of
16 an inhalation POD. However, a relatively high confidence cRfC of 0.5 ppm was developed from
17 a study in rats (Waseem et al., 2001). Two cRfDs within a narrow range of 0.7-1.7 mg/kg/d
18 were developed based on two oral studies reporting psychomotor effects (Nunes et al., 2001;
19 Moser et al., 1995), although varying in degree of confidence.
20 For the other neurological effects, the estimated cRfCs and cRfDs were more uncertain,
21 as there were fewer studies available for any particular endpoint, and the PODs from several
22 studies required more adjustment to arrive at a cRfC or cRfD. However, the endpoints in these
23 studies also tended to be indicative of more sensitive effects and, therefore, they need to be
24 considered. The lower cRfCs fall in the range 0.01-0.1 ppm and were based on effects on visual
25 function in rabbits (Blain et al., 1994), wakefulness in rats (Arito et al., 1994), and regeneration
26 of the sciatic nerve in mice and rats (Kjellstrand et al., 1987). Of these, altered wakefulness
27 (Arito et al., 1994) has both the lowest POD and the lowest cRfC. There is relatively high
28 confidence in this study, as it shows a clear dose-response trend, with effects persisting
29 postexposure. For the subchronic-to-chronic UF, a value of 3 was used because, even though it
30 was just a 6-week study, there was no evidence of a greater impact on wakefulness following
31 6 weeks of exposure than there was following 2 weeks of exposure at the LOAEL, although
32 there was an effect of repeated exposure on the postexposure period impacts of higher exposure
9U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with
a composite UF of greater than 3,000; however, composite UFs exceeding 3,000 are considered here because the
derivation of the cRfCs and cRfDs is part of a screening process and the subsequent application of the PBPK model
for candidate critical effects will reduce the values of some of the individual UFs.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-10 DRAFT—DO NOT CITE OR QUOTE
-------
1 levels. The cRfDs, in the range 0.005-0.07, were based on demyelination in the hippocampus
2 (Isaacson et al., 1990) and degeneration of dopaminergic neurons (Gash et al., 2007), both in
3 rats. In both these cases, adjusting for study design characteristics led to a composite uncertainty
4 factor of 10,000,10 so the confidence in these cRfDs is lower. However, no other studies of these
5 effects are available.
6 In summary, although there is high confidence both in the hazard and in the cRfCs and
7 cRfDs for trigeminal nerve, auditory, or psychomotor effects, the available data suggest that the
8 more sensitive indicators of TCE neurotoxicity are changes in wakefulness, regeneration of the
9 sciatic nerve, demyelination in the hippocampus and degeneration of dopaminergic neurons.
10 Therefore, these more sensitive effects are considered the candidate critical effects for
11 neurotoxicity, albeit with more uncertainty in the corresponding cRfCs and cRfDs. Of these
12 more sensitive effects, for the reasons discussed above, there is greater confidence in the changes
13 in wakefulness reported by Arito et al. (1994). In addition, trigeminal nerve effects are
14 considered a candidate critical effect because this is the only type of neurological effect for
15 which human data are available, and the POD for this effect is similar to that from the most
16 sensitive rodent study (Arito et al., 1994, for changes in wakefulness). Between the two human
17 studies of trigeminal nerve effects, Ruitjen et al. (1991) is preferred for deriving noncancer
18 reference values because its exposure characterization is considered more reliable.
19
20 5.1.2.2. Candidate Critical Kidney Effects on the Basis of Applied Dose
21 As summarized in Section 4.11.1.2, multiple lines of evidence support TCE
22 nephrotoxicity in the form of tubular toxicity, mediated predominantly through the glutathione
23 (GSH) conjugation product dichlorovinyl cysteine (DCVC). Available human studies, while
24 providing evidence of hazard, did not have adequate exposure information for quantitative
25 estimates of PODs. Several studies in rodents, some of chronic duration, have shown
26 histological changes, nephropathy, or increased kidney/body weight ratios, and were suitable for
27 deriving cRfCs and cRfDs, shown in Table 5-2.
10U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values
with a composite UF of greater than 3,000; however, composite UFs exceeding 3,000 are considered here because
the derivation of the cRfCs and cRfDs is part of a screening process and the subsequent application of the PBPK
model for candidate critical effects will reduce the values of some of the individual UFs.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-11 DRAFT—DO NOT CITE OR QUOTE
-------
to
Table 5-2. Kidney, liver, and body weight effects in studies suitable for dose-response, and corresponding cRfCs
and cRfDs
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UFloael
UFdb
[1C b
"-"^comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Histological changes in kidney
Maltoni, 1986
NTP, 1990
NCI, 1976
NTP, 1988
Rat
Rat
Rat
Mouse
Rat
BMDL
BMDL
LOAEL
LOAEL
BMDL
40.2
34
360
620
9.45
1
1
1
1
1
3
10
10
10
10
10
10
10
10
10
1
1
10
30
1
1
1
1
1
1
30
100
1,000
3,000
100
1.3
0.34
0.36
0.21
0.0945
meganucleocytosis; BMR = 10% extra
risk
meganucleocytosis; BMR = 10% extra
risk
cytomegaly & karyomegaly; considered
minimally adverse, but UF|0aei = 10 due to
high response rate (>98%) at LOAEL;
also in mice, but use NCI (1976) for that
species
toxic nephrosis; UFloael = 30 due to
>90% response at LOAEL for severe
effect
toxic nephropathy; female Marshall (most
sensitive sex/strain); BMR = 5% extra risk
| kidney/body weight ratio
Kjellstrand et al.,
1983b
Woolhiser et al.,
2006
Mouse
Rat
BMDL
BMDL
34.7
15.7
1
1
3
3
10
10
1
1
1
1
30
30
1.2
0.52
BMR = 10% increase; 30 d, but 120 d @
120 ppm not more severe so UFsc = 1 ;
results are for males, which were slightly
more sensitive, and yielded better fit to
variance model
BMR = 10% increase; UFsc = 1 based on
Kjellstrand et al. (1983b) result
| liver/body weight ratio
Kjellstrand et al.,
1983b
Woolhiser et al.,
2006
Buben and
O'Flaherty, 1985
Mouse
Rat
Mouse
BMDL
BMDL
BMDL
21.6
25.2
81.5
1
1
1
3
3
10
10
10
10
1
1
1
1
1
1
30
30
100
0.72
0.84
0.82
BMR = 10% increase; UFsc = 1 based on
not more severe at 4 months
BMR = 10% increase; UFsc = 1 based on
Kjellstrand et al. (1983b) result
BMR = 10% increase; UFsc = 1 based on
Kjellstrand et al. (1983b) result
TO
Co
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
O 1
to
- a
o §"
vo 5'
&•
o
I
§
^.
S'
a
I
^t>
>
i
TO'
^s
s
1
rl
OJ o
S
^
I
gfr
r* TO
C ^
^rl «s
H O
O §
°§
i-?1 a-
•^H XL"
g|
O^
HH Oq
H TO
SI
^^
o ^
B*
H
W
Table 5-2. Kidney, liver, and body weight effects in studies suitable for dose-response, and corresponding cRfCs
and cRfDs (continued)
Effect tvoe
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UFloael
UFdb
[1C b
"-"^comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Decreased body weight
NTP, 1990
NCI, 1976
Mouse
Rat
LOAEL
LOAEL
710
360
1
1
10
10
10
10
10
10
1
1
1,000
1,000
0.71
0.36
Reflects several, but not all,
strains/sexes.
"Adjusted to continuous exposure unless otherwise noted. For inhalation studies, adjustments yield a POD that is a human equivalent concentration as
recommended for a Category 3 gas in U.S. EPA (1994) in the absence of PBPK modeling. Same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFloael = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded studies/endpoints were selected as candidate critical effects/studies.
-------
1 The cRfCs developed from three suitable inhalation studies, one reporting
2 meganucleocytosis in rats (Maltoni et al., 1986) and two others reporting increased kidney
3 weights in mice (Kjellstand et al., 1983b) and rats (Woolhiser et al., 2006),n are in a narrow
4 range of 0.5-1.3 ppm. All three utilized BMD modeling and, thus, take into account statistical
5 limitations of the Woolhiser et al. (2006) and Kjellstrand et al. (1983b) studies, such as
6 variability in responses or the use of low numbers of animals in the experiment. The response
7 used for kidney weight increases was the organ weight as a percentage of body weight, to
8 account for any commensurate decreases in body weight, although the results did not generally
9 differ much when absolute weights were used instead. Although the two studies reporting
10 kidney weight changes were subchronic, longer-term experiments by Kjellstrand et al. (1983b)
11 did not report increased severity, so no subchronic-to-chronic uncertainty factor was used in the
12 derivation of the cRfC. The high response level of 73% at the lowest dose for
13 meganucleocytosis in the chronic study of Maltoni et al. (1986) implies more uncertainty in the
14 low-dose extrapolation. However, strengths of this study include the presence of
15 histopathological analysis and relatively high numbers of animals per dose group.
16 The suitable oral studies give cRfDs within a narrow range of 0.09-0.4 mg/kg/d, as
17 shown in Table 5-2, although the degree of confidence in the cRfDs varies considerably. For
18 cRfDs based on National Toxicology Program (NTP, 1990) and National Cancer Institute (NCI,
19 1976) chronic studies in rodents, extremely high response rates of >90% precluded BMD
20 modeling. An UF of 10 was applied for extrapolation from a LOAEL to a NOAEL in the NTP
21 (1990) study because the effect (cytomegaly and karyomegaly), although minimally adverse, was
22 observed at such a high incidence. An UF of 30 was applied for extrapolation from a LOAEL to
23 a NOAEL in the NCI (1976) study because of the high incidence of a clearly adverse effect
24 (toxic nephrosis). There is more confidence in the cRfDs based on meganucleocytosis reported
25 in Maltoni et al. (1986) and toxic nephropathy NTP (1988), as BMD modeling was used to
26 estimate BMDLs. Because these two oral studies measured somewhat different endpoints, but
27 both were sensitive markers of nephrotoxic responses, they were considered to have similarly
28 strong weight. For meganucleocytosis, a BMR of 10% extra risk was selected because the effect
29 was considered to be minimally adverse. For toxic nephropathy, a BMR of 5% extra risk was
30 used because toxic nephropathy is a severe toxic effect. This BMR required substantial
31 extrapolation below the observed responses (about 60%); however, the response level seemed
32 warranted for this type of effect and the ratio of the BMD to the BMDL was not large (1.56).
"Woolhiser et al. (2006) is an Organisation for Economic Co-operation and Development guideline immunotoxicity
study performed by the Dow Chemical Company, certified by Dow as conforming to Good Laboratory Practices as
published by the U.S. EPA for the Toxic Substances Control Act.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 In summary, there is high confidence in both the hazard and the cRfCs and cRfDs for
2 histopathological and weight changes in the kidney. These effects are considered to be candidate
3 critical effects for several reasons. First, they appear to be the most sensitive indicators of
4 toxicity that are available for the kidney. In addition, as discussed in Section 3.5,
5 pharmacokinetic data indicate substantially more production of GSH-conjugates thought to
6 mediate TCE kidney effects in humans relative to rats and mice. As discussed above, several
7 studies are considered reliable for developing cRfCs and cRfDs for these endpoints. For
8 histopathological changes, the most sensitive were selected as candidate critical studies. These
9 were the only available inhalation study (Maltoni et al., 1986), the NTP (1988) study in rats, and
10 the NCI (1976) study in mice. While the NCI (1976) study has greater uncertainty, as discussed
11 above, with a high response incidence at the POD that necessitates greater low-dose
12 extrapolation, it is included to add a second species to the set of candidate critical effects. For
13 kidney weight changes, both available studies were chosen as candidate critical studies.
14
15 5.1.2.3. Candidate Critical Liver Effects on the Basis of Applied Dose
16 As summarized in Section 4.11.1.3, while there is only limited epidemiologic evidence of
17 TCE hepatotoxicity, TCE clearly leads to liver toxicity in laboratory animals, likely through its
18 oxidative metabolites. Available human studies contribute to the overall weight of evidence of
19 hazard, but did not have adequate exposure information for quantitative estimates of PODs. In
20 rodent studies, TCE causes a wide array of hepatotoxic endpoints: increased liver weight, small
21 transient increases in DNA synthesis, changes in ploidy, cytomegaly, increased nuclear size, and
22 proliferation of peroxisomes. Increased liver weight (hepatomegaly, or specifically increased
23 liver/body weight ratio) has been the most studied endpoint across a range of studies in both
24 sexes of rats and mice, with a variety of exposure routes and durations. Hepatomegaly was
25 selected as the critical liver effect for multiple reasons. First, it has been consistently reported in
26 multiple studies in rats and mice following both inhalation and oral routes of exposure. In
27 addition, it appears to accompany the other hepatic effects at the doses tested, and hence
28 constitutes a hepatotoxicity marker of similar sensitivity to the other effects. Finally, in several
29 studies, there are good dose-response data for BMD modeling.
30 As shown in Table 5-2, cRfCs for hepatomegaly developed from the two most suitable
31 subchronic inhalation studies (Woolhiser et al., 2006; Kjellstrand et al., 1983b), while in
32 different species (rats and mice, respectively), are both based on similar PODs derived from
33 BMD modeling, have the same composite uncertainty factor of 30, and result in similar cRfC
34 estimates of about 0.8 ppm. The cRfD for hepatomegaly developed from the oral study of Buben
35 and O'Flaherty (1985) in mice also was based on a POD derived from BMD modeling and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-15 DRAFT—DO NOT CITE OR QUOTE
-------
1 resulted in a cRfD estimate of 0.8 mg/kg/d. Among the studies reporting liver weight changes
2 (reviewed in Section 4.5 and Appendix E), this study had by far the most extensive
3 dose-response data. The response used in each case was the liver weight as a percentage of body
4 weight, to account for any commensurate decreases in body weight, although the results did not
5 generally differ much when absolute weights were used instead.
6 There is high confidence in all these candidate reference values. BMD modeling takes
7 into account statistical limitations such as variability in response or low numbers of animals and
8 standardizes the response rate at the POD. Although the studies were subchronic, hepatomegaly
9 occurs rapidly with TCE exposure, and the degree of hepatomegaly does not increase with
10 chronic exposure (Kjellstrand et al., 1983b), so no subchronic-to-chronic uncertainty factor was
11 used.
12 In summary, there is high confidence both in the hazard and the cRfCs and cRfDs for
13 hepatomegaly. Hepatomegaly also appears to be the most sensitive indicator of toxicity that is
14 available for the liver and is therefore considered a candidate critical effect. As discussed above,
15 several studies are considered reliable for developing cRfCs and cRfDs for this endpoint, and,
16 since they all indicated similar sensitivity but represented different species and/or routes of
17 exposure, were all considered candidate critical studies.
18
19 5.1.2.4. Candidate Critical Body Weight Effects on the Basis of Applied Dose
20 The chronic oral bioassays NCI (1976) and NTP (1990) reported decreased body weight
21 with TCE exposure, as shown in Table 5-2. However, the lowest doses in these studies were
22 quite high, even on an adjusted basis (see PODs in Table 5-2). These were not considered
23 critical effects because they are not likely to be the most sensitive noncancer endpoints, and were
24 not considered candidate critical effects.
25
26 5.1.2.5. Candidate Critical Immunological Effects on the Basis of Applied Dose
27 As summarized in Section 4.11.1.4, the human and experimental animal studies of TCE
28 and immune-related effects provide strong evidence for a role of TCE in autoimmune disease
29 and in a specific type of generalized hypersensitivity syndrome, while there are fewer data
30 pertaining to immunosuppressive effects. Available human studies, while providing evidence of
31 hazard, did not have adequate exposure information for quantitative estimates of PODs. Several
32 studies in rodents were available on autoimmune and immunosuppressive effects that were
33 adequate for deriving cRfCs and cRfDs, which are summarized in Table 5-3.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-16 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-3. Immunological effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
to
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
'J'comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
I thvmus weight
Keil et al., 2009
Mouse
LOAEL
0.35
1
10
10
10
1
1,000
0.00035
J, thymus weight; corresponding decrease in
total thymic cellularity reported at 1 0x higher
dose
Autoimmunity
Kaneko et al., 2000
Keil et al., 2009
Griffin etal.,2000
Cai et al., 2008
Mouse
(MRL-
Ipr/lpr)
Mouse
Mouse
(MRL+/+)
Mouse
(MRL+/+)
LOAEL
LOAEL
BMDL
LOAEL
70
0.35
13.4
60
10
1
1
1
3
10
10
10
3
10
3
3
10
1
1
10
1
1
1
1
1,000
100
30
300
0.070
0.0035
0.45
0.20
Changes in immunoreactive organs — liver (incl.
sporadic necrosis in hepatic lobules), spleen;
UFh = 3 due to autoimmune-prone mouse
| anti-dsDNA and anti-ssDNA Abs (early
markers for SLE) (B6C3F1 mouse);
UFIoael = 1 due to early marker
Various signs of autoimmune hepatitis;
BMR = 1 0% extra risk for> minimal effects
Inflammation in liver, kidney, lungs, and
pancreas, which may lead to SLE-like disease;
UFh = 3 due to autoimmune-prone mouse;
UFIoael = 1 0 since some hepatic necrosis
Immunosuppression
Woolhiseretal., 2006
Sanders etal., 1982
Rat
Mouse
Mouse
Mouse
BMDL
NOAEL
LOAEL
LOAEL
31.2
190
18
18
10
1
1
1
3
10
10
10
10
10
10
10
1
1
3
3
1
1
1
1
300
100
300
300
0.10
1.9
0.060
0.060
J, RFC response; BMR = 1 SD change
J, humoral response to sRBC; largely transient
during exposure
J, stem cell bone marrow recolonization
(sustained); females more sensitive
J, cell-mediated response to sRBC (largely
transient during exposure); females more
sensitive
TO
Co
o §
H I
O >
HH Oq
H TO
si
H
W
""Adjusted to continuous exposure unless otherwise noted. For inhalation studies, adjustments yield a POD that is a human equivalent concentration as
recommended for a Category 3 gas in U.S. EPA (1994) in the absence of PBPK modeling. Same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFloael = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded studies/endpoints were selected as candidate critical effects/studies.
-------
1 For decreased thymus weights, a cRfD from the only suitable study (Keil et al., 2009) is
2 0.00035 mg/kg/d based on results from nonautoimmune-prone B6C3F1 mice, with a composite
3 uncertainty factor of 1,000 for a POD that is a LOAEL (the dose-response relationship is
4 sufficiently supralinear that attempts at BMD modeling did not result in adequate fits to these
5 data). Thymus weights were not significantly affected in autoimmune prone mice in the same
6 study, consistent with the results reported by Kaneko et al. (2000) in autoimmune-prone mice. In
7 addition, Keil et al. (2009) and Peden-Adams et al. (2008) reported that for several
8 immunotoxicity endpoints associated with TCE, the autoimmune-prone strain appeared to be less
9 sensitive than the nonautoimmune prone B6C3F1 strain. In rats, Woolhiser et al. (2006) reported
10 no significant change in thymus weights in the Sprague-Dawley (S-D) strain. These data are
11 consistent with normal mice being sensitive to this effect as compared to autoimmune-prone
12 mice or S-D rats, so the results of Keil et al. (2009) are not necessarily discordant with the other
13 studies
14 For autoimmune effects, the cRfC from the only suitable inhalation study (Kaneko et al.,
15 2000) is 0.07 ppm. This study reported changes in immunoreactive organs (i.e., liver and spleen)
16 in autoimmune-prone mice. BMD modeling was not feasible, so a LOAEL was used as the
17 POD. The standard value of 10 was used for the LOAEL-to-NOAEL UF because the
18 inflammation was reported to include sporadic necrosis in the hepatic lobules at the LOAEL, so
19 this was considered an adverse effect. A value of 3 was used for the human (intraspecies)
20 variability UF because the effect was induced in autoimmune-prone mice, a sensitive mouse
21 strain for such an effect. The cRfDs from the oral studies (Keil et al., 2009; Griffin et al., 2000;
22 Cai et al., 2008) spanned about a 100-fold range from 0.004-0.5 mg/kg/d. Each of the studies
23 used different markers for autoimmune effects, which may explain the over 100-fold range of
24 PODs (0.4-60 mg/kg/d). The most sensitive endpoint, reported by Keil et al. (2009), was
25 increases in anti-dsDNA and anti-ssDNA antibodies, early markers for systemic lupus
26 erythematosus (SLE), in B6C3Flmice exposed to the lowest tested dose of 0.35 mg/kg/d,
27 yielding a cRfD of 0.004 mg/kg/d. Therefore, the results of Keil et al. (2009) are not discordant
28 with the higher PODs and cRfDs derived from the other oral studies that examined more frank
29 autoimmune effects.
30 For immunosuppressive effects, the only suitable inhalation study (Woolhiser et al.,
31 2006) gave a cRfC of 0.08 ppm. The cRfDs from the only suitable oral study (Sanders et al.,
32 1982) ranged from 0.06 mg/kg/d to 2 mg/kg/d, based on different markers for
33 immunosuppression. Woolhiser et al. (2006) reported decreased PFC response in rats. Data
34 from Woolhiser et al. (2006) were amenable to BMD modeling, but there is notable uncertainty
35 in the modeling. First, it is unclear what should constitute the cut-point for characterizing the
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-18 DRAFT—DO NOT CITE OR QUOTE
-------
1 change as minimally biologically significant, so a BMR of 1 control SD change was used. In
2 addition, the dose-response relationship is supralinear, and the highest exposure group was
3 dropped to improve the fit to the low-dose data points. Nonetheless, the uncertainty in the BMD
4 modeling is no greater than the uncertainty inherent in the use of a LOAEL or NOAEL. The
5 more sensitive endpoints reported by Sanders et al. (1982), both of which were in female mice
6 exposed to a LOAEL of 18 mg/kg/d TCE in drinking water for 4 months, were decreased
7 cell-mediated response to sheep red blood cells (sRBC) and decreased stem cell bone
8 recolonization, a sign of impaired bone marrow function. The cRfD based on these endpoints is
9 0.06 mg/kg/d, with a LOAEL-to-NOAEL UF of 3 because, although the immunosuppressive
10 effects may not be adverse in and of themselves, multiple effects were observed suggesting
11 potentially less resilience to an insult requiring an immunological response.
12 In summary, there is high qualitative confidence for TCE immunotoxicity and moderate
13 confidence in the cRfCs and cRfDs that can be derived from the available studies. Decreased
14 thymus weight reported at relatively low exposures in nonautoimmune-prone mice is a clear
15 indicator of immunotoxicity (Keil et al., 2009), and is therefore considered a candidate critical
16 effect. A number of studies have also reported changes in markers of immunotoxicity at
17 relatively low exposures. Therefore, among markers for autoimmune effects, the more sensitive
18 measures of autoimmune changes in liver and spleen (Kaneko et al., 2000) and increased
19 anti-dsDNA and anti-ssDNA antibodies (Keil et al., 2009) are considered the candidate critical
20 effects. Similarly, for markers of immunosuppression, the more sensitive measures of decreased
21 PFC response (Woolhiser et al., 2006), decreased stem cell bone marrow recolonization, and
22 decreased cell-mediated response to sRBC (both from Sanders et al., 1982) are considered the
23 candidate critical effects.
24
25 5.1.2.6. Candidate Critical Respiratory Tract Effects on the Basis of Applied Dose
26 As summarized in Section 4.11.1.5, available data are suggestive of TCE causing
27 respiratory tract toxicity, based primarily on short-term studies in mice and rats. However, these
28 studies are generally at high inhalation exposures and over durations of less than 2 weeks. Thus,
29 these were not considered critical effects because such data are not necessarily indicators of
30 longer-term effects at lower exposure and are not likely to be the most sensitive noncancer
31 endpoints for chronic exposures. Therefore, cRfCs and cRfDs were not developed for them.
32
33 5.1.2.7. Candidate Critical Reproductive Effects on the Basis of Applied Dose
34 As summarized in Section 4.11.1.6, both human and experimental animal studies have
35 associated TCE exposure with adverse reproductive effects. The strongest evidence of hazard is
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-19 DRAFT—DO NOT CITE OR QUOTE
-------
1 for effects on sperm and male reproductive outcomes, with evidence from multiple human
2 studies and several experimental animal studies. There is also substantial evidence for effects on
3 the male reproductive tract and male serum hormone levels, as well as evidence for effects on
4 male reproductive behavior. There are fewer data and more limited support for effects on female
5 reproduction. The PODs, UFs, and resulting cRfDs and cRfCs for the effects from the suitable
6 reproductive studies are summarized in Table 5-4.
7
8 5.1.2.7.1. Male reproductive effects (effects on sperm and reproductive tract). A number of
9 available studies have reported functional and structural changes in sperm and male reproductive
10 organs and effects on male reproductive outcomes following TCE exposure (see Table 5-4). A
11 cRfC of 0.014 ppm was derived based on hyperzoospermia reported in the available human
12 study (Chia et al., 1996), but there is substantial uncertainty in this estimate due to multiple
13 issues.12 Among the rodent inhalation studies, the cRfC of 0.2 ppm based on increased abnormal
14 sperm in the mouse reported by Land et al. (1981) is considered relatively reliable because it is
15 based on BMD modeling rather than a LOAEL or NOAEL. However, increased sperm
16 abnormalities do not appear to be the most sensitive effect, as Kumar et al. (2000a, b, 2001)
17 reported a similar POD to be a LOAEL for reported multiple effects on sperm and testes, as well
18 as altered testicular enzyme markers in the rat. Although there are greater uncertainties
19 associated with the cRfC of 0.02 ppm for this effect and a composite UF of 3,000 was applied to
20 the POD, the uncertainties are generally typical of those encountered in RfC derivations.
21
12Mean exposure estimates for the exposure groups were limited because they were defined in terms of ranges and
because they were based on mean urinary TCA (mg/g creatinine). There is substantial uncertainty in the conversion
of urinary TCA to TCE exposure level (see discussion of Mhiri et al. [2004], for neurotoxicity, above). In addition,
there was uncertainty about the adversity of the effect being measured. While rodent evidence supports effects of
TCE on sperm, and hyperzoospermia has reportedly been associated with infertility, the adversity of the
hyperzoospermia (i.e., high sperm density) outcome measured in the Chia et al. (1996) study is unclear.
Furthermore, the cut-point used to define hyperzoospermia in this study (i.e., >120 million sperm per mL ejaculate)
is lower than some other reported cut-points, such as 200 and 250 million sperm/mL. A BMR of 10% extra risk was
used on the assumption that this is a minimally adverse effect, but biological significance of this effect level is
unclear.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-20 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-4. Reproductive effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
to
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
'J'comp
cRfC
(ppm)
cRfD
(mg/kg/d)
I
Effect; comments
Effects on sperm, male reproductive outcomes
Chiaet al., 1996
Landetal., 1981
Kan et al., 2007
Xuetal.,2004
Kumar etal., 2000a,
2001 b
George etal., 1985
DuTeauxetal., 2004
Human
Mouse
Mouse
Mouse
Rat
Rat
Mouse
Rat
BMDL
BMDL
LOAEL
LOAEL
LOAEL
LOAEL
NOAEL
LOAEL
1.43
46.9
180
180
45
45
362
141
10
10
10
10
10
1
1
10
1
3
3
3
3
3
10
10
10
10
10
10
10
10
10
10
1
1
10
10
10
10
1
10
1
1
1
1
1
1
1
1
100
300
3,000
3,000
3,000
300
100
10,000C
0.014
0.16
0.060
0.060
0.015
0.15
3.6
0.014
Hyperzoospermia; exposure I
estimates based on U-TCA from
Ikeda et al. (1972); BMR = 10%
extra risk
| abnormal sperm; BMR = 0.5 SD
| abnormal sperm; Land et al.
(1981) cRfC preferred due to BMD
modeling
J, fertilization
Multiple sperm effects, increasing
severity from 12 to 24 weeks
Pre- and postimplantation losses;
UFsc = 1 due to exposure covered
time period for sperm development;
higher response for preimplantation
losses
J, sperm motility
J, ability of sperm to fertilize in vitro
Male reproductive tract effects
Forkert etal., 2002 ;
Kan etal. ,2007
Kumar etal., 2000a
2001 b
George etal., 1985
George etal., 1986
Mouse
Rat
Mouse
Rat
LOAEL
LOAEL
NOAEL
NOAEL
180
45
362
186
10
10
1
1
3
3
10
10
10
10
10
10
10
10
1
1
1
1
1
1
3,000
3,000
100
100
0.060
0.015
3.6
1.9
Effects on epididymis epithelium
Testes effects, altered testicular
enzyme markers, increasing severity
from 1 2 to 24 weeks
J, testis/seminal vesicle weights
| testis/epididymis weights
TO
Co
Y1
to ^
~
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-4. Reproductive effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
(continued)
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UFloael
UFdb
[1C b
"-"^comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Female maternal weight gain
Carney et al., 2006
Schwetz et al., 1975
Narotsky etal., 1995
Manson et al., 1984
George etal., 1986
Rat
Rat
Rat
Rat
Rat
BMDL
LOAEL
BMDL
NOAEL
NOAEL
10.5
88
108
100
186
1
1
1
1
1
3
3
10
10
10
10
10
10
10
10
1
10
1
1
1
1
1
1
1
1
30
300
100
100
100
0.35
0.29
1.1
1.0
1.9
J, BWgain; BMR = 10% decrease
J, mat BW; Carney et al. (2006) cRfC
preferred due to BMD modeling
I BWgain; BMR = 10% decrease
J, BWgain; Narotsky et al. (1995)
preferred due to BMD modeling
(different strain)
J, postpartum BW; Narotsky et al.
(1995) cRfD preferred due to BMD
modeling
Female reproductive outcomes
Narotsky et al., 1995
Rat
LOAEL
475
1
10
10
10
1
1,000
0.48
Delayed parturition
Reproductive behavior
Zenicketal., 1984
George etal., 1986
Rat
Rat
NOAEL
LOAEL
100
389
1
1
10
10
10
10
1
10
1
1
100
1,000
1.0
0.39
J, copulatory performance in males
J, mating (both sexes exposed)
Reproductive effects from exposure to both sexes
George etal., 1986
Rat
Rat
BMDL
BMDL
179
152
1
1
10
10
10
10
1
1
1
1
100
100
1.8
1.5
J, # litters/pair; BMR = 0.5 SD
J, live pups/litter; BMR = 0.5 SD
TO
Co
Y1
to ^
to o
I
o §
H I
O >
HH Oq
H TO
si
H
W
""Adjusted to continuous exposure unless otherwise noted. For inhalation studies, adjustments yield a POD that is a human equivalent concentration as
recommended for a Category 3 gas in U.S. EPA (1994) in the absence of PBPK modeling. Same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual UFs.
°U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with a composite UF of greater than 3,000;
however, composite UFs exceeding 3,000 are considered here because the derivation of the cRfCs and cRfDs is part of a screening process and the subsequent
application of the PBPK model for candidate critical effects will reduce the values of some of the individual UFs.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded studies/endpoints were selected as candidate critical effects/studies.
-------
1 Standard values of 3, 10, and 10 were used for the interspecies UF, the human variability UF,
2 and the LOAEL-to-NOAEL UF, respectively. In addition, although the study would have
3 qualified as a chronic exposure study based on its duration of 24 weeks (i.e., >10% of lifetime),
4 statistically significant decreases in testicular weight and in sperm count and motility were
5 already observed from subchronic exposure (12 weeks) to the same TCE exposure concentration
6 and these effects became more severe after 24 weeks of exposure. Moreover, several testicular
7 enzyme markers associated with spermatogenesis and germ cell maturation had significantly
8 altered activities after 12 weeks of exposure, with more severe alterations at 24 weeks, and
9 histological changes were also observed in the testes at 12 weeks, with the testes being severely
10 deteriorated by 24 weeks. Thus, since the single exposure level used was already a LOAEL from
11 subchronic exposure, and the testes were even more seriously affected by longer exposures, a
12 subchronic-to-chronic UF of 10 was applied.13 Note that for the cRfC derived for pre and
13 postimplantation losses reported by Kumar et al. (2000a), the subchronic-to-chronic UF was not
14 applied because the exposure covered the time period for sperm development. This cRfC was
15 0.2 ppm, similar to that derived from Land et al. (1981) based on BMD modeling of increases in
16 abnormal sperm.
17 At a higher inhalation POD, Xu et al. (2004) reported decreased fertilization following
18 exposure in male mice, and Forkert et al. (2002) and Kan et al. (2007) reported effects on the
19 epididymal epithelium in male mice. Kan et al. (2007) reported degenerative effects on the
20 epididymis as early as 1 week into exposure that became more severe at 4 weeks of exposure
21 when the study ended; increases in abnormal sperm were also observed. As with the cRfC
22 developed from the Kumar et al. (2000a, b, 2001), a composite UF of 3,000 was applied to these
23 data, but the uncertainties are again typical of those encountered in RfC derivations. Standard
24 values of 3 for the interspecies UF, 10 for the human variability UF, 10 for the
25 LOAEL-to-NOAEL UF, and 10 for the subchronic-to-chronic UF were applied to each of the
26 study PODs.
27 Among the oral studies, cRfDs derived for decreased sperm motility and changes in
28 reproductive organ weights in rodents reported by George et al. (1985, 1986) were relatively
29 high (2-4 mg/kg/d), and these effects were not considered candidate critical effects. The
30 remaining available oral study of male reproductive effects is DuTeaux et al. (2004b), which
31 reported decreased ability of sperm from TCE-exposed rats to fertilize eggs in vitro. This effect
32 occurred in the absence of changes in combined testes/epididymes weight, sperm concentration
^Alternatively, the value of the LOAEL-to-NOAEL UF could have been increased above 10 to reflect the extreme
severity of the effects at the LOAEL after 24 weeks; however, the comparison of the 12-week and 24-week results
gives such a clear depiction of the progression of the effects, it was more compelling to frame the issue as a
subchronic-to-chronic extrapolation issue.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-23 DRAFT—DO NOT CITE OR QUOTE
-------
1 or motility, or histological changes in the testes or epididymes. DuTeaux et al. (2004b)
2 hypothesize that the effect is due to oxidative damage to the sperm. A LOAEL was used as the
3 POD, and the standard UF values of 10 were used for each of the UFs, i.e., the subchronic-to-
4 chronic UF (14-day study; substantially less than the 70-day time period for sperm
5 development), the interspecies UF for oral exposures, the human variability UF, and the
6 LOAEL-to-NOAEL UF. The resulting composite UF was 10,000,14 and this yielded a cRfD of
7 0.01 mg/kg/d. The excessive magnitude of the composite UF, however, highlights the
8 uncertainty in this estimate.
9 In summary, there is high qualitative confidence for TCE male reproductive tract toxicity
10 and lower confidence in the cRfCs and cRfDs that can be derived from the available studies.
11 Relatively high PODs are derived from several studies reporting less sensitive endpoints
12 (George et al., 1985, 1986; Land et al., 1981), and correspondingly higher cRfCs and cRfDs
13 suggest that they are not likely to be critical effects. The studies reporting more sensitive
14 endpoints also tend to have greater uncertainty. For the human study by Chia et al. (1996), as
15 discussed above, there are uncertainties in the characterization of exposure and the adversity of
16 the effect measured in the study. For the Kumar et al. (2000a, b, 2001), Forkert et al. (2002) and
17 Kan et al. (2007) studies, the severity of the sperm and testes effects appears to be continuing to
18 increase with duration even at the end of the study, so it is plausible that a lower exposure for a
19 longer duration may elicit similar effects. For the DuTeaux et al. (2004b) study, there is also
20 duration- and low-dose extrapolation uncertainty due to the short duration of the study in
21 comparison to the time period for sperm development as well as the lack of a NOAEL at the
22 tested doses. Overall, even though there are limitations in the quantitative assessment, there
23 remains sufficient evidence to consider these to be candidate critical effects.
24
25 5.1.2.7.2. Other reproductive effects. With respect to female reproductive effects, several
26 studies reporting decreased maternal weight gain were suitable for deriving candidate reference
27 values (see Table 5-4). The cRfCs from the two inhalation studies (Carney et al., 2006; Schwetz
28 et al., 1975) yielded virtually the same estimate (0.3-0.4 ppm), although the Carney et al. (2006)
29 result is preferred due to the use of BMD modeling, which obviates the need for the 10-fold
30 LOAEL-to-NOAEL UF used for Schwetz et al. (1975) (the other UFs, with a product of 30, were
31 the same). The cRfDs for this endpoint from the three oral studies were within 3-fold of each
32 other (1-3 mg/kg/d), with the same composite UFs of 100. The most sensitive estimate of
14U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values
with a composite UF of greater than 3,000; however, composite UFs exceeding 3,000 are considered here because
the derivation of the cRfCs and cRfDs is part of a screening process and the subsequent application of the PBPK
model for candidate critical effects will reduce the values of some of the individual UFs.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 Narotsky et al. (1995) is preferred due to the use of BMD modeling and the apparent greater
2 sensitivity of the rat strain used.
3 With respect to other reproductive effects, the most reliable cRfD estimates of about
4 2 mg/kg/d, derived from BMD modeling with composite UFs of 100, are based on decreased
5 litters/pair and decreased live pups/litter in rats reported in the continuous breeding study of
6 George et al. (1986). Both of these effects were considered severe adverse effects, so a BMR of
7 a 0.5 control SD shift from the control mean was used. Somewhat lower cRfDs of
8 0.4-1 mg/kg/d were derived based on delayed parturition in females (Narotsky et al., 1995),
9 decreased copulatory performance in males (Zenick et al., 1984), and decreased mating for both
10 exposed males and females in cross-over mating trials (George et al., 1986), all with composite
11 UFs of 100 or 1,000 depending on whether a LOAEL or NOAEL was used.
12 In summary, there is moderate confidence both in the hazard and the cRfCs and cRfDs
13 for reproductive effects other than the male reproductive effects discussed previously. While
14 there are multiple studies suggesting decreased maternal body weight with TCE exposure, this
15 systemic change may not be indicative of more sensitive reproductive effects. None of the
16 estimates developed from other reproductive effects is particularly uncertain or unreliable.
17 Therefore, delayed parturition (Narotsky et al., 1995) and decreased mating (George et al.,
18 1986), which yielded the lowest cRfDs, were considered candidate critical effects. These effects
19 were also included so that candidate critical reproductive effects from oral studies would not
20 include only that reported by DuTeaux et al. (2004b), from which deriving the cRfD entailed a
21 higher degree of uncertainty.
22
23 5.1.2.8. Candidate Critical Developmental Effects on the Basis of Applied Dose
24 As summarized in Section 4.11.1.7, both human and experimental animal studies have
25 associated TCE exposure with adverse developmental effects. Weakly suggestive epidemiologic
26 data and fairly consistent experimental animal data support TCE exposure posing a hazard for
27 increased prenatal or postnatal mortality and decreased pre or postnatal growth. In addition,
28 congenital malformations following maternal TCE exposure have been reported in a number of
29 epidemiologic and experimental animal studies. There is also some support for TCE effects on
30 neurological and immunological development. Available human studies, while indicative of
31 hazard, did not have adequate exposure information for quantitative estimates of PODs, so only
32 experimental animal studies are considered here. The PODs, UFs, and resulting cRfDs and
33 cRfCs for the effects from the suitable developmental studies are summarized in Table 5-5.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-25 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-5. Developmental effects in studies suitable for dose-response, and corresponding cRfCs and cRfDs
to
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
'J'comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Pre and postnatal mortality
George etal., 1985
Narotsky etal., 1995
Manson et al., 1984
Healeyetal., 1982
Narotsky etal., 1995
Mouse
Rat
Rat
Rat
Rat
Rat
NOAEL
LOAEL
NOAEL
LOAEL
BMDL
BMDL
362
475
100
17
469
32.2
1
1
1
1
1
1
10
10
10
3
10
10
10
10
10
10
10
10
1
10
1
10
1
1
1
1
1
1
1
1
100
1,000
100
300
100
100
0.057
3.6
0.48
1.0
4.7
0.32
| perinatal mortality
Postnatal mortality; Manson et al.
(1 984) cRfD preferred for same
endpoint due to NOAEL vs. LOAEL
| neonatal death
Resorptions
Prenatal loss; BMR = 1% extra risk
Resorptions; BMR = 1% extra risk
Pre and postnatal growth
Healeyetal., 1982
Narotsky etal., 1995
George etal., 1985
George etal., 1986
Rat
Rat
Mouse
Rat
LOAEL
NOAEL
NOAEL
BMDL
17
844
362
79.7
1
1
1
1
3
10
10
10
10
10
10
10
10
1
1
1
1
1
1
1
300
100
100
100
0.057
8.4
3.6
0.80
J, fetal weight; skeletal effects
J, fetal weight
J, fetal weight
j BW at d21 ; BMR = 5% decrease
Congenital defects
Narotsky etal., 1995
Johnson et al., 2003
Rat
Rat
Rat
BMDL
BMDL
BMDL
60
0.0146
0.0207
1
1
1
10
10
10
10
10
10
1
1
1
1
1
1
100
100
100
0.60
0.00015
0.00021
Eye defects; low BMR (1%), but
severe effect and low bkgd. rate
(<1%)
Heart malformations (litters);
BMR = 10% extra risk (only -1/10
from each litter affected); highest-
dose group (1,000-fold higher than
next highest) dropped to improve
model fit.
Heart malformations (pups);
BMR = 1% extra risk; preferred due
to accounting for intralitter effects via
nested model and pups being the
unit of measure; highest-dose group
(1,000-fold higher than next highest)
dropped to improve model fit
TO
Co
Y1
to ^
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5 5. Developmental effects in studies suitable for dose-response, and corresponding cRfCs
and cRfDs (continued)
Effect type
Supporting studies
Species
POD
type
PODa
UFSC
UFis
UFh
UFloael
UFdb
[1C b
"-"^comp
cRfC
(ppm)
cRfD
(mg/kg/d)
Effect; comments
Developmental neurotoxicity
George etal., 1986
Fredricksson et al.,
1993
Taylor etal., 1985
Isaacson and Taylor,
1989
Rat
Mouse
Rat
Rat
BMDL
LOAEL
LOAEL
LOAEL
72.6
50
45
16
1
3
1
1
10
10
10
10
10
10
10
10
1
10
10
10
1
1
1
1
100
3,000
1,000
1,000
0.73
0.017
0.045
0.016
J, locomotor activity; BMR = doubling
of traverse time; results from
females (males similar with
BMDL = 92)
J, rearing postexposure; pup gavage
dose; No effect at tested doses on
locomotion behavior; UFsc = 3
because exposure only during
PND10-16
| exploration postexposure;
estimated dam dose; Less sensitive
than Isaacson and Taylor (1989), but
included because exposure is
preweaning, so can utilize PBPK
model
J, myelination in hippocampus;
estimated dam dose
Developmental immunotoxicity
Peden-Adams et al.,
2006
Mouse
LOAEL
0.37
1
10
10
10
1
1,000
0.00037
j PFC, tDTH; POD is estimated
dam dose (exposure throughout
gestation and lactation + to 3 or 8
wks of age); UF LOAEL = 1 0 since |
DTH and also multiple immuno.
effects
Co
Y1 %
to co
^ §
^
I
gfr
r* TO
r> <*i
H |
O §
°§
^ s-
g|
o^
HH Oq
H TO
SI
51
B*
H
W
""Adjusted to continuous exposure unless otherwise noted. For inhalation studies, adjustments yield a POD that is a human equivalent concentration as
recommended for a Category 3 gas in U.S. EPA (1994) in the absence of PBPK modeling. Same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded studies/endpoints were selected as candidate critical effects/studies.
-------
1 For pre and postnatal mortality and growth, a cRfC of 0.06 ppm for resorptions,
2 decreased fetal weight, and variations in skeletal development indicative of delays in ossification
3 was developed based on the single available (rat) inhalation study considered (Healy et al., 1982)
4 and utilizing the composite UF of 300 for an inhalation POD that is a LOAEL. The cRfDs for
5 pre and postnatal mortality derived from oral studies were within about a 10-fold range of
6 0.4-5 mg/kg/d, depending on the study and specific endpoint assessed. Of these, the estimate
7 based on Narotsky et al. (1995) rat data was both the most sensitive and most reliable cRfD. The
8 dose response for increased full-litter resorptions from this study is based on BMD modeling.
9 Because of the severe nature of this effect, a BMR of 1% extra risk was used. The ratio of the
10 resulting BMD to the BMDL was 5.7, which is on the high side, but given the severity of the
11 effect and the low background response, a judgment was made to use 1% extra risk.
12 Alternatively, a 10% extra risk could have been used, in which case the POD would have been
13 considered more analogous to a LOAEL than a NOAEL, and a LOAEL-to-NOAEL UF of 10
14 would have been applied, ultimately resulting in the same cRfD estimate. The cRfDs for altered
15 pre and postnatal growth developed from the oral studies ranged about 10-fold from
16 0.8-8 mg/kg/d, all utilizing the composite UFs for the corresponding type of POD. The cRfDs
17 for decreased fetal weight, both of which were based on NOAELs, were consistent, being about
18 2-fold apart (Narotsky et al., 1995; George et al., 1985). The cRfD based on postnatal growth at
19 21 days, reported in George et al. (1986), was lower and is preferred because it was based on
20 BMD modeling. A BMR of 5% decrease in weight was used for postnatal growth at 21 days
21 because decreases in weight gain so early in life were considered similar to effects on fetal
22 weight.
23 For congenital defects, there is relatively high confidence in the cRfD for eye defects in
24 rats reported in Narotsky et al. (1995), derived using a composite UF of 100 for BMD modeling
25 in a study of duration that encompasses the full window of eye development. However, the most
26 sensitive developmental effect by far was heart malformations in the rat reported by
27 Johnson et al. (2003), yielding a cRfD estimate of 0.0002 mg/kg/d, also with a composite UF of
28 100. As discussed in detail in Section 4.8 and summarized in Section 4.11.1.7, although this
29 study has important limitations, the overall weight of evidence supports an effect of TCE on
30 cardiac development, and this is the only study of heart malformations available for conducting
31 dose-response analysis. Individual data were kindly provided by Dr. Johnson (personal
32 communication from Paula Johnson, University of Arizona, to Susan Makris, U.S. EPA,
33 25 August 2008), and, for analyses for which the pup was the unit of measure, BMD modeling
34 was done using nested models because accounting for the intralitter correlation improved model
35 fit. For these latter analyses, a 1% extra risk of a pup having a heart malformation was used as
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 the BMR because of the severity of the effect, since, for example, some of the types of
2 malformations observed could have been fatal. The ratio of the resulting BMD to the BMDL
3 was about three.
4 For developmental neurotoxicity, the cRfD estimates based on the four oral studies span a
5 wide range from 0.02 to 0.8 mg/kg/d. The most reliable estimate, with a composite UF of 100, is
6 based on BMD modeling of decreased locomotor activity in rats reported in George et al. (1986),
7 although a nonstandard BMR of a 2-fold change was selected because the control SD appeared
8 unusually small. The cRfDs developed for decreased rearing postexposure in mice (Fredricksson
9 et al., 1993), increased exploration postexposure in rats (Taylor et al., 1985) and decreased
10 myelination in the hippocampus of rats (Isaacson and Taylor, 1989), while being more than
11 10-fold lower, are all within a 3-fold range of 0.02-0.05 mg/kg/d. Importantly, there is some
12 evidence from adult neurotoxicity studies of TCE causing demyelination, so there is additional
13 biological support for the latter effect. There is greater uncertainty in the Fredricksson et al.
14 (1993), the cRfD for which utilized a subchronic-to-chronic UF of three rather than one, because
15 exposure during postnatal day (PND) 10-16 does not cover the full developmental window (Rice
16 and Barone, 2000). The cRfDs derived from Taylor et al. (1985) and (Isaacson and Taylor,
17 1989) used the composite UF of 1,000 for a POD that is a LOAEL. While there is greater
18 uncertainty in these endpoints, none of the uncertainties is particularly high, and they also appear
19 to be more sensitive indicators of developmental neurotoxicity than that from George et al.
20 (1986).
21 A cRfD of 0.0004 mg/kg/d was developed from the study (Peden-Adams et al., 2006)
22 that reported developmental immunotoxicity. The main effects observed were significantly
23 decreased PFC response and increased delayed-type hypersensitivity. The data on these effects
24 were kindly provided by Dr. Peden-Adams (personal communication from Margie
25 Peden-Adams, Medical University of South Carolina, to Jennifer Jinot, U.S. EPA,
26 26 August 2008); however, the dose-response relationships were sufficiently supralinear that
27 attempts at BMD modeling did not result in adequate fits to these data. Thus, the LOAEL was
28 used as the POD. Although decreased PFC response may not be considered adverse in and of
29 itself, a LOAEL-to-NOAEL UF of 10 was used because of the increased delayed-type
30 hypersensitivity at the same dose. While there is uncertainty in this estimate, it is notable that
31 decreased PFC response was also observed in an immunotoxicity study in adult animals
32 (Woolhiser et al., 2006), lending biological plausibility to the effect.
33 In summary, there is moderate-to-high confidence both in the hazard and the cRfCs and
34 cRfDs for developmental effects of TCE. It is also noteworthy that the PODs for the more
35 sensitive developmental effects were similar to or, in most cases, lower than the PODs for the
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 more sensitive reproductive effects, suggesting that developmental effects are not a result of
2 paternal or maternal toxicity. Among inhalation studies, cRfCs were only developed for effects
3 in rats reported in Healy et al. (1982), so the effects of resorptions, decreased fetal weight, and
4 delayed skeletal ossification were considered candidate critical developmental effects. Because
5 resorptions were also reported in oral studies, the most sensitive (rat) oral study (and most
6 reliable for dose-response analysis) of Narotsky et al. (1995) was also selected as a candidate
7 critical study for this effect. The confidence in the oral studies and candidate reference values
8 developed for more sensitive endpoints is more moderate, but still sufficient for consideration as
9 candidate critical effects. The most sensitive endpoints by far are the increased fetal heart
10 malformations in rats reported by Johnson et al. (2003) and the developmental immunotoxicity in
11 mice reported by Peden-Adams et al. (2006), and these are both considered candidate critical
12 effects. Neurodevelopmental effects are a distinct type among developmental effects. Thus, the
13 next most sensitive endpoints of decreased rearing postexposure in mice (Fredricksson et al.,
14 1993), increased exploration postexposure in rats (Taylor et al., 1985) and decreased myelination
15 in the hippocampus of rats (Isaacson and Taylor, 1989) are also considered candidate critical
16 effects.
17
18 5.1.2.9. Summary ofcRfCs, cRfDs, and Candidate Critical Effects
19 An overall summary of the cRfCs, cRfDs, and candidate critical effects across the health
20 effect domains is shown in Tables 5-6-5-7. These tables present, for each type of noncancer
21 effect, the relative ranges of the cRfC and cRfD developed for the different endpoints. The
22 candidate critical effects selected above for each effect domain are shown in bold. As discussed
23 above, these effects were generally selected to represent the most sensitive endpoints, across
24 species where possible. From these candidate critical effects, candidate reference values based
25 on internal dose metrics from the PBPK model (p-cRfCs and p-cRfDs) were developed where
26 possible. Application of the PBPK model is discussed in the next section.
27
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-30 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-6. Ranges of cRfCs based on applied dose for various noncancer effects associated with inhalation TCE
exposure
cRfC range
(ppm)
10-100
1-10
0.1-1
0.01-0.1
Neurological
Impaired visual
discrimination (rat)
Ototoxicity (rat)
Hyperactivity (rat)
Changes in locomotor activity
(rat)
Trigeminal nerve effects
(human)
Impaired visual function
(rabbit)
J, regeneration of sciatic
nerve (rat)
J, regeneration of sciatic
nerve (mouse)
Disturbed wakefulness (rat)
Systemic/organ-specific
Kidney
meganucleocytosis
(rat)
| kidney weight
(mouse)
| liver weight (rat)
| liver weight (mouse)
| kidney weight (rat)
Immunological
J, PFC response (rat)
Autoimmune changes
(MRI^lpr/lpr
mouse)
Reproductive
J, maternal body weight gain
(rat)
t abnormal sperm (mouse)
pre/postimplantation losses
(male rat exp)
Effects on epididymis
epithelium (mouse)
J, fertilization (male mouse
exp)
Testes and sperm effects (rat)
Hyperzoospermia (human)
Developmental
Resorptions (female rat)
J, fetal weight (rat)
Skeletal effects (rat)
Endpoints in bold were selected as candidate critical effects (see Sections 5.1.2.1-5.1.2.8).
-------
Table 5-7. Ranges of cRfDs based on applied dose for various noncancer effects associated with oral TCE
exposure
cRfD range
(mg/kg/d)
1-10
0.1-1
0.01-0.1
0.001-0.01
io-4-o.ooi
Neurological
t neuromuscular changes
(rat)
t # rears (rat)
t foot splaying (rat)
Trigeminal nerve effect
(rat)
Degeneration of
dopaminergic
neurons (rat)
Demyelination in
hippocampus (rat)
Systemic/organ-specific
4 BW (mouse)
t liver weight (mouse)
| BW (mouse)
| BW (rat)
Toxic nephropathy &
meganucleocytosis (other
rat strains/sexes & mouse)
Toxic nephropathy (female
Marshall rat)
Immunological
4 humoral response to
sRBC (mouse)
Signs of autoimmune
hepatitis (MRL +/+
mouse)
Inflamm. in various tissues
(MRL +/+ mouse)
4 cell-mediated response
to sRBC (mouse)
4 stem cell bone marrow
recolonization (mouse)
t anti-dsDNA & anti-
ssDNA Abs (early
marker for SLE)
(mouse)
4 thymus weight (mouse)
Reproductive
4 testis/seminal vesicle
weight (mouse)
J, sperm motility (mouse)
t testis/epididymis weight
(rat)
4 litters/pair (rat)
4 live pups/litter (rat)
4 BW gain (rat)
4 copulatory performance
(rat)
Delayed parturition (rat)
4 mating (rat)
4 ability of sperm to
fertilize (rat)
Developmental
4 fetal weight (rat)
Prenatal loss (rat)
4 fetal weight (mouse)
t neonatal mortality
(mouse, rat)
4BWatPND21(rat)
4 locomotor activity (rat)
Eye defects (rat)
Resorptions (rat)
t exploration (postexp.)
(rat)
4 rearing (postexp.)
(mouse)
4 myelination in
hippocampus (rat)
Immunotox (4 PFC, |
DTK) (B6C3F1 mouse)
Heart malformations (rat)
Endpoints in bold were selected as candidate critical effects (see Sections 5.1.2.1-5.1.2.8).
-------
1 5.1.3. Application of Physiologically Based Pharmacokinetic (PBPK) Model to Inter- and
2 Intraspecies Extrapolation for Candidate Critical Effects
3 For the candidate critical effects, the use of PBPK modeling of internal doses could
4 justify, where appropriate, replacement of the uncertainty factors for pharmacokinetic inter and
5 intraspecies extrapolation. For more details on PBPK modeling used to estimate levels of dose
6 metrics corresponding to different exposure scenarios in rodents and humans, as well as a
7 qualitative discussion of the uncertainties and limitations of the model, see Section 3.5.
8 Quantitative analyses of the PBPK modeling uncertainties and their implications for dose-
9 response assessment, utilizing the results of the Bayesian analysis of the PBPK model, are
10 discussed separately in Section 5.1.4.
11
12 5.1.3.1. Selection of Dose Metrics for Different Endpoints
13 One area of scientific uncertainty in noncancer dose-response assessment is the
14 appropriate scaling between rodent and human doses for equivalent responses. Another way one
15 could regard the UF for interspecies extrapolation discussed above for applied dose is that it
16 reflects the combination of an adjustment factor due to the expected scaling of
17 toxicologically-equivalent doses across species (commonly attributed to pharmacokinetics) and a
18 factor accounting for uncertainty in the appropriate interspecies extrapolation for specific
19 noncancer effects from a specific chemical exposure (commonly attributed to
20 pharmacodynamics). For considering how to scale internal doses predicted by a PBPK model
21 across species, it is useful to consider two possible interpretations of the "adjustment"
22 component (UFis.adj), and their consequent implications for the remaining "uncertainty"
23 component (UFis.unc) of the interspecies UF.
24 The first (denoted "empirical dosimetry") interpretation is that the "adjustment" is based
25 on the empirical finding that scaling the delivered dose rate by body weight to the % power
26 results in equivalent toxicity (e.g., Travis and White, 1988; U.S. EPA, 1992), since the 3-fold
27 factor comprising this UFis.adj component is similar to what would result from body weight
28 -% power-scaling from rats to humans (an adjustment of mg/kg/d dose by (70/0.4) " = 3.6). The
29 scaling of dose by body weight to the 3/4 power is supported biologically by data showing that the
30 rates of both kinetic and dynamic physiologic processes are generally consistent with 3/4 power of
31 body weight scaling across species (U.S. EPA, 1992). Note also that this applies to inhalation
32 exposure because the delivered dose rate in that case is the air concentration multiplied by the
33 ventilation rate, which scales by body weight to the 3/4 power. Applying this interpretation to
34 internal doses would imply that the dose rate of the active moiety delivered to the target tissue,
35 scaled by body weight to the 3/4 power, would be assumed to result in equivalent responses.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 Under this interpretation, the "uncertainty" component, UFis.unc, of the interspecies UF (which is
2 still retained for reference values using PBPK modeling) reflects the possible deviations from the
3 empirically-based "adjustment" due to the kinetics or dynamics for a particular noncancer effect
4 for a particular chemical in the particular species from which human risk is being extrapolated.
5 The second (denoted "concentration equivalence dosimetry") interpretation is consistent
6 with the further hypothesis that the empirical finding (and hence the "adjustment" component of
7 the interspecies UF) is largely pharmacokinetically-driven, so UFis.adj = UFis.pk (e.g.,
8 IPCS, 2005). Under this interpretation, it is hypothesized that, due to the body weight to the %
9 scaling of physiologic flows (cardiac output, ventilation rate, glomerular filtration, etc.) and
10 metabolic rates (enzyme-mediated biotransformation), the "adjustment" component is intended
11 to result in average internal concentrations of the active moiety at the target tissue, which in turn
12 results in equivalent toxicity (NRC, 1986, 1987). Applying this interpretation to internal doses
13 would imply that equal (average) concentrations of the active moiety or moieties at the target
14 tissue would result in equivalent responses. Under this interpretation, the "uncertainty"
15 component of the interspecies UF (which is still retained for reference values using PBPK
16 modeling) reflects the possible deviations from the empirically-based "adjustment" due to the
17 pharmacodynamics (and not pharmacokinetics) for a particular noncancer effect for a particular
18 chemical in the particular species from which human risk is being extrapolated, so
19 UFIs.unc = UFIs.pd.
20 To the extent that production and clearance of the active moiety or moieties all scale by
21 body weight to the 3A power, these two dosimetry interpretations both lead to the same dose
22 metrics and quantitative results. However, these interpretations may lead to different
23 quantitative results when there are deviations of the underlying physiologic or metabolic
24 processes from body weight to the 3A power scaling. For instance, as discussed in Section 3.5,
25 the PBPK model predictions for the area-under-the-curve (AUC) of TCE in blood deviate from
26 the body weight to the % scaling (the scaling is closer to mg/kg/d than mg/kg Yd), so use of this
27 dose metric implicitly assumes the "concentration equivalence dosimetry." In addition, as
28 discussed below, in most cases involving TCE metabolites, only the rate of production of the
29 active moiety(ies) or the rate of transformation through a particular metabolic pathway can be
30 estimated using the PBPK model, and the actual concentration of the active moiety(ies) cannot
31 be estimated due to data limitations. Under "empirical dosimetry," these metabolism rates,
32 which are estimates of the systemic or tissue-specific delivery of the active moiety(ies), would be
33 scaled by body weight to the 3A power to yield equivalent toxicological response. Under
34 "concentration equivalence dosimetry," additional assumptions about the rate of clearance are
35 necessary to specify the scaling that would yield concentration equivalence. In the absence of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-34 DRAFT—DO NOT CITE OR QUOTE
-------
1 data, active metabolites are assumed to be sufficiently stable so that clearance is via enzyme-
2 catalyzed transformation or systemic excretion (e.g., blood flow, glomerular filtration), which
3 scale approximately by body weight to the % power. Therefore, under "concentration
4 equivalence dosimetry," the metabolism rates would also be scaled by body weight to the
5 % power in the absence of additional data.
6 For toxicity that is associated with local (in situ) production of "reactive" metabolites
7 whose concentrations cannot be directly measured in the target tissue, an alternative approach,
8 under "concentration equivalence dosimetry," of scaling by unit tissue mass has been proposed
9 (e.g., Andersen et al., 1987). As discussed by Travis (1990), scaling the rate of local metabolism
10 across species and individuals by tissue mass is appropriate if the metabolites are sufficiently
11 reactive and are cleared by "spontaneous" deactivation (i.e., changes in chemical structure
12 without the need of biological influences). Thus, use of this alternative scaling approach requires
13 that (1) the active moiety or moieties do not leave the target tissue in appreciable quantities (i.e.,
14 are cleared primarily by in situ transformation to other chemical species and/or binding
15 to/reactions with cellular components); and (2) the clearance of the active moieties from the
16 target tissue is governed by biochemical reactions whose rates are independent of body weight
17 (e.g., purely chemical reactions). If these conditions are met, then under the "concentration
18 equivalence dosimetry," the relevant metabolism rates estimated by the PBPK model would be
19 scaled by tissue mass, rather than by body weight to the 3A power.
20 To summarize, the internal dose metric for equivalent toxicological responses across
21 species can be specified by invoking one of two alternative interpretations of the "adjustment"
22 component of the interspecies UF: "empirical dosimetry" based on the rate at which the active
23 moiety(ies) is(are) delivered to the target tissue scaled by body weight to the % power or
24 "concentration equivalence dosimetry" based on matching internal concentrations of the active
25 moiety(ies) in the target tissue. If the active moiety(ies) is TCE itself or a putatively reactive
26 metabolite, the choice of interpretation will affect the choice of internal dose metric. In the
27 discussions of dose metric selections for the individual endpoints below, the implications of both
28 "empirical dosimetry" and "concentration equivalence dosimetry" are discussed.
29 The use of these dose metrics was then also deemed to obviate the need for the
30 pharmacokinetic component, UFh-pk, of the UF for human (intraspecies) variability. Because all
31 the dose metrics used for TCE are for adults, and the dose metrics are not very sensitive to the
32 plausible range of adult body weight, for convenience the body weight % scaling used for
33 interspecies extrapolation was retained for characterization of human variability. However, it
34 should be emphasized that this intraspecies characterization is of pharmacokinetics only, and not
35 pharmacodynamics.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-35 DRAFT—DO NOT CITE OR QUOTE
-------
1 In general, an attempt was made to use tissue-specific dose metrics representing
2 particular pathways or metabolites identified from available data on the role of metabolism in
3 toxicity for each endpoint (discussed in more detail below). The selection was limited to dose
4 metrics for which uncertainty and variability could be adequately characterized by the PBPK
5 model (see Section 3.5). For most endpoints, sufficient information on the role of metabolites or
6 MOA was not available to identify likely relevant dose metrics, and more "upstream" metrics
7 representing either parent compound or total metabolism had to be used. The "primary" or
8 "preferred" dose metric referred to in subsequent tables has the greater biological support for its
9 involvement in toxicity, whereas "alternative" dose metrics are those that may also be plausibly
10 involved (discussed further below). A discussion of the dose metrics selected for particular
11 noncancer endpoints follows.
12
13 5.1.3.1.1. Kidney toxicity (meganucleuocytosis, increased kidney weight, toxic nephropathy).
14 As discussed in Sections 4.4.6-4.4.7, there is sufficient evidence to conclude that TCE-induced
15 kidney toxicity is caused predominantly by GSH conjugation metabolites either produced in situ
16 in or delivered systemically to the kidney. As discussed in Section 3.3.3.2, bioactivation of
17 S-dichlorovinyl glutathione (DCVG), DCVC, and N-acetyl-S-(l,2-dichlrovinyl)-L-cysteine
18 (NAcDCVC) within the kidney, either by beta-lyase, flavin mono-oxygenase (FMO), or
19 cytochrome P450 (CYP), produces reactive species, any or all of which may cause
20 nephrotoxicity. Therefore, multiple lines of evidence support the conclusion that renal
21 bioactivation of DCVC is the preferred basis for internal dose extrapolations for TCE-induced
22 kidney toxicity. However, uncertainties remain as to the relative contribution from each
23 bioactivation pathway; and quantitative clearance data necessary to calculate the concentration of
24 each species are lacking.
25 Under "empirical dosimetry," the rate of renal bioactivation of DCVC would be scaled by
26 body weight to the % power. As discussed above, under "concentration equivalence dosimetry,"
27 when the concentration of the active moiety cannot be estimated, qualitative data on the nature of
28 clearance of the active moiety or moieties can be used to inform whether to scale the rate of
29 metabolism by body weight to the % power or by the target tissue weight. For the beta-lyase
30 pathway, Dekant et al. (1988) reported in trapping experiments that the postulated reactive
31 metabolites decompose to stable (unreactive) metabolites in the presence of water. Moreover,
32 the necessity of a chemical trapping mechanism to detect the reactive metabolites suggests a very
33 rapid reaction such that it is unlikely that the reactive metabolites leave the site of production.
34 Therefore, these data support the conclusion that, for this bioactivation pathway, clearance is
35 chemical in nature and hence species-independent. If this were the only bioactivation pathway,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-36 DRAFT—DO NOT CITE OR QUOTE
-------
1 then scaling by kidney weight would be supported. With respect to the FMO bioactivation
2 pathway, Sausen and Elfarra (1991) reported that after direct dosing of the postulated reactive
3 sulfoxide (DCVC sulfoxide), the sulfoxide was detected as an excretion product in bile. These
4 data suggest that reactivity in the tissue to which the sulfoxide was delivered (the liver, in this
5 case) is insufficient to rule out a significant role for enzymatic or systemic clearance. Therefore,
6 according to the criteria outlined above, for this bioactivation pathway, the data support scaling
7 the rate of metabolism by body weight to the 3/4 power. For P450-mediated bioactivation
8 producing NAcDCVC sulfoxide, the only relevant data on clearance are from a study of the
9 structural analogue to DCVC, fluoromethyl-2,2-difluoro-l-(trifluoromethyl)vinyl ether (FDVE;
10 Sheffels et al., 2004), which reported that the postulated reactive sulfoxide was detected in urine.
11 This suggests that the sulfoxide is sufficiently stable to be excreted by the kidney and supports
12 the scaling of the rate of metabolism by body weight to the % power.
13 Therefore, because the contributions to TCE-induced nephrotoxicity from each possible
14 bioactivation pathway are not clear, and, even under "concentration equivalence dosimetry," the
15 scaling by body weight to the % power is supported for two of the three bioactivation pathways,
16 it is decided here to scale the DCVC bioactivation rate by body weight to the 3/4 power. The
17 primary internal dose metric for TCE-induced kidney tumors is thus, the weekly rate of DCVC
18 bioactivation per unit body weight to the 3/4 power (ABioactDCVCBW34 [mg/kgy7week]).
19 However, it should be noted that due to the larger relative kidney weight in rats as compared to
20 humans, scaling by kidney weight instead of body weight to the % power would only change the
21 quantitative interspecies extrapolation by about 2-fold,15 so the sensitivity of the results to the
22 scaling choice is relatively small. In addition, quantitative estimates for this dose metric are only
23 available in rats and humans, and not in mice. Accordingly, this metric was only used for
24 extrapolating results from rat toxicity studies.
25 To summarize, under the "empirical dosimetry" approach, the underlying assumption for
26 the ABioactDCVCBW34 dose metric is that equalizing the rate of renal bioactivation of DCVC
27 (i.e., local production of active moiety(ies) in the target tissue), scaled by the 3/4 power of body
28 weight, accounts for the "adjustment" component of the interspecies UF and the
29 "pharmacokinetic" component of the intraspecies UF. Under "concentration equivalence
30 dosimetry," the underlying assumptions for the ABioactDCVCBW34 dose metric are that
31 (1) matching the average concentration of reactive species in the kidney accounts for the
32 "adjustment" component of the interspecies UF and the "pharmacokinetic" component of the
15The range of the difference is 2.1-2.4-fold using the posterior medians for the relative kidney weight in rats and
humans from the PBPK model described in Section 3.5 (see Table 3-36), and body weights of 0.3-0.4 kg for rats
and 60-70 kg for humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-37 DRAFT—DO NOT CITE OR QUOTE
-------
1 intraspecies UF ; and (2) the rates of clearance of these reactive species scale by the 3/4 power of
2 body weight (e.g., assumed for enzyme-activity or blood-flow).
3 An alternative dose metric that also involves the GSH conjugation pathway is the amount
4 of GSH conjugation scaled by the 3/4 power of body weight (AMetGSHBW34 [mg/kgy7week]).
5 This dose metric uses the total flux of GSH conjugation as the toxicologically-relevant dose, and,
6 thus, incorporates any direct contributions from DCVG and DCVC, which are not addressed in
7 the DCVC bioactivation metric. Under the "empirical dosimetry" approach, the underlying
8 assumption for the AMetGSHBW34 dose metric is that equalizing the (whole body) rate of
9 production of GSH conjugation metabolites (i.e., systemic production of active moiety[ies]),
10 scaled by the 3/4 power of body weight, accounts for the "adjustment" component of the
11 interspecies UF and the "pharmacokinetic" component of the intraspecies UF. Under
12 "concentration equivalence dosimetry," the AMetGSHBW34 dose metric is consistent with the
13 assumptions that (1) matching the same average concentration of the (relatively) stable upstream
14 metabolites DCVG or DCVC in the kidney (the PBPK model assumes all DCVG and DCVC
15 produced translocates to the kidney) accounts for the "adjustment" component of the interspecies
16 UF and the "pharmacokinetic" component of the intraspecies UF; and (2) the rate of clearance of
17 DCVG or DCVC scales by the % power of body weight (as is assumed for enzyme activity or
18 blood flow). Because of the lack of availability of the DCVC bioactivation dose metric in mice,
19 the GSH conjugation metric is used as the primary dose metric for the nephrotoxicity endpoint in
20 studies of mice.
21 Another alternative dose metric is the total amount of TCE metabolism (oxidation and
22 GSH conjugation together) scaled by the 3/4 power of body weight (TotMetabBW34
23 [mg/kgy7week]). This dose metric uses the total flux of TCE metabolism as the lexicologically
24 relevant dose, and, thus, incorporates the possible involvement of oxidative metabolites, acting
25 either additively or interactively, in addition to GSH conjugation metabolites in nephrotoxicity
26 (see Section 4.4.6). However, this dose metric is given less weight than those involving GSH
27 conjugation because, as discussed in Sections 4.4.6, the weight of evidence supports the
28 conclusion that GSH conjugation metabolites play a predominant role in nephrotoxicity. Under
29 the "empirical dosimetry" approach, the underlying assumption for the TotMetabBW34 dose
30 metric is that equalizing the (whole body) rate of production of all metabolites (i.e., systemic
31 production (and distribution) of active moiety [ies]), scaled by the 3/4 power of body weight,
32 accounts for the "adjustment" component of the interspecies UF and the "pharmacokinetic"
33 component of the intraspecies UF. Under "concentration equivalence dosimetry," the
34 TotMetabBW34 dose metric is consistent with the assumptions that (1) the relative proportions
35 and blood:tissue partitioning of the active metabolites is similar across species; (2) matching the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-38 DRAFT—DO NOT CITE OR QUOTE
-------
1 average concentration of one or more metabolites in the kidney accounts for the "adjustment"
2 component of the interspecies UF and the "pharmacokinetic" component of the intraspecies UF;
3 and (3) the rate of clearance of active metabolites scales by the % power of body weight (e.g.,
4 assumed for enzyme-activity or blood-flow).
5
6 5.1.3.1.2. Liver weight increases (hepatomegaly). As discussed in Section 4.5.6, there is
7 substantial evidence that oxidative metabolism is involved in TCE hepatotoxicity, based
8 primarily on similarities in noncancer effects with a number of oxidative metabolites of TCE
9 (e.g., chloral hydrate [CH], TCA, and dichloroacetic acid [DCA]). While TCA is a stable,
10 circulating metabolite, CH and DCA are relatively short-lived, although enzymatically cleared
11 (see Section 3.3.3.1). As discussed in Section 4.5.6.2.1, there is substantial evidence that TCA
12 alone does not adequately account for the hepatomegaly induced by TCE; therefore, unlike in
13 previous dose-response analyses (Barton and Clewell, 2000, Clewell and Andersen, 2004), the
14 AUC of TCA in plasma or in liver were not considered as dose metrics. However, there are
15 inadequate data across species to quantify the dosimetry of CH and DCA, and other
16 intermediates of oxidative metabolism (such as TCE-oxide or dichloroacetylchloride) may be
17 involved in hepatomegaly. Thus, due to uncertainties as to the active moiety(ies), but given the
18 strong evidence associating TCE liver effects with oxidative metabolism in the liver, hepatic
19 oxidative metabolism is the preferred basis for internal dose extrapolations of TCE-induced liver
20 weight increases. Under "empirical dosimetry," the rate of hepatic oxidative metabolism would
21 be scaled by body weight to the 3A power. As discussed above, under "concentration equivalence
22 dosimetry," when the concentration of the active moiety cannot be estimated, qualitative data on
23 the nature of clearance of the active moiety or moieties can be used to inform whether to scale
24 the rate of metabolism by body weight to the 3A power or by the target tissue weight. However,
25 several of the oxidative metabolites are stable and systemically available, and several of those
26 that are cleared rapidly are metabolized enzymatically, so, according to the criteria discussed
27 above, there are insufficient data to support the conclusions that the active moiety or moieties do
28 not leave the target tissue in appreciable quantities and are cleared by mechanisms whose rates
29 are independent of body weight. Thus, scaling the rate of oxidative metabolism by body weight
30 to the 3/4 power would also be supported under "concentration equivalence dosimetry."
31 Therefore, the primary internal dose metric for TCE-induced liver weight changes is selected to
32 be the weekly rate of hepatic oxidation per unit body weight to the % power (AMetLivlBW34
33 [mg/kg3/4/week]). The use of this dose metric is also supported by the analysis in
34 Section 4.5.6.2.1 showing much more consistency in the dose-response relationships for TCE-
35 induced hepatomegaly across studies and routes of exposure using this metric and the total
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 oxidative metabolism dose metric (discussed below) as compared to the AUC of TCE in blood.
2 It should be noted that due to the larger relative liver weight in mice as compared to humans,
3 scaling by liver weight instead of body weight to the % power would only change the
4 quantitative interspecies extrapolation by about 4-fold,16 so the sensitivity of the results to the
5 scaling choice is relatively modest.
6 To summarize, under the "empirical dosimetry" approach, the underlying assumption for
7 the AMetLivlBW34 dose metric is that equalizing the rate of hepatic oxidation of TCE (i.e.,
8 local production of active moiety(ies) in the target tissue), scaled by the % power of body weight,
9 accounts for the "adjustment" component of the interspecies UF and the "pharmacokinetic"
10 component of the intraspecies UF. Under "concentration equivalence dosimetry," the
11 AMetLivlBW34 dose metric is consistent with the assumptions that (1) oxidative metabolites
12 are primarily generated in situ in the liver; (2) the relative proportions and blood:tissue
13 partitioning of the active oxidative metabolites are similar across species; (3) matching the
14 average concentration of the active oxidative metabolites in the liver accounts for the
15 "adjustment" component of the interspecies UF and the "pharmacokinetic" component of the
16 intraspecies UF; and (4) the rates of clearance of the active oxidative metabolites scale by the
17 % power of body weight (e.g., assumed for enzyme-activity or blood-flow).
18 It is also known that the lung has substantial capacity for oxidative metabolism, with
19 some proportion of the oxidative metabolites produced there entering systemic circulation. Thus,
20 it is possible that extrahepatic oxidative metabolism can contribute to TCE-induced
21 hepatomegaly. Therefore, the total amount of oxidative metabolism of TCE scaled by the
22 3/4 power of body weight (TotOxMetabBW34 [mg/kgy7week]) was selected as an alternative
23 dose metric (the justification for the body weight to the % power scaling is analogous to that for
24 hepatic oxidative metabolism, above). Under the "empirical dosimetry" approach, the
25 underlying assumption for the TotOxMetabBW34 dose metric is that equalizing the rate of total
26 oxidation of TCE (i.e., systemic production of active moiety[ies]), scaled by the % power of
27 body weight, accounts for the "adjustment" component of the interspecies UF and the
28 "pharmacokinetic" component of the intraspecies UF. Under "concentration equivalence
29 dosimetry," this dose metric is consistent with the assumptions that (1) oxidative metabolites
30 may be generated in situ in the liver or delivered to the liver via systemic circulation; (2) the
31 relative proportions and blood:tissue partitioning of the active oxidative metabolites is similar
32 across species; (3) matching the average concentration of the active oxidative metabolites in the
16The range of the difference is 3.5-3.9-fold using the posterior medians for the relative liver weight in mice and
humans from the PBPK model described in Section 3.5 (see Table 3-36), and body weights of 0.03-0.04 kg for mice
and 60-70 kg for humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-40 DRAFT—DO NOT CITE OR QUOTE
-------
1 liver accounts for the "adjustment" component of the interspecies UF and the "pharmacokinetic"
2 component of the intraspecies UF; and (4) the rates of clearance of the active oxidative
3 metabolites scale by the % power of body weight (e.g., enzyme-activity or blood-flow).
4
5 5.1.3.1.3. Developmental toxicity—heart malformations. As discussed in Section 4.8.3.2.1,
6 several studies have reported that the prenatal exposure to TCE oxidative metabolites TCA or
7 DCA also induces heart malformations, suggesting that oxidative metabolism is involved in
8 TCE-induced heart malformations. However, there are inadequate data across species to
9 quantify the dosimetry of DCA, and it is unclear if other products of TCE oxidative metabolism
10 are involved. Therefore, the total amount of oxidative metabolism of TCE scaled by the
11 % power of body weight (TotOxMetabBW34 [mg/kg Vweek]) was selected as the primary dose
12 metric. Under the "empirical dosimetry" approach, the underlying assumption for the
13 TotOxMetabBW34 dose metric is that equalizing the rate of total oxidation of TCE (i.e.,
14 systemic production of active moiety(ies), the same proportion of which is assumed to be
15 delivered to the fetus across species/individuals), scaled by the 3/4 power of body weight,
16 accounts for the "adjustment" component of the interspecies UF and the "pharmacokinetic"
17 component of the intraspecies UF. Under "concentration equivalence dosimetry," this dose
18 metric is consistent with the assumptions that (1) oxidative metabolites are delivered to the fetus
19 via systemic circulation; (2) the relative proportions and blood:tissue partitioning of the active
20 oxidative metabolites is similar across species; (3) matching the average concentration of the
21 active oxidative metabolites in the fetus accounts for the "adjustment" component of the
22 interspecies UF and the "pharmacokinetic" component of the intraspecies UF; and (4) the rates
23 of clearance of the active oxidative metabolites scale by the % power of body weight (e.g.,
24 enzyme-activity or blood-flow).
25 An alternative dose metric that is considered here is the AUC of TCE in (maternal) blood
26 (AUCCBld [mg-hour/L/day]). Under either "empirical dosimetry" or "concentration
27 equivalence dosimetry," this dose metric would account for the possible role of local
28 metabolism, which is determined by TCE delivered in blood via systemic circulation to the target
29 tissue (the flow rate of which scales as body weight to the % power). Moreover, the placenta is a
30 highly perfused tissue, and TCE is known to cross the placenta to the fetus, with rats showing
31 similar (within 2-fold) maternal and fetal blood TCE concentrations (see Section 3.2). Under the
32 "concentration equivalence dosimetry," this dose metric also accounts for the possible role of
33 TCE itself. This dose metric of AUC of TCE in blood is therefore consistent with the
34 assumptions that (1) maternal blood:fetal partitioning of TCE is similar across species, so that
35 similar blood concentrations imply similar fetal concentrations; (2) to the extent that local
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-41 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolism in the placenta or fetus is involved, both in situ metabolism of TCE and clearance of
2 active oxidative metabolites scale by the % power of (adult) body weight (e.g., enzyme-activity
3 or blood-flow); and therefore, (3) matching the average concentrations of TCE in blood accounts
4 for the "adjustment" component of the interspecies UF and the "pharmacokinetic" component of
5 the intraspecies UF.
6
7 5.1.3.1.4. Reproductive toxicity—decreased ability ofsperm to fertilize oocytes. The
8 decreased ability of sperm to fertilize oocytes observed by DuTeaux et al. (2004) occurred in the
9 absence of changes in combined testes/epididymes weight, sperm concentration or motility, or
10 histological changes in the testes or epididymes. However, there was evidence of oxidative
11 damage to the sperm, and DuTeaux et al. (2003) previously reported the ability of the rat
12 epididymis and efferent ducts to metabolize TCE oxidatively. Based on this evidence, DuTeaux
13 et al. (2004) hypothesize that the decreased ability to fertilize is due to oxidative damage to the
14 sperm from local metabolism. Thus, the primary dose metric for this endpoint is selected to be
15 the AUC of TCE in blood (AUCCBld [mg-hour/L/day]), based on the assumption that in situ
16 oxidation of systemically-delivered TCE (the flow rate of which scales as body weight to the
17 % power) is the determinant of toxicity. Under either "empirical dosimetry" or "concentration
18 equivalence dosimetry," this dose metric is therefore consistent with the assumptions that
19 (1) blood:tissue partitioning of TCE is similar across species, so that similar blood concentrations
20 imply similar tissue concentrations; (2) in situ oxidation of TCE and clearance of active
21 oxidative metabolites scale by the 3/4 power of body weight (e.g., enzyme-activity or blood-flow);
22 and, therefore, (3) matching the average concentrations of TCE in blood accounts for the
23 "adjustment" component of the interspecies UF and the "pharmacokinetic" component of the
24 intraspecies UF.
25 Because metabolites causing oxidative damage may be delivered systemically to the
26 target tissue, an alternative dose metric that is considered here is total oxidative metabolism of
27 TCE scaled by the 3/4 power of body weight (TotOxMetabBW34 [mg/kg3/4/day]). Under the
28 "empirical dosimetry" approach, the underlying assumption for the TotOxMetabBW34 dose
29 metric is that equalizing the rate of total oxidation of TCE (i.e., systemic production of active
30 moiety(ies), the same proportion of which is assumed to be delivered to the target tissue across
31 species/individuals), scaled by the % power of body weight, accounts for the "adjustment"
32 component of the interspecies UF and the "pharmacokinetic" component of the intraspecies UF.
33 Under "concentration equivalence dosimetry," this dose metric is consistent with the
34 assumptions that (1) oxidative metabolites are delivered to the target tissue via systemic
35 circulation; (2) the relative proportions and blood:tissue partitioning of the active oxidative
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-42 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolites is similar across species; (3) matching the average concentrations of the active
2 oxidative metabolites in the target tissue accounts for the "adjustment" component of the
3 interspecies UF and the "pharmacokinetic" component of the intraspecies UF; and (4) the rates
4 of clearance of the active oxidative metabolites scale by the 3/4 power of body weight (e.g.,
5 enzyme-activity or blood-flow). Because oxidative metabolites make up the majority of TCE
6 metabolism, total metabolism gives very similar results (within 1.2-fold) to total oxidative
7 metabolism and is therefore not included as a dose metric.
8
9 5.1.3.1.5. Other reproductive and developmental effects and neurological effects and
10 immunologic effects. For all other candidate critical endpoints listed in Tables 5-6-5-7,
11 including developmental effects other than heart malformations and reproductive effects other
12 than decreased ability of sperm to fertilize, there is insufficient information for site-specific
13 determinations of an appropriate dose metric. While TCE metabolites and/or metabolizing
14 enzymes have been reported in some of these tissues (e.g., male reproductive tract), their general
15 roles in toxicity in the respective tissues have not been established. The choice of total
16 metabolism as the primary dose metric is based on the observation that, in general, TCE toxicity
17 is associated with metabolism rather than the parent compound. It is acknowledged that there is
18 no compelling evidence that definitively establishes one metric as more plausible than the other
19 in any particular case. Nonetheless, as a general inference in the absence of specific data, total
20 metabolism is viewed as more likely to be involved in toxicity than the concentration of TCE
21 itself.
22 Therefore, given that the majority of the toxic and carcinogenic responses in many tissues
23 to TCE appears to be associated with metabolism, the primary dose metric is selected to be total
24 metabolism of TCE scaled by the % power of body weight (TotMetabBW34 [mg/kg Yd]). Under
25 the "empirical dosimetry" approach, the underlying assumption for the TotOxMetabBW34 dose
26 metric is that equalizing the rate of total oxidation of TCE (i.e., systemic production of active
27 moiety(ies), the same proportion of which is assumed to be delivered to the target tissue across
28 species/individuals), scaled by the 3/4 power of body weight, accounts for the "adjustment"
29 component of the interspecies UF and the "pharmacokinetic" component of the intraspecies UF.
30 Under "concentration equivalence dosimetry," this dose metric is consistent with the
31 assumptions that (1) metabolites are delivered to the target tissue via systemic circulation; (2) the
32 relative proportions and blood:tissue partitioning of the active metabolites is similar across
33 species; (3) matching the average concentrations of the active metabolites in the target tissue
34 accounts for the "adjustment" component of the interspecies UF and the "pharmacokinetic"
35 component of the intraspecies UF; and (4) the rates of clearance of the active metabolites scale
36 by the 3/4 power of body weight (e.g., enzyme-activity or blood-flow). Because oxidative
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-43 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolites make up the majority of TCE metabolism, total oxidative metabolism gives very
2 similar results (within 1.2-fold) to total metabolism and is therefore not included as a dose
3 metric.
4 An alternative dose metric that is considered here is the AUC of TCE in blood
5 (AUCCBld [mg-hour/L/day]). Under either "empirical dosimetry" or "concentration
6 equivalence dosimetry," this dose metric would account for the possible role of local
7 metabolism, which is determined by TCE delivered in blood via systemic circulation to the target
8 tissue (the flow rate of which scales as body weight to the % power). Under the "concentration
9 equivalence dosimetry," this dose metric also accounts for the possible role of TCE itself. This
10 dose metric is consistent with the assumption that matching the average concentrations of TCE in
11 blood accounts for the "adjustment" component of the interspecies UF and the
12 "pharmacokinetic" component of the intraspecies UF. This dose metric would also be most
13 applicable to tissues that have similar tissue:blood partition coefficients across and within
14 species.
15 Because the PBPK model described in Section 3.5 did not include a fetal compartment,
16 the maternal internal dose metric is taken as a surrogate for developmental effects in which
17 exposure was before or during pregnancy (Taylor et al., 1985; Fredricksson et al., 1993;
18 Narotsky et al., 1995; Johnson et al., 2003). This was considered reasonable because TCE and
19 the major circulating metabolites (TCA and trichloroethanol [TCOH]) appear to cross the
20 placenta (see Sections 3.2, 3.3, and 4.10 [Ghantous et al., 1986; Fisher et al., 1989]), and
21 maternal metabolizing capacity is generally greater than that of the fetus (see Section 4.10). In
22 the cases where exposure continues after birth (Issacson and Taylor, 1989; Peden-Adams et al.,
23 2006), no PBPK model-based internal dose was used. Because of the complicated fetus/neonate
24 dosing that includes transplacental, lactational, and direct (if dosing continues postweaning)
25 exposure, the maternal internal dose is no more accurate a surrogate than applied dose in this
26 case.
27
28
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-44 DRAFT—DO NOT CITE OR QUOTE
-------
1 5.1.3.2. Methods for Inter- and Intraspecies Extrapolation Using Internal Doses17
2 As shown in Figures 5-2 and 5-3, the general approach taken to use the internal dose
3 metrics in deriving HECs and HEDs was to first apply the rodent PBPK model to get rodent
4 values for the dose metrics corresponding to the applied doses in a study reporting noncancer
5 effects. The idPOD is then obtained either directly from the internal dose corresponding to the
6 applied dose LOAEL or NOAEL, or by dose-response modeling of responses with respect to the
7 internal doses to derive a BMDL in terms of internal dose. Separately, the human PBPK model
8 is run for a range of continuous exposures from 10"1 to 2 x 103 ppm or mg/kg/d to obtain the
9 relationship between human exposure and internal dose for the same dose metric used for the
10 rodent. The human equivalent exposure (HEC or HED) corresponding to the idPOD is derived
11 by interpolation. It should be noted that median values of dose metrics were used for rodents,
12 whereas both median and 99th percentile values were used for humans. As discussed in
13 Section 3.5, the rodent population model characterizes study-to-study variation, while, within a
14 study, animals with the same sex/species/strain combination were assumed to be identical
15 pharmacokinetically and represented by the group average (typically the only data reported).
16 Therefore, use of median dose metric values can be interpreted as assuming that the animals in
17 the noncancer toxicity study were all "typical" animals and the idPOD is for a rodent that is
18 pharmacokinetically "typical." In practice, the use of median or mean internal doses for rodents
19 did not make much difference except when the uncertainty in the rodent dose metric was high.
20 The impact of the uncertainty in the rodent PBPK dose metrics is analyzed quantitatively in
21 Section 5.1.4.2.
17An alternative approach (e.g., Clewell et al., 2002) applies the UFs to the internal dose prior to using the human
PBPK model to derive a human exposure level. As noted by Barton and Clewell (2000) for previous TCE PBPK
models, because the human PBPK model for TCE is linear for all the dose metrics over very broad dose and
concentration ranges, essentially identical results would be obtained using this alternative approach. Specifically,
for all the primary dose metrics, the difference in the two approaches is less than 2-fold, with the results from the
critical studies differing by <0.1%. For some studies using AUCBld as an alternative dose metric, the difference
ranged from 3- to 7-fold. Overall, use of the alternative approach would not significantly change the noncancer
dose-response assessment of TCE, and the derived RfC and RfD would be identical.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-45 DRAFT—DO NOT CITE OR QUOTE
-------
Rodent
model
parameters
[distribution (combined
ncertainty and variability)
Human
model
parameters
idistribution
[distribution (separate
ncertainty and variability)
irnedian
Dose-Response Model
or
Human
Internal dose
as function of
applied dose
LOAEUNOAEL
V
;
idPOD (internal
dose unit) =
BMDL or
LOAEL or
NOAEL
X
invert functions of dose
or concentration
Overall /
median^'
"Typical"
human internal
dose as
function
of applied
dose
^l
-n
^
"Typical"
human
equivalent
dose or
concentration
\ Overall
\499thpercentile
"Sensitive"
human internal
dose as
function
of applied
dose
1
"Sensitive"
human
equivalent
dose or
concentration
HEC50 or
HEC99 or
(replaces
POD/UFis.adj)
99
HED
[replaces
POD/(UFis.adj*UFh.pk)]
1
2
3
4
5
Figure 5-2 Flow-chart for dose-response analyses of rodent noncancer
effects using PBPK model-based dose metrics. Square nodes indicate point
values, circle nodes indicate distributions, and the inverted triangle indicates a
(deterministic) functional relationship.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-46 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Uncertainty*
variability
distribution
Human internal
dose
Human inhalation
exposure (ppm)
Rodent internal
dose
Uncertainty &
variability
ribution
Human internal
dose
dose groups
LOAEL/
NOAEL
Human oral exposure
(mg/kg/d)
Lower 99th
percentile
Figure 5-3. Schematic of combined interspecies, intraspecies, and route-to-
route extrapolation from a rodent study LOAEL or NOAEL. In the case
where BMD modeling is performed, the applied dose values are replaced by the
corresponding median internal dose estimate, and the idPOD is the modeled
BMDL in internal dose units.
The human population model characterizes individual-to-individual variation, in addition
to its uncertainty. The "median" value for the HEC or HED was calculated as a point of
comparison but was not actually used for derivation of candidate reference values. Because the
RfC and RfD are intended to characterize the dose below which a sensitive individual would
likely not experience adverse effects, the overall 99th percentile of the combined uncertainty and
variability distribution was used for deriving the HEC and HED (denoted HECgg and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-47 DRAFT—DO NOT CITE OR QUOTE
-------
1 from each idPOD.18 As shown in Figures 5-2 and 5-3, the HECgg or HED99 replaces the quantity
2 POD/(UFis.adj x UFh-pk) in the calculation of the RfC or RfD, i.e., the pharmacokinetic
3 components of the UFs representing interspecies extrapolation and human interindividual
4 variability. As calculated, the extrapolated HECgg and HED99 can be interpreted as being the
5 dose or exposure for which there is 99% likelihood that a randomly selected individual will have
6 an internal dose less than or equal to the idPOD derived from the rodent study. The separate
7 contributions of uncertainty and variability in the human PBPK model are analyzed
8 quantitatively, along with the uncertainty in the rodent PBPK dose metrics as mentioned above,
9 in Section 5.1.4.2.
10 Because they are derived from rodent internal dose estimates, the HEC and HED are
11 derived in the same manner independent of the route of administration of the original rodent
12 study. Therefore, a route-to-route extrapolation from an oral (inhalation) study in rodents to a
13 HEC (HED) in humans is straight-forward. As shown in Tables 5-8-5-13, route-to-route
14 extrapolation was performed for a number of endpoints with low cRfCs and cRfDs to derive
15 p-cRfDs and p-cRfCs.
16
17
18While for uncertainty, a 95th percentile is often selected by convention, there is no explicit guidance on the
selection of the percentile for human toxicokinetic variability. Ideally, all sources of uncertainty and variability
would be included, and percentile selected that is more in line with the levels of risk at which cancer dose-response
is typically characterized (e.g., 106 to 104) along with a level of confidence. However, only toxicokinetic
uncertainty and variability is assessed quantitatively. Because the distribution here incorporates both uncertainty
and variability simultaneously, a percentile higher than the 95th (a conventional choice for uncertainty only) was
selected. However, percentiles greater than the 99th are likely to be progressively less reliable due to the unknown
shape of the tail of the input uncertainty and variability distributions for the PBPK model parameters (which were
largely assumed to be normal or lognormal), and the fact that only 42 individuals were incorporated in the PBPK
model for characterization of uncertainty and inter-individual variability (see Section 3.5). This concern is
somewhat ameliorated because the candidate reference values also incorporate use of UFs to account for inter- and
intraspecies toxicodynamic sensitivity.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-48 DRAFT—DO NOT CITE OR QUOTE
-------
to
Table 5-8. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical neurological effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Trigeminal nerve effects
Ruitjen etal., 1991
Human
LOAEL
HECgg
HECgg
HEDgg
HEDgg
14
5.3
8.3
7.3
14
1
1
1
1
1
1
1
1
1
1
10
3
3
3
3
3
3
3
3
3
1
1
1
1
30
10
10
10
10
0.47
0.53
0.83
0.73
1.4
Trigeminal nerve effects
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Cognitive effects
Isaacson et al., 1990
Rat
LOAEL
HEDgg
HEDgg
HECgg
HECgg
47
9.2
4.3
7.1
2.3
10
10
10
10
10
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
10,000C
1,000
1,000
1,000
1,000
0.0071
0.0023
0.0047
0.0092
0.0043
demyelination in hippocampus
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Mood and sleep disorders
Aritoet al., 1994
Rat
LOAEL
HECgg
HECgg
HEDgg
HEDgg
12
4.8
9.0
6.5
15
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1,000
300
300
300
300
0.012
0.016
0.030
0.022
0.051
Changes in wakefulness
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
TO
Co
Y1
4^ Co
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-8. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical neurological effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Other neurological effects
Kjellstrand et al., 1987
Gash et al., 2007
Rat
Mouse
Rat
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
300
93
257
97
142
150
120
108
120
76
710
53
192
47
363
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3,000
1,000
1,000
1,000
1,000
3,000
1,000
1,000
1,000
1,000
10,000C
1,000
1,000
1,000
1,000
0.10
0.093
0.26
0.050
0.12
0.11
0.047
0.36
0.097
0.14
0.12
0.076
0.071
0.053
0.19
I regeneration of sciatic nerve
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
I regeneration of sciatic nerve
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
degeneration of dopaminergic
neurons
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, 3,000, or 10,000 [see Footnote c below].
°U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with a composite UF of greater than 3,000;
however, composite UFs exceeding 3,000 are considered here because the derivation of the cRfCs and cRfDs is part of a screening process and the application
of the PBPK model for candidate critical effects reduces the values of some of the individual UFs for the p-cRfCs and p-cRfDs.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric.
-------
to
Table 5-9. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical kidney effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Histological changes in kidney
Maltoni, 1986
NCI, 1976
NTP, 1988
Rat
Mouse
rat
BMDL
HECgg
HECgg
HECgg
HEDgg
HEDgg
HEDgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
BMDL
HEDgg
HEDgg
HEDgg
HECgg
HECgg
HECgg
40.2
0.038
0.058
15.3
0.023
0.036
19
620
0.30
48
0.50
42
9.45
0.0034
0.0053
0.74
0.0056
0.0087
0.51
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
3
3
10
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
3
3
1
1
1
1
1
1
1
30
30
30
30
30
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
30
10
10
10
10
10
10
3,000
300
300
300
300
100
10
10
10
10
10
10
1.3
0.0038
0.0058
1.5
0.00165
0.140
0.00056
0.00087
0.051
0.0023
0.0036
1.9
0.21
0.00101
0.160
0.0945
0.00034
0.00053
0.074
meganucleocytosis; BMR = 10%
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34] (route-to-
route)
[AMetGSHBW34] (route-to- route)
[TotMetabBW34] (route-to-route)
toxic nephrosis
[AMetGSHBW34]
[TotMetabBW34]
[AMetGSHBW34] (route-to- route)
[TotMetabBW34] (route-to-route)
toxic nephropathy; BMR = 5%;
female Marshall (most sensitive
sex/strain)
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34] (route-to-
route)
[AMetGSHBW34] (route-to- route)
[TotMetabBW34] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-9. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical kidney effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
| kidney/body weight ratio
Kjellstrand et al.,
1983b
Woolhiser et al.,
2006
Mouse
Rat
BMDL
HECgg
HECgg
HEDgg
HEDgg
BMDL
HECgg
HECgg
HECgg
HEDgg
HEDgg
HEDgg
34.7
0.12
21
0.070
25
15.7
0.013
0.022
11
0.0079
0.013
14
1
1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
3
3
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
30
10
10
10
10
30
10
10
10
10
10
10
1.2
0.012
2.1
0.52
0.0013
0.0022
1.1
0.0070
2.5
0.00079
0.0013
1.4
BMR=10%
[AMetGSHBW34]
[TotMetabBW34]
[AMetGSHBW34] (route-to- route)
[TotMetabBW34] (route-to-route)
BMR = 10%
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34] (route-to-
[AMetGSHBW34] (route-to- route)
[TotMetabBW34] (route-to-route)
TO
Co
Y1
I
Co
§
H I
O ^
HH Oq
H TO
si
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC or cRfD.
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, or 3,000.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric.
H
W
-------
to
Table 5-10. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical liver effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
| liver/body weight ratio
Kjellstrand et al.,
1983b
Woolhiser et al.,
2006
Buben and
O'Flaherty, 1985
Mouse
Rat
Mouse
BMDL
HECgg
HECgg
HEDgg
HEDgg
BMDL
HECgg
HECgg
HEDgg
HEDgg
BMDL
HEDgg
HEDgg
HECgg
HECgg
21.6
9.1
24.9
7.9
25.7
25
19
16
16
17
82
10
13
11
11
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
1
1
1
1
13
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
30
10
10
10
10
30
10
10
10
10
100
10
10
10
10
0.72
0.91
2.5
0.83
1.9
1.6
1.1
1.1
0.79
2.6
1.6
1.7
0.82
1.0
1.3
BMR= 10% increase
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (route-to- route)
[TotOxMetabBW34] (route-to- route)
BMR= 10% increase
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (route-to- route)
[TotOxMetabBW34] (route-to-route)
BMR= 10% increase
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (route-to- route)
[TotOxMetabBW34] (route-to-route)
TO
Co
Y1
I
Co
§
H I
O ^
HH Oq
H TO
si
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, or 3,000.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric.
H
W
-------
to
Table 5-11. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical immunological effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
J, thymus weight
Keil et al., 2009
Mouse
LOAEL
HED99
HED99
HECgg
HECgg
0.35
0.048
0.016
0.033
0.0082
1
1
1
1
1
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1,000
100
100
100
100
0.00033
0.000082
0.00035
0.00048
0.00016
J, thymus weight
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Autoimmunity
Kaneko et al., 2000
Keil et al., 2009
Mouse
Mouse
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
70
37
69
42
57
0.35
0.048
0.016
0.033
0.0082
10
10
10
10
10
1
1
1
1
1
3
3
3
3
3
10
3
3
3
3
3
1
1
1
1
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1,000
300
300
300
300
100
10
10
10
10
0.070
0.12
0.23
0.0033
0.00082
0.14
0.19
0.0035
0.0048
0.0016
Changes in immunoreactive organs -
liver (including sporadic necrosis in
hepatic lobules), spleen; UFh = 3
due to autoimmune-prone mouse
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
t anti-dsDNA & anti-ssDNA Abs
(early markers for SLE)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-11. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical immunological effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
"-"COITIP
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Immunosuppression
Woolhiser et al.,
2006
Sanders et al.,
1982
Rat
Mouse
BMDL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
24.9
11
140
14
91
18
2.5
0.84
1.7
0.43
10
10
10
10
10
1
1
1
1
1
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
1
1
1
1
1
3
3
3
3
3
1
1
1
1
1
1
1
1
1
1
300
100
100
100
100
300
30
30
30
30
0.083
0.11
1.4
0.057
0.014
0.14
0.91
0.060
0.083
0.028
1 RFC response; BMR = 1 SD
change; dropped highest dose
[TotMetabBW34]; all does groups
[AUCCBId] ; all does groups
[TotMetabBW34] (route-to-route) ; all
does groups
[AUCCBId] (route-to-route) ; all does
groups
I stem cell bone marrow
recolonization (sustained); J, cell-
mediated response to sRBC (largely
transient during exposure); females
more sensitive
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, or 3,000.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric
H
W
-------
to
Table 5-12. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical reproductive effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
•Jrcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Effects on sperm, male reproductive outcomes
Chiaetal., 1996
Xuetal.,2004
Kumar et al.,
2000a, 2001 b
DuTeauxet al.,
2004
Human
Mouse
Rat
Rat
BMDL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
1.4
0.50
0.83
0.73
1.6
180
67
170
73
104
45
13
53
16
49
141
16
42
9.3
43
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
1
1
1
1
1
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
1
1
1
1
1
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
100
30
30
30
30
3,000
1,000
1,000
1,000
1,000
3,000
1,000
1,000
1,000
1,000
10,000C
1,000
1,000
1,000
1,000
0.014
0.0017
0.0028
0.060
0.067
0.17
0.015
0.013
0.053
0.0093
0.043
0.024
0.053
0.073
0.10
0.016
0.049
0.014
0.016
0.042
Hyperzoospermia; BMR = 10%
extra risk
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
I fertilization
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Multiple sperm effects, increasing
severity from 1 2 to 24 weeks
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
I ability of sperm to fertilize in vitro
[AUCCBId]
[TotOxMetabBW34]
[AUCCBId] (route-to-route)
[TotOxMetabBW34] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-12. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical reproductive effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
•Jrcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Male reproductive tract effects
Forkert et al.,
2002; Kan etal.,
2007
Kumar et al.,
2000a, 2001 b
Mouse
Rat
LOAEL
HECgg
HECgg
HEDgg
HEDgg
LOAEL
HECgg
HECgg
HEDgg
HEDgg
180
67
170
73
104
45
13
53
16
49
10
10
10
10
10
10
10
10
10
10
3
3
3
3
3
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
10
10
10
10
10
1
1
1
1
1
1
1
1
1
1
3,000
1,000
1,000
1,000
1,000
3,000
1,000
1,000
1,000
1,000
0.060
0.067
0.17
0.015
0.013
0.053
0.073
0.10
0.016
0.049
Effects on epididymis epithelium
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Testes effects, testicular enzyme
markers, increasing severity from 12
to 24 weeks
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Female reproductive outcomes
Narotsky et al.,
1995
Rat
LOAEL
HEDgg
HEDgg
HECgg
HECgg
475
44
114
37
190
1
1
1
1
1
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1,000
100
100
100
100
0.37
1.9
0.48
0.44
1.1
Delayed parturition
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-12. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical reproductive effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
urcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Reproductive behavior
George etal., 1986
Rat
LOAEL
HEDgg
HEDgg
HECgg
HECgg
389
77
52
71
60
1
1
1
1
1
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1,000
100
100
100
100
0.71
0.60
0.39
0.77
0.52
J, mating (both sexes exposed)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to-route)
[AUCCBId] (route-to-route)
Co
^3
SS
1
Y1 1
(j\ co
00 o
-2
^
I
gfr
r* TO
r> co
H |
o §
°§
^ s-
g|
o >
HH Oq
H TO
SI
Sl
B*
H
W
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, 3,000, or 10,000 (see footnote [c] below).
°U.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with a composite UF of greater than 3,000;
however, composite UFs exceeding 3,000 are considered here because the derivation of the cRfCs and cRfDs is part of a screening process and the application
of the PBPK model for candidate critical effects reduces the values of some of the individual UFs for the p-cRfCs and p-cRfDs.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric.
-------
to
Table 5-13. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical developmental effects
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
•Jrcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Pre and postnatal mortality
Healyetal., 1982
Narotsky et al.,
1995
Rat
Rat
LOAEL
HECgg
HECgg
HEDgg
HEDgg
BMDL
HEDgg
HEDgg
HECgg
HECgg
17
6.2
14
8.5
20
32.2
28
29
23
24
1
1
1
1
1
1
1
1
1
1
3
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
1
—
1
1
1
1
1
1
1
1
300
100
100
100
100
100
10
10
10
10
0.057
0.062
0.14
2.3
2.4
0.085
0.20
0.32
2.8
2.9
Resorptions
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to- route)
[AUCCBId] (route-to-route)
Resorptions; BMR = 1% extra
risk
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to- route)
[AUCCBId] (route-to-route)
Pre and postnatal growth
Healyetal., 1982
Rat
LOAEL
HECgg
HECgg
HEDgg
HEDgg
17
6.2
14
8.5
20
1
1
1
1
1
3
3
3
3
3
10
3
3
3
3
10
10
10
10
10
1
1
1
1
1
300
100
100
100
100
0.057
0.062
0.14
0.085
0.20
I fetal weight; skeletal effects
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to- route)
[AUCCBId] (route-to-route)
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-13. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical developmental effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae,
UFdb
UF b
•Jrcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Congenital defects
Johnson et al.,
2003
Rat
BMDL
HEDgg
HEDgg
HECgg
HECgg
0.0207
0.0052
0.0017
0.0037
0.00093
1
1
1
1
1
10
3
3
3
3
10
3
3
3
3
1
1
1
1
1
1
1
1
1
1
100
10
10
10
10
0.00037
0.000093
0.00021
0.00052
0.00017
Heart malformations (pups);
BMR = 1% extra risk; highest-
dose group (1,000-fold higher
than next highest) dropped to
improve model fit
[TotOxMetabBW34]
[AUCCBId]
[TotOxMetabBW34] (route-to-
route)
[AUCCBId] (route-to-route)
Developmental neurotoxicity
Fredricksson et al.,
1993
Taylor etal., 1985
Isaacson and
Taylor, 1989
Mouse
Rat
Rat
LOAEL
HEDgg
HEDgg
HECgg
HECgg
LOAEL
HEDgg
HEDgg
HECgg
HECgg
LOAEL
50
4.1
3.5
3.0
1.8
45
11
4.1
8.4
2.2
16
3
^_
3
3
3
1
1
1
1
1
1
10
3
3
3
3
10
3
3
3
3
10
10
3
3
3
3
10
3
3
3
3
10
10
10
10
10
10
10
10
10
10
10
10
1
1
1
1
1
1
1
1
1
1
1
3,000
300
300
300
300
1,000
100
100
100
100
1,000
0.010
0.0061
0.084
0.022
0.017
0.014
0.012
0.045
0.11
0.041
0.016
J, rearing postexposure; pup
gavage dose
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to- route)
[AUCCBId] (route-to-route)
| exploration postexposure;
estimated dam dose
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (route-to- route)
[AUCCBId] (route-to-route)
J, myelination in hippocampus;
estimated dam dose
TO
Co
Y1
I
o §
H I
O >
HH Oq
H TO
si
H
W
-------
to
Table 5-13. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs (based on PBPK modeled
internal dose metrics) for candidate critical developmental effects (continued)
Effect type
Candidate critical
studies
Species
POD
type
POD,
HEC99,
or
HED99a
UFSC
UFis
UFh
UF,oae|
UFdb
UP b
urcomp
cRfC or
p-cRfC
(ppm)
cRfD or
p-cRfD
(mg/kg/d)
Candidate critical effect;
comments [dose metric]
Developmental immunotoxicity
Peden-Adams et
al., 2006
Mouse
LOAEL
0.37
1
10
10
10
1
1,000
0.00037
J, PFC, |DTH; POD is estimated
dam dose (exposure throughout
gestation and lactation + to 3 or
8 wks of age)
Co
aApplied dose POD adjusted to continuous exposure unless otherwise noted. POD, HEC99, and HED99 have same units as cRfC (ppm) or cRfD (mg/kg/d).
bProduct of individual uncertainty factors, rounded to 3, 10, 30, 100, 300, 1,000, or 3,000.
UFSC = subchronic-to-chronic UF; UF1S = interspecies UF; UFh = human variability UF; UFioaei = LOAEL-to-NOAEL UF; UFdb = database UF.
Shaded rows represent the p-cRfC or p-cRfD using the preferred PBPK model dose metric or, in the cases where the PBPK model was not used, the cRfD or
cRfC based on applied dose.
H £
-------
1 5.1.3.3. Results and Discussion ofp-RfCs andp-RfDsfor Candidate Critical Effects
2 Tables 5-8-5-13 present the p-cRfCs and p-cRfDs developed using the PBPK internal
3 dose metrics, along with the cRfCs and cRfDs based on applied dose for comparison, for each
4 health effect domain.
5 The greatest impact of using the PBPK model was, as expected, for kidney effects, since
6 as discussed in Sections 3.3 and 3.5, toxicokinetic data indicate substantially more GSH
7 conjugation of TCE and subsequent bioactivation of GSH-conjugates in humans relative to rats
8 or mice. In addition, as discussed in Sections 3.3 and 3.5, the available in vivo data indicate high
9 interindividual variability in the amount of TCE conjugated with GSH. The overall impact is
10 that the p-cRfCs and p-cRfDs based on the preferred dose metric of bioactivated DCVC are
11 300- to 400-fold lower than the corresponding cRfCs and cRfDs based on applied dose. As
12 shown in Figure 3-14 in Section 3.5, for this dose metric there is about a 30- to 100-fold
13 difference (depending on exposure route and level) between rats and humans in the "central
14 estimates" of interspecies differences for the fraction of TCE that is bioactivated as DCVC. The
15 uncertainty in the human central estimate is only on the order of 2-fold (in either direction),
16 while that in the rat central estimate is substantially greater, about 10-fold (in either direction).
17 In addition, the interindividual variability about the human median estimate is on the order of
18 10-fold (in either direction). Because of the high confidence in the PBPK model's
19 characterization of the uncertainty and variability in internal dose metrics, as well as the high
20 confidence in GSH conjugation and subsequent bioactivation being the appropriate dose metric
21 for TCE kidney effects, there is also high confidence in the p-cRfCs and p-RfDs for these effects.
22 In addition, in two cases in which BMD modeling was employed, using internal dose
23 metrics led to a sufficiently different dose-response shape so as to change the resulting reference
24 value by greater than 5-fold. For the Woolhiser et al. (2006) decreased PFC response, this
25 occurred with the AUC of TCE in blood dose metric, leading to a p-cRfC 17-fold higher than
26 thecRfC based on applied dose. However, the model fit for this effect using this metric was
27 substantially worse than the fit using the preferred metric of Total oxidative metabolism.
28 Moreover, whereas an adequate fit was obtained with applied dose only with the highest-dose
29 group dropped, all the dose groups were included when the total oxidative metabolism dose
30 metric was used while still resulting in a good model fit. Therefore, it appears that using this
31 metric resolves some of the low-dose supralinearity in the dose-response curve. Nonetheless, the
32 overall impact of the preferred metric was minimal, as the p-cRfC based on the Total oxidative
33 metabolism metric was less than 1.4-fold larger than the cRfC based on applied dose. The
34 second case in which BMD modeling based on internal doses changed the candidate reference
35 value by more than 5-fold was for resorptions reported by Narotsky et al. (1995). Here, the
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 p-cRfDs were 7- to 8-fold larger than the corresponding cRfD based on applied dose. However,
2 for applied dose there is substantial uncertainty in the low-dose curvature of the dose-response
3 curve. This uncertainty persisted with the use of internal dose metrics, so the BMD remains
4 somewhat uncertain (see figures in Appendix F).
5 In the remaining cases, which generally involved the "generic" dose metrics of total
6 metabolism and AUC of TCE in blood, the p-cRfCs and p-cRfDs were within 5-fold of the
7 corresponding cRfC or cRfD based on applied dose, with the vast majority within 3-fold. This
8 suggests that the standard UFs for inter and intraspecies pharmacokinetic variability are fairly
9 accurate in capturing these differences for these TCE studies.
10
11 5.1.4. Uncertainties in cRfCs and cRfDs
12 5.1.4.1. Qualitative Uncertainties
13 An underlying assumption in deriving reference values for noncancer effects is that the
14 dose-response relationship for these effects has a threshold. Thus, a fundamental uncertainty is
15 the validity of that assumption. For some effects, in particular effects on very sensitive processes
16 (e.g., developmental processes) or effects for which there is a nontrivial background level and
17 even small exposures may contribute to background disease processes in more susceptible
18 people, a practical threshold (i.e., a threshold within the range of environmental exposure levels
19 of regulatory concern) may not exist.
20 Nonetheless, under the assumption of a threshold, the desired exposure level to have as a
21 reference value is the maximum level at which there is no appreciable risk for an adverse effect
22 in (nonnegligible) sensitive subgroups (of humans). However, because it is not possible to know
23 what this level is, "uncertainty factors" are used to attempt to address quantitatively various
24 aspects, depending on the data set, of qualitative uncertainty.
25 First there is uncertainty about the "point of departure" for the application of UFs.
26 Conceptually, the POD should represent the maximum exposure level at which there is no
27 appreciable risk for an adverse effect in the study population under study conditions (i.e., the
28 threshold in the dose-response relationship). Then, the application of the relevant UFs is
29 intended to convey that exposure level to the corresponding exposure level for sensitive human
30 subgroups exposed continuously for a lifetime. In fact, it is again not possible to know that
31 exposure level even for a laboratory study because of experimental limitations (e.g., the power to
32 detect an effect, dose spacing, measurement errors, etc.), and crude approximations like the
33 NOAEL or a BMDL are used. If a LOAEL is used as the POD, the LOAEL-to-NOAEL UF is
34 applied as an adjustment factor to get a better approximation of the desired exposure level
35 (threshold), but the necessary extent of adjustment is unknown.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-63 DRAFT—DO NOT CITE OR QUOTE
-------
1 If a BMDL is used as the POD, there are uncertainties regarding the appropriate dose-
2 response model to apply to the data, but these should be minimal if the modeling is in the
3 observable range of the data. There are also uncertainties about what BMR to use to best
4 approximate the desired exposure level (threshold, see above). For continuous endpoints, in
5 particular, it is often difficult to identify the level of change that constitutes the "cut-point" for an
6 adverse effect. Sometimes, to better approximate the desired exposure level, a BMR somewhat
7 below the observable range of the data is selected. In such cases, the model uncertainty is
8 increased, but this is a trade-off to reduce the uncertainty about the POD not being a good
9 approximation for the desired exposure level.
10 For each of these types of PODs, there are additional uncertainties pertaining to
11 adjustments to the administered exposures (doses). Typically, administered exposures (doses)
12 are converted to equivalent continuous exposures (daily doses) over the study exposure period
13 under the assumption that the effects are related to concentration x time, independent of the daily
14 (or weekly) exposure regimen (i.e., a daily exposure of 6 hours to 4 ppm is considered equivalent
15 to 24 hours of exposure to 1 ppm). However, the validity of this assumption is generally
16 unknown, and, if there are dose-rate effects, the assumption of C x t equivalence would tend to
17 bias the POD downwards. Where there is evidence that administered exposure better correlates
18 to the effect than equivalent continuous exposure averaged over the study exposure period (e.g.,
19 visual effects), administered exposure was not adjusted. For the PBPK analyses in this
20 assessment, the actual administered exposures are taken into account in the PBPK modeling, and
21 equivalent daily values (averaged over the study exposure period) for the dose metrics are
22 obtained (see above, Section 5.1.3.2). Additional uncertainties about the PBPK-based estimates
23 include uncertainties about the appropriate dose metric for each effect, although for some effects
24 there was better information about relevant dose metrics than for others (see Section 5.1.3.1).
25 Second, there is uncertainty about the UFs. The human variability UF is to some extent
26 an adjustment factor because for more sensitive people, the dose-response relationship shifts to
27 lower exposures. However, there is uncertainty about the extent of the adjustment required, i.e.,
28 about the distribution of human susceptibility. Therefore, in the absence of data on a more
29 sensitive population(s) or on the distribution of susceptibility in the general population, an UF of
30 10 is generally used, in part for pharmacokinetic variability and in part for pharmacodynamic
31 variability. The PBPK analyses in this assessment attempt to account for the pharmacokinetic
32 portion of human variability using human data on pharmacokinetic variability. A quantitative
33 uncertainty analysis of the PBPK-derived dose metrics used in the assessment is presented in
34 Section 5.1.4.2 below. There is still uncertainty regarding the susceptible subgroups for TCE
35 exposure and the extent of pharmacodynamic variability.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-64 DRAFT—DO NOT CITE OR QUOTE
-------
1 If the data used to determine a particular POD are from laboratory animals, an
2 interspecies extrapolation UF is used. This UF is also to some extent an adjustment factor for the
3 expected scaling for toxicologically-equivalent doses across species (i.e., according to body
4 weight to the 3/4 power for oral exposure). However, there is also uncertainty about the true
5 extent of interspecies differences for specific noncancer effects from specific chemical
6 exposures. Often, the "adjustment" component of this UF has been attributed to
7 pharmacokinetics, while the "uncertainty" component has been attributed to pharmacodynamics,
8 but as discussed above in Section 5.1.3.1, this is not the only interpretation supported. For oral
9 exposures, the standard value for the interspecies UF is 10, which can be viewed as breaking
10 down (approximately) to a factor of three for the "adjustment" (nominally pharmacokinetics) and
11 a factor of three for the "uncertainty" (nominally pharmacodynamics). For inhalation exposures,
12 no adjustment across species is generally assumed for fixed air concentrations (ppm
13 equivalence), and the standard value for the interspecies UF is 3 reflects "uncertainty"
14 (nominally pharmacodynamics only). The PBPK analyses in this assessment attempt to account
15 for the "adjustment" portion of interspecies extrapolation using rodent pharmacokinetic data to
16 estimate internal doses for various dose metrics. With respect to the "uncertainty" component,
17 quantitative uncertainty analyses of the PBPK-derived dose metrics used in the assessment are
18 presented in Section 5.1.4.2 below. However, these only address the pharmacokinetic
19 uncertainties in a particular dose metric, and there is still uncertainty regarding the true dose
20 metrics. Nor do the PBPK analyses address the uncertainty in either cross-species
21 pharmacodynamic differences (i.e., about the assumption that equal doses of the appropriate dose
22 metric convey equivalent risk across species for a particular endpoint from a specific chemical
23 exposure) or in cross-species pharmacokinetic differences not accounted for by the PBPK model
24 dose metrics (e.g., departures from the assumed interspecies scaling of clearance of the active
25 moiety, in the cases where only its production is estimated). A value of 3 is typically used for
26 the "uncertainty" about cross-species differences, and this generally represents true uncertainty
27 because it is usually unknown, even after adjustments have been made to account for the
28 expected interspecies differences, whether humans have more or less susceptibility, and to what
29 degree, than the laboratory species in question.
30 If only subchronic data are available, the subchronic-to-chronic UF is to some extent an
31 adjustment factor because, if the effect becomes more severe with increasing exposure, then
32 chronic exposure would shift the dose-response relationship to lower exposures. However, the
33 true extent of the shift is unknown.
34 Sometimes a database UF is also applied to address limitations or uncertainties in the
35 database. The overall database for TCE is quite extensive, with studies for many different types
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-65 DRAFT—DO NOT CITE OR QUOTE
-------
1 of effects, including 2-generation reproductive studies, as well as neurological, immunological,
2 and developmental immunological studies. In addition, there were sufficient data to develop a
3 reliable PBPK model to estimate route-to-route extrapolated doses for some candidate critical
4 effects for which data were only available for one route of exposure. Thus, there is a high degree
5 of confidence that the TCE database was sufficient to identify some sensitive endpoints.
6
7 5.1.4.2. Quantitative Uncertainty Analysis of Physiologically Based Pharmacokinetic (PBPK)
8 Model-Based Dose Metrics for Lowest-Observed-Adverse-Effect Level (LOAEL) or
9 No-Observed-Adverse-Effect Level (NOAEL)-Based Point of Departures (PODs)
10 The Bayesian analysis of the PBPK model for TCE generates distributions of uncertainty
11 and variability in the internal dose metrics that can be readily used for characterizing the
12 uncertainty and variability in the PBPK model-based derivations of the HEC and HED. As
13 shown in Figure 5-4, the overall approach taken for the uncertainty analysis is similar to that
14 used for the point estimates except for the carrying through of distributions rather than median or
15 expected values at various points. Because of a lack of tested software and limitations of time
16 and resources, this analysis was not performed for idPODs based on BMD modeling, and was
17 only performed for idPODs derived from a LOAEL or NOAEL. However, for those endpoints
18 for which BMD modeling was performed, for the purposes of this uncertainty analysis, an
19 alternative idPOD was used based on the study LOAEL or NOAEL.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-66 DRAFT—DO NOT CITE OR QUOTE
-------
•distribution
[distribution (combined
ncertainty and variability)
0.1-2000 ppm
in air or
0.1-2000
mg/kg-d
continuous
exposure
/^~~^
I Human
I model
V parameters
^^
fixed*-
distribution
fixed*-
Idistribution
Dose-Response Model
or
LOAEUNOAEL
idPOD=
LOAEL,
or NOAEL
(internal
dose unit
Jdistribution (separate
ncertainty and variability)
invert functions of dose
or concentration
Uncertainty distribution
of population median
Human
dose or
concentratio
at idPOD
listribution of functions
dose or concentration
Uncertainty distribution
of population 95th
percentile
Typical
human
equivalent
Sensitive
Human
equivalent
distribution
distribution
2 Figure 5-4. Flow-chart for uncertainty analysis of HECs and HEDs derived
3 using PBPK model-based dose metrics. Square nodes indicate point values,
4 circle nodes indicate distributions, and the inverted triangle indicates a
5 (deterministic) functional relationship.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-67 DRAFT—DO NOT CITE OR QUOTE
-------
1 In brief, the distribution of rodent PBPK model parameters is carried through to a
2 distribution of idPODs, reflecting combined uncertainty and variability in the rodent internal
3 dosimetry. Separately, for each set of human population parameters, a set of individual PBPK
4 model parameters is generated, and the human PBPK model is run for a range of continuous
5 exposures from 10"1 to 2x 103 ppm or mg/kg/d to obtain the distribution of the relationship
6 between human exposure and internal dose. For a given set of (1) an idPOD sampled from the
7 rodent distribution, (2) a human population sampled from the distribution of populations, and
8 (3) an individual sampled from this population, a human equivalent exposure (HEC or HED)
9 corresponding to the idPOD is derived by interpolation. Within each population, a HEC or HED
10 corresponding to the median and 99th percentile individuals are derived, resulting in two
11 distributions (both reflecting uncertainty): one of "typical" individuals represented by the
12 distribution of population medians, and one of "sensitive" individuals represented by the
13 distribution of an upper percentile of the population (e.g., 99th percentile). Note that because a
14 distribution of rodent-derived idPODs was used, the uncertainty distribution includes the
15 contribution from the uncertainty in the rodent internal dose. Thus, for selected quantiles of the
16 population and level of confidence (e.g., Xth percentile individual at Yth% confidence), the
17 interpretation is that at the resulting HEC or HED, there is Y% confidence that X% of the
18 population has an internal dose less than that of the rodent in the toxicity study.
19 As shown in Tables 5-14-5-18, the HECgg and HEDgg derived using the rodent median
20 dose metrics and the combined uncertainty and variability in human dose metrics is generally
21 near (within 1.3-fold of) the median confidence level estimate of the HEC and HED for the
22 99th percentile individual. Therefore, the interpretation is that there is about 50% confidence that
23 human exposure at the HECgg or HEDgg will, in 99% of the human population, lead to an internal
24 dose less than or equal to that in the subjects (rodent or human) exposed at the POD in the
25 corresponding study.
26 In several cases, the uncertainty, as reflected in the ratio between the 95% and 50%
27 confidence bounds on the 99th percentile individual, was rather high (e.g., >5-fold), and reflected
28 primarily uncertainty in the rodent internal dose estimates, discussed previously in Section 3.5.7.
29 The largest uncertainties (ratios between 95% to 50% confidence bounds of 8- to 10-fold) were
30 for kidney effects in mice using the AMetGSHBW34 dose metric (Kjellstrand et al., 1983; NCI,
31 1976). More moderate uncertainties (ratios between 95% to 50% confidence bounds of 5- to
32 8-fold) were evident in some oral studies using the AUCCBld dose metric (Sanders et al., 1982;
33 George et al., 1986; Fredricksson et al., 1993; Keil et al., 2009), as well as in studies reporting
34 kidney effects in rats in which the ABioactDCVCBW34 or AMetGSHBW34 dose metrics were
35 used (Woolhiser et al., 2006; NTP, 1988). Therefore, in these cases, a POD that is protective of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-68 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
the 99th percentile individual at a confidence level higher than 50% could be as much as an order
of magnitude lower.
Table 5-14. Comparison of "sensitive individual" HECs or HEDs for
neurological effects based on PBPK modeled internal dose metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL
Candidate critical effect
Candidate critical study
(species)
POD
type
Ratio
HEC/D50:
HEC/D99
HECX or HEDX
X = 99
X = 99,
median
X = 99,
95lcb
[Dose metric]
Neurological
Trigeminal nerve effects
Ruitjen etal., 1991 (human)
Demyelination in hippocampus
Isaacson et al., 1990 (rat)
Changes in wakefulness
Aritoetal., 1994 (rat)
J, regeneration of sciatic nerve
Kjellstrand etal., 1987 (rat)
J, regeneration of sciatic nerve
Kjellstrand etal., 1987
(mouse)
Degeneration of dopaminergic
neurons
Gash etal., 2007 (rat)
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HED
HED
HEC
2.62
1.68
1.02
4.31
1.02
7.20
2.59
1.68
2.65
1.67
1.02
4.25
2.94
1.90
1.13
3.08
3.16
1.84
1.21
2.13
1.06
2.98
2.70
5.4
8.3
7.3
14
9.21
4.29
7.09
2.29
4.79
9
6.46
15.2
93.1
257
97.1
142
120
108
120
75.8
53
192
46.8
5.4
8.3
7.2
16
9.20
5.28
6.77
2.42
4.86
9.10
6.50
18.0
93.6
266
96.8
147
125
111
121
79.1
53.8
199
47.9
2.6
4.9
3.8
8.0
7.39
2.52
4.94
0.606
2.37
4.63
3.39
8.33
38.6
114
43.4
78.0
48.8
59.7
57.0
53.4
17.1
94.7
14.2
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
9
10
11
12
13
14
15
HEC99 = the 99th percentile of the combined human uncertainty and variability distribution of continuous exposure
concentrations that lead to the (fixed) median estimate of the rodent internal dose at the POD.
HEC99median (or HEC99 95icb) = the median (or 95th percentile lower confidence bound) estimate of the uncertainty
distribution of continuous exposure concentrations for which the 99th percentile individual has an internal dose
less than the (uncertain) rodent internal dose at the POD.
rtr = route-to-route extrapolation using PBPK model and the specified dose metric.
Shaded rows denote results for the primary dose metric.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-69 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table 5-15. Comparison of "sensitive individual" HECs or HEDs for kidney
and liver effects based on PBPK modeled internal dose metrics at different
levels of confidence and sensitivity, at the NOAEL or LOAEL
Candidate critical effect
Candidate critical study
(species)
POD
type
Ratio
HEC/D50:
HEC/D99
HECX or HEDX
X = 99
X = 99,
median
X = 99,
95lcb
[Dose metric]
Kidney
Meganucleocytosis
[NOAEL]*
Maltoni, 1986 (rat)
Toxic nephrosis
NCI, 1976 (mouse)
Toxic nephropathy [LOAEL]*
NTP, 1988 (rat)
| kidney/body weight ratio
[NOAEL]*
Kjellstrandetal., 1983b
(mouse)
| kidney/body weight ratio
[NOAEL]*
Woolhiser et al., 2006 (rat)
HEC
HEC
HEC
HED
HED
HED
HED
HED
HEC
HEC
HED
HED
HED
HEC
HEC
HEC
HEC
HEC
HED
HED
HEC
HEC
HEC
HED
HED
HED
7.53
7.70
2.57
9.86
9.83
1.02
9.51
1.05
7.78
2.67
9.75
9.64
1.03
7.55
7.75
2.59
7.69
2.63
9.78
1.03
7.53
7.70
2.54
9.84
9.81
1.02
0.0233
0.0364
8.31
0.0140
0.0223
10.6
0.30
48
0.50
42
0.121
0.193
33.1
0.201
0.314
28.2
0.111
34.5
0.068
39.9
0.0438
0.0724
16.1
0.0264
0.0444
19.5
0.0260
0.0411
7.97
0.0156
0.0242
10.7
0.32
48.9
0.514
43.5
0.126
0.210
33.1
0.204
0.353
27.2
0.103
33.7
0.00641
39.2
0.0481
0.0827
15.2
0.0282
0.0488
19.2
0.00366
0.00992
4.03
0.00216
0.00597
5.75
0.044
16.2
0.0703
13.7
0.0177
0.0379
11.1
0.0269
0.0676
8.77
0.00809
13.5
0.00497
17.9
0.00737
0.0179
7.56
0.00447
0.0111
10.5
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34]
(rtr)
[AMetGSHBW34] (rtr)
[TotMetabBW34] (rtr)
[AMetGSHBW34]
[TotMetabBW34]
[AMetGSHBW34] (rtr)
[TotMetabBW34] (rtr)
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34]
(rtr)
[AMetGSHBW34] (rtr)
[TotMetabBW34] (rtr)
[AMetGSHBW34]
[TotMetabBW34]
[AMetGSHBW34] (rtr)
[TotMetabBW34] (rtr)
[ABioactDCVCBW34]
[AMetGSHBW34]
[TotMetabBW34]
[ABioactDCVCBW34]
(rtr)
[AMetGSHBW34] (rtr)
[TotMetabBW34] (rtr)
Liver
| liver/body weight ratio
[LOAEL]*
Kjellstrandetal., 1983b
(mouse)
HEC
HEC
HED
HED
2.85
3.63
1.16
1.53
16.2
40.9
14.1
40.1
16.3
38.1
14.1
39.4
6.92
15.0
5.85
17.9
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (rtr)
[TotOxMetabBW34]
(rtr)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-70 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-15. Comparison of "sensitive individual" HECs or HEDs for kidney
and liver effects based on PBPK modeled internal dose metrics at different
levels of confidence and sensitivity, at the NOAEL or LOAEL (continued)
Candidate critical effect
Candidate critical study
(species)
| liver/body weight ratio
[NOAEL]*
Woolhiseretal., 2006 (rat)
| liver/body weight ratio
[LOAEL]*
Buben and O'Flaherty,
1985 (mouse)
POD
type
HEC
HEC
HED
HED
HED
HED
HEC
HEC
Ratio
HEC/D50:
HEC/D99
2.86
2.94
1.20
1.21
1.14
1.14
2.80
3.13
HECX or HEDX
X = 99
20.7
18.2
17.8
19.6
8.82
9.64
10.1
7.83
X = 99,
median
21.0
17.1
17.7
19.3
8.95
9.78
9.97
7.65
X = 99,
95lcb
11.0
8.20
9.94
10.5
4.17
5.28
4.83
4.23
[Dose metric]
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (rtr)
[TotOxMetabBW34]
(rtr)
[AMetLivl BW34]
[TotOxMetabBW34]
[AMetLivl BW34] (rtr)
[TotOxMetabBW34]
(rtr)
2
O
4
5
6
7
8
9
10
11
*BMDL used for p-cRfC or p-cRfD, but LOAEL or NOAEL (as noted) used for uncertainty analysis.
HEC99 = the 99th percentile of the combined human uncertainty and variability distribution of continuous exposure
concentrations that lead to the (fixed) median estimate of the rodent internal dose at the POD.
HEC99median (or HEC99 95icb) = the median (or 95th percentile lower confidence bound) estimate of the uncertainty
distribution of continuous exposure concentrations for which the 99th percentile individual has an internal dose
less than the (uncertain) rodent internal dose at the POD.
rtr = route-to-route extrapolation using PBPK model and the specified dose metric.
Shaded rows denote results for the primary dose metric.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-71 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table 5-16. Comparison of "sensitive individual" HECs or HEDs for
immunological effects based on PBPK modeled internal dose metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL
Candidate critical effect
Candidate critical study
(species)
POD
type
Ratio
HEC/D50:
HEC/D99
HECX or HEDX
X = 99
X = 99,
median
X = 99,
95lcb
[Dose metric]
Immunological
Changes in immunoreactive
organs — liver (including
sporatic necrosis in hepatic
lobules), spleen
Kanekoetal., 2000
(mouse)
t anti-dsDNA & anti-ssDNA
Abs (early markers for SLE);
J, thymus weight
Keil etal., 2009 (mouse)
| RFC response [NOAEL]*
Woolhiser etal., 2006 (rat)
J, stem cell bone marrow
recolonization; J, cell-
mediated response to sRBC
Sanders etal., 1982
(mouse)
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HEC
HEC
HED
HED
HED
HED
HEC
HEC
2.65
1.75
1.04
3.21
1.02
12.1
2.77
1.69
2.54
1.73
1.02
3.21
1.02
10.5
2.77
1.68
36.7
68.9
42.3
56.5
0.0482
0.0161
0.0332
0.00821
16.1
59.6
19.5
52
2.48
0.838
1.72
0.43
38.3
70.0
43.3
59.0
0.0483
0.0189
0.0337
0.00787
15.2
60.1
19.2
55.9
2.48
0.967
1.75
0.412
16.0
37.1
21.3
39.8
0.0380
0.00363
0.0246
0.00199
7.56
26.2
10.5
33.0
1.94
0.187
1.28
0.103
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
5
6
7
8
9
10
11
12
13
14
*BMDL used for p-cRfC or p-cRfD, but LOAEL or NOAEL (as noted) used for uncertainty analysis.
HEC99 = the 99th percentile of the combined human uncertainty and variability distribution of continuous exposure
concentrations that lead to the (fixed) median estimate of the rodent internal dose at the POD.
HEC99median (or HEC99 95icb) = the median (or 95th percentile lower confidence bound) estimate of the uncertainty
distribution of continuous exposure concentrations for which the 99th percentile individual has an internal dose
less than the (uncertain) rodent internal dose at the POD.
rtr = route-to-route extrapolation using PBPK model and the specified dose metric.
Shaded rows denote results for the primary dose metric.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-72 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table 5-17. Comparison of "sensitive individual" HECs or HEDs for
reproductive effects based on PBPK modeled internal dose metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL
Candidate critical effect
Candidate critical study
(species)
POD
type
Ratio
HEC/D50:
HEC/D99
HECX or HEDX
X = 99
X = 99,
median
X = 99,
95lcb
[Dose metric]
Reproductive
Hyperzoospermia
Chia etal., 1996 (human)
J, fertilization
Xu et al., 2004 (mouse)
Multiple sperm effects,
testicular enzyme markers
Kumar etal., 2000a,
2001 b (rat)
J, ability of sperm to fertilize
in vitro
DuTeaux et al., 2004 (rat)
Effects on epididymis
epithelium
Forkert etal. ,2002; Kan
etal., 2007 (mouse)
Testes effects
Kumar etal., 2000a,
2001 b (rat)
Delayed parturition
Narotskyetal., 1995 (rat)
J, mating (both sexes
exposed)
George etal., 1986 (rat)
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HED
HED
HEC
HEC
2.78
1.68
1.02
9.69
2.85
1.89
1.09
3.11
2.53
1.72
1.02
3.21
4.20
1.57
1.67
3.75
2.85
1.89
1.09
3.11
2.53
1.72
1.02
3.21
1.06
3.07
2.66
1.91
1.10
3.21
2.86
1.73
0.50
0.83
0.73
1.6
66.6
170
73.3
104
12.8
53.2
15.8
48.8
15.6
41.7
9.3
42.5
66.6
170
73.3
104
12.8
53.2
15.8
48.8
44.3
114
36.9
190
77.4
51.9
71.1
59.5
0.53
0.83
0.71
2.0
72.3
171
76.9
109
12.2
54.4
15.7
52.6
18.1
41.9
10.1
55.6
72.3
171
76.9
109
12.2
54.4
15.7
52.6
43.9
119
35.3
197
77.1
55.8
70.0
63.3
0.25
0.49
0.37
0.92
26.6
97.1
32.9
67.9
6.20
23.2
8.60
30.6
4.07
32.0
2.09
39.1
26.6
97.1
32.9
67.9
6.20
23.2
8.60
30.6
15.1
47.7
11.6
48.1
34.2
14.7
29.5
8.14
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[AUCCBId]
[TotOxMetabBW34]
[AUCCBId] (rtr)
[TotOxMetabBW34] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
5
6
7
HEC99 = the 99th percentile of the combined human uncertainty and variability distribution of continuous exposure
concentrations that lead to the (fixed) median estimate of the rodent internal dose at the POD.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-73 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-17. Comparison of "sensitive individual" HECs or HEDs for
2 reproductive effects based on PBPK modeled internal dose metrics at
3 different levels of confidence and sensitivity, at the NOAEL or LOAEL
4 (continued)
5
6 HEC99)median (or HEC99)95icb) = the median (or 95th percentile lower confidence bound) estimate of the uncertainty
7 distribution of continuous exposure concentrations for which the 99th percentile individual has an internal dose
8 less than the (uncertain) rodent internal dose at the POD.
9 rtr = route-to-route extrapolation using PBPK model and the specified dose metric.
10 Shaded rows denote results for the primary dose metric.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-74 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table 5-18. Comparison of "sensitive individual" HECs or HEDs for
developmental effects based on PBPK modeled internal dose metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL
Candidate critical effect
Candidate critical study
(species)
POD
type
Ratio
HEC/D50:
HEC/D99
HECX or HEDX
X = 99
X = 95,
median
X = 95,
95lcb
[Dose metric]
Developmental
Resorptions
Healyetal., 1982 (rat)
Resorptions [LOAEL]*
Narotsky etal., 1995
(rat)
J, fetal weight; skeletal
effects
Healyetal., 1982 (rat)
Heart malformations
(pups) [LOAEL]*
Johnson etal., 2003
(rat)
J, rearing postexposure
Fredricksson et al., 1993
(mouse)
| exploration
postexposure
Taylor etal., 1985 (rat)
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HEC
HEC
HED
HED
HED
HED
HEC
HEC
HED
HED
HEC
HEC
HED
HED
HEC
HEC
2.58
1.69
1.02
3.68
1.06
3.07
2.66
1.91
2.58
1.69
1.02
3.68
1.02
11.6
2.75
1.70
1.02
7.69
2.71
1.68
1.02
7.29
2.57
1.68
6.19
13.7
8.5
19.7
44.3
114
36.9
190
6.19
13.7
8.5
19.7
0.012
0.00382
0.00848
0.00216
4.13
3.46
2.96
1.84
10.7
4.11
8.36
2.19
6.02
13.9
8.50
22.4
43.9
119
35.3
197
6.02
13.9
8.50
22.4
0.012
0.00476
0.00866
0.00221
4.19
4.21
2.96
1.81
10.7
5.08
7.94
2.31
3.13
7.27
4.61
11.5
15.1
47.7
11.6
48.1
3.13
7.27
4.61
11.5
0.0102
0.00112
0.00632
0.000578
2.22
0.592
1.48
0.302
8.86
1.16
5.95
0.580
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotOxMetabBW34]
[AUCCBId]
[TotOxMetabBW34]
(rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
[TotMetabBW34]
[AUCCBId]
[TotMetabBW34] (rtr)
[AUCCBId] (rtr)
5
6
7
8
9
10
11
12
13
14
*BMDL used for p-cRfC or p-cRfD, but LOAEL or NOAEL (as noted) used for uncertainty analysis.
HEC99 = the 99th percentile of the combined human uncertainty and variability distribution of continuous exposure
concentrations that lead to the (fixed) median estimate of the rodent internal dose at the POD.
HEC99)medmn (or HEC99)95icb) = the median (or 95th percentile lower confidence bound) estimate of the uncertainty
distribution of continuous exposure concentrations for which the 99th percentile individual has an internal dose
less than the (uncertain) rodent internal dose at the POD.
rtr = route-to-route extrapolation using PBPK model and the specified dose metric.
Shaded rows denote results for the primary dose metric.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-75 DRAFT—DO NOT CITE OR QUOTE
-------
1 For comparison, Tables 5-14 and 5-18 also show the ratios of the overall 50th percentile
2 to the overall 99th percentile HECs and HEDs, reflecting combined human uncertainty and
3 variability at the median study/endpoint idPOD. The smallest ratios (up to 1.2-fold) are for total,
4 oxidative, and hepatic oxidative metabolism dose metrics from oral exposures, due to the large
5 hepatic first-pass effect resulting in virtually all of the oral intake being metabolized before
6 systemic circulation. Conversely, the large hepatic first-pass results in high variability in the
7 blood concentration of TCE following oral exposures, with ratios up to 12-fold at low exposures
8 (e.g., 90 vs. 99% first-pass would result in amounts metabolized differing by about 10% but TCE
9 blood concentrations differing by about 10-fold). From inhalation exposures, there is moderate
10 variability in these metrics, about 2- to 3-fold. For GSH conjugation and bioactivated DCVC,
11 however, variability is high (8- to 10-fold) for both exposure routes, which follows from the
12 incorporation in the PBPK model analysis of the data from Lash et al. (1999b) showing
13 substantial interindividual variability in GSH conjugation in humans.
14 Finally, it is important to emphasize that this analysis only addresses pharmacokinetic
15 uncertainty and variability, so other aspects of extrapolation addressed in the UFs (e.g., LOAEL
16 to NOAEL, subchronic to chronic, and pharmacodynamic differences), discussed above, are not
17 included in the level of confidence.
18
19 5.1.5. Summary of Noncancer Reference Values
20 5.1.5.1. Preferred Candidate Reference Values (cRfCs, cRfD,p-cRfCs andp-cRfDs) for
21 Candidate Critical Effects
22 The candidate critical effects that yielded the lowest p-cRfC or p-cRfD for each type of
23 effect, based on the primary dose metric, are summarized in Tables 5-19 (p-cRfCs) and 5-20
24 (p-cRfDs). These results are extracted from Tables 5-8-5-13. In cases where a route-to-route
25 extrapolated p-cRfC (p-cRfD) is lower than the lowest p-cRfC (p-cRfD) from an inhalation
26 (oral) study, both values are presented in the table. In addition, if there is greater than usual
27 uncertainty associated with the lowest p-cRfC or p-cRfD for a type of effect, then the endpoint
28 with the next lowest value is also presented. Furthermore, given those selections, the same sets
29 of critical effects and studies are displayed across both tables, with the exception of two oral
30 studies for which route-to-route extrapolation was not performed. Tables 5-19 and 5-20 are
31 further summarized in Tables 5-21 and 5-22 to present the overall preferred p-cRfC and p-cRfD
32 for each type of noncancer effect. The purpose of these summary tables is to show the most
33 sensitive endpoints for each type of effect and the apparent relative sensitivities (based on
34 reference value estimates) of the different types of effects.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-76 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-19. Lowest p-cRfCs or cRfCs for different effect domains
Effect domain
Effect type
Candidate critical effect
(Species/Critical Study)
p-cRfC or cRfC in ppm
(composite uncertainty factor)
Preferred
dose
metric3
Default
methodology
Alternative dose
metrics/studies
(Tables 5-8-5-1 3)
Neurologic
Trigeminal nerve
effects
Cognitive effects
Mood/sleep
changes
Trigeminal nerve effects
(human/Ruitjen et al., 1991)
Demyelination in hippocampus
(rat/Isaacson et al., 1990)
Changes in wakefulness
(rat/Aritoetal., 1994)
0.54
(10)
0.0071
(1 ,000)
0.016
(300)
0.47
(30)
[rtr]
0.012
(1,000)
0.83
(10)
0.0023
(1 ,000)
0.030
(300)
Kidney
Histological
changes
| kidney weight
Toxic nephropathy
(rat/NTP, 1988)
Toxic nephrosis
(mouse/NCI, 1976)
| kidney weight
(rat/Woolhiser et al., 2006)
0.00056
(10)
0.0017
(300)
0.0013
(10)
[rtr]
[rtr]
0.52
(30)
0.00087-1.3
(10-300)
0.0022-2.1
(1 0-30)
Liver
| liver weight
| liver weight
(mouse/Kjellstrand etal., 1983b)
0.91
(10)
0.72
(30)
0.83-2.5
(1 0-30)
Immunologic
J, thymus weight
Immuno-
suppression
Autoimmunity
J, thymus weight
(mouse/Keil et al., 2009)
J, stem cell recolonization
(mouse/Sanders et al., 1982)
Decreased RFC response
(rat/Woolhiser et al., 2006)
t anti-dsDNA & anti-ssDNA Abs
(mouse/Keil et al., 2009)
Autoimmune organ changes
(mouse/Kaneko et al., 2000)
0.00033
(100)
0.057
(30)
0.11
(100)
0.0033
(10)
0.12
(300)
[rtr]
[rtr]
0.083
(300)
[rtr]
0.070
(1,000)
0.000082
(100)
0.014-1.4
(30-100)
0.00082-0.23
(10-300)
Reproductive
Effects on sperm
and testes
J, ability of sperm to fertilize
(rat/DuTeaux et al., 2004)
Multiple effects
(rat/Kumar et al., 2000a, 2001 b)
Hyperzoospermia
(human/Chiaetal., 1996)b
0.0093
(1 ,000)
0.013
(1 ,000)
0.017
(30)
[rtr]
0.015
(3,000)
0.014
(100)
0.028-0.17
(30-1,000)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-77 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-19. Lowest p-cRfCs or cRfCs for different effect domains
(continued)
Effect domain
Effect type
Candidate critical effect
(Species/Critical Study)
p-cRfC or cRfC in ppm
(composite uncertainty factor)
Preferred
dose
metric3
Default
methodology
Alternative dose
metrics/studies
(Tables 5-8-5-1 3)
Developmental
Congenital
defects
Develop.
neurotox.
P re/postnatal
mortality/growth
Heart malformations
(rat/Johnson et al., 2003)
I rearing postexposure
(rat/Fredricksson et al., 1993)
Resorptions/j, fetal weight/
skeletal effects
(rat/Healyetal., 1982)
0.00037
(10)
0.028
(300)
0.062
(100)
[rtr]
[rtr]
0.057
(300)
0.000093
(10)
0.0077-0.084
(100-300)
0.14-2.4
(10-100)
4
5
aThe critical effects/studies and p-cRfCs supporting the RfC are in bold.
bgreater than usual degree of uncertainty (see Section 5.1.2).
rtr = route-to-route extrapolated result.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-78 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-20. Lowest p-cRfDs or cRfDs for different effect domains
Effect domain
Effect type
Candidate critical effect
(Species/Critical Study)
p-cRfD or cRfD in mg/kg/d
(composite uncertainty factor)
Preferred
dose
metric3
Default
methodology
Alternative dose
metrics/studies
(Tables 5-8-5-1 3)
Neurologic
Trigeminal nerve
effects
Cognitive effects
Mood/sleep
changes
Trigeminal nerve effects
(human/Ruitjen et al., 1991)
Demyelination in hippocampus
(rat/Isaacson et al., 1990)
Changes in wakefulness
(rat/Aritoetal., 1994)
0.73
(10)
0.0092
(1 ,000)
0.022
(300)
[rtr]
0.0047
(10,000b)
[rtr]
1.4
(10)
0.0043
(1 ,000)
0.051
(300)
Kidney
Histological
changes
| kidney weight
Toxic nephropathy
(rat/NTP, 1988)
Toxic nephrosis
(mouse/NCI, 1976)
| kidney weight
(rat/Woolhiser et al., 2006)
0.00034
(10)
0.0010
(300)
0.00079
(10)
0.0945
(100)
[rtr]
0.00053-1.9
(10-300)
0.0013-2.5
(10)
Liver
| liver weight
| liver weight
(mouse/Kjellstrand etal., 1983b)
0.79
(10)
[rtr]
0.82-2.6
(10-100)
Immunologic
J, thymus weight
Immuno-
suppression
Autoimmunity
J, thymus weight
(mouse/Keil et al., 2009)
J, stem cell recolonization
(mouse/Sanders et al., 1982)
Decreased RFC response
(rat/Woolhiser et al., 2006)
t anti-dsDNA & anti-ssDNA Abs
(mouse/Keil et al., 2009)
Autoimmune organ changes
(mouse/Kaneko et al., 2000)
0.00048
(100)
0.083
(30)
0.14
(100)
0.0048
(10)
0.14
(300)
0.00035
(1,000)
0.060
(300)
[rtr]
0.0035
(100)
[rtr]
0.00016
(100)
0.028-0.91
(30-100)
0.0016-0.19
(10-300)
Reproductive
Effects on sperm
and testes
J, ability of sperm to fertilize
(rat/DuTeaux et al., 2004)
Multiple effects
(rat/Kumar et al., 2000a, 2001 b)
Hyperzoospermia
(human/Chiaetal., 1996)c
0.016
(1 ,000)
0.016
(1 ,000)
0.024
(30)
0.014
(10,000b)
[rtr]
[rtr]
0.042-0.10
(30-1,000)
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-79 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-20. Lowest p-cRfDs or cRfDs for different effect domains
(continued)
Effect domain
Effect type
Candidate critical effect
(Species/Critical Study)
p-cRfD or cRfD in mg/kg/d
(composite uncertainty factor)
Preferred
dose
metric3
Default
methodology
Alternative dose
metrics/studies
(Tables 5-8-5-1 3)
Developmental
Develop.
immunotox.
Congenital
defects
Develop.
neurotox.
P re/postnatal
mortality/growth
| RFC, t DTH
(rat/Peden-Adams et al., 2006)d
Heart malformations
(rat/Johnson et al., 2003)
I rearing postexposure
(rat/Fredricksson et al., 1993)d
Resorptions/j, fetal weight/
skeletal effects
(rat/Healyetal., 1982)
0.00037
(1 ,000)
0.00052
(10)
0.016
(1 ,000)
0.085
(100)
Same as
preferred
0.00021
(100)
Same as
preferred
[rtr]
—
0.00017
(10)
0.017-0.11
(100-3,000)
0.70-2.9
(10-100)
2
O
4
5
6
1
8
9
10
11
"The critical effects/studies and p-cRfDs or cRfDs supporting the RfD are in bold.
bU.S. EPA's report on the RfC and RfD processes (U.S. EPA, 2002) recommends not deriving reference values with
a composite UF of greater than 3,000; however, composite UFs exceeding 3,000 are considered here because the
derivation of the cRfCs and cRfDs is part of a screening process and the application of the PBPK model for
candidate critical effects reduces the values of some of the individual UFs for the p-cRfCs and p-cRfDs.
'Greater than usual degree of uncertainty (see Section 5.1.2).
dNo PBPK model based analyses were done, so cRfD on the basis of applied dose only.
rtr = route-to-route extrapolated result (no value for default methodology).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-80 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-21. Lowest p-cRfCs for candidate critical effects for different types
of effect based on primary dose metric
Type of effect
Neurological
Kidney
Liver
Immunological
Reproductive
Developmental
Effect
(primary dose metric)
Demyelination in hippocampus in rats
(TotMetabBW34)
Toxic nephropathy in rats
(ABioactDCVCBW34)
Increased liver weight in mice
(AMetLivlBW34)
Decreased thymus weight in mice
(TotMetabBW34)
Decreased ability of rat sperm to fertilize
(AUCCBld)
Heart malformations in rats
(TotOxMetabBW34)
p-cRfC (ppm)
0.007 (rtr)
0.0006 (rtr)
0.9
0.0003 (rtr)
0.009 (rtr)*
0.0004 (rtr)
4
5
6
7
*This value is supported by the p-cRfC value of 0.01 ppm for multiple testes and sperm effects from an inhalation
study in rats.
rtr = route-to-route extrapolated result.
9
10
11
12
Table 5-22. Lowest p-cRfDs for candidate critical effects for different types
of effect based on primary dose metric
Type of effect
Neurological
Kidney
Liver
Immunological
Reproductive
Developmental
Effect
(primary dose metric)
Demyelination in hippocampus in rats
(TotMetabBW34)
Toxic nephropathy in rats
(ABioactDCVCBW34)
Increased liver weight in mice
(AMetLivlBW34)
Decreased thymus weight in mice
(TotMetabBW34)
Decreased ability of rat sperm to fertilize (AUCCBld) &
multiple testes and sperm effects (TotMetabBW34)a
Heart malformations in rats
(TotOxMetabBW34)
p-cRfD
(mg/kg/d)
0.009
0.0003
0.8 (rtr)
0.0005
0.02
0.0005b
13
14
15
16
17
18
aEndpoints from two different studies yielded the same p-cRfD value.
bThis value is supported by the cRfD value of 0.0004 mg/kg/d derived for developmental immunotoxicity effects in
mice (Peden-Adams et al., 2006); however, no PBPK analyses were done for this latter effect, so the value of
0.0004 mg/kg/d is based on applied dose.
rtr = route-to-route extrapolated result.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-81 DRAFT—DO NOT CITE OR QUOTE
-------
1 For neurological, kidney, immunological, and developmental effects, the lowest p-cRfCs
2 were derived from oral studies by route-to-route extrapolation. This appears to be a function of
3 the lack of comparable inhalation studies for many effects studied via the oral exposure route, for
4 which there is a larger database of studies. For the liver and reproductive effects, inhalation
5 studies yielded a p-cRfC lower than the lowest route-to-route extrapolated p-cRfC for that type
6 of effect. Conversely, the lowest p-cRfDs were derived from oral studies with the exception of
7 reproductive effects, for which route-to-route extrapolation from an inhalation study in humans
8 also yielded among the lowest p-cRfDs. The only effect for which there were comparable
9 studies for comparing a p-cRfC from an inhalation study with a p-cRfC estimated by
10 route-to-route extrapolation from an oral study was increased liver weight in the mouse. The
11 primary dose metric of amount of TCE oxidized in the liver yielded similar p-cRfCs of 1.0 and
12 1.1 ppm for the inhalation result and the route-to-route extrapolated result, respectively (see
13 Table 5-10).
14 As can be seen in these tables, the most sensitive types of effects (the types with the
15 lowest p-cRfCs and p-cRfDs) appear to be developmental, kidney, and immunological (adult and
16 developmental) effects, and then neurological and reproductive effects, in that order. Lastly, the
17 liver effects have p-cRfC and p-cRfD values that are about 3l/2 orders of magnitude higher than
18 those for developmental, kidney, and immunological effects.
19
20 5.1.5.2. Reference Concentration
21 The goal is to select an overall RfC that is well supported by the available data (i.e.,
22 without excessive uncertainty given the extensive database) and protective for all the candidate
23 critical effects, recognizing that individual candidate RfC values are by nature somewhat
24 imprecise. The lowest candidate RfC values within each health effect category span a 3000-fold
25 range from 0.0003-0.9 ppm (see Table 5-21). One approach to selecting a RfC would be to
26 select the lowest calculated value of 0.0003 ppm for decreased thymus weight in mice.
27 However, as can be seen in Table 5-19, six p-cRfCs from both oral and inhalation studies are in
28 the relatively narrow range of 0.0003-0.003 ppm at the low end of the overall range. Given the
29 somewhat imprecise nature of the individual candidate RfC values, and the fact that multiple
30 effects/studies lead to similar candidate RfC values, the approach taken in this assessment is to
31 select a RfC supported by multiple effects/studies. The advantages of this approach, which is
32 only possible when there is a relatively large database of studies/effects and when multiple
33 candidate values happen to fall within a narrow range at the low end of the overall range, are that
34 it leads to a more robust RfC (less sensitive to limitations of individual studies) and that it
35 provides the important characterization that the RfC exposure level is similar for multiple
36 noncancer effects rather than being based on a sole explicit critical effect.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-82 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-23 summarizes the PODs and UFs for the six critical studies/effects
2 corresponding to the p-cRfCs that have been chosen to support the RfC for TCE noncancer
3 effects. Five of the lowest candidate p-cRfCs, ranging from 0.0003-0.003 ppm, for
4 developmental, kidney, and immunologic effects, are values derived from route-to-route
5 extrapolation using the PBPK model. The lowest p-cRfC estimate (for a primary dose metric)
6 from an inhalation studies is 0.001 ppm for kidney effects. For all six candidate RfCs, the PBPK
7 model was used for inter and intraspecies extrapolation, based on the preferred dose metric for
8 each endpoints. There is high confidence in the p-cRfCs for kidney effects (see Section 5.1.2.2)
9 for the following reasons: they are based on clearly adverse effects, two of the values are derived
10 from chronic studies, and the extrapolation to humans is based on dose metrics clearly related to
11 toxicity estimated with high confidence with the PBPK model developed in Section 3.5. There is
12 somewhat less confidence in the lowest p-cRfC for developmental effects (heart malformations)
13 (see Section 5.1.2.8) and the lowest p-cRfC estimates for immunological effects (see
14 Section 5.1.2.5). Thus, this assessment does not rely on any single estimate alone; however,
15 each estimate is supported by estimates of similar magnitude from other effects.
16 As a whole, the estimates support a preferred RfC estimate of 0.001 ppm (1 ppb or
17 5 ug/m3). This estimate is within approximately a factor of three of the lowest estimates of
18 0.0003 ppm for decreased thymus weight in mice, 0.0004 ppm for heart malformations in rats,
19 0.0006 ppm for toxic nephropathy in rats, 0.001 ppm for increased kidney weight in rats,
20 0.002 ppm for toxic nephrosis in mice, and 0.003 ppm for increased anti-dsDNA antibodies in
21 mice. Thus, there is robust support for a RfC of 0.001 ppm provided by estimates for multiple
22 effects from multiple studies. The estimates are based on PBPK model-based estimates of
23 internal dose for interspecies, intraspecies, and/or route-to-route extrapolation, and there is
24 sufficient confidence in the PBPK model, as well as support from mechanistic data for some of
25 the dose metrics (specifically TotOxMetabBW34 for the heart malformations and
26 ABioactDCVCBW34 and AMetGSHBW34 for toxic nephropathy) (see Section 5.1.3.1). Note
27 that there is some human evidence of developmental heart defects from TCE exposure in
28 community studies (see Section 4.8.3.1.1) and of kidney toxicity in TCE-exposed workers (see
29 Section 4.4.1).
30 In summary, the preferred RfC estimate is 0.001 ppm (1 ppb or 5 ug/m3) based on route-
31 to-route extrapolated results from oral studies for the critical effects of heart malformations
32 (rats), immunotoxicity (mice), and toxic nephropathy (rats, mice), and an inhalation study for the
33 critical effect of increased kidney weight (rats).
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-83 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-23. Summary of critical studies, effects, PODs, and UFs supporting
2 the RfC
NTP (1988)—Toxic nephropathy in female Marshall rats exposed for 104 weeks by oral gavage (5 d/wk).
• idPOD = 0.0132 mg DCVC bioactivated/kg3/7d, which is the BMDL from BMD modeling using
PBPK model-predicted internal doses, BMR = 5% (clearly toxic effect), and Log-logistic model
(see Appendix F, Section F.6.1).
• HEC99 = 0.0056 ppm (lifetime continuous exposure) derived from combined interspecies,
intraspecies, and route-to-route extrapolation using PBPK model.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
• p-cRfC = 0.0056/10 = 0.00056 ppm (3 ug/m3).
NCI (1976)—Toxic nephrosis in female B3C3F1 mice exposed for 78 weeks by oral gavage (5 d/wk).
• idPOD = 0.735 mg TCE conjugated with GSH/kgyVd, which is the PBPK model-predicted
internal dose at the applied dose LOAEL of 869 mg/kg/d (5 d/wk) (BMD modeling failed due to
almost maximal response at lowest dose) (see Appendix F, Section F.6.2).
• HEC99 = 0.50 ppm (lifetime continuous exposure) derived from combined interspecies,
intraspecies, and route-to-route extrapolation using PBPK model.
• UFioaei = 30 because POD is a LOAEL for an adverse effect with a response >90%.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
• p-cRfC = 0.50/300 = 0.0017 ppm (0.9 ug/m3).
Woolhiser et al. (2006)—Increased kidney weight in female S-D rats exposed for 4 weeks by inhalation
(6 h/d, 5 d/wk).
• idPOD = 0.0309 mg DCVC bioactivated/kgy7d, which is the BMDL from BMD modeling using
PBPK model-predicted internal doses, BMR = 10%, and Hill model with constant variance (see
Appendix F, Section F.6.3).
• HEC99 = 0.013 ppm (lifetime continuous exposure) derived from combined interspecies and
intraspecies extrapolation using PBPK model.
• UFSC = 1 because Kjellstrand et al. (1983b) reported that in mice, kidney effects after exposure for
120 d was no more severe than those after 30 d exposure.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
• p-cRfC = 0.013/10 = 0.0013 ppm (7 ug/m3).
Keil et al. (2009)—Decreased thymus weight in female B6C3F1 mice exposed for 30 weeks by drinking
water.
• idPOD = 0.139 mg TCE metabolized/kg3/7d, which is the PBPK model-predicted internal dose at
the applied dose LOAEL of 0.35 mg/kg/d (continuous) (no BMD modeling due to inadequate
model fit caused by supralinear dose-response shape) (see Appendix F, Section F.6.4).
• HEC99 = 0.033 ppm (lifetime continuous exposure) derived from combined interspecies,
intraspecies, and route-to-route extrapolation using PBPK model.
• UFioaei =10 because POD is a LOAEL for an adverse effect.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
• p-cRfC = 0.033/100 = 0.00033 ppm (2 ug/m3).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-84 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-23. Summary of critical studies, effects, PODs, and UFs supporting
the RfC (continued)
Keil et al. (2009)—Increased anti-dsDNA and anti-ssDNA antibodies in female B6C3F1 mice exposed
for 30 weeks by drinking water.
• idPOD = 0.139 mg TCE metabolized/kg3/7d, which is the PBPK model-predicted internal dose at
the applied dose LOAEL of 0.35 mg/kg/d (continuous) (no BMD modeling due to inadequate
model fit caused by supralinear dose-response shape) (see Appendix F, Section F.6.4).
• HEC99 = 0.033 ppm (lifetime continuous exposure) derived from combined interspecies,
intraspecies, and route-to-route extrapolation using PBPK model.
• UFioaei = 1 because POD is a LOAEL for an early marker for an adverse effect.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability
• p-cRfC = 0.033/10 = 0.0033 ppm (18 ug/m3).
Johnson et al. (2003)—fetal heart malformations in S-D rats exposed from GD 1-22 by drinking water.
• idPOD = 0.0142 mg TCE metabolized by oxidation/kgy7d, which is the BMDL from BMD
modeling using PBPK model-predicted internal doses, with highest-dose group (1,000-fold
higher than next highest-dose group) dropped, pup as unit of analysis, BMR =1% (due to
severity of defects, some of which could have been fatal), and a nested Log-logistic model to
account for intralitter correlation (see Appendix F, Section F.6.5).
• HEC99 = 0.0037 ppm (lifetime continuous exposure) derived from combined interspecies,
intraspecies, and route-to-route extrapolation using PBPK model.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
• p-cRfC = 0.0037/10 = 0.00037 ppm (2 ug/m3).
3 GD = gestation day.
4
5
6 5.1.5.3. Reference Dose
1 As with the RfC determination above, the goal is to select an overall RfD that is well
8 supported by the available data (i.e., without excessive uncertainty given the extensive database)
9 and protective for all the candidate critical effects, recognizing that individual candidate RfD
10 values are by nature somewhat imprecise. The lowest candidate RfD values within each health
11 effect category span a nearly 3,000-fold range from 0.0003-0.8 mg/kg/d (see Table 5-21). One
12 approach to selecting a RfC would be to select the lowest calculated value of 0.0003 ppm for
13 toxic nephropathy in rats. However, as can be seen in Table 5-20, multiple p-cRfDs or cRfDs
14 from oral studies are in the relatively narrow range of 0.0003-0.0005 mg/kg/d at the low end of
15 the overall range. Given the somewhat imprecise nature of the individual candidate RfD values,
16 and the fact that multiple effects/studies lead to similar candidate RfD values, the approach taken
17 in this assessment is to select a RfD supported by multiple effects/studies. The advantages of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-85 DRAFT—DO NOT CITE OR QUOTE
-------
1 this approach, which is only possible when there is a relatively large database of studies/effects
2 and when multiple candidate values happen to fall within a narrow range at the low end of the
3 overall range, are that it leads to a more robust RfD (less sensitive to limitations of individual
4 studies) and that it provides the important characterization that the RfD exposure level is similar
5 for multiple noncancer effects rather than being based on a sole explicit critical effect.
6 Table 5-24 summarizes the PODs and UFs for the four critical studies/effects
7 corresponding to the p-cRfDs or cRfDs that have been chosen to support the RfD for TCE
8 noncancer effects. Three of the lowest p-cRfDs for the primary dose metrics—0.0003 mg/kg/d
9 for toxic nephropathy in rats and 0.0005 mg/kg/d for heart malformations in rats and decreased
10 thymus weights in mice—are derived using the PBPK model for inter and intraspecies
11 extrapolation. The other of these lowest values—0.0004 mg/kg/d for developmental
12 immunotoxicity (decreased PFC response and increased delayed-type hypersensitivity) in
13 mice—is based on applied dose. There is high confidence in the p-cRfD for kidney effects (see
14 Section 5.1.2.2), which is based on clearly adverse effects, derived from a chronic study, and
15 extrapolated to humans based on a dose metric clearly related to toxicity estimated with high
16 confidence with the PBPK model developed in Section 3.5. There is somewhat less confidence
17 in the p-cRfDs for decreased thymus weights (see Section 5.1.2.5) and heart malformations and
18 developmental immunological effects (see Section 5.1.2.8). Thus, this assessment does not rely
19 on any single estimate alone; however, each estimate is supported by estimates of similar
20 magnitude from other effects.
21 As a whole, the estimates support a preferred RfD of 0.0004 mg/kg/d. This estimate is
22 within 25% of the lowest estimates of 0.0003 for toxic nephropathy in rats, 0.0004 mg/kg/d for
23 developmental immunotoxicity (decreased PFC and increased delayed-type hypersensitivity) in
24 mice, and 0.0005 mg/kg/d for heart malformations in rats and decreased thymus weights in mice.
25 Thus, there is strong, robust support for a RfD of 0.0004 mg/kg/d provided by the concordance
26 of estimates derived from multiple effects from multiple studies. The estimates for kidney
27 effects, thymus effects, and developmental heart malformations are based on PBPK model-based
28 estimates of internal dose for interspecies and intraspecies extrapolation, and there is sufficient
29 confidence in the PBPK model, as well as support from mechanistic data for some of the dose
30 metrics (specifically TotOxMetabBW34 for the heart malformations and ABioactDCVCBW34
31 for toxic nephropathy) (see Section 5.1.3.1). Note that there is some human evidence of
32 developmental heart defects from TCE exposure in community studies (see Section 4.8.3.1.1)
33 and of kidney toxicity in TCE-exposed workers (see Section 4.4.1).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-86 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-24. Summary of critical studies, effects, PODs, and UFs supporting
2 the RfD
NTP (1988)—Toxic nephropathy in female Marshall rats exposed for 104 weeks by oral gavage (5 d/wk).
• idPOD = 0.0132 mg DCVC bioactivated/kgy7d, which is the BMDL from BMD modeling using
PBPK model-predicted internal doses, BMR = 5% (clearly toxic effect), and Log-logistic model
(see Appendix F, Section F.6.1).
• HED99 = 0.0034 mg/kg/d (lifetime continuous exposure) derived from combined interspecies and
intraspecies extrapolation using PBPK model.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
» p-cRfD = 0.0034/10 = 0.00034 mg/kg/d.
Keil et al. (2009)—Decreased thymus weight in female B6C3F1 mice exposed for 30 weeks by drinking
water.
• idPOD = 0.139 mg TCE metabolized/kg3/7d, which is the PBPK model-predicted internal dose at
the applied dose LOAEL of 0.35 mg/kg/d (continuous) (no BMD modeling due to inadequate
model fit caused by supralinear dose-response shape) (see Appendix F, Section F.6.4).
• HED99 = 0.048 mg/kg/d (lifetime continuous exposure) derived from combined interspecies and
intraspecies extrapolation using PBPK model.
• UFioaei =10 because POD is a LOAEL for an adverse effect.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
» p-cRfD = 0.048/100 = 0.00048 mg/kg/d.
Peden-Adams et al. (2006)—Decreased PFC response (3 and 8 weeks), increased delayed-type
hypersensitivity (8 weeks) in pups exposed from GD 0 to 3- or 8-weeks-of-age through drinking water
(placental and lactational transfer, and pup ingestion).
• POD = 0.37 mg/kg/d is the applied dose LOAEL (estimated daily dam dose) (no BMD modeling
due to inadequate model fit caused by supralinear dose-response shape). No PBPK modeling was
attempted due to lack of appropriate models/parameters to account for complicated fetal/pup
exposure pattern (see Appendix F, Section F.6.6).
• UFi0aei =10 because POD is a LOAEL for multiple adverse effects.
• UF1S = 10 for interspecies extrapolation because PBPK model was not used.
• UFh = 10 for human variability because PBPK model was not used.
» cRfD = 0.37/1000 = 0.00037 mg/kg/d.
Johnson et al. (2003)—fetal heart malformations in S-D rats exposed from GD 1-22 by drinking water
• idPOD = 0.0142 mg TCE metabolized by oxidation/kgy7d, which is the BMDL from BMD
modeling using PBPK model-predicted internal doses, with highest-dose group (1,000-fold
higher than next highest-dose group) dropped, pup as unit of analysis, BMR =1% (due to
severity of defects, some of which could have been fatal), and a nested Log-logistic model to
account for intralitter correlation (see Appendix F, Section F.6.5).
• HED99 = 0.0051 mg/kg/d (lifetime continuous exposure) derived from combined interspecies and
intraspecies extrapolation using PBPK model.
• UF1S = 3.16 because the PBPK model was used for interspecies extrapolation.
• UFh = 3.16 because the PBPK model was used to characterize human toxicokinetic variability.
» p-cRfD = 0.0051/10 = 0.00051 mg/kg/d.
4
5 GD = gestation day.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-87 DRAFT—DO NOT CITE OR QUOTE
-------
1 In summary, the preferred RfD estimate is 0.0004 mg/kg/d based on the critical effects of
2 heart malformations (rats), adult immunological effects (mice), developmental immunotoxicity
3 (mice), and toxic nephropathy (rats).
4
5 5.2. DOSE-RESPONSE ANALYSIS FOR CANCER ENDPOINTS
6 This section describes the dose-response analysis for cancer endpoints. Section 5.2.1
7 discusses the analyses of data from chronic rodent bioassays. Section 5.2.2 discusses the
8 analyses of human epidemiologic data. Section 5.2.3 discusses the choice of the preferred
9 inhalation unit risk and oral unit risk estimates, as well as the application of age-dependent
10 adjustment factors to the unit risk estimates.
11
12 5.2.1. Dose-Response Analyses: Rodent Bioassays
13 This section describes the calculation of cancer unit risk estimates based on rodent
14 bioassays. First, all the available studies (i.e., chronic rodent bioassays) were considered, and
15 those suitable for dose-response modeling were selected for analysis (see Section 5.2.1.1). Then
16 dose-response modeling using the linearized multistage model was performed using applied
17 doses (default dosimetry) as well as PBPK model-based internal doses (see Section 5.2.1.2).
18 Bioassays for which time-to-tumor data were available were analyzed using poly-3 adjustment
19 techniques and using a Multistage Weibull model. In addition, a cancer potency estimate for
20 different tumor types combined was derived from bioassays in which there was more than one
21 type of tumor response in the same sex and species. Unit risk estimates based on PBPK model -
22 estimated internal doses were then extrapolated to human population unit risk estimates using the
23 human PBPK model. From these results (see Section 5.2.1.3), estimates from the most sensitive
24 bioassay (i.e., that with the greatest unit risk estimate) for each combination of administration
25 route, sex, and species, based on the PBPK model-estimated internal doses, were considered as
26 candidate unit risk estimates for TCE. Uncertainties in the rodent-based dose-response analyses
27 are described in Section 5.2.1.4.
28
29 5.2.1.1. Rodent Dose-Response Analyses: Studies and Modeling Approaches
30 The rodent cancer bioassays that were identified for consideration for dose-response
31 analysis are listed in Tables 5-25 (inhalation bioassays) and 5-26 (oral bioassays) for each
32 sex/species combination. The bioassays selected for dose-response analysis are marked with an
33 asterisk; rationales for rejecting the bioassays that were not selected are provided in the
34 "Comments" columns of the tables. For the selected bioassays, the tissues/organs that exhibited
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-88 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
a TCE-associated carcinogenic response and for which dose-response modeling was performed
are listed in the "Tissue/Organ" columns.
Table 5-25. Inhalation bioassays
Study
Strain
Tissue/Organ
Comments
Female mice
*Fukuda et al., 1983
* Henschleretal., 1980
*Maltoni et al., 1986
Maltoni et al., 1986
Crj:CD-l (ICR)
Han:NMRI
B6C3F1
Swiss
Lung
Lymphoma
Liver, Lung
-
No dose-response
Male mice
Henschleretal., 1980
Maltoni et al., 1986
Maltoni et al., 1986
* Maltoni et al., 1986
Han:NMRI
B6C3F1
B6C3F1
Swiss
-
Liver
Liver
Liver
No dose-response
Exp #BT306: excessive fighting
Exp #BT306bis. Results similar
to Swiss mice
Female rats
Fukudaetal., 1983
Henschleretal., 1980
Maltoni et al., 1986
Sprague-Dawley
Wistar
Sprague-Dawley
-
-
-
No dose-response
No dose-response
No dose-response
Male rats
Henschleretal., 1980
* Maltoni et al., 1986
Wistar
Sprague-Dawley
-
Kidney, Leydig
cell, Leukemia
No dose-response
6
7
* Selected for dose-response analysis.
"No dose-response" = no tumor incidence data suitable for dose-response modeling.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-89 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-26. Oral bioassays
Study
Strain
Tissue/organ
Comments
Female mice
Henschleretal., 1984
*NCI, 1976
NTP, 1990
VanDuren et al., 1979
Han:NMRI
B6C3F1
B6C3F1
Swiss
-
Liver, lung, sarcomas
and lymphomas
Liver, lung,
lymphomas
Liver
Toxicity, no dose-response
Single dose
Single dose, no dose-response
Male mice
Annaetal., 1994
Bull et al., 2002
Henschleretal., 1984
*NCI, 1976
NTP, 1990
VanDuren et al., 1979
B6C3F1
B6C3F1
Han:NMRI
B6C3F1
B6C3F1
Swiss
Liver
Liver
-
Liver
Liver
-
Single dose
Single dose
Toxicity, no dose-response
Single dose
Single dose, no dose-response
Female rats
NCI, 1976
NTP, 1988
*NTP, 1988
NTP, 1988
NTP, 1988
NTP, 1990
Osborne-Mendel
ACI
August
Marshall
Osborne-Mendel
F344/7V
-
-
Leukemia
-
Adrenal cortex
-
Toxicity, no dose-response
No dose-response
No dose-response
Adenomas only
No dose-response
Male rats
NCI, 1976
NTP, 1988
*NTP, 1988
*NTP, 1988
*NTP, 1988
*NTP, 1990
Osborne-Mendel
ACI
August
Marshall
Osborne-Mendel
F344/7V
-
-
Subcutaneous tissue
sarcomas
Testes
Kidney
Kidney
Toxicity, no dose-response
No dose-response
3
4
5
6
7
8
9
10
* Selected for dose-response analysis.
"No dose-response" = no tumor incidence data suitable for dose-response modeling.
The general approach used was to model each sex/species/bioassay tumor response to
determine the most sensitive bioassay response (in terms of human equivalent exposure or dose)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-90 DRAFT—DO NOT CITE OR QUOTE
-------
1 for each sex/species combination. The various modeling approaches, model selection, and unit
2 risk derivation are discussed below. Modeling was done using the applied dose or exposure
3 (default dosimetry) and several internal dose metrics. The dose metrics used in the dose-
4 response modeling are discussed in Section 5.2.1.2. Because of the large volume of analyses and
5 results, detailed discussions about how the data were modeled using the various dosimetry and
6 modeling approaches and results for individual data sets are provided in Appendix G. The
7 overall results are summarized and discussed in Section 5.2.1.3.
8 Most tumor responses were modeled using the multistage model in U.S. EPA's BMDS
9 (www.epa.gov/ncea/bmds). The multistage model is a flexible model, capable of fitting most
10 cancer bioassay data, and it is U.S. EPA's long-standing model for the modeling of such cancer
11 data. The multistage model has the general form
12
13 P(d)= 1 - exp[-(q0+qid + q2d2 + - +qkd")], (Eq. 5-1)
14
15 where P(d) represents the lifetime risk (probability) of cancer at dose d, and parameters qt > 0,
16 for /' = 0, 1, ..., k. For each data set, the multistage model was evaluated for one stage and (n - 1)
17 stages, where n is the number of dose groups in the bioassay. A detailed description of how the
18 data were modeled, as well as tables of the dose-response input data and figures of the multistage
19 modeling results, is provided in Appendix G.
20 Only models with acceptable fit (p > 0.05) were considered. If 1-parameter and
21 2-parameter models were both acceptable (in no case was there a 3-parameter model), the more
22 parsimonious model (i.e., the 1-parameter model) was selected unless the inclusion of the
23 2nd parameter resulted in a statistically significant19 improvement in fit. If two different
24 1-parameter models were available (e.g., a 1-stage model and a 3-stage model with /?; and $2
25 both equal to 0), the one with the best fit, as indicated by the lowest AIC value, was selected. If
26 the AIC values were the same (to three significant figures), then the lower-stage model was
27 selected. Visual fit and scaled chi-square residuals were also considered for confirmation in
28 model selection. For two data sets, the highest-dose group was dropped to improve the fit in the
29 lower dose range.
30 From the selected model for each data set, the maximum likelihood estimate (MLE) for
31 the dose corresponding to a specified level of risk (i.e., the benchmark dose, or BMD) and its
32 95% lower confidence bound (BMDL) were estimated.20 In most cases, the risk level, or BMR,
19Using a standard criterion for nested models, that the difference in -2*log-likelihood exceeds 3.84 (the
20BMDS estimates confidence intervals using the profile likelihood method.
This document is a draft for review purposes only and does not constitute Agency policy.
1 0/20/09 5-9 1 DRAFT— DO NOT CITE OR QUOTE
-------
1 was 10% extra risk;21 however, in a few cases with low response rates, a BMR of 5%, or even
2 1%, extra risk was used to avoid extrapolation above the range of the data. As discussed in
3 Section 4.4, there is sufficient evidence to conclude that a mutagenic MOA is operative for TCE-
4 induced kidney tumors, so linear extrapolation from the BMDL to the origin was used to derive
5 unit risk estimates (or "slope factors" for oral exposures) for this site. For all other tumor types,
6 the available evidence supports the conclusion that the MOA(s) for TCE-induced rodent tumors
7 is unknown, as discussed in Sections 4.5-4.10 and summarized in Section 4.11.2.3. Therefore,
8 linear extrapolation was also used based on the general principles outlined in U.S. EPA's
9 Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) and reviewed below in
10 Section 5.2.1.4.1. Thus, for all TCE-associated rodent tumors, unit risk estimates are equal to
11 BMR/BMDL (e.g., 0.10/BMDLio for a BMR of 10%). See Section 5.2.1.3 for a summary of the
12 unit risk estimates for each sex/species/bioassay/tumor type.
13 Some of the bioassays exhibited differential early mortality across the dose groups, and,
14 for three such male rat studies (identified with checkmarks in the "Time-to-tumor" column of
15 Table 5-27), analyses that take individual animal survival times into account were performed.
16 (For bioassays with differential early mortality occurring primarily before the time of the
17 1st tumor [or 52 weeks, whichever came first], the effects of early mortality were largely
18 accounted for by adjusting the tumor incidence for animals at risk, as described in Appendix G,
19 and the dose-response data were modeled using the regular multistage model, as discussed
20 above, rather than approaches that account for individual animal survival times.) Two
21 approaches were used to take individual survival times into account. First, U.S. EPA's
22 Multistage Weibull (MSW) software22 was used for time-to-tumor modeling. The Multistage
23 Weibull time-to-tumor model has the general form
24
25 P(d,t) = l ~ exp[-( 1,
28 t0 > 0, and qt > 0 for /' = 0,1 ,...,&, where k = the number of dose groups; the parameter t0 represents
29 the time between when a potentially fatal tumor becomes observable and when it causes death.
30 (All of our analyses used the model for incidental tumors, which has no t0 term.) Although the
31 fit of the MSW model can be assessed visually using the plot feature of the MSW software,
32 because there is no applicable goodness-of-fit statistic with a well-defined asymptotic
21Extra risk over the background tumor rate is defined as [P(d) - P(0)]/[l - P(0)], where P(d) represents the lifetime
risk (probability) of cancer at dose d.
22This software has been thoroughly tested and externally reviewed. In February 2009, it will become available on
U.S. EPA's Web site.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-92 DRAFT—DO NOT CITE OR QUOTE
-------
1 distribution, an alternative survival-adjustment technique, "poly-3 adjustment," was also applied
2 (Portier and Bailer, 1989). This technique was used to adjust the tumor incidence denominators
3 based on the individual animal survival times.23 The adjusted incidence data then served as
4 inputs for U.S. EPA's BMDS multistage model, and model (i.e., stage) selection was conducted
5 as already described above. Under both survival-adjustment approaches, BMDs and BMDLs
6 were obtained and unit risks derived as discussed above for the standard multistage model
7 approach. See Appendix G for a more detailed description of the MSW modeling and for the
8 results of both the MSW and poly-3 approaches for the individual data sets. A comparison of the
9 results for the three different data sets and the various dose metrics used is presented in
10 Section 5.2.1.3.
11
23Each tumorless animal is weighted by its fractional survival time (number of days on study divided by 728 days,
the typical number of days in a 2-year bioassay) raised to the power of 3 to reflect the fact that animals are at greater
risk of cancer at older ages. Animals with tumors are given a weight of 1. The sum of the weights of all the animals
in an exposure group yields the effective survival-adjusted denominator.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-93 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-27. Specific dose-response analyses performed and dose metrics used
to
Bioassay
Strain
Endpoint
Applied
dose
PBPK-based—
primary dose metric
PBPK-based—
alternative dose
metric(s)
Time-
to-
tumor
INHALATION
Female mice
Fukudaetal., 1983
Henschler et al.,
1980
Maltoni et al., 1986
Crj:CD-l (ICR)
Han:NMRI
B6C3F1
Lung adenomas and carcinomas
Lymphoma
Liver hepatomas
Lung adenomas and carcinomas
Combined risk
V
V
V
V
V
AMetLngBW34
TotMetabBW34
AMetLivlBW34
AMetLngBW34
TotOxMetabBW34
AUCCBld
AUCCBld
TotOxMetabBW34
TotOxMetabBW34
AUCCBld
Male mice
Maltoni et al., 1986
Swiss
Liver hepatomas
V
AMetLivlBW34
TotOxMetabBW34
Female rats
None selected
Male rats
Maltoni et al., 1986
Sprague-
Dawley
Kidney adenomas and carcinomas
Leydig cell tumors
Leukemias
Combined risk
V
V
V
V
ABioactDCVCBW34
TotMetabBW34
TotMetabBW34
AMetGSHBW34
TotMetabBW34
AUCCBld
AUCCBld
§
i
TO'
Y1
VO ^
I
H I
O >
HH Oq
H TO
O
H
W
-------
Table 5-27. Specific dose-response analyses performed and dose metrics used (continued)
Bioassay
Strain
Endpoint
Applied
dose
PBPK-based—
primary dose metric
PBPK-based—
alternative dose
metric(s)
Time-
to-
tumor
ORAL
Female mice
NCI, 1976
B6C3F1
Liver carcinomas
Lung adenomas and carcinomas
Multiple sarcomas/lymphomas
Combined risk
V
V
V
V
AMetLivlBW34
AMetLngBW34
TotMetabBW34
TotOxMetabBW34
TotOxMetabBW34
AUCCBld
AUCCBld
Male mice
NCI, 1976
B6C3F1
Liver carcinomas
V
AMetLivlBW34
TotOxMetabBW34
Female rats
NTP, 1988
August
Leukemia
V
TotMetabBW34
AUCCBld
Male rats
NTP, 1988
NTP, 1988
NTP, 1988
NTP, 1990
August
Marshall
Osborne-
Mendel
F344/7V
Subcutaneous tissue sarcomas
Testicular interstitial cell tumors
Kidney adenomas and carcinomas
Kidney adenomas and carcinomas
V
V
V
V
TotMetabBW34
TotMetabBW34
ABioactDCVCBW34
ABioactDCVCBW34
AUCCBld
AUCCBld
AMetGSHBW34
TotMetabBW34
AMetGSHBW34
TotMetabBW34
V
V
V
PBPK-based dose metric abbreviations:
ABioactDCVCBW34 = Amount of DCVC bioactivated in the kidney per unit body weight54 (mg DCVC/kgy7week).
AMetGSHBW34 = Amount of TCE conjugated with GSH per unit body weight'7' (mg TCE/kgyVweek).
AMetLivlBW34 = Amount of TCE oxidized per unit body weight54 (mg TCE/kgyYweek).
AMetLngBW34 = Amount of TCE oxidized in the respiratory tract per unit body weight54 (mg TCE/kgyVweek).
AUCCBld = Area under the curve of the venous blood concentration of TCE (mg-hour/L/week).
TotMetabBW34 = Total amount of TCE metabolized per unit body weight54 (mg TCE/kgy7week).
TotOxMetabBW34 = Total amount of TCE oxidized per unit body weight54 (mg TCE/kgyVweek).
-------
1 For bioassays that exhibited more than one type of tumor response in the same sex and
2 species (these studies have a row for "combined risk" in the "Endpoint" column of Table 5-27),
3 the cancer potency for the different tumor types combined was estimated. The combined tumor
4 risk estimate describes the risk of developing tumors for any (not all together) of the tumor types
5 that exhibited a TCE-associated tumor response; this estimate then represents the total excess
6 cancer risk. The model for the combined tumor risk is also multistage, with the sum of the stage-
7 specific multistage coefficients from the individual tumor models serving as the stage-specific
8 coefficients for the combined risk model (i.e., for each qf, qi[COmbined\ = qn + qt2 + ••• + qtk, where
9 the <7;S are the coefficients for the powers of dose and k is the number of tumor types being
10 combined) (Bogen, 1990; NRC, 1994). This model assumes that the occurrences of two or more
11 tumor types are independent. Although the resulting model equation can be readily solved for a
12 given BMR to obtain an MLE (BMD) for the combined risk, the confidence bounds for the
13 combined risk estimate are not calculated by available modeling software. Therefore, the
14 confidence bounds on the combined BMD were estimated using a Bayesian approach, computed
15 using Markov chain Monte Carlo techniques and implemented using the freely available
16 WinBugs software (Spiegelhalter et al., 2003). Use of WinBugs for derivation of a distribution
17 of BMDs for a single multistage model has been demonstrated by Kopylev et al. (2007), and this
18 approach can be straightforwardly generalized to derive the distribution of BMDs for the
19 combined tumor load. For further details on the implementation of this approach and for the
20 results of the analyses, see Appendix G.
21
22 5.2.1.2. Rodent Dose-Response Analyses: Dosimetry
23 In modeling the applied doses (or exposures), default dosimetry procedures were applied
24 to convert applied rodent doses to human equivalent doses. Essentially, for inhalation exposures,
25 "ppm equivalence" across species was assumed. For oral doses, 3/4-power body-weight scaling
26 was used, with a default average human body weight of 70 kg. See Appendix G for more details
27 on the default dosimetry procedures.
28 In addition to applied doses, several internal dose metrics were used in the dose-response
29 modeling for each tumor type. Use of internal dose metrics in dose-response modeling is
30 described here briefly. For more details on the PBPK modeling used to estimate the levels of the
31 dose metrics corresponding to different exposure scenarios in rodents and humans, as well as a
32 qualitative discussion of the uncertainties and limitations of the model, see Section 3.5; for a
33 more detailed discussion of how the dose metrics were used in dose-response modeling, see
34 Appendix G. Quantitative analyses of the uncertainties and their implications for dose-response
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-96 DRAFT—DO NOT CITE OR QUOTE
-------
1 assessment, utilizing the results of the Bayesian analysis of the PBPK model, are discussed
2 separately in Section 5.2.1.4.2.
O
4 5.2.1.2.1. Selection of dose metrics for different tumor types. One area of scientific
5 uncertainty in cancer dose-response assessment is the appropriate scaling between rodent and
6 human doses for equivalent responses. As discussed above, for applied dose, the standard
7 dosimetry assumptions for equal lifetime carcinogenic risk are, for inhalation exposure, the same
8 lifetime exposure concentration in air, and, for oral exposure, the same lifetime daily dose scaled
9 by body weight to the % power. For scaling internal doses, it is useful to consider two possible
10 interpretations of these standard dosimetry assumptions. The first (denoted "empirical
11 dosimetry") interpretation is that standard dosimetry is based on the empirical finding that
12 scaling the delivered dose rate by body weight to the % power results in equivalent toxicity (e.g.,
13 Travis and White, 1988; U.S. EPA, 1992). This is supported biologically by data showing that
14 rates of both kinetic and dynamic physiologic processes are generally consistent with % power of
15 body weight scaling across species (U.S. EPA, 1992). Note also that this applies to inhalation
16 exposure because the delivered dose rate in that case is the air concentration multiplied by the
17 ventilation rate, which scales by body weight to the % power. Applying this interpretation to
18 internal doses would imply that the dose rate of the active moiety delivered to the target tissue,
19 scaled by body weight to the 3/4 power, would be assumed to result in equivalent responses. The
20 second (denoted "concentration equivalence dosimetry") interpretation hypothesizes that the
21 empirical finding is pharmacokinetically-driven, due to the body weight to the 3/4 scaling of
22 physiologic flows (cardiac output, ventilation rate, glomerular filtration, etc.) and metabolic rates
23 (enzyme-mediated biotransformation). Therefore, the standard dosimetry assumptions yield
24 equivalent average internal concentrations, which in turn yield equivalent carcinogenic risk
25 (NRC, 1986, 1987). Applying this dosimetry interpretation to internal doses would imply that
26 equivalent carcinogenic risk should be based on equal (average) concentrations of the active
27 moiety or moieties at the target tissue.
28 To the extent that production and clearance of the active moiety or moieties all scale by
29 body weight to the % power, these two dosimetry interpretations both lead to the same
30 quantitative results. However, these interpretations may lead to different quantitative results
31 when there are deviations of the underlying physiologic or metabolic processes from body
32 weight to the % power scaling. For instance, as discussed in Section 3.5, the PBPK model
33 predictions for AUC of TCE in blood deviate from the body weight to the 3/4 scaling (the scaling
34 is closer to mg/kg/d than mg/kg Yd), so use of this dose metric when TCE is the active moiety
35 implicitly assumes the "concentration equivalence dosimetry." In addition, as discussed below,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-97 DRAFT—DO NOT CITE OR QUOTE
-------
1 in most cases involving TCE metabolites, only the rate of production of the active moiety(ies) or
2 the rate of transformation through a particular metabolic pathway can be estimated using the
3 PBPK model, and the actual concentration of the active moiety(ies) cannot be estimated due to
4 data limitations. Under "empirical dosimetry," these metabolism rates, which are estimates of
5 the systemic or tissue-specific delivery of the active moiety(ies), would be scaled by body weight
6 to the % power to yield equivalent carcinogenic risk. Under "concentration equivalence
7 dosimetry," additional assumptions about the rate of clearance are necessary to specify the
8 scaling that would yield concentration equivalence. In the absence of data, active metabolites are
9 assumed to be sufficiently stable so that clearance is via enzyme-catalyzed transformation or
10 systemic excretion (e.g., blood flow, glomerular filtration), which scale approximately by body
11 weight to the % power. Therefore, under "concentration equivalence dosimetry," the metabolism
12 rates would also be scaled by body weight to the % power in the absence of additional data.
13 For toxicity that is associated with local (in situ) production of "reactive" metabolites
14 whose concentrations cannot be directly measured in the target tissue, an alternative approach,
15 under "concentration equivalence dosimetry," of scaling by unit tissue mass has been proposed
16 (e.g., Andersen et al., 1987). As discussed by Travis (1990), in this situation, scaling the rate of
17 local metabolism across species and individuals by tissue mass is appropriate if the metabolites
18 are sufficiently reactive and are cleared by "spontaneous" deactivation (i.e., changes in chemical
19 structure without the need of biological influences). Thus, use of this alternative scaling
20 approach requires that (1) the active moiety or moieties do not leave the target tissue in
21 appreciable quantities (i.e., are cleared primarily by in situ transformation to other chemical
22 species and/or binding to/reactions with cellular components); and (2) the clearance of the active
23 moieties from the target tissue is governed by biochemical reactions whose rates are independent
24 of body weight (e.g., purely chemical reactions). If these conditions are met, then under the
25 "concentration equivalence dosimetry," the relevant metabolism rates estimated by the PBPK
26 model would be scaled by tissue mass, rather than by body weight to the % power.
27 To summarize, the appropriate internal dose metric for equivalent carcinogenic responses
28 can be specified by invoking one of two alternative interpretations of the standard dosimetry for
29 applied dose: "empirical dosimetry" based on the rate at which the active moiety(ies) is(are)
30 delivered to the target tissue scaled by body weight to the 3/4 power or "concentration equivalence
31 dosimetry" based on matching internal concentrations of the active moiety (ies) in the target
32 tissue. If the active moiety(ies) is TCE itself or a putatively reactive metabolite, the choice of
33 interpretation will affect the choice of internal dose metric. In the discussions of dose metric
34 selections for the individual tumors sites below, the implications of both "empirical dosimetry"
35 and "concentration equivalence dosimetry" are discussed. Additionally, an attempt was made to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-98 DRAFT—DO NOT CITE OR QUOTE
-------
1 use tissue-specific dose metrics representing particular pathways or metabolites identified from
2 available data as having a likely role in the induction of a tissue-specific cancer. Where
3 insufficient information was available to establish particular metabolites or pathways of likely
4 relevance to a tissue-specific cancer, more general "upstream" metrics representing either parent
5 compound or total metabolism had to be used. In addition, the selection of dose metrics was
6 limited to metrics that could be adequately estimated by the PBPK model (see Section 3.5). The
7 (PBPK-based) dose metrics used for the different tumor types are listed in Table 5-27. For each
8 tumor type, the "primary" dose metric referred to in Table 5-27 is the metric representing the
9 particular metabolite or pathway whose involvement in carcinogenicity has the greatest
10 biological support, whereas "alternative" dose metrics represent upstream metabolic pathways
11 (or TCE distribution, in the case of AUCCBld) that may be more generally involved.
12
13 5.2.1.2.1.1. Kidney. As discussed in Sections 4.4.6-4.4.7, there is sufficient evidence to
14 conclude that TCE-induced kidney tumors in rats are primarily caused by GSH-conjugation
15 metabolites either produced in situ in or delivered systemically to the kidney. As discussed in
16 Section 3.3.3.2, bioactivation of these metabolites within the kidney, either by beta-lyase, FMO,
17 or P450s, produces reactive species. Therefore, multiple lines of evidence support the
18 conclusion that renal bioactivation of DCVC is the preferred basis for internal dose
19 extrapolations of TCE-induced kidney tumors. However, uncertainties remain as to the relative
20 contributions from each bioactivation pathway, and quantitative clearance data necessary to
21 calculate the concentration of each species are lacking.
22 Under "empirical dosimetry," the rate of renal bioactivation of DCVC would be scaled by
23 body weight to the % power. As discussed above, under "concentration equivalence dosimetry,"
24 when the concentration of the active moiety cannot be estimated, qualitative data on the nature of
25 clearance of the active moiety or moieties can be used to inform whether to scale the rate of
26 metabolism by body weight to the 3A power or by the target tissue weight. For the beta-lyase
27 pathway, Dekant et al. (1988) reported in trapping experiments that the postulated reactive
28 metabolites decompose to stable (unreactive) metabolites in the presence of water. Moreover,
29 the necessity of a chemical trapping mechanism to detect the reactive metabolites suggests a very
30 rapid reaction such that it is unlikely that the reactive metabolites leave the site of production.
31 Therefore, these data support the conclusion that, for this bioactivation pathway, clearance is
32 chemical in nature and hence species-independent. If this were the only bioactivation pathway,
33 then the scaling by kidney weight would be supported. With respect to the FMO bioactivation
34 pathway, Sausen and Elfarra (1991) reported that after direct dosing of the postulated reactive
35 sulfoxide, the sulfoxide was detected as an excretion product in bile. These data suggest that
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-99 DRAFT—DO NOT CITE OR QUOTE
-------
1 reactivity in the tissue to which the sulfoxide was delivered (the liver, in this case) is insufficient
2 to rule out a significant role for enzymatic or systemic clearance. Therefore, according to the
3 criteria outlined above, for this bioactivation pathway, the data support scaling the rate of
4 metabolism by body weight to the 3/4 power. For P450-mediated bioactivation producing
5 NAcDCVC sulfoxide, the only relevant data on clearance are from a study of the structural
6 analogue to DCVC, FDVE (Sheffels et al., 2004), which reported that the postulated reactive
7 sulfoxide was detected in urine. This suggests that the sulfoxide is sufficiently stable to be
8 excreted by the kidney and supports the scaling of the rate of metabolism by body weight to the
9 % power.
10 Therefore, because the contributions to TCE-induced nephrocarcinogenicity from each
11 possible bioactivation pathway are not clear, and, even under "concentration equivalence
12 dosimetry," the scaling by body weight to the % power is supported for two of the three
13 bioactivation pathways, it is decided here to scale the DCVC bioactivation rate by body weight
14 to the % power. The primary internal dose metric for TCE-induced kidney tumors is, thus, the
15 weekly rate of DCVC bioactivation per unit body weight to the % power (ABioactDCVCBW34
16 [mg/kgy7week]). However, it should be noted that due to the larger relative kidney weight in
17 rats as compared to humans, scaling by kidney weight instead of body weight to the % power
18 would only change the quantitative interspecies extrapolation by about 2-fold,24 so the sensitivity
19 of the results to the scaling choice is relatively small.
20 To summarize, under the "empirical dosimetry" approach, the underlying assumption for
21 the ABioactDCVCBW34 dose metric is that equalizing the rate of renal bioactivation of DCVC
22 (i.e., local production of active moiety(ies) in the target tissue), scaled by the 3/4 power of body
23 weight, yields equivalent lifetime cancer risk across species. Under "concentration equivalence
24 dosimetry," the underlying assumptions for the ABioactDCVCBW34 dose metric are that (1) the
25 same average concentration of reactive species produced from DCVC in the kidney leads to a
26 similar lifetime cancer risk across species; and (2) the rate of clearance of these reactive species
27 scales by the 3/4 power of body weight (e.g., assumed for enzyme-activity or blood-flow).
28 An alternative dose metric that also involves the GSH conjugation pathway is the amount
29 of GSH conjugation scaled by the 3/4 power of body weight (AMetGSHBW34 [mg/kgy7week]).
30 This dose metric uses the total flux of GSH conjugation as the toxicologically-relevant dose, and,
31 thus, incorporates any direct contributions from DCVG and DCVC, which are not addressed in
32 the DCVC bioactivation metric. Under the "empirical dosimetry" approach, the underlying
24The range of the difference is 2.1-2.4-fold using the posterior medians for the relative kidney weight in rats and
humans from the PBPK model described in Section 3.5 (see Table 3-36) and body weights of 0.3-0.4 kg for rats and
60-70 kg for humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-100 DRAFT—DO NOT CITE OR QUOTE
-------
1 assumption for the AMetGSHBW34 dose metric is that equalizing the (whole body) rate of
2 production of GSH conjugation metabolites (i.e., systemic production of active moiety[ies]),
3 scaled by the % power of body weight, yields equivalent lifetime cancer risk across species.
4 Under "concentration equivalence dosimetry," the AMetGSHBW34 dose metric is consistent
5 with the assumptions that (1) the same average concentration of the (relatively) stable upstream
6 metabolites DCVG and (subsequently) DCVC in the kidney (the PBPK model assumes all
7 DCVG and DCVC produced translocates to the kidney) leads to the same lifetime cancer risk
8 across species; and (2) the rates of clearance of DCVG and (subsequently) DCVC scale by the
9 % power of body weight (as is assumed for enzyme activity or blood flow).
10 Another alternative dose metric is the total amount of TCE metabolism (oxidation and
11 GSH conjugation together) scaled by the % power of body weight (TotMetabBW34
12 [mg/kgy7week]). This dose metric uses the total flux of TCE metabolism as the lexicologically
13 relevant dose, and, thus, incorporates the possible involvement of oxidative metabolites, acting
14 either additively or interactively, in addition to GSH conjugation metabolites in
15 nephrocarcinogenicity (see Section 4.4.6). While there is no evidence that TCE oxidative
16 metabolites can on their own induce kidney cancer, some nephrotoxic effects attributable to
17 oxidative metabolites (e.g., peroxisome proliferation) may modulate the nephrocarcinogenic
18 potency of GSH metabolites. However, this dose metric is given less weight than those
19 involving GSH conjugation because, as discussed in Sections 4.4.6 and 4.4.7, the weight of
20 evidence supports the conclusion that GSH conjugation metabolites play a predominant role in
21 nephrocarcinogenicity. Under the "empirical dosimetry" approach, the underlying assumption
22 for the TotMetabBW34 dose metric is that equalizing the (whole body) rate of production of all
23 metabolites (i.e., systemic production and distribution of active moiety[ies]), scaled by the
24 3/4 power of body weight, yields equivalent lifetime cancer risk across species. Under
25 "concentration equivalence dosimetry," the TotMetabBW34 dose metric is consistent with the
26 assumptions that (1) the relative proportions and blood:tissue partitioning of the active
27 metabolites is similar across species; (2) the same average concentration of one or more active
28 metabolites in the kidney leads to a similar lifetime cancer risk across species; and (3) the rates
29 of clearance of active metabolites scale by the % power of body weight (e.g., as is assumed for
30 enzyme activity or blood flow).
31
32 5.2.1.2.1.2. Liver. As discussed in Section 4.5.6, there is substantial evidence that oxidative
33 metabolism is involved in TCE hepatocarcinogenicity, based primarily on noncancer and cancer
34 effects similar to those observed with TCE being observed with a number of oxidative
35 metabolites of TCE (e.g., CH, TCA, and DCA). While TCA is a stable, circulating metabolite,
36 CH and DCA are relatively short-lived, although enzymatically cleared (see Section 3.3.3.1). As
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-101 DRAFT—DO NOT CITE OR QUOTE
-------
1 discussed in Sections 4.5.6 and 4.5.7, there is now substantial evidence that TCA does not
2 adequately account for the hepatocarcinogenicity of TCE; therefore, unlike in previous dose-
3 response analyses (Rhomberg, 2000; Clewell and Andersen, 2004), the AUC of TCA in plasma
4 and in liver were not considered as dose metrics. However, there are inadequate data across
5 species to quantify the dosimetry of CH and DCA, and other intermediates of oxidative
6 metabolism (such as TCE-oxide or dichloroacetylchloride) also may be involved in
7 carcinogenicity. Thus, due to uncertainties as to the active moiety(ies), but the strong evidence
8 associating TCE liver effects with oxidative metabolism in the liver, hepatic oxidative
9 metabolism is the preferred basis for internal dose extrapolations of TCE-induced liver tumors.
10 Under "empirical dosimetry," the rate of hepatic oxidative metabolism would be scaled by body
11 weight to the % power. As discussed above, under "concentration equivalence dosimetry," when
12 the concentration of the active moiety cannot be estimated, qualitative data on the nature of
13 clearance of the active moiety or moieties can be used to inform whether to scale the rate of
14 metabolism by body weight to the 3/4 power or by the target tissue weight. However, several of
15 the oxidative metabolites are stable and systemically available, and several of those that are
16 cleared rapidly are metabolized enzymatically, so, according to the criteria discussed above,
17 there are insufficient data to support the conclusions that the active moiety or moieties do not
18 leave the target tissue in appreciable quantities and are cleared by mechanisms whose rates are
19 independent of body weight. Thus, scaling the rate of oxidative metabolism by body weight to
20 the 3/4 power would also be supported under "concentration equivalence dosimetry." Therefore,
21 the primary internal dose metric for TCE-induced liver tumors is selected to be the weekly rate
22 of hepatic oxidation per unit body weight to the % power (AMetLivlBW34 [mg/kg Vweek]). It
23 should be noted that due to the larger relative liver weight in mice as compared to humans,
24 scaling by liver weight instead of body weight to the % power would only change the
25 quantitative interspecies extrapolation by about 4-fold,25 so the sensitivity of the results to the
26 scaling choice is relatively modest.
27 To summarize, under the "empirical dosimetry" approach, the underlying assumption for
28 the AMetLivlBW34 dose metric is that equalizing the rate of hepatic oxidation of TCE (i.e.,
29 local production of active moiety(ies) in the target tissue), scaled by the % power of body weight,
30 yields equivalent lifetime cancer risk across species. Under "concentration equivalence
31 dosimetry," the AMetLivlBW34 dose metric is consistent with the assumptions that (1) the same
32 average concentrations of the active oxidative metabolites in the liver leads to a similar lifetime
25The range of the difference is 3.5-3.9-fold using the posterior medians for the relative liver weight in mice and
humans from the PBPK model described in Section 3.5 (see Table 3-36), and body weights of 0.03-0.04 kg for mice
and 60-70 kg for humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-102 DRAFT—DO NOT CITE OR QUOTE
-------
1 cancer risk across species; (2) active metabolites are primarily generated in situ in the liver; (3)
2 the relative proportions of the active oxidative metabolites are similar across species; and (4) the
3 rates of clearance of the active oxidative metabolites scale by the % power of body weight (e.g.,
4 enzyme-activity or blood-flow).
5 It is also known that the lung has substantial capacity for oxidative metabolism, with
6 some proportion of the oxidative metabolites produced there entering systemic circulation. Thus,
7 it is possible that extrahepatic oxidative metabolism can contribute to TCE
8 hepatocarcinogenicity. Therefore, the total amount of oxidative metabolism of TCE scaled by
9 the % power of body weight (TotOxMetabBW34 [mg/kgy7week]) was selected as an alternative
10 dose metric (the justification for the body weight to the 3/4 power scaling is analogous to that for
11 hepatic oxidative metabolism, above). Under the "empirical dosimetry" approach, the
12 underlying assumption for the TotOxMetabBW34 dose metric is that equalizing the rate of total
13 oxidation of TCE (i.e., systemic production of active moiety[ies]), scaled by the 3/4 power of
14 body weight, yields equivalent lifetime cancer risk across species. Under "concentration
15 equivalence dosimetry," this dose metric is consistent with the assumptions that (1) active
16 metabolites may be generated in situ in the liver or delivered to the liver via systemic circulation;
17 (2) the relative proportions and blood:tissue partitioning of the active oxidative metabolites are
18 similar across species; (3) the same average concentrations of the active oxidative metabolites in
19 the liver leads to a similar lifetime cancer risk across species; and (4) the rates of clearance of the
20 active oxidative metabolites scale by the % power of body weight (e.g., as is assumed for enzyme
21 activity or blood flow).
22
23 5.2.1.2.1.3. Lung. As discussed in Section 4.7.3, in situ oxidative metabolism in the
24 respiratory tract may be more important to lung toxicity than systemically delivered metabolites,
25 at least as evidenced by acute pulmonary toxicity. While chloral was originally implicated as the
26 active metabolite, based on either acute toxicity or mutagenicity of chloral and/or chloral
27 hydrate, more recent evidence suggests that other oxidative metabolites may also contribute to
28 lung toxicity. These data include the identification of dichloroacetyl lysine adducts in Clara cells
29 (Forkert et al., 2006), and the induction of pulmonary toxicity by TCE in CYP2El-null mice,
30 which may generate a different spectrum of oxidative metabolites as compared to wild-type mice
31 (respiratory tract tissue also contains P450s from the CYP2F family). Overall, the weight of
32 evidence supports the selection of respiratory tract oxidation of TCE as the preferred basis for
33 internal dose extrapolations of TCE-induced lung tumors. However, uncertainties remain as to
34 the relative contributions from different oxidative metabolites, and quantitative clearance data
35 necessary to calculate the concentration of each species are lacking.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-103 DRAFT—DO NOT CITE OR QUOTE
-------
1 Under "empirical dosimetry," the rate of respiratory tract oxidation would be scaled by
2 body weight to the % power. As discussed above, under "concentration equivalence dosimetry,"
3 when the concentration of the active moiety cannot be estimated, qualitative data on the nature of
4 clearance of the active moiety or moieties can be used to inform whether to scale the rate of
5 metabolism by body weight to the % power or by the target tissue weight. For chloral, as
6 discussed in Section 4.7.3, the reporting of substantial TCOH but no detectable chloral hydrate in
7 blood following TCE exposure from experiments in isolated, perfused lungs (Dalby and
8 Bingham, 1978) support the conclusion that chloral does not leave the target tissue in substantial
9 quantities, but that there is substantial clearance by enzyme-mediated biotransformation.
10 Dichloroacetyl chloride is a relatively-short-lived intermediate from aqueous (nonenzymatic)
11 decomposition of TCE-oxide that can be trapped with lysine or degrade further to form DC A,
12 among other products (Cai and Guengerich, 1999). Cai and Guengerich (1999) reported a half-
13 life of TCE-oxide under aqueous conditions of 12 s at 23 °C, a time-scale that would be shorter at
14 physiological conditions (37°C) and that includes formation of dichloroacetyl chloride as well as
15 its decomposition. Therefore, evidence for this metabolite suggests its clearance both is
16 sufficiently rapid so that it would remain at the site of formation and is nonenzymatically
17 mediated so that its rate would be independent of body weight. Other oxidative metabolites may
18 also play a role, but, because they have not been identified, no inferences can be made as to their
19 clearance.
20 Therefore, because it is not clear what the contributions to TCE-induced lung tumors are
21 from different oxidative metabolites produced in situ and, even under "concentration equivalence
22 dosimetry," the scaling by body weight to the 3/4 power is supported for at least one of the
23 possible active moieties, it was decided here to scale the rate of respiratory tract tissue oxidation
24 of TCE by body weight to the % power. The primary internal dose metric for TCE-induced lung
25 tumors is, thus, the weekly rate of respiratory tract oxidation per unit body weight to the 3/4 power
26 (AMetLngBW34 [mg/kgy7week]). It should be noted that, due to the larger relative respiratory
27 tract tissue weight in mice as compared to humans, scaling by tissue weight instead of body
28 weight to the 3/4 power would change the quantitative interspecies extrapolation by less than
29 2-fold,26 so the sensitivity of the results to the scaling choice is relatively small.
30 To summarize, under the "empirical dosimetry" approach, the underlying assumption for
31 the AMetLngBW34 dose metric is that equalizing the rate of respiratory tract oxidation of TCE
32 (i.e., local production of active moiety(ies) in the target tissue), scaled by the % power of body
26The range of the difference is 1.6-1.8-fold using the posterior medians for the relative respiratory tract tissue
weight in mice and humans from the PBPK model described in Section 3.5 (see Table 3-36), and body weights of
0.03-0.04 kg for mice and 60-70 kg for humans.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-104 DRAFT—DO NOT CITE OR QUOTE
-------
1 weight, yields equivalent lifetime cancer risk across species. Under "concentration equivalence
2 dosimetry," the use of the AMetLngBW34 dose metric is consistent with the assumptions that
3 (1) the proportion of respiratory tract oxidative metabolism to active metabolites are similar
4 across species (2) the same average concentration of the active moiety(ies) in the metabolizing
5 respiratory tract tissue leads to a similar lifetime cancer risk across species; and (3) the rates of
6 clearance of these reactive species scale by the % power of body weight (e.g., enzyme-activity or
7 blood-flow).
8 While there is substantial evidence that acute pulmonary toxicity is related to pulmonary
9 oxidative metabolism, for carcinogenicity, it is possible that, in addition to locally produced
10 metabolites, systemically-delivered oxidative metabolites also play a role. Therefore, total
11 oxidative metabolism scaled by the % power of body weight (TotOxMetabBW34
12 [mg/kgy7week]) was selected as an alternative dose metric (the justification for the body weight
13 to the 3/4 power scaling is analogous to that for respiratory tract oxidative metabolism, above).
14 Under the "empirical dosimetry" approach, the underlying assumption for the
15 TotOxMetabBW34 dose metric is that equalizing the rate of total oxidation of TCE (i.e.,
16 systemic production of oxidative metabolites), scaled by the 3/4 power of body weight, yields
17 equivalent lifetime cancer risk across species. Under "concentration equivalence dosimetry,"
18 this dose metric is consistent with the assumptions that (1) active oxidative metabolites may be
19 generated in situ in the lung or delivered to the lung via systemic circulation; (2) the relative
20 proportions and blood:tissue partitioning of the active oxidative metabolites are similar across
21 species; (3) the same average concentrations of the active oxidative metabolites in the lung leads
22 to a similar lifetime cancer risk across species; and (4) the rates of clearance of the active
23 oxidative metabolites scale by the % power of body weight (e.g., as is assumed for enzyme
24 activity or blood flow).
25 Another alternative dose metric considered here is the AUC of TCE in blood (AUCCBld
26 [mg-hour/L/week]). Under either the "empirical dosimetry" or "concentration equivalence"
27 approach, this dose metric would account for the possibility that local metabolism is determined
28 primarily by TCE delivered in blood via systemic circulation to pulmonary tissue (the flow rate
29 of which scales as body weight to the % power), as assumed in previous PBPK models, rather
30 than TCE delivered in air via diffusion to the respiratory tract, as is assumed in the PBPK model
31 described in Section 3.5. However, as discussed in Section 3.5 and Appendix A, the available
32 pharmacokinetic data provide greater support for the updated model structure. Under
33 "concentration equivalence dosimetry," this dose metric also accounts for the possible role of
34 TCE itself in pulmonary carcinogenicity (consistent with the assumption that the same average
35 concentration of TCE in blood will lead to a similar lifetime cancer risk across species).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-105 DRAFT—DO NOT CITE OR QUOTE
-------
1 5.2.1.2.1.4. Other sites. For all other sites listed in Table 5-27, there is insufficient information
2 for site-specific determinations of appropriate dose metrics. While TCE metabolites and/or
3 metabolizing enzymes have been reported in some of these tissues (e.g., male reproductive tract),
4 their roles in carcinogenicity for these specific sites have not been established. Although
5 "primary" and "alternative" dose metrics are defined, they do not differ appreciably in their
6 degrees of plausibility.
7 Given that the majority of the toxic and carcinogenic responses to TCE appear to be
8 associated with metabolism, total metabolism of TCE scaled by the % power of body weight was
9 selected as the primary dose metric (TotMetabBW34 [mg/kgy7week]). This dose metric uses
10 the total flux of TCE metabolism as the toxicologically-relevant dose, and, thus, incorporates the
11 possible involvement of any TCE metabolite in carcinogenicity. Under the "empirical
12 dosimetry" approach, the underlying assumption for the TotMetabBW34 dose metric is that
13 equalizing the (whole body) rate of production of all metabolites (i.e., systemic production of
14 active moiety[ies]), scaled by the % power of body weight, yields equivalent lifetime cancer risk
15 across species. Under "concentration equivalence dosimetry," the TotMetabBW34 dose metric
16 is consistent with the assumptions that (1) active metabolites are delivered to the target tissue via
17 systemic circulation; (2) the relative proportions and blood:tissue partitioning of the active
18 metabolites is similar across species; (3) the same average concentrations of the active
19 metabolites in the target tissue leads to a similar lifetime cancer risk across species; and (4) the
20 rates of clearance of the active metabolites scale by the % power of body weight (e.g., as is
21 assumed for enzyme activity or blood flow).
22 An alternative dose metric considered here is the AUC of TCE in blood. Under either the
23 "empirical dosimetry" or "concentration equivalence" approach, this dose metric would account
24 for the possibility that the determinant of carcinogenicity is local metabolism, governed
25 primarily by TCE delivered in blood via systemic circulation to the target tissue (the flow rate of
26 which scales as body weight to the % power). Under "concentration equivalence dosimetry,"
27 this dose metric also accounts for the possible role of TCE itself in carcinogenicity (consistent
28 with the assumption that the same average concentration of TCE in blood will lead to a similar
29 lifetime cancer risk across species).
30
31 5.2.1.2.2. Methods for dose-response analyses using internal dose metrics. As shown in
32 Figure 5-5, the general approach taken for the use of internal dose metrics in dose-response
33 modeling was to first apply the rodent PBPK model to obtain rodent values for the dose metrics
34 corresponding to the applied doses in a bioassay. Then, dose-response modeling for a tumor
35 response was performed using the internal dose metrics and the multistage model or the survival -
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-106 DRAFT—DO NOT CITE OR QUOTE
-------
1 adjusted modeling approaches described above to obtain a BMD and BMDL in terms of the dose
2 metric. On an internal dose basis, humans and rodents are presumed to have similar lifetime
3 cancer risks, and the relationship between human internal and external doses is essentially linear
4 at low doses up to 0.1 mg/kg/d or 0.1 ppm, and nearly linear up to 10 mg/kg/d or 10 ppm.
5 Therefore, the BMD and BMDL were then converted human equivalent doses (or exposures)
6 using conversion ratios estimated from the human PBPK model at 0.001 mg/kg/d or 0.001 ppm
7 (see Table 5-28). Because the male and female conversions differed by less than 11%, the
8 human BMDLs were derived using the mean of the sex-specific conversion factors (except for
9 testicular tumors, for which only male conversion factors were used). Finally, a unit risk
10 estimate for that tumor response was derived from the human "BMDLs" as described above (i.e.,
11 BMR/BMDL). Note that the converted "BMDs" and "BMDLs" are not actually human
12 equivalent BMDs and BMDLs corresponding to the BMR because the conversion was not made
13 in the dose range of the BMD; the converted BMDs and BMDLs are merely intermediaries to
14 obtain a converted unit risk estimate. In addition, it should be noted that median values of dose
15 metrics were used for rodents, whereas mean values were used for humans. Because the rodent
16 population model characterizes study-to-study variation, animals of the same sex/species/strain
17 combination within a study were assumed to be identical. Therefore, use of median dose metric
18 values for rodents can be interpreted as assuming that the animals in the bioassay were all
19 "typical" animals and the dose-response model is estimating a "risk to the typical rodent." In
20 practice, the use of median or mean internal doses for rodents did not make much difference
21 except when the uncertainty in the dose metric was high (e.g., AMetLungBW34 dose metric in
22 mice). A quantitative analysis of the impact of the uncertainty in the rodent PBPK dose metrics
23 is included in Section 5.2.1.4.2. On the other hand, the human population model characterizes
24 individual-to-individual variation. Because the quantity of interest is the human population
25 mean risk, the expected value (averaging over the uncertainty) of the population mean (averaging
26 over the variability) dose metric was used for the conversion to human unit risks. Therefore, the
27 extrapolated unit risk estimates can be interpreted as the expected "average risk" across the
28 population based on rodent bioassays.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-107 DRAFT—DO NOT CITE OR QUOTE
-------
•distribution
[distribution (combined
ncertainty and variability)
(distribution
distribution (separate
ncertainty and variability)
Site-specific
cancer unit
risk=BMR/
BMDL
(per internal
dose unit)
population
irnean
i
irnean
Human site-
specific cancer
unit risk
(per ppm or
per mg/kg-d)
2
3
4
5
Figure 5-5. Flow-chart for dose-response analyses of rodent bioassays using
PBPK model-based dose metrics. Square nodes indicate point values, circular
nodes indicate distributions, and the inverted triangles indicate a (deterministic)
functional relationship.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
5-108
DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table 5-28. Mean PBPK model predictions for weekly internal dose in
humans exposed continuously to low levels of TCE via inhalation (ppm) or
orally (mg/kg/d)
Dose metric
ABioactDCVCBW34
AMetGSHBW34
AMetLivlBW34
AMetLngBW34
AUCCBld
TotMetabBW34
TotOxMetabBW34
0.001 ppm
Female
0.00324
0.00200
0.00703
0.00281
0.00288
0.0118
0.00984
Male
0.00324
0.00200
0.00683
0.00287
0.00298
0.0117
0.00970
0.001 mg/kg/d
Female
0.00493
0.00304
0.0157
6.60xlO'5
0.000411
0.0188
0.0157
Male
0.00515
0.00318
0.0164
6.08xlO'5
0.000372
0.0196
0.0164
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
See note to Table 5-27 for dose metric abbreviations. Values represent the mean of the (uncertainty) distribution of
population means for each sex and exposure scenario, generated from Monte Carlo simulation of 500 populations of
500 individuals each.
5.2.1.3. Rodent Dose-Response Analyses: Results
A summary of the PODs and unit risk estimates for each sex/species/bioassay/tumor type
is presented in Tables 5-29 (inhalation studies) and 5-30 (oral studies). The PODs for individual
tumor types were extracted from the modeling results in the figures in Appendix G. For the
applied dose (default dosimetry) analyses, the POD is the BMDL from the male human ("M")
BMDL entry at the top of the figure for the selected model; male results were extracted because
the default weight for males in the PBPK modeling is 70 kg, which is the overall human weight
in U.S. EPA's default dosimetry methods (for inhalation, male and female results are identical).
As described in Section 5.2.1.2 above, for internal dose metrics, male and female results were
averaged, and the converted human "BMDLs" are not true BMDLs because they were converted
outside the linear range of the PBPK models. It can be seen in Appendix G that the male and
female results were similar for all the dose metrics.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-109 DRAFT—DO NOT CITE OR QUOTE
-------
Table 5-29. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor type (inhalation)
to
Study
Tumor type
BMR
PODs (ppm, in human equivalent exposures)3
Applied
dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Female mouse
Fukuda
Henschler
Maltoni
Lung AD + CARC
Lymphoma
Lung AD + CARC
Liver
Combined
0.1
0.1
0.1
0.05
0.05
26.3
11. Ob
44.6
37.1
15.7
55.5
b
96.6
9.84
31.3
51.4
45.8
20.7
38.8
55.7
41.9
Male mouse
Maltoni
Liver
0.1
O A O
34.3
51
37.9
Male rat
Maltoni
Study
Leukemia
Kidney AD + CARC
Leydig cell
Combined
Tumor type
0.05
0.01
0.1
0.01
28.2C
22.7
18.6C
1.44
b
d
28.3
13.7
18.1
1.37
0.197
0.121
Unit risk estimate (ppm 1)e
Applied dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Female mouse
Fukuda
Henschler
Maltoni
Lung AD + CARC
Lymphoma
Lung AD + CARC
Liver
Combined
3.8 x 10'3
9.1 x 10'3
2.2 x 10'3
1.3 x lO'3
3.2 x ID'3
1.8 x 10'3
1.0 x 10'3
1.0 x 10 2
3.2 x 10'3
1.9 x 10'3
1.1 x lO'3
2.4 x 10 3
2.6 x 10 3
1.8 x 10 3
1.2 x 10 3
Male mouse
Maltoni
Liver
2.9 x lO'3
2.0 x lO'3
2.6 x 10 3
§
i
TO'
I
TO
H I
O ^
HH Oq
H TO
O
H
W
-------
Table 5-29. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor type (inhalation)
(continued)
Study
Tumor type
Unit risk estimate (ppm 1)e
Applied dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Male rat
Maltoni
Leukemia
Kidney AD + CARC
Leydig cell
Combined
1.8 x ID'3
4.4 x ID'4
5.4 x ID'3
7.0 x ID'3
1.8 x 10 3
7.3 x ID'4
5.5 x 10 3
7.3 x ID'3
5.1 x 1Q-2
8.3 x 10 2
Tor the applied doses, the PODs are BMDLs. However, for the internal dose metrics, the PODs are not actually human equivalent BMDLs corresponding to the
BMR because the interspecies conversion does not apply to the dose range of the BMDL; the converted BMDLs are merely intermediaries to obtain a converted
unit risk estimate. The calculation that was done is equivalent to using linear extrapolation from the BMDLs in terms of the internal dose metric to get a unit
risk estimate for low-dose risk in terms of the internal dose metric and then converting that estimate to a unit risk estimate in terms of human equivalent
exposures. The PODs reported here are what one would get if one then used the unit risk estimate to calculate the human exposure level corresponding to a 10%
extra risk, but the unit risk estimate is not intended to be extrapolated upward out of the low-dose range, e.g., above 10~4 risk. In addition, for the internal dose
metrics, the PODs are the average of the male and female human "BMDL" results presented in Appendix G.
blnadequate fit to control group, but the primary metric, TotMetabBW34, fits adequately.
Dropped highest-dose group to improve model fit.
Inadequate overall fit.
eUnit risk estimate = BMR/POD. Results for the primary dose metric are in bold.
AD = adenoma, CARC = carcinoma.
-------
Table 5-30. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor type (oral)
to
Study
Tumor type
BMR
PODs (mg/kg/d, in human equivalent doses)3
Applied
dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Female mouse
NCI
Liver care
Lung AD + CARC
Leukemias + sarcomas
Combined
0.1
0.1
0.1
0.05
26.5
41.1
43.1
7.43
682
733
20.6
17.6
24.7
5.38
757
14.1
Male mouse
NCI
Liver care
0.1
8.23
4.34
3.45
Female rat
NTP, 1988
Leukemia
0.05
72.3
3,220
21.7
Male rat
NTP, 1990C
NTP, 1988
Marshall
August
Osborne-Mendel0
Study
Kidney AD + CARC
Testicular
Subcut sarcoma
Kidney AD + CARC
Tumor type
0.1
0.1
0.05
0.1
32
3.95
60.2
41.5
167
2,560
11.5
1.41
21.5
14.3
0.471
0.648
0.292
0.402
Unit risk estimate (mg/kg/d) 1)b
Applied dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Female mouse
NCI
Liver care
Lung AD + CARC
Leukemias + sarcomas
Combined
3.8 x 1(T3
2.4 x 1(T3
2.3 x lO'3
6.7 x ID'3
1.5 x lO'4
1.4 x 10"4
4.9 x 10 3
5.7 x 1(T3
4.0 x lO'3
9.3 x 10 3
1.3 x 10 4
7.1 x 10 3
Male mouse
NCI
Liver care
1.2 x lO'2
2.3 x ID'2
2.9 x 10 2
§
i
TO'
I
H I
O ^
HH Oq
H ^
O
H
W
-------
Table 5-30. Summary of PODs and unit risk estimates for each sex/species/bioassay/tumor type (oral)
(continued)
Study
Tumor type
Unit risk estimate (mg/kg/d) 1)b
Applied dose
AUC
CBld
TotMetab
BW34
TotOxMetab
BW34
AMetLng
BW34
AMetLivl
BW34
AMetGSH
BW34
ABioact
DCVCBW34
Female rat
NTP, 1988
Leukemia
6.9 x 1Q-4
1.6 x ID'5
2.3 x 10 3
Male rat
NTP, 1990C
NTP, 1988
Marshall"1
August
Osborne-Mendel0
Kidney AD + CARC
Testicular
Subcut sarcoma
Kidney AD + CARC
1.6 x ID'3
2.5 x ID'2
8.3 x ID'4
2.4 x ID'3
6.0 x ID'4
2.0 x ID'5
4.3 x ID'3
7.1 x 10 2
2.3 x 10 3
7.0 x ID'3
1.1 x ID'1
1.5 x ID'1
1.7 x 10 *
2.5 x 10 *
Tor the applied doses, the PODs are BMDLs. However, for the internal dose metrics, the PODs are not actually human equivalent BMDLs corresponding to the
BMR because the interspecies conversion does not apply to the dose range of the BMDL; the converted BMDLs are merely intermediaries to obtain a converted
unit risk estimate. The calculation that was done is equivalent to using linear extrapolation from the BMDLs in terms of the internal dose metric to get a unit
risk estimate for low-dose risk in terms of the internal dose metric and then converting that estimate to a unit risk (slope factor) estimate in terms of human
equivalent doses. The PODs reported here are what one would get if one then used the unit risk estimate to calculate the human dose level corresponding to a
10% extra risk, but the unit risk estimate is not intended to be extrapolated upward out of the low-dose range, e.g., above 10"4 risk. In addition, for the internal
dose metrics, the PODs are the average of the male and female human "BMDL" results presented in Appendix G.
bUnit risk estimate = BMR/POD. Results for the primary dose metric are in bold.
°Using MSW adjusted incidences (see text and Table 5-31).
dUsing poly-3 adjusted incidences (see text and Table 5-31).
AD = adenoma, CARC = carcinoma.
-------
1 For two data sets, the highest dose (exposure) group was dropped to get a better fit when
2 using applied doses. This technique can improve the fit when the response tends to plateau with
3 increasing dose. Plateauing typically occurs when metabolic saturation alters the pattern of
4 metabolite formation or when survival is impacted at higher doses, and it is assumed that these
5 high-dose responses are less relevant to low-dose risk. The highest-dose group was not dropped
6 to improve the fit for any of the internal dose metrics because it was felt that if the dose metric
7 was an appropriate reflection of internal dose of the reactive metabolite(s), then use of the dose
8 metric should have ameliorated the plateauing in the dose-response relationship (note that
9 survival-impacted data sets were addressed using survival adjustment techniques). For a 3rd data
10 set (Henschler lymphomas), it might have helped to drop the highest exposure group, but there
11 were only two exposure groups, so this was not done. As a result, the selected model, although it
12 had an adequate fit overall, did not fit the control group very well (the model estimated a higher
13 background response than was observed); thus, the BMD and BMDL were likely overestimated
14 and the risk underestimated. The estimates from the NCI (1976) oral male mouse liver cancer
15 data set are also somewhat more uncertain because the response rate was extrapolated down from
16 a response rate of about 50% extra risk to the BMR of 10% extra risk.
17 Some general patterns can be observed in Tables 5-29 and 5-30. For inhalation, the unit
18 risk estimates for different dose metrics were generally similar (within about 2.5-fold) for most
19 tumor types. The exception was for kidney cancer, where the estimates varied by over 2 orders
20 of magnitude, with the AMetGSHBW34 and ABioactDC VCBW34 metrics yielding the highest
21 estimates. This occurs because pharmacokinetic data indicate, and the PBPK model predicts,
22 substantially more GSH conjugation (as a fraction of intake), and hence subsequent
23 bioactivation, in humans relative to rats. The range of the risk estimates for individual tumor
24 types overall (across tumor types and dose metrics) was encompassed by the range of estimates
25 across the dose metrics for kidney cancer in the male rat, which was from 4.4 x 10"4 per ppm
26 (applied dose) to 8.3 x 10'2 per ppm (ABioactDCVCBW34).
27 For oral exposure, the unit risk (slope factor) estimates are more variable across dose
28 metrics because of first-pass effects in the liver (median estimates for the fraction of TCE
29 metabolized in one pass through the liver in mice, rats, and humans are >0.8). Here, the
30 exception is for the risk estimates for cancer of the liver itself, which are also within about a
31 2.5-fold range, because the liver gets the full dose of all the metrics during that "first pass." For
32 the other tumor types, the range of estimates across dose metrics varies from about 30-fold to
33 over 2 orders of magnitude, with the estimates based on AUCCBld and AMetLngBW34 being at
34 the low end and those based on AMetGSHBW34 and ABioactDCVCBW34 again being at the
35 high end. For AUCCBld, the PBPK model predicted the blood concentrations to scale more
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-114 DRAFT—DO NOT CITE OR QUOTE
-------
1 closely to body weight rather than the 3/4 power of body weight, so the extrapolated human unit
2 risks using this dose metric are smaller than those obtained by applied dose or other dose metrics
3 that included % power body weight scaling. For AMetLngBW34, pharmacokinetic data indicate,
4 and the PBPK model predicts, that the human respiratory tract metabolizes a lower fraction of
5 total TCE intake than the mouse respiratory tract, so the extrapolated risk to humans based on
6 this metric is lower than that obtained using applied dose or other dose metrics. Overall, the oral
7 unit risk estimates for individual tumor types ranged from 1.6 x 10"5 per mg/kg/d (female rat
8 leukemia, AUCCBld) to 2.5 x 10"1 per mg/kg/d (male Osborne-Mendel rat kidney,
9 ABioactDCVCBW34), a range of over 4 orders of magnitude. It must be recognized, however,
10 that not all dose metrics are equally credible, and, as will be presented below, the unit risk
11 estimates for total cancer risk for the most sensitive bioassay response for each sex/species
12 combination using the primary (preferred) dose metrics fall within a very narrow range.
13 Results for survival-adjusted analyses are summarized in Table 5-31. For the time-
14 independent (BMDS) multistage model, the risk estimates using poly-3 adjustment are higher
15 than those without poly-3 adjustment. This is to be expected because the poly-3 adjustment
16 decreases denominators when accounting for early mortality, and, for these data sets, the higher-
17 dose groups had greater early mortality. The difference was fairly modest for the kidney cancer
18 data sets (about 30% higher) but somewhat larger for the testicular cancer data set (about 150%
19 higher).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-115 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table 5-31. Comparison of survival-adjusted results for 3 oral male rat data
sets*
Dose metric
Adjustment
method
BMR
POD
(mg/kg/d)
BMD:BMDL
Unit risk estimate
(per mg/kg/d)
NTP, 1990 F344 rat kidney AD + CARC
Applied dose
TotMetabBW34
AMetGSHBW34
ABioactDCVCBW34
unadj BMDS
poly -3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
0.05
0.1
0.05
0.05
0.1
0.05
0.05
0.1
0.05
0.05
0.1
0.05
56.9
89.2
32.0
20.2
31.8
11.5
0.841
1.32
0.471
0.522
0.817
0.292
1.9
1.9
2.6
2.1
1.7
3.1
1.9
1.9
2.4
1.9
1.9
2.4
8.8 x 10'4
1.1 x 10'3
1.6 x 10'3
2.5 x 10'3
3.1 x 1Q-3
4.3 x 1Q-3
5.9 x 1Q-2
7.6 x 1Q-2
1.1 x 1Q-1
9.6 x lO'2
1.2 x 10'1
1.7 x lO'1
NTP, 1988 Osborne-Mendel rat kidney AD + CARC
Applied dose
TotMetabBW34
AMetGSHBW34
ABioactDCVCBW34
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
86.6
65.9
41.5
30.4
23.1
14.3
1.35
1.03
0.648
0.835
0.636
0.402
1.7
1.7
2.0
1.7
1.7
2.0
1.7
1.7
2.0
1.7
1.7
2.0
1.2 x 1Q-3
1.5 x 1Q-3
2.4 x 1Q-3
3.3 x 1Q-3
4.3 x 1Q-3
7.0 x lO'3
7.4 x 10'2
9.7 x 10'2
1.5 x 10'1
1.2 x 10'1
1.6 x 1Q-1
2.5 x 10 *
NTP, 1988 Marshall rat testicular tumors
Applied dose
AUCCBld
TotMetabBW34
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
unadj BMDS
poly-3 BMDS
MSW
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
9.94
3.95
1.64
427
167
60.4
3.53
1.41
0.73
1.4
1.5
5.2
1.4
1.6
2.6
4.3
1.5
9.4
1.0 x 1Q-2
2.5 x 1Q-2
6.1 x lO'2
2.3 x 10'4
6.0 x 10'4
1.7 x 10'3
2.8 x 10'2
7.1 x 10 2
1.4 x 1Q-1
4
5
6
7
8
9
10
*For the applied doses, the PODs are BMDLs. However, for the internal dose metrics, the PODs are not actually
human equivalent BMDLs corresponding to the BMR because the interspecies conversion does not apply to the
dose range of the BMDL; the converted BMDLs are merely intermediaries to obtain a converted unit risk estimate.
Results for the primary dose metric are in bold.
AD = adenoma, CARC = carcinoma.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-116 DRAFT—DO NOT CITE OR QUOTE
-------
1 In addition, the MSW time-to-tumor model generated higher risk estimates than the poly-
2 3 adjustment technique. The MSW results were about 40% higher for the NTP F344 rat kidney
3 cancer data sets and about 60% higher for the NTP Osborne-Mendel rat kidney cancer data sets.
4 For the NTP Marshall rat testicular cancer data set, the discrepancies were greater; the results
5 ranged from about 100% to 180% higher for the different dose metrics. As discussed in
6 Section 5.2.1.1, these two approaches differ in the way they take early mortality into account.
7 The poly-3 technique merely adjusts the tumor incidence denominators, using a constant power 3
8 of time, to reflect the fact that animals are at greater risk of cancer at older ages. The MSW
9 model estimates risk as a function of time (and dose), and it estimates the power (of time)
10 parameter for each data set.27 For the NTP F344 rat kidney cancer and NTP Marshall rat
11 testicular cancer data sets, the estimated power parameter was close to 3 in each case, ranging
12 from 3.0 to 3.7; for the NTP Osborne-Mendel rat kidney cancer data sets, however, the estimated
13 power parameter was about 10 for each of the dose metrics, presumably reflecting the fact that
14 these were late-occurring tumors (the earliest occurred at 92 weeks). Using a higher power
15 parameter than 3 in the poly-3 adjustment would give even less weight to nontumor-bearing
16 animals that die early and would, thus, increase the adjusted incidence even more in the highest-
17 dose groups where the early mortality is most pronounced, increasing the unit risk estimate.
18 Nonetheless, as noted above, the MSW results were only about 60% higher for the NTP
19 Osborne-Mendel rat kidney cancer data sets for which MSW estimated a power parameter of
20 about 10.
21 In general, the risk estimates from the MSW model would be preferred because, as
22 discussed above, this model incorporates more information (e.g., tumor context) and estimates
23 the power parameter rather than using a constant value of three. From Table 5-31, it can be seen
24 that the results from MSW yielded higher BMD:BMDL ratios than the results from the poly-3
25 technique. These ratios were only slightly higher and not unusually large for MSW model
26 analyses of the NTP (1988, 1990) kidney tumor estimates, and this, along with the adequate fit
27 (assessed visually) of the MSW model, supports using the unit risk estimates from the MSW
28 modeling of rat kidney tumor incidence. On the other hand, the BMD:BMDL ratio was
29 relatively large for the applied dose analysis and, in particular, for the preferred dose metric
30 analysis (9.4-fold) of the NTP Marshall rat testicular tumor data set. Therefore, for this
31 endpoint, the poly-3-adjusted results were used, although they may underestimate risk somewhat
32 as compared to the MSW model.
27Conceptually, the approaches differ most when different tumor contexts (incidental or fatal) are considered,
because the poly-3 technique only accounts for time of death, while the MSW model can account for the tumor
context and attempt to estimate an induction time (to), although this was not done for any of the datasets in this
assessment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-117 DRAFT—DO NOT CITE OR QUOTE
-------
1 In addition to the results from dose-response modeling of individual tumor types, the
2 results of the combined tumor risk analyses for the three bioassays in which the rodents exhibited
3 increased risks at multiple sites are also presented in Tables 5-29 and 5-30, in the rows labeled
4 "combined" under the column heading "Tumor Type." These results were extracted from the
5 detailed results in Appendix G. Note that, because of the computational complexity of the
6 combined tumor analyses, dose-response modeling was only done using applied dose and a
7 common upstream internal dose metric, rather than using the different preferred dose metrics for
8 each tumor type within a combined tumor analysis.
9 For the Maltoni female mouse inhalation bioassay, the combined tumor risk estimates are
10 bounded by the highest individual tumor risk estimates and the sums of the individual tumor
11 risks estimates (the risk estimates are upper bounds, so the combined risk estimate, i.e., the upper
12 bound on the sum of the individual central tendency estimates, should be less than the sum of the
13 individual upper bound estimates), as one would expect. The common upstream internal dose
14 metric used for the combined analysis was TotOxMetabBW34, which is not the primary metric
15 for either of the individual tumor types. For the liver tumors, the primary metric was
16 AMetLivlBW34, but as can be seen in Table 5-29, it yields results similar to those for
17 TotOxMetabBW34. Likewise, for the lung tumors, the primary metric was AMetLngBW34,
18 which yields a unit risk estimate slightly smaller that for TotOxMetabBW34. Thus, the results of
19 the combined analysis using TotOxMetabBW34 as a common metric is not likely to substantially
20 over- or underestimate the combined risk based on preferred metrics for each of the tumor types.
21 For the Maltoni male rat inhalation bioassay, the combined risk estimates are also
22 reasonably bounded, as expected. The common upstream internal dose metric used for the
23 combined analysis was TotMetabBW34, which is the primary metric for two of the three
24 individual tumor types. However, as can be seen in Table 5-29, the risk estimate for the
25 preferred dose metric for the third tumor type, ABioactDCVCBW34 for the kidney tumors, is
26 substantially higher than the risk estimates for the primary dose metrics for the other two tumor
27 types and would dominate a combined tumor risk estimate across primary dose metrics; thus, the
28 ABioactDCVCBW34-based kidney tumor risk estimate alone can reasonably be used to
29 represent the total cancer risk for the bioassay using preferred internal dose metrics, although it
30 would underestimate the combined risk to some extent (e.g., the kidney-based estimate is
31 8.3 x 10"2 per ppm; the combined estimate would be about 9 x 10"2 per ppm, rounded to one
32 significant figure).
33 For the third bioassay (NCI female mouse oral bioassay), the combined tumor risk
34 estimates are once again reasonably bounded. The common upstream internal dose metric used
35 for the combined analysis was TotOxMetabBW34, which is not the primary metric for any of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-118 DRAFT—DO NOT CITE OR QUOTE
-------
1 three individual tumor types but was considered to be the most suitable metric to apply as a basis
2 for combining risk across these different tumor types. The unit risk estimate for the lung based
3 on the primary dose metric for that site becomes negligible compared to the estimates for the
4 other two tumor types (see Table 5-30). However, the unit risk estimates for the remaining two
5 tumor types are both somewhat underestimated using the TotOxMetabBW34 metric rather than
6 the primary metrics for those tumors (the TotOxMetabBW34-based estimate for leukemias +
7 sarcomas, which is not presented in Table 5-30 because, in the absence of better mechanistic
8 information, more upstream metrics were used for that individual tumor type, is 4.1 x 10"3 per
9 mg/kg/d). Thus, overall, the combined estimate based on TotOxMetabBW34 is probably a
10 reasonable estimate for the total tumor risk in this bioassay, although it might overestimate risk
11 slightly.
12 The most sensitive sex/species results are extracted from Tables 5-29 and 5-30 and
13 presented in Tables 5-32 (inhalation) and 5-33 (oral) below. The BMD:BMDL ratios for all the
14 results corresponding to the unit risk estimates based on the preferred dose metrics ranged from
15 1.3-2.1. For inhalation, the most sensitive bioassay responses based on the preferred dose
16 metrics ranged from 2.6 x 10~3 per ppm to 8.3 x 10~2 per ppm across the sex/species
17 combinations (with the exception of the female rat, which exhibited no apparent TCE-associated
18 response in the 3 available bioassays). For oral exposure, the most sensitive bioassay responses
19 based on the preferred dose metrics ranged from 2.3 x 10~3 per mg/kg/d to 2.5 x 10"1 per
20 mg/kg/d across the sex/species combinations. For both routes of exposure, the most sensitive
21 sex/species response was (or was dominated by, in the case of the combined tumors in the male
22 rat by inhalation) male rat kidney cancer based on the preferred dose metric of
23 ABioactDCVCBW34.
24
25 5.2.1.4. Uncertainties in Dose-Response Analyses of Rodent Bioassays
26 5.2.1.4.1. Qualitative discussion of uncertainties. All risk assessments involve uncertainty, as
27 study data are extrapolated to make inferences about potential effects in humans from
28 environmental exposure. The largest sources of uncertainty in the TCE rodent-based cancer risk
29 estimates are interspecies extrapolation and low-dose extrapolation. Some limited human
30 (occupational) data from which to estimate human cancer risk are available, and cancer risk
31 estimates based on these data are developed in Section 5.2.2 below. In addition, some
32 quantitative uncertainty analyses of the interspecies differences in pharmacokinetics were
33 conducted and are presented in Section 5.2.1.4.2.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-119 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table 5-32. Inhalation: most sensitive bioassay for each sex/species
combination*
Sex/species
Female
mouse
Male mouse
Female rat
Male rat
Endpoint
(study)
Lymphoma
(Henschler et al., 1980)
Liver hepatoma
(Maltoni et al., 1986)
-
Leukemia+
Kidney AD & CARC+
Leydig cell tumors
(Maltoni et al., 1986)
Unit risk per ppm
Preferred
dose metric
l.Ox 1(T2
2.6 x 1(T3
-
8.3 x 1(T2
Default
methodology
9.1 x 1(T3
2.9 x 1(T3
-
7.0 x 10~3
Alternative dose
metrics, studies,
or endpoints
1 x 10~3~4x 10~3
2 x 10~3
-
4x 10~4~5 x 10~2
[individual site
results]
4
5
6
7
8
9
10
11
*Results extracted from Table 5-29.
AD = adenoma, CARC = carcinoma.
Table 5-33. Oral: most sensitive bioassay for each sex/species combination3
Sex/species
Female
mouse
Male mouse
Female rat
Male rat
Endpoint
(Study)
Liver CARC +
lung AD & CARC+
sarcomas + leukemias
(NCI, 1976)
Liver CARC
(NCI, 1976)
Leukemia
(NTP, 1988)
Kidney AD + CARC
(NTP, 1988, Osborne-
Mendel)
Unit risk per mg/kg/d
Preferred
dose metric
9.3 x 10~3
2.9 x 10~2
2.3 x 10~3
2.5 x 10"1
Default
methodology
6.7 x 10~3
1.2x 10~2
6.9 x 10^
2.4 x 10~3b
Alternative dose
metrics, studies,
or endpoints
1 x 10~4~7x 10~3
[individual site
results]
2 x 10~2
2 x 10~5
2x 10~5~2x 10"1
12
13
14
15
16
17
"Results extracted from Table 5-30.
bMost sensitive male rat result using default methodology is 2.5 x 1CT2 per mg/kg/d for NTP (1988) Marshall rat
testicular tumors.
AD = adenoma, CARC = carcinoma.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-120 DRAFT—DO NOT CITE OR QUOTE
-------
1 The rodent bioassay data offer conclusive evidence of carcinogenicity in both rats and
2 mice, and the available epidemiologic and mechanistic data support the relevance to humans of
3 the TCE-induced carcinogenicity observed in rodents. The epidemiologic data provide sufficient
4 evidence that TCE is carcinogenic to humans (see Section 4.11). There is even some evidence of
5 site concordance with the rodent findings, although site concordance is not essential to human
6 relevance and, in fact, is not observed across TCE-exposed rats and mice. The strongest
7 evidence in humans is for TCE-induced kidney tumors, with fairly strong evidence for
8 lymphomas and some lesser support for liver tumors; each of these tumor types has also been
9 observed in TCE rodent bioassays. Furthermore, the mechanistic data are supportive of human
10 relevance because, while the exact reactive species associated with TCE-induced tumors are not
11 known, the metabolic pathways for TCE are qualitatively similar for rats, mice, and humans (see
12 Section 3.3). The impact of uncertainties with respect to quantitative differences in TCE
13 metabolism is discussed in Section 5.2.1.4.2.
14 Typically, the cancer risk estimated is for the total cancer burden from all sites that
15 demonstrate an increased tumor incidence for the most sensitive experimental species and sex. It
16 is expected that this approach is protective of the human population, which is more diverse but is
17 exposed to lower exposure levels.
18 For the inhalation unit risk estimates, the preferred estimate from the most sensitive
19 species and sex was the estimate of 8.3 x 1CT2 per ppm for the male rat, which was based on
20 multiple tumors observed in this sex/species but was dominated by the kidney tumor risk
21 estimated with the dose metric for bioactivated DCVC. This estimate was the high end of the
22 range of estimates (see Table 5-32) but was within an order of magnitude of other estimates,
23 such as the preferred estimate for the female mouse and the male rat kidney estimate based on
24 the GSH conjugation dose metric, which provide additional support for an estimate of this
25 magnitude. The preferred estimate for the male mouse was about an order of magnitude and a
26 half lower. The female rat showed no apparent TCE-associated tumor response in the 3 available
27 inhalation bioassays; however, this apparent absence of response is inconsistent with the
28 observations of increased cancer risk in occupationally exposed humans and in female rats in
29 oral bioassays. In Section 5.2.2.2, an inhalation unit risk estimate based on the human data is
30 derived and can be compared to the rodent-based estimate.
31 For the oral unit risk (slope factor) estimate, the preferred estimate from the most
32 sensitive species and sex was the estimate of 2.5 x 10"1 per mg/kg/d, again for the male rat,
33 based on the kidney tumor risk estimated with the dose metric for bioactivated DCVC. This
34 estimate was at the high end of the range of estimates (see Table 5-33) but was within an order of
35 magnitude of other estimates, such as the preferred male mouse estimate and the male rat kidney
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-121 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimate based on the GSH conjugation dose metric, which provide additional support for an
2 estimate of this magnitude. The preferred estimates for the female mouse and the female rat
3 were about another order of magnitude lower. Some of the oral unit risk estimates based on the
4 alternative dose metric of AUC for TCE in the blood were as much as 3 orders of magnitude
5 lower, but these estimates were considered less credible than those based on the preferred dose
6 metrics. In Section 5.2.2.3, an oral unit risk estimate based on the human (inhalation) data is
7 derived using the PBPK model for route-to-route extrapolation; this estimate can be compared to
8 the rodent-based estimate.
9 Furthermore, the male rat kidney tumor estimates from the inhalation (Maltoni et al.,
10 1986) and oral (NTP, 1988) studies were consistent on the basis of internal dose using the dose
11 metric for bioactivated DCVC. In particular, the linearly extrapolated slope (i.e., the
12 BMR/BMDL) per unit of internal dose derived from Maltoni et al. (1986) male rat kidney tumor
13 data was 2.4 x 1CT1 per weekly mg DCVC bioactivated per unit body weight34, while the
14 analogous slope derived from NTP (1988) male rat kidney tumor data was 9.3 x 1CT2 per weekly
15 mg DCVC bioactivated per unit body weight4 (MSW-modeled results), a difference of less than
16 3-fold.28 These results also suggest that differences between routes of administration are
17 adequately accounted for by the PBPK model using this dose metric.
18 Regarding low-dose extrapolation, a key consideration in determining what extrapolation
19 approach to use is the MOA(s). However, MOA data are lacking or limited for each of the
20 cancer responses associated with TCE exposure, with the exception of the kidney tumors (see
21 Section 4.11). For the kidney tumors, the weight of the available evidence supports the
22 conclusion that a mutagenic MOA is operative (see Section 4.4); this MOA supports linear low-
23 dose extrapolation. For the other TCE-induced tumors, the MOA(s) is unknown. When the
24 MOA(s) cannot be clearly defined, U.S. EPA generally uses a linear approach to estimate low-
25 dose risk (U.S. EPA, 2005a), based on the following general principles:
26
27 • A chemical's carcinogenic effects may act additively to ongoing biological processes,
28 given that diverse human populations are already exposed to other agents and have
29 substantial background incidences of various cancers.
28For the Maltoni et al. (1986) male rat kidney tumors, the unit risk estimate of 8.3 x 10~2 per ppm using the
ABioactDCVCBW34 dose metric, from Table 5-29, is divided by the average male and female internal doses at
0.001 ppm, (0.0034/0.001), from Table 5-28, to yield a unit risk in internal dose units of 2.4 x 10"2. For the
NTP (1988) male rat kidney tumors, the unit risk estimate of 2.5 x 10"1 per mg/kg/d using the ABioactDCVCBW34
dose metric, from Table 5-30, is divided by the average male and female internal doses at 0.001 mg/kg/d,
(0.0027/0.001), from Table 5-28, to yield a unit risk in internal dose units of 9.3 x 10"2. Note that the original
BMDLs and unit risks from BMD modeling were in internal dose units that were then converted to applied dose
units using the values in Table 5-28, so this calculation reverses that conversion.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-122 DRAFT—DO NOT CITE OR QUOTE
-------
1 • A broadening of the dose-response curve (i.e., less rapid fall-off of response with
2 decreasing dose) in diverse human populations and, accordingly, a greater potential for
3 risks from low-dose exposures (Ziese et al., 1987; Lutz et al., 2005) is expected for two
4 reasons: First, even if there is a "threshold" concentration for effects at the cellular level,
5 that threshold is expected to differ across individuals. Second, greater variability in
6 response to exposures would be anticipated in heterogeneous populations than in inbred
7 laboratory species under controlled conditions (due to, e.g., genetic variability, disease
8 status, age, nutrition, and smoking status).
9 • The general use of linear extrapolation provides reasonable upper-bound estimates that
10 are believed to be health-protective (U.S. EPA, 2005a) and also provides consistency
11 across assessments.
12
13 Additional uncertainties arise from the specific dosimetry assumptions, the model
14 structures and parameter estimates in the PBPK models, the dose-response modeling of data in
15 the observable range, and the application of the results to potentially sensitive human
16 populations. As discussed in Section 5.2.1.2.1, one uncertainty in the tissue-specific dose
17 metrics used here is whether to scale the rate of metabolism by tissue mass or body weight to the
18 3/4 in the absence of specific data on clearance; however, in the cases where this is an issue (the
19 lung, liver, and kidney), the impact of this choice is relatively modest (less than 2-fold to about
20 4-fold). An additional dosimetry assumption inherent in this analysis is that equal concentrations
21 of the active moiety over a lifetime yield equivalent lifetime risk of cancer across species, and
22 the extent to which this is true for TCE is unknown. Furthermore, it should be noted that use of
23 tissue-specific dosimetry inherently presumes site concordance of tumors across species.
24 With respect to uncertainties in the estimates of internal dose themselves, a quantitative
25 analysis of the uncertainty and variability in the PBPK model-predicted dose metric estimates
26 and their impacts on cancer risk estimates is presented in Section 5.2.1.4.2. Additional
27 uncertainties in the PBPK model were discussed in Section 3.5. Furthermore, this assessment
28 examined a variety of dose metrics for the different tumor types using PBPK models for rats,
29 mice, and humans, so the impact of dose metric selection can be assessed. As discussed in
30 Section 5.2.1.2.1, there is strong support for the primary dose metrics selected for kidney, liver,
31 and, to a lesser extent, lung. For the other tumor sites, there is more uncertainty about dose
32 metric selection. The cancer unit risk estimates obtained using the preferred dose metrics were
33 generally similar (within about 3-fold) to those derived using default dosimetry assumptions
34 (e.g., equal risks result from equal cumulative equivalent exposures or doses), with the exception
35 of the bioactivated DCVC dose metric for rat kidney tumors and the metric for the amount of
36 TCE oxidized in the respiratory tract for mouse lung tumors occurring from oral exposure (see
37 Tables 5-32 and 5-33). The higher risk estimates for kidney tumors based on the bioactivated
38 DCVC dose metric are to be expected because pharmacokinetic data indicate, and the PBPK
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-123 DRAFT—DO NOT CITE OR QUOTE
-------
1 model predicts, substantially more GSH conjugation (as a fraction of intake), and hence
2 subsequent bioactivation, in humans relative to rats. The lower risk estimates for lung tumors
3 from oral TCE exposure based on the metric for the amount of TCE oxidized in the respiratory
4 tract are because there is a greater first-pass effect in human liver relative to mouse liver
5 following oral exposure and because the gavage dosing used in rodent studies leads to a large
6 bolus dose that potentially overwhelms liver metabolism to a greater extent than a more graded
7 oral exposure. Both of these effects result in relatively more TCE being available for
8 metabolism in the lung for mice than for humans. In addition, mice have greater respiratory
9 metabolism relative to humans. However, because oxidative metabolites produced in the liver
10 may contribute to respiratory tract effects, using respiratory tract metabolism alone as a dose
11 metric may underestimate lung tumor risk. The unit risk estimates obtained using the alternative
12 dose metrics were also generally similar to those derived using default dosimetry assumptions,
13 with the exception of the metric for the amount of TCE conjugated with GSH for rat kidney
14 tumors, again because humans have greater GSH conjugation, and the AUC of TCE in blood for
15 all the tumor types resulting from oral exposure, again because of first-pass effects.
16 With respect to uncertainties in the dose-response modeling, the two-step approach of
17 modeling only in the observable range, as put forth in U.S. EPA's cancer assessment guidelines
18 (U.S. EPA, 2005a), is designed in part to minimize model dependence. The ratios of the BMDs
19 to the BMDLs give some indication of the uncertainties in the dose-response modeling. These
20 ratios did not exceed a value of 2.5 for all the primary analyses used in this assessment. Thus,
21 overall, modeling uncertainties in the observable range are considered to be negligible. Some
22 additional uncertainty is conveyed by uncertainties in the survival adjustments made to some of
23 the bioassay data; however, their impact is also believed to be minimal relative to the
24 uncertainties already discussed (i.e., interspecies and low-dose extrapolations).
25 Regarding the cancer risks to potentially sensitive human populations or life stages,
26 pharmacokinetic data on 42 individuals were used in the Bayesian population analysis of the
27 PBPK model discussed in Section 3.5. The impacts of these data on the predicted population
28 mean are incorporated in the quantitative uncertainty analyses presented in Section 5.2.1.4.2.
29 These data do not, however, reflect the full range of metabolic variability in the human
30 population (they are all from healthy, mostly male, human volunteers) and do not address
31 specific potentially sensitive subgroups (see Section 4.10). Moreover, there is inadequate
32 information about disease status, coexposures, and other factors that make humans vary in their
33 responses to TCE. It will be a challenge for future research to quantify the differential risk
34 indicated by different risk factors or exposure scenarios.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-124 DRAFT—DO NOT CITE OR QUOTE
-------
1 5.2.1.4.2. Quantitative uncertainty analysis of physiologically based pharmacokinetic (PBPK)
2 model-based dose metrics. The Bayesian analysis of the PBPK model for TCE generates
3 distributions of uncertainty and variability in the internal dose metrics than can be readily fed
4 into dose-response analysis. As shown in Figure 5-6, the overall approach taken for the
5 uncertainty analysis is similar to that used for the point estimates except that distributions are
6 carried through the analysis rather than median or expected values. In particular, the PBPK
7 model-based rodent internal doses are carried through to a distribution of BMDs (which also
8 includes sampling variance from the number of responding and at risk animals in the bioassay).
9 This distribution of BMDs generates a distribution of cancer slope factors based on internal dose,
10 which then is combined with the (uncertainty) distribution of the human population mean
11 conversion to applied dose or exposure. The resulting distribution for the human population
12 mean risk per unit dose or exposure accounts for uncertainty in the PBPK model parameters
13 (rodent and human) and the binomial sampling error in the bioassays. These distributions can
14 then be compared with the point estimates, based on median rodent dose metrics and mean
15 human population dose metrics, reported in Tables 5-29 and 5-30. Details of the implementation
16 of this uncertainty analysis, which used the WinBugs software in conjugation with the
17 R statistical package, are reported in Appendix G.
18 Overall, as shown in Tables 5-34 and 5-35, the 95% confidence upper bound of the
19 distributions for the linearly extrapolated risk per unit dose or exposure ranged from 1- to 8-fold
20 higher than the point unit risks derived using the BMDLs reported in Tables 5-29 and 5-30. The
21 largest differences, up to 4-fold, for rat kidney tumors and 8-fold for mouse lung tumors,
22 primarily reflect the substantial uncertainty in the internal dose metrics for rat kidney DCVC and
23 GSH conjugation and for mouse lung oxidation (see Section 3.5). Additionally, despite the
24 differences in the degree of uncertainty due to the PBPK model across endpoints and dose
25 metrics, the only case where the choice of the most sensitive bioassay for each sex/species
26 combination would change based on the 95% confidence upper bounds reported in Tables 5-34
27 and 5-35 would be for female mouse inhalation bioassays. Even in this case, the difference
28 between unit risk estimate for the most sensitive and next most sensitive study/endpoint was only
29 2-fold.
30
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-125 DRAFT—DO NOT CITE OR QUOTE
-------
fixed^
Idistribution
[distribution (combined
ncertainty and variability)
Rodent
internal
dose
Dose-
esponse
Model
Cancer slope
factor =
BMR/BMD
(per internal
dose unit))
Human cancer
slope factor
(per ppm or
per mg/kg-d)
listribution
[distribution (separate
ncertainty and variability)
Uncertainty
idistribution of
[population mean
Population
mean human
internal dose
distribution
4
5
6
7
Figure 5-6. Flow-chart for uncertainty analysis of dose-response analyses of
rodent bioassays using PBPK model-based dose metrics. Square nodes
indicate point values, circular nodes indicate distributions, and the inverted
triangles indicate a (deterministic) functional relationship.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-126 DRAFT—DO NOT CITE OR QUOTE
-------
to
Table 5-34. Summary of PBPK model-based uncertainty analysis of unit risk estimates for each
sex/species/bioassay/tumor type (inhalation)
Study
Tumor Type
BMR
Dose Metric
Unit risk estimates (mg/kg-d)"1)
From
Table
5-29
Summary statistics of unit risk distribution
Mean
5% lower
bound
Median
95% upper
bound
Female mouse
Fukuda
Henschler
Maltoni
Lung AD + CARCa
Lymphomab
Lung AD + CARCa
Liver
0.1
0.1
0.1
0.05
AMetLngBW34
TotOxMetabBW34
AUCCBld
TotMetabBW34
AMetLngBW34
TotOxMetabBW34
AUCCBld
AMetLivlBW34
TotOxMetabBW34
2.6 x 10 3
3.2 x lO'3
1.8 x ID'3
1.0 x 10 2
1.8 x 10 3
1.9 x 1Q-3
1.0 x lO'3
1.2 x 10 3
1.1 x ID'3
5.65 x ID'3
1.88 x ID'3
1.01 x ID'3
4.38 x ID'3
3.88 x ID'3
1.10 x 1Q-3
5.25 x ID'4
6.27 x lO'4
5.98 x ID'4
2.34 x lO'4
3.27 x lO'4
1.54 x lO'4
6.06 x lO'4
1.48 x ID'4
3.73 x ID'4
1.63 x ID'4
2.18 x ID'4
1.81 x ID'4
1.49 x lO'3
1.52 x lO'3
8.36 x lO'4
3.49 x lO'3
1.04 x 1Q-3
9.52 x 1Q-4
4.64 x lO'4
5.39 x lO'4
5.07 x lO'4
2.18 x lO'2
4.59 x lO'3
2.44 x lO'3
1.11 x ID'2
1.52 x 1Q-2
2.32 x 1Q-3
l.lOx lO'3
1.32 x lO'3
1.31 x ID'3
Male mouse
Maltoni
Liver
0.1
AMetLivlBW34
TotOxMetabBW34
2.6 x 10 3
2.0 x lO'3
1.35 x ID'3
1.23 x ID'3
4.28 x ID'4
4.24 x lO'4
1.16 x 1Q-3
1.06 x lO'3
2.93 x ID'3
2.60 x lO'3
Male rat
Maltoni
Leukemia*3
Kidney AD +
CARC
Leydig cellb
0.05
0.01
0.1
TotMetabBW34
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
TotMetabBW34
1.8 x 10 3
8.3 x 1Q-2
5.1 x ID'2
7.3 x ID'4
5.5 x 10 3
9.38 x ID'4
9.07 x lO'2
3.90 x lO'2
3.94 x 1Q-4
4.34 x lO'3
1.26 x lO'4
3.66 x lO'3
2.71 x ID'3
8.74 x 1Q-5
1.99 x lO'3
7.86 x lO'4
3.64 x lO'2
2.20 x lO'2
3.42 x 1Q-4
3.98 x ID'3
2.25 x ID'3
3.21 x lO'1
1.30 x 10'1
8.74 x 1Q-4
7.87 x lO'3
§
i
TO'
I
TO
H I
O ^
HH Oq
H TO
O
H
W
aWinBUGS dose-response analyses did not adequately converge for the AMetLngBW34 dose metric using the 3rd-order multistage model (used for results in
Table 5-29), but did converge when the 2nd-order model was used. Summary statistics reflect results of 2nd-order model calculations.
bPoor dose-response fits in point estimates for AUCCBld, so not included in uncertainty analysis.
AD = adenoma, CARC = carcinoma.
-------
to
Table 5-35. Summary of PBPK model-based uncertainty analysis of unit risk estimates for each
sex/species/bioassay/tumor type (oral)
Study
Tumor type
BMR
Dose metric
Unit risk estimates (mg/kg-d)"1)
From
Table 5-30
or 5-31
Summary statistics of distribution
Mean
5% lower
bound
Median
95% upper
bound
Female mouse
NCI
Liver CARC
Lung AD + CARCa
Leukemias + sarcomas
0.1
0.1
0.1
AMetLivlBW34
TotOxMetabBW34
AMetLngBW34
TotOxMetabBW34
AUCCBld
TotMetabBW34
AUCCBld
7.1 x 10 3
5.7 x lO'3
1.3 x 10 4
4.0 x lO'3
1.5 x 1Q-4
4.9 x 10 3
1.4 x lO'4
3.26 x 10'3
2.63 x 10'3
1.28 x lO'4
1.84 x lO'3
7.16x 1Q-5
1.60 x 1Q-3
6.36 x lO'5
9.35 x lO'4
8.76 x lO'4
6.73 x ID'6
5.29 x lO'4
4.40 x ID'6
1.42 x ID'4
3.10 x lO'6
2.44 x lO'3
2.01 x ID'3
4.12x lO'5
1.39 x lO'3
3.39 x ID'5
1.13 x ID'3
2.90 x lO'5
8.35 x ID'3
6.60 x lO'3
4.62 x lO'4
4.73 x ID'3
2.18 x ID'4
4.65 x ID'3
1.94 x lO'4
Male mouse
NCI
Liver CARC
0.1
AMetLivlBW34
TotOxMetabBW34
2.9 x W2
23 x m2
1.65 x lO'2
1.32 x lO'2
4.70 x lO'3
4.41 x ID'3
1.25 x ID'2
1.01 x ID'2
4.25 x ID'2
3.29 x lO'2
Female rat
NTP, 1988
Leukemia
0.05
TotMetabBW34
AUCCBld
2.3 x 10 3
1.6 x lO'5
1.89 x lO'3
1.56 x lO'5
5.09 x lO'4
3.39 x lO'6
1.43 x ID'3
1.07 x lO'5
4.69 x lO'3
3.98 x ID'5
Male rat
NTP, 1990
Kidney AD + CARCb
0.1
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
1.2 x 10'1
7.6 x \Q-2
3.1 x 1Q-3
1.40 x 10'1
6.18 x lO'2
2.49 x 1Q-3
5.69 x lO'3
4.00 x lO'3
7.14 x ID'4
5.24 x lO'2
3.27 x lO'2
1.96 x ID'3
5.18 x lO'1
2.11 x lO'1
5.96 x ID'3
§
i
TO'
I
TO
H I
O ^
HH Oq
H TO
O
H
W
-------
to
Table 5-35. Summary of PBPK model-based uncertainty analysis of unit risk estimates for each
sex/species/bioassay/tumor type (oral) (continued)
Study
Tumor type
BMR
Dose metric
Unit risk estimates (mg/kg-d)"1)
From
Table 5-30
or 5-31
Summary statistics of distribution
Mean
5% lower
bound
Median
95% upper
bound
NTP, 1988
Marshall
August
Osborne-Mendel
Testicularb
Subcut sarcoma
Kidney AD + CARCb
0.1
0.05
0.1
TotMetabBW34
AUCCBld
TotMetabBW34
AUCCBld
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
7.1 x 102
6.0 x 1Q-4
2.3 x 103
2.0 x lO'5
1.6 x lO'1
9.7 x lO'2
4.3 x ID'3
6.18 x ID'2
5.45 x ID'4
1.65 x ID'3
1.35 x ID'5
1.61 x lO'1
7.47 x lO'2
2.73 x ID'3
1.92 x lO'2
1.18 x ID'4
4.58 x ID'4
1.53 x ID'6
5.45 x ID'3
3.90 x lO'3
5.40 x lO'4
4.89 x lO'2
3.70 x 1Q-4
1.27 x 1Q-3
8.34 x lO'6
6.35 x ID'2
3.85 x ID'2
2.10 x lO'3
1.45 x lO'1
1.44 x 1Q-3
4.04 x 1Q-3
3.73 x ID'5
6.02 x 10'1
2.54 x 10'1
6.89 x lO'3
§
i
TO'
Ss
I
TO
aWinBUGS dose-response analyses did not adequately converge for AMetLngBW34 dose metric using the 3rd-order multistage model (used for results in
Table 5-30), but did converge when the 2nd-order model was used. Summary statistics reflect results of 2nd-order model calculations.
bUsing poly-3 adjusted incidences from Table 5-31 (software for WinBUGS-based analyses using the MSW model was not developed).
AD = adenoma, CARC = carcinoma.
H I
O >
HH Oq
H TO
O
H
W
-------
1 5.2.2. Dose-Response Analyses: Human Epidemiologic Data
2 Of the epidemiological studies of TCE and cancer, only one had sufficient exposure-
3 response information for dose-response analysis. This was the Charbotel et al. (2006) case-
4 control study of TCE and kidney cancer incidence, which was used to derive an inhalation unit
5 risk estimate for that endpoint (see Section 5.2.2.1). Other epidemiological studies were used in
6 Section 5.2.2.2 below to provide information for a comparison of relative risk (RR) estimates
7 across cancer types. These epidemiologic data were used to derive an adjusted inhalation unit
8 risk estimate for the combined risk of developing kidney cancer, non-Hodgkin's lymphoma
9 (NHL), or liver cancer. The human PBPK model was then used to perform route-to-route
10 extrapolation to derive an oral unit risk estimate for the combined risk of kidney cancer, NHL, or
11 liver cancer (see Section 5.2.2.3).
12
13 5.2.2.1. Inhalation Unit Risk Estimate for Renal Cell Carcinoma Derived from Charbotel et
14 al. (2006) Data
15 The Charbotel et al. (2006) case-control study of 86 incident renal cell carcinoma (RCC)
16 cases and 316 age- and sex-matched controls, with individual cumulative exposure estimates for
17 TCE for each subject, provides a sufficient human data set for deriving quantitative cancer risk
18 estimates for RCC in humans. The study is a high-quality study that used a detailed exposure
19 assessment (Fevotte et al., 2006) and took numerous potential confounding factors, including
20 exposure to other chemicals, into account (see Section 4.4). A significant dose-response
21 relationship was reported for cumulative TCE exposure and RCC (Charbotel et al., 2006).
22 The derivation of an inhalation unit risk estimate, defined as the plausible upper bound
23 lifetime risk of cancer from chronic inhalation of TCE per unit of air concentration, for RCC
24 incidence in the U.S. population, based on results of the Charbotel et al. (2006) case-control
25 study, is presented in the following subsections.
26
27 5.2.2.1.1. Renal cell carcinoma (RCC) results from the Charbotel et al. (2006) study.
28 Charbotel et al. (2006) analyzed their data using conditional logistic regression, matching on sex
29 and age, and reported results (odds ratios [ORs]) for cumulative TCE exposure categories,
30 adjusted for tobacco smoking and body mass index (Charbotel et al., 2006, Table 6). The
31 exposure categories were constructed as tertiles based on the cumulative exposure levels in the
32 exposed control subjects. The results are summarized in Table 5-36, with mean exposure levels
33 kindly provided by Dr. Charbotel (personal communication from Barbara Charbotel, University
34 of Lyon, to Cheryl Scott, U.S. EPA, 11 April 2008).
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-130 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-36. Results from Charbotel et al. on relationship between TCE
exposure and RCC
Cumulative exposure
category
Nonexposed
Low
Medium
High
Mean Cumulative exposure
(ppm x yrs)
62.4
253.2
925.0
Adjusted OR
(95% CI)
1
1.62(0.75,3.47)
1.15(0.47,2.77)
2.16(1.02,4.60)
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
CI = confidence interval.
For additional details and discussion of the Charbotel et al. (2006) study, see Section 4.4
and Appendix B.
5.2.2.1.2. Prediction of lifetime extra risk of renal cell carcinoma (RCC) incidence from
trichloroethylene (TCE) exposure. The categorical results summarized in Table 5-36 were used
for predicting the extra risk of RCC incidence from continuous environmental exposure to TCE.
Extra risk is defined as
Extra risk = (Rx - Ro)l(\ - Ro\
(Eq. 5-3)
where Rx is the lifetime risk in the exposed population and Ro is the lifetime risk in an
unexposed population (i.e., the background risk). Because kidney cancer is a rare event, the ORs
in Table 5-36 can be used as estimates of the relative risk ratio, RR = RxIRo (Rothman and
Greenland, 1998). A weighted linear regression model was used to model the dose-response data
in Table 5-36 to obtain a slope estimate (regression coefficient) for RR of RCC versus
cumulative exposure. Use of a linear model in the observable range of the data is often a good
general approach for epidemiological data because such data are frequently too limited (i.e.,
imprecise), as is the case here, to clearly identify an alternate model (U.S. EPA, 2005a). This
linear dose-response function was then used to calculate lifetime extra risks in an actuarial
program (life-table analysis) that accounts for age-specific rates of death and background
disease, under the assumption that the RR is independent of age.29
29
This program is an adaptation of the approach previously used by the Committee on the Biological Effects of
Ionizing Radiation (BEIR, 1988). The same methodology was also used in U.S. EPA's 1,3-butadiene health risk
assessment (U.S. EPA, 2002). A spreadsheet illustrating the extra risk calculation for the derivation of the LECM
for RCC incidence is presented in Appendix H.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-131 DRAFT—DO NOT CITE OR QUOTE
-------
1 For the weighted linear regression, the weights used for the RR estimates were the
2 inverses of the variances, which were calculated from the confidence intervals. Using this
3 approach,30 a linear regression coefficient of 0.001205 per ppm x year
4 (standard error = 0.0008195 per ppm x year) was obtained from the categorical results.
5 For the life-table analysis, U.S. age-specific all-cause mortality rates for 2004 for both
6 sexes and all race groups combined (NCHS, 2007) were used to specify the all-cause background
7 mortality rates in the actuarial program. Because the goal is to estimate the unit risk for extra
8 risk of cancer incidence, not mortality, and because the Charbotel et al. data are incidence data,
9 RCC incidence rates were used for the cause-specific background "mortality" rates in the life-
10 table analysis.31 Surveillance, Epidemiology, and End Results (SEER) 2001-2005 cause-
11 specific background incidence rates for RCC were obtained from the SEER public-use
12 database.32 SEER collects good-quality cancer incidence data from a variety of geographical
13 areas in the United States. The incidence data used here are from SEER 17, a registry of
14 17 states, cities, or regions covering about 26% of the United States population
15 (http://seer.cancer.gov). The risks were computed up to age 85 years for continuous exposures to
16 TCE.33 Conversions between occupational TCE exposures and continuous environmental
17 exposures were made to account for differences in the number of days exposed per year (240 vs.
18 365 days) and in the amount of air inhaled per day (10 vs. 20 m3; U.S. EPA, 1994). The standard
19 error for the regression coefficient from the weighted linear regression calculation described
20 above was used to compute the 95% upper confidence limit (UCL) for the slope estimate, and
21 this value was used to derive 95% UCLs for risk estimates (or 95% LCLs for corresponding
22 exposure estimates), based on a normal approximation.
23 Point estimates and one-sided 95% UCLs for the extra risk of RCC incidence associated
24 with varying levels of environmental exposure to TCE based on linear regression of the
25 Charbotel et al. (2006) categorical results were determined by the actuarial program; the results
26 are presented in Section 5.2.13. The models based on cumulative exposure yield extra risk
27 estimates that are fairly linear for exposures up to 1 ppm or so.
Equations for this weighted linear regression approach are presented in Rothman (1986) and summarized in
Appendix H.
31No adjustment was made for using RCC incidence rates rather than mortality rates to represent cause-specific
mortality in the actuarial program because the RCC incidence rates are negligible in comparison to the all-cause
mortality rates. Otherwise, all-cause mortality rates for each age interval would have been adjusted to reflect people
dying of a cause other than RCC or being diagnosed with RCC.
32In accordance with the "SEER Program Coding and Staging Manual 2007"
(http://seer.cancer.gov/manuals/2007/SPCSM_2007_AppendixC_p6.pdf), pages C-831 to C-833, RRC was
specified as ICD-0-3 histological types coded 8312, 8260, 8310, 8316-8320, 8510, 8959, and 8255 (mixed types).
33Rates above age 85 years are not included because cause-specific disease rates are less stable for those ages. Note
that 85 years is not employed here as an average lifespan but, rather, as a cut-off point for the life-table analysis,
which uses actual age-specific mortality rates.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-132 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Consistent with U.S. EPA's Guidelines for Carcinogen Risk Assessment
(U.S. EPA, 2005a), the same data and methodology were also used to estimate the exposure level
(ECX: "effective concentration corresponding to an extra risk of x%") and the associated 95%
lower confidence limit of the effective concentration corresponding to an extra risk of 1%
(LECX, x = 0.01). A 1% extra risk level is commonly used for the determination of the POD for
epidemiological data. Use of a 1% extra risk level for these data is supported by the fact that,
based on the actuarial program, the risk ratio (i.e., Rx/Ro) for an extra risk of 1% for RCC
incidence is 1.9, which is in the range of the ORs reported by Charbotel et al. (see Table 5-36).
Thus, 1% extra risk was selected for determination of the POD, and, consistent with the
Guidelines for Carcinogen Risk Assessment, the lowest effective concentration (LEG) value
corresponding to that risk level was used as the actual POD. For the linear model that was
selected, the unit risk is independent of the benchmark risk level used to determine the POD (at
low exposures/risk levels; see Table 5-37); however, selection of a benchmark risk level is
generally useful for comparisons across models.
Table 5-37. Extra risk estimates for RCC incidence from various levels of
lifetime exposure to TCE, using linear cumulative exposure model
Exposure concentration (ppm)
0.001
0.01
0.1
1.0
10.0
MLE of extra risk
2.603 x 10'6
2.603 x 10'5
2.602 x 10'4
2.598 x 10'3
2.562 x 10'2
95% UCL on extra risk
5.514 x 10'6
5.514 x 10'5
5.512 x 10'4
5.496 x 10'3
5.333 x 10'2
19
20
21
22
23
24
25
26
As discussed in Section 4.4, there is sufficient evidence to conclude that a mutagenic
MO A is operative for TCE-induced kidney tumors, which supports the use of linear low-dose
extrapolation from the POD. The ECoi, LECoi, and inhalation unit risk estimates for RCC
incidence using the linear cumulative exposure model are presented in Table 5-38. Converting
the units, 5.49 x 10"3 per ppm corresponds to a unit risk of 1.02 x 10"6 per ug/m3 for RCC
incidence.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-133 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-38. EC0i, LEC0i, and unit risk estimates for RCC incidence, using
2 linear cumulative exposure model
O
ECoi (ppm)
3.87
LECoi (ppm)
1.82
unit risk (per ppm)*
5.49 x 10'3
4
5 *Unit risk = 0.01/LECM.
6
7
8 5.2.2.1.3. Uncertainties in the renal cell carcinoma (RCC) unit risk estimate. The two major
9 sources of uncertainty in quantitative cancer risk estimates are generally interspecies
10 extrapolation and high-dose to low-dose extrapolation. The unit risk estimate for RCC incidence
11 derived from the Charbotel et al. (2006) results is not subject to interspecies uncertainty because
12 it is based on human data. A major uncertainty remains in the extrapolation from occupational
13 exposures to lower environmental exposures. There was some evidence of a contribution to
14 increased RCC risk from peak exposures; however, there remained an apparent dose-response
15 relationship for RCC risk with increasing cumulative exposure without peaks, and the OR for
16 exposure with peaks compared to exposure without peaks was not significantly elevated
17 (Charbotel et al., 2006). Although the actual exposure-response relationship at low exposure
18 levels is unknown, the conclusion that a mutagenic MOA is operative for TCE-induced kidney
19 tumors supports the linear low-dose extrapolation that was used (U.S. EPA, 2005a).
20 Another notable source of uncertainty in the cancer unit risk estimate is the dose-response
21 model used to model the study data to estimate the POD. A weighted linear regression across the
22 categorical ORs was used to obtain a slope estimate; use of a linear model in the observable
23 range of the data is often a good general approach for human data because epidemiological data
24 are frequently too limited (i.e., imprecise) to clearly identify an alternate model (U.S. EPA,
25 2005a). The Charbotel et al. study is a relatively small case-control study, with only 86 RCC
26 cases, 37 of which had TCE exposure; thus, the dose-response data upon which to specify a
27 model are indeed limited.
28 In accordance with U.S. EPA's Guidelines for Carcinogen Risk Assessment, the lower
29 bound on the ECoi is used as the POD; this acknowledges some of the uncertainty in estimating
30 the POD from the available dose-response data. In this case, the statistical uncertainty associated
31 with the ECoi is relatively small, as the ratio between the ECoi and the LECoi is about 2-fold.
32 The inhalation unit risk estimate of 5.49 x 10"3 per ppm presented above, which is calculated
33 based on a linear extrapolation from the POD (LECoi), is expected to provide an upper bound on
34 the risk of cancer incidence. However, for certain applications, such as benefit-cost analyses,
35 estimates of "central tendency" for the risk below the POD are desired. Because a linear dose-
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-134 DRAFT—DO NOT CITE OR QUOTE
-------
1 response model was used in the observable range of the human data and the POD was within the
2 low-dose linear range for extra risk as a function of exposure, linear extrapolation below the
3 LECoi has virtually the same slope as the 95% UCL on the actual (linear) dose-response model
4 in the low-dose range (i.e., below the POD). This is illustrated in Table 5-37, where the 95%
5 UCL on extra risk for RCC incidence predicted by the dose-response model is about 5.51 x 10"3
6 per ppm for exposures at or below about 0.1 ppm, which is virtually equivalent to the unit risk
7 estimate of 5.49 x 10"3 per ppm derived from the LECoi (see Table 5-38). The same holds for
8 the central tendency (weighted least squares) estimates of the extra risk from the (linear) dose-
9 response model (i.e., the dose-response model prediction of 2.60 x 10"3 per ppm from Table 5-37
10 is virtually identical to the value of 2.58 x 10"3 per ppm obtained from linear extrapolation below
11 the ECoi, i.e., by dividing 0.01 extra risk by the EC0i of 3.87 from Table 5-38). In other words,
12 because the dose-response model that was used to model the data in the observable range is
13 already low-dose linear near the POD, if one assumes that the same linear model is valid for the
14 low-dose range, one can use the central tendency (weighted least squares) estimates from the
15 model to derive a statistical "best estimate" of the slope rather than relying on an extrapolated
16 risk estimates (0.01/ECoi). [The extrapolated risk estimates are not generally central tendency
17 estimates in any statistical sense because once risk is extrapolated below the ECoi using the
18 formulation 0.01/ECoi, it is no longer a function of the original model which generated the ECoiS
19 and the LECois.]
20 An important source of uncertainty in the underlying Charbotel et al. (2006) study is the
21 retrospective estimation of TCE exposures in the study subjects. This case-control study was
22 conducted in the Arve Valley in France, a region with a high concentration of workshops
23 devoted to screw cutting, which involves the use of TCE and other degreasing agents. Since the
24 1960s, occupational physicians of the region have collected a large quantity of well-documented
25 measurements, including TCE air concentrations and urinary metabolite levels (Fevotte et al.,
26 2006). The study investigators conducted a comprehensive exposure assessment to estimate
27 cumulative TCE exposures for the individual study subjects, using a detailed occupational
28 questionnaire with a customized task-exposure matrix for the screw-cutting workers and a more
29 general occupational questionnaire for workers exposed to TCE in other industries
30 (Fevotte et al., 2006). The exposure assessment even attempted to take dermal exposure from
31 hand-dipping practices into account by equating it with an equivalent airborne concentration
32 based on biological monitoring data. Despite the appreciable effort of the investigators,
33 considerable uncertainty associated with any retrospective exposure assessment is inevitable, and
34 some exposure misclassification is unavoidable. Such exposure misclassification was most
35 likely for the 19 deceased cases and their matched controls, for which proxy respondents were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-135 DRAFT—DO NOT CITE OR QUOTE
-------
1 used, and for exposures outside the screw-cutting industry (295 of 1,486 identified job periods
2 involved TCE exposure; 120 of these were not in the screw-cutting industry).
3 Another noteworthy source of uncertainty in the Charbotel et al. (2006) study is the
4 possible influence of potential confounding or modifying factors. This study population, with a
5 high prevalence of metal-working, also had relatively high prevalences of exposure to petroleum
6 oils, cadmium, petroleum solvents, welding fumes, and asbestos (Fevotte et al., 2006). Other
7 exposures assessed included other solvents (including other chlorinated solvents), lead, and
8 ionizing radiation. None of these exposures was found to be significantly associated with RCC
9 atap = 0.05 significance level. Cutting fluids and other petroleum oils were associated with
10 RCC at ap = 0.1 significance level; however, further modeling suggested no association with
11 RCC when other significant factors were taken into account (Charbotel et al., 2006). The
12 medical questionnaire included familial kidney disease and medical history, such as kidney
13 stones, infection, chronic dialysis, hypertension, and use of anti-hypertensive drugs, diuretics,
14 and analgesics. Body mass index (BMI) was also calculated, and lifestyle information such as
15 smoking habits and coffee consumption was collected. Univariate analyses found high levels of
16 smoking and BMI to be associated with increased odds of RCC, and these two variables were
17 included in the conditional logistic regressions. Thus, although impacts of other factors are
18 possible, this study took great pains to attempt to account for potential confounding or modifying
19 factors.
20 Some other sources of uncertainty associated with the epidemiological data are the dose
21 metric and lag period. As discussed above, there was some evidence of a contribution to
22 increased RCC risk from peak TCE exposures; however, there appeared to be an independent
23 effect of cumulative exposure without peaks. Cumulative exposure is considered a good
24 measure of total exposure because it integrates exposure (levels) over time. If there is a
25 contributing effect of peak exposures, not already taken into account in the cumulative exposure
26 metric, the linear slope may be overestimated to some extent. Sometimes cancer data are
27 modeled with the inclusion of a lag period to discount more recent exposures not likely to have
28 contributed to the onset of cancer. In an unpublished report (Charbotel et al., 2005), Charbotel
29 et al. also present the results of a conditional logistic regression with a 10-year lag period, and
30 these results are very similar to the unlagged results reported in their published paper, suggesting
31 that the lag period might not be an important factor in this study.
32 Some additional sources of uncertainty are not so much inherent in the exposure-response
33 modeling or in the epidemiologic data themselves but, rather, arise in the process of obtaining
34 more general Agency risk estimates from the epidemiologic results. U.S. EPA cancer risk
35 estimates are typically derived to represent an upper bound on increased risk of cancer incidence
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-136 DRAFT—DO NOT CITE OR QUOTE
-------
1 for all sites affected by an agent for the general population. From experimental animal studies,
2 this is accomplished by using tumor incidence data and summing across all the tumor sites that
3 demonstrate significantly increased incidences, customarily for the most sensitive sex and
4 species, to attempt to be protective of the general human population. However, in estimating
5 comparable risks from the Charbotel et al. (2006) epidemiologic data, certain limitations are
6 encountered. For one thing, these epidemiology data represent a geographically limited (Arve
7 Valley, France) and likely not very diverse population of working adults. Thus, there is
8 uncertainty about the applicability of the results to a more diverse general population.
9 Additionally, the Charbotel et al. (2006) study was a study of RCC only, and so the risk estimate
10 derived from it does not represent all the tumor sites that may be affected by TCE. The issue of
11 cancer risk at other sites is addressed in the next section (see Section 5.2.2.2).
12
13 5.2.2.1.4. Conclusions regarding the renal cell carcinoma (RCC) unit risk estimate. An ECoi
14 of 3.9 ppm was calculated using a life-table analysis and linear modeling of the categorical
15 conditional logistic regression results for RCC incidence reported in a high-quality case-control
16 study. Linear low-dose extrapolation from the LECoi yielded a lifetime extra RCC incidence
17 unit risk estimate of 5.5 x 10"3 per ppm (1.0 x 10"6 per |ig/m3) of continuous TCE exposure. The
18 assumption of low-dose linearity is supported by the conclusion that a mutagenic MOA is
19 operative for TCE-induced kidney tumors. The inhalation unit risk estimate is expected to
20 provide an upper bound on the risk of RCC incidence; however, this is just the risk estimate for
21 RCC. A risk estimate for total cancer risk to humans would need to include the risk for other
22 potential TCE-associated cancers.
23
24 5.2.2.2. Adjustment of the Inhalation Unit Risk Estimate for Multiple Sites
25 Human data on TCE exposure and cancer risk sufficient for dose-response modeling are
26 only available for RCC, yet human and rodent data suggest that TCE exposure increases the risk
27 of cancer at other sites as well. In particular, there is evidence from human (and rodent) studies
28 for increased risks of lymphoma and liver cancer (see Section 4.11). Therefore, the inhalation
29 unit risk estimate derived from human data for RCC incidence was adjusted to account for
30 potential increased risk of those tumor types. To make this adjustment, a factor accounting for
31 the relative contributions to the extra risk for cancer incidence from TCE exposure for these
32 three tumor types combined versus the extra risk for RCC alone was estimated, and this factor
33 was applied to the unit risk estimate for RCC to obtain a unit risk estimate for the three tumor
34 types combined (i.e., lifetime extra risk for developing any of the 3 types of tumor). This
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-137 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimate is considered a better estimate of total cancer risk from TCE exposure than the estimate
2 for RCC alone.
3 Although only the Charbotel et al. (2006) study was found adequate for direct estimation
4 of inhalation unit risks, the available epidemiologic data provide sufficient information for
5 estimating the relative potency of TCE across tumor sites. In particular, the relative
6 contributions to extra risk (for cancer incidence) were calculated from two different data sets to
7 derive the adjustment factor for adjusting the unit risk estimate for RCC to a unit risk estimate
8 for the 3 types of cancers (RCC, lymphoma, and liver) combined. The first calculation is based
9 on the results of the meta-analyses of human epidemiologic data for the three tumor types (see
10 Appendix C); the second calculation is based on the results of the Raaschou-Nielsen et al. (2003)
11 study, the largest single human epidemiologic study by far with RR estimates for all three tumor
12 types. The approach for each calculation was to use the RR estimates and estimates of the
13 lifetime background risk in an unexposed population, Ro, to calculate the lifetime risk in the
14 exposed population, Rx, where Rx = RR x Ro, for each tumor type. Then, the extra risk from
15 TCE exposure for each tumor type could be calculated using the equation in Section 5.2.2.1.2.
16 Finally, the extra risks were summed across the three tumor types and the ratio of the sum of the
17 extra risks to the extra risk for RCC was derived. For the first calculation, the pooled relative
18 risk estimates (RRps) from the meta-analyses for lymphoma, kidney cancer, and liver (and
19 biliary) cancer were used as the RR estimates. For the second calculation, the SIR estimates
20 from the Raaschou-Nielsen et al. (2003) study were used. For both calculations, Ro for RCC
21 was taken from the life-table analysis described in Section 5.2.2.1.2 and presented in
22 Appendix H, which estimated a lifetime risk for RCC incidence up to age 85 years. For Ro
23 values for the other 2 sites, SEER statistics for the lifetime risk of developing cancer were used
24 (http://seer.cancer.gov/statfacts/html/nhl.html and
25 http://seer.cancer.gov/statfacts/html/livibd.html).
26 In both cases, an underlying assumption in deriving the relative potencies is that the
27 relative values of the age-specific background incidence risks for the person-years from the
28 epidemiologic studies for each tumor type approximate the relative values of the lifetime
29 background incidence risks for those tumor types. In other words, at least on a proportional
30 basis, the lifetime background incidence risks (for the United States population) for each site
31 approximate the age-specific background incidence risks for the study populations. A further
32 assumption is that the lifetime risk of RCC up to 85 years is an adequate approximation to the
33 full lifetime risk, which is what was used for the other two tumor types. The first calculation,
34 based on the results of the meta-analyses for the three tumor types, has the advantage of being
35 based on a large data set, incorporating data from many different studies. However, this
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-138 DRAFT—DO NOT CITE OR QUOTE
-------
1 calculation relies on a number of additional assumptions. First, it is assumed that the RRps from
2 the meta-analyses, which are based on different groups of studies, reflect similar overall TCE
3 exposures, i.e., that the overall TCE exposures are similar across the different groups of studies
4 that went into the different meta-analyses for the three tumor types. Second, it is assumed that
5 the RRps, which incorporate RR estimates for both mortality and incidence, represent good
6 estimates for cancer incidence risk from TCE exposure. In addition, it is assumed that the RRp
7 for kidney cancer, for which RCC estimates from individual studies were used when available, is
8 a good estimate for the overall RR for RCC and that the RRp estimate for lymphoma, for which
9 different studies used different classification schemes, is a good estimate for the overall RR for
10 NHL. The second calculation, based on the results of the Raaschou-Nielsen et al. (2003) study,
11 the largest single study with RR estimates for all three tumor types, has the advantage of having
12 RR estimates that are directly comparable. In addition, the Raaschou-Nielsen et al. study
13 provided data for the precise tumor types of interest for the calculation, i.e., RCC, NHL, and
14 liver (and biliary) cancer.
15 The input data and results of the calculations are presented in Table 5-39. The value for
16 the ratio of the sum of the extra risks to the extra risk for RCC alone was 3.83 in calculation #1
17 and 4.36 in calculation #2, which together suggest that 4 is a reasonable factor to use to adjust
18 the inhalation unit risk estimate based on RCC for multiple sites to obtain a total cancer unit risk
19 estimate. Using this factor to adjust the unit risk estimate based on RCCs entails the further
20 fundamental assumption that the dose-response relationships for the other two tumor types (NHL
21 and liver cancer) are similarly linear, i.e., that the relative potencies are roughly maintained at
22 lower exposure levels. This assumption is consistent with U.S. EPA's Guidelines for
23 Carcinogen Risk Assessment (U.S. EPA, 2005a), which recommends low-dose linear
24 extrapolation in the absence of sufficient evidence to support a nonlinear MOA.
25 Applying the factor of four to the lifetime extra RCC incidence unit risk estimate of
26 5.49 x 10"3 per ppm (1.0 x 10"6 per |ig/m3) of continuous TCE exposure yields a cancer unit risk
27 estimate of 2.2 x 10"2 per ppm (4.1 x 10"6 per |ig/m3). Table 5-39 also presents calculations for
28 just kidney and lymphoma extra risks combined, because the strongest human evidence is for
29 those two tumor types. For those two tumor types, the calculations support a factor of three.
30 Applying this factor to the RCC unit risk estimate yields an estimate of 1.6 x 10"2 per ppm,
31 which results in the same estimate as for the three tumor types combined when finally rounded to
32 one significant figure, i.e., 2 x 10"2 per ppm (or 3 x 10"6 per |ig/m3, which is still similar to the
33 three-tumor-type estimate in those units).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-139 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-39. Relative contributions to extra risk for cancer incidence from
TCE exposure for multiple tumor types
RR
Ro
Rx
Extra risk
Ratio to
kidney value
Calculation #1: using RR estimates from the meta-analyses
Kidney (RCC)
Lymphoma (NHL)
Liver (& biliary) cancer
Kidney + NHL only
1.25
1.23
1.33
0.0107
0.0202
0.0066
0.01338
0.02485
0.008778
sum
sum
0.002704
0.004742
0.002192
0.01077
0.008379
1
1.75
0.81
3.56
2.75
Calculation #2: using RR estimates from Rasschou-Nielsen et al. (2003)
Kidney (RCC)
Lymphoma (NHL)
Liver (& biliary) cancer
Kidney + NHL only
1.20
1.24
1.35
0.0107
0.0202
0.0066
0.01284
0.02505
0.008910
sum
sum
0.002163
0.004948
0.002325
0.009436
0.007111
1
2.29
1.07
4.36
3.29
4
5
6 In addition to the uncertainties in the underlying RCC estimate, there are uncertainties
7 related to the assumptions inherent in these calculations for adjusting to multiple sites, as
8 detailed above. Nonetheless, the fact that the calculations based on two different data sets
9 yielded comparable values for the adjustment factor provides more robust support for the use of
10 the factor of four. Additional uncertainties pertain to the weight of evidence supporting the
11 association of TCE exposure with increased risk of cancer for the three tumor types. As
12 discussed in Section 4.11.2, it was found that the weight of evidence for kidney cancer was
13 sufficient to classify TCE as "carcinogenic to humans." It was also concluded that there was
14 strong evidence that TCE causes NHL as well, although the evidence for liver cancer was more
15 limited. In addition, the rodent studies demonstrate clear evidence of multisite carcinogenicity,
16 with tumor types including those for which associations with TCE exposure are observed in
17 human studies, i.e., liver and kidney cancers and lymphomas. Overall, the evidence was found
18 to be sufficiently persuasive to support the use of the adjustment factor of four based on these
19 three tumor types, resulting in a cancer inhalation unit risk estimate of 2.2 x 10"2 per ppm (4.1 x
20 10"6 per |ig/m3). Alternatively, if one were to use the factor based only on the two tumor types
21 with the strongest evidence, the cancer inhalation unit risk estimate would be only slightly
22 reduced (25%).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-140 DRAFT—DO NOT CITE OR QUOTE
-------
1 5.2.2.3. Route-to-Route Extrapolation Using Physiologically Based Pharmacokinetic (PBPK)
2 Model
3 Route-to-route extrapolation of the inhalation unit risk estimate was performed using the
4 PBPK model described in Section 3.5. The (partial) unit risk estimates for lymphoma and liver
5 cancer were derived as for the total cancer inhalation unit risk estimate in Section 5.2.2.2 above,
6 except that the ratios of extra risk for the individual tumor types relative to kidney cancer were
7 used as adjustment factors rather than the ratio of the sum. As presented in Table 5-39, for
8 lymphoma, the ratios from the two different calculations were 1.75 and 2.29, so a factor of two
9 was used; for liver cancer, the ratios were 0.81 and 1.07, so a factor of one was used. With the
10 ratio of one for kidney cancer itself, the combined adjustment factor is four, consistent with the
11 factor of four used to estimate the total cancer unit risk from the multiple sites in Section 5.2.2.2.
12 Because different internal dose metrics are preferred for each target tissue site, a separate
13 route-to-route extrapolation was performed for each site-specific unit risk estimate calculated in
14 Sections 5.2.2.1 and 5.2.2.2. As shown in Figure 5-7, the approach taken to apply the human
15 PBPK model in the low-dose range where external and internal doses are linearly related to
16 derive a conversion that is the ratio of internal dose per mg/kg/d to internal dose per ppm. The
17 expected value of the population mean for this conversion factor (in ppm per mg/kg/d) was used
18 to extrapolate each inhalation unit risk in units of risk per ppm to an oral slope factor in units of
19 risk per mg/kg/d. Note that this conversion is the mean of the ratio of internal dose predictions,
20 and is not the same as taking the ratio of the mean of internal dose predictions in Table 5-28.34
34For route-to-route extrapolation based on dose-response analysis performed on internal dose, as is the case for
rodent bioassays, it would be appropriate to use the values in Table 5-28 to first "unconvert" the unit risk based on
one route, and then recover! to a unit risk based on the other route.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-141 DRAFT—DO NOT CITE OR QUOTE
-------
istribution
[internal dos
per mg/kg/d]/
[internal dos
per ppm]
[distribution (separate
ncertainty and variability)
IpJop ulation
imean
fixed*.
nean
Expected
site-specific
human
cancer unit risk
per mg/kg-d
2
3
4
5
6
7
Figure 5-7. Flow-chart for route-to-route extrapolation of human site-
specific cancer inhalation unit risks to oral slope factors. Square nodes
indicate point values, circle nodes indicate distributions, and the inverted triangle
indicates a (deterministic) functional relationship.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-142 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-40 shows the results of this route-to-route extrapolation for the "primary" and
2 "alternative" dose metrics. For reference, route-to-route extrapolation based on total intake (i.e.,
3 ventilation rate x air concentration = oral dose x BW) using the parameters in the PBPK model
4 would yield an expected population average conversion of 0.95 ppm per mg/kg/d. For
5 TotMetabBW34, TotOxMetabBW34, and AMetLivlBW34, the conversion is 2.0-2.8 ppm per
6 mg/kg/d, greater than that based on intake. This is because of the greater metabolic first pass in
7 the liver, which leads to a higher percentage of intake being metabolized via oral exposure
8 relative to inhalation exposure for the same intake. Conversely, for the AUC in blood, the
9 conversion is 0.14 ppm per mg/kg/d, less than that based on intake—the greater first pass in the
10 liver means lower blood levels of parent compound via oral exposure relative to inhalation for
11 the same intake. The conversion for the primary dose metric for the kidney,
12 ABioactDCVCBW34, is 1.7 ppm per mg/kg/d, less than that for total, oxidative, or liver
13 oxidative metabolism. This is because the majority of metabolism in first pass through the liver
14 is via oxidation, whereas with inhalation exposure, more parent compound reaches the kidney, in
15 which metabolism is via GSH conjugation.
16 When one sums the oral slope factor estimates based on the primary (preferred) dose
17 metrics for the 3 individual tumor types shown in Table 5-40, the resulting total cancer oral unit
18 risk (slope factor) estimate is 4.63 x io~2 per mg/kg/d. In the case of the oral route-extrapolated
19 results, the ratio of the risk estimate for the three tumor types combined to the risk estimate for
20 kidney cancer alone is 5.0. This value differs from the factor of four used for the total cancer
21 inhalation unit risk estimate because of the different dose metrics used for the different tumor
22 types when the route-to-route extrapolation is performed. If only the kidney cancer and NHL
23 results, for which the evidence is strongest, were combined, the resulting total cancer oral unit
24 risk estimate would be 3.08 x 10"2 per mg/kg/d, and the ratio of this risk estimate to that for
25 ki dney cancer al one woul d b e 3.3.
26 If one were to use some of the risk estimates based on alternative dose metrics in
27 Table 5-40, the total cancer risk estimate would vary depending on for which tumor type(s) an
28 alternative metric was used. The most extreme difference would occur when the alternative
29 metric is used for NHL and liver tumors; in that case, the resulting total cancer oral unit risk
30 estimate would be 2.20 x 10"2 per mg/kg/d, and the ratio of this risk estimate to that for kidney
31 cancer alone (based on the primary dose metric of ABioactDCVCBW34) would be 2.4.
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-143 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-40. Route-to-route extrapolation of site-specific inhalation unit risks
to oral slope factors
Inhalation unit risk
(risk per ppm)
Primary dose metric
ppm per mg/kg/db
Oral slope factor
(risk per mg/kg/d)
Alternative dose metric
ppm per mg/kg/db
Oral slope factor
(risk per mg/kg/d)
Kidney
5.49 x 10'3
ABioactDCVCBW34a
1.70
9.33 x 10'3
TotMetabBW34
1.97
1.08 x 10'2
NHL
1.09 x 10'2
TotMetabBW34
1.97
2.15 x 10'2
AUCCBld
0.137
1.49 x 10'3
Liver
5.49 x 10'3
AMetLivlBW34
2.82
1.55 x 10'2
TotOxMetabBW34
2.04
1.12 x 10'2
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
aThe AMetGSHBW34 dose metric gives the same route-to-route conversion because there is no route dependence in
the pathway between GSH conjugation and DCVC bioactivation.
bAverage of expected population mean of males and females. Male and female estimates differed by <1% for
ABioactDCVCBW34; TotMetabBW34, AMetLivlBW34, and TotOxMetabBW34, and <15% for AUCCBld.
Uncertainty on the population mean route-to-route conversion, expressed as the ratio between the 97.5% quantile
the 2.5% quantile, is about 2.6-fold for ABioactDCVCBW34, 1.5-fold for TotMetabBW34, AMetLivlBW34, and
TotOxMetabBW34, and about 3.4-fold for AUCCBld.
The uncertainties in these conversions are relatively modest. As discussed in the note to
Table 5-40, the 95% confidence range for the route-to-route conversions at its greatest spans
3.4-fold. The greatest uncertainty is in the selection of the dose metric for NHL, since the use of
the alternative dose metric of AUCCBld yields a converted oral slope factor that is 14-fold lower
than that using the primary dose metric of TotMetabBW34. However, for the other two tumor
sites, the range of conversions is tighter, and lies within 3-fold of the conversion based solely on
intake.
5.2.3. Summary of Unit Risk Estimates
5.2.3.1. Inhalation Unit Risk Estimate
The inhalation unit risk for TCE is defined as a plausible upper bound lifetime extra risk
of cancer from chronic inhalation of TCE per unit of air concentration. The preferred estimate of
the inhalation unit risk for TCE is 2.20 x 10~2 per ppm (2 x 10~2 per ppm [4 x 10~6 per ug/m3]
rounded to 1 significant figure), based on human kidney cancer risks reported by Charbotel et al.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-144 DRAFT—DO NOT CITE OR QUOTE
-------
1 (2006) and adjusted for potential risk for tumors at multiple sites. This estimate is based on
2 good-quality human data, thus avoiding the uncertainties inherent in interspecies extrapolation.
3 This value is supported by inhalation unit risk estimates from multiple rodent bioassays,
4 the most sensitive of which range from 1 x 10~2 to 2 x 10"1 per ppm [2 x 10~6 to
5 3 x 10~5 per jig/m3]. From the inhalation bioassays selected for analysis in Section 5.2.1.1, and
6 using the preferred PBPK model-based dose metrics, the inhalation unit risk estimate for the
7 most sensitive sex/species is 8 x 10~2 per ppm [2 x 10~5 per |ig/m3], based on kidney adenomas
8 and carcinomas reported by Maltoni et al. (1986) for male Sprague-Dawley rats. Leukemias and
9 Leydig cell tumors were also increased in these rats, and, although a combined analysis for these
10 tumor types that incorporated the different site-specific preferred dose metrics was not
11 performed, the result of such an analysis is expected to be similar, about 9 x 10~2 per ppm
12 [2 x 10~5 per |ig/m3]. The next most sensitive sex/species from the inhalation bioassays is the
13 female mouse, for which lymphomas were reported by Henschler et al. (1980); these data yield a
14 unit risk estimate of 1.0 x 10~2 per ppm [2 x 10~6 per |ig/m3]. In addition, the 90% confidence
15 intervals reported in Table 5-34 for male rat kidney tumors from Maltoni et al. (1986) and female
16 mouse lymphomas from Henschler et al. (1980), derived from the quantitative analysis of PBPK
17 model uncertainty, both included the estimate based on human data of 2 x 10~2 per ppm.
18 Furthermore, PBPK model-based route-to-route extrapolation of the results for the most sensitive
19 sex/species from the oral bioassays, kidney tumors in male Osborne-Mendel rats and testicular
20 tumors in Marshall rats (NTP, 1988), leads to inhalation unit risk estimates of 2 x 10"1 per ppm
21 [3 x 10~5 per |ig/m3] and 4 x 10~2 per ppm [8 x 10~6 per |ig/m3], respectively, with the preferred
22 estimate based on human data falling within the route-to-route extrapolation of the 90%
23 confidence intervals reported in Table 5-35.35 Finally, for all these estimates, the ratios of
24 BMDs to the BMDLs did not exceed a value of 3, indicating that the uncertainties in the dose-
25 response modeling for determining the POD in the observable range are small.
26 Although there are uncertainties in these various estimates, as discussed in
27 Sections 5.2.1.4, 5.2.2.1.3, and 5.2.2.2, confidence in the proposed inhalation unit risk estimate
28 of 2 x 10~2 per ppm [4 x 10~6 per |ig/m3], based on human kidney cancer risks reported by
35For oral-to-inhalation extrapolation of NTP (1988) male rat kidney tumors, the unit risk estimate of 2.5 x 10"1 per
mg/kg/d using the ABioactDCVCBW34 dose metric, from Table 5-30, is divided by the average male and female
internal doses at 0.001 mg/kg/d, (0.00504/0.001), and then multiplied by the average male and female internal doses
at 0.001 ppm, (0.00324/0.001), both from Table 5-28, to yield a unit risk of 1.6 x 10"1 [3.0 x 10~5 per ug/m3]. For
oral-to-inhalation extrapolation of NTP (1988) male rat testicular tumors, the unit risk estimate of 7.1 x 10"2 per
mg/kg/d using the TotMetabBW34 dose metric, from Table 5-30, is divided by the male internal dose at
0.001 mg/kg/d, (0.0192/0.001), and then multiplied by the male internal doses at 0.001 ppm, (0.0118/0.001), both
from Table 5-28, to yield a unit risk of 4.4 x 10"2 [8.1 x 10^ per ug/m3].
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-145 DRAFT—DO NOT CITE OR QUOTE
-------
1 Charbotel et al. (2006) and adjusted for potential risk for tumors at multiple sites (as discussed in
2 Section 5.2.2.2), is further increased by the similarity of this estimate to estimates based on
3 multiple rodent data sets.
4
5 5.2.3.2. Oral Unit Risk Estimate
6 The oral unit risk (or slope factor) for TCE is defined as a plausible upper bound lifetime
7 extra risk of cancer from chronic ingestion of TCE per mg/kg/d oral dose. The preferred
8 estimate of the oral unit risk is 4.63 x 10~2 per mg/kg/d (5 x 10~2 per mg/kg/d rounded to
9 1 significant figure), resulting from PBPK model-based route-to-route extrapolation of the
10 inhalation unit risk estimate based on the human kidney cancer risks reported in Charbotel et al.
11 (2006) and adjusted for potential risk for tumors at multiple sites. This estimate is based on
12 good-quality human data, thus avoiding uncertainties inherent in interspecies extrapolation. In
13 addition, uncertainty in the PBPK model-based route-to-route extrapolation is relatively low
14 (Chiu and White, 2006; Chiu, 2006). In this particular case, extrapolation using different dose
15 metrics yielded expected population mean risks within about a 2-fold range, and, for any
16 particular dose metric, the 95% confidence interval for the extrapolated population mean risks
17 for each site spanned a range of no more than about 3-fold.
18 This value is supported by oral unit risk estimates from multiple rodent bioassays, the
19 most sensitive of which range from 3 x 10~2 to 3 x 10"1 per mg/kg/d. From the oral bioassays
20 selected for analysis in Section 5.2.1.1, and using the preferred PBPK model-based dose metrics,
21 the oral unit risk estimate for the most sensitive sex/species is 3 x 10"1 per mg/kg/d, based on
22 kidney tumors in male Osborne-Mendel rats (NTP, 1988). The oral unit risk estimate for
23 testicular tumors in male Marshall rats (NTP, 1988) is somewhat lower at 7 x 10~2 per mg/kg/d.
24 The next most sensitive sex/species result from the oral studies is for male mouse liver tumors
25 (NCI, 1976), with an oral unit risk estimate of 3 x 10~2 per mg/kg/d. In addition, the 90%
26 confidence intervals reported in Table 5-35 for male Osborne-Mendel rat kidney tumors (NTP,
27 1988), male F344 rat kidney tumors (NTP, 1990), and male Marshall rat testicular tumors (NTP,
28 1988), derived from the quantitative analysis of PBPK model uncertainty, all included the
29 estimate based on human data of 5 x 10~2 per mg/kg/d, while the upper 95% confidence bound
30 for male mouse liver tumors from NCI (1976) was slightly below this value at 4 x 10~2 per
31 mg/kg/d. Furthermore, PBPK model-based route-to-route extrapolation of the most sensitive
32 endpoint from the inhalation bioassays, male rat kidney tumors from Maltoni et al. (1986), leads
33 to an oral unit risk estimate of 1 x 10"1 per mg/kg/d, with the preferred estimate based on human
34 data falling within the route-to-route extrapolation of the 90% confidence interval reported in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-146 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table 5-34.36 Finally, for all these estimates, the ratios of BMDs to the BMDLs did not exceed a
2 value of 3, indicating that the uncertainties in the dose-response modeling for determining the
3 POD in the observable range are small.
4 Although there are uncertainties in these various estimates, as discussed in
5 Sections 5.2.1.4, 5.2.2.1.3, 5.2.2.2, and 5.2.2.3, confidence in the proposed oral unit risk estimate
6 of 5 x 1CT2 per mg/kg/d, resulting from PBPK model-based route-to-route extrapolation of the
7 inhalation unit risk estimate based on the human kidney cancer risks reported in
8 Charbotel et al. (2006) and adjusted for potential risk for tumors at multiple sites (as discussed in
9 Section 5.2.2.2), is further increased by the similarity of this estimate to estimates based on
10 multiple rodent data sets.
11
12 5.2.3.3. Application of Age-Dependent Adjustment Factors
13 When there is sufficient weight of evidence to conclude that a carcinogen operates
14 through a mutagenic MO A, and in the absence of chemical-specific data on age-specific
15 susceptibility, U.S. EPA's Supplemental Guidance for Assessing Susceptibility from Early-Life
16 Exposure to Carcinogens (U.S. EPA, 2005b) advises that increased early-life susceptibility be
17 assumed and recommends that default age-dependent adjustment factors (ADAFs) be applied to
18 adjust for this potential increased susceptibility from early-life exposure. As discussed in
19 Section 4.4, there is sufficient evidence to conclude that a mutagenic MOA is operative for TCE-
20 induced kidney tumors. In addition, as described in Section 4.10, TCE-specific data are
21 inadequate for quantification of early-life susceptibility to TCE carcinogenicity. Therefore, as
22 recommended in the Supplemental Guidance, the default ADAFs are applied.
23 See the Supplemental Guidance for detailed information on the general application of
24 these adjustment factors. In brief, the Supplemental Guidance establishes ADAFs for three
25 specific age groups. The current ADAFs and their age groupings are 10 for <2 years, 3 for 2 to
26 <16 years, and 1 for 16 years and above (U.S. EPA, 2005b). For risk assessments based on
27 specific exposure assessments, the 10-fold and 3-fold adjustments to the unit risk estimates are to
28 be combined with age-specific exposure estimates when estimating cancer risks from early-life
29 (<16-years-of-age) exposure. The ADAFs and their age groups may be revised over time. The
30 most current information on the application of ADAFs for cancer risk assessment can be found at
31 www.epa.gov/cancerguidelines.
36For the Maltoni et al. (1986) male rat kidney tumors, the unit risk estimate of 8.3 x 10~2 per ppm using the
ABioactDCVCBW34 dose metric, from Table 5-29, is divided by the average male and female internal doses at
0.001 ppm, (0.00324/0.001) and then multiplied by the average male and female internal doses at 0.001 mg/kg/d,
(0.00504/0.001), both from Table 5-28, to yield a unit risk of 1.3 x 10"1 per mg/kg/d.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-147 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
In the case of TCE, the inhalation and oral unit risk estimates reflect lifetime risk for
cancer at multiple sites, and a mutagenic MOA has been established for one of these sites, the
kidney. The following subsections illustrate how one might apply the default ADAFs to the
kidney-cancer component of the inhalation and oral unit risk estimates for TCE. These are
sample calculations, and individual risk assessors should use exposure-related parameters (e.g.,
age-specific water ingestion rates) that are appropriate for their particular risk assessment
applications.
In addition to the uncertainties discussed above for the inhalation and oral total cancer
unit risk estimates, there are uncertainties in the application of ADAFs to adjust for potential
increased early-life susceptibility. For one thing, the adjustment is made only for the kidney-
cancer component of total cancer risk because that is the tumor type for which the weight of
evidence was sufficient to conclude that TCE-induced carcinogenesis operates through a
mutagenic MOA. However, it may be that TCE operates through a mutagenic MOA for other
tumor types as well or that it operates through other MO As that might also convey increased
early-life susceptibility. Additionally, the ADAFs are general default factors, and it is uncertain
to what extent they reflect increased early-life susceptibility for exposure to TCE, if increased
early-life susceptibility occurs.
5.2.3.3.1. Example application of age-dependent adjustment factors (ADAFs) for inhalation
exposures. For inhalation exposures, assuming ppm equivalence across age groups, i.e.,
equivalent risk from equivalent exposure levels, independent of body size, the calculation is
fairly straightforward. The ADAF-adjusted lifetime cancer unit risk estimate for kidney cancer
alone is calculated as follows:
kidney cancer risk from exposure to constant TCE exposure level of
1 ug/m3 from ages 0-70:
Age group
0-<2 years
2-<16 years
>16 years
ADAF
10
O
1
unit risk
(per ug/m3)
1.0 x 10'6
1.0 x 10'6
1.0 x 10'6
exposure
cone. (ug/m3)
1
1
1
duration partial
adjustment risk
2 years/70 years 2.9 x 10
14 years/70 years 6.0 x 10
54 years/70 years 7.7 x 10
total risk = 1.7 x 10
Note that the partial risk for each age group is the product of the values in columns 2-5 [e.g.,
10 x (1.0 x 10"6) x 1 x 2/70 = 2.9 x 10"7], and the total risk is the sum of the partial risks. This
70-year risk estimate for a constant exposure of 1 ug/m3 is equivalent to a lifetime unit risk of
1.7 x 10"6 per ug/m3, adjusted for early-life susceptibility, assuming a 70-year lifetime and
constant exposure across age groups.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-148 DRAFT—DO NOT CITE OR QUOTE
-------
1 In other words, the lifetime unit risk estimate for kidney cancer alone, adjusted for
2 potential increased early-life susceptibility is 1.7-times the unadjusted unit risk estimate. Adding
3 a 3-fold factor to the unadjusted unit risk estimate to account for potential risk at multiple sites
4 ("1-fold" of the factor of four for multiple sites is already included in the 1.7-times adjustment
5 for early-life susceptibility) yields a total adjustment factor of 4.7. Applying a factor of 4.7 to
6 the unit risk estimate based on kidney cancer alone results in a total cancer unit risk estimate of
7 2.6 x 10"2 per ppm (4.8 x 10"6 per ug/m3) of constant lifetime TCE exposure, adjusted for
8 potential early-life susceptibility.
9 Note that the above calculation for adjusting the ADAF-adjusted lifetime unit risk
10 estimate for multiple sites is equivalent to adjusting each ADAF by adding a factor of three and
11 applying those factors as age-specific adjustment factors for both early-life susceptibility and
12 multiple sites to the unadjusted kidney cancer unit risk estimate (i.e., 13, 6, and 4 for <2 years,
13 2 to <16 years, and >16 years, respectively). The total cancer risk estimate of 4.7 x 10"6 per
14 ug/m3, adjusted for potential increased early-life susceptibility, derived below for a constant
15 exposure of 1 ug/m3 differs from the unit risk estimate of 4.8 x 10"6 per ug/m3 presented above
16 only because of round-off error.
17
18 total cancer risk from exposure to constant TCE exposure level of
19 1 ug/m3 from ages 0-70
20
21
22 combined
23 adjustment unit risk exposure duration partial
24 Age group factor (per ug/m3) cone (ug/m3) adjustment risk
25 0-<2 years 13 1.0 x 10'6 1 2 years/70 years 3.7 x 10'7
26 2-<16 years 6 1.0 x 10'6 1 14 years/70 years 1.2 x 10'6
27 >16 years 4 1.0 x 10'6 1 54 years/70 years 3.1 x 10'6
28 total risk = 4.7 x 10'6
29
30 Note that the partial risk for each age group is the product of the values in columns 2-5 [e.g.,
31 13 x (1.0 x 10"6) x 1 x 2/70 = 3.7 x 10"7], and the total risk is the sum of the partial risks. This
32 70-year risk estimate for a constant exposure of 1 ug/m3 is equivalent to a lifetime unit risk of
33 4.7 x 10"6 per ug/m3, adjusted for early-life susceptibility, assuming a 70-year lifetime and
34 constant exposure across age groups.
35
36 This total cancer unit risk estimate of 2.6 x 10"2 per ppm (4.8 x 10"6 per ug/m3), adjusted
37 for potential increased early-life susceptibility, is only minimally (17.5%) increased over the
38 unadjusted total cancer unit risk estimate because the kidney cancer risk estimate that gets
39 adjusted for potential increased early-life susceptibility is only part of the total cancer risk
40 estimate. Thus, foregoing the ADAF adjustment in the case of full lifetime calculations will not
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-149 DRAFT—DO NOT CITE OR QUOTE
-------
1 seriously impact the resulting risk estimate. For less-than-lifetime exposure calculations, the
2 impact of applying the ADAFs will increase as the proportion of time at older ages decreases.
3 The maximum impact will be when exposure is for only the first 2 years of life, in which case the
4 partial lifetime total cancer risk estimate for exposure to 1 ug/m3 adjusted for potential increased
5 early-life susceptibility is 13 x (1 ug/m3) x (1.0 x 1CT6 per ug/m3) x (2/70), or 3.7 x 10"7, which
6 is over 3 times greater than the unadjusted partial lifetime total cancer risk estimate for exposure
7 to 1 ug/m3 of 4 x (1 ug/m3) x (1.0 x 1(T6 per ug/m3) x (2/70), or 1.1 x lO'7.
8
9 5.2.3.3.2. Example application ofage-dpendent adjustment factors (ADAFs) for oral
10 exposures. For oral exposures, the calculation of risk estimates adjusted for potential increased
11 early-life susceptibility is complicated by the fact that for a constant exposure level, e.g., a
12 constant concentration of TCE in drinking water, doses will vary by age because of different age-
13 specific uptake rates, e.g., drinking water consumption rates. Different U.S. EPA Program or
14 Regional Offices may have different default age-specific uptake rates that they use for risk
15 assessments for specific exposure scenarios, and the calculations presented below are merely to
16 illustrate the general approach to applying ADAFs for oral TCE exposures, using lifetime
17 exposure to 1 ug/L of TCE in drinking water as an example.
18 Age-specific water ingestion rates in L/kg/day were taken from U.S. EPA's Child-
19 Specific Exposure Factors Handbook (U.S. EPA, 2008). Values for the 90th percentile were
20 taken from Table 3-19 (consumers-only estimates of combined direct and indirect water
21 ingestion from community water). The 90th percentile was based on the policy in the U.S. EPA
22 Office of Water for determining risk through direct and indirect consumption of drinking water.
23 Community water was used in the illustration because U.S. EPA only regulates community water
24 sources and not private wells and cisterns or bottled water. Data for "consumers only" (i.e.,
25 excluding individuals who did not ingest community water) were used because formula-fed
26 infants (as opposed to breast-fed infants, who consume very little community water), children,
27 and young adolescents are often the population of concern with respect to water consumption.
28 For the 16+ age group, the standard default rate for adults was used (i.e., 2 L/day -=- 70 kg, or
29 0.029 L/kg/day) (U.S. EPA, 1997, page 3-1), which is identical to the 90th percentile for the 18 to
30 <21 age group. For the purposes of this illustration, the different age-specific rates were
31 collapsed into the same age groupings as the ADAFs using a time-weighted averaging. These
32 age-specific water ingestion rates are presented in Table 5-41.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-150 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table 5-41. Estimates of age-specific water ingestion rates (90 percentile)"
3
4
5
6
9
10
11
12
13
14
15
16
17
18
Age
Birth to <1 month
1 to <3 months
3 to <6 months
6 to <12 months
1 to <2 years
0 to <2 years
2 to <3 years
3 to <6 years
6 to <11 years
11 to <16 years
2 to <16 years
>16 yearsb
Ingestion rate (L/kg/d)
0.238
0.228
0.148
0.112
0.056
0.103
0.052
0.049
0.035
0.026
0.036
0.029
"Values in bold are time-weighted averages corresponding to the ADAF age groupings.
'Tor this age grouping, the standard adult default rate is presented.
For simplicity, the adjustments for potential cancer risk at multiple sites and for potential
increased early-life susceptibility are made simultaneously using age-specific combined
adjustment factors, as was done in the second (equivalent) lifetime risk calculation for inhalation
exposures in Section 5.2.3.3.1. In the case of oral cancer risk, however, the ratio for total risk
relative to kidney cancer risk was about five (see Section 5.2.2.3); thus, a factor of four is added
to each of the ADAFs to account for risk of tumor types other than kidney cancer. The
calculations for the combined adjustment are shown in Table 5-42.
Because the TCE intake is not constant across age groups, one does not calculate a
lifetime unit risk estimate in terms of risk per mg/kg/d adjusted for potential increased early-life
susceptibility. One could calculate a unit risk estimate for TCE in drinking water in terms of
ug/L from the result in Table 5-42, but this is not something that is commonly reported, and it is
dependent on the water ingestion rates used.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-151 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
Table 5-42. Sample calculation for total lifetime cancer risk based on the
kidney unit risk estimate, adjusting for potential risk at multiple sites and for
potential increased early-life susceptibility and assuming a constant lifetime
exposure to 1 ug/mL of TCE in drinking water
Age group
(years)
0 to <2 years
2 to <16 years
>16 years
Combined
adjustment
factor
14
7
5
Unit risk3
(per
mg/kg/d)
9.33 x lO'3
9.33 x lO'3
9.33 x lO'3
Exposure
conc.b
(mg/L)
0.001
0.001
0.001
Water
ingestion
rate
(L/kg/d)
0.103
0.036
0.029
Duration
adjustment
(fraction of
years)
2/70
14/70
54/70
Total lifetime riskd
Partial riskc
3.8 x 1Q-7
4.7 x lO'7
1.04 x lO'6
1.9 x 10 6
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
aUnit risk estimate for kidney cancer based on primary dose metric, from Table 5-40.
bFrom Table 5-41.
°The partial risk for each tumor type is the product of the values in columns 2-6.
dThe total lifetime risk estimate is the sum of the partial risks.
As with the adjusted inhalation risk estimate in Section 5.2.3.3.1, the lifetime total cancer
risk estimate of 1.9 x 10"6 calculated for lifetime exposure to 1 ug/L of TCE in drinking water
adjusted for potential increased early-life susceptibility is only minimally (25%) increased over
the unadjusted total cancer unit risk estimate. (This calculation is not shown, but if one uses just
the factor of five for potential cancer risk at multiple sites for each of the age groups in
Table 5-42, the resulting total lifetime risk estimate is 1.5 x 10"6.) Unlike with inhalation
exposure under the assumption of ppm equivalence, the oral intake rates are higher in the
potentially more susceptible younger age groups. This would tend to yield a larger relative
impact of adjusting for potential increased early-life susceptibility for oral risk estimates
compared to inhalation risk estimates. In the case of TCE, however, this impact is partially
offset by the lesser proportion of the total oral cancer risk that is accounted for by the kidney
cancer risk, which is the component of total risk that is being adjusted for potential increased
early-life susceptibility, based on the primary dose metrics (1/5 vs. 1/4 for inhalation). Thus, as
with lifetime inhalation risk, foregoing the ADAF adjustment in the case of full lifetime
calculations will not seriously impact the resulting risk estimate. For less-than-lifetime exposure
calculations, the impact of applying the ADAFs will increase as the proportion of time at older
ages decreases. The maximum impact will be when exposure is for only the first 2 years of life,
in which case the partial lifetime total cancer risk estimate for exposure to 1 ug/L adjusted for
potential increased early-life susceptibility is 3.8 x 10"7 (from Table 5-42), which is almost 3
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-152 DRAFT—DO NOT CITE OR QUOTE
-------
1 times greater than the unadjusted partial lifetime total cancer risk estimate for exposure to 1 ug/L
2 of 5 x (0.001 mg/L) x (0.103 L/kg/day) x (9.33 x 10~3 per mg/kg/d) x (2/70), or 1.4 x 10'7.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 5-153 DRAFT—DO NOT CITE OR QUOTE
-------
1 6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF
2 HAZARD AND DOSE RESPONSE
O
4
5 6.1. HUMAN HAZARD POTENTIAL
6 This section summarizes the human hazard potential for trichloroethylene (TCE). For
7 extensive discussions and references, see Chapter 2 for Exposure, Chapter 3 for toxicokinetics
8 and physiologically-based pharmacokinetic (PBPK) modeling, and Sections 4.1-4.9 for the
9 epidemiologic and experimental studies of TCE noncancer and cancer toxicity. Section 4.10
10 summarizes information on susceptibility, and Section 4.11 provides a more detailed summary
11 and references for noncancer toxicity and carcinogenicity.
12
13 6.1.1. Exposure (see Chapter 2)
14 TCE is a volatile compound with moderate water solubility. Most TCE produced today
15 is used for metal degreasing. The highest environmental releases are to the air. Ambient air
16 monitoring data suggests that mean levels have remained fairly constant since 1999 at about
17 0.3 ug/m3 (0.06 ppb). As discussed in Chapter 2, in 2006, ambient air monitors (n = 258) had
18 annual means ranging from 0.03 to 7.73 ug/m3 with a median of 0.13 and an overall average of
19 0.23 ug/m3. Indoor levels are commonly three or more times higher than outdoor levels due to
20 releases from building materials and consumer products. TCE is among the most common
21 groundwater contaminants. The median level of TCE in groundwater, based on a large survey
22 by the United States Geological Survey for 1985-2001, is 0.15 ug/L. It has also been detected in
23 a wide variety of foods in the 1-100 ug/kg range. None of the environmental sampling has been
24 done using statistically based national surveys. However, a substantial amount of air and
25 groundwater data have been collected allowing reasonably well supported estimates of typical
26 daily intakes by the general population: inhalation—13 ug/day and water ingestion—0.2 ug/day.
27 The limited food data suggests an intake of about 5 ug/day, but this must be considered
28 preliminary.
29 Much higher exposures have occurred to various occupational groups. For example, past
30 studies of aircraft workers have shown short term peak exposures in the hundreds of ppm
31 (>500,000 ug/m3) and long term exposures in the low tens of ppm (>50,000 ug/m3).
32 Occupational exposures have likely decreased in recent years due to better release controls and
33 improvements in worker protection.
34 Exposure to a variety of TCE related compounds, which include metabolites of TCE and
35 other parent compounds that produce similar metabolites, can alter or enhance TCE metabolism
36 and toxicity by generating higher internal metabolite concentrations than would result from TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-1 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure by itself. Available estimates suggest that exposures to most of these TCE-related
2 compounds are comparable to or greater than TCE itself.
O
4 6.1.2. Toxicokinetics and Physiologically-Based Pharmacokinetic (PBPK) Modeling (see
5 Chapter 3 and Appendix A)
6 TCE is a lipophilic compound that readily crosses biological membranes. Exposures may
7 occur via the oral, dermal, and inhalation route, with evidence for systemic availability from
8 each route. TCE can also be transferred transplacentally and through breast milk ingestion. TCE
9 is rapidly and nearly completely absorbed from the gut following oral administration, and animal
10 studies indicate that exposure vehicle may impact the time course of absorption: oily vehicles
11 may delay absorption whereas aqueous vehicles result in a more rapid increase in blood
12 concentrations. See Section 3.1 for additional discussion of TCE absorption.
13 Following absorption to the systemic circulation, TCE distributes from blood to solid
14 tissues by each organ's solubility. This process is mainly determined by the blood:tissue
15 partition coefficients, which are largely determined by tissue lipid content. Adipose partitioning
16 is high, so adipose tissue may serve as a reservoir for TCE, and accumulation into adipose tissue
17 may prolong internal exposures. TCE attains high concentrations relative to blood in the brain,
18 kidney, and liver—all of which are important target organs of toxicity. TCE is cleared via
19 metabolism mainly in three organs: the kidney, liver, and lungs. See Section 3.2 for additional
20 discussion of TCE distribution.
21 The metabolism of TCE is an important determinant of its toxicity. Metabolites are
22 generally thought to be responsible for toxicity—especially for the liver and kidney. Initially,
23 TCE may be oxidized via cytochrome P450 (CYP) isoforms or conjugated with glutathione by
24 glutathione S-transferase enzymes. While CYP2E1 is generally accepted to be the CYP isoform
25 most responsible for TCE oxidation, others forms may also contribute. There are conflicting
26 data as to which glutathione-S-transferase (GST) isoforms are responsible for TCE conjugation,
27 with one rat study indicating alpha-class GSTs and another rat study indicating mu and pi-class
28 GST. The balance between oxidative and conjugative metabolites generally favors the oxidative
29 pathway, especially at lower concentrations, and inhibition of CYP-dependent oxidation in vitro
30 increases glutathione (GSH) conjugation in renal preparations. However, in humans, direct
31 comparison of in vitro rates of oxidation and conjugation, as well as in vivo data on the amount
32 of the TCE GSH conjugation to dichlorovinyl glutathione in blood, support a flux through the
33 GSH pathway that may be one or more orders of magnitude greater than the <0.1% inferred from
34 excretion of GSH conjugation derived urinary mercapturates. See Section 3.3 for additional
35 discussion of TCE metabolism.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 Once absorbed, TCE is excreted primarily either in breath as unchanged TCE or carbon
2 dioxide [CO2], or in urine as metabolites. Minor pathways of elimination include excretion of
3 metabolites in saliva, sweat, and feces. Following oral administration or upon cessation of
4 inhalation exposure, exhalation of unmetabolized TCE is a major elimination pathway. Initially,
5 elimination of TCE upon cessation of inhalation exposure demonstrates a steep concentration-
6 time profile: TCE is rapidly eliminated in the minutes and hours postexposure, and then the rate
7 of elimination via exhalation decreases. Following oral or inhalation exposure, urinary
8 elimination of parent TCE is minimal, with urinary elimination of the metabolites trichloroacetic
9 acid and trichloroethanol accounting for the bulk of the absorbed dose of TCE. See Section 3.4
10 for additional discussion of TCE excretion.
11 As part of this assessment, a comprehensive Bayesian PBPK model-based analysis of the
12 population toxicokinetics of TCE and its metabolites was developed in mice, rats, and humans
13 (also reported in Chiu et al., 2009). This analysis considered a wider range of physiological,
14 chemical, in vitro., and in vivo data than any previously published analysis of TCE. The
15 toxicokinetics of the "population average," its population variability, and their uncertainties are
16 characterized and estimates of experimental variability and uncertainty are included in this
17 analysis. The experimental database included separate sets for model calibration and evaluation
18 for rats and humans; fewer data were available in mice, and were all used for model calibration.
19 The total combination of these approaches and PBPK analysis substantially supports the model
20 predictions. In addition, the approach employed yields an accurate characterization of the
21 uncertainty in metabolic pathways for which available data were sparse or relatively indirect,
22 such as GSH conjugation and respiratory tract metabolism. Key conclusions from the model
23 predictions include (1) as expected, TCE is substantially metabolized, primarily by oxidation at
24 doses below saturation; (2) GSH conjugation and subsequent bioactivation in humans appears to
25 be 10- to 100-fold greater than previously estimated; and (3) mice had the greatest rate of
26 respiratory tract oxidative metabolism compared to rats and humans. The predictions of the
27 PBPK model are subsequently used in noncancer and cancer dose-response analyses for inter-
28 and intraspecies extrapolation of toxicokinetics (see below). See Section 3.5 and Appendix A for
29 additional discussion of and details about PBPK modeling of TCE and metabolites.
30
31 6.1.3. Noncancer Toxicity
32 This section summarizes the weight of evidence for TCE noncancer toxicity. Based on
33 the available human epidemiologic data and experimental and mechanistic studies, it is
34 concluded that TCE poses a potential human health hazard for noncancer toxicity to the central
35 nervous system, the kidney, the liver, the immune system, the male reproductive system, and the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 developing fetus. The evidence is more limited for TCE toxicity to the respiratory tract and
2 female reproductive system. The conclusions pertaining to specific endpoints within these
3 tissues and systems are summarized below.
4
5 6.1.3.1. Neurological Effects (see Sections 4.3 and 4.11.1.1 and Appendix D)
6 Both human and animal studies have associated TCE exposure with effects on several
7 neurological domains. Multiple epidemiologic studies in different populations have reported
8 abnormalities in trigeminal nerve function in association with TCE exposure. Two small studies
9 did not report an association between TCE exposure and trigeminal nerve function. However,
10 statistical power was limited, exposure misclassification was possible, and, in one case, methods
11 for assessing trigeminal nerve function were not available. As a result, these studies do not
12 provide substantial evidence against a causal relationship between TCE exposure and trigeminal
13 nerve impairment. Laboratory animal studies have also demonstrated TCE-induced changes in
14 the morphology of the trigeminal nerve following short-term exposures in rats. However, one
15 study reported no significant changes in trigeminal somatosensory evoked potential in rats
16 exposed to TCE for 13 weeks. See Section 4.3.1 for additional discussion of studies of
17 alterations in nerve conduction and trigeminal nerve effects. Human chamber, occupational, and
18 geographic based/drinking water studies have consistently reported subjective symptoms such as
19 headaches, dizziness, and nausea which are suggestive of vestibular system impairments. One
20 study reported changes in nystagmus threshold (a measure of vestibular system function)
21 following an acute TCE exposure. There are only a few laboratory animal studies relevant to
22 this neurological domain, with reports of changes in nystagmus, balance, and handling reactivity.
23 See Section 4.3.3 for additional discussion of TCE effects on vestibular function. Fewer and
24 more limited epidemiologic studies are suggestive of TCE exposure being associated with
25 delayed motor function, and changes in auditory, visual, and cognitive function or performance
26 (see Sections 4.3.2, 4.3.4, 4.3.5, and 4.3.6). Acute and subchronic animal studies show
27 disruption of the auditory system, changes in visual evoked responses to patterns or flash
28 stimulus, and neurochemical and molecular changes. Animal studies suggest that while the
29 effects on the auditory system lead to permanent function impairments and histopathology,
30 effects on the visual system may be reversible with termination of exposure. Additional acute
31 studies reported structural or functional changes in hippocampus, such as decreased myelination
32 or decreased excitability of hippocampal CA1 neurons, although the relationship of these effects
33 to overall cognitive function is not established (see Section 4.3.9). An association between TCE
34 exposure and sleep changes has also been demonstrated in rats (see Section 4.3.7). Some
35 evidence exists for motor-related changes in rats/mice exposed acutely/subchronically to TCE,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 but these effects have not been reported consistently across all studies (see Section 4.3.6).
2 Gestational exposure to TCE in humans has been reported to be associated with
3 neurodevelopmental abnormalities including neural tube defects, encephalopathy, impaired
4 cognition, aggressive behavior, and speech and hearing impairment. Developmental
5 neurotoxicological changes have also been observed in animals including aggressive behaviors
6 following an in utero exposure to TCE and a suggestion of impaired cognition as noted by
7 decreased myelination in the CA1 hippocampal region of the brain. See Section 4.3.8 for
8 additional discussion of developmental neurological effects of TCE. Therefore, overall, the
9 strongest neurological evidence of human toxicological hazard is for changes in trigeminal nerve
10 function or morphology and impairment of vestibular function, based on both human and
11 experimental studies, while fewer and more limited evidence exists for delayed motor function,
12 changes in auditory, visual, and cognitive function or performance, and neurodevelopmental
13 outcomes.
14
15 6.1.3.2. Kidney Effects (see Sections 4.4.1, 4.4.4, 4.4.6, and4.11.1.2)
16 Kidney toxicity has also been associated with TCE exposure in both human and animal
17 studies. There are few human data pertaining to TCE-related noncancer kidney toxicity;
18 however, several available studies reported elevated excretion of urinary proteins, considered
19 nonspecific markers of nephrotoxicity, among TCE-exposed subjects compared to unexposed
20 controls. While some of these studies include subjects previously diagnosed with kidney cancer,
21 other studies report similar results in subjects that are disease free. Some additional support for
22 TCE nephrotoxicity in humans is provided by a study reporting a greater incidence of end-stage
23 renal disease in TCE-exposed workers as compared to unexposed controls, although some
24 subjects in this study were also exposed to hydrocarbons, JP-4 gasoline, and multiple solvents,
25 including TCE and 1,1,1-trichloroethane. See Section 4.4.1 for additional discussion of human
26 data on the noncancer kidney effects of TCE. Laboratory animal and in vitro data provide
27 additional support for TCE nephrotoxicity. TCE causes renal toxicity in the form of cytomegaly
28 and karyomegaly of the renal tubules in male and female rats and mice following either oral or
29 inhalation exposure. In rats, the pathology of TCE-induced nephrotoxicity appears distinct from
30 age-related nephropathy. Increased kidney weights have also been reported in some rodent
31 studies. See Section 4.4.4 for additional discussion of laboratory animal data on the noncancer
32 kidney effects of TCE. Further studies with TCE metabolites have demonstrated a potential role
33 for dichlorovinyl cysteine (DCVC), trichloroethanol, and trichloroacetic acid (TCA) in TCE-
34 induced nephrotoxicity. Of these, available data suggest that DCVC induced renal effects are
35 most similar to those of TCE and that DCVC is formed in sufficient amounts following TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-5 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure to account for these effects. TCE or DCVC have also been shown to be cytotoxic to
2 primary cultures of rat and human renal tubular cells. See Section 4.4.6 for additional discussion
3 on the role of metabolism in the noncancer kidney effects of TCE. Overall, multiple lines of
4 evidence support the conclusion that TCE causes nephrotoxicity in the form of tubular toxicity,
5 mediated predominantly through the TCE GSH conjugation product DCVC.
6
7 6.1.3.3. Liver Effects (see Sections 4.5.1, 4.5.3, 4.5.4, 4.5.6, and 4.11.1.3, and Appendix E)
8 Liver toxicity has also been associated with TCE exposure in both human and animal
9 studies. Although there are few human studies on liver toxicity and TCE exposure, several
10 available studies have reported TCE exposure to be associated with significant changes in serum
11 liver function tests, widely used in clinical settings in part to identify patients with liver disease,
12 or changes in plasma or serum bile acids. Additional, more limited human evidence for TCE
13 induced liver toxicity includes reports suggesting an association between TCE exposure and liver
14 disorders, and case reports of liver toxicity including hepatitis accompanying immune-related
15 generalized skin diseases, jaundice, hepatomegaly, hepatosplenomegaly, and liver failure in
16 TCE-exposed workers. Cohort studies examining cirrhosis mortality and either TCE exposure or
17 solvent exposure are generally null, but these studies cannot rule out an association with TCE
18 because of their use of death certificates where there is a high degree (up to 50%) of
19 underreporting. Overall, while some evidence exists of liver toxicity as assessed from liver
20 function tests, the data are inadequate for making conclusions regarding causality. See
21 Section 4.5.1 for additional discussion of human data on the noncancer liver effects of TCE. In
22 rats and mice, TCE exposure causes hepatomegaly without concurrent cytotoxicity. Like
23 humans, laboratory animals exposed to TCE have been observed to have increased serum bile
24 acids, although the toxicological importance of this effect is unclear. Other effects in the rodent
25 liver include small transient increases in DNA synthesis, cytomegaly in the form of "swollen" or
26 enlarged hepatocytes, increased nuclear size probably reflecting polyploidization, and
27 proliferation of peroxisomes. Available data also suggest that TCE does not induce substantial
28 cytotoxicity, necrosis, or regenerative hyperplasia, as only isolated, focal necroses and mild to
29 moderate changes in serum and liver enzyme toxicity markers having been reported. These
30 effects are consistently observed across rodent species and strains, although the degree of
31 response at a given mg/kg/d dose appears to be highly variable across strains, with mice on
32 average appearing to be more sensitive. See Sections 4.5.3 and 4.5.4 for additional discussion of
33 laboratory animal data on the noncancer liver effects of TCE. While it is likely that oxidative
34 metabolism is necessary for TCE-induced effects in the liver, the specific metabolite or
35 metabolites responsible is less clear. However, the available data are strongly inconsistent with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-6 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCA being the sole or predominant active moiety for TCE-induced liver effects, particularly
2 with respect to hepatomegaly. See Section 4.5.6 for additional discussion on the role of
3 metabolism in the noncancer liver effects of TCE. Overall, TCE, likely through its oxidative
4 metabolites, clearly leads to liver toxicity in laboratory animals, with mice appearing to be more
5 sensitive than other laboratory animal species, but there is only limited epidemiologic evidence
6 of hepatotoxicity being associated with TCE exposure.
7
8 6.1.3.4. ImmunologicalEffects (seeSections 4.6.1.1, 4.6.2, and4.11.1.4)
9 Effects related the immune system have also been associated with TCE exposure in both
10 human and animal studies. A relationship between systemic autoimmune diseases, such as
11 scleroderma, and occupational exposure to TCE has been reported in several recent studies, and a
12 meta-analysis of scleroderma studies resulted in a statistically significant combined odds ratio for
13 any exposure in men (odds ratio [OR]: 2.5, 95% confidence interval [CI]: 1.1, 5.4), with a lower
14 relative risk seen in women in women (OR: 1.2, 95% CI: 0.58, 2.6). The human data at this time
15 do not allow a determination of whether the difference in effect estimates between men and
16 women reflects the relatively low background risk of scleroderma in men, gender-related
17 differences in exposure prevalence or in the reliability of exposure assessment, a gender-related
18 difference in susceptibility to the effects of TCE, or chance. Additional human evidence for the
19 immunological effects of TCE includes studies reporting TCE-associated changes in levels of
20 inflammatory cytokines in occupationally-exposed workers and infants exposed via indoor air at
21 air concentrations typical of such exposure scenarios (see Section 6.1.1, above); a large number
22 of case reports (mentioned above) of a severe hypersensitivity skin disorder, distinct from
23 contact dermatitis and often accompanied by hepatitis; and a reported association between
24 increased history of infections and exposure to TCE contaminated drinking water. See
25 Section 4.6.1.1 for additional discussion of human data on the immunological effects of TCE.
26 Immunotoxicity has also been reported in experimental rodent studies of TCE. Numerous
27 studies have demonstrated accelerated autoimmune responses in autoimmune-prone mice,
28 including changes in cytokine levels similar to those reported in human studies, with more severe
29 effects, including autoimmune hepatitis, inflammatory skin lesions, and alopecia, manifesting at
30 longer exposure periods. Immunotoxic effects have been also reported in B6C3F1 mice, which
31 do not have a known particular susceptibility to autoimmune disease. Developmental
32 immunotoxicity in the form of hypersensitivity responses have been reported in TCE-treated
33 guinea pigs and mice via drinking water pre- and postnatally. Evidence of localized
34 immunosuppression has also been reported in mice and rats. See Section 4.6.2 for additional
35 discussion of laboratory animal data on the immunological effects of TCE. Overall, the human
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-7 DRAFT—DO NOT CITE OR QUOTE
-------
1 and animal studies of TCE and immune-related effects provide strong evidence for a role of TCE
2 in autoimmune disease and in a specific type of generalized hypersensitivity syndrome, while
3 there are less data pertaining to immunosuppressive effects.
4
5 6.1.3.5. Respiratory Tract Effects (see Sections 4.7.1.1, 4.7.2.1, 4.7.3, and 4.11.1.5)
6 The very few human data on TCE and pulmonary toxicity are too limited for drawing
7 conclusions (see Section 4.7.1.1), but laboratory studies in mice and rats have shown toxicity in
8 the bronchial epithelium, primarily in Clara cells, following acute exposures to TCE (see
9 Section 4.7.2.1). A few studies of longer duration have reported more generalized toxicity, such
10 as pulmonary fibrosis in mice and pulmonary vasculitis in rats. However, respiratory tract
11 effects were not reported in other longer-term studies. Acute pulmonary toxicity appears to be
12 dependent on oxidative metabolism, although the particular active moiety is not known. While
13 earlier studies implicated chloral produced in situ by CYP enzymes in respiratory tract tissue in
14 toxicity, the evidence is inconsistent and several other possibilities are viable. Although humans
15 appear to have lower overall capacity for enzymatic oxidation in the lung relative to mice, CYP
16 enzymes do reside in human respiratory tract tissue, suggesting that, qualitatively, the respiratory
17 tract toxicity observed in rodents is biologically plausible in humans. See Section 4.7.3 for
18 additional discussion of the role of metabolism in the noncancer respiratory tract toxicity of
19 TCE. Therefore, overall, data are suggestive of TCE causing respiratory tract toxicity, based
20 primarily on short-term studies in mice and rats, with available human data too few and limited
21 to add to the weight of evidence for pulmonary toxicity.
22
23 6.1.3.6. Reproductive Effects (see Sections 4.8.1 and4.11.1.6)
24 A number of human and laboratory animal studies suggest that TCE exposure has the
25 potential for male reproductive toxicity, with a more limited number of studies examining female
26 reproductive toxicity. Human studies have reported TCE exposure to be associated (in all but
27 one case statistically-significantly) with increased sperm density and decreased sperm quality,
28 altered sexual drive or function, or altered serum endocrine levels. Measures of male fertility,
29 however, were either not reported or reported to be unchanged with TCE exposure, though the
30 statistical power of the available studies is quite limited. Epidemiologic studies have identified
31 possible associations of TCE exposure with effects on female fertility and with menstrual cycle
32 disturbances, but these data are fewer than those available for male reproductive toxicity. See
33 Section 4.8.1.1 for additional discussion of human data on the reproductive effects of TCE.
34 Evidence of similar effects, particularly for male reproductive toxicity, is provided by several
35 laboratory animal studies that reported effects on sperm, libido/copulatory behavior, and serum
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-8 DRAFT—DO NOT CITE OR QUOTE
-------
1 hormone levels, although some studies that assessed sperm measures did not report treatment-
2 related alterations. Additional adverse effects on male reproduction have also been reported,
3 including histopathological lesions in the testes or epididymides and altered in vitro sperm-
4 oocyte binding or in vivo fertilization due to TCE or metabolites. While reduced fertility in
5 rodents was only observed in one study, this is not surprising given the redundancy and
6 efficiency of rodent reproductive capabilities. In addition, although the reduced fertility
7 observed in the rodent study was originally attributed to systemic toxicity, the database as a
8 whole suggests that TCE does induce reproductive toxicity independent of systemic effects.
9 Fewer data are available in rodents on female reproductive toxicity. While in vitro oocyte
10 fertilizability has been reported to be reduced as a result of TCE exposure in rats, a number of
11 other laboratory animal studies did not report adverse effects on female reproductive function.
12 See Section 4.8.1.2 for additional discussion of laboratory animal data on the reproductive
13 effects of TCE. Very limited data are available to elucidate the mode of action (MOA) for these
14 effects, though some aspects of a putative MOA (e.g., perturbations in testosterone biosynthesis)
15 appear to have some commonalities between humans and animals (see Section 4.8.1.3.2).
16 Together, the human and laboratory animal data support the conclusion that TCE exposure poses
17 a potential hazard to the male reproductive system, but are more limited with regard to the
18 potential hazard to the female reproductive system.
19
20 6.1.3.7. Developmental Effects (see Sections 4.8.3 and 4.11.1.7)
21 The relationship between TCE exposure (direct or parental) and developmental toxicity
22 has been investigated in a number of epidemiologic and laboratory animal studies. Postnatal
23 developmental outcomes examined include developmental neurotoxicity (addressed above with
24 neurotoxicity), developmental immunotoxicity (addressed above with immunotoxicity), and
25 childhood cancers. Prenatal effects examined include death (spontaneous abortion, perinatal
26 death, pre- or postimplantation loss, resorptions), decreased growth (low birth weight, small for
27 gestational age, intrauterine growth restriction, decreased postnatal growth), and congenital
28 malformations, in particular cardiac defects. Some epidemiological studies have reported
29 associations between parental exposure to TCE and spontaneous abortion or perinatal death, and
30 decreased birth weight or small for gestational age, although other studies reported mixed or null
31 findings. While comprising both occupational and environmental exposures, these studies are
32 overall not highly informative due to the small numbers of cases and limited exposure
33 characterization or to the fact that exposures were to a mixture of solvents. See Section 4.8.3.1
34 for additional discussion of human data on the developmental effects of TCE. However,
35 multiple well conducted studies in rats and mice show analogous effects of TCE exposure: pre-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-9 DRAFT—DO NOT CITE OR QUOTE
-------
1 or postimplantation losses, increased resorptions, perinatal death, and decreased birth weight.
2 Interestingly, the rat studies reporting these effects used Fischer 344 or Wistar rats, while several
3 other studies, all of which used Sprague-Dawley rats, reported no increased risk in these
4 developmental measures, suggesting a strain difference in susceptibility. See Section 4.8.3.2 for
5 additional discussion of laboratory animal data on the developmental effects of TCE. Therefore,
6 overall, based on weakly suggestive epidemiologic data and fairly consistent laboratory animal
7 data, it can be concluded that TCE exposure poses a potential hazard for prenatal losses and
8 decreased growth or birth weight of offspring.
9 With respect to congenital malformations, epidemiology and experimental animal studies
10 of TCE have reported increases in total birth defects, central nervous system defects, oral cleft
11 defects, eye/ear defects, kidney/urinary tract disorders, musculoskeletal birth anomalies,
12 lung/respiratory tract disorders, skeletal defects, and cardiac defects. Human occupational cohort
13 studies, while not consistently reporting positive results, are generally limited by the small
14 number of observed or expected cases of birth defects. While only one of the epidemiological
15 studies specifically reported observations of eye anomalies, studies in rats have identified
16 increases in the incidence of fetal eye defects following oral exposures during the period of
17 organogenesis with TCE or its oxidative metabolites dichloroacetic acid (DCA) and TCA. The
18 epidemiological studies, while individually limited, as a whole show relatively consistent
19 elevations, some of which were statistically significant, in the incidence of cardiac defects in
20 TCE-exposed populations compared to reference groups. In laboratory animal models, avian
21 studies were the first to identify adverse effects of TCE exposure on cardiac development, and
22 the initial findings have been confirmed multiple times. Additionally, administration of TCE and
23 its metabolites TCA and DCA in maternal drinking water during gestation has been reported to
24 induce cardiac malformations in rat fetuses. It is notable that a number of other studies, several
25 of which were well-conducted, did not report induction of cardiac defects in rats, mice, or rabbits
26 in which TCE was administered by inhalation or gavage. However, many of these studies used a
27 traditional free-hand section technique on fixed fetal specimens, and a fresh dissection technique
28 that can enhance detection of anomalies was used in the positive studies by Dawson et al. (1993)
29 and Johnson et al. (2003, 2005). Nonetheless, two studies that used the same or similar fresh
30 dissection technique did not report cardiac anomalies. Differences in other aspects of
31 experimental design may have been contributing factors to the differences in observed response.
32 In addition, mechanistic studies, such as the treatment-related alterations in endothelial cushion
33 development observed in avian in ovo and in vitro studies, provide a plausible mechanistic basis
34 for defects in septal and valvular morphogenesis observed in rodents, and consequently support
35 the plausibility of cardiac defects induced by TCE in humans. Therefore, while the studies by
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-10 DRAFT—DO NOT CITE OR QUOTE
-------
1 Dawson et al. (1993) and Johnson et al. (2003, 2005) have significant limitations, including the
2 lack of clear dose-response relationship for the incidence of any specific cardiac anomaly and the
3 pooling of data collected over an extended period, there is insufficient reason to dismiss their
4 findings. See Section 4.8.3.3.2 for additional discussion of the conclusions with respect to TCE-
5 induced cardiac malformations. Therefore, overall, based on weakly suggestive, but overall
6 consistent, epidemiologic data, in combination with evidence from experimental animal and
7 mechanistic studies, it can be concluded that TCE exposure poses a potential hazard for
8 congenital malformations, including cardiac defects, in offspring.
9
10 6.1.4. Carcinogenicity (see Sections 4.1, 4.2, 4.4.2, 4.4.5, 4.4.7, 4.5.2, 4.5.5, 4.5.6, 4.5.7,
11 4.6.1.2, 4.6.2.4, 4.7.1.2, 4.7.2.2, 4.7.4, 4.8.2, 4.9, and 4.11.2, and Appendices B and C)
12 In 1995, International Agency for Research on Cancer (IARC) concluded that
13 trichloroethylene is "probably carcinogenic to humans" (IARC, 1995). In 2000, National
14 Toxicology Program (NTP) concluded that trichloroethylene is "reasonably anticipated to be a
15 human carcinogen." (NTP, 2000). In 2001, the draft U.S. Environmental Protection Agency
16 (U.S. EPA) health risk assessment of TCE concluded that TCE was "highly likely" to be
17 carcinogenic in humans. In 2006, a committee of the National Research Council stated that
18 "findings of experimental, mechanistic, and epidemiologic studies lead to the conclusion that
19 trichloroethylene can be considered a potential human carcinogen" (NRC, 2006).
20 Following U.S. EPA (2005a) Guidelines for Carcinogen Risk Assessment, based on the
21 available data as of 2009, TCE is characterized as "Carcinogenic to Humans" by all routes of
22 exposure. This conclusion is based on convincing evidence of a causal association between TCE
23 exposure in humans and kidney cancer. The consistency of increased kidney cancer relative risk
24 estimates across a large number of independent studies of different designs and populations from
25 different countries and industries provides compelling evidence given the difficulty, a priori, in
26 detecting effects in epidemiologic studies when the relative risks are modest, the cancers are
27 relatively rare, and therefore, individual studies have limited statistical power. This strong
28 consistency of the epidemiologic data on TCE and kidney cancer argues against chance, bias,
29 and confounding as explanations for the elevated kidney cancer risks. In addition, statistically
30 significant exposure-response trends are observed in high-quality studies. These studies were
31 designed to examine kidney cancer in populations with high TCE exposure intensity. These
32 studies addressed important potential confounders and biases, further supporting the observed
33 associations with kidney cancer as causal. See Section 4.4.2 for additional discussion of the
34 human epidemiologic data on TCE exposure and kidney cancer. In a meta-analysis of 14 high-
35 quality studies, a statistically significant pooled relative risk estimate was observed for overall
36 TCE exposure (RRp: 1.25 [95% CI: 1.11, 1.41]). The pooled relative risk estimate was greater
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-11 DRAFT—DO NOT CITE OR QUOTE
-------
1 for the highest TCE exposure groups (RRp: 1.53 [95% CI: 1.23, 1.91]; n = 12 studies). Meta-
2 analyses investigating the influence of individual studies and the sensitivity of the results to
3 alternate relative risk estimate selections found the pooled relative risk estimates to be highly
4 robust. Furthermore, there was no indication of publication bias or significant heterogeneity. It
5 would require a substantial amount of high-quality negative data to contradict this observed
6 association. See Section 4.4.2.5 and Appendix C for additional discussion of the kidney cancer
7 meta-analysis.
8 The human evidence of carcinogenicity from epidemiologic studies of TCE exposure is
9 compelling for lymphoma but less convincing than for kidney cancer. High quality studies
10 generally reported excess relative risk estimates, with statistically significant increases in three
11 studies, and a statistically significant trend with TCE exposure in one study (see Section 4.6.1.2).
12 The consistency of the association between TCE exposure and lymphoma is further supported by
13 the results of meta-analyses (see Section 4.6.1.2.2 and Appendix C). A statistically significant
14 pooled relative risk estimate was observed for overall TCE exposure (RRp: 1.23 [95% CI: 1.04,
15 1.44]), and, as with kidney cancer, the pooled relative risk estimate was greater for the highest
16 TCE exposure groups (RRp: 1.57 [95% CI: 1.27, 1.94]) than for overall TCE exposure.
17 Sensitivity analyses indicated that this result and its statistical significance were not overly
18 influenced by most individual studies or choice of individual (study-specific) risk estimates, and
19 in only one case was the resulting pooled relative risk estimates not statistically significant
20 (lower confidence bound of 1.00). Some heterogeneity was observed, particularly between
21 cohort and case-control studies, but it was not statistically significant. Notably, no heterogeneity
22 was observed in the meta-analysis of the highest exposure group, providing some evidence of
23 exposure misclassification as a source of heterogeneity in the overall analysis. In addition, there
24 was some evidence of potential publication bias. Thus, while the evidence is strong for
25 lymphoma, issues of study heterogeneity, potential publication bias, and weaker exposure-
26 response results contribute greater uncertainty.
27 The evidence is more limited for liver and biliary tract cancer mainly because only cohort
28 studies are available and most of these studies have small numbers of cases due the comparative
29 rarity of liver and biliary tract cancer. While most high quality studies reported excess relative
30 risk estimates, they were generally based on small numbers of cases or deaths, with the result of
31 wide confidence intervals on the estimates. The low number of liver cancer cases in the
32 available studies made assessing exposure-response relationships difficult. See Section 4.5.2 for
33 additional discussion of the human epidemiologic data on TCE exposure and liver cancer. A
34 consistency of the association between TCE exposure and liver cancer is supported by the results
35 of meta-analyses (see Section 4.5.2 and Appendix C). These meta-analyses found a statistically
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-12 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant increased pooled relative risk estimate for liver and biliary tract cancer of 1.33 (95%
2 CI: 1.09, 1.64) with overall TCE exposure; but the meta-analyses using only the highest
3 exposure groups yielded a lower, and nonstatistically significant, pooled estimate for primary
4 liver cancer (1.28 [95% CI: 0.93, 1.77]). Although there was no evidence of heterogeneity or
5 publication bias and the pooled estimates were fairly insensitive to the use of alternative relative
6 risk estimates, the statistical significance of the pooled estimates depends heavily on the one
7 large study by Raaschou-Nielsen et al. (2003). There were fewer adequate, high quality studies
8 available for meta-analysis of liver cancer (9 versus 16 for lymphoma and 14 for kidney), leading
9 to lower statistical power, even with pooling. Thus, while there is epidemiologic evidence of an
10 association between TCE exposure and liver cancer, the much more limited database, both in
11 terms of number of available studies and number of cases upon which the studies are based,
12 contributes to greater uncertainty as compared to the evidence for kidney cancer or lymphoma.
13 There are several other lines of supporting evidence for TCE carcinogenicity in humans
14 by all routes of exposure. First, multiple chronic bioassays in rats and mice have reported
15 increased incidences of tumors with TCE treatment via inhalation and oral gavage, including
16 tumors in the kidney, liver, and lymphoid tissues—target tissues of TCE carcinogenicity also
17 seen in epidemiological studies. Of particular note is the site-concordant finding of low, but
18 biologically and sometimes statistically significant, increases in the incidence of kidney tumors
19 in multiple strains of rats treated with TCE by either inhalation or corn oil gavage (see
20 Section 4.4.5). The increased incidences were only detected at the highest tested doses, and were
21 greater in male than female rats; although, notably, pooled incidences in females from five rat
22 strains tested by NTP (1988, 1990) resulted in a statistically significant trend. Although these
23 studies have shown limited increases in kidney tumors, and several individual studies have a
24 number of limitations, given the rarity of these tumors as assessed by historical controls and the
25 repeatability of this result across studies and strains, these are considered biologically significant.
26 Therefore, while individual studies provide only suggestive evidence of renal carcinogenicity,
27 the database as a whole supports the conclusion that TCE is a kidney carcinogen in rats, with
28 males being more sensitive than females. No other tested laboratory species (i.e., mice and
29 hamsters) have exhibited increased kidney tumors, with no adequate explanation for these
30 species differences (particularly with mice, which have been extensively tested). With respect to
31 the liver, TCE and its oxidative metabolites chloral hydrate (CH), TCA, and DCA are clearly
32 carcinogenic in mice, with strain and sex differences in potency that appear to parallel,
33 qualitatively, differences in background tumor incidence. Data in other laboratory animal
34 species are limited; thus, except for DCA which is carcinogenic in rats, inadequate evidence
35 exists to evaluate the hepatocarcinogenicity of these compounds in rats or hamsters. However, to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-13 DRAFT—DO NOT CITE OR QUOTE
-------
1 the extent that there is hepatocarcinogenic potential in rats, TCE is clearly less potent in the
2 strains tested in this species than in B6C3F1 and Swiss mice. See Section 4.5.5 for additional
3 discussion of laboratory animal data on TCE-induced liver tumors. Additionally, there is more
4 limited evidence for TCE-induced lymphatic cancers in rats and mice, lung tumors in mice, and
5 testicular tumors in rats. With respect to the lymphatic cancers, two studies in mice reported
6 increased incidences of lymphomas in females of two different strains, and two studies in rats
7 reported leukemias in males of one strain and females of another. However, these tumors had
8 relatively modest increases in incidence with treatment, and were not reported to be increased in
9 other studies. See Section 4.6.2.4 for additional discussion of laboratory animal data on TCE-
10 induced lymphatic tumors. With respect to lung tumors, rodent bioassays have demonstrated a
11 statistically significant increase in pulmonary tumors in mice following chronic inhalation
12 exposure to TCE, and nonstatistically significant increases in mice exposed orally; but
13 pulmonary tumors were not reported in other species tested (i.e., rats and hamsters) (see
14 Section 4.7.2.2). Finally, increased testicular (interstitial or Leydig cell) tumors have been
15 observed in multiple studies of rats exposed by inhalation and gavage, although in some cases
16 high (>75%) control rates of testicular tumors in rats limited the ability to detect a treatment
17 effect. See Section 4.8.2.2 for additional discussion of laboratory animal data on TCE-induced
18 tumors of the reproductive system. Overall, TCE is clearly carcinogenic in rats and mice. The
19 apparent lack of site concordance across laboratory animal studies may be due to limitations in
20 design or conduct in a number of rat bioassays and/or genuine interspecies differences in
21 qualitative or quantitative sensitivity (i.e., potency). Nonetheless, these studies have shown
22 carcinogenic effects across different strains, sexes, and routes of exposure, and site-concordance
23 is not necessarily expected for carcinogens.
24 A second line of supporting evidence for TCE carcinogenicity in humans consists of
25 toxicokinetic data indicating that TCE is well absorbed by all routes of exposure, and that TCE
26 absorption, distribution, metabolism, and excretion are qualitatively similar in humans and
27 rodents. As summarized above, there is evidence that TCE is systemically available, distributes
28 to organs and tissues, and undergoes systemic metabolism from all routes of exposure.
29 Therefore, although the strongest evidence from epidemiologic studies largely involves
30 inhalation exposures, the evidence supports TCE carcinogenicity being applicable to all routes of
31 exposure. In addition, there is no evidence of major qualitative differences across species in
32 TCE absorption, distribution, metabolism, and excretion. Extensive in vivo and in vitro data
33 show that mice, rats, and humans all metabolize TCE via two primary pathways: oxidation by
34 CYPs and conjugation with glutathione via GSTs. Several metabolites and excretion products
35 from both pathways have been detected in blood and urine from exposed humans was well as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 from at least one rodent species. In addition, the subsequent distribution, metabolism, and
2 excretion of TCE metabolites are qualitatively similar among species. Therefore, humans
3 possess the metabolic pathways that produce the TCE metabolites thought to be involved in the
4 induction of rat kidney and mouse liver tumors, and internal target tissues of both humans and
5 rodents experience a similar mix of TCE and metabolites. See Sections 3.1-3.4 for additional
6 discussion of TCE toxicokinetics. Quantitative interspecies differences in toxicokinetics do
7 exist, and are addressed through PBPK modeling (see Section 3.5 and Appendix A).
8 Importantly, these quantitative differences affect only interspecies extrapolations of carcinogenic
9 potency, and do not affect inferences as to the carcinogenic hazard for TCE.
10 Finally, available mechanistic data do not suggest a lack of human carcinogenic hazard
11 from TCE exposure. In particular, these data do not suggest qualitative differences between
12 humans and test animals that would preclude any of the hypothesized key events in the
13 carcinogenic MO A in rodents from occurring in humans. For the kidney, the predominance of
14 positive genotoxicity data in the database of available studies of TCE metabolites derived from
15 GSH conjugation (in particular DCVC), together with toxicokinetic data consistent with their
16 systemic delivery to and in situ formation in the kidney, supports the conclusion that a mutagenic
17 MO A is operative in TCE-induced kidney tumors. While supporting the biological plausibility
18 of this hypothesized MO A, available data on the von Hippel-Lindau (VHL) gene in humans or
19 transgenic animals do not conclusively elucidate the role of VHL mutation in TCE-induced renal
20 carcinogenesis. Cytotoxicity and compensatory cell proliferation, similarly presumed to be
21 mediated through metabolites formed after GSH-conjugation of TCE, have also been suggested
22 to play a role in the MO A for renal carcinogenesis, as high incidences of nephrotoxicity have
23 been observed in animals at doses that induce kidney tumors. Human studies have reported
24 markers for nephrotoxicity at current occupational exposures, although data are lacking at lower
25 exposures. Nephrotoxicity is observed in both mice and rats, in some cases with nearly 100%
26 incidence in all dose groups, but kidney tumors are only observed at low incidences in rats at the
27 highest tested doses. Therefore, nephrotoxicity alone appears to be insufficient, or at least not
28 rate-limiting, for rodent renal carcinogenesis, since maximal levels of toxicity are reached before
29 the onset of tumors. In addition, nephrotoxicity has not been shown to be necessary for kidney
30 tumor induction by TCE in rodents. In particular, there is a lack of experimental support for
31 causal links, such as compensatory cellular proliferation or clonal expansion of initiated cells,
32 between nephrotoxicity and kidney tumors induced by TCE. Furthermore, it is not clear if
33 nephrotoxicity is one of several key events in a MO A, if it is a marker for an "upstream" key
34 event (such as oxidative stress) that may contribute independently to both nephrotoxicity and
35 renal carcinogenesis, or if it is incidental to kidney tumor induction. Moreover, while
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-15 DRAFT—DO NOT CITE OR QUOTE
-------
1 toxicokinetic differences in the GSH conjugation pathway along with their uncertainty are
2 addressed through PBPK modeling, no data suggest that any of the proposed key events for
3 TCE-induced kidney tumors in rats are precluded in humans. See Section 4.4.7 for additional
4 discussion of the MOA for TCE-induced kidney tumors. Therefore, TCE-induced rat kidney
5 tumors provide additional support for the convincing human evidence of TCE-induced kidney
6 cancer, with mechanistic data supportive of a mutagenic MOA.
7 With respect to other tumor sites, data are insufficient to conclude that any of the other
8 hypothesized MO As are operant. In the liver, a mutagenic MOA mediated by CH, which has
9 evidence for genotoxic effects, or some other oxidative metabolite of TCE cannot be ruled out,
10 but data are insufficient to conclude it is operant. A second MOA hypothesis for TCE-induced
11 liver tumors involves activation of the peroxisome proliferator activated receptor alpha (PPARa)
12 receptor. Clearly, in vivo administration of TCE leads to activation of PPARa in rodents and
13 likely does so in humans as well. However, the evidence as a whole does not support the view
14 that PPARa is the sole operant MOA mediating TCE hepatocarcinogenesis. Rather, there is
15 evidential support for multiple TCE metabolites and multiple toxicity pathways contributing to
16 TCE-induced liver tumors. Furthermore, recent experiments have demonstrated that PPARa
17 activation and the sequence of key events in the hypothesized MOA are not sufficient to induce
18 hepatocarcinogenesis (Yang et al., 2007). Moreover, the demonstration that the PPARa agonist
19 di(2-ethylhexyl) phthalate induces tumors in PPARa-null mice supports the view that the events
20 comprising the hypothesized PPARa activation MOA are not necessary for liver tumor induction
21 in mice by this PPARa agonist (Ito et al., 2007). See Section 4.5.7 for additional discussion of
22 the MOA for TCE-induced liver tumors. For mouse lung tumors, as with the liver, a mutagenic
23 MOA involving CH has also been hypothesized, but there are insufficient data to conclude that it
24 is operant. A second MOA hypothesis for mouse lung tumors has been posited involving other
25 effects of oxidative metabolites including cytotoxicity and regenerative cell proliferation, but
26 experimental support remains limited, with no data on proposed key events in experiments of
27 duration 2 weeks or longer. See Section 4.7.4 for additional discussion of the MOA for TCE-
28 induced lung tumors. A MOA subsequent to in situ oxidative metabolism, whether involving
29 mutagenicity, cytotoxicity, or other key events, may also be relevant to other tissues where TCE
30 would undergo CYP metabolism. For instance, CYP2E1, oxidative metabolites, and protein
31 adducts have been reported in the testes of rats exposed to TCE, and, in some rat bioassays, TCE
32 exposure increased the incidence of rat testicular tumors. However, inadequate data exist to
33 adequately define a MOA hypothesis for this tumor site (see Section 4.8.2.3 for additional
34 discussion of the MOA for TCE-induced testicular tumors).
35
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-16 DRAFT—DO NOT CITE OR QUOTE
-------
1 6.1.5. Susceptibility (see Sections 4.10 and 4.11.3)
2 There is some evidence that certain populations may be more susceptible to exposure to
3 TCE. Factors affecting susceptibility examined include lifestage, gender, genetic
4 polymorphisms, race/ethnicity, preexisting health status, and lifestyle factors and nutrition status.
5 Factors that impact early lifestage susceptibility include exposures such as transplacental transfer
6 and breast milk ingestion, early lifestage-specific toxicokinetics, and differential outcomes in
7 early lifestages such as developmental cardiac defects (see Section 4.10.1). Because the weight
8 of evidence supports a mutagenic MOA being operative for TCE carcinogenicity in the kidney
9 (see Section 4.4.7), and there is an absence of chemical-specific data to evaluate differences in
10 carcinogenic susceptibility, early-life susceptibility should be assumed and the age-dependent
11 adjustment factors (ADAFs) should be applied, in accordance with the Supplemental Guidance
12 (see summary below in Section 6.2.2.5). Fewer data are available on later lifestages, although
13 there is suggestive evidence to indicate that older adults may experience increased adverse
14 effects than younger adults due to greater tissue distribution of TCE. In general, more studies
15 specifically designed to evaluate effects in early and later lifestages are needed in order to more
16 fully characterize potential life stage-related TCE toxicity. Gender-specific (see
17 Section 4.10.2.1) differences also exist in toxicokinetics (e.g., cardiac outputs, percent body fat,
18 expression of metabolizing enzymes) and susceptibility to toxic endpoints (e.g., gender-specific
19 effects on the reproductive system, gender differences in baseline risks to endpoints such as
20 scleroderma or liver cancer). Genetic variation (see Section 4.10.2.2) likely has an effect on the
21 toxicokinetics of TCE. Increased CYP2E1 activity and GST polymorphisms may influence
22 susceptibility of TCE due to effects on production of toxic metabolites or may play a role in
23 variability in toxic response. Differences in genetic polymorphisms related to the metabolism of
24 TCE have also been observed among various race/ethnic groups (see Section 4.10.2.3).
25 Preexisting diminished health status (see Section 4.10.2.4) may alter the response to TCE
26 exposure. Individuals with increased body mass may have an altered toxicokinetic response due
27 to the increased uptake of TCE into fat. Other conditions that may alter the response to TCE
28 exposure include diabetes and hypertension, and lifestyle and nutrition factors (see
29 Section 4.10.2.5) such alcohol consumption, tobacco smoking, nutritional status, physical
30 activity, and socioeconomic status. Alcohol intake has been associated with inhibition of TCE
31 metabolism in both humans and experimental animals. In addition, such conditions have been
32 associated with increased baseline risks for health effects also associated with TCE, such as
33 kidney cancer and liver cancer. However, the interaction between TCE and known risk factors
34 for human diseases is not known, and further evaluation of the effects due to these factors is
35 needed.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-17 DRAFT—DO NOT CITE OR QUOTE
-------
1 In sum, there is some evidence that certain populations may be more susceptible to
2 exposure to TCE. Factors affecting susceptibility examined include lifestage, gender, genetic
3 polymorphisms, race/ethnicity, preexisting health status, and lifestyle factors and nutrition status.
4 However, except in the case of toxicokinetic variability characterized using the PBPK model
5 described in Section 3.5, there are inadequate chemical-specific data to quantify the degree of
6 differential susceptibility due to such factors.
7
8 6.2. DOSE-RESPONSE ASSESSMENT
9 This section summarizes the major conclusions of the dose-response analysis for TCE
10 noncancer effects and carcinogenicity, with more detailed discussions in Chapter 5.
11
12 6.2.1. Noncancer Effects (see Section 5.1)
13 6.2.1.1. Background and Methods
14 As summarized above, based on the available human epidemiologic data and
15 experimental and mechanistic studies, it is concluded that TCE poses a potential human health
16 hazard for noncancer toxicity to the central nervous system, the kidney, the liver, the immune
17 system, the male reproductive system, and the developing fetus. The evidence is more limited
18 for TCE toxicity to the respiratory tract and female reproductive system.
19 Dose-response analysis for a noncancer endpoint generally involves two steps: (1) the
20 determination of a point of departure (POD) derived from a benchmark dose (BMD)1, a
21 no-observed-adverse-effect level (NOAEL), or a lowest-observed-adverse-effect level (LOAEL);
22 and (2) adjustment of the POD by endpoint/study-specific "uncertainty factors" (UFs),
23 accounting for adjustments and uncertainties in the extrapolation from the study conditions to
24 conditions of human exposure.
25 Because of the large number of noncancer health effects associated with TCE exposure
26 and the large number of studies reporting on these effects, in contrast to toxicological reviews for
27 chemicals with smaller databases of studies, a formal, quantitative screening process (see
28 Section 5.1) was used to reduce the number of endpoints and studies to those that would best
29 inform the selection of the critical effects for the inhalation reference concentration (RfC) and
30 oral reference dose (RfD).2 As described in Section 5.1, for all studies described in Chapter 4
1 more precisely, it is the benchmark dose lower bound (BMDL), i.e., the (one-sided) 95% lower confidence bound
on the dose corresponding to the benchmark response (BMR) for the effect, that is used as the POD
2 In U.S. EPA noncancer health assessments, the RfC [RfD] is an estimate (with uncertainty spanning perhaps an
order of magnitude) of a continuous inhalation [daily oral] exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. It can be derived
from a NOAEL, LOAEL, or benchmark concentration [dose], with uncertainty factors generally applied to reflect
limitations of the data used.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-18 DRAFT—DO NOT CITE OR QUOTE
-------
1 which report adverse noncancer health effects and provided quantitative dose-response data,
2 PODs on the basis of applied dose, adjusted by endpoint/study-specific UFs, were used to
3 develop candidate RfCs (cRfCs) and candidate RfDs (cRfDs) intended to be protective for each
4 endpoint individually. Candidate critical effects—those with the lowest cRfCs and cRfDs taking
5 into account the confidence in each estimate—were selected within each of the following health
6 effect domains: (1) neurological, (2) systemic/organ system; (3) immunological; (4)
7 reproductive; and (5) developmental. For each of these candidate critical effects, the PBPK
8 model developed in Section 3.5 was used for interspecies, intraspecies, and route-to-route
9 extrapolation on the basis of internal dose to develop PBPK model-based PODs. Plausible
10 internal dose metrics were selected based on what is understood about the role of different TCE
11 metabolites in toxicity and the MOA for toxicity. These PODs were then adjusted by
12 endpoint/study-specific UFs, taking into account the use of the PBPK model, to develop PBPK
13 model-based candidate RfCs (p-cRfCs) and candidate RfDs (p-cRfDs). The most sensitive
14 cRfCs, p-cRfCs, cRfDs, and p-cRfDs were then evaluated, taking into account the confidence in
15 each estimate, to arrive at overall candidate RfCs and RfDs for each health effect type. Then, the
16 RfC and RfD for TCE were selected so as to be protective of the most sensitive effects. In
17 contrast to the approach used in most assessments, in which the RfC and RfD are each based on
18 a single critical effect, the final RfC and RfD for TCE were based on multiple critical effects that
19 resulted in very similar candidate RfC and RfD values at the low end of the full range of values.
20 This approach was taken here because it provides robust estimates of the RfC and RfD and
21 because it highlights the multiple effects that are all yielding very similar candidate values.
22
23 6.2.1.2. Uncertainties and Application of Uncertainty Factors (UFs) (see Section 5.1.1 and
24 5.1.4)
25 An underlying assumption in deriving reference values for noncancer effects is that the
26 dose-response relationship for these effects has a threshold. Thus, a fundamental uncertainty is
27 the validity of that assumption. For some effects, in particular effects on very sensitive processes
28 (e.g., developmental processes) or effects for which there is a nontrivial background level and
29 even small exposures may contribute to background disease processes in more susceptible
30 people, a practical threshold (i.e., a threshold within the range of environmental exposure levels
31 of regulatory concern) may not exist.
32 Nonetheless, under the assumption of a threshold, the desired exposure level to have as a
33 reference value is the maximum level at which there is no appreciable risk for an adverse effect
34 in sensitive subgroups (of humans). However, because it is not possible to know what this level
35 is, "uncertainty factors" are used to attempt to address quantitatively various aspects, depending
36 on the data set, of qualitative uncertainty.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-19 DRAFT—DO NOT CITE OR QUOTE
-------
1 First there is uncertainty about the "point of departure" for the application of UFs.
2 Conceptually, the POD should represent the maximum exposure level at which there is no
3 appreciable risk for an adverse effect in the study population under study conditions (i.e., the
4 threshold in the dose-response relationship). Then, the application of the relevant UFs is
5 intended to convey that exposure level to the corresponding exposure level for sensitive human
6 subgroups exposed continuously for a lifetime. In fact, it is again not possible to know that
7 exposure level even for a laboratory study because of experimental limitations (e.g. the power to
8 detect an effect, dose spacing, measurement errors, etc.), and crude approximations like the
9 NOAEL or a BMDL are used. If a LOAEL is used as the POD, the LOAEL-to-NOAEL UF is
10 applied as an adjustment factor to better approximate the desired exposure level (threshold),
11 although the necessary extent of adjustment is unknown. The standard value for the LOAEL-to-
12 NOAEL UF is 10, although sometimes a value of 3 is used if the effect is considered minimally
13 adverse at the response level observed at the LOAEL or even 1 if the effect is an early marker for
14 an adverse effect. For one POD in this assessment, a value of 30 was used for the LOAEL-to-
15 NOAEL UF because the incidence rate for the adverse effect was >90% at the LOAEL.
16 If a BMDL is used as the POD, there are uncertainties regarding the appropriate dose-
17 response model to apply to the data, but these should be minimal if the modeling is in the
18 observable range of the data. There are also uncertainties about what BMR to use to best
19 approximate the desired exposure level (threshold, see above). For continuous endpoints, in
20 particular, it is often difficult to identify the level of change that constitutes the "cut-point" for an
21 adverse effect. Sometimes, to better approximate the desired exposure level, a BMR somewhat
22 below the observable range of the data is selected. In such cases, the model uncertainty is
23 increased, but this is a trade-off to reduce the uncertainty about the POD not being a good
24 approximation for the desired exposure level.
25 For each of these types of PODs, there are additional uncertainties pertaining to
26 adjustments to the administered exposures (doses). Typically, administered exposures (doses)
27 are converted to equivalent continuous exposures (daily doses) over the study exposure period
28 under the assumption that the effects are related to concentration x time, independent of the daily
29 (or weekly) exposure regimen (i.e., a daily exposure of 6 hours to 4 ppm is considered equivalent
30 to 24 hours of exposure to 1 ppm). However, the validity of this assumption is generally
31 unknown, and, if there are dose-rate effects, the assumption of C x t equivalence would tend to
32 bias the POD downwards. Where there is evidence that administered exposure better correlates
33 to the effect than equivalent continuous exposure averaged over the study exposure period (e.g.,
34 visual effects), administered exposure was not adjusted. For the PBPK analyses in this
35 assessment, the actual administered exposures are taken into account in the PBPK modeling, and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-20 DRAFT—DO NOT CITE OR QUOTE
-------
1 equivalent daily values (averaged over the study exposure period) for the dose metrics are
2 obtained (see above, 5.1.3.2). Additional uncertainties about the PBPK based estimates include
3 uncertainties about the appropriate dose metric for each effect, although for some effects there
4 was better information about relevant dose metrics than for others (see Section 5.1.3.1).
5 There is also uncertainty about the other UFs. The human variability UF is to some
6 extent an adjustment factor because for more sensitive people, the dose-response relationship
7 shifts to lower exposures. But there is uncertainty about the extent of the adjustment required,
8 i.e., about the distribution of human susceptibility. Therefore, in the absence of data on a
9 susceptible population(s) or on the distribution of susceptibility in the general population, an UF
10 of 10 is generally used, which breaks down (approximately) to a factor of 3 for pharmacokinetic
11 variability and a factor of 3 for pharmacodynamic variability. This standard value was used for
12 all the PODs based on applied dose in this assessment with the exception of the PODs for a few
13 immunological effects that were based on data from a sensitive (autoimmune-prone) mouse
14 strain. For those PODs, an UF of 3 (reflecting pharmacokinetics only) was used for human
15 variability. The PBPK analyses in this assessment attempt to account for the pharmacokinetic
16 portion of human variability using human data on pharmacokinetic variability. For PBPK
17 model-based candidate reference values, the pharmacokinetic component of this UF was omitted.
18 A quantitative uncertainty analysis of the PBPK derived dose metrics used in the assessment is
19 presented in Section 5.1.4.2 in Chapter 5. There is still uncertainty regarding the susceptible
20 subgroups for TCE exposure and the extent of pharmacodynamic variability.
21 If the data used to determine a particular POD are from laboratory animals, an
22 interspecies extrapolation UF is used. This UF is also to some extent an adjustment factor for the
23 expected scaling for toxicologically-equivalent doses across species (i.e., according to body
24 weight to the % power for oral exposures). However, there is also uncertainty about the true
25 extent of interspecies differences for specific noncancer effects from specific chemical
26 exposures. For oral exposures, the standard value for the interspecies UF is 10, which can be
27 viewed as breaking down (approximately) to a factor of 3 for the "adjustment" (nominally
28 pharmacokinetics) and a factor of 3 for the "uncertainty" (nominally pharmacodynamics). For
29 inhalation exposures for systemic toxicants such as TCE, no adjustment across species is
30 generally assumed for fixed air concentrations (ppm equivalence), and the standard value for the
31 interspecies UF is 3 reflects "uncertainty" (nominally pharmacodynamics only). The PBPK
32 analyses in this assessment attempt to account for the "adjustment" portion of interspecies
33 extrapolation using rodent pharmacokinetic data to estimate internal doses for various dose
34 metrics. Equal doses of these dose metrics, appropriately scaled, are then assumed to convey
35 equivalent risk across species. For PBPK model-based candidate reference values, the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-21 DRAFT—DO NOT CITE OR QUOTE
-------
1 "adjustment" component of this UF was omitted. With respect to the "uncertainty" component,
2 quantitative uncertainty analyses of the PBPK derived dose metrics used in the assessment are
3 presented in Section 5.1.4.2 in Chapter 5. However, these only address the pharmacokinetic
4 uncertainties in a particular dose metric, and there is still uncertainty regarding the true dose
5 metrics. Nor do the PBPK analyses address the uncertainty in either cross-species
6 pharmacodynamic differences (i.e., about the assumption that equal doses of the appropriate dose
7 metric convey equivalent risk across species for a particular endpoint from a specific chemical
8 exposure) or in cross-species pharmacokinetic differences not accounted for by the PBPK model
9 dose metrics (e.g., departures from the assumed interspecies scaling of clearance of the active
10 moiety, in the cases where only its production is estimated). A value of 3 is typically used for
11 the "uncertainty" about cross-species differences, and this generally represents true uncertainty
12 because it is usually unknown, even after adjustments have been made to account for the
13 expected interspecies differences, whether humans have more or less susceptibility, and to what
14 degree, than the laboratory species in question.
15 RfCs and RfDs apply to lifetime exposure, but sometimes the best (or only) available
16 data come from less-than-lifetime studies. Lifetime exposure can induce effects that may not be
17 apparent or as large in magnitude in a shorter study; consequently, a dose that elicits a specific
18 level of response from a lifetime exposure may be less than the dose eliciting the same level of
19 response from a shorter exposure period. If the effect becomes more severe with increasing
20 exposure, then chronic exposure would shift the dose-response relationship to lower exposures,
21 although the true extent of the shift is unknown. PODs based on subchronic exposure data are
22 generally divided by a subchronic-to-chronic UF, which has a standard value of 10. If there is
23 evidence suggesting that exposure for longer time periods does not increase the magnitude of an
24 effect, a lower value of 3 or 1 might be used. For some reproductive and developmental effects,
25 chronic exposure is that which covers a specific window of exposure that is relevant for eliciting
26 the effect, and subchronic exposure would correspond to an exposure that is notably less than the
27 full window of exposure.
28 Sometimes a database UF is also applied to address limitations or uncertainties in the
29 database. The overall database for TCE is quite extensive, with studies for many different types
30 of effects, including 2-generation reproductive studies, as well as neurological and
31 immunological studies. In addition, there were sufficient data to develop a reliable PBPK model
32 to estimate route-to-route extrapolated doses for some candidate critical effects for which data
33 were only available for one route of exposure. Thus, there is a high degree of confidence that the
34 TCE database was sufficient to identify some sensitive endpoints, and no database UF was used
35 in this assessment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 6.2.1.3. Candidate Critical Effects and Reference Values (see Sections 5.1.2 and 5.1.3)
2 A large number of endpoints and studies were considered within each health effect
3 domain. Chapter 5 contains a comprehensive discussion of all endpoints/studies which were
4 considered for developing candidate reference values (cRfCs, cRfDs, p-cRfCs, and p-cRfDs),
5 their PODs, and the UFs applied. The summary below reviews the selection of candidate critical
6 effects for each health effect domain, the confidence in the reference values, the selection of
7 PBPK model-based dose metrics, and the impact of PBPK modeling on the candidate reference
8 values.
9
10 6.2.1.3.1. Neurological effects. Candidate reference values were developed for several
11 neurological domains for which there was evidence of hazard (see Tables 5-1 and 5-8). There is
12 higher confidence in the candidate reference values for trigeminal nerve, auditory, or
13 psychomotor effects, but the available data suggest that the more sensitive indicators of TCE
14 neurotoxicity are changes in wakefulness, regeneration of the sciatic nerve, demyelination in the
15 hippocampus and degeneration of dopaminergic neurons. Therefore, these more sensitive effects
16 are considered the candidate critical effects for neurotoxicity, albeit with more uncertainty in the
17 corresponding candidate reference values. Of these more sensitive effects, there is greater
18 confidence in the changes in wakefulness reported by Arito et al. (1994). In addition, trigeminal
19 nerve effects are considered a candidate critical effect because this is the only type of
20 neurological effect for which human data are available, and the POD for this effect is similar to
21 that from the most sensitive rodent study (Arito et al., 1994, for changes in wakefulness).
22 Between the two human studies of trigeminal nerve effects, Ruitjen et al. (1991) is preferred for
23 deriving noncancer reference values because its exposure characterization is considered more
24 reliable.
25 Because of the lack of specific data as to the metabolites involved and the MO A for the
26 candidate critical neurologic effects, PBPK model predictions of total metabolism (scaled by
27 body weight to the % power) were selected as the preferred dose metric based on the general
28 observation that TCE toxicity is associated with metabolism. The area-under-the-curve (AUC)
29 of TCE in blood was used as an alternative dose metric. With these dose metrics, the candidate
30 reference values derived using the PBPK model were only modestly (~3-fold or less) different
31 than those derived on the basis of applied dose.
32
33 6.2.1.3.2. Kidney effects. High-confidence candidate reference values were developed for
34 histopathological and weight changes in the kidney (see Tables 5-2 and 5-9), and these are
35 considered to be candidate critical effects for several reasons. First, they appear to be the most
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-23 DRAFT—DO NOT CITE OR QUOTE
-------
1 sensitive indicators of toxicity that are available for the kidney. In addition, as discussed in
2 Sections 3.3 and 3.5, both in vitro and in vivo pharmacokinetic data indicate substantially more
3 production of GSH-conjugates thought to mediate TCE kidney effects in humans relative to rats
4 and mice. Several studies are considered reliable for developing candidate reference values for
5 these endpoints. For histopathological changes, these were the only available inhalation study
6 (Maltoni et al., 1986), the NTP (1988) study in rats, and the National Cancer Institute (NCI,
7 1976) study in mice. For kidney weight changes, both available studies (Kjellstrand et al.,
8 1983b; Woolhiser et al., 2006) were chosen as candidate critical studies.
9 Due to the substantial evidence supporting the role of GSH conjugation metabolites in
10 TCE-induced nephrotoxicity, the preferred PBPK model dose metrics for kidney effects were the
11 amount of DCVC bioactivated in the kidney for rat studies and the amount of GSH conjugation
12 (both scaled by body weight to the 3/4 power) for mouse studies (inadequate toxicokinetic data are
13 available in mice for predicting the amount of DCVC bioactivation). With these dose metrics,
14 the candidate reference values derived using the PBPK model were 300- to 400-fold lower than
15 those derived on the basis of applied dose. As discussed above and in Chapter 3, this is due to
16 the available in vivo and in vitro data supporting not only substantially more GSH conjugation in
17 humans than in rodents, but also substantial interindividual toxicokinetic variability.
18
19 6.2.1.3.3. Liver effects. Hepatomegaly appears to be the most sensitive indicator of toxicity that
20 is available for the liver and is therefore, considered a candidate critical effect. Several studies
21 are considered reliable for developing high confidence candidate reference values for this
22 endpoint. Since they all indicated similar sensitivity but represented different species and/or
23 routes of exposure, they were all considered candidate critical studies (see Tables 5-2 and 5-9).
24 Due to the substantial evidence supporting the role of oxidative metabolism in TCE-
25 induced hepatomegaly (and evidence against TC A being the sole mediator of TCE-induced
26 hepatomegaly [Evans et al., 2009]), the preferred PBPK model dose metric for liver effects was
27 the amount of hepatic oxidative metabolism (scaled by body weight to the % power). Total
28 (hepatic and extrahepatic) oxidative metabolism (scaled by body weight to the 3/4 power) was
29 used as an alternative dose metric. With these dose metrics, the candidate reference values
30 derived using the PBPK model were only modestly (~3-fold or less) different than those derived
31 on the basis of applied dose.
32
33 6.2.1.3.4. Immunological effects. There is high qualitative confidence for TCE immunotoxicity
34 and moderate confidence in the candidate reference values that can be derived from the available
35 studies (see Tables 5-3 and 5-11). Decreased thymus weight reported at relatively low exposures
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 in nonautoimmune-prone mice is a clear indicator of immunotoxicity (Keil et al., 2009), and is
2 therefore, considered a candidate critical effect. A number of studies have also reported changes
3 in markers of immunotoxicity at relatively low exposures. Among markers for autoimmune
4 effects, the more sensitive measures of autoimmune changes in liver and spleen (Kaneko et al.,
5 2000) and increased anti-dsDNA and anti-ssDNA antibodies (early markers for systemic lupus
6 erythematosus) (Keil et al., 2009) are considered the candidate critical effects. For markers of
7 immunosuppression, the more sensitive measures of decreased PFC response (Woolhiser et al.,
8 2006), decreased stem cell bone marrow recolonization, and decreased cell-mediated response to
9 sRBC (both from Sanders et al., 1982) are considered the candidate critical effects.
10 Developmental immunological effects are discussed below as part of the summary of
11 developmental effects (see Section 6.2.1.3.6).
12 Because of the lack of specific data as to the metabolites involved and the MO A for the
13 candidate critical immunologic effects, PBPK model predictions of total metabolism (scaled by
14 body weight to the 3/4 power) was selected as the preferred dose metric based on the general
15 observation that TCE toxicity is associated with metabolism. The AUC of TCE in blood was
16 used as an alternative dose metric. With these dose metrics, the candidate reference values
17 derived using the PBPK model were, with one exception, only modestly (~3-fold or less)
18 different than those derived on the basis of applied dose. For the Woolhiser et al. (2006)
19 decreased PFC response, with the alternative dose metric of AUC of TCE in blood, BMD
20 modeling based on internal doses changed the candidate reference value by 17-fold higher than
21 the cRfC based on applied dose. However, the dose-response model fit for this effect using this
22 metric was substantially worse than the fit using the preferred metric of total oxidative
23 metabolism, with which the change in candidate reference value was only 1.3-fold.
24
25 6.2.1.3.5. Reproductive effects. While there is high qualitative confidence in the male
26 reproductive hazard posed by TCE, there is lower confidence in the reference values that can be
27 derived from the available studies of these effects (see Tables 5-4 and 5-12). Relatively high
28 PODs are derived from several studies reporting less sensitive endpoints (George et al., 1985,
29 1986; Land et al., 1981), and correspondingly higher cRfCs and cRfDs suggest that they are not
30 likely to be critical effects. The studies reporting more sensitive endpoints also tend to have
31 greater uncertainty. For the human study by Chia et al. (1996), there are uncertainties in the
32 characterization of exposure and the adversity of the effect measured in the study. For the
33 Kumar et al. (2000a, b, 2001), Forkert et al. (2002) and Kan et al. (2007) studies, the severity of
34 the sperm and testes effects appears to be continuing to increase with duration even at the end of
35 the study, so it is plausible that a lower exposure for a longer duration may elicit similar effects.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-25 DRAFT—DO NOT CITE OR QUOTE
-------
1 For the DuTeaux et al. (2004b) study, there is also duration- and low-dose extrapolation
2 uncertainty due to the short duration of the study in comparison to the time period for sperm
3 development as well as the lack of a NOAEL at the tested doses. Overall, even though there are
4 limitations in the quantitative assessment, there remains sufficient evidence to consider these to
5 be candidate critical effects.
6 There is moderate confidence both in the hazard and the candidate reference values for
7 reproductive effects other than male reproductive effects. While there are multiple studies
8 suggesting decreased maternal body weight with TCE exposure, this systemic change may not be
9 indicative of more sensitive reproductive effects. None of the estimates developed from other
10 reproductive effects is particularly uncertain or unreliable. Therefore, delayed parturition
11 (Narotsky et al., 1995) and decreased mating (George et al., 1986), which yielded the lowest
12 cRfDs, were considered candidate critical effects. These effects were also included so that
13 candidate critical reproductive effects from oral studies would not include only that reported by
14 DuTeaux et al. (2004b), from which deriving the cRfD entailed a higher degree of uncertainly.
15 Because of the general lack of specific data as to the metabolites involved and the MOA
16 for the candidate critical reproductive effects, PBPK model predictions of total metabolism
17 (scaled by body weight to the 3/4 power) was selected as the preferred dose metric based on the
18 general observation that TCE toxicity is associated with metabolism. The AUC of TCE in blood
19 was used as an alternative dose metric. The only exception to this was for the DuTeaux et al.
20 (2004) study, which suggested that local oxidative metabolism of TCE in the male reproductive
21 tract was involved in the effects reported. Therefore, in this case, AUC of TCE in blood was
22 considered the preferred dose metric, while total oxidative metabolism (scaled by body weight to
23 the 3/4 power) was considered the alternative metric. With these dose metrics, the candidate
24 reference values derived using the PBPK model were only modestly (~3.5-fold or less) different
25 than those derived on the basis of applied dose.
26
27 6.2.1.3.6. Developmental effects. There is moderate-to-high confidence both in the hazard and
28 the candidate reference values for developmental effects of TCE (see Tables 5-5 and 5-13). It is
29 also noteworthy that the PODs for the more sensitive developmental effects were similar to or, in
30 most cases, lower than the PODs for the more sensitive reproductive effects, suggesting that
31 developmental effects are not a result of paternal or maternal toxicity. Among inhalation studies,
32 candidate reference values were only developed for effects in rats reported in Healy et al. (1982),
33 of resorptions, decreased fetal weight, and delayed skeletal ossification. These were all
34 considered candidate critical developmental effects. Because resorptions were also reported in
35 oral studies, the most sensitive (rat) oral study for this effect (and most reliable for dose-response
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-26 DRAFT—DO NOT CITE OR QUOTE
-------
1 analysis) of Narotsky et al. (1995) was also selected as a candidate critical study. The
2 confidence in the oral studies and candidate reference values developed for more sensitive
3 endpoints is more moderate, but still sufficient for consideration as candidate critical effects.
4 The most sensitive endpoints by far are the increased fetal heart malformations in rats reported
5 by Johnson et al. (2003) and the developmental immunotoxicity in mice reported by Peden-
6 Adams et al. (2006), and these are both considered candidate critical effects.
7 Neurodevelopmental effects are a distinct type among developmental effects. Thus, the next
8 most sensitive endpoints of decreased rearing postexposure in mice (Fredricksson et al., 1993),
9 increased exploration postexposure in rats (Taylor et al., 1985) and decreased myelination in the
10 hippocampus of rats (Isaacson and Taylor, 1989) are also considered candidate critical effects.
11 Because of the general lack of specific data as to the metabolites involved and the MOA
12 for the candidate critical reproductive effects, PBPK model predictions of total metabolism
13 (scaled by body weight to the % power) was selected as the preferred dose metric based on the
14 general observation that TCE toxicity is associated with metabolism. The AUC of TCE in blood
15 was used as an alternative dose metric. The only exception to this was for the Johnson et al.
16 (2003) study, which suggested that oxidative metabolites were involved in the effects reported
17 based on similar effects being reported from TCA and DCA exposure. Therefore, in this case,
18 total oxidative metabolism (scaled by body weight to the % power) was considered the preferred
19 dose metric, while AUC of TCE in blood was considered the alternative metric. With these dose
20 metrics, the candidate reference values derived using the PBPK model were, with one exception,
21 only modestly (~3-fold or less) different than those derived on the basis of applied dose. For
22 resorptions reported by Narotsky et al. (1995), BMD modeling based on internal doses changed
23 the candidate reference value by 7- to 8-fold larger than the corresponding cRfD based on
24 applied dose. However, there is substantial uncertainty in the low-dose curvature of the dose-
25 response curve for modeling both with applied and internal dose, so the BMD remains somewhat
26 uncertain for this endpoint/study. Finally, for two studies (Isaacson and Taylor, 1989; Peden-
27 Adams et al., 2006), PBPK modeling of internal doses was not performed due to the inability to
28 model the complicated exposure pattern (in utero, followed by lactational transfer, followed by
29 drinking water postweaning).
30
31 6.2.1.3.7. Summary of most sensitive candidate reference values. As shown in Section 5.1.3
32 and 5.1.5, the most sensitive candidate reference values are for developmental effects of heart
33 malformations in rats (candidate RfC of 0.0004 ppm and candidate RfD of 0.0005 mg/kg/d),
34 developmental immunotoxicity in mice exposed pre- and postnatally (candidate RfD of
35 0.0004 mg/kg/d), immunological effects in mice (lowest candidate RfCs of 0.0003-0.003 ppm
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 and lowest candidate RfDs of 0.0005-0.005 mg/kg/d), and kidney effects in rats and mice
2 (candidate RfCs of 0.0006-0.002 ppm and candidate RfDs of 0.0003-0.001 mg/kg/d). The most
3 sensitive candidate reference values also generally have low composite uncertainty factors (with
4 the exception of some mouse immunological and kidney effects), so are expected to be reflective
5 of the most sensitive effects as well. Thus, the most sensitive candidate references values for
6 multiple effects span about an order of magnitude for both inhalation (0.0003-0.003 ppm
7 [0.002-0.02 mg/m3]) and oral (0.0004-0.005 mg/kg/d) exposures. The most sensitive candidate
8 references values for neurological and reproductive effects are about an order of magnitude
9 higher (lowest candidate RfCs of 0.007-0.02 ppm [0.04-0.1 mg/m3] and lowest candidate RfDs
10 of 0.009-0.02 mg/kg/d). Lastly, the liver effects have candidate reference values that are another
11 2 orders of magnitude higher (candidate RfCs of 1-2 ppm [6-10 mg/m3] and candidate RfDs of
12 0.9-2 mg/kg/d).
13
14 6.2.1.4. Noncancer Reference Values (see Section 5.1.5)
15 6.2.1.4.1. Reference concentration. The goal is to select an overall RfC that is well supported
16 by the available data (i.e., without excessive uncertainty given the extensive database) and
17 protective for all the candidate critical effects, recognizing that individual candidate RfC values
18 are by nature somewhat imprecise. As discussed in Section 5.1 in Chapter 5, the lowest
19 candidate RfC values within each health effect category span a 3000-fold range from 0.0003-
20 0.9 ppm (see Table 5-21). One approach to selecting a RfC would be to select the lowest
21 calculated value of 0.0003 ppm for decreased thymus weight in mice. However, six candidate
22 RfCs (cRfCs and p-cRfCs) from both oral and inhalation studies are in the relatively narrow
23 range of 0.0003-0.003 ppm at the low end of the overall range (see Table 5-19). Given the
24 somewhat imprecise nature of the individual candidate RfC values, and the fact that multiple
25 effects/studies lead to similar candidate RfC values, the approach taken in this assessment is to
26 select a RfC supported by multiple effects/studies. The advantages of this approach, which is
27 only possible when there is a relatively large database of studies/effects and when multiple
28 candidate values happen to fall within a narrow range at the low end of the overall range, are that
29 it leads to a more robust RfC (less sensitive to limitations of individual studies) and that it
30 provides the important characterization that the RfC exposure level is similar for multiple
31 noncancer effects rather than being based on a sole explicit critical effect.
32 Therefore, six critical studies/effects were chosen to support the RfC for TCE noncancer
33 effects (see Table 5-23). Five of the lowest candidate RfCs, ranging from 0.0003-0.003 ppm for
34 developmental, kidney, and immunologic effects, are values derived from route-to-route
35 extrapolation using the PBPK model. The lowest candidate RfC estimate from an inhalation
36 study is 0.001 ppm for kidney effects. For all six candidate RfCs, the PBPK model was used for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 inter- and intraspecies extrapolation, based on the preferred dose metric for each endpoint. There
2 is high confidence in the candidate RfCs for kidney effects for the following reasons: they are
3 based on clearly adverse effects, two of the values are derived from chronic studies, and the
4 extrapolation to humans is based on dose metrics clearly related to toxicity estimated with high
5 confidence with the PBPK model developed in Section 3.5. There is somewhat less confidence
6 in the lowest candidate RfC for developmental effects (heart malformations) (see
7 Section 5.1.2.8), and the lowest candidate RfC estimates for immunological effects (see
8 Section 5.1.2.5). Thus, this assessment does not rely on any single estimate alone; however,
9 each estimate is supported by estimates of similar magnitude from other effects.
10 As a whole, the estimates support a preferred RfC estimate of 0.001 ppm (1 ppb or
11 5 ug/m3). This estimate is within approximately a factor of 3 of the lowest estimates of
12 0.0003 ppm for decreased thymus weight in mice, 0.0004 ppm for heart malformations in rats,
13 0.0006 ppm for toxic nephropathy in rats, 0.001 ppm for increased kidney weight in rats,
14 0.002 ppm for toxic nephrosis in mice, and 0.003 ppm for increased anti-dsDNA antibodies in
15 mice. Thus, there is robust support for an RfC of 0.001 ppm provided by estimates for multiple
16 effects from multiple studies. The estimates are based on PBPK model-based estimates of
17 internal dose for interspecies, intraspecies, and/or route-to-route extrapolation, and there is
18 sufficient confidence in the PBPK model, as well as support from mechanistic data for some of
19 the dose metrics (specifically total oxidative metabolism for the heart malformations and
20 bioactivation ofDCVC and total GSH metabolism for toxic nephropathy) (see Section 5.1.3.1).
21 Note that there is some human evidence of developmental heart defects from TCE exposure in
22 community studies (see Section 4.8.3.1.1) and of kidney toxicity in TCE-exposed workers (see
23 Section 4.4.1).
24 In summary, the preferred RfC estimate is 0.001 ppm (1 ppb or 5 ug/m3) based on route-
25 to-route extrapolated results from oral studies for the critical effects of heart malformations
26 (rats), immunotoxicity (mice), and toxic nephropathy (rats, mice), and an inhalation study for the
27 critical effect of increased kidney weight (rats).
28
29 6.2.1.4.2. Reference dose. As with the RfC determination above, the goal is to select an overall
30 RfD that is well supported by the available data (i.e., without excessive uncertainty given the
31 extensive database) and protective for all the candidate critical effects, recognizing that
32 individual candidate RfD values are by nature somewhat imprecise. As discussed in Section 5.1
33 in Chapter 5, the lowest candidate RfD values (cRfDs and p-cRfDs) within each health effect
34 category span a nearly 3000-fold range from 0.0003-0.8 mg/kg/d (see Table 5-21). However,
35 four candidate RfDs from oral studies are in the relatively narrow range of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 0.0003-0.0005 mg/kg/d at the low end of the overall range. Given the somewhat imprecise
2 nature of the individual candidate RfD values, and the fact that multiple effects/studies lead to
3 similar candidate RfD values, the approach taken in this assessment is to select a RfD supported
4 by multiple effects/studies. The advantages of this approach, which is only possible when there
5 is a relatively large database of studies/effects and when multiple candidate values happen to fall
6 within a narrow range at the low end of the overall range, are that it leads to a more robust RfD
7 (less sensitive to limitations of individual studies) and that it provides the important
8 characterization that the RfD exposure level is similar for multiple noncancer effects rather than
9 being based on a sole explicit critical effect.
10 Therefore, four critical studies/effects were chosen to support the RfD for TCE noncancer
11 effects (see Table 5-24). Three of the lowest candidate RfDs—0.0003 mg/kg/d for toxic
12 nephropathy in rats, and 0.0005 mg/kg/d for heart malformations in rats and decreased thymus
13 weights in mice—are derived using the PBPK model for inter- and intraspecies extrapolation,
14 based on the preferred dose metric for each endpoint. The other of these lowest candidate
15 RfDs—0.0004 mg/kg/d for developmental immunotoxicity (decreased PFC response and
16 increased delayed-type hypersensitivity) in mice—is based on applied dose. There is high
17 confidence in the candidate RfD for kidney effects (see Section 5.1.2.2), which is based on
18 clearly adverse effects, derived from a chronic study, and extrapolated to humans based on a
19 dose metric clearly related to toxicity estimated with high confidence with the PBPK model
20 developed in Section 3.5. There is somewhat less confidence in the candidate RfDs for
21 decreased thymus weights (see Section 5.1.2.5) and heart malformations and developmental
22 immunological effects (see Section 5.1.2.8). Thus, this assessment does not rely on any single
23 estimate alone; however, each estimate is supported by estimates of similar magnitude from
24 other effects. As a whole, the estimates support a preferred RfD of 0.0004 mg/kg/d. This
25 estimate is within 25% of the lowest estimates of 0.0003 for toxic nephropathy in rats,
26 0.0004 mg/kg/d for developmental immunotoxicity (decreased PFC and increased delayed-type
27 hypersensitivity) in mice, and 0.0005 mg/kg/d for heart malformations in rats and decreased
28 thymus weights in mice. Thus, there is strong, robust support for an RfD of 0.0004 mg/kg/d
29 provided by the concordance of estimates derived from multiple effects from multiple studies.
30 The estimates for kidney effects, thymus effects, and developmental heart malformations are
31 based on PBPK model-based estimates of internal dose for interspecies and intraspecies
32 extrapolation, and there is sufficient confidence in the PBPK model, as well as support from
33 mechanistic data for some of the dose metrics (specifically total oxidative metabolism for the
34 heart malformations and bioactivation of DCVC for toxic nephropathy) (see Section 5.1.3.1).
35 Note that there is some human evidence of developmental heart defects from TCE exposure in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-30 DRAFT—DO NOT CITE OR QUOTE
-------
1 community studies (see Section 4.8.3.1.1) and of kidney toxicity in TCE-exposed workers (see
2 Section 4.4.1).
3 In summary, the preferred RfD estimate is 0.0004 mg/kg/d based on the critical effects of
4 heart malformations (rats), adult immunological effects (mice), developmental immunotoxicity
5 (mice), and toxic nephropathy (rats).
6
7 6.2.2. Cancer (see Section 5.2)
8 6.2.2.1. Background and Methods (rodent: see Section 5.2.1.1; human: see Section 5.2.2.1)
9 As summarized above, following U.S. EPA (2005a) Guidelines for Carcinogen Risk
10 Assessment, TCE is characterized as "Carcinogenic to Humans" by all routes of exposure, based
11 on convincing evidence of a causal association between TCE exposure in humans and kidney
12 cancer, but there is also human evidence of TCE carcinogenicity in the liver and lymphoid
13 tissues. This conclusion is further supported by rodent bioassay data indicating carcinogenicity
14 of TCE in rats and mice at tumor sites that include those identified in human epidemiologic
15 studies. Therefore, both human epidemiologic studies as well as rodent bioassays were
16 considered for deriving PODs for dose-response assessment of cancer endpoints. For PODs
17 derived from rodent bioassays, default dosimetry procedures were applied to convert applied
18 rodent doses to human equivalent doses. Essentially, for inhalation exposures, "ppm
19 equivalence" across species was assumed. For oral doses, 3/4-power body-weight scaling was
20 used, with a default average human body weight of 70 kg. In addition to applied doses, several
21 internal dose metrics estimated using a PBPK model for TCE and its metabolites were used in
22 the dose-response modeling for each tumor type. In general, an attempt was made to use tissue-
23 specific dose metrics representing particular pathways or metabolites identified from available
24 data as having a likely role in the induction of a tissue-specific cancer. Where insufficient
25 information was available to establish particular metabolites or pathways of likely relevance to a
26 tissue-specific cancer, more general "upstream" metrics had to be used. In addition, the selection
27 of dose metrics was limited to metrics that could be adequately estimated by the PBPK model.
28 Regarding low-dose extrapolation, a key consideration in determining what extrapolation
29 approach to use is the MOA(s). However, MOA data are lacking or limited for each of the
30 cancer responses associated with TCE exposure, with the exception of the kidney tumors. For
31 the kidney tumors, the weight of the available evidence supports the conclusion that a mutagenic
32 MOA is operative; this MOA supports linear low-dose extrapolation. For the other TCE-induced
33 tumors, the MOA(s) is unknown. When the MOA(s) cannot be clearly defined, U.S. EPA
34 generally uses a linear approach to estimate low-dose risk (U.S. EPA, 2005a), based on the
35 following general principles:
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 • A chemical's carcinogenic effects may act additively to ongoing biological processes,
2 given that diverse human populations are already exposed to other agents and have
3 substantial background incidences of various cancers.
4 • A broadening of the dose-response curve (i.e., less rapid fall-off of response with
5 decreasing dose) in diverse human populations and, accordingly, a greater potential for
6 risks from low-dose exposures (Ziese et al., 1987; Lutz et al., 2005) is expected for two
7 reasons: First, even if there is a "threshold" concentration for effects at the cellular level,
8 that threshold is expected to differ across individuals. Second, greater variability in
9 response to exposures would be anticipated in heterogeneous populations than in inbred
10 laboratory species under controlled conditions (due to, e.g., genetic variability, disease
11 status, age, nutrition, and smoking status).
12 • The general use of linear extrapolation provides reasonable upper-bound estimates that
13 are believed to be health-protective (U.S. EPA, 2005a) and also provides consistency
14 across assessments.
15 6.2.2.2. Inhalation Unit Risk Estimate (rodent: see Section 5.2.1.3; human: see
16 Section 5.2.2.1 and 5.2.2.2)
17 The inhalation unit risk for TCE is defined as a plausible upper bound lifetime extra risk
18 of cancer from chronic inhalation of TCE per unit of air concentration. The preferred estimate of
19 the inhalation unit risk for TCE is 2.20 x 10~2 per ppm (2 x 10~2 per ppm [4 x 10~6 per jig/m3]
20 rounded to 1 significant figure), based on human kidney cancer risks reported by Charbotel et al.
21 (2006) and adjusted for potential risk for tumors at multiple sites. This estimate is based on
22 good-quality human data, thus, avoiding the uncertainties inherent in interspecies extrapolation.
23 The Charbotel et al. (2006) case-control study of 86 incident renal cell carcinoma (RCC) cases
24 and 316 age- and sex-matched controls, with individual cumulative exposure estimates for TCE
25 inhalation for each subject, provides a sufficient human data set for deriving quantitative cancer
26 risk estimates for RCC in humans. The study is a high-quality study which used a detailed
27 exposure assessment (Fevotte et al., 2006) and took numerous potential confounding factors,
28 including exposure to other chemicals, into account. A significant dose-response relationship
29 was reported for cumulative TCE exposure and RCC (Charbotel et al., 2006). Human data on
30 TCE exposure and cancer risk sufficient for dose-response modeling are only available for RCC,
31 yet human and rodent data suggest that TCE exposure increases the risk of cancer at other sites
32 as well. In particular, there is evidence from human (and rodent) studies for increased risks of
33 lymphoma and liver cancer. Therefore, the inhalation unit risk estimate derived from human
34 data for RCC incidence was adjusted to account for potential increased risk of those tumor types.
35 To make this adjustment, a factor accounting for the relative contributions to the extra risk for
36 cancer incidence from TCE exposure for these three tumor types combined versus the extra risk
37 for RCC alone was estimated, and this factor was applied to the unit risk estimate for RCC to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-32 DRAFT—DO NOT CITE OR QUOTE
-------
1 obtain a unit risk estimate for the three tumor types combined (i.e., lifetime extra risk for
2 developing any of the 3 types of tumor). This estimate is considered a better estimate of total
3 cancer risk from TCE exposure than the estimate for RCC alone. Although only the Charbotel et
4 al. (2006) study was found adequate for direct estimation of inhalation unit risks, the available
5 epidemiologic data provide sufficient information for estimating the relative potency of TCE
6 across tumor sites. In particular, the relative contributions to extra risk (for cancer incidence)
7 were calculated from two different data sets to derive the adjustment factor for adjusting the unit
8 risk estimate for RCC to a unit risk estimate for the 3 types of cancers (RCC, lymphoma, and
9 liver) combined. The first calculation is based on the results of the meta-analyses of human
10 epidemiologic data for the 3 tumor types; the second calculation is based on the results of the
11 Raaschou-Nielsen et al. (2003) study, the largest single human epidemiologic study by far with
12 RR estimates for all 3 tumor types. Both calculations support a 4-fold adjustment factor.
13 The preferred estimate of the inhalation unit risk based on human epidemiologic data is
14 supported by inhalation unit risk estimates from multiple rodent bioassays, the most sensitive of
15 which range from 1 x 10~2 to 2 x 10"1 per ppm [2 x 10"6 to 3 x 10~5 per jig/m3]. From the
16 inhalation bioassays selected for analysis in Section 5.2.1.1, and using the preferred PBPK
17 model-based dose metrics, the inhalation unit risk estimate for the most sensitive sex/species is
18 8 x 10~2 per ppm [2 x 10~5 per |ig/m3], based on kidney adenomas and carcinomas reported by
19 Maltoni et al. (1986) for male Sprague-Dawley rats. Leukemias and Leydig cell tumors were
20 also increased in these rats, and, although a combined analysis for these tumor types which
21 incorporated the different site-specific preferred dose metrics was not performed, the result of
22 such an analysis is expected to be similar, about 9 x 10~2 per ppm [2 x 10~5 per |ig/m3]. The next
23 most sensitive sex/species from the inhalation bioassays is the female mouse, for which
24 lymphomas were reported by Henschler et al. (1980); these data yield a unit risk estimate of
25 1.0 x 10~2 per ppm [2 x KT6 per |ig/m3]. In addition, the 90% confidence intervals (i.e., 5% to
26 95% bounds) reported in Table 5-34 for male rat kidney tumors from Maltoni et al. (1986) and
27 female mouse lymphomas from Henschler et al. (1980), derived from the quantitative analysis of
28 PBPK model uncertainty, both included the estimate based on human data of 2 x 10~2 per ppm.
29 Furthermore, PBPK model-based route-to-route extrapolation of the results for the most sensitive
30 sex/species from the oral bioassays, kidney tumors in male Osborne-Mendel rats and testicular
31 tumors in Marshall rats (NTP, 1988), leads to inhalation unit risk estimates of 2 x 10"1 per ppm
32 [3 x 10~5 per |ig/m3] and 4 x 10~2 per ppm [8 x 10"6 per |ig/m3], respectively, with the preferred
33 estimate based on human data falling within the route-to-route extrapolation of the 90%
34 confidence intervals reported in Table 5-35. Finally, for all these estimates, the ratios of BMDs
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 to the BMDLs did not exceed a value of 3, indicating that the uncertainties in the dose-response
2 modeling for determining the POD in the observable range are small.
3 Although there are uncertainties in these various estimates, confidence in the proposed
4 inhalation unit risk estimate of 2 x 1CT2 per ppm [4 x 1CT6 per |ig/m3], based on human kidney
5 cancer risks reported by Charbotel et al. (2006) and adjusted for potential risk for tumors at
6 multiple sites (as summarized above in Section 6.1.4), is further increased by the similarity of
7 this estimate to estimates based on multiple rodent data sets. Application of the ADAF for
8 kidney cancer risks due to the weight of evidence supporting a mutagenic MOA for this endpoint
9 is summarized below in Section 6.2.2.5.
10
11 6.2.2.3. Oral Unit Risk Estimate (rodent: see Section 5.2.1.3; human: see Section 5.2.2.3)
12 The oral unit risk (or slope factor) for TCE is defined as a plausible upper bound lifetime
13 extra risk of cancer from chronic ingestion of TCE per mg/kg/d oral dose. The preferred
14 estimate of the oral unit risk is 4.63 x 10~2 per mg/kg/d (5 x 10~2 per mg/kg/d rounded to
15 1 significant figure), resulting from PBPK model-based route-to-route extrapolation of the
16 inhalation unit risk estimate based on the human kidney cancer risks reported in Charbotel et al.
17 (2006) and adjusted for potential risk for tumors at multiple sites. This estimate is based on
18 good-quality human data, thus, avoiding uncertainties inherent in interspecies extrapolation. In
19 addition, uncertainty in the PBPK model-based route-to-route extrapolation is relatively low
20 (Chiu and White, 2006; Chiu, 2006). In this particular case, extrapolation using different dose
21 metrics yielded expected population mean risks within about a 2-fold range, and, for any
22 particular dose metric, the 95% confidence interval for the extrapolated population mean risks
23 for each site spanned a range of no more than about 3-fold.
24 This value is supported by oral unit risk estimates from multiple rodent bioassays, the
25 most sensitive of which range from 3 x 10~2 to 3 x 10"1 per mg/kg/d. From the oral bioassays
26 selected for analysis in Section 5.2.1.1, and using the preferred PBPK model-based dose metrics,
27 the oral unit risk estimate for the most sensitive sex/species is 3 x 10"1 per mg/kg/d, based on
28 kidney tumors in male Osborne-Mendel rats (NTP, 1988). The oral unit risk estimate for
29 testicular tumors in male Marshall rats (NTP, 1988) is somewhat lower at 7 x 10~2 per mg/kg/d.
30 The next most sensitive sex/species result from the oral studies is for male mouse liver tumors
31 (NCI, 1976), with an oral unit risk estimate of 3 x 10~2 per mg/kg/d. In addition, the 90%
32 confidence intervals reported in Table 5-35 for male Osborne-Mendel rat kidney tumors (NTP,
33 1988), male F344 rat kidney tumors (NTP, 1990), and male Marshall rat testicular tumors (NTP,
34 1988), derived from the quantitative analysis of PBPK model uncertainty, all included the
35 estimate based on human data of 5 x 10~2 per mg/kg/d, while the upper 95% confidence bound
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-34 DRAFT—DO NOT CITE OR QUOTE
-------
1 for male mouse liver tumors from NCI (1976) was slightly below this value at 4 x 10 2 per
2 mg/kg/d. Furthermore, PBPK model-based route-to-route extrapolation of the most sensitive
3 endpoint from the inhalation bioassays, male rat kidney tumors from Maltoni et al. (1986), leads
4 to an oral unit risk estimate of 1 x 1CT1 per mg/kg/d, with the preferred estimate based on human
5 data falling within the route-to-route extrapolation of the 90% confidence interval reported in
6 Table 5-34. Finally, for all these estimates, the ratios of BMDs to the BMDLs did not exceed a
7 value of 3, indicating that the uncertainties in the dose-response modeling for determining the
8 POD in the observable range are small.
9 Although there are uncertainties in these various estimates, confidence in the proposed
10 oral unit risk estimate of 5 x 10~2 per mg/kg/d, resulting from PBPK model-based route-to-route
11 extrapolation of the inhalation unit risk estimate based on the human kidney cancer risks reported
12 in Charbotel et al. (2006) and adjusted for potential risk for tumors at multiple sites (as
13 summarized above), is further increased by the similarity of this estimate to estimates based on
14 multiple rodent data sets. Application of the ADAF for kidney cancer risks due to the weight of
15 evidence supporting a mutagenic MOA for this endpoint is summarized below in Section 6.2.2.5.
16
17 6.2.2.4. Uncertainties in Cancer Dose-Response Assessment
18 6.2.2.4.1. Uncertainties in estimates based on human epidemiologic data (see
19 Section 5.2.2.1.3). All risk assessments involve uncertainty, as study data are extrapolated to
20 make inferences about potential effects in humans from environmental exposure. The preferred
21 values for the unit risk estimates are based on good quality human data, which avoids
22 interspecies extrapolation, one of the major sources of uncertainty in quantitative cancer risk
23 estimates.
24 A remaining major uncertainty in the unit risk estimate for RCC incidence derived from
25 the Charbotel et al. (2006) is the extrapolation from occupational exposures to lower
26 environmental exposures. There was some evidence of a contribution to increased RCC risk
27 from peak exposures; however, there remained an apparent dose-response relationship for RCC
28 risk with increasing cumulative exposure without peaks, and the OR for exposure with peaks
29 compared to exposure without peaks was not significantly elevated (Charbotel et al., 2006).
30 Although the actual exposure-response relationship at low exposure levels is unknown, the
31 conclusion that a mutagenic MOA is operative for TCE-induced kidney tumors supports the
32 linear low-dose extrapolation that was used (U.S. EPA, 2005a). Additional support for use of
33 linear extrapolation is discussed above in Section 6.2.2.1.
34 In addition, because a linear model was used in the observable range of the human data
35 and the POD was within the low-dose linear range for extra risk as a function of exposure, linear
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 extrapolation below the 95% lower confidence limit of the effective concentration for a 1%
2 response (LECoi) is virtually a straight continuation of the 95% upper confidence limit on the
3 linear model used above the LECoi. Thus, the use of linear extrapolation from the POD differed
4 negligibly from extrapolation of the dose-response model itself to low dose.
5 With respect to uncertainties in the dose-response modeling, the two-step approach of
6 modeling only in the observable range, as put forth in U.S. EPA's Guidelines for Carcinogen
1 Risk Assessment (U.S. EPA, 2005a), is designed in part to minimize model dependence. The
8 ratio of the maximum likelihood estimate of the effective concentration for a 1% response (ECoi)
9 to the LECoi, which gives some indication of the uncertainties in the dose-response modeling,
10 was about a factor of 2. Thus, overall, modeling uncertainties in the observable range are
11 considered to be negligible.
12 An important source of uncertainty in the underlying Charbotel et al. (2006) study is the
13 retrospective estimation of TCE exposures in the study subjects. This case-control study was
14 conducted in the Arve Valley in France, a region with a high concentration of screw cutting
15 workshops using TCE and other degreasing agents. Since the 1960s, occupational physicians of
16 the region have collected a large quantity of well-documented measurements, including TCE air
17 concentrations and urinary metabolite levels (Fevotte et al., 2006). The study investigators
18 conducted a comprehensive exposure assessment to estimate cumulative TCE exposures for the
19 individual study subjects, using a detailed occupational questionnaire with a customized task-
20 exposure matrix for the screw-cutting workers and a more general occupational questionnaire for
21 workers exposed to TCE in other industries (Fevotte et al., 2006). The exposure assessment also
22 attempted to take dermal exposure from hand-dipping practices into account by equating it with
23 an equivalent airborne concentration based on biological monitoring data. Despite the
24 appreciable effort of the investigators, considerable uncertainty associated with any retrospective
25 exposure assessment is inevitable, and some exposure misclassification is unavoidable. Such
26 exposure misclassification was most likely for the 19 deceased cases and their matched controls,
27 for which proxy respondents were used, and for exposures outside the screw-cutting industry
28 (295 of 1,486 identified job periods involved TCE exposure; 120 of these were not in the screw-
29 cutting industry).
30 Another noteworthy source of uncertainty in the Charbotel et al. (2006) study is the
31 possible influence of potential confounding or modifying factors. This study population, with a
32 high prevalence of metal-working, also had relatively high prevalences of exposure to petroleum
33 oils, cadmium, petroleum solvents, welding fumes, and asbestos (Fevotte et al., 2006). Other
34 exposures assessed included other solvents (including other chlorinated solvents), lead, and
35 ionizing radiation. None of these exposures was found to be significantly associated with RCC
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-36 DRAFT—DO NOT CITE OR QUOTE
-------
1 atap = 0.05 significance level. Cutting fluids and other petroleum oils were associated with
2 RCC at ap = 0.1 significance level; however, further modeling suggested no association with
3 RCC when other significant factors were taken into account (Charbotel et al., 2006). The
4 medical questionnaire included familial kidney disease and medical history, such as kidney
5 stones, infection, chronic dialysis, hypertension, and use of anti-hypertensive drugs, diuretics,
6 and analgesics. Body mass index (BMI) was also calculated, and lifestyle information such as
7 smoking habits and coffee consumption was collected. Univariate analyses found high levels of
8 smoking and BMI to be associated with increased odds of RCC, and these two variables were
9 included in the conditional logistic regressions. Thus, although impacts of other factors are
10 possible, this study took great pains to attempt to account for potential confounding or modifying
11 factors.
12 Some other sources of uncertainty associated with the epidemiological data are the dose
13 metric and lag period. As discussed above, there was some evidence of a contribution to
14 increased RCC risk from peak TCE exposures; however, there appeared to be an independent
15 effect of cumulative exposure without peaks. Cumulative exposure is considered a good
16 measure of total exposure because it integrates exposure (levels) over time. If there is a
17 contributing effect of peak exposures, not already taken into account in the cumulative exposure
18 metric, the linear slope may be overestimated to some extent. Sometimes cancer data are
19 modeled with the inclusion of a lag period to discount more recent exposures not likely to have
20 contributed to the onset of cancer. In an unpublished report (Charbotel et al., 2005), Charbotel et
21 al. also present the results of a conditional logistic regression with a 10-year lag period, and these
22 results are very similar to the unlagged results reported in their published paper, suggesting that
23 the lag period might not be an important factor in this study.
24 Some additional sources of uncertainty are not so much inherent in the exposure-response
25 modeling or in the epidemiologic data themselves but, rather, arise in the process of obtaining
26 more general Agency risk estimates from the epidemiologic results. U.S. EPA cancer risk
27 estimates are typically derived to represent an upper bound on increased risk of cancer incidence
28 for all sites affected by an agent for the general population. From experimental animal studies,
29 this is accomplished by using tumor incidence data and summing across all the tumor sites that
30 demonstrate significantly increased incidences, customarily for the most sensitive sex and
31 species, to attempt to be protective of the general human population. However, in estimating
32 comparable risks from the Charbotel et al. (2006) epidemiologic data, certain limitations are
33 encountered. For one thing, these epidemiology data represent a geographically limited (Arve
34 Valley, France) and likely not very diverse population of working adults. Thus, there is
35 uncertainty about the applicability of the results to a more diverse general population.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-37 DRAFT—DO NOT CITE OR QUOTE
-------
1 Additionally, the Charbotel et al. (2006) study was a study of RCC only, and so the risk
2 estimate derived from it does not represent all the tumor sites that may be affected by TCE. This
3 uncertainty was addressed by adjusting the RCC estimate to multiple sites, but there are also
4 uncertainties related to the assumptions inherent in the calculations for this adjustment. As
5 discussed in Section 5.2.2.2, adequate quantitative dose-response data were only available for
6 one cancer site in humans, so other human data were used to adjust the estimate derived for RCC
7 to include risk for other cancers with substantial human evidence of hazard (lymphoma and liver
8 cancer). The relative contributions to extra risk (for cancer incidence) were calculated from two
9 different data sets to derive an adjustment factor. The first calculation is based on the results of
10 the meta-analyses for the 3 tumor types; the second calculation is based on the results of the
11 Raaschou-Nielsen et al. (2003) study, the largest single study by far with RR estimates for all 3
12 tumor types. The fact that the calculations based on 2 different data sets yielded comparable
13 values for the adjustment factor provides more robust support for the use of the factor of 4.
14 Additional uncertainties pertain to the weight of evidence supporting the association of TCE
15 exposure with increased risk of cancer for the 3 tumor types. As discussed in Section 4.11.2, it is
16 concluded that the weight of evidence for kidney cancer is sufficient to classify TCE as
17 "Carcinogenic to Humans." It is also concluded that there is strong evidence that TCE causes
18 lymphoma as well, although the evidence for liver cancer was more limited. In addition, the
19 rodent studies demonstrate clear evidence of multisite carcinogenicity, with tumor types
20 including those for which associations with TCE exposure are observed in human studies, i.e.,
21 liver and kidney cancers and lymphomas. Overall, the evidence is sufficiently persuasive to
22 support the use of the adjustment factor of 4 based on these 3 tumor types. Alternatively, if one
23 were to use the factor based only on the 2 tumor types with the strongest evidence, the cancer
24 inhalation unit risk estimate would be only slightly reduced (25%).
25 Finally, the preferred value for the oral unit risk estimate was based on route-to-route
26 extrapolation of the inhalation unit risk based on human data using predictions from the PBPK
27 model. Because different internal dose metrics are preferred for each target tissue site, a separate
28 route-to-route extrapolation was performed for each site-specific unit risk estimate. As discussed
29 above, uncertainty in the PBPK model-based route-to-route extrapolation is relatively low (Chiu
30 and White, 2006; Chiu, 2006). In this particular case, extrapolation using different dose metrics
31 yielded expected population mean risks within about a 2-fold range, and, for any particular dose
32 metric, the 95% confidence interval for the extrapolated population mean risks for each site
33 spanned a range of no more than about 3-fold.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 6.2.2.4.2. Uncertainties in estimates based on rodent bioassays (see Section 5.2.1.4). With
2 respect to rodent-based cancer risk estimates, the cancer risk is typically estimated from the total
3 cancer burden from all sites that demonstrate an increased tumor incidence for the most sensitive
4 experimental species and sex. It is expected that this approach is protective of the human
5 population, which is more diverse but is exposed to lower exposure levels. In the case of TCE,
6 the impact of selection of the bioassay is limited, since, as discussed in Sections 5.2.1.3 and
7 5.2.3, estimates based on the two or three most sensitive bioassays are within an order of
8 magnitude of each other, and are consistent across routes of exposure when extrapolated using
9 the PBPK model.
10 Another source of uncertainty in the TCE rodent-based cancer risk estimates is
11 interspecies extrapolation. Several plausible PBPK model-based dose metrics were used for
12 extrapolation of toxicokinetics, but the cancer unit risk estimates obtained using the preferred
13 dose metrics were generally similar (within about 3-fold) to those derived using default
14 dosimetry assumptions, with the exception of the bioactivated DCVC dose metric for rat kidney
15 tumors and the metric for the amount of TCE oxidized in the respiratory tract for mouse lung
16 tumors occurring from oral exposure. However, there is greater biological support for these
17 selected dose metrics. The uncertainty in the PBPK model predictions themselves were analyzed
18 quantitatively through an analysis of the impact of parameter uncertainties in the PBPK model.
19 The 95% lower bounds on the BMD including parameter uncertainties in the PBPK model were
20 no more than 4-fold lower than those based on central estimates of the PBPK model predictions.
21 The greatest uncertainly was for unit risks derived from rat kidney tumors, primarily reflecting
22 the substantial uncertainty in the rat internal dose.
23 Regarding low-dose extrapolation, a key consideration in determining what extrapolation
24 approach to use is the MOA(s). However, MOA data are lacking or limited for each of the
25 cancer responses associated with TCE exposure, with the exception of the kidney tumors. For
26 the kidney tumors, the weight of the available evidence supports the conclusion that a mutagenic
27 MOA is operative; this MOA supports linear low-dose extrapolation. For the other TCE-induced
28 tumors, the MOA(s) is unknown. When the MOA(s) cannot be clearly defined, U.S. EPA
29 generally uses a linear approach to estimate low-dose risk (U.S. EPA, 2005a), based on the
30 general principles discussed above.
31 With respect to uncertainties in the dose-response modeling, the two-step approach of
32 modeling only in the observable range, as put forth in U.S. EPA's Guidelines for Carcinogen
33 Risk Assessment (U.S. EPA, 2005a), is designed in part to minimize model dependence. The
34 ratios of the BMDs to the BMDLs, which give some indication of the uncertainties in the dose-
35 response modeling, did not exceed a value of 2.5 for all the primary analyses used in this
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 assessment. Thus, overall, modeling uncertainties in the observable range are considered to be
2 negligible. Some additional uncertainty is conveyed by uncertainties in the survival adjustments
3 made to some of the bioassay data; however, a comparison of the results of two different survival
4 adjustment methods suggest that their impact is minimal relative to the uncertainties already
5 discussed.
6
7 6.2.2.5. Application of Age-Dependent Adjustment Factors (see Section 5.2.3.3)
8 When there is sufficient weight of evidence to conclude that a carcinogen operates
9 through a mutagenic MO A, and in the absence of chemical-specific data on age-specific
10 susceptibility, U.S. EPA's Supplemental Guidance for Assessing Susceptibility from Early-Life
11 Exposure to Carcinogens (U.S. EPA, 2005b) recommends the application of default ADAFs to
12 adjust for potential increased susceptibility from early-life exposure. See the Supplemental
13 Guidance for detailed information on the general application of these adjustment factors. In
14 brief, the Supplemental Guidance establishes ADAFs for three specific age groups. The current
15 ADAFs and their age groupings are 10 for <2 years, 3 for 2 to <16 years, and 1 for 16 years and
16 above (U.S. EPA, 2005b). For risk assessments based on specific exposure assessments, the
17 10-fold and 3-fold adjustments to the unit risk estimates are to be combined with age-specific
18 exposure estimates when estimating cancer risks from early-life (<16 years age) exposure. The
19 ADAFs and their age groups may be revised over time. The most current information on the
20 application of ADAFs for cancer risk assessment can be found at
21 www.epa.gov/cancerguidelines.
22 In the case of TCE, the inhalation and oral unit risk estimates reflect lifetime risk for
23 cancer at multiple sites, and a mutagenic MOA has been established for one of these sites, the
24 kidney. In addition, as discussed in Section 4.10, inadequate TCE-specific data exists to quantify
25 early-life susceptibility to TCE carcinogenicity; therefore, as recommended in the Supplemental
26 Guidance, the default ADAFs are used. As illustrated in the example calculations in
27 Section 5.2.3.3, application of the default ADAFs to the kidney cancer inhalation and oral unit
28 risk estimates for TCE is likely to have minimal impact on the total cancer risk except when
29 exposure is primarily during early life.
30 In addition to the uncertainties discussed above for the inhalation and oral total cancer
31 unit risk estimates, there are uncertainties in the application of ADAFs to adjust for potential
32 increased early-life susceptibility. The adjustment is made only for the kidney cancer component
33 of total cancer risk because that is the tumor type for which the weight of evidence was sufficient
34 to conclude that TCE-induced carcinogenesis operates through a mutagenic MOA. However, it
35 may be that TCE operates through a mutagenic MOA for other tumor types as well or that it
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-40 DRAFT—DO NOT CITE OR QUOTE
-------
1 operates through other MO As that might also convey increased early-life susceptibility.
2 Additionally, the ADAFs are general default factors, and it is uncertain to what extent they
3 reflect increased early-life susceptibility for exposure to TCE, if increased early-life
4 susceptibility occurs.
5
6 6.3. OVERALL CHARACTERIZATION OF TCE HAZARD AND DOSE RESPONSE
7 There is substantial potential for human exposure to TCE, as it has a widespread presence
8 in ambient air, indoor air, soil, and groundwater. At the same time, humans are likely to be
9 exposed to a variety of compounds that are either metabolites of TCE or which have common
10 metabolites or targets of toxicity. Once exposed, humans, as well as laboratory animal species,
11 rapidly absorb TCE, which is then distributed to tissues via systemic circulation, extensively
12 metabolized, and then excreted primarily in breath as unchanged TCE or CO2, or in urine as
13 metabolites.
14 Based on the available human epidemiologic data and experimental and mechanistic
15 studies, it is concluded that TCE poses a potential human health hazard for noncancer toxicity to
16 the central nervous system, the kidney, the liver, the immune system, the male reproductive
17 system, and the developing fetus. The evidence is more limited for TCE toxicity to the
18 respiratory tract and female reproductive system. Following U.S. EPA (2005a) Guidelines for
19 Carcinogen Risk Assessment, TCE is characterized as "Carcinogenic to Humans " by all routes
20 of exposure. This conclusion is based on convincing evidence of a causal association between
21 TCE exposure in humans and kidney cancer. The human evidence of carcinogenicity from
22 epidemiologic studies of TCE exposure is compelling for lymphoma, but less convincing than
23 for kidney cancer, and more limited for liver and biliary tract cancer. Further support for the
24 characterization of TCE as "Carcinogenic to Humans " by all routes of exposure is derived from
25 positive results in multiple rodent cancer bioassays in rats and mice of both sexes, similar
26 toxicokinetics between rodents and humans, mechanistic data supporting a mutagenic MOA for
27 kidney tumors, and the lack of mechanistic data supporting the conclusion that any of the
28 MOA(s) for TCE-induced rodent tumors are irrelevant to humans.
29 As TCE toxicity and carcinogenicity are generally associated with TCE metabolism,
30 susceptibility to TCE health effects may be modulated by factors affecting toxicokinetics,
31 including lifestage, gender, genetic polymorphisms, race/ethnicity, preexisting health status,
32 lifestyle, and nutrition status. In addition, while some of these factors are known risk factors for
33 effects associated with TCE exposure, it is not known how TCE interacts with known risk factors
34 for human diseases.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-41 DRAFT—DO NOT CITE OR QUOTE
-------
1 For noncancer effects, the most sensitive types of effects, based either on human
2 equivalent concentrations/doses or on candidate RfCs/RfDs, appear to be developmental, kidney,
3 and immunological (adult and developmental) effects. The neurological and reproductive effects
4 appear to be about an order of magnitude less sensitive, with liver effects another two orders of
5 magnitude less sensitive. The preferred RfC estimate of 0.001 ppm (1 ppb or 5 ug/m3) is based
6 on route-to-route extrapolated results from oral studies for the critical effects of heart
7 malformations (rats), immunotoxicity (mice), and toxic nephropathy (rats, mice), and an
8 inhalation study for the critical effect of increased kidney weight (rats). Similarly, the preferred
9 RfD estimate for noncancer effects of 0.0004 mg/kg/d is based on the critical effects of heart
10 malformations (rats), adult immunological effects (mice), developmental immunotoxicity (mice),
11 and toxic nephropathy (rats). There is high confidence in these preferred noncancer reference
12 values, as they are supported by moderate- to high-confidence estimates for multiple effects from
13 multiple studies.
14 For cancer, the preferred estimate of the inhalation unit risk is 2 x 10~2 per ppm
15 [4 x 10~6 per jig/m3], based on human kidney cancer risks reported by Charbotel et al. (2006)
16 and adjusted, using human epidemiologic data, for potential risk for tumors at multiple sites.
17 The preferred estimate of the oral unit risk for cancer is 5 x 10~2 per mg/kg/d, resulting from
18 PBPK model-based route-to-route extrapolation of the inhalation unit risk estimate based on the
19 human kidney cancer risks reported in Charbotel et al. (2006) and adjusted, using human
20 epidemiologic data, for potential risk for tumors at multiple sites. There is high confidence in
21 these unit risks for cancer, as they are based on good quality human data, as well as being similar
22 to unit risk estimates based on multiple rodent bioassays. Because there is both sufficient weight
23 of evidence to conclude that TCE operates through a mutagenic MOA for kidney tumors and a
24 lack of TCE-specific quantitative data on early-life susceptibility, the default ADAFs can be
25 applied for the kidney cancer component of the unit risks for cancer; however, the application of
26 ADAFs is likely to have a minimal impact on the total cancer risk except when exposures are
27 primarily during early life.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 6-42 DRAFT—DO NOT CITE OR QUOTE
-------
REFERENCES
Abbas, R; Fisher, JW. (1997) A physiologically based pharmacokinetic model for trichloroethylene and its
metabolites, chloral hydrate, trichloroacetate, dichloroacetate, trichloroethanol, and trichloroethanol glucuronide in
B6C3F1 mice. Toxicol Appl Pharmacol 147:15-30.
Abbas, R; Seckel, CS; MacMahon, KL; et al. (1997) Determination of kinetic rate constants for chloral hydrate,
trichloroethanol, trichloroacetic acid, and dichloroacetic acid~a physiologically based modeling approach.
Toxicologist 36:32-33.
Abbas, RR; Seckel, CS; Kidney, JK; et al.(1996) Pharmacokinetic analysis of chloral hydrate and its metabolism in
B6C3F1 mice. Drug Metab Dispos 24:1340-1346.
Abraham, DG; Patel, P; Cooper, AJL. (1995) Isolation from rat kidney of a cytosolic high molecular weight
cysteine-S'-conjugate (3-lyase with activity toward leukotriene E4. J Biol Chem 270(1): 180-188.
Abraham, DG; Thomas, RJ; Cooper, AJL. (1995b) Glutamine transaminase K is not a major cysteine S-conjugate (3-
lyase of rat kidney mitochondria: Evidence that a high-molecular weight enzyme fulfills this role. J Pharmacol Exp
Ther 48:855-860.
ACS (American Cancer Society). (2008) What are the risk factors for cervical cancer? Available from:
http://www.cancer.org/docroot/CRI/content/CRI 24 2X What are the risk factors for cervical cancer S.asp.
Adachi, H; Kubota, I; Okamura, N; et al. (1989) Purification and characterization of human microsomal dipeptidase.
JBiocheml05(6):957-961.
Adami, HO; Chow, WH; Nyrene, O; et al. (1996) Excess risk of primary liver cancer in patients with diabetes
mellitus. J Natl Cancer Inst 88:1472-1477.
Adams, C; Keil, D; Meyers, K; et al. (2003) Lifetime exposure to trichloroethylene (TCE) modulates immune
function. Toxicologist 72(S-1):375.
Adgate, JL; Church, TR; Ryan, AD; et al. (2004b) Outdoor, indoor, and personal exposure to VOCs in children.
Environ Health Perspect 112(14): 1386-1392.
Adgate, JL; Eberly, LE; Stroebel, C; et al. (2004a) Personal, indoor, and outdoor VOC exposures in a probability
sample of children. J Expo Anal Environ Epidemiol. 14(Suppl 1):S4-S13.
ADHS (Arizona Department of Health Services). (1990) The incidence of childhood leukemia and testicular cancer
in Pima County: 1970-1986. Prepared for the Arizona Department of Environmental Quality under the Water
Quality Assurance Revolving Fund and by the Arizona Department of Health Services, Division of Disease
Prevention, Office of Risk Assessment and Investigation, Office of Chronic Disease Epidemiology. September 17,
1990.
ADHS (Arizona Department of Health Services). .(1995) Update of the incidence of childhood cancers and testicular
cancer in Southwest Tucson, 1987-1991. Prepared by the Arizona Department of Health Services, Office of Risk
Assessment and Investigations, Disease Prevention Services Arizona Department of Health Services. June, 6, 1995.
ADHS (Arizona Department of Health Services). (1988) Report on mortality in Maricopa County 1966-1986.
ADHS (Arizona Department of Health Services). (1990a) Incidence study of childhood cancer in Maricopa County:
1965-1986.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-l DRAFT—DO NOT CITE OR QUOTE
-------
ADHS (Arizona Department of Health Services). (1990b) Cancer incidence and mortality in an East Phoenix area
overlaying groundwater contaminated with volatile organic compounds.
ADHS (Arizona Department of Health Services). (1990c) The incidence of childhood leukemia and testicular cancer
in Pima County: 1970-1986.
ADHS (Arizona Department of Health Services). (1997) Case-referent study of childhood leukemia in Maricopa
County, Arizona 1965-1990.
Aickin, M; Chapin, CA; Flood, TJ; et al. (1992) Assessment of the spatial occurrence of childhood leukaemia
mortality using standardized rate ratios with a simple linear Poisson model. Int J Epidemiol 21:649-655.
Aickin, M. (2004) Bayes without priors. J Clin Epidemiol 57:4-13.
Albanese, RA; Banks, HT; Evans, MV; et al. (2002) Physiologically based pharmacokinetic models for the transport
of trichloroethylene in adipose tissue. Bull Math Biol 64:97-131.
Albee, RR; Nitschke, KD; Mattsson, JL; et al. (1997) Dichloroacetylene: Effects on the rat trigeminal nerve
somatosensory evoked potential. Neurotoxicol Teratol 19(l):27-37.
Albee, RR; Spencer, PJ; Johnson, KA; et al. (2006) Lack of trigeminal nerve toxicity in rats exposed to
trichloroethylene vapor for 13 weeks. Int J Toxicol 25(6):531-540.
Alberati-Giani, D; Malherbe, P; Kohler, C; et al. (1995) Cloning and characterization of a soluble kynurenine
aminotransferase from rat brain: Identity with kidney cysteine conjugate (3-lyase. J Neurochem 64:1448-1455.
Albert, DA; Albert, AN; Vernace, M; et al. (2005) Analysis of a cluster of cases of Wegener granulomatosis. J Clin
Rheumatol 11:188-193.
Alder, ID. (1993) Synopsis of the in vivo results obtained with the 10 known or suspected aneugens tested in the
CEC collaborative study. Mutat Res 287(1): 131-137.
Alexander, DD; Kelsh, MA; Mink, PJ; et al. (2007) A meta-analysis of occupational trichloroethylene exposure and
liver cancer. Int Arch Occup Environ Health 81 (2): 127-43.
Alexander, DD; Mink, PJ; Adami, HO; et al. (2007) The non-Hodgkin lymphomas: a review of the epidemiologic
literature. Int J Cancer 120 Suppl 12:1-39.
Alexander, DD; Mink, PJ; Mandel, JH; et al. (2006) A meta-analysis of occupational trichloroethylene exposure and
multiple myeloma or leukaemia. Occup Med (Lond) 56:485-493.
Alimov, A; Kost-Alimova, M; Liu, J; et al. (2000) Combined LOH/CGH analysis proves the existence of interstitial
3p deletions in renal cell carcinoma. Oncogene 19:1392-1399.
Allen, BC; Fisher, JW. (1993) Pharmacokinetic modeling of trichloroethylene and trichloroacetic acid in humans.
Risk Anal 13(l):71-86.
Allen, BC; Kavlock, RJ; Kimmel, CA; et al. (1994) Dose-response assessment for developmental toxicity. Fundam
Appl Toxicol 23:487-495.
Allen, DG; Pearse, G; Haseman, JK; et al. (2004) Prediction of rodent carcinogenesis: an evaluation of prechronic
liver lesions as forecasters of liver tumors inNTP carcinogenicity studies. Toxicol Pathol 32:393-401.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-2 DRAFT—DO NOT CITE OR QUOTE
-------
Allen, JW; Collins, BW; Evansky, PA. (1994) Spermatid micronucleus analyses of trichloroethylene and chloral
hydrate effects in mice. Mutat Res 323(l-2):81-88.
Altuntas, TG; Park, SB; Kharasch, ED. (2004) Sulfoxidation of cysteine and mercapturic acid conjugates of the
sevoflurane degradation product fluoromethyl-2,2-difluoro-l-(trifluoromethyl)vinyl ether (compound A). Chem Res
Toxicol 17(3):435-445.
Amacher, DE; Zelljadt, I. (1983) The morphological transformation of Syrian hamster embryo cells by chemicals
reportedly nonmutagenic to Salmonella typhimurium. Carcinogenesis 4:291-295.
Andelman, JB. (1985) Human exposure to volatile halogenated organic chemicals in indoor and outdoor air.
Environ Health Perspect. 62:313-318.
Anders, MW; Jakobson, I. (1985) Biotransformation of halogenated solvents. Scand J Work Env Health. ll(Suppl.
l):23-32.
Andersen, ME. (1987) Tissue dosimetxy in risk assessment, or what's the problem here anyway? In: Drinking water
and health, pharmacokinetics in risk assessment. Vol. 8. Washington DC: National Academy Press.
Andersen, ME; Gargas, ML; Clewell, HJ, 3rd; et al. (1987) Quantitative evaluation of the metabolic interactions
between trichloroethylene and 1,1-dichloroethylene in vivo using gas uptake methods. Toxicol Appl Pharmacol
89:149-157.
Andersson, L; Heraldsson, B; Johansson, C; et al. (2008) Methodological issues on the use of urinary alpha-1-
microglobulin in epidemiological studies. Nephrol Dial Transplant 23:1252-1256.
Anna, CH; Maronpot, RR; Pereira, MA; et al. (1994) ras proto-oncogene activation in dichloroacetic-,
trichloroethylene-and tetrachloroethylene-induced liver tumors in B6C3F1 mice. Carcinogenesis 15:2255-2261.
Anttila, A; Pukkala, E; Sallmen, M; et al. (1995) Cancer incidence among Finnish workers exposed to halogenated
hydrocarbons. J Occup Environ Med 37:797-806.
Aranyi, C; O'Shea, WJ; Graham, JA; et al. (1986) The effects of inhalation of organic chemical air contaminants on
murine lung host defenses. Fundam Appl Toxicol 6:713-720.
Arif, AA; Shah, SM. (2007) Association between personal exposure to volatile organic compounds and asthma
among US adult population. Int Arch Occup Environ Health 80:711-719.
Arito, H; Takahashi, M; Ishikawa, T. (1994) Effect of subchronic inhalation exposure to low-level trichloroethylene
on heart rate and wakefulness-sleep in freely moving rats. Jpn J Ind Health (Sangyo Igaku) 36(1): 1-8.
Aryal, BK; Khuder, SA; Schaub, EA. (2001) Meta-analysis of systemic sclerosis and exposure to solvents. Am J
Ind Med 40:271-274.
Asal, NR; Geyer, JR; Risser, DR; et al. (1988) Risk factors in renal cell carcinoma. II. Medical history, occupation,
multivariate analysis and conclusions. Cancer Detect Prev 13:263-279.
Asal, NR; Risser, DR; Kadamani, S; et al. (1988a) Risk factors in renal cell carcinoma: I. Methodology,
demographics, tobacco, beverage use, and obesity. Cancer Detect Prev 11(3-6):359-377.
Ashley, DL; Bonin, MA; Cardinali, FL; et al. (1994) Blood concentrations of volatile organic compounds in a
nonoccupationally exposed US population and in groups with suspected exposure. Clin Chem 40(7): 1401-1404.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-3 DRAFT—DO NOT CITE OR QUOTE
-------
Astrand, I; Ovrum, P. (1976) Exposure to trichloroethylene I. Uptake and distribution in man. Scand J Work
Environ Health 2(4): 199-211.
Astrand, I. (1975) Uptake of solvents in the blood and tissues of man. A review. Scand J Work Environ
Health.l(4):199-218.
ATSDR (Agency for Toxic Substances and Disease Registry). (1990) Toxicological profile for 1,1-dichloroethane.
U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, ATSDR, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (1995) Toxicological profile for 1,1,1-trichloroethane.
Update. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, ATSDR,
Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (1996a) Toxicological profile for 1,2-dichloroethene.
Update. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, ATSDR,
Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (1996b) Biennial report to congress (1991 and 1992)
U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, ATSDR, Atlanta, GA.
Available online at www.dhhs.gov.
ATSDR (Agency for Toxic Substances and Disease Registry). (1997) Toxicological Profile for Trichloroethylene.
U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, ATSDR, Atlanta, GA.
Available online at http://www.atsdr.cdc.gov/toxprofiles/tpl9.html.
ATSDR (Agency for Toxic Substances and Disease Registry). (1998) Volatile organic compounds in drinking water
and adverse pregnancy outcomes. U.S. Department of Health and Human Services, Centers for Disease Control and
Prevention, ATSDR, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2001) Final Report: evaluation of priority health
conditions in a community with historical contamination by trichloroethylene. U.S. Department of Health and
Human Services, Centers for Disease Control and Prevention, ATSDR. Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2003) Health Consultation. View-Master Factory
Supply Well (a.k.a. Mattel Portland Operations) Beaverton, Washington County, Oregon. Prepared by Oregon
Department of Human Services under Cooperative Agreement with the Agency for Toxic Substances and Disease
Registry. October 20, 2003.
ATSDR (Agency for Toxic Substances and Disease Registry). (2003) Impact of trichloroethylene exposure on oral
motor, speech and hearing in children. U.S. Department of Health and Human Services, July 2003, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2003b) Progress report: survey of specific childhood
cancers and birth defects among children whose mothers were pregnant while living at U.S. Marine Corps Base
Camp Lejeune, North Carolina, 1968-1985. U.S. Department of Health and Human Services, July 2003, Atlanta,
GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2003b) Survey of specific childhood cancers and
birth defects among children whose mothers were pregnant while living at US Marine Corps Base Camp Lejeune,
North Carolina, 1968-1985. U.S. Department of Health and Human Services, July 2003, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2004) Feasibility Investigation of Worker Exposure
to Trichloroethylene at the View-Master Factory in Beaverton, Oregon. Prepared by the Oregon Department of
Human Services. Report No. PB2005-102173.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-4 DRAFT—DO NOT CITE OR QUOTE
-------
ATSDR (Agency for Toxic Substances and Disease Registry). (2006) Cancer and birth outcome analysis, Endicott
Area, Town of Union, Broome County, New York. U.S. Department of Health and Human Services, May 2006,
Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2006) Health Consultation. Cancer and Birth
Outcome Analysis, Endicott Area, Town of Union, Broome County, New York. Health Statistics Review. Atlanta,
GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and
Disease Registry. May 26, 2006.
ATSDR (Agency for Toxic Substances and Disease Registry). (2007) Analyses of groundwater flow, contaminant
fate and transport and distribution of drinking water at Tarawa Terrace and vicinity, U.S. Marine Corps Base Camp
Lejeune, North Carolina: historical reconstruction and present-day conditions. U.S. Department of Health and
Human Services, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2008) Health statistics review follow-up: cancer and
birth outcome analysis. Endicott Area Investigation, Endicott Area, Town of Union, Broome County, New York.
U.S. Department of Health and Human Services, May 2008, Atlanta, GA.
ATSDR (Agency for Toxic Substances and Disease Registry). (2009) Study on birth defects and childhood cancers.
U.S. Department of Health and Human Services, Atlanta, GA. Available online at
http://www.atsdr.cdc.gov/sites/lejeune/update.html (accessed September 17, 2009).
Auperin,A; Asal, NR; Geyer, JR; et al. (1988) Risk factors in renal cell carcinoma. II. Medical history, occupation,
multivariate analysis, and conclusions. Cancer Detect Preven 13:263-279.
Auperin, A; Benhamou, S; Ory-Paoletti, C; et al. (1994) Occupational risk factors for renal cell carcinoma: a case-
control study. Occup Environ Med 51:426-428.
Austin, EW; Parrish, JM; Kinder, DH; et al. (1996) Lipid peroxidation and formation of 8-hydroxydeoxygaunosine
from acute doses of halogenated acetic acids. Fund Appl Toxicol 31:77-82.
Austin, H; Delzell, E; Grufferman, S; et al. (1987) Case-control study of hepatocellular carcinoma, occupation, and
chemical exposures. J Occup Med 29:665-669.
Axelson, O; Selden, A; Andersson, K; et al. (1994) Updated and expanded Swedish cohort study on
trichloroethylene and cancer risk. J Occup Environ 36:556-562.
Baden,JM; Kelley, M; Mazze, RI; et al. (1979) Mutagenicity of inhalation anaesthetics: Trichloroethylene, divinyl
ether, nitrous oxide, and cyclopropane. Br J Anaesth 51:417-421.
Bai, CL; Canfield, PJ; Stacey, NH. (1992a) Effects of hexachloro-1,3- butadiene and 1,1,2,2-tetrachloroethylene on
individual serum bile acids. Toxicol Ind Health 8:191-203.
Bai, CL; Canfield, PJ; Stacey, NH. (1992b) Individual serum bile acids as early indicators of carbon tetrachloride-
and chloroform-induced liver injury. Toxicology 75:221-234.
Bailly, M; Bain, C; Favrot, M-C; et al. (1995) Somatic mutations of von Hippel-Lindau (VHL) tumor-suppressor
gene in European kidney cancer. Int J Cancer 63:660-664.
Bakke, B; Stewart, PA; Waters, MA. (2007) Uses of and exposure to trichloroethylene in U.S. Industry: a
systematic literature review. J Occup Environ Hgy 4:375-390.
Ballestar, E; Esteller, M. (2002) The impact of chromatin in human cancer: linking DNA methylation to gene
silencing. Carcinogenesis 23:1103-1109
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-5 DRAFT—DO NOT CITE OR QUOTE
-------
Balluz, L; Philen, R; Ortega, L; et al. (2001) Investigation of systemic lupus erythematosus in Nogales, Arizona.
Am JEpidemiol 154:1029-1036.
Band, PR; Le, ND; Fang, R; et al. (2000) Identification of occupational cancer risks in British Columbia. J Occup
Environ Med 42:284-310.
Banerjee, S; Van Duuren, BL. (1978) Covalent binding of the carcinogen trichloroethylene to hepatic microsomal
proteins and to exogenous DNA in vitro. Cancer Res 38:776-780.
Bang, KM; Kim, JH. (2001) Prevalence of cigarette smoking by occupation and industry in the United States. Am J
Ind Med 40:233-239.
Banki, K; Elfarra, AA; Lash, LH; et al. (1986) Metabolism of S-(2-cWoro-l,l,2-trifluoroethyl)-l-cysteine to
hydrogen sulfide and the role of hydrogen suffide inS-(2-cmoro-l,l,2-trifluoroethyl)-l-cysteine-induced
mitochondria! toxicity. BiochemBiophRes Co 138:707-713.
Banks, PM. (1992) Changes in diagnosis of non-Hodgkin's lymphomas over time. Cancer Res 52:5453s-5455s.
Bannasch, P; Nehrbass, D; Kopp-Schneider, A. (2001) Significance of hepatic preneoplasia for cancer
chemoprevention. IARC SciPubl 154:223-240.
Bannasch, P. (2001) Hormonal and hormone-like effects eliciting hepatocarcinogenesis. Folia Histochem Cytobio
139(Suppl 2):28-29.
Bannerjee, S; van Duuren, BL. (1978) Covalent binding of the carcinogen trichloroethylene to hepatic microsomal
proteins and to exogenous DNA in vitro. Cancer Res 38:776-780.
Barak, Y; Sadovsky, Y; Shalom-Barak, T. (2008) PPAR signaling in placental development and function. PPAR
Res. 2008; 2008: 142082. . Doi: 10.1155/2008/142082. Bardodej, Z; Vyskocil, J. (1956) The problem of
trichloroethylene in occupational medicine; trichloroethylene metabolism and its effect on the nervous system
evaluated as a means of hygienic control. AMA Arch Ind Health 13(6):581-592.
Barret, L; Faure, J; Guilland, B; et al. (1984) Trichloroethylene occupational exposure: elements for better
prevention. Int Arch Occup Environ Health 53(4):283-289.
Barret, L; Arsac, P; Vincent, M; et al. (1982) Evoked trigeminal nerve potential in chronic trichloroethylene
intoxication. J Toxicol Clin Toxicol 19(4) :419-423.
Barret, L; Garrel, S; Danel, V; et al. (1987) Chronic trichloroethylene intoxication: a new approach by trigeminal-
evoked potentials? Arch Environ Health 42(5):297-302.
Barret, L; Torch, S; Leray, CL; et al. (1992) Morphometric and Biochemical Studies in Trigeminal Nerves of Rat
after Trichlorethylene or Dichloracetylene Oral Administration. Neurotoxicology 13:601-614.
Barret, L; Torch, S; Usson, Y; et al. (1991) A morphometric evaluation of the effects of trichloroethylene and
dichloroacetylene on the rat mental nerve. Preliminary results. NeurosciLett 131(2): 141-144.
Barter, ZE.; Bayliss, MK; Beaune, PH; et al. (2007) Scaling factors for the extrapolation of in vivo metabolic drug
clearance from in vitro data: reaching a consensus on values of human microsomal protein and hepatocellularity per
gram of liver. Curr Drug Metab 8:33-45.
Barton, HA; Bull, R; Schultz, I; et al. (1999) Dichloroacetate (DCA) dosimetry: interpreting DCA-induced liver
cancer dose response and the potential for DCA to contribute to trichloroethylene-induced liver cancer. Toxicol Lett
106:9-21.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-6 DRAFT—DO NOT CITE OR QUOTE
-------
Barton, HA; Chiu, WA; Woodrow Setzer, R; et al. (2007) Characterizing uncertainty and variability in
physiologically based pharmacokinetic models: state of the science and needs for research and implementation.
Toxicol Sci 99:395-402.
Barton, HA; Clewell, 3rd, HJ. (2000) Evaluating noncancer effects of trichloroethylene: dosimetry, mode of action,
and risk assessment. Environ Health Perspect Suppl 10:323-334.
Barton, HA; Flemming, CD; Lipscomb, JC. (1986) Evaluating human variability in chemical risk assessment:
hazard identification and dose-response assessment for noncancer oral toxicity of trichloroethylene. Toxicology
lll(l-3):271-87.
Barton, H; Das, S. (1996) Alternatives for a risk assessment on chronic noncancer effects from oral exposure to
trichloroethylene. Regul Toxicol Pharmacol 24:269-285.
Barton, HA; Creech, JR; Godin, CS; et al. (1995) Chloroethylene mixtures: pharmacokinetic modeling and in vitro
metabolism of vinyl chloride, trichloroethylene, and trans-1,2-dichloroethylene in rat. Toxicol Appl Pharm
130(2):237-247.
Bartonicek V. (1962) Metabolism and excretion of trichloroethylene after inhalation by human subjects. Brit J Ind
Med 19:134-141.
Bartonicek, V; Teisinger, J. (1962) Effect of tetraethyl thiuram disulphide (disulfiram) on metabolism of
trichloroethylene in man. Brit J Industr Med 19:216-221.
Bartosiewicz, M; Perm, S; Buckpitt, A. (2001) Applications of gene arrays in environmental toxicology: fingerprints
of gene regulation associated with cadmium chloride, benzo(a)pyrene, and trichloroethylene. Environ Health
Perspect 109:71-74.
Bartsch, H; Malaveille, C; Barbin, A; et al. (1979) Mutagenic and alkylating metabolites of halo-ethylenes,
chlorobutadienes and dichlorobutenes produced by rodent or human liver tissues. Evidence for oxirane formation by
P450-linked microsomal mono-oxygenases. Arch Toxicol. 41(4):249-77.
Beaudreuil, S; Lasfargues, G; Laueriere, L; et al. (2005) Occupational exposure in ANCA-positive patients: a case-
control study. Kidney Int 67:1961-1966.
Becker, N; Deeg, E; Nieters, A. (2004) Population-based study of lymphoma in Germany: rationale, study design
and first results. Leuk Res 28:713-724.
Beckstead, MJ; Weiner, JL; Eger, 2nd, El; et al. (2000) Glycine and gamma-aminobutyric acid(A) receptor function
is enhanced by inhaled drugs of abuse. Mol Pharmacol. 57:1199-205.
BEIR ([Committee on the] Biological Effects of Ionizing Radiation). (1988) Health risks of radon and other
internally deposited alpha emitters. BEIR IV. Washington, DC: National Academy Press.
Beland, FA. (1999) NTP technical report on the toxicity and metabolism studies of chloral hydrate (CAS No. 302-
17-0). Administered by gavage to F344/N rats and B6C3F1 mice. Toxic Rep Ser. 59:1-66, A1-E7.
Bellies, RP; Brucik, DJ; Mecler, FJ; et al. (1980) Teratogenic-mutagenic risk of workplace contaminants:
Trichloroethylene, perchloroethylene, and carbon disulfide. U.S. Department of Health, Education and Welfare.
Contract No. 210-77-0047.
Benane, SG; Blackman, CF; House, DE. (1996) Effect of perchloroethylene and its metabolites on intercellular
communication in clone 9 rat liver cells. J Toxicol Environ Health 48(5):427-37.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-7 DRAFT—DO NOT CITE OR QUOTE
-------
Benedetti, MS; Whomsley, R; Canning, M. (2007) Drag metabolism in the paediatric population and in the elderly.
DragDiscov Today 12(15-16):599-610.
Benichou J; Chow, WH; McLaughlin, JK; et al. (1998) Population attributable risk of renal cell cancer in Minnesota.
Am JEpidemiol 148(5):424-430.
Benson, R. (2004) Correspondence from Robert Benson (U.S. Environmental Protection Agency, Region 8, Denver,
CO 80202) to Dr. P.O. Johnson (University of Arizona, UAC, Tucson, AZ 85721), December 27, 2004.
Beppu, K. (1968) Transmission of the anesthetic agents through the placenta in painless delivery and their effects on
newborn infants. Keio JMed 17(2):81-107.
Berger, J; Daxenbichler, G. (2002) DNA methylation of nuclear receptor genes—possible role in malignancy. J
Steroid Biochem Mol Bio 180:1-11.
Berger, T; Horner, CM. (2003) In vivo exposure of female rats to toxicants may affect oocyte quality. Reprod
Toxicol 17:273-281.
Bergeron, A; Jorquera, R; Tanguay, RM. (2003) [Hereditary tyrosinemia: an endoplasmic reticulum stress disorder?].
Med Sci (Paris) 19:976-980.
Bergman, K. (1983) Interaction of trichloroethylene with DNA in vitro and with RNA and DNA of various mouse
tissues in vivo. Arch Toxicol 54:181-193.
Berman, E; Schlicht, M; Moser, VC; et al. (1995) A Multidisciplinary Approach to Toxicological Screening: I.
Systemic Toxicity. J Toxicol Environ Health 45:127-143.
Bernad, PG; Newell, S; Spyker, DA. (1987) Neurotoxicity and behavior abnormalities in a cohort chronically
exposed to trichloroethylene. Abstract. Vet Hum Toxicol 29(6):475.
Bernauer, U; Birner, G; Dekant, W; et al. (1996) Biotransformation of trichloroethene: dose-dependent excretion of
2,2,2-trichloro-metabolites and mercapturic acids in rats and humans after inhalation. Arch Toxicol 70(6):338-346.
Bernillon, P; Bois, FY. (2000) Statistical issues in toxicokinetic modeling: a Bayesian perspective. Environ Health
Perspect 108(Suppl. 5):883-893.
Beroud, C; Joly, D; Gallou, C; et al. (1998) Software and database for the analysis of mutation in the VHL gene.
Nucleic Acids Res 26:256-258.
Bhunya, SP; Jena, GB. (1996) The evaluation of clastogenic potential of trichloroacetic acid (TCA) in chick in vivo
test system. MutatRes. 367(4):254-259.
Bhunya, SP; Behera, BC. (1987) Relative genotoxicity of trichloroacetic acid (TCA) as revealed by different
cytogenetic assays: bone marrow chromosome aberration, micronucleus and sperm-head abnormality in the mouse.
MutatRes 188:215-221.
Bignami, M; Conti, G; Conti, L; et al. (1980) Mutagenicity of halogenated aliphatic hydrocarbons in Salmonella
typhimurium, Streptomyces coelicolor and Aspergillus nidulans. Chem Biol Interact. 30(l):9-23.
Birner, G; Vamvakas, S; Dekant, W; et al. (1993) Nephrotoxic and genotoxic N-acetyl-S-dichlorovinyl-L-cysteine is
a urinary metabolite after occupational 1,1,2-trichloroethene exposure in humans: implications for the risk of
trichloroethene exposure. Environ Health Perspect. 99:281-284.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-8 DRAFT—DO NOT CITE OR QUOTE
-------
Birner, G; Bernauer, U; Werner, M; et al. (1997) Biotransformation, excretion and nephrotoxicity of haloalkene-
derived cysteine S-conjugates. Arch Toxicol 72(1): 1-8.
Birner, G; Rutkowska, A; Dekant, W. (1996) N-acetyl-S-(l,2,2-trichlorovinyl)-L-cysteine and 2,2,2-trichloroethanol:
two novel metabolites of tetrachloroethene in humans after occupational exposure. Drug Metab Dispos 24:41-48.
Birner, G; Richling, C; Henschler, D; et al. (1994) Metabolism of tetrachloroethene in rats: identification of N
epsilon-(dichloroacetyl)-L-lysine and N epsilon-(trichloroacetyl)-L-lysine as protein adducts. Chem Res Toxicol
7(6):724-732.
Birner, G; Werner, M; Rosner, E; et al. (1998) Biotransformation, excretion, and nephrotoxicity of the
hexachlorobutadiene metabolite (E)-N-acetyl-S-(l,2,3,4, 4-pentachlorobutadienyl)-L-cysteine sulfoxide. Chem Res
Toxicol ll(7):750-777.
Blackburn, AC; Coggan, M; Tzeng, HF; et al. (2001) GSTZld: a new allele of glutathione transferase zeta and
maleylacetoacetate isomerase. Pharmacogenetics 11:671-678.
Blackburn, AC; Tzeng, HF; Anders, MW; et al. (2000) Discovery of a functional polymorphism in human
glutathione transferase zeta by expressed sequence tag database analysis. Pharmacogenetics 10:49-57.
Blackshaw, AW. (1970) Histochemical localization of testicular enzyme. In Johnson AD; Gomes, WR;
Vandermark, NL; eds. The Testes. New York: Academic Press, pp 73-173.
Blain, L; Lachapelle, P; Molotchnikoff, S. (1994) Electroretinal responses are modified by chronic exposure to
trichloroethylene. Neurotoxicology 15(3):627-631.
Blair, A; Hartge, P; Stewart, PA; et al. (1998) Mortality and cancer incidence of aircraft maintenance workers
exposed to trichloroethylene and other organic solvents and chemicals: extended follow up. Occup Environ Med
55:161-171.
Blair, A; Linos, A; Stewart, PA; et al. (1993) Evaluation of risks for non-Hodgkin's lymphoma by occupation and
industry exposures from a case-control study. Am J Ind Med 23:301-312.
Blair, A; Stewart, P; Lubin, JH; et al. (2007) Methodological issues regarding confounding and exposure
misclassification in epidemiological studies of occupational exposures. Am J Ind Med 50:199-207.
Blake, JE; Compton, KV; Schmidt, W;et al. (1988) Accuracy of death certificates in the diagnosis of alcoholic liver
cirrhosis. Alcohol Clin Exp Res 12:168-172.
Blake, MJ; Castro, L; Leeder, JS; et al. (2005) Ontogeny of drug metabolizing enzymes in the neonate. Semin Fetal
Neonatal Med 10(2):123-138.
Blechzcz, BRA; Gores, GJ. (2008) Cholangiocarcinoma. Clin Liver Disease 12:131-150.
Bloemen, L; Tomenson, J. (1995) Increased incidence of renal cell tumours in a cohort of cardboard workers
exposed to trichloroethylene [Letter]. Arch Toxicol 70:129-130.
Bloemen, LJ; Monster, AC; Kezic, S; et al. (2001) Study on the cytochrome P-450- and glutathione-dependent
biotransformation of trichloroethylene in humans. Int Arch Occup Environ Health 74:102-108.
Blossom, SJ; Doss, JC; Gilbert, KM. (2007) Chronic exposure to a trichloroethylene metabolite in autoimmune-
prone MRL+/+ mice promotes immune modulation and alopecia. Toxicol Sci 95(2):401-411.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-9 DRAFT—DO NOT CITE OR QUOTE
-------
Blossom, SJ; Doss, JC; Mannings, LJ; et al. (2008) Developmental exposure to trichloroethylene promotes CD4(+)
T cell differentiation and hyperactivity in association with oxidative stress and neurobehavioral deficits in MRL+/+
mice. Toxicol Appl Pharmacol 231(3):344-353.
Blossom, SJ; Doss, JC. (2007) Trichloroethylene alters central and peripheral immune function in autoimmune-
prone MRL+/+ mice following continuous developmental and early life exposure. J Immunotoxicol 4:129-141.
Blossom, SJ; Gilbert, KM. (2006) Exposure to a metabolite of the environmental toxicant, trichloroethylene,
attenuates CD4+ T cell activation-induced cell death by metalloproteinase-dependent FasL shedding. Toxicol Sci
92:103-114.
Blossom, SJ; Pumford, NR; Gilbert, KM. (2004) Activation and attenuation of apoptosis of CD4+ T cells following
in vivo exposure to two common environmental toxicants, trichloroacetaldehyde hydrate and trichloroacetic acid. J
Autoimmun 23:211-220.
Board, PG; Chelvanayagam, G; Jermiin, LS; et al. (2001) Identification of novel glutathione transferases and
polymorphic variants by expressed sequence tag database analysis. Drug Metab Dispos 29:544-547.
Boffetta, P; de Vocht, F. (2007) Occupation and the risk of non-Hodgkin lymphoma. Cancer Epidemiol Biomarkers
Prev 16:369-372.
Bogen, KT. (1990) Uncertainty in environmental health risk assessment. London: Taylor & Francis.
Bogen, KT; Colston, BW, Jr; MaChicao, LK. (1992) Dermal absorption of dilute aqueous chloroform,
trichloroethylene, and tetrachloroethylene in hairless guinea pigs. Fund Appl Toxicol 18(l):30-39.
Boice, JD, Jr.; Marano, DE; Fryzek, JP; et al. (1999) Mortality among aircraft manufacturing workers. Occup
Environ Med 56:581-597.
Boice, JD; Marano, DE; Cohen, SS; et al. (2006) Mortality among Rocketdyne worker who tested rocket engines,
1948-1999. J Occup Environ Med 48:1070-1092. .
Boice, JD; Marano, DE; Fryzek, JP; et al. (1999) Mortality among aircraft manufacturing workers. Occup Environ
Med 56:581-597.
Bois, FY. (2000a) Statistical analysis of Fisher et al. PBPK model of trichloroethylene kinetics. Environ Health
Perspect 108(Suppl 2):275-282.
Bois, FY. (2000b) Statistical analysis of Clewell et al. PBPK model of trichloroethylene kinetics. Environ Health
Perspect 108(Suppl 2):307-316.
Bolt, H; Lammert, M; Selinski, S; et al. (2004) Urinary al-microglobulin excretion as biomarker of renal toxicity in
trichloroethylene-exposed persons. Int Arch Occup Environ Health 77:186-190.
Bolt, HM; Filser, JG. (1977) Irreversible binding of chlorinated ethylenes to macromolecules. Environ Health Persp
21:107-112.
Bonatti, S; Cavalieri, Z; Viaggi, S; et al. (1992)The analysis of 10 potential spindle poisons for their ability to induce
CREST-positive micronuclei in human diploid fibroblasts. Mutagenesis. 7(2): 111-4
Bong, M; Laskowska-Klita, T; Szymczyk, T. (1985) Effect of the benzene fraction of petroleum on protein content
in rat liver and kidney. Bull Environ Contam Toxicol 34(l):45-54.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-10 DRAFT—DO NOT CITE OR QUOTE
-------
Boublik, T; Fried, V; Hala, E. (1984) The vapour pressures of pure substances. Second Revised Edition.
Amsterdam: Elsevier.
Bove, FJ; Fulcomer, MC; Klotz, JB; et al. (1995) Public drinking water contamination and birth outcomes. Am J
Epidemiol 141(9):850-862.
Bove, FJ. (1996) Public drinking water contamination and birthweight, prematurity, fetal deaths, and birth defects.
Toxicol Ind Health 12(2):255-266.
Bovenzi, M; Barbone, F; Pisa, FE; et al. (2004) A case-control study of occupational exposures and systemic
sclerosis. Int Arch Occup Environ Health 77:10-16.
Boyer, AS; Finch, WT; Runyan, RB. (2000) Trichloroethylene inhibits development of embryonic heart valve
precursors in vitro. Toxicol Sci 53:109-117.
Boyes, WK; Bercegeay, M; Ali, JS; et al. (2003) Dose-based duration adjustments for the effects of inhaled
trichloroethylene on rat visual function. Toxicol Sci 76(1): 121-130.
Boyes, WK; Bercegeay, M; Krantz, T; et al. (2005) Momentary brain concentration of trichloroethylene predicts the
effects on rat visual function. Toxicol Sci 87(1): 187-196.
Boyes, WK; Bushnell, PJ; Crofton, KM; et al. (2000) Neurotoxic and pharmacokinetic responses to
trichloroethylene as a function of exposure scenario. Environ Health Perspect 108(Suppl 2):317-322.
Braissant, O; Wahli, W. (1998) Differential expression of peroxisome proliferator-activated receptor-o, -(3, and -y
during rat embryonic development. Endocrinology 139(6):2748-2754.
Brauch, H; Weirich, G; Hornauer, MA; et al. (1999) Trichloroethylene exposure and specific somatic mutations in
patients with renal cell carcinoma. J Natl Cancer Inst 91:854:861.
Brauch, H; Weirich, G; Klein, B; et al. (2004) VHL mutations in renal cell cancer: does occupational exposure to
trichloroethylene make a difference? Toxicol Lett 151(1):301-310.
Briving, C; Jacobson, I; Hamberger, A; et al. (1986) Chronic effects of perchloroethylene and trichloroethylene on
the gerbil brain amino acids and glutathione. Neurotoxicology 7(1):101-108.
Brodsky, WY; Uryvaeva, IV. (1977) Cell polyploidy: its relation to tissue growth and function. Int Rev Cytol 50:
275-332.
Bronley-DeLancey, A; McMillan, DC; McMillan, JM; et al. (2006) Application of cryopreserved human
hepatocytes in trichloroethylene risk assessment: relative disposition of chloral hydrate to trichloroacetate and
trichloroethanol. Environ Health Persp 114(8): 1237-1242.
Bronzetti, G; Corsi, C; Nieri, R. (1980) Use of the D7 strain of S. cerevisiae in the determination of environmental
risk. 1. Genetic effects of trichloroethylene. Boll Soc Ital Biol Sper 56(12): 1315-21.
Bronzetti, G; Galli, A; Corsi, C; et al. (1984) Genetic and biochemical investigation on chloral hydrate in vitro and
in vivo. MutatRes. 141(1): 19-22.
Bross, G; DiFranceisco, D; Desmond, ME. (1983) The effects of low dosages of trichloroethylene on chick
development. Toxicology 28:283-294.
Brown, HS; Bishop, DR; Rowan, CA. (1984) The role of skin absorption as a route of exposure for volatile organic
compounds in drinking water. Am J Pub Health 74(5):479-484.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-l 1 DRAFT—DO NOT CITE OR QUOTE
-------
Brown, RP; Delp, MD; Lindstedt, SL; Rhomberg, LR; Bellies, RP. (1997) Physiological parameter values for
physiologically based pharmacokinetic models. Toxicol Ind Health 13:407-484.
Brownson, RC. (1988) A case-control study of renal cell carcinoma in relation to occupation, smoking, and alcohol
consumption. Arch Environ Health 43:238-241.
Bruning, T; Lammert, M; Kempkes, M; et al. (1997) Influence of polymorphisms of GSTM1 and GSTT1 for risk of
renal cell cancer in workers with long-term high occupational exposure to trichloroethene. Arch Toxicol
71(9):596-599.
Bruning, T; Mann, H; Melzer, H; et al. (1999a) Pathological excretion patterns of urinary proteins in renal cell
cancer patients exposed to trichloroethylene. Occup Med 29:299-305.
Bruning, T; Pesch, B; Wiesenhiitter, B; et al. (2003) Renal cell cancer risk and occupational exposure to
trichloroethylene: results of a consecutive case-control study in Arnsberg, Germany. Am J Ind Med 43:274-285.
Bruning, T; Sundberg, AGM; Birner, G; et al. (1999b) Glutathione transferase alpha as a marker for tubular damage
after trichloroethylene exposure. Arch Toxicol 73:246-254.
Bruning, T; Vamvakas, S; Makropoulos, V; etal. (1998) Acute intoxication with trichloroethene: clinical symptoms,
toxicokinetics, metabolism, and development of biochemical parameters for renal damage. Toxicol Sci
41(2):157-165.
Bruning, T; Weirich, G; Hornauer, MA; et al. (1997a) Renal cell carcinomas in trichloroethene TRI) exposed
persons are associated with somatic mutations in the von Hippel-Lindau (VHL) tumour suppressor gene. Arch
Toxicol 71:332-335.
Brzezinski, MR;, Boutelet-Bochan, H; Person, RE; et al. (1999) Catalytic activity and quantitation of cytochrome P-
450 2E1 in prenatal human brain. J Pharmacol Exp Ther289(3):1648-1653.
Buben, JA; O'Flaherty, EJ. (1985) Delineation of the role of metabolism in the hepatotoxicity of trichloroethylene
and perchloroethylene: a dose-effect study. Toxicol Appl Pharmacol 78:105-122.
Buckpitt, A; Chang, A-M; Weir, A; et al. (1995) Relationship of cytochrome P450 activity to Clara cell cytotoxicity.
Mol Pharmacol 47:74-81.
Bull, RJ; Sanchez, IM; Nelson, MA; et al. (1990) Liver tumor induction in B6C3F1 mice by dichloroacetate and
trichloroacetate. Toxicology 63:341-359.
Bull, RJ. (2004) Trichloroethylene and liver tumors in mice. In: Proceedings of new scientific research related to the
health effects of trichloroethylene, 26-27 February 2004, Washington DC. Washington, DC:U.S. Environmental
Protection Agency. Available online at http://cfpub.epa.gov/ncea/cfm/recordisply.cfm?deid=75934 [accessed 3
August 2006].
Bull, RJ. (2000) Mode of action of liver tumor induction by trichloroethylene and its metabolites, trichloroacetate
and dichloroacetate. Environ Health Persp 108(Suppl 2):241-259.
Bull, RJ; Orner, GA; Cheng, RS; et al. (2002) Contribution of dichloroacetate and trichloroacetate to liver tumor
induction in mice by trichloroethylene. Toxicol Appl Pharmacol 182:55-65.
Bunce, NJ; Schneider, UA. (1994) Chemical lifetimes of chlorinated aliphatic priority pollutants in the Canadian
troposphere. J Photochem Photobiol A Chem A: Chemistry 81:93-101.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-12 DRAFT—DO NOT CITE OR QUOTE
-------
Burg, JA; Gist, GL. (1997) The potential impact on women from environmental exposures. J Womens Health
6(2):159-161.
Burg, JR; Gist, GL; Allred, SL; et al. (1995) The National Exposure Registry - Morbidity analyses of noncancer
outcomes from the trichloroethylene subregistry baseline data. Int J Occup Med Toxicol 4(2):237-257.
Burg, JR; Gist, GL. (1999) Health effects of environmental contaminant exposure: an intrafile comparison of the
Trichloroethylene Subregistry. Arch Environ Health 54(4):231-241.
Burmaster, DE. (1982) The new pollution-groundwater contamination. Environment 24(2):6-36.
Butler, T.C. (1949) Metabolic transformations of trichloroethylene. Presented in part at the 39th Annual Meeting of
the American Society for Pharmacology and Experimental Therapeutics, April 22, 1949. Fed Proc 8:278.
Buxton, P.H.; Hay ward, M. (1967) Polyneuritis cranialis associated with industrial trichloroethylene poisoning. J
Neurol Neurosurg Psych 30:511.
Byers, VS; Levin, AS; Ozonoff, DM; et al. (1988) Association between clinical symptoms and lymphocyte
abnormalities in a population with chronic domestic exposure to industrial solvent-contaminated domestic water
supply and a high incidence of leukaemia. Cancer Immunol Immunother 27(1):77-81.
Cai, H; Guengerich, P. (2001) Reaction of trichloroethylene oxide with proteins andDNA: instability of adducts and
modulation of functions. Chem Res Toxicol 14:54-61.
Cai, H; Guengerich, P. (1999) Mechanism of aqueous decomposition of trichloroethylene oxide. J Am Chem Soc
121:11656-11663.
Cai, P; Konig, R; Boor, PJ; et al. (2008) Chronic exposure to trichloroethene causes early onset of SLE-like disease
in female MRL +/+ mice. Toxicol Appl Pharmacol 228:68-75.
Cai, P; Konig, R; Khan, MF; et al. (2007) Differential immune responses to albumin adducts of reactive
intermediates of trichloroethene inMRL+/+ mice. Toxicol Appl Pharmacol 220:278-283.
Cai, P; Konig, R; Khan, MF; et al. (2006) Autoimmune response in MRL+/+ mice following treatment with
dichloroacetyl chloride or dichloroacetic anhydride. Toxicol Appl Pharmacol 216:248-255.
Cai, H; Guengerich, FP. (2001) Reaction of trichloroethylene and trichloroethylene oxide with cytochrome P450
enzymes: Inactivation and sites of modification. Chem Res Toxicol 14:451-458.
Cakmak, A; Ekici, A; Ekici, M; et al. (2004) Respiratory findings in gun factory workers exposed to solvents.
Respirat Med 98:52-56.
Caldwell, JC; Keshava, N; Evans, MV. (2008b) Difficulty of mode of action determination for trichloroethylene: an
example of complex interactions of metabolites and other chemical exposures. Environ Mol Mutagen 49(2): 142-54.
Caldwell, JC; Keshava, N. (2006) Key issues in the modes of action and effects of trichloroethylene metabolites for
liver and kidney tumorigenesis. Environ Health Perspect 114:1457-1463.
Caldwell, PT; Thorne, PA; Johnson, PD; et al. (2008) Trichloroethylene disrupts cardiac gene expression and
homeostasis in rat myocytes. Toxicol Sci 104(1):135-143.
Callen, DF; Wolf, CR; Philpot, RM. (1980) Cytochrome P-450-mediated genetic activity and cytotoxicity of seven
halogenated aliphatic hydrocarbons in Succharomyces cerevisiae. Mutat Res 77:55-63.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-13 DRAFT—DO NOT CITE OR QUOTE
-------
Calvisi, DF; Ladu, S; Gorden, A; et al, SS. (2007) Mechanistic and prognostic significance of aberrant methylation
in the molecular pathogenesis of human hepatocellular carcinoma. J Clin Invest 117(9):2713-2711.
Cao, K; Hollenbach, J; Shi, X; et al. (2001) Analysis of the frequencies of HLA-A, B, and C alleles and haplotypes
in the five major ethnic groups of the United States reveals high levels of diversity in these loci and contrasting
distribution patterns in these populations. Hum Immunol 62(9): 1009-30.
Carney, EW; Thorsrud, BA; Dugard, PH; et al. (2006) Developmental toxicity studies in Crl:Cd (SD) rats following
inhalation exposure to trichloroethylene and perchloroethylene. Birth Defects Research (Part B) 77:405-412.
Carpenter, SP; Lasker, JM; Raucy, JL. (1996) Expression, induction, and catalytic activity of the ethanol-inducible
cytochrome P450 (CYP2E1) in human fetal liver and hepatocytes. Mol Pharmacol 49(2):260-268.
Carrieri, M; Magosso, D; Piccoli, P; et al. (2007) Acute nonfatal intoxication with trichloroethylene. Arch Toxicol
81:529-32.
Carter, JH; Carter, HW; DeAngelo, AB. (1995) Biochemical pathologic and morphometric alterations induced in
male B6C3F1 mouse liver by short-term exposure to dichloroacetic acid. Toxicol Lett 81:55-71.
Carter, JH; Carter, HW; Deddens, JA; et al. (2003) A 2-year dose-response study of lesion sequences during
hepatocellular carcinogenesis in the male B6C3F1 mouse given the drinking water chemical dichloroacetic acid.
Environ Health Perspect 11 l(l):53-64.
CDHS (California Department of Health Services). (1986) Organic chemical contamination of large public water
systems in California. California Department of Health Services, Sacramento, CA.
Chakrabarti, SK; Tuchweber, B. (1988) Studies of acute nephrotoxic potential of trichloroethylene in Fischer 344
rats. J Toxicol Environ Health 23:147-158.
Chalupa, B;, Synkova, J.; Sevcik, M. (1960) The assessment of electroencephalographic changes in memory
disturbances in acute intoxications with industrial poisons. Brit J Ind Med 17:238.
Chang, LW; Daniel, FB; DeAngelo, AB. (1992) Analysis of DNA strand breaks induced in rodent liver in vivo,
hepatocytes in primary culture, and a human cell line by chlorinated acetic acids and chlorinated acetaldehydes.
Environ Mol Mutagen 20L:277-288.
Chang, Y-M; Tai, C-F; Yang, S-C; et al. (2003) A cohort mortality study of workers exposed to chlorinated organic
solvents in Taiwan. AnnEpidemiol 13:652-660.
Chang,Y-M; Tai, C-F; Yang, S-C; et al. (2005) Cancer incidence among workers potentially exposed to chlorinated
solvents in an electronics factory. J Occup Health 47:171-180.
Chang, LW; Daniel, FB; DeAngelo, AB. (1992) Analysis of DNA strand breaks induced in rodent liver in vivo,
hepatocytes in primary culture, and a human cell line by chloroacetic acids and chloroacetaldehydes. Environ Mol
Mutagen 20:277-288.
Channel, SR; Latendresse, JR; Kidney, JK; et al. (1998) A subchronic exposure to trichloroethylene causes lipid
peroxidation and hepatocellular proliferation in male B6C3F1 mouse liver. Toxicol Sci 43:145-154.
Chapin, RE; Norton, RM; Popp, JA; et al. (1982) The effect of 2,5-hexanedione on reproductive hormones and
testicular enzyme activities in the F344 rat. Toxicol Appl Pharmacol 62:262-272.
Chapin, RE; Sloane, RA. (1996) Reproductive assessment by continuous breeding: evolving study design and
summaries of eighty-eight studies. Environ Health Perspect 104(Suppl. 2): 1-7.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-14 DRAFT—DO NOT CITE OR QUOTE
-------
Charbonneau Mm Couture, J; Plaa, GL. (1991) Inhalation versus oral administration of acetone: effect of the vehicle
on the potentiation of CCL4- induced liver injury. Toxicol Lett 57:47-54.
Charbotel, B; Fevotte, J; Hours, M; et al. (2005) Case-control study on renal cell cancer and occupational
trichloroethylene exposure, in the Arve valley (France). Final report. Submitted to U.S. EPA by Paul Dugard,
Ph.D., Halogenated Solvents Industry Alliance. #8EHQ-1105-16030.
Charbotel, B; Fevotte, J; Hours, M; et al. (2006) Case-control study on renal cell cancer and occupational exposure
to trichloroethylene. Part II: Epidemiological aspects. Ann Occup Hyg 50(8):777-787.
Charbotel, B; Fevotte, J; Martin, JL; et al. (2009) Cancer du rein et expositions au trichloroethylene: les valeurs
limites d'exposition professionnelle fracaises en vigueur sont-elles adaptees. Rev Epidemiol Sante Publique
57:41-47.
Charbotel, B; Gad, S; Cai'ola, D; et al. (2007) Trichloroethylene exposure and somatic mutations of the VHL gene in
patients with Renal Cell Carcinoma. J Occup Med Toxicol 2:13.
Chemical Marketing Reporter (1997) Chemical profile trichloroethylene. Dec 8, 1997. NY, NY: Schnell Pub Co.
Chen, C; Cheung, ST; So, S; et al. (2002) Gene expression patters in human liver cancers. Mol Biol Cell
13:1929-1939.
Chen, CW; (2000) Biologically based dose-response model for liver tumors induced by trichloroethylene. Environ
Health Perspect 108(Suppl 2):335-342.
Chen, SJ; Wang, JL; Chen, JH; et al. (2002) Possible involvement of glutathione and p53 in trichloroethylene- and
perchloroethylene-induced lipid peroxidation and apoptosis in human lung cancer cells. Free Radic Biol Med
33(4):464-472.
Chen, Y ; Cai, J ; Anders, MW ; et al. (2001) Role of mitochondrial dysfunction in S-(l,2-dichlorovinyl)-L-cysteine-
induced apoptosis. Toxicol Appl Pharmacol 170:172-180.
Cherrie, JW; Kromhout, H; Semple, S. (2001) The importance of reliable exposure estimates in deciding whether
trichloroethylene can cause kidney cancer. J Cancer Res Clin Oncol 127:400-404.
Chia, SE; Goh, VH; Ong, CN. (1997) Endocrine profiles of male workers with exposure to trichloroethylene. Am J
IndMed32(3):217-222.
Chia, SE; Ong, CN; Tsakok, MF; et al. (1996) Semen parameters in workers exposed to trichloroethylene. Reprod
Toxicol 10(4):295-299.
Chieco, P; Moslen, MT; Reynolds, ES. (1981) Effect of administration vehicle on oral 1,1-dichloroethane toxicity.
Toxicol Appl Pharm 57:146-155.
Chiesa, R; Donate, F; Tagger, A; et al. (2000) Etiology of hepatocellular carcinoma in Italian patients with and
without cirrhosis. Cancer Epidemiol Biomarkers Prev 9:213-216.
Chiu, BC; Weisenburger, DD. (2003) An update of the epidemiology of non-Hodgkin's lymphoma. Clin
Lymphoma 4:161-168.
Chiu, WA; Bois, FY. (2007) An approximate method for population toxicokinetic analysis with aggregated data. J
Agric Biol Environ Stat 12:346-363.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-15 DRAFT—DO NOT CITE OR QUOTE
-------
Chiu, WA; Okino, MS; Lipscomb, JC; et al. (2006) Issues in the pharmacokinetics of trichloroethylene and its
metabolites. Environ Health Perspect 1141450-1456.
Chiu, WA; White, P. (2006) Steady-state solutions to PBPK models and their applications to risk assessment I:
Route-to-route extrapolation of volatile chemicals. Risk Anal 26:769-780.
Chiu, WA. (2006) Statistical issues in physiologically-based pharmacokinetic modeling. Toxicokinetics and Risk
Assessment, ed. Lipscomb, JC, Ohanian, EV. New York: Informa Healthcare, p. 269-284.
Chiu, WA; Micallef, S;, Monster, AC; et al. (2007) Toxicokinetics of inhaled trichloroethylene and
tetrachloroethylene in humans at 1 ppm: empirical results and comparisons with previous studies. Toxicol Sci
95:23-36.
Chiu, WA; Okino, MS; Lipscomb, JC; et al. (2006) Issues in the pharmacokinetics of trichloroethylene and its
metabolites. Environ Health Perspect 114:1450-1456.
Chiu, WA; Okino, MS; Evans, MV. (2009) Characterizing uncertainty and population variability in the
toxicokinetics of trichloroethylene and metabolites in mice, rats, and humans using an updated database,
physiologically based pharmacokinetic (PBPK) model, and Bayesian approach. Toxicol Appl Pharmacol.
doi:10.1016/j.taap.2009.07.032. (In Press)
Chow, WH; Devesa, SS. (2008) Contemporary epidemiology of renal cell cancer. Cancer J. 14(5):288-301.
Chow, WH; Gridley, G; McLaughlin, JK; et al. (1994) Protein intake and risk of renal cell cancer. J Natl Cancer
Inst 86:1131-1139.
Cicmanec, JL; Condie, LW; Olson, GR; et al. (1991) 90-Day toxicity study of dichloroacetate in dogs. Fundam
Appl Toxicol 17:376-389.
Clapp, RW; Hoffman, K. (2008) Cancer mortality in IBM Endicott IBM workers, 1969-2001; an update on a NY
production plant. Environ Health 7:13.
Clay, P (2008) Assessment of the genotoxicity of trichloroethylene and its metabolite, S-(l,2-dichlorovinyl)-L-
cysteine (DCVC), in the comet assay in rat kidney. Mutagenesis 23(l):27-33.
Clegg, ED; Cook, JC, Chapin, RE; et al. (1997) Leydig cell hyperplasia and adenoma formation: mechanisms and
relevance to humans. Reprod Toxicol 11(1):107-121.
Clewell, HJ; Anderson, ME. (2004) Applying mode-of-action and pharmacokinetic considerations in contemporary
cancer risk assessments: an example with trichloroethylene. Crit Rev Toxicol 34(5):385-445.
Clewell, HJ, 3rd; Gentry, PR; Covington, TR; et al. (2000) Development of a physiologically based pharmacokinetic
model of trichloroethylene and its metabolites for use in risk assessment. Environ Health Persp 108(Suppl 2):283-
305.
Clewell, HJ, 3rd; Andersen, ME; Barton, HA. (2002) A consistent approach for the application of pharmacokinetic
modeling in cancer and noncancer risk assessment. Environ Health Perspect 110:85-93.
Coberly, S; Oudiz, DJ; Overstreet, JW; et al. (1992) Effects of maternal exposure to trichloroethylene (TCE) on cell
proliferation in the mouse preimplantation embryo. Reprod Toxicol 6:241-245.
Cohn, P; Klotz, J; Bove, F; et al. (1994) Drinking water contamination and the incidence of leukemia and non-
Hodgkin's lymphoma. Environ Health Perspect 102(6-7):556-561.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-16 DRAFT—DO NOT CITE OR QUOTE
-------
Cojocel, C; Beuter, W; Muller, W; et al. (1989) Lipid peroxidation: a possible mechanism of trichloroethylene-
induced nephrotoxicity. Toxicology 55:131-141.
Collier, JM; Selmin, O; Johnson, PD; et al. (2003) Trichloroethylene effects on gene expression during cardiac
development. Birth Defects Res A Clin Mol Teratol 67:488-495.
Colt, JS; Blair, A. (1998) Parental occupational exposures and risk of childhood cancer. Environ Health Perspect
106(Suppl 3):909-925.
Commandeur, JN; Vermeulen, NP. (1990) Identification of A7-acetyl(2,2-dichlorovinyl)- and A^-acety 1(1,2-
dichlorovinyl)-L-cysteine as two regioisomeric mercapturic acids of trichloroethylene in the rat. Chem Res Toxicol
3(3):212-218.
Cook, JC; Klinefelter, GR; Hardisty, JF; et al. (1999) Rodent Leydig cell tumorigenesis: a review of the physiology,
pathology, mechanisms, and relevance to humans. Crit Rev Toxicol 29:169-261.
Cooper, GS; Miller, FW; Pandey, JP. (1999) The role of genetic factors in autoimmune disease: implications for
environmental research. Environ Health Perspect 107(Suppl 5):693-700.
Cooper, GS; Parks, CG; Treadwell, EL; et al. (2004) Occupational risk factors for the development of systemic
lupus erythematosus. J Rheumatol 31:1928-193 3.
Cooper, GS, Stroehla, BC. (2003) The epidemiology of autoimmune diseases. AutoimmunRev 2:119-125.
Cooper, AJ. (2004) The role of glutamine transaminase k (GTK) in sulfur and alpha-keto acid metabolism in the
brain, and in the possible bioactivation of neurotoxicants. Neurochem Int 44(8):557-577.
Coopman, VAE; Cordonnier, JACM; De Letter, EA; et al. (2003) Tissue distribution of trichloroethylene in a case
of accidental acute intoxication by inhalation. Forensic Sci Int 134:115-119.
Cornett, R; James, MO; Henderson, GN; et al. (1999) Inhibition of glutathione S-transferase zeta and tyrosine
metabolism by dichloroacetate: a potential unifying mechanism for its altered biotransformation and toxicity.
Biochem Biophys Res Commun 262:752-756.
Gorton JC; Lapinskas, PJ. (2005) Peroxisome proliferator receptors: mediators of phthalate ester-induced effects in
the male reproductive tract? Toxicol Sci 83:4-17.
Gorton, JC. (2008) Evaluation of the role of peroxisome proliferator-activated receptor alpha (PPARalpha) in mouse
liver tumor induction by trichloroethylene and metabolites. Crit Rev Toxicol 38(10):857-875.
Cosby, NC; Dukelow, WR. (1992) Toxicology of maternally ingested trichloroethylene (TCE) on embryonal and
fetal development in mice and of TCE metabolites on in vitro fertilization. Fundam Appl Toxicol 19:268-274.
Costa, AK; Ivanetich, KM. (1984) Chlorinated ethylenes: their metabolism and effect on DNA repair in rat
hepatocytes. Carcinogenesis 5(12): 1629-1636.
Costa, G;, Merletti, F; Segnan, N. (1989) A mortality cohort study in a north Italian aircraft factory. Br J Ind Med
46:738-743.
Costantini, AS; Benvenuti, A; Vineis, P; et al. (2008) Risk of leukemia and multiple myeloma associated with
exposure to benzene and other organic solvents: Evidence from the Italian Multicenter Case-control study. Am J
Ind Med 51:803-811.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-17 DRAFT—DO NOT CITE OR QUOTE
-------
Costas, K; Knorr, RS; Condon, SK. (2002) A case-control study of childhood leukemia in Woburn, Massachusetts:
the relationship between leukemia incidence and exposure to public drinking water. Sci Total Environ
300(l-3):23-35.
Counts, JL; Sarmiento, JI; Harbison, ML; et al. (1996) Cell proliferation and global methylation status in mouse
liver after phenobarital and/or choline devoid, methioninedeficient diet administration. Carcinogenesis
17(6):1251-1257.
Crebelli, R; Bignami, M; Conti, L; et al. (1982) Mutagenicity of trichloroethylene in Salmonella typhimurium
TA100. Ann 1st Super Sanita 18(1):117-121.
Crebelli, R; Carere, A. (1989) Genetic toxicology of 1,1,2-trichloroethylene. MutatRes 221:11-37.
Crebelli, R; Conti, G; Conti, L; et al. (1991) In vitro studies with nine known or suspected spindle poisons: results in
tests for chromosome malsegregation in Aspergillus nidulans. Mutagenesis. 6(2): 131-136.
Crofton, KM; Lassiter, TL; Rebert, CS. (1994) Solvent-induced ototoxicity in rats: an atypical selective mid-
frequency hearing deficit. Hear Res 80:25-30.
Crofton, KM; Zhao, X. (1997) The ototoxicity of trichloroethylene: extrapolation and relevance of high-
concentration, short-duration animal exposure data. Fundam Appl Toxicol 38(1):101-106.
Cummings, BS; Parker, JC; Lash, LH. (2001) Cytochrome P450-dependent metabolism of trichloroethylene in rat
kidney. Toxicol Sci 60:11-19.
Cummings, BS; Lash, LH. (2000) Metabolism and toxicity of trichloroethylene and S'-(l,2-dichlorovinyl)--cysteine
in freshly isolated human proximal tubular cells. Toxicol Sci 53(2):458-466.
Cummings, BS; Lasker, JM; Lash, LH. (2000a) Expression of glutathione-dependent enzymes and cytochrome
P450s in freshly isolated and primary cultures of proximal tubular cells from human kidney. J Pharmacol Exp Ther
293:677-685.
Cummings, BS; Parker, JC; Lash, LH. (2000a) Role of cytochrome P450 and glutathione S-transferase alpha in the
metabolism and cytotoxicity of trichloroethylene in rat kidney. Biochem Pharmacol 59(5):531-543.
Cummings, BS; Zangar, RC; Novak, RF; et al. (2000b) Cytotoxicity of trichloroethylene and S-(l, 2-dichlorovinyl)-
L-cysteine in primary cultures of rat renal proximal tubular and distal tubular cells. Toxicology 150(l-3):83-98.
Cutler, JJ; Parker, GS; Rosen, S; et al. (1986) Childhood leukemia in Woburn, Massachusetts. Public Health Rep
101(2):201-205.
D'Souza, RW; Bruckner, JV; Feldman, S. (1985) Oral and intravenous trichloroethylene pharmacokinetics in the rat.
J Toxicol Env Health 15(5):587-601.
Dahlgren, J; Takhar,H; Anderson-Mahoney, P; et al, R. (2007) Cluster of systemic lupus erythematosus (SLE)
associated with an oil field waste site: a cross sectional study. Environ Health 6:8.
Dai, Y; Leng, S; Li, L; et al. (2004) Genetic polymorphisms of cytokine genes and risk for trichloroethylene-induced
severe generalized dermatitis: a case-control study. Biomarkers 9(6):470-478.
Dalbey, W; Bingham, E. (1978) Metabolism of trichloroethylene by the isolated perfused lung. Toxicol Appl
Pharmacol 43:267-277.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-18 DRAFT—DO NOT CITE OR QUOTE
-------
Dallas, CE; Gallo, JG; Ramanathan, R; et al. (1991) Physiological pharmacokinetic modeling of inhaled
trichloroethylene in rats. Toxicol Appl Pharm 110:303-314.
Damani, LA; Houdi, AA. (1988) Cytochrome P-450 and FAD-monoxygenase mediated S- and N-oxygenations.
Drag Metabol Drag Interact 6(3-4):235-244.
Daniel, FB; DeAngelo, AB; Stober, JA; et al. (1992) Hepatocarcinogenicity of chloral hydrate, 2-
chloroacetaldehyde, and dichloroacetic acid in the male B6C3F1 mouse. Fundam Appl Toxicol 19:159-168.
Darnerad, P; Brandt, I; Feil, V; et al. (1989) Dichlorovinyl cysteine (DCVC) in the mouse kidney: tissue-binding
and toxicity after glutathione depletion and probenecid treatment. Arch Toxicol 63:345-350.
Das, RM; Scott, JE. (1994) Trichloroethylene-induced pneumotoxicity in fetal and neonatal mice. Toxicol Lett
73:227-239.
David, RM; Clewell, HJ; Gentry, PR; et al. (2006) Revised assessment of cancer risk to dichloromethane II.
Application of probabilistic methods to cancer risk determinations. Regul Toxicol Pharmacol 45:55-65.
Davidson, IW; Beliles, RP. (1991) Consideration of the target organ toxicity of trichloroethylene in terms of
metabolite toxicity and pharmacokinetics. Drag Metab Rev 23(5-6):493-599.
Davis, SI; Laszlo Pallos, L; Wu, JQ; et al. (2005) ATSDR's trichloroethylene subregistry methods and results:
1989-2000. Arch Environ Occup Health 60(3): 130-139.
Davis, RR; Murphy, WJ; Snawder, JE; et al. (2002) Susceptibility to the ototoxic properties of toluene is species
specific. Hearing Res 166(l-2):24-32.
Dawson, BV; Johnson, PD; Goldberg, SJ; et al. (1990) Cardiac teratogenesis of trichloroethylene and
dichloroethylene in a mammalian model. J Am Coll Cardiol 16(5): 1304-1309.
Dawson, BV, Johnson, PD, Goldberg, SJ, et al. (1993) Cardiac teratogenesis of halogenated hydrocarbon-
contaminated drinking water. J Am Coll Cardiol 21(6): 1466-1472.
De Baere, S; Meyer, E; Dirinck, I; et al. (1997) Tissue distribution of trichloroethylene and its metabolites in a
forensic case. J Anal Toxicol 21:223-227.
de Jong, PE; Brenner, BM. (2004) From secondary to primary prevention of progressive renal disease: the case for
screening for albuminuria. Kidney Int. 66(6):2109-2118.
De Roos, AJ; Olshan, AF; Teschke, K; et al. (2001) Parental occupational exposures to chemicals and incidence of
neuroblastoma in offspring. AmJEpidemiol 154(2): 106-114. .
Deane, M; Swan, SH; Harris, JA; et al. (1989) Adverse pregnancy outcomes in relation to water contamination,
Santa Clara County, California, 1980-1981. AmJEpidemiol 129(5):894-904.
DeAngelo, AB, Daniel, FB; Wong, DM; et al. (2008) The induction of hepatocellular neoplasiaby trichloroacetic
acid administered in the drinking water of the male B6C3F1 mouse. J Toxicol Environ Health Part A 71:1056-1068.
DeAngelo, AB; George, MH; House, DE. (1999) Hepatocarcinogenicity in the male B6C3F1 mouse following a
lifetime exposure to dichloroacetic acid in the drinking water: dose-response determination and modes of action. J
Toxicol Environ Health 58:485-507.
DeAngelo, AB; Daniel, FB; McMillan, L; et al. (1989) Species and strain sensitivity to the induction of peroxisome
proliferation by chloroacetic acids. Toxicol Appl Pharmacol 101:285-298.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-19 DRAFT—DO NOT CITE OR QUOTE
-------
DeAngelo, AB; Daniel, FB; Most, BM; et al. (1996) The carcinogenicity of dichloroacetic acid in the male Fischer
344 rat. Toxicology 114:207-221.
DeAngelo, AB; Daniel, FB; Most, BM; et al. (1997) Failure of monochloroacetic acid and trichloroacetic acid
administered in the drinking water to produce liver cancer in male F344/N rats. J Toxicol Environ Health
52:425-445.
Deerberg F; Muller-Peddinghaus, R. (1970) [Comparative studies on the occurrence of spontaneous diseases and
spontaneous deaths in conventional and specific germ-free female NMRI-HAN mice]. Z Versuchstierkd
12:341-347.
Deerberg, F; Pittermann, W; Rapp, K. (1974) [Lung tumor in the mouse: a progressive tumor]. Vet Pathol
11:430-441.
Dees, C; Travis, C. (1993) The mitogenic potential of trichloroethylene inB6C3Fl mice. Toxicol Lett 69:129-137.
Degrassi, F; Tanzarella, C. (1988) Immunofluorescent staining of kinetochores in micronuclei: a new assay for the
detection of aneuploidy. Mutat Res 203(5):339-345.
Dehon, B; Humbert, L; Devisme, M; et al. (2000) Tetrachloroethylene and trichloroethylene fatality: case report and
simple headspace SPME-capillary gas chromatographic determination in tissues. J Anal Toxicol 24:22-26.
Dekant, W; Vamvakas, S; Berthold, K; et al. (1986) Bacterial beta-lyase mediated cleavage and mutagenicity of
cysteine conjugates derived from the nephrocarcinogenic alkenes trichloroethylene, tetrachloroethylene and
hexachlorobutadiene. Chem Biol Interact. 60(l):31-45.
Dekant, W; Berthold, K; Vamvakas, K; et al. (1988) Thioacylating intermediates as metabolites of S-(l,2-
dichlorovinyl)-L-cysteine and S'-(l,2,2-trichlorovinyl)--cysteine formed by cysteine conjugate (3-lyase. Chem Res
Toxicol 1:175-178.
Dekant, W; Koob, M; Henschler, D. (1990) Metabolism of trichloroethene - in vivo and in vitro evidence for
activation by glutathione conjugation. Chem Biol Interact 73:89-101.
Dekant, W; Metzler, M; Henschler, D. (1984) Novel metabolites of trichloroethylene through dechlorination
reactions in rats, mice and humans. Biochem Pharmacol 33(13):2021-2027.
Dekant, W; Metzler, M; Henschler, D. (1986b) Identification of S'-l,2-dichlorovinyl-A/-acetyl-cysteine as a urinary
metabolite of trichloroethylene: a possible explanation for its nephrocarcinogenicity in male rats. Biochem
Pharmacol 35(15):2455-2458.
Dekant, W; Schulz, A; Metzler, M; et al. (1986a) Absorption, elimination and metabolism of trichloroethylene: a
quantitative comparison between rats and mice. Xenobiotica 16(2): 143-152.
Delahunt, B; Bethwaite, PB; Nacey, JN. (1995) Occupational risk for renal cell carcinoma: A case-control study
based on the New Zealand Cancer Registry. Br J Urol 75:578-582.
DeMarini, DM; Perry, E; Shelton, ML. (1994) Dichloroacetic acid and related compounds: induction of prophage in
E. coli and mutagenicity and mutation spectra in Salmonella TA100. Mutagenesis 9(5):429-437.
DerSimonian, R; Laird, N. (1986) Meta-analysis in clinical trials. Control Clin Trials 7:177-188.
Ding, R; Shen, T; Zhu, QX. (2006) [Effects of TCE and PCE on cultured human keratinocyte lipid peroxidation and
protective effect of vitamin E on it]. ZhonghuaLao Dong Wei ShengZhi Ye BingZaZhi 24(ll):662-665.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-20 DRAFT—DO NOT CITE OR QUOTE
-------
Diot, E; Lesire, V; Guilmot, JL; et al. 2002. Systemic sclerosis and occupational risk factors: a case-control study.
Occup Environ Med 59:545-549.
DiRenzo, AB; Gandolfi, AJ; Sipes, IG. (1982) Microsomal bioactivation and covalent binding of aliphatic halides to
DNA. T oxicolLett 11:243-252.
Dix, KJ; Kedderis, GL; Borghoff, SJ. (1997) Vehicle-dependent oral absorption and target tissue dosimetry of
chloroform in male rats and female mice. Toxicol Lett 91:197-209.
Dohn, DR; Anders, MW. (1982) Assay of cysteine conjugate (3-lyase activity with5'-(2-benzothiazolyl)cysteine as
the substrate. AnalBiochem 120:379-386.
Dorfmueller, MA; Henne, SP; York, G; et al. (1979) Evaluation of teratogenicity and behavioral toxicity with
inhalation exposure of maternal rats to trichloroethylene. Toxicology 14:153-166.
Dome, JL; Walton, K; Renwick, AG. (2005) Human variability in xenobiotic metabolism and pathway-related
uncertainty factors for chemical risk assessment: a review. Food Chem Toxicol 43(2):203-216.
Dosemeci, M; Cocco, P; Chow, WH. (1999) Gender differences in risk of renal cell carcinoma and occupational
exposures to chlorinated aliphatic hydrocarbons. Am J Ind Med 36(l):54-59.
Dossing, M; Petersen, KT; Vyberg, M; et al. (1997) Liver cancer among employees in Denmark. Am J Ind Med
32:248-254.
Douglas, GR; Gingerich, JD; Soper, LM; et al. (1999) Evidence for the lack of base-change and small-deletion
mutation induction by trichloroethylene in lacZ transgenic mice. Environ Mol Mutagen. 34(2-3): 190-194.
Dow, JL; Green, T. (2000) Trichloroethylene induced vitamin B(12) and folate deficiency leads to increased formic
acid excretion in the rat. Toxicology 146:123-136.
Dowsley, TF; Ulreich, JB; Bolton, JL; etal. (1996) CYP2E1-dependent bioactivation of 1,1-dichloroethylene in
murine lung: formation of reactive intermediates and glutathione conjugates. Toxicol Appl Pharm 139(l):42-48.
Drake, VJ; Koprowski, SL; Hu, N; et al. (2006b) Cardiogenic effects of trichloroethylene and trichloroacetic acid
following exposure during heart specification of avian development. Toxicol Sci 94(1): 153-162.
Drake, VJ; Koprowski, SL; Lough, J; et al. (2006a) Trichloroethylene exposure during cardiac valvuloseptal
morphogenesis alters cushion formation and cardiac hemodynamics in the avian embryo. Environ Health Perspect
114(6):842-847.
Driscoll, TR; Harridan, HH; Wang, G; et al. (1992) Concentrations of individual serum or plasma bile acids in
workers exposed to chlorinated aliphatic hydrocarbons. Br J Ind Med 49:700-705.
Dryver, E; Brandt, L; Kauppinen, T; et al. (2004) Occupational exposures and non-Hodgkin's lymphoma in
Southern Sweden. Int J Occup Environ Health 10:13-21.
D'Souza, RW; Bruckner, JV; Feldman, S. (1985) Oral and intravenous trichloroethylene pharmacokinetics in the rat.
J Toxicol Environ Health 15(5):587-601.
Duffell, MW; Jakobi, WB. (1982) Cysteine s-conjugate w-acetyltransferase from rat kidney microsomes. Mol
Pharmacol 21:444-448.
Dugard, PH. (2000) Letters to the Editor. Toxicol Sci 56:437-438.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-21 DRAFT—DO NOT CITE OR QUOTE
-------
DuTeaux, SB; Berger, T; Hess, RA; et al. (2004b) Male reproductive toxicity of trichloroethylene: sperm protein
oxidation and decreased fertilizing ability. Biol Reprod 70:1518-1526.
DuTeaux, SB; Berger, T; Hess, RA; et al. (2008) CORRECTION: Male reproductive toxicity of trichloroethylene:
sperm protein oxidation and decreased fertilizing ability. Biol Reprod 79:787. (dosage error in text).
DuTeaux, SB; Hengel, MJ; DeGroot, DE; et al. (2003) Evidence for trichloroethylene bioactivation and adduct
formation in the rat epididymis and efferent ducts. Biol Reprod 69:771-779.
DuTeaux, SB; Newman, JW; Morisseau, C; et al. (2004a) Epoxide hydrolases in the rat epididymis: possible roles in
xenobiotic and endogenous fatty acid metabolism. Toxicol Sci 78:187-195.
Eastmond, DA; Tucker, JD. (1989) Identification of aneuploidy-inducing agents using cytokinesis-blocked human
lymphocytes and an antikinetochore antibody. Environ Mol Mutagen 13(l):34-43.
Eastmond, DA; Hartwig, A; Anderson, D; et al. (2009) Mutagenicity testing for chemical risk assessment: update of
the WHO/IPCS Harmonized Scheme. Mutagenesis: 24(4):341-349.
Eichenlaub-Ritter, U; Baart, E.; Yin, H; et al. (1996) Mechanisms of spontaneous and chemically-induced
aneuploidy in mammalian oogenesis: Basis of sex-specific differences in response to aneugens and the necessity for
further tests. Mutat Res 372:279-294.
El Ghawabi, SM; Mansoor, MB; El Gamel, MS; et al. (1973) Chronic trichloroethylene exposure. J Egypt Med
Assoc56(ll-12):715-724.
Elcombe, CR; Rose, MS; Pratt, IS. (1985) Biochemical, histological, and ultrastructural changes in rat and mouse
liver following the administration of trichloroethylene: possible relevance to species differences in
hepatocarcinogenicity. Toxicol Appl Pharmacol 79:365-376.
Elfarra, AA; Krause, RJ; Last, AR; et al. (1998) Species- and sex-related differences in metabolism of
trichloroethylene to yield chloral and trichloroethanol in mouse, rat, and human liver microsomes. Drug Metab
Dispos 26(8):779-785.
Elfarra, AA;Hwang, IY. (1990) In vivo metabolites of 5'-(2-benzothiozolyl)~cysteine as markers of in vivo cysteine
conjugate (3-lyase and thiol glucuronosyl transferase activities. Drug Metab Dispos 18(6):917-922.
Elfarra, AA; Anders, MW. (1984) Commentary. Renal processing of glutathione conjugates role in nephrotoxicity.
Biochem Pharmacol 33(23):3729-3732..
Elfarra, AA;Lash, LH; Anders, MW. (1986b) a-Ketoacids stimulate rat renal cysteine conjugate (3-lyase activity and
potentiate the cytotoxicity of ^-(l^-dichloroviny^-L-cysteine. Mol Pharmacol 31:208-212.
Elfarra, AA; Jakobson, I; Anders, MW. (1986a) Mechamsmof S-(l,2-dicUorovinyl)grutathione-induced
nephrotoxicity. Biochem Pharmacol 35(2):283-288.
El-Ghawabi, SM; Mansoor, MB; El Gamel, MS; et al. (1973) Chronic trichloroethylene exposure. J Egypt Med
Assoc 56(ll-12):715-724.
Elizur, A; Adair-Kirk, TL; Kelley, DG; et al. (2008) Tumor necrosis factor-alpha from macrophages enhances LPS-
induced clara cell expression of keratinocyte-derived chemokine. Am J Respir Cell Mol Biol 38:8-15 (Epub).
Elovaara, G; Hemminki, K; Vainio, H. (1979) Effects of methylene chloride, trichlororethane, trichloroethylene,
tetrachloroethylene and toluene on the development of chick embryos. Toxicology 12:111-119.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-22 DRAFT—DO NOT CITE OR QUOTE
-------
El-Serag, HB; Rudolph, KL. (2007) Hepatocellular carcinoma: epidemiology and molecular carcinogenesis.
Gastroenterology 132:2557-2576.
El-Serag, HB. (2007) Epidemiology of hepatocellular carcinoma in USA. Hepatology Res 37(Suppl 2):S88-S94.
Engel, LS; Vaughan, TL; Gammon, MD; et al. (2002) Occupational risk of esophageal and gastric cardia
adenocarcinoma. Am J Ind Med 42:11-22.
Environmental Health Strategies.. (1997) Final Report. Correspondence from Paul A. Cammer, Ph.D.,
Trichloroethylene Issues Group, to Cheryl Siegel Scott, US Environmental Protection Agency. December 22, 1997.
Epstein, DL; Nolen, GA; Randall, JL; et al. (1992) Cardiopathic effects of dichloroacetate in the fetal Long-Evans
rat. Teratology 46:225-235.
Ertle, T; Henschler, D; Mtiller, G; et al. (1972) Metabolism of trichloroethylene in man. I. The significance of
trichloroethanol in long-term exposure conditions. Arch Toxikol 29:171-188.
Evans, MV; Chiu, WA; Okino, MS; et al. Development of an updated PBPK model for trichloroethylene and
metabolites in mice, and its application to discern the role of oxidative metabolism in TCE-induced hepatomegaly.
Toxicol Appl Pharmacol 236:329-340.
Evert, M; Schneider-Stock, R; Dombrowski, F. (2003) Apolipoprotein A-IV mRNA overexpression in early
preneoplastic hepatic foci induced by low-number pancreatic islet transplants in streptozotocin-diabetic rats. Pathol
Res Pract 199(6):373-379.
Eyre, PJ; Stevens, DK; Parker, JC; et al. (1995a) Acid-labile adducts to protein can be used as indicators of the
cysteine S-conjugate pathway of trichloroethene metabolism. J Toxicol Environ Health 46(4):443-464.
Eyre, PJ; Stevens, DK; Parker, JC; et al. (1995a) Renal activation of trichloroethene and S-(l,2-dichlorovinyl)-L-
cysteine and cell proliferative responses in the kidneys of F344 rats and B6C3F1 mice. J Toxicol Environ Health
46:465-481.
Fagliano, J; Berry, M; Bove, F; et al. 1990 Drinking water contamination and the incidence of leukemia: an ecologic
study. Am J Public Health 80(10):1209-1212.
Fahrig, R; Madle, S; Baumann, H. (1995) Genetic toxicology of trichloroethylene (TCE). Mutat Res 340:1-36.
Fairman, MJ; Scott, BB; Toothill, C; et al. (1977) Jejunal mucosal gamma glutamyl transferase activity in coeliac
disease. Gut. 18(6):484-487.
Falk Filipsson, A; Victorin, K. (2003) Comparison of available benchmark dose softwares and models using
trichloroethylene as a model substance. Regul Toxicol Pharmacol 37(3):343-55.
Fan, AM. (1988) Trichloroethylene: water contamination and health risk assessment. Rev Environ Contam Toxicol
101:55-92.
Fattovich, G; Stroffolini, T; Zagni, I; et al. (2004) Hepatocellular carcinoma in cirrhosis: incidence and risk factors.
Gastroenterology 127:835-850.
Fechter, LD; Liu, Y; Herr, et al. (1998) Trichloroethylene ototoxicity: Evidence for a cochlear origin. Toxicol Sci
42:28-35.
Feitelson, MA;, Sun, B; Satiroglu-Tufan, NL; et al. (2002) Genetic mechanisms of hepatocarcinogenesis. Oncogene
21:2593-2604.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-23 DRAFT—DO NOT CITE OR QUOTE
-------
Feldman, RG; Chirico-Post, J; Proctor, SP. (1988) Blink reflex latency after exposure to trichloroethylene in well
water. Arch Environ Health 43(2): 143-148.
Feldman, RG; Niles, C; Proctor, SP; et al. (1992) Blink reflex measurement of effects of trichloroethylene exposure
on the trigeminal nerve. Muscle Nerve 15(4):490-495.
Ferlay, J; Bray, F; Pisani, P; et al. (2004) GLOBOCAN 2002: cancer incidence, mortality and prevalence worldwide.
IARC CancerBase No. 5, Version 2.0. Lyon, France: IARC Press.
Fernandez, JG; Droz, PO; Humbert BE; et al. (1977) Trichloroethylene exposure. Simulation of uptake, excretion,
and metabolism using a mathematical model. Br J Ind Med 34:43-55.
Ferreira-Gonzalez DeAngelo, AB; Nasim, S; Garret, CT. (1995) Ras oncogene activation during
hepatocarcinogenesis in B6 C3F1 male mice by dichloroacetic and trichloroacetic acids. Carcinogenesis
16(3):495-500.
Fevotte, J; Charbotel, B; Muller-Beaute, P; et al. (2006) Case-control study on renal cell cancer and occupational
exposure to trichloroethylene. Part I: Exposure assessment. Ann Occup Hyg 50(8):765-775.
Figgs, LW; Dosemeci, M; Blair, A. (1995) United States non-Hodgkin's lymphoma surveillance by occupation
1984-1989: a twenty-four state death certificate study. Am J Ind Med 27:817-835.
Filser, JG; Bolt, HM. (1979) Pharmacokinetics of halogenated etylenes in rats. Arch Toxicol 42:123-136.
Finckh, A; Cooper, GS; Chibnik, LB; et al. (2006) Occupational silica and solvent exposures and risk of systemic
lupus erythematosus in urban women. Arthritis Rheum 54:3648-3654.
Fiserova-Bergerova, V; Tichy, M; Di Carlo, FJ. (1984) Effects of biosolubility on pulmonary uptake and disposition
of gases and vapors of lipophilic chemicals. Drug Metab Rev 15(5-6): 1033-1070.
Fisher J; Mahle, D; Bankston, L; et al. (1997) Lactational transfer of volatile chemicals in breast milk. Am Ind Hyg
AssocJ58(6):425-431.
Fisher, JW; Allen, BC. (1993) Evaluating the risk of liver cancer in humans exposed to trichloroethylene using
physiological models. Risk Anal 13(l):87-95.
Fisher, JW; Mahle, D; Abbas, R. (1998) A human physiologically based pharmacokinetic model for
trichloroethylene and its metabolites, trichloroacetic acid and free trichloroethanol. Toxicol Appl Pharmacol
152(2):339-359.
Fisher, JW; Whittaker, TA; Taylor, DH; et al. (1990) Physiologically based pharmacokinetic modeling of the
lactating rat and nursing pup: a multiroute exposure model for trichloroethylene and its metabolite, trichloroacetic
acid. Toxicol Appl Pharmacol 102(3):497-513.
Fisher, JW; Whittaker, TA; Taylor, DH; et al. (1989) Physiologically based pharmacokinetic modeling of the
pregnant rat: a multiroute exposure model for trichloroethylene and its metabolite, trichloroacetic acid. Toxicol
Appl Pharmacol 99(3):395-414.
Fisher JW. (2000) Physiologically based pharmacokinetic models for trichloroethylene and its oxidative metabolites.
Environ Health Perspect 108(Suppl 2):265-273.
Fisher, JW; Channel, SR; Eggers, JS; et al. (2001) Trichloroethylene, trichloroacetic acid, and dichloroacetic acid:
do they affect fetal rat heart development? Int J Toxicol 20:257-267.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-24 DRAFT—DO NOT CITE OR QUOTE
-------
Fisher, JW; Gargas, ML; Allen, BC; et al. (1991) Physiologically based pharmacokinetic modeling with
trichloroethylene and its metabolite, trichloroacetic acid, in the rat and mouse. Toxicol Appl Pharm 109:183-195.
Fleming-Jones, ME; Smith, RE. (2003) Volatile organic compounds in foods: A five year study. J Agric Food
Chem 51:8120-8127.
Ford, ES; Rhodes, S; McDiarmid, M; et al. (1995) Deaths from acute exposure to trichloroethylene. J Occup
Environ Med 37(6):750-754.
Forkert, PG; Baldwin, RM; Millen, B; et al. (2005) Pulmonary bioactivation of trichloroethylene to chloral hydrate:
relative contributions of CYP2E1, CYP2F and CYP2B1. DrugMetab Dispos 33:1429-1437.
Forkert, PG; Birch, DW. (1989) Pulmonary toxicity of trichloroethylene in mice. Drug Metabolism and Disposition
17:106-113.
Forkert, PG; Forkert, L. (1994) Trichloroethylene induces pulmonary fibrosis in mice. Can J Physiol Pharmacol 72:
205-210.
Forkert, PG; Lash, L; Tardif, R; et al. (2003) Identification of trichloroethylene and its metabolites in human seminal
fluid of workers exposed to trichloroethylene. Drug Metab Dispos 31(3):306-311.
Forkert, PG; Millen, B; Lash, LH; et al. (2006) Pulmonary bronchiolar cytotoxicity and formation of dichloroacetyl
lysine protein adducts in mice treated with trichloroethylene. J Pharmacol Exp Ther 316:520-529.
Forkert, PG; Sylvestre, PL; Poland, JS. (1985) Lung injury induced by trichloroethylene. Toxicology 35:143-160.
Forkert, PG. (1999a) 1,1-Dichloroethylene-induced Clara cell damage is associated with in situ formation of the
reactive epoxide. Immunohistochemical detection of its glutathione conjugate. Am J Resp Cell Mol Biol
20(6):1310-1318.
Forkert, PG. (1999b) In vivo formation and localization of 1,1-dichloroethylene epoxide in murine liver:
identification of its glutathione conjugate 2-5'-glutathionyl acetate. J Pharmacol Exp Ther 290(3): 1299-1306.
Forkert, P-G; Baldwin, RM, Millen, B; et al. (2005) Pulmonary bioactivation of trichloroethylene to chloral hydrate:
Relative contributions of CYP2E1, CYP2F, and CYP2B1. DrugMetab Disp 33(10):1429-1437.
Forkert, P-G; Lash, LH; Nadeau V; et al. (2002) Metabolism and toxicity of trichloroethylene in epididymis and
testis. Toxicol Appl Pharmacol 182:244-254.
Forkert, PG; Malkinson, AM; Rice, P; et al. (1999) Diminished CYP2E1 expression and formation of 2-5*-
glutathionyl acetate, a glutathione conjugate derived from 1,1-dichloroethylene epoxide, in murine lung tumors.
Drug Metab Dispos 27(l):68-73.
Forkert, PG; Millen, B; Lash, LH; et al. (2006) Pulmonary bronchiolar cytotoxicity and formation of dichloroacetyl
lysine protein adducts in mice treated with trichloroethylene. J Pharmacol Exp Ther 316(2):520-529.
Fort, DJ; Rayburn, JR; DeYoung, DJ; et al. (1991) Assessing the efficacy of an aroclor 1254-induced exogenous
metabolic activation system for FETAX. Drug Chem Toxicol 14(1&2): 143-160.
Fort, DJ; Stover, EL; Rayburn, JR; et al. (1993) Evaluation of the developmental toxicity of trichloroethylene and
detoxification metabolites using Xenopus. Teratog Carcinog Mutagen 13:35-45.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-25 DRAFT—DO NOT CITE OR QUOTE
-------
Fox, TR; Schumann, AM; Watanabe, PG; et al. (1990) Mutational analysis of H-ras oncogene in spontaneous
C57B176 X C3H/He mouse liver tumors and tumors induced with genotoxic and nongenotoxic hepatocarcinogens.
Cancer Res 50:4014-4019.
Fraga, MF; Ballestar, E; Villar-Garea, A; et al. (2005) Loss of acetylation at Lysl6 and trimethylation at Lys20 of
histone H4 is a common hallmark of human cancer. Nat Genet 37:391-400.
Fraga, MF; Herranz, M; Espada, J; et al. (2004) A mouse skin multistage carcinogenesis model reflects the aberrant
DNA methylation patterns of human tumors. Cancer Res 64:5527-5534.
Franco, A; Costoya, MA; Roca, E. (2007) Estimating risk during showering exposure to VOCs of workers in a
metal-degreasing facility. J Toxicol Environ Health A 70(7):627-637.
Franco, JL; Teixeira, A; Meotti, FC; et al. (2006) Cerebellar thiol status and motor deficit after lactational exposure
to methylmercury. Environ Res 102:22-28.
Fredriksson, A; Danielsson, BRG; Eriksson, P. (1993) Altered behavior in adult mice orally exposed to tri- and
tetrachloroethylene as neonates. Toxicol Lett 66:13-19.
Fuduka, K; Takemoto, K; Tsuruta, H. (1983) Inhalation carcinogenicity of trichloroethylene in mice and rats. Ind
Health 21:243-254.
Purge, KA; Dykema, K; Petillo, D; et al. (2007) Combining differential expression, chromosomal and pathway
analyses for the molecular characterization of renal cell carcinoma. Can Urol Assoc J(CUAJ) 1(2 Suppl):S21-S27.
Furnus, CC ; Ulrich, MA ; Terreros, MC ; et al. (1990) The induction of aneuploidy in cultured Chinese hamster
cells by propionaldehyde and chloral hydrate. Mutagenesis 5(4):323-326.
Fuscoe, JC; Afshari, AJ; George, MH; etal. (1996) In vivo genotoxicity of dichloroacetic acid: evaluation with the
mouse peripheral blood micronucleus assay and the single cell gel assay. Environ Mol Mutagen 27(1): 1-9.
Gallou, C; Longuemaux, S; Delomenie, C; et al. (2001) Association of GSTT1 non-null and NAT1 slow/rapid
genotypes with von Hippel-Lindau tumour suppressor gene tranversions in sporadic renal cell carcinoma.
Pharmacogenetics 11:521-535.
Galloway, SM; Armstrong, MJ; Reuben, C; et al. (1987) Chromosome aberrations and sister chromatid exchanges in
Chinese hamster ovary cells: evaluations of 108 chemicals. Environ Mol Mutagen 10(Suppl 10): 1-175.
Gama, C; Meira, JB. (1978) Occupational acro-osteolysis. J Bone Joint Surg Am 60:86-90.
Gamberale, F; Annwall, G; Olson, BA. (1976) Exposure to trichloroethylene III. Psychological functions. Scand J
Work Environ Health 2(4):220-224.
Gandhi, M; Aweeka, F; Greenblatt, RM; et al. (2004) Sex differences in pharmacokinetics and pharmacodynamics.
Annu Rev Pharmacol Toxicol 44:499-523.
GAO (U.S. Government Accountability Office). (2007a) Defense Health Care: Activities related to past drinking
water contamination at Marine Corps Base Camp Lejeune. Report to Congressional Committees. GAO-07-276.
GAO (U.S. Government Accountability Office). (2007b) Defense Health Care: Issues related to past drinking water
contamination at Marine Corps Base Camp Lejeune. Statement of Marcia Crosse, Director, Health Care. Testimony
before the Subcommittee on Oversight and Investigations, Committee on Energy and Commerce, House of
Representatives. GAO-07-933T.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-26 DRAFT—DO NOT CITE OR QUOTE
-------
Garabrant, DH; Held, J; Langholz, B; et al. (1988) Mortality of aircraft manufacturing workers in southern
California. Am J Ind Med 13:683-693.
Garabrant, DH; Lacey, JV, Jr; Laing, TJ; et al. (2003) Scleroderma and solvent exposure among women. Am J
Epidemiol 157(6):493-500.
Gargas, ML; Seybold, PG; Andersen, ME. (1988) Modeling the tissue solubilities and metabolic rate constant
(Vmax) of halogenated methanes, ethanes, and ethylenes. Toxicol Lett 43:235-256.
Gargas, ML; Burgess, RJ; Voisard, DE; et al. (1989) Partition coefficients of low-molecular-weight volatile
chemicals in various liquids and tissues. Toxicol Appl Pharm 98(l):87-99.
Garte, S; Gaspari, L; Alexandrie, AK; et al (2001) Metabolic gene polymorphism frequencies in control populations.
Cancer Epidemiol Biomarkers Prev 10(12): 1239-1248.
Gash, DM; Rutland, K; Hudson, NL; et al. (2008) Trichloroethylene: Parkinsonism and complex 1 mitochondria!
neurotoxicity. Ann Neurol 63(2): 184-192.
Ge, R; Tao, L; Kramer, PM; et al (2002) Effect of peroxisome proliferators on the methylation and protein level of
c-myc protooncogene in B6C3F1 mice liver. J Biochem Mol Toxicol 16(l):41-47.
Ge, R; Yang, S; Kramer, PM; etal. (2001) The effect of dichloroacetic acid and trichloroacetic acidonDNA
methylation and cell proliferation in B6C3F1 mice. J Biochem Mol Toxicol 15:100-106.
Gelman, A; Bois, F; Jiang, J. (1996) Physiological pharmacokinetic analysis using population modeling and
infromatibe prior distributions. J Am Stat Assoc 91:1400-1412.
Gelman, A; Carlin, JB; Stern, HS; et al. (2004) Bayesian Data Analysis. New York: Chapman & Hall/CRC.
George, J; Byth, K; Farrell, GC. (1995) Age but not gender selectively affects expression of individual cytochrome
P450 proteins in human liver. Biochem Pharmacol 50(5):727-730.
George, MH; Moore, T; Kilburn, S; et al. (2000) Carcinogenicity of Chloral hydrate administered in drinking water
to the Male F344/N rat and Male B6C3F1 mouse. Toxicol Pathol 28(4):610-618.
George, JD; Reel, JR; Myers, CB; et al. (1986) Trichloroethylene: reproduction and fertility assessment in F344 rats
when administered in the feed. RTI Project No. 310-2344, NTP-86-085. National Institute of Environmental Health
Sciences, National Toxicology Program, RTF, NC.
Ghantous, H; Danielsson, BR; Dencker, L; et al. (1986) Trichloroacetic acid accumulates in murine amniotic fluid
after tri- and tetrachloroethylene inhalation. Acta Pharmacol Toxicol (Copenh) 58(2): 105-114.
Ghoshal, AK; Farber, E. (1984) The induction of liver cancer by dietary deficiency of choline and methionine
without added carcinogens. Carcinogenesis(lO): 1367-1370.
Giardino, NJ; Andelman, JB. (1996) Characterization of the emissions of trichloroethylene, chloroform, and 1,2-
dibromo-3-chloropropane in a full-size, experimental shower. J Expo Anal Environ Epidemiol 6(4):413-423.
Gibson, DP; Aardema, MJ; Kerckaert, GA; et al. (1995) Detection of aneuploidy-inducing carcinogens in the Syrian
hamster embryo (SHE) cell transformation assay. Mutat Res 343(l):7-24.
Gilbert, KM; Griffin, JM; Pumford, NR. (1999) Trichloroethylene activates CD4+ T cells: potential role in an
autoimmune response. DrugMetab Rev 31:901-916.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-27 DRAFT—DO NOT CITE OR QUOTE
-------
Gilbert, KM; Pumford, NR; Blossom, SJ. (2006) Environmental contaminant trichloroethylene promotes
autoimmune disease and inhibits T-cell apoptosis in MRL(+/+) Mice. J Immunotoxicol 3:263-267.
Gilbert, KM; Whitlow, AB; Pumford, NR. (2004) Environmental contaminant and disinfection by-product
trichloroacetaldehyde stimulates T cells in vitro. Int Immunopharmacol 4:25-36.
Gilkeson, GS; Keil, D; Peden-Adams, MM. (2004) Immune effects of trichloroethylene on autoimmune disease in
mice. In: Trichloroethylene - the scientific basis of risk assessment (Mohr LC, Hoel DG, Jollow D, eds). Charleston,
S.C.: The Medical University of South Carolina Press, 87-98.
Giller, S; Le Curieux, F; Gauthier, L; Erb, F; Marzin, D. (1995) Genotoxicity assay of chloral hydrate and
chloropicrine. MutatRes 348(4): 147-52.
Giller, S; Le Curieux, F; Erb, F; et al. (1997) Comparative genotoxicity of halogenated acetic acids found in
drinking water. Mutagenesis 12(5):321-328.
Ginsberg, G; Hattis, D; Miller, R; et al. (2004a) Pediatric pharmacokinetic data: implications for environmental risk
assessment for children. Pediatrics 113(Suppl. 4):973-983.
Ginsberg, G; Hattis, D; Russ, A; et al. (2005) Pharmacokinetic and pharmacodynamic factors that can affect
sensitivity to neurotoxic sequelae in elderly individuals. Environ Health Perspect 113(9): 1243-1249.
Ginsberg, G; Hattis, D; Sonawane, B; et al. (2002) Evaluation of child/adult pharmacokinetic differences from a
database derived from the therapeutic drug literature. Toxicol Sci 66(2): 185-200.
Ginsberg, G; Hattis, D; Sonawane, B. (2004b) Incorporating pharmacokinetic differences between children and
adults in assessing children's risks to environmental toxicants. Toxicol Appl Pharmacol 198(2):164-83.
Gochfeld, M. (2007) Framework for gender differences in human and animal toxicology. Environ Res 104(1):4-21.
Goel, SK; Rao, GS; Pandya, KP; et al. (1992) Trichloroethylene toxicity in mice: a biochemical, hematological and
pathological assessment. Indian J Exp Biol 30:402-406.
Goeptar, AR; Commandeur, JN; van Ommen, B; et al. (1995) Metabolism and kinetics of trichloroethylene in
relation to toxicity and carcinogenicity. Relevance of the mercapturic acid pathway. Chem Res Toxicol 8(1):3-21.
Goh, VH; Chia, SE; Ong, CN. (1998) Effects of chronic exposure to low doses of trichloroethylene on steroid
hormone and insulin levels in normal men. Environ Health Perspect 106(l):41-44.
Goldberg, SJ; Lebowitz, MD; Graver, EJ; et al. (1990) An association of human congenital cardiac malformations
and drinking water contaminants. J Am Coll Cardiol 16(1): 155-164.
Goldsworthy, TL; Lyght, O; Burnett, VL; et al. (1988) Potential role of a-2u-globulin, protein droplet accumulation,
and cell replication in the renal carcinogenicity of rats exposed to trichloroethylene, perchloroethylene, and
pentachloroethane. Toxicol Appl Pharmacol 96:367-379.
Goldsworthy, TL; Popp, JA. (1987) Chlorinated hydrocarbon-induced peroxisomal enzyme activity in relation to
species and organ carcinogenicity. Toxicol Appl Pharmacol 88:225-233.
Gonzalez-Leon, A; Merdink, JL; Bull, RJ; et al. (1999) Effect of pre-treatment with dichloroacetic or trichloroacetic
acid in drinking water on the pharmacokinetics of a subsequent challenge dose in B6C3F1 mice. Chem Biol Interact
123(3):239-253.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-28 DRAFT—DO NOT CITE OR QUOTE
-------
Grandjean, E; Mtinchinger, R; Turrian, V; et al. (1955) Investigations into the effects of exposure to
trichloroethylene in mechanical engineering. Brit J Industr Med 12:131-142.
Grasl-Kraupp, B; Waldhor, T; Huber, W; et al. (1993) Glutathione S-transferase isoenzyme patters in different
subtypes of enzyme-altered rat liver foci treated with the peroxisome proliferator nafenopin or with Phenobarbital.
Carcinogenesis 14:2407-2412.
Grawe, J; Musse, M; Adler, ID. (1997) Quantative and qualitative studies of micronucleus induction in mouse
erythrocytes using flow cytometry. I. Measurement of micronucleus induction in peripheral blood polychromatic
erythrocytes by chemicals with known and suspected genotoxicity. Mutagenesis 12:1-8.
Green, T; Dow, J; Ong, CN; et al. (2004) Biological monitoring of kidney function among workers occupationally
exposed to trichloroethylene. Occup Environ Med 61:312-317.
Green, LC; Lash, TL. (1999) Re: "Renal cell cancer correlated with occupational exposure to trichloroethylene". J
Cancer Res ClinOncol 125:430-432.
Green, T; Mainwaring, GW; Foster, JR. (1997) Trichloroethylene-induced mouse lung tumors: Studies of the mode
of action and comparisons between species. Fundam Appl Toxicol 37:125-130.
Green, T. (2000) Pulmonary toxicity and carcinogenicity of trichloroethylene: species differences and modes of
action. Environ Health Perspect 108(Suppl 2):261-264.
Green, T; Prout, MS. (1985) Species differences in response to trichloroethylene. II. Biotransformation in rats and
mice. Toxicol Appl Pharmacol 79:401-411.
Green, T; Odum, J; Nash, JA; et al. (1990) Perchloroethylene-induced rat kidney tumors: An investigation of the
mechanisms involved and their relevance to humans. Toxicol Appl Pharm 103:77-89.
Green, T; Dow, J; Ellis, MK; et al. (1997a) The role of glutathione conjugation in the development of kidney
tumours in rats exposed to trichloroethylene. Chem Biol Interact 105(2):99-117.
Green, T; Dow, J; Foster, J. (2003) Increased formic acid excretion and the development of kidney toxicity in rats
following chronic dosing with trichloroethanol, a major metabolite of trichloroethylene. Toxicology 191:109-119.
Green, T; Dow, J; Foster, JR; et al. (1998) Formic acid excretion in rats exposed to trichloroethylene: a possible
explanation for renal toxicity in long-term studies. Toxicology 127:39-47.
Green, T; Prout, MS. (1985) Species differences in response to trichloroethylene. II. Biotransformation in rats and
mice. Toxicol Appl Pharmacol 79:401-411.
Greenberg, MS; Burton, GA; Fisher, JW. (1999) Physiologically based pharmacokinetic modeling of inhaled
trichloroethylene and its oxidative metabolites in B6C3F1 mice. Toxicol Appl Pharmacol 154:264-178.
Greenland, S; Salvan, A; Wegman, DH; et al. (1994) A case-control study of cancer mortality at the transformer-
assembly facility. Int Arch Occup Environ Health 66:49-54.
Greim, H; Bonse, G; Radwan, Z; et al. (1975) Mutagenicity in vitro and potential carcinogenicity of chlorinated
ethylenes as a function of metabolic oxirane formation. Biochem Pharmacol 24:2013-2017.
Griffin, JM; Blossom, SJ; Jackson, SK; et al. (2000a) Trichloroethylene accelerates an autoimmune response by Thl
T cell activation in MRL +/+ mice. Immunopharmacology 46:123-137.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-29 DRAFT—DO NOT CITE OR QUOTE
-------
Griffin, JM; Gilbert, KM; Lamps, LW; et al. (2000b) CD4(+) T-cell activation and induction of autoimmune
hepatitis following trichloroethylene treatment in MRL+/+ mice. Toxicol Sci 57:345-352.
Griffin, JM; Gilbert, KM; Pumford, MR. (2000c) Inhibition of CYP2E1 reverses CD4+ T-cell alterations in
trichloroethylene-treated MRL+/+ mice. Toxicol Sci 54:384-389.
Groves, CE; Hayden, PJ; Lock, EA; et al. (1993) Differential cellular effects in the toxicity of haloalkene and
haloalkane cysteine conjugates to rabbit renal proximal tubules. J Biochem Toxicol 8:49-56.
Gu, ZW; Sele, B; Chmara, D; et al. (1981a) Effects of trichloroethylene and its metabolites on the rate of sister
chromatid exchange. In vivo and in vitro study on the human lymphocytes (author's translation). Sem Hop
57(37-38):1511-1512.
Gu, ZW; Sele, B; Jalbert, P; et al. (1981b) Induction of sister chromatid exchange by trichloroethylene and its
metabolites (author's translation) Toxicol EurRes 3(2):63-67.
Gualandi, G. (1987) Use of alpha- and beta-tubulin mutants for the study of spontaneous and induced chromosomal
mis-distribution in Aspergillus nidulans. MutatRes 178(1):33-41.
Guehl, D; Bezard, E; Dovero, S; et al. (1999) Trichloroethylene and parkinsonism: a human and experimental
observation. Eur J Neural 6(5):609-611.
Guengerich, FP; Crawford, WM; Watanabe, PG. (1979) Activation of vinyl chloride to covalently bound
metabolites: roles of 2-chloroethylene and 2-chloroacetaldehyde. Biochemistry 18:5177-5182.
Guengerich, FP; Shimada, T. (1991) Oxidation of toxic and carcinogenic chemicals by human cytochrome P-450
enzymes. Chem Res Toxicol 4(4):391-407.
Guengerich, FP. (2004) Metabolism of trichloroethylene and covalent binding of reaction products. In: Proceedings
of symposium on new scientific research related to the health effects of trichloroethylene, Feb. 26-27, 2004,
Washington, DC. Available online at http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=75934.
Guengerich, FP; Kim, DH; Iwasaki, M. (1991) Role of human cytochrome P-450 IIE1 in the oxidation of many low
molecular weight cancer suspects. Chem Res Toxicol 4(2): 168-179.
Gun, RT; Grygorcewicz, C; Nettelbeck, TJ. (1978) Choice reaction time in workers using trichloroethylene. Med J
Aust l(10):535-536.
Guo, H; Lee, SC; Chan, LY; et al. (2004) Risk assessment of exposure to volatile organic compounds in different
indoor environments. Environ Res 94(l):57-66.
Gupta, S. (2000) Hepatic polyploidy and liver growth control. Seminars Cancer Biol. 10:161-171.
Gustavsson, P; Jakobsson, R; Johansson, H; et al. (1998)Occupational exposure and squamous cell carcinoma of the
oral cavity, pharynx, larynx, and oesophagus: a case-control study in Sweden. Occup Environ Med 55:393-400.
Guyton, KZ; Barone, S, Jr; Brown, RC; et al. (2008) Mode of action frameworks: a critical analysis. J Toxicol
Environ Health B 11(1):16-31.
Guyton, KZ; Chiu, WA; Bateson, TF; et al. (2009) A Reexamination of the PPAR-a Activation Mode of Action as a
Basis for Assessing Human Cancer Risks of Environmental Contaminants. Environ Health Perspect.
doi:10.1289/ehp.0900758 available online at http://dx.doi.org/ [accessed 15 May 2009]
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-30 DRAFT—DO NOT CITE OR QUOTE
-------
Habib, GM; Barrios, R; Shi, ZZ; et al. (1996) Four distinct membrane-bound dipeptidase RNAs are differentially
expressed and show discordant regulation with gamma-glutamyl transpeptidase. J Biol Chem
271(27):16273-16280.
Habib, GM; Shi, ZZ; Cuevas, AA; et al. (1998) Leukotriene D4 and cystinyl-bis-glycine metabolism in membrane-
bound dipeptidase-deficient mice. Proc Natl Acad Sci USA 95(9):4859-4863.
Hack, CE; Chiu, WA; Zhao, QJ; et al.(2006) Bayesian population analysis of a harmonized physiologically based
pharmacokinetic model of trichloroethylene and its metabolites. Regul Toxicol Pharmacol 46:63-83.
Hakkola, J; Tanaka, E; Pelkonen, O. (1998b) Developmental expression of cytochrome P450 enzymes in human
liver. Pharmacol Toxicol 82(5):209-217.
Hakkola, J; Pasanen, M; Hukkanen, J; et al. (1996a) Expression of xenobiotic-metabolizing cytochrome P450 forms
in human full-term placenta. Biochem Pharmacol 51:403-411.
Hakkola, J; Pelkonen, O; Pasanen, M; et al. (1998a) Xenobiotic-metabolizing cytochrome P450 enzymes in the
human feto- placenta! unit: role in intrauterine toxicity. Crit Rev Toxicol 28:35-72.
Hakkola, J; Raunio, H; Purkunen, R; et al. (1996b) Detection of cytochrome P450 gene expression in human
placenta in first trimester of pregnancy. Biochem Pharmacol 52:379-383.
Halmes, NC; Perkins, EJ; McMillan, DC; et al. (1997) Detection of trichloroethylene-protein adducts in rat liver and
plasma. Toxicol Lett 92:187-194.
Hamada, T; Tanaka, H. (1995) Transfer of methyl chloroform, trichloroethylene and tetrachloroethylene to milk,
tissues and expired air following intraruminal or oral administration in lactating goats and milk-fed kids. Environ
Pollut87(3):313-8.
Hamdan, H; Stacey, NH. (1993) Mechanisms of trichloroethylene-induced elevation of individual serum bile acids: I.
Correlation of trichloroethylene concentrations to bile acids in rat serum. Toxicol Appl Pharmacol 121:291-295.
Hansch, C; Leo, A; Hoekman, D. (1995) Exploring QSAR - Hydrophobic, electronic, and steric constants.
Washington, DC: American Chemical Society.
Hansen, J; Raaschou-Nielsen, O; Christensen, JM; et al. (2001) Cancer incidence among Danish workers exposed to
trichloroethylene. J Occup Environ Med 43:133-139.
Hardell, L; Bengtsson, NO; Johsson, U; et al. (1984) Aetiological aspects o primary liver cancer with special regard
to alcohol, organic solvents and acute intermittent porphyria - an epidemiological investigation. Br J Cancer
50:389-297.
Hardell, L; Eriksson, M; Degerman, A. (1994) Exposure to phenoxyacetic acids, chlorophenols, or organic solvents
in relation to histopathology, stage, and anatomical localization of non-Hodgkin's lymphoma. Cancer Res
54:2386-2389.
Hardell, L; Eriksson, M; Lenner, P; et al. (1981) Malignant lymphoma and exposure to chemicals, especially organic
solvents, chlorophenols and phenoxy acids: a case-control study. Br J Cancer 43:169-176.
Hardin, BD; Bond, GP; Sikov, MR; et al. (1981) Testing of selected work place chemicals for teratogenic potential.
Scand J Work Environ Hlth 7(Suppl 45):66-75.
Hardin, BD; Kelman BJ; Brent RL. (2004) Trichloroethylene and cardiac malformations. Environ Health Perspect
112(11):A607-A608.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-31 DRAFT—DO NOT CITE OR QUOTE
-------
Hardin, BD; Kelman, BJ; Brent, RL. (2005) Trichloroethylene and dichloroethylene: a critical review of
teratogenicity. Birth Defects Res A Clin Mol Teratol 73:931-955.
Harrington, JM; Whitby, H; Gray, CN; et al. (1989) Renal disease and occupational exposure to organic solvents: a
case referent approach. Br J Ind Med 46:643-650.
Harrington-Brock, K; Doerr, CL; Moore, MM. (1998) Mutagenicity of three disinfection by-products: di- and
trichloroacetic acid and chloral hydrate in L5178Y/TK+/- (-) 3.7.2C mouse lymphoma cells. Mutat Res
413(3):265-276.
Harris, NL; Jaffe, ES; Diebold, J; et al. (2000) Lymphoma classification-from controversy to consensus: the
R.E.A.L. and WHO Classification of lymphoid neoplasms. Ann Oncol 1 l(Suppl 1):3-10.
Hasmall, SC; Roberts, RA. (2000) The nongenotoxic hepatocarcinogen diethylhexylphthalate and
methylclofenebrate induce DNA synthesis preferentially in octoploid rat hepatocytes. Toxicol Pathol
28(4):503-509.
Hassall, CD; Brendel, K; Gandolfi, AJ. (1983) Regulation of a S(trans-l,2-dichlorovinyl)-L-cysteine-induced renal
tubular toxicity by glutathione. J Appl Toxicol 3:321-325.
Hattis, D; Ginsberg, G; Sonawane, B; et al. (2003) Differences in pharmacokinetics between children and adults~II.
Children's variability in drug elimination half-lives and in some parameters needed for physiologically-based
pharmacokinetic modeling. Risk Anal 23(1): 117-142.
Haworth, S; Lawlor, T; Mortelmans, K; et al. (1983) Salmonella mutagenicity test results for 250 chemicals.
Environ Mutagen 5(Suppl 1): 1-142.
Healy, TEJ; Poole, TR; Hopper, A. (1982) Rat fetal development and maternal exposure to trichloroethylene 100
p.p.m. Br JAnaesth 54:337-340..
Hegi, ME; Fox, TR; Belinsky, SA; etal. (1993) Analysis of activated protooncogenes in B6C3F1 mouse liver
tumors induced by ciprofibrate, a potent peroxisome proliferator. Carcinogenesis 14(1): 145-149.
Heineman, EF; Cocco, P; Gomez, MR; et al. (1994) Occupational exposure to chlorinated aliphatic hydrocarbons
and risk of astrocytic brain cancer. Am J Ind Med 26:155-169.
Heinemann, K; Willich, SN; Heinemann, AJ; et al. (2000) Occupational exposure and liver cancer in women:
results of the Multicentre International Liver Tumour Study (MILTS). Occup Med 50:422-429.
Helliwell, PJ; Hutton, AM. (1950) Trichloroethylene anaesthesia. Anaesthesia 5(1):4-13.
Henderson, GN; Yan, Z; James, MO; et al. (1997) Kinetics and metabolism of chloral hydrate in children:
Identification of dichloroacetate as a metabolite. Biochem Biophys Res Commun 235(3):695-698.
Hendler, AH; Crow, WL. (1992) Preliminary results of the Chemical Manufacturers Association urban baseline
VOC measurement program. Proc Annu Meet Air Waste Manage Assoc, 85th Meeting, Kansas City, MS, 92-75.05,
94-08743-2-9.
Hennig, UG; Rudd, NL; Hoar, DI. (1988) Kinetochore immunofluorescence in micronuclei: a rapid method for the
in situ detection of aneuploidy and chromosome breakage in human fibroblasts. Mutat Res 203(6):405-414.
Henning, SM; Swendseid, ME. (1996) The role of folate, choline, and methionine in carcinogenesis induced by
methyl-deficient diets. Adv Exp Med Biol 399:143-155.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-32 DRAFT—DO NOT CITE OR QUOTE
-------
Henschler, D; Eder, E; Neudecker, et al. (1977) Carcinogenicity of trichloroethylene: fact or artifact? Arch Toxicol
37(3):233-236.
Henschler, D; Elsasser, H; Romen, et al. (1984) Carcinogenicity study of trichloroethylene, with and without
epoxide stabilizers, in mice. J Cancer Res Clin Oncol 107:149-156..,
Henschler, D; Romen, W; Elsasser, HM; et al. (1980) Carcinogenicity study of trichloroethylene by longterm
inhalation in three animal species. Arch Toxicol 43:237-248.
Henschler, D; Vamvakas, S; Lammert, M; et al. (1995) Increased incidence of renal cell tumors in a cohort of
cardboard workers exposed to trichloroethene. Arch Toxicol 69:291-299.
Henschler, D; Elsasser H; Romen, W; et al. (1984) Carcinogenicity study of trichloroethylene, with and without
epoxide stabilizers, in mice. J Cancer Res Clin Oncol 107:149-156.
Herbert, V; Gardner, A; Colman, N. (1980) Mutagenicity of dichloroacetate, an ingredient of some formulations of
pangamic acid (trade-named "vitamin B15"). Am J Clin Nutr 33(6): 1179-1182.
Herman, JG; Umar, A; Polyak, K; et al. (1998) Incidence and functional consequences of hMLHl promoter
hypermethylation in colorectal carcinoma. Proc Natl Acad Sci USA 95(12):6870-6785.
Hernberg, S; Kauppinen, T; Riala, R; et al. (1988) Increased risk for primary liver cancer among women exposed to
solvents. Scand J Work Environ Health 14:356-365.
Hernberg, S; Korkala, M-L; Asikainen, U; et al. (1984) Primary liver cancer and exposure to solvents. Int Arch
Occup Environ Health 54:147-153.
Herren-Freund, SL; Pereiira, MA; Khoury, MD; et al. (1987) The Carcinogenicity of trichloroethylene and its
metabolites trichloroacetic acid and dichloroacetic acid, in mouse liver. Toxicol Appl Pharmacol 90:183-189.
Hers, I; Zapf-Gilje, R; Li, L; et al. (2001) The use of indoor air measurements to evaluate intrusion of subsurface
VOC vapors into buildings. J Air Waste Manage Assoc 51:1318-1331.
Hill, AB. (1965) The environment and disease: association or causation? Proc R Soc Med 58:295-300.
Hinchman, CA; Ballatori, N. (1990) Glutathione-degrading capacities of liver and kidney in different species.
Biochem Pharmacol 40(5): 1131-1135.
Hines RN, McCarverDG. (2002) The ontogeny of human drug-metabolizing enzymes: phase I oxidative enzymes.
J Pharmacol Exp Ther 300(2):355-360.
Hirsch, AR; Rankin, KM; Panelli, PP. (1996) Trichloroethylene exposure and headache. Headache Q 7(2): 126-138.
Hissink, EM; Bogaards, JJP; Freidg, AP; et al. (2002) The use of in vitro metabolic parameters and physiologically
based pharmacokinetic (PBPK) modeling to explore the risk assessment of trichloroethylene. Environ Toxicol
Pharmacol 11:259-271.
Hobara, T; Kobayashi, H; Higashihara, E; et al. (1984) Acute effects of 1,1,1-trichloroethane, trichloroethylene, and
toluene on the hematologic parameters in dogs. Arch Environ Contam Toxicol 13:589-593.
Hodgen, GD; Sherins, RJ. (1973) Enzymes as markers of testicular growth and development in the rat.
Endocrinology 93(4):985-989.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-33 DRAFT—DO NOT CITE OR QUOTE
-------
Hoffman, S; Mishima, N; Krug, EL. (2004) An improved model for evaluating trichloroethylene and its metabolites
as cardiac-specific teratogens. In: Mohr, LC; Hoel, DG; Jollow, D; ed. Trichloroethylene - the scientific basis of
risk assessment. Charleston, SC: The Medical University of South Carolina Press; pp. 69-79.
Hoivik, DJ; Quails, CW; Mirabile, RC; et al. (2004) Fibrates induce hepatic peroxisome and mitochondrial
proliferation without evidence of cellular proliferation and oxidative stress in cynomolgus monkeys. Carcinogenesis
25(9): 1757-1769.
Hooper, NM; Hesp, RJ; Tieku, S. (1994) Metabolism of aspartame by human and pig intestinal microvillar
peptidases. Biochem J. 298(Pt 3):635-639.
Horesovsky, G; Ginsler, J; Everitt, et al. (1994) Increased susceptibility to in vitro transformation of cells.
Carcinogenesis 15:2183-2187.
Hotvath, AL; Getzen, FW; Maczynska, Z. (1999) IUPAC-NIST solubility data series 67. halogenated ethanes and
ethenes with water. J Phys Chern Ref Data 28:395-627.
Howard, PH; Boethling, RS; Jarvis, WF; et al. (1991) Handbook of environmental degradation rates. Chelsea, MI:
Lewis Publishers.
Hrelia, P; Maffei, F; Vigagni, F; et al. (1994) Interactive effects between trichloroethylene and pesticides at
metabolic and genetic level in mice. Environ Health Perspect 102(Suppl 9): 31-34.
HSDB (Hazardous Substances Data Bank). (2002) Trichloroethylene. National Library of Medicine. National
Institutes of Health. Washington, DC: U.S. Department of Health and Human Services.
Hu, C; Jiang, L; Geng, C; et al. (2008) Possible involvement of oxidative stress in trichloroethylene-induced
genotoxicity in human GepG2 cells. Mutat Res 652:88-94.
Hu, Y; Hakkola, J; Oscarson, M; et al. (1999) Structural and functional characterization of the 5'-flanking region of
the rat and human cytochrome P450 2E1 genes: identification of a polymorphic repeat in the human gene. Biochem
Biophys Res Commun 263:286-293.
Huang, HL; Li, LY; Chen, B; et al. (2002) New problems caused by occupational trichlorethylene exposure. Int J
Immunopath Pharm 15:30-32.
Hunter, ES; Rogers, EH; Schmid, et al. (1996) Comparative effects of haloacetic acids in whole embryo culture.
Teratology 54:57-64.
IARC (International Agency for Research on Cancer). (1995) Dry Cleaning, Some Chlorinated Solvents and Other
Industrial Chemicals. Summary of Data Reported and Evaluation. IARC Monographs on the Evaluation of
Carcinogenic Risks to Humans, Volume 63. International Agency for Research on Cancer, World Health
Organization, Lyon, France.
lavicoli, I; Marinaccio, A; Carelli, G. (2005) Effects of occupational trichloroethylene exposure on cytokine levels
in workers. J Occup Environ Med 47:453-457.
ICRP (The International Commission on Radiological Protection). (2002) Basic anatomical andphysiological data
for use in radiological protection: reference values. A report of age- and gender-related differences in the anatomical
and physiological characteristics of reference individuals.ICRP Publication 89. Ann ICRP 32(3-4):5-265.
Ikbal, M; Tastekin, A; Dogan, H; et al. (2004) The assessment of genotoxic effects in lymphocyte cultures of infants
treated with chloral hydrate. Mutat Res 564:159-164.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-34 DRAFT—DO NOT CITE OR QUOTE
-------
Ikeda, M; Otsuji, H; Imamura, et al. (1972) Urinary excretion of total trichloro-compounds, trichloroethanol, and
trichloroacetic acid as a measure of exposure to trichloroethylene and tetrachloroethylene. Br J Ind Med
29(3):328-333.
Ikeda, M. (1977) Metabolism of trichloroethylene and tetrachloroethylene in human subjects. Environ Health Persp
21:239-245.
Ikeda, M; Imamura, T. (1973) Biological half-life of trichloroethylene and tetrachloroethylene in human subjects.
Int Arch Arbeitsmed 31:209-224.
Ikeda, M; Ohtsuji, H; Kawai, H; et al. (1971) Excretion kinetics of urinary metabolites in a patient addicted to
trichloroethylene. Br J Ind Med 28(2):203-206.
ILSI (International Life Sciences Institute). (1992) Similarities and differences between children and adults:
implications for risk assessment. New York: ILSI Press.
Inoue, O; Seiji, K; Kawai, T; et al. (1989) Relationship between vapor exposure and urinary metabolite excretion
among workers exposed to trichloroethylene. Am J Ind Med 15(1): 103-10.
IOM (Institute of Medicine). (2003) Gulf War and Health, Vol 2. Insecticides and Solvents. Washington DC: The
National Academies Press.
IPCS (International Programme on Chemical Safety). (2005) Chemical-specific adjustment factors for interspecies
differences and human variability : guidance document for use of data in dose/concentration-response assessment.
Geneva: World Health Organization.
Isaacson, LG; Spohler, SA; Taylor, DH. (1990) Trichloroethylene affects learning and decreases myelin in the rat
hippocampus. Neurotoxicol Teratol 12(4):375-381.
Isaacson, LG; Taylor, DH. (1989) Maternal exposure to 1,1,2-trichloroethylene affects myelin in the hippocampal
formation of the developing rat. Brain Res 488(1-2):403-407.
Isacson, P; Bean, JA; Splinter, R; et al. (1985) Drinking water and cancer incidence in Iowa. III. Association of
cancer with indices of contamination. AmJEpidemiol 121:856-869.
Ito, Y; Yamanoshita, O; Asaeda, N; et al. (2007) Di(2-ethylhexyl)phthalate induced hepatic tumorigenesis through a
peroxisome proliferator-activated receptor a-independent pathway. J Occup Health 49:172-182..
Iyer, RA; Frink, EJ; Ebert, TJ; et al. (1998) Cysteine conjugate p-lyase-dependent metabolism of compound A (2-
[fluoromethoxy]-l,l,3,3,3-pentafluoro-l-propene) in human subjects anesthetized with sevoflurane and in rats given
compound A. Anesthesiology 88:611-618.
Jaffe, DR; Hassall, CD; Gandolfi, AJ; et al. (1985) Production of DNA single strand breaks in rabbit renal tissue
after exposure to 1,2-dichlorovinylcysteine. Toxicology 35(l):25-33.
Jaffe, DR; Gandolfi, AJ; Nagle, RB. (1984) Chronic toxicity of S-(trans-l,2-dichlorovinyl)-L-cysteine in mice. J
ApplToxicol4:315-319.
Jakobson, I; Holmberg, B; Ekner, A. (1986) Venous blood levels of inhaled trichloroethylene in female rats and
changes induced by interacting agents. Acta Pharmacol Toxicol (Copenh) 59:135-143.
Jakobson, I; Wahlberg, JE; Holmberg, B; et al. (1982) Uptake via the blood and elimination of 10 organic solvents
following epicutaneous exposure of anesthetized guinea pigs. Toxicol Appl Pharm 63:181-187.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
Jakubowski, M; Wieczorek, H. (1988) The effects of physical effort on pulmonary uptake of selected organic
compound vapours. Pol J Occup Med 1(1):62-71.
James, M.O., Cornett, R; Yan, Z; et al. (1997) Glutathione-dependent conversion to glyoxylate, a major pathway of
dichloroacetate biotransformation in hepatic cytosol from humans and rats, is reduced in dichloroacetate-treated rats.
Drug Metab Disp 25(11): 1223-1227.
Jansson, C; Johansson, AL; Bergdahl, IA; et al. (2006a) Occupational exposures and risk of esophageal and gastric
cardia cancers among male Swedish construction workers. Cancer Causes Control 16:755-764.
Jansson, C; Plate, N; Johansson, AL; et al. (2006b) Airborne occupational exposures and risk of oesophageal and
cardia adenocarcinoma. Occup Environ Med 63:107-112.
Jasinska, M. (1965) [General anesthesia with trichloroethylene in children with congenital developmental defects of
the jaw and face]. Czas Stomatol 18(8):935-40. Translated.
Jaspers, RMA; Muijser, H; Lammers, JH; et al. (1993) Mid-frequency hearing loss and reduction of acoustic startle
responding in rats following trichloroethylene exposure. Neurotoxicol Teratol 15(6):407-412.
Jemal, A; Siegel, R; Ward, E; et al. (2008) Cancer statistics, 2008. CA Cancer J Clin 58:71-96.
Jensen, OM; Knudsen, JB; McLaughlin, JK; et al. (1988) The Copenhagen case-control study of renal pelvis and
ureter cancer: role of smoking and occupational exposures. Int J Cancer 41:557-561.
Ji, B-T; Blair, A; Shu, X-O; et al. (2008) Occupation and breast cancer risk among Shanghai women in a population-
based cohort study. Am J Ind Med 51:100-110.
Ji J; Hemminki, K. (2005) Variation in the risk for liver and gallbladder cancers in socioeconomic and occupational
groups in Sweden with etiological implications. Int Arch Occup Environ Health 78:641-649.
Jia, M; Coats, B; Chadha, M; et al. (2006) Human kinetics of orally and intravenously administered low-dose 1,2-
13C-dichloroacetate. J Clin Pharmacol 46:1449-1459.
Johnson, AD; VanDemark WL; Gomes, WR; et al. (1967) The effect of antispermatogenic and hormone treatments
on testicular cholesterol. Fed Proc 26:645.
Johnson, PD. (2004) Trichloroethylene: Johnson et al. 's response. Environ Health Perspect 112(11):A608-A609.
Johnson, PD; Dawson, BV; Goldberg, SJ. (1998a) Cardiac teratogenicity of trichloroethylene metabolites. J Am
Coll Cardiol 72(2):540-545.
Johnson, PD; Dawson, BV; Goldberg, SJ. (1998b) A review: trichloroethylene metabolites: potential cardiac
teratogens. Environ Health Perspect 106(Suppl. 4):995-999.
Johnson, PD; Goldberg, SJ; Mays, MZ; et al. (2003) Threshold of trichloroethylene contamination in maternal
drinking waters affecting fetal heart development in the rat. Environ Health Perspect 111(3):289-292.
Johnson, PD; Goldberg, SJ; Mays, MZ; et al. (2005) Errata. Threshold of trichloroethylene contamination in
maternal drinking waters affecting fetal heart development in the rat. Environ Health Perspect 113(1):A18..
Johnsrud, EK; Koukouritaki, SB; Divakaran, K; et al. (2003) Human hepatic CYP2E1 expression during
development. J Pharmacol Exp Ther 307(1):402-407.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-36 DRAFT—DO NOT CITE OR QUOTE
-------
Jones, TW; Qin, C; Schaeffer, VH; et al. (1988) Immunohistochemical localization of glutamine transaminase K, a
rat kidney cysteine conjugate (3-lyase, and the relationship to the segment specificity of cysteine conjugate
nephrotoxicity. Mol Pharmacol 34:621-627.
Jorquera, R; Tanguay, RM. (2001) Fumarylacetoacetate, the metabolite accumulating in hereditary tyrosinemia,
activates the ERK pathway and induces mitotic abnormalities and genomic instability. Hum Mol Genet
10:1741-1752.
Juran, BD; Lazaridis, KN. (2006) Genomics and complex liver disease: challenges and Opportunities. Hepatology
44:1380-1390.
Kadry, AM; Turkall, RM; Skowronski, GA; et al. (1991) Soil adsorption alters kinetics and bioavailability of
trichloroethylene in orally exposed female rats. Toxicol Lett 58:337-346.
Kafer, E. (1986) Tests which distinguish induced crossing-over and aneuploidy from secondary segregation in
Aspergillus treated with chloral hydrate or gamma-rays. Mutat Res. 164(3): 145-166.
Kamijima, M; Hisanaga, N; Wang, H; et al. (2007) Occupational trichloroethylene exposure as a cause of
idiosyncratic generalized skin disorders and accompanying hepatitis similar to drug hypersensitivities. Int Arch
Occup Environ Health 80:357-370.
Kamijima, M; Wang, H; Huang, H; et al. (2008) Trichloroethylene Causes Generalized Hypersensitivity Skin
Disorders Complicated by Hepatitis. J Occup Health 50(4):328-338.
Kan, FW; Forkert, PG; Wade, MG. (2007) Trichloroethylene exposure elicits damage in epididymal epithelium and
spermatozoa in mice. Histol Histopathol 22(9):977-988.
Kaneko, T; Saegusa, M; Tasaka, K; et al. (2000) Immunotoxicity of trichloroethylene: a study with MRL-lpr/lpr
mice. JAppl Toxicol 20:471-475.
Kaneko, T ; Wang, P-Y ; Sato, A. (1994) Enzymes induced by ethanol differently affect the pharmacokinetics of
trichloroethylene and 1,1,1-trichloroethane. Occup Environ Med 51:113-119.
Kaneko, T; Saegusa, M; Tasaka, K; et al. (2000) Immunotoxicity of trichloroethylene: a study with MRL-lpr/lpr
mice. JAppl Toxicol 20:471-475.
Kanisawa, M; Suzuki, S. (1978) Induction of renal and hepatic tumors in mice by ochratoxin A, a mycotoxin. Gann.
69:599-600.
Kappas, A. (1989) On the mechanisms of induced aneuploidy in Aspergillus nidulans and validation of tests for
genomic mutations. Prog Clin Biol Res 318:377-384.
Kargalioglu, Y; McMillan, B J; Minear, RA; et al. (2002) Analysis of the cytotoxicity and mutagenicity of drinking
water disinfection by-products in Salmonella typhimurium. Teratog Carcinog Mutagen 22:113-128.
Karunanayake, CP; McDuffie, HH; Bosnian, JA; et al. (2008) Occupational exposures and non-Hodgkin's
lymphoma: Canadian case-control study. Environ Health 7:44.
Kato-Weinstein, J; Stauber, AJ; Orner, GA; et al. (2001) Differential effects of dihalogenated and trihalogenated
acetates in the liver of B6C3 Fl mice. Journal of Applied Toxicology 21:81-89.
Kato-Weinstein, J; Lingohr, MK; Orner, GA; et al. (1998) Effects of dichloroacetate on glycogen metabolism in
B6C3Flmice. Toxicology 130:141-154.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-37 DRAFT—DO NOT CITE OR QUOTE
-------
Kauffmann, BM; White, KL, Jr.; Sanders, VM; et al. (1982) Humoral and cell-mediated immune status in mice
exposed to chloral hydrate. Environ Health Perspect 44:147-151.
Kautiainen, A; Vogel, JS; Turteltaub, KW. (1997) Dose-dependent binding of trichloroethylene to hepatic DNA and
protein at low doses in mice. Chem-Biol Interact 106:109-121.
Kawamoto, T; Hobara, T; Nakamura, K; et al. (1988) Induction of cytochrome P450, cytochrome b-5, NADPH
cytochrome c reductase and change in cytochrome P-450 isozymes with long-term trichloroethylene treatment.
Toxicology 53:239-249.
Kawamoto, TT; Hobara, T; Aoki, et al. (1988) Metabolism of chloral hydrate in the anoxic perfused liver. Toxicol
Lett 40:225-231.
Keil, DE; Peden-Adams, MM; Wallace, S; et al. (2009) Assessment of trichloroethylene (TCE) exposure in murine
strains genetically-prone and non-prone to develop autoimmune disease. J Environ Sci Health A Environ Sci Eng
Toxic 44:443-453.
Keller, DA; Heck, HD. (1988) Mechanistic studies on chloral toxicity: Relationship to trichloroethylene
carcinogenesis. Toxicol Lett 42:183-191.
Kelsh, MA; Weihgart, M; Mandel, J; et al. (2005) A meta-analysis of epidemiology studies of occupational TCE
exposure and selected cancers. Presented to National Research Committee on Health Risks of Trichloroethylene.
June 9, 2005.
Kenck, C; Wilhelm, M; Bugert, P; et al. (1996) Mutation of the VHL gene is associated exclusively with the
development of non-papillary renal cell carcinomas. J Pathol 179(2):157-161.
Keshava, N; Caldwell, JC. (2006) Key issues in the role of peroxisome proliferator-activated receptor agonism and
cell signaling in trichloroethylene toxicity. Environ Health Perspect 114:1464-1470.
Ketcha, MM; Stevens, DK; Warren, DA; et al. (1996) Conversion of trichloroacetic acid to dichloroacetic acid in
biological samples. J Anal Toxicol 20:236-241.
Keys, DA; Bruckner, JV; Muralidhara, S; et al. (2003) Tissue dosimetry expansion and cross-validation of rat and
mouse physiologically based pharmacokinetic models for trichloroethylene. Toxicol Sci 76:35-50.
Keys, DA; Schultz, IR; Mahle, DA; et al. (2004) A quantitative description of suicide inhibition of dichloroacetic
acid in rats and mice. Toxicol Sci 82(2):381-393.
Kezic, S; Monster, AC; Kriise, J; et al. (2000) Skin absorption of some vaporous solvents in volunteers. Int Arch
OccEnvHea 73:415-422.
Kezic, S; Monster, AC; van de Gevel, IA; et al. (2001) Dermal absorption of neat liquid solvents on brief exposures
in volunteers. Am Ind Hyg AssocJ62(l):12-18.
Khan, MF; Kaphalia, BS; Prabhakar, BS; et al. (1995) Trichloroethene-induced autoimmune response in female
MRL +/+ mice. Toxicol Appl Pharmacol 134:155-160.
Khan, MF; Wu, X; Ansari, GA. (2001) Anti-malondialdehyde antibodies in MRL+/+ mice treated with
trichloroethene and dichloroacetyl chloride: possible role of lipid peroxidation in autoimmunity. Toxicol Appl
Pharmacol 170:88-92.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
Kilburn, KH; Warshaw, RH. (1992) Prevalence of symptoms of systemic lupus erythematosus (SLE) and of
fluorescent antinuclear antibodies associated with chronic exposure to trichloroethylene and other chemicals in well
water. Environ Res 57:1-9.
Kilburn, KH; Warshaw, RH. (1993) Neurobehavioral testing of subjects exposed residentially to groundwater
contaminated from an aluminum die-casting plant and local referents. J Toxicol Environ Health 39:483-496.
Kilburn, KH; Warshaw, R; Thorton, JC. (1987) Formaldehyde impairs memory, equilibrium, and dexterity in
histology technicians: effects which persist for days after exposure. Arch Environ Health 42:117-120.
Kilburn, KH; Warshaw, R. (1995a) Neurotoxic effects from residential exposure to chemicals from an oil
reprocessing facility and superfund site. Neurotoxicol Teratol 17:89-102.
Kilburn, KH; Warshaw, RH. (1995b) Hydrogen sulfide and reduced-sulfur gases adversely affect
neurophysiological functions. Toxicol Ind Health 11:185-197.
Kilburn, KH. (2002a) Is neurotoxicity associated with environmental trichloroethylene (TCE)? Arch Environ
Health 57(2): 113-120.
Kilburn, KH. (2002b) Do duration of exposure, proximity to electronic manufacturing plants, and involvement in a
lawsuit affect chlorinated solvent toxicity? Arch Environ Health 57(2): 121-126.
Kilburn, KH; Thornton, JC. (1996) Prediction equations for simple and visual two-choice reactions times in
environmental neurotoxicology. Arch Environ Health 51(6):439-444.
Kilburn, KH; Warshaw, RH. (1992) Prospective study of neurobehavioral effects of formaldehyde and solvents on
histology technicians. Environ Res 58:134-146.
Kilburn, KH; Warshaw, RH. (1993) Effects on neurobehavioral performance of chronic exposure to chemically
contaminated well water. Toxicol Ind Health 9(3):391-404.
Kim, CF; Jackson, EL; Woolfenden, AE; et al. (2005) Identification of bronchioalveolar stem cells in normal lung
and lung cancer. Cell 121(6):823-35.
Kim, D; Ghanayem, BI. (2006) Comparative metabolism and disposition of trichloroethylene in Cyp2el-/-and wild-
type mice. Drug Metab Dispos 34(12):2020-2027.
Kim, HJ; Odend'hal, S; Bruckner, JV. (1990) Effect of oral dosing vehicles on the acute hepatotoxicity of carbon
tetrachloride in rats. Toxicol Appl Pharmacol 102:34-49.
Kim, HJ; Bruckner, JV; Dallas, CE; et al. (1990) Effect of dosing vehicles on the pharmacokinetics of orally
administered carbon tetrachloride in rats. Toxicol Appl Pharm 102:50-60.
Kim, S; Kim, D; Pollack, GM; et al. (2009) Pharmacokinetic analysis of trichloroethylene metabolism in male
B6C3F1 mice: Formation and disposition of trichloroacetic acid, dichloroacetic acid, S-(l,2-
dichlorovinyl)glutathione and S-(l,2-dichlorovinyl)-L-cysteine. Toxicol Appl Pharmacol 238(l):90-99.
Kim, SZ; Kupke, KG; lerardi-Curto, L; et al. (2000) Hepatocellular carcinoma despite long-term survival in chronic
tyrosinaemia I. J Inherit Metab Dis 23:791-804.
Kimmerle, G; Eben, A. (1973) Metabolism, excretion and toxicology of trichloroethylene after inhalation. 2.
Experimental human exposure. Arch Toxikol 30(2): 127-38.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-39 DRAFT—DO NOT CITE OR QUOTE
-------
Kimmerle, G; Eben, A. (1973a) Metabolism, excretion and toxicology of trichloroethylene after inhalation. I.
Experimental exposure on rats. Arch Toxikol 30:115-126.
King, GS; Smialek, JE; Troutman, WG. (1985) Sudden death in adolescents resulting from the inhalation of
typewriter correction fluid. J AmMed Assoc 253(11): 1604-1606.
Kishi, R; Harabuchi, I; Ikeda, T; et al. (1993) Acute effects of trichloroethylene on blood concentrations and
performance decrements in rats and their relevance to humans. Br J Ind Med 50(5):470-480.
Kjellstrand, P; Holmquist, B; Aim, P; et al. (1983a) Trichloroethylene: further studies of the effects on body and
organ weights and plasma butyrylcholinesterase activity in mice. Acta Pharmacol Toxicol (Copenh) 53(5):375-384.
Kjellstrand, P; Holmquist, B; Mandahl, N; et al. (1983b) Effects of continuous trichloroethylene inhalation on
different strains of mice. Acta Pharmacol Toxicol (Copenh) 53(5):369-374.
Kjellstrand, P; Kanje, M; Mansson, L; etal. (1981) Trichloroethylene: effects on body and organ weights in mice,
rats and gerbils. Toxicology 21:105-115.
Kjellstrand, P; Holmquist, B; Aim, P; et al. (1983a) Trichloroethylene: further studies of the effects on body and
organ weights and plasma butyrylcholinesterase activity in mice. Acta Pharmacol Toxicol (Copenh) 53:375-384.
Kjellstrand, P; Kanje, M; Bjerkemo, M. (1987) Regeneration of the sciatic nerve in mice and rats exposed to
trichloroethylene. Toxicol Lett 38(1-2):187-191.
Kjellstrand, P; Lanke, J; Bjerkemo, M; et al. (1980) Irreversible effects of trichloroethylene exposure on the central
nervous system. Scand J Work Environ Health 6(l):40-47.
Klaunig, JE; Ruch, RJ; Lin, EL. (1989) Effects of trichloroethylene and its metabolites on rodent hepatocyte
intercellular communication. Toxicol Appl Pharmacol 99(3):454-465.
Klaunig, JE; Babich, MA; Baetcke, KP; et al. (2003) PPARalpha agonist-induced rodent tumors: modes of action
and human relevance. Crit Rev Toxicol 33:655-780.
Kligerman, AD; Bryant, MF; Doerr, CL; et al. (1994) Inhalation studies of the genotoxicity of trichloroethylene to
rodents. MutatRes 322(2):87-96.
Koch, R; Schlegehnilch, R; Wolf, HU. (1988) Genetic effects of chlorinated ethylenes in the yeast Saccharomyces
cerevisiae. MutatRes 206:209-216.
Koizumi, A. (1989) Potential of physiologically based pharmacokinetics to amalgamate kinetic data of
trichloroethylene and tetrachloroethyolene obtained in rats and man. Br J Ind Med 46(4):239-249.
Konietzko, H; Elster, I; Sayer, H; et al. (1975) [EEG changes under standardized trichloroethylene load (author's
transl)]. Int Arch Occup Environ Hlth 35(3-4): 257-264. Originally published: Zentralnervose schaden durch
trichlorathylen. Staub, Reinhaltung Der Luft 35(6):240-241.
Kopylev, L; Chen, C; White, P. (2007) Towards quantitative uncertainty assessment for cancer risks: central
estimates and probability distributions of risk in dose-response modeling. Regul Toxicol Pharmacol 49(3):203-207.
Korrapati, MC; Lock, EA; Mehendale, HM. (2005) Molecular mechanisms of enhanced renal cell division in
protection against S-l,2-dichlorovinyl-L-cysteine-induced acute renal failure and death. Am J Physiol Renal
Physiol 289:F175-F185.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-40 DRAFT—DO NOT CITE OR QUOTE
-------
Kozak, EM; Tate, SS. (1982) Glutathione-degrading enzymes of microvillus membranes. J Biol Chem
257(11):6322-6327.
Krasowski, MD; Harrison, ML. (2000) The actions of ether, alcohol and alkane general anaesthetics on GABAA and
glycine receptors and the effects of TM2 and TM3 mutations. Br J Pharmacol 129(4):731-743.
Kraupp-Grasl, B; Huber, W; Putz, B; et al. (1990) Tumor promotion by peroxisome proliferator nafenopin involving
specific subtype of altered foci in rat liver. Cancer Res 50:3701-3706.
Krause, RJ; Lash, LH; Elfarra, AA. (2003) Human kidney flavin-containing monooxygenases and their potential
roles in cysteine s-conjugate metabolism and nephrotoxicity. J Pharmacol Exp Ther 304(1): 185-191.
Krejci, ME; Ridgewell, RE; Koechel, DA. (1991) Acute effects of the D-isomer of S-(l,2-dichlorovinyl)cysteine on
renal function and ultrastructure in the pentobarbital-anesthetized dog: site-specific toxicity involving the SI and S2
cells of the proximal tubule. Toxicology 69:151-164.
Krishnadasan, A; Kennedy, N; Zhao, Y; et al. (2007) Nested case-control study of occupational chemical exposures
and prostate cancer in aerospace and radiation workers. Am J Ind Med 50:383-390.
Krishnadasan, A; Kennedy, N; Zhao, Y; et al. (2008) Nested case-control study of occupational physical activity and
prostate cancer among workers using a job exposure matrix. Cancer Causes Control 19(1): 107-114.
Kroneld, R. (1989) Volatile pollutants in the environment and human tissues. B Environ Contam Tox 42:873-877.
Kubo, Y; Mitani, H; Hino, O. (1994) Allelic loss at the predisposing gene locus in spontaneous and chemically
induced renal cell carcinomas in the Eker rat. Cancer Res 54(10):2633-2635.
Kulig, BM. (1987) The effects of chronic trichloroethylene exposure on neurobehavioral functioning in the rat.
Neurotoxicol Teratol 9(2):171-178.
Kumar, M; Kumar, R; Hissar, SS; et al. (2007) Risk factors analysis for hepatocellular carcinoma in patients with
and without cirrhosis: A case-control study of 213 hepatocellular carcinoma patients from India. J Gastroent
Hepatol 22:1104-1 111.
Kumar, P; Prasad, AK; Dutta, KK. (2000b) Steroidogenic alterations in testes and sera of rats exposed to
trichloroethylene (TCE)by inhalation. Hum Exp Toxicol 19:117-121.
Kumar, P; Prasad, AK; Maji, BK; et al. (2001) Hepatotoxic alterations induced by inhalation of trichloroethylene
(TCE) in rats. Biomed Environ Sci 14:325-332.
Kumar, P; Prasad, AK; Mani, U; et al. (2001) Trichloroethylene induced testicular toxicity in rats exposed by
inhalation. Hum Exp Toxicol 20:585-589.
Kumar, P; Prasad, AK; Saxena, DK; et al. (2000a) Fertility and general reproduction studies in trichloroethylene
exposed rats. Indian J Occup Health 43(3): 117-125.
Kung, H-C; Hoyert, DL; Xu, J; et al. (2008) Deaths: Final Data for 2005. National Vital Statistics Reports, U.S.
Department of Health and Human Services, Centers for Disease Control and Prevention. Vol 56, Issue 10. Available
online from ; http://www.cdc.gov/nchs/data/nvsr/nvsr56/nvsr56 10.pdf Accessed May 29, 2008]
Kurasawa, K. (1988) Selective damage of pulmonary nonciliated bronchiolar epithelial (Clara) cells by
trichloroethylene in rats. Jpn J Ind Health 30:121-133 (abstract only, Japanese).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-41 DRAFT—DO NOT CITE OR QUOTE
-------
Kvam, BM; Romundstad, PR; Boffestta, P; et al. (2005) Cancer in the Norwegian printing industry. Scand J Work
Environ Health 31:3 6-43.
Kylin, B; Axell, K; Samuel, HE; et al. (1967) Effect of inhaled trichloroethylene on the ens as measured by
optokinetic nystagmus. Arch Environ Hlth 15:48-52.
Lacey, JV, Jr.; Garabrant, DH; Laing, TJ; et al. (1999) Petroleum distillate solvents as risk factors for
undifferentiated connective tissue disease (UCTD). Am J Epidemiol 149:761-770.
Lagakos, SW; Wessen, B; Zelen, M. (1986) Analysis of contaminated well water and health effect in Woburn,
Massachusetts. J Am Stat Assoc 81:583-596.
Laham, S. (1970) Studies on placenta! transfer: Trichloroethylene. Ind Med 39:46-49.
Lamb, JC; Hentz, KL. (2006) Toxicological review of male reproductive effects and trichloroethylene exposure:
Assessing the relevance to human male reproductive health. Reprod Toxicol 22:557-563.
Land, PC; Owen, EL; Linde, HW. (1981) Morphologic changes in mouse spermatozoa after exposure to inhalational
anesthetics during early spermatogenesis. Anesthesiology 54:53-56.
Landrigan, PJ; Kominsky, JR; Stein, GF; et al. (1987) Common-source community and industrial exposure to
trichloroethylene. Arch Environ Health 42(6):327-332.
Lane, SE; Watts, RA; Bentham, G; et al. (2003) Are environmental factors important in primary systemic vasculitis?
A case-control study. Arthritis Rheum 48:814-823.
Lapare, S; Tardif, R; Brodeur, J. (1995) Effect of various exposure scenarios on the biological monitoring of
organicsolvents in alveolar air. II. 1,1,1-Trichloroethane and trichloroethylene. Int Arch Occup Environ Health
67:375-394.
Larsen, GL; Stevens, JL. (1986) Cysteine conjugate beta-lyase in the gastrointestinal bacterium Eubacterium
limosum. MolPharmacol 29:97-103.
Larson, JL; Bull, RJ. (1992) Metabolism of lipoperoxidative activity of trichloroacetate and dichloroacetate in rats
and mice. Toxicol Appl Pharmacol 115(2):268-277.
Larson, JL; Bull, RJ. (1989) Effect of ethanol on the metabolism of trichloroethylene. J Toxicol Environ Health
28(4):395-406.
Larson, JL; Bull, RJ. (1992a) Species differences in the metabolism of trichloroethylene to the carcinogenic
metabolites trichloroacetate and dichloroacetate. Toxicol Appl Pharm 115(2):278-285.
Lash, LH; Anders, MW. (1986) Cytotoxicity of S-(l,2-dichlorovinyl)glutathione and S-(l,2-dichlorovinyl)-L-
cysteine in isolated rat kidney cells. JBiol Chem 261:13076-13081.
Lash, TL; Green, LC. (1993) Blink reflex measurement of effects of trichloroethylene. Muscle Nerve 16:217-218.
Lash, LH; Elfarra, AA; Anders, MW. (1986) Renal cysteine conjugate (3-lyase. Bioactivation of nephrotoxic
cysteine S-conjugates in mitochondria! outer membrane. J Biol Chem 261(13):5930-5935.
Lash, LH; Jones, DP; Anders, MW. (1988) Glutathione homeostasis and glutathione S-conjugate toxicity in kidney.
Rev Biochem Toxicol 9:29-67.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-42 DRAFT—DO NOT CITE OR QUOTE
-------
Lash, LH; Nelson, RM; Van Dyke, RA; et al. (1990a) Purification and characterization of human kidney cytosolic
cysteine conjugate (3-lyase activity. DrugMetab Disp 18(l):50-54.
Lash, LH; Elfarra, AA; Rakiewicz-Nemeth, D; et al. (1990b) Bioactivation mechanism of cytotoxic homocysteine S-
conjugates. Arch Biochem Biophys 276(2):322-330.
Lash, LH; Sausen, PJ; Duescher, RJ; Cooley, AJ; Elfarra, AA. (1994) Roles of cysteine conjugate beta-lyase and S-
oxidase in nephrotoxicity: studies with S-(l,2-dichlorovinyl)-L-cysteine and S-(l,2-dichlorovinyl)-L-cysteine
sulfoxide. J Pharmacol Exp Ther 269(l):374-383.
Lash, LH; Xu, Y; Elfarra, AA; et al. (1995) Glutathione-dependent metabolism of trichloroethylene in isolated liver
and kidney cells of rats and its role in mitochondria! and cellular toxicity. DrugMetab Dispos 23:846-853.
Lash, LH; Qian, W; Putt, DA; et al. (1998) Glutathione conjugation of trichloroethylene in rats and mice: sex-,
species-, and tissue-dependent differences. Drug Metab Dispos 26(1): 12-19.
Lash, L H; Visarius, TM; Sail, JM; et al. (1998) Cellular and subcellular heterogeneity of glutathione metabolism
and transport in rat kidney cells. Toxicology 130:1-15.
Lash, LH; Lipscomb, JC; Putt, DA; et al. (1999a) Glutathione conjugation of trichloroethylene in human liver and
kidney: kinetics and individual variation. Drug Metab Dispos 27(3):351-359.
Lash, LH; Putt, DA; Brashear, WT; et al. (1999b) Identification of S-(l,2-dichlorovinyl)glutathione in the blood of
human volunteers exposed to trichloroethylene. J Toxicol Environ Health A 56:1-21.
Lash, LH; Fisher, JW; Lipscomb, JC; et al. (2000a) Metabolism of trichloroethylene. Environ Health Perspect
108(Suppl. 2): 177-200.
Lash, LH; Parker, JC; Scott, CS. (2000b) Modes of action of trichloroethylene for kidney tumorigenesis. Environ
Health Perspect 108(Suppl. 2):225-240.
Lash, LH; Qian, W; Putt, DA; et al. (2001) Renal and hepatic toxicity of trichloroethylene and its glutathione-
derived metabolites in rats and mice: sex-, species-, and tissue-dependent differences. J Pharmacol Exp Ther
297:155-164.
Lash, LH; Putt, DA; Hueni, SE; et al. (2003) Roles of necrosis, apoptosis, and mitochondria! dysfunction in S-(l,2-
dichlorovinyl)-L-cysteine sulfoxide-induced cytotoxicity in primary cultures of human renal proximal tubular cells.
J Pharmacol Exp Ther 305(3): 1163-1172.
Lash, LH; Putt, DA; Parker, JC. (2006) Metabolism and tissue distribution of orally administered trichloroethylene
in male and female rats: identification of glutathione- and cytochrome P-450-derived metabolites in liver, kidney,
blood, and urine. J Toxicol Env Health A 69(13):1285-1309.
Lash, LH; Putt, DA; Huang, P; et al. (2007) Modulation of hepatic and renal metabolism and toxicity of
trichloroethylene and perchloroethylene by alterations in status of cytochrome P450 and glutathione. Toxicology
235(l-2):ll-26.
Laslo-Baker, D; Barrera, M; Knittel-Keren, D; et al. (2004) Child neurodevelopmental outcome and maternal
occupational exposure to solvents. Arch Pediatr Adolesc Med 158(10):956-61.
Latendresse, JR; Pereira, MA. (1997) Dissimilar characteristics of N-methyl-N-nitrosourea-initiated foci and tumors
promoted by dichloroacetic acid or trichloroacetic acid in the liver of female B6C3F1 mice. Toxicol Pathol
25(5):433-40.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-43 DRAFT—DO NOT CITE OR QUOTE
-------
Lau, SS; Kleiner, HE; Monks, TJ. (1995) Metabolism as a determinant of sepecies susceptibility to 2,3,5-
(triglutatMon-S-YL)hydroquinone-mediated nephrotoxicity. The role of TV-acetylation and TV-deacetylation. Drug
Metab Disp 23(10): 1136-1142.
Laughter, AR; Dunn, CS; Swanson, CL; et al. (2004) Role of the peroxisome proliferator-activated receptor alpha
(PPARalpha) in responses to trichloroethylene and metabolites, trichloroacetate and dichloroacetate in mouse liver.
Toxicology 203:83-98.
Laureno, R. (1988) Trichloroethylene poisoning mimicking multiple sclerosis. Can J Neurol Sci 15:354.
Laureno, R. (1993) Blink reflex measurement of effects of trichloroethylene. Muscle Nerve 16:217.
LaVecchia, C; Negri, E; Franceschi, S; et al. (1994) A case-control study of diabetes mellitus and cancer risk. Br J
Cancer 70:950-953.
Lawrence, T; Hagemann, T; Balkwill, F. (2007) Sex, cytokines, and cancer. Science 317:51-52.
Le Mesurier, SM; Lykke, AWJ; Stewart, BW. (1980) Reduced yield of pulmonary surfactant: Patterns of response
following administration of chemicals to rats by inhalation. Toxicol Lett 5:89-93.
Le Mesurier, SM; Stewart, BW; O'Connell, PJ; et al. (1979) Pulmonary responses to atmospheric pollutants. II.
Effect of petrol vapour inhalation on secretion of pulmonary surfactant. Pathology 11:81-87.
Leakey, JEA; Seng, JE; Allaben, WT. (2003b) Body weight considerations in the B6C3F1 mouse and the use of
dietary control to standardize background tumor incidence in chronic bioassays. Toxicol Appl Pharmacol
193:237-265.
Leakey, JEA ; Seng, JE ; Lantendresse, JR ; et al. (2003a) Dietary controlled carcinogenicity study of chloral
hydrate in male B6C3F1 mice. Toxicol Appl Pharmacol 193:266-280.
Leavitt, SA; DeAngelo, AB; George, MH; et al (1997) Assessment of the mutagenicity of dichloroacetic acid in lacl
transgenic B6C3F1 mouse liver. Carcinogenesis 18(11):2101-2106.
Lee, JH; Chung, CW; Ma, YC; et al. (2003) Increased mortality odds ratio of male liver cancer in a community
contaminated by chlorinated hydrocarbons in groundwater. Occup Environ Med 60:364-369.
Lee, LJ; Chan, CC; Chung, CW; et al. (2002) Health risk assessment on residents exposed to chlorinated
hydrocarbons contaminated in groundwater of a hazardous waste site. J Toxicol Environ Health A 65(3-4):219-35.
Lee, KM; Bruckner, JV; Muralidhara, S; et al. (1996) Characterization of presystemic elimination of
trichloroethylene and its nonlinear kinetics in rats. Toxicol Appl Pharmacol 139:262-271.
Lee, KM; Muralidhara, S; Schnellmann, RG; Bruckner, JV. (2000a) Contribution of direct solvent injury to the
dose-dependent kinetics of trichloroethylene: portal vein administration to rats. Toxicol Appl Pharmacol 164:46-54.
Lee, KM; Muralidhara, S; White, CA; et al. (2000b) Mechanisms of the dose-dependent kinetics of trichloroethylene:
oral bolus dosing of rats. Toxicol Appl Pharmacol 164:55-64.
Lehmann, I; Rehwagen, M; Diez, U; et al. (2001) Enhanced in vivo IgE production and T cell polarization toward
the type 2 phenotype in association with indoor exposure to VOC: results of the LARS study. Int J Hyg Environ
Health 204(4):211-221.
Lehmann, I; Thoelke, A; Rehwagen, M; et al. (2002) The influence of maternal exposure to volatile organic
compounds on the cytokine secretion profile of neonatal T cells. Environ Toxicol 17(3):203-210.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-44 DRAFT—DO NOT CITE OR QUOTE
-------
Leigh JP. 1996. Occupations, cigarette smoking, and lung cancer in the epidemiological follow-up to the
NHANES I and the California occupational mortality study. Bull NY Acad Med 73:370-397.
Leigh, JP; Jiang, W. (1993) Liver cirrhosis deaths within occupations and industries in the California occupational
mortality study. Addiction 88:767-779.
Leighton, DT, Jr; Calo, JM. (1981) Distribution coefficients of chlorinated hydrocarbons in dilute air-water systems
for groundwater contamination applications. J Chem Eng Data 26:382-385.
Lemke, LE; McQueen, CA. (1995) Acetylation and its role in the mutagenicity of the antihypertensive agent
hydralazine. Drug Metab Dispos 23:559-565.
Leopardi, P; Zijno, A; Bassani, B; et al. (1993) In vivo studies on chemically induced aneuploidy in mouse somatic
and germinal cells. MutatRes 287(1): 119-130.
Leuschner, J; Beuscher, N. (1998) Studies on the mutagenic and carcinogenic potential of chloral hydrate.
Arzneimittelforschung48(10):961-968.
Leuschner, J; Leuschner, F. (1991) Evaluation of the mutagenicity of chloral hydrate in vitro and in vivo.
Arzneimittelforschung41(10):1101-1103.
Lewis, R. (2001) Hawley's Condensed Chemical Dictionary, 14th Edition. New York, NY: John Wiley and Sons.
Li, H; Dai, Y; Huang, H; et al. (2007) HLA-B*1301 as a biomarker for genetic susceptibility to hypersensitivity
dermatitis induced by trichloroethylene among workers in China. Environ Health Perspect 115(11): 1553-1556.
Li, T; Schultz, I; Keys, DA;et al. (2008) Quantitative evaluation of dichloroacetic acid kinetics in human—a
physiologically based pharmacokinetic modeling investigation. Toxicology 245:35-48.
Liang, JC; Brinkley, BR. (1985) Chemical probes and possible targets for the induction of aneuploidy. Basic Life
Sci 36:491-505
Liang, JC; Pacchierotti, F. (1988) Cytogenetic investigation of chemically-induced aneuploidy in mouse
spermatocytes. MutatRes201(2):325-35.
Liangpunsakul, S; Kolwankar, D; Pinto, A; et al. N. (2005) Activity of CYP2E1 and CYP3A enzymes in adults with
moderate alcohol consumption: a comparison with nonalcoholics. Hepatology 41(5): 1144-1150.
Lieber, CS. (2004) The discovery of the microsomal ethanol oxidizing system and its physiologic and pathologic
role. Drug Metab Rev 36(3-4):511-529.
Ligocki, MP; Leuenberger, C; Pankow, JF. (1985) Trace organic compounds in rain-II. Gas scavenging of neutral
organic compounds. Atmos Environ 19:1609-1617.
Lilly, PD; Simmons, JE; Pegram, RA. (1994) Dose-dependent vehicle differences in the acute toxicity of
bromodichloromethane. Fundam Appl Toxicol 23:132-140.
Lindbohm, ML; Sallmen, M; Kyyronen, P; et al. (2009) Risk of liver cancer and exposure to organic solvents and
gasoline vapors among Finnish workers. Int J Cancer EPUB DOI10.1002/ijc.24309
Lindbohm, ML; Taskinen, H; Sallmen, M; et al. (1990) Spontaneous abortions among women exposed to organic
solvents. Am JIndMed 17(4):449-463.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-45 DRAFT—DO NOT CITE OR QUOTE
-------
Lingohr, MK; Thrall, BD; Bull, RJ. (2001) Effects of dichloroacetate (DCA) on serum insulin levels and insulin-
controlled signaling proteins in livers of male B6C3F1 mice. Toxicol Sci 59:178-184.
Lipscomb, JC; Mahle, DA; Brashear, WT; et al. (1995) Dichloroacetic acid: Metabolism in cytosol. Drug Metab
Disp23(ll):1202-1205.
Lipscomb, JC; Mahle, DA; Brashear, WT; et al. (1996) A species comparison of chloral hydrate metabolism in
blood and liver. Biochem Biophys Res Commun 227: 340-350.
Lipscomb, JC; Garrett, CM; Snawder, JE. (1997) Cytochrome P450-dependent metabolism of trichloroethylene:
interindividual differences in humans. Toxicol Appl Pharmacol 142(2):311-318.
Lipscomb, JC; Fisher, JW; Confer, PD; et al. (1998a) In vitro to in vivo extrapolation for trichloroethylene
metabolism in humans. Toxicol Appl Pharmacol 152:376-387.
Lipscomb, JC; Garrett, CM; Snawder, JE. (1998b) Use of kinetic and mechanistic data in species extrapolation of
bioactivation: Cytochrome P-450 dependent trichloroethylene metabolism at occupationally relevant concentrations.
JOccup Health 40:110-117.
Lipscomb, JC; Teuschler, LK; Swartout, J; et al. (2003) The impact of Cytochrome P450 2El-dependent metabolic
variance on a risk-relevant pharmacokinetic outcome in humans. Risk Anal 23(6):1221-1238.
Little, JC; Daisey, JM; Nazaroff, WW. (1992) Transport of subsurface contaminants into buildings, an exposure
pathway for volatile organics. Environ Sci Technol 26:2058-2066.
Liu, E; Jelinek, J; Pastore, YD; et al. (2003) Discrimination of polycythemias and thrombocytoses by novel, simple,
accurate clonality assays and comparison with PRV-1 expression and BFU-E response to erythropoietin. Blood
101:3294-3301.
Liu, E; Percy, MJ; Amos, CI; et al. (2004) The worldwide distribution of the VHL 598C>T mutation indicates a
single founding event. Blood 103(5):1937-1940.
Liu, Y; Bartlett, MG; White, CA; et al. (2009) Presystemic elimination of trichloroethylene in rats following
environmentally-relevant oral exposures. Drug Metab Dispos (In Press).
Liu, YT; Jin, C; Chen, Z; et al. (1988) Increased subjective symptom prevalence among workers exposed to
trichloroethylene at sub-OEL levels. Tohoku J Exp Med 155(2): 183-195.
Lock, EA; Reed, CJ. (2006) Trichloroethylene: Mechanisms of Renal Toxicity and Renal Cancer and Relevance to
Risk Assessment. Toxicol Sci 91:313-331.
Lock, EA; Reed, CJ; McMillan, JM; et al. (2007) Lack of formic acid production in rat hepatocytes and human renal
proximal tubule cells exposed to chloral hydrate or trichloroacetic acid. Toxicology 230:234-243.
Lockey, JE; Kelly, CR; Cannon, GW; et al. (1987) Progressive systemic sclerosis associated with exposure to
trichloroethylene. J Occup Med 29:493-496.
Loeber, CP; Hendrix, MJC; de Pinos, SD; et al. (1988) Trichloroethylene: a cardiac teratogen in developing chick
embryos. PediatrRes24(6):740-744.
Loeber, CP; Runyan, RB. (1990) A comparison of fibronectin, laminin, and galactosyltransferase adhesion
mechanisms during embryonic cardiac mesenchymal cell migration in vitro. Dev Biol 140:401-412.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-46 DRAFT—DO NOT CITE OR QUOTE
-------
Lopreato, GF; Phelan, R; Borghese, CM; et al. (2003) Inhaled drags of abuse enhance serotonin-3 receptor function.
Drag Alcohol Depend 70(1): 11-5.
Lorente. C; Cordier, S; Bergeret, A; et al. (2000) Maternal occupational risk factors for oral clefts. Occupational
Exposure and Congenital Malformation Working Group. Scand J Work Environ Health 26(2): 137-145.
Lowengart, RA; Peters, JM; Cicioni, C; et al. (1987) Childhood leukemia and parents' occupational and home
exposures. J Natl Cancer Inst 79(l):39-46.
Luebke, RW; Chen, DH; Dietert, R; et al. (2006) The comparative immunotoxicity of five selected compounds
following developmental or adult exposure. J Toxicol Environ Health B 9:1-26.
Lumpkin, MH; Bruckner, JV; Campbell, JL; et al. (2003) Plasma binding of trichloroacetic acid in mice, rats, and
humans under cancer bioassay and environmental exposure conditions. Drag Metab Dispos 31(10): 1203-1207.
Lundberg, I; Alfredsson, L; Plato, N; et al. (1994) Occupation, occupational exposure to chemicals and
rheumatological disease. A register based cohort study. Scand J Rheumatol 23:305-310.
Lutz, WK; Gay lor, DW; Conolly, RB; et al. (2005) Nonlinearity and thresholds in dose-response relationships for
carcinogenicity due to sampling variation, logarithmic dose scaling, or small differences in individual susceptibility.
Toxicol Appl Pharmacol 207(Suppl. 2):565-569.
Lybarger, JA; Lichtveld, MY; Amler, RW. (1999) Biomedical testing of the kidney for persons exposed to
hazardous substances in the environment. Renal Failure 21:263-274.
Lynch, AM; Parry, JM. (1993) The cytochalasin-B micronucleus/kinetochore assay in vitro: studies with 10
suspected aneugens. Mutat Res 287(l):71-86.
Lynge, E; Anttila, A; Hemminki, K. (1997) Organic solvents and cancer. Cancer Cause Control 8:406-419.
MacFarlane, M; Foster, JR; Gibson, GG; et al. (1989) Cysteine conjugate beta-lyase of rat kidney cytosol:
characterization, immunocytochemical localization, and correlation with hexachlorobutadiene nephrotoxicity.
Toxicol Appl Pharm 98(2): 185-197.
Mackay, JM; Fox, V; Griffiths, K; et al. (1995) Trichloroacetic acid: investigation into the mechanism of
chromosomal damage in the in vitro human lymphocyte cytogenetic assay and the mouse bone marrow
micronucleus test. Carcinogenesis 16(5):1127-1133.
MADPH (Massachusetts Department of Public Health). (1997) Woburn childhood leukemia follow-up study.
Volume 1: Analyses. MADPH, Boston, MA.
Magnusson, B; Kligman, AM. (1969) The identification of contact allergens by animal assay. The guinea pig
maximization test. J Invest Dermatol 52:268-276.
Mahle, DA; Gearhart, JM; Grigsby, CC; et al. (2007) Age-dependent partition coefficients for a mixture of volatile
organic solvents in Sprague-Dawley rats and humans. J Toxicol Env Health A 70:1745-1751.
Mahr, AD; Neogi, T; Merkel, PA. (2006) Epidemiology of Wegener's granulomatosis: lessons from descriptive
studies and analyses of genetic and environmental risk determinants. Clin Exp Rheumatol 24:S82-S91.
Maihes, JB; Aardema, MJ; Marchetti, F. (1993) Investigation of aneuploidy induction in mouse oocytes following
exposure to vinblastine-sulfate, pyrimethamine, diethylstilbestrol diphosphate, or chloral hydrate. Environ Mol
Mutagen 22(2): 107-114.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-47 DRAFT—DO NOT CITE OR QUOTE
-------
Maitre, A; Hours, M; Bonneterre, V; et al. (2004) Systemic sclerosis and occupational risk factors: role of solvents
and cleaning products. JRheumatol 31:2395-2401.
Malherbe, P; Alberati-Giani, D; Kohler, C; et al. (1995) Identification of a mitochondria! form of kynurenine
aminotransferase/glutamine transaminase K from rat brain. FEES Lett 367:141-144.
Mallin, K. (1990) Investigation of a bladder cancer cluster in Northwestern Illinois. Am J Epidemiol
132(Suppl):S96-S106.
Mally, A; Walker, CL; Everitt, JI; et al. (2006) Analysis of renal cell transformation following exposure to
trichloroethene in vivo and its metabolite S-(dichlorovinyl)-L-cysteine in vitro. Toxicology. 224(1-2): 108-118.
Maloney, EK; Waxman, DJ. (1999) trans-Activation of PPARalpha and PPARgamma by structurally diverse
environmental chemicals. Toxicol Appl Pharmacol 161:209-218.
Maltoni, C; Lefemine, G; Cotti, G; et al. (1988) Long-term carcinogenicity bioassays on trichloroethylene
administered by inhalation to Sprague-Dawley rats and Swiss andB6C3Fl mice. In: Maltoni; C, Selikoff, IJ; eds.
Living in a chemical world. Annals of the New York Academy of Sciences, Volume 534. New York, NY: New
York Academy of Sciences..
Maltoni, C; Lefemine, G; Cotti, G. (1986) Experimental research on trichloroethylene carcinogenesis. In: Maltoni, C;
Mehlman MA, eds. Archives of research on industrial carcinogenesis, Vol. 5. Princeton, NJ: Princeton Scientific
Publishing; pp 316-342.
Mandel, JH; Kelsh, MA; Mink, PJ; et al. (2006) Occupational trichloroethylene exposure and non-Hodgkin's
lymphoma: a meta-analysis and review. Occup Environ Med 63:597-607.
Mandel, JS; McLaughlin, JK; Schlehofer, B; et al. (1995) Internal renal-cell cancer study. IV. Occupation. Int J
Cancer 61:601-605.
Mandel, JS. (2001) Renal cell cancer correlated with occupational exposure to trichloroethylene. J Cancer Res Clin
Oncol 127:265-268.
Mannetje, A; Dryson, E; Walls, C; et al. (2008) High risk occupations for non-Hodgkin's lymphoma in New Zealand:
Case-control study. Occup Environ Med 65:354-363.
Manson, JM; Murphy, M; Richdale, N; et al. (1984) Effects of oral exposure to trichloroethylene on female
reproductive function. Toxicology 32:229-242.
Markwald, R; Eisenberg, C; Eisenberg, L; et al. (1996) Epithelial-mesenchymal transformations in early avian heart
development. Acta Anatomica 156:173-186.
Markwald, RR; Runyan, RB; Kitten, GT; et al. (1984) Use of collagen gel cultures to study heart development:
proteoglycan and glycoprotein interactions during the formation of endocardial cushion tissue. In: The role of
extracellular matrix in development. New York Alan R. Liss, Inc.; pp 323-350.
Maronpot, RR; Fox, T; Malarkey, DE; et al. (1995) Mutations in the ras proto-oncogene clues to etiology and
molecular pathogenesis of mouse liver tumors. Toxicology 101:125-156.
Marrazzini, A; Betti, C; Bernacchi, F; et al. (1994) Micronucleus test and metaphase analyses in mice exposed to
known and suspected spindle poisons. Mutagenesis 9(6): 505-515
Marshall, EK; Owens, AH. (1955) Absorption, excretion and metabolic fate of chloral hydrate and trichloroethanol.
Bull Johns Hopkins Hosp 95:1-18.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-48 DRAFT—DO NOT CITE OR QUOTE
-------
Marsman, DS; Cattley, RC; Conway, JG; et al. (1988) Relationship of hepatic peroxisome proliferation and
replicative DNA synthesis to the hepatocarcinogenicity of the peroxisome proliferators di(2-ethylhexyl)phthalate
and [4-chloro-6-(2,3-xylidino)-2-pyrimidinylthio]acetic acid (Wy-14,643) in rats. Cancer Res 48:6739-6744.
Mather, GG; Exon, JH; Koller, LD. (1990) Subchronic 90 day toxicity of dichloroacetic and trichloroacetic acid in
rats. Toxicology 64:71-80.
Mattioli, S; Truffelli, D; Baldasseroni, A; et al. (2002) Occupational risk factors for renal cell cancer: a case-control
study in northern Italy. J Occup Environ Med 44:1028-1036.
Mazzullo, M; Bartoli, S; Bonora, B; et al. (1992) In vivo and in vitro interaction of trichloroethylene with
macromolecules from various organs of rat and mouse. Res Commun Chem Path 76(2): 192-208.
McCarthy, TB; Jones, RD. (1983) Industrial gassing poisonings due to trichlorethylene, perchlorethylene, and 1-1-1
trichloroethane, 1961-80. Br JIndMed40(4):450-455.
McCarver, DG; Byun, R; Hines, RN; et al. (1998) A genetic polymorphism in the regulatory sequences of human
CYP2E1: association with increased chlorzoxazone hydroxylation in the presence of obesity and ethanol intake.
Toxicol Appl Pharmacol 152(1):276-281.
McCarver, DG; Hines, RN. (2002) The ontogeny of human drug-metabolizing enzymes: phase II conjugation
enzymes and regulatory mechanisms. J Pharmacol Exp Ther 300(2):361-366.
McConnell, G; Ferguson, DM; Pearson, CR. (1975) Chlorinated hydrocarbons and the environment. Endeavour
34:13-18.
McCredie, M; Stewart, JH. (1993) Risk factors for kidney cancer in New South Wales. IV. Occupation. Br J Ind
Med 50:349-354.
McCunney, RJ. (1988) Diverse manifestations of trichloroethylene. Br J Ind Med 45:122-126.
McDaniel, TV; Martin, PA; Ross, N; et al. (2004) Effects of chlorinated solvents on four species of North American
amphibians. Arch Environ Contam Toxicol 47:101-109.
McGregor, DB; Reynolds, DM; Zeiger, E. (1989) Conditions affecting the mutagenicity of trichloroethylene in
Salmonella. Environ Mol Mutagen 13:197-202.
McKinney, PA; Alexander, FE; Cartwright, RA; et al. (1991) Parental occupations of children with leukaemia in
west Cumbria, north Humberside, and Gateshead. Brit J Med 302:681-687.
McLaughlin, JK; Blot, WJ. (1997) A critical review of epidemiology studies of trichloroethylene and
perchloroethylene and risk of renal-cell cancer. Int Arch Occup Environ Health 70:222-31.
McLaughlin, JK; Lipworth, L; Tarone, RE. (2006) Epidemiologic aspects of renal cell carcinoma. Semin Oncol
33:527-33.
Melchiorri, C; Chieco, P; Zedda, Al; et al. (1993) Ploidy and nuclearity of rat hepatocytes after compensatory
regeneration or mitogen-induced liver growth. Carcinogenesis 14(9): 1825-1830.
Mellemgaard, A; Engholm, G; McLaughlin, JK; et al. (1994) Occupational risk factors for renal-cell carcinoma in
Denmark. Scand J Work Environ Health 20:160-165.
Melnick, RL; Jameson, CW; Goehl, TJ; etal. (1987) Application of microencapsulation for toxicology studies. II.
Toxicity of microencapsulated trichloroethylene in Fischer 344 rats. Fundam Appl Toxicol 8:432-442.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-49 DRAFT—DO NOT CITE OR QUOTE
-------
Mensing, T; Welge, P; Voss, B; et al. (2002) Renal toxicity after chronic inhalation exposure of rats to
trichloroethylene. ToxicolLett 128:243-247.
Merdink, JL; Robison, LM; Stevens, DK; et al. (2008) Kinetics of chloral hydrate and its metabolites in male human
volunteers. Toxicology 245:130-40.
Merdink, JL; Stenner, RD; Stevens, DK; et al. (1999) Effect of enterohepatic circulation on the pharmacokinetics of
chloral hydrate and its metabolites in F344 rats. J Toxicol Environ Health A 57:357-368.
Merdink, JL; Bull, RJ; Schultz, IR. (2000) Trapping and identification of the dichloroacetate radical from the
reductive dehalogenation of trichloroacetate by mouse and rat liver microsomes. Free Radic Biol Med 29(2): 125-
130.
Merdink, JL; Gonzalez-Leon, A; Bull, RJ; et al. (1998) The extent of dichloroacetate formation from
trichloroethylene, chloral hydrate, trichloroacetate, and trichloroethanol inB6C3Fl mice. Toxicol Sci 45(1):33-41.
Merrick, BA; Robinson, M; Condi, LW. (1989) Differing hepatotoxicity and lethality after subacute
trichloroethylene exposure in aqueous or corn oil gavage vehicles in B6C3F1 mice. JAppl Toxicol 9(1): 15-21.
Messing, K; Punnett, L; Bond, M; et al. (2003) Be the fairest of them all: challenges and recommendations for the
treatment of gender in occupational health research. Am J Ind Med 43:618-629.
Mester, B; Nieters, A; Deeg, E; et al. (2006) Occupation and malignant lymphoma: a population based case control
study in Germany. Occup Environ Med 63:17-26.
Mhiri, C; Choyakh, F; Ben, HM; et al. (2004) Trigeminal somatosensory evoked potentials in trichloroethylene-
exposed workers. Neurosciences 9(2): 102-107.
Michalik, L; Desvergne, B; Dreyer, C; et al. (2002) PPAR expression and function during vertebrate development.
Int J Develop Biol 46:105-114.
Michel, C; Desdouets, C; Slaoui, M; et al. (2007) Diethylnitrosamine initiation does not alter clofibric acid-induced
hepatocarcinogenesis in the rat. Toxicol Sci 99(l):58-69.
Mikol, YB; Hoover, KL; Creasia, D; et al. (1983) Hepatocarcinogenesis in rats fed methyl-deficient, amino acid-
defined diets. Carcinogenesis 4(12):1619-29.
Miligi, L; Costantini, AS; Benvenuti, A; et al. (2006) Occupational exposure to solvents and the risk of lymphomas.
Epidemiology 17:552-561.
Miller, BM; Adler, I-D. (1992) Aneuploidy induction in mouse spermatocytes. Mutagenesis 7:69-76
Miller, RT; Shah, RS; Cattley, RC; et al. (1996) The peroxisome proliferator WY-14,643 and methylclofenapate
induce hepatocellular ploidy alterations and ploidy-specific DNA synthesis in F344 rats. Toxicol Appl Pharmacol
138:317-323.
Miller, RE; Guengerich, FP. (1982) Oxidation of trichloroethylene by liver microsomal cytochrome P-450:
Evidence for chlorine migration in a transition state not involving trichloroethylene oxide. Biochemistry
21:1090-1097.
Miller, RE; Guengerich, FP. (1983) Metabolism of trichloroethylene in isolated hepatocytes, microsomes, and
reconstituted enzyme systems containing cytochrome P-450. Cancer Res 43(3): 1145-1152.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-50 DRAFT—DO NOT CITE OR QUOTE
-------
Mills, NC; Means, AR. (1970) Sorbitol dehydrogenase of rat testis changes activity during development after
hypophysectomy and following gonadotropic hormone administration. Endocrinology 91(1): 147-156.
Mirsalis, JC; Tyson, CK; Steinmetz, KL; et al. (1989) Measurement of unscheduled DNA synthesis and S-phase
synthesis in rodent hepatocytes following in vivo treatment: testing of 24 compounds. Environ Mol Mutagen
14:155-164.
Mishima, N; Hoffman, S; Hill, EG; et al. (2006) Chick embryos exposed to trichloroethylene in an ex ovo culture
model show selective defects in early endocardia! cushion tissue formation. Birth Defects Res A Clin Mol Teratol
76:517-527.
Mitchell, ABS; Parsons-Smith, BG. (1969) Trichloroethylene neuropathy. BritMed. J 1:422-423.
Mjaatvedt, CH; Krug, EL; Markwald, RR. (1991) An antiserum (ESI) against a paniculate form of extracellular
matrix blocks the transition of cardiac endothelium into mesenchyme in culture. DevBiol 145:219-230.
Mjaatvedt, CH; Lepera, RC; Markwald, RR. (1987) Myocardial specificity for initiating endothelial-mesenchymal
cell transition in embryonic chick heart correlates with a paniculate distribution of fibronectin. Dev Biol 119:59-67.
Moghaddam, AP; Abbas, R; Fisher, JW; et al. (1997) The role of mouse intestinal microflora in the metabolism of
trichloroethylene, an in vivo study. Human Exp Toxicol 16:629-635.
Moghaddam, AP; Abbas, R; Fisher, JW; et al. (1996) Formation of dichloroacetic acid by rat and mouse gut
microflora, an in vitro study. Biochem Biophys Res Comm 228:639-645.
Monster, AC; Boersma, G; Duba, WC. (1979) Kinetics of trichloroethylene in repeated exposure of volunteers. Int
Arch Occup Environ Health 42:283-292.
Monster, AC. (1979) Difference in uptake, elimination, and metabolism in exposure to trichloroethylene, 1,1,1-
trichloroethane and tetrachloroethylene. Int Arch Occup Environ Health 42(3-4):311-317.
Monster, AC; Boersma, G; Duba, WC. (1976) Pharmacokinetics of trichloroethylene in volunteers, influence of
workload and exposure concentration. Int Arch Occup Environ Health 38(2):87-102.
Moore, LE; Wilson, RT; Campleman, SL. (2005) Lifestyle factors, exposures, genetic susceptibility, and renal cell
cancer risk: a review. Cancer Invest 23:240-255.
Moore, MM; Harrington-Brock, K. (2000) Mutagenicity of trichloroethylene and its metabolites: implications for
the risk assessment of trichloroethylene. Environ Health Perspect 108(Suppl. 2):215-223.
Morgan, JW; Cassady, RE. (2002) Community cancer assessment in response to long-time exposure to perchlorate
and trichloroethylene in drinking water. J Occup Environ Med 44(7):616-621.
Morgan, RW; Kelsh, MA; Zhao, K; et al. (1998) Mortality of aerospace workers exposed to trichloroethylene.
Epidemiology 9:424-431. .
Morgan, RW; Kelsh, MA; Zhao, K; et al. (2000) Mortality of aerospace workers exposed to trichloroethylene
[erratum]. Epidemiology 11:360.
Morimura, K; Cheung, C; Ward, JM; et al. (2006) Differential susceptibility of mice humanized for peroxisome
proliferator-activated receptor {alpha} to Wy-14,643 induced liver tumorigenesis. Carcinogenesis. 27:1074-1080.
Mortelmans, K; Haworth, S; Lawlor, T; et al. (1986) Salmonella mutagenicity tests: II. Results from the testing of
270 chemicals. Environ Mutagen 8(Suppl. 7): 1-119.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-51 DRAFT—DO NOT CITE OR QUOTE
-------
Morton, LD; Youssef, AF; Lloyd, E; et al. (2002) Evaluation of carcinogenic responses in the Eker rat following
short-term exposure to selected nephrotoxins and carcinogens. Toxicol Pathol 30:559-564.
Morton, LM; Turner, JJ; Cerhan, JR; et al. (2007) Proposed classification of lymphoid neoplasms for epidemiologic
research from the Pathology Working Group of the International Lymphoma Epidemiology Consortium
(InterLymph). Blood 110:695-708.
Morton, LM; Wang, SS; Devesa, SS; et al. (2006) Lymphoma incidence patterns by WHO subtype in the United
States, 1992-2001. Blood 107:265-276.
Moser, VC; Cheek, BM; MacPhail, RC. (1995) A multidisciplinary approach to lexicological screening III:
Neurobehavioral toxicity. J Toxicol Environ Health 45(2): 173-210.
Moser, VC; MacPhail, RC; Gennings, C. (2003) Neurobehavioral evaluations of mixtures of trichloroethylene,
heptachlor, and di(2-ethylhexyl)phthlate in a full-factorial design. Toxicology 188(2-3): 125-137.
Moslen, MT; Reynolds, ES; Boor, PJ; et al. (1977) Trichloroethylene-induced deactivation of cytochrome P-450 and
loss of liver glutathione in vivo. Res Commun Chem Pathol Pharmacol 16(1): 109-120.
Mueller, WF; Coulston, F; Korte, F. (1982) Comparative metabolism of 14C TCE in chimpanzees, baboons, and
rhesus monkeys. Chemosphere 11:215-218.
Muijser, H; Lammers, J; Kulig, BM. (2000) Effects of exposure to trichloroethylene and noise on hearing in rats.
Noise Health 6:57-66.
Mtiller, G; Spassowski, M; Henschler, D. (1975) Metabolism of trichloroethylene in man. III. Interaction of
trichloroethylene and ethanol. Arch Toxicol 33(3): 173-89.
Nagaya, T; Ishikawa, N; Hata, H. (1989) Sister chromatid exchanges in lymphocytes of workers exposed to
trichloroethylene. MutatRes 222:279-282.
Nagaya, T; Ishikawa, N; Hata, H; et al. (1993) Subclinical and reversible hepatic effects of occupational exposure to
trichloroethylene. Int Arch Occup Environ Health 64:561-563.
Nagaya, T; Ishikawa, N; Hata, H. (1989) Urinary total protein and p2-Microglobulin in workers exposed to
trichloroethylene. Environ Res 50:86-92.
Nagaya, T; Ishikawa, N; Hata, H. (1990) No change in serum dopamine-beta-hydroxylase activity in workers
exposed to trichloroethylene. Toxicol Lett 54(2/3):221-227.
Nakai, JS; Stathopulos, PB; Campbell, GL; et al. (1999) Penetration of chloroform, trichloroethylene, and
tetrachloroethylene through human skin. J Toxicol Environ Health A. 58(3): 157-170.
Nakajima, T; Aoyama, T. (2000) Polymorphism of drug-metabolizing enzymes in relation to individual
susceptibility to industrial chemicals. Ind Health 38(2): 143-152.
Nakajima, T; Kamijo, Y; Usuda, N; et al. (2000) Sex-dependent regulation of hepatic peroxisome proliferation in
mice by trichloroethylene via peroxisome proliferator-activated receptor alpha (PPARalpha). Carcinogenesis
21(4):677-682.
Nakajima, T; Koyama, Y; Sato, A. (1982) Dietary modification of metabolism and toxicity of chemical substances--
with special reference to carbohydrate. Biochem Pharmacol 31(6):1005-1011.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-52 DRAFT—DO NOT CITE OR QUOTE
-------
Nakajima, T; Okino, T; Okuyama, S; et al. (1988) Ethanol-induced enhancement of trichloroethylene metabolism
and hepatotoxicity: difference from the effect of phenobarbital. Toxicol Appl Pharmacol 94(2):227-237.
Nakajima, T; Wang, RS; Elovaara, E; et al. (1992b) A comparative study on the contribution of cytochrome P450
isozymes to metabolism of benzene, toluene and trichloroethylene in rat liver. Biochem Pharmacol 43(2):251-257.
Nakajima, T; Wang, RS; Elovaara, E; et al. (1993) Cytochrome P450-related differences between rats and mice in
the metabolism of benzene, toluene and trichloroethylene in liver microsomes. Biochem Pharmacol
45(5):1079-1085.
Nakajima, T; Wang, RS; Katakura, Y; et al. (1992a) Sex-, age- and pregnancy-induced changes in the metabolism of
toluene and trichloroethylene in rat liver in relation to the regulation of cytochrome P450IIE1 and P450IIC11
content. J Pharmacol Exp Ther 261(3):869-874.
Nakajima, T; Wang, RS; Murayama, N; et al. (1990) Three forms of trichloroethylene-metabolizing enzymes in rat
liver induced by ethanol, phenobarbital, and 3-methylcholanthrene. Toxicol Appl Pharmacol 102(3):546-552.
Nakajima, T; Yamanoshita, O; Kamijima, M; et al. (2003) Generalized skin reactions in relation to trichloroethylene
exposure: a review from the viewpoint of drug-metabolizing enzymes. J Occup Health 45(1):8-14.
Nakajima, T; Okino, T; Okuyama, S; et al. (1988) Ethanol-induced enhancement of trichloroethylene metabolism
and hepatotoxicity: difference from the effect of phenobarbital. Toxicol Appl Pharmacol 94:227-237.
Nakajima, T; Wang, RS; Elovaara, E; et al. (1993) Cytochrome P450-related differences between rats and mice in
the metabolism of benzene, toluene and trichloroethylene in liver microsomes. Biochem Pharmacol
45(5):1079-1085.
Nakajima, Y; Mironov, V; Yamagishi, T; et al. (1997) Expression of smooth muscle alpha-actin in mesenchymal
cells during formation of avian endocardia! cushion tissue: a role for transforming growth factor (33. Dev Dyn
209:296-309.
Nakanishi, S; Shiohara, E; Tsukada, M; et al. (1978) Acetaldehyde level in the blood and liver aldehyde
dehydrogenase activities in trichloroethylene-treated rats. Arch Toxicol 41(3):207-214.
Narotsky, MG; Kavlock, RJ. (1995) A multidisciplinary approach to lexicological screening: II. Developmental
toxicity. J Toxicol Environ Health 45:145-171.
Narotsky, MG; Weller, EA; Chinchilli, VM; et al (1995) Nonadditive developmental toxicity in mixtures of
trichloroethylene, di(2-ethylhexyl) phthalate, and heptachlor in a 5 x 5 x 5 design. Fundam Appl Toxicol
27:203-216.
Natarajan, AT; Duivenvoorden, WC; Meijers, M; et al. (1993) Induction of mitotic aneuploidy using Chinese
hamster primary embryonic cells. Test results of 10 chemicals. Mutat Res 287(l):47-56.
Naugler, WE; Sakurai, T; Kim, S; et al. (2007) Gender disparity in liver cancer due to sex differences in MyD88-
dependent IL-6 production. Science 317: 121-124.
NCHS (National Center for Health Statistics). (2007) National Vital Statistics Reports, vol. 55, no. 19; August 21,
2007, Table 3. National Center for Health Statistics, Hyattsville, MD.
NCI (National Cancer Institute). (1976) Carcinogenesis bioassay of trichloroethylene. Division of Cancer Cause
and Prevention, National Cancer Institute, U.S. Department of Health, Education, and Welfare, DHEW Publication
No. (NIH) 76-802, Technical Report Series No. 2, 218 pages; NCI-CG-TR-2; NTIS PB254122. Available online at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-53 DRAFT—DO NOT CITE OR QUOTE
-------
Neghab, M; Qu, S; Bai, CL; et al. (1997) Raised concentration of serum bile acids following occupational exposure
to halogenated solvents, l,l,2-tricUoro-l,2,2-trifluoroethane and trichloroethylene. Int Arch Occup Environ Health
70:187-194.
Neghab, M; Stacey, NH. (1997) Toluene-induced elevation of serum bile acids: relationship to bile acid transport. J
Toxicol Environ Health 52:249-268.
Nelson, MA; Bull, RJ. (1988) Induction of strand breaks in DNA by trichloroethylene and metabolites in rat and
mouse liver in vivo. Toxicol Appl Pharmacol 94:45-54.
Nelson, MA; Lansing, AJ; Sanches, AM; et al. (1989) Dichloroacetic acid and trichloroacetic acid-induced DNA
strand breaks are independent of peroxisome proliferation. Toxicology 58:239-248.
Nelson, MA; Sanchez, IM; Bull, RJ; et al. (1990) Increased expression of c-myc and c-H-ras in dichloroacetate and
trichloroacetate-induced liver tumors in B6C3F1 mice. Toxicology 64:47-57.
Nelson, GM; Swank, AE; Brooks, LR; et al. (2001) Metabolism, microflora effects, and genotoxicity in haloacetic
acid-treated cultures of rat cecal microbiota. Toxicol Sci 60(2):232-241.
Nelson, JL; Zenick, H. (1986) The effect of trichloroethylene on male sexual behavior: possible opioid role.
Neurotoxicol Teratol 8(5):441-445.
Nelson, MA; Bull, RJ. (1988) Induction of strand breaks in DNA by trichloroethylene and metabolites in rat and
mouse liver in vivo. Toxicol Appl Pharmacol 94:45-54.
Nesslany, F; Marzin, D. (1999) A micromethod for the in vitro micronucleus assay. Mutagenesis 14(4):403-410.
NYSDEC (New York State Department of Environmental Conservation). (2006a)
http://www.dec.state.nv.us/website/der/projects/712006.html
NYSDEC (New York State Department of Environmental Conservation). (2006b)
http://www.dec.state.nv.us/website/der/projects/endicott/endicottsampling.htm
Ni, YC; Kadlubar, FF; Fu, PP. (1995) Formation of malondialdehyde-modified 2'-deoxyguanosinyl adduct from
metabolism of chloral hydrate by mouse liver microsomes. Biochem Biophys Res Commun216(3):1110-1117.
Ni, YC; Wong, TY; Kadlubar, FF; et al (1994) Hepatic metabolism of chloral hydrate to free radical(s) and
induction of lipid peroxidation. Biochem Biophys Res Commun. 204(2):937-943.
Ni, Y-C; Wong, T-Y; Lloyd, RV; et al. (1996) Mouse liver microsomal metabolism of chloral hydrate,
trichloroacetic acid, and trichloroethanol leading to induction of lipid peroxidation via a free radical mechanism.
Drug Metab Disp 24(1):81-90.
NIAMS (National Institute of Arthritis and Musculoskeletal Diseases). (2007) The future directions of lupus
research. Available online at http://www.niams.nih.gov/About_Us/Mission_and_Purpose/lupus_plan.pdf
Nickerson, ML; Jaeger, E; Shi, Y; et al. (2008) Improved identification of von Hippel-Lindau gene alterations in
clear cell renal tumors. Clin Cancer Res 14(15):4726-4734.
Niederlehner, BR; Cairns, J, Jr.; Smith, EP. (1998) Modeling acute and chronic toxicity of nonpolar narcotic
chemicals and mixtures to Ceriodaphnia dubia. Ecotoxicol Environ Saf 39:136-146.
Nietert, PJ; Sutherland, SE; Silver, RM; et al. (1998) Is occupational organic solvent exposure a risk factor for
scleroderma? Arthritis Rheum 41 (6): 1111 -8. Erratum in: Arthritis Rheum 41 (8): 1512.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-54 DRAFT—DO NOT CITE OR QUOTE
-------
Niklasson, M; Tham, R; Larsby, B; et al. (1993) Effects of toluene, styrene, trichloroethylene and trichloroethane on
the vestibule and opto-oculo motor system in rats. Neurotoxicol Teratol 15(5):327-334.
Noland-Gerbec, EA; Pfohl, RJ; Taylor, DH; et al. (1986) 2-Deoxyglucose uptake in the developing rat brain upon
pre- and postnatal exposure to trichloroethylene. Neurotoxicology 7(3): 157-164.
Nomiyama, K; Nomiyama, H. (1971) Metabolism of trichloroethylene in human. Sex difference in urinary excretion
of trichloroacetic acid and trichloroethanol. Int Arch Arbeitsmed 28(l):37-48.
Nomiyama, H; Nomiyama, K. (1979) Pathway and rate of metabolism of TCE in rats and rabbits. Ind Health 17:29
37.
Nordstrom, M; Hardell, L; Magnuson, A; et al. (1998) Occupational exposures, animal exposure and smoking as
risk factors for hairy cell leukaemia evaluated in a case-control study. Br J Cancer 77:2048-2052.
NRC (National Research Council). (1986) Drinking water and health, Vol. 6. Washington DC: National Academy
Press.
NRC (National Research Council). (1987) Drinking water and health, pharmacokinerics in risk assessment, Vol. 8.
Washington DC: National Academy Press.
NRC (National Research Council) (2006) Assessing the human health risks of trichloroethylene: key scientific
issues. Committee on Human Health Risks of Trichloroethylene. Washington, DC: National Academies Press.
NRC (National Research Council). (1983) Risk assessment in the federal government: managing the process.
Washington, DC: National Academy Press.
NRC (National Research Council). (1993) Pesticides in the diets of infants and children. Washington, DC: National
Academy Press.
NRC (National Research Council). (2006) Key scientific issues in the health risk assessment of trichloroethylene.
Washington: National Academy Press.
NRC (National Research Council). 1994. Science and judgment in risk assessment. Washington, DC: National
Academy Press
NTP (National Toxicology Program). (1988) Toxicology and carcinogenesis studies of trichloroethylene (CAS no.
79-01-6) in four strains of rats (ACI, August, Marshall, Osborne-Mendel) (gavage studies). Public Health Service,
U.S. Department of Health and Human Services; NTP TR-273; NIH Publication No. 88-2529. Available from the
National Institute of Environmental Health Sciences, Research Triangle Park, NC, and the National Technical
Information Service, Springfield, VA; PB88-218896. Available online at
http://ntp.niehs.nih.gov/ntp/htdocs/LT rpts/tr273.pdf.
NTP (National Toxicology Program). (1990) Carcinogenesis studies of trichloroethylene (without epichlorohydrin)
in F344/N rats and B6C3F1 mice (gavage studies). Public Health Service, U.S. Department of Health and Human
Services; NTP TR-243; NIH Publication No. 90-1779. Available from the National Institute of Environmental
Health Sciences, Research Triangle Park, NC. Available online at
http://ntp.niehs.nih.gov/ntp/htdocs/LT rptsAr243.pdf.
Nunes, J; Enrich, M; Robertson, J. (2001) Toxicosis associated with dual oral exposure of rats to lead and
trichloroethylene. Toxicol Pathol 29(4):451-457.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-55 DRAFT—DO NOT CITE OR QUOTE
-------
Nutley, EV; Tcheong, AC; Allen, JW; et al. (1996) Micronuclei induced in round spermatids of mice after stem-cell
treatment with chloral hydrate: evaluations with centromeric DNA probes and kinetochore antibodies. Environ Mol
Mutagen28(2):80-89.
O'Donnell, L; McLachlan, RI; Wreford, NG; et al. (1994) Testosterone promotes the conversion of round spermatids
between stages VII and VIII of the rat spermatogenic cycle. Endocrinology 135:2608-2614.
Odum, J; Foster, JR; Green, T. (1992) A mechanism for the development of clara cell lesions in the mouse lung after
exposure to trichloroethylene. Chem Biol Interact 83:135-153.
Ogata, M; Takatsuka, Y; Tomokuni, K. (1971) Excretion of organic chlorine compounds in the urine of persons
exposed to vapours of trichloroethylene and tetrachloroethylene. Br J Ind Med 28(4):386-391.
Ohta, M; Saito, T; Saito, K; et al. (2001) Effect of trichloroethylene on spatiotemporal pattern of LTP in mouse
hippocampal slices. Int J Neurosci 111(3-4):257-271.
Okino T, Nakajima T, Nakano M. (1991) Morphological and biochemical analyses of trichloroethylene
hepatotoxicity: differences in ethanol- and phenobarbital-pretreated rats. Toxicol Appl Pharmacol 108(3):379-389.
Olsson, AR; Skogh, T; Axelson, O; et al. (2004) Occupations and exposures in the work environment as
determinants for rheumatoid arthritis. Occup Environ Med 61:233-238.
Olsson, AR; Skogh, T; Wingren, G. (2000) Occupational determinants for rheumatoid arthritis. Scand J Work
Environ Health 26:243-249.
Ono, Y; Somiya, I; Kawamura, M. (1991) The evaluation of genotoxicity using DNA repairing test for chemicals
produced in chlorination and ozonation processes. Water Sci Technol 23(l-3):329-338.
Opdam, JJ. (1989) Intra and interindividual variability in the kinetics of a poorly and highly metabolizing solvent.
Br J Ind Med 46(12):831-845.
Ou, J; Ou, Z; McCarver, G; et al. (2003) Trichloroethylene decreases heat shock protein 90 interactions with
endothelial nitric oxide synthase: implications for endothelial cell proliferation. Toxicol Sci 73:90-97.
Palmer, CN; Hsu, MH; Griffin, KJ; et al. (1998) Peroxisome proliferator activated receptor-alpha expression in
human liver. Mol Pharmacol 53(l):14-22.
Parent, ME; Hua, Y; Siemiatycki. (2000) Occupational risk factors for renal cell carcinoma in Montreal. Am J Ind
Med 38:609-618.
Parent, M-E; Siemiatycki, J; Fritschi, L. 2000 Workplace exposure and oesophageal cancer. Occup Environ Med
57:325-324.
Park, PA; Gilmour, MI; Selgrade, MK. (1993) Pulmonary defenses to streptococcal infection following acute
exposure to trichloroethylene (TCE) or chloroform (CHC13). Toxicologist 13:107.
Park, SB; Osterich, JD; Vamvakas, S; et al. (1992) Flavin-containing monooxygenase-dependent stereoselective S-
oxygenation and cytotoxicity of cysteine S-conjugates and mercapturates. Chem Res Toxicol 5(2): 193-201.
Parker, MJ; Tarlow, MJ; Anderson, JM. (1984) Glue sniffing and cerebral infarction. ArchDis Child
59(7):675-677.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-56 DRAFT—DO NOT CITE OR QUOTE
-------
Parkinson, A; Mudra, DR; Johnson, C; et al. 2004) The effects of gender, age, ethnicity, and liver cirrhosis on
cytochrome P450 enzyme activity in human liver microsomes and inducibility in cultured human hepatocytes.
Toxicol Appl Pharmacol 199(3): 193-209.
Parnell, MJ; Koller, LD; Exon, JH; et al. (1986) Trichloroacetic acid effects on rat liver peroxisomes and enzyme-
altered foci. Environ Health Perspect 69:73-79.
Parrish, JM; Austin, EW; Stevens, DK; et al. (1996) Haloacetate-induced oxidative damage to DNA in the liver of
maleB6C3Fl mice. Toxicology 110:103-111.
Parry, JM; Parry, EM; Bourner, R; et al. (1996) The detection and evaluation of aneugenic chemicals. Mutat Res
353(l-2):ll-46.
Partenen, T; Heikkila, P; Hernberg, S; et al.. (1991) Renal cell cancer and occupational exposure to chemical agents.
Scand J Work Environ Health 17:231-239.
Pastino, GM; Yap, WY; Carroquino, M. (2000) Human variability and susceptibility to trichloroethylene. Environ
Health Persp 108(Suppl. 2):201-214.
Paycok, ZV; Powell, JF. (1945) The excretion of sodium trichloroacetate. J Pharmacol Exper Therapeut
85:289-293.
Peden-Adams, MM; Eudaly, JG; Heesemann, LM; et al. (2006) Developmental immunotoxicity of trichloroethylene
(TCE): studies in B6C3F1 mice. J Environ Sci Health A Tox Hazard Subst Environ Eng 41:249-271.
Peden-Adams, MM; EuDaly, JG; Lee, AM; et al. (2008) Developmental effects of trichloroethylene (TCE) on
immunity: studies in MRL (+/+) mice. J Environ Sci Health A Tox Hazard Subst Environ Eng A43(12): 1402-1409.
Peden-Adams, MM; EuDaly, JG; Heesemann, LM; et al. (2006) Developmental immunotoxicity of
trichloroethylene (TCE): Studies in B6C3F1 mice. J Environ Sci Health A 41:249-271.
Pelekis, M; Gephart, LA; Lerman, SE. (2001) Physiological-model-based derivation of the adult and child
pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Regul Toxicol Pharmacol
33(1): 12-20.
Pellizzari, ED; Hartwell, TD; Harris, BS, 3rd; et al. (1982) Purgeable organic compounds in mother's milk. Bull
Environ Contam Toxicol 28(3):322-328.
Perbellini, L; Olivato, D; Zedde, A; et al. (1991) Acute trichloroethylene poisoning by ingestion: clinical and
pharmacokinetic aspects. Intens Care Med 17:234-235.
Percy, C; Stanek, E; Gloeckler, L. (1981) Accuracy of cancer death certificates and its effect on cancer motqality
statistics. Am J Public Health 71:242-250.
Perdue, MP; Severson, RK; Colt, JS; et al. (2009) Degreasing and risk of non-Hodgkin lymphoma. Occup Environ
Med66(8):557-560.
Pereira, MA; Kramer, PM; Conran, PB; et al. (2001) Effect of chloroform on dichloroacetic acid and trichloroacetic
acid-induced hypomethylation and expression of the c-myc gene and on their promotion of liver and kidney tumors
inmice. Carcinogenesis 22(9):1511-1519.
Pereira, MA; Phelps, BJ. (1996) Promotion by dichloroacetic acid and trichloroacetic acid of N-methyl-N-
nitrosurea-initiated cancer in the liver of female B6C3F1 mice. Cancer Lett 102:133-141.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-57 DRAFT—DO NOT CITE OR QUOTE
-------
Pereira, MA; Wang, W; Kramer, PM; et al. (2004) Prevention by methionine of dichloroacetic acid-induced liver
cancer and DNA hypomethylation in mice. Toxicol Sci 77:243-248.
Perocco, P; Prodi, G. (1981) DNA damage by haloalkanes in human lymphocytes cultured in vitro. Cancer Lett
13:213-218.
Perrot, N; Nalpas, B; Yang, CS; et al. (1989) Modulation of cytochrome P450 isozymes in human liver, by ethanol
and drug intake. Eur J Clin Invest 19(6):549-555.
Perry, SJ; Schofield, MA; Macfarlane, M; et al. (1993) Isolation and expression of a cDNA coding for rat kidney
cytosolic cysteine conjugate (3-lyase. Mol Pharmacol 43:660-665.
Persson, B; Dahlander, A-M; Fredriksson, M; et al. (1989) Malignant lymphomas and occupational exposures. Br J
IndMed 46:516-520.
Persson, B; Fredrikson, M. (1999) Some risk factors for non-Hodgkin's lymphoma. Int J Occup Med Environ
Health 12:135-142.
Persson, B; Fredriksson, M; Olsen, K; et al. (1993) Some occupational exposure as risk factors for malignant
lymphomas. Cancer 72:1773-1778.
Pesch, B; Haerting, J; Ranft, U; et al.. (2000) Occupational risk factors for renal cell carcinoma: agent-specific
results from a case-control study in Germany. MURC Study Group. Multicenter urothelial and renal cancer study.
Int J Epidemiol 29(6):1014-1024.
Peter, R; Bocker, R; Beaune, PH; et al. (1990) Hydroxylation of chlorzoxazone as a specific probe for human liver
cytochrome P-450IIE1. ChemRes Toxicol 3:566-573.
Peters, FM; Preston-Martin, S; Yu, MC. (1981) Brain tumors in children and occupational exposure of parents.
Science 213(4504):235-237.
Peters, JM; Cattley, RC; Gonzalez, FJ. (1997) Role of PPAR alpha in the mechanism of action of the nongenotoxic
carcinogen and peroxisome proliferator Wy-14,643. Carcinogenesis. 18:2029-2033.
Peters, JM; Garabrant, DH; Wright, WE; et al. (1985) Uses of a cancer registry in the assessment of occupational
cancer risks. Natl Cancer Inst Monogr 69:157-161. .
Peters, JM; Preston-Martin, S; Yu, MC. (1981) Brain tumors in children and occupational exposure of parents.
Science 213(4504):235-237.
Pfaffenberger, CD; Peoples, AJ; Enos, HF. (1980) Distribution of volatile halogenated organic compounds between
rat blood serum and adipose tissue. Int J Environ Anal Chem 8:55-65.
Phillips, TJ; Macdonald, RR. (1971) Comparative effect of pethidine, trichloroethylene, and Entonox on fetal and
neonatal acid-base and PO2. Br Med J 3(5774):558-560.
Plewa, MJ; Kargalioglu, Y; Vankerk, D; et al. (2002) Mammalian cell cytotoxicity and genotoxicity analysis of
drinking water disinfection by-products. Environ Mol Mutagen 40:134-142.
Poet, TS; Corley, RA; Thrall, KD; et al. (2000) Assessment of the percutaneous absorption of trichloroethylene in
rats and humans using MS/MS real-time breath analysis and physiologically based pharmacokinetic modeling.
Toxicol Sci 56:61-72.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-58 DRAFT—DO NOT CITE OR QUOTE
-------
Pogribny, IP; Tryndyak, VP; Woods, CG; et al. (2007) Epigenetic effects of the continuous exposure to peroxisome
proliferator WY- 14,643 in mouse liver are dependent on peroxisome proliferator activated receptor a. Mutat Res
625:62-71.
Porru, S; Placidi, D; Carta, A; et al. (2001) Primary liver cancer and occupation in men: A case-control study in a
high-incidence area in Northern Italy. Int J Cancer 94:878-883.
Portier, CJ; Bailer, AJ. (1989) Testing for increased carcinogenicity using a survival-adjusted quanta! response test.
Fund Appl Toxicol 12:731-737.
Povey, A; Guppy, MJ; Wood, M; et al. (2001) Cytochrome P2 polymorphisms and susceptibility to scleroderma
following exposure to organic solvents. Arthritis Rheum 44(3):662-665.
Powell, JF. (1945) Trichlorethylene: Absorption, elimination and metabolism. Br J Ind Med (2): 142-145.
Prasanna, HR; Hart, RW; Magee, PN. (1989) Effect of food restriction on the metabolism of dimethylnitrosamine
(NDMA) in rats. Anticancer Res 9:93-96.
Pravecek, TL; Channel, SR; Schmidt, WJ; et al. (1996) Cytotoxicity and metabolism of dichloroacetic and
trichloroacetic acid in B6C3F1 mouse liver tissue. Toxicol In Vitro 9(3):261-269.
Prendergast, JA; Jones, RA; Jenkins, LJ; et al. (1967) Effects on experimental animals of long-term inhalation of
trichloroethylene, carbon tetrachloride, 1,1,1-trichloroethane, dichlorodifluoromethane, and 1,1-dichloroethylene.
Toxicol Appl Pharmacol 10:270-289.
Price, PJ; Hassett, CM; Mansfield, JI. (1978) Transforming activities of trichloroethylene and proposed industrial
alternatives. In Vitro Cell Dev Biol Plant 14(3):290-293.
Price, RG; Taylor, SA; Olivers, I; et al. (1996) Development and validation of new screening tests for nephrotoxic
effects. Human Environ Toxicol 15(Suppl. 1):510-519.
Price, RG; Wedeen, R; Lichtveld, MY; et al. (1999) Urinary biomarkers: roles in risk assessment to environmental
and occupational nephrotoxins: monitoring of effects and evaluation of mechanisms of toxicity . Renal Failure
Prout, MS; Provan, WM; Green, T. (1985) Species differences in response to trichloroethylene. I. Pharmacokinetics
in rats and mice. Toxicol Appl Pharm 79(3):389-400.
Purdue, MP; Severson, RK; Colt, JS; et al. (2009) Degreasing and risk of non-Hodgkin lymphoma. Occup Environ
Med 66:557-560.
Raaschou-Nielsen, O; Hansen, J; McLaughlin, JK; et al. (2003) Cancer risk among workers at Danish companies
using trichloroethylene: a cohort study. AmJEpidemiol 158(12): 1182-1 192.
Radican, L; Blair, A; Stewart, P; et al. (2008) Mortality of aircraft maintenance workers exposed to
trichloroethylene and other hydrocarbons and chemicals: extended follow-up. J Occup Environ Med 50: 1306-13 19.
Radican, L; Wartenberg, D; Rhoads, GG; et al. (2006) A retrospective occupational cohort study of endOstage renal
disease in aircraft workers exposed to trichloroethylene and other hydrocarbons. J Occup Environ Med 48: 1-12.
Raijmakers, MT; Steegers, EA; Peters, WH. (2001) Glutathione S-transferases andthiol concentrations in
embryonic and early fetal tissues. Hum Reprod 16(1 1):2445-2450.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-59 DRAFT—DO NOT CITE OR QUOTE
-------
Rake, JP; Visser, G; Labrune, P; et al. (2002) Glycogen storage disease type I: diagnosis, management, clinical
course and outcome. Results of the European Study on Glycogen Storage Disease Type I (ESGSD I). Eur J Pediatr
161(Suppl 1):S20-S34.
Rakoff-Nahoum, S; Medzhitov, R. (2007) Regulation of spontaneous intestinal tumorigenesis through the adaptor
protein MyD88. Science 317(5834): 124-127.
Ramanakumar, A; Nadon, L; Siemiatycki, J. (2008) Exposure in painting related occupations and risk of selected
cancers: results from a case-control study in Montreal. Am J Ind Med 51:419-427.
Ramdhan, DH; Kamijima, M; Yamada, N; et al. (2008) Molecular mechanisms of trichloroethylene-induced
hepatotoxicity mediated by CYP2E1. Toxicol Appl Pharmacol 231:300-307.
Ramsdell, AF; Markwald. RR. (1997) Induction of endocardial cushion tissue in the avian heart is regulated, in part,
by TGFp-3-mediated autocrine signaling. Develop Biol 188:64-74.
Ramsey, JC; Andersen, ME. (1984) A physiologically based description of the inhalation pharmacokinetics of
styrene in rats and humans. Toxicol Appl Pharmacol 72:159-175.
Rapson, WH; Nazar, MA; Butsky, W. (1980) Mutagenicity produced by aqueous chlorination of organic
compounds. Bull Environ Contam Toxicol 24:590-596.
Rasmussen K, Sabroe S, Wohlert M, et al. (1988) A genotoxic study of metal workers exposed to trichloroethylene.
Sperm parameters and chromosome aberrations in lymphocytes. Int Arch Occup Environ Health 60(6):419-423.
Rasmussen, K; Arlien-Soborg, P; Sabroe, S. (1993a) Clinical neurological findings among metal degreasers exposed
to chlorinated solvents. Acta Neurol Scand 87(3):200-204.
Rasmussen, K; Brogren, C-H; Sabroe, S. (1993b) Subclinical affection of liver and kidney function and solvent
exposure. Int Arch Occup Environ Health 64:445-448.
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (1993c) Solvent-induced chronic toxic encephalopathy. Am J IndustrMed
23:779-792.
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (1993d) Psychometric tests for assessment of brain function after solvent
exposure. Am J Industr Med 24:553-565.
Rasmussen, K; Arlien-Soborg, P; Sabroe, S. (1993e) Clinical neurological findings among metal degreasers exposed
to chlorinated solvents. Acta Neurol Scand 87(3):200-204.
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (19931) Solvent-induced chronic toxic encephalopathy. Am J Ind Med
23:779-792.
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (1993g) Psychometric tests for assessment of brain function after solvent
exposure. Am J Ind Med 24(5):553-565.:
Raunio, H; Hokkola, J; Pelkonen, O; et al. (1998). Expression of xenobiotic metabolising cytochrome P450's in
human pulmonary tissue. Arch Toxicol. 20:465-469.
Ray, A; Chatterjee, S; Bagchi, P; et al. (1987) Effect of Quinalphos on testicular steroidogenesis in rats. Andrologia
19(3):163-168.
Rebert, CS; Boyes, WK; Pryor, GT; et al (1993) Combined effects of solvents on the rat's auditory system: styrene
and trichloroethylene. Int J Psychophysiol 14(l):49-59.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-60 DRAFT—DO NOT CITE OR QUOTE
-------
Rebert, CS; Day, VL; Matteucci, et al. (1991) Sensory-evoked potentials in rats chronically exposed to
trichloroethylene: predominant auditory dysfunction. Neurotoxicol Teratol 13(1):83-90.
Rebert, CS; Schwartz, RW; Svendsgaard, DJ; et al. (1995) Combined effects of paired solvents on the rat's auditory
system. Toxicology 105(2-3):345-54.
Reif, JS; Burch, JB; Nuckols, JR; et al. (2003) Neurobehavioral effects of exposure to trichloroethylene through a
municipal water supply. Environ Res 93(3):248-58.
Rennix, CP; Quinn, MM; Amoroso, PJ; et al. (2005) Risk of breast cancer among enlisted Army women
occupational exposed to volatile organic compounds. Am J Ind Med 48:157-167.
Reznik, G; Ward, JM; Hardisty, JF; et al. (1979) Renal carcinogenic and nephrotoxic effects of the flame retardant
tris(2,3-dibromopropyl) phosphate in F344 rats and (C57B1/6N x C3H/HeN)Fl mice. J Natl Cancer Inst
63(1):205-212.
Rhee, I; Bachman, KE; Park, BH; et al. (2002) DNMT1 and DNMT3b cooperate to silence genes in human cancer
cells. Nature 416(6880):552-556.
Rhomberg, LR. (2000) Dose-response analyses of the carcinogenic effects of trichloroethylene in experimental
animals. Environ Health Perspect 108(Suppl. 2):343-358.
Rice, D; Barone, S, Jr. (2000) Critical periods of vulnerability for the developing nervous system: evidence from
humans and animal models. Environ Health Perspect 108(Suppl. 3):511-533.
Richard, AM; Hunter, ES. (1996) Quantitative structure-activity relationships for the developmental toxicity of
haloacetic acids in mammalian whole embryo culture. Teratology 53:352-360.
Richardson, DB; Terschuren, C; Hoffmann, W. (2008) Occupational risk factors for non-Hodgkin's lymphoma: a
population-based case-control study in Northern Germany. Am J Ind Med 51:258-268.
Richmond, RE; Carter, JH; Carter, HW; et al. (1995) Immunohistochemical analysis of dichloroacetic acid (DCA)-
induced hepatocarcinogenesis in male Fischer (F344) rats. Cancer Lett 92:67-76.
Ries, LAG; Melbert, D; Krapcho, M; et al; eds. (2008) SEER Cancer Statistics Review, 1975-2005. National
Cancer Institute. Bethesda, MD. Available online at http://seer.cancer.gov/csr/1975_2005 [Accessed July 9, 2008.]
Ripp, SL; Overby, LH; Philpot, RM; et al. (1997) Oxidation of cysteine S-conjugates by rabbit liver microsomes and
cDNA-expressed flavin-containing mono-oxygenases: Studies with ^-(l^-dichloroviny^-L-cysteine, £-(1,2,2-
trichlorovinyl)-L-cysteine, S-allyl-L-cysteine, and S-benzyl-L-cysteine. Mol Pharmacol 51:507-515.
Ripp, SL; Itagaki, K; Philpot, RM; et al. (1999) Species and sex differences in expression of flavin-containing
monooxygenase form 3 in liver and kidney microsomes. Drug Metab Dispos 27(l):46-52.
Ristoff, E; Larsson, A. (2007) Inborn errors in the metabolism of glutathione. Orphanet J Rare Dis 2:16.
Ritz, B; Borgenstern, H; Froines, J; et al. (1999) Effects of exposure to external ionizing radiation on cancer
mortality in nuclear workers monitored for radiation at Rocketdyne/Atomics International. Am J Ind Med 35:21-31.
Ritz, B. (1999) Cancer mortality among workers exposed to chemicals during uranium processing. J Occup Environ
Med 41:556-566.
Robbiano, L; Baroni, D; Carrozzino, R; et al. (2004) DNA damage and micronuclei induced in rat and human
kidney cells by six chemicals carcinogenic to the rat kidney. Toxicology 204(2-3): 187-195.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-61 DRAFT—DO NOT CITE OR QUOTE
-------
Robbiano, L; Mereto, E; Migliazzi-Morando, A; et al. (1998) Increased frequency of micronucleated kidney cells in
rats exposed to halogenated anaesthetics. Mutat Res 413(l):l-6.
Rodenbeck, SE; Sanderson, LM; Rene, A. (2000) Maternal exposure to trichloroethylene in drinking water and
birth-weight outcomes. Arch Environ Health 55(3): 188-194.
Rodriguez, CE; Mahle, DA; Gearhart, JM; et al. (2007) Predicting age-appropriate pharmacokinetics of six volatile
organic compounds in the rat utilizing physiologically based pharmacokinetic modeling. Toxicol Sci 98(l):43-56.
Rossi, AM; Migliore, L; Barali, R; et al. (1983) In vivo and in vitro mutagenicity studies of a possible carcinogen,
trichloroethylene, and its two stabilizers, epichlorohydrin and 1,2-epoxybutane. Teratog Carcinog Mutagen 3:75-87.
Rothman, KJ (1986) Modern epidemiology, first edition. Boston: Little, Brown and Company.
Rothman, KJ; Greenland, S. (1998) Modern epidemilogy, second edition. Philadelphia: Lippincott Williams &
Wilkins.
Rothman, KJ; Greenland, S; Lash, TL. (2008) Case-control studies, Chapter 2. In: Modern Epidemiology, 3rd ed.
Phildelphia, PA: Wolters Kluwer/Lippincott Williams and Wilkins.
Rufer, ES; Hacker, T; Lough, J; et al. (2008) Low-dose trichloroethylene exposure during valvuloseptal
morphogenesis causes ventricular septal defects in hatched chicks. Toxicologist 102(S-1):316.
Ruijten, M; Verberk, MM; Salle, HJ. (1991) Nerve function in workers with long term exposure to trichloroethene.
BrJIndMed48(2):87-92.
Russo, A; Levis, AG. (1992a) Detection of aneuploidy in male germ cells of mice by means of a meiotic
micronucleus assay. Mutat Res 281(3):187-191.
Russo, A; Levis, AG. (1992b) Further evidence for the aneuploidogenic properties of chelating agents: induction of
micronuclei in mouse male germ cells by EDTA. Environ Mol Mutagen 19(2): 125-131.
Russo, A; Pacchierotti, F; Metalli, PN. (1984) Nondisjunction induced in mouse spermatogenesis by chloral hydrate,
a metabolite of trichloroethylene. Environ Mutagen 6:695-703.
Russo, A; Stocco, A; Majone, F. (1992) Identification of kinetochore-containing (CREST+) micronuclei in mouse
bone marrow erythrocytes. Mutagenesis 7(3): 195-197.
Rusyn, I; Bradham, CA; Cohn, L; et al. (1999) Corn oil rapidly activates nuclear factor - KB in hepatic Kupffer cells
by oxidant-dependent mechanisms. Carcinogenesis 20(11):2095-2100.
Rusyn, I; Peters, JM; Cunningham, ML. (2006) Modes of action and species-specific effects of di-(2-
ethylhexl)phthalate in the liver. Crit Rev Toxicol 36:459-479.
Sabel; GV; Clark, TP. (1984) Volatile organic compounds as indicators of municipal solid waste leachate
contamination. Waste Manag Res 2:119-130.
Sagawa, K; Nishitani, H; Kawai, H; Kuge, Y; Ikeda, M. (1973) Transverse lesion of spinal cord after accidental
exposure to trichloroethylene. Int Arch Arbeitsmed 31(4):257-64
Saghir, S A; Schultz, IR. (2002) Low-dose pharmacokinetics and oral bioavailability of dichloroacetate in naive and
GSTC-depleted rats. Environ Health Persp 110(8):757-763.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-62 DRAFT—DO NOT CITE OR QUOTE
-------
Saihan, EM; Burton, JL; Heaton, KW. (1978) A new syndrome with pigmentation, scleroderma, gynaecomastia,
Raynaud's phenomenon and peripheral neuropathy. Br J Dermatol 99(4):437-440.
Saillenfait, AM; Langonne, I; Sabate, JP. (1995) Developmental toxicity of trichloroethylene, tetrachloroethylene
and four of their metabolites in rat whole embryo culture. Arch Toxicol 70:71-82.
Sallmen, M; Lindbohm, ML; Anttila, A; Kyyronen, P; Taskinen, H; Nykyri, E; Hemminki, K. (1998) Time to
pregnancy among the wives of men exposed to organic solvents. Occup Environ Med 55(1):24-30.
Sallmen, M; Lindbohm, ML; Kyyronen, P; et al. (1995) Reduced fertility among women exposed to organic
solvents. Am J Ind Med 27(5):699-713.
Salmon, AG; Kizer, KW; Zeise, L; et al. (1995) Potential carcinogenicity of chloral hydrate~a review. J Toxicol
Clin Toxicol 3 3 (2): 115-21.
Salvan, A; Stayner, L; Steenland, K; et al. (1995) Selecting an exposure lag period. Epidemiology 6:387-390.
Salvini, M; Binasch, S; Riva, M. (1971) Evaluation of the psychophysiological functions in humans exposed to
trichloroethylene. Br J Ind Med 28:293-295.
Sanchez, IM; Bull, RJ. (1990) Early induction of reparative hyperplasia in the liver of B6C3F1 mice treated with
dichloroacetate and trichloroacetate. Toxicology 64:33-46.
Sanders, VM; Tucker, AN; White, KL, Jr; et al. (1982) Humoral and cell-mediated immune status in mice exposed
to trichloroethylene in the drinking water. Toxicol Appl Pharmacol 62:358-368.
Sanders, VM; Kauffmann, BM; White, KL, Jr; et al. (1982) Toxicology of chloral hydrate in the mouse. Environ
Health Perspect 44:137-146
Santibanez, M; Vioque, J; Alguiacil, J; et al. (2008) [In Press] Occupational exposures and risk of oesophageal
cancer by histological type: A case control study in eastern Spain. Occup Environ Med DOI:
10.1135/oem.2007.037929.
Sapkota, A; Williams, D; Buckley, T. (2005) Tollbooth workers and mobile source-related hazardous air pollutants:
How protective is the indoor environment? Environ Sci Technol 39:2936-2943.
Sarangapani, R; Gentry, PR; Covington, TR; et al. (2003) Evaluation of the potential impact of age- and gender-
specific lung morphology and ventilation rate on the dosimetry of vapors. Inhal Toxicol 15:987-1016.
Sato, A; Endoh, K; Kaneko, T; et al. (1991b) A simulation study of physiological factors affecting pharmacokinetic
behaviour of organic solvent vapours. Br J Ind Med 48(5):342-347.
Sato, A; Nakajima, T; Fujiwara, Y; et al. (1977) A pharmacokinetic model to study the excretion of
trichloroethylene and its metabolites after an inhalation exposure. Brit J Ind Med 34:56-63.
Sato, A; Nakajima, T; Koyama, Y. (1980) Effects of chronic ethanol consumption on hepatic metabolism of
aromatic and chlorinated hydrocarbons in rats. Br J Ind Med 37(4):382-386.
Sato, A; Nakajima, T; Koyama, Y. (1981) Dose-related effects of a single dose of ethanol on the metabolism in rat
liver of some aromatic and chlorinated hydrocarbons. Toxicol Appl Pharmacol 60(1):8-15.
Sato, A; Nakajima, T; Koyama, Y. (1983) Interaction between ethanol and carbohydrate on the metabolism in rat
liver of aromatic and chlorinated hydrocarbons. Toxicol Appl Pharmacol 68(2):242-249.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-63 DRAFT—DO NOT CITE OR QUOTE
-------
Sato, A; Nakajima, T. (1985) Enhanced metabolism of volatile hydrocarbons in rat liver following food deprivation,
restricted carbohydrate intake, and administration of ethanol, phenobarbital, polychlorinated biphenyl and 3-
methylcholanthrene: a comparative study. Xenobiotica 15(l):67-75.
Sato, A; Nakajima, T. (1978) Differences following skin or inhalation exposure in the absorption and excretion
kinetics of trichloroethylene and toluene. Brit J Ind Med 35:43-49.
Sato, A. (1993) Confounding factors in biological monitoring of exposure to organic solvents. Int Arch Occup
Environ Health 65(Suppl. 1):S61-S67.
Sato, A; Nakajima, T; Fujiwara, Y; et al. (1977) A pharmacokinetic model to study the excretion of
trichloroethylene and its metabolites after an inhalation exposure. Br J Ind Med 34:56-63.
Sato, A; Nakajima, T. (1979) A vial-equilibration method to evaluate the drug-metabolizing enzyme activity for
volatile hydrocarbons. Toxicol Appl Pharm 47(l):41-46.
Saunders, RA. (1967) A new hazard in closed environmental atmospheres. Arch Environ Health 14:380-384.
Sausen, PJ; Elfarra, AA. (1990) Cysteine conjugate S-oxidase. Characterization of a novel enzymatic activity in rat
hepatic and renal microsomes. J Biol Chem 265(1 1):6139-6145.
Sausen, PJ; Elfarra, AA. (1991) Reactivity of cysteine s-conjugate sulfoxides: formation of ^-[
glutathionyl)vinyl]-L-cysteine sulfoxideby the reaction of ^-(l^-dichloroviny^-L-cycsteine sulfoxide with
glutathione. Chem Res Toxicol 4:655-660.
Savolainen, H; Pfaffli, P; Tengen, M; et al. (1977) Trichloroethylene and 1,1,1-trichloroethane: effects on brain and
liver after five days intermittent inhalation. Arch Toxicol 38:229-237.
Saxena, DK; Srivasta, RS; Lai, B; et al. (1987) The effect of lead exposure on the testis of growing rats. Exp Pathol
31:249-252.
Say gun, M; Cakmak, A; Ekici, A; et al. (2007) Five annual observations of respiratory findings in gun factory
workers exposed to solvents. J Occup Environ Med 49:909-912.
Sbrana, I; Di Sibio, A; Lomi, A; et al.. (1993) C-mitosis and numerical chromosome aberration analyses in human
lymphocytes: 10 known or suspected spindle poisons. Mutat Res 281:57-70.
Scharf, JG; Dombrowski, F; Ramadori, G. (2001) The IGF axis and hepatocarcinogenesis. Mol Pathol 54:138-144.
Schatten, H; Chakrabarti, A. (1998) Centrosome structure and function is altered by chloral hydrate and diazepam
during the first reproductive cell cycles in sea urchin eggs. Eur J Cell Biol 75(1):9-20.
Scheff, PA; Wadden, RA. (1993) Receptor modeling of volatile organic compounds. 1. Emission inventory and
validation. Environ Sci Technol 27:617-625.
Schenk M; Purdue, MP; Colt, JS; et al. (2009) Occupation/industry and risk of non-Hodgkin's lymphoma in the
United States. Occup Environ Med 66:23-3 1.
Schlehofer, B; Heuer, C; Blettner, M; et al. (1995) Occupation, smoking and demographic factors, and renal cell
carcinoma in Germany. Int J Epidemiol 24(l):51-7.
Schraml, P; Kononen, J; Bubendorf, L; et al. (1999) Tissue microarrays for gene amplification surveys in many
different tumor types. Clin Cancer Res 5(8): 1966-1975.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-64 DRAFT— DO NOT CITE OR QUOTE
-------
Schraml, P; Zhou, M; Richter, J; et al. (1999) Analysis of renal tumors in trichloroethylene-exposed workers by
comparative genomic hybridization and DNA sequencing analysis. Verh Dtsch Ges Pathol 83:218-224.
Schultz, IR; Merdink, JL; Gonzalez-Leon, A; Bull, RJ. (1999) Comparative toxicokinetics of chlorinated and
brominated haloacetates in F3444 rats. Toxicol Appl Pharm 158(2): 103-114.
Schultz, IR; Merdink, JL; Gonzalez-Leon, A; et al. (2002) Dichloroacetate toxicokinetics and disruption of tyrosine
catabolism in B6C3F1 mice: dose-response relationships and age as a modifying factor. Toxicology 173:229-247.
Schwetz, BA; Leong, BKJ; Gehring, PJ. (1975) The effect of maternally inhaled trichloroethylene,
perchloroethylene, methyl chloroform, and methylene chloride on embryonal and fetal development in mice and rats.
Toxicol Appl Pharmacol 32:84-96.
Scott, JE; Forkert, PG; Oulton, M; et al. (1988) Pulmonary toxicity of trichloroethylene: induction of changes in
surfactant phospholipids and phospholipase A2 activity in the mouse lung. Exp Mol Pathol 49:141-150.
Seelbach, A; Fissler, B; Madle, S. (1993) Further evaluation of a modified micronucleus assay with V79 cells for
detection of aneugenic effects. Mutat Res 303(4): 163-169.
Seidler, A; Mohner, M; Berger, J; et al. (2007) Solvent exposure and malignant lymphoma: a population-based case-
control study in Germany. J Occup Med Toxicol 2:2.
Selden, A; Hultberg, B; Ulander, A; et al. (1993) Trichloroethylene exposure in vapour degreasing and the urinary
excretion of N-acetyl-p-glucosaminidase. Arch Toxicol 67:224-226.
Sellers, EM; Lang-Sellers, BS; Koch-Weser, J. (1978) Comparative metabolism of chloral hydrate and triclofos.
From the Division of Clinical Pharmacology, Clinical Institute, Addiction Research Foundation and Toronto
Western Hospital and Departments of Pharmacology and Medicine, University of Toronto, Toronto, and the Centre
de Recherche Merrell International, Strasbourg, France.
Sellers, EM; Lang, M; Koch-Weser, J; et al. (1972) Interaction of chloral hydrate and ethanol in man. I: metabolism.
Clin Pharmacol Ther 13:37-49.
Selmin, O; Thorne, PA; Caldwell, PT; et al. (2008) Trichloroethylene and trichloroacetic acid regulate calcium
signaling pathways in murine embryonal carcinoma cells P19. Cardiovasc Toxicol. 8(2):47-56. Doi: 10.1007/s
12012-008-9014-2.
Seng, JE; Agrawal, N; Horsley, ETM; et al. (2003) Toxicokinetics of chloral hydrate in ad libitum-fed, dietary
controlled and calorically restricted male B6C3F1 mice following short-term exposure. Toxicol Appl Pharmacol
193(2):281-292.
Seo, M; Ikeda, K; Okamura, T; et al. (2008) Enhancing effect of chlorinated organic solvents on histamine release
and inflammatory mediator production. Toxicology 243:75-83.
Sexton, K; Adgate, JL; Church, TR; et al. (2005) Children's exposure to volatile organic compounds as determined
by longitudinal measurements in blood. Environ Health Perspect 113(3):342-349.
Sexton, K; Mongin, S J; Adgate, JL; et al. (2007) Estimating volatile organic compound concentrations in selected
microenvironments using time-activity and personal exposure data. J Toxicol Environ Health A 70(5):465-476.
Sexton, K; Adgate, JL; Ramachandran, G; et al. (2004) Comparison of personal, indoor and outdoor exposures to
hazardous air pollutants in three urban communities. Environ Sci Technol 38:423-430.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-65 DRAFT—DO NOT CITE OR QUOTE
-------
Shafer, TJ; Bushnell, PJ; Benignus, VA; et al. (2005) Perturbation of voltage-sensitive Ca2+ channel function by
volatile organic solvents. J Pharmacol Exp Ther 315(3): 1109-1118.
Shah, YM; Morimura, K; Yang, Q; et al. (2007) Peroxisome proliferator-activated receptor a regulates a
microRNA-mediated signaling cascade responsible for Hepatocellular proliferation. Mol Cell Biol
27(12):4238-4247.
Shah, JJ; Singh, HB. (1988) Distribution of volatile organic chemicals in outdoor and indoor air: a national VOCs
database. Environ Sci Technol 22:1381-1388.
Shahin, MM; vonBorstel, RC (1977) Mutagenic and lethal effects of a-benzene hexachloride, dibutyl-phthalate and
trichloroethylene in Saccharomyces cerevisiae. Mutat Res 48:173-180.
Shaib, Y; El-Serag, HB. (2004) The epidemiology of cholangiocarcinoma. Semin Liver Dis 24:115-125.
Shankar, K; Vaidya, VS; Gorton, JJ; et al. (2003) Activation of PPAR-a in streptozotocin-induced diabetes is
essential for resistance against acetaminophen toxicity. FASEB J 17(12): 1748-1750
Shao, J; Stapleton, PL; Lin, YS; et al. (2007) Cytochrome p450 and glutathione s-transferase mRNA expression in
human fetal liver hematopoietic stem cells. Drug Metab Dispos 35(1): 168-175.
Sharp, CR; Rochon, JE; Adam, JM; Suissa, S. (1989) Case-control study of hydrocarbon exposure in patients with
renal cell carcinoma. CMAJ 140:1309-1318.
Sheffels, P; Schroeder, JL; Altuntas, TG; et al. (2004) Role of cytochrome P4503A in cysteine S-conjugates
sulfoxidation and the nephrotoxicity of the sevoflurane degradation product fluoromethyl-2,2-difluoro-l-
(trifluoromethyl)vinyl ether (compound A) in rats. Chem Res Toxicol 17(9): 1177-1189.
Sherins, RJ; Hodgen, GD. (1976) Testicular GT and SDH indices of Sertoli and germ cell function in man. J
ReprodFertil 48:191-193.
Shiao, YH; Kamata, SI; Li, LM; et al. (2002) Mutations in the VHL gene from potassium bromate-induced rat clear
cell renal tumors. Cancer Lett 187:207-214.
Shiao, YH; Rice, JM; Anderson, LM; et al. (1998) von Hippel-Lindau gene mutations in N-nitrosodimethylamine-
induced rat renal epithelial tumors. J Natl Cancer Inst 90(22):1720-1723.
Shih, CL; Chen, HH; Chiu, TH. (2001) Acute exposure to trichloroethylene differentially alters the susceptibility to
chemoconvulsants in mice. Toxicology 162:35-42.
Shimada, T; Swanson, AF; Leber, P; et al. (1985) Activities of chlorinated ethane and ethylene compounds in the
Salmonella microsome mutagenesis and rat hepatocyte/DNA repair assays under vapor phase exposure conditions.
Cell Biol Toxicol 1:159-179.
Shimada, T; Yamazaki, H; Mimura, M; et al. (1994) Interindividual variations in human liver cytochrome P-450
enzymes involved in the oxidation of drugs, carcinogens and toxic chemicals: studies with liver microsomes of 30
Japanese and 30 Caucasians. J Pharmacol Exp Ther 270(l):414-423.
Shindell, S; Ulrich, S. (1985) A cohort study of employees of a manufacturing plant using trichloroethylene. J
OccupMed 27:577-579.
Shroads, AL; Guo, X; Dixit, V; et al. (2008) Age-dependent kinetics and metabolism of dichloroacetate: possible
relevance to toxicity. J Pharmacol Exp Ther 324(3):1163-1171.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-66 DRAFT—DO NOT CITE OR QUOTE
-------
Shu, XO; Perentesis, JP; Wen, W; et al. (2004) Parental exposure to medications and hydrocarbons and ras
mutations in children with acute lymphoblastic leukemia: a report from the Children's Oncology Group. Cancer
Epidemiol Biomarkers Prev 13:1230-1235.
Shu, XO; Stewart, P; Wen, WQ; et al. (1999) Parental occupational exposure to hydrocarbons and risk of acute
lymphocytic leukemia in offspring. Cancer Epidemiol Biomarkers Prev 8(9):783-791.
Shultz, J; Weiner, H. (1979) Alteration of the enzymology of chloral hydrate reduction in the presence of ethanol.
Biochem Pharmacol 28:3379-3384.
Siemiatycki, J. (1991) Risk factors for cancer in the workplace. Boca Raton: CRC Press.
Siemiatycki, J; Dewar, R; Nadon, L; et al (1994) Occupational risk factors for bladder cancer: results from a case-
control study in Montreal, Quebec, Canada. Am JEpidemiol 140:1061-1080.
Siemiatycki, J; Wacholder, S; Dewar, R; et al. (1988) Degree of confounding bias related to smoking, ethnic group,
and socioeconomic status in estimates of the associations between occupational and cancer. J Occup Med
30:617-625.
Simmons, JE; Boyes, WK; Bushnell, PJ; et al. (2002) A physiologically based pharmacokinetic model for
trichloroethylene in the male long-evans rat. Toxicol Sci 69:3-15.
Singh, M; Sinha, U. (1976) Chloral hydrate induced haploidization in Aspergillus nidulans. Experientia.
32(9): 1144-1145.
Sinks, T; Lushniak, B; Haussler, BJ; et al. (1992) Renal cell cancer among paperboard printing workers.
Epidemiology 3:483-489.
Skender, LJ; Karacic, V; Prpic-Majic, D. (1991) A comparative study of human levels of trichloroethylene and
tetrachloroethylene after occupational exposure. Arch Environ Health 46(3): 174-178.
Slikker, W, Jr; Andersen, ME; Bogdanffy, MS; et al. (2004a) Dose-dependent transitions in mechanisms of toxicity.
Toxicol Appl Pharmacol 201:203-225.
Slikker, W, Jr; Andersen, ME; Bogdanffy, MS; et al. (2004b) Dose-dependent transitions in mechanisms of toxicity:
case studies. Toxicol Appl Pharmacol 201:226-294.
Smith, MK; Randall, JL; Read, EJ; et al. (1989) Teratogenic activity of trichloroacetic acid in the rat. Teratology
40:445-451.
Smith, MK; Randall, JL; Read, EJ; et al. (1992) Developmental toxicity of dichloroacetate in the rat. Teratology
46:217-223.
Smith, GF. (1970) The investigation of the mental effects of trichlorethylene. Ergonomics 13(5):580-586.
Snyder, RD; Pullman, J; Carter, JH; et al. (1995) In vivo administration of dichloroacetic acid suppresses
spontaneous apoptosis in murine hepatocytes. Cancer Res 55:3702-3705.
Song, BJ; Veech, RL; Saenger, P. (1990) Cytochrome P450IIE1 is elevated in lymphocytes from poorly controlled
insulin-dependent diabetics. J Clin Endocrinol Metab 71(4): 1036-1040.
Sonnenfeld, N; Hertz-Picciotto, I; Kaye, WE. (2001) Tetrachloroethylene in drinking water and birth outcomes at
the US Marine Corps Base at Camp Lejeune, North Carolina. Am J Epidemiol 154(10):902-908.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-67 DRAFT—DO NOT CITE OR QUOTE
-------
Sora, S; Carbone, MLA. (1987) Chloral hydrate, methylmercury hydroxide and ethidium bromide affect
chromosomal segregation during meiosis of Saccharomyces cerevisiae. Mutat Res 190(1): 13-17.
Soucek, B; Vlachova, D. (1960) Excretion of trichloroethylene metabolites in human urine. Brit J Ind Med 17:60-
64.
Spiegelhalter, D; Thomas, A; Best, N; et al. (2003) WinBUGS user manual. Version 1.4. Available online at
www.mrc-bsu.cam.ac.uk/bugs/winbugs/manuall4.pdf.
Spielberg, SP. (1996) N-acetyltransferases: pharmacogenetics and clinical consequences of polymorphic drug
metabolism. J Pharmacokinet Biopharm 24:509-519.
Squillace, PJ; Moran, MJ; Price, CV. (2004) VOCs in shallow groundwater in new residential/commercial areas of
the US. Environ Sci Technol 38:5327-5338.
SRI, 1992. (1992) Directory of chemical producers-United States of America. Menlo Park, CA: SRI International,
1032.
Srivastava, S; Srivastava, SP. (1991) Effect of di(2-ethylhexyl)phthalate on 17 (3-hydroxy steroid dehydrogenase.
Toxicol Lett 57:235-239.
Staats, DA; Fisher, JW; Connolly, RB. (1991) Gastrointestinal absorption of xenobiotics in physiologically based
pharmacokinetic models. A two-compartment description. Drug Metab Dispos 19(1): 144-148.
Stadler, JC; Kennedy, GL, Jr. (1996) Evaluation of the sensory irritation potential of volatile organic chemicals from
carpets—alone and in combination. Food Chem Toxicol 34(11-12): 1125-1130.
Stanley, LA; Blackburn, DR; Devereaux, S; et al. (1994) Ras mutations in methylclofenapate-induced B6C3F1 and
C57BL/10J mouse liver tumors. Carcinogenesis 15(6):1125-1131.
Staples, RE. (1974) Detection of visceral alterations in mammalian fetuses. Teratology 9:A37-A38.
Stauber, AJ; Bull, RJ; Thrall, BD. (1998) Dichloroacetate and trichloroacetate promote clonal expansion of
anchorage-independent hepatocytes in vivo and in vitro. Toxicol Appl Pharmacol 150:287-294.
Stauber, AJ; Bull, PJ. (1997) Differences in phenotype and cell replicative behavior of hepatic tumors induced by
dichloroacetate (DCA) and trichloroacetate (TCA). Toxicol Appl Pharmacol 144:235-246.
Steinberg, AD. (1993) Should chloral hydrate be banned? Pediatrics 92(3):442-446.
Stemhagen, A; Slade, J; Altman, R; et al. (1983) Occupational risk factors and liver cancer. Amer J Epidemiol
117:443-454.
Stemmer, K; Ellinger-Ziegelbauer, H; Ahr, HJ; et al. (2007) Carcinogen-specific gene expression profiles in short-
term treated Eker and wild-type rats indicative of pathways involved in renal tumorigenesis. Cancer Res
67(9):4052-68.
Stenner, RD; Merdink, JL; Stevens, DK; et al. (1997) Enterohepatic recirculation of trichloroethanol glucuronide as
a significant source of trichloroacetic acid. Metabolites of trichloroethylene. Drug Metab Dispos 25(5):529-535.
Stephens, EA; Taylor, JA; Kaplan, N; et al. (1994) Ethnic variation in the CYP2E1 gene: polymorphism analysis of
695 African-Americans, European-Americans and Taiwanese. Pharmacogenetics 4(4): 185-192.
Stephenson, RP. (1956) A modification of receptor theory. BrJPharmacol 11:379-393.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-68 DRAFT—DO NOT CITE OR QUOTE
-------
Stevens, JL; Robbins, JD; Byrd, RA. (1986) A purified cysteine conjugate B-lyase from rat kidney cytosol. Journal
of BiolChem 261:15529-15537.
Stevens, J. (1985a) Cysteine conjugate (3-lyase activities in rat kidney cortex: subcellular localization and
relationship to the hepatic enzyme. Biochem Biophys Res Commun 129(2):499-504.
Stevens, J. (1985b) Isolation of characterization of a rat liver enzyme with both cysteine conjugate (3-lyase and
kynureninase activity. J Biol Chem 260(13):7945-7950.
Stevens, J; Jakoby, WB. (1983) Cysteine conjugate (3-lyase. Mol Pharmacol 23:761-765.
Stevens, JL; Robbins, JD; Byrd, RA. (1986) A purified cycsteine conjugate (3-lyase from rat kidney cytosol. J Biol
Chem 261(33): 15529-15537.
Stevens, JL; Ayoubi, N; Robbins, JD. (1988) The role of mitochondrial matrix enzymes in the metabolism and
toxicity of cysteine conjugates. J Biol Chem 263(7):3395-3401.
Stevens, JL; Hatzinger, PB; Hayden, PJ. (1989) Quantitation of multiple pathways for the metabolism of
Mephrotoxic cycsteine conjugates using selective inhibitors of L-oc-hydroxy acid oxidase (L-amino acid oxidase)
and cysteine conjugate (3-lyase. Drug Metab Dispos 17(3):297-303.
Stewart, BW; Le Mesurier, SM; Lykke, AWJ. (1979) Correlation of biochemical and morphological changes
induced by chemical injury to the lung. Chem Biol Interact 26:321-338.
Stewart, PA; Lee, JS; Marano, DE; et al. (1991) Retrospective cohort mortality study of workers at an aircraft
maintenance facility. II. Exposures and their assessment. Br J Ind Med 48:531-537.
Stewart, RD; Hake, CL; Peterson, JE. (1974) "Degreaser's flush". Arch Environ Health 29(1): 1-5.
Stewart, P; Dosemeci, M. (2005) Exposure estimation in occupational epidemiology studies. Presentation at the
Third Meeting on Assessing Human Health Risks of Trichloroethylene, June 9, 2005, Irvine, CA.
Stewart, RD; Dodd, HC; Gay, HH; et al. (1970) Experimental human exposure to trichloroethylene. Arch Environ
Health 20(1):64-71.
Storer, RD; McKelvey, TW; Kraynak, AR; et al. (1996) Revalidation of the in vitro alkaline elution/rat hepatocyte
assay for DNA damage: improved criteria for assessment of cytotoxicity and genotoxicity and results for 81
compounds. MutatRes 368(2):59-101.
Stott, WT; Quast, JF; Watanabe, PG. (1982) The pharmacokinetics and macromolecular interactions of
trichloroethylene in mice and rats. Toxicol Appl Pharmacol 62:137-151.
Stuckhardt, JL; Poppe, SM. (1984) Fresh visceral examination of rat and rabbit fetuses used in teratogenicity testing.
Teratog Carcinog Mutagen 4:181-188.
Styles, JA; Kelly, M; Pritchard, NR;et al. (1988) A species comparison of acute hyperplasia induced by peroxisome
proliferatory methylcofenapate: involvement of the binucleated hepatocyte. Carcinogenesis 9:1647-1655.
Styles, JA; Wyatt, I; Coutts, C. (1991) Trichloroacetic acid: studies on uptake and effects on hepatic DNA and liver
growth in mouse. Carcinogenesis 12(9):1715-1719.
Sung, T-I; Chen P-C; Lee LJ-H; et al. (2007) Increased standardized incidence ratio of breast cancer in female
electronics workers. BMC Public Health 7:102. Available online at
http://www.biomedcentral.com/content/pdf/1471-2458-7-102.pdf. [Accessed July 9, 2007]
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-69 DRAFT—DO NOT CITE OR QUOTE
-------
Sung, TI; Wang, JD; Chen, PC. (2008) Increased risk of cancer in the offspring of female electronics workers.
Reprod Toxicol 25:115-119.
Sverdrup, B; Kallberg, H; Bengtsson, C; et al. (2005) Association between occupational exposure to mineral oil and
rheumatoid arthritis: results from the Swedish EIRA case-control study. Arthritis Res Ther 7:R1296-R1303.
Swaen, GM. (1995) Increased incidence of renal cell tumours in a cohort of cardboard workers exposed to
trichloroethylene. Comments onHenschler, D; et al.. Arch Toxicol 70:127-128, 131-133.
Swan, SH; Shaw, G; Harris, JA; et al. (1989) Congenital cardiac anomalies in relation to water contamination, Santa
Clara County, California, 1981-1983. Am JEpidemiol 129(5):885-893.
Sweeney, C; Farrow, DC; Schwartz, SM; et al. (2000) Glutathione S-transferase Ml, Tl, and PI polymorphisms as
risk factors for renal cell carcinoma: a case-control study. Cancer Epidemiol Biomarkers Prev 9(4):449-454.
Sweeney, LM; Kirman, CR; Gargas, ML; et al. (2009) Contribution of trichloroacetic acid to liver tumors observed
inperchloroethylene (perc)-exposed mice. Toxicology 260(l-3):77-83.
Sweet, CW; Vermette, SJ (1992) Toxic volatile organic compounds in urban air in Illinois. Environ Sci Technol
26:165-173.
't Mannetje, A; Dryson, E; Walls, C; et al. (2008) High risk occupations for non-Hodgkin's lymphoma in New
Zealand: case-control study. Occup EnvironMed 65:354-363.
Takashima, K; Ito, Y; Gonzalez, FJ; et al. (2008) Different mechanisms of DEHP-induced hepatocellular adenoma
tumorigenesis in wild-type and PPAR alpha-null mice. J Occup Health 50(2): 169-180.
Tang, X; Que, B; Song, X; et al. (2008) Characterization of liver injury associated with hypersensitive skin reactions
induced by trichloroethylene in the guinea pig maximization test. J Occup Health 50:114-121.
Tang, XJ; Li, LY; Huang, et al. (2002) Guinea pig maximization test for trichloroethylene and its metabolites.
Biomed Environ Sci 15:113-118.
Tanguay, RM; Jorquera, R; Poudrier, J; et al. (1996) Tyrosine and its catabolites: from disease to cancer. Acta
Biochim Pol 43:209-216.
Tao, L; Li, Y; Kramer, PM; et al. (2004) Hypomethylation of DNA and the insulin-like growth factor-II gene in
dichloroacetic and trichloroacetic acid-promoted mouse liver tumors. Toxicology 196:127-136.
Tao, L; Yan, S; Xie, M; et al. (2000) Effect of trichloroethylene and its metabolites, dichloroacetic acid and
trichloroacetic acid, on the methylation and expression of c-Jun and c-Myc protooncogenes in mouse liver:
prevention by methionine. Toxicological Sciences 54:399-407.
Taskinen, H; Anttila, A; Lindbohm, ML; et al. (1989) Spontaneous abortions and congenital malformations among
the wives of men occupationally exposed to organic solvents. Scand J Work Environ Health 15(5):345-352.
Taskinen, H; Kyyronen, P; Hemminki, K; et al. (1994) Laboratory work and pregnancy outcome. J Occup Med
36(3):311-319.
Tateishi, M; Suzuki, S; Shimizu, H. (1978) Cysteine conjugate (3-lyase in rat liver. J Biol Chem 253(24):8854-8859.
Tateishi, T; Nakura, H; Asoh, M; et al. (1997) A comparison of hepatic cytochrome P450 protein expression
between infancy and postinfancy. Life Sci 61(21):2567-2574.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-70 DRAFT—DO NOT CITE OR QUOTE
-------
Tatham, L; Tolbert, P; Kjeldsberg, C. (1997) Occupational risk factors for subgroups of non-Hodgkin's lymphoma.
Epidemiology 8:551-558.
Taylor, DH; Lagory, KE; Zaccaro, DJ; et al. (1985) Effect of trichloroethylene on the exploratory and locomotor
activity of rats exposed during development. Sci Total Environ 47:415-420.
Templin, MV; Parker, JC; Bull, PJ. (1993) Relative formation of dichloroacetate and trichloroacetate from
trichloroethylene in male B6C3F1 mice. Toxicol Appl Pharm 123(1): 1-8.
Templin, MV; Stevens, DK; Stenner, RD; etal. (1995a). Factors affecting species differences in the kinetics of
metabolites of trichloroethylene. J Toxicol Env Health 44(4):435-447.
Terracini, B; Parker, V. (1965a) A pathological study on the toxicity of s-dichlorovinyl-L-cysteine. Food Cosmet
Toxicol 42:67-74.
Tham, R; Bunnfors, I; Eriksson, B; et al. (1984) Vestibule ocular disturbances in rats exposed to organic solvents.
Acta Pharmacol Toxicol 54(l):58-63.
Tham, R; Larsby, B; Odkvist, LM; et al. (1979) The influence of trichloroethylene and related drugs on the
vestibular system. Acta Pharmacol Toxicol (Copenh) 44(5):336-342.
Thiele, DL; Eigenbrodt, EH; Ware, AJ. (1982) Cirrhosis after reported trichloroethylene and 1, 1, 1-trichlorethane
exposure. Gastroenterol 83:926-929.
Thrall, KD; Poet, TS. (2000) Determination of biokinetic interactions in chemical mixtures using real-time breath
analysis and physiologically based pharmacokinetic modeling. J Toxicol Environ Health A 59:653-670.
Till, C; Westall, CA; Rovet, JF; et al. (2001) Effects of maternal occupational exposure to organic solvents on
offspring visual functioning: a prospective controlled study. Teratology 64:134-141.
Tola, S; Vilhunen, R; Jarvinen, E; et al. (1980) A cohort study on workers exposed to trichloroethylene. J Occup
Med22(ll):737-740.
Tola, S; Vilhunen, R; Jarvinen, E; et al. (1980) A cohort study on workers exposed to trichloroethylene. J Occup
Med22(ll):737-740.
Toma, MI; Grosser, M; Herr, A; et al. (2008) Loss of heterozygosity and copy number abnormality in clear cell
renal cell carcinoma discovered by high-density affymetrix 10K single nucleotide polymorphism mapping array.
Neoplasia 10(7):634-642.
Tomisawa, H; Ichihara, S; Fukazawa, H; et al. (1986) Purification and characterization of human hepatic cysteine-
conjugate (3-lyase. Biochem J 235:569-575.
Tomisawa, H; Suzuki, S; Ichihara, S; et al. (1984) Purification and characterization of C-S lyase from fusobacterium
varium. J Biol Chem 259(4):2588-2593.
Tong, Z; Board, PG; Anders, MW. (1998) Glutathione transferase zeta catalyses the oxygenation of the carcinogen
dichloroacetic acid to glyoxylic acid. Biochem J 33 l(Pt. 2):371-374.
Toraason, M; Clark, J; Dankovic, D; Mathias, P; Skaggs, S; Walker, C; Werren, D. (1999) Oxidative stress and
DNA damage in Fischer rats following acute exposure to trichloroethylene or perchloroethylene. Toxicology
138:43-53.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-71 DRAFT—DO NOT CITE OR QUOTE
-------
Toth, B; Hornung, D; Scholz, C; et al. (2007) Peroxisome proliferator-activated receptors: new players in the field of
reproduction. Am J Reprod Immunol 58:298-310.
Travis CC, White RK. (1988) Interspecific scaling of toxicity data. Risk Anal. 8:119-125.
Travis, CC. (1990) Tissue dosimetry for reactive metabolites. Risk Anal 10:317-21.
Triebig, G; Essing, HG; Schaller, KH; et al. (1976) Biochemical and psychological examinations of
trichloroethylene exposed volunteers. Zentralbl Bakteriol [OrigB] 163(5-6):383-416. (translated from German)
Triebig, G; Lehrl, S; Kinzel, W; et al. (1977a) Psycho-pathometric results of follow-up studies of persons exposed to
trichloroethylene. Zbl Bakt Hyg, I Abt Orig B 164, 314-3271 (translated from German).
Triebig, G; Schaller, KH; Erzigkeit, H; et al. (1977b). Biochemical investigations and psychological studies of
persons chronically exposed to trichloroethylene with regard to non-exposure intervals. Int Arch Occup Environ
Health 38(3): 149-162 (translated from German).
Triebig, G; Trautner, P; Weltle, D; et al. (1982) Investigations on the Neurotoxicity of processing materials III.
Measurement of Motor and Sensory Nerve Conduction Velocity in Persons with Job-Related Trichloroethylene
Exposure. Int Arch Occup Environ Health (1982) 51: 25-34 (translated from German).
Triebig, G; Bestler, W; Baumeister, P; et al. (1983) Investigations on the neurotoxicity of chemical substances at the
workplace, Part IV: Determination of the motor and sensory nerve conduction velocity in persons occupationally
exposed to a mixture of organic solvents. Int Arch Occup Environ Health 52:139-150 (translated from German).
Troster, Al; Ruff, RM. (1990) Neuropsychological sequelae of exposure to the chlorinated hydrocarbon solvents
trichloroethylene and trichloroethane. Arch Clin Neuropsychol 5(1):31-47.
Tryndyak, VP; Muskhelishvili, L; Kovalchuk, O; et al. (2006) Effect of long-term tamoxifen exposure on genotoxic
and epigenetic changes in rat liver; implications for tamoxifen-induced hepatocarcinogenesis. Carcinogenesis 27(8)
1713-1720.
Tu, AS; Murray, TA; Hatch, KM; et al. (1985) In vitro transformation of BALBZC-3T3 cells by chlorinated ethanes
and ethylenes. Cancer Lett 28:85-92.
Tucker, AN; Sanders, VM; Barnes, DW; et al. (1982) Toxicology of trichloroethylene in the mouse. Toxicol Appl
Pharmacol 62:351-3 57.
Tugwood, JD; Aldridge, TC; Lambe, KG; et al. (1996) Peroxisome proliferator-activated receptors: structures and
function. Ann N Y Acad Sci 804:252-265.
Tzeng, HF; Blackburn, AC; Board, PG; et al. (2000) Polymorphism- and species-dependent inactivation of
glutathione transferase zeta by dichloroacetate. Chem Res Toxicol 13:231-236.
U.S. EPA (Environmental Protection Agency). (1985) Assessment of trichloroethylene as a potentially toxic air
pollutant; proposed rule. Federal Register; Part IV 50(246):52422-52425.
U.S. EPA (Environmental Protection Agency). (1986a) Guidelines for the health risk assessment of chemical
mixtures. Federal Register 51(185):34014-34025. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Environmental Protection Agency). (1986b) Guidelines for mutagenicity risk assessment. Federal
Register 51(185):34006-34012. Available online at http://www.epa.gov/iris/backgr-d.htm.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-72 DRAFT—DO NOT CITE OR QUOTE
-------
U.S. EPA (Environmental Protection Agency). (1987) The total exposure assessment methodology (TEAM) study:
summary and analysis, Volume 1. EPA/600/6-87/002a.
U.S. EPA (Environmental Protection Agency). (1988) Recommendations for and documentation of biological values
for use in risk assessment. Prepared by the Environmental Criteria and Assessment Office, Office of Health and
Environmental Assessment, Cincinnati, OH for the Office of Solid Waste and Emergency Response, Washington,
DC; EPA 600/6-87/008. Available from: .
U.S. EPA (Environmental Protection Agency). (1991) Guidelines for developmental toxicity risk assessment.
Federal Register 56(234):63798-63826. Available from: .
U.S. EPA (Environmental Protection Agency). (1992) Draft report: A cross-species scaling factor for carcinogen
risk assessment based on equivalence of mg/kg3/4/Day; Notice. Federal Register 57(109): 24152-2424173. Friday,
June 5, 1992.
U.S. EPA (Environmental Protection Agency). (1994a) Interim policy for particle size and limit concentration issues
in inhalation toxicity studies. Federal Register 59(206):53799. Available from:
.
U.S. EPA (Environmental Protection Agency). (1994b) Methods for derivation of inhalation reference
concentrations and application of inhalation dosimetry. Office of Research and Development, Washington, DC;
EPA/600/8-90/066F. Available from: .
U.S. EPA (Environmental Protection Agency). (1995) The national public water systems supervision program. The
FY 1994 Compliance Report. Office of Water. EPA 812-R-94-001. Washington: U.S. Environmental Protection
Agency.
U.S. EPA (Environmental Protection Agency). (1995) Use of the benchmark dose approach in health risk
assessment. Risk Assessment Forum, Washington, DC; EPA/630/R-94/007. Available from:
.
U.S. EPA (Environmental Protection Agency). (1996) Guidelines for reproductive toxicity risk assessment. Federal
Register 61(212):56274-56322. Available from: .
U.S. EPA (Environmental Protection Agency). (1997) Exposure factors handbook, Volume I - General Factors.
National Center for Environmental Assessment, Office of Research and Development, Washington, DC,
EPA/600/P-95/002Fa
U.S. EPA (Environmental Protection Agency). (1997) Exposure factors handbook. EPA/600/P-95/002F
U.S. EPA (Environmental Protection Agency). (1998) Guidelines for neurotoxicity risk assessment. Federal
Register 63(93):26926-26954. Available from: .
U.S. EPA (Environmental Protection Agency). (1998) EPA data file. Office of Groundwater and Drinking Water.
Washington, D.C.; EPA.
U.S. EPA (Environmental Protection Agency). (2000a) Science policy council handbook: risk characterization.
Office of Science Policy, Office of Research and Development. Washington, DC; EPA 100-B-00-002. Available
from: .
U.S. EPA (Environmental Protection Agency). (2000b) Benchmark dose technical guidance document [external
review draft]. Risk Assessment Forum, Washington, DC; EPA/630/R-00/001. Available from:
.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-73 DRAFT—DO NOT CITE OR QUOTE
-------
U.S. EPA (Environmental Protection Agency). (2000c) Supplementary guidance for conducting for health risk
assessment of chemical mixtures. Risk Assessment Forum, Washington, DC; EPA/630/R-00/002. Available from:
.
U.S. EPA (Environmental Protection Agency). (2001) Trichloroethylene health risk assessment: synthesis and
characterization (external review draft). Office of Research and Development, National Center for Environmental
Assessment, Washington, DC, EPA/600/P-01/002A.
U.S. EPA (Environmental Protection Agency). (2002) A review of the reference dose and reference concentration
processes. Risk Assessment Forum, Washington, DC; EPA/630/P-02/0002F. Available from:
.
U.S. EPA (Environmental Protection Agency). (2002) OSWER draft guidance for evaluating the vapor intrusion to
indoor air pathway from groundwater and soils. Available online at
http://epa.gov/osw/hazard/correctiveaction/eis/vapor/guidance.pdf.
U.S. EPA (Environmental Protection Agency). (2002a) Health assessment of 1,3-butadiene. Office of Research and
Development. Washington, DC; EPA/600/P-98/001F.
U.S. EPA. (2003) Occurrence estimation methodology and occurrence findings report of the six-year review of
existing National Primary Drinking Water Regulations. EPA/815/R-03/006.
U.S. EPA (Environmental Protection Agency). (2004a) Risk assessment guidance for superfund. Volume I: Human
health evaluation manual (Part E, Supplemental guidance for dermal risk assessment). EPA/540/R/99/005.
U.S. EPA. (2004b) Estimated per capita water ingestion and body weight in the United States-an update.
EPA/822/R-00/001. Office of Water, Washington, DC.
U.S. EPA (Environmental Protection Agency). (2005) The HAPEM5 user's guide. Hazardous Air Pollutant
Exposure Model, Version 5. EPA Office of Air Quality Planning and Standards, Research Triangle Park, North
Carolina.
U.S. EPA (Environmental Protection Agency). (2005a) Guidelines for carcinogen risk assessment. Risk Assessment
Forum, Washington, DC; EPA/630/P-03/001B. Available online at http://www.epa.gov/iris/backgr-d.htm. ES,
U.S. EPA (Environmental Protection Agency). (2005b) Supplemental guidance for assessing susceptibility from
early-life exposure to carcinogens. Risk Assessment Forum, Washington, DC; EPA/630/R-03/003F. Available
online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Environmental Protection Agency). (2006a) Science policy council handbook: peer review. Third edition.
Office of Science Policy, Office of Research and Development, Washington, DC; EPA/ 100/B-06/002. Available
online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Environmental Protection Agency). (2006b) A framework for assessing health risk of environmental
exposures to children. National Center for Environmental Assessment, Washington, DC, EPA/600/R-05/093F.
Available online at http://cfpub.epa. gov/ncea/cfm/recordisplav.cfm?deid= 158363.
U.S. EPA (Environmental Protection Agency). (2007a) Technology transfer network -1999 National-Scale Air
Toxics Assessment. Availabe online at http://www.epa.gov/ttn/atw/natal999/nsata99.html.
U.S. EPA (Environmental Protection Agency). (2007b) EPA's air quality system database at the AirData website.
Office of Air Quality Planning & Standards. Research Triangle Park, NC. Available online at
http://www.epa.gov/air/data/index.html.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-74 DRAFT—DO NOT CITE OR QUOTE
-------
U.S. EPA (Environmental Protection Agency). (2008a) Child-specific exposure factors handbook. Washington, DC;
EPA/600/R-06/096F, 2008.
U.S. EPA (Environmental Protection Agency). (2008b) 1,1,2-Trichloro-l, 2, 2-trifluoroethane. IRIS (Integrated
Risk Information System). Available online at http://www.epa.gOv/IRIS/subst/0123.htm. [Accessed June 2, 2008]
U.S. EPA (Environmental Protection Agency). (2008c) Benchmark dose technical guidance (Inter-Agency Review
Draft).
U.S. EPA. (2008d). FACTOIDS: Drinking water and ground water statistics for 2008. EPA/816/K-08/004.
U.S. EPA (Environmental Protection Agency). (2009) Submission to Docket EPA-ORD-2008-0461 by Mr. M.
Partain, February 18, 2009, EPA-ORD-2008-0451-0021. Available online at
http://www.regulations.gov/search/Regs/home.html#docketDetail?R=EPA-HQ-ORD-2008-0461. accessed August
20, 2009.
USGS (U.S. Geologic Survey). (2006) Volatile organic compounds in the nation's ground water and drinking-water
supply wells. Circular 1292. Reston, VA.
Umezu, T; Yonemoto, J; Soma, Y; et al. (1997) Behavioral effects of trichloroethylene and tetrachloroethylene in
mice. Pharmacol Biochem Behav 58(3):665-671.
USAF (U.S. Air Force). (2004) Determination of partition coefficients for a mixture of volatile organic compounds
in rats and humans at different life stages. Air Force Research Laboratory, Wright-Patterson Air Force Base, OH.
AFRL-HE-WP-TR-2005-0012.
Vagnarelli, P; De Sario, A; De Carli, L. (1990) Aneuploidy induced by chloral hydrate detected in human
lymphocytes with the Y97 probe. Mutagenesis 5(6):591-592.
Vaidya, VS; Shankar, K; Lock, EA; et al. (2003a) Renal injury and repair following S-l, 2 dichlorovinyl-L-cysteine
administration to mice. Toxicol Appl Pharmacol 188:110-121.
Vaidya, VS; Shankar, K; Lock, EA; et al. (2003c) Molecular mechanisms of renal tissue repair in survival from
acute renal tubule necrosis: role of ERK1/2 pathway. Toxicol Pathol 31:604-618.
Valic, E; Waldhor, T; Konnaris, C; et al. (1997) Acquired dyschromatopsia in combined exposure to solvents and
alcohol. Int Arch Occup Environ Health 70(6):403-406.
Vamvakas, S; Bittner, D; Koster, U. (1993) Enhanced expression of the protooncogenes c-myc and c-fos in normal
and malignant renal growth. Toxicol Lett 67(l-3):161-72.
Vamvakas, S; Bruning, T; Thomasson, B; et al. (1998) Renal cell cancer correlated with occupational exposure to
trichloroethene. J Cancer Res Clin Oncol 124:374-382.
Vamvakas, S; Dekant, W; Berthold, K; et al. (1987) Enzymatic transformation of metcapturic acids derived from
halogenated alkenes to reactive mutagenic intermediates. Biochem Pharmacol 36(17):2741-2748.
Vamvakas, S; Dekant, W; Henschler, D. (1989) Assessment of unscheduled DNA synthesis in a cultured line of
renal epithelial cells exposed to cysteine S-conjugates of haloalkenes and haloalkanes. Mutat Res 222(4):329-335.
Vamvakas, S; Dekant, W; Schiffmann, D; et al. (1988b) Induction of unscheduled DNA synthesis and micronucleus
formation in Syrian hamster embryo fibroblasts treated with cysteine S-conjugates of chlorinated hydrocarbons.
Cell Biol Toxicol (4):393-403.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-75 DRAFT—DO NOT CITE OR QUOTE
-------
Vamvakas, S; Elfarra, AA; Dekant, W; et al. (1988a) Mutagenicity of amino acid and glutathione S-conjugates in
the Ames test. Mutat Res 206(1):83-90.
Vamvakas, S; Richter, H; Bittner, D. (1996) Induction of dedifferentiated clones of LLC-PK1 cells upon long-term
exposure to dichlorovinylcysteine. Toxicology 106(l-3):65—74.
van Bree, L; Commandeur, J; Lamberts, B; et al. (1990) Induction of drug metabolism enzymes by dihalogenated
biphenyls. J Biochem Toxicol 5:57-63.
VanDuuren, BL; Goldschmidt, BM; Loewengart, G; et al. (1979) Carcinogenicity of halogenated olefmic and
aliphatic hydrocarbons in mice. J Natl Cancer Inst 63(6): 1433-1439.
Van Hummelen, P; Kirsch-Volders, M. (1992) Analysis of eight known or suspected aneugens by the in vitro human
lymphocyte micronucleus test. Mutagenesis 7(6):447-455.
van Lieshout, EM; Knapen, MF; Lange, WP; et al. (1998) Localization of glutathione S-transferases alpha and pi in
human embryonic tissues at 8 weeks gestational age. Hum Reprod 13(5): 1380-1386.
van Lieshout, EM; Peters, WH. (1998) Age and gender dependent levels of glutathione and glutathione S-
transferases in human lymphocytes. Carcinogenesis 19(10): 1873-1875.
Vartiainen, T; Pukkala, E; Rienoja, T; et al. (1993) Population exposure to tri-and tetrachloroethene and cancer risk:
two cases of drinking water pollution. Chemosphere 27:1171-1181.
Veeramachaneni, DNR; Palmer, JS; Amann, RP. (2001) Long-term effects on male reproduction of early exposure
to common chemical contaminants in drinking water. Hum Reprod 16(5):979-987.
Verma, Y; Rana, SV. (2003) Gender differences in the metabolism of benzene, toluene and trichloroethylene in rat
with special reference to certain biochemical parameters. J Environ Biol 24(2): 135-140.
Vernon, RJ; Ferguson, RK. (1969) Effects of trichloroethylene on visual-motor performance. Arch Environ Health
18(6):894-900.
Vesterberg, O; Astrand, I. (1976) Exposure to trichloroethylene monitored by analysis of metabolites in blood and
urine. JOccupMed 18(4):224-226.
Vesterberg, O; Gorczak, J; Krasts, M. (1976) Exposure to trichloroethylene II. Metabolites in blood and urine.
Scand J Work Environ Health 4:212-219.
Vickers, AEM; Lucier, GW. (1996) Estrogen receptor levels and occupancy in hepatic sinusoidal endothelial and
Kupffer cells are enhanced by initiation with diethylnitroamine and promotion with 17 a-ethinylestradiol in rats.
Carcinogenesis 17(6): 1235-1242.
Vieira, I; Sonnier, M; Cresteil, T. (1996) Developmental expression of CYP2E1 in the human liver.
Hypermethylation control of gene expression during the neonatal period. Eur J Biochem 238(2):476-483.
Villaschi, S; Giovanetti, A; Lombardi, CC; et al. (1991) Damage and repair of mouse bronchial epithelium
following acute inhalation of trichloroethylene. Exp Lung Res 15:601-614.
Vineis, P; Miligi, L; Costantini, AS. (2007) Exposure to solvents and risk of non-Hodgkin lymphoma: clues on
putative mechanisms. Cancer Epidemiol Biomarkers Prev 16:381-384.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-76 DRAFT—DO NOT CITE OR QUOTE
-------
Von Tungeln, LS; Yi, P; Bucci, TJ; et al. (2002) Tumorigenicity of chloral hydrate, trichloroacetic acid,
trichloroethanol, malondialdehyde, 4-hydroxy-2-nonenal, crotonaldehyde, and acrolein in the B6C3F(1) neonatal
mouse. Cancer Lett 185(1): 13-9.
Voss, C; Zerban, H; Bannasch, P; et al. (2005) Lifelong exposure to di-(2-ethylhexyl)-phthalated induced tumors in
liver and testes of Sprague-Dawley rats. Toxicology 2006:359-371.
Voss, KA; Riley, R; Dunn, C; et al. (2006) The role of tumor necrosis factor alpha and the peroxisome proliferator-
activated receptor alpha in modulating the effects of fumonisin in mouse liver. Toxicology 222:165-174.
Wainfan, E; Poirier, LA. (1992) Methyl groups in carcinogenesis: effects on DNA methylation and gene expression.
Cancer Res 52(Suppl. 7):2071s-2077s.
Walgren, JE; Kurtz, DT; McMillan, JM. (2000a) Expression of PPAR(alpha) in human hepatocytes and activation
by trichloroacetate and dichloroacetate. Res Commun Mol Pathol Pharmacol 108:116-132.
Walgren, JE; Kurtz, DT; McMillan, JM. (2000b) The effect of the trichloroethylene metabolites trichloroacetate and
dichloroacetate on peroxisome proliferation and DNA synthesis in cultured human hepatocytes. Cell Biol Toxicol
16:257-273.
Wallace, LA; Pellizzari, E; Hartwell, T; et al. (1984) Personal exposure to volatile organic compounds. I. Direct
measurements in breathing-zone air, drinking water, food, and exhaled breath. Environ Res 35(l):293-319.
Walles, SA. (1986) Induction of single-strand breaks in DNA of mice by trichloroethylene and tetrachloroethylene.
Toxicol Lett 31:31-35.
Wang, G; Ansari, GA; Khan, MF. (2007a) Involvement of lipid peroxidation-derived aldehyde-protein adducts in
autoimmunity mediated by trichloroethene. J Toxicol Environ Health A 70:1977-1985.
Wang, G; Cai, P; Ansari, GA; et al. (2007b) Oxidative and nitrosative stress in trichloroethene-mediated
autoimmune response. Toxicology 229:186-193.
Wang, G; Konig, R; Ansari, GA; et al. (2008) Lipid peroxidation-derived aldehyde-protein adducts contribute to
trichloroethene-mediated autoimmunity via activation of CD4(+) T cells. Free Radic Biol Med44(7): 1475-1482.
Wang, G; Stacey, NH. (1990) Elevation of individual serum bile acids on exposure to trichloroethylene or a-
naphthylisothioncyanate. Toxicol and Appl Pharmacol 105:209-215.
Wang, JL; Chen, WL; Tsai, SY; et al. (2001) An in vitro model for evaluation of vaporous toxicity of
trichloroethylene and tetrachloroethylene to CHO-K1 cells. ChemBiol Interact 137(2): 139-154.
Wang, R; Zhang, Y; Lan, Q; et al. (2009) Occupational exposure to solvents and risk of non-Hodgkin lymphoma in
Connecticut women. Am J Epidemiol 169:176-185.
Wang, Z; Hall, SD; Maya, JF; et al. (2003) Diabetes mellitus increases the in vivo activity of cytochrome P450 2E1
in humans. Br J Clin Pharmacol 55(l):77-85.
Wang, G-Z; Takano, T; Tomita, K; et al. (1993) Interaction of trichloroethylene and isopropyl alcohol in the
perfused rat liver. Jpn J Hyg 48:1000-1005.
Ward, EM; Thun, MJ; Hannan, LM; et al. (2006) Interpreting cancer trends. Ann Ny Acad Sci 1076:29-53.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-77 DRAFT—DO NOT CITE OR QUOTE
-------
Ward, JM; Peters, JM; Perella, CM; et al. (1998) Receptor and nonreceptor-mediated organ-specific toxicity of di(2-
ethylhexyl)phthalate (DEHP) in peroxisome proliferator-activated receptor alpha-null mice. Toxicol Pathol
26(2):240-246.
Warr, TJ; Parry, EM; Parry, JM. (1993) A comparison of two in vitro mammalian cell cytogenetic assays for the
detection of mitotic aneuploidy using 10 known or suspected aneugens. Mutat Res 287(l):29-46.
Warren, DA; Graeter, LJ; Channel, SR; et al. (2006) Trichloroethylene, trichloroacetic acid, and dichloroacetic acid:
do they affect eye development in the Sprague-Dawley rat? Int J Toxicol 25(4):279-284.
Wartenberg, D; Reyner, D; Scott, CS. (2000) Trichloroethylene and cancer: epidemiologic evidence. Environ
Health Perspect 108(Suppl. 2): 161-176.
Waseem, M; Ali, M; Dogra, S; et al. (2001) Toxicity of trichloroethylene following inhalation and drinking
contaminated water. J Appl Toxicol 21(6):441-444.
Waskell, L. (1978) A study of the mutagenicity of anesthetics and their metabolites. Mutat Res 57:141-153.
Watanabe, S; Fukui, T. (2000) Suppressive effect of curcumin on trichloroethylene-induced oxidative stress. J Nutr
Sci Vitaminol (Tokyo) 46:230-234.
Watson, RE; Jacobson, CF; Williams, AL; et al. (2006) Trichloroethylene-contaminated drinking water and
congenital heart defects: a critical analysis of the literature. Reprod Toxicol 21:117-147.
Weiderpass, E; Pukkala, E; Kauppinen, T; et al. (1999) Breast cancer and occupational exposure in women in
Finland. Am J Ind Med 36:48-53.
Weiderpass, E; Vainion, H; Kauppinen, T; et al. (2003) Occupational exposure and gastrointestinal cancers among
Finnish women. J Occup Environ Med 45:305-315.
Weisel, CP; Jo, WK. (1996) Ingestion, inhalation, and dermal exposures to chloroform and trichloroethene from tap
water. Environ Health Perspect 104(1):48-51.
Weisenburger, DD. (1992) Pathological classification of non-Hodgkin's lymphoma for epidemiological studies.
Cancer Res 52:5456s-5464s.
Weiss, JR; Moysich, KB; Swede, H. (2005) Epidemiology of male breast cancer. Cancer Epidemiol Biomarkers
Prev 14:20-26.
Wells, GM; Schroth, W; Brauch, H; et al. (2009) Bilateral renal-cell carcinoma associated with anacquired VHL
mutation and long-term trichloroethylene exposure. Clin Nephrol 71:708-713.
Welzel, TM; McGlynn, KA; Hsing, AW; et al. (2006) Impact of classification of Hilar Cholangiocarcinomas
(Klatskin Tumros) on the incidence of intra- and extrahepatic cholangiocarcinoma in the United States. J Natl
Cancer Inst 98:873-875.
Wendling, O; Chambon, P; Mark, M. (1999) Retinoid X receptors are essential for early mouse development and
placentogenesis. Proc Nat Acad Sci USA 96:547-551.
Werner, H; Le Roith, D. (2000) New concepts in regulation and function of the insulin-like growth factors:
implications for understanding normal growth and neoplasia. Cell Mol Life Sci 57:932-942.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-78 DRAFT—DO NOT CITE OR QUOTE
-------
Werner, M; Birner, G; Dekant, W. (1995a) The role of cytochrome P4503A1/2 in the sex-specific sulfoxidation of
the hexachlorobutadiene metabolite, N-acetyl-S-(pentachlorobutadienyl)-L-cysteine in rats. Drug Metab Dispos
23(8):861-868.
Werner, M; Birner, G; Dekant, W. (1996) Sulfoxidation of mercapturic acids derived from tri- and tetrachloroethene
by cytochromes P450 3 A: a bioactivation reaction in addition to deacetylation and cysteine conjugate beta-lyase
mediated cleavage. Chem Res Toxicol 9(l):41-49.
Werner, M; Guo, Z; Birner, G; et al. (1995b) The sulfoxidation of the hexachlorobutadiene metabolite N-acetyl-S-
(l,2,3,4,4-pentachlorobutadienyl)-L-cysteine is catalyzed by human cytochrome P450 3 A enzymes. Chem Res
Toxicol 8(7):917-923.
West, GB; Woodruff, WH; Brown, JH. (2002) Allometric scaling of metabolic rate from molecules and
mitochondria to cells and mammals. Proc Natl Acad Sci USA 99:2473-2478.
Westergren, I; Kjellstrand, P; Linder, LE; et al. (1984) Reduction of brain specific gravity in mice prenatally
exposed to trichloroethylene. Toxicol Lett 23:223-226.
Wheeler, CW; Guenthener, TM. (1990) Spectroscoptic quantitation of cytochrome P-450 in human lung
microsomes. J Biochem Toxicol 5:1-4.
Wheeler, CW; Wrighton, SA; Guenthner, TM. (1992) Detection of human lung cytochromes P450 that are
immunochemically related to cytochrome P450IIE1 and cytochrome P450IIIA. Biochem Pharmacol 44:183-186.
White, AE; Eger, El; Wolff, S; et al. (1979) Sister chromatid exchange induced by inhaled anesthetics.
Anesthesiology 50:426-430.
White, JF; Carlson, GP. (1981) Epinephrine-induced cardiac arrhythmias in rabbits exposed to trichloroethylene:
potentiationby ethanol. Toxicol Appl Pharmacol 60(3):466-471.
White, KL; David, DW; Butterworth, LF; et al .(2000) Assessment of autoimmunity-inducing potential using the
brown Norway rat challenge model. Toxicol Lett 112-113:443-451.
White, RF; Feldman, RG; Eviator, II; et al. (1997) Hazardous waste and neurobehavioral effects: a developmental
perspective. Environ Res 73(1-2): 113-124.
WHO (World Health Organization). (1985) Environmental health criteria: trichloroethylene. Geneva, Switzerland:
World Health Organization.
Wideroff, L; Gridley, G; Mellemkjaer, L; et al. (1997) Cancer incidence in a population-based cohort of patients
hospitalized with diabetes mellitus in Denmark. J Natl Cancer Inst 89:1360-1365.
Wiesenhutter, B; Selinski, S; Golka, K; et al. (2007) Re-assessment of the influence of polymorphisms of phase-II
metabolic enzymes on renal cell cancer risk of trichloroethylene-exposed workers. Int Arch Occup Environ Health
81:247-251.
Wigle, DT; Turner, MC; Gomes, J; et al. (2008) Role of hormonal and other factors in human prostate cancer. J
Toxicol Environ Health B CritRev 11:242-259.
Wilcosky, TC; Checkoway, H; Marshall, EG; et al. (1984) Cancer mortality and solvent exposures in the rubber
industry. Am Ind Hyg Assoc J 45(12):809-811.
Williams, F; Meenagh, A; Darke, C; et al. (2001) Analysis of the distribution of HLA-B alleles in populations from
five continents. Hum Immunol 62(6):645-650.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-79 DRAFT—DO NOT CITE OR QUOTE
-------
Williams, AL; DeSesso, JM. (2008) Trichloroethylene and ocular malformations: analysis of extant literature. Int J
Toxicol 27:81-95.
Williams, P; Benton, L; Warmerdam, J; et al. (2002) Comparative risk analysis of six volatile organic compounds in
California drinking water. Environ Sci Technol 36:4721-4728.
Wilson, RT; Donahue, M; Gridley, G; et al. (2008) Shared occupational risks for transitional cell cancer of the
bladder and renal pelvis among men and women in Sweden. Am J Ind Med 51:83-99.
Wilson, JG. (1965) Methods for administering agents and detecting malformations in experimental animals. In:
Wilson, JG; Warkany, T; eds. Teratology principles and methods. (Chicago, IL: The University of Chicago Press ;
pp. 262-267.
Windemuller, FJ; Ettema, JH. (1978) Effects of combined exposure to trichloroethylene and alcohol on mental
capacity. Int Arch Occup Environ Health 41(2):77-85.
Windham, GC; Shusterman, D; Swan, SH; et al. (1991) Exposure to organic solvents and adverse pregnancy
outcome. Am JIndMed20(2):241-259.
Windham, GC; Zhang, L; Gunier, R; et al. (2006) Autism spectrum disorders in relation to distribution of hazardous
air pollutants in the San Francisco Bay area. Environ Health Perspect 114(9): 1438-1444.
Winneke, G. (1987) Acute behavioral effects of exposure to some organic solvents-psychophysiological aspects.
Acta Neurol Scand 32:117-129.
Withey, JR; Karpinski, K. (1985) The fetal distribution of some aliphatic chlorinated hydrocarbons in the rat after
vapor phase exposure. Biol Res Pregnancy Perinatal 6(2):79-88.
Withey, JR; Collins, BT; Collins, PG. (1983) Effect of vehicle on the pharmacokinetics and uptake of four
halogenated hydrocarbons from the gastrointestinal tract of the rat. J Appl Toxicol 3(5):249-253.
Wolf, DC; Goldsworthy, TL; Janszen, DB; et al. (2000) Promotion by sodium barbital induces early development
but does not increase the multiplicity of hereditary renal tumors in Eker rats. Carcinogenesis 21 (8): 1553-1558.
Wolff, DL; Siegmund, R. (1978) [The circadian-dependent effect of trichloroethylene on spontaneous locomotor
activity and of tetrachloroethane on mortality in mice.] Biol Zentralbl 97(3):345-351.(German)
Wolfgang, GH; Gandolfi, AJ; Stevens, JL; et al. (1989b) In vitro and in vivo nephrotoxicity of the L and D isomers
of S-(l,2-dichlorovinyl)-cysteine. Toxicology 58:33-42.
Wolfgang, GHI; Gandolfi, AJ; Stevens, JL; et al. (1989) jV-acetyl S'-(l,2-dichlorovinyl)--cysteine produces a similar
toxicity to ^-(l^-dichloroviny^-L-cysteine in rabbit renal slices: differential transport and metabolism. Toxicol
ApplPharm 101:205-219.
Woods, CG; Burns, AM; Bradford, BU; et al. (2007) Wy-14,643-induced cell proliferation and oxidative stress in
mouse liver are independent of NADPH oxidase. Toxicol Sci 98(2): 366-374.
Woolhiser, MR; Krieger, SM; Thomas, J; et al. (2006) Trichloroethylene (TCE): Immunotoxicity potential in CD
rats following a 4-week vapor inhalation exposure. Dow Chemical Company, Toxicology & Environmental
Research and Consulting, Midland, MI, Study ID 031020, July 5, 2006, unpublished.
Wright, PF; Thomas, WD; Stacey, NH. (1991) Effects of trichloroethylene on hepatic and splenic lymphocytotoxic
activities in rodents. Toxicology 70:231-242.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-80 DRAFT—DO NOT CITE OR QUOTE
-------
Wu, C; Schaum, J. (2000) Exposure assessment of trichloroethylene. Environ Health Perspect 108 (suppl
2):359-363.
Wu, C; Schaum, J. (2001) Sources, emissions and exposures for trichloroethylene (TCE) and related chemicals.
Washington, DC: US Environmental Protection Agency, EPA/600/R-00/099.
Wu, KL; Berger, T. (2007) Trichloroethylene metabolism in the rat ovary reduces oocyte fertilizibility. Chemico-
Biological Interactions 170:20-30.
Wu, KL; Berger, T. (2008) Ovarian gene expression is stable after exposure to trichloroethylene. Toxicol Lett
177:59-65.
Xu, W; Adler, ID. (1990) Clastogenic effects of known and suspect spindle poisons studied by chromosome analysis
in mouse bone marrow cells. Mutagenesis 5(4):371-374.
Xu, X; Yang, R; Wu, N; et al. (2009) Severe hypersensitivity dermatitis and liver dysfunction induced by
occupational exposure to trichloroethylene. Ind Health 47:107-112.
Xu, H; Tanphaichitr, N; Forkert, P-G; et al. (2004) Exposure to trichloroethylene and its metabolites causes
impairment of sperm fertilizing ability in mice. Toxicol Sci 82:590-597.
Yamamura, K; Ikeda, I; Sadamoto, T; et al. (1983) Effects of trichloroethylene exposure on hearing. An
investigation of cochlear microphonics and action potential of the guinea pig. Eur J Appl Physiol Occup Physiol
52:47-50.
Yang, Q; Ito, S; Gonzalez, FJ. (2007) Hepatocyte-restricted constitutive activation of PPAR alpha induces
hepatoproliferation but not hepatocarcinogenesis. Carcinogenesis 28(6): 1171-1177.
Yang, Y; Heng, ZC. (2006) Study on rat's DNA damage of p53 induced by chlorinated acetic acids of drinking
water disinfection by-products (article in Chinese). Wei Sheng Yan Jiu. 35(6):683-685.
Yauck, JS; Malloy, ME; Blair, K; et al. (2004) Proximity of residence to trichloroethylene-emitting sites and
increased risk of offspring congenital heart defects among older women. Birth Defects Res A Clin Mol Teratol
70(10):808-814.
Yeh, MM; Larson, AM; Campbell, JS; et al. (2007) The expression of transforming growth factor -a in cirrhosis,
dysplastic nodules, and hepatocellular carcinoma; an immunohistochemical study in 70 cases. Am J Surg Pathol
31(5):681-689.
Yeung, RS; Xiao, GH; Everitt, JI; et al. (1995) Allelic loss at the tuberous sclerosis 2 locus in spontaneous tumors in
the Eker rat. Mol Carcinog 14(l):28-36.
Yoon, HS; Monks, TJ; Walker, CL; et al. (2001) Transformation of kidney epithelial cells by a quinol thioether via
inactivation of the tuberous sclerosis-2 tumor suppressor gene. Mol Carcinog 3 l(l):37-45.
Yoon, JS; Mason, JM; Valencia, R; et al. (1985) Chemical mutagenesis testing in Drosophila. IV. Results of 45
coded compounds tested for the National Toxicology Program. Environ Mutagen 7(3):349-367.
Yoon, M; Madden, MC; Barton, HA. (2007) Extrahepatic metabolism by CYP2E1 inPBPK modeling of lipophilic
volatile organic chemicals: impacts on metabolic parameter estimation and prediction of dose metrics. J Toxicol
Environ Health A 70(18):1527-1541.
Yoshida, M; Fukabori, S; Kara, K; et al. (1996) Concentrations of trichloroethylene and its metabolites in blood and
urine after acute poisoning by ingestion. Hum Exp Toxicol 15:254-258.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-81 DRAFT—DO NOT CITE OR QUOTE
-------
Yoshioka, T; Krauser, JA; Guengerich, FP. (2002) Microsomal oxidation of tribromoethylene and reactions of
tribromoethylene oxide. Chem Res Toxicol 15(11): 1414-1420.
Yu, MC; Garabrant, DH; Peters, JM; et al. (1988) Tobacco, alcohol, diet, occupation, and carcinoma of the
esophagus. Cancer Res 48:3843-3848.
Yu, KO; Barton, HA; Mahle, DA; et al. (2000) In vivo kinetics of tricholoracetate in male Fischer 344 rats. Toxicol
Sci 54:302-311.
Yu, S; Rao, S; Reddy, JK. (2003) Peroxisome proliferator-activated receptors, fatty acid oxidation, steatohepatitis
and hepatocarcinogenesis. Curr Mol Med 3:561-572.
Yuan, JM; Castelao, JE; Gago-Dominguez, M; et al. (1998) Tobacco use in relation to renal cell carcinoma. Cancer
Epidemiol Biomarkers Prev 7(5):429-33.
Yue, F; Huang, HL; Huang, JX; et al. (2007) HLA-DM polymorphism and risk of trichloroethylene induced
medicamentosa-like dermatitis. Biomed Environ Sci 20(6):506-511.
Zandi-Nejad, K; Eddy, AA; Glassock, RJ; et al. (2004) Why is proteinuria an ominous biomarker of progressive
kidney disease? Kidney Int 66:876-89.
Zanelli, U; Puccini, P; Acerbi, D; et al. (1996) Induction of peroxisomal beta-oxidation and P-450 4°-dependent
activities by pivalic and trichloroacetic acid in rat liver and kidney. Arch Toxicol 70:145-149.
Zeise, L; Wilson, R; Crouch, EA. (1987) Dose-response relationships for carcinogens: a review. Environ Health
Perspect 73:259-306.
Zenick, H; Blackburn, K; Hope, E; et al. (1984) Effects of trichloroethylene exposure on male reproductive function
in rats. Toxicology 31(3-4):237-50.
Zeppa, P; Benincasa, G; Troncone, G; et al. (1998) Retrospective evaluation of DNA ploidy of hepatocarcinoma on
cytological samples. Diagn Cytopathol 19(5):323-329.
Zhang, Y; Cantor, KP; Lynch, CF; et al. (2004) A population-based case-control study of occupation and renal cell
carcinoma risk in Iowa. J Occup Environ Med 46:235-240.
Zhang, G; Stevens, JL. (1989) Transport and activation of ^-(l^-dichloroviny^-L-cysteine andA^-acetyl-5'-(l,2-
dichlrovinyl)-L-cysteine in rat kidney proximal tubules. Toxicol Appl Pharm 100:51-61.
Zhao, Y; Krishnadasan, A; Kennedy, N; et al. (2005) Estimated effects of solvents and mineral oils on cancer
incidence and Mortality in a cohort of aerospace workers. Am J Ind Med 48:249-258.
Zhao, Y; Krishnadasan, A; Kennedy, N; et al. (2005) Estimated effects of solvents and mineral oils on cancer
incidence and mortality in a cohort of aerospace workers. Am J Ind Med 48:249-258.
Zhou, YC; Waxman, DJ. (1998) Activation of peroxisome proliferator-activated receptors by chlorinated
hydrocarbons and endogenous steroids. Environ Health Perspect 106(Suppl. 4):983-988 .
Zhu, J; Newhook, R; Marro, L; et al. (2005) Selected volatile organic compounds in residential air in the city of
Ottawa, Canada. Environ Sci Technol 39(11):3964-3971.
Zielinska, B; Fujita, E; Sagebiel, J; et al. (1998) Arizona hazardous air pollutants monitoring program. J Air Waste
Manage Assoc 48:1038-1050.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-82 DRAFT—DO NOT CITE OR QUOTE
-------
Zielinski, A. (1973) General health state of women professionally exposed to trichloroethylene vapours. Med Pr
24:263-271.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 R-83 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX A
PBPK Modeling of TCE and Metabolites
Detailed Methods and Results
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix: PBPK Modeling of TCE and Metabolites—Detailed Methods
and Results
LIST OF TABLES A-iv
LIST OF FIGURES A-v
APPENDIX A. PBPK MODELING OF TCE AND METABOLITES-DETAILED
METHODS AND RESULTS A-l
A. 1. THE HIERARCHICAL BAYESIAN APPROACH TO CHARACTERIZING
PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL
UNCERTAINTY AND VARIABILITY A-l
A.2. EVALUATION OF THE HACK ET AL. (2006) PHYSIOLOGICALLY
BASED PHARMACOKINETIC (PBPK) MODEL A-4
A.2.1. Convergence A-4
A.2.2. Evaluation of Posterior Distributions for Population Parameters A-6
A.2.3. Comparison of Model Predictions With Data A-7
A.2.3.1. Mouse Model A-8
A.2.3.2. Rat Model A-15
A.2.3.3. Human model A-23
A.3. PRELIMINARY ANALYSIS OF MOUSE GAS UPTAKE DATA:
MOTIVATION FOR MODIFICATION OF RESPIRATORY METABOLISM A-33
A.4. DETAILS OF THE UPDATED PHYSIOLOGICALLY BASED
PHARMACOKINETIC (PBPK) MODEL FOR TRICHLOROETHYLENE
(TCE) AND ITS METABOLITES A-38
A.4.1. Model Parameters and Baseline Values A-38
A.4.2. Statistical Distributions for Parameter Uncertainty and Variability A-46
A.4.2.1. Initial Prior Uncertainly in Population Mean Parameters A-46
A.4.2.2. Interspecies Scaling to Update Selected Prior Distributions
in the Rat and Human A-46
A.4.2.3. Population Variance: Prior Central Estimates and Uncertainty A-56
A.4.2.4. Prior distributions for Residual Error Estimates A-60
A. 5. RESULTS OF UPDATED PHYSIOLOGICALLY BASED
PHARMACOKINETIC (PBPK) MODEL A-63
A.5.1. Convergence and Posterior Distributions of Sampled Parameters A-63
A.5.2. Comparison of Model Predictions With Data A-63
A.5.2.1. Mouse Model A-63
A.5.2.2. Rat Model A-63
A.5.2.3. Human Model A-74
A.6. EVALUATION OF RECENTLY PUBLISHED TOXICOKINETIC DATA A-74
A.6.1. TCE Metabolite Toxicokinetics in Mice: Kim et al. (2009) A-74
A.6.2. TCE Toxicokinetics in Rats: Liu et al. (2009) A-78
A.6.3. TCA Toxicokinetics in Mice and Rats: Mahle et al. (2001) and Green
(2003a, b) A-79
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-ii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
A.7. UPDATED PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL
CODE A-81
A.8. REFERENCES A-106
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-iii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
A-l. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in mice .... A-10
A-2. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in rats A-16
A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in
humans A-24
A-4. PBPK model parameters, baseline values, and scaling relationships A-39
A-5. Uncertainty distributions for the population mean of the PBPK model parameters A-47
A-6. Updated prior distributions for selected parameters in the rat and human A-52
A-7. Uncertainty distributions for the population variance of the PBPK model
parameters A-57
A-8. Measurements used for calibration A-61
A-9. Posterior distributions for mouse PBPK model population parameters A-64
A-10. Posterior distributions for mouse residual errors A-66
A-11. Posterior distributions for rat PBPK model population parameters A-67
A-12. Posterior distributions for rat residual errors A-69
A-13. Posterior distributions for human PBPK model population parameters A-71
A-14. Posterior distributions for human residual errors A-73
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-iv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
A-l. Hierarchical population statistical model for PBPK model parameter uncertainty
and variability (see Gelman etal., 1996) A-2
A-2. Schematic of how posterior predictions were generated for comparison with
experimental data A-8
A-3. Limited optimization results for male closed chamber data from Fisher et al.
(1991) without (top) and with (bottom) respiratory metabolism A-36
A-4. Limited optimization results for female closed chamber data from Fisher et al.
(1991) without (top) and with (bottom) respiratory metabolism A-37
A-5. Respiratory metabolism model for updated PBPK model A-38
A-6. Updated hierarchical structure for rat and human models A-62
A-7. Comparison of best-fitting (out of 50,000 posterior samples) PBPK model
prediction and Kim et al. (2009) TCA blood concentration data for mice
gavagedwith2,140mg/kgTCE A-75
A-8. Comparison of best-fitting (out of 50,000 posterior samples) PBPK model
prediction and Kim et al. (2009) DCVG blood concentration data for mice
gavagedwith2,140mg/kgTCE A-76
A-9. PBPK model predictions for the fraction of intake undergoing GSH conjugation
in mice continuosly exposed orally to TCE A-77
A-10. PBPK model predictions for the fraction of intake undergoing GSH conjugation
in mice continuosly exposed via inhalation to TCE A-78
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-v DRAFT—DO NOT CITE OR QUOTE
-------
1 APPENDIX A. PBPK MODELING OF TCE AND METABOLITES-DETAILED
2 METHODS AND RESULTS
O
4
5 A.l. THE HIERARCHICAL BAYESIAN APPROACH TO CHARACTERIZING
6 PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODEL
7 UNCERTAINTY AND VARIABILITY
8 The Bayesian approach for characterizing uncertainty and variability in PBPK model
9 parameters, used previously for trichloroethylene (TCE) in Bois (2000a, b) and Hack et al.
10 (2006), is briefly described here as background. Once a physiologically based pharmacokinetic
11 (PBPK) model structure is specified, characterizing the model reduces to calibrating and making
12 inferences about model parameters. The use of least-squares point estimators is limited by the
13 large number of parameters and small amounts of data. The use of least-squares estimation is
14 reported after imposing constraints for several parameters (Fisher, 2000; Clewell et al., 2000).
15 This is reasonable for a first estimate, but it is important to follow-up with a more refined
16 treatment. This is implemented by a Bayesian approach to estimate posterior distributions on the
17 unknown parameters, a natural choice, and almost a compulsory consequence given the large
18 number of parameters and relatively small amount of data, and given the difficulties of
19 frequentist estimation in this setting.
20 As described by Gelman et al. (1996), the Bayesian approach to population PBPK
21 modeling involves setting up the overall model in several stages. A nonlinear PBPK model, with
22 predictions denoted/ describes the absorption, distribution, metabolism, and excretion of a
23 compound and its metabolites in the body. This model depends on several, usually known,
24 parameters such as measurement times t, exposure E, and measured covariates (p. Additionally,
25 each subject /' in a population has a set of unmeasured parameters 0,. A random effects model
26 describes their population variability P(0, u, E2), and a prior distribution /"(a, E2) on the
27 population mean [j, and covariance E2 (often assumed to be diagonal) incorporates existing
28 scientific knowledge about them. Finally, a "measurement error" model P(y |_/[0, (p, E, t], a2)
29 describes deviations (with variance a2) between the data^ and model predict!ons/(which of
30 course depends on the unmeasured parameters 0, and the measured parameters t, E, and (p). This
31 "measurement error" level of the hierarchical model typically also encompasses intraindividual
32 variability as well as model misspecification, but for notational convenience we refer to it here as
33 "measurement error." Because these other sources of variance are lumped into a single
34 "measurement error," a prior distribution of its variance a2 must be specified even if the actual
35 analytic measurement error is known. All these components are illustrated graphically in
36 Figure A-1.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-1 DRAFT—DO NOT CITE OR QUOTE
-------
1
2 Figure A-l. Hierarchical population statistical model for PBPK model
3 parameter uncertainty and variability (see Gelman et al., 1996). Square nodes
4 denote fixed or observed quantities; circle notes represent uncertain or unobserved
5 quantities, and the nonlinear model outputs are denoted by the inverted triangle.
6 Solid arrows denote a stochastic relationship represented by a conditional
7 distribution [A-^B means B ~ P(B\A)], while dashed arrows represent a function
8 relationship [B =f[A)]. The population consists of groups (or subjects) /', each of
9 which undergoes one or more experiments y with exposure parameters Ey with
10 dataytj collected at times %. The PBPK model produces outputs/y for comparison
11 with the datayy. The difference between them ("measurement error") has
12 variance a2, with a fixed prior distribution Pr, which in this case is the same for
13 the entire population. The PBPK model also depends on measured covariates (j),
14 (e.g., body weight) and unobserved model parameters 9, (e.g., VMAX)- The
15 parameters 9, are drawn from a population with mean (j, and variance Z2, each of
16 which is uncertain and has a prior distribution assigned to it.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 The posterior distribution for the unknown parameters is obtained in the usual manner by
2 multiplying (1) the prior distribution for the population mean and variance and the
3 "measurement" error P(\ji, E2) /"(a2), (2) the population distribution for the individual parameters
4 ,P(0 u, E2), and (3) the likelihood P(y 0, a2), where for notational convenience, the dependence
5 on/ cp, E, and t (which are taken as fixed for a given dataset) is dropped:
6
7 P(Q, u, E2, o2 | y) oc P(p, E2) /"(a2) P(0 | u, E2) P(y 0, a2) (Eq. A-l)
8
9 Here, each subject's parameters 0, have the same sampling distribution (i.e., they are
10 independently and identically distributed), so their joint prior distribution is
11
12 P(0 u, E2) = n/=i...» P(Qi I u, 22) (Eq. A-2)
13
14 Different experiments7 = !...«,- may have different exposure and different data collected and
15 different time points. In addition, different types of measurements k= \...n^ (e.g., TCE blood,
16 TCE breath, trichloroacetic acid [TCA] blood, etc.) may have different errors, but errors are
17 otherwise assumed to be iid. Since the individuals are treated as independent given Q\...n, the
18 total likelihood function is simply
19
Of! EV-ii I fl ^2\ — TT TT TT TT EV-n I fl r* ^ i ^ (T!n A T\
•^U r(y \ 0, G ) — [[j=l...n [\j'=l...nij IlA=l...m Y\l=\...Nijk f \Jijkl \ o/, Ofe , Ijjkl) wl- A~JJ
21
22 where n is the number of subjects, ntj is the number of experiments in that subject, m is the
23 number of different types of measurements, Nyk is the number (possibly 0) of measurements
24 (e.g., time points) for subject / of type k in experiment j, and %« are the times at which
25 measurements for individual /' of type k were made in experiment/
26 Given the large number of parameters, complex likelihood functions, and nonlinear
27 PBPK model, Markov chain Monte Carlo (MCMC) simulation was used to generate samples
28 from the posterior distribution. An important practical advantage of MCMC sampling is the
29 ability to implement inference in nearly any probability model and the possibility to report
30 inference on any event of interest. MCMC simulation was introduced by Gelfand and Smith
31 (1990) as a generic tool for posterior inference. See Gilks et al. (1996) for a review. In addition,
32 because many parameters are allowed to vary simultaneously, the local parameter sensitivity
33 analyses often performed with PBPK models (in which the changes in model predictions are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 assessed with each parameter varied by a small amount) are unnecessary.l In the context of
2 PBPK models, the MCMC simulation can be carried out as described by Hack et al. (2006). The
3 simulation program MCSim (version 5.0.0) was used to implement MCMC posterior simulation,
4 with analysis of the results performed using the R statistical package. Simulation-based
5 parameter estimation with MCMC posterior simulation gives rise to an additional source of
6 uncertainty. For instance, averages computed from the MCMC simulation output represent the
7 desired posterior means only asymptotically, in the limit as the number of iterations goes to
8 infinity. Any implementation needs to include a convergence diagnostic to judge practical
9 convergence. The potential scale-reduction-factor convergence diagnostic R of Gelman et al.
10 (1996) was used here, as it was in Hack et al. (2006).
11
12 A.2. EVALUATION OF THE HACK ET AL. (2006) PHYSIOLOGICALLY BASED
13 PHARMACOKINETIC (PBPK) MODEL
14 U.S. Environmental Protection Agency (U.S. EPA) obtained the original model code for
15 the version of the TCE PBPK model published in Hack et al. (2006) and conducted a detailed
16 evaluation of the model, focusing on the following areas: convergence, posterior estimates for
17 model parameters, and comparison of model predictions with in vivo data.
18
19 A.2.1. Convergence
20 As noted in Hack et al. (2006), the diagnostics for the MCMC simulations (3 chains of
21 length 20,000-25,000 for each species) indicated that additional samples might further improve
22 convergence. A recent analysis of tetrachloroethylene pharmacokinetics indicated the need to be
23 especially careful in ensuring convergence (Chiu and Bois, 2006). Therefore, the number of
24 MCMC samples per chain was increased to 75,000 for rats (first 25,000 discarded) and 175,000
25 for mice and humans (first 75,000 discarded). Using these chain lengths, the vast majority of the
26 parameters had potential scale reduction factors R < 1.01, and all population parameters had
27 R < 1.05, indicating that longer chains would be expected to reduce the standard deviation (or
28 other measure of scale, such as a confidence interval) of the posterior distribution by less than
29 this factor (Gelman et al., 2004).
1 In particular, local sensitivity analyses are typically used to assess the impact of alternative parameter estimates on
model predictions, inform experimental design, or assist prioritizing risk assessment research. Only the first purpose
is relevant here; however, the full uncertainty and variability analysis allows for a more comprehensive assessment
than can be done with sensitivity analyses. Separately, such analyses could be done to design experiments and
prioritize research that would be most likely to help reduce the remaining uncertainties in TCE toxicokinetics, but
that is beyond the scope of this assessment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 In addition, analysis of autocorrelation within chains using the R-CODA package
2 (Plumber et al., 2008) indicated that there was significant serial correlation, so additional
3 "thinning" of the chains was performed in order to reduce serial correlations. In particular, for
4 rats, for each of three chains, every 100th sample from the last 50,000 samples was used; and for
5 mice and humans, for each of three chains, every 200th sample from the last 100,000 samples
6 was used. This thinning resulted in a total of 1,500 samples for each species available for use for
7 posterior inference.
8 Finally, an evaluation was made of the "convergence" of dose metric predictions—that is,
9 the extent to which the standard deviation or confidence intervals for these predictions would be
10 reduced with additional samples. This is analogous to a "sensitivity analysis" performed so that
11 most effort is spent on parameters that are most influential in the result. In this case, the purpose
12 is to evaluate whether one can sample chains only long enough to ensure convergence of
13 predictions of interest, even if certain more poorly identified parameters take longer chains to
14 converge. The motivation for this analysis is that for a more complex model, running chains
15 until all parameters have R < 1.01 or 1.05 may be infeasible given the available time and
16 resource. In addition, as some of the model parameters had prior distributions derived from
17 "visual fitting" to the same data, replacing those distributions with less informative distributions
18 (in order to reduce bias from "using the same data twice") may require even longer chains for
19 convergence.
20 Indeed, it was found that ^-values for dose metric predictions approached one more
21 quickly than PBPK model input parameters. The most informative simulations were for mice,
22 which converged the slowest and, thus, had the most potential for convergence-related error.
23 Results for rats could not be assessed because the model converged so rapidly, and results for
24 humans were similar to those in mice, though the deviations were all less because of the more
25 rapid convergence. In the mouse model, after 25,000 iterations, many PBPK model parameters
26 had ^-values >2, with more than 25% greater than 1.2. However, all dose metric predictions had
27 R < 1.4, with the more than 96% of then <1.2 and the majority of them <1.01. In addition, when
28 compared to the results of the last 100,000 iterations (after the total of 175,000 iterations), more
29 than 90% of the medians estimates shifted by less than 20%, with the largest shifts less than 40%
30 (for glutathione [GSH] metabolism dose metrics, which had no relevant calibration data). Tail
31 quantiles had somewhat larger shifts, which was expected given the limited number of samples
32 in the tail, but still more than 90% of the 2.5 and 97.5 percentile quantiles had shifts of less than
33 40%. Again, the largest shifts, on order of 2-fold, were for GSH-related dose metrics that had
34 high uncertainty, so the relative impact of limited sample size is small.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-5 DRAFT—DO NOT CITE OR QUOTE
-------
1 Therefore, the additional simulations performed in this evaluation, with 3- to 7-fold
2 longer chains, did not result in much change in risk assessment predictions from the original
3 Hack et al. (2006) results. Thus, assessing prediction convergence appears sufficient for
4 assessing convergence of the TCE PBPK model for the purposes of risk assessment prediction.
5
6 A.2.2. Evaluation of Posterior Distributions for Population Parameters
7 Posterior distributions for the population parameters were first checked for whether they
8 appeared reasonable given the prior distributions. Inconsistency between the prior and posterior
9 distributions may indicate an insufficiently broad prior distribution (i.e., overconfidence in their
10 specification), a mis-specification of the model structure, or an error in the data. Parameters that
11 were flagged for further investigation were those for which the interquartile ranges (intervals
12 bounded by the 25th and 75th percentiles) of the prior and posterior distributions did not overlap.
13 In addition, lumped metabolism and clearance parameters for TCA, trichloroethanol (TCOH),
14 and trichloroethanol-glucuronide conjugate (TCOG) were checked to make sure that they
15 remained physiological—e.g., metabolic clearance was not more than hepatic blood flow and
16 urinary clearance not more than kidney blood flow (constraints that were not present in the Hack
17 et al., 2006 priors).
18 In mice, population mean parameters that had lack of overlap between priors and
19 posteriors included the affinity of oxidative metabolism (InKM), the TCA plasma-blood
20 concentration ratio (InTCAPlas), the TCE stomach to duodenum transfer coefficient (InKTSD),
21 and the urinary excretion rates of TCA and TCOG (InkUrnTCAC and InkUrnTCOGC). For KM,
22 this is not unexpected, as previous investigators have noted inconsistency in the KM values
23 between in vitro values (upon which the prior distribution was based) and in vivo values derived
24 from oral and inhalation exposures in mice (Abbas and Fisher, 1997; Greenberg et al., 1999).
25 For the other mean parameters, the central estimates were based on visual fits, without any other
26 a priori data, so it is reasonable to assume that the inconsistency is due to insufficiently broad
27 prior distributions. In addition, the population variance for the TCE absorption coefficient from
28 the duodenum (kAD) was rather large compared to the prior distribution, likely due to the fact
29 that oral studies included TCE in both oil and aqueous solutions, which are known to have very
30 different absorption properties. Thus, the larger population variance was required to
31 accommodate both of them. Finally, the estimated clearance rate for glucurondiation of TCOH
32 was substantially greater than hepatic blood flow. This is an artifact of the one-compartment
33 model used for TCOH and TCOG, and suggests that first pass effects are important for TCOH
34 glucurondiation. Therefore, the model would benefit from the additional of a separate liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-6 DRAFT—DO NOT CITE OR QUOTE
-------
1 compartment so that first pass effects can be accounted for, particularly when comparing across
2 dose-routes.
3 In rats, the only population mean or variance parameter for which the posterior
4 distribution was somewhat inconsistent with the prior distribution was the population mean for
5 the InKM. While the interquartile regions did not overlap, the 95 percentile regions did, so the
6 discordance was relatively minor. However, as with mice, the estimated clearance rate for
7 glucurondiation of TCOH was substantially greater than hepatic blood flow.
8 In humans, some of the chemical-specific parameters for which priors were established
9 using visual fits had posterior distributions that were somewhat inconsistent, including the
10 oxidative split between TCA and TCOH, biliary excretion of TCOG (InkBileC), and the TCOH
11 distribution volume (VBodC). More concerning was the fact that the posterior distributions for
12 several physiological volumes and flows were rather strongly discordant with the priors and/or
13 near their truncation limits, including gut, liver, and slowly perfused blood flow, the volumes of
14 the liver and rapidly perfused compartments. In addition, a number of tissue partition
15 coefficients were somewhat inconsistent with their priors, including those for TCE in the gut,
16 rapidly perfused, and slowly perfused tissues, and TCA in the body and liver. Finally, a number
17 of population variances (for TCOH clearance [C1TCOHC], urinary excretion of TCOG
18 [kUrnTCOGC], ventilation-perfusion ratio [VPR], cardiac output [QCC], fat blood flow and
19 volume [QFatC and VFatC], and TCE blood-air partition coefficient [PB])were somewhat high
20 compared to their prior distributions, indicating much greater population variability than
21 expected.
22
23 A.2.3. Comparison of Model Predictions With Data
24 A schematic of the comparisons between model predictions and data are shown in
25 Figure A-2. In the hierarchical population model, group-specific parameters were estimated for
26 each dataset used in calibrating the model (posterior group-specific 0, in Figure A-2). Because
27 these parameters are in a sense "optimized" to the experimental data themselves, the group-
28 specific predictions (posterior group-specific ytj in Figure A-2) using these parameters should be
29 accurate by design. Poor fits to the data using these group-parameters may indicate a
30 misspecification of the model structure, prior parameter distributions, or an error in the data. In
31 addition, it is useful to generate "population-based" parameters (posterior population 0) using
32 only the posterior distributions for the population means (u) and variances (E2), instead of the
33 estimated group-specific parameters. These population predictions provide a sense as to whether
34 the model and the predicted degree of population uncertainty and variability adequately account
35 for the range of heterogeneity in the experimental data. Furthermore, assuming the group-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-7 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
specific predictions are accurate, the population-based predictions are useful to identify whether
one or more if the datasets are "outliers" with respect to the predicted population. In addition, a
substantial number of in vivo datasets was available in all three species that were not previously
used for calibration. Thus, it is informative to compare the population-based model predictions,
discussed above, to these additional "validation" data in order to assess the predictive power of
the PBPK model.
MCMC outputs
Posterior
Posterior I2
Posterior group-
specific
9,
Posterior population
9
Posterior population
prediction^
Yii
-------
1 some experiments (this is partially responsible for the slower convergence). In particular, the
2 predictions for TCE and TCOH concentrations for the Abbas and Fisher (1997) data were poor.
3 In addition, TCE blood concentrations for the Greenberg et al. (1999) data were consistently
4 overpredicted. These data are discussed further in Table A-l.
5 Next, only samples of the population parameters (means and variances) were used, and
6 "new groups" were sampled from appropriate distributions using these population means and
7 variances. These "new groups" then represent the predicted population distribution,
8 incorporating both variability in the population as well as uncertainty in the population means
9 and variances. These "population-based" predictions were then compared to both the data used
10 in calibration, as well as the additional data identified that was not used in calibration. The
11 PBPK model was modified to accommodate some of the different outputs (e.g., tissue
12 concentrations) and exposure routes (TCE, TCA, and TCOH intravenous [i.v.]) used in the
13 "noncalibration" data, but otherwise it is unchanged.
14
15 A.2.3.1.1.1. Group-specific predictions and calibration data. [See
16 Appendix, linked. files\AppA.2.3.1.1.1 .Hack.mouse.group.calib.TCE.DRAFT.pdf.1
17
18 A.2.3.1.1.2. Population-based predictions and calibration and additional evaluation data.
19 rSeeAppendix.linked.files\AppA.2.3.1.1.2.Hack.mouse.pop.calib.eval.TCE.DRAFT.pdf.1
20
21 A.2.3.1.2. Conclusions regarding mouse model.
22 A.2.3.1.2.1. Trichloroethylene (TCE) concentrations in blood and tissues not well-predicted.
23 The PBPK model for the parent compound does not appear to be robust. Even group-specific
24 fits to datasets used for calibration were not always accurate. For oral dosing data, there is
25 clearly high variability in oral uptake parameters, and the addition of uptake through the first
26 (stomach) compartment should improve the fit. Unfortunately, inaccurate TCE uptake
27 parameters may lead to inaccurately estimated kinetic parameters for metabolites TCA and
28 TCOH, even if current fits are adequate.
29
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-9 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-l. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in mice
vo £3'
Reference
Simulation #
Calibration
data
Discussion
Abbas etal., 1997
41-42
These data are only published as an abstract. They consist of TCA and TCOH blood and urine data from
TCA and TCOH i.v. dosing. Blood levels of TCA and TCOH are fairly accurately predicted. From
TCOH dosing, urinary TCOG excretion is substantially overpredicted, and from TCA dosing, urinary
TCA excretion is substantially overpredicted.
TO
Abbas and Fisher,
1997
3-6
Results for these data were mixed. TCA levels were the best fit. The calibration data included TCA
blood and liver data, which were well predicted except at the earliest time-point. In addition, TCA
concentrations in the kidney were fairly consistent with the surrogate TCA body concentrations predicted
by the model. Urinary TCA was well predicted at the lower two and highest doses, but somewhat
underpredicted (though still in the 95% confidence region) at 1,200 mg/kg.
TCE levels were in general not well fit. Calibration data included blood, fat, and liver concentrations,
which were predicted poorly particularly at early and late times. One reason for this is probably the
representation of oral uptake. Although both the current model and the original Abbas and Fisher (1997)
model had two-compartments representing oral absorption, in the current model uptake can only occur
from the second compartment. By contrast, the Abbas and Fisher (1997) model had uptake from both
compartments, with the majority occurring from the first compartment. Thus, the explanation for the poor
fit, particularly of blood and liver concentrations, at early times is probably simply due to differences in
modeling oral uptake. This is also supported by the fact that the oral uptake parameters tended to be
among those that took the longest to converge.
Group-specific blood TCOH predictions were poor, with under-prediction at early times and
overprediction at late times. Population-based blood TCOH predictions tended to be underpredicted,
though generally within the 95% confidence region. Group-specific urinary TCOG predictions were
fairly accurate except at the highest dose. These predictions are also probably affected by the apparent
misrepresentation of oral uptake. In addition, a problem as found in the calibration data in that data on
free TCOH was calibrated against predictions of total TCOH (TCOH+TCOG).
A number of TCOH and TCOG measurements were not included in the calibration—among them
tissue concentrations of TCOH and tissue and blood concentrations of TCOG. Blood concentrations (the
only available surrogate) were poor predictors of tissue concentrations of TCOH and TCOG (model
generally under-predicted). For TCOG, this may be due in part to the model assumption that the
distribution volume of TCOG is equal to that of TCOH.
O £?
-------
This document is a draft for review purposes only and does not constitute Agency policy
20/09 A-1 1 DRAFT: DO NOT CITE OR QUOTE
Table A-1. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in mice (continued)
Reference
Fisher etal., 1991
Fisher and Allen,
1993
Green and Prout,
1985
Simulation #
1-2
(open
chamber)
7-16 (closed
chamber)
21-26 (open
chamber,
additional
exposures)
31-36
40
Calibration
data
A/
A/
Discussion
Venous blood TCE concentrations were somewhat underpredicted (a common issue with inhalation
exposures in mice — see discussion of Greenberg et al., 1999 below), but within the 95% confidence
region of both group-specific and population-based predictions. Plasma TCA levels were well predicted,
with most of the data near the interquartile region of both group-specific and population-based predictions
(but with substantial scatter in the male mice). However, it should be noted that only a single exposure
concentration for each sex was used in calibration, with 6 additional exposures (3 for each sex) not
included (see simulations 21-26, below).
Good posterior fits were obtained for these data — closed chamber data with initial concentrations from
300 to 10,000 ppm. Some variability in VMAX, however, was noted in the posterior distributions for that
parameter. Using group-specific VMAX values resulted in better fits to these data. However, there appears
to be a systematic trend of lower estimated apparent VMAX at higher exposures. Similarly, posterior
estimates of cardiac output and the ventilation-perfusion ratio declined (slightly) with higher exposures.
These could be related to documented physiological changes (e.g., reduced ventilation rate and body
temperature) in mice when exposed to some volatile organics.
Data from three additional exposures for each sex were available for comparison to model predictions.
Plasma TCA levels were generally well predicted, though the predictions for female mice data showed
some systematic over-prediction, particularly at late times (i.e., data showed shorter apparent half-life).
Blood TCE concentrations were consistently overpredicted, sometimes by almost an order of magnitude,
except in the case of female mice at 236 ppm, for which predictions were fairly accurate.
Predictions for these gavage data were generally fairly accurate. There was a slight tendency to
overpredict TCA plasma concentrations, with predictions tending to be worse in the female mice. Blood
levels of TCE were adequately predicted, though there was some systematic underprediction at 2-6 h after
dosing.
This datum consists of a single measurement of urinary excretion of TCA at 24 h as a fraction of dose,
from TCA i.v. dosing. The model substantially over-predicts the amount excreted. Whereas Green and
Prout (1985) measured 35% excreted at 24 h, the model predicts virtually complete excretion at 24 h.
-------
Table A-l. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in mice (continued)
Reference
Greenberg et al.,
1999
Larson and Bull,
1992b
Proutetal., 1985
Templin et al., 1993
Simulation #
17-18
37-39
19
27-30
(urinary
excretion at
different
doses)
20
Calibration
data
A/
A/
A/
Discussion
The calibration data included blood TCE, TCOH, and TCA data. Fits to blood TCA and TCOH were
adequate, but as with the Fisher et al. (1991) inhalation data, TCE levels were overpredicted (outside the
95% confidence region during and shortly after exposure).
As with Abbas and Fisher (1997), there were additional data in the study that was not used in
calibration, including blood levels of TCOG and tissue levels of TCE, TCA, TCOH, and TCOG. Tissue
levels of TCE were somewhat overpredicted, but generally within the 95% confidence region. TCA
levels were adequately predicted, and mostly in or near the interquartile region. TCOH levels were
somewhat underpredicted, though within the 95% confidence region. TCOG levels, for which blood
served as a surrogate for all tissues, were well predicted in blood and the lung, generally within the
interquartile region. However, blood TCOG predictions underpredicted liver and kidney concentrations.
Blood TCA predictions were fairly accurate for these data. However, TCE and TCOH blood
concentrations were underpredicted by up to an order of magnitude (outside the 95% confidence region).
Part of this may be due to uncertain oral dosing parameters. Urinary TCA and TCOG were also generally
underpredicted, in some cases outside of the 95% confidence region.
Fits to these data were generally adequate — within or near the interquartile region.
These data consisted of mass balance studies of the amount excreted in urine and exhaled unchanged at
doses from 10 to 2,000 mg/kg. TCA excretion was consistently overpredicted, except at the highest dose.
TCOG excretion was generally well predicted — within the interquartile range. The amount exhaled was
somewhat overpredicted, with a 4-fold difference (but still within 95% confidence) at the highest dose.
Blood TCA levels from these data were well predicted by the model. Blood TCE and TCOH levels were
well predicted using group-specific parameters, but did not appear representative using population-derived
parameters. However, this is probably a result of the group-specific oral absorption parameter, which was
substantially different than the population mean.
-------
1 The TCE data from inhalation experiments also are not well estimated, particularly blood
2 levels of TCE. While fractional uptake has been hypothesized, direct evidence for this is
3 lacking. In addition, physiologic responses to TCE vapors (reduced ventilation rates, lowered
4 body temperature) are a possibility. These are weakly supported by the closed chamber data, but
5 the amount of the changes is not sufficient to account for the low blood levels of TCE observed
6 in the open chamber experiments. It is also not clear what role presystemic elimination due to
7 local metabolism in the lung may play. It is known that the mouse lung has a high capacity to
8 metabolize TCE (Green et al., 1997). However, in the Hack et al. (2006) model, lung
9 metabolism is limited by flow to the tracheobronchial region. An alternative formulation for
10 lung metabolism in which TCE is available for metabolism directly from inhaled air (similar to
11 that used for styrene, Sarangapani et al., 2003), may allow for greater presystemic elimination of
12 TCE, as well as for evaluating the possibility of wash-in/wash-out effects. Furthermore, the
13 potential impact of other extrahepatic metabolism has not been evaluated. Curiously, predictions
14 for the tissue concentrations of TCE observed by Greenberg et al. (1999) were not as discrepant
15 as those for blood. A number of these hypotheses could be tested; however, the existing data
16 may not be sufficient to distinguish them. The Merdink et al. (1998) study, in which TCE was
17 given by i.v. (thereby avoiding both first pass in the liver and any fractional uptake issue in the
18 lung), may be somewhat helpful, but unfortunately only oxidative metabolite concentrations
19 were reported, not TCE concentrations.
20
21 A.2.3.1.2.2. Trichloroacetic acid (TCA) blood concentrations well predicted folio-wins
22 trichloroethylene (TCE) exposures, but TCA flux and disposition may not be accurate. TCA
23 blood and plasma concentrations following TCE exposure are consistently well predicted.
24 However, the total flux of TCA may not be correct, as evidenced by the varying degrees of
25 consistency with urinary excretion data. Of particular importance are TCA dosing studies, none
26 of which were included in the calibration. In these studies, total recovery of urinary TCA was
27 found to be substantially less than the administered dose. However, the current model assumes
28 that urinary excretion is the only source of clearance of TCA, leading to overestimation of
29 urinary excretion. This fact, combined with the observation that under TCE dosing, the model
30 appears to give accurate predictions of TCA urinary excretion for several datasets, strongly
31 suggests a discrepancy in the amount of TCA formed from TCE. That is, since the model
32 appears to overpredict the fraction of TCA that appears in urine, it may be reducing TCA
33 production to compensate. Inclusion of the TCA dosing studies (including some oral dosing
34 studies), along with inclusion of a nonrenal clearance pathway, would probably be helpful in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-13 DRAFT—DO NOT CITE OR QUOTE
-------
1 reducing these discrepancies. Finally, improvements in the TCOH/TCOG submodel, below,
2 should also help to ensure accurate estimates of TCA kinetics.
O
4 A.2.3.1.2.3. Trichloroethanol-trichloroethanol-slucuronide conjugate (TCOH/TCOG)
5 submodel requires revision and recalibration. Blood levels of TCOH and TCOG were
6 inconsistently predicted. Part of this is due to the problems with oral uptake, as discussed above.
7 In addition, the problems identified with the use of the Abbas and Fisher (1997) data (i.e., free
8 TCOH vs. total TCOH), mean that this submodel is not likely to be robust.
9 An additional concern is the over-prediction of urinary TCOG from the Abbas et al.
10 (1997) TCOH i.v. data. Like the case of TCA, this indicates that some other source of TCOH
11 clearance (not to TCA or urine—e.g., to dichloroacetic acid [DCA] or some other untracked
12 metabolite) is possible. This pathway can be considered for inclusion, and limits can be placed
13 on it using the available data.
14 Also, like for TCA, the fact that blood and urine are relatively well predicted from TCE
15 dosing strongly suggests a discrepancy in the amount of TCOH formed from TCE. That is, since
16 the model appears to overpredict the fraction of TCOH that appears in urine, it may be reducing
17 TCOH production to compensate. Including the TCOH dosing data would likely be helpful in
18 reducing these discrepancies.
19 Finally, as with the rat, the model needs to ensure that any first pass effect is accounted
20 for appropriately. Importantly, the estimated clearance rate for glucuronidation of TCOH is
21 substantially greater than hepatic blood flow. As was shown in Okino et al. (2005), in such a
22 situation, the use of a single compartment model across dose routes will be misleading because it
23 implies a substantial first-pass effect in the liver that cannot be modeled in a single compartment
24 model. That is, since TCOH is formed in the liver from TCE, and TCOH is also glucuronidated
25 in the liver to TCOG, a substantial portion of the TCOH may be glucuronidated before reaching
26 systemic circulation. This suggests that a liver compartment for TCOH is necessary.
27 Furthermore, because substantial TCOG can be excreted in bile from the liver prior to systemic
28 circulation, a liver compartment for TCOG may also be necessary to address that first pass
29 effect.
30 The addition of the liver compartment will necessitate several changes to model
31 parameters. The distribution volume for TCOH will be replaced by two parameters: the
32 liverblood and body:blood partition coefficients. Similarly for TCOG, liverblood and
33 body :blood partition coefficients will need to be added. Clearance of TCOH to TCA and TCOG
34 can be redefined as occurring in the liver, and urinary clearance can be redefined as coming from
35 the rest of the body. Fortunately, there are substantial data on circulating TCOG that has not
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 been included in the calibration. These data should be extremely informative in better estimating
2 the TCOH/TCOG submodel parameters.
O
4 A.2.3.1.2.4. Uncertainty in estimates of total metabolism. Closed chamber data are generally
5 thought to provide a good indicator of total metabolism. Both group-specific and population-
6 based predictions of the only available closed chamber data (Fisher et al., 1991) were fairly
7 accurate. Unfortunately, no additional closed chamber data were available. In addition, the
8 discrepancies in observed and predicted TCE blood concentrations following inhalation
9 exposures remain unresolved. Hypothesized explanations such as fractional uptake or
10 presystemic elimination could have a substantial impact on estimates of total metabolism.
11 In addition, no data are directly informative as to the fraction of total metabolism in the
12 lung, the amount of "untracked" hepatic oxidative metabolism (parameterized as "FracDCA"), or
13 any other extrahepatic metabolism. The lung metabolism as currently modeled could just as well
14 be located in other extrahepatic tissues, with little change in calibration. In addition, it is
15 difficult to distinguish between untracked hepatic oxidative metabolism and GSH conjugation,
16 particularly at low doses.
17 A.2.3.2. Rat Model
18 A.2.3.2.1. Group-specific and population-based predictions. As with the mouse mode,
19 initially, the sampled group-specific parameters were used to generate predictions for
20 comparison to the calibration data. Because these parameters were "optimized" for each group,
21 these "group-specific" predictions should be accurate by design, and indeed they were, as
22 discussed in more detail in Table A-2.
23 Next, as with the mouse, only samples of the population parameters (means and
24 variances) were used, and "new groups" were sampled from appropriate distribution using these
25 population means and variances. These "new groups" then represent the predicted population
26 distribution, incorporating both variability in the population as well as uncertainty in the
27 population means and variances. These "population-based" predictions were then compared to
28 both the data used in calibration, as well as the additional data identified that was not used in
29 calibration. The Hack et al. (2006) PBPK model used for prediction was modified to
30 accommodate some of the different outputs (e.g., tissue concentrations) and exposure routes (i.v.,
31 intra-arterial [i.a.], and intraperivenous [p.v.]) used in the "noncalibration" data, but otherwise
32 unchanged.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-15 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
Table A-2 Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in rats
Reference
Simulation #
Calibration
data
Discussion
Andersen et al..
1987
7-11
A/
Good posterior fits were obtained for these data-
4,640 ppm.
closed chamber data with initial concentrations from 100 to
a
^s
3
TO'
•3
§
s.
§
Barton et al..
1995
17-20
It was assumed that the closed chamber volume was the same as for Andersen et al. (1987). However, the
initial chamber concentrations are not clear in the paper. The values that were used in the simulations do not
appear to be correct, since in many cases the time-course is inaccurately predicted even at the earliest time-
points. Conclusions as to these data need to await definitive values for the initial chamber concentrations,
which were not available.
Bernauer et al.,
1996
1-3
Urinary time-course data (Fig 6-7) for TCA, TCOG, and NAcDCVC was given in concentration units (mg/mg
creat-h), whereas total excretion at 48 h (Table 2) was given in molar units (mmol excreted). In the original
calibration files, the conversion from concentration to cumulative excretion was not consistent-i.e., the amount
excreted at 48 h was different. The data were revised using a conversion that forced consistency. One
concern, however, is that this conversion amounts to 6.2 mg creatinine over 48 h, or 1.14 micromol/h. This
seems very low for rats; Trevisan et al. (2001), in samples from 195 male control rats, found a median value of
4.95 micromol/h, a mean of 5.39 micromol/h, and a 1-99 percentile range of 2.56-10.46 micromol/h.
In addition, the NAcDCVC data were revised in include both 1,2- and 2,2-isomers, since the goal of the
GSH pathway is primarily to constrain the total flux. Furthermore, because of the extensive interorgan
processing of GSH conjugates, and the fact that excretion was still ongoing at the end of the study (48 h), the
amount of NAcDCVC recovered can only be a lower bound on the amount ultimately excreted in urine.
However, the model does not attempt to represent the excretion time-course of GSH conjugates—it merely
models the total flux. This is evinced by the fact that the model predicts complete excretion by the first time
point of 12 h, whereas in the data, there is still substantial excretion occurring at 48 h.
Posterior fits to these data were poor in all cases except urinary TCA at the highest dose. In all other
cases, TCOH/TCOG and TCA excretion was substantially overpredicted, though this is due to the revision of
the data (i.e., the different assumptions about creatinine excretion). Unfortunately, of the original calibration
data, this is the only one with TCA and TCOH/TCOG urinary excretion. Therefore, that part of the model is
poorly calibrated. On the other hand, NAcDCVC was underpredicted for a number of reasons, as noted above.
Because of the incomplete capture of NAcDCVC in urine, unless the model can accurately portray the
time-course of NAcDCVC in urine, it should probably not be used for calibration of the GSH pathway, except
perhaps as a lower bound.
O £?
-------
to
O k^j
o ^
Table A-2. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in rats (continued)
Reference
Simulation #
Calibration
data
Discussion
a
Birner et al.,
1993
21-22
These data only showed urine concentrations, so a conversion was made to cumulative excretion based on an
assumed urine flow rate of 22.5 mL/d. Based on this, urinary NAcDCVC was underestimated by 100- to
1,000-fold. Urinary TCA was underestimated by about 2-fold in females (barely within the 95% confidence
interval), and was accurately estimated in males. Note that data on urinary flow rate from Trevisan et al.
(2001) in samples from 195 male control rats showed high variability, with a geometric standard deviation of
1.75, so this may explain the discrepancy in urinary TCA. However, the underestimation of urinary
NAcDCVC cannot be explained this way.
TO'
•3
O
o
a
Dallas et al.,
1991
23-24
At the lower (50 ppm) exposure, arterial blood concentrations were consistently overpredicted by about 2.5-
fold, while at the higher (500 ppm) exposure, arterial blood was overpredicted by 1.5- to 2-fold, but within the
range of variability. Exhaled breath concentrations were in the middle of the predicted range of variability at
both exposure levels. The ratio of exhaled breath and arterial blood should depend largely on the blood-air
partition coefficient, with minor dependence on the assumed dead space. This suggests the possibility of some
unaccounted-for variability in the partition coefficient (e.g., posterior mean estimated to be 15.7; in vitro
measured values from the literature are as follows: 25.82 [Sato et al., 1977], 21.9 [Gargas et al., 1989], 25.8
[Koizumi, 1989], 13.2 [Fisher et al., 1989], posterior). Alternatively, there may be a systematic error in these
data, since, as discussed below, the fit of the model to the arterial blood data of Keys et al. (2003) was highly
accurate.
Fisher et al.,
1989
25-28
Good posterior fits were obtained for these data (in females)—closed chamber data with initial concentrations
from 300 to 5,100 ppm. There was some slight overprediction of chamber concentrations (i.e., data showed
more uptake/metabolism) at the lower doses, but still within the 95% confidence interval.
o§
O s
H ^
O t-k-
HH Oq
H TO
gl
O ~r
O^
H
W
Fisher et al.,
1991
4-6
Good posterior fits were obtained from these data — plasma levels of TCA and venous blood levels of TCE.
Green and Prout,
1985
29-30
In naive rats at 500 mg/kg, urinary excretion of TCOH/TCOG and TCA at 24 h was underpredicted (2-fold),
although within the 95% confidence interval. With bile-cannulated rats at the same dose, the amount of
TCOG in bile was well within the 95% confidence interval. Urinary TCOH/TCOG was still underpredicted by
about 2-fold, but again still within the 95% confidence interval.
Jakobsonetal.,
1986
31
The only data from the experiment (500 ppm in female rats) were venous blood concentrations during
exposure. There were somewhat overpredicted at early times (outside of 95% confidence interval for first
30 min) but was well predicted at the termination of exposure. This suggests some discrepancies in uptake to
tissues that reach equilibrium quickly—the model approaches the peak concentration at a faster rate than the
data suggest.
-------
to
VO Co'
S
§
a
"§>
TO
TO'
Q
2-fold; outside 95% confidence interval), while excretion of TCA was accurately predicted.
In addition, elimination by exhaled breath was substantially overpredicted at the lowest exposure. Blood
TCOH levels were accurately predicted, but blood TCE levels were overpredicted at the 55 ppm. Part of the
discrepancies may be due to limited analytic sensitivities at the lower exposures.
Larson and Bull,
1992b
12-14
The digitization in the calibration file did not appear to be accurate, as there was a 10-fold discrepancy with
the original paper in the TCOH data. The data were replaced this those used by Clewell et al. (2000) and Bois
(2000b). Except for the TCOH data, differences between the digitizations were 20% or less.
Adequate posterior predictions were obtained for these data (oral dosing from 200 mg/kg to 3,000 mg/kg). All
predictions were within the 95% confidence interval of posterior predictions. Better fits were obtained using
group-specific posterior parameters, for which gut absorption and TCA urinary excretion parameters were
more highly identified.
-------
Table A-2. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in rats (continued)
vo £3'
Reference
Simulation #
Calibration
data
Discussion
Lash etal., 2006
45-46
In these corn-oil gavage experiments, almost all of the measurements appeared to be systematically low,
sometimes by many orders of magnitude. For example, at the lowest dose (263 mg/kg), urinary excretion of
TCOH/TCOG and TCA, and blood concentrations of TCOH were overpredicted by the model by around >105-
fold. TCE concentrations in blood and tissues at 2, 4, and 8 h were underpredicted by 103- to 104-fold. Many
studies, including those using the corn oil gavage (Green and Prout, 1985; Hissink et al., 2002), with similar
ranges of oral doses show good agreement with the model, it seems likely that these data are aberrant.
Leeetal., 1996
47-61
This extensive set of experiments involved multiroute administration of TCE (oral, i.v., i.a., or portal vein),
with serial measurements of arterial blood concentrations. For the oral route (8 mg/kg-64 mg/kg), the GI
absorption parameters had to be modified. The values from Keys et al. (2003) were used, and the resulting
predictions were quite accurate, albeit a more prominent peak was predicted. Predictions >30 min after dosing
were highly accurate.
For the i.v. route (0.71 mg/kg-64 mg/kg), predictions were also highly accurate in almost all cases. At the
lower doses (0.71 mg/kg and 2 mg/kg), there was slight overprediction in the first 30 min after dosing. At
highest dose (64 mg/kg), there was slight underprediction between 1 and 2 h after dosing. In all cases, the
values were within the 95% confidence interval.
For the i.a. route (0.71 mg/kg-16 mg/kg), all predictions were very accurate.
For the p.v. route (0.71 mg/kg-64 mg/kg), predictions still remained in the 95% confidence interval,
although there was more variation. At the lowest dose, there was overprediction in the first 30 min after
dosing. At the highest two doses (16 mg/kg and 64 mg/kg), there was slight underprediction between 1 and
5 h after dosing. This may in part be because a pharmacodynamic change in metabolism (e.g., via direct
solvent injury proposed by Lee et al., 2000).
Lee et al., 2000
62-69
In the p.v. and i.v. exposures, blood and liver concentrations were accurately predicted. For oral exposures,
the GI absorption parameters needed to be changed. While the values from Keys et al. (2003) led to accurate
predictions for lower doses (2 mg/kg-16 mg/kg), at the higher doses (48 mg/kg-432 mg/kg), much slower
absorption was evident. Comparisons at these higher dose are not meaningful without calibration of
absorption parameters.
Proutetal., 1985
15
Adequate posterior fits were obtained for these data—rat dosing at 1,000 mg/kg in corn oil. All predictions
were within the 95% confidence interval of posterior predictions. Better fits were obtained using group-
specific posterior parameters, for which gut absorption and TCA urinary excretion parameters were more
highly identified.
O £?
-------
Table A-2. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in rats (continued)
Reference
Simulation #
Calibration
data
Discussion
Stenner et al..
1997
70
As with other oral exposures, different GI absorption parameters were necessary. Again, the values from Keys
et al. (2003) were used, with some success. Blood TCA levels were accurately predicted, while TCOH blood
levels were systematically under-predicted (up to 10-fold).
Additional data with TCOH and TCA dosing, including naive and bile-cannulated rats, can be added when
those exposure routes are added to the model. These could be useful in better calibrating the enterohepatic
recirculation parameters.
Templinetal..
1995
16
Adequate posterior fits were obtained for blood TCA from these data—oral dosing at 100 mg/kg in Tween.
Blood levels of TCOH were underpredicted, while the time-course of TCE in blood exhibited an earlier peak.
Better fits were obtained using group-specific posterior parameters, for which gut absorption and TCA urinary
excretion parameters (and to a lesser extent glucuronidation of TCOH and biliary excretion of TCOG) were
more highly identified.
GI = gastrointestinal, NAc-l,2-DCVC = N-acetyl-S-(l,2-dichlrovinyl)-L-cysteine, NAc-2,2-DCVC = N-acetyl-S-(2,2-dichlrovinyl)-L-cysteine, NAcDCVC =
NAc-l,2-DCVC andNAc-2,2-DCVC.
-------
1 A.2.3.2.1.1. Group-specific predictions and calibration data. [See
2 Appendix.linked.files\AppA.2.3.2.1.1.Hack.rat.group.calib.TCE.DRAFT.pdf.1
O
4 A.2.3.2.1.2. Population-based predictions and calibration and additional evaluation data.
5 rSeeAppendix.linked.files\AppA.2.3.2.1.2.Hack.rat.pop.calib.eval.TCE.DRAFT.pdf. 1
6
7 A.2.3.2.2. Conclusions regarding rat model.
8 A.2.3.2.2.1. Trichloroethylene (TCE) concentrations in blood and tissues generally well-
9 predicted. The PBPK model for the parent compound appears to be robust. Multiple datasets
10 not used for calibration with TCE measurements in blood and tissues were simulated, and overall
11 the model gave very accurate predictions. A few datasets seemed somewhat anomalous—Dallas
12 et al. (1991), Kimmerle and Eben (1973a), Lash et al. (2006). However, data from Kaneko et al.
13 (1994), Keys et al. (2003), and Lee et al. (1996, 2000) were all well simulated, and corroborated
14 the data used for calibration (Fisher et al., 1991; Larson and Bull, 1992b; Prout et al., 1985;
15 Templin et al., 1995). Particularly important is the fact that tissue concentrations from
16 Keys et al. (2003) were well simulated.
17
18 A.2.3.2.2.2. Total metabolism probably well simulated, but ultimate disposition is less certain.
19 Closed chamber data are generally thought to provide a good indicator of total metabolism. Two
20 closed chamber studies not used for calibration were available—Barton et al. (1995) and Fisher
21 et al. (1989). Additional experimental information is required to analyze the Barton et al. (1995)
22 data, but the predictions for the Fisher et al. (1989) data were quite accurate.
23 However, the ultimate disposition of metabolized TCE is much less certain. Clearly, the
24 flux through the GSH pathway is not well constrained, with apparent discrepancies between the
25 N-acetyl-S-(l,2-dichlorovinyl)-L-cysteine (NAc-l,2-DCVC) data of Bernauer et al. (1996) and
26 Birner et al. (1993). Moreover, each of these data has limitations—in particular, the Bernauer et
27 al. (1996) data show that excretion is still substantial at the end of the reporting period, so that
28 the total flux of mercapturates has not been collected. Moreover, there is some question as to the
29 consistency of the Bernauer et al. (1996) data (Table 2 vs. Figures 6 and 7), since a direct
30 comparison seems to imply a very low creatinine excretion rate. The Birner et al. (1993) data
31 only report concentrations—not total excretion—so a urinary flow rate needs to be assumed.
32 In addition, no data are directly informative as to the fraction of total metabolism in the
33 lung or the amount of "untracked" hepatic oxidative metabolism (parameterized as "FracDCA").
34 The lung metabolism could just as well be located in other extrahepatic tissues, with little change
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-21 DRAFT—DO NOT CITE OR QUOTE
-------
1 in calibration. In addition, there is a degeneracy between untracked hepatic oxidative
2 metabolism and GSH conjugation, particularly at low doses.
3 The ultimate disposition of TCE as excreted TCOH/TCOG or TCA is also poorly
4 estimated in some cases, as discussed in more detail below.
5
6 A.2.3.2.2.3. Trichloroethanol-trichlorethanol-slucuronide conjugate (TCOH/TCOG)
1 submodel requires revision and recalibration. TCOH blood levels of TCOH were
8 inconsistently predicted in noncalibration datasets (well predicted for Larson and Bull [1992b];
9 Kimmerle and Eben [1973a]; but not Stenner et al. [1997] or Lash et al. [2006]), and the amount
10 of TCE ultimately excreted as TCOG/TCOH also appeared to be poorly predicted. The model
11 generally underpredicted TCOG/TCOH urinary excretion (underpredicted Green and Prout
12 [1985], overpredicted Kaneko et al. [1994], Kimmerle and Eben [1973a], and Lash et al. [2006]).
13 This may in part be due to discrepancies in the Bernauer et al. (1996) data as to the conversion of
14 excretion relative to creatinine.
15 Moreover, there are relatively sparse data on TCOH in combination with a relatively
16 complex model, so the identifiability of various pathways—conversion to TCA, enterohepatic
17 recirculation, and excretion in urine—is questionable.
18 This could be improved by the ability to incorporate TCOH dosing data from Merdink et
19 al. (1999) and Stenner et al. (1997), the latter of which included bile duct cannulation to better
20 estimate enterohepatic recirculation parameters. However, the TCOH dosing in these studies is
21 by the intravenous route, whereas with TCE dosing, TCOH first appears in the liver. Thus, the
22 model needs to ensure that any first pass effect is accounted for appropriately. Importantly, the
23 estimated clearance rate for glucuronidation of TCOH is substantially greater than hepatic blood
24 flow. That is, since TCOH is formed in the liver from TCE, and TCOH is also glucuronidated in
25 the liver to TCOG, a substantial portion of the TCOH may be glucuronidated before reaching
26 systemic circulation. Thus, suggests that a liver compartment for TCOH is necessary.
27 Furthermore, because substantial TCOG can be excreted in bile from the liver prior to systemic
28 circulation, a liver compartment for TCOG may also be necessary to address that first pass
29 effect.
30 The addition of the liver compartment will necessitate several changes to model
31 parameters. The distribution volume for TCOH will be replaced by two parameters: the
32 liverblood and body:blood partition coefficients. Similarly for TCOG, liverblood and
33 body :blood partition coefficients will need to be added. Clearance of TCOH to TCA and TCOG
34 can be redefined as occurring in the liver, and urinary clearance can be redefined as coming from
35 the rest of the body.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 Finally, additional clearance of TCOH (not to TCA or urine—e.g., to DCA or some other
2 untracked metabolite) is possible. This may in part explain the discrepancy between the accurate
3 predictions to blood data along with poor predictions to urinary excretion (i.e., there is a missing
4 pathway). This pathway can be considered for inclusion, and limits can be placed on it using the
5 available data.
6
7 A.2.3.2.2.4. Trichloroacetic acid (TCA) submodel would benefit from revised
8 trichloroethanol/trichloroethanol-slucuronide conjugate (TCOH/TCOG) submodel and
9 incorporating TCA dosing studies. While blood levels of TCA were well predicted in the one
10 noncalibration dataset (Stenner et al., 1997), the urinary excretion of TCA was inconsistently
11 predicted (underpredicted in Green and Prout [1985]; overpredicted in Kaneko et al. [1994] and
12 Lash et al. [2006]; accurately predicted in Kimmerle and Eben [1973a]). Because TCA is in part
13 derived from TCOH, a more accurate TCOH/TCOG submodel would probably improve the TCA
14 submodel.
15 In addition, there are a number of TCA dosing studies that could be used to isolate the
16 TCA kinetics from the complexities of TCE and TCOH. These could be readily incorporated
17 into the TCA submodel.
18 Finally, as with TCOH, additional clearance of TCA (not to urine—e.g., to DCA or some
19 other untracked metabolite) is possible. This may in part explain the discrepancy between the
20 accurate predictions to blood data along with poor predictions to urinary excretion (i.e., there is a
21 missing pathway). As with TCOH, this pathway can be considered for inclusion, and limits can
22 be placed on it using the available data.
23
24 A.2.3.3. Human model.
25 A.2.3.3.1. Individual-specific and population-based predictions. As with the mouse and rat
26 models, initially, the sampled individual-specific parameters (the term "individual" instead of
27 "group" is used since human variability was at the individual level) were used to generate
28 predictions for comparison to the calibration data. Because these parameters were "optimized"
29 for each individual, these "individual-specific" predictions should be accurate by design.
30 However, unlike for the rat, this was not the case for some experiments (this is partially
31 responsible for the slower convergence), although the inaccuracies were generally less than those
32 in the mouse. For example, alveolar air concentrations were systematically overpredicted for
33 several datasets. There was also variability in the ability to predict the precise time-course of
34 TCA and TCOH blood levels, with a few datasets more difficult for the model to accommodate.
35 These data are discussed further in Table A-3.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-23 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans
to
vo £;•
1
§
a
"§>
TO
TO'
Reference
Simulation #
Calibration
data
Discussion
<»
Bartonicek, 1962
38-45
The measured minute-volume was multiplied by a factor of 0.7 to obtain an estimate for alveolar ventilation
rate, which was fixed for each individual. These data are difficult to interpret because they consist of many
single data points. It is easiest to go through the measurements one at a time:
Alveolar retention (1—exhaled dose/inhaled dose during exposure) and Retained dose (inhaled dose—exhaled
dose during exposure): Curiously, retention was generally under-predicted, which in many cases retained dose
was accurately predicted. However, alveolar retention was an adjustment of the observed total retention:
TotRet = (CInh - CExh)/CInh = QAlv x (CInh - CAlv)/(MV x CInh), so that
AlvRet = TotRet x (QAlv/MV), with QAlv/MV assumed to be 0.7
Because retained dose is the more relevant quantity, and is less sensitive to assumptions about QAlv/MV, then
this is the better quantity to use for calibration.
Urinary TCOG: This was generally underpredicted, although generally within the 95% confidence
interval. Thus, these data will be informative as to interindividual variability.
Urinary TCA: Total collection (at 528 h) was accurately predicted, although the amount collected at 72 h
was generally under-predicted, sometimes substantially so.
Plasma TCA: Generally well predicted.
to
TO
*?> is:
•^H S<
H I
O t-k-
HH Oq
H TO
W S
Bernauer et al.,
1996
1-3
Individual-specific predictions were good for the time-courses of urinary TCOG and TCA, but poor for total
urinary TCOG+TCA and for urinary NAc-l,2-DCVC. One reason for the discrepancy in urinary excretion of
TCA and TCOG is that the urinary time-course data (see Figures 4-5 in the manuscript) for TCA, TCOG, and
NAc-l,2-DCVC was given in concentration units (mg/mg creat-h), whereas total excretion at 48 h (Table 2 in
the manuscript) was given in molar units (mmol excreted). In the original calibration files, the conversion
from concentration to cumulative excretion was not consistent—i.e., the amount excreted at 48 h was
different. For population-based predictions, the data were revised using a conversion that forced consistency.
One concern, however, is that this conversion amounts to 400-500 mg creatinine over 48 h, or 200-250 mg/d,
which seems rather low. For instance, Araki (1978) reported creatinine excretion of 11.5+/-1.8 mmol/24 h
(mean +/- SD) in 9 individuals, corresponding to 1,300 +/- 200 mg/d.
In addition, for population-based predictions, the data were revised include both the NAc-l,2-DCVC and
the N acetyl-S-(2,2-dichlorovinyl)-L-cysteine isomer (the combination denoted NAcDCVC), since the goal of
the GSH pathway is primarily to constrain the total flux. Furthermore, because of the extensive interorgan
processing of GSH conjugates, and the fact that excretion was still ongoing at the end of the study (48 h), the
amount of NAcDCVC recovered can only be a lower bound on the amount ultimately excreted in urine.
However, the model does not attempt to represent the excretion time-course of GSH conjugates—it merely
models the total flux. This is evinced by the fact that the model predicts complete excretion by the first time
point of 12 h, whereas in the data, there is still substantial excretion occurring at 48 h.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans (continued)
Reference
Bernauer et al.,
1996 (continued)
Bloemenetal.,
2001
Chiuetal.,2007
Fernandez et al.,
1977
Simulation #
1-3
(continued)
72-75
66-71
Calibration
data
Discussion
Population-based posterior fits to these data were quite good for urinary TCA and TCOH, but not for
NAcDCVC in urine. Because of the incomplete capture of NAcDCVC in urine, unless the model can
accurately portray the time-course of NAcDCVC in urine, it should probably not be used for calibration of the
GSH pathway, except perhaps as a lower bound.
Like Bartonicek (1962), these data are more difficult to interpret due to their being single data points for each
individual and exposure. However, in general, posterior population-based estimates of retained dose, urinary
TCOG, and urinary TCA were fairly accurate, staying within the 95% confidence interval, and mostly inside
the interquartile range. The data on GSH mercapturates are limited — first they are all nondetects. In addition,
because of the 48-56 h collection period, excretion of GSH mercapturates is probably incomplete, as noted
above in the discussion of Bernauer et al. (1996).
The measured minute-volume was multiplied by a factor of 0.7 to obtain an estimate for alveolar ventilation
rate, which was fixed for each individual. Alveolar air concentrations of TCE were generally well predicted,
especially during the exposure period. Postexposure, the initial drop in TCE concentration was generally
further than predicted, but the slope of the terminal phase was similar. Blood concentrations of TCE were
consistently overpredicted for all subjects and occasions.
Blood concentrations of TCA were consistently over-predicted, though mostly staying in the lower 95%
confidence region. Blood TCOH (free) levels were generally over-predicted, in many cases falling below the
95% confidence region, though in some cases the predictions were accurate. On the other hand, total TCOH
(free+glucuronidated) was well predicted (or even under-predicted) in most cases — in the cases where free
TCOH was accurately predicted, total TCOH was underpredicted. The free and total TCOH data reflect the
higher fraction of TCOH as TCOG than previously reported (e.g., Fisher et al. [1998] reported no detectable
TCOG in blood).
Data on urinary TCA and TCOG were complicated by some measurements being saturated, as well as the
intermittent nature of urine collection after Day 3. Thus, only the nonsaturated measurements for which the
time since the last voiding was known were included for direct comparison to the model predictions.
Saturated measurements were kept track of separately for comparison, but were considered only rough lower
bounds. TCA excretion was generally over-predicted, whether looking at unsaturated or saturated
measurements (the latter, would of course, be expected). Urinary excretion of TCOG generally stayed within
the 95% confidence range.
Alveolar air concentrations are somewhat overestimated. Other measurements are fairly well predicted.
to
§
I
TO
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £3'
s
§
a
<§»
TO
TO'
Tab\e A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans (continued)
Reference
Simulation #
Calibration
data
Discussion
to
Fisher et al.,
1998
13-33
A/
The majority of the data used in the calibration (both in terms of experiments and data points) came from this
study. In general, the individual-specific fits to these data were good, with the exception of alveolar air
concentrations, which were consistently over-predicted. In addition, for some individuals, the shape of the
TCOH time-course deviated from the predictions (#14, #24, #29, and #30)—the predicted peak was too
"sharp," with underprediction at early times. Simulation #23 showed the most deviation from predictions,
with substantial inaccuracies in blood TCA, TCOH, and urinary TCA.
Interestingly, in the population-based predictions, in same cases the predictions were not very accurate—
indicating that the full range of population variability is not accounted for in the posterior simulations. This is
particularly the case with venous blood TCE concentrations, which are generally under-predicted in
population estimates (although in some cases the predictions are accurate).
One issue with the way in which these data were utilized in the calibration is that in some cases, the same
individual was exposed to two different concentrations, but in the calibration, they were treated as separate
"individuals." Thus, parameters were allowed to vary between exposures, mixing interindividual and
interoccasion variability. It is recommended that in subsequent calibrations, the different occasions with the
same individual be modeled together. This will also allow identification of any dose-related changes in
parameters (e.g., saturation).
TO
^ s$
•^H XL"
O |
^H c^
o ^
HH Oq
H TO
W S
O ~r
O^
H
W
Kimmerle and
Eben, 1973b
46-57
Blood TCE levels are generally over-predicted for both single and multiexposure experiments. However,
levels at the end of exposure are rapidly changing, so some of those values may be better predicted if the
"exact" time after cessation of exposure were known.
Blood TCOH levels are fairly accurately predicted, although in some individuals in single exposure
experiments, there is a tendency to overpredict at early times and underpredict at late times. In multiexposure
experiments, the decline after the last exposure was somewhat steeper than predicted. Urinary excretion of
TCA and TCOH was well predicted.
Only grouped data on alveolar air concentrations were available, so they were not used.
Lapare et al..
1995
62-65
(individual
data)
Predictions for these data were not accurate. However, there was an error in some of the exposure
concentrations used in the original calibration. In addition, the last exposure "occasion" in these experiments
involved exercise/workload, and so should be excluded. Finally, individual data are available for these
experiments.
Taking into account these changes, population-based predictions were somewhat more accurate. However,
alveolar air concentrations and venous blood TCE concentrations were still over-predicted.
-------
to
vo £3'
s
§
a
"§>
TO
TO'
Tab\e A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans (continued)
Reference
Simulation #
Calibration
data
Discussion
to
o
a
TO
O
Monster et al.,
1976
5-6
(summary
data)
A/
58-61
(individual
data)
Individual-specific predictions were quite good, except that for blood TCA concentrations exhibited a higher
peak that predicted. However, TCOH values were entered as free TCOH, whereas the TCOH data were
actually total (free+glucuronidated) TCOH. Therefore, for population-based predictions, this change was
made. In addition, as with the Monster et al. (1979) data, minute-volume and exhaled air concentrations were
measured and incorporated for population-based predictions. Finally, individual-specific data are available, so
in those data should replace the grouped data in any revised calibration. These individual data also included
estimates of retained dose based on complete inhaled and exhaled air samples during exposure.
For population-based predictions, as with the Monster et al. (1979) data, grouped urinary and blood
TCOH/TCOG was somewhat under-predicted in the population-based predictions, and grouped alveolar and
blood TCE concentrations were somewhat over-predicted.
The results for the individual data were similar, but exhibited substantially greater variability that predicted.
For instance, in subject A, blood TCOH levels were generally greater than the 95% confidence interval at both
70 and 140 ppm, whereas predictions for blood TCOH in subject D were quite good. In another example, for
blood TCE levels, predictions for subject B were quite good, but those for subject D were poor (substantially
overpredicted). Thus, it is anticipated that adding these individual data will be substantially informative as to
interindividual variability, especially since all 4 individuals were exposed at 2 different doses.
Monster et al.,
1979
Individual-specific predictions for these data were quite good. However, TCA values were entered as plasma,
whereas the TCA data were actually in whole blood. Therefore, for population-based predictions, this change
was made. In addition, two additional time-courses were available that were not used in calibration: exhaled
air concentrations and total TCOH blood concentrations. These were added for population-based predictions.
In addition, the original article had data on ventilation rate, which as incorporated into the model. The
minute volume needed to be converted to alveolar ventilation rate for the model, but this required adjusted for
an extra dead space volume of 0.15 L due to use of a mask, as suggested in the article. The measured mean
minute volume was 11 L/min, and with a breathing rate of 14 breaths/min (assumed in the article), this
corresponding to a total volume of 0.79 L. Subtracting the 0.15 L of mask dead space and 0.15 L of
physiological dead space (suggested in the article) gives 0.49 L of total physiological dead space. Thus, the
minute volume of 11 L/min was adjusted by the factor 0.49/0.79 to give an alveolar ventilation rate of 6.8
L/min, which is a reasonably typical value at rest.
Due to extra nonphysiological dead space issue, some adjustment to the exhaled air predictions also
needed to be made. The alveolar air concentration CAlv was, therefore, estimated based on the formula
O £?
-------
Table A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans (continued)
Reference
Simulation #
Calibration
data
Discussion
Monster et al.,
1979 (continued)
4 (continued)
CAlv = (CExh x VTot - CInh x VDs)/VAlv
where CExh is the measured exhaled air concentration, VTot is the total volume (alveolar space VAlv of
0.49 L, physiological dead space of 0.15 L, and mask dead space of 0.15 L), VDs is the total dead space of
0.3 L, and CInh is the inhaled concentration.
Population-based predictions for these data lead to slight underestimation urinary TCOG and blood
TCOH levels, as well as some over-prediction of alveolar air and venous blood concentrations by factors of
3~10-fold.
Muller et al..
1972, 1974,
1975
7-10
Individual-specific predictions for these data were good, except for alveolar air concentrations. However,
several problems were found with these data as utilized in the original calibration:
• Digitization problems, particular with the time axis in the multiday exposure study (Simulation 9) that led
to measurements taken prior to an exposure modeled as occurring during the exposure. The original
digitization from Bois (2000b) and Clewell et al. (2000) was used for population-based estimates.
• Original article showed TCA as measured in plasma, not blood as was assumed in the calibration.
• Blood was taken from the earlobe, which is thought to be indicative of arterial blood concentrations, rather
than venous blood concentrations.
• TCOH in blood was free, not total, as Ertle et al. (1972 [cited in Methods]) had no use of beta-
glucuronidase in analyzing blood samples. Separate free and total measurements were done in plasma (not
whole blood), but these data were not included.
• Simulation 9, contiguous data on urinary excretion were only available out to 6 d, so only that data should
be included.
• Simulation 10, is actually the same as the first day of simulation 9, from Muller et al. (1972, 1975) (the
data were reported in both papers), and, thus, should be deleted.
These were corrected in the population-based estimates. Alveolar air concentration measurements remained
over-predicted, while the change to arterial blood led to over-prediction of those measurements during
exposure (but postexposure predictions were accurate).
-------
Table A-3. Evaluation of Hack et al. (2006) PBPK model predictions for in vivo data in humans (continued)
Reference
Muller et al.,
1974
Paycok and
Powell, 1945
Sato et al., 1977
Stewart et al.,
1970
Triebig et al.,
1976
Simulation #
81-82 (TCA
and TCOH
dosing)
35-37
76
11
12
Calibration
data
A/
A/
Discussion
The experiment with TCA showed somewhat more rapid decline in plasma levels than predicted, but still well
within the 95% confidence range. Urinary excretion was well predicted, but only accounted for 60% of the
administered dose — this is not consistent with the rapid decline in TCA plasma levels (10-fold lower than
peak at the end of exposure), which would seem to suggest the majority of TCA has been eliminated. With
TCOH dosing, blood levels of TCOH were over-predicted in the first 5 hours, perhaps due to slower oral
absorption (the augmented model used instantaneous and complete absorption). TCA plasma and urinary
excretion levels were fairly well predicted. However, urinary excretion of TCOG was near the bottom of the
95% confidence interval; while, in the same individuals with TCE dosing (Simulation 7), urinary excretion of
TCOG was substantially greater (near slightly above the interquartile region). Furthermore, total TCA and
TCOG urinary excretion accounted for <40% of the administered dose.
Population-based fits were good, within the inner quartile region.
Both alveolar air and blood concentrations are over-predicted in this model. Urinary TCA and TCOG, on the
other hand, are well predicted.
Individual-specific predictions for these data were good, except for some alveolar air concentrations.
However, a couple of problems were found with these data as utilized in the original calibration:
• The original article noted that individual took a lunch break during which there was no exposure. This was
not accounted for in the calibration runs, which a assumed a continuous 7-h exposure. The exposures
were, therefore, revised with a 3-h morning exposure (9-12), a 1 hour lunch break (12-1), and 4-h
afternoon exposure (1-5), to mimic a typical workday. The times of the measurements had to be revised
as well, since the article gave "relative" rather than "absolute" times (e.g., x hours postexposure).
• Contiguous data on urinary excretion were only available out to 1 1 d, so only that data should be included
(Table 2).
With these changes, population-based predictions of urinary TCA and TCOG were still accurate, but alveolar
air concentrations were over-predicted.
Only two data points are available for alveolar air, and blood TCA and TCOH. Only one data point is
available on blood TCE. Alveolar air was underpredicted at 24 h. Blood TCA and TCOH were within the
95% confidence ranges. Blood TCE was over-predicted substantially (outside 95% confidence range).
SD = standard deviation.
-------
1 Next, only samples of the population parameters (means and variances) were used, and
2 "new individuals" were sampled from appropriate distribution using these population means and
3 variances. These "new individuals" then represent the predicted population distribution,
4 incorporating both variability as well as uncertainty in the population means and variances.
5 These "population-based" predictions were then compared to both the data used in calibration, as
6 well as the additional data identified that was not used in calibration. The Hack et al. (2006)
7 PBPK model was modified to accommodate some of the different outputs (e.g., arterial blood,
8 intermittently collected urine, retained dose) and exposure routes (TCA i.v., oral TCA, and
9 TCOH) used in the "noncalibration" data, but otherwise unchanged.
10
11 A.2.3.3.1.1. Individual-specific predictions and calibration data. [See
12 Appendix, linked. files\AppA.2.3.3. l.l.Hack.human.indiv.calib.TCE.DRAFT.pdf.l
13
14 A.2.3.3.1.2. Population-based predictions and calibration and additional evaluation data.
15 [See Appendix.linked.files\AppA.2.3.3.1.2.Hack.human.pop.calib.eval.TCE.DRAFT.pdfl
16
17 A.2.3.3.2. Conclusions regarding human model.
18 A.2.3.3.2.1. Trichloroethylene (TCE) concentrations in blood and air are often not well-
19 predicted. Except for the Chiu et al. (2007) during exposure, TCE alveolar air levels were
20 consistently overpredicted. Even in Chiu et al. (2007), TCE levels postexposure were over-
21 predicted, as the drop-off after the end of exposure was further than predicted. Because
22 predictions for retained dose appear to be fairly accurate, this implies that less clearance is
23 occurring via exhalation than predicted by the model. This could be the result of additional
24 metabolism or storage not accounted for by the model.
25 Except for the Fisher et al. (1998) data, TCE blood levels were consistently
26 overpredicted. Because the majority of the data used for calibration was from Fisher et al.
27 (1998), this implies that the Fisher et al. (1998) data had blood concentrations that were
28 consistently higher than the other studies. This could be due to differences in metabolism and/or
29 distribution among studies.
30 Interestingly, the mouse inhalation data also exhibited inaccurate prediction of blood
31 TCE levels. Hypotheses such as fractional uptake or presystemic elimination due to local
32 metabolism in the lung have not been tested experimentally, nor is it clear that they can explain
33 the discrepancies.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-30 DRAFT—DO NOT CITE OR QUOTE
-------
1 Due to the difficulty in accurately predicted blood and air concentrations, there may be
2 substantial uncertainty in tissue concentrations of TCE. However, such potential model errors
3 can be characterized estimated and estimated as part of a revised calibration.
4
5 A.2.3.3.2.2. Trichloroacetic acid (TCA) blood concentrations well predicted folio-wins
6 trichloroethylene (TCE) exposures, but some uncertainty in TCA flux and disposition. TCA
7 blood and plasma concentrations and urinary excretion, following TCE exposure, are generally
8 well predicted. Even though the model's central estimates over-predicted the Chiu et al. (2007)
9 TCA data, the confidence intervals were still wide enough to encompass those data.
10 However, the total flux of TCA may not be correct, as evidenced by TCA dosing studies,
11 none of which were included in the calibration. In these studies, total recovery of urinary TCA
12 was found to be substantially less than the administered dose. However, the current model
13 assumes that urinary excretion is the only source of clearance of TCA. This leads to
14 overestimation of urinary excretion. This fact, combined with the observation that under TCE
15 dosing, the model appears to give accurate predictions of TCA urinary excretion for several
16 datasets, strongly suggests a discrepancy in the amount of TCA formed from TCE. That is, since
17 the model appears to overpredict the fraction of TCA that appears in urine, it may be reducing
18 TCA production to compensate. Inclusion of the TCA dosing studies, along with inclusion of a
19 nonrenal clearance pathway, would probably be helpful in reducing these discrepancies. Finally,
20 improvements in the TCOH/TCOG submodel, below, should also help to insure accurate
21 estimates of TCA kinetics.
22
23 A.2.3.3.2.3. Trichloroethanol-trichlorethanol-slucuronide conjugate (TCOH/TCOG)
24 submodel requires revision and recalibration. Blood levels of TCOH and urinary excretion of
25 TCOG were generally well predicted. Additional individual data show substantial
26 interindividual variability than can be incorporated into the calibration. Several errors as to the
27 measurement of free or total TCOH in blood need to be corrected.
28 A few inconsistencies with noncalibration datasets stand out. The presence of substantial
29 TCOG in blood in the Chiu et al. (2007) data are not predicted by the model. Interestingly, only
30 two studies that included measurements of TCOG in blood (rather than just total TCOH or just
31 free TCOH)—Muller et al. (1975), which found about 17% of total TCOH to be TCOG, and
32 Fisher et al. (1998), who could not detect TCOG. Both of these studies had exposures at
33 100 ppm. Interestingly Muller et al. (1975) reported increased TCOG (as fraction of total
34 TCOH) with ethanol consumption, hypothesizing the inhibition of a glucuronyl transferase that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 slowed glucuronidation. This also would result in a greater half-life for TCOH in blood with
2 ethanol consumptions, which was observed.
3 An additional concern is the over-prediction of urinary TCOG following TCOH
4 administration from the Muller et al. (1974) data. Like the case of TCA, this indicates that some
5 other source of TCOH clearance (not to TCA or urine—e.g., to DCA or some other untracked
6 metabolite) is possible. This pathway can be considered for inclusion, and limits can be placed
7 on it using the available data.
8 Also, as for TCA, the fact that blood and urine are relatively well predicted from TCE
9 dosing strongly suggests a discrepancy in the amount of TCOH formed from TCE. That is, since
10 the model appears to overpredict the fraction of TCOH that appears in urine, it may be reducing
11 TCOH production to compensate.
12 Finally, as with the rat and mice, the model needs to ensure that any first pass effect is
13 accounted for appropriately. Particularly for the Chiu et al. (2007) data, in which substantial
14 TCOG appears in blood, since TCOH is formed in the liver from TCE, and TCOH is also
15 glucuronidated in the liver to TCOG, a substantial portion of the TCOH may be glucuronidated
16 before reaching systemic circulation. Thus, suggests that a liver compartment for TCOH is
17 necessary. Furthermore, because substantial TCOG can be excreted in bile from the liver prior
18 to systemic circulation, a liver compartment for TCOG may also be necessary to address that
19 first pass effect. In addition, in light of the Chiu et al. (2007) data, it may be useful to expand the
20 prior range for the KM of TCOH glucuronidation.
21 The addition of the liver compartment will necessitate several changes to model
22 parameters. The distribution volume for TCOH will be replaced by two parameters: the
23 liver:blood and body:blood partition coefficients. Similarly for TCOG, liver:blood and
24 body :blood partition coefficients will need to be added. Clearance of TCOH to TCA and TCOG
25 can be redefined as occurring in the liver, and urinary clearance can be redefined as coming from
26 the rest of the body. Fortunately, there are in vitro partition coefficients for TCOH. It may be
27 important to incorporate the fact that Fisher et al. (1998) found no TCOG in blood. This can be
28 included by having the TCOH data be used for both free and total TCOH (particularly since that
29 is how the estimation of TCOG was made—by taking the difference between total and free).
30
31 A.2.3.3.2.4. Uncertainty in estimates of total metabolism. Estimates of total recovery after
32 TCE exposure (TCE in exhaled air, TCA and TCOG in urine) have been found to be only
33 60-70% (Monster et al., 1976, 1979; Chiu et al., 2007). Even estimates of total recovery after
34 TCA and TCOH dosing have found 25-50% unaccounted for in urinary excretion (Paycok and
35 Powell, 1945; Muller et al., 1974). Bartonicek found some TCOH and TCA in feces, but this
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-32 DRAFT—DO NOT CITE OR QUOTE
-------
1 was about 10-fold less than that found in urine, so this cannot account for the discrepancy.
2 Therefore, it is likely that additional metabolism of TCE, TCOH, and/or TCA are occurring.
3 Additional metabolism of TCE could account for the consistent overestimation of TCE in blood
4 and exhaled breath found in many studies. However, no data are directly informative as to the
5 fraction of total metabolism in the lung, the amount of "untracked" hepatic oxidative metabolism
6 (parameterized as "FracDCA"), or any other extrahepatic metabolism. The lung (TB)
7 metabolism as currently modeled could just as well be located in other extrahepatic tissues, with
8 little change in calibration. In addition, it is difficult to distinguish between untracked hepatic
9 oxidative metabolism and GSH conjugation, particularly at low doses.
10
11 A.3. PRELIMINARY ANALYSIS OF MOUSE GAS UPTAKE DATA: MOTIVATION
12 FOR MODIFICATION OF RESPIRATORY METABOLISM
13 Potential different model structures can be investigated using the core PBPK model
14 containing averaged input parameters, since this approach saves computational time and is more
15 efficient when testing different structural hypotheses. This approach is particularly helpful for
16 quick comparisons of data with model predictions. During the calibration process, this approach
17 was used for different routes of exposure and across all three species. For both mice and rats, the
18 closed chamber inhalation data resulted in fits that were considered not optimal when visually
19 examined. Although closed chamber inhalation usually combines multiple animals per
20 experiment, and may not be as useful in differentiating between individual and experimental
21 uncertainty (Hack et al., 2006), closed chamber data do describe in vivo metabolism and have
22 been historically used to quantify averaged in vivo Michaelis-Menten kinetics in rodents.
23 There are several assumptions used when combining PBPK modeling and closed
24 chamber data to estimate metabolism via regression. The key experimental principles require a
25 tight, sealed, or air-closed system where all chamber variables are controlled to known set points
26 or monitored, that is all except for metabolism. For example, the inhalation chamber is
27 calibrated without an animal, to determine normal absorption to the empty system. This empty
28 chamber calibration is then followed with a dead animal experiment, identical in every way to
29 the in vivo exposure, and is meant to account for every factor other than metabolism, which is
30 zero in the dead animal. When the live animal(s) are placed in the chamber, oxygen is provided
31 for, and carbon dioxide accumulated during breathing is removed by absorption with a chemical
32 scrubber. A bolus injection of the parent chemical, TCE, is given and this injection time starts
33 the inhalation exposure. The chemical inside the chamber will decrease with time, as it is
34 absorbed by the system and the metabolic process inside the rodent. Since all known processes
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 contributing to the decline are quantified, except for metabolism, the metabolic parameters can
2 be extracted from the total chamber concentration decline using regression techniques.
3 The basic structure for the PBPK model that is linked to closed chamber inhalation data
4 has the same basic structure as described before. The one major difference is the inclusion of
5 one additional equation that accounts for mass balance changes inside the inhalation chamber or
6 system, and connects the chamber with the inhaled and exhaled concentrations breathed in and
7 out by the animal:
8
ch (Eq.A-4)
ch
10
1 1 where
12 RATS = number of animals in the chamber
13 QP = alveolar ventilation rate
14 Cx = exhaled concentration
15 ACH = net amount of chemical inside chamber
16 Vch = volume of chamber
17 KLOSS = loss rate constant to glassware.
18
19 An updated model was developed that included updated physiological and chemical -
20 specific parameters as well as GSH metabolism in the liver and kidney, as discussed in
21 Chapter 3. The PBPK model code was translated from MCSim to use in Matlab® (version
22 7.2.0.232, R2006a, Natick, MA) using their m language. This PBPK model made use of fixed or
23 constant, averaged values for physiological, chemical and other input parameters; there were no
24 statistical distributions attached to each average value. As an additional step in quality control,
25 mass balance was checked for the MCSim code, and comparisons across both sets of code were
26 made to ensure that both sets of predictions were the same.
27 The resulting simulations were compared to mice gas uptake data (Fisher et al., 1991)
28 after some adjustments of the fat compartment volumes and flows based on visual fits, and
29 limited least-squares optimization of just VMAX (different for males and females) and KM (same
30 for males and females). The results are shown in the top panels of Figures A-3-A-4, which
3 1 showed poor fits particularly at lower chamber concentrations. In particular, metabolism is
32 observed to be faster than predicted by simulation. This is directly related to metabolism of TCE
33 being limited by hepatic blood flow at these exposures. Indeed, Fisher et al. (1991) was able to
34 obtain adequate fits to these data by using cardiac output and ventilation rates that were about
35 2-fold higher than is typical for mice. Although their later publication reporting inhalation
36 experiments (Greenberg et al., 1999) used the lower values from Brown et al. (1997) for these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-34 DRAFT— DO NOT CITE OR QUOTE
-------
1 parameters, they did not revisit the Fisher et al. (1991) data with the updated model. In addition,
2 the Hack et al. (2006) model estimated the cardiac output and ventilation rate and for these
3 experiments to be about 2-fold higher than typical. However, it seems unlikely that cardiac
4 output and ventilation rate were really as high as used in these models, since TCE and other
5 solvents typically have central nervous system-depressing effects. In the mouse, after the liver,
6 the lung has the highest rate of oxidative metabolism, as assessed by in vitro methods (see
7 footnote in Section 3.5.4.2 for a discussion of why kidney oxidative metabolism is likely to be
8 minor quantitatively). In addition, TCE administered via inhalation is available to the lung
9 directly, as well as through blood flow. Therefore, it was hypothesized that a more refined
10 treatment of respiratory metabolism may be necessary to account for the additional metabolism.
11 The structure of the updated respiratory metabolism model is shown in Figure A-5, with
12 the mathematical formulation shown in the model code in Section A.6, where the "D" is the
13 diffusion rate, "concentrations" and "amounts" are related by the compartment volume, and the
14 other symbols have their standard meanings in the context of PBPK modeling. In brief, this is a
15 more highly "lumped" version of the Sarangapani et al. (2003) respiratory metabolism model for
16 styrene combined with a "continuous breathing" model to account for a possible wash-in/wash-
17 out effect. In brief, upon inhalation (at a rate equal to the full minute volume, not just the
18 alveolar ventilation), TCE can either (1) diffuse between the respiratory tract lumen and the
19 respiratory tract tissue; (2) remain in the dead space, or (3) enter the gas exchange region. In the
20 respiratory tract tissue, TCE can either be "stored" temporarily until exhalation, during which it
21 diffuses to the "exhalation" respiratory tract lumen, or be metabolized. In the dead space, TCE is
22 transferred directly to the "exhalation" respiratory tract lumen at a rate equal to the minute-
23 volume minus the alveolar ventilation rate, where it mixes with the other sources. In the gas
24 exchange region, it undergoes transfer to and from blood, as is standard for PBPK models of
25 volatile organics. Therefore, if respiratory metabolism is absent (VMAxClara = 0), then the
26 model reduces to a wash-in/wash-out effect where TCE is temporarily adsorbed to the
27 respiratory tract tissue, the amount of which depends on the diffusion rate, the volume of the
28 tissue, and the partition coefficients.
29 The results of the same limited optimization, now with additional parameters VMAxClara,
30 KMClara, and D being estimated simultaneously with the hepatic VMAX and KM, are shown in the
31 bottom panels of Figures A-2 and A-3. The improvement in the model fits is obvious, and these
32 results served as a motivation to include this respiratory metabolism model for analysis by the
33 more formal Bayesian methods.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
10
E 10
Q.
Q.
c
o
'-4-J
(O
O
O
E
Jz 1
6 10
10
234
time (h)
10
1
2
3
10
Q.
Q.
o
'-4-J
(O
o
O
L_
O)
J3
E
™ 1
6 10
10
234
time (h)
Figure A-3. Limited optimization results for male closed chamber data from
Fisher et al. (1991) without (top) and with (bottom) respiratory metabolism.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-36 DRAFT—DO NOT CITE OR QUOTE
-------
10
E 10
Q.
Q.
O
'-*->
ro
o
O
6 10
10
234
time (h)
1
2
3
4
10
i" 10
Q.
Q.
O
O
E
™ 1
o 10
10
234
time (h)
Figure A-4. Limited optimization results for female closed chamber data
from Fisher et al. (1991) without (top) and with (bottom) respiratory
metabolism.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-37 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
QM*CInh
I
Respiratory
Tract During
Inhalation
(AInhResp)
nhResp
i
r
QM*CExhResp
t
D* CResp
D*CInhResp
VMaxClara*C
(QM
Respiratory
Tract Tissue
(AResp)
D* CResp R
* Tr
* E
D*CExhResp /^
:Resp/(KMClara + CResp)
- QP)*CInhResp
espiratory
act During
-xhalation
lExhResp)
1 L
QP*CA
Alveolar (Gas Exchange) Region
QC*CVen
|QC*
CArt
A.4.
Figure A-5. Respiratory metabolism model for updated PBPK model.
DETAILS OF THE UPDATED PHYSIOLOGICALLY BASED
PHARMACOKINETIC (PBPK) MODEL FOR TRICHLOROETHYLENE (TCE)
AND ITS METABOLITES
The structure of the updated PBPK model and the statistical population model are shown
graphically in Chapter 3, with the model code shown below in Section A.6. Details as to its
parameter values and their prior distributions are given below.
A.4.1. Model Parameters and Baseline Values
The multipage Table A-4 below describes all the parameters of the updated PBPK model,
their baseline values (which are used as central estimates in the prior distributions for the
Bayesian analysis), and any scaling relationship used in their calculation. More detailed notes
are included in the comments of the model code (see Section A.6).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-4. PBPK model parameters, baseline values, and scaling relationships
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
2
•••2
I
TO
Model parameter
Body weight (kg)
Abbreviation
BW
Baseline value (if applicable)
Mouse
0.03
Rat
0.3
Human
Female
(or both)
60
Male
70
Scaling (Sampled)
Parameter
Additional
scaling
(if any)
Notes/
source
a
Flows
Cardiac output (L/h)
Alveolar ventilation (L/h)
Respiratory lumen:tissue diffusion flow
rate (L/h)
QC
QP
DResp
11.6
2.5
13.3
1.9
16
0.96
InQCC
InVPRC
InDRespC
BW374
QC
QP
b
c
d
Physiological blood flows to tissues
Fat blood flow
Gut blood flow (portal vein)
Liver blood flow (hepatic artery)
Slowly perfused blood flow
Kidney blood flow
Rapidly perfused blood flow
Fraction of blood that is plasma
QFat
QGut
QLiv
QSIw
QKid
QRap
FracPlas
0.07
0.141
0.02
0.217
0.091
0.52
0.07
0.153
0.021
0.336
0.141
0.53
0.085
0.21
0.065
0.17
0.17
0.615
0.05
0.19
0.22
0.19
0.567
QFatC
QGutC
QLivC
QSIwC
QKidC
FracPlasC
QC
QC
QC
QC
QC
e
e
e
e
e
e
f
Physiological volumes
Fat compartment volume (L)
Gut compartment volume (L)
Liver compartment volume (L)
Rapidly perfused compartment volume (L)
Volume of respiratory lumen (L air)
Effective volume for respiratory tissue
(L air)
Kidney compartment volume (L)
Blood compartment volume (L)
Total perfused volume (L)
VFat
VGut
VLiv
VRap
VRespLum
VRespEff
VKid
VBId
VPerf
0.07
0.049
0.055
0.1
0.004667
0.0007
0.017
0.049
0.8897
0.07
0.032
0.034
0.088
0.004667
0.0005
0.007
0.074
0.8995
0.317
0.022
0.023
0.093
0.002386
0.00018
0.0046
0.068
0.85778
0.199
0.02
0.025
0.088
0.00018
0.0043
0.077
0.8560
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
BW
BW
BW
BW
BW
BWxPResp
xPB
BW
BW
BW
g
g
g
g
g
g
g
g
g
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
s
TO'
I
TO
Co
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
Model parameter
Slowly perfused compartment volume (L)
Plasma compartment volume (L)
TCA body compartment volume (L)
TCOH/G body compartment volume (L)
Abbreviation
VSIw
VPIas
VBod
VBodTCOH
Baseline value (if applicable)
Mouse
Rat
Human
Female
(or both)
Male
Scaling (Sampled)
Parameter
Additional
scaling
(if any)
Notes/
source
g
h
i
j
TCE distribution/partitioning
TCE blood/air partition coefficient
TCE fat/blood partition coefficient
TCE gut/blood partition coefficient
TCE liver/blood partition coefficient
TCE rapidly perfused/blood partition
coefficient
TCE respiratory tissue:air partition
coefficient
TCE kidney/blood partition coefficient
TCE slowly perfused/blood partition
coefficient
PB
PFat
PGut
PLiv
PRap
PResp
PKid
PSIw
15
36
1.9
1.7
1.9
2.6
2.1
2.4
22
27
1.4
1.5
1.3
1
1.3
0.58
9.5
67
2.6
4.1
2.6
1.3
1.6
2.1
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
k
'
m
n
0
P
q
r
TCA distribution/partitioning
TCA blood/plasma concentration ratio
Free TCA body/blood plasma partition
coefficient
Free TCA liver/blood plasma partition
coefficient
TCAPIas
PBodTCA
PLivTCA
0.5
0.88
1.18
0.5
0.88
1.18
0.5
0.52
0.66
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
See note
s
t
t
TCA plasma binding
Protein/TCA dissociation constant
(umol/L)
Protein concentration (umole/L)
kDissoc
BMax
107
0.88
275
1.22
182
4.62
InkDissocC
InBMaxkDC
u
u
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
I
TO
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
Model parameter
Abbreviation
Baseline value (if applicable)
Mouse
Rat
Human
Female
(or both)
Male
Scaling (Sampled)
Parameter
Additional
scaling
(if any)
Notes/
source
TCOH and TCOG distribution/partitioning
TCOH body/blood partition coefficient
TCOH liver/body partition coefficient
TCOG body/blood partition coefficient
TCOG liver/body partition coefficient
PBodTCOH
PLivTCOH
PBodTCOG
PLivTCOG
1.11
1.3
1.11
1.3
1.11
1.3
1.11
1.3
0.91
0.59
0.91
0.59
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
V
V
w
w
DCVG distribution/partitioning
DCVG effective volume of distribution
VDCVG
InPeffDCVG
See note
X
TCE metabolism
VMAX for hepatic TCE oxidation (mg/h)
KM for hepatic TCE oxidation (mg/L)
Fraction of hepatic TCE oxidation not to
TCA+TCOH
Fraction of hepatic TCE oxidation to TCA
VMAX for hepatic TCE GSH conjugation
(mg/h)
KM for hepatic TCE GSH conjugation
(mg/L)
VMAX for renal TCE GSH conjugation
(mg/h)
KM for renal TCE GSH conjugation (mg/L)
VMAX
KM
FracOther
FracTCA
VMAXDCVG
KMDCVG
VMAxKidDCVG
KMKidDCVG
2,700
36
0.32
300
1.53
60
0.34
600
21
0.32
66
0.25
6
0.026
255
66
0.32
19
2.9
230
2.7
InVMAxC
lnKMC
InCIC
InFracOtherC
InFracTCAC
InVMAxDCVGC
InCIDCVGC
InKMDCVGC
InVMAxKidDCVGC
InCIKidDCVGC
InKMKidDCVGC
VLiv
See note
See note
See note
VLiv
VKid
y
y
y
z
aa
bb
bb
bb
bb
bb
bb
TCE metabolism (respiratory tract)
VMAX fortracheo-bronchial TCE oxidation
(mg/h)
KM fortracheo-bronchial TCE oxidation
(mg/L air)
VMAxClara
KMClara
0.070102
0.014347
0.027273
0.025253
InVMAxLungLivC
lnKMClara
VMAX
cc
cc
H I
O ^
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
I
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
Model parameter
Fraction of respiratory oxidation entering
systemic circulation
Abbreviation
FracLungSys
Baseline value (if applicable)
Mouse
Rat
Human
Female
(or both)
Male
Scaling (Sampled)
Parameter
InFracLungSysC
Additional
scaling
(if any)
See note
Notes/
source
dd
TCOH metabolism
VMAX for hepatic TCOH->TCA (mg/h)
KM for hepatic TCOH->TCA (mg/L)
VMAX for hepatic TCOH->TCOG (mg/h)
KM for hepatic TCOH->TCOG (mg/L)
Rate constant for hepatic TCOH->other
(/h)
VMAxTCOH
K,v,TCOH
VMAxGlUC
KMGluc
kMetTCOH
InVMAxTCOHC
InCITCOHC
InKivJCOH
InVMAxGlucC
InCIGIucC
lnKMGluc
InkMetTCOHC
BW374
BW374
BW374
BW374
BW174
TCA metabolism/clearance
Rate constant for TCA plasma->urine (/h)
Rate constant for hepatic TCA->other (/h)
kUrnTCA
kMetTCA
0.6
0.522
0.108
InkUrnTCAC
InkMetTCAC
VPIas'1
BW174
ee
TCOG metabolism/clearance
Rate constant for TCOG liver->bile (/h)
Lumped rate constant for TCOG bile-
>TCOH liver (/h)
Rate constant forTCOG->urine (/h)
kBile
kEHR
kUrnTCOG
0.6
0.522
0.108
InkBileC
InkEHRC
InkUrnTCOGC
BW174
BW174
VBId'1
ee
DCVG metabolism
Rate constant for hepatic DCVG->DCVC
(/h)
kDCVG
InkDCVGC
BW174
ff
H I
O >
HH Oq
H TO
O
H
W
-------
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
Model parameter
Abbreviation
Baseline value (if applicable)
Mouse
Rat
Human
Female
(or both)
Male
Scaling (Sampled)
Parameter
Additional
scaling
(if any)
Notes/
source
DCVC metabolism/clearance
Lumped rate constant for DCVC->Urinary
NAcDCVC (/h)
Rate constant for DCVC bioactivation (/h)
kNAT
kKidBioact
InkNATC
InkKidBioactC
BW1/4
BW1/4
gg
gg
Oral uptake/transfer coefficients
TCE Stomach-duodenum transfer
coefficient (/h)
TCE stomach absorption coefficient (/h)
TCE duodenum absorption coefficient (/h)
TCA stomach absorption coefficient (/h)
TCOH stomach absorption coefficient (/h)
kTSD
kAS
kAD
kASTCA
kASTCOH
InkTSD
InkAS
InkAD
InkASTCA
InkASTCOH
hh
hh
hh
hh
hh
Explanatory note. Unless otherwise noted, the model parameter is obtained by multiplying (1) the "baseline value" (equals 1 if not specified) times (2) the
scaling parameter [or for those beginning with "In," which are natural-log transformed, exp(lnXX)] times (3) any additional scaling as noted in the second to last
column. Unless otherwise noted, all log-transformed scaling parameters have baseline value of 0 [i.e., exp(lnXX) has baseline value of 1] and all other scaling
parameters have baseline parameters of 1.
"Use measured value if available.
blf QP is measured, then scale by QP using VPR. Baseline values are from Brown et al. (1997) (mouse and rat) and ICRP (International Commission on
Radiological Protection) Publication 89 (2003) (human).
°Use measured QP, if available; otherwise scale by QC using alveolar VPR. Baseline values are from Brown et al. (1997) (mouse and rat) and ICRP Publication
89 (2003) (human).
dScaling parameter is relative to alveolar ventilation rate.
Tat represents adipose tissue only. Gut is the gastro-intestinal tract, pancreas, and spleen (all drain to the portal vein). Slowly perfused tissue is the muscle and
skin. Rapidly perfused tissue is the rest of the organs, plus the bone marrow and lymph nodes, the blood flow for which is calculated as the difference between
QC and the sum of the other blood flows. Baseline values are from Brown et al. (1997) (mouse and rat) and ICRP Publication 89 (2003) (human).
fThis is equal to 1 minus the hematocrit (measured value used if available). Baseline values from control animals in Hejtmancik et al. (2002) (mouse and rat) and
ICRP Publication 89 (2003) (human).
-------
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
at represents adipose tissue only, and the measured value is used, if available. Gut is the gastro-intestinal tract, pancreas, and spleen (all drain to the portal
vein). Rapidly perfused tissue is the rest of the organs, plus the bone marrow and lymph nodes, minus the tracheobronchial region. The respiratory tissue
o volume is tracheobronchial region, with an effective air volume given by multiplying by its tissue:air partition coefficient (= tissue:blood times blood:air). The
§ slowly perfused tissue is the muscle and skin. This leaves a small (10-15% of body weight [BW]) unperfused volume that consists mostly of bone (minus
H marrow) and the gastro-intestinal tract contents. Baseline values are from Brown et al. (1997) (mouse and rat) and ICRP Publication 89 (2003) (human),
31 except for volumes of the respiratory lumen, which are from Sarangapani et al. (2003).
S' hDerived from blood volume using FracPlas.
a 'Sum of all compartments except the blood and liver.
g" J Sum of all compartments except the liver.
^j> kMouse value is from pooling Abbas and Fisher (1997) and Fisher et al. (1991). Rat value is from pooling Sato et al. (1977), Gargas et al. (1989), Barton et al.
^ (1995), Simmons et al. (2002), Koizumi (1989), and Fisher et al. (1989). Human value is from pooling Sato and Nakajima (1979), Sato et al. (1977), Gargas et
^ al. (1989), Fiserova-Bergerova et al. (1984), Fisher et al. (1998), and Koizumi (1989).
^ 'Mouse value is from Abbas and Fisher (1997). Rat value is from pooling Barton et al. (1995), Sato et al. (1977), and Fisher et al. (1989). Human value is from
I' pooling Fiserova-Bergerova et al. (1984), Fisher et al. (1998), and Sato et al. (1977).
^ mValue is the geometric mean of liver and kidney (relatively high uncertainty) values.
s "Mouse value is from Fisher et al. (1991). Rat value is from pooling Barton et al. (1995), Sato et al. (1977), and Fisher et al. (1989). Human value is from
^ pooling Fiserova-Bergerova et al. (1984) and Fisher et al. (1998).
% "Mouse value is geometric mean of liver and kidney values. Rat value is the brain value from Sato et al. (1977). Human value is the brain value from Fiserova-
£ Bergerova et al. (1984).
g^ pMouse value is the lung value from Abbas and Fisher (1997). Rat value is the lung value from Sato et al. (1977). Human value is from pooling lung values
j? from Fiserova-Bergerova et al. (1984) and Fisher et al. (1998).
a qMouse value is from Abbas and Fisher (1997). Rat value is from pooling Barton et al. (1995) and Sato et al. (1977). Human value is from pooling Fiserova-
^ Bergerova et al. (1984) and Fisher et al. (1998).
^ 'Mouse value is the muscle value from Abbas and Fisher (1997). Rat value is the muscle value from pooling Barton et al. (1995), Sato et al. (1977), and Fisher et
^ al. (1989). Human value is the muscle value from pooling Fiserova-Bergerova et al. (1984) and Fisher et al. (1998).
o "Scaling parameter is the effective partition coefficient between red blood cells and plasma. Thus, the TCA blood-plasma concentration ratio depends on the
g plasma fraction. Baseline value is based on the blood-plasma concentration ratio of 0.76 in rats (Schultz et al., 1999).
S '/« vitro partition coefficients were determined at high concentration, when plasma binding is saturated, so should reflect the free blood:tissue partition
». coefficient. To get the plasma partition coefficient, the partition coefficient is multiplied by the blood:plasma concentration ratio (TCAPlas). In vitro values
s were from Abbas and Fisher (1997) in the mouse (used for both mouse and rat) and from Fisher et al. (1998). Body values based on measurements in muscle.
? "Values are based on the geometric mean of estimates based on data from Lumpkin et al. (2003), Schultz et al. (1999), Templin et al. (1993, 1995), and Yu et al.
Oq (2000). Scaling parameter for BMAX is actually the ratio of BMAx/kD, which determines the binding at low concentrations.
~3 vData are from Abbas and Fisher (1997) in the mouse (used for the mouse and rat) and Fisher et al. (1998) (human).
^ "Used in vitro measurements in TCOH as a proxy, but higher uncertainty is noted.
§ "The scaling parameter (only used in the human model) is the effective partition coefficient for the "body" (nonblood) compartment, so that the distribution
^ volume VDCVG is given by VBld + exp(lnPeffDCVG) x (VBod + VLiv).
-------
Table A-4. PBPK model parameters, baseline values, and scaling relationships (continued)
yBaseline values have the following units: for VMax, mg/hour/kg liver; for KM, mg/L blood; and for clearance (Cl), L/hour/kg liver (in humans, KM is calculated
from KM = VMax/(exp(lnClC) x Vliv). Values are based on in vitro (microsomal and hepatocellular preparations) from Elfarra et al. (1998), Lipscomb et al.
(1997, 1998a, b). Scaling from in vitro data based on 32 mg microsomal protein/g liver and 99 x 106 hepatocytes/g liver (Barter et al., 2007). Scaling of KM
from microsomes were based on two methods: (1) assuming microsomal concentrations equal to liver tissue concentrations and (2) using the measured
microsome:air partition coefficient and a central estimate of the blood:air partition coefficient. For KM from human hepatocyte preparations, the measured
hepatocyte:air partition coefficient and a central estimate of the blood:air partition coefficient was used.
zScaling parameter is ratio of "DCA" to "non-DCA" oxidative pathway (where DCA is a proxy for oxidative metabolism not producing TCA or TCOH).
Fraction of "other" oxidation is exp(lnFracOtherC)/(l + exp[lnFracOtherC]).
aaScaling parameter is ratio of TCA to TCOH pathways. Baseline value based on geometric mean of Lipscomb et al. (1998b) using fresh hepatocytes and
Bronley-DeLancey et al. (2006) using cryogenically-preserved hepatocytes. Fraction of oxidation to TCA is
(1 -FracOther) x exp(lnFracTCAC)/(l +exp[lnFracTCAC]).
bbBaseline values are based on in vitro data. In the mouse and rat, the only in vitro data are at 1 or 2 mM (Lash et al., 1995, 1998). In most cases, rates at 2 mM
were increased over the same sex/species at 1 mM, indicating VMax has not yet been reached. These data therefore put lower bounds on both VMax (in units of
mg/hour/kg tissue) and clearance (in units of L/hour/kg tissue), so those are the scaling parameters used, with those bounds used as baseline values. For
humans, data from Lash et al. (1999a) in the liver (hepatocytes) and the kidney (cytosol) and Green et al. (1997) (liver cytosol) was used to estimate the
clearance in units of L/hour/kg tissue and KM in units of mg/L in blood.
ccScaling parameter is the ratio of the lung to liver VMax (each in units of mg/hour), with baseline values based on microsomal preparations (mg/hour/mg protein)
assayed at ~1 mM (Green et al., 1997), further adjusted by the ratio of lung to liver tissue masses (Brown et al., 1997; ICRP Publication 89 [2003]).
ddScaling parameter is the ratio of respiratory oxidation entering systemic circulation (translocated to the liver) to that locally cleared in the lung. Fraction of
respiratory oxidation entering systemic circulation is exp(lnFracLungSysC)/(l + exp[lnFracLungSysC]).
eeBaseline parameters for urinary clearance (L/hour) were based on glomular filtration rate per unit body weight (L/hour/kg BW) from Lin (1995), multiplied by
the body weights cited in the study. For TCA, these were scaled by plasma volume to obtain the rate constant (/hour), since the model clears TCA from
plasma. For TCOG, these were scaled by the effective distribution volume of the body (VBodTCOH x PBodTCOG) to obtain the rate constant (/hour), since
the model clears TCOG from the body compartment.
ffHuman model only.
ggRat and human models only.
^Baseline value for oral absorption scaling parameter are as follows: kTSD and kAS, 1.4/hour, based on human stomach half time of 0.5 hour; kAD, kASTCA,
and kASTCOH, 0.75/hour, based on human small intestine transit time of 4 hours (ICRP Publication 89, 2003). These are noted to have very high uncertainty.
DCVG = S-dichlorovinyl glutathione.
-------
1 A.4.2. Statistical Distributions for Parameter Uncertainty and Variability
2 A.4.2.1. Initial Prior Uncertainty in Population Mean Parameters
3 The following multipage Table A-5 describes the initial prior distributions for the
4 population mean of the PBPK model parameters. For selected parameters, rat prior distributions
5 were subsequently updated using the mouse posterior distributions, and human prior distributions
6 were then updated using mouse and rat posterior distributions (see Section A.4.2.2).
7
8 A.4.2.2. Interspecies Scaling to Update Selected Prior Distributions in the Rat and Human
9 As shown in Table A-5, for several parameters, there is little or no in vitro or other prior
10 information available to develop informative prior distributions, so many parameters had
11 lognormal or log-uniform priors that spanned a wide range. Initially, the PBPK model for each
12 species was run with the initial prior distributions in Table A-5, but, in the time available for
13 analysis (up to about 100,000 iterations), only for the mouse did all these parameters achieve
14 adequate convergence. Additional preliminary runs indicated replacing the log-uniform priors
15 with lognormal priors and/or requiring more consistency between species could lead to adequate
16 convergence. However, an objective method of "centering" the lognormal distributions that did
17 not rely on the in vivo data (e.g., via visual fitting or limited optimization) being calibrated
18 against was necessary in order to minimize potential bias.
19 Therefore, the approach taken was to consider three species sequentially, from mouse to
20 rat to human, and to use a model for interspecies scaling to update the prior distributions across
21 species (the original prior distributions define the prior bounds). This sequence was chosen
22 because the models are essentially "nested" in this order—the rat model adds to the mouse model
23 the "downstream" GSH conjugation pathways, and the human model adds to the rat model the
24 intermediary S-dichlorovinyl glutathione (DCVG) compartment. Therefore, for those
25 parameters with little or no independent data only, the mouse posteriors were used to update the
26 rat priors, and both the mouse and rat posteriors were used to update the human priors. A list of
27 the parameters for which this scaling was used to update prior distributions is contained in
28 Table A-6, with the updated prior distributions. The correspondence between the "scaling
29 parameters" and the physical parameters generally follows standard practice, and were explicitly
30 described in Table A-4. For instance, VMAX and clearance rates are scaled by body weight to the
31 % power, whereas KM values are assumed to have no scaling, and rate constants (inverse time
32 units) are scaled by body weight to the -!/4 power.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-46 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-5. Uncertainty distributions for the population mean of the PBPK model parameters
to
VO C?'
I
I
§
***.
£3'
1
TO'
I
TO
Scaling (sampled)
parameter
Mouse
Distribution3
SD or
Min
Truncation
(±nxSD) or
Max
Rat
Distribution
SD or
Min
Truncation
(±nxSD) or
Max
Human
Distribution
SD or
Min
Truncation
(±nxSD) or
Max
Notes/
Source
Flows
InQCC
InVPRC
InDRespC
TruncNormal
TruncNormal
Uniform
0.2
0.2
-11.513
4
4
2.303
TruncNormal
TruncNormal
Uniform
0.14
0.3
-11.513
4
4
2.303
TruncNormal
TruncNormal
Uniform
0.2
0.2
-11.513
4
4
2.303
a
a
b
Physiological blood flows to tissues
QFatC
QGutC
QLivC
QSIwC
QKidC
FracPlasC
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.46
0.17
0.17
0.29
0.32
0.2
2
2
2
2
2
3
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.46
0.17
0.17
0.3
0.13
0.2
2
2
2
2
2
3
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.46
0.18
0.45
0.32
0.12
0.05
2
2
2
2
2
3
a
a
a
a
a
c
Physiological volumes
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.45
0.13
0.24
0.1
0.11
0.11
0.1
0.12
2
2
2
2
2
2
2
2
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.45
0.13
0.18
0.12
0.18
0.18
0.15
0.12
2
2
2
2
2
2
2
2
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.45
0.08
0.23
0.08
0.2
0.2
0.17
0.12
2
2
2
2
2
2
2
2
a
a
a
a
a
a
a
a
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
00 O
•••2
I
TO
Table A-5. Uncertainty distributions for the population mean of the PBPK model parameters (continued)
Scaling (sampled)
parameter
Mouse
Distribution3
SD or
Min
Truncation
(±nxSD) or
Max
Rat
Distribution
SD or
Min
Truncation
(±nxSD) or
Max
Human
Distribution
SD or
Min
Truncation
(±nxSD) or
Max
Notes/
Source
TCE distribution/partitioning
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.25
0.3
0.4
0.4
0.4
0.4
0.4
0.4
3
3
3
3
3
3
3
3
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.25
0.3
0.4
0.15
0.4
0.4
0.3
0.3
3
3
3
3
3
3
3
3
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
0.2
0.2
0.4
0.4
0.4
0.4
0.2
0.3
3
3
3
3
3
3
3
3
d
TCA distribution/partitioning
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
Uniform
TruncNormal
TruncNormal
-4.605
0.336
0.336
4.605
3
3
TruncNormal
TruncNormal
TruncNormal
0.336
0.693
0.693
3
3
3
Uniform
TruncNormal
TruncNormal
-4.605
0.336
0.336
4.605
3
3
e
f
TCA plasma binding
InkDissocC
InBMaxkDC
TruncNormal
TruncNormal
1.191
0.495
3
3
TruncNormal
TruncNormal
0.61
0.47
3
3
TruncNormal
TruncNormal
0.06
0.182
3
3
g
TCOH and TCOG distribution/partitioning
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
TruncNormal
TruncNormal
Uniform
Uniform
0.336
0.336
-4.605
-4.605
3
3
4.605
4.605
TruncNormal
TruncNormal
Uniform
Uniform
0.693
0.693
-4.605
-4.605
3
3
4.605
4.605
TruncNormal
TruncNormal
Uniform
Uniform
0.336
0.336
-4.605
-4.605
3
3
4.605
4.605
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
2
•••2
I
TO
Table A-5. Uncertainty distributions for the population mean of the PBPK model parameters (continued)
Scaling (sampled)
parameter
Mouse
Distribution3
SDor
Min
Truncation
(±nxSD) or
Max
Rat
Distribution
SDor
Min
Truncation
(±nxSD) or
Max
Human
Distribution
SDor
Min
Truncation
(±nxSD) or
Max
Notes/
Source
DCVG distribution/partitioning
InPeffDCVG
Uniform
-6.908
6.908
Uniform
-6.908
6.908
Uniform
-6.908
6.908
h
TCE Metabolism
InVMAxC
lnKMC
InCIC
InFracOtherC
InFracTCAC
lnVMAXDCVGC
InCIDCVGC
lnK,v,DCVGC
InVMAxKidDCVGC
InCIKidDCVGC
lnKMKidDCVGC
InVMAxLungLivC
lnKMClara
InFracLungSysC
TruncNormal
TruncNormal
Uniform
TruncNormal
Uniform
Uniform
Uniform
Uniform
TruncNormal
Uniform
Uniform
0.693
1.386
-6.908
1.163
-4.605
-4.605
-4.605
-4.605
1.099
-6.908
-6.908
3
3
6.908
3
9.21
9.21
9.21
9.21
3
6.908
6.908
TruncNormal
TruncNormal
Uniform
TruncNormal
Uniform
Uniform
Uniform
Uniform
TruncNormal
Uniform
Uniform
0.693
1.386
-6.908
1.163
-4.605
-4.605
-4.605
-4.605
1.099
-6.908
-6.908
3
3
6.908
3
9.21
9.21
9.21
9.21
3
6.908
6.908
TruncNormal
TruncNormal
Uniform
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
TruncNormal
Uniform
Uniform
0.693
1.386
-6.908
1.163
4.605
1.386
4.605
1.386
1.099
-6.908
-6.908
3
3
6.908
3
3
3
3
3
3
6.908
6.908
i
i
i
h
J
k
k
k
k
k
k
'
h
h
TCOH metabolism
lnVMAXTCOHC
InCITCOHC
InKivJCOH
InVMAxGlucC
InCIGIucC
lnKMGluc
InkMetTCOHC
Uniform
Uniform
Uniform
Uniform
Uniform
-9.21
-9.21
-9.21
-6.908
-11.513
9.21
9.21
9.21
6.908
6.908
Uniform
Uniform
Uniform
Uniform
Uniform
-9.21
-9.21
-9.21
-6.908
-11.513
9.21
9.21
9.21
6.908
6.908
Uniform
Uniform
Uniform
Uniform
Uniform
-11.513
-9.21
-9.21
-6.908
-11.513
6.908
9.21
4.605
6.908
6.908
h
h
H I
O >
HH Oq
H TO
O
H
W
-------
to
VO Lo'
I
I
§
***.
£3'
1
TO'
°§
I
TO
Table A-5. Uncertainty distributions for the population mean of the PBPK model parameters (continued)
Scaling (sampled)
parameter
Mouse
Distribution3
SDor
Min
Truncation
(±nxSD) or
Max
Rat
Distribution
SDor
Min
Truncation
(±nxSD) or
Max
Human
Distribution
SDor
Min
Truncation
(±nxSD) or
Max
Notes/
Source
TCA metabolism/clearance
InkUrnTCAC
InkMetTCAC
Uniform
Uniform
-4.605
-9.21
4.605
4.605
Uniform
Uniform
-4.605
-9.21
4.605
4.605
Uniform
Uniform
-4.605
-9.21
4.605
4.605
h
TCOG metabolism/clearance
InkBileC
InkEHRC
InkUrnTCOGC
Uniform
Uniform
Uniform
-9.21
-9.21
-6.908
4.605
4.605
6.908
Uniform
Uniform
Uniform
-9.21
-9.21
-6.908
4.605
4.605
6.908
Uniform
Uniform
Uniform
-9.21
-9.21
-6.908
4.605
4.605
6.908
h
DCVG metabolism
InFracKidDCVCC
InkDCVGC
Uniform
Uniform
-6.908
-9.21
6.908
4.605
Uniform
Uniform
-6.908
-9.21
6.908
4.605
Uniform
Uniform
-6.908
-9.21
6.908
4.605
h
DCVC metabolism/clearance
InkNATC
InkKidBioactC
Uniform
Uniform
-9.21
-9.21
4.605
4.605
Uniform
Uniform
-9.21
-9.21
4.605
4.605
Uniform
Uniform
-9.21
-9.21
4.605
4.605
h
Oral uptake/transfer coefficients
InkTSD
InkAS
InkTD
InkAD
InkASTCA
InkASTCOH
Uniform
Uniform
Uniform
Uniform
Uniform
Uniform
-4.269
-6.571
-4.605
-7.195
-7.195
-7.195
4.942
7.244
0
6.62
6.62
6.62
Uniform
Uniform
Uniform
Uniform
Uniform
Uniform
-4.269
-6.571
-4.605
-7.195
-7.195
-7.195
4.942
7.244
0
6.62
6.62
6.62
Uniform
Uniform
Uniform
Uniform
Uniform
Uniform
-4.269
-6.571
-4.605
-7.195
-7.195
-7.195
4.942
7.244
0
6.62
6.62
6.62
h
h
H I
O >
HH Oq
H TO
O
Explanatory note. All population mean parameters have either truncated normal (TruncNormal) or uniform distributions. For those with TruncNormal
distributions, the mean for the population mean is 0 for natural-log transformed parameters (parameter name starting with "In") and 1 for untransformed
parameters, with the truncation at the specified number (n) of standard deviations (SD). All uniformly distributed parameters are natural-log transformed, so
their untransformed minimum and maximum are exp(Min) and exp(Max), respectively.
H
W
-------
o Table A-5. Uncertainty distributions for the population mean of the PBPK model parameters (continued)
to
O ko
o ^ ""Uncertainty based on CV or range of values in Brown et al. (1997) (mouse and rat) and a comparison of values from ICRP Publication 89 (2003), Brown et al.
"° I" (1997), and Price etal. (2003) (human).
o" bNoninformative prior distribution intended to span a wide range of possibilities because no independent data are available on these parameters. These priors for
§ the rat and human were subsequently updated (see Section A.4.2.2).
H 'Because of potential strain differences, uncertainty in mice and rat assumed to be 20%. In humans, Price et al. (2003) reported variability of about 5%, and this
31 is also used for the uncertainty in the mean.
S3' dFor partition coefficients, it is not clear whether interstudy variability is due to interindividual or assay variability, so uncertainty in the mean is based on
a interstudy variability among in vitro measurements. For single measurements, uncertainty SD of 0.3 was used for fat (mouse) and 0.4 for other tissues was
g" used. In addition, where measurements were from a surrogate tissue (e.g., gut was based on liver and kidney), an uncertainty SD 0.4 was used.
^j> eSingle in vitro data point available in rats, so a geometric standard deviation (GSD) of 1.4 was used. In mice and humans, where no in vitro data was available,
^ a noninformative prior was used.
^ fSingle in vitro data points available in mice and humans, so a GSD of 1.4 was used. In rats, where the mouse data was used as a surrogate, a GSD of 2.0 was
^ used, based on the difference between mice and rats in vitro.
g' 8GSD for uncertainty based on different estimates from different in vitro studies.
^ hNoninformative prior distribution.
s 'Assume 2-fold uncertainty GSD in VMax, based on observed variability and uncertainties of in vitro-to-in vivo scaling. For KM and C1C, the uncertainty is
^ assumed to be 4-fold, due to the different methods for scaling of concentrations from TCE in the in vitro medium to TCE in blood.
i> % Uncertainty GSD of 3.2-fold reflects difference between in vitro measurements from Lipscomb et al. (1998b) and Bronley-DeLancey et al. (2006).
^ ^ kln mice and rats, the baseline values are notional lower-limits on VMax and clearance, however, the lower bound of the prior distribution is set to 100-fold less
g^ because of uncertainty in in vitro-in vivo extrapolation, and because Green et al. (1997) reported values 100-fold smaller than Lash et al. (1995, 1998). In
j? humans, the uncertainty GSD in clearance is assumed to be 100-fold, due to the difference between Lash et al. (1998) and Green et al. (1997). For KM, the
a uncertainty GSD of 4-fold is based on differences between concentrations in cells and cytosol.
Q 5^ 'Uncertainty GSD of 3-fold was assumed due to possible differences in microsomal protein content, the fact that measurements were at a single concentration,
5d ^ and the fact that the human baseline values was based on the limit of detection.
r> co
H O DCVG = S-dichlorovinyl glutathione, SD = standard deviation.
-------
1
2
Table A-6. Updated prior distributions for selected parameters in the rat
and human
Scaling parameter
InDRespC
InPBodTCOGC
InPLivTCOGC
InFracOtherC
lnVMAXDCVGC
InCIDCVGC
InVMAxKidDCVGC
InCIKidDCVGC
InViuAxLungLivC
lnKMClara
InFracLungSysC
lnVMAXTCOHC
InCITCOHC
InKivJCOH
InVMAxGlucC
InCIGIucC
lnKMGluc
InkMetTCOHC
InkUrnTCAC
InkMetTCAC
InkBileC
InkEHRC
InkUrnTCOGC
InkNATC
InkKidBioactC
Initial prior bounds
exp(min)
1.00E-05
1.00E-02
1.00E-02
1.00E-03
1.00E-02
1.00E-02
1.00E-02
1.00E-02
3.70E-02
1.00E-03
1.00E-03
1.00E-04
1.00E-05
1.00E-04
1.00E-04
1.00E-04
1.00E-03
1.00E-05
1.00E-02
1.00E-04
1.00E-04
1.00E-04
1.00E-03
1.00E-04
1.00E-04
exp(max)
1.00E+01
1.00E+02
1.00E+02
1.00E+03
1.00E+04
1.00E+04
1.00E+04
1.00E+04
2.70E+01
1.00E+03
1.00E+03
1.00E+04
1.00E+03
1.00E+04
1.00E+04
1.00E+02
1.00E+03
1.00E+03
1.00E+02
1.00E+02
1.00E+02
1.00E+02
1.00E+03
1.00E+02
1.00E+02
Updated rat prior
exp(M)
1.22
0.42
1.01
0.02
2.61
0.36
2.56
1.22
2.77
0.01
4.39
1.65
0.93
69.41
30.57
3.35
0.11
0.61
1.01
0.01
8.58
exp(a)
5.21
5.47
5.31
6.82
42.52
15.03
22.65
15.03
6.17
6.69
11.13
5.42
5.64
5.58
6.11
5.87
5.42
5.37
5.70
6.62
6.05
Updated human prior
exp(M)
1.84
0.81
2.92
0.14
2.80
0.02
3.10
0.37
4.81
3.39
11.13
2.39
0.09
0.45
3.39
0.22
16.12
0.00
0.01
exp(a)
4.18
5.10
4.31
4.76
4.71
4.85
8.08
4.44
4.53
4.35
4.57
4.62
4.22
4.26
4.44
4.71
4.81
6.11
6.49
4
5
6
7
8
9
10
11
12
13
14
15
16
Notes: updated rat prior is based on the mouse posterior; and the updated human priors are based on combining the
mouse and rat posteriors, except in the case of InkNATC and InkKidBioactC, which are unidentified in the mouse
model. Columns labeled exp(min) and exp(max) are the exponentiated prior bounds; columns labeled exp(u) and
exp(o) are the exponentiated mean and standard deviation of the updated prior distributions, which are normal
distributions truncated at the prior bounds.
The scaling model is given explicitly as follows. If 0, are the "scaling" parameters
(usually also natural-log-transformed) that are actually estimated, and A is the "universal"
(species-independent) parameter, then 9, = A + s/, where et is the species-specific "departure"
from the scaling relationship, assumed to be normally distributed with variance oe2. This
"scatter" in the interspecies scaling relationship is assumed to have a standard deviation of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-52 DRAFT—DO NOT CITE OR QUOTE
-------
1 1.15 = ln(3.16), so that the un-logarithmically transformed 95% confidence interval spans about
2 100-fold (i.e., exp(2o) = 10). This implies that 95% of the time, the species-specific scaling
3 parameter is assumed be within 10-fold higher or lower than the "species-independent" value.
4 However, the prior bounds, which generally span a wider range, are maintained so that if the data
5 strongly imply an extreme species-specific value, it can be accommodated.
6 Therefore, the mouse model gives an initial estimate of "A," which is used to update the
7 prior distribution for 0r = A + sr in the rat (alternatively, since there is only one species at this
8 stage, one could think of this as estimating the rat parameter using the mouse parameter, but with
9 a cross-species variance is twice the allometric scatter variance). The rat and mouse together
10 then give a "better" estimate of A, which is used to update the prior distribution for 0/, = A + s/, in
11 the human, with the assumed distribution for s/,. This approach is implemented by
12 approximating the posterior distributions by normal distributions, deriving heuristic "data" for
13 the specific-specific parameters, and then using these pseudo-data to derive updated prior
14 distributions for the other species parameters. Specifically, the procedure is as follows:
15
16 1. Run the mouse model.
17 2. Use the mouse posterior to derive the mouse "pseudo-data" Dm (equal to the posterior
18 mean) and its uncertainty om2 (equal to the posterior variance).
19 3. Use the Dm as the prior mean for the rat. The prior variance for the rat is 2oe2 + om2,
20 which accounts for two components of species-specific departure from "species-
21 independence" (one each for mouse and rat), and the mouse posterior uncertainty.
22 4. Match the rat posterior mean and variance to the values derived from the normal
23 approximation (posterior mean = (Dm/(2oe2 + om2) + Dr/or2}/{ l/(2oe2 + om2) + l/or2};
24 posterior variance = {l/(2oe2 + om2) + I/Or2}"1), and solve for the rat "data" Dr and its
25 uncertainty or2.
26 5. Use, om2, and or2 to derive the updated prior mean and variance for the human model.
27 For the mean (={Dm/(oe2 + om2) + Dr/(oe2 + or2)}/{ l/(oe2 + om2) + l/(oe2 + or2)}), it is the
28 weighted average of the mouse and rat, with each weight including both posterior
29 uncertainty and departure from "species-independence." For the variance (={ l/(oe2 +
30 om2) + l/(oe2 + Or2)}"1 + Ge2), it is the variance in the weighted average of the mouse and
31 rat plus an additional component of species-specific departure from "species-
32 independence."
33
34 Formally, then, the probability of 0, given A can be written as
35
36 P(Qi | A) = 9(0, -A, GE2) (Eq. A-5)
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-53 DRAFT—DO NOT CITE OR QUOTE
-------
1 where (p(x, G ) is the normal density centered on 0 with variance G . Let D, be
a heuristic
2 "datum" for species /', so the likelihood given 0, is adequately approximated by
3
4 P(Dt \ 0;) = 9(0,- - 0,-, G,2)
5
(Eq. A-6)
6 Therefore, considering A to have a uniform prior distribution, then running the mouse model
7 gives a posted or of the form
9 P(A, Qm\Dm)xP(A)P(Qm A)P(Dm 0m) ex: cp(0m - A, GE2) cp(Dm - 0m,
10
1 1 From the MCMC posterior, the values of Dm and Gm2 are simply the mean and
12 scaled parameter 0m.
13
14 Now, adding the rat data gives
15
16 P(A, 0m, 0r Dm, Dr) oc P(A) P(0m A) P(Vm 0m) P(Qr A) P(Dr \ 0r)
17 QC cp(0m - A, Ge2) (p(Dm - 0m, cm2) (p(0r - A, Ge2) (p(Dr - 0r, cr2)
18
19 Dr and Gr2 can be derived by marginalizing first over 0m and then over A:
20
21 \P(A, 0m, Qr\Dm,Dr)dQmdA
22 oc [J P(A) (I P(0m A) P(Dm 0m) d0m} P(0r A) dA ]P(Dr \ 0r)
23 =[\P(A}P(Dm ^)P(0r ^)cy]P(Dr 0r)
24 oc [J P(A \ Dm) P(Qr A) dA] P(Dr \ 0r)
25 = P(0r | Dm) P(Dr | 0r)
26
27 So P(Qr Dm) can be identified as the prior for 0r based on the mouse data, and
28 rat-specific likelihood. The updated prior for the rats is then
29
30 P(Qr \ Dm) oc J (p(0m -A, GE2) cp(Dm - 0m, Gm2) 9(0r - A, GE2) d0m dA
31 =J (p(Dm - ^, ce2 + Gm2) 9(0r -A, Ge2) dA
32 = 9(Dm - 0r, 2ce2 + Gm2)
33
34 Therefore, for the "mouse-based" prior, use the mean Dm from the mouse, and
35 from the mouse Gm2 plus twice the "allometric scatter" variance Ge2.
Gm2) (Eq. A-7)
variance of the
(Eq. A-8)
(Eq. A-9)
(Eq. A- 10)
(Eq. A-ll)
(Eq. A- 12)
(Eq. A-13)
P(Dr 0r)asthe
(Eq. A- 14)
(Eq. A- 15)
(Eq. A- 16)
then the variance
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-54 DRAFT—DO NOT CITE OR QUOTE
-------
1 The rat "data" and variance, assuming conditional independence of the rat and mouse
2 data," is thus
3
4 P(Qr\Dm,Dr)ccP(Qr Dm)P(Dr 0r)
5 oc cp(Dm - 0r, 2oE2 + om2) cp(Dr - 0r, or2)
6
7 This distribution is also normal with
9 E(0r) = (Dm/(2oe2 + om2) + Dr/or2}/ { l/(2oe2 + om2) + l/or2} = weighted mean of Dr
10 VAR(0r) = { l/(2oe2 + om2) + I/O,2}"1 = harmonic mean of variances
11
"pseudo-
(Eq. A- 17)
(Eq. A- 18)
(Eq. A- 19)
(Eq. A-20)
12 Thus, using the mean and variance of the posterior distribution from the MCMC analysis,
13 Dr and or2 can be derived.
14 Now, Dm, om2, Dr, and or2 are known, so the analogous "mouse+rat" based prior used in
15 the human model can be derived. As with the rat prior, the human prior is based on a
normal
16 approximation of the posterior for^4, and then incorporates a random term for cross-species
17 variation (allometric scatter).
18
19 P(A, 0m, 0r, Qh \ Dm, Dr, D/0
20 ccP(A)P(Qm\A)P(Dm\Qm)P(Qr\A)P(pr 0r) P(Qh \ A) P(Dh 0,)
21 oc y(Qm-A, GE2) q>(Dm - 0m, om2) q>(0r -A, oe2) q>(Dr - 0r, or2)
22 9(6^-^4, oe2) ^(D/, - 0fe a/,2)
23
24 Consider marginalizing first over 0m, then over 0r, and then over A:
25
26 J P(A, 0m, 0r, 0/, | Dm, Dr, D/,) d0m d0r dA
27 ex: [J P(A) (I P(0m ^) P(Dm 0m) d0m} (J P(0r ^) P(Dr \ 0r) d0r} P(0A A) dA
28 P(D, 0,)
29 =[|P(A)P(Dm A)P(Dr A)P(0, A) dA ] P(D, 0,)
30 K[$P(A\DmDr)P(Qh\A)dA]P(Dh 0,)
31 =P(Qh\DmDr)P(Dh\Qh)
32
33 So /"(0/j Dm Dr) is the prior for 0/, based on the mouse and rat data, and P(D/,
(Eq. A-21)
(Eq. A-22)
(Eq. A-23)
(Eq. A-24)
(Eq. A-25)
(Eq. A-26)
Qh) as the
34 human-specific likelihood. The prior is used in the MCMC analysis for the humans, and it is
35 derived to be
36
772/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-55 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
P(Qh
9(0
Dm Dr) oc J 9(0m -A, Ge2) 9(Dm - 0m, Gm2) 9(0r -A, Ge2) 9(Dr - 0r, Gr2)
h-A, ce2) d0m d0r cL4
= J [9(Dm -A, GE2 + Gm2)
-------
1
2
Table A-7. Uncertainty distributions for the population variance of the
PBPK model parameters
Scaling (sampled)
parameter
Mouse
CV
CU
Rat
CV
CU
Human
CV
CU
Notes/
source
Flows
InQCC
InVPRC
InDRespC
0.2
0.2
0.2
2
2
0.5
0.14
0.3
0.2
2
2
0.5
0.2
0.2
0.2
2
2
0.5
a
Physiological blood flows to tissues
QFatC
QGutC
QLivC
QSIwC
QKidC
FracPlasC
0.46
0.17
0.17
0.29
0.32
0.2
0.5
0.5
0.5
0.5
0.5
0.5
0.46
0.17
0.17
0.3
0.13
0.2
0.5
0.5
0.5
0.5
0.5
0.5
0.46
0.18
0.45
0.32
0.12
0.05
0.5
0.5
0.5
0.5
0.5
0.5
a
Physiological volumes
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
0.45
0.13
0.24
0.1
0.11
0.11
0.1
0.12
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.45
0.13
0.18
0.12
0.18
0.18
0.15
0.12
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.45
0.08
0.23
0.08
0.2
0.2
0.17
0.12
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
a
TCE distribution/partitioning
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
0.25
0.3
0.4
0.4
0.4
0.4
0.4
0.4
2
2
2
2
2
2
2
2
0.25
0.3
0.4
0.15
0.4
0.4
0.3
0.3
0.333
0.333
2
0.333
2
2
0.577
0.333
0.185
0.2
0.4
0.4
0.4
0.4
0.2
0.3
0.333
1
2
1.414
2
2
1.414
1.414
b
TCA distribution/partitioning
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
0.336
0.336
0.336
2
2
2
0.336
0.693
0.693
2
2
2
0.336
0.336
0.336
2
2
2
c
b
TCA plasma binding
InkDissocC
InBMaxkDC
1.191
0.495
2
2
0.61
0.47
2
2
0.06
0.182
2
2
b
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-57 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-7. Uncertainty distributions for the population variance of the
PBPK model parameters (continued)
Scaling (sampled)
parameter
Mouse
CV
cu
Rat
CV
CU
Human
CV
CU
Notes/
source
TCOH and TCOG distribution/partitioning
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
0.336
0.336
0.4
0.4
2
2
2
2
0.693
0.693
0.4
0.4
2
2
2
2
0.336
0.336
0.4
0.4
2
2
2
2
b
b
d
d
DCVG distribution/partitioning
InPeffDCVG
0.4
2
0.4
2
0.4
2
b
TCE metabolism
InVMAxC
lnKMC
InCIC
InFracOtherC
InFracTCAC
lnVMAXDCVGC
InCIDCVGC
lnK,v,DCVGC
InVMAxKidDCVGC
InCIKidDCVGC
lnKMKidDCVGC
InVMAxLungLivC
lnKMClara
InFracLungSysC
0.824
0.270
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
1
1
2
2
2
2
2
2
2
2
2
0.806
1.200
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
1
1
2
2
2
2
2
2
2
2
2
0.708
0.944
0.5
1.8
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.26
1.41
2
2
2
2
2
2
2
2
2
e
f
g
f
TCOH metabolism
InVMAxTCOHC
InCITCOHC
InKwJCOH
InVMAxGlucC
InCIGIucC
lnKMGluc
InkMetTCOHC
0.5
0.5
0.5
0.5
0.5
2
2
2
2
2
0.5
0.5
0.5
0.5
0.5
2
2
2
2
2
0.5
0.5
0.5
0.5
0.5
2
2
2
2
2
f
TCA metabolism/clearance
InkUrnTCAC
InkMetTCAC
0.5
0.5
2
2
0.5
0.5
2
2
0.5
0.5
2
2
f
TCOG metabolism/clearance
InkBileC
InkEHRC
0.5
0.5
2
2
0.5
0.5
2
2
0.5
0.5
2
2
f
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-58 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-7. Uncertainty distributions for the population variance of the
PBPK model parameters (continued)
Scaling (sampled)
parameter
InkUrnTCOGC
Mouse
CV
0.5
CU
2
Rat
CV
0.5
CU
2
Human
CV
0.5
CU
2
Notes/
source
t
DCVG metabolism/clearance
InFracKidDCVCC
InkDCVGC
0.5
0.5
2
2
0.5
0.5
2
2
0.5
0.5
2
2
f
DCVC metabolism/clearance
InkNATC
InkKidBioactC
0.5
0.5
2
2
0.5
0.5
2
2
0.5
0.5
2
2
f
Oral uptake/transfer coefficients
InkTSD
InkAS
InkTD
InkAD
InkASTCA
InkASTCOH
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Explanatory note. All population variance parameters (V_pname, for parameter "pname") have Inverse-Gamma
distributions, with the expected value given by CV and coefficient of uncertainty given by CU (i.e., standard
deviation of V_pname divided by expected value of V_pname) (notation the same as Hack et al. [2006]). Under
these conditions, the Inverse-Gamma distribution has a shape parameter is given by a = 2 + 1/CU2 and scale
parameter p = (a - 1) CV2. In addition, it should be noted that, under a normal distribution and a uniform prior
distribution on the population variance, the posterior distribution for the variance given n data points with a sample
variance s2 is given by and Inverse-Gamma distribution with a = (« - l)/2 and fi = as2. Therefore, the "effective"
number of data points is given by n = 5 + 2/CU2 and the "effective" sample variance is s2 = CV2 oc/(oc - 1).
Tor physiological parameters, CV values generally taken to be equal to the uncertainty SD in the population mean,
most of which were based on variability between studies (i.e., not clear whether variability represents uncertainty
or variability). Given this uncertainty, CU of 2 assigned to cardiac output and ventilation-perfusion, while CU of
0.5 assigned to the remaining physiological parameters.
bAs discussed above, it is not clear whether interstudy variability is due to interindividual or assay variability, so the
same central were assigned to the uncertainty in the population mean as to the central estimate of the population
variance. In the cases were direct measurements were available, the CU for the uncertainty in the population
variance is based on the actual sample n, with the derivation discussed in the notes preceding this table.
Otherwise, a CU of 2 was assigned, reflecting high uncertainty.
°Used value from uncertainty in population in mean in rats for all species with high uncertainty.
dNo data, so assumed CV of 0.4 with high uncertainty.
Tor mice and rats, based on variability in results from Lipscomb et al. (1998a) and Elfarra et al. (1998) in
microsomes. Since only pooled or mean values are available, CU of 1 was assigned (moderate uncertainty). For
humans, based on variability in individual samples from Lipscomb et al. (1997) (microsomes), Elfarra et al.
(1998) (microsomes) and Lipscomb et al. (1998a) (freshly isolated hepatocytes). High uncertainty in clearance
(InCIC) reflects two different methods for scaling concentrations in microsomal preparations to blood
concentrations: (1) assuming microsomal concentration equals liver concentration and then using the measured
liverblood partition coefficient to convert to blood and (2) using the measured microsome:air partition coefficient
and then using the measured blood:air partition coefficient to convert to blood.
fNo data on variability, so a CV of 0.5 was assigned, with a CU of 2.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-59 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table A-7. Uncertainty distributions for the population variance of the
2 PBPK model parameters (continued)
O
4 gFor mice and rats, no data on variability, so a CV of 0.5 was assigned, with a CU of 2. For humans, 6-fold
5 variability based on in vitro data from Bronley-DeLancy et al. (2006), but with high uncertainty.
6 hNo data on variability, so a CV of 2 was assigned (larger than assumed for metabolism due to possible vehicle
7 effects), with a CU of 2.
8
9
10 A.4.2.4. Prior distributions for Residual Error Estimates
11 In all cases except one, the likelihood was assumed to be lognormal, which requires
12 specification of the variance of the "residual error." This error may include variability due to
13 measurement error, intraindividual and intrastudy heterogeneity, as well as model
14 misspecification. The available in vivo measurements to which the model was calibrated are
15 listed in Table A-8. The variances for each of the corresponding residual errors were given log-
16 uniform distributions. For all measurements, the bounds on the log-uniform distribution was
17 0.01 and 3.3, corresponding to geometric standard deviations bounded by 1.11 and6.15. The
18 lower bound was set to prevent "over-fitting," as was done in Bois (2000a) and Hack et al.
19 (2006).
20 Nondetects of DCVG from Lash et al. (1999b) were also included in the data, at it was
21 found that these data were needed to place constraints on the clearance rate of DCVG from
22 blood. The detection limit reported in the study was LD = 0.05 pmol/mL= 5 x 10"5 mmol/L. It
23 was assumed, as is standard in analytical chemistry, that the detection limit represents a response
24 from a blank sample at 3-standard deviations. Because detector responses near the detection
25 limit are generally normally distributed, the likelihood for observing a nondetect given a model-
26 predicted value of yp is equal to />(ND[y/,) =
-------
1
2
Table A-8 Measurements used for calibration
Measurement
abbreviation
RetDose
CAIvPPM
ClnhPPM
CArt
CVen
CBIdMix
CFat
CGut
CKid
CLiv
CMus
AExhpost
CTCOH
CLivTCOH
CPIasTCA
CBIdTCA
CLivTCA
AUrnTCA
AUrnTCA_collect
ABileTCOG
CTCOG
CTCOGTCOH
CLivTCOGTCOH
AUrnTCOGTCOH
AUrnTCOGTCOH_
collect
CDCVGmol
CDCVG_ND
AUrnNDCVC
AUrnTCTotMole
TotCTCOH
Mouse
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
Rat
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
V
Human
V
V
V
V
V
V
V
V
V
V
V
V
V
V
Measurement description
Retained TCE dose (mg)
TCE concentration in alveolar air (ppm)
TCE concentration in closed chamber (ppm)
TCE concentration in arterial blood (mg/L)
TCE concentration in venous blood (mg/L)
TCE concentration in mixed arterial and venous blood
(mg/L)
TCE concentration in fat (mg/L)
TCE concentration in gut (mg/L)
TCE concentration in kidney (mg/L)
TCE concentration in liver (mg/L)
TCE concentration in muscle (mg/L)
Amount of TCE exhaled postexposure (mg)
Free TCOH concentration in blood (mg/L)
Free TCOH concentration in liver (mg/L)
TCA concentration in plasma (mg/L)
TCA concentration in blood (mg/L)
TCA concentration in liver (mg/L)
Cumulative amount of TCA excreted in urine (mg)
Cumulative amount of TCA collected in urine
(noncontinuous sampling) (mg)
Cumulative amount of bound TCOH excreted in bile (mg)
Bound TCOH concentration in blood (mg/L)
Bound TCOH concentration in blood in free TCOH
equivalents (mg/L)
Bound TCOH concentration in liver in free TCOH
equivalents (mg/L)
Cumulative amount of total TCOH excreted in urine (mg)
Cumulative amount of total TCOH collected in urine
(noncontinuous sampling) (mg)
DCVG concentration in blood (mmol/L)
DCVG nondetects from Lash et al. (1999b)
Cumulative amount of NAcDCVC excreted in urine (mg)
Cumulative amount of TCA+total TCOH excreted in urine
(mmol)
Total TCOH concentration in blood (mg/L)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-61 DRAFT—DO NOT CITE OR QUOTE
-------
1 Rat
Human
4
5
6
7
8
9
10
11
12
13
14
15
Figure A-6. Updated hierarchical structure for rat and human models.
Symbols have the same meaning as Figure A-l, with modifications for the rat and
human. In particular, in the rat, each "group" consists of animals (usually
comprising multiple dose groups) of the same sex, species, and strain within a
study (possibly reported in more than one publication, but reasonably presumed to
be of animals in the same "lot"). Animals within each group are presumed to be
"identical," with the same PBPK model parameters, and each such group is
assigned its own set of "residual" error variances a2. In humans, each
"individual" is a single person, possibly exposed in multiple experiments, and
each individual is assigned a set of PBPK model parameters drawn from the
population. However, in humans, "residual" error variances are assigned at the
"study" level, rather than the individual or the population level.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 A.5. RESULTS OF UPDATED PHYSIOLOGICALLY BASED PHARMACOKINETIC
2 (PBPK) MODEL
3 The evaluation of the updated PBPK model was discussed in Chapter 3. Detailed results
4 in the form of tables and figures are provided in this section.
5
6 A.5.1. Convergence and Posterior Distributions of Sampled Parameters
7 For each sampled parameter (population mean and variance and the variance for residual
8 errors), summary statistics (median, [2.5%, 97.5%] confidence interval) for the posterior
9 distribution are tabulated in Tables A-9-A-14 below. In addition, the potential scale reduction
10 factor,/?, calculated from comparing four independent chains, is given.
11 In addition, posterior distributions for the group- or individual-specific parameters are
12 summarized in supplementary figures accessible here:
13
14 • Mouse: Appendix.linked.files\AppA.5.1.Mouse.posteriors.by.group.pdf
15 • Rat: Appendix.linked.files\AppA.5.1.Rat.posteriors.by.group.pdf
16 • Human: Appendix.linked.files\AppA.5.1.Human.posteriors.by.group.or.individual.pdf
17
18 A.5.2. Comparison of Model Predictions With Data
19 A.5.2.1. Mouse Model
20 A.5.2.1.1. Group-specific predictions and calibration data. [See
21 Appendix.linked.files\AppA.5.2.1.1. Updated.mouse.group.calib.TCE.DRAFT.pdf 1
22
23 A.5.2.1.2. Population-based predictions and calibration data. [See
24 Appendix.linked.files\AppA.5.2.1.2.Updated.mouse.pop.calib.TCE.DRAFT.pdf.1
25
26 A.5.2.2. Rat Model
27 A.5.2.2.1. Group-specific predictions and calibration data. [See
28 Appendix.linked.files\AppA.5.2.2.1.Updated.rat.group.calib.TCE.DRAFT.pdf.1
29
30 A.5.2.2.2. Population-based predictions and calibration data. [See
31 Appendix.linked.files\AppA.5.2.2.2.Updated.rat.pop.calib.TCE.DRAFT.pdf.1
32
33 A.5.2.2.3. Population-based predictions and additional evaluation data. [See
34 Appendix.linked.files\AppA.5.2.2.3.Updated.rat.pop.eval.TCE.DRAFT.pdf.1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-63 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-9. Posterior distributions for mouse PBPK model population
parameters
Sampled parameter*
InQCC
InVPRC
QFatC
QGutC
QLivC
QSIwC
InDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
InkDissocC
InBMaxkDC
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
InPeffDCVG
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
1.237(0.8972, 1.602)
0.8076 (0.6434, 1 .022)
1.034(0.5235, 1.55)
1.183(1.002, 1.322)
1.035(0.8002, 1.256)
0.9828 (0.6043, 1 .378)
1.214(0.7167,2.149)
0.995 (0.5642, 1 .425)
0.8707(0.5979, 1.152)
1.329(0.8537, 1.784)
0.9871 (0.817, 1.162)
0.8035(0.5609, 1.093)
0.997(0.8627, 1.131)
0.9995(0.8536, 1.145)
1 (0.8537, 1.148)
1.001 (0.8676, 1.134)
0.9916(0.8341, 1.153)
0.9259(0.647, 1.369)
0.9828(0.7039, 1.431)
0.805(0.4735, 1.418)
1.297(0.7687,2.039)
0.9529(0.5336, 1.721)
0.9918(0.5566, 1.773)
1.277(0.7274,2.089)
0.92(0.5585, 1.586)
2.495(1.144,5.138)
0.8816(0.6219, 1.29)
0.8003(0.5696, 1.15)
1.214(0.2527,4.896)
1.25(0.6793,2.162)
0.8025(0.5607, 1.174)
1.526(0.9099,2.245)
0.4241 (0.1555, 1.053)
1.013(0.492,2.025)
0.9807(0.008098, 149.6)
R
1
1
1
1
1
1
1.002
1
1.001
1.002
1
1.013
1
1
1.001
1
1.001
1
1.001
1
1
1
1.001
1
1.001
1.001
1.003
1.003
1.003
1.002
1
1
1.004
1.002
1.041
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
1.402(1.183,2.283)
1.224(1.108, 1.63)
0.436 (0.3057, 0.6935)
0.1548(0.1101,0.2421)
0.1593(0.1107,0.2581)
0.275(0.1915,0.4425)
1.215(1.143, 1.375)
0.3001 (0.21,0.48)
0.1903(0.1327,0.3039)
0.4123(0.2928,0.6414)
0.1219(0.085,0.1965)
0.2216(0.1552,0.3488)
0.09384(0.06519,0.1512)
0.1027(0.07172,0.1639)
0.1032(0.07176,0.1652)
0.09365(0.06523,0.1494)
0.1126(0.07835,0.1817)
1.644(1.278,3.682)
1.321 (1.16,2.002)
1.375(1.198,2.062)
1.415(1.21,2.342)
1.378(1.203,2.141)
1.378(1.2,2.066)
1.554(1.265,2.872)
1.411 (1.209,2.3)
1.398(1.178,2.623)
1.27(1.158, 1.609)
1.278(1.157, 1.641)
2.71 (1.765,8.973)
1.474(1.253,2.383)
1.314(1.17, 1.85)
1.399(1.194,2.352)
1.398(1.207,2.156)
1.554(1.279,2.526)
1.406(1.206,2.379)
R
1
1.001
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1.001
1
1
1
1
1
1.001
1.001
1
1.001
1
1
1.001
1
1
1
1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-64 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-9. Posterior distributions for mouse PBPK model population
parameters (continued)
Sampled parameter*
InkTSD
InkAS
InkTD
InkAD
InkASTCA
InkASTCOH
InVMAxC
lnKMC
InFracOtherC
InFracTCAC
lnVMAXDCVGC
InCIDCVGC
InVMAxKidDCVGC
InCIKidDCVGC
InVMAxLungLivC
lnKMClara
InFracLungSysC
lnVMAXTCOHC
InKivJCOH
InVMAxGlucC
lnKMGluc
InkMetTCOHC
InkUrnTCAC
InkMetTCAC
InkBileC
InkEHRC
InkUrnTCOGC
InFracKidDCVCC
InkDCVGC
InkNATC
InkKidBioactC
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
5.187(0.3909,69.34)
1.711 (0.3729, 11.23)
0.1002(0.01304,0.7688)
0.2665(0.05143, 1.483)
3.986(0.1048, 141.9)
0.7308 (0.006338, 89.75)
0.6693(0.4093, 1.106)
0.07148(0.0323,0.1882)
0.02384(0.003244,0.1611)
0.4875 (0.2764, 0.8444)
1.517(0.02376, 1,421)
0.1794(0.02333,79.69)
1.424(0.04313,704.9)
0.827(0.04059, 167.2)
2.903(0.487, 12.1)
0.01123(0.001983,0.09537)
3.304(0.2619, 182.1)
1.645(0.6986,3.915)
0.9594 (0.2867, 2.778)
65.59 (27.58, 232.5)
31.16(6.122, 137.3)
3.629 (0.7248, 9.535)
0.1126(0.04083,0.2423)
0.6175(0.2702, 1.305)
0.9954(0.316,3.952)
0.01553(0.001001,0.0432)
7.874 (2.408, 50.28)
1.931 (0.01084, 113.7)
0.2266(0.001104, 16.46)
0.1175(0.0008506, 14.34)
0.07506(0.0009418, 12.35)
R
1.001
1.001
1
1.003
1
1.001
1.005
1
1.006
1.002
1.001
1.013
1.014
1.019
1.001
1.012
1.011
1.005
1.007
1.018
1.015
1.009
1.012
1.027
1.003
1.008
1
1.018
1.011
1.024
1.035
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
5.858(2.614,80)
4.203(2.379, 18.15)
5.16(2.478,60.24)
4.282 (2.378, 20.21)
5.187(2.516,58.72)
5.047 (2.496, 54.8)
1.793(1.49,2.675)
2.203(1.535,4.536)
1.532(1.265,2.971)
1.474(1.258,2.111)
1.53(1.263,2.795)
1.528(1.261,2.922)
1.533(1.262,2.854)
1.527(1.263,2.874)
4.157(1.778,29.01)
1.629(1.278,5.955)
1.543(1.266,3.102)
1.603(1.28,2.918)
1.521 (1.264,2.626)
1.487(1.254,2.335)
1.781 (1.299,5.667)
1.527(1.265,2.626)
1.757(1.318,3.281)
1.508(1.262,2.352)
1.502(1.26,2.453)
1.534(1.264,2.767)
3.156(1.783,12.18)
1.53(1.264,2.77)
1.525(1.263,2.855)
1.528(1.264,2.851)
1.527(1.263,2.84)
R
1
1
1
1
1
1
1
1.001
1
1
1
1
1
1
1.018
1.003
1.001
1
1
1
1.002
1
1.003
1.002
1
1
1.001
1
1
1
1.001
2
O
4
5
6
These "sampled parameters" are scaled one or more times (see Table A-4) to obtain a biologically-meaningful
parameter, posterior distributions of which are summarized in Tables 3-36 through 3-40). For natural log
transformed parameters (name starting with "In"), values are for the population geometric means and standard
deviations.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-65 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-10. Posterior distributions for mouse residual errors
3
4
Measurement
ClnhPPM
CVen
CBIdMix
CFat
CKid
CLiv
AExhpost
CTCOH
CLivTCOH
CPIasTCA
CBIdTCA
CLivTCA
AUrnTCA
CTCOGTCOH
CLivTCOGTCOH
AUrnTCOGTCOH
TotCTCOH
Residual error geometric standard deviation
Median (2.5%, 97.5%)
1.177(1.16, 1.198)
2.678(2.354,3.146)
1.606(1.415, 1.96)
2.486(2.08,3.195)
2.23(1.908,2.796)
1.712(1.543, 1.993)
1.234(1.159, 1.359)
1.543(1.424, 1.725)
1.591 (1.454, 1.818)
1.396(1.338, 1.467)
1.488(1.423, 1.572)
1.337(1.271, 1.43)
1.338(1.259, 1.467)
1.493(1.38, 1.674)
1.63(1.457, 1.924)
1.263(1.203, 1.355)
1.846(1.506,2.509)
R
1.001
1.001
1.001
1
1
1
1
1
1
1.001
1.001
1
1
1.001
1
1
1.002
Note: the hierarchical statistical model for residual errors did not separate by group.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-66 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-ll. Posterior distributions for rat PBPK model population
parameters
Sampled parameter
InQCC
InVPRC
QFatC
QGutC
QLivC
QSIwC
InDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
InkDissocC
InBMaxkDC
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
InkTSD
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
1.195(0.9285, 1.448)
0.6304 (0.4788, 0.8607)
1.167(0.8321, 1.561)
1.154(0.988, 1.306)
1.029(0.8322, 1.223)
0.9086(0.5738, 1.251)
2.765(1.391,5.262)
1.002(0.8519, 1.152)
1.037(0.8071, 1.259)
0.9728(0.593, 1.378)
0.9826(0.8321, 1.137)
0.9608(0.7493, 1.19)
0.9929(0.8563, 1.133)
1.001 (0.7924, 1.21)
0.999(0.7921, 1.208)
0.999(0.8263, 1.169)
1.002(0.8617, 1.141)
0.8551 (0.6854, 1.065)
1.17(0.8705, 1.595)
0.8197(0.5649, 1.227)
1.046(0.8886, 1.234)
1.021 (0.6239, 1.675)
0.993(0.5964, 1.645)
0.9209(0.6728, 1.281)
1.258(0.9228, 1.711)
0.9763(0.6761, 1.353)
1.136(0.6737, 1.953)
1.283(0.6425,2.491)
1.01 (0.5052,2.017)
0.9654(0.5716, 1.733)
0.9454 (0.4533, 1 .884)
0.926(0.3916,2.196)
1.968(0.09185, 14.44)
7.484 (2.389, 26.92)
3.747 (0.2263, 62.58)
R
1.034
1.012
1
1
1.002
1.001
1.018
1.001
1.002
1
1
1.015
1.001
1
1.001
1
1
1.001
1.003
1
1.001
1.002
1.001
1
1.001
1
1.008
1.008
1.002
1.02
1.045
1.013
1.031
1.017
1.01
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
1.298(1.123,2.041)
1.446(1.247,2.011)
0.4119(0.2934,0.6438)
0.1613(0.1132, 0.2542)
0.1551 (0.1092,0.2483)
0.2817(0.1968,0.4493)
1.21 (1.142, 1.358)
0.1185(0.08284,0.1871)
0.1785(0.1272,0.2723)
0.4139(0.2924, 0.6552)
0.1187(0.08296,0.1873)
0.1682(0.1168,0.2718)
0.1093(0.07693,0.175)
0.1636(0.116,0.2601)
0.1635(0.1161,0.2598)
0.1361 (0.09617,0.2167)
0.1096(0.07755,0.176)
1.317(1.232, 1.462)
1.333(1.247, 1.481)
1.362(1.198, 1.895)
1.152(1.115, 1.214)
1.373(1.201, 1.988)
1.356(1.197, 1.948)
1.304(1.201, 1.536)
1 .364 (1 .263, 1 .544)
1.276(1.159, 1.634)
1.631 (1.364,2.351)
1.651 (1.356,2.658)
1.596(1.315,2.774)
1.412(1.234,2.01)
1.734(1.39,3.151)
1.785(1.382,4.142)
1.414(1.208,2.571)
1.41 (1.208,2.108)
6.777 (2.844, 87.29)
R
1.031
1.005
1
1
1
1
1.001
1
1
1.002
1
1.001
1
1
1
1
1
1.001
1.001
1
1
1
1
1
1
1
1.003
1
1
1
1.002
1.003
1
1
1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-67 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-ll. Posterior distributions for rat PBPK model population
parameters (continued)
Sampled parameter
InkAS
InkAD
InkASTCA
InkASTCOH
InVMAxC
lnKMC
InFracOtherC
InFracTCAC
lnVMAXDCVGC
InCIDCVGC
InVMAxKidDCVGC
InCIKidDCVGC
InVMAxLungLivC
lnKMClara
InFracLungSysC
lnVMAXTCOHC
InKivJCOH
InVMAxGlucC
lnKMGluc
InkMetTCOHC
InkUrnTCAC
InkMetTCAC
InkBileC
InkEHRC
InkUrnTCOGC
InkNATC
InkKidBioactC
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
2.474 (0.2542, 28.35)
0.1731 (0.04001,0.7841)
1.513(0.1401, 17.19)
0.6896(0.01534,25.81)
0.8948 (0.6377, 1 .293)
0.0239(0.01602,0.04993)
0.344(0.0206, 1.228)
0.2348(0.122,0.4616)
7.749 (0.2332, 458.8)
0.3556(0.06631,2.242)
0.2089(0.04229, 1.14)
184(26.29, 1312)
2.673(0.4019, 14.16)
0.02563(0.005231,0.197)
2.729(0.04124,63.27)
1.832(0.6673,6.885)
22.09(3.075, 131.9)
28.72(10.02,86.33)
6.579(1.378,23.57)
2.354(0.3445, 15.83)
0.07112(0.03934,0.1329)
0.3554(0.1195,0.8715)
8.7(1.939,26.71)
1.396(0.2711,6.624)
20.65(2.437,138)
0.002035 (0.0004799, 0.01019)
0.006618 (0.0009409, 0.0367)
R
1.004
1.018
1.002
1.001
1.028
1.001
1.442
1.028
1.088
1.018
1.011
1.02
1.002
1.01
1.027
1.041
1.186
1.225
1.119
1.287
1.076
1.036
1.05
1.091
1.041
1.01
1.039
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
10.16(4.085, 143.7)
4.069(2.373, 14.19)
4.376 (2.43, 22.83)
4.734 (2.444, 35.2)
1.646(1.424,2.146)
2.402(1.812,4.056)
3(1.332, 10.04)
1.517(1.264,2.393)
1.534(1.262,2.804)
1.509(1.261,2.553)
1.542(1.263,2.923)
1.527(1.265,2.873)
4.833(1.599,48.32)
1.66(1.279, 18.74)
1.536(1.267,2.868)
1.667(1.292,3.148)
1.629(1.276,3.773)
2.331 (1.364,5.891)
2.046(1.309, 10.3)
1.876(1.283, 11.82)
1.513(1.27,2.327)
1.528(1.263,2.444)
1.65(1.282,5.494)
1.647(1.277,5.582)
1.595(1.269,5.257)
1.523(1.261,2.593)
1.52(1.261,2.674)
R
1
1.009
1
1.001
1.021
1.001
1.353
1.001
1.001
1
1.001
1.001
1.002
1.002
1.001
1.002
1.017
1.126
1.125
1.182
1.003
1.001
1.017
1.005
1.026
1.001
1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-68 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-12. Posterior distributions for rat residual errors
Measurement
ClnhPPM
CMixExh
CArt
CVen
CBIdMix
CFat
CGut
CKid
CLiv
CMus
AExhpost
CTCOH
CPIasTCA
Group
Group 3
Group 16
Group 2
Group 2
Group 6
Group 4
Group 7
Group 8
Group 9
Group 10
Group 11
Group 16
Group 18
Group 12
Group 9
Group 16
Group 9
Group 9
Group 9
Group 12
Group 16
Group 9
Group 6
Group 10
Group 14
Group 15
Group 6
Group 10
Group 11
Group 13
Group 17
Group 18
Group 4
Group 5
Group 11
Group 19
Residual error geometric standard deviation
Median (2.5%, 97.5%)
1.124(1.108, 1.147)
1.106(1.105, 1.111)
1.501 (1.398, 1.65)
1.174(1.142, 1.222)
1.523(1.321, 1.918)
1.22(1.111, 1.877)
1.668(1.489,1.986)
1.45(1.234,2.065)
1.571 (1.426,1.811)
4.459 (2.754, 6.009)
1.587(1.347,2.296)
1.874(1.466,2.964)
1.676(1.188,3.486)
1.498(1.268,2.189)
1.846(1.635,2.184)
2.658(1.861,4.728)
1.855(1.622,2.243)
1.469(1.354,1.648)
1.783(1.554,2.157)
1.744(1.401,2.892)
1.665(1.376,2.411)
1.653(1.494, 1.919)
1.142(1.108, 1.239)
1.117(1.106, 1.184)
1.166(1.107, 1.475)
1.125(1.106, 1.237)
1.635(1.455, 1.983)
1.259(1.122, 1.868)
1.497(1.299, 1.923)
1.611 (1.216,3.556)
1.45(1.213,2.208)
1.142(1.107,1.268)
1.134(1.106,1.254)
1.141 (1.107,1.291)
1.213(1.136,1.381)
1.201 (1.145,1.305)
R
1
1
1
1
1.002
1
1.001
1.014
1
1
1.002
1.011
1.003
1
1
1.001
1
1
1
1
1.001
1
1.003
1.004
1
1
1.002
1.009
1.01
1.001
1.004
1
1
1
1
1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-69 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-12. Posterior distributions for rat residual errors (continued)
2
3
4
5
6
7
Measurement
CBIdTCA
CLivTCA
AUrnTCA
ABileTCOG
CTCOG
AUrnTCOGTCOH
AUrnNDCVC
AUrnTCTotMole
TotCTCOH
Group
Group 4
Group 5
Group 6
Group 11
Group 17
Group 18
Group 19
Group 19
Group 1
Group 6
Group 8
Group 10
Group 17
Group 19
Group 6
Group 17
Group 1
Group 6
Group 8
Group 10
Group 17
Group 1
Group 6
Group 7
Group 14
Group 15
Group 17
Residual error geometric standard deviation
Median (2.5%, 97.5%)
1.134(1.106, 1.258)
1.14(1.107, 1.289)
1.59(1.431, 1.878)
1.429(1.292, 1.701)
1.432(1.282, 1.675)
1.193(1.12, 1.358)
1.214(1.153, 1.327)
1.666(1.443,2.104)
1.498(1.125,2.18)
1.95(1.124,5.264)
1.221 (1.146,1.375)
1.18(1.108, 1.444)
1.753(1.163,4.337)
1.333(1.201,1.707)
2.129(1.128,5.363)
2.758(1.664,5.734)
1.129(1.106,1.232)
1.483(1.113,4.791)
1.115(1.106,1.162)
1.145(1.107,1.305)
2.27(1.53,4.956)
1.168(1.11, 1.33)
1.538(1.182,3.868)
1.117(1.106, 1.153)
1.121 (1.106, 1.207)
1.162(1.108, 1.358)
1.488(1.172,2.366)
R
1
1
1.001
1.001
1.03
1.004
1
1
1.135
1.003
1.003
1.007
1.001
1
1.003
1.028
1.004
1.002
1
1
1.009
1.002
1.002
1.001
1
1
1.015
The nineteen groups are (1) Bernauer et al, 1996; (2) Dallas et al, 1991; (3) Fisher et al, 1989
females; (4) Fisher et al., 1991 females; (5) Fisher et al., 1991 males; (6) Green and Prout, 1985,
Prout et al., 1985, male OA rats; (7) Hissink et al., 2002; (8) Kaneko et al., 1994; (9) Keys et al.,
2003; (10) Kimmerle and Eben, 1973a; (11) Larson and Bull, 1992a, b; (12) Lee et al., 2000; (13)
Merdink et al., 1999; (14) Prout et al., 1985 AP rats; (15) Prout et al., 1985 OM rats; (16)
Simmons et al., 2002; (17) Stenner et al., 1997; (18) Templin et al., 1995; (19) Yu et al., 2000.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-70 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-13. Posterior distributions for human PBPK model population
parameters
Sampled parameter
InQCC
InVPRC
QFatC
QGutC
QLivC
QSIwC
InDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBIdC
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC
InPKidC
InPSIwC
InPRBCPIasTCAC
InPBodTCAC
InPLivTCAC
InkDissocC
InBMaxkDC
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
InPeffDCVG
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
0.837(0.6761, 1.022)
1.519(1.261, 1.884)
0.7781 (0.405, 1.143)
0.7917(0.6631,0.925)
0.5099(0.1737,0.8386)
0.7261 (0.4864, 0.9234)
0.626(0.3063, 1.013)
1.007(0.9137, 1.103)
1.001 (0.9544, 1.047)
0.788(0.48, 1.056)
1 (0.937, 1.067)
1.043(0.8683, 1.23)
0.9959(0.9311, 1.06)
1.003(0.8461, 1.164)
1 (0.8383, 1.159)
0.9965(0.8551, 1.14)
1.013(0.9177, 1.108)
0.9704(0.8529, 1.101)
0.8498 (0.7334, 0.9976)
1.095(0.7377, 1.585)
0.9907(0.6679, 1.441)
0.93(0.6589,1.28)
1.018(0.6773, 1.5)
0.9993(0.8236, 1.219)
1.157(0.8468, 1.59)
0.3223 (0.04876, 0.8378)
1.194(0.929, 1.481)
1.202(0.8429, 1.634)
0.9932(0.9387, 1.053)
0.8806(0.7492, 1.047)
1.703(1.439,2.172)
1.069(0.7643, 1.485)
0.7264(0.1237,2.54)
6.671 (1.545,24.87)
0.01007 (0.003264, 0.03264)
R
1.038
1.007
1.014
1.017
1.031
1.011
1.197
1.009
1.01
1.005
1.007
1.047
1.006
1.001
1.001
1.007
1.003
1.001
1.002
1.029
1.01
1.003
1.015
1.003
1.018
1.007
1.043
1.046
1.012
1.038
1.019
1.028
1.003
1.225
1.004
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
1.457(1.271, 1.996)
1.497(1.317, 1.851)
0.6272(0.4431,0.9773)
0.1693(0.1199,0.2559)
0.4167(0.2943,0.6324)
0.3166(0.2254,0.4802)
1.291 (1.158,2.006)
0.1004(0.07307,0.1545)
0.04275(0.03155,0.06305)
0.3666 (0.2696, 0.5542)
0.06745(0.04923,0.1038)
0.1959(0.1424,0.3017)
0.06692(0.04843,0.1027)
0.1671 (0.1209,0.255)
0.1672(0.1215,0.259)
0.1425(0.1037,0.2183)
0.1005(0.07265,0.1564)
1.216(1.161, 1.307)
1.188(1.113, 1.366)
1.413(1.214,2.05)
1.338(1.203, 1.683)
1.528(1.248,2.472)
1.32(1.192, 1.656)
1.155(1.097, 1.287)
1.69(1.383,3.157)
5.507(3.047, 19.88)
1.327(1.185, 1.67)
1.285(1.162, 1.648)
1.043(1.026, 1.076)
1.157(1.085, 1.37)
1.409(1.267, 1.678)
1.288(1.165, 1.629)
11.98(5.037, 185.3)
5.954 (2.653, 23.68)
1.385(1.201,2.03)
R
1.036
1.008
1
1.019
1.009
1.005
1.083
1
1
1
1
1.003
1
1
1
1
1
1.002
1.002
1.002
1
1.001
1
1
1.008
1.003
1.018
1.007
1.003
1.012
1.011
1.002
1.017
1.052
1.001
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-71 DRAFT—DO NOT CITE OR QUOTE
-------
Table A-13. Posterior distributions for human PBPK model population
parameters (continued)
Sampled parameter
InkASTCA
InkASTCOH
lnV,v,AxC
InCIC
InFracOtherC
InFracTCAC
InCIDCVGC
lnK,v,DCVGC
InCIKidDCVGC
lnK,v,KidDCVGC
InVMAxLungLivC
lnKMClara
InFracLungSysC
InCITCOHC
InKivJCOH
InCIGIucC
lnKMGluc
InkMetTCOHC
InkUrnTCAC
InkMetTCAC
InkBileC
InkEHRC
InkUrnTCOGC
InkDCVGC
InkNATC
InkKidBioactC
Posterior distributions reflecting uncertainty in population distribution
Population (geometric) mean
Median (2.5%, 97.5%)
4.511 (0.04731,465.7)
8.262 (0.0677, 347.9)
0.3759(0.2218,0.5882)
12.64(5.207,39.96)
0.1186(0.02298,0.2989)
0.1315(0.07115,0.197)
2.786(1.326,5.769)
1.213(0.3908,4.707)
0.04538(0.001311,0.1945)
0.2802(0.1096, 1.778)
3.772(0.8319,9.157)
0.2726(0.02144, 1.411)
24.08(6.276,81.14)
0.1767(0.1374,0.2257)
2.221 (1.296,4.575)
0.2796(0.2132,0.3807)
133.4(51.56,277.2)
0.7546(0.1427,2.13)
0.04565 (0.0324, 0.06029)
0.2812(0.1293,0.5359)
6.855(3.016,20.69)
0.1561 (0.09511,0.2608)
15.78(6.135,72.5)
7.123(5.429,9.702)
0.0003157 (0.0001087, 0.002305)
0.06516(0.01763,0.1743)
R
1
1
1.026
1.028
1.061
1.026
1.08
1.029
1.204
1.097
1.035
1.041
1.016
1.011
1.02
1.056
1.02
1.007
1.005
1.004
1.464
1.1
1.007
1.026
1.008
1.001
Population (geometric) standard
deviation
Median (2.5%, 97.5%)
5.467(2.523,71.06)
5.481 (2.513,67.86)
2.21 (1.862,2.848)
4.325 (2.672, 9.003)
3.449(1.392,9.146)
2.467(1.916,3.778)
2.789(1.867,4.877)
4.43(2.396, 18.56)
3.338(1.295,30.46)
1.496(1.263,2.317)
2.228(1.335,21.89)
11.63(1.877,682.7)
1.496(1.263,2.439)
1.888(1.624,2.307)
2.578(1.782,4.584)
1.955(1.583,2.418)
1.573(1.266,4.968)
5.011 (2.668, 15.71)
1.878(1.589,2.48)
2.529(1.78,4.211)
1.589(1.27,3.358)
1.699(1.348,2.498)
9.351 (4.93, 29.96)
1.507(1.311, 1.897)
1.54(1.261,3.306)
1.523(1.262,2.987)
R
1
1
1.003
1.016
1.102
1.01
1.02
1.035
1.095
1.001
1.014
1.041
1.001
1.01
1.015
1.079
1.011
1.002
1.006
1.002
1.015
1.015
1.003
1.008
1
1
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-72 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table A-14. Posterior distributions for human residual errors
Measurement
RetDose
CAIvPPM
CVen
CTCOH
CPIasTCA
CBIdTCA
zAUrnTCA
zAUrnTCA_collect
AUrnTCOGTCOH
AUrnTCOGTCOH_collect
CDCVGmol
zAUrnNDCVC
TotCTCOH
Group
Group 4
Group 1
Group 4
Group 5
Group 1
Group 3
Group 4
Group 5
Group 1
Group 3
Group 5
Group 7
Group 2
Group 7
Group 1
Group 2
Group 4
Group 5
Group 1
Group 2
Group 3
Group 4
Group 6
Group 7
Group 3
Group 5
Group 1
Group 3
Group 4
Group 6
Group 7
Group 3
Group 5
Group 1
Group 6
Group 1
Group 4
Group 5
Residual error geometric standard
deviation
Median (2.5%, 97.5%)
1.131 (1.106,1.25)
1.832(1.509,2.376)
1.515(1.378, 1.738)
1.44(1.413, 1.471)
1.875(1.683,2.129)
1.618(1.462, 1.862)
1.716(1.513,2.057)
2.948 (2.423, 3.8)
1.205(1.185, 1.227)
1.213(1.187, 1.247)
2.101 (1.826,2.571)
1.144(1.106,2.887)
1.117(1.106,1.17)
1.168(1.123, 1.242)
1.138(1.126, 1.152)
1.119(1.106, 1.178)
1.488(1.351, 1.646)
1.438(1.367, 1.537)
1.448(1.414, 1.485)
1.113(1.105, 1.149)
1.242(1.197, 1.301)
1 .538 (1 .441 , 1 .67)
1.158(1.118, 1.228)
1.119(1.106, 1.181)
1.999(1.178,3.903)
2.787(2.134,4.23)
1.106(1.105, 1.112)
1.11 (1.105, 1.125)
1.124(1.107, 1.151)
1.117(1.106, 1.157)
1.134(1.106, 1.348)
1.3(1.111,2.333)
1 .626 (1 .524, 1 .767)
1.53(1.436, 1.656)
1.167(1.124, 1.244)
1.204(1.185, 1.226)
1.247(1.177, 1.366)
1.689(1.552, 1.9)
R
1.001
1.015
1
1
1.018
1
1.001
1.007
1.012
1
1.001
1.123
1.001
1
1.003
1
1.018
1.002
1.001
1.001
1.001
1
1
1
1.003
1.001
1.001
1
1.001
1.001
1.003
1.004
1
1.009
1
1.011
1.009
1.001
The seven groups are (1) Fisher et al., 1998; (2) Paycok and Powell, 1945; (3) Kimmerle and Eben, 1973b;
(4) Monster et al., 1976; (5) Chiu et al., 2007; (6) Bernauer et al., 1996; (7) Muller et al., 1974.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-73 DRAFT—DO NOT CITE OR QUOTE
-------
1 A.5.2.3. Human Model
2 A.5.2.3.1. Individual-specific predictions and calibration data, [See
3 Appendix.linked.files\AppA.5.2.3.1.Updated.human.indiv.calib.TCE.DRAFT.pdf.1
4
5 A.5.2.3.2. Population-based predictions and calibration data. [See
6 Appendix.linked.files\AppA.5.2.3.2.Updated.human.pop.calib.TCE.DRAFT.pdf.1
7
8 A.5.2.3.3. Population-based predictions and additional evaluation data. [See
9 Appendix.linked.files\AppA.5.2.3.3.Updated.hutnan.pop.eval.TCE.DRAFT.pdf.]
10
11 A.6. EVALUATION OF RECENTLY PUBLISHED TOXICOKINETIC DATA
12 Several in vivo toxicokinetic studies were published or became available during internal
13 U.S. EPA review and Interagency Consultation, and were not evaluated as part of the originally
14 planned analyses. Preliminary analyses of these data are summarized here. The general
15 approach is the same as that used for the evaluation data in the primary analysis—population
16 predictions from the PBPK model are compared visually with the toxicokinetic data. Figures
17 with the population-based predictions and these recently published data are in the following
18 linked files:
19
20 • Mouse (Kim et al., 2009; Mahle et al., 2001; Green, 2003a, b):
21 Appendix.linked.files\AppA.6.Updated.mouse.pop.eval.TCE.DRAFT.pdf.
22 • Rat (Liu et al., 2009; Mahle et al., 2001):
23 Appendix.linked.files\AppA.6.Updated.rat.pop.eval.TCE.DRAFT.pdf.
24
25 A.6.1. TCE Metabolite Toxicokinetics in Mice: Kim et al. (2009)
26 Kim et al. (2009) measured TCA, DC A, DCVG, and DCVC in blood of male B6C3F1
27 mice following a single gavage dose of 2,140 mg/kg. Of these data, only TCA and DCVG blood
28 concentrations are predicted by the updated PBPK model, so only those data are compared with
29 PBPK model predictions (prior values for the distribution volume and elimination rate constant
30 of DCVG were used, as there were no calibration data informing those parameters). These data
31 were within the inter-quartile region of the PBPK model population predictions.
32 An assessment was made as to whether these data are informative as to the flux of GSH
33 conjugation in mice. First, the best fitting parameter sample (least squares on TCA and DCVG
34 in blood, weighted by inverse of the observed variance) from the posterior distribution was
35 selected out of 50,000 samples generated by Monte Carlo (see Figures A-7 and A-8 for the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-74 DRAFT—DO NOT CITE OR QUOTE
-------
1 comparison with predictions with data). This parameter sample was then used to calculate the
2 fraction of intake that is predicted by the PBPK model to undergo GSH metabolism for
3 continuous oral and continuous inhalation exposure, and this point estimate compared to the full
4 posterior distribution (see Figures A-9 and A-10). The predictions for this "best fitting"
5 parameter set was similar (within 3-fold) of the median of the full posterior distribution. While a
6 formal assessment of the impact of these new data (i.e., including its uncertainty and variability)
7 would require a re-running of the Bayesian analysis, it appears that the median estimates for the
8 mouse GSH conjugation dose metric used in the dose-response assessment (see Chapter 5) are
9 reasonably consistent with the Kim et al. (2009) data.
10
11
12
13
14
O
O
o
oo
o
CD
o
CM
O -
\
0
\
5
10
15
20
t(hr)
Figure A-7. Comparison of best-fitting (out of 50,000 posterior samples)
PBPK model prediction and Kim et al. (2009) TCA blood concentration data
for mice gavaged with 2,140 mg/kg TCE. Full population distributions are
shown in a separate linked file (see text).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-75 DRAFT—DO NOT CITE OR QUOTE
-------
E,
O
O
Q
O
CD
o -
q
ci
CM
o
o
o -\
0
\
2
\
4
\
6
\
8
10
12
1
2
3
4
5
t(hr)
Figure A-8. Comparison of best-fitting (out of 50,000 posterior samples)
PBPK model prediction and Kim et al. (2009) DCVG blood concentration
data for mice gavaged with 2,140 mg/kg TCE. Full population distributions are
shown in a separate linked file (see text).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-76 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
c
o
"-I—•
03
O)
D
'c
O
O
X
C/3
O
c
g
ti
03
O -
7
o
CO
O
10
-1
T
1 10 10
oral exposure (mg/kg/d continuous)
Figure A-9. PBPK model predictions for the fraction of intake undergoing
GSH conjugation in mice continuously exposed orally to TCE. Lines and
error bars represent the median and 95th percentile confidence interval for the
posterior predictions, respectively (also reported in Section 3.5.7.2.1). Filled
circles represent the predictions from the sample (out of 50,000 total posterior
samples) which provides the best fit to the Kim et al. (2009) TCA and DCVG
blood concentration data for mice gavaged with 2,140 mg/kg TCE.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-77 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
-
"CD
O)
D
'c
O
O
C/3
O
c
O
o
O 1
T _,
o -^
10
-1
10'
Inhalation exposure (ppm continuous)
icr
Figure A-10. PBPK model predictions for the fraction of intake undergoing
GSH conjugation in mice continuously exposed via inhalation to TCE. Lines
and error bars represent the median and 95th percentile confidence interval for the
posterior predictions, respectively (also reported in Section 3.5.7.2.1). Filled
circles represent the predictions from the sample (out of 50,000 total posterior
samples) which provides the best fit to the Kim et al. (2009) TCA and DCVG
blood concentration data for mice gavaged with 2,140 mg/kg TCE.
An additional note of interest from the Kim et al. (2009) data is the inter-study variability
in TCA kinetics. In particular, the TCA blood concentrations reported by Kim et al. (2009) are
2-fold lower than those reported by Abbas and Fisher (1997) in the same sex and strain of
mouse, with a very similar corn oil gavage dose of 2,000 mg/kg (as compared to 2,140 mg/kg
used in Kim et al., 2009).
A.6.2. TCE Toxicokinetics in Rats: Liu et al. (2009)
Liu et al. (2009) measured TCE in blood of male rats after treatment with TCE by i.v.
injection (0.1, 1.0, or 2.5 mg/kg) or aqueous gavage (0.0001, 0.001, 0.01, 0.1, 1, 2.5, 5, or
10 mg/kg). Almost all of the data from gavage exposures were within the inter-quartile region of
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-78 DRAFT—DO NOT CITE OR QUOTE
-------
1 the PBPK model population predictions, with all of it within the 95% confidence interval. For
2 i.v. exposures, the data at 1 and 2.5 mg/kg were well simulated, but the time-course data at
3 0.1 mg/kg were substantially different in shape from that predicted by the PBPK model, with a
4 lower initial concentration and longer half-life. The slower elimination rat at 0.1 mg/kg was
5 noted by the study authors through use of noncompartamental analysis. There is no clear
6 explanation for this discrepancy, particularly since the gavage data at this and even lower doses
7 were well predicted by the PBPK model.
8
9 A.6.3. TCA Toxicokinetics in Mice and Rats: Mahle et al. (2001) and Green (2003a, b)
10 Three technical reports (Mahle et al., 2001; Green, 2003a, b) described by Sweeney et al.
11 (2009) contained data on TCA toxicokinetics in mice and rats exposed to TCA in drinking water.
12 These technical reports were provided to U.S. EPA by the Sweeney et al. (2009) authors.
13 TCA blood and liver concentrations were reported by Mahle et al. (2001) for male
14 B6C3F1 mice and male Fischer 344 rats exposed to 0.1 g/L to 2 g/L TCA in drinking water for 3
15 or 14 days (12 to 270 mg/kg/d in mice and 7 to 150 mg/kg/d in rats). For mice, these data were
16 all within the 95% confidence interval of PBPK model population predictions, with about half of
17 these data within the interquartile region. For rats, all these data, except those for the 3-day
18 exposure at 0.1 g/L, were within the 95% confidence interval of the PBPK model predictions. In
19 addition, the median rat predictions were consistently higher than the data, although this could be
20 explained by inter-study (strain, lot, etc.) variability.
21 TCA blood concentrations were reported by Green (2003a) for male and female B6C3F1
22 mice exposed to 0.5 g/L to 2.5 g/L TCA in drinking water for 5 days (130 to 600 mg/kg/d in
23 males and 160 to 750 mg/kg/d in females). Notably, these animals consumed around twice as
24 much water per day as compared to the mice reported by Mahle et al. (2001), and therefore
25 received comparatively higher doses of TCA for the same TCE concentration in drinking water.
26 In male mice, the data at the lower two doses (130 and 250 mg/kg/d) were within the inter-
27 quartile region of the PBPK model predictions. The data for male mice at the highest dose
28 (600 mg/kg/d) were below the inter-quartile region, but within the 95% confidence interval of
29 the PBPK model predictions. In females, the data at the lower two doses (160 and 360 mg/kg/d)
30 were mostly below the inter-quartile region, but within the 95% confidence interval of the PBPK
31 model predictions, while about half the data at the highest dose were just below the 95%
32 confidence interval.
33 TCA blood, plasma, and liver concentrations were reported by Green (2003b) for male
34 PPARa-null mice, male 129/sv mice (the background strain of the PPARa-null mice), and male
35 and female B6C3F1 mice, exposed to 1.0 g/L or 2.5 g/L TCA in drinking water for 5 days (male
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-79 DRAFT—DO NOT CITE OR QUOTE
-------
1 B6C3F1 only) to 14 days.2 In male PPARa-null mice, plasma and blood concentrations were
2 within the inter-quartile region of the PBPK model predictions, while liver concentrations were
3 below the inter-quartile region but within the 95% confidence interval. In male 129/sv mice, the
4 plasma concentrations were within the inter-quartile region of the PBPK model predictions,
5 while blood and liver concentrations were below the inter-quartile region but within the 95%
6 confidence interval. In male B6C3F1 mice, all data were within the 95% confidence intervals of
7 the PBPK model predictions, with about half within the inter-quartile region, and the rest above
8 (plasma concentrations at the lower dose) or below (liver concentrations at all but the lowest
9 dose at 5 days). In female B6C3F1 mice, plasma concentrations were below the inter-quartile
10 region but within the 95% confidence region, while liver and blood concentrations were at or
11 below the lower 95% confidence bound.
12 Overall, the predictions of the TCA submodel of the updated TCE PBPK model appear
13 consistent with these data on the toxicokinetics of TCA after drinking water exposure in male
14 rats and male mice. In female mice, the reported concentrations tends to be at the low end of or
15 lower than those predicted by the PBPK model. Importantly, the data used for calibrating the
16 mouse PBPK model parameters were predominantly in males, with only Fisher et al. (1991,
17 1993) reporting TCA plasma levels in female mice after TCE exposure. In addition, median
18 PBPK model predictions at higher doses (>300 mg/kg/d), even in males, tended to be higher than
19 the concentrations reported. While TCA kinetics after TCE exposure includes predicted internal
20 production at these higher levels, previously published data on TCA kinetics alone only included
21 doses up to 100 mg/kg, and only in males. Therefore, these results suggest that the median
22 predictions of the TCA sub-model of the updated TCE PBPK model are somewhat less accurate
23 for female mice and for higher doses of TCA (>300 mg/kg/d) in mice, though the 95%
24 confidence intervals still cover the majority of the reported data. Finally, the ratio of blood to
25 liver concentrations of ~1.4 reported in the mouse experiments in Mahle et al. (2001) were
26 significantly different from the ratios of-2.3 reported by Green (2003b), a difference for which
27 there is no clear explanation given the similar experimental designs and common use the
28 B6C3F1 mouse strain. Because median PBPK model predictions for the blood to liver
29 concentration ratio for these studies are -1.3, they are more consistent with the Mahle et al.
30 (2001) data than with the Green (2003b) data.
31 Sweeney et al. (2009) also suggested that the available data, in conjunction with
32 deterministic modeling using the TCA portion of the Hack et al. (2006) TCE PBPK model,
2Sweeney et al. (2009) reported that blood concentrations in Green (2003b) were incorrect due to an arithmetic error
owing to a change in chemical analytic methodology, and should have been multiplied by 2. This correction was
included in the present analysis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-80 DRAFT—DO NOT CITE OR QUOTE
-------
1 supported a hypothesis that the bioavailability of TCA in drinking water in mice is substantially
2 less than 100%. Classically, oral bioavailability is assessed by comparing blood concentration
3 profiles from oral and i.v. dosing experiments, because blood concentration data from oral
4 dosing alone cannot distinguish fractional uptake from metabolism. Schultz et al. (1999) made
5 this comparison in rats at a single dose of 82 mg/kg, and reported an empirical bioavailability of
6 116%, consistent with complete absorption. A priori, there would not seem to be a strong reason
7 to suspect that oral absorption in mice would be significantly different from that in rats. As
8 discussed above in the evaluation of Hack et al. (2006) model, available data strongly support
9 clearance of TCA in addition to urinary excretion, based on the finding of less than 100%
10 recovery in urine after i.v. dosing. In addition, as the current TCE PBPK model assumes 100%
11 absorption for orally-administered TCA, and the PBPK model predictions are consistent with
12 these data, it is likely that the limited bioavailability determined by Sweeney et al. (2009) was
13 confounded by this additional clearance pathway unaccounted for by Hack et al. (2006).
14 Therefore, the data are consistent with the combination of 100% absorption for orally-
15 administered TCA and an additional clearance pathway for TCA other than urinary excretion in
16 both rats and mice. This hypothesis could be further tested with additional experiments in mice
17 directly comparing of TCA toxicokinetics (blood or plasma concentrations and urinary
18 excretion) between i.v. and oral dosing.
19
20 A.7. UPDATED PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK)
21 MODEL CODE
22 The following pages contain the updated PBPK model code for the MCSim software
23 (version 5.0.0). Additional details on baseline parameter derivations are included as inline
24 documentation. Example simulation files containing prior distributions and experimental
25 calibration data are available electronically:
26
27 • Mouse: Appendix.linked.filesVTCE. 1.2.3.3.Mouse.pop.example.in
28 • Rat: Appendix.linked.filesVTCE. 1.2.3.3.Rat.pop.example.in
29 • Human: Appendix.linked.filesVTCE. 1.2.3.3.Human.pop.example.in.
30
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-81 DRAFT—DO NOT CITE OR QUOTE
-------
#
#### HISTORY OF HACK ET AL. (2006) MODEL
# Model code to correspond to the block diagram version of the model
# Edited by Deborah Keys to incorporate Lapare et al. 1995 data
# Last edited: August 6, 2004
# Translated into MCSim from acslXtreme CSL file by Eric Hack, started 31Aug2004
# Removed nonessential differential eguations (i.e., AUCCBld) for MCMC runs.
# Changed QRap and QSlw calculations and added QTot to scale fractional flows
# back to 1 after sampling.
# Finished translating and verifying results on 15Sep2004.
# Changed QSlw calculation and removed QTot 21Sep2004.
# Removed diffusion-limited fat uptake 24Sep2004.
#### HISTORY OF U.S. EPA (2009) MODEL (CHIU ET AL., 2009)
# Extensively revised by U.S. EPA June 2007-June 2008
# - Fixed hepatic plasma flow for TCA-submodel to include
# portal vein (i.e., QGutLivPlas -- originally was just
# QLivPlas, which was only hepatic artery).
# - Clearer coding and in-line documentation
# - Single model for 3 species
# - Revised physiological parameters, with discussion of
# uncertainty and variability,
# - In vitro data used for default metabolism parameters,
# with discussion of uncertainty and variability
# - added TCE blood compartment
# - added TCE kidney compartment, with GSH metabolism
# - added DCVG compartment
# - added additional outputs available from in vivo data
# - removed DCA compartment
# - added IA and PV dosing (for rats)
# - Version 1.1 -- fixed urinary parameter scaling
# — fixed VBod in kUrnTCOG (should be VBodTCOH)
# - Version 1.1.1 -- changed some truncation limits (in commments only)
# - Version 1.2 --
# -- removed TB compartment as currently coded
# -- added respiratory oxidative metabolism:
# 3 states: AInhResp, AResp, AExhResp
# -- removed clearance from respiratory metabolism
# - Version 1.2.1 -- changed oral dosing to be similar to IV
# - Version 1.2.2 -- fixed default lung metabolism (additional
# scaling by lung/liver weight ratio)
# - Version 1.2.3 — fixed FracKidDCVC scaling
# - Version 1.2.3.1 -- added output CDCVG ND (no new dynamics)
# for non-detects of DCVG in blood
# - Version 1.2.3.2 -- Exact version of non-detects likelihood
# - Version 1.2.3.3 -- Error variances changed to "Ve xxx"
# NOTE -- lines with comment "(vrisk)" are used only for
# calculating dose metrics, and are commented out
# when doing MCMC runs.
States = {
##— TCE uptake
AStom,
ADuod,
AExc,
AO,
# Amount of TCE in stomach
# oral gavage absorption -- mice and rats only
#(vrisk) excreted in feces from gavage (currently 0)
#(vrisk) total absorbed
InhDose, # Amount inhaled
#— TCE in the body
ARap,
ASlw,
AFat,
AGut,
ALiv,
AKid,
ABld,
# Amount in rapidly perfused tissues
# Amount in slowly perfused tissues
# Amount in fat
# Amount in gut
# Amount in liver
# Amount in Kidney -- previously in Rap tissue
# Amount in Blood -- previously in Rap tissue
AInhResp, # Amount in respiratory lumen during inhalation
AResp, # Amount in respiratory tissue
AExhResp, # Amount in respiratory lumen during exhalation
##— TCA in the body
AOTCA, #(vrisk)
AStomTCA, # Amount of TCA in stomach
APlasTCA, # Amount of TCA in plasma #comment out for
ABodTCA, # Amount of TCA in lumped body compartment
ALivTCA, # Amount of TCA in liver
##— TCA metabolized
AUrnTCA, # Cumulative Amount of TCA excreted in urine
AUrnTCA_sat, # Amount of TCA excreted that during times that had
# saturated measurements (for lower bounds)
AUrnTCA collect,# Cumulative Amount of TCA excreted in urine during
# collection times (for intermittent collection)
##— TCOH in body
AOTCOH,
AStomTCOH,
ABodTCOH,
ALivTCOH,
##— TCOG in body
ABodTCOG,
ALivTCOG,
ABileTCOG,
ARecircTCOG,
##— TCOG excreted
AUrnTCOG, # Amount of TCOG excreted in urine
AUrnTCOG_sat, # Amount of TCOG excreted that during times that had
# saturated measurements (for lower bounds)
AUrnTCOG_collect,# Cumulative Amount of TCA excreted in urine during
# collection times (for intermittent collection)
##— DCVG in body
ADCVGIn, #(vrisk)
ADCVGmol, # Amount of DCVG in body in mmoles
AMetDCVG, #(vrisk)
##— DCVC in body
ADCVCIn, #(vrisk)
ADCVC, # Amount of DCVC in body
#(vrisk)
# Amount of TCOH in stomach
# Amount of TCOH in lumped body compartment
# Amount of TCOH in liver
# Amount of TCOG in lumped body compartment
# Amount of TCOG in liver
# Amount of TCOG in bile (incl. gut)
#(vrisk)
-------
to
O k^j
j*"
vo £3'
I
£5'
a
§-
TO
$>
O
I
O §
O g
1-7* ""-k
O 3"
H TO"
W
O £?
H
W
ABioactDCVC,
##— NAcDCVC excreted
AUrnNDCVC,
##— Other states for TCE
ACh,
AExh,
AExhExp,
##— Metabolism
AMetLivl, #(vrisk) Amount metabolized by P450 in liver
AMetLiv2, #(vrisk) Amount metabolized by GSH conjugation in liver
AMetLng, #(vrisk) Amount metabolized in the lung
AMetKid, #(vrisk)
AMetTCOHTCA, #(vrisk) Amount of TCOH metabolized to TCA
AMetTCOHGluc, #(vrisk) Amount of TCOH glucuronidated
AMetTCOHOther, #(vrisk)
AMetTCA, #(vrisk) Amount of TCA metabolized
##-- Other Dose metrics
AUCCBld, #(vrisk)
AUCCLiv, #(vrisk)
AUCCKid, #(vrisk)
AUCCRap, #(vrisk)
AUCCTCOH, #(vrisk)
AUCCBodTCOH,
AUCTotCTCOH,
AUCPlasTCAFree,
AUCPlasTCA,
AUCLivTCA,
AUCCDCVG #(vrisk)
Input
## —
TCE dosing
Cone,
IVDose,
PDose,
Drink,
lADose,
PVDose,
TCA dosing
IVDoseTCA,
PODoseTCA,
TCOH dosing
IVDoseTCOH,
PODoseTCOH,
Potentially time-var^
QPmeas,
TCAUrnSat,
TCOGUrnSat,
UrnMissing
# Inhalation exposure cone. (ppm)
# IV dose (mg/kg)
# Oral gavage dose (mg/kg)
# Drinking water dose (mg/kg/day)
# Inter-arterial
# Portal Vein
# IV dose (mg/kg) of TCOH
# Oral dose (mg/kg) of TCOH
ing parameters
# Measured value of Alveolar ventilation QP
# Flag for saturated TCA urine
# Flag for saturated TCOG urine
# Flag for missing urine collection times
#*** Outputs for mass balance check
MassBalTCE,
TotDose,
TotTissue,
MassBalTCOH,
TotTCOHIn,
TotTCOHDose,
TotTissueTCOH,
TotMetabTCOH,
MassBalTCA,
TotTCAIn,
TotTissueTCA,
MassBalTCOG,
TotTCOGIn,
TotTissueTCOG,
MassBalDCVG,
MassBalDCVC,
AUrnNDCVCequiv,
#*** Outputs that are potential dose metrics
TotMetab, #(vrisk) Total metabolism
TotMetabBW34, #(vrisk) Total metabolism/BW"3/4
ATotMetLiv, #(vrisk) Total metabolism in liver
AMetLivlLiv, #(vrisk) Total oxidation in liver/liver volume
AMetLivOther, #(vrisk) Total "other" oxidation in liver
AMetLivOtherLiv, #(vrisk) Total "other" oxidation in liver/liver vol
AMetLngResp, #(vrisk) oxiation in lung/respiratory tissue volume
AMetGSH, #(vrisk) total GSH conjugation
AMetGSHBW34, #(vrisk) total GSH conjugation/BW"3/4
ABioactDCVCKid, #(vrisk) Amount of DCVC bioactivated/kidney volume
# NEW
TotDoseBW34, # (vrisk) mg intake / BW"3/4
AMetLivlBW34, #(vrisk) mg hepatic oxidative metabolism / BW"3/4
TotOxMetabBW34, #(vrisk) mg oxidative metabolism / BW"3/4
TotTCAInBW, #(vrisk) TCA production / BW
AMetLngBW34, #(vrisk) oxiation in lung/BW"3/4
ABioactDCVCBW34, #(vrisk) Amount of DCVC bioactivated/BW"3/4
AMetLivOtherBW34, # (vrisk) Total "other" oxidation in liver/BW'v3/4
ft******************************************************************************
#*** Outputs for comparison to in vivo data
# TCE
RetDose, # human - = (InhDose - AExhExp)
CAlv, # needed for CAlvPPM
CAlvPPM, # human
CInhPPM, # mouse, rat
-------
K>
O K^
O
•~K
°
n5
CFat,
CGut,
CRap,
CSlw,
CHrt,
CKld,
CLlv,
CLung,
CMus,
CSpl,
CBrn,
zAExh,
zAExhpost
O
# mouse - TCOG concentration in blood (in TCOH-eguiv)
# mouse - TCOG concentration in kidney (in TCOH-eguiv)
# mouse - TCOG concentration in liver (in TCOH-eguiv)
# mouse - TCOG concentration in lung (in TCOH-eguiv)
# mouse, rat, human - Cumulative Urinary TCOG (in TCOH-eguiv)
# human - TCOG (in TCOH-eguiv) measurements for
# intermittent collection
at, # human - Saturated TCOG (in TCOH-eguiv) measurements
# Other
CDCVGmol, # concentration of DCVG (mmol/1)
CDCVGmolO, # Dummy variable without likelihood (for plotting)#(vl.2.3.1)
CDCVG_ND, # Non-detect of DCVG (<0.05 pmol/ml= 5e-5 mmol/1 )#(vl.2.3.1)
# Output -ln(likelihood)#(vl.2.3.1)
zAUrnNDCVC, # rat, human - Cumulative urinary NAcDCVC
AUrnTCTotMole, # rat, human - Cumulative urinary TCOH+TCA in mmoles
TotCTCOH, # mouse, human - TCOH+TCOG Concentration (in TCOH-eguiv)
TotCTCOHcomp, # ONLY FOR COMPARISON WITH HACK
ATCOG, # ONLY FOR COMPARISON WITH HACK
QPsamp, # human - sampled value of alveolar ventilation rate
## PARAMETERS #(vrisk)
QCnow, # (vrisk) #Cardiac output (L/hr)
QP, # (vrisk) #Alveolar ventilation (L/hr)
QFatCtmp, # (vrisk) #Scaled fat blood flow
QGutCtmp, # (vrisk) #Scaled gut blood flow
QLivCtmp, # (vrisk) #Scaled liver blood flow
QSlwCtmp, # (vrisk) #Scaled slowly perfused blood flow
QRapCtmp, # (vrisk) #Scaled rapidly perfused blood flow
QKidCtmp, # (vrisk) #Scaled kidney blood flow
DResp, # (vrisk) #Respiratory lumen:tissue diffusive clearance rate
VFatCtmp, # (vrisk) #Fat fractional compartment volume
VGutCtmp, # (vrisk) #Gut fractional compartment volume
VLivCtmp, # (vrisk) #Liver fractional compartment volume
VRapCtmp, # (vrisk) #Rapidly perfused fractional compartment volume
VRespLumCtmp, # (vrisk) # Fractional volume of respiratory lumen
VRespEffCtmp, # (vrisk) #Effective fractional volume of respiratory tissue
VKidCtmp, # (vrisk) #Kidney fractional compartment volume
VBldCtmp, # (vrisk) #Blood fractional compartment volume
VSlwCtmp, # (vrisk) #Slowly perfused fractional compartment volume
VPlasCtmp, # (vrisk) #Plasma fractional compartment volume
VBodCtmp, # (vrisk) #TCA Body fractional compartment volume [not incl.
blood+liver]
VBodTCOHCtmp, # (vrisk) #TCOH/G Body fractional compartment volume [not incl.
liver]
PB, # (vrisk) #TCE Blood/air partition coefficient
PFat, # (vrisk) #TCE Fat/Blood partition coefficient
PGut, # (vrisk) #TCE Gut/Blood partition coefficient
PLiv, # (vrisk) #TCE Liver/Blood partition coefficient
PRap, # (vrisk) #TCE Rapidly perfused/Blood partition coefficient
PResp, # (vrisk) #TCE Respiratory tissue:air partition coefficient
PKid, # (vrisk) #TCE Kidney/Blood partition coefficient
PSlw, # (vrisk) #TCE Slowly perfused/Blood partition coefficient
TCAPlas, # (vrisk) #TCA blood/plasma concentration ratio
PBodTCA, # (vrisk) #Free TCA Body/blood plasma partition coefficient
PLivTCA, # (vrisk) #Free TCA Liver/blood plasma partition coefficient
kDissoc, # (vrisk) #Protein/TCA dissociation constant (umole/L)
BMax, # (vrisk) #Maximum binding concentration (umole/L)
PBodTCOH, # (vrisk) #TCOH body/blood partition coefficient
PLivTCOH, # (vrisk) #TCOH liver/body partition coefficient
PBodTCOG, # (vrisk) #TCOG body/blood partition coefficient
PLivTCOG, # (vrisk) #TCOG liver/body partition coefficient
VDCVG, # (vrisk) #DCVG effective volume of distribution
kAS, # (vrisk) #TCE Stomach absorption coefficient (/hr)
kTSD, # (vrisk) #TCE Stomach-duodenum transfer coefficient (/hr)
-------
kAD, # (vrisk) #TCE Duodenum absorption coefficient (/hr)
kTD, # (vrisk) #TCE Duodenum-feces transfer coefficient (/hr)
kASTCA, # (vrisk) #TCA Stomach absorption coefficient (/hr)
kASTCOH, # (vrisk) #TCOH Stomach absorption coefficient (/hr)
VMax, # (vrisk) #VMax for hepatic TCE oxidation (mg/hr)
KM, # (vrisk) #KM for hepatic TCE oxidation (mg/L)
FracOther, # (vrisk) #Fraction of hepatic TCE oxidation not to TCA+TCOH
FracTCA, # (vrisk) #Fraction of hepatic TCE oxidation to TCA
VMaxDCVG, # (vrisk) #VMax for hepatic TCE GSH conjugation (mg/hr)
KMDCVG, # (vrisk) #KM for hepatic TCE GSH conjugation (mg/L)
VMaxKidDCVG, # (vrisk) #VMax for renal TCE GSH conjugation (mg/hr)
KMKidDCVG, # (vrisk) #KM for renal TCE GSH conjugation (mg/L)
FracKidDCVC, # (vrisk) #Fraction of renal TCE GSH conj. "directly" to DCVC
# (vrisk) #(i.e., via first pass)
VMaxClara, # (vrisk) #VMax for Tracheo-bronchial TCE oxidation (mg/hr)
KMClara, # (vrisk) #KM for Tracheo-bronchial TCE oxidation (mg/L)
FracLungSys, # (vrisk) #Fraction of respiratory metabolism to systemic circ
VMaxTCOH, # (vrisk) #VMax for hepatic TCOH->TCA (mg/hr)
KMTCOH, # (vrisk) #KM for hepatic TCOH->TCA (mg/L)
VMaxGluc, # (vrisk) #VMax for hepatic TCOH->TCOG (mg/hr)
kMetTCOH, # (vrisk) #Rate constant for hepatic TCOH->other (/hr)
kUrnTCA, # (vrisk) #Rate constant for TCA plasma->urine (/hr)
kMetTCA, # (vrisk) #Rate constant for hepatic TCA->other (/hr)
kBile, # (vrisk) #Rate constant for TCOG liver->bile (/hr)
kEHR, # (vrisk) #Lumped rate constant for TCOG bile->TCOH liver (/hr)
kUrnTCOG, # (vrisk) #Rate constant for TCOG->urine (/hr)
kDCVG, # (vrisk) #Rate constant for hepatic DCVG->DCVC (/hr)
kNAT, # (vrisk) #Lumped rate constant for DCVC->Urinary NAcDCVC (/hr)
kKidBioact, # (vrisk) #Rate constant for DCVC bioactivation (/hr)
# Misc
RUrnTCA, #(vrisk)
RUrnTCOGTCOH, #(vrisk)
RUrnNDCVC, #(vrisk)
RAO,
CVenMole,
CPlasTCAMole,
CPlasTCAFreeMole
# Molecular Weights
MWTCE = 131.39; # TCE
MWDCA = 129.0; # DCA
MWDCVC =216.1; # DCVC
MWTCA =163.5; # TCA
MWChlor = 147.5; # Chloral
MWTCOH =149.5; # TCOH
MWTCOHGluc = 325.53; # TCOH-Gluc
# Stoichiometry
StochChlorTCE =
StochTCATCE =
StochTCATCOH =
StochTCOHTCE =
StochGlucTCOH =
StochTCOHGluc =
StochTCEGluc =
StochDCVCTCE =
StochN =
StochDCATCE =
MWChlor / MWTCE;
MWTCA / MWTCE;
MWTCA / MWTCOH;
MWTCOH / MWTCE;
MWTCOHGluc / MWTCOH;
MWTCOH / MWTCOHGluc;
MWTCE / MWTCOHGluc;
MWDCVC / MWTCE;
MWNADCVC / MWDCVC;
MWDCA / MWTCE;
# Flows
QC
QPsamp
VPR
QFatCtmp
QGutCtmp
QLivCtmp
QSlwCtmp
DResptmp
[scaled to QP]
QKidCtmp = 1 ;
FracPlas = 1;
1;
1;
1;
1;
1;
1 ;
1;
1;
# Cardiac output (L/hr)
# Alveolar ventilation (L/hr)
# Alveolar venti1ation-perfusion ratio
# Scaled fat blood flow
# Scaled gut blood flow
# Scaled liver blood flow
# Scaled slowly perfused blood flow
# Respiratory lumen:tissue diffusive clearance rate (L/hr)
VGut
VLiv
VRap
VRespLum
VRespEfftmp
VRespEff = 1;
Resp:Air partition
VBodTCOH = 1;
# Fat compartment volume (L)
# Gut compartment volume (L)
# Liver compartment volume (L)
# Rapidly perfused compartment volume (L)
# Volume of respiratory lumen (L air)
= 1; #(vrisk) volume for respiratory tissue (L)
# Effective volume for respiratory tissue (L air) = V(tissue)
coefficient
# Kidney compartment volume (L)
# Blood compartment volume (L)
# Slowly perfused compartment volume (L)
# Plasma compartment volume [fraction of blood] (L)
# TCA Body compartment volume [not incl. blood+liver]
# TCOH/G Body compartment volume [not incl. liver] (1
-------
PFat
PGut
PLiv
PRap
PResp
PKid
PSlw
TCAPlas
PBodTCA
PLivTCA
kDissoc
BMax
PBodTCOH
PLivTCOH
PBodTCOG
PLivTCOG
VDCVG
TCE Fat/Blood partition coefficient
TCE Gut/Blood partition coefficient
TCE Liver/Blood partition coefficient
TCE Rapidly perfused/Blood partition coefficient
TCE Respiratory tissue:air partition coefficient
TCE Kidney/Blood partition coefficient
TCE Slowly perfused/Blood partition coefficient
TCA blood/plasma concentration ratio
Free TCA Body/blood plasma partition coefficient
Free TCA Liver/blood plasma partition coefficient
Protein/TCA dissociation constant (umole/L)
Protein concentration (UNITS?)
TCOH body/blood partition coefficient
TCOH liver/body partition coefficient
TCOG body/blood partition coefficient
TCOG liver/body partition coefficient
DCVG effective volume of distribution
# TCE Stomach-duodenum transfer coefficient (/hi
# TCE Stomach absorption coefficient (/hr)
# TCE Duodenum-feces transfer coefficient (/hr)
# TCE Duodenum absorption coefficient (/hr)
# TCA Stomach absorption coefficient (/hr)
# TCOH Stomach absorption coefficient (/hr)
# TCE Metabolism
VMax = 1;
KM = 1;
FracOther = 1;
FracTCA = 1;
VMaxDCVG = 1 ;
KMDCVG = 1;
VMaxKidDCVG
KMKidDCVG = 1;
VMaxClara = 1;
KMClara = 1 ;
# VMax for hepatic TCE oxidation (mg/hr)
# KM for hepatic TCE oxidation (mg/L)
# Fraction of hepatic TCE oxidation not to TCA+TCOH
# Fraction of hepatic TCE oxidation to TCA
# VMax for hepatic TCE GSH conjugation (mg/hr)
# KM for hepatic TCE GSH conjugation (mg/L)
= 1; # VMax for renal TCE GSH conjugation (mg/hr)
# KM for renal TCE GSH conjugation (mg/L)
# VMax for Tracheo-bronchial TCE oxidation (mg/hr)
# KM for Tracheo-bronchial TCE oxidation (mg/L)
# but in units of air concentration
= 1; # Fraction of respiratory oxidative metabolism that
culation
# TCOH metabolism
# VMax for hepatic TCOH->TCA (mg/hr)
# KM for hepatic TCOH->TCA (mg/L)
# VMax for hepatic TCOH->TCOG (mg/hr)
# KM for hepatic TCOH->TCOG (mg/L)
# Rate constant for hepatic TCOH->other (/hr)
# TCOG metabolism/clearance
kBile = 1; # Rate constant for TCOG liver->bile (/hr)
kEHR = 1; # Lumped rate constant for TCOG bile->TCOH liver (/hr)
kUrnTCOG = 1; # Rate constant for TCOG->urine (/hr)
#++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
# DCVG metabolism
kDCVG = 1; # Rate constant for hepatic DCVG->DCVC (/hr)
FracKidDCVC =1; # Fraction of renal TCE GSH conj. "directly" to DCVC
(i.e., via first pass)
^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^^i^^.
# Closed chamber and other exposure parameters
Rodents = 1; # Number of rodents in closed chamber data
VCh = 1; # Chamber volume for closed chamber data
kLoss = 1; # Rate constant for closed chamber air loss
CC =0.0; # Initial chamber concentration (ppm)
TChng = 0.003; # IV infusion duration (hour)
## Flag for species, sex -- these are global parameters
BW = 0.0; # Species-specific defaults during initialization
BW75 = 0.0; #(vrisk) Variable for BW"3/4
Male =1.0; # 1 = male, 0 = female
Species = 1.0; # 1 = human, 2 = rat, 3 = mouse
BWmeas =0.0; # Body weight
VFatCmeas = 0.0; # Fractional volume fat
PBmeas =0.0; # Measured blood-air partition coefficient
Hematocritmeas = 0.0; # Measured hematocrit -- used for FracPlas = 1 - HCt
CDCVGmolLD = 5e-5; # Detection limit of CDCVGmol#(vl.2.3.1)
# These parameters are potentially sampled/calibrated in the MCMC or MC
# analyses. The default values here are used if no sampled value is given.
# M_ indicates population mean parameters used only in MC sampling
# V indicates a population variance parameter used in MC and MCMC sampling
# Fractional Blood Flows to Tissues (fraction of cardiac output)
QFatC = 1.0; # Scaled to species-specific central estimates
QGutC = 1.0; # Scaled to species-specific central estimates
QLivC = 1.0; # Scaled to species-specific central estimates
QSlwC = 1.0; # Scaled to species-specific central estimates
-------
o
to
vo £;•
§
a
Fractional Tissue Volumes (fraction of BW)
VFatC = 1.0;
VGutC = 1.0;
VLivC = 1.0;
VRapC = 1.0;
VRespLumC = 1.0;
VRespEffC = 1.0;
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
<§»
^>K
°
TO
•^
TO'
^
^3
5
# Partition
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPRespC =
InPKidC
InPSlwC
Coef
0.0;
0.0;
0.0;
0.0;
0.0;
0.0;
0.0;
0.0;
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Partition Coefficients for TCA
InPRBCPlasTCAC = 0.0; # Scaled to species-specific central estimates
InPBodTCAC = 0.0; # Scaled to species-specific central estimates
InPLivTCAC = 0.0; # Scaled to species-specific central estimates
# Plasma Binding for TCA
InkDissocC = 0.0; # Scaled to species-specific central estimates
InBMaxkDC = 0.0; # Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# Scaled to species-specific central estimates
# TCE Metabolism
InVMaxC =0.0; # Scaled by liver weight and species-specific central estimates
InKMC = 0.0; # Scaled to species-specific central estimates
InCIC = 0.0; # Scaled to species-specific central estimates
InFracOtherC
InFracTCAC
InVMaxDCVGC
estimates
InClDCVGC = 0.0;
InKMDCVGC = 0.0;
InVMaxKidDCVGC
estimates
InClKidDCVGC
InKMKidDCVGC
InVMaxLungLivC
= 0.0; # Scaled to species-specific central estimates
= 0.0; # Scaled to species-specific central estimates
=0.0; # Ratio of lung Vmax to liver Vmax,
# Scaled to species-specific central estimates
# now in units of air concentration
# Clearance in lung
InFracLungSysC
oxidation
InVMaxTCOHC
InClTCOHC = 0.0;
InKMTCOH = 0.0;
InVMaxGlucC
InClGlucC = 0.0;
InKMGluc = 0.0;
InkMetTCOHC
# TCA Metabolism/clearance
InkUrnTCAC = 0.0;
central estimates
InkMetTCAC = 0.0;
# TCOG excretion and reabsorption
InkBileC =0.0; # Scaled by BW"-0.25
InkEHRC = 0.0;
InkUrnTCOGC
# Closed chamber parameters
NRodents = 1; #
VChC = 1; t
InkLossC =0; #
# Population means
-------
# These are given truncated normal or uniform distributions, depending on
prior information is available. Note that these distributions
reflect uncertainty in the population mean, not inter-individual
ty. Normal distributions are truncated at 2, 3, or 4 SD.
For fractional volumes and flows, 2xSD
For plasma fraction, 3xSD
For cardiac output and venti1ation-perfusion ratio, 4xSD
For all others, 3xSD
For uniform distributions, range of Ie2 to Ie8 fold, centered on
central estimate.
to
o
o
VO
£>•
1
oo
oo
O
r>-
ITJ
H
O
^
2|
Q
H
O
HH
H
W
Q
hj
0
s
H
W
£3'
o"
s
s
§
~^
£5'
a
ft*
2
^Js
^
•s
TO
^;
TO'
^
^3
§;
o
Co
O
•3
V:
s
a.
o
TO
3
^.
§
Co
w.
S
TO^
ffe'
TO
S
%^
^3
§r
^
# what prii
# reflect i
# variabil.
#
#
t
#
# For unifi
#
#
M InQCC = 1.0;
M InVPRC = 1.0;
M QFatC = 1.0;
M QGutC = 1.0;
M QLivC = 1.0;
M QSlwC = 1.0;
M QKidC = 1.0;
M FracPlasC
M InDRespC = 1.0;
M VFatC = 1.0;
M VGutC = 1.0;
M VLivC = 1.0;
M VRapC = 1.0;
M VRespLumC = 1.0;
M VRespEffC = 1.0;
M VKidC = 1.0;
M VBldC = 1.0;
M InPBC = 1.0;
M InPFatC = 1.0;
M InPGutC = 1.0;
M InPLivC = 1.0;
M InPRapC = 1.0;
M InPRespC
M InPKidC = 1.0;
M InPSlwC = 1.0;
M InPRBCPlasTCAC
M InPBodTCAC
M InPLivTCAC
M InkDissocC
M InBMaxkDC
M InPBodTCOHC
M InPLivTCOHC
M InPBodTCOGC
M InPLivTCOGC
M InPeffDCVG
M InkTSD = 1.0;
M InkAS = 1.0;
M InkTD = 1.0;
M InkAD = 1.0;
M InkASTCA
M InkASTCOH
M_lnVMaxC = 1.0;
M_lnKMC = 1.0;
M_lnClC = 1.0;
M_lnFracOtherC
M_lnFracTCAC
M_lnVMaxDCVGC
M_lnClDCVGC
M_lnKMDCVGC
M_lnVMaxKldDCVGC
M_lnClKldDCVGC
M_lnKMKldDCVGC
M_lnVMaxLungLivC
M_lnKMClara
M InFracLungSysC
M_lnVMaxTCOHC
M_lnClTCOHC
M_lnKMTCOH
M_lnVMaxGlucC
M_lnClGlucC
M_lnKMGluc
M_lnkMetTCOHC
M_lnkUrnTCAC
M_lnkMetTCAC
M_lnkBlleC
M_lnkEHRC = 1.0;
M_lnkUrnTCOGC
M_lnFracKldDCVCC
M_lnkDCVGC
M_lnkNATC = 1.0;
M InkKldBloactC
Population Variances
These are given InvGamma(alpha,beta) distributions. The parameterization
for alpha and beta is given by:
alpha = (n-1)/2
beta = s"2* (n-1)/2
where n = number of data points, and sr'2 is the sample variance
Sum(x_i"2)/n - ~2.
Generally, for parameters for which there is no direct data, assume a
value of n = 5 (alpha = 2). For a sample variance s~2, this gives
an expected value for the standard deviation = 0.9*s,
a median [2.58,97.58] of 1.1*s [ 0.6*s,2.9*s] .
V_lnQCC =1.0
V_lnVPRC =1.0
V_QFatC =1.0
V_QGutC =1.0
V_QLivC =1.0
V_QSlwC =1.0
V_QKidC =1.0
-------
1— >
^
to
-^
^o
^.
1
^>K
°
TO
•^
TO'
^
^
^
^3
o
TO
O
s
V:
s
a.
o
TO
S
^*.
n
2
<*,
'••C
s^
TO"
h^>-
^
s
%^
^3
§r
^
V FracPlasC
V InDRespC = 1.0;
V VFatC = 1.0;
V VGutC = 1.0;
V VLlvC = 1.0;
V VRapC = 1.0;
V VRespLumC = 1.0;
V VRespEffC = 1.0;
V VKldC = 1.0;
V VBldC = 1.0;
V InPBC = 1.0;
V InPFatC = 1.0;
V InPGutC = 1.0;
V InPLlvC = 1.0;
V InPRapC = 1.0;
V InPRespC
V InPKldC = 1.0;
V InPSlwC = 1.0;
V InPRBCPlasTCAC
V InPBodTCAC
V InPLlvTCAC
V InkDissocC
V InBMaxkDC
V InPBodTCOHC
V InPLlvTCOHC
V InPBodTCOGC
V InPLlvTCOGC
V InPeffDCVG
V InkTSD = 1.0;
V InkAS = 1.0;
V InkTD = 1.0;
V InkAD = 1.0;
V InkASTCA
V InkASTCOH
V InVMaxC = 1.0;
V InKMC = 1.0;
V InCIC = 1.0;
V InFracOtherC
V InFracTCAC
V InVMaxDCVGC
V InClDCVGC
V InKMDCVGC
V InVMaxKldDCVGC
V InClKldDCVGC
V InKMKldDCVGC
V InVMaxLungLivC
V InKMClara
V InFracLungSysC
V InVMaxTCOHC
V InClTCOHC
V InKMTCOH
V InVMaxGlucC
V InClGlucC
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
= 1.0
V_lnKMGluc
V_lnkMetTCOHC
V_lnkUrnTCAC
V_lnkMetTCAC
V_lnkBlleC
V_lnkEHRC = 1.0;
V_lnkUrnTCOGC
V_lnFracKldDCVCC
V_lnkDCVGC
V_lnkNATC = 1.0;
V InkKidBioactC
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
Ve
RetDose
CAlv = 1;
CAlvPPM
CInhPPM
CInh = 1;
CMlxExh
CArt = 1;
CVen = 1 ;
CBldMlx
CFat = 1;
CGut = 1;
CRap = 1;
CSlw = 1;
CHrt = 1;
CKld = 1;
CLlv = 1;
CLung = 1 ;
CMus = 1;
CSpl = 1;
CBrn = 1;
zAExh = 1 ;
= 1
= 1
= 1
= 1
= 1
Ve_zAExhpost
Ve_CPlasTCA
Ve_CBldTCA
Ve_CBodTCA
Ve_CKldTCA
Ve_CLlvTCA
Ve_CLungTCA
Ve zAUrnTCA
-------
Ve_zAUrnTCA_collect = 1;
Ve zAUrnTCA sat = 1 ;
Ve_zABileTCOG
Ve_CTCOG = 1;
Ve_CTCOGTCOH
Ve_CKidTCOGTCOH
Ve_CLivTCOGTCOH
Ve_CLungTCOGTCOH
Ve AUrnTCOGTCOH
Ve_CDCVGmol
Ve_zAUrnNDCVC
Ve_AUrnTCTotMole
Ve_TotCTCOH
Ve QPsamp = 1;
Defaults for input parameters
# Inhalation exposure cone. (ppm)
# IV dose (ing/kg)
# Oral gavage dose (ing/kg)
# Drinking water dose (mg/kg/day)
# Intraarterial dose (ing/kg)
# Portal vein dose (ing/kg)
##— TCE dosing
Cone = 0.0;
IVDose = 0.0;
PDose = 0.0;
Drink = 0.0;
lADose = 0.0;
PVDose = 0.0;
##— TCA dosing
IVDoseTCA = 0.0;# IV dose (ing/kg) of TCA
PODoseTCA = 0.0;# Oral dose (mg/kg) of TCA
##— TCOH dosing
IVDoseTCOH = 0.0;# IV dose (mg/kg) of TCOH
PODoseTCOH = 0.0;# Oral dose (mg/kg) of TCOH
##-- Potentially time-varying parameters
QPmeas =0.0; # Measured value of Alveolar ventilation QP
TCAUrnSat = 0.0;# Flag for saturated TCA urine
TCOGUrnSat = 0.0;# Flag for saturated TCOG urine
UrnMissing = 0.0;# Flag for missing urine collection times
Model Parameters (used in dynamics):
QC Cardiac output (L/hr)
VPR Venti1ation-perfusion ratio
QPsamp Alveolar ventilation (L/hr)
QFatCtmp Scaled fat blood flow
QGutCtmp Scaled gut blood flow
QLivCtmp Scaled liver blood flow
QSlwCtmp Scaled slowly perfused blood flow
DResptmp Respiratory lumen: tissue diffusive clearance rate
QKidCtmp Scaled kidney blood flow
FracPlas Fraction of blood that is plasma ( 1-hematocrit )
VFat Fat compartment volume (L)
VGut Gut compartment volume (L)
VLiv Liver compartment volume (L)
VRap Rapidly perfused compartment volume (L)
VRespLum Volume of respiratory lumen (L air)
VRespEf f Effective volume of respiratory tissue (L air)
VKid Kidney compartment volume (L)
VBld Blood compartment volume (L)
VSlw Slowly perfused compartment volume (L)
VPlas Plasma compartment volume [ fraction of blood] (L)
VBod TCA Body compartment volume [not incl . blood+ liver]
VBodTCOH TCOH/G Body compartment volume [not incl. liver] (L)
PB TCE Blood/ air partition coefficient
PFat TCE Fat/Blood partition coefficient
PGut TCE Gut/Blood partition coefficient
PLiv TCE Liver/Blood partition coefficient
PRap TCE Rapidly perfused/Blood partition coefficient
PResp TCE Respiratory tissue : air partition coefficient
PKid TCE Kidney/Blood partition coefficient
PSlw TCE Slowly perfused/Blood partition coefficient
T CAP las TCA blood/ plasma concentration ratio
PBodTCA Free TCA Body/blood plasma partition coefficient
PLivTCA Free TCA Liver/blood plasma partition coefficient
kDissoc Protein/ TCA dissociation constant (umole/L)
BMax Maximum binding concentration (umole/L)
PBodTCOH TCOH body/blood partition coefficient
PLivTCOH TCOH liver/body partition coefficient
PLivTCOG TCOG liver/body partition coefficient
TCE Stomach absorption coefficient (/hr)
TCE Stomach-duodenum transfer coefficient (/hr)
TCE Duodenum absorption coefficient (/hr)
TCE Duodenum-feces transfer coefficient (/hr)
TCA Stomach absorption coefficient (/hr)
TCOH Stomach absorption coefficient (/hr)
VMax for hepatic TCE oxidation (mg/hr)
KM for hepatic TCE oxidation (mg/L)
FracOther Fraction of hepatic TCE oxidation not to TCA+TCOH
FracTCA Fraction of hepatic TCE oxidation to TCA
VMaxDCVG VMax for hepatic TCE GSH conjugation (mg/hr)
KMDCVG KM for hepatic TCE GSH conjugation (mg/L)
VMaxKidDCVG VMax for renal TCE GSH conjugation (mg/hr)
KMKidDCVG KM for renal TCE GSH conjugation (mg/L)
VMaxClara VMax for Tracheo-bronchial TCE oxidation (mg/hr)
KMClara KM for Tracheo-bronchial TCE oxidation (mg/L)
-------
FracLungSys Fraction of respiratory metabolism to systemic circ.
VMaxTCOH VMax for hepatic TCOH->TCA (mg/hr)
KMTCOH KM for hepatic TCOH->TCA (mg/L)
VMaxGluc VMax for hepatic TCOH->TCOG (mg/hr)
KMGluc KM for hepatic TCOH->TCOG (mg/L)
kMetTCOH Rate constant for hepatic TCOH->other (/hr)
kUrnTCA Rate constant for TCA plasma->urine (/hr)
kMetTCA Rate constant for hepatic TCA->other (/hr)
kBile Rate constant for TCOG liver->bile (/hr)
kEHR Lumped rate constant for TCOG bile->TCOH liver (/hr)
kUrnTCOG Rate constant for TCOG->urine (/hr)
kDCVG Rate constant for hepatic DCVG->DCVC (/hr)
FracKidDCVC Fraction of renal TCE GSH conj. "directly" to DCVC
(i.e., via first pass)
DCVG effective volume of distribution
Lumped rate constant for DCVC->Urinary NAcDCVC (/hr)
Rate constant for DCVC bioactivation (/hr)
Number of rodents in closed chamber data
Chamber volume for closed chamber data
Rate constant for closed chamber air loss
Parameters used (not assigned here)
BW Body weight in kg
Species 1 = human (default), 2 = rat, 3 = mouse
Male 0 = female, 1 (default) = male
CC Closed chamber initial concentration
Sampling/scaling parameters (assigned or sampled)
InQCC
InVPRC
InDRespC
QFatC
QGutC
QLivC
QSlwC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEffC
VKidC
VBldC
InPBC
InPFatC
InPGutC
InPLivC
InPRapC
InPSlwC
InPRespC
InPKidC
InPRBCPlasTCAC
InPBodTCAC
InPLivTCAC
InkDissocC
InBMaxkDC
InPBodTCOHC
InPLivTCOHC
InPBodTCOGC
InPLivTCOGC
InPeffDCVG
InkTSD
InkAS
InkTD
InkAD
InkASTCA
InkASTCOH
InVMaxC
InKMC
InCIC
InFracOtherC
InFracTCAC
InVMaxDCVGC
InClDCVGC
InKMDCVGC
InVMaxKidDCVGC
InClKidDCVGC
InKMKidDCVGC
InVMaxLungLivC
InKMClara
InFracLungSysC
InVMaxTCOHC
InClTCOHC
InKMTCOH
InVMaxGlucC
InClGlucC
InKMGluc
InkMetTCOHC
InkUrnTCAC
InkMetTCAC
InkBileC
InkEHRC
InkUrnTCOGC
InFracKidDCVCC
InkDCVGC
InkNATC
InkKidBioactC
NRodents
VChC
InkLossC
Input parameters
none
Notes:
-------
to
'
I
I
§
Co
1
to O
•••2
I
H I
O >
HH Oq
H TO
O
H
W
# Cardiac Output and alveolar ventilation (L/hr)
QC = exp(lnQCC) * BW75 * # Mouse, Rat, Human (default)
(Species == 3 ? 11.6 : (Species == 2 ? 13.3 : 16.0 ));
# Mouse: 00=13.98 +/- 2.85 ml/min, BW=30 g (Brown et al. 1997, Tab. 22)
# Uncertainty CV is 0.20
# Rat: 00=110.4 ml/min +/- 15.6, BW=396 g (Brown et al. 1997, Tab. 22,
# p 441). Uncertainty CV is 0.14.
# Human: Average of Male 00=6.5 1/min, BW=73 kg
# and female 00= 5.9 1/min, BW=60 kg (ICRP #89, sitting at rest)
# From Price et al. 2003, estimates of human perfusion rate were
# 4.7-6.5 for females and 5.5-7.1 1/min for males (note
# portal blood was double-counted, and subtracted off here)
# Thus for uncertainty use CV of 0.2, truncated at 4xCV
# Variability from Price et al. (2003) had CV of 0.14-0.20,
# so use 0.2 as central estimate
VPR = exp(lnVPRC)*
(Species == 3 ? 2.5 : (Species == 2 ? 1.9 : 0.96 ) ) ;
# Mouse: QP/BW=116.5 ml/min/100 g (Brown et al. 1997, Tab. 31), VPR=2.5
# Assume uncertainty CV of 0.2 similar to QC, truncated at 4xCV
# Consistent with range of QP in Tab. 31
# Rat: QP/BW=52.9 ml/min/100 g (Brown et al. 1997, Tab. 31), VPR=1.9
# Assume uncertainty CV of 0.3 similar to QC, truncated at 4xCV
# Used larger CV because Tab. 31 shows a very large range of QP
# Human: Average of Male VE=9 1/min, resp. rate=12 /min,
# dead space=0.15 1 (QP=7.2 1/min), and Female
# VE=6.5 1/min, resp. rate=14 /min, dead space=0.12 1
# (QP=4.8 1/min), VPR =0.96
# Assume uncertainty CV of 0.2 similar to QC, truncated at 4xCV
# Consistent with range of QP in Tab. 31
QPsamp = QC^VPR;
# Respiratory diffusion flow rate
# Will be scaled by QP in dynamics
# Use log-uniform distribution from le-5 to 10
DResptmp = exp(InDRespC);
# Fractional Flows scaled to the appropriate species
# Fat = Adipose only
# Gut = GI tract + pancreas + spleen (all drain to portal vein)
# Liv = Liver, hepatic artery
# Slw = Muscle + Skin
# Kid = Kidney
# Rap = Rapidly perfused (rest of organs, plus bone marrow, lymph, etc.),
# derived by difference in dynamics
QLivCtmp = QLivC*
(Species == 3 ? 0.02 : (Species == 2 ? 0.021 : 0.065 ));
QSlwCtmp = QSlwC*
(Species == 3 ? 0.217 : (Species == 2 ? 0.336 : (Male == 0
# Plasma Flows to Tissues (L/hr)
## Mice and rats from Hejtmancik et al. 2002,
## control F344 rats and B6C3F1 mice at 19 weeks of age
## However, there appear to be significant strain differences in rodents, so
## assume uncertainty CV=0.2 and variability CV=0.2.
## Human central estimate from ICRP. Well measured in humans, from Price et al.,
## human SD in hematocrit was 0.029 in females, 0.027 in males,
## corresponding to FracPlas CV of 0.047 in females and
## 0.048 in males. Use rounded CV = 0.05 for both uncertainty and
variability
## Use measured 1-hematocrit if available
## Truncate distributions at 3xCV to encompass clinical "normal range"
FracPlas = (Hematocritmeas > 0.0 ? (1-Hematocritmeas) : (FracPlasC *
(Species == 3 ? 0.52 : (Species == 2 ? 0.53 : (Male == 0 ? 0.615 :
0.567)))));
# Tissue Volumes (L)
# Fat = Adipose only
# Gut = GI tract (not contents) + pancreas + spleen (all drain to portal vein)
# Liv = Liver
# Rap = Brain + Heart + (Lungs-TB) + Bone marrow + "Rest of the body"
# VResp = Tracheobroneial region (trachea+broncial basal+
# broneial secretory+bronchiolar)
# Kid = Kidney
# Bid = Blood
# Slw = Muscle + Skin, derived by difference
# residual (assumed unperfused) = (Bone-Marrow)+GI contents+other
VFat = BW * (VFatCmeas > 0.0 ? VFatCmeas : (VFatC * (Species ==3 ? 0.07 :
(Species == 2 ? 0.07 : (Male == 0 ? 0.317 : 0.199) ))));
VGut = VGutC * BW *
(Species == 3 ? 0.049 : (Species == 2 ? 0.032 : (Male == 0 ? 0.022 :
0.020) )) ;
VLiv = VLivC * BW *
-------
(Species == 3 ? 0.055 : (Species == 2 ? 0.034 : (Male == 0 ? 0.023 :
0.025) ));
VRap = VRapC * BW *
(Species == 3 ? 0.100 : (Species == 2 ? 0.088 : (Male == 0 ? 0.093 :
0.088) ));
VRespLum = VRespLumC * BW *
(Species == 3 ? (0.00014/0.03) : (Species == 2 ? (0.0014/0.3) : (0.167/70)
)); # Lumenal volumes from Styrene model (Sarangapani et al. 2002)
VRespEfftmp = VRespEffC * BW *
(Species == 3 ? 0.0007 : (Species == 2 ? 0.0005 : 0.00018 ));
# Respiratory tract volume is TB region
# will be multiplied by partition coef. below
VKid = VKidC * BW *
(Species == 3 ? 0.017 : (Species == 2 ? 0.007 : (Male == 0 ? 0.0046 :
0.0043) ) ) ;
VBld = VBldC * BW *
(Species == 3 ? 0.049 : (Species == 2 ? 0.074 : (Male == 0 ? 0.068 :
0.077) ));
VSlw = (Species == 3 ? 0.8897 : (Species == 2 ? 0.8995 : (Male == 0 ?
0.85778 : 0.856))) * BW
- VFat - VGut - VLiv - VRap - VRespEfftmp - VKid - VBld;
# Slowly perfused:
# Baseline mouse: 0.8897-0.049-0.017-0.0007-0.1-0.055-0.049-0.07= 0.549
# Baseline rat: 0.8995 -0.074-0.007-0.0005-0.088-0.034-0.032-0.07= 0.594
# Baseline human F: 0.85778-0.068-0.0046-0.00018-0.093-0.023-0.022-0.317= 0.33
# Baseline human M: 0.856-0.077-0.0043-0.00018-0.088-0.025-0.02-0.199= 0.4425
Partition coefficients
PB = (PBmeas > 0.0 ? PBmeas : (exp(InPBC) * (Species == 3 ? 15. : (Species
? 22. : 9.5 )))); # Blood-air
# Mice: pooling Abbas and Fisher 1997, Fisher et al. 1991
# each a single measurement, with overall CV = 0.07.
# Given small number of measurements, and variability
# in rat, use CV of 0.25 for uncertainty and variability.
# Rats: pooling Sato et al. 1977, Gargas et al. 1989,
# Barton et al. 1995, Simmons et al. 2002, Koizumi 1989,
# Fisher et al. 1989. Fisher et al. measurement substantially
# smaller than others (15 vs. 21-26). Recent article
# by Rodriguez et al. 2007 shows significant change with
# age (13.1 at PND10, 17.5 at adult, 21.8 at aged), also seems
# to favor lower values than previously reported. Therefore
# use CV = 0.25 for uncertainty and variability.
# Humans: pooling Sato and Nakaj ima 1979, Sato et al. 1977,
# Gargas et al. 1989, Fiserova-Bergerova et al. 1984,
# Fisher et al. 1998, Koizumi 1989
# Overall variability CV = 0.185. Consistent with
# within study inter-individual variability CV = 0.07-0.22.
# Study-to-study, sex-specific means range 8.1-11, so
# uncertainty CV = 0.2.
PFat = exp(lnPFatC) * # Fat/blood
(Species == 3 ? 36. : (Species == 2 ? 27. : 67. ));
# Mice: Abbas and Fisher 1997. Single measurement. Use
# rat uncertainty of CV = 0.3.
# Rats: Pooling Barton et al. 1995, Sato et al. 1977,
# Fisher et al. 1989. Recent article by Rodriguez et al.
# (2007) shows higher value of 36., so assume uncertainty
# CV of 0.3.
# Humans: Pooling Fiserova-Bergerova et al. 1984, Fisher et al. 1998,
# Sato et al. 1977. Variability in Fat:Air has CV = 0.07.
# For uncertainty, dominated by PB uncertainty CV = 0.2
# For variability, add CVs in guadrature for
# sgrt(0.07^2+0.185"2)=0.20
PGut = exp(lnPGutC) * # Gut/blood
(Species ==3 ? 1.9 : (Species ==2 ? 1.4 : 2.6 ));
# Mice: Geometric mean of liver, kidney
# Rats: Geometric mean of liver, kidney
# Humans: Geometric mean of liver, kidney
# Uncertainty of CV = 0.4 due to tissue extrapolation
PLiv = exp(lnPLivC) * # Liver/blood
(Species == 3 ? 1.7 : (Species == 2 ? 1.5 : 4.1 ));
# Mice: Fisher et al. 1991, single datum, so assumed uncert CV = 0.4
# Rats: Pooling Barton et al. 1995, Sato et al. 1977,
# Fisher et al. 1989, with little variation (range 1.3-1.7).
# Recent article by Rodriguez et al.reports 1.34. Use
# uncertainty CV = 0.15.
# Humans: Pooling Fiserova-Bergerova et al. 1984, Fisher et al. 1998
# almost 2-fold difference in Liver:Air values, so uncertainty
# CV = 0.4
PRap = exp(lnPRapC) * # Rapidly perfused/blood
(Species == 3 ? 1.9 : (Species == 2 ? 1.3 : 2.6 ));
# Mice: Similar to liver, kidney. Uncertainty CV = 0.4 due to
# tissue extrapolation
# Rats: Use brain values Sato et al. 1977. Recent article by
# Rodriguez et al. (2007) reports 0.99 for brain. Uncertainty
# CV of 0.4 due to tissue extrapolation.
# Humans: Use brain from Fiserova-Bergerova et al. 1984
# Uncertainty of CV = 0.4 due to tissue extrapolation
PResp = exp(InPRespC) * # Resp/blood =
(Species ==3 ? 2.6 : (Species ==2 ? 1.0 : 1.3 ));
# Mice: Abbas and Fisher 1997, single datum, so assumed uncert CV = 0.4
# Rats: Sato et al. 1977, single datum, so assumed uncert CV = 0.4
# Humans: Pooling Fiserova-Bergerova et al. 1984, Fisher et al. 1998
# > 2-fold difference in lung:air values, so uncertainty
# CV = 0.4
VRespEff = VRespEfftmp * PResp * PB; # Effective air volume
PKid = exp(lnPKidC) * # Slowly perfused/blood
(Species == 3 ? 2.1 : (Species == 2 ? 1.3 : 1.6 ));
# Mice: Abbas and Fisher 1997, single datum, so assumed uncert CV = 0.4
# Rats: Pooling Barton et al. 1995, Sato et al. 1977. Recent article
# by Rodriguez et al. (2007) reports 1.01, so use uncertainty
# CV of 0.3. Pooled variability CV = 0.39.
# Humans: Pooling Fiserova-Bergerova et al. 1984, Fisher et al. 1998
-------
# For uncertainty, dominated by PB uncertainty CV = 0.2
# Variability in kidney:air CV = 0.23, so add to PB variability
# in quadrature sqrt(0.23"2+0.185"2)=0.30
PSlw = exp(InPSlwC) * # Slowly perfused/blood
(Species == 3 ? 2.4 : (Species == 2 ? 0.58 : 2.1 ));
# Mice: Muscle - Abbas and Fisher 1997, single datum, so assumed
# uncert CV = 0.4
# Rats: Pooling Barton et al. 1995, Sato et al. 1977,
# Fisher et al. 1989. Recent article by Rodriguez et al. (2007)
# reported 0.72, so use uncertainty CV of 0.25. Variability
# in Muscle:air and muscle:blood - CV = 0.3
# Humans: Pooling Fiserova-Bergerova et al. 1984, Fisher et al. 1998
# Range of values 1.4-2.4, so uncertainty CV = 0.3
# Variability in muscle:air CV =0.3, so add to PB variability
# in quadrature sqrt(0.3"2+0.185"2)=0.35
# TCA partitioning
TCAPlas = FracPlas + (1 - FracPlas) * 0.5 * exp(InPRBCPlasTCAC);
# Blood/Plasma concentration ratio. Note dependence
# on fraction of blood that is plasma. Here
# exp(InPRBCPlasTCA) = partition coefficient
# C(blood minus plasma)/C(plasma)
# Default of 0.5, corresponding to Blood/Plasma
# concentration ratio of 0.76 in
# rats (Schultz et al 1999)
# For rats, Normal uncertainty with GSD = 1.4
# For mice and humans, diffuse prior uncertainty of
# 100-fold up/down
PBodTCA = TCAPlas * exp(InPBodTCAC) *
(Species == 3 ? 0.88 : (Species == 2 ? 0.88 : 0.52 ));
# Note -- these were done at 10-20 microg/ml (Abbas and Fisher 1997),
# which is 1.635-3.27 mmol/ml (1.635-3.27 x 10^6 microM).
# At this high concentration, plasma binding should be
# saturated -- e.g., plasma albumin concentration was
# measured to be P=190-239 microM in mouse, rat, and human
# plasma by Lumpkin et al. 2003, or > 6800 molecules of
# TCA per molecule of albumin. So the measured partition
# coefficients should reflect free blood-tissue partitioning.
# Used muscle values, multiplied by blood:plasma ratio to get
# Body:Plasma partition coefficient
# Rats = mice from Abbas and Fisher 1997
# Humans from Fisher et al. 1998
# Uncertainty in mice, humans GSD = 1.4
# For rats, GSD = 2.0, based on difference between mice
# and humans.
PLivTCA = TCAPlas * exp(InPLivTCAC) *
(Species == 3 ? 1.18 : (Species == 2 ? 1.18 : 0.66 ));
# Multiplied by blood:plasma ratio to get Liver:Plasma
# Rats = mice from Abbas and Fisher 1997
# Humans from Fisher et al. 1998
# Uncertainty in mice, humans GSD = 1.4
# For rats, GSD = 2.0, based on difference between mice
# and humans.
# Binding Parameters for TCA
# GM of Lumpkin et al . 2003; Schultz et al . 1999;
# Templin et al . 1993, 1995; Yu et al . 2000
# Protein/ TCA dissociation constant (umole/L)
# note - GSD = 3.29, 1.84, and 1.062 for mouse, rat, human
kDissoc = exp (InkDissocC) *
(Species == 3 ? 107. : (Species == 2 ? 275. : 182. ));
# BMax = NSites * Protein concentration. Sampled parameter is
# BMax/kD (determines binding at low concentrations )
# note - GSD = 1.64, 1.60, 1.20 for mouse, rat, human
BMax = kDissoc * exp (InBMaxkDC) *
(Species == 3 ? 0.88 : (Species == 2 ? 1.22 : 4.62 ));
# TCOH partitioning
# Data from Abbas and Fisher 1997 (mouse) and Fisher et al .
# 1998 (human). For rat, used mouse values.
# Uncertainty in mice, humans GSD = 1.4
# For rats , GSD = 2 . 0 , based on difference between mice
# and humans .
PBodTCOH = exp(lnPBodTCOHC) *
(Species == 3 ? 1.11 : (Species == 2 ? 1.11 : 0.91 ));
PLivTCOH = exp(lnPLivTCOHC) *
(Species == 3 ? 1.3 : (Species == 2 ? 1.3 : 0.59 ));
# TCOG partitioning
# Use TCOH as a proxy, but uncertainty much greater
# (e.g., use uniform prior, 100-fold up/ down)
PBodTCOG = exp(lnPBodTCOGC) *
(Species == 3 ? 1.11 : (Species == 2 ? 1.11 : 0.91 ));
PLivTCOG = exp(lnPLivTCOGC) *
(Species == 3 ? 1.3 : (Species == 2 ? 1.3 : 0.59 ));
# DCVG distribution volume
# exp (InPef fDCVG) is the effective partition coefficient for
# the "body" (non-blood) compartment
# Diffuse prior distribution: logunif orm le-3 to Ie3
VDCVG = VBld + # blood plus body (with "effective" PC)
exp (InPef fDCVG) * (VBod + VLiv) ;
# Absorption Rate Constants (/hr)
# All priors are diffuse (log) uniform distributions
# transfer from stomach centered on 1 . 4/hr , range up or down 100-fold,
# based on human stomach half-time of 0.5 hr .
kTSD = exp (IrikTSD) ;
# stomach absorption centered on 1 . 4/hr , range up or down 1000-fold
kAS = exp (InkAS) ;
# assume no fecal excretion -- 100% absorption
kTD = 0.0 * exp(lnkTD) ;
# intestinal absorption centered on 0 . 75/hr , range up or down
# 1000-fold, based on human transit time of small intestine
# of 4 hr (95§ throughput in 4 hr)
-------
TO
O
TCE
For
For
All
For
Oxidative Metabolism Constants
rodents, in vitro microsomal data define priors (pooled).
human, combined in vitro microsomoal+hepatocellular individual data
define priors.
data from Elfarra et al. 1998; Lipscomb et al. 1997, 1998a,b
VMax, scaling from in vitro data were (Barter et al. 2007):
32 mg microsomal protein/g liver
99 x Ie6 hepatocytes/g liver
Here, human data assumed representative of mouse and rats.
KM, two different scaling methods were used for microsomes:
Assume microsomal concentration = liver concentration, and
use central estimate of liver:blood PC (see above)
Use measured microsome:air partition coefficient (1.78) and
central estimate of blood:air PC (see above)
human KM from hepatocytes, used measured human hepatocyte:air
partition coefficient (21.62, Lipscomb et al. 1998), and
central estimate of blood:air PC.
Note that to that the hepatocyte:air PC is similar to that
found in liver homogenates (human: 29.4+/-5.1 from Fiserova-
Bergerova et al. 1984, and 54 for Fisher et al. 1998; rat:
27.2+7-3.4 from Gargas et al. 1989, 62.7 from Koisumi 1989,
43.6 from Sato et al. 1977; mouse: 23.2 from Fisher et al. 1991).
humans, sampled parameters are VMax and C1C (VMax/KM), due to
improved convergence. VMax is kept as a parameter because it
appears less uncertain (i.e., more consistent across microsomal
and hepatocyte data).
# Central estimate of VMax is 342, 76.2, and 32.3 (micromol/min/
# kg liver) for mouse, rat, human. Converting to /hr by
# * (60 min/hr * 0.1314 mg/micromol) gives
# 2700, 600, and 255 mg/hr/kg liver
# Observed variability of about 2-fold GSD. Assume 2-fold GSD for
# both uncertainty and variability
VMax = VLiv^exp(InVMaxC)*
(Species == 3 ? 2700. : (Species == 2 ? 600. : 255.));
# For mouse and rat central estimates for KM are 0.068-1.088 and
# 0.039-0.679 mmol/1 in blood, depending on the scaling
# method used. Taking the geometric mean, and converting
# to mg/1 by 131.4 mg/mmol gives 36. and 21. mg/1 in blood.
# For human, central estimate
# for Cl are 0.306-3.95 1/min/kg liver. Taking the geometric
# mean and converting to /hr gives a central estimate of
# 66. 1/hr/kg.
# KM is then derived from KM = VMax/(Cl*Vliv) (central estimate
# of
# Note uncertainty due to scaling is about 4-fold.
# Variability is about 3-fold in mice, 1.3-fold in rats, and
# 2- to 4- fold in humans (depending on scaling).
# Oxidative metabolism splits
# Fractional split of TCE to DCA
# exp(InFracOtherC) = ratio of DCA to non-DCA
# Diffuse prior distribution: loguniform le-4 to Ie2
FracOther = exp(InFracOtherC)/(1+exp(InFracOtherC));
# Fractional split of TCE to TCA
# exp(lnFracTCAC) = ratio of TCA to TCOH
# TCA/TCOH = 0.1 from Lipscomb et al. 1998 using fresh hepatocytes,
# but TCA/TCOH - 1 from Bronley-DeLancey et al 2006
# GM = 0.32, GSD =3.2
FracTCA = 0.32*exp(InFracTCAC)*(1-FracOther)/(1+0.32*exp(InFracTCAC));
# TCE GSH Metabolism Constants
# Human in vitro data from Lash et al. 1999, define human priors.
# VMax (nmol/min/ KM (mM) CLeff (ml/min/
# gtissue) gtissue)
#
Human liver cytosol:
Human liver cytosol+
microsomes
* estimated visually from Fig 1, Lash et al. 1999
** Fig 1A, data from 50-500 ppm headspace at 60 min
and Fig IB, data at 100-5000 ppm in headspace for 120 min
*** Fig IB, 30-100 ppm headspace, converted to blood concentration
using blood:air PC of 9.5
**** Fig 1A, data at 50 ppm headspace at 120 min and Fig IB, data at
25 and 50 ppm headspace at 120 min.
Overall, human liver hepatocytes are probably most like the
intact liver (e.g., accounting for the competition between
GSH conjugation and oxidation). So central estimates based
on those: CLeff - 0.32 ml/min/g tissue, KM - 0.022 mM in blood.
CLeff converted to 19 1/hr/kg; KM converted to 2.9 mg/1 in blood
However, uncertainty in CLeff is large (values in cytosol
-100-fold larger). Moreover, Green et al. 1997 reported
DCVG formation in cytosol that was -30,000-fold smaller
than Lash et al. (1998) in cytosol, which would be a VMax
-300-fold smaller than Lash et al. (1998) in hepatocytes.
Uncertainty in KM appears smaller (-4-fold)
CLC: GM = 19., GSD = 100; KM: GM = 2.9., GSD = 4.
In addition, at a single concentration, the variability
in human liver cytosol samples had a GSD=1.3.
For the human kidney, the kidney cytosol values are used, with the same
uncertainty as for the liver. Note that the DCVG formation rates
in rat kidney cortical cells and rat cytosol are guite similar
(see below).
CLC: GM = 230., GSD = 100; KM: GM = 2.7., GSD = 4.
-------
Rat and mouse in vitro data from Lash et al. 1995,1998 define rat and mouse
priors. However, rats and mice are only assayed at 1 and 2 mM
providing only a bound on VMax and very little data on KM.
Rate at 2 mM Equivalent CLeff
blood cone. at 2 mM
(nmol/min/ (mM) (ml/min/
g tissue) g tissue)
hepatocytes :
liver cytosol:
kidney cells :
kidney cytosol:
liver cytosol:
kidney cytosol:
4.4-16
8.0-12
0.79-1.1 2.2
0.53-0.75 1.1-2.0
36-40
6.2-9.3
2.0
1.7-2.0
0.00036-0.
0.00027-0.
1.1-2.0
0.91-2.0 0.0031-0.0
0.0022-0.0079
0.0040-0.0072
0.00049
00068
0.018-0.036
# In most cases, rates were increased over the same sex/species at 1 mM,
# indicating VMax has not yet been reached. The values between cells
# and cytosol are more much consistent that in the human data.
# These data therefore put a lower bound on VMax and a lower bound
# on CLC. To account for in vitro-in vivo uncertainty, the lower
# bound of the prior distribution is set 100-fold below the central
# estimate of the measurements here. In addition, Green et al.
# (1997) found values 100-fold smaller than Lash et al. 1995, 1998.
# Therefore diffuse prior distributions set to le-2-le4.
# Rat liver: Bound on VMax of 4.4-16, with GM of 8.4. Converting to
# mg/hr/kg tissue (* 131.4 ng/nmol * 60 min/hr * Ie3 g/kg / Ie6 mg/ng)
# gives a central estimate of 66. mg/hr/kg tissue. Bound on CL of
# 0.0022-0.0079, with GM of 0.0042. Converting to 1/hr/kg tissue
# (* 60 min/hr) gives 0.25 1/hr/kg tissue.
# Rat kidney: Bound on VMax of 0.53-1.1, with GM of 0.76. Converting
# to mg/hr/kg tissue gives a central estimate of 6.0 mg/hr/kg.
# Bound on CL of 0.00027-0.00068, with GM of 0.00043. Converting
# tol/hr/kg tissue gives 0.026 1/hr/kg tissue.
# Mouse liver: Bound on VMax of 36-40, with GM of 38. Converting
# to mg/hr/kg tissue gives a central estimate of 300. mg/hr/kg.
# Bound on CL of 0.018-0.036, with GM of 0.025. Converting
# tol/hr/kg tissue gives 1.53 1/hr/kg tissue.
# Mouse kidney: Bound on VMax of 6.2-9.3, with GM of 7.6. Converting
# to mg/hr/kg tissue gives a central estimate of 60. mg/hr/kg.
# Bound on CL of 0.0031-0.0102, with GM of 0.0056. Converting
# tol/hr/kg tissue gives 0.34 1/hr/kg tissue.
VMaxDCVG = VLiv*(Species == 3 ? (300.*exp(InVMaxDCVGC))
(66.*exp(InVMaxDCVGC)) : (2.9*19.*exp(InClDCVGC+lnKMDCVGC))));
KMDCVG = (Species == 3 ? (VMaxDCVG/(VLiv*1.53*exp(InClDCVGC)))
2 ? (VMaxDCVG/(VLiv*0.25*exp(InClDCVGC))) : 2.9* exp(InKMDCVGC)));
VMaxKidDCVG = VKid*(Species == 3 ? (60.*exp(InVMaxKidDCVGC)) : (Species
2 ? (6.0*exp(InVMaxKidDCVGC)) : (2.7*230.*exp(InClKidDCVGC+lnKMKidDCVGC))));
KMKidDCVG = (Species == 3 ? (VMaxKidDCVG/(VKid*0.34*exp(InClKidDCVGC))) :
(Species == 2 ? (VMaxKidDCVG/(VKid*0.Q26*exp(InClKidDCVGC))) :
2.7* exp(InKMKidDCVGC)));
# Scaled to liver VMax using data from Green et al. (1997)
# in microsomal preparations (nmol/min/mg protein) at -1 mM.
# For humans, used detection limit of 0.03
# Additional scaling by lung/liver weight ratio
# from Brown et al. Table 21 (mouse and rat) or
# ICRP Pub 89 Table 2.8 (Human female and male)
# Uncertainty - 3-fold truncated at 3 GSD
VMaxClara = exp(InVMaxLungLivC) * VMax *
(Species == 3 ? (1. 03/1. 87*0.7/5.5): (Species == 2 ?
(0.08/0.82*0.5/3.4) : (0.03/0.33*(Male == 0 ? (0.42/1.4) : (0.5/1.8)))));
KMClara = exp(InKMClara);
# Fraction of Respiratory Metabolism that goes to system circulation
# (translocated to the liver)
FracLungSys = exp(InFracLungSysC)/(1 + exp(InFracLungSysC));
# TCOH Metabolism Constants (mg/hr)
# No in vitro data. So use diffuse priors of
# le-4 to Ie4 mg/hr/kg"0.75 for VMax
# (4e-5 to 4000 mg/hr for rat),
# le-4 to Ie4 mg/1 for KM,
# and le-5 to Ie3 l/hr/kg^O.75 for Cl
# (2e-4 to 2.4e4 1/hr for human)
VMaxTCOH = BW75*
(Species == 3 ? (exp(InVMaxTCOHC)) : (Species == 2 ?
(exp(lnVMaxTCOHC)) : (exp(InClTCOHC+lnKMTCOH))));
KMTCOH = exp(lnKMTCOH);
VMaxGluc = BW75*
(Species == 3 ? (exp(InVMaxGlucC)) : (Species == 2 ?
(exp(InVMaxGlucC)) : (exp(InClGlucC+lnKMGluc))));
KMGluc = exp(lnKMGluc);
# No in vitro data. So use diffuse priors of
# le-5 to Ie3 kg"0.25/hr (3.5e-6/hr to 3.5e2/hr for human)
kMetTCOH = exp(InkMetTCOHC) / BW25;
# TCA kinetic parameters
# Central estimate based on GFR clearance per unit body weight
# 10.0,8.7,1.8 ml/min/kg for mouse, rat, human
# (= 0.6, 0.522, 0.108 1/hr/kg) from Lin 1995.
# = CL_GFR / BW (BW=0.02 for mouse, 0.265 for rat, 70 for human)
# kUrn = CL_GFR / VPlas
# Diffuse prior with uncertainty of up,down 100-fold
kUrnTCA = exp(InkUrnTCAC) * BW / VPlas *
(Species == 3 ? 0.6 : (Species == 2 ? 0.522 : 0.108));
# No in vitro data. So use diffuse priors of
# le-4 to Ie2 /hr/kg"0.25 (0.3/hr to 35/hr for human)
kMetTCA = exp(InkMetTCAC) / BW25;
# TCOG kinetic parameters
# No in vitro data. So use diffuse priors of
# le-4 to Ie2 /hr/kg"0.25 (0.3/hr to 35/hr for human)
kBile = exp(lnkBileC) / BW25;
kEHR = exp(lnkEHRC) / BW25;
# Central estimate based on GFR clearance per unit body weight
-------
o
to
o a"
vo £3"
§
Co
a
1
>1
\o <*>
I
# 10.0,8.7,1.8 ml/min/kg for mouse, rat, human
# (= 0.6, 0.522, 0.108 1/hr/kg) from Lin 1995.
# = CL_GFR / BW (BW=0.02 for mouse, 0.265 for rat, 70 for human)
# kUrn = CL_GFR / VBld
# Diffuse prior with Uncertainty of up,down 1000-fold
kUrnTCOG = exp(InkUrnTCOGC) * BW / (VBodTCOH * PBodTCOG) *
(Species == 3 ? 0.6 : (Species == 2 ? 0.522 : 0.108));
# DCVG Kinetics (/hr)
# Fraction of renal TCE GSH conj. "directly" to DCVC via "first pass"
# exp(InFracOtherCC) = ratio of direct/non-direct
# Diffuse prior distribution: loguniform le-3 to Ie3
# FIXED in vl.2.3
# In ".in" files, set to 1, so that all kidney GSH conjugation
# is assumed to directly produce DCVC (model lacks identiflability
# otherwise).
FracKidDCVC = exp(InFracKidDCVCC)/(1 + exp(InFracKidDCVCC));
# No in vitro data. So use diffuse priors of
# le-4 to Ie2 /hr/kg"0.25 (0.3/hr to 35/hr for human)
kDCVG = exp(lnkDCVGC) / BW25;
# DCVC Kinetics in Kidney (/hr)
# No in vitro data. So use diffuse priors of
# le-4 to Ie2 /hr/kg"0.25 (0.3/hr to 35/hr for human)
kNAT = exp(lnkNATC) / BW25;
kKidBioact = exp(InkKidBioactC) / BW25;
# CC data initialization
Rodents = (CC > 0 ? NRodents : 0.0); # Closed chamber simulation
VCh = (CC > 0 ? VChC - (Rodents * BW) : 1.0);
# Calculate net chamber volume
kLoss = (CC > 0 ? exp(lnkLossC) : 0.0);
Other State Variables and Global Parameters
QC
VPR
DResptmp
QPsamp
QFatCtmp
QGutCtmp
QLivCtmp
QSlwCtmp
QKidCtmp
FracPlas
Temporary variables used:
none
Temporary variables assigned:
QP
DResp
QCnow
QFat
QGut
QLiv
QSlw
QKid
QGutLiv
QRap
QCPlas
QBodPlas
QGutLivPlas
Notes:
# QP uses QPmeas if value is > 0, otherwise uses sampled value
QP = (QPmeas > 0 ? QPmeas : QPsamp);
DResp = DResptmp * QP;
O §
o I
^—^ ""-k
of
^H c^
o ^
O £?
Initial amount in chamber
State Variables with dynamics:
v , \—*
^^ ^ # none
-i # Input Variables :
[-[-] # QPmeas
# These done here in dynamics in case QCnow changes
# Blood Flows to Tissues (L/hr)
QFat = (QFatCtmp) * QCnow; #
QGut = (QGutCtmp) * QCnow; #
QLiv = (QLivCtmp) * QCnow; #
QSlw = (QSlwCtmp) * QCnow; #
QKid = (QKidCtmp) * QCnow; #
QGutLiv = QGut + QLiv; #
QRap = QCnow - QFat - QGut - QLiv - QSlw - QKid;
QRapCtmp = QRap/QCnow; #(vrisk)
QBod = QCnow - QGutLiv;
-------
QGutLivPlas = FracPlas * QGutLiv;
State Variables with dynamics:
AS torn
ADuod
AStomTCA
AStomTCOH
Input Variables:
IVDose
PDose
Drink
Cone
IVDoseTCA
PODoseTCA
IVDoseTCOH
PODoseTCOH
Other State Variables and Global Parameters:
ACh
CC
MWTCE
BW
TChng
kAS
kTSD
kAD
kTD
kASTCA
kASTCOH
Temporary variables used:
none
Temporary variables assigned:
klV - rate into CVen
klA - rate into CArt
kPV - rate into portal vein
kStom - rate into stomach
kDrink - incorporated into RAO
RAO - rate into gut (oral absorption - both gavage and drinking water)
CInh - inhalation exposure concentration
klVTCA - rate into blood
kStomTCA - rate into stomach
kPOTCA - rate into liver (oral absorption)
klVTCOH - rate into blood
kStomTCOH - rate into stomach
kPOTCOH - rate into liver (oral absorption)
Notes:
For oral dosing, using "Spikes" for instantaneous inputs
Inhalation Concentration (mg/L)
CInh uses Cone when open chamber (CC=0) and
ACh/VCh when closed chamber COO.
BW) / TChng;# IV infusion rate (mg/hr)
# (IVDose constant for duration TChng)
klA = (lADose * BW) / TChng; # IA infusion rate (mg/hr)
kPV = (PVDose * BW) / TChng; # PV infusion rate (mg/hr)
kStom = (PDose * BW) / TChng;# PO dose rate (into stomach) (mg/hr)
# Amount of TCE in duodenum -- for oral dosing only (mg)
dt(ADuod) = (kTSD * AStom) - (kAD + kTD) * ADuod;
# Rate of absorption from drinking water
kDrink = (Drink * BW) / 24.0; #Ingestion rate via drinking water (mg/hr)
# Total rate of absorption including gavage and drinking water
RAO = kDrink + (kAS * AStom) + (kAD * ADuod);
## Inhalation route
CInh = (CC > 0 ? ACh/VCh : Conc*MWTCE/24450.0); t in mg/1
tttt TCA Dosing
klVTCA = (IVDoseTCA * BW) / TChng; t TCA IV infusion rate (mg/hr)
kStomTCA = (PODoseTCA * BW) / TChng; # TCA PO dose rate into stomach
dt(AStomTCA) = kStomTCA - AStomTCA * kASTCA;
kPOTCA = AStomTCA * kASTCA; t TCA oral absorption rate (mg/hr)
tttt TCOH Dosing
klVTCOH = (IVDoseTCOH * BW) / TChng;#TCOH IV infusion rate (mg/hr)
kStomTCOH = (PODoseTCOH * BW) / TChng; t TCOH PO dose rate into stomach
dt(AStomTCOH) = kStomTCOH - AStomTCOH * kASTCOH;
kPOTCOH = AStomTCOH * kASTCOH;! TCOH oral absorption rate (mg/hr)
ARap,
ASlw,
AFat,
AGut,
ALiv,
AInhResp,
AResp,
AExhResp,
AKid,
ABld,
ACh,
'ariables :
# Amount in rapidly perfused tissues
# Amount in slowly perfused tissues
# Amount in fat
# Amount in gut
# Amount in liver
-------
H TO"
O 5?
VRap
PRap
PSlw
VFat
PFat
VGut
PGut
VLiv
PLiv
VRespLum
VRespEff
FracLungSys
VKid
PKid
VBld
VMaxClara
KMClara
PB
Rodents
VCh
kLoss
VMax
KM
VMaxDCVG
KMDCVG
VMaxKidDCVG
KMKidDCVG
Temporary variables used:
QM
QFat
QGutLiv
QSlw
QRap
QKid
kiv
QCnow
CInh
QP
RAO
Temporary variables assigned:
QM
CRap
CSlw
CFat
CGut
CLiv
CInhResp
CResp
CExhResp
ExhFactor
CMixExh
CKid
CVRap
CVSlw
CVFat
CVGut
CVLiv
CVTB
CVKid
CVen
RAMetLng
CArt_tmp
CArt
CAlv
RAMetLivl
RAMetLiv2
RAMetKid
# Tissue Concentrations (mg/L)
CRap = ARap/VRap;
CSlw = ASlw/VSlw;
CFat = AFat/VFat;
CGut = AGut/VGut;
CLiv = ALiv/VLiv;
CKid = AKid/VKid;
# Venous Concentrations (mg/L)
CVRap = CRap / PRap
CVSlw = CSlw / PSlw
CVFat = CFat / PFat
CVGut = CGut / PGut
CVLiv = CLiv / PLiv
CVKid = CKid / PKid
# Concentration of TCE in mixed venous blood (mg/L)
CVen = ABld/VBld;
# Dynamics for blood
dt(ABld) = (QFat^CVFat + QGutLiv*CVLiv + QSlw*CVSlw +
QRap*CVRap + QKid*CVKid + klV) - CVen * QCn<
QM = QP/0.7; # Minute-volume
CInhResp = AInhResp/VRespLum;
CResp = AResp/VRespEff;
CExhResp = AExhResp/VRespLum;
dt(AInhResp) = (QM*CInh + DResp*(CResp-CInhResp) - QM*CInhResp);
RAMetLng = VMaxClara * CResp/(KMClara + CResp);
dt(AResp) = (DResp*(CInhResp + CExhResp - 2*CResp) - RAMetLng);
CArt_tmp = (QCnow*CVen + QP*CInhResp)/(QCnow + (QP/PB));
dt(AExhResp) = (QM*(CInhResp-CExhResp) + QP*(CArt_tmp/PB-CInhResp) +
DResp*(CResp-CExhResp));
CMixExh = (CExhResp > 0 ? CExhResp : le-15); # mixed exhaled breath
-------
# Concentration in alveolar air (mg/L)
# Correction factor for exhaled air to account for
# absorption/desorption/metabolism in respiratory tissue
t = 1 if DResp = 0
ExhFactor_den = (QP * CArt_tmp / PB + (QM-QP)*CInhResp);
ExhFactor = (ExhFactor_den > 0) ? (
QM * CMixExh / ExhFactor_den) : 1;
# End-exhaled breath (corrected for absorption/
# desorption/metabolism in respiratory tissue)
CAlv = CArt_tmp / PB * ExhFactor;
# Concentration in arterial blood entering circulation (mg/L)
CArt = CArt_tmp + klA/QCnow; # add inter-arterial dose
dt(ARap) = QRap * ,
# Amount of TCE in slowly perfused tissues
dt(ASlw) = QSlw * (CArt - CVSlw);
# Amount of TCE in fat tissue (mg)
dt(AFat) = QFat*(CArt - CVFat);
ount of TCE in gut compartment (mg)
#**** Liver *******************************************************************
t Rate of TCE oxidation by P450 to TCA, TCOH, and other (DCA) in liver (mg/hr)
RAMetLivl = (VMax * CVLiv) / (KM + CVLiv);
# Rate of TCE metabolized to DCVG in liver (mg)
RAMetLiv2 = (VMaxDCVG * CVLiv) / (KMDCVG + CVLiv);
# Dynamics for amount of TCE in liver (mg)
dt(ALiv) = (QLiv * (CArt - CVLiv)) + (QGut * (CVGut - CVLiv))
- RAMetLivl - RAMetLiv2 + kPV; # added PV dose
#* * * * Kidney ******************************************************************
# Rate of TCE metabolized to DCVG in kidney (mg) #
RAMetKid = (VMaxKidDCVG * CVKid) / (KMKidDCVG + CVKid);
# Amount of TCE in kidney compartment (mg)
dt(AKid) = (QKid * (CArt - CVKid)) - RAMetKid;
State Variables with dynamics:
ABodTCOH
ALivTCOH
Input Variables:
none
Other State Variables and Global Parameters:
ABileTCOG
t kEHR
t VBodTCOH
# PBodTCOH
# VLiv
t PLivTCOH
t VMaxTCOH
# KMTCOH
# VMaxGluc
# KMGluc
t kMetTCOH - hepatic metabolism of TCOH (e.g., to DCA)
t FracOther
t FracTCA
# StochTCOHTCE
# StochTCOHGluc
# FracLungSys
# Temporary variables used:
# QBod
# QGutLiv
# QCnow
t kPOTCOH
t RAMetLivl
# RAMetLng
# Temporary variables assigned:
# CVBodTCOH
t CVLivTCOH
t CTCOH
# RAMetTCOHTCA
# RAMetTCOHGluc
# RAMetTCOH
t RARecircTCOG
#**** Blood (venous=arterial) *********************************************
# Venous Concentrations (mg/L)
CVBodTCOH = ABodTCOH / VBodTCOH / PBodTCOH;
CVLivTCOH = ALivTCOH / VLiv / PLivTCOH;
CTCOH = (QBod * CVBodTCOH + QGutLiv * CVLivTCOH + klVTCOH)/QCnow;
# Rate of oxidation of TCOH to TCA (mg/hr)
RAMetTCOHTCA = (VMaxTCOH * CVLivTCOH) / (KMTCOH + CVLivTCOH);
# Amount of glucuronidation to TCOG (mg/hr)
RAMetTCOHGluc = (VMaxGluc * CVLivTCOH) / (KMGluc + CVLivTCOH);
# Amount of TCOH metabolized to other (e.g., DCA)
RAMetTCOH = kMetTCOH * ALivTCOH;
# Amount of TCOH-Gluc recirculated (mg)
RARecircTCOG = kEHR * ABileTCOG;
# Amount of TCOH in liver (mg)
-------
dt(ALivTCOH) = kPOTCOH + QGutLiv * (CTCOH - CVLivTCOH)
- RAMetTCOH - RAMetTCOHTCA - RAMetTCOHGluc
+ ((1.0 - FracOther - FracTCA) * StochTCOHTCE
(RAMetLivl + FracLungSys*RAMetLng))
+ (StochTCOHGluc * RARecircTCOG);
O 5?
State Variables with dynamics:
APlasTCA
ABodTCA
ALivTCA
AUrnTCA
AUrnTCA_sat
AUrnTCA_collect
Input Variables:
TCAUrnSat
UrnMissing
Other State Variables and Global Parameters:
VPlas
MWTCA
kDissoc
BMax
kMetTCA — hepatic metabolism of TCA (e.g., to DCA)
VBod
PBodTCA
PLivTCA
kUrnTCA
FracTCA
StochTCATCE
StochTCATCOH
FracLungSys
Temporary variables used:
klVTCA
kPOTCA
QBodPlas
QGutLivPlas
QCPlas
RAMetLivl
RAMetTCOHTCA
RAMetLng
Temporary variables assigned:
CPlasTCA
CPLasTCAMole
a, b, c
CPlasTCAFreeMole
CPlasTCAFree
APlasTCAFree
CPlasTCABnd
CBodTCAFree
CLivTCAFree
CBodTCA
CLivTCA
CVBodTCA
CVLivTCA
RUrnTCA
RAMetTCA
Concentration of TCA in plasma (umoles/L)
CPlasTCA = (APlasTCA<1.0e-15 ? l.Oe-15 : APlasTCA/VPlas);
Concentration of free TCA in plasma in (umoles/L)
CPlasTCAMole = (CPlasTCA / MWTCA) * 1000.0;
a = kDissoc+BMax-CPlasTCAMole;
b = 4.0*kDissoc*CPlasTCAMole;
c = (b < 0.01*a*a ? b/2.0/a : sgrt(a*a+b)-a);
CPlasTCAFreeMole = 0.5*c;
Concentration of free TCA in plasma (mg/L)
CPlasTCAFree = (CPlasTCAFreeMole * MWTCA) / 1000.0;
APlasTCAFree = CPlasTCAFree * VPlas;
Concentration of bound TCA in plasma (mg/L)
CPlasTCABnd = (CPlasTCA
-------
o
to
o a"
vo £3'
§
a
TO'
to
I
O
TO
o §
O g
"r-^ ""-k
O 3"
H TO"
O k-
w i
o ~r
o^
H
W
State Variables with dynamics:
ABodTCOG
ALivTCOG
ABileTCOG
AUrnTCOG
AUrnTCOG_sat
AUrnTCOG_collect
Input Variables:
TCOGUrnSat
UrnMissing
Other State Variables and Global Parameters:
VBodTCOH
VLiv
PBodTCOG
PLivTCOG
kUrnTCOG
kBile
StochGlucTCOH
Temporary variables used:
QBod
QGutLiv
QCnow
RAMetTCOHGluc
RARecircTCOG
Temporary variables assigned:
CVBodTCOG
CVLivTCOG
CTCOG
RUrnTCOG
RBileTCOG
Notes:
Venous Concentrations (mg/L)
CVBodTCOG = ABodTCOG / VBodTCOH / PBodTCOG;
CVLivTCOG = ALivTCOG / VLiv / PLivTCOG;
CTCOG = (QBod * CVBodTCOG + QGutLiv * CVLivTCOG)/QCnow;
* * * * Body *******************************************************
Amount of TCOG in body
RUrnTCOG = kUrnTCOG * ABodTCOG;
dt(ABodTCOG) = QBod * (CTCOG - CVBodTCOG) - RUrnTCOG;
RUrnTCOGTCOH = RUrnTCOG*StochTCOHGluc; #(vrisk)
Amount of TCOG in liver (mg)
RBileTCOG = kBile * ALivTCOG;
dt(ALivTCOG) = QGutLiv * (CTCOG - CVLivTCOG)
+ (StochGlucTCOH * RAMetTCOHGluc) - RBileTCOG;
Amount of TCOH-Gluc excreted in urine (mg)
dt(AUrnTCOG) = RUrnTCOG;
dt(AUrnTCOG_sat) = TCOGUrnSat*(1-UrnMissing)*RUrnTCOG;
# Saturated, but not missing collection times
dt(AUrnTCOG_collect) = (1-TCOGUrnSat)*(1-UrnMissing)^RUrnTCOG;
# Not saturated and not missing collection times
State Variables with dynamics:
ADCVGmol
Input Variables:
none
Other State Variables and Global Parameters:
kDCVG
FracKidDCVC # Fraction of kidney DCVG going to DCVC in first pass
VDCVG
Temporary variables used:
RAMetLiv2
RAMetKid
Temporary variables assigned:
RAMetDCVGmol
CDCVGmol
Notes:
Assume negligible GGT activity in liver as compared to kidney,
supported by in vitro data on GGT (even accounting for 5x
greater liver mass relative to kidney mass), as well as lack
of DCVC detected in blood.
"FracKidDCVC" Needed to account for "first pass" in
kidney (TCE->DCVG->DCVC without systemic circulation of DCVG).
Rate of metabolism of DCVG to DCVC
RAMetDCVGmol = kDCVG * ADCVGmol;
Dynamics for DCVG in blood
dt(ADCVGmol) = (RAMetLiv2 + RAMetKid*(1-FracKidDCVC)) / MWTCE
- RAMetDCVGmol;
Concentration of DCVG in blood (in mmoles/1)
CDCVGmol = ADCVGmol / VDCVG;
# State Variables with dynamics:
# ADCVC
# AUrnNDCVC
# Input Variables:
-------
to
'
I
I
§
Co
1
TO'
> s
I
o a.
o §
H I
O
H
W
none
Other State Variables and Global Parameters:
MWDCVC
FracKidDCVC
StochDCVCTCE
kNAT
kKidBioact
StochN
Temporary variables used:
RAMetDCVGmol
RAMetKid
Temporary variables assigned:
RAUrnDCVC
Notes:
Cannot detect DCVC in blood, so assume all is locally generated
and excreted or bioactivated in kidney.
Amount of DCVC in kidney (mg)
dt(ADCVC) = RAMetDCVGmol * MWDCVC
+ RAMetKid * FracKidDCVC * StochDCVCTCE
- ((kNAT + kKidBioact) * ADCVC);
Rate of NAcDCVC excretion into urine (mg)
RAUrnDCVC = kNAT * ADCVC;
Dynamics for amount of N Acetyl DCVC excreted (mg)
dt(AUrnNDCVC) = StochN * RAUrnDCVC;
RUrnNDCVC = StochN * RAUrnDCVC; #(vrisk)
*** Total Mass Balance
# Total intake from inhalation (mg)
RInhDose = QM * CInh;
dt(InhDose) = RInhDose;
# Amount of TCE absorbed by non-inhalation routes (mg)
dt(AO) = RAO + klV + klA + kPV; #(vrisk)
# Total dose
TotDose = InhDose + AO; #(vrisk)
# Total in tissues
TotTissue = #(vrisk)
ARap + ASlw + AFat + AGut + ALiv + AKid + ABld + #(vrisk)
AInhResp + AResp + AExhResp; #(vrisk)
# Total metabolized
dt(AMetLng) = RAMetLng; #(vrisk)
dt(AMetLivl) = RAMetLivl; #(vrisk)
dt(AMetLiv2) = RAMetLiv2; #(vrisk)
dt(AMetKid) = RAMetKid; #(vrisk)
ATotMetLiv = AMetLivl + AMetLiv2; #(vrisk)
TotMetab = AMetLng + ATotMetLiv + AMetKid; #(vrisk)
AMetLivOther = AMetLivl * FracOther; #(vrisk)
AMetGSH = AMetLiv2 + AMetKid; #(vrisk)
# Amount of TCE excreted in feces (mg)
RAExc = kTD * ADuod; #(vrisk)
dt(AExc) = RAExc; #(vrisk)
# Amount exhaled (mg)
RAExh = QM * CMixExh;
dt(AExh) = RAExh;
# Mass balance
TCEDiff = TotDose - TotTissue - TotMetab; #(vrisk)
MassBalTCE = TCEDiff - AExc - AExh; #(vrisk)
# Total production/intake of TCOH
dt(ARecircTCOG) = RARecircTCOG; #(vrisk)
dt(AOTCOH) = kPOTCOH + klVTCOH; #(vrisk)
TotTCOHIn = AOTCOH + ((1.0 - FracOther - FracTCA) * #(vrisk)
StochTCOHTCE * (AMetLivl + FracLungSys*AMetLng)) + #(vrisk)
(StochTCOHGluc * ARecircTCOG); #(vrisk)
TotTCOHDose = AOTCOH + ((1.0 - FracOther - FracTCA) * #(vrisk)
StochTCOHTCE * (AMetLivl + FracLungSys^AMetLng)); #(vrisk)
# Total in tissues
TotTissueTCOH = ABodTCOH + ALivTCOH; #(vrisk)
# Total metabolism of TCOH
dt(AMetTCOHTCA) = RAMetTCOHTCA; #(vrisk)
dt(AMetTCOHGluc) = RAMetTCOHGluc; #(vrisk)
dt(AMetTCOHOther) = RAMetTCOH; #(vrisk)
TotMetabTCOH = AMetTCOHTCA + AMetTCOHGluc + AMetTCOHOther; #(vrisk)
# Mass balance
MassBalTCOH = TotTCOHIn - TotTissueTCOH - TotMetabTCOH; #(vrisk)
Total production/intake of TCA
dt(AOTCA) = kPOTCA + klVTCA; #(vrisk)
TotTCAIn = AOTCA + (FracTCA*StochTCATCE*(AMetLivl + #(vrisk)
FracLungSys*AMetLng)) + (StochTCATCOH*AMetTCOHTCA); #(vrisk)
Total in tissues
TotTissueTCA = APlasTCA + ABodTCA + ALivTCA; #(vrisk)
Total metabolism of TCA
dt(AMetTCA) = RAMetTCA; #(vrisk)
Mass balance
TCADiff = TotTCAIn - TotTissueTCA - AMetTCA; #(vrisk)
MassBalTCA = TCADiff - AUrnTCA; #(vrisk)
**** Mass Balance for TCOG ***************************************************
Total production of TCOG
TotTCOGIn = StochGlucTCOH * AMetTCOHGluc; #(vrisk)
Total in tissues
TotTissueTCOG = ABodTCOG + ALivTCOG + ABileTCOG; #(vrisk)
Mass balance
MassBalTCOG = TotTCOGIn - TotTissueTCOG - #(vrisk)
ARecircTCOG - AUrnTCOG; #(vrisk)
**** Mass Balance for DCVG ***************************************************
Total production of DCVG
dt(ADCVGIn) = (RAMetLiv2 + RAMetKid*(1-FracKidDCVC)) / MWTCE; #(vrisk)
Metabolism of DCVG
dt(AMetDCVG) = RAMetDCVGmol; #(vrisk)
-------
H £
#**** Mass Balance for DCVC ***************************************************
t Total production of DCVC
dt(ADCVCIn) = RAMetDCVGmol * MWDCVC #(vrisk)
+ RAMetKid * FracKidDCVC * StochDCVCTCE;#(vrisk)
# Bioactivation of DCVC
dt(ABioactDCVC) = (kKidBioact * ADCVC);#(vrisk)
# Mass balance
AUrnNDCVCequiv = AUrnNDCVC/StochN;
MassBalDCVC = ADCVCIn - ADCVC - ABioactDCVC - AUrnNDCVCequiv;#(vrisk)
#**** AUCs in mg-hr/L unless otherwise noted ***************
#AUC of TCE in arterial blood
dt(AUCCBld) = CArt; #(vrisk)
#AUC of TCE in liver
dt(AUCCLiv) = CLiv; f(vrisk)
#AUC of TCE in kidney
dt(AUCCKid) = CKid; #(vrisk)
#AUC of TCE in rapidly perfused
dt(AUCCRap) = CRap; #(vrisk)
#AUC of TCOH in blood
dt(AUCCTCOH) = CTCOH; f(vrisk)
#AUC of TCOH in body
dt(AUCCBodTCOH) = ABodTCOH / VBodTCOH; #(vrisk)
#AUC of free TCA in the plasma (mg/L * hr)
dt(AUCPlasTCAFree) = CPlasTCAFree; #(vrisk)
#AUC of total TCA in plasma (mg/L * hr)
dt(AUCPlasTCA) = CPlasTCA; #(vrisk)
#AUC of TCA in liver (mg/L * hr)
dt(AUCLivTCA) = CLivTCA; #(vrisk)
#AUC of total TCOH (free+gluc) in TCOH-equiv in blood (mg/L
dt(AUCTotCTCOH) = CTCOH + CTCOGTCOH; #(vrisk)
#AUC of DCVG in blood (mmol/L * hr) — NOTE moles, not mg
dt(AUCCDCVG) = CDCVGmol; #(vrisk)
};
End of Dynamics
l.Oe-15 : CFat);
RetDose = ((InhDose-AExhExp) > 0 ? (InhDose - AExhExp) : le-15);
CAlvPPM = (CAlv < l.Oe-15 ? l.Oe-15 : CAlv * (24450.0 / MWTCE));
CInhPPM = (ACh< l.Oe-15 ? l.Oe-15 : ACh/VCh*24450.0/MWTCE);
# CInhPPM Only used for CC inhalation
CArt = (CArt < l.Oe-15 ? l.Oe-15 : CArt);
CVen = (CVen < l.Oe-15 ? l.Oe-15
CBldMix = (CArt+CVen)/2;
CFat = (CFat < l.Oe-15
CGut = (CGut < l.Oe-15
CRap = (CRap < l.Oe-15
CSlw = (CSlw < l.Oe-15
CHrt = CRap;
CKid = (CKid < l.Oe-15
CLiv = (CLiv < l.Oe-15
CLung = CRap;
CMus = (CSlw < l.Oe-15
CSpl = CRap;
CBrn = CRap;
zAExh = (AExh < l.Oe-15
zAExhpost = ((AExh - AExhExp) < l.Oe-15 ? l.Oe-15 : AExh - AExhExp);
CTCOH = (CTCOH < l.Oe-15 ? l.Oe-15 : CTCOH);
CBodTCOH = (ABodTCOH < l.Oe-15 ? l.Oe-15 : ABodTCOH/VBodTCOH);
CKidTCOH = CBodTCOH;
CLivTCOH = (ALivTCOH < l.Oe-15
CLungTCOH = CBodTCOH;
CPlasTCA = (CPlasTCA < l.Oe-15 ? l.Oe-15 : CPlasTCA);
CBldTCA = CPlasTCA*TCAPlas;
CBodTCA = (CBodTCA < l.Oe-15
CLivTCA = (CLivTCA < l.Oe-15
CKidTCA = CBodTCA;
CLungTCA = CBodTCA;
zAUrnTCA = (AUrnTCA < l.Oe-15 ? l.Oe-15 : AUrnTCA);
zAUrnTCA_sat = (AUrnTCA_sat < l.Oe-15 ? l.Oe-15 : AUrnTCA_sat);
zAUrnTCA_collect = (AUrnTCA_collect < l.Oe-15 ? l.Oe-15 :
AUrnTCA_collect) ;
# TCOG
zABileTCOG = (ABileTCOG < l.Oe-15 ? l.Oe-15 : ABileTCOG);
# Concentrations are in TCOH-equivalents
CTCOG = (CTCOG < l.Oe-15 ? l.Oe-15 : CTCOG);
CTCOGTCOH = (CTCOG < l.Oe-15
CBodTCOGTCOH = (ABodTCOG < l.Oe-15 ?
StochTCOHGluc*ABodTCOG/VBodTCOH);
CKidTCOGTCOH = CBodTCOGTCOH;
CLivTCOGTCOH = (ALivTCOG < l.Oe-15 ?
StochTCOHGluc*ALivTCOG/VLiv);
CLungTCOGTCOH = CBodTCOGTCOH;
AUrnTCOGTCOH = (AUrnTCOG < l.Oe-15 ?
AUrnTCOGTCOH_sat = (AUrnTCOG_sat < l.Oe-1
StochTCOHGluc*AUrnTCOG_sat);
AUrnTCOGTCOH_collect = (AUrnTCOG_collect
StochTCOHGluc*AUrnTCOG collect);
-------
to
vo £3'
I
§
t Other
CDCVGmol = (CDCVGmol < l.Oe-15 ? l.Oe-15 : CDCVGmol);
CDCVGmolO = CDCVGmol; #(vl.2.3.2)
CDCVG_NDtmp = CDFNormal(3*(1-CDCVGmol/CDCVGmolLD));
# Assuming LD = 3*sigma_blank, Normally distributed
CDCVG_ND = ( CDCVG_NDtmp < 1.0 ? ( CDCVG_NDtmp >= le-100 ? -
log(CDCVG_NDtmp) : -log(le-100)) : le-100 );
#(vl.2.3.2)
zAUrnNDCVC =(AUrnNDCVC < l.Oe-15 ? l.Oe-15 : AUrnNDCVC);
AUrnTCTotMole = zAUrnTCA / MWTCA + AUrnTCOGTCOH / MWTCOH;
TotCTCOH = CTCOH + CTCOGTCOH;
TotCTCOHcomp = CTCOH + CTCOG; # ONLY FOR COMPARISON WITH HACK
ATCOG = ABodTCOG + ALivTCOG; # ONLY FOR COMPARISON WITH HACK
# Misc
CVenMole = CVen / MWTCE;
CPlasTCAMole = (CPlasTCAMole < l.Oe-15 ? l.Oe-15 : CPlasTCAMole)
CPlasTCAFreeMole = (CPlasTCAFreeMole < l.Oe-15 ? l.Oe-15 :
CPlasTCAFreeMole);
#**** Additional Dose Metrics ********************************************
TotTCAInBW = TotTCAIn/BW;#(vrisk)
Scaled by BW"3/4
TotMetabBW34 = TotMetab/BW75;#(vrisk)
AMetGSHBW34 = AMetGSH/BW75;#(vrisk)
TotDoseBW34 = TotDose/BW75;#(vrisk)
AMetLivlBW34 = AMetLivl/BW75;#(vrisk)
TotOxMetabBW34 = (AMetLng+AMetLivl)/BW75;#(vrisk)
Scaled by tissue volume
AMetLivlLiv = AMetLivl/VLiv; #(vrisk)
AMetLivOtherLiv = AMetLivOther/VLiv; #(vrisk)
AMetLngResp = AMetLng/VRespEfftmp; #(vrisk)
ABioactDCVCKid = ABioactDCVC/VKid;#(vrisk)
VFatCtmp = VFat/BW; t(vrisk)
VGutCtmp = VGut/BW; #(vrisk)
VLivCtmp = VLiv/BW; #(vrisk)
VRapCtmp = VRap/BW; #(vrisk)
VRespLumCtmp = VRespLum/BW; #(vrisk)
VRespEffCtmp = VRespEfftmp/BW; #(vrisk)
VKidCtmp = VKid/BW; #(vrisk)
VBldCtmp = VBld/BW; #(vrisk)
VSlwCtmp = VSlw/BW; #(vrisk)
VPlasCtmp = VPlas/BW; #(vrisk)
VBodCtmp = VBod/BW; #(vrisk)
VBodTCOHCtmp = VBodTCOH/BW; #(vrisk)
-------
1 A.8. REFERENCES
2 Abbas R, Seckel CS, MacMahon KL, Fisher JW. (1997). Determination of kinetic rate constants for chloral hydrate,
3 trichloroethanol, trichloroacetic acid, and dichloroacetic acid~a physiologically based modeling approach.
4 Toxicologist 36:32-33.
5 Abbas R, Fisher JW. (1997). physiologically based pharmacokinetic model for trichloroethylene and its metabolites,
6 chloral hydrate, trichloroacetate, dichloroacetate, trichloroethanol, and trichloroethanol glucuronide in B6C3F1
7 mice. ToxicolApplPharmacol. 147:15-30.
8 Andersen, M. E., Gargas, M. L., Clewell, H. I, 3rd, and Severyn, K. M. (1987). Quantitative evaluation of the
9 metabolic interactions between trichloroethylene and 1,1-dichloroethylene in vivo using gas uptake methods. Toxicol
10 Appl Pharmacol 89, 149-157.
11 Araki, S. (1978). The effects of water restriction and water loading on urinary excretion of lead, delta-
12 aminolevulinic acid and coproporphyrin. Br J Ind Med. 35(4):312-7.
13 Barter, Z. E., Bayliss, M. K., Beaune, P. H., Boobis, A. R., Carlile, D. J., Edwards, R. J., Houston, J. B., Lake, B. G.,
14 Lipscomb, J. C., Pelkonen, O. R., Tucker, G. T., and Rostami-Hodjegan, A. (2007). Scaling factors for the
15 extrapolation of in vivo metabolic drug clearance from in vitro data: reaching a consensus on values of human
16 microsomal protein and hepatocellularity per gram of liver. Curr DrugMetab 8, 33-45.
17 Barton, H. A., Creech, J. R., Godin, C. S., Randall, G. M., and Seckel, C. S. (1995). Chloroethylene mixtures:
18 pharmacokinetic modeling and in vitro metabolism of vinyl chloride, trichloroethylene, and trans-1,2-
19 dichloroethylene in rat. Toxicol Appl Pharmacol 130, 237-247.
20 Bartonicek, V. (1962). Metabolism and excretion of trichloroethylene after inhalation by human subjects. Br J
21 IndustMed. 19:134-141.
22 Bernauer, U., Birner, G., Dekant, W., and Henschler, D. (1996). Biotransformation of trichloroethene: dose-
23 dependent excretion of 2,2,2-trichloro-metabolites and mercapturic acids in rats and humans after inhalation. Arch
24 Toxicol 70, 338-346.
25 Birner G, Vamvakas S, Dekant W, Henschler D. (1993). Nephrotoxic and genotoxic N-acetyl-S-dichlorovinyl-L-
26 cysteine is a urinary metabolite after occupational 1,1,2-trichloroethene exposure in humans: implications for the
27 risk of trichloroethene exposure. Environ Health Perspect. 99:281-4.
28 Bloemen, L. J., Monster, A. C., Kezic, S., Commandeur, J. N., Veulemans, H., Vermeulen, N. P., and Wilmer, J. W.
29 (2001). Study on the cytochrome P-450- and glutathione-dependent biotransformation of trichloroethylene in
30 humans. IntArch Occup Environ Health 74, 102-108.
31 Bois FY. (2000a). Statistical analysis of Fisher et al. PBPK model of trichloroethylene kinetics. Environ Health
32 Perspect. 108 Suppl 2:275-82.
33 Bois FY. (2000b). Statistical analysis of Clewell et al. PBPK model of trichloroethylene kinetics. Environ Health
34 Perspect. 108 Suppl 2:307-16.
35 Bronley-DeLancey, A., McMillan, D. C., McMillan, J. M., Jollow, D. J., Mohr, L. C., and Hoel, D. G. (2006).
3 6 Application of cryopreserved human hepatocytes in trichloroethylene risk assessment: relative disposition of chloral
37 hydrate to trichloroacetate and trichloroethanol. Environ Health Perspect 114, 1237-1242.
3 8 Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, Beliles RP. (1997). Physiological parameter values for
39 physiologically based pharmacokinetic models. Toxicol Ind Health. 13:407-84.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-106 DRAFT—DO NOT CITE OR QUOTE
-------
1 Chiu, WA; Bois, RY. (2006) Revisiting the population toxicokinetics of tetrachloroethylene. Arch Toxicol
2 80(6):382-385.
3 Chiu, W. A., Micallef, S., Monster, A. C., and Bois, F. Y. (2007). Toxicokinetics of inhaled trichloroethylene and
4 tetrachloroethylene in humans at 1 ppm: empirical results and comparisons with previous studies. Toxicol Sci 95,
5 23-36.
6 Clewell HJ 3rd, Gentry PR, Covington TR, Gearhart JM. (2000). Development of a physiologically based
7 pharmacokinetic model of trichloroethylene and its metabolites for use in risk assessment. Environ Health Perspect.
8 108 Suppl 2:283-305.
9 Dallas, C. E., Gallo, J. M, Ramanathan, R., Muralidhara, S., and Bruckner, J. V. (1991). Physiological
10 pharmacokinetic modeling of inhaled trichloroethylene in rats. Toxicol Appl Pharmacol 110, 303-314.
11 Elfarra, A. A., Krause, R. J., Last, A. R., Lash, L. H., and Parker, J. C. (1998). Species- and sex-related differences
12 in metabolism of trichloroethylene to yield chloral and trichloroethanol in mouse, rat, and human liver microsomes.
13 Drug Me tab Dispos 26, 779-785.
14 Ertle T, HenschlerD, MtillerG, Spassowski M. (1972). Metabolism of trichloroethylene in man. I. The significance
15 of trichloroethanol in long-term exposure conditions. Arch Toxikol. 29:171-88.
16 Fernandez, J. G., Droz, P. O., Humbert, B. E., and Caperos, J. R. (1977). Trichloroethylene exposure. Simulation of
17 uptake, excretion, and metabolism using a mathematical model. Br JIndMed 34, 43-55.
18 Fiserova-Bergerova, V., Tichy, M., and Di Carlo, F. J. (1984). Effects of biosolubility on pulmonary uptake and
19 disposition of gases and vapors of lipophilic chemicals. DrugMetab Rev 15, 1033-1070.
20 Fisher JW. (2000). Physiologically based pharmacokinetic models for trichloroethylene and its oxidative
21 metabolites. Environ Health Perspect. 108 Suppl 2:265-73.
22 Fisher JW, Allen BC. (1993). Evaluating the risk of liver cancer in humans exposed to trichloroethylene using
23 physiological models. Risk Anal. 13(l):87-95.
24 Fisher, J. W., Whittaker, T. A., Taylor, D. H., Clewell, H. J., 3rd, and Andersen, M. E. (1989). Physiologically based
25 pharmacokinetic modeling of the pregnant rat: a multiroute exposure model for trichloroethylene and its metabolite,
26 trichloroacetic acid. Toxicol Appl Pharmacol 99, 395-414.
27 Fisher JW, Gargas ML, Allen BC, Andersen ME. (1991). Physiologically based pharmacokinetic modeling with
28 trichloroethylene and its metabolite, trichloroacetic acid, in the rat and mouse. Toxicol Appl Pharmacol.
29 109:183-95.
30 Fisher, J. W., Mahle, D., and Abbas, R. (1998). A human physiologically based pharmacokinetic model for
31 trichloroethylene and its metabolites, trichloroacetic acid and free trichloroethanol. Toxicol Appl Pharmacol 152,
32 339-359.
33 Gargas, M. L., Burgess, R. J., Voisard, D. E., Cason, G. H., and Andersen, M. E. (1989). Partition coefficients of
34 low-molecular-weight volatile chemicals in various liquids and tissues. Toxicol Appl Pharmacol 98, 87-99.
35 Gelfand, A.E., Smith, A.F.M. (1990). Sampling-based approaches to calculating marginal densities. J Am Stat
36 Assoc. 85:398-409.
37 Gelman A, Bois F, Jiang J. (1996). Physiological pharmacokinetic analysis using population modeling and
3 8 informative prior distributions. J Am Stat Assoc. 91:1400-1412.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-107 DRAFT—DO NOT CITE OR QUOTE
-------
1 Gelman A, Carlin JB, Stern HS, Rubin DB. (2004). Bayesian Data Analysis. New York: Chapman & Hall/CRC.
2 Gilks, W.R., Richardson, S., Spiegelhalter, D.J., eds. (1996) Markov Chain Monte Carlo in Practice. New York:
3 Chapman & Hall/CRC Press.
4 Green, T. (2003 a) The concentrations of trichloroacetic acid in blood following administration of trichloroacetic
5 acid in drinking water. Central Toxicology Laboratory, Alderley Park Macclesfield, Cheshire, UK.
6 Green, T. (2003b) The concentrations of trichloroacetic acid in blood following administration of trichloroacetic
7 acid in drinking water. Central Toxicology Laboratory, Alderley Park Macclesfield, Cheshire, UK.
8 Green, T., and Prout, M. S. (1985). Species differences in response to trichloroethylene. II. Biotransformation in rats
9 and mice. ToxicolAppl Pharmacol 79, 401-411.
10 Green, T., Mainwaring, G. W., and Foster, J. R. (1997). Trichloroethylene-induced mouse lung tumors: studies of
11 the mode of action and comparisons between species. Fundam Appl Toxicol 37, 125-130.
12 Greenberg, M. S., Burton, G. A., and Fisher, J. W. (1999). Physiologically based pharmacokinetic modeling of
13 inhaled trichloroethylene and its oxidative metabolites in B6C3F1 mice. ToxicolAppl Pharmacol 154, 264-278.
14 Hack CE, Chiu WA, Jay Zhao Q, Clewell HJ.(2006). Bayesian population analysis of a harmonized physiologically
15 based pharmacokinetic model of trichloroethylene and its metabolites. Regul Toxicol Pharmacol. 46:63-83.
16 Hejtmancik MR, Trela BA, Kurtz PJ, Persing RL, Ryan MJ, Yarrington JT, Chhabra RS. (2002). Comparative
17 gavage subchronic toxicity studies of o-chloroaniline and m-chloroaniline in F344 rats and B6C3F1 mice. Toxicol
18 Sci. 69:234-43.
19 Hissink, E. M., Bogaards, J. J. P., Freidg, A. P., Commandeur, J. N. M., Vermeulen, N. P. E., and van Bladeren, P. J.
20 (2002). The use of in vitro metabolic parameters and physiologically based pharmacokinetic (PBPK) modeling to
21 explore the risk assessment of trichloroethylene. Environmental Toxicology and Pharmacology 11, 259-271.
22 ICRP (International Commission on Radiological Protection). (2003) Basic anatomical and physiological data for
23 use in radiological protection: Reference values, 89. Oxford; New York: Pergamon Press.
24 Jakobson, I., Holmberg, B., and Ekner, A. (1986). Venous blood levels of inhaled trichloroethylene in female rats
25 and changes induced by interacting agents. Acta Pharmacol Toxicol (Copenh) 59, 135-143.
26 Kaneko, T., Wang, P. Y., and Sato, A. (1994). Enzymes induced by ethanol differently affect the pharmacokinetics
27 of trichloroethylene and 1,1,1-trichloroethane. Occup Environ Med 51, 113-119.
28 Keys, D. A., Bruckner, J. V., Muralidhara, S., and Fisher, J. W. (2003). Tissue dosimetry expansion and cross-
29 validation of rat and mouse physiologically based pharmacokinetic models for trichloroethylene. Toxicol Sci 76,
30 35-50.
31 Kim, S; Kim, D; Pollack, GM; et al. (2009) Pharmacokinetic analysis of trichloroethylene metabolism in male
32 B6C3F1 mice: Formation and disposition of trichloroacetic acid, dichloroacetic acid.
33 S-(1.2-dichlorovinyl)glutathioneandS-(1.2-dichlorovinyl)-L-cysteine. Toxicol Appl Pharmacol 238(l):90-99.
34 Epub2009May3.
35 Kimmerle, G., Eben, A. (1973a). Metabolism, excretion and toxicology of trichloroethylene after inhalation. I.
36 Experimental exposure on rats. Arch Toxikol 3Q:115-126.
37 Kimmerle, G., Eben, A. (1973b). Metabolism, excretion and toxicology of trichloroethylene after inhalation. II.
38 Experimental human exposure. Arch Toxikol 30:127-138.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-108 DRAFT—DO NOT CITE OR QUOTE
-------
1 Koizumi, A. (1989). Potential of physiologically based pharmacokinetics to amalgamate kinetic data of
2 trichloroethylene and tetrachloroethylene obtained in rats and man. Br JIndMed 46, 239-249.
3 Lapare S, Tardif R, Brodeur 1(1995). Effect of various exposure scenarios on the biological monitoring of
4 organicsolvents in alveolar air. II. 1,1,1-Trichloroethane and trichloroethylene. IntArch Occup Environ Health.
5 67:375-94.
6 Larson, J. L., and Bull, R. J. (1992a). Metabolism and lipoperoxidative activity of trichloroacetate and
7 dichloroacetate in rats and mice. ToxicolAppl Pharmacol 115, 268-277.
8 Larson, J. L., and Bull, R. J. (1992b). Species differences in the metabolism of trichloroethylene to the carcinogenic
9 metabolites trichloroacetate and dichloroacetate. ToxicolAppl Pharmacol 115, 278-285.
10 Lash, L. H., Tokarz, J. J., and Pegouske, D. M. (1995). Susceptibility of primary cultures of proximal tubular and
11 distal tubular cells from rat kidney to chemically induced toxicity. Toxicology 103, 85-103.
12 Lash, L. H., Visarius, T. M., Sail, J. M., Qian, W., and Tokarz, J. J. (1998). Cellular and subcellular heterogeneity of
13 glutathione metabolism and transport in rat kidney cells. Toxicology 130, 1-15.
14 Lash LH, Lipscomb JC, Putt DC, Parker JC. (1999a). Glutathione conjugation of trichloroethylene in human liver
15 and kidney: kinetics and individual variation. Drug Metab Dispos. 27:351-359.
16 Lash LH, Putt DA, Brashear WT, Abbas R, Parker JC, Fisher JW. (1999b). Identification of S-(l,2-
17 dichlorovinyl)glutathione in the blood of human volunteers exposed to trichloroethylene. J Toxicol Environ Health
18 .4.56:1-21.
19 Lash LH, Putt DA, Parker JC. (2006). Metabolism and tissue distribution of orally administered trichloroethylene in
20 male and female rats: identification of glutathione- and cytochrome P-450-derived metabolites in liver, kidney,
21 blood, and urine. J Toxicol Environ Health A. 69:1285-309.
22 Lee, K. M., Bruckner, J. V., Muralidhara, S., and Gallo, J. M. (1996). Characterization of presystemic elimination of
23 trichloroethylene and its nonlinear kinetics in rats. ToxicolAppl Pharmacol 139, 262-271.
24 Lee, K. M., Muralidhara, S., Schnellmann, R. G., and Bruckner, J. V. (2000). Contribution of direct solvent injury to
25 the dose-dependent kinetics of trichloroethylene: portal vein administration to rats. ToxicolAppl Pharmacol 164,
26 46-54.
27 LinJH. (1995). Species similarities and differences in pharmacokinetics. DrugMetab Dispos. 23:1008-21.
28 Lipscomb, J. C., Garrett, C. M., and Snawder, J. E. (1997). Cytochrome P450-dependent metabolism of
29 trichloroethylene: interindividual differences in humans. ToxicolAppl Pharmacol 142, 311-318.
30 Lipscomb, J. C., Garrett, C. M., and Snawder, J. E. (1998a). Use of Kinetic and Mechanistic Data in Species
31 Extrapolation of Bioactivation: Cytochrome-P450 Dependent Trichloroethylene Metabolism at Occupationally
32 Relevant Concentrations. J Occup Health 40, 110-117.
33 Lipscomb, J. C., Fisher, J. W., Confer, P. D., and Byczkowski, J. Z. (1998b). In vitro to in vivo extrapolation for
34 trichloroethylene metabolism in humans. ToxicolAppl Pharmacol 152, 376-387.
35 Liu, Y; Bartlett, MG; White, CA; et al. (2009) Presystemic elimination of trichloroethylene in rats following
36 environmentally relevant oral exposures. Drug Metab Dispos (10):1994-1998. Epub 2009 Jul 6.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-109 DRAFT—DO NOT CITE OR QUOTE
-------
1 Lumpkin, M. H., Bruckner, J. V., Campbell, J. L., Dallas, C. E., White, C. A., and Fisher, J. W. (2003). Plasma
2 binding of trichloroacetic acid in mice, rats, and humans under cancer bioassay and environmental exposure
3 conditions. DrugMetab Dispos 31, 1203-1207.
4 Mahle, DA; Godfrey, RJ; Buttler, GW; et al. (2001) Pharmacokinetics and metabolism of dichloroacetic acid and
5 trichloroacetic acid administered in drinking water in rats and mice. United States Air Force Research Laboratory,
6 Wright Patterson Air Force Base, OH. AFRL-HE-WP-TR-2001-0059.
7 Merdink, J. L., Gonzalez-Leon, A., Bull, R. J., and Schultz, I. R. (1998). The extent of dichloroacetate formation
8 from trichloroethylene, chloral hydrate, trichloroacetate, and trichloroethanol in B6C3F1 mice. Toxicol Sci 45,
9 33-41.
10 Merdink, J. L., Stenner, R. D., Stevens, D. K., Parker, J. C., and Bull, R. J. (1999). Effect of enterohepatic
11 circulation on the pharmacokinetics of chloral hydrate and its metabolites in F344 rats. J Toxicol Environ Health A
12 57,357-368.
13 Monster, A. C., Boersma, G., andDuba, W. C. (1976). Pharmacokinetics of trichloroethylene in volunteers,
14 influence of workload and exposure concentration. IntArch Occup Environ Health 38, 87-102.
15 Monster, A. C., Boersma, G., and Duba, W. C. (1979). Kinetics of trichloroethylene in repeated exposure of
16 volunteers. IntArch Occup Environ Health 42, 283-292.
17 Mtiller G, Spassowski M, Henschler D. (1972). Trichloroethylene exposure and trichloroethylene metabolites in
18 urine and blood. Arch. Toxikol. 29:335-340.
19 Muller, G., Spassovski, M., and Henschler, D. (1974). Metabolism of trichloroethylene in man. II. Pharmacokinetics
20 of metabolites. Arch Toxicol 32, 283-295.
21 Muller, G., Spassowski, M., and Henschler, D. (1975). Metabolism of trichloroethylene in man. III. Interaction of
22 trichloroethylene and ethanol. Arch Toxicol 33, 173-189.
23 Okino MS, Chiu WA, Evans MV, Power FW, Lipscomb JC, Tornero-Velez R, Dary CC, Blancato JN, Chen C .
24 2005. Suitability of Using In Vitro and Computationally Estimated Parameters in Simplified Pharmacokinetic
25 Models. Drug Metab Rev 2005; 37(suppl. 2): 162.
26 Paycok, Z. V., Powell, J.F. (1945). The excretion of sodium trichloroacetate. J Pharmacol Exper Therapeut. 85:
27 289-293.
28 Plummer M, Best N, Cowles K, Vines K. (2008). coda: Output analysis and diagnostics for MCMC. R package
29 version 0.13-3.
30 Price PS, Conolly RB, Chaisson CF, Gross EA, Young JS, Mathis ET, Tedder DR. (2003). Modeling interindividual
31 variation in physiological factors used in PBPK models of humans. Crit Rev Toxicol. 33:469-503.
32 Prout, M. S., Provan, W. M., and Green, T. (1985). Species differences in response to trichloroethylene. I.
33 Pharmacokinetics in rats and mice. Toxicol Appl Pharmacol 79, 389-400.
34 Rodriguez, CE; Mahle, DA; Gearhart, JM; et al. (2007) Predicting age-appropriate pharmacokinetics of six volatile
3 5 organic compounds in the rat utilizing physiologically based pharmacokinetic modeling. Toxicol Sci 98(1) :43-56.
3 6 Sarangapani R, Gentry PR, Covington TR, Teeguarden JG, Clewell HJ 3rd. (2003). Evaluation of the potential
3 7 impact of age- and gender-specific lung morphology and ventilation rate on the dosimetry of vapors. Inhal Toxicol.
38 15:987-1016.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-110 DRAFT—DO NOT CITE OR QUOTE
-------
1 Sato, A., Nakajima, T., Fujiwara, Y., and Murayama, N. (1977). A pharmacokinetic model to study the excretion of
2 trichloroethylene and its metabolites after an inhalation exposure. Br JIndMed 34, 56-63.
3 Sato, A., and Nakajima, T. (1979). Partition coefficients of some aromatic hydrocarbons and ketones in water, blood
4 and oil. Br J IndMed 36, 231-234.
5 Schultz, I. R., Merdink, J. L., Gonzalez-Leon, A., and Bull, R. J. (1999). Comparative toxicokinetics of chlorinated
6 and brominated haloacetates in F344 rats. ToxicolAppl Pharmacol 158, 103-114.
7 Simmons, J. E., Boyes, W. K., Bushnell, P. J., Raymer, J. H., Limsakun, T., McDonald, A., Sey, Y. M, and Evans,
8 M. V. (2002). A physiologically based pharmacokinetic model for trichloroethylene in the male long-Evans rat.
9 Toxicol Sci 69, 3-15.
10 Stenner, R. D., Merdink, J. L., Stevens, D. K., Springer, D. L., and Bull, R. J. (1997). Enterohepatic recirculation of
11 trichloroethanol glucuronide as a significant source of trichloroacetic acid. Metabolites of trichloroethylene. Drug
12 Metab Dispos 25, 529-535.
13 Stewart, R. D., Dodd, H. C., Gay, H. H., and Erley, D. S. (1970). Experimental human exposure to trichloroethylene.
14 Arch Environ Health 20, 64-71.
15 Sweeney, LM; Kirman, CR; Gargas, ML; et al. (2009) Contribution of trichloroacetic acid to liver tumors observed
16 inperchloroethylene (perc)-exposed mice. Toxicology 260(l-3):77-83. Epub 2009 Mar 24.
17 Templin, M. V., Parker, J. C., and Bull, R. J. (1993). Relative formation of dichloroacetate and trichloroacetate from
18 trichloroethylene in male B6C3F1 mice. ToxicolAppl Pharmacol 123, 1-8.
19 Templin, M. V., Stevens, D. K., Stenner, R. D., Bonate, P. L., Tuman, D., and Bull, R. J. (1995). Factors affecting
20 species differences in the kinetics of metabolites of trichloroethylene. J Toxicol Environ Health 44, 435-447.
21 Trevisan A, Giraldo M, Borella M, Maso S. (2001). Historical control data on urinary and renal tissue biomarkers in
22 naive male Wistar rats. JAppl Toxicol. 21 (5) :409-13.
23 Triebig, G., Essing, H.-G., Schaller, K.-H., Valentin, H. (1976). Biochemical and psychological examinations of test
24 subjects exposed to trichloroethylene. Zbl. Bakt. Hyg., I. Abt Orig. B 163: 383-416.
25 Yu, K. O., Barton, H. A., Mahle, D. A., and Frazier, J. M. (2000). In vivo kinetics of trichloroacetate in male Fischer
26 344 rats. Toxicol Sci 54, 302-311.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 A-111 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX B
Systematic Review of Epidemiologic Studies
on Cancer and Trichloroethylene (TCE)
Exposure
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix B: Systematic Review of Epidemiologic Studies on Cancer and
Trichloroethylene (TCE) Exposure
LIST OF TABLES B-iv
APPENDIX B. SYSTEMATIC REVIEW OF EPIDEMIOLOGIC STUDIES ON
CANCER AND TRICHLOROETHYLENE (TCE) EXPOSURE B-l
B.I. INTRODUCTION B-l
B.2. METHODOLOGIC REVIEW OF EPIDEMIOLOGIC STUDIES ON
CANCER AND TRICHLOROETHYLENE B-l
B.2.1. Study Designs and Characteristics B-25
B.2.2. Outcomes Assessed in Trichloroethylene (TCE) Epidemiologic
Studies B-28
B.2.3. Disease Classifications Adopted in Trichloroethylene (TCE)
Epidemiologic Studies B-29
B.2.4. Exposure Classification B-31
B.2.5. Follow-up in Trichloroethylene (TCE) Cohort Studies B-33
B.2.6. Interview Approaches in Case-Control Studies of Cancer and
Trichloroethylene (TCE) Exposure B-34
B.2.7. Sample Size and Approximate Statistical Power B-36
B.2.8. Statistical Analysis and Result Documentation B-38
B.2.9. Systematic Review for Identifying Cancer Hazards and
Trichloroethylene (TCE) Exposure B-46
B.2.9.1. Cohort Studies B-47
B.2.9.2. Case-Control Studies B-51
B.2.9.3. Geographic-Based Studies B-54
B.2.9.4. Recommendation of Studies for Treatment Using
Meta-Analysis Approaches B-55
B.3. INDIVIDUAL STUDY REVIEWS AND ABSTRACTS B-56
B.3.1. Cohort Studies B-56
B.3.1.1. Studies of Aerospace Workers B-56
B.3.1.2. Cancer Incidence Studies Using Biological Monitoring
Databases B-96
B.3.1.3. Studies in the Taoyuan Region of Taiwan B-108
B.3.1.4. Studies of Other Cohorts B-122
B.3.2. Case-Control Studies B-171
B.3.2.1. Bladder Cancer Case-Control Studies B-171
B.3.2.2. Central Nervous System Cancers Case-Control Studies B-180
B.3.2.3. Colon and Rectal Cancers Case-Control Studies B-188
B.3.2.4. Esophageal Cancer Case-Control Studies B-200
B.3.2.5. Liver Cancer Case-Control Studies B-204
B.3.2.6. Lymphoma Case-Control Studies B-208
B.3.2.7. Childhood Leukemia B-238
B.3.2.8. Melanoma Case-Control Studies B-257
B.3.2.9. Pancreatic Cancer Case-Control Studies B-261
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-ii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
B.3.2.10. Prostatic Cancer Case-Control Studies B-265
B.3.2.11. Renal Cell Carcinoma Case-Control Studies—Arnsberg
Region of Germany B-269
B.3.2.12. Renal Cell Carcinoma Case-Control Studies—Arve
Valley Region of France B-284
B.3.2.13. Renal Cell Carcinoma Case-Control Studies in Other
Regions B-292
B.3.2.14. Other Cancer Site Case-Control Studies B-302
B.3.3. Geographic-Based Studies B-306
B.3.3.1. Coyle et al. (2005) B-306
B.3.3.2. Morgan and Cassady (2002) B-310
B.3.3.3. Cohnetal. (1994) B-314
B.3.3.4. Vartiainen et al. (1993) B-319
B.3.3.5. Mallin(1990) B-322
B.3.3.6. Issacson et al. (1985) B-326
B.3.3.7. Studies in the Endicott Area of New York B-331
B.3.3.8. Studies in Arizona B-342
B.4. REFERENCES B-351
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-iii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE
exposure B-3
B-2. Case-control epidemiologic studies examining cancer and TCE exposure B-9
B-3. Geographic-based studies assessing cancer and TCE exposure B-19
B-4. Approximate statistical power (%) in cohort and geographic-based studies to
detect an RR = 2 B-39
B-5. Summary of rationale for study selection for meta-analysis B-57
B-6. Characteristics of epidemiologic investigations of Rocketdyne workers B-71
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-iv DRAFT—DO NOT CITE OR QUOTE
-------
1 APPENDIX B. SYSTEMATIC REVIEW OF EPIDEMIOLOGIC STUDIES ON
2 CANCER AND TRICHLOROETHYLENE (TCE) EXPOSURE
O
4
5 B.I. INTRODUCTION
6 The epidemiologic evidence on trichloroethylene (TCE) is large with over 50 studies
7 identified as of June 2009 and includes occupational cohort studies, case-control studies, both
8 nested within a cohort (nested case-control study) or population based, and geographic based
9 studies. The analysis of epidemiologic studies on cancer and TCE serves to document essential
10 design features, exposure assessment approaches, statistical analyses, and potential sources of
11 confounding and bias. These studies are described below and reviewed according to criteria to
12 assess (1) their ability to inform weight of evidence evaluation for TCE exposure and a cancer
13 hazard and (2) their utility for examination using meta-analysis approaches. A secondary goal of
14 the qualitative review is to provide transparency on study strengths and weaknesses, providing
15 background for inclusion or exclusion of individual studies for quantitative treatment using meta-
16 analysis approaches. Individual study qualities are discussed according to specific criteria in
17 Section B.2.1 to B.2.8., and rationale for studies examined using meta-analysis approaches, the
18 systematic review, contained in Section B.2.9. Appendix C contains a full discussion of the
19 meta-analysis, its analytical methodology, including sensitivity analyses, and findings. This
20 analysis supports discussion of site-specific cancer observations in Chapter 4 where a
21 presentation may be found of study findings with assessment and discussion of observations
22 according to a study's weight of evidence and potential for alternative explanations, including
23 bias and confounding.
24
25 B.2. METHODOLOGIC REVIEW OF EPIDEMIOLOGIC STUDIES ON CANCER
26 AND TRICHLOROETHYLENE
27 Epidemiologic studies considered in this analysis assess the relationship between TCE
28 exposure and cancer, and are identified using several sources and their utility for characterizing
29 hazard and quantitative treatment is based on recommendations in National Research Council
30 (NRC, 2006). A thorough search of the literature was carried out through June 2009 without
31 restriction on year of publication or language using the following approaches: a search of the
32 bibliographic database PubMed (http://www.ncbi.nlm.nih.gov/ pubmed/), TOXNET
33 (http://toxnet.nlm.nih.gov/) and EMBASE (http://www.embase.com/) using the terms
34 "trichloroethylene cancer epidemiology" and ancillary terms, "degreasers," "aircraft, aerospace
35 or aircraft maintenance workers," "metal workers," and "electronic workers," "trichloroethylene
36 and cohort," or, "trichloroethylene and case-control;" bibliographies of reviews of the TCE
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-1 DRAFT—DO NOT CITE OR QUOTE
-------
1 epidemiologic literature such as those of the Institute of Medicine (IOM, 2003), NRC (2006,
2 2009) and Scott and Chiu (2006) and, review of bibliographies of individual studies for relevant
3 studies not identified in the previous two approaches. NRC (2006) noted "a full review of the
4 literature should identify all published studies in which there was a possibility that
5 trichloroethylene was investigated, even though results per se may not have been reported."
6 Additional steps of U.S. Environmental Protection Agency (U.S. EPA) staff to identify
7 studies not published in the literature included contacting primary investigators for case-control
8 studies of liver, kidney and lymphoma and occupation, asking for information on analyses
9 examining trichloroethylene uniquely and a review of Agency for Toxic Substances and Disease
10 Registry (ATSDR) or state health department community health surveys or statistics reviews for
11 information on TCE exposure and cancer incidence or mortality.
12 The breadth of the available epidemiologic database on trichloroethylene and cancer is
13 wide compared to that available for other chemicals assessed by U.S. EPA. However, few
14 studies were designed with the sole, or primary, objective of this report—to characterize the
15 magnitude of underlying association, if such exists, between TCE and cancer. Yet, many studies
16 in the body of evidence can provide information for identifying cancer hazard and dose-response
17 inferences. The weight a study contributes to the overall evidence on TCE and cancer depends
18 on a number of characteristics regarding the design, exposure assessment, and analysis
19 approaches. Epidemiologic studies were most informative for analysis if they approached ideals
20 described below, as evaluated using objective criteria for identifying a cancer hazard.
21 Seventy-five studies potentially relevant to health assessment of TCE exposure and
22 cancer and identified from the above comprehensive search are presented in Tables B-l, B-2, and
23 B-3. The studies vary widely in their approaches to study design, exposure assessment, and
24 statistical analysis; for these reasons, studies vary in their usefulness for identifying cancer
25 hazard. Studies are reviewed according to a set of a priori guidelines of their utility for assessing
26 TCE exposure and cancer according to the below criteria. Studies approaching criteria ideals
27 contribute greater weight in the weight of evidence analysis than studies with significant
28 deficiencies. These criteria are not meant to be used to "accept" or "reject" a particular study for
29 identifying cancer hazard. Rather, they are to be used as measurement tools for evaluating a
30 study's ability to identify TCE exposure and cancer outcomes. Studies suitable for meta-analysis
31 treatment are selected according to specific criteria identified in B.2.9.4. Individual study
32 descriptions and abstract sheets according to these criteria are found in Section B.3. Appendix C
33 describes meta-analysis methods and findings.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-2 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-l Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
to
vo £;•
I
I
§
***.
£3'
1
TO'
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Aircraft and aerospace workers
Radican et al.
(2008), Blair
etal. (1998)
Civilian aircraft-maintenance
workers with at least 1 yr in
1952-1956 at Hill Air Force Base,
UT. Vital status (VS) to 1990
(Blair et al. 1998) or 2000 (Radican
et al., 2008); cancer incidence
1973-1990 (Blairetal., 1998).
14,457 (7,204 ever exposed to
TCE).
Incidence (Blair et al., 1998) and
mortality rates (Blair et al., 1998;
Radican etal., 2008) of
nonchemical exposed subjects.
Most subjects (n = 10,718) with potential exposure to 1 to 25
solvents. Cumulative TCE assigned to individual subjects using
JEM. Exposure-response patterns assessed using cumulative
exposure, continuous or intermittent exposures, and peak exposure.
TCE replaced in 1968 with 1,1,1-trichloroethane and was
discontinued in 1978 in vapor degreasing activities. Median TCE
exposures were about 10 ppm for rag and bucket; 100-200 ppm for
vapor degreasing. Poisson regression analyses controlled for age,
calendar time, sex (Blair et al., 1998) or Cox proportional hazard
model for age and race.
Krishnadasan
et al. (2007)
Nested case-control study within a
cohort of 7,618 workers employed
for between 1950 and 1992, or who
had started employment before
1980 at Boeing/Rockwell/
Rocketdyne (Santa Susana Field
Laboratory, [the UCLA cohort of
Morgensternetal., 1997]). Cancer
incidence 1988-1999.
326 prostate cancer cases, 1,805
controls.
Response rate:
Cases, 69%; Controls, 60%.
JEM for TCE, hydrazine, PAHs, benzene, mineral oil constructed
from company records, walk-through, or interviews. Lifestyle factors
obtained from living subjects through mail and telephone surveys.
Conditional logistic regression controlled for cohort, age at diagnosis,
physical activity, SES and other occupational exposure (benzene,
PAHs, mineral oil, hydrazine).
^§
H I
O >
HH Oq
H TO
O
Zhao et al.
(2005); Ritz
etal. (1999)
Aerospace workers with >2 yrs of
employment at Rockwell/
Rocketdyne (now Boeing) and who
worked at Santa Susana Field
Laboratory, Ventura, CA, from
1950-1993 (the UCLA cohort of
Morgenstern etal. [1997]). Cancer
mortality as of December 31, 2001.
Cancer incidence 1988-2000 for
subjects alive as of 1988.
6,044 (2,689 with high cumulative
exposure to TCE). Mortality rates of
subjects in lowest TCE exposure
category.
5,049 (2,227 with high cumulative
exposure to TCE). Incidence rates of
subjects in lowest TCE exposure
category.
JEM for TCE, hydrazine, PAHs, mineral oil, and benzene. IH
ranked each job title ranked for presumptive TCE exposure as high
(3), medium (2), low (1), or no (0) exposure for 3 time periods
(1951-1969, 1970-1979, 1980-1989). Cumulative TCE score: low
(up to 3), medium (over 3 up to 12), high (over 12) assigned to
individual subjects using JEM. Cox proportional hazard, controlled
for time, since 1st employment, SES, age at diagnosis and
hydrazine.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
(continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Boice et al.
(2006a)
Aerospace workers with >6 months
employment at Rockwell/
Rocketdyne (Santa Susana Field
Laboratory and nearby facilities)
from 1948-1999 (IEI cohort, IEI
[2005]). VS to 1999.
41,351, 1,642 male hourly test stand
mechanics (1,111 with potential TCE
exposure).
Mortality rates of United States
population and California
population. Internal referent groups
including male hourly
nonadministrative Rocketdyne
workers; male hourly,
nonadministrative SSFL workers;
and test stand mechanics with no
potential exposure to TCE.
Potential TCE exposure assigned to test stands workers only whose
tasks included the cleaning or flushing of rocket engines (engine
flush) (n = 639) or for general utility cleaning (n = 472); potential
for exposure to large quantities of TCE was much greater during
engine flush than when TCE used as a utility solvent. JEM for TCE
and hydrazine without semiquantitative intensity estimates.
Exposure to other solvents not evaluated due to low potential for
confounding (few exposed, low exposure intensity, or not
carcinogenic). Exposure metrics included employment duration,
employment decade, years worked with potential TCE exposure, and
years worked with potential TCE exposure via engine cleaning,
weighted by number of tests. Lifetable (SMR); Cox proportional
hazard controlling for birth year, hire year, and hydrazine exposure.
^§
H I
O >
HH Oq
H TO
O
Boice et al.
(1999)
Aircraft-manufacturing workers
with at least 1 yr >1960 at
Lockheed Martin (Burbank, CA).
VS to 1996.
77,965 (2,267 with potential routine
TCE exposures and 3,016 with
routine or intermittent TCE
exposure).
Mortality rates of United States
population (routine TCE exposed
subjects) and non-exposed internal
referents (routine and intermittent
TCE exposed subjects).
12% with potential routine mixed solvent exposure and 30% with
route or intermittent solvent exposure. JEM for potential TCE
exposure on (1) routine basis or (2) intermittent or routine basis
without semiquantitative intensity estimate. Exposure-response
patterns assessed by any exposure or duration of exposure and
internal control group. Vapor degreasing with TCE before 1966 and
PCE, afterwards. Lifetable analyses (SMR); Poisson regression
analysis adjusting for birth date, starting employment date, finishing
employment date, sex and race.
Morgan et al.
(1998)
Aerospace workers with >6 months
1950-1985 at Hughes (Tucson,
AZ). VS to 1993.
20,508 (4,733 with TCE exposures).
Mortality rates of United States
population for overall TCE exposure;
mortality rates of all-other cohort
subjects (internal referents) for
exposure-response analyses.
TCE exposure intensity assigned using JEM. Exposure-response
patterns assessed using cumulative exposure (low versus high) and
job with highest TCE exposure rating (peak, medium/high exposure
versus no/low exposure). "High exposure" job classification defined
as >50 ppm. Vapor degreasing with TCE 1952-1977, but limited IH
data <1975. Limited IH data before 1975 and medium/ low rankings
likely misclassified given temporal changes in exposure intensity not
fully considered (NRC, 2006).
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
(continued)
Reference
Costa et al.
(1989)
Garabrant et
al. (1988)
Description
Aircraft manufacturing workers
employed 1954-1981at plant in
Italy. VS to 1981.
Aircraft manufacturing workers >4
yrs employment and who had
worked at least 1 d at San Diego,
CA, plant 1958-1982. VS to 1982.
Study group (N)
Comparison group (N)
8,626 subjects
Mortality rates of the Italian
population.
14,067
Mortality rates of United States
population.
Exposure assessment and other information
No exposure assessment to TCE and job titles grouped into one of
four categories: blue- and white-collar workers, technical staff, and
administrative clerks. Lifetable (SMR).
TCE exposure assessment for 70 of 14,067 subjects; 14 cases of
esophageal cancer and 56 matched controls. For these 70 subjects,
company work records identified 37% with job title with potential
TCE exposure without quantitative estimates. Lifetable (SMR).
Cohorts Identified From Biological Monitoring (U-TCA)
Hansen et al.
(2001)
Anttila et al.
(1995)
Axelson et al.
(1994)
Workers biological monitored using
U-TCA and air-TCE, 1947-1989.
Cancer incidence from 1964-1996.
Workers biological monitored using
U-TCA, 1965-1982. VS
1965-1991 and cancer incidence
1967-1992.
Workers biological monitored using
U-TCA, 1955-1975. VS to 1986
and cancer incidence 1958-1987.
803 total
Cancer incidence rates of the Danish
population.
3,974 total (3,089 with U-TCA
measurements]).
Mortality and cancer incidence rates
of the Finnish population.
1,4,21 males
Mortality and cancer incidence rates
of Swedish male population.
712 with U-TCA, 89 with air-TCE measurement records, 2 with
records of both types. U-TCA from 1947-1989; air TCE
measurements from 1974. Historic median exposures estimated
from the U-TCA concentrations were: 9 ppm for 1947 to 1964,
5 ppm for 1965 to 1973, 4 ppm for 1974 to 1979, and 0.7 ppm for
1980 to 1989. Air TCE measurements from 1974 onward were
19 ppm (mean) and 5 ppm (median). Overall, median TCE
exposure to cohort as extrapolated from air TCE and U-TCA
measurements was 4 ppm (arithmetic mean, 12 ppm). Exposure
metrics: year 1st employed, employment duration, mean exposure,
cumulative exposure. Exposure metrics: employment duration,
average TCE intensity, cumulative TCE, period 1st employment.
Lifetable analysis (SIR).
Median U-TCA, 63 umol/L for females and 48 umol/L for males;
mean U-TCA was 100 umol/L. Average 2.5 U-TCA measurements
per individual. Using the Ikeda et al. (1972) relationship for TCE
exposure to U-TCA, TCE exposures were roughly 4 ppm (median)
and 6 ppm (mean). Exposure metrics: years since 1st measurement.
Lifetable analysis (SMR, SIR).
Biological monitoring for U-TCA from 1955 and 1975. Roughly %
of cohort had U-TCA concentrations equivalent to <20 ppm TCE.
Exposure metrics: duration exposure, mean U-TCA. Lifetable
analysis (SMR, SIR).
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
"I
"§
i
^§
H I
O >
HH Oq
H TO
O
Table B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
(continued)
Reference Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Other Cohorts
Clapp and
Hoffman
(2008)
Deaths between 1969-2001 among
employees >5 yrs employment
duration at an IBM facility
(Endicott, NY).
360 deaths
Proportion of deaths among New
York residents during 1979 to 1998.
No exposure assessment to TCE. PMR analysis.
Sung et al.
(2007, 2008)
Female workers 1st employed
1973-1997 at an electronics (RCA)
manufacturing factory (Taoyuan,
Taiwan). Cancer incidence 1979-
2001 (Sung et al., 2007).
Childhood leukemia 1979-2001
among first born of female subjects
in Sung et al. (2007, 2008).
63,982 females and 40,647 females
with 1st live born offspring.
Cancer incidence rates of Taiwan
population (Sung et al., 2007).
Childhood leukemia incidence rates
of first born live births of Taiwan
population (Sung et al., 2007).
No exposure assessment. Chlorinated solvents including TCE and
PCE found in soil and groundwater at factory site. Company records
indicated TCE not used 1975-1991 and PCE 1975-1991 and PCE
after 1981. No information for other time periods. Exposure-
response using employment duration. Lifetable analysis (SMR, SIR)
(Chang et al., 2003, 2005; Sung et al., 2007) or Poisson regression
adjusting for maternal age, education, sex, and birth year (Sung et
al., 2008).
Chang et al.
(2005),
Chang et al.
(2003)
Male and female workers employed
1978-1997 at electronics factory as
studied by Sung et al. (2007). VS
from 1985-1997 and cancer
incidence 1979-1997.
86,868 total
Incidence (Chang et al., 2005) or
mortality (Chang et al., 2003) rates
Taiwan population.
ATSDR
(2004)
Workers 1952-1980 at the View-
Master factory (Beaverton, OR).
616 deaths 1989-2001
Proportion of deaths between
1989-2001 in Oregon population.
No exposure information on individual subjects. TCE and other
VOCs detected in well water at the time of the plant closure in 1998
were TCE, 1,220-1,670 ug/L; 1,1-DCE, up to 33 ug/L; and, PCE up
to 56 ng/L. PMR analysis.
Raaschou-
Nielsen et al.
(2003)
Blue-collar workers employed
>1968 at 347 Danish TCE-using
companies. Cancer incidence
through 1997.
40,049 total (14,360 with presumably
higher level exposure to TCE).
Cancer incidence rates of the Danish
population.
Employers had documented TCE usage but no information on
individual subjects. Blue-collar versus white-collar workers and
companies with <200 workers were variables identified as increasing
the likelihood for TCE exposure. Subjects from iron and metal,
electronics, painting, printing, chemical, and dry cleaning industries.
Median exposures to trichloroethylene were 40-60 ppm for the years
before 1970, 10-20 ppm for 1970 to 1979, and approximately 4 ppm
for 1980 to 1989. Exposure metrics: employment duration, year 1st
employed, and # employees in company. Lifetable (SIR).
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
"I
^§
i
^§
H I
O >
HH Oq
H TO
Table B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
(continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Ritz (1999a)
Male uranium-processing plant
workers >3 months employment
1951-1972 at DOE facility
(Fernald, OH). VS 1951-1989,
cancer.
3,814 white males monitored for
radiation (2,971 with potential TCE
exposure).
Mortality rates of the United States
population; Non-TCE exposed
internal controls for TCE exposure-
response analyses.
JEM for TCE, cutting fluids, kerosene, and radiation generated by
employees and industrial hygienists. Subjects assigned potential
TCE according to intensity: light (2,792 subjects), moderate
(179 subjects), heavy (no subjects). Lifetable (SMR) and
conditional logistic regression adjusted for pay status, date first hire,
radiation.
Henschler et
al. (1995)
Male workers > 1 yr 1956-1975 at
cardboard factory (Arnsberg region,
Germany). VS to 1992.
169 exposed; 190 unexposed
Mortality rates from German
Democratic Republic (broad
categories) or renal cell carcinoma
incidence rates from Danish
population, German Democratic, or
non-TCE exposed subjects.
Walk-through surveys and employee interviews used to identify
work areas with TCE exposure. TCE exposure assigned to renal
cancer cases using workman's compensation files. Lifetable (SMR,
SIR) or Mantel-Haenszel.
Greenland et
al. (1994)
Cancer deaths, 1969-1984, among
pensioned workers employed
<1984 at GE transformer
manufacturing plant (Pittsfield,
MA), and who had job history
record; controls were noncancer
deaths among pensioned workers.
512 cases, 1,202 controls.
Response rate:
Cases, 69%;
Controls, 60%.
Industrial hygienist assessment from interviews and position
descriptions. TCE (no/any exposure) assigned to individual subjects
using JEM. Logistic regression.
Sinks et al.
(1992)
Workers employed 1957-1980 at a
paperboard container
manufacturing and printing plant
(Newnan, GA). VS to 1988.
Kidney and bladder cancer
incidence through 1990.
2,050 total
Mortality rates of the United States
population, bladder and kidney
cancer incidence rates from the
Atlanta-SEER registry for the years
1973-1977.
No exposure assessment to TCE; analyses of all plant employees
including white- and blue-collar employees. Assignment of work
department in case-control study based upon work history; Material
Safety Data Sheets identified chemical usage by department.
Lifetable (SMR, SIR) or conditional logistic regression adjusted for
hire date and age at hire, and using 5- and 10-year lagged
employment duration.
O
H
W
-------
to
O k^j
o ^
"I
I
§
^.
Co'
1
TO'
td
Table B-l. Description of epidemiologic cohort and PMR studies assessing cancer and TCE exposure
(continued)
Reference
Description
Study group (N)
Comparison group (N)
Exposure assessment and other information
Blair et al.
(1989)
Workers employed 1942-1970 in
U.S. Coast. VS to 1980.
3,781 males of whom 1,767 were
marine inspectors (48%).
Mortality rates of the United States
population. Mortality rates of marine
inspectors also compared to that of
noninspectors.
No exposure assessment to TCE. Marine inspectors worked in
confined spaces and had exposure potential to multiple chemicals.
TCE was identified as one of 10 potential chemical exposures.
Lifetable (SMR) and directly adjusted relative risks.
Shannon et
al. (1988)
Workers employed >6 months at
GE lamp manufacturing plant,
1960-1975. Cancer incidence from
1964-1982.
1,870 males and females, 249 (13%)
in coiling and wire-drawing area.
Cancer incidence rates from Ontario
Cancer Registry.
No exposure assessment to TCE. Workers in coiling and wire
drawing (CWD) had potential exposure to many chemicals including
metals and solvents. A 1955-dated engineering instruction sheet
identified trichloroethylene used as degreasing solvent in CWD.
Lifetable (SMR).
Co
§
oo
Shindell and
Ulrich(1985)
Workers employed >3 months at a
TCE manufacturing plant 1957-
1983. VS to 1983.
2,646 males and females
Mortality rates of the United States
population.
No exposure assessment to TCE; job titles categorized as either
white- or blue-collar. Lifetable analysis (SMR).
i
^§
Wilcosky et
al. (1984)
Respiratory, stomach, prostate,
lymphosarcoma, and lymphatic
leukemia cancer deaths 1964-1972
among 6,678 active and retired
production workers at a rubber
plant (Akron, OH); controls were a
20% age-stratified random sample
of the cohort.
183 cases (101 respiratory,
33 prostate, 30 stomach, 9
lymphosarcoma and 10 lymphatic
leukemia cancer deaths).
JEM without quantitative intensity estimates for 20 exposures
including TCE. Exposure metric: ever held job with potential TCE
exposure.
H I
O >
HH Oq
H TO
DCE = dichloroethylene, DOE = U.S. Department of Energy, IEI = International Epidemiology Institute, JEM = job-exposure matrix, NRC = National Research
Council, PCE = perchloroethylene, PMR = proportionate mortality ratio, SIR = standardized incidence ratio, SMR = standardized mortality ratio, SSFL = Santa
Susanna Field Laboratory, U-TCA = urinary trichloroacetic acid, UCLA = University of California, Los Angeles, VOCs = volatile organic compounds, VS =
vital status.
O
H
W
-------
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
?§
"§
i
^§
H I
O >
HH Oq
H TO
O
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Bladder
Pesch et al.
(2000a)
Histologically confirmed
urothelial cancer (bladder, ureter,
renal pelvis) cases from German
hospitals (5 regions) in
1991-1995; controls randomly
selected from residency registries
matched on region, sex, and age.
1,035 cases
4,298 controls
Cases, 84%; Controls, 71%
Occupational history using job title or serf-reported exposure. JEM and
]TEM to assign exposure potential to metals and solvents (chlorinated
solvents, TCE, PCE). Lifetime exposure to TCE exposure examined as 30th,
60th, and 90th percentiles (medium, high, and substantial) of exposed control
exposure index. Duration used to examine occupational title and job task
duties and defined as 30th, 60th, and 90th percentiles (medium, long, and
very long) of exposed control durations.
Logistic regression with covariates for age, study center, and smoking.
Siemiatycki et
al. (1994),
Siemiatycki
(1991)
Male bladder cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and random digit
dialing (ROD).
484 cases
533 population controls;
740 other cancer controls
Cases, 78%; Controls, 72%
JEM to assign 294 exposures including TCE on semiquantitative scales
categorized as any or substantial exposure. Other exposure metrics included
exposure duration in occupation or job title.
Logistic regression adjusted for age, ethnic origin, socioeconomic status,
smoking, coffee consumption, and respondent status [occupation or job title]
or Mantel-Haenszel stratified on age, income, index for cigarette smoking,
coffee consumption, and respondent status (TCE).
Brain
De Roos et al.
(2001)
Olshan et al.
(1999)
Neuroblastoma cases in children
of <19 yrs selected from
Children's Cancer Group and
Pediatric Oncology Group with
diagnosis in 1992-1994;
population controls (RDD)
matched to control on birth date.
504 cases
504 controls
Cases, 73%; Controls, 74%
Telephone interview with parent using questionnaire to assess parental
occupation and serf-reported exposure history and judgment-based attribution
of exposure to chemical classes (halogenated solvents) and specific solvents
(TCE). Exposure metric was any potential exposure.
Logistic regression with covariate for child's age and material race, age, and
education.
Heineman et
al. (1994)
White, male cases, age >30 yrs,
identified from death certificates
in 1978-1981; controls identified
from death certificates and
matched for age, year of death and
study area.
300 cases
386 controls
Cases, 74%; Controls, 63%
In-person interview with next-of-kin; questionnaire assessing lifetime
occupational history using job title and JEM of Gomez et al. (1994).
Cumulative exposure metric (low, medium or and high) based on weighted
probability and duration.
Logistic regression with covariates for age and study area.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
Dl
Co
o §
H I
O >
HH Oq
H TO
O
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Colon and Rectum
Goldberg et al.
(2001),
Siemiatycki
(1991)
Male colon cancer cases, 35-75
yrs, from 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and random digit
dialing (ROD).
497 cases
533 population controls and
740 cancer controls
Cases, 82%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales);
potential TCE exposure defined as any or substantial exposure.
Logistic regression adjusted for age, ethnic origin, birthplace, education,
income, parent's occupation, smoking, alcohol consumption, tea
consumption, respondent status, heating source socioeconomic status,
smoking, coffee consumption, and respondent status [occupation, some
chemical agents] or Mantel-Haenszel stratified on age, income, index for
cigarette smoking, coffee consumption, and respondent status [TCE].
Dumas et al.
(2000),
Simeiatycki
(1991)
Male rectal cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
292 cases
533 population controls and
740 other cancer controls
Cases, 78%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales);
potential TCE exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, respondent status, cigarette
smoking, beer consumption and body mass index [TCE] or Mantel-Haenszel
stratified on age, income, index for cigarette smoking, coffee consumption,
ethnic origin, and beer consumption [TCE].
Fredriksson et
al. (1989)
Colon cancer cases aged 30-75
yrs identified through the Swedish
Cancer Registry among patients
diagnosed in 1980-1983;
population-based controls were
frequency-matched on age and sex
and were randomly selected from
a population register.
329 cases
658 controls
Not available
Mailed questionnaire assessing occupational history with telephone interview
follow-up. Serf-reported exposure to TCE defined as any exposure.
Mantel-Haenszel stratified on age, sex, and physical activity.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Esophagus
Parent et al.
(2000a),
Siemiatycki
(1991)
Male esophageal cancer cases,
35-75 yrs, diagnosed in 19 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
292 cases
533 population controls;
740 subjects with other
cancers
Cases, 78%; controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales);
potential TCE exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, respondent status, cigarette
smoking, beer consumption and body mass index [solvents] or Mantel-
Haenszel stratified on age, income, index for cigarette smoking, coffee
consumption, ethnic origin, and beer consumption [TCE].
Lymphoma
Wang et al.
(2009)
Cases among females aged 21 and
84 yrs with NHL in 1996-2000
and identified from Connecticut
Cancer Registry; population-based
female controls (1) if <65 yrs of
age, having Connecticut address
stratified by 5-yr age groups
identified from random digit
dialing or (2) >65 yrs of age, by
random selection from Centers for
Medicare and Medicaid Service
files.
601 cases
717 controls
Cases, 72%; Controls, 69%
(<65 yrs), 47% (>65 yrs)
In-person interview with using questionnaire assessment specific jobs held
for >1 yr. Intensity and probability of exposure to broad category of organic
solvents and to individual solvents, including TCE, estimated using JEM
(Gomez et al, 1994; Dosemeci et al., 1994) and assigned blinded. Exposure
metric of any exposure, exposure intensity (low, medium/high), and exposure
probability (low, medium/high).
Logistic regression adjusted for age, family history of hematopoietic cancer,
alcohol consumption and race.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
i
to O
I
-
o §
H I
O >
HH Oq
H TO
O
H
W
Constantini et
al. (2008),
Miligi et al.
(2006)
Cases aged 20-74 with NHL,
including CLL, all forms of
leukemia, or multiple myeloma
(MM) in 1991-1993 and
identified through surveys of
hospital and pathology
departments in study areas and in
specialized hematology centers in
8 areas in Italy; population-based
controls stratified by 5-yr age
groups and by sex selected
through random sampling of
demographic or of National Health
Service files.
1,428 NHL + CLL, 586
Leukemia,
263, MM
1,278 controls (leukemia
analysis)
1,100 controls (MM
analysis)
Cases, 83%; Controls, 73%
In-person interview primarily at interviewee's home (not blinded) using
questionnaire assessing specific jobs, extra occupational exposure to solvents
and pesticides, residential history, and medical history. Occupational
exposure assessed by job-specific or industry-specific questionnaires. JEM
used to assign TCE exposure and assessed using intensity (2 categories) and
exposure duration (2 categories). All NHL diagnoses and 20% sample of all
cases confirmed by panel of 3 pathologists.
Logistic regression with covariates for sex, age, region, and education.
Logistic regression for specific NHL included an additional covariate for
smoking.
Seidler et al.
(2007)
Mester et al.
(2006)
Becker et al.
(2004)
NHL and Hodgkin's disease cases
aged 18-80 yrs identified through
all hospitals and ambulatory
physicians in six regions of
Germany between 1998 and 2003;
population controls were
identified from population
registers and matched on age, sex,
and region.
710 cases
710 controls
Cases, 87%; Controls, 44%
In-person interview using questionnaire assessing personal characteristics,
lifestyle, medical history, UV light exposure, and occupational history of all
jobs held for >1 yr. Exposure of a priori interest were assessed using job
task-specific supplementary questionnaires. JEM used to assign cumulative
quantitative TCE exposure metric, categorized according to the distribution
among the control persons (50th and 90th percentile of the exposed controls).
Conditional logistic regression adjusted for age, sex, region, smoking and
alcohol consumption.
Persson and
Fredriksson
(1999)
Combined
analysis of
NHL cases in
Persson et al.
(1993),
Persson et al.
(1989)
Histologically confirmed cases of
B-cell NHL, age 20-79 yrs,
identified in two hospitals in
Sweden: Oreboro in 1964-1986
(Persson et al., 1989) and in
Linkoping between 1975-1984
(Persson et al., 1993); controls
were identified from previous
studies and were randomly
selected from population registers.
NHL cases, 199
479 controls
Cases, 96% (Oreboro),
90% (Linkoping);
controls, not reported
Mailed questionnaire to assess self reported occupational exposures to TCE
and other solvents.
Unadjusted Mantel-Haenszel chi-square.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Nordstrom et
al. (1998)
Histologically-confirmed cases in
males of hairy-cell leukemia
reported to Swedish Cancer
Registry in 1987-1992 (includes
one case latter identified with an
incorrect diagnosis date);
population-based controls
identified from the National
Population Registry and matched
(1:4 ratio) to cases for age and
county.
Ill cases
400 controls
Cases, 91%; Controls, 83%
Mailed questionnaire to assess serf reported working history, specific
exposure, and leisure time activities.
Univariate analysis for chemical-specific exposures (any TCE exposure).
Fritschi and
Siemiatycki,
1996a),
Siemiatycki
(1991)
Male NHL cases, age 35-75 yrs,
diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and ROD.
215 cases
533 population controls
(Group 1) and
1,900 subjects with other
cancers (Group 2)
Cases, 83%; Controls, 71%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales).
Exposure metric defined as any or substantial exposure.
Logistic regression adjusted for age, proxy status, income, and ethnicity
[solvents] or Mantel-Haenszel stratified by age, body mass index, and
cigarette smoking [TCE].
^§
H I
O ^
HH Oq
H TO
O
H
W
Hardell et al.
(1994, 1981)
Histologically-confirmed cases of
NHL in males, age 25-85 yrs,
admitted to Swedish (Umea)
hospital between 1974-1978;
living controls (1:2 ratio) from the
National Population Register,
matched to living cases on sex,
age, and place of residence;
deceased controls from the
National Registry for Causes of
Death, matched (1:2 ratio) to dead
cases on sex, age, place of
residence, and year of death.
105 cases
335 controls
Response rate not available
Serf-administered questionnaire assessing self-reported solvent exposure;
phone follow-up with subject, if necessary.
Unadjusted Mantel-Haenszel chi-square.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
Dl
Co
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Persson et al.
(1993),
Persson et al.
(1989)
Histologically confirmed cases of
Hodgkin's disease, age 20-80 yrs,
identified in two hospitals in
Sweden: Oreboro in 1964-1986
(Persson et al., 1989) and in
Linkoping between 1975-1984
(Persson et al., 1993); controls
randomly selected from
population registers.
54 cases (1989 study);
3 leases (1993 study)
275 controls (1989 study);
204 controls (1993 study)
Response rate not available
Mailed questionnaire to assess self reported occupational exposures to TCE
and other solvents.
Logistic regression with adjustment for age and other exposure; unadjusted
Mantel-Haenszel chi-square.
Childhood Leukemia
Shu et al.
(2004, 1999)
Childhood leukemia cases, <15
yrs, diagnosed between 1989 and
1993 by a Children's Cancer
Group member or affiliated
institute; population controls
(random digit dialing), matched
for age, race, and telephone area
code and exchange.
1,842 cases
1,986 controls
Cases, 92%; controls, 77%
Telephone interview with mother, and whenever available, fathers using
questionnaire to assess occupation using job-industry title and serf-reported
exposure history. Questionnaire included questions specific for solvent,
degreaser or cleaning agent exposures.
Logistic regression with adjustment for maternal or paternal education, race,
and family income. Analyses of paternal exposure also included age and sex
of the index child.
H I
O >
HH Oq
H TO
O
H
W
Costas et al.
(2002), MA
DPH (1997)
Childhood leukemia (<19 yrs age)
diagnosed in 1969-1989 and who
were resident of Woburn. MA;
controls randomly selected from
Woburn public School records,
matched for age.
19 cases
37 controls
Cases, 91%; Controls, not
available
Questionnaire administered to parents separately assessing demographic and
lifestyle characteristics, medical history information, environmental and
occupational exposure and use of public drinking water in the home.
Hydraulic mixing model used to infer delivery of TCE and other solvents
water to residence.
Logistic regression with composite covariate, a weighted variable of
individual covariates.
McKinney et
al. (1991)
Incident childhood leukemia and
non-Hodgkin's lymphoma cases,
1974-1988, ages not identified,
from three geographical areas in
England; controls randomly
selected from children of residents
in the three areas and matched for
sex and birth health district.
109 cases
206 controls
Cases, 72%; Controls, 77%
In-person interview with questionnaire with mother to assess maternal
occupational exposure history, and with father and mother, as surrogate, to
assess paternal occupational exposure history. No information provided in
paper whether interviewer was blinded as to case and control status.
Matched pair design using logistic regression for univariate and multivariate
analysis.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Lowengart et
al. (1987)
Population
Childhood leukemia cases aged
<10 yrs and identified from the
Los Angeles (CA) Cancer
Surveillance Program in
1980-1984; controls selected from
RDD or from friends of cases and
matched on age, sex, and race.
Study group (N)
Comparison group (N)
Response rates
123 cases
123 controls
Cases, 79%; Controls,
not available
Exposure assessment and other information
Telephone interview with questionnaire to assess parental occupational and
self-reported exposure history.
Matched (discordant) pair analysis.
Melanoma
Fritschi and
Siemiatycki
(1996b),
Siemiatycki
(1991)
Male melanoma cases, age 35-75
yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
103 cases
533 population controls and
533 other cancer controls
Cases, 78%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales);
potential TCE exposure defined as any or substantial exposure.
Logistic regression adjusted for age, education, and ethic origin [TCE] or
Mantel-Haenszel stratified on age, income, index for cigarette smoking, and
ethnic origin [TCE].
Prostate
Aronson et al.
(1996),
Siemiatycki
(1991)
Male prostate cancer cases, age
35-75 yrs, diagnosed in 16 large
Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
449 cases
533 population controls
(Group 1) and
other cancer cases from
same study (Group 2)
Cases, 81%; Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales).
Logistic regression adjusted for age, ethnic origin, socioeconomic status,
Quetlet, and respondent status [occupation] or Mantel-Haenszel stratified on
age, income, index for cigarette smoking, ethnic origin, and respondent status
[TCE].
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Renal Cell
Charbotel et
al. (2006,
2009)
Cases from Arve Valley region in
France identified from local
urologists files and from area
teaching hospitals; age- and sex-
matched controls chosen from file
of same urologist as who treated
case or recruited among the
patients of the case's general
practitioner.
87 cases
316 controls
Cases, 74%; controls, 78%
Telephone interview with case or control, or, if deceased, with next-of-kin
(22% cases, 2% controls). Questionnaire assessing occupational history,
particularly, employment in the screw cutting jobs, and medical history.
Semiquantitative TCE exposure assigned to subjects using a task/TCE-
Exposure Matrix designed using information obtained from questionnaires
and routine atmospheric monitoring of work shops or biological monitoring
(U-TCA) of workers carried out since the 1960s. Cumulative exposure,
cumulative exposure with peaks, and TWA.
Conditional logistic regression with covariates for tobacco smoking and body
mass index.
i
ON O
I
-
Briining et al.
(2003)
Histologically-confirmed cases
1992-2000 from German
hospitals (Arnsberg); hospital
controls (urology department)
serving area, and local geriatric
department, for older controls,
matched by sex and age.
134 cases
401 controls
Cases, 83%; Controls, not
available
In-person interviews with case or next-of-kin; questionnaire assessing
occupational history using job title. Exposure metrics included longest job
held, JEM of Pannett et al. (1985) to assign cumulative exposure to TCE and
PCE, and exposure duration.
Logistic regression with covariates for age, sex, and smoking.
Pesch et al.
(2000b)
Histologically-confirmed cases
from German hospitals (5 regions)
in 1991-1995; controls randomly
selected from residency registries
matched on region, sex, and age.
935 cases
4,298 controls
Cases, 88%; Controls, 71%
In-person interview with case or next-of-kin; questionnaire assessing
occupational history using job title (JEM approach), self-reported exposure,
or job task (JTEM approach) to assign TCE and other exposures.
Logistic regression with covariates for age, study center, and smoking.
H I
O >
HH Oq
H TO
O
H
W
Parent et al.
(2000b),
Siemiatycki
(1991)
Male renal cell carcinoma cases,
age 35-75 yrs, diagnosed in 16
large Montreal-area hospitals in
1979-1985 and histologically
confirmed; controls identified
concurrently at 18 other cancer
sites; age-matched, population-
based controls identified from
electoral lists and RDD.
142 cases
533 population controls
(Group 1) and
other cancer controls
(excluding lung and bladder
cancers) (Group 2)
Cases, 82%; Controls, 71%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (about 300 exposures on semiquantitative scales);
TCE defined as any or substantial exposure.
Mantel-Haenszel stratified by age, body mass index, and cigarette smoking
[TCE] or logistic regression adjusted for respondent status, age, smoking,
and body mass index [occupation, job title].
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
Dl
Co
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Dosemeci et
al. (1999)
Histologically-confirmed cases,
1988-1990, white males and
females, 20-85 yrs, from
Minnesota Cancer Registry;
controls stratified for age and sex
using RDD, 21-64 yrs, or from
HCFA records, 64-85 yrs.
438 cases
687 controls
Cases, 87%; Controls, 86%
In-person interviews with case or next-of-kin; questionnaire assessing
occupational history of TCE using job title and JEM of Gomez et al. (1994).
Exposure metric was any TCE exposure.
Logistic regression with covariates for age, smoking, hypertension, and body
mass index.
Vamvakas et
al. (1998)
Cases who underwent
nephrectomy in 1987-1992 in a
hospital in Arnsberg region of
Germany; controls selected
accident wards from nearby
hospital in 1992.
58 cases
84 controls
Cases, 83%; Controls, 75%
In-person interview with case or next-of-kin; questionnaire assessing
occupational history using job title or self-reported exposure to assign TCE
and PCE exposure.
Logistic regression with covariates for age, smoking, body mass index,
hypertension, and diuretic intake.
Multiple or Other Sites
Lee et al.
(2003)
Liver, lung, stomach, colorectal
cancer deaths in males and
females between 1966-1997 from
two villages in Taiwan; controls
were cardiovascular and cerebral-
vascular disease deaths from same
underlying area as cases.
53 liver,
39 stomach,
26 colorectal,
41 lung cancer cases
286 controls
Response rate not reported
Residence as recorded on death certificate.
Mantel-Haenszel stratified by age, sex, and time period.
H I
O >
HH Oq
H TO
Kernan et al.
(1999)
Pancreatic deaths, 1984-1993, in
24 states; non-cancer death and
non-pancreatic disease death
controls, frequency matched to
cases by age, gender, race and
state.
63,097 pancreatic cancer
cases
252,386 non-cancer
population controls
Response rate not reported
Usual occupation and industry on death certificate coded to standardized
occupation codes and industry codes for 1980 U. S. census. Potential
exposure to 11 chlorinated hydrocarbons, including TCE, assessed using job-
exposure matrix of Gomez et al. (1994).
Logistic regression adjusted for age, marital status, gender, race, and
metropolitan and residential status.
O
H
W
-------
o
to
Table B-2. Case-control epidemiologic studies examining cancer and TCE exposure (continued)
vo
§•
§
Reference
Population
Study group (N)
Comparison group (N)
Response rates
Exposure assessment and other information
Siemiatycki
(1991)
Male cancer cases, 1979-1985,
35-75 yrs, diagnosed in
16 Montreal-area hospitals,
histologically confirmed; cancer
controls identified concurrently;
age-matched, population-based
controls identified from electoral
lists and ROD.
857 lung and
117 pancreatic cancer cases
533 population controls
(Group 1) and other cancer
cases from same study
(Group 2)
Cases, 79% (lung), 71%
(pancreas); Controls, 72%
In-person interviews (direct or proxy) with segments on work histories (job
titles and serf-reported exposures); analyzed and coded by a team of chemists
and industrial hygienists (294 exposures on semiquantitative scales); TCE
defined as any or substantial exposure.
Mantel-Haenszel stratified on age, income, index for cigarette smoking,
ethnic origin, and respondent status (lung cancer) and age, income, index for
cigarette smoking, and respondent status (pancreatic cancer).
TO
TO'
•3
, O
oo o
§
H TO1
W S
HCFA = Health Care Financing Administration, JEM = job-expo sure matrix, ITEM = job-task-expo sure matrix, NCI = National Cancer Institute,
PCE = perchloroethylene, RDD = random digit dialing, U-TCA = urinary trichloroacetic acid, UV = ultra-violet.
O
H
W
-------
Table B-3. Geographic-based studies assessing cancer and TCE exposure
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
VO O
•••2
a
o §
H I
O >
HH Oq
H TO
O
H
W
Reference
Description
Analysis approach
Exposure assessment
Broome County, NY Studies
ATSDR
(2006a, 2008)
Total, 22 site-specific, and
childhood cancer incidence
from 1980-2001 among
residents in 2 areas in
Endicott, NY.
SIR among all subjects (ATSDR,
2006a) or among white subjects
only (ATSDR, 2008) with expected
numbers of cancers derived using
age-specific cancer incidence rates
for New York State, excluding New
York City. Limited assessment of
smoking and occupation using
medical and other records in lung
and kidney cancer subjects
(ATSDR, 2008).
Two study areas, Eastern and Western study areas, identified based on
potential for soil vapor intrusion exposures as defined by the extent of
likely soil vapor contamination. Contour lines of modeled VOC soil vapor
contamination levels based on exposure model using GIS mapping and soil
vapor sampling results taken in 2003. The study areas were defined by
2000 Census block boundaries to conform to model predicted areas of soil
vapor contamination. TCE was the most commonly found contaminant in
indoor air in Eastern study area at levels ranging from 0. 18 to 140 ug/m3 ,
with tetrachloroethylene, cis-l,2-dichloroethene, 1,1,1-trichloroethane, 1,1-
dichloroethylene, 1,1-dichloroethane, and Freon 113 detected at lower
levels. PCE was most common contaminant in indoor air in Western study
area with other VOCs detected at lower levels.
Maricopa County, AZ Studies
Aickin et al.
(1992) Aickin
(2004)
Cancer deaths, including
leukemia, 1966-1986, and
childhood (<19yrs old)
leukemia incident cases
(1965-1986), Maricopa
County, AZ.
Standardized mortality RR from
Poisson regression modeling.
Childhood leukemia incidence data
evaluated using Bayes methods and
Poisson regression modeling.
Location of residency in Maricopa County, AZ, at the time of death as
surrogate for exposure. Some analyses examined residency in West Central
Phoenix and cancer. Exposure information is limited to TCE concentration
in two drinking water wells in 1982.
Pima County, AZ Studies
AZDHS(1990,
1995)
Cancer incidence in
children (<19 yrs old) and
testicular cancer in
1970-1986 and
1987-1991, Pima County,
AZ.
Standardized incidence RR from
Poisson regression modeling using
method of Aickin et al. (1992).
Analysis compares incidence in
Tucson Airport Area to rate for rest
of Pima County.
Location of residency in Pima, County, AZ, at the time of diagnosis or
death as surrogate for exposure. Exposure information is limited to
monitoring since 1981 and includes VOCs in soil gas samples (TCE, PCE,
1,1-dichloroethylene, 1,1,1-trichloroacetic acid); PCBs in soil samples, and
TCE in municipal water supply wells.
Other
Coyle et al.
(2005)
Incident breast cancer
cases among men and
women, 1995-2000,
reported to Texas Cancer
Registry
Correlation study using rank order
statistics of mean average annual
breast cancer rate among women
and men and atmospheric release of
12 hazardous air pollutants.
Reporting to EPA Toxic Release Inventory the number of pounds released
for 12 hazardous air pollutants, (carbon tetrachloride, formaldehyde,
methylene chloride, styrene, tetrachloroethylene, trichloroethylene, arsenic,
cadmium, chromium, cobalt, copper, and nickel).
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-3. Geographic-based studies assessing cancer and TCE exposure (continued)
Reference
Description
Analysis approach
Exposure assessment
Morgan and
Cassady (2002)
Incident cancer cases,
1988-1989, among
residents of 13 census
tracts in Redlands area,
San Bernardino County,
CA.
SIR for all cancer sites and 16 site-
specific cancers; expected numbers
using incidence rates of site-specific
cancer of a four-county region
between 1988-1992.
TCE and perchlorate detected in some county wells; no information on
location of wells to residents, distribution of contaminated water, or TCE
exposure potential to individual residents in studied census tracts.
Vartiainen et al.
(1993)
Total cancer and site-
specific cancer cases
(lymphoma sites and liver)
from 1953-1991 in two
Finnish municipalities.
SIR with expected number of
cancers and site-specific cancers
derived from incidence of the
Finnish population.
Monitoring data from 1992 indicated presence of TCE, tetrachloroethylene
and 1,1,1,-trichloroethane in drinking water supplies in largest towns in
municipalities. Residence in town used to infer exposure to TCE.
^§
H I
O ^
HH Oq
H TO
Cohn et al.
(1994)
Fagliano et al.
(1990)
Incident leukemia and
NHL cases, 1979-1987,
from 75 municipalities and
identified from the New
Jersey State Cancer
Registry. Histological
type classified using WHO
scheme and the
classification of NIH
Working Formulation
Group for grading NHL.
Logistic regression modeling
adjusted for age.
Monitoring data from 1984-1985 on TCE, THM, and VOCs concentrations
in public water supplies, and historical monitoring data conducted in
1978-1984.
Mallin(1990)
Incident bladder cancer
cases and deaths,
1978-1985, among
residents of 9 NW Illinois
counties.
SIR and SMR by county of
residence and zip code; expected
numbers of bladder cancers using
age-race-sex specific incidence
rates from SEER or bladder cancer
mortality rates of the United States
population from 1978-1985.
Exposure data are lacking for the study population with the exception of
noting one of two zip code areas with observed elevated bladder cancer
rates also had groundwater supplies contaminated with TCE, PCE and other
solvents.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
Table B-3. Geographic-based studies assessing cancer and TCE exposure (continued)
Reference
Description
Analysis approach
Exposure assessment
Isacson et al.
(1985)
Incident bladder, breast,
prostate, colon, lung and
rectal cancer cases
reported to Iowa cancer
registry between
1969-1981.
Age-adjusted site-specific cancer
incidence in Iowa towns with
populations of 1,000-10,000 and
who were serviced by a public
drinking water supply.
Monitoring data of drinking water at treatment plant in each Iowa
municipality with populations of 1,000-10,000 used to infer TCE and other
volatile organic compound concentrations in finished drinking water
supplies.
GIS = geographic information system, NW = Northwestern, PCE = perchloroethylene, RR = rate ratio, SEER = Surveillance, Epidemiology, and End Results,
SIR = standardized incidence ratio, SMR = standardized mortality ratio, VOCs = volatile organic compounds, WHO = World Health Organization.
TO'
^§
H I
O ^
HH Oq
H TO
O
H
W
-------
1 Category A: Study Design
2
3 • Clear articulation of study objectives or hypothesis. The ideal is a clearly stated
4 hypothesis or study objectives and the study is designed to achieve the identified
5 objectives.
6 • Selection and characterization in cohort studies of exposure and control groups and of
7 cases and controls (case-control studies) is adequate. The ideal is for selection of cohort
8 and referents from the same underlying population and differences between these groups
9 are due to TCE exposure or level of TCE exposure and not to physiological, health status,
10 or lifestyle factors. Controls or referents are assumed to lack or to have background
11 exposure to TCE. These factors may lead to a downward bias including one of which is
12 known as "healthy worker bias," often introduced in analyses when mortality or
13 incidence rates from a large population such as the U.S. population are used to derive
14 expected numbers of events. The ideal in case-control studies is cases and controls are
15 derived from the same population and are representative of all cases and controls in that
16 population. Any differences between controls and cases are due to exposure to TCE
17 itself and not to confounding factors related to both TCE exposure and disease.
18 Additionally, the ideal is for controls to be free of any disease related to TCE exposure.
19 In this latter case, potential bias is toward the null hypothesis.
20
21 Category B: Endpoint Measured
22
23 • Levels of health outcome assessed. Three levels of health outcomes are considered in
24 assessing the human health risks associated with exposure to TCE: biomarkers of effects
25 and susceptibility, morbidity, and mortality. Both morbidity as enumerated by incidence
26 and mortality as identified from death certificates are useful indicators in risk assessment
27 for hazard identification. The ideal is for accurate and predictive indicator of disease.
28 Incidence rates are generally considered to provide an accurate indication of disease in a
29 population and cancer incidence is generally enumerated with a high degree of accuracy
30 in cancer registries. Death certifications are readily available and have complete national
31 coverage but diagnostic accuracy is reduced and can vary by specific diagnosis.
32 Furthermore, diagnostic inaccuracies can contribute to death certificates as a poor
33 surrogate for disease incidence. Incidence, when obtained from population-based cancer
34 registries, is preferred for identifying cancer hazards.
35 • Changes in diagnostic coding systems for lymphoma, particularly non-Hodgkin's
36 lymphoma. Classification of lymphomas today is based on morphologic,
37 immunophenotypic, genotypic, and clinical features and is based upon the World Health
38 Organization (WHO) classification, introduced in 2001, and incorporation of WHO
39 terminology into International Classification of Disease (ICD)-0-3. ICD Versions 7 and
40 earlier had rubrics for general types of lymphatic and hematopoietic cancer, but no
41 categories for distinguishing specific types of cancers, such as acute leukemia.
42 Epidemiologic studies based on causes of deaths as coded using these older ICD
43 classifications typically grouped together lymphatic neoplasms instead of examining
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 individual types of cancer or specific cell types. Before the use of immunophenotyping,
2 these grouping of ambiguous diseases such as non-Hodgkin's lymphoma and Hodgkin's
3 lymphoma may be have misclassified. Lymphatic tumors coding, starting in 1994 with
4 the introduction of the Revised European-American Lymphoma classification, the basis
5 of the current WHO classification, was more similar to that presently used.
6 Misclassification of specific types of cancer, if unrelated to exposure, would have
7 attenuated estimate of relative risk and reduced statistical power to detect associations.
8 When the outcome was mortality, rather than incidence, misclassification would be
9 greater because of the errors in the coding of underlying causes of death on death
10 certificates (IOM, 2003). Older studies that combined all lymphatic and hematopoietic
11 neoplasms must be interpreted with care.
12
13 Category C: TCE-Exposure Criteria
14
15 • Adequate characterization of exposure. The ideal is for TCE exposure potential known
16 for each subject and quantitative assessment (job-exposure-matrix approach) of TCE
17 exposure assessment for each subject as a function of job title, year exposed, duration,
18 and intensity. Consideration of job task as additional information supplementing job title
19 strengthens assessment increases specificity of TCE assignment. The assessment
20 approach is accurate for assigning TCE intensity (TCE concentration or a time-weighted
21 average) to individual study subjects and estimates of TCE intensity are validated using
22 monitoring data from the time period. The objective for cohort and case-controls studies
23 is to differentiate TCE exposed subjects from subjects with little or no TCE exposure. A
24 variety of dose metrics may be used to quantify or classify exposures for an
25 epidemiologic study. They include precise summaries of quantitative exposure,
26 concentrations of biomarkers, cumulative exposure, and simple qualitative assessments of
27 whether exposure occurred (yes or no). Each method has implicit assumptions and
28 potential problems that may lead to misclassification. Exposure assessment approaches
29 in which it was unclear that the study population was actually exposed to TCE are
30 considered inferior since there may be a lower likelihood or degree of exposure to study
31 subjects compared to approaches which assign known TCE exposure potential to each
32 subject.
33
34 Category D: Follow-up (Cohort)
35
36 • Loss to follow-up. The ideal is complete follow-up of all subjects; however, this is not
37 achievable in practice, but it seems reasonable to expect loss to follow-up not to exceed
38 10%. The bias from loss to follow-up is indeterminate. Random loss may have less
39 effect than if subjects who are not followed have some significant characteristics in
40 common.
41 • Follow-up period allows full latency period for over 50% of the cohort. The ideal to
42 follow all study subjects until death. Short of the ideal, a sufficient follow-up period to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-23 DRAFT—DO NOT CITE OR QUOTE
-------
1 allow for cancer induction period or latency over 15 or 20 years is desired for a large
2 percentage of cohort subjects.
3
4 Category E: Interview Type (Case-control)
5
6 • Interview approach. The ideal interviewing technique is face-to-face by trained
7 interviewers with more than 90% of interviews with cases and control subjects conduced
8 face-to-face. The effect on the quality of information from other types of data collection
9 is unclear, but telephone interviews and mail-in questionnaires probably increase the rate
10 of misclassification of subject information. The bias is toward the null hypothesis if the
11 proportion of interview by type is the same for case and control, and of indeterminate
12 direction otherwise.
13 • Blinded interviewer. The ideal is for the interviewer to be unaware whether the subject is
14 among the cases or controls and the subject to be unaware of the purpose and intended
15 use of the information collected. Blinding of the interviewer is generally not possible in a
16 face-to-face interview. In face-to-face and telephone interviews, potential bias may arise
17 from the interviewer expects regarding the relationship between exposure and cancer
18 incidence. The potential for bias from face-to-face interviews is probably less than with
19 mail-in interviews. Some studies have assigned exposure status in a blinded manner
20 using a job-exposure matrix and information collected in the unblinded interview. The
21 potential for bias in this situation is probably less with this approach than for nonblinded
22 assignment of exposure status.
23
24 Category F: Proxy Respondents
25
26 • Proxy respondents. The ideal is for data to be supplied by the subject because the subject
27 generally would be expected to be the most reliable source; less than 10% of either total
28 cases or total controls for case-control studies. A subject may be either deceased or too
29 ill to participate, however, making the use of proxy responses unavoidable if those
30 subjects are to be included in the study. The direction and magnitude of bias from use of
31 proxies is unclear, and may be inconsistent across studies.
32
33 Category G: Sample Size
34
35 • The ideal is for the sample size is large enough to provide sufficient statistical power to
36 ensure that any elevation of effect in the exposure group, if present, would be found, and
37 to ensure that the confidence bounds placed on relative risk estimates can be
38 well-characterized.
39
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 Category H: Analysis Issues
2
3 • Control for potentially confounding factors of importance in analysis. The ideal in cohort
4 studies is to derive expected numbers of cases based on age-sex- and time-specific cancer
5 rates in the referent population and in case-control studies by matching on age and sex in
6 the design and then adjusting for age in the analysis of data. Age and sex are likely
7 correlated with exposure and are also risk factors for cancer development. Similarly,
8 other factors such as cigarette smoking and alcohol consumption are risk factors for
9 several site-specific cancers reported as associative with TCE exposure. To be a
10 confounder of TCE, exposure to the other factor must be correlated, and the association
11 of the factor with the site-specific cancer must be causal. The expected effect from
12 controlling for confounders is to move the estimated relative risk estimate closer to the
13 true value.
14 • Statistical methods are appropriate. The ideal is that conclusions are drawn from the
15 application of statistical methods that are appropriate to the problem and accurately
16 interpreted.
17 • Evaluation of exposure-response. The ideal is an examination of a linear
18 exposure-response as assessed with a quantitative exposure metric such as cumulative
19 exposure. Some studies, absent quantitative exposure metrics, examine exposure
20 response relationships using a semiquantitative exposure metric or by duration of
21 exposure. A positive dose-response relationship is usually more convincing of an
22 association as causal than a simple excess of disease using TCE dose metric. However, a
23 number of reasons have been identified for a lack of linear exposure-response finding and
24 the failure to find such a relationship means little from an etiological viewpoint and does
25 not minimize an observed association with overall TCE exposure.
26 • Documentation of results. The ideal is for analysis observations to be completely and
27 clearly documented and discussed in the published paper, or provided in supplementary
28 materials accompanying publication.
29
30 B.2.1. Study Designs and Characteristics
31 The epidemiologic designs investigating TCE exposure and cancer include cohort studies
32 of occupationally exposure populations, population case-control studies, and geographic studies
33 of residents in communities with TCE in water supplies or ambient air. Analytical
34 epidemiologic studies, which include case-control and cohort designs, are generally relied on for
35 identifying a causal association between human exposure and adverse health effects (U.S. EPA,
36 2005) due to their clear ability to show exposure precedes disease occurrence. In contrast,
37 ecologic studies such as health surveys of cancer incidence or mortality in a community during a
38 specified time period, i.e., geographic-based studies identified in Appendix B, Table B-3,
39 provide correlations between rates of cancer and exposure measured at the geographic level.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-25 DRAFT—DO NOT CITE OR QUOTE
-------
1 An epidemiologic study's ability to inform a question on TCE and cancer depends on
2 clear articulation of study objective or hypothesis and adequate selection of exposed and control
3 group in cohort studies and cases and controls in case-control studies are important. As the body
4 of evidence on trichloroethylene has grown over the past 20 years, so has the number of studies
5 with clearly articulated hypothesis. All Nordic cohort studies (Axelson et al., 1994; Anttila et al.,
6 1995; Hansen et al., 2001; Raaschou-Nielsen et al., 2003) are designed to examine cancer and
7 TCE, albeit some with limited statistical power, as are recent cohort studies of United States
8 occupationally exposed populations (Ritz, 1999a; Blair et al., 1998; Morgan et al., 1998; Boice et
9 al., 1999, 2006a; Zhao et al., 2005; Radican et al, 2008). Exposure assessment approaches in
10 these studies distinguished subjects with varying potentials for TCE exposure, and in some cases,
11 assigned a semi quantitative TCE exposure surrogate to individual study subjects. Three case-
12 control studies nested in cohorts, furthermore, examined TCE exposure and site-specific cancer,
13 albeit a subject's potential and overall prevalence of TCE exposure greatly varied between these
14 studies (Wilcosky et al., 1984; Greenland et al., 1994; Krishnadasan et al., 2007). Typically,
15 studies of all workers at a plant or manufacturing facility (Shindell and Ulrich, 1985; Shannon et
16 al., 1988; Blair et al., 1989; Sinks et al., 1992; Garabrant et al., 1988; Costa et al., 1989; ATSDR,
17 2004; Chang et al., 2003, 2005; Sung et al., 2007, 2008; Clapp and Hoffman, 2008) are not
18 designed to evaluate cancer and TCE specifically, given their inability to identify varying TCE
19 exposure potential for individual study subjects; rather, such studies evaluate the health status of
20 the entire population working at that facility. Bias associated with exposure misclassification is
21 greater in these studies, and for this and other reasons more fully discussed below, they are of
22 limited utility for informing evaluations on TCE exposure and cancer.
23 Recent case-control studies with hypotheses specific for TCE exposure include the
24 kidney cancer case-control studies of Vamvakas et al. (1998), Briining et al. (2003), and
25 Charbotel et al. (2006, 2009). More common, population-based case-control studies assess
26 occupational exposure to organic solvents, using a job-exposure matrix approach for exposure
27 assessment to examine organic solvent categories, i.e., aliphatic hydrocarbons, or specific
28 solvents such as TCE. The case-control studies of Costas et al. (2002; childhood leukemia) and
29 Lee et al. (2003; liver cancer) were also designed to examine possible association with
30 contaminated drinking water containing trichloroethylene and other solvents detected at lower
31 concentrations. The hypothesis of Siemiatycki (1991) and ancillary publications (Siemiatycki et
32 al., 1994; Fritschi and Siemiatycki, 1996a, b; Dumas et al., 2000; Parent et al., 2000a, b;
33 Goldberg et al., 2001) explored possible association between 20 site-specific cancers and
34 occupational title or chemical exposures, including TCE exposure, using a contemporary
35 exposure assessment approach for more focused research investigation.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-26 DRAFT—DO NOT CITE OR QUOTE
-------
1 Cases and control selection in most population-based case-control studies of TCE
2 exposure are considered a random sample and representative of the source population
3 (Siemiatycki, 1991 [and related publications, Siemiatycki et al., 1994; Aronson et al., 1996;
4 Fritchi and Siemiatycki, 1996a, b; Dumas et al., 2000; Parent et al., 2000a, b; Goldberg et al.,
5 2001]; Lowengart et al., 1987; McKinney et al., 1991; Hardell et al., 1994; Heineman et al.,
6 1994; Nordstrom et al., 1998; Dosemeci et al., 1999; Kernan et al., 1999; Persson and
7 Fredriksson, 1999; Pesch et al., 2000a, b; De Roos et al., 2001; Costas et al., 2002; Briining et
8 al., 2003; Lee et al., 2003; Shu et al., 2004; Charbotel et al., 2006, 2009; Miligi et al., 2006;
9 Seidler et al., 2007; Constantini et al., 2008; Wang et al., 2009]). Case and control selection in
10 Vamvakas et al. (1998), a study conducted in the Arnsberg area of Germany, is subject to
11 criticism regarding possible selection bias resulting from differences in selection criteria, cases
12 worked in small industries and controls from a wider universe of industries; differences in age,
13 controls being younger than cases with possible lower exposure potentials; and temporal
14 difference in case and control selection, controls selected only during the last year of the study
15 period with possible lower exposure potential if exposure has decreased over period of the study
16 (NRC, 2006). The potential for selection bias in Briining et al. (2003), another study in the same
17 area as Vamvakas et al. (1998) but of later period of observation, was likely reduced compared to
18 Vamvakas et al. (1998) due to the broader region of southern Germany from which cases were
19 identified and interviewing cases and controls during the same time. One case-control study
20 nested in a cohort (Greenland et al., 1994) included subjects whose deaths were reported to and
21 known by the employer, e.g., occurred among vested or pensioned employees or among
22 currently employees. A 10- to 15-year employment period was required for employees in this
23 study to receive a pension; deaths among employees who left employment before this time were
24 not known to the employer and not included the study. Survivor bias, a selection bias, may be
25 introduced by excluding nonpensioned workers or those who leave employment before
26 becoming vested in a company's retirement plan is more likely than in a study of all employees
27 with complete follow-up. The use of pensioned deaths as controls, as was done in this study,
28 would reduce potential bias if both cases and control had the same likelihood of becoming
29 pensioned. That is, the probability for becoming a pensioned worker is similar for all deaths and
30 unrelated to the likelihood of exposure or magnitude of exposure and disease. No information
31 was available in Greenland et al. (1994) to evaluate this assumption.
32 Geographic-based and ecological studies of TCE contaminated water supplies typically
33 focus on estimating cancer or other disease rates in geographically circumscribed populations
34 who are geospatially located with a source containing TCE, e.g., a hazardous waste site, well
35 water, or air. These studies are often less informative for studying cancer due to their inability to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimate incidence rate ratios, essential for causal inferences, inferior exposure assessment
2 approach, and to possible selection biases. Ecological studies also are subject to bias known as
3 "ecological fallacy" since variables of exposure and outcome measured on an aggregate level do
4 not represent association at the individual level. Consideration of this bias is important for
5 diseases with more than one risk factor, such as the site-specific cancers evaluated in this
6 assessment.
7
8 B.2.2. Outcomes Assessed in Trichloroethylene (TCE) Epidemiologic Studies
9 The epidemiologic studies consider at least three levels of health outcomes in their
10 examinations of human health risks associated with exposure to trichloroethylene: biomarkers of
11 effects and susceptibility, morbidity, and mortality (NRC, 2006). Few susceptibility biomarkers
12 have been examined and these are not specific to trichloroethylene (NRC, 2006). By far, the
13 bulk of the literature on cancer and trichloroethylene exposure is of cancer morbidity (Isacson et
14 al., 1985; Lowengart et al., 1987; Shannon et al., 1988; Fredriksson et al., 1989; AZ DHS, 1990,
15 1995; McKinney et al., 1991; Siemiatycki, 1991; Persson et al., 1993; Persson and Fredriksson,
16 1999; Vartiainen et al., 1993; Axelson et al., 1994; Cohn et al., 1994; Hardell et al., 1994; Anttila
17 et al., 1995; Nordstrom et al., 1998; Vamvakas et al., 1998; Dosemeci et al., 1999; Dumas et al.,
18 2000; Pesch et al., 2000a, b; De Roos et al., 2001; Hansen et al., 2001; Costas et al., 2002;
19 Morgan and Cassady, 2002; Briining et al., 2003; Rasschou-Nielsen et al., 2003; Aickin, 2004;
20 Shu et al., 2004; Coyle et al., 2005; ATSDR, 2006a; Charbotel et al., 2006, 2009; Miligi et al.,
21 2006; Seidler et al., 2007; Sung et al., 2008; Wang et al., 2009), mortality (Wilcosky et al., 1984;
22 Shindell and Ulrich, 1985; Garabrant et al., 1988; Blair et al., 1989; Costa et al., 1989; Kernan et
23 al., 1999; Aickin et al., 1992; Greenland et al., 1994; Heineman et al., 1994; Morgan et al., 1998;
24 Boice et al., 1999, 2006a; Ritz, 1999a; Lee et al., 2003; ATSDR, 2004;; Clapp and Hoffman,
25 2008, Radican et al, 2008) or both (Sinks et al., 1992; Henschler et al., 1995; Blair et al., 1998;
26 Chang et al., 2003, 2005; Sung et al., 2007; Zhao et al., 2005).
27 Mortality is readily identified from death certificates; however, diagnostic accuracy from
28 death certificates varies by the specific diagnosis (Brenner and Gefeller, 1993). Incident cancer
29 cases are enumerated more accurately by tumor registries and by hospital pathology records and
30 cases identified from these sources are considered to have less bias resulting from disease
31 misclassification than cause or underlying cause of death as noted on death certificates. Studies
32 of incidence are preferred, particularly for examining association with site-specific cancers
33 having high 5-year survival rates or which may be misclassified on death certificate.
34 Misclassification of the cause of death as noted on death certificates attenuates statistical power
35 through errors of outcome identification. This nondifferential misclassification of outcome in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 cohort studies will lead to attenuation of rate ratios, although the magnitude of is difficult to
2 predict (NRC, 2006). Cancer registries are used for cases diagnosed in more recent time periods
3 and cohorts whose entrance dates are 30 or 40 years may miss many incident cancers and
4 reduced statistical power as a consequence. Two studies examine both cancer incidence and
5 mortality (Blair et al., 1998; Zhao et al., 2005). The lapse of 20 or more years in Blair et al.
6 (1998) and 38 years in Zhao et al. (2005) between date of cohort identification and cancer
7 incidence ascertainment suggests these studies are missing cases and limits incidence
8 examinations.
9
10 B.2.3. Disease Classifications Adopted in Trichloroethylene (TCE) Epidemiologic Studies
11 Disease coding and changes over time are important in epidemiologic evaluations,
12 particularly in evaluation of heterogeneity or consistency of observations from a body of
13 evidence. The ICD, published by WHO, is used to code underlying and contributing cause of
14 death on death certificates and is updated periodically, adding to diagnostic inconsistency for
15 cross-study comparisons (NRC, 2006). Tumor registries use the International Classification of
16 Diseases-Oncology (ICD-O) for coding the site and the histology of neoplasms, principally
17 obtained from a pathology report.
18 The epidemiologic studies of TCE exposure have used a number of different
19 classification systems (Scott and Chiu, 2006). A number of studies classified neoplasms
20 according to ICD-O (Siemiatycki, 1991; Costas et al., 2002) or to ICD-9 (Nordstrom et al., 1998;
21 Kernan et al., 1999; Ritz, 1999a; Chang et al., 2005; Zhao et al., 2005). Other ICD revisions
22 used in recent studies include ICDA-8 (Blair et al., 1989; Greenland et al., 1994; Blair et al.,
23 1998), ICD-7 (Axelson et al., 1994; Anttila et al., 1995; Hansen et al., 2001; Raaschou-Nielsen et
24 al., 2003), or several ICD revisions, whichever was in effect at the date of death (Garabrant et al.,
25 1988; Morgan et al., 1998; Boice et al., 1999, 2006a; Radican et al., 2008). In this latter case,
26 changes in disease classification over revisions are not harmonized or receded to a common
27 classification; and, diagnostic inconsistencies and disease misclassification errors leads to a
28 greater likelihood for bias in these studies. Greatest weight is placed on studies where all cases
29 or deaths are classified using current classification systems. However, association in studies
30 adopting older revisions, ICD 7 (Axelson et al., 1994; Anttila et al., 1995; Hansen et al., 2001;
31 Raaschou-Nielsen et al., 2003), for example, is noteworthy given the narrow consideration of
32 lymphoid neoplasms compared to contemporary classification systems. Consistency
33 examinations of the overall body of evidence using meta-analysis methods and examination of
34 heterogeneity will need to consider study differences in coding in interpreting findings.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 A major shift in thinking occurred around 1995 with the Revised European-American
2 Lymphoma (REAL) classification of grouping diseases of the blood and lymphatic tissues along
3 their cell lines compared to previous approaches to group lymphomas by a cell's physical
4 characteristics. It was increasing recognized that some lymphomas and corresponding lymphoid
5 leukemias were different phases (solid and circulating) of the same disease entity (Morton et al.,
6 2007). Many concepts of contemporary knowledge of lymphomas are incorporated in the WHO
7 Classification of Neoplastic Diseases of the Hematopoietic and Lymphoid Tissues, an
8 international consensus scheme for classifying leukemia and lymphoma now in use and the
9 predecessor to REAL (Jaffe et al., 2001). Both the ICD-O, 3rd edition, and ICD-10 have adopted
10 the WHO classification framework.
11 The only study coding lymphomas using the WHO classification is Seidler et al. (2007).
12 Other lymphoma studies have adopted older lymphoma classification systems, either the
13 National Cancer Institute's (NCI) Working Formulation (Miligi et al., 2006; Costantini et al.,
14 2008) or other systems coding lymphomas according to NCI's Working Formulation, i.e.,
15 International Classification of Disease-Oncology, 2nd Edition (Wang et al., 2009), that divided
16 lymphomas into low-grade, intermediate-grade and high grade, with subgroups based on cell
17 type and presentation, or Rappaport (Hardell et al., 1994, 1981), with groupings based on
18 microscopic morphology (Lymphoma Information Network, 2008). Lowengart et al. (1987),
19 Persson et al. (1989, 1993), McKinney et al. (1991) nor Persson and Fredriksson (1999) provide
20 information in their published articles on lymphomas classification systems used in these studies.
21 Implications of classification changes are most significant for lymphoma. As noted by
22 the IOM (2003), in Revision 7 and earlier editions of the ICD, all lymphatic and hematopoietic
23 neoplasms were grouped together instead of treated as individual types of cancer (such as
24 Hodgkin's disease) or specific cell types (such as acute lymphocytic leukemia). One limitation
25 of this treatment was the amalgamation of these relatively rare cancers would increase the
26 apparent sample size but could also result in diluted estimates of effect if etiologic heterogeneity
27 of different lymphoma subtypes existed, i.e., different sites of cancer were not associated in
28 similar ways with the exposures of interest. Additionally, immunophenotyping was not
29 available, leading to decreased ability to distinguish ambiguous diseases, and diagnoses of these
30 cancers may have been misclassified; for example, NHL may have been grouped with other
31 lymphatic and hematopoietic cancers to increase statistical power or misclassified as Hodgkin's
32 disease, for example. Examination of distinct lymphoma subtypes is expected to reduce disease
33 misclassification bias. Two case-control studies on non-Hodgkin's lymphoma (NHL) include
34 analysis of lymphoma subtype and trichloroethylene exposure (Miligi et al., 2006; Seidler et al.,
35 2007).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-30 DRAFT—DO NOT CITE OR QUOTE
-------
1 A change in liver cancer coding occurred between ICDA-8 and ICD-9 and is important to
2 consider in examinations of liver cancer observations across the TCE studies. With ICD-9, liver
3 cancer "not specified as primary or secondary" was moved from the grouping of secondary
4 malignant neoplasms and added to the larger class of malignant liver neoplasms. Thus, a similar
5 grouping of liver cancer causes is necessary to cross-study comparisons. For example, an
6 examination of liver cancer, based on ICDA-8, would need to include codes for liver and
7 intrahepatic bile duct (code 155) and liver, not specified as primary or secondary (code 197.8),
8 but, for ICD-9, would include liver and intrahepatic bile duct (code 155) only. The effect of
9 adding "liver cancer, not specified as primary or secondary" to the larger liver and intrahepatic
10 bile duct category in ICD-9 was a 2-fold increase in the overall liver cancer mortality rate (Percy
11 etal., 1990).
12
13 B.2.4. Exposure Classification
14 Adequacy of exposure assessment approaches and their supporting data are a critical
15 determinant of a study's contribution in a weight-of-evidence evaluation (Checkoway et al.,
16 1989). Exposure assessment approaches in studies of TCE and cancer vary greatly. At one
17 extreme, studies assume subjects are exposed by residence in a defined geographic area (Isacson
18 et al., 1985; AZ DHS, 1990, 1995; Aickin et al., 1992, Aickin, 2004; Vartiainen et al., 1993;
19 Cohn et al., 1994; Morgan and Cassidy, 2002; Lee et al., 2003; Coyle et al., 2005; ATSDR,
20 2006a, 2008) or by employment in a plant or job title (Shindell and Ulrich, 1985; Garabrant et
21 al., 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Chang et al., 2003, 2005;
22 ATSDR, 2004; Sung et al., 2007, 2008; Clapp and Hoffman, 2008). This is a poor exposure
23 surrogate given potential for TCE exposure can vary in these broad categories depending on job
24 function, year, use of personal protection, and, for residential exposure, pollutant fate and
25 transport, water system distribution characteristics, percent of time per day in residence, presence
26 of mitigation devices, drinking water consumption rates, and showering times. Another example
27 comprises measurement from a subset of workers with jobs where TCE is routinely used to infer
28 TCE exposure and TCE intensity to all subjects. In both examples, exposure misclassification
29 potential may be extensive and with a downward bias in risk estimates.
30 At the other extreme and preferred given a reduced likelihood for misclassification bias,
31 quantitative exposure assessment based upon a subject' s job history, job title, and monitoring
32 data are used to develop estimates of TCE intensity and cumulative exposure (quantitative
33 exposure metrics or measures) and is know as job-exposure matrix (JEM) approaches. Peak
34 exposure is also well characterized. Addition to JEM approaches of information on job tasks
35 (JTEM) associated with exposure such as that done by Pesch et al. (2000a, b) is expected to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 reduce potential exposure misclassification. In between these two extremes, semi quantitative
2 estimates of low, medium, and high TCE exposure are assigned to subjects. Twelve studies
3 assigned a quantitative or semiquantitive TCE surrogate metrics to individual subjects using a
4 JEM or job-task-exposure-matrix (ITEM): Siemiatycki (1991 [and related publications,
5 Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and Siemiatycki, 1996a, b; Dumas et al.,
6 2000; Parent et al., 2000a, b; Goldberg et al., 2001]), Blair et al. (1998) and follow-up by
7 Radican et al. (2008), Morgan et al. (1998), Vamvakas et al. (1998), Kernan et al. (1999), Ritz
8 (1999a), Pesch et al. (2000a, b), Briining et al. (2003), Zhao et al. (2005), Charbotel et al. (2006,
9 2009), Krishnadansen et al. (2007), Seidler et al. (2007), and Wang et al. (2009).
10 Fifteen other studies assigned a qualitative TCE surrogate metric (ever exposed or never
11 exposed), less preferred to a semi-quantitative exposure surrogate given greater likelihood for
12 error associated exposure misclassification, using general job classification of job title by
13 reference to industrial hygiene records indicating a high probability of TCE use, individual
14 biomarkers, job exposure matrices, water distribution models, for cohort studies, or obtained
15 from subjects using questionnaire for case-control studies. The 15 studies were: Wilcosky et al.
16 (1984), Lowengart et al. (1987), McKinney et al. (1991), Greenland et al. (1994), Hardell et al.
17 (1994), Nordstrom et al. (1998), Shu et al. (1999, 2004), Boice et al. (1999, 2006a), Dosemeci et
18 al. (1999), Persson and Fredriksson (1999), Costas et al. (2002), Raaschou-Nielsen et al. (2003),
19 Miligi et al. (2006), and Costantini et al. (2008). Without quantitative measures, however, it is
20 not possible to quantify exposure difference between groupings nor is it possible to compare
21 similarly named categories across studies. Exposure misclassification for dichotomous exposure
22 defined in these studies, if nondifferential, would downward bias resulting risk estimates.
23 Zhao et al. (2005), Krishnadansen et al. (2007), and Boice et al. (2006a) are studies with
24 overlap in some subjects, but with different exposure assessment approaches, more fully
25 discussed in B.3.1.1., with implication on study ability to identify cancer hazard. While these
26 studies used job title to assign TCE exposure potential, Zhao et al. (2005) and Krishnadansen et
27 al. (2007) developed a semi quantitative estimate of TCE exposure potential, whereas, Boice et
28 al. (2006a) classified subjects as either "exposed" or "unexposed" using a qualitative surrogate.
29 These studies, furthermore, identify TCE exposure potentially differently for possibly similar job
30 titles. For example, jobs as instrument mechanics, inspectors, test stand engineers, and research
31 engineers are identified with medium potential exposure in Zhao et al. (2005) and Krishnadansen
32 et al. (2007); however, these job titles were considered in Boice et al. (2006a) as having
33 background exposure and were combined with unexposed subjects, the referent population in
34 Cox Proportional Hazard analyses.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-32 DRAFT—DO NOT CITE OR QUOTE
-------
1 Three Nordic cohorts have TCE exposure as indicated from biological markers, assigning
2 TCE exposure to subjects using either concentration of trichloroacetic acid (TCA) in urine or
3 TCE in blood (Axelson et al., 1994; Anttila et al., 1995; Hansen et al., 2001). The utility of a
4 biomarker depends on it selectivity and the exposure situation. Urinary TCA (U-TCA) is a
5 nonselective marker since other chlorinated solvents besides TCE are metabolized to TCA and
6 resultant urinary elimination. If only TCE is the only exposure, urinary TCE may be a useful
7 marker; however, in setting with mixed exposure, urinary TCA may serve as an integrated
8 exposure marker of several chlorinated solvents. The Nordic studies used the linear relationship
9 found for average inhaled trichloroethylene versus U-TCA: trichloroethylene (mg/m3) = 1.96;
10 U-TCA (mg/L) = 0.7 for exposures lower than 375 mg/m3 (69.8 ppm) (Ikeda et al., 1972). This
11 relationship shows considerable variability among individuals, which reflects variation in urinary
12 output and activity of metabolic enzymes. Therefore, the estimated inhalation exposures are
13 only approximate for individuals but can provide reasonable estimates of group exposures.
14 There is evidence of nonlinear formation of U-TCA above about 400 mg/m3 or 75 ppm of
15 trichloroethylene. The half-life of U-TCA is about 100 hours. Therefore, the U-TCA value
16 represents roughly the weekly average of exposure from all sources, including skin absorption.
17 The Ikeda et al. (1972) relationship can be used to convert urinary values into approximate
18 airborne concentration, which can lead to misclassification if tetrachloroethylene and
19 1,1,1-trichloroethane are also being used because they also produce U-TCA. In most cases, the
20 Ikeda et al. (1972) relationship provides a rough upper boundary of exposure to
21 trichloroethylene.
22
23 B.2.5. Follow-up in Trichloroethylene (TCE) Cohort Studies
24 Cohort studies are most informative if vital status is ascertained for all cohort subjects
25 and if the period of time for disease ascertainment is sufficient to allow for long latencies,
26 particularly for cancer detection and death, in the case of mortality studies. Inability to ascertain
27 vital status for all subjects, or, conversely, subjects who are loss-to-follow-up, can affect the
28 validity of observations and lead to biased results. Both power and rate ratios estimated in
29 cohort studies can be underestimated due to bias introduced if the follow-up period was not long
30 enough to account for latency (NRC, 2006). The probability of loss to follow-up may be related
31 to exposure, disease, or both. The multiple-stage process of cancer development occurs over
32 decades after first exposure and studies with full latent periods are considered to provide greater
33 weight to the evaluation compared to cohort studies with shortened follow-up period and lower
34 percentage of subjects whose vital status was known on the date follow-up ended. Vital status
35 ascertainment for over 90% of all cohort studies and long mean follow-up periods, say 15 years
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 of longer, characterized many occupational cohort studies on trichloroethylene and cancer
2 (Garabrant et al., 1988; Costa et al., 1989; Anttila et al., 1995; Blair et al., 1998 and the
3 follow-up study of Radican et al., 2008; Morgan et al., 1998; Boice et al., 1999, 2006a; Ritz,
4 1999a; Raaschou-Nielsen et al., 2003; Zhao et al., 2005). Information is lacking in two
5 biomarker studies (Axelson et al., 1994; Hansen et al., 2001), additionally, to estimate the mean
6 follow-up period for TCE-exposed subjects; although, Hansen et al. (2001) state "some workers
7 were followed for as long as 50 years after their exposure, which allowed the detection of
8 cancers with long latency periods." Other studies of trichloroethylene and cancer did not
9 identify a latent period, information for calculating a latent period, or contained other
10 deficiencies in follow-up criteria (Wilcosky et al., 1984; Shannon et al., 1988; Blair et al., 1989;
11 Costa et al., 1989; Sinks et al., 1992; Henschler et al., 1995; Chang et al., 2005; Sung et al.,
12 2007). Proportionate mortality ratio studies, based only on deaths and which lack information on
13 person-year structure as cohort studies, by definition, do not contain information on cancer latent
14 periods or follow-up (ATSDR, 2004; Clapp and Hoffman, 2008).
15
16 B.2.6. Interview Approaches in Case-Control Studies of Cancer and Trichloroethylene
17 (TCE) Exposure
18 Interview approaches and the percentage of subjects with information obtained from
19 proxy or next-of-kin respondents need consideration in interpreting population and hospital -
20 based case-control studies in light of possible biases. Biases resulting from proxy respondent or
21 from low participation related to mailed questionnaires are not relevant to cohort or geographic
22 studies since information is obtained from local, national, or corporate records. Both face-to-
23 face and telephone interviews are common and valid approaches used in population or
24 hospital-based case-control studies. Important to each is the use of a structured questionnaires
25 combined with intensive training as ways to minimize a high potential for biases often associated
26 with mailed questionnaires (Schlesselman, 1982; Blatter et al., 1997). Studies with information
27 limited to job title, type of business and dates of employment and aided with computer or
28 job-exposure-matrix approaches are preferred to studies of job title only; the added approaches
29 can reduce exposure misclassification bias and improve disease risk estimates (Stewart et al.,
30 1996). Moreover, interview with respondents other than the individual case or control, through
31 proxy or next-of-kin respondents, may also introduce bias in case-control studies. Proxy
32 respondents are used when cases or control are either too sick to respond or if deceased. This
33 bias would dampen observed associations if proxy respondents did not fully provide accurate
34 information. Boyle et al. (1992), for example, in their study of several site-specific cancers and
35 occupational exposures found low sensitivity, or correct reporting, for occupational exposure to
36 solvents among proxy respondents. The weight of evidence analysis on trichloroethylene and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-34 DRAFT—DO NOT CITE OR QUOTE
-------
1 cancer, for this reason, places greatest weight on observations from studies which obtain
2 information on personal, medical, and occupational histories from each case and control with
3 lesser weight is placed on studies where 10 percent or more of interviews are with proxy
4 respondents.
5 Many of the more recent case-control studies include face-to-face (McKinney et al.,
6 1991; Siemiatycki, 1991; Vamvakas et al., 1998; Dosemeci et al., 1999; Costas et al., 2002;
7 Pesch et al., 2000a, b; Briining et al., 2003; Miligi et al., 2006; Seidler et al., 2007; Wang et al.,
8 2009) or telephone (Lowengart et al., 1987; Shu et al., 1999, 2004; Charbotel et al., 2006, 2009)
9 interviews. Few of these studies included interviewers who were blinded or did not know the
10 identity of who is a case and who is a control; although, many studies assigned exposure to cases
11 and controls in a blinded manner. Information obtained from mailed questionnaire
12 predominantly characterized older Nordic studies (Hardell et al., 1981, 1994; Fredriksson et al.,
13 1989; Persson et al., 1989, 1993; Persson and Fredriksson, 1999; Nordstrom et al., 1998). One
14 case-control study did not ascertain information from a questionnaire or through interviews,
15 instead using occupation coded on death certificates to infer TCE exposure potential (Kernan et
16 al., 1999). In all studies except Costas et al. (2002) and Kernan et al. (1999), assignment of
17 potential TCE exposure to cases and controls, to different degrees depending on each study, is
18 based on self-reported information on job title, and in some cases, to specific chemicals.
19 More common to the case-control studies on trichloroethylene and cancer was possible
20 bias related to a higher percentage of proxy interviews. Four studies (Dosemeci et al., 1999;
21 Pesch et al., 2000a, b; Wang et al., 2009) excluded subjects with proxy interviews and the
22 percentage of proxy interview among subjects in one other study is less than 10 percent
23 (Nordstrom et al., 1998). Charbotel et al. (2006, 2009) furthermore presents analyses for data
24 they considered as better quality, including higher confidence exposure information and
25 excluding proxy respondents, in addition to analyses using both living and proxy respondents. A
26 consideration of proxy interviews in studies of childhood cancers which include an examination
27 of paternal occupational exposure is needed given a greater likelihood for bias if fathers are not
28 directly interviewed and the father's occupational information is provided only by the child's
29 mother. A good practice is for statistical analyses examining paternal occupational exposure to
30 included only cases and controls with direct information provided by the fathers, such as
31 De Roos et al. (2001), the only childhood cancer study (neuroblastoma) to exclude the use of
32 proxy information.
33
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.2.7. Sample Size and Approximate Statistical Power
2 Cancer is generally considered a rare disease compared to more common health outcomes
3 such as cardiovascular disease. Of all site-specific cancers, endocrine cancers of the breast
4 prostate and lung cancer are most common, with age-adjusted incidence rates of 126.0 per
5 100,000 women (breast), 163 per 100,000 men (prostate), and 63.9 per 100,000 men and women
6 (lung) (Ries et al., 2008). Several site-specific cancers including kidney cancer, liver cancer, and
7 lymphoma that are of interest to trichloroethylene are rarer and consideration of study size and
8 the influence on statistical power are factors for judging a study's validity and assessment of a
9 study's contribution to the overall weight-of-evidence for identifying a hazard. For example, the
10 age-adjusted incidence rates of non-Hodgkin's lymphoma, liver and intrahepatic bile duct
11 cancer, and kidney and renal pelvis cancer in the United States population are 19.5 per 100,000,
12 6.4 per 100,000, and 13.2 per 100,000; rates vary by sex and race. Age-adjusted mortality rates
13 for these cancers are lower: 7.3 per 100,000 (NHL), 5.0 per 100,000 (liver and intrahepatic bile
14 duct), 4.2 per 100,000 (kidney and renal pelvis). Rates of the childhood cancer, acute
15 lymphocytic leukemia, are even lower: 1.6 (incidence) and 0.5 (mortality) per 100,000 (Ries et
16 al., 2008).
17 Only very large cohort or case-control studies would have a sufficient number of cases
18 and statistical power to estimate excess risks and exposure-response relationships (NRC, 2006).
19 Observations from studies with large numbers of TCE-exposed subjects, given consideration of
20 exposure conditions and other criteria discussed in this section, can provide useful information
21 on hazard and may provide quantitative information on possible upper bound trichloroethylene
22 cancer risks. Alternatively, studies of small numbers of subjects or cases and controls, typically,
23 studies with statistical power less than 80% to detect risk of a magnitude of 2 or less, are not
24 likely to provide useful evidence for or against the hypothesis that trichloroethylene is a human
25 carcinogen.
26 Studies with either a large number of TCE-exposed subjects or with large numbers of
27 total deaths, cancer deaths, or cancer cases among TCE-exposed subjects are the cohort studies
28 of Blair et al. (1998), Raaschou-Nielsen et al. (2003), and Zhao et al. (2005), and the case-control
29 studies of Pesch et al. (2000a), Shu et al. (1999, 2004 [paternal exposure assessment, only]),
30 Miligi et al. (2006), and Seidler et al. (2007). The cohorts of Boice et al. (1999, 2006a) and
31 Morgan et al. (1998), like that of Blair et al. (1998), comprised over 10,000 subjects both with
32 and without potential TCE exposure; however, the number of subjects and the percentage of the
33 larger cohort identified with TCE exposure in these studies was less than that in Blair et al.
34 (1998); 23% of all subjects in Morgan et al. (1998), 3% in Boice et al. (1999), 2% in Boice et al.
35 (2006a) compared to 50% in Blair et al. (1998). Moreover, although the cohorts of Garabrant et
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-36 DRAFT—DO NOT CITE OR QUOTE
-------
1 al. (1988), Chang et al. (2005) and Sung et al. (2007) are also of population sizes greater than
2 10,000, these studies of employees at one manufacturing facility lack assignment of potential
3 TCE exposure to individual subjects and include subjects with varying exposure potential, some
4 of whom are likely with very low to no exposure potential to TCE. Rate ratios estimated from
5 cohorts that include unexposed subjects would be underestimated due; although the magnitude of
6 this bias can not be calculated given the absence in individual studies of information on the
7 percentage of subjects lacking potential TCE exposure.
8 Examination of the statistical power or ability to detect a rate ratio magnitude for site-
9 specific cancer in an epidemiologic study informs weight-of-evidence evaluations and provides
10 perspective on a study's validity and robustness of observations. Although statistical power
11 calculations are traditionally carried out during the design phase for sample size estimation,
12 examination of a study's statistical power post hoc is one of several tools to evaluate a study's
13 validity; however, such calculations must be interpreted in context of exposure conditions in the
14 study. Given the lower average exposure concentrations in the cohort studies and in population
15 case-control studies, an assumption of low relative risks is plausible. Approximate statistical
16 power to detect a relative risk of 2.0 with a = 0.05 was calculated for site-specific cancers in
17 cohort and geographic-based studies according to the methods of Beaumont and Breslow (1981),
18 as suggested by NRC (2006), and are found in Table B-4. Approximate statistical power was
19 calculated for kidney, NHL, and liver cancers as examples. Radican et al. (2008), the previously
20 follow-up of this cohort by Blair et al. (1998), and Raaschou-Nielsen et al. (2003) have over 80%
21 statistical power to detect relative risk of 2.0 for kidney and liver cancers and NHL and overall
22 TCE exposure. However, while these studies may appear sufficient for examining overall TCE
23 exposure and relative risks of 2.0, they have a greatly reduced ability to detect underlying risks
24 of this magnitude in analyses using rank-ordered exposure- or duration-response analyses. Other
25 studies with fewer TCE-exposed subjects and of similar or lower exposure conditions as Blair et
26 al. (1998) will decreased statistical power to detect most site-specific cancer risks of less than
27 2.0. Statistical power in Morgan et al. (1998, 2000) and Boice et al. (1999) approaches that in
28 Blair et al. (1999) and Raaschou-Nielsen et al. (2003). As further identified in Table B-4,
29 Garabrant et al. (1988) and Morgan and Cassady each had over 80% statistical power to detect
30 relative risks of 2.0 for liver and kidney cancer and reflects the number of subjects in each of
31 these studies. However, underlying risk in both studies and other studies such as these which
32 lack characterization of TCE exposure to individual subjects is likely lower than 2.0 because of
33 inclusion of subjects with varying exposure potential, including low exposure potential. Case-
34 control studies such as Charbotel et al. (2006) and Briining et al. (2003) examine higher level
35 exposure to TCE than average exposure in the population case-control studies, and although
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-37 DRAFT—DO NOT CITE OR QUOTE
-------
1 these two studies contain fewer subjects than population case-control studies such as Seidler et
2 al. (2007), a higher statistical power is expected related to the different and higher exposure
3 conditions and to the higher prevalence of exposure.
4 Overall, except for a few studies noted above, the body of evidence has limited statistical
5 power for evaluating low level cancer risk and trichloroethylene. For this reason, studies
6 reporting statistically significant association between trichloroethylene and site-specific cancer
7 are noteworthy if positive biases such as confounding are minimal.
8
9 B.2.8. Statistical Analysis and Result Documentation
10 Appropriate analysis approaches characterize most cohort and case-control studies on
11 trichloroethylene cancer. Many studies clearly documented statistical analyses, evaluated
12 possible confounding factors, and included an examination of exposure-response. In
13 occupational cohort studies, potential confounding factors other than age, sex, race, and calendar
14 year are, generally, not evaluated. Expected numbers of outcomes (deaths or incident cancers)
15 were calculated using life table analysis and an external comparison group, national or regional
16 population mortality or incidence rates (Shindell and Ulrich, 1985; Garabrant et al., 1988;
17 Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Sinks et al., 1992; Axelson et al.,
18 1994; Anttila et al., 1995; Henschler et al., 1995; Morgan et al., 1998; Blair et al., 1998; Boice et
19 al., 1999, 2006a; Raaschou-Nielsen et al., 2003; Chang et al., 2003, 2005; ATSDR, 2004; Sung
20 et al., 2007). Risk ratios are also presented in some cohort studies using proportional hazard and
21 logistic regression statistical methods using mortality or incidence rates of non-TCE exposed
22 cohort subjects as referent or internal controls (Ritz, 1999a; Blair et al., 1998; Boice et al., 1999,
23 2006a; Zhao et al., 2005, Radican et al., 2008). Use of a non-TCE exposed referent group
24 employed at the same facility as exposed generally reduces downward bias or bias having
25 potential associations masked by a healthy worker work or other factors that may be more
26 similar within an occupational cohort than between the cohort and the general population.
27 However, the advantage is minimized if subjects with lower TCE exposure potential are included
28 in the referent group as in Boice et al. (2006a). One referent group (the SSFL group) of Boice et
29 al. (2006a) included individuals with low TCE potential, a treatment different from the
30 overlapping study of Zhao et al. (2005) whose exposure assessment adopted a semi-quantitative
31 approach, grouping subjects identified with low TCE exposure potential separately from subjects
32 with no TCE exposure potential. A second referent group of all Rocketdyne workers in Boice et
33 al. (2006a) for whom TCE exposure potential was not examined may, also, have potential for
34 greater than background exposure since TCE use was widespread and rocket engine cleaning
35 occurred at other locations besides at test sites (Morgenstern et al., 1999).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
o
•••2
i
Exposure group
NHL
Kidney
Liver
Reference
Cohort studies — incidence
Aerospace workers (Rocketdyne)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
Not reported
Referent
97.0
58.2
Not reported
Referent
43.8
18.7
Not reported
Referent
Not reported
Not reported
All employees at electronics factory (Taiwan)
Males
Females
Not reported
Not reported
Not reported
92. la
16.9
15.4
Danish blue-collar worker with TCE exposure
Any exposure, all subjects
Employment duration, males
5 yrs
Employment duration, females
5 yrs
100.0
98.4
99.4
97.7
40.3
48.4
39.6
100.0
96.6
98.4
97.0
30.1
37.1
31.9
100.0
85.2
92.7
93.1
27.3
34.1
30.5
Zhao et al., 2005
Chang et al., 2005
Raaschou-Nielsen et al., 2003
^§
H I
O
H
W
-------
to
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
(continued)
I
I
§
***.
£3'
1
TO'
Exposure group
NHL
Kidney
Liver
Biologically-monitored Danish workers
Any TCE exposure
Cumulative exposure (Ikeda)
<17 ppm-yr
>17 ppm-yr
Mean concentration (Ikeda)
<4ppm
4+ppm
Employment duration
<6.25 yr
>6.25
37.9
17.9
20.3
21.0
23.6
18.3
20.1
47.9
Not reported
Not reported
Not reported
35.7
Not reported
Not reported
Not reported
Aircraft maintenance workers from Hill Air Force Base
TCE subcohort
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
Not reported
Referent
79.5
63.1
70.8
Referent
28.2
0 cases
34.1
Not reported
Referent
67.8
49.4
58.4
Referent
0 cases
0 cases
Not reported
Referent
58.2
44.7
47.4
Referent
0 cases
0 cases
0 cases
Reference
Hansenetal., 2001
Blair etal., 1998
^§
***.
o§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
(continued)
Exposure group
NHL
Kidney
Liver
Biologically-monitored Finnish workers
All subjects
Mean air-TCE (Ikeda extrapolation)
<6ppm
6+ppm
53.8
36.8
25.6
70.4
Not reported
Not reported
56.5
23.2
17.4
Cardboard manufacturing workers in Arnsberg, Germany
Exposed workers
Not reported
16.3
Not reported
Biologically-monitored Swedish workers
Any TCE exposure, males
Any TCE exposure, females
43.5
Not reported
59.6
Not reported
0.05
Not reported
Cardboard manufacturing workers, Atlanta area, GA
All subjects
Not reported
27.9
Not reported
Reference
Anttila et al., 1995
Henschler et al., 1995
Axelson et al., 1994
Sinks etal., 1992
Cohort studies — mortality
Aerospace workers (Rocketdyne)
Any TCE (utility/engine flush)
Any exposure to TCE
Low cumulative TCE score
Medium cumulative TCE score
High TCE score
56.0
Not reported
Referent
97.0
55.4
43.5
Not reported
Referent
57.6
26.4
42.6
Not reported
Referent
Not reported
Not reported
View-Master employees
Males
Females
40.9
74.1
17.3
24.1
23.4
0 deaths
Boice et al., 2006a
Zhao et al., 2005
ATSDR, 2004
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
(continued)
Exposure group
NHL
Kidney
Liver
All employees at electronics factory (Taiwan)
Males
Females
49.8
79.0
0 deaths
37.5
16.9
15.4
United States uranium-processing workers (Fernald)
Any TCE exposure
Light TCE exposure, >2 yrs duration
Mod. TCE exposure, >2 yrs duration
91.6b
20. 9b
59.7C
0 deaths'
10.1
0.08
Aerospace workers (Lockheed)
Routine exposure
Duration of exposure, routine-intermittent
Oyrs
5 yrs
p for trend
88.4
Referent
81.7
73.5
78.5
71.3
Referent
66.3
60.3
63.8
72.9
Referent
73.6
63.5
67.3
Aerospace workers (Hughes)
TCE subcohort
Low intensity (<50 ppm)
High intensity (>50 ppm)
42.6, 79.6d
22.1
31.8
65.5
33.3
50.1
65.6
34.7
49.2
Reference
Chang et al., 2003
Ritz, 1999a
Boiceetal, 1999
Morgan etal., 1998
l
to O
•••2
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
l
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
(continued)
Exposure group
NHL
Kidney
Liver
Aircraft maintenance workers (Hill AFB, UT)
TCE subcohort
92.7
81.5
87.9
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
62.1
43.1
54.8
50.7
37.1
44.9
61.4
44.7
52.8
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
TCE subcohort
18.2
0 deaths
22.0
99.9
0 deaths
8.4
11.5
94.4
0 deaths
0 deaths
19.1
99.7
Males, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
83.0
64.9
75.7
43.8
53.0
33.4
59.4
70.6
50.9
Females, cumulative exposure
0
<5 ppm-yr
5-25 ppm-yr
>25 ppm-yr
38.9
0 deaths
49.2
0 deaths
12.4
21.1
25.9
0 deaths
32.2
Cardboard manufacturing workers in Arnsberg, Germany
TCE exposed workers
19.6b
16.0
Not reported
Reference
Blair etal., 1998
Radican et al., 2008
Henschler et al., 1995
^§
"*4-
o§
H I
O ^
HH Oq
H TO
O
H
W
-------
to
O k^j
o J5"
vo £3'
I
I
§
***.
&
1
TO'
^
w 1
_L $
Table B-4. Approximate statistical power (%) in cohort and geographic-based studies to detect an RR = 2
(continued)
Exposure group
Cardboard manufacturing workers, Atlanta area, GA
NHL
45.3b
Kidney
17.3
Liver
Not reported
Coast Guard employees (US)
Marine inspectors
31.8
31.8
38.6
Aircraft manufacturing plant employees (Italy)
All subjects
94. lb
Not reported
63.1
Aircraft manufacturing plant employees (San Diego, CA)
All subjects
95. le, 74.2f
90.9
77.9
Reference
Sinks etal., 1992
Blair etal., 1989
Costa etal., 1989
Garabrant et al., 1988
Geographic based studies
Residents in two study areas in Endicott, NY
Residents of 13 census tracts in Redlands, CA
90.8
100
41.7
100.0
31.8
98.7
Finnish residents
Residents of Hausjarvi
Residents of Huttula
98.8
98.7
Not reported
Not reported
84.2
83.2
ATSDR, 2006
Morgan and Cassady, 2002
Vartiainen et al., 1993
H I
O
"Kidney cancer and other urinary organs, excluding bladder, as reported in Sung et al. (2008).
bAll cancers of hematopoietic and lymphatic tissues.
°Bladder and kidney cancer, as reported in NRC (2006).
dBased on number of observed cases of NHL reported in Mandel et al. (2006).
eLymphosarcoma and reticulosarcoma.
fOther lymphatic and hematopoietic tissue neoplasms.
H
W
-------
1 Cohort studies additionally evaluate a limited number of other factors associated with
2 employment which could be easily obtained from company and other records such as hire date,
3 time since first employment, socioeconomic status or pay status, and termination date (Greenland
4 et al., 1994; Boice et al., 1999, 2006a; Zhao et al., 2005), and three studies (Ritz, 1999a; Zhao et
5 al., 2005; Boice et al., 2006a) included a limited evaluation of smoking using information
6 collected by survey on smoking patterns from a subgroup of subjects. Neither Morgan et al.
7 (1998) nor Zhao et al. (2005) control for race in analyses, although Morgan et al. (1998) stated
8 that "data concerning race were too sparse to use." The direction of any bias introduced depends
9 on proportion of nonwhites in the referent (internal) group compared to TCE-exposed and on
10 differences between racial groups in site-specific cancer incidence and mortality rates. Blair et
11 al. (1998), furthermore, presumed all subjects of unknown race were white, an assumption with
12 little associated error as shown later by Radican et al. (2008) whose relative risk estimates were
13 adjusted for race in follow-up analysis of this cohort.
14 The case-control studies on trichloroethylene are better able than cohort studies to
15 evaluate other possible confounders besides age and sex using logistic regression approaches
16 since such information can be obtained directly through interview and questionnaires. The case-
17 control studies of Hardell et al. (1994), Nordstrom et al. (1998) and Persson and Fredriksson
18 (1999) lack evaluation of possible confounding factors other than age, sex and other
19 demographic information used to match control subjects to case subjects. Renal cell carcinoma
20 (RCC) case-control studies included evaluation of suggested risk factors for RCC such as
21 smoking (Siemiatycki, 1991; Vamvakas et al., 1998; Pesch et al., 2000a; Briining et al., 2003;
22 Charbotel et al., 2006), weight, or obesity (Dosemeci et al., 1999; Charbotel et al., 2006), and
23 diuretics (Vamvakas et al., 1998; Dosemeci et al., 1999). NHL and childhood leukemia case-
24 control studies included evaluation and control for possible confounding due to smoking
25 (Siemiatycki, 1991; Costas et al., 2002; Seidler et al., 2007), alcohol consumption (Costas et al.,
26 2002; Seidler et al., 2007), education (Miligi et al., 2006; Costantini et al., 2008), although
27 etiological factors for these cancers are not well identified other than a suggestion of a role of
28 immune function and some infectious agents in NHL (Alexander et al., 2007).
29 Mineral oils such as cutting fluids or hydrazine common to some job titles with potential
30 TCE exposure as machinists, metal workers, and test stand mechanics are included as covariates
31 in statistical analyses of Zhao et al. (2005), Boice et al. (2006a) and Charbotel et al. (2006,
32 2009). In all cases, exposure to cutting oils or to hydrazine did not greatly affect magnitude of
33 risk estimates for TCE exposure.
34 Geographical studies do not examine possible confounding factors other than sex, age
35 and calendar year. These studies are generally health surveys using publically-available records
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-45 DRAFT—DO NOT CITE OR QUOTE
-------
1 such as death certificates and lack information on other risk factors such as smoking and
2 exposure to viruses, important to Lee et al. (2003), introduces uncertainties for informing
3 evaluations of trichloroethylene and cancer.
4 B.2.9. Systematic Review for Identifying Cancer Hazards and Trichloroethylene (TCE)
5 Exposure
6 The epidemiological studies on cancer and trichloroethylene are reviewed systematically
7 and transparently using criteria to identify studies for meta-analysis. Section B.3 contains a
8 description of and comment on 75 studies of varying qualities for identifying cancer hazard, a
9 question complementary but separate from that examined using meta-analysis. This section
10 identifies of the studies reviewed, studies in which there is a high likelihood of TCE exposure in
11 individual study subjects (e.g., based on job-exposure matrices, biomarker monitoring, or
12 industrial hygiene data indicating a high probability of TCE use) and were judged to have met
13 the inclusion criteria identified below. Lack of inclusion of an individual study in the meta-
14 analysis does not necessarily imply an inability to identify cancer hazard. Not all questions
15 associated with identifying a cancer hazard are addressed using meta-analyses and the 75 studies
16 with varying abilities approached, to sufficient degrees, the standards of epidemiologic design
17 and analysis, identified in the beginning of Section B.2.
18 The NRC (2006) suggested U.S. EPA conduct a new meta-analysis of the epidemiologic
19 data on trichloroethylene to synthesize the epidemiologic data on TCE exposure. Meta-analysis
20 approaches are feasible for examining cancers of the liver, kidney, and lymphoma given most
21 studies presented risks for these sites in their published papers and these cancer sites are of
22 interest given observations in the animal studies. Examination of site-specific cancers other than
23 kidney cancer, liver cancer, and lymphoma, such as for childhood leukemia, is more difficult and
24 not recommended due to few available high-quality studies. NRC (2006) specifically suggested
25 EPA to:
26
27 1. Document essential design features, exposure, and results from the epidemiologic
28 studies—Information on study design, exposure assessment approach, statistical
29 analysis, and other aspects important to interpreting observations in a weight of
30 evidence evaluation for individual studies is found in Section B.3. and
31 site-specific estimated relative risks or measures of association are presented in
32 Section 4;
33 2. Analyze the epidemiologic studies to discriminate the amount of exposure
34 experience by the study population; exclude studies in meta-analysis based on
35 objective criteria (e.g., studies in which it was unclear that the study population
36 was exposed)—Appendix B.3. describes exposure assessment approach for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-46 DRAFT—DO NOT CITE OR QUOTE
-------
1 individual studies and inclusion criteria for identifying studies for meta-analysis
2 are identified below;
3 3. Classify studies in terms of objective characteristics, such as on the basis of the
4 study's design characteristics or documentation of exposure —Section B.3.
5 groups studies by study design, analytical designs and geographic-based designs,
6 with discussion of factors important to study design, endpoint measured, exposure
7 assessment approach, study size, and statistical analysis methods including
8 adjustment for potential confounding exposures;
9 4. Assess statistical power of each study—Table B.3 presents power calculations for
10 cohort studies;
11 5. Combine case-control and cohort studies in the analysis, unless it introduces
12 substantial heterogeneity—Appendix C discusses the meta-analysis statistical
13 methods and findings;
14 6. Testing of heterogeneity (e.g., fixed or random effect models)—Appendix C
15 discusses the meta-analysis statistical methods and findings;
16 7. Perform a sensitivity analysis in which each study is excluded from the analysis to
17 determine whether any study significantly influences the finding—Appendix C
18 discusses the meta-analysis statistical methods and findings.
19
20 Studies selected for inclusion in the meta-analysis met the following criteria: (1) cohort
21 or case-control designs; (2) evaluation of incidence or mortality; (3) adequate selection in cohort
22 studies of exposure and control groups and of cases and controls in case-control studies; (4) TCE
23 exposure potential inferred to each subject and quantitative assessment of TCE exposure for each
24 subject by reference to industrial hygiene records indicating a high probability of TCE use,
25 individual biomarkers, job exposure matrices, water distribution models, or obtained from
26 subjects using questionnaire (case-control studies); (5) relative risk estimates for kidney cancer,
27 liver cancer, or lymphoma adjusted, at minimum, for possible confounding of age, sex, and race.
28 Table B-5 in Section B.2.9.4 identifies studies included in the meta-analysis and studies that did
29 not meet the inclusion criteria and the primary reasons for their deficiencies.
30
31 B.2.9.1. Cohort Studies
32 The cohort studies (Wilcosky et al., 1984; Shindell and Ulrich, 1985; Garabrant et al.,
33 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Sinks et al., 1992; Axelson et
34 al., 1994; Greenland et al., 1994; Anttila et al., 1995; Henschler et al., 1995; Ritz, 1999a; Blair et
35 al., 1998; Morgan et al., 1998; Boice et al., 1999, 2006a; Hansen et al., 2001; Raaschou-Nielsen
36 et al., 2003; Chang et al., 2003, 2005; Zhao et al., 2005; Krishnadasan et al., 2007; Sung et al.,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-47 DRAFT—DO NOT CITE OR QUOTE
-------
1 2007, 2008; Radican et al., 2008) with data on the incidence or morality of site-specific cancer in
2 relation to trichloroethylene exposure range in size (803 [Hansen et al., 2001] to 86,868 [Chang
3 et al., 2003, 2005]), and were conducted in Denmark, Sweden, Finland, Germany, Taiwan and
4 the United States (see Table B-l). Three case-control studies nested within cohorts (Wilcosky et
5 al., 1984; Greenland et al., 1994; Krishnadasan et al., 2007) are considered as cohort studies
6 because the summary risk estimate from a nested case-control study, the odds ratio, was
7 estimated from incidence density sampling and is considered an unbiased estimate of the hazard
8 ratio, similar to a relative risk estimate from a cohort study. Two studies of deaths within a
9 cohort were included in the group, but these studies lacked information on the person-year
10 structure; i.e., both are proportionate mortality ratio studies, and did not satisfy the meta-analysis
11 inclusion criteria for analytical study design (ATSDR, 2004; Clapp and Hoffman, 2008).
12 Cohort and nested case-control study designs are analytical epidemiologic studies and are
13 generally relied on for identifying a causal association between human exposure and adverse
14 health effects (U.S. EPA, 2005). Some subjects in the Hansen et al. study are also included in a
15 study reported by Raaschou-Nielsen et al. (2003); however, any contribution from the former to
16 the latter are minimal given the large differences in cohort sizes of these studies (Hansen et al.,
17 2001; Raaschou-Nielson et al., 2003). Similarly, some females in Chang et al. (2003, 2005), a
18 large cohort of 70,735 female and 16,133 male subjects, are included in Sung et al. (2007), a
19 cohort of 63,982 female electronic workers from the same factory who were followed an
20 additional 4-year period than subjects in Chang et al. (2003, 2005). Cancer observations for
21 female subjects in these studies are considered as equivalent since they are derived from
22 essentially the same population. Krishnadasan et al. (2007) is a nested case-control study of
23 prostate cancer with cases and controls drawn from subjects in a large cohort of aerospace
24 workers as subjects in Zhao et al. (2005), who did not report on prostate cancer, and met all the
25 inclusion criteria except that for reporting a relative risk estimate for cancer of the kidney, liver
26 or lymphoma.
27 Ten of the cohort studies met all five inclusion criteria: the cohorts of Blair et al. (1998)
28 and its further follow-up by Radican et al. (2008), Morgan et al. (1998), Boice et al. (1999,
29 2006a), and Zhao et al. (2005) of aerospace workers or aircraft mechanics; Axelson et al. (1994),
30 Anttila et al. (1995), Hansen et al. (2001), and Raaschou-Nielsen et al. (2003) of Nordic workers
31 in multiple industries with TCE exposure; and Greenland et al. (1994) of electrical
32 manufacturing workers. All ten cohort studies adopted statistical methods, e.g., life table
33 analysis, Poisson regression analysis, or Cox Proportional Hazard analysis, that met
34 epidemiologic standards, and were able to control for age, race, sex, and calendar time trends in
35 cancer rates. Statistical analyses in Boice et al. (1999) adjusted for demographic variable such as
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-48 DRAFT—DO NOT CITE OR QUOTE
-------
1 age, race, and sex, and, also, included date of first employment and terminating date of
2 employments, which may have decreased the statistical power of their analyses due to colinearity
3 between age, first and last employment dates. Statistical analyses in Zhao et al. (2005) and
4 Boice et al. (2006a) adjusted for potential effects by other occupational exposures on cancer and
5 both Raaschou-Nielsen et al. (2003) and Zhao et al. (2005) examined possible confounding by
6 smoking on TCE exposure and cancer risks using indirect approaches.
7 Of the ten studies, two studies reported risk estimates for both site-specific cancer
8 incidence and mortality (Blair et al., 1998; its follow-up by Radican et al., (2008); Zhao et al.,
9 2005), four studies reported risk estimates for cancer incidence only (Axelson et al., 1994;
10 Anttila et al., 1995; Hansen et al., 2001; Raaschou-Nielsen et al., 2003; Krishnadasan et al.,
11 2007) and three studies reported risk estimates for mortality only (Morgan et al., 1998; Boice et
12 al., 1999, 2006a). Incidence ascertainment in two cohorts began 21 (Blair et al., 1998) and
13 38 years (Zhao et al., 2005) after the inception of the cohort. Specifically, Zhao et al. (2005)
14 note "results may not accurately reflect the effects of carcinogenic exposure that resulted in
15 nonfatal cancers before 1988." Because of the issues concerning case ascertainment raised by
16 this incomplete coverage, incidence observations must be interpreted in light of possible bias
17 reflecting incomplete ascertainment of incident cases. Furthermore, use of an internal referent
18 population, nonexposed subjects drawn from the same or near-by facilities as exposed workers,
19 in Blair et al. (1998) and Radican et al. (2008) for overall TCE exposure, and in Blair et al.
20 (1998), Morgan et al. (1998), Boice et al. (1999), Zhao et al. (2005), Boice et al. (2006a), and
21 Radican et al. (2008) for rank-ordered TCE exposure is expected to reduce bias associated with
22 the healthy worker effect. Morgan et al. (1998) presents risk estimates for overall TCE exposure
23 comparing mortality in their TCE subcohort to that expected using mortality rate of the U.S.
24 population in an Environmental Health Strategies Final Report and sent to U.S. EPA by Paul
25 Cammer, Ph.D., on behalf of the Trichloroethylene Issues Group (Environmental Health
26 Strategies, 1997). The final report also contained risk estimates from internal analyses of rank-
27 order TCE exposure and published as Morgan et al. (1998). Both internal cohort analyses of the
28 rank-ordered exposure, presented in both the final report of Environment Health Strategies
29 (1997) and Morgan et al. (1998), and overall TCE exposure, available in the final report or upon
30 request, are based on the same group of internal referents, nonexposed TCE subjects employed at
31 the same facility.
32 Subjects in these studies had a high likelihood or potential for TCE exposure, although
33 estimated average exposure intensity for overall TCE exposure in some cohorts was considered
34 as less than 10 or 20 ppm (time-weighted average). The exposure assessment techniques used in
35 these cohort studies included a detailed job-exposure matrix (Greenland et al., 1994; Blair et al.,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-49 DRAFT—DO NOT CITE OR QUOTE
-------
1 1998; its follow-up by Radican et al., 2008; Morgan et al., 1998; Boice et al., 1999, 2006a; Zhao
2 et al., 2005; Radican et al. (2008), biomonitoring data (Axelson et al., 1994; Anttila et al., 1995;
3 Hansen et al., 2001), or use of industrial hygiene data on TCE exposure patterns and factors that
4 affect such exposure (Raaschou-Nielsen et al., 2003), with high probability of TCE exposure
5 potential to individual subjects. The job-exposure matrix in six studies provided rank-ordered
6 surrogate metrics for TCE exposure (Axelson et al., 1994; Anttila et al., 1995; Hansen et al.,
7 2001; Blair et al., 1998 and its follow-up by Radican et al., 2008; Zhao et al., 2005), a strength
8 compared to use of duration of employment as an exposure surrogate, e.g., Boice et al. (1999,
9 2006a) or Raachou-Nielsen et al. (2003), which is a poorer exposure metric given subjects may
10 have differing exposure intensity with similar exposure duration (NRC, 2006). Rank-ordered
11 TCE dose surrogates for low and medium exposure from the job-exposure matrix of Morgan et
12 al. (1998) are uncertain because of a lack on information on frequency of exposure-related tasks
13 and on temporal changes (NRC, 2006); only the high category for TCE exposure is
14 unambiguous. The nested case-control study of Greenland et al. (1994) examined TCE as one of
15 seven exposures and potential assigned to individual cases and controls using a job-exposure-
16 matrix approach. However, the low exposure prevalence, missing job history information for
17 34% of eligible subjects, and study of pensioned workers only were other factors judged to lower
18 this study's sensitivity for cancer hazard identification.
19 The remaining cohort studies (Wilcosky et al., 1984; Shindell and Ulrich, 1985;
20 Garabrant et al., 1988; Shannon et al., 1988; Blair et al., 1989; Costa et al., 1989; Sinks et al.,
21 1992; Henschler et al., 1995; Ritz, 1999a; Chang et al., 2003, 2005; Sung et al., 2007, 2008) less
22 satisfactorily meet inclusion criteria. These studies, while not meeting the meta-analysis
23 inclusion criteria, can inform the hazard analysis although their findings are weighted less than
24 for observations in higher-quality studies, and observations may have alternative causes.
25 Reasons for study insufficiencies varied. Nine studies do not assign TCE exposure potential to
26 individual subjects (Shindell and Ulrich, 1985; Garabrant et al., 1988; Costa et al., 1989; Sinks et
27 al., 1992; Chang et al., 2003, 2005; ATSDR, 2004; Sung et al., 2007, 2008; Clapp and Hoffman,
28 2008); all subjects are presumed as "exposed" because of employment in the plant or facility
29 although individual subjects would be expected to have differing exposure potentials.
30 TCE exposure potential is ambiguous in both Wilcosky et al. (1984) and Ritz (1999a),
31 two studies of low potential, low intensity TCE exposure compared to studies using exposure
32 assessment approaches supported by information on job titles, tasks, and industrial hygiene
33 monitoring data. Furthermore, high correlation in Ritz (1999a) between TCE and other
34 exposures, particularly cutting fluids and radiation, may not have been sufficiently controlled in
35 statistical analyses. Ritz et al. (1999a), furthermore, did not report estimated relative risks for
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-50 DRAFT—DO NOT CITE OR QUOTE
-------
1 kidney or lymphoma separately; rather, presenting relative risk estimates for kidney and bladder
2 cancer combined and for all hemato- and lymphopoietic cancers.
3 Two studies do not sufficiently define the underlying cohort or there is uncertainty in
4 cancer case or death ascertainment (Shindell and Ulrich, 1985; Henschler et al., 1995).
5 Furthermore, magnitude of observed risk in Henschler et al. (1995), ATSDR (2004) and Clapp
6 and Hoffman (2008) must be interpreted in a weight-of-evidence evaluation in light of possible
7 bias introduced through use of analysis of proportion of deaths (proportionate mortality ratio) in
8 ATSDR (2004) and Clapp and Hoffman (2008), or to inclusion of index kidney cancer cases in
9 Henschler et al. (1995).
10
11 B.2.9.2. Case-Control Studies
12 Case-control studies on TCE exposure are of several site-specific cancers and include
13 bladder (Siemiatycki, 1991; Siemiatycki et al., 1994; Pesch et al., 2000a); brain (Heineman et al.,
14 1994; De Roos et al., 2001; childhood lymphoma or leukemia (Lowengart et al., 1987;
15 McKinney et al., 1991; Shu et al., 1999, 2004; Costas et al., 2002); colon cancer (Siemiatycki,
16 1991; Goldberg et al., 2001); esophageal cancer (Siemiatycki, 1991; Parent et al., 2000a); liver
17 cancer (Lee et al., 2003); lung (Siemiatycki, 1991), lymphoma (Hardell et al., 1994 [NHL,
18 Hodgkin lymphoma]; Siemiatycki, 1991; Fritschi and Siemiatycki, 1996a; Nordstrom et al.,
19 1998; [hairy cell leukemia]; Persson and Fredriksson, 1999 [NHL]; Miligi et al., 2006 [NHL and
20 chronic lymphocytic leukemia (CLL)]; Seidler et al., 2007 [NHL, Hodgkin lymphoma];
21 Constantini et al., 2008 [leukemia types, CLL included in Miligi et al., 2006]; Wang et al., 2009
22 [NHL]); melanoma (Siemiatycki, 1991; Fritchi and Siemiatycki, 1996b); rectal cancer
23 (Siemiatycki, 1991; Dumas et al., 2000); renal cell carcinoma, a form of kidney cancer
24 (Siemiatycki, 1991; Parent et al., 2000b; Vamvakas et al., 1998; Dosemeci et al., 1999; Pesch et
25 al., 2000b; Briining et al., 2003; Charbotel et al., 2006, 2009); pancreatic cancer (Siemiatycki,
26 1991); and prostate cancer (Siemiatycki, 1991; Aronson et al., 1996). No case-control studies of
27 reproductive cancers (breast or cervix) and TCE exposure were found in the peer-reviewed
28 literature.
29 Several of the above publications are studies of cases and controls drawn from the same
30 underlying population with a common control series. Miligi et al. (2006) and Costantini et al.
31 (2008) presented observations from the Italian multicenter lymphoma population case-control
32 study; Miligi et al. (2006) on occupation or specific solvent exposures and NHL, and who also
33 included CLL and Hodgkin's lymphoma in the overall NHL category, and Costantini et al.
34 (2008) who examined leukemia subtypes, and included CLL as a separate disease outcome.
35 Pesch et al. (2000a, b), a multiple center population case- control study of urothelial cancers in
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-51 DRAFT—DO NOT CITE OR QUOTE
-------
1 Germany, presented observations on TCE and bladder cancer, including cancer of the ureter and
2 renal pelvis, in Pesch et al. (2000a) and renal cell carcinoma in Pesch et al. (2000b). Siemiatycki
3 (1991), a case-control of occupational exposures and several site-specific cancers (bladder,
4 colon, esophagus, lung, rectum, pancreas, and prostate) and designed to generate hypotheses
5 about possible occupational carcinogens, presents risk estimates associated with TCE exposure
6 using Mantel-Haentszel methods. Subsequent publications examine either TCE exposure
7 (analyses of melanoma and colon cancers) or job title/occupation (all other cancer sites) using
8 logistic regression methods (Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and
9 Siemiatycki, 1996a, b; Dumas et al., 2000; Parent et al., 2000a, b; Goldberg et al., 2001).
10 The population case-control studies with data on cancer incidence (Siemiatycki, 1991
11 [and related publications, Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and Siemiatycki,
12 1996a, b; Dumas et al., 2000; Parent et al., 2000a, b; Goldberg et al., 2001]; Lowengart et al.,
13 1987; McKinney et al., 1991; Hardell et al., 1994; Nordstrom et al., 1998; Vamvakas et al., 1998;
14 Dosemeci et al., 1999; Kernan et al., 1999; Persson and Fredriksson, 1999; Pesch et al., 2000a, b;
15 De Roos et al., 2001; Costas et al., 2002; Briining et al., 2003; Shu et al., 2004; Charbotel et al.,
16 2006, 2009; Miligi et al., 2006; Seidler et al., 2007; Constantini et al., 2008; Wang et al., 2009)
17 or mortality (Heineman et al., 1994; Lee et al., 2003) in relation to trichloroethylene exposure
18 range in size, from small studies with less than 100 cases and control (Costas et al., 2002) to
19 multiple-center studies large-scale studies of over 2,000 cases and controls (Shu et al., 1999,
20 2004; Pesch et al., 2000a, b; Miligi et al., 2006; Costantini et al., 2008), and were conducted in
21 Sweden, Germany, Italy, Taiwan, Canada and the United States (see Table B-2).
22 Thirteen of the case-control studies met the meta-analysis inclusion criteria identified in
23 Section B.2.9 (Siemiatycki, 1991; Hardell et al., 1994; Nordstrom et al., 1998; Dosemeci et al.,
24 1999; Persson and Fredriksson, 1999; Pesch et al., 2000 b; Briining et al., 2003; Miligi et al.,
25 2006; Charbotel et al., 2006, 2009; Seidler et al., 2007; Constantini et al., 2008, Wang et al.,
26 2009). They were of analytical study design, cases and controls were considered to represent
27 underlying populations and selected with minimal potential for bias; exposure assessment
28 approaches included assignment of TCE exposure potential to individual subjects using
29 information obtained from face-to- face, mailed, or telephone interviews; analyses methods were
30 appropriate, well-documented, included adjustment for potential confounding exposures, with
31 relative risk estimates and associated confidence intervals reported for kidney cancer, liver
32 cancer or lymphoma.
33 All thirteen studies evaluated TCE exposure potential to individual cases and controls and
34 a structured questionnaire sought information on self-reported occupational history and specific
35 exposures such as TCE. Three studies assigned TCE exposure potential to cases and controls
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-52 DRAFT—DO NOT CITE OR QUOTE
-------
1 using self-reported information (Hardell et al., 1994; Nordstrom et al., 1998; Persson and
2 Fredriksson, 1999) and two of these studies used judgment to assign potential exposure intensity
3 (Nordstrom et al., 1998; Persson and Fredriksson, 1999). Persson and Fredriksson (1999) also
4 assigned TCE exposure potential from both occupational and leisure use, the only study to do so.
5 The ten other studies assigned TCE exposure potential using self-reported job title and
6 occupational history, a superior approach compared to use of a job exposure matrix (JEM)
7 supported by expert judgment and information on only self-reported information given its expect
8 greater specificity (Siemiatycki, 1991; Dosemeci et al., 1999; Pesch et al., 2000b; Briining et al.,
9 2003; Miligi et al., 2006; Charbotel et al., 2006, 2009; Seidler et al., 2007; Constantini et al.,
10 2008, Wang et al., 2009). Pesch et al. (2000b) assigned TCE exposure potential using both job
11 exposure matrix and job-task exposure matrix (ITEM). The inclusion of task information is
12 considered superior to exposure assignment using only job title since it likely reduces potential
13 misclassification and, for this reason, relative risk estimates in Pesch et al. (2000b) for TCE from
14 a ITEM are preferred. All studies except Hardell et al. (1994) and Dosemeci et al. (1999)
15 developed a semi quantitative or quantitative TCE exposure surrogate.
16 These studies to varying degrees were considered as high-quality studies for weight-of
17 evidence characterization of hazard. Both Briining et al. (2003) and Charbotel et al. (2006,
18 2009) had a priori hypotheses for examining renal cell carcinoma and TCE exposure. Strengths
19 of both studies are in their examination of populations with potential for high exposure intensity
20 and in areas with high frequency of TCE usage and their assessment of TCE potential. An
21 important feature of the exposure assessment approach of Charbotel et al. (2006) is their use of a
22 large number of studies on biological monitoring of workers in the screw-cutting industry a
23 predominant industry with documented TCE exposures as support. The other studies were either
24 large multiple-center studies (Pesch et al., 2000a, b; Miligi et al., 2006; Constantini et al., 2008;
25 Wang et al., 2009) or reporting from one location of a larger international study (Dosemeci et al.,
26 1999; Seidler et al., 2007). In contrast to Briining et al. (2003) and Charbotel et al. (2006, 2009),
27 two studies conducted in geographical areas with widespread TCE usage and potential for
28 exposure to higher intensity, a lower exposure prevalence to TCE is found [any TCE exposure:
29 15% of cases (Dosemeci et al., 1999); 6% of cases (Miligi et al., 2006); 13% of cases (Seidler et
30 al., 2007); 13% of cases (Wang et al., 2008)] and most subjects identified as exposed to TCE
31 probably had minimal contact [3% of cases with moderate/high TCE exposure (Miligi et al.,
32 2006); 1% of cases with high cumulative TCE (Seidler et al., 2007); 2% of cases with high
33 intensity, but of low probability TCE exposure (Wang et al., 2008)]. This pattern of lower
34 exposure prevalence and intensity is common to community-based population case-control
35 studies (Teschke et al., 2002).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-53 DRAFT—DO NOT CITE OR QUOTE
-------
1 Thirteen case-control studies did not meet specific inclusion criterion (Siemiatycki et al.,
2 1994; Aronson et al., 1996; Fritchi and Siemiatycki, 1996b; Dumas et al., 2000; Parent et al.,
3 2000a; Goldberg et al., 2001; Vamvakas et al., 1998; Kernan et al., 1999; Shu et al., 1999, 2004;
4 Pesch et al., 2000a; Costas et al., 2002; Lee et al., 2003). Vamvakas et al. (1998) has been
5 subject of considerable controversy (Bloemen and Tomenson, 1995; Swaen, 1995; McLaughlin
6 and Blot, 1997; Green and Lash, 1999; Cherrie et al., 2001; Mandel, 2001) with questions raised
7 on potential for selection bias related to the study's controls. This study was deficient in the
8 criterion for adequacy of case and control selection. Briining et al. (2003), a study from the same
9 region as Vamvakas et al. (1998), is considered a stronger study for identifying cancer hazard
10 since it addresses many of the deficiencies of Vamvakas et al. (1998). Lee et al. (2003) in their
11 study of hepatocellular cancer assigns one level of exposure to all subjects in a geographic area,
12 and inherent measurement error and misclassification bias because not all subjects are exposed
13 uniformly. Additionally, statistical analyses in this study did not control for hepatitis viral
14 infection, a known risk factor for hepatocellular cancer and of high prevalence in the study area,
15 Ten of twelve studies reported relative risk estimates for site-specific cancers other than kidney,
16 liver, and lymphomas (Siemiatycki et al., 1994; Aronson et al., 1996; Fritchi and Siemiatycki,
17 1996b; Kernan et al., 1999; Dumas et al., 2000; Parent et al., 2000a; Pesch et al., 2000a;
18 Goldberg et al., 2001; Shu et al., 1999, 2004; Costas et al., 2002).
19
20 B.2.9.3. Geographic-Based Studies
21 The geographic-based studies (Isacson et al., 1985; AZ DHS, 1990, 1995; Aickin et al.,
22 1992; Aickin, 2004; Mallin, 1990; Vartiainen et al., 1993; Cohn et al., 1994, Morgan and
23 Cassady, 2002; ATSDR, 2006a, 2008) with data on cancer incidence (all studies) are correlation
24 studies to examine cancer outcomes of residents living in communities with TCE and other
25 chemicals detected in groundwater wells or in municipal drinking water supplies. These eight
26 studies did not meet inclusion criteria and were deficient in a number of criteria.
27 All geographic-based studies are surveys of cancer rates for a defined time period among
28 residents in geographic areas with TCE contamination in groundwater or drinking water
29 supplies, or soil and are not of analytical designs such as cohort and case-control designs. A
30 major shortcoming in all studies is, also, their low level of detail to individual subjects for TCE
31 potential. The exposure surrogate is assigned to a community, town, or a geographically-defined
32 area such as a contiguous grouping of census tracts as an aggregate level, typically based on
33 limited number of water monitoring data from a recent time period and is a poor exposure
34 surrogate because potential for TCE exposure can vary in these broad categories depending on
35 job function, year, use of personal protection, and, for residential exposure, pollutant fate and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-54 DRAFT—DO NOT CITE OR QUOTE
-------
1 transport, water system distribution characteristics, percent of time per day in residence, presence
2 of mitigation devices, drinking water consumption rates, and showering times. Additionally,
3 ATSDR (2008), the only geographic-based study to examine other possible risk factors on
4 individual subjects, reported smoking patterns and occupational exposures may partly contribute
5 to the observed elevated rates of kidney and renal pelvis cancer and lung cancer in subjects living
6 in a community with contaminated groundwater and with TCE exposure potential from vapor
7 intrusion into residences.
8
9 B.2.9.4. Recommendation of Studies for Treatment Using Meta-Analysis Approaches
10 All studies are initially considered for inclusion in the meta-analysis; however, as
11 discussed through-out this section, some studies are better than others for inclusion in a
12 quantitative examination of cancer and trichloroethylene. Studies included in the meta-analysis
13 (statistical methods and findings discussed in Appendix C) met the following five inclusion
14 criteria: (1) cohort or case-control designs; (2) evaluation of incidence or mortality; (3) adequate
15 selection in cohort studies of exposure and control groups and of cases and controls in case-
16 control studies; (4) TCE exposure potential inferred to each subject and quantitative assessment
17 of TCE exposure assessment for each subject by reference to industrial hygiene records
18 indicating a high probability of TCE use, individual biomarkers, job exposure matrices, water
19 distribution models, or obtained from subjects using questionnaire (case-control studies); (5)
20 relative risk estimates for kidney cancer, liver cancer, or lymphoma adjusted, at minimum, for
21 possible confounding of age, sex, and race. The twenty-three studies that met these inclusion
22 are: Siemiatycki (1991), Axelson et al. (1994), Greenland et al. (1994), Hardell et al. (1994),
23 Anttila et al. (1995), Blair et al. (1998), Morgan et al. (1998), Nordstrom et al. (1998), Dosemeci
24 et al. (1999), Boice et al. (1999, 2006a), Persson and Fredriksson (1999), Pesch et al. (2000b),
25 Hansen et al. (2001), Briining et al. (2003), Raaschou-Nielsen et al. (2003), Zhao et al. (2005),
26 Miligi et al. (2006), Charbotel et al. (2006, 2009), Seidler et al. (2007), Radican et al. (2008), and
27 Wang et al. (2009). Table B-5 identifies studies included in the meta-analysis and studies that
28 did not meet the inclusion criteria and the primary reasons for their deficiencies.
29 There is some overlap between the cohorts of Zhao et al. (2005) and Boice et al. (2006a),
30 each cohort is identified from a population of workers, but these studies differ on cohort
31 definition, cohort identification dates, disease outcome examined, and exposure assessment
32 approach. Zhao et al. (2005) who adopted a semi quantitative approach for TCE exposure
33 assessment is preferred to Boice et al. (2006a), whose TCE subcohort included subjects with a
34 lower likelihood for TCE exposure and duration of exposure, a poor exposure metric given
35 subjects may have differing exposure intensity with similar exposure duration (NRC, 2006).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-55 DRAFT—DO NOT CITE OR QUOTE
-------
1 Additionally, a larger number of site-specific cancer deaths identified with potential TCE
2 exposure is observed by Zhao et al. (2005) compared to Boice et al. (2006a); e. g., 95 lung
3 cancer cases with medium or high TCE exposure (Zhao et al., 2005) and 51 lung cancer cases
4 with any TCE exposure (Boice et al., 2006a) (see further discussion in B.3.1.1.1.3.). Radican et
5 al. (2008) studied the same subjects as Blair et al. (1998), adding an additional 10 years of
6 follow-up and updating mortality. Observed site-specific cancer mortality risk estimates in
7 Radican et al. (2008) did not change appreciably and were consistent with those reported in Blair
8 et al. (1998) and is preferred. Blair et al. (1998) who also presented incidence relative risk
9 estimates is recommended for inclusion in sensitivity analyses.
10
11 B.3. INDIVIDUAL STUDY REVIEWS AND ABSTRACTS
12 B.3.1. Cohort Studies
13 B.3.1.1. Studies of Aerospace Workers
14 Seven papers reported on cohort studies of aerospace or aircraft maintenance and
15 manufacturing workers in large facilities.
16
17 B.3.1.1.1. Studies of Santa Susanna Field Laboratory workers. Trichloroethylene exposure
18 to workers at Santa Susanna Field Laboratory (SSFL), an aerospace facility located nearby Los
19 Angeles, California, operated by Rocketdyne/Atomics International, formerly a division of
20 Boeing and currently owned by Pratt-Whitney, is subject of two research efforts: (1) the
21 University of California at Los Angeles (UCLA) study, overseen by the California Department
22 of Health Services and funded by the U.S. Department of Energy (DOE) (Morgenstern et al.,
23 1997, 1999; Ritz et al., 1999) with two publications on trichloroethylene exposure and cancer
24 incidence (Zhao et al., 2005; Krishnadasan et al., 2007) and mortality (Zhao et al., 2005); and,
25 (2) the International Epidemiology Institute study (IEI), funded by Boeing after publication of
26 the initial UCLA reports, of all Rocketdyne employees which included a mortality analysis of
27 trichloroethylene exposure in a subcohort of SSFL test stand mechanics (Boice et al., 2006a). In
28 addition to chemical exposure, both groups examine radiation exposure and cancer among
29 Rocketdyne workers monitored for radiation (Ritz et al., 2000; Boice et al., 2006b).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-56 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table B-5. Summary of rationale for study selection for meta-analysis
2
Decision
Outcome
Studies
Primary reason(s)
Studies Recommended for Meta-analysis:
Siemiatycki, 1991; Axelsonetal., 1994;
Hardell, 1994; Greenland etal., 1994;
Anttila et al., 1995; Morgan et al, 1998;
Nordstrom et al., 1998; Boice et al., 1999,
2006a; Dosemeci et al., 1999; Persson and
Fredriksson, 1999; Pesch et al., 2000b;
Hansen et al., 2001; Briining et al., 2003;
Raaschou-Nielsen et al., 2003; Zhao et al.,
2005; Miligi et al., 2006; Seidler et al.,
2007; Charbotel et al., 2006, 2009; Radican
etal., 2008 [Blair etal., 1998, incidence];
Wang etal., 2009
Analytical study designs of cohort or case-control approaches;
Evaluation of cancer incidence or cancer mortality;
Specifically identified TCE exposure potential to individual
study subjects by reference to industrial hygiene records,
individual biomarkers, job exposure matrices, water distribution
models, industrial hygiene data indicating a high probability of
TCE use (cohort studies), or obtained information on TCE
exposure from subjects using questionnaire (case-control
studies);
Reported results for kidney cancer, liver cancer, or lymphoma
with relative risk estimates and corresponding confidence
intervals (or information to allow calculation).
Studies Not Recommended for Meta-analysis:
ATSDR, 2004; Clapp and Hoffman, 2008
Conn etal., 1994
Wilcosky et al., 1984; Isacson et al., 1985;
Shindell and Ulrich, 1985; Garabrant et al.,
1988; Shannon et al., 1988; Blair et al.,
1989; Costa et al., 1989; AZ DHS, 1990,
1995; Mallin, 1990; Aickin et al., 1992;
Sinks et al., 1992; Vartiainen et al., 1993;
Morgan and Cassady, 2002; Lee et al.,
2003; Aickin, 2004; Chang et al., 2003,
2005; Coyle etal., 2005; ATSDR, 2006a,
2008; Sung et al., 2007, 2008;
Lowengart et al., 1987; Fredriksson et al.,
1989; McKinney et al., 1991; Heineman et
al., 1994; Siemiatycki etal., 1994; Aronson
et al., 1996; Fritchi and Siemiatycki,
1996b; Dumas et al., 2000; Kernan et al.,
1999; Shu et al., 1999, 2004; Parent et al.,
2000a; Pesch et al., 2000a; De Roos et al.,
2001; Goldberg et al., 2001; Costas et al.,
2002; Krishnadasan et al., 2007;
Ritz, 1999a
Henschleretal., 1995
Vamvakas et al., 1998
Weakness with respect to analytical study design (i.e.,
geographic -based, ecological or proportional mortality ratio
design)
TCE exposure potential not assigned to individual subjects using
job exposure matrix, individual biomarkers, water distribution
models, or industrial hygiene data indicating a high probability
of TCE use (cohort studies)
Cancer incidence or mortality reported for cancers other than
kidney, liver, or lymphoma
Subjects monitored for radiation exposure with likelihood for
potential confounding;
Cancer mortality and TCE exposure not reported for kidney
cancer and all hemato- and lymphopoietic cancer reported as
broad category
Incomplete identification of cohort and index kidney cancer
cases included in case series
Control selection may not represent case series with potential for
selection bias
3
4
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-57 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.3.1.1.1.1. International epidemiology institute study of Rocketdyne workers.
2 B.3.1.1.1.1.1. Boiceetal. (2006a).
3 B.3.1.1.1.1.1.1. Author's abstract.
4
5 Objective: The objective of this study was to evaluate potential health risks
6 associated with testing rocket engines. Methods: A retrospective cohort mortality
7 study was conducted of 8372 Rocketdyne workers employed 1948 to 1999 at the
8 Santa Susana Field Laboratory (SSFL). Standardized mortality ratios (SMRs) and
9 95% confidence intervals (CIs) were calculated for all workers, including those
10 employed at specific test areas where particular fuels, solvents, and chemicals were
11 used. Dose-response trends were evaluated using Cox proportional hazards
12 models. Results: SMRs for all cancers were close to population expects among
13 SSFL workers overall (SMR = 0.89; CI = 0.82-0.96) and test stand mechanics in
14 particular (n = 1651; SMR= 1.00; CI = 0.86-1.1.6), including those likely
15 exposure to hydrazines (n = 315; SMR= 1.09; CI = 0.75-1.52) or trichloroethylene
16 (TCE) (n=l111; SMR = 1.00; CI = 0.83-1.19). Nonsignificant associations were
17 seen between kidney cancer and TCE, lung cancer and hy drazines, and stomach
18 cancer and years worked as a test stand mechanic. No trends over exposure
19 categories were statistically significant. Conclusion: Work at the SSFL rocket
20 engine test facility or as a test stand mechanic was not associated with a significant
21 increase in cancer mortality overall or for any specific cancer.
22
23 B.3.1.1.1.1.1.2. Study description and comment. Boice et al. (2006a) examined all cause, all
24 cancer and site-specific mortality in a subcohort of 1,651 male and female test stand mechanics
25 who had been employed on or after 1949 to 1999, the end of follow-up, for at least 6 months at
26 SSFL. Subjects were identified from 41,345 male and female Rocketdyne workers at SSFL
27 (n = 8.372) and two nearby facilities (32,979). Of the 1,642 male test stand mechanics,
28 9 females were excluded due to few numbers, personnel listing in company phone directories
29 were used to identify test stand assignments (and infer potential specific chemical exposures) for
30 1,440 subjects, and of this group, 1,111 male test stand mechanics were identified with potential
31 trichloroethylene exposure either from the cleaning of rocket engines between tests or from more
32 generalized use as a utility degreasing solvent. Cause-specific mortality is compared to several
33 referents: (1) morality rates of the U.S. population, (2) mortality rates of California residents,
34 (3) hourly nonadministrative workers at SSFL and two nearby facilities, and (4) 1,598 SSFL
35 hourly workers; however, the published paper does not clearly present details of all analyses.
36 For example, the referent population is not identified for the standardized mortality ratio (SMR)
37 analysis of the 1,111 male subjects with TCE potential exposure and analyses examining
38 exposure duration present point estimates and p-values from tests of linear trend, but not always
39 confidence intervals (e.g., Boice et al. [2006a, Table 7] table footnotes).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-58 DRAFT—DO NOT CITE OR QUOTE
-------
1 Exposure assessment to trichloroethylene is qualitative without attempt to characterize
2 exposure level as was done in the exposure assessment approach of Zhao et al. (2005) and
3 Krishnadsen et al. (2007). Test stand mechanics were nonadministrative hourly positions and
4 had the greatest potential for chemical exposures to TCE and hydrazine. Potential exposure to
5 chemicals also existed for other subjects associated with test stand work such as instrument
6 mechanics, inspectors, test stand engineers, and research engineers potential for chemical
7 exposure, although Boice et al. (2006a) considered their exposure potential lower compared to
8 that received by test stand mechanics and, thus, were not included in the cohort. Like that
9 encountered by UCLA researchers, work history information in the personnel file was not
10 specific to identify work location and test stand and Boice et al. (2006a) adopted ancillary
11 information, company phone directories, as an aid to identify subjects with greater potential for
12 TCE exposure. From these aids, investigators identified rocket stand assignment for 1,440 or
13 87% of the SSFL test stand mechanics. Bias is introduced through missing information on the
14 other 211 subjects or if phone directories were not available for the full period of the study. Test
15 stand mechanics, if exposed, had the likelihood for exposure to high TCE concentrations
16 associated with flushing or cleaning of rocket engines; 593 of the 1,111 subjects (53%) were
17 identified as having potential TCE exposure through rocket engine cleaning. The removal or
18 flushing of hydrocarbon deposits in fuel j ackets and in liquid oxygen dome of large engines
19 entailed the use of 5 to 100 gallons of TCE, with TCE use starting around 1956 and ceased by
20 the late 1960's at all test stands except one which continued until 1994. No information was
21 provided on test stand and working conditions or the frequency of exposure-related tasks, and no
22 atmospheric monitoring data were available on TCE. A small number of these subjects (121)
23 also had potential exposure to hydrazines. The remaining 518 subjects in the TCE subcohort
24 were presumed exposed to TCE as a utility solvent. Information on use of TCE as a utility
25 solvent is lacking except that TCE as a utility solvent was discontinued in 1974 except at one test
26 stand where it was used until 1984. These subjects have a lower likelihood of exposure
27 compared to subjects with TCE exposure from cleaning rocket engines.
28 Several study design and analysis aspects limit this study for assessing risks associated
29 with trichloroethylene exposure. Overall, exposures were likely substantially misclassified and
30 their frequency likely low, particularly for subjects identified with TCE use as a utility solvent
31 who comprise roughly 50% of the TCE subcohort. Analyses examining number of years
32 employed at SSFL or worked as test stand mechanic as a surrogate for cumulative exposure has a
33 large potential for misclassification bias due to the lack of air monitoring data and inability to
34 account to temporal changes in TCE usage. Moreover, the exposure metric used in some dose-
35 response analyses is weighted by the number of workers without rationale provided and would
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-59 DRAFT—DO NOT CITE OR QUOTE
-------
1 introduce bias if the workforce changed over the period covered by this study. Some information
2 suggests this was likely (1) the number of cohort subjects entering the cohort decreased over the
3 time period of this study, as much as a 20% decrease between 1960's and 1970s, and
4 (2) ancillary information (http://www.thewednesdayreport.com/twr/twr48v7.htm, accessed
5 March 11, 2008; DOE Closure Project, http://www.etec.energy.gov/Reading-
6 Room/DeSoto.html, accessed March 11, 2008). Study investigators did not carry out exposure
7 assessment for referents and no information is provided on potential trichloroethylene exposure.
8 If referents had more than background exposure, likely for other hourly subjects with direct
9 association with test stand work but with a j ob title other than test stand mechanic, the bias
10 introduced leads to an underestimation of risk. TCE use at SSFL was widespread and rocket
11 engine cleaning occurred at other locations besides at test sites (Morgenstern et al., 1999),
12 locations from which the referent population arose.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-60 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Boice JD, Marano DE, Cohen SS, Mumma MT, Blott WJ, Brill AB, Fryzek JP, Henderson BE, McLaughlin JK. 2006a.
Mortality among Rocketdyne workers who tested rocket engines, 1948-1999. J Occup Environ Med 48:1070-1092.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract "objective of this study was to evaluate potential health risks
associated with testing rocket engines."
54,384 Rocketdyne workers of which 41,351 were employed on or after 1-1-1948
and for at least 6 mos at Santa Susana Field Laboratory or nearby facilities. Of the
41,351 subjects, 1,651 were identified as having a job title of test stand mechanic and
exposure assignments could be made for 1,440 of these subjects.
Site-specific mortality rates of U.S. population and of all-other Rocketdyne
employees. Potential TCE exposures of all other subjects (referents) not documented
but investigators assumed referents are unexposed to TCE.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality from 1948 to 12-31-1999.
Coding to ICD in use at time of death.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Qualitative exposure assessment, any TCE exposure. No quantitative information on
TCE intensity by job title or to individual subjects or referents.
Missing exposure potential to 12% of test stand mechanics; potential exposure
hydrazine and/or TCE assigned to 1,440 of 1,651 test stand mechanics. Of 1,440 test
stand mechanics, 1,111* identified with potential TCE exposure, 518 of the
1,111 identified as having presumed high intensity exposure from the cleaning of
rocket engines. The remaining 593 subjects with potential exposure to TCE through
use as "utility solvent," a job task with low likelihood or potential for TCE exposure.
H 3
§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
to o
•••2
i
^§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
0.4% for test stand mechanic cohort (1,651 subjects).
35 years average follow-up; 88% of 1,651 test stand mechanics >20 yr follow-up.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
TCE exposed subcohort — 391 total deaths, 121 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
SMR analysis restricted to male hourly test stand mechanics using U.S. population
rates as referent — no adjustment of potential confounders other than age and
calendar-year.
Cox proportional hazard models examining TCE exposure adjusted for birth year,
year of hire and potential hydrazine exposure. Race was not included in Cox
proportional hazard analysis.
SMR analysis and Cox proportional hazard.
Duration of exposure (employment): 2-sided tests for linear trend.
All analyses are not presented in published paper. Follow-up correspondence
Scott, U.S. EPA, to J. Boice, of 12-31-06 and 02-28-07 remain unanswered as
November 15,2007.
ofC
of
*Zhao et al. (2005), whose study period and base population overlaps that of Boice et al. (2006a), identified a larger number of subjects with potential TCE
exposures; 2,689 subjects with TCE score > 3, a group having medium to high cumulative TCE exposure.
-------
1 B.3.1.1.1.2. University of California at Los Angeles (UCLA) studies ofRocketdyne workers.
2 B.3.1.1.1.2.1. Krishnadasan et al. (2007).
3 B.3.1.1.1.2.1.1. Author's abstract.
4
5 Background To date, little is known about the potential contributions of
6 occupational exposure to chemicals to the etiology of prostate cancer. Previous
7 studies examining associations suffered from limitations including the reliance on
8 mortality data and inadequate exposure assessment. Methods We conducted a
9 nested case-control study of 362 cases and 1,805 matched controls to examine the
10 association between occupational chemical exposures and prostate cancer
11 incidence. Workers were employed between 1950 and 1992 at a nuclear energy
12 and rocket engine-testing facility in Southern California. We obtained cancer
13 incidence data from the California Cancer Registry and seven other state cancer
14 registries. Data from company records were used to construct a job exposure
15 matrix (JEM) for occupational exposures to hydrazine, trichloroethylene (TCE),
16 poly cyclic aromatic hydrocarbons (PAHs), benzene, and mineral oil.
17 Associations between chemical exposures and prostate cancer incidence were
18 assessed in conditional logistic regression models. Results With adjustment for
19 occupational confounders, including socioeconomic status, occupational physical
20 activity, and exposure to the other chemicals evaluated, the odds ratio for
21 low/moderate TCE exposure was 1.3; 95%CI=0.8 to 2.1, and for high TCE
22 exposure was 2.1; 95%CI=1.2 to 3.9. Furthermore, we noted a positive trend
23 between increasing levels of TCE exposure and prostate cancer (p-value for
24 trend=0.02). Conclusion Our results suggest that high levels of TCE exposure
25 are associated with prostate cancer among workers in our study population.
26
27 B.3.1.1.1.2.2. Zhao et al. (2005).
28 B.3.1.1.1.2.2.1. Author's abstract.
29
30 Background A retrospective cohort study of workers employed at a California
31 aerospace company between 1950 and 1993 was conducted; it examined cancer
32 mortality from exposures to the rocket fuel hydrazine. Methods In this study, we
33 employed a job exposure matrix (JEM) to assess exposures to other known or
34 suspected carcinogens—including trichloroethylene (TCE), polycyclic aromatic
35 hydrocarbons (PAHs), mineral oils, and benzene—on cancer mortality
36 (1960-2001) and incidence (1988-2000) in 6,107 male workers. We derived
37 rate- (hazard-) ratios estimates from Cox proportional hazard models with time-
38 dependent exposures. Results High levels of TCE exposure were positively
39 associated with cancer incidence of the bladder (rate ratio (RR): 1.98, 95%
40 confidence interval (CI) 0.93-4.22) and kidney (4.90; 1.23-19.6). High levels of
41 exposure to mineral oils increased mortality and incidence of lung cancer (1.56;
42 1.02-2.39 and 1.99; 1.03-3.85), and incidence of melanoma (3.32; 1.20-9.24).
43 Mineral oil exposures also contributed to incidence and mortality of esophageal
44 and stomach cancers and of non-Hodgkin's lymphoma and leukemia when
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-63 DRAFT—DO NOT CITE OR QUOTE
-------
1 adjusting for other chemical exposures. Lagging exposure measures by 20 years
2 changed effect estimates only minimally. No associations were observed for
3 benzene or PAH exposures in this cohort. Conclusions Our findings suggest that
4 these aerospace workers who were highly exposed to mineral oils experienced an
5 increased risk of developing and/or dying from cancers of the lung, melanoma,
6 and possibly from cancers of the esophagus and stomach and non-Hodgkin's
7 lymphoma and leukemia. These results and the increases we observed for TCE
8 and kidney cancers are consistent with findings of previous studies.
9
10 B.3.1.1.1.2.3. Study description and comment. The source population for Krishnadasen et al.
11 (2007) and Zhao et al. (2005) is the UCLA chemical cohort of 6,044 male workers with 2 or
12 more years of employment Rocketdyne between 1950 and 1993, who engaged in rocket testing
13 at SSFL before 1980 and who have never been monitored for radiation. Zhao et al. (2005)
14 examined cancer mortality between 1960-2001, an additional 7 years from earlier analyses of
15 the chemical subcohort (Morgenstern et al., 1999; Ritz et al., 1999), and cancer incidence
16 (5,049 subjects) between 1988-2000, matching cohort subjects to names in California's Cancer
17 Registry and eight other state cancer registries. Deaths before 1998 are coded using ICD, 9th
18 revision, and ICD-10 after this date; ICD-0 was used to code cancer incidence with leukemia,
19 lymphoma, and other lymphopoietic tumors grouped on the basis of morphology codes. A total
20 of 600 cancer deaths and 691 incident cancers were identified during the study period.
21 Krishnadasen et al. (2007) adopted a nested case-control design to examine occupational
22 exposure to several chemicals and prostate cancer incidence in a cohort which included the SSFL
23 chemically-exposed subjects and an additional 4,607 workers in the larger cohort who were
24 enrolled in the company's radiation monitoring program. A total of 362 incident prostate
25 cancers were identified between 1988 and 12-31-1999. Controls were randomly selected from
26 the original cohorts using risk-set sampling and a 5:1 matching ratio on age at start of
27 employment, age at diagnosis, and cohort.
28 Both studies are based on the same exposure assessment approach. Walk-through visits,
29 interviews with managers and workers, job descriptions manual, and historical facility reports
30 supported the development of a JEM with jobs ranked on a scale of 0 (no exposure) to 3 (highly
31 exposure) on presumptive exposure reflecting relative intensity of that exposure over 3 temporal
32 periods: 1950-1960, 1970s, 1980-1990. Of the 6,044 subjects, 2,689 had TCE exposure scores
33 of >3 and 2,643 with an exposure score 3 or greater for hydrazine. Workers with job titles
34 indicating technical or mechanical work on rocket engines were presumed to have high
35 hydrazine rocket fuel exposure and high TCE exposure, which was used in cleaning rocket
36 engines and parts. Although fewer subjects had exposure to benzene (819 subjects) or mineral
37 oil (1,499 subjects), a high percentage of these subjects were also exposed to TCE. TCE use was
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-64 DRAFT—DO NOT CITE OR QUOTE
-------
1 widespread at the facility and other mechanics, maintenance and utility workers, and machinists
2 were presumed as having exposure. No details were provided for job titles other than rocket test
3 stand mechanics for assigning TCE exposure intensity and historical trends in TCE usage. Air
4 monitoring data was absent for any chemicals prior to 1985 and investigators could not link
5 study subjects to specific work locations and rocket-engine test stands. As a result, exposures
6 were probably substantially misclassified, particularly those with low to moderate TCE
7 exposure. Cumulative intensity score was the sum of the job-and time-specific intensity score
8 and years in job. Exposure classification was assigned blinded to survival status and cause of
9 death.
10 Proportional hazards modeling in calendar time with both fixed and time-depend
11 predictors was used by Zhao et al. (2005) to estimate exposure effects on site-specific cancer
12 incidence and mortality for a combined exposure group of medium and high exposure intensity
13 with workers with no to low exposure intensity as referents. Variables in the proportional hazard
14 model included time since first employment, socioeconomic status, age at diagnosis or death, and
15 exposure to other chemical agents including benzene, poly cyclic aromatic hydrocarbons (PAHs)
16 mineral oil, and hydrazine. Krishnadasen et al. (2007) fit conditional logistic regression model
17 to their data adjusting of cohort, age at diagnosis, occupation physical activity, socioeconomic
18 status and all other chemical exposure levels. Both publications include exposure-response
19 analysis and present p-values for linear trend. Race was not controlled in either study given the
20 lack of recording on personnel records. Smoking histories was available for only a small
21 percentage of the cohort; for those subjects reporting smoking information, mean cumulative
22 TCE score did not differ between smokers and nonsmokers.
23 This study develops semi quantitative exposure levels and is strength of the exposure
24 assessment. However, potential for exposure misclassification exists and would be of a
25 nondifferential direction. Rocket engine test stand mechanics had likely exposure to TCE,
26 kerosene, and hydrazine fuels; no information is available as to exposure concentrations.
27 Statistical analyses in both Zhao et al. (2005) and Krishnadansan et al. (2007) present risk
28 estimates for TCE that were adjusted for these other chemical exposures. Other strengths of this
29 study include a long follow-up period for mortality, greater than an average time of 29 years of
30 which 16 at SSFL, use of internal referents and the examination of cancer incidence, although
31 under ascertainment of cases is likely given only 8 state cancer registries were used to identify
32 cases and incidence ascertained after 1981, 40 years after the cohort's initial definition date.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-65 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
ON O
•••2
i
Krishnadasan A, Kennedy N, Zhao Y, Morgenstern H, Ritz B. 2007. Nested case-control study of occupational chemical
exposures and prostate cancer in aerospace and radiation workers. Am J Ind Med 50:383-390.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Nested case-control study of the UCLA chemical and radiation cohorts (Morgenstern
et al., 1997, 1999) to assess occupational exposures including TCE and prostate
cancer.
4,607 radiation cohort + 6,107 Santa Susana chemical cohort (Ritz et al., 1999; Zhao
et al., 2005), excluded 1,410 deaths before 1988 (date of cancer incidence follow-up).
Incident prostate cancer cases identified from eight State cancer registries (California,
Nevada, Arizona, Texas, Washington Florida, Arkansas, and Oregon). Controls were
randomly selected from the original cohorts using risk-set sampling.
362 cases and 1,805 controls (100% participation rate).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Prostate cancer incidence.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
TCE exposure assigned to cases and controls based on longest job held at company as
identified from personnel records. Cumulative exposure — ranked exposure intensity
score for TCE by 3 time periods — using method of Zhao et al. (2005).
Blinded ranking of exposure status.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
§
i
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Employment records were used to assign exposure. 734 subjects (249 cases and
485 controls, or 33% of all cases and controls) were interviewed via telephone or sent
a mailed questionnaire to obtain medical history, education and personal information
on physical activity level and smoking history.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No proxy interviews.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Any TCE exposure: 135 cases (37%) and 668 controls (37%).
High cumulative TCE exposure: 45 cases (12%) and 124 controls (7%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Cohort, age at diagnosis, occupational physical activity, SES, other chemical
exposures (benzene, PAHs, mineral oil, hydrazine). No adjustment for race due to
lacking information; affect of race on OR examined using information from survey
workers still alive in 1999. Few African American workers (n = 7), TCE levels did
not vary greatly with race.
of
Crude and adjusted conditional logistic regression.
/7-value for trend with exposure lag (0 yrs, 20 yr).
Adequate.
o §
H I
O >
HH Oq
H TO
OR=odds ratio. SES= socio-economic status.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
oo o
i
H 3
§
H I
O >
HH Oq
H TO
O
Zhao Y, Krishnadasan A, Kennedy N, Morgenstern H, Ritz B. 2005. Estimated effects of solvents and mineral oils on cancer
incidence and Mortality in a cohort of aerospace workers. Am J Ind Med 48:249-258.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort studies
of exposure and control groups and of cases and
controls in case-control studies is adequate
From introduction "one aim of this new investigation was to determine whether
these aerospace workers also developed cancers from exposures to other chemicals
including trichloroethylene (TCE), polycyclic aromatic hydrocarbons (PAHs),
mineral oils, and benzene."
6,107 male workers employed for 2 or more years and before 1980 at Santa Susana
Field Laboratory. Internal referents (no or low TCE exposure).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence between 1988-2000.
Mortality between 1950-2001.
ICD-0 for cancer incidence. Leukemia, lymphomas, and other lymphopoietic
malignancies grouped on the basis of morphology codes.
Mortality: ICD-9, before 1998, and ICD-10 thereafter. Incidence: ICD-Oncology
Lymphoma and leukemia grouping includes lymphosarcoma and reticulosarcoma,
Hodgkin's disease, other malignant neoplasm of the lymphoid and histiocytic tissue,
multiple myeloma and immunoproliferative neoplasms, and all leukemias except
chronic lymphoid leukemia. The following incident tumors were also included:
Hodgkin's disease, leukemia, polycythemia vera, chronic myeloproliferative
disease, myelosclerosis, eosinophilic conditions, platelet diseases, and red blood cell
diseases.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Cumulative exposure — ranked exposure intensity score for TCE by 3 time periods
Blinded ranking of exposure status.
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
o
•••2
i
^§
H I
O >
HH Oq
H TO
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
99% follow-up for mortality (6,044 of 6,107 subjects).
Average latency = 29 yrs (Ritz et al., 1999).
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
600 cancer deaths, 621 cancer cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Time since first employment, SES, age (at incidence or mortality), exposure to other
carcinogens, including hydrazine. No adjustment for race. Indirectly assessment of
smoking through examination of smoking distribution by chemical exposure. Mean
TCE cumulative exposure scores of smokers and nonsmokers is not statistically
significant different.
Cox proportional hazards modeling in calendar time with both fixed and
time-dependent predictors.
Exposure lagged 10 and 20 yrs.
Test for monotonic trend of cumulative exposure, two-sided p-value for trend.
Liver cancer results are not reported in published paper.
O
SES = socio-economic status.
H
W
-------
1 B.3.1.1.1.3. Comment on the Santa Susanna Field Laboratory (SSFL) studies. Rocketdyne
2 workers at SSFL are subject of two separate and independent studies. Both research groups draw
3 subjects from the same underlying source population, Rocketdyne workers including those at
4 SSFL, however, the methods adopted to identify study subjects and to define TCE exposure
5 differ with each study. A subset of SSFL workers is common to both studies; however, no
6 information exist in final published reports (Morgenstern et al., 1997, 1999; LEI, 2005) to
7 indicate the percentage overlap between cohorts or between observed number of site-specific
8 events.
9 Notable differences in both study design and analysis including cohort identification,
10 endpoint, exposure assessment approaches, and statistical methods exist between Zhao et al.
11 (2005) and Krishnadasan et al. (2007), whose source population is the UCLA cohort, and Boice
12 et al. (2006a) whose source population is the IEI cohort. A perspective of each study's
13 characteristics may be obtained from Table B-6, below.
14
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-70 DRAFT—DO NOT CITE OR QUOTE
-------
Table B-6. Characteristics of epidemiologic investigations of Rocketdyne workers
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
^
*l
^§
Study
Source population
TCE subcohort
Pay-type (hourly)
Job title with
potential TCE
exposure
Exposure metric
Endpoint
Statistical analysis
Observed number
of deaths:
Total cancer
Lung
Kidney
Bladder
NHL/Leukemia
Boice et al. (2006a)
41,351 administrative/scientific and nonadministrative male
and female employees between 1949-1999 at Rocketdyne
SSFL and two nearby facilities
1,111 male test stand mechanics with potential TCE exposure
100% of TCE subcohort
Test stand mechanics identified with greatest potential for
TCE exposure
Other job titles with direct association with test stand work —
instrument mechanics, inspectors, test stand engineers, and
research engineers — identified with lower exposure potential
to TCE and included in referent population
Qualitative, yes/no, and employment duration
Mortality as of 1999
Standardized mortality ratio
Proportional hazards modeling with covariates for birth year,
hire year, and potential exposure to hydrazine.
121
51
7
5
6
Zhao et al. (2005)
-55,000 subjects of SSFL and two nearby facilities employed between 1950 and
1993
6,107 males working at SSFL before 1980 and identified as test stand personnel,
of whom 2,689 males had exposure scores greater than no- to low-TCE
exposure potential
11.3%
High potential exposure group included job titles as propulsion/test mechanics
or technicians; Medium potential exposure group included propulsion/test
inspector, test or research engineer, and instrumentation mechanic;
Low -exposure potential included employees who, according to job title may
have been present during engine test firings but without direct contact
Cumulative exposure score = £ (exposure score (0-3) x number of years in job)
Mortality as of 2001 and Incidence as of 2000
Proportional hazards modeling with covariates for time since first employment,
socioeconomic status, age at event, and exposure to all other carcinogens,
including hydrazine
600
No/low, 99
Medium, 62
High, 33
No/low, 7
Medium, 7
High, 3
No/low, 8
Medium, 6
High, 3
No/low, 27
Medium, 27
High, 6
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 A number of strengths and limitations underlie these studies. First, the Zhao et al. (2005)
2 and Krishnadasan et al. (2007) analyses is of a larger population and of more cancer cases or
3 deaths; 600 cancer deaths and 691 cancer cases in Zhao et al. (2005) compared to 121 cancer
4 deaths in the TCE subcohort of Boice et al. (2006a), and for prostatic cancer among all
5 Rocketdyne workers, 362 incident prostatic cancer cases in Krishnandasan et al. (2007)
6 compared to 193 deaths in Boice et al. (2006a). Second, exposed populations appear
7 appropriately selected in the three studies although questions exist regarding the referent
8 population in Boice et al. (2006a) whose referent population included subjects with some direct
9 association with test stand work but whose job title was other than test stand mechanic. As a
10 result, it appears that these studies identify TCE exposure potential different for possibly similar
11 job titles. For example, jobs as instrument mechanics, inspectors, test stand engineers, and
12 research engineers are identified with medium potential exposure in Zhao et al. (2005). Boice et
13 al. (2006a) on the other hand included these subjects in the referent population and assumed they
14 had background exposure. TCE use at SSFL was also widespread and rocket engine cleaning
15 occurred at other locations besides at test sites (Morgenstern et al., 1999), locations from which
16 the referent population in Boice et al. (2006a) arose. If referents in Boice et al. (2006a) had more
17 than background exposure, the bias introduced leads to an underestimation of risk. Third, Zhao
18 et al. (2005) and Krishnadasan et al. (2007) studies include an examination of incidence, and are
19 likely to have a smaller bias associated with disease misclassification than Boice et al. (2006a)
20 who examines only mortality. Fourth, use of cumulative exposure score although still subject to
21 biases is preferred to qualitative approach for exposure assessment. Last, all three studies
22 adjusted for potentially confounding factors such as smoking, socioeconomic status, and other
23 carcinogenic exposures using different approaches either in the design of the study, such as
24 Boice et al. (2006a) limitation to only hourly workers, or in the statistical analysis such as Zhao
25 et al. (2005) and Krishnadansen et al. (2007). For this reason, the large difference in hourly
26 workers between the UCLA cohort and Boice et al. (2006a) is not likely to greatly impact
27 observations.
28
29 B.3.1.1.2. Blair et al (1998), Radican et al (2008).
30 B.3.1.1.2.1. Radican et al (2008) abstract.
31
32 OBJECTIVE: To extend follow-up of 14,455 workers from 1990 to 2000, and
33 evaluate mortality risk from exposure to trichloroethylene (TCE) and other
34 chemicals. METHODS: Multivariable Cox models were used to estimate relative
35 risk (RR) for exposed versus unexposed workers based on previously developed
36 exposure surrogates. RESULTS: Among TCE-exposed workers, there was no
37 statistically significant increased risk of all-cause mortality (RR = 1.04) or death
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-72 DRAFT—DO NOT CITE OR QUOTE
-------
1 from all cancers (RR = 1.03). Exposure-response gradients for TCE were
2 relatively flat and did not materially change since 1990. Statistically significant
3 excesses were found for several chemical exposure subgroups and causes and
4 were generally consistent with the previous follow-up. CONCLUSIONS: Patterns
5 of mortality have not changed substantially since 1990. Although positive
6 associations with several cancers were observed, and are consistent with the
7 published literature, interpretation is limited due to the small numbers of events
8 for specific exposures.
9
10 B.3.1.1.2.2. Blair et al (1998) abstract.
11
12 OBJECTIVES: To extend the follow up of a cohort of 14,457 aircraft
13 maintenance workers to the end of 1990 to evaluate cancer risks from potential
14 exposure to trichloroethylene and other chemicals. METHODS: The cohort
15 comprised civilians employed for at least one year between 1952 and 1956, of
16 whom 5727 had died by 31 December 1990. Analyses compared the mortality of
17 the cohort with the general population of Utah and the mortality and cancer
18 incidence of exposed workers with those unexposed to chemicals, while adjusting
19 for age, sex, and calendar time. RESULTS: In the combined follow up period
20 (1952-90), mortality from all causes and all cancer was close to expected
21 (standardized mortality ratios (SMRs) 97 and 96, respectively). Significant
22 excesses occurred for ischemic heart disease (SMR 108), asthma (SMR 160), and
23 cancer of the bone (SMR 227), whereas significant deficits occurred for
24 cerebrovascular disease (SMR 88), accidents (SMR 70), and cancer of the central
25 nervous system (SMR 64). Workers exposed to trichloroethylene showed non-
26 significant excesses for non-Hodgkin's lymphoma (relative risk (RR) 2.0), and
27 cancers of the oesophagus (RR 5.6), colon (RR 1.4), primary liver (RR 1.7),
28 breast (RR 1.8), cervix (RR 1.8), kidney (RR 1.6), and bone (RR 2.1). None of
29 these cancers showed an exposure-response gradient and RRs among workers
30 exposed to other chemicals but not trichloroethylene often had RRs as large as
31 workers exposed to trichloroethylene. Workers exposed to solvents other than
32 trichloroethylene had slightly increased mortality from asthma, non-Hodgkin's
33 lymphoma, multiple myeloma, and breast cancer. CONCLUSION: These
34 findings do not strongly support a causal link with trichloroethylene because the
35 associations were not significant, not clearly dose-related, and inconsistent
36 between men and women. Because findings from experimental investigations and
37 other epidemiological studies on solvents other than trichloroethylene provide
38 some biological plausibility, the suggested links between these chemicals and
39 non-Hodgkin's lymphoma, multiple myeloma, and breast cancer found here
40 deserve further attention. Although this extended follow up cannot rule out a
41 connection between exposures to solvents and some diseases, it seems clear that
42 these workers have not experienced a major increase in cancer mortality or cancer
43 incidence.
44
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-73 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.3.1.1.2.3. Study description and comment. This historical cohort study of 14,457
2 (9,400 male and 3,138 female) civilian personnel employed at least one year between 1942 and
3 1956 at Hill Air Force Base in Utah examines mortality to the end of 1982 (Spirtas et al., 1991)
4 to the end of 1990 (Blair et al., 1998), or to the end of 2000 (Radican et al., 2008). About half of
5 the cohort was identified with exposure to TCE (6,153 white men and 1,051 white women).
6 One-fourth of subjects were born before 1909 with an attained age of 43 years at cohort's
7 identification date of 1952 and whose first exposure could have been as early as 1939, a cohort
8 considered as a "survivor cohort."
9 As of December 2008, the end of follow-up in Radican et al. (2008), 8,580 deaths (3,628
10 in TCE subcohort) were identified, an increase of 2,853 deaths with the additional 8 years
11 follow-up period compared to Blair et al. (1998) (5,727 total deaths, 2,813 among TCE
12 subcohort subjects), with a larger proportion deaths among non-TCE exposed subjects (58%) as
13 of December 2008 compared to the December 2000 (51%). Approximately 50% of
14 TCE-exposed subjects and 60% of all cohort subjects had died, with mean age of 75 years for
15 TCE-exposed subjects still alive and 45 or more years since the cohort's definition (1953 to
16 1955), a time period longer than that typically considered for an induction or latent window for
17 detecting an adverse outcome like cancer. Blair et al. (1998) additionally examined cancer
18 incidence among white TCE-exposed workers alive on 1-1-1973, a period of 31 years after the
19 cohort's inception date, to the end of 1990. Incident cancer cases are likely under ascertained for
20 this reason.
21 Statistical analyses in Spirtas et al. (1991) and Blair et al. (1998) focus on site-specific
22 mortality for white subjects or subjects with unknown race who were assumed to as white since
23 97% of all subjects with know race were white. SMRs are presented with expected numbers of
24 deaths based upon age-, race- and year-specific mortality rates of the Utah population (Spirtas et
25 al., 1991; Blair et al., 1998) or rate ratios for mortality or cancer incidence for the TCE subcohort
26 from Poisson regression models, adjusting for date of birth, calendar year of death, and sex
27 where appropriate, and an internal standard of mortality rates of the cohort's nonchemical
28 exposed subjects (internal referents) (Blair et al., 1998). Blair et al. (1998), in addition to their
29 presentation in the published papers of risk estimates associated with TCE exposure, also,
30 presented risk estimates for subjects with an aggregated category of "any solvent exposure" (ever
31 exposed) and for exposure to 14 solvents. To compare with risk ratios from Poisson regression
32 models of Blair et al. (1998), Radican et al. (2008) adopted Cox proportional hazard models to
33 reanalyze mortality observations of follow-up through 1990. For most site-specific cancers,
34 Radican et al. (2008) did not observe large differences between the Cox hazard ratio and Poisson
35 rate ratio of Blair et al. (1998), although difference between risk estimates from Cox proportional
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-74 DRAFT—DO NOT CITE OR QUOTE
-------
1 hazard and Poisson regression of 20% or larger was observed for kidney cancer (increased risk
2 estimate) and primary liver cancer (decreased risk estimate). Radican et al. (2008), furthermore,
3 noted hazard ratios for all subjects were similar to results for white subjects only; therefore, their
4 analyses of follow-up through 2000 included all subjects.
5 The original exposure assessment of Stewart et al. (1991) who conducted a detailed
6 exposure assessment of TCE exposures at Hill Air Force Base was used by Radican et al. (2008),
7 Blair et al. (1999), and Spirtas et al. (1991). Their was limited for linking subjects with
8 exposures principally because solvent exposures were associated with work in "shops," but work
9 records listed only broad job titles and administrative units. As a result, exposures were
10 probably substantially misclassified, particularly in "mixed solvent group." Trichloroethylene
11 was used principally for degreasing and hand cleaning in work areas during 1955-1968. TCE
12 was the predominant solvent used in the few available vapor degreasers located in the
13 electroplating (main hanger), propeller, and engine repair shops before the mid-1950 and,
14 afterwards, as a cold state solvent, replacing Stoddard solvent. Solvents, notably TCE after
15 1955, were used primarily by aircraft mechanics with short but high exposures and sheet metal
16 workers for spot clean aircraft surfaces. The investigators determined that 32% had "frequent"
17 exposures to peak concentrations (one or two daily peaks of about 15 minutes to
18 trichloroethylene at 200-600 ppm) during vapor degreasing. Work areas were located in very
19 large buildings with few internal partitions, which aided dispersion of trichloroethylene. While
20 TCE exposures were less controlled in the 1950s, by the end of 1960s, TCE exposure had been
21 reduced significantly. Only a small number of subjects with "high" exposure had long-duration
22 exposures, no more than 16%. Few workers were exposed only to trichloroethylene; most had
23 mixed exposures to other chlorinated and nonchlorinated solvents. Person-years of exposure
24 were computed from date of first exposure, which could have been as early as 1939, to the end of
25 1982.
26 Overall, Blair et al. (1998) and Radican et al. (2008) are high quality studies with
27 approximately half of the larger cohort identified as having some potential for TCE exposure (the
28 TCE subcohort) and calculation of cancer risk estimates for TCE exposure, either risk ratios in
29 Blair et al. (1998) or hazard ratios in Radican et al. (2008), using workers in the cohort without
30 any chemical exposures as referent population, superior to standardized mortality ratios of
31 Spirtas et al. (1991) who first reported on mortality and TCE exposure. Use of an internal
32 referent population of workers from the same company or plant, but lacking the exposure of
33 interest, is considered to reduce bias associated with the healthy worker effect. For follow-up in
34 Radican et al. (2008) who examined mortality 45 years after first exposure and likely at the tail
35 of or beyond a window for cancer induction time, any influence on exposure on disease
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-75 DRAFT—DO NOT CITE OR QUOTE
-------
1 development or detection times would be diminshed or less evident if exposures like TCE
2 shortened induction time, e.g., if exposure shortened the natural course of disease development,
3 which would become evident in an unexposed subjects with longer follow-up periods. The
4 induction time of 35 years in Blair et al. (1998) may also fall outside a cancer induction window;
5 however, it is more consistent with cancer induction times observed with other chemical
6 carcinogens such as aromatic amines (Weistenhofer et al., 2008) and vinyl chloride (Du and
7 Wang, 1998). A strong exposure assessment was performed, but precision in the exposure
8 assignment was limited by vague personnel data. The cohort had a modest number of highly
9 exposed (about 100 ppm) subjects, but overall most were exposed to low concentrations (about
10 10 ppm) of tri chl oroethy 1 ene.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-76 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
?l
Radican L, Blair A, Stewart P, Wartenberg D. 2008. Mortality of aircraft maintenance workers exposed to trichloroethylene
and other hydrocarbons and chemicals: extended follow-up. J Occup Environ Med 50:1306-1319.
Blair A, Hartge P, Stewart PA, McAdams M, Lubin J. 1998. Mortality and cancer incidence of aircraft maintenance workers
exposed to trichloroethylene and other organic solvents and chemicals: extended follow-up. Occup Environ Med 55:161-171.
Spirtas R, Stewart PA, Lee JS, Marano DE, Forbes CD, Grauman DJ, Pettigrew HM, Blair A, Hoover RN, Cohen JL. 1991.
Retrospective cohort mortality study of workers at an aircraft maintenance facility. I. Epidemiological results. Br J Ind Med
48:515-530.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Abstract: "...to evaluate cancer risks from potential exposure to trichloroethylene and
other chemicals."
All civilians employed at Hill AFB for >1 yr between 1-1-1952 and 12-31-1956;
cohort of 14,457 workers identified form earnings records.
TCE subcohort — 7,204 white males and females (50%).
External referents, all civilian cohort — Utah population rates, 1953-1990.
Internal referents, TCE subcohort analysis of mortality (Blair et al., 1998; Radican
al., 2008) and incidence (Blair et al., 1998) — workers without chemical exposures.
et
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality, all civilian cohort and TCE subcohort.
Incidence, TCE subcohort.
Underlying and contributing causes of deaths as coded to ICDA 8.
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
51
oo o
H I
O >
HH Oq
H ^
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Detailed records on setting and job activities, worker interviews; work done in large
open shops; shops not recorded in personnel records, link of job with IH data was
weak. Limited exposure IH measurements for TCE between 1960-1990. Plant JEM,
rank order assignments by history; determined exposure duration during vapor
degreasing tasks about 2,000 ppm-h and hard degreasing about 20 ppm-h. Median
exposure were about 10 ppm for rag and bucket (cold degreasing process);
100-200 ppm for vapor degreasing (Stewart et al., 1991). Cherrie et al. (2001)
estimated long-term exposure as ~50 ppm with short-term excursion up to
-600 ppm. NRC (2006) concluded the cohort had a modest number of highly
exposed (about 100 ppm) subjects, but overall most were exposed to low TCE
concentrations (about 10 ppm).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
97% of cohort traced successfully to 12-31-1982.
Yes, all subjects followed minimum of 35 yrs (Blair et al., 1998) or 45 yrs (Radican et
al., 2008).
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
TCE subcohort — 2,813 deaths (39%), 528 cancer deaths, and 549 incident cancers
(1973-1990) (Blair et al., 1998); 3,628 deaths (50%). 729 cancer deaths (Radican et
al., 2008).
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
SMR analysis evaluates age, sex, and calendar year (Spirtas et al., 1991).
Date of hire, calendar year of death, and sex in Poisson regression analysis (Blair et
al., 1998).
Age, gender, and race (to compare with RR of Blair et al.,[1998], or age and gender
for follow-up to 2000] in Cox proportional hazard analysis (Radican et al., 2008).
External analysis is restricted to Caucasian subjects — Life table analysis for mortality
(Spirtas et al., 1991).
Internal analysis restricted to Caucasian subjects or subject of unknown race assumed
to be Caucasian and followed to 1990 — Poisson regression (Blair et al., 1998) or Cox
Proportional Hazard (Radican et al., 2008).
Internal analysis — all subjects followed to 2000 (Radican et al., 2008).
Risk ratios from Poisson regression model and hazard ratios from Cox Proportional
Hazard model for exposure rankings but no formal statistical trend test presented in
papers.
Adequate.
•3
1
td
VO O
•••2
i
RR = relative risk.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.1.3. Boice et al (1999).
2 B.3.1.1.3.1. Author's abstract.
3
4 OBJECTIVES: To evaluate the risk of cancer and other diseases among workers
5 engaged in aircraft manufacturing and potentially exposed to compounds
6 containing chromate, trichloroethylene (TCE), perchloroethylene (PCE), and
7 mixed solvents. METHODS: A retrospective cohort mortality study was
8 conducted of workers employed for at least 1 year at a large aircraft
9 manufacturing facility in California on or after 1 January 1960. The mortality
10 experience of these workers was determined by examination of national, state,
11 and company records to the end of 1996. Standardized mortality ratios (SMRs)
12 were evaluated comparing the observed numbers of deaths among workers with
13 those expected in the general population adjusting for age, sex, race, and calendar
14 year. The SMRs for 40 causes of death categories were computed for the total
15 cohort and for subgroups defined by sex, race, and position in the factory, work
16 duration, year of first employment, latency, and broad occupational groups.
17 Factory job titles were classified as to likely use of chemicals, and internal
18 Poisson regression analyses were used to compute mortality risk ratios for
19 categories of years of exposure to chromate, TCE, PCE, and mixed solvents, with
20 unexposed factory workers serving as referents. RESULTS: The study cohort
21 comprised 77,965 workers who accrued nearly 1.9 million person-years of follow
22 up (mean 24.2 years). Mortality follow-up, estimated as 99% complete, showed
23 that 20,236 workers had died by 31 December 1996, with cause of death obtained
24 for 98%. Workers experienced low overall mortality (all causes of death SMR
25 0.83) and low cancer mortality (SMR 0.90). No significant increases in risk were
26 found for any of the 40 specific causes of death categories, whereas for several
27 causes the numbers of deaths were significantly below expectation. Analyses by
28 occupational group and specific job titles showed no remarkable mortality
29 patterns. Factory workers estimated to have been routinely exposed to chromate
30 were not at increased risk of total cancer (SMR 0.93) or of lung cancer (SMR
31 1.02). Workers routinely exposed to TCE, PCE, or a mixture of solvents also were
32 not at increased risk of total cancer (SMRs 0.86, 1.07, and 0.89, respectively), and
33 the numbers of deaths for specific cancer sites were close to expected values.
34 Slight to moderately increased rates of non-Hodgkin's lymphoma were found
35 among workers exposed to TCE or PCE, but none was significant. A significant
36 increase in testicular cancer was found among those with exposure to mixed
37 solvents, but the excess was based on only six deaths and could not be linked to
38 any particular solvent or job activity. Internal cohort analyses showed no
39 significant trends of increased risk for any cancer with increasing years of
40 exposure to chromate or solvents.
41 The results from this large scale cohort study of workers followed up for over
42 3 decades provide no clear evidence that occupational exposures at the aircraft
43 manufacturing factory resulted in increases in the risk of death from cancer or
44 other diseases. Our findings support previous studies of aircraft workers in which
45 cancer risks were generally at or below expected levels.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-80 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.3.1.1.3.2. Study description and comment. This study was conducted on an aircraft
2 manufacturing worker cohort employed at Lockheed-Martin in Burbank, California with
3 exposure assessment described by Marano et al. (2000). This large cohort study of
4 77,965 subject workers with at least 1 year employment on or after 1-1-1960, examined causes
5 of mortality in the entire cohort, but also by broad job titles and for selected chemical exposures
6 including TCE. Mortality was assessed as of 12-31-1996, with subjects lacking death certificates
7 presumed alive at end of follow-up. Exposure assessment developed using a method of exposure
8 assignment by j ob categories based on j ob histories (Kardex cards) and the judgment of
9 long-term employees. Job histories were not available for every worker, and, if missing,
10 auxiliary sources of job information were used to broadly classify workers into various job
11 categories. Only subjects with job histories as recorded on Kardex cards are included in
12 exposure duration analyses. TCE was used for vapor degreasing on routine basis prior to 1966
13 and, given the cohort beginning date of 1960, only a small percentage of the total cohort was
14 identified as having potential TCE exposure. The investigators determined that 5,443 factory
15 workers had potential TCE exposure. Of these subjects, 3% (2,267 out of 77,965 subjects) had
16 "routine" defined as use of TCE as part of daily job activities and an additional 3,176 subjects
17 (4%) had potential "intermittent" based upon j ob title and judgment of nonroutine or nondaily
18 TCE usage and were included in the mortality analysis. No information was provided on
19 building and working conditions or the frequency of exposure-related tasks, and no atmospheric
20 monitoring data were available on TCE, although some limited data were available after 1970 on
21 other solvents such as perchloroethylene, which replaced TCE in 1966 in vapor degreasing,
22 methylene chloride, and 1,1,1-trichloroethane. Without more information, it is not possible to
23 determine the quality of some of the TCE assignments. This study had limited ability to detect
24 exposure-related effects given its use of duration of exposure, a poor exposure metric given
25 subjects may have differing exposure intensity with similar exposure duration (NRC, 2006).
26 Lacking monitoring information, analyses examining the number of years of routine and
27 intermittent TCE exposure are likely biased due to exposure misclassification related to inability
28 to account for changes in process and chemical usage patterns over time. Stewart et al. (1991)
29 show atmospheric TCE concentrations decreased over time. Similarly, an observation of inverse
30 relationship between some site-specific causes of death and duration of exposure may be due to
31 selection bias or to misallocation of person-years of follow-up (NYS DOH, 2006).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-81 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
Boice JD, Marano DE, Fryzek JP, Sadler CJ, McLaughlin JK. 1999. Mortality among aircraft manufacturing workers.
Occup Environ Med 56:581-597.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract: "To evaluate the risk of cancer and other diseases among workers
engaged in aircraft manufacturing and potentially exposed to compounds containing
chromate, trichloroethylene (TCE), perchloroethylene (PCE), and mixed solvents."
All workers employed on or after 1-1-1960 for at least 1 yr at Lockheed Martin
aircraft manufacturing factories in California.
Control population: U.S. mortality rates or factory workers no exposed to any solvent
(internal referents).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD code in use at the time of death.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Qualitative. Few exposure measurements existed prior to the late 1970s, a period
after TCE had been discontinued at Lockheed-Martin aircraft manufacturing
factories.
Subjects are categorized as potentially TCE exposed received on a routine basis
(2,075 subjects), daily job activity, or routine and intermittent basis (3,016 subjects),
nonroutine or nondaily TCE usage, based on information on Service Record and
Permanent Employment Record (Kardex) and other sources of job history
information for subjects lacking Kardex cards.
•3
1
oo ^
to O
*5
td
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
W 1
oo ^
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
This study does not adopt methods to verify vital status of employees. All workers
for which death certificate were not found are assumed to be alive until end of
follow-up.
Average follow-up of TCE cohort was 29 yrs.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,100 total deaths and 277 cancer deaths in TCE subcohort.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
SMR analysis — age, sex and calendar-time.
Poisson regression using internal referents — birth date, date first employed, date of
finishing employment, race, and sex.
SMR for routine TCE exposure subcohort.
Poisson regression for routine and intermittent TCE exposure subcohort.
Duration of exposure for subjects with Kardex cards only —
2-sides test for linear trend.
Adequate.
H I
O >
HH Oq
H ^
O
H
W
-------
1 B.3.1.1.4. Morgan et al (1998, 2000).
2 B.3.1.1.4.1. Author's abstract.
3
4 We measured mortality rates in a cohort of 20,508 aerospace workers who were
5 followed up over the period 1950-1993. A total of 4,733 workers had
6 occupational exposure to trichloroethylene. In addition, trichloroethylene was
7 present in some of the washing and drinking water used at the work site. We
8 developed a job-exposure matrix to classify all jobs by trichloroethylene exposure
9 levels into four categories ranging from "none" to "high" exposure. We calculated
10 standardized mortality ratios for the entire cohort and the trichloroethylene
11 exposed subcohort. In the standardized mortality ratio analyses, we observed a
12 consistent elevation for nonmalignant respiratory disease, which we attribute
13 primarily to the higher background rates of respiratory disease in this region. We
14 also compared trichloroethylene-exposed workers with workers in the "low" and
15 "none" exposure categories. Mortality rate ratios for nonmalignant respiratory
16 disease were near or less than 1.00 for trichloroethylene exposure groups. We
17 observed elevated rare ratios for ovarian cancer among those with peak exposure
18 at medium and high levels] relative risk (RR) = 2.74; 95% confidence interval
19 (CI) = 0.84-8.99] and among women with high cumulative exposure (RR = 7.09;
20 95% CI = 2.14-23.54). Among those with peak exposures at medium and high
21 levels, we observed slightly elevated rate ratios for cancers of the kidney (RR =
22 1.89; 95% CI = 0.85-4.23), bladder (RR = 1.41; 95% CI = 0.52-3.81), and
23 prostate (RR = 1.47; 95% CI = 0.85-2.55). Our findings do not indicate an
24 association between trichloroethylene exposure and respiratory cancer, liver
25 cancer, leukemia or lymphoma, or all cancers combined.
26
27 Erratum:
28
29 One of the authors of the article entitled Mortality of aerospace workers exposed
30 to trichloroethylene, by Robert W. Morgan, Michael A. Kelsh, Ke Zhao, and
31 Shirley Heringer, published in Epidemiology 1998;9:424-431, informed us of
32 some errors in one of the tables. In Table 5, the authors had inadvertently included
33 both genders in counting person-years, rather than presenting gender-specific risk
34 ratios for prostate and ovarian cancer. In addition, one subject, in the high
35 trichloroethylene (TCE) exposure category, had been incorrectly classified with a
36 diagnosis of ovarian cancer, instead of other female genital cancer. The authors
37 report that correction of these errors did not change the overall conclusions of the
38 study. The correct estimates of effect for prostate and ovarian cancer are
39 presented in the Table below.
40
41 B.3.1.1.4.2. Study description and comment. This study of a cohort of 20,508 aircraft
42 manufacturing workers employed for at least 6 months between 1950 and 1985 at Hughes
43 Aircraft in Arizona was followed through 1993 for mortality. Cause-specific SMRs are resented
44 for the entire cohort and the TCE-subcohort using U.S. Mortality rates from 1950-1992 as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-84 DRAFT—DO NOT CITE OR QUOTE
-------
1 referents. Additionally, internal cohort analyses fitting Cox proportional hazards models are
2 presented comparing risks for those with TCE exposure to never-exposed subjects. Morgan et al.
3 (1998, 2000) do not identify job titles of individuals in the never-exposed group; however, it is
4 assumed these individuals were likely white-collar workers, administrative staff, or other
5 blue-collar worker with chemical or solvents exposures other than TCE.
6 The company conducted a limited semi quantitative assessment of TCE exposure based
7 on the judgment of long-term employees. Most TCE exposure occurred in vapor degreasing
8 units between 1952 and 1977. No details were provided on the protocol for processing the jobs
9 in the work histories into job classifications; no examples were provided. Additionally, no
10 information is provided other chemical exposures that may also have been used in the different
11 jobs. Of the 20,508 subjects, 4,733 were identified with TCE exposure. Exposure categories
12 were assigned to job classifications: high = worked on degreasers (industrial hygiene reported
13 exposures were >50 ppm); medium = worked near degreasers; and low = work location was
14 away from degreasers but "occasional contact with (trichloroethylene)." There was also a "no
15 exposure" category. No data were provided on the frequency of exposure-related tasks. Without
16 more information, it is not possible to determine the quality of some of these assignments. Only
17 the high category is an unambiguous setting. Depending on how the degreasers were operated,
18 operator exposure to trichloroethylene might have been substantially greater than 50 ppm.
19 Furthermore, TCE intensity likely changed over time with changes in degreaser operations and
20 exposure assignment based on job title only is able to correctly place subjects with a similar job
21 title but held at different time periods. Furthermore, there are too many possible situations in
22 which an exposure category of medium or low might be assigned to determine whether the
23 ranking is useful. Therefore, the medium and low rankings are likely to be highly misclassified.
24 Deficiencies in job rankings are further magnified in the cumulative exposure groupings.
25 Internal analyses examine TCE exposed, defined as low and high cumulative exposure,
26 compared to never-TCE exposed subjects. Low cumulative exposure group includes any
27 workers with the equivalent of up to 5 years of exposure at jobs at low exposure or 1.4 years of
28 medium exposure; all other workers were placed in the high cumulative exposure grouping.
29 Ambiguity in low and medium job rankings and the lack of exposure data to define "medium"
30 and "low" precludes meaningful analysis of cumulative exposure, specifically, and
31 exposure-response, generally.
32 The development of exposure assignments in this study was insufficient to define
33 exposures of the cohort and bias related to exposure misclassification is likely great. The
34 inability to account for changes in TCE use and exposure potential over time introduces bias and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-85 DRAFT—DO NOT CITE OR QUOTE
-------
1 may dampen observed risks. This study had limited ability to detect exposure-related effects
2 and, overall, limited ability to provide insight on TCE exposure and cancer outcomes.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-86 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
S'
1
TO'
•3
» 1
oo ^
Morgan RW, Kelsh MA, Zhao K, Heringer S. 1998. Mortality of aerospace workers exposure to trichloroethylene.
Epidemiol 9:424-431.
Morgan RW, Kelsh MA, Zhao K, Heringer S. 2000. Mortality of aerospace workers exposed to trichloroethylene. Erratum.
Epidemiology 9:424-431.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
"measured mortality rates in a cohort of aerospace workers, comparing TCE workers
with workers in low and none exposure categories."
20,508 male and female workers are identified using company records and who were
employed at plant for at least 6 mos between 1-1-1950 and 12-31-1985.
TCE subcohort — 4,733 (23%) male and female subjects.
External referents — U.S. population rates, 1950-1992.
Internal referents — Analysis of peak exposure, Low or no TCE exposure; analysis of
cumulative exposure, never exposed to TCE. Internal referents are likely white-collar
workers, administrative staff, and blue-collar workers with chemical exposure other
than TCE. White-collar and administrative staff subjects are not representative of
blue-collar workers due to SES and sex differences. Also, the never-TCE exposed
blue-collar workers may potentially have other chlorinated solvents exposures,
exposures that may be associated with a similar array of targets as TCE. These
individuals may not be representative of a nonchemical exposed population as that
used in Blair et al. (1998).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality
No, ICD in use at time of death (ICD 7, 8, 9).
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
1
oo ^
oo o
td
H I
O >
HH Oq
H ^
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Semi quantitative. Limited IH measurements before 1975. Jobs ranked into high,
medium, or low intensity exposure categories; categories are undefined as to TCE
intensity. Jobs with high intensity exposure rating involved work on degreaser
machines with TCE exposure equivalent to 50 ppm; assigned exposure score of 9
Job with medium rating were near (distance undefined in published paper) degreasing
area and a score of 4. Jobs with low rating were away (undefined distance) from
degreasing area and assigned score of 1 . Cumulative exposure score = X (duration
exposure * score). Peak exposure defined by job with highest ranking score.
CATEGORY D: FOLLOW-UP (Cohort)
More than 10% loss to follow-up
>50% cohort with full latency
No, 27 subjects were excluded from analysis due to missing information.
Average 22 yrs of follow-up for TCE subcohort.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
TCE subcohort — 917 total deaths (19%) of subcohort, 270 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age, race, sex, and calendar year in SMR analysis.
Internal analysis- age (for bladder, prostate, ovarian cancers) and, age and sex (liver,
kidney cancers).
Life table analysis (SMR).
Cox proportional hazards modeling (unexposed subjects as internal referents) — peak
and two-levels of cumulative exposure (Environmental Health Strategies, 1997;
Morgan et al., 1998); any TCE exposure (Environmental Health Strategies, 1997).
-------
to
Exposure-response analysis presented in
published paper
Documentation of results
Qualitative presentation, only;
no formal
statistical test for linear trend.
Adequate.
'
I
I
§
SES = socio-economic status.
1
TO'
td
I
OO
•3
1
§
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.1.5. Costa et al (1989).
2 B.3.1.1.5.1. Author's abstract.
3
4 Mortality in a cohort of 8626 workers employed between 1954 and 1981 in an
5 aircraft manufacturing factory in northern Italy was studied. Total follow up was
6 132,042 person-years, with 76% accumulated in the age range 15 to 54. Median
7 duration of follow up from the date of first employment was 16 years. Vital status
8 was ascertained for 98.5% of the cohort. Standardized mortality ratios were
9 calculated based on Italian national mortality rates. Altogether 685 deaths
10 occurred (SMR = 85). There was a significant excess of mortality for melanoma
11 (6 cases, SMR = 561). Six deaths certified as due to pleural tumors occurred. No
12 significant excess of mortality was found in specific jobs or work areas.
13
14 B.3.1.1.5.2. Study description and comment. This study assesses mortality in a small cohort
15 of 8,626 aircraft manufacturing workers employed between 1954 and the end of follow-up in
16 June, 1981. A period of minimum employment duration before accumulating person-years was
17 not a prerequisite for cohort definition. The cohort included employees identified as blue collar
18 workers, technical staff, administrative clerks, and white-collar workers. Blue-collar workers
19 comprised 7,105 of the 8,626 cohort subjects. Mortality was examined for all workers and
20 included job title of blue collar workers, technical staff members, administrative clerks, and
21 white collar workers- not otherwise specified. No exposure assessment was used and the
22 published paper does not identify chemical exposures. In fact, Costa et al. (1989) do not even
23 mention TCE in the paper.
24 Overall, the lack of exposure assessment, the inability to identify TCE as an exposure to
25 this cohort, and the inclusion of subjects who likely do not have potential TCE exposure are
26 reasons why this study is not useful for determining whether trichloroethylene may cause
27 increased risk of disease.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-90 DRAFT—DO NOT CITE OR QUOTE
-------
Costas G, Merletti F, Segnan N. 1989. A mortality study in a north Italian aircraft factory. Br J Ind Med 46:738-743.
to
vo £;•
I
I
§
***.
£3'
1
TO'
w 1
vb 8
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The 1st paragraph of the paper identified this study was carried out to investigate an
apparently high number of malignant tumors among employees that were brought to
the attention of the local health authority by staff representative. This study was not
designed to examine TCE exposure and cancer outcomes.
Cohort is defined as all workers every employed between 1-1-1954 and 6-30-1981
(end of follow-up) at a north Italian aircraft manufacturing factory. Cohort include
8.626 subjects: 950 women (636 clerks, 314 blue-collar workers/technical staff) and
7,676 men (5,625 blue collar workers, 965 technical staff, 571 administrative clerks,
and 515 white collar workers).
External referent — Age, year (5-yr periods over 1955-1981)-sex and cause-specific
death rates of Italian population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
Causes and underlying causes of death coded to ICD rule in effect at the time of
death and grouped into categories consistent with ICD 8th revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Exposure is defined as employment in the factory. TCE is not mentioned in
published paper and no exposure assessment was carried out by study investigators.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Vital status ascertained for 98% of cohort; 2% could not be traced (1% unknown and
1% had emigrated).
Average mean follow-up: males, 17 yrs; females, 13 yrs.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
W 1
vb £
to O
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
642 total deaths, 168 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex and calendar year.
SMR.
No.
Adequate.
i
^§
H I
O
H
W
-------
1 B.3.1.1.6. Garabrant et al (1988).
2 B.3.1.1.6.1. Author's abstract.
3
4 A retrospective cohort mortality study was conducted among men and women
5 employed for four or more years, between 1958 and 1982, at an aircraft
6 manufacturing company in San Diego County. Specific causes of death under
7 investigation included cancer of the brain and nervous system, malignant
8 melanoma, and cancer of the testicle, which previous reports have suggested to be
9 associated with work in aircraft manufacturing. Follow-up of the cohort of 14,067
10 subjects for a mean duration of 15.8 yr from the date of first employment resulted
11 in successful tracing of 95% of the cohort and found 1,804 deaths through 1982.
12 Standardized mortality ratios (SMRs) were calculated based on U. S. national
13 mortality rates and separately based on San Diego County mortality rates.
14 Mortality due to all causes was significantly low (SMR = 75), as was mortality
15 due to all cancer (SMR = 84). There was no significant excess of cancer of the
16 brain, malignant melanoma, cancer of the testicle, any other cancer site, or any
17 other category of death. Additional analyses of cancer sites for which at least ten
18 deaths were found and for which the SMR was at least 110 showed no increase in
19 risk with increasing duration of work or in any specific calendar period. Although
20 this study found no significant excesses in cause-specific mortality, excess risks
21 cannot be ruled out for those diseases that have latency periods in excess of 20 to
22 30 yr, or for exposures that might be restricted to a small proportion of the cohort.
23
24 B.3.1.1.6.2. Study description and comment. This study reported on the overall mortality of a
25 cohort of workers in the aircraft manufacturing industry in southern California who had worked
26 1 day at the facility and had at least 4 years duration of employment. Fifty-four (54) percent of
27 cohort entered cohort at beginning date (1-1-1958). This is a survivor cohort. This study lacks
28 exposure assessment for study subjects. The only exposure metric was years of work.
29 Examination of jobs held by 70 study subjects, no details provided in paper on subject selection
30 criteria, identified 37% as having possible trichloroethylene TCE exposure, but no information
31 was presented on how they were exposed, frequency or duration of exposure, or job titles
32 associated with exposure. No information is provided on possible trichloroethylene exposure to
33 the remaining -14,000 subjects in this cohort. The exposure assignment in this study was
34 insufficient to define exposures of the cohort and the frequency of exposures was likely low.
35 Given the enormous misclassification on exposure, the effect of exposure would have to be very
36 large to be detected as an overall risk for the population. Null findings are to be expected due to
37 bias likely associated with a survivor cohort and to exposure misclassification. Therefore, this
38 study provides little information on whether trichloroethylene is related to disease risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-93 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
w 1
vb 8
Garabrant DH, Held J, Langholz B, Bernstein L. 1988. Mortality of Aircraft Manufacturing Workers in Southern California.
Am J Ind Med 13:683-693.
Langholz B, Goldstein L. 1996. Risk Set Sampling in Epidemiologic Cohort Studies. Stat Sci 11:35-53.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
"Our objects were to evaluate the oval mortality among the [aircraft manufacturing]
workers and to test the hypotheses that brain tumors, malignant melanoma, and
testicular neoplasms are associated with work in this industry." [Introduction]
This study was not designed to evaluate any specific exposure, but rather
employment in aircraft manufacturing industry.
14,067 males and females working at least 4 yrs with a large aircraft manufacturing
company and who had worked for at least 1 day at a factory in San Diego County,
CA. Person-year accrued from the anniversary date of an individual's 4th yr of
service or from 1-1-1958 to end of follow-up 12-31-1982.
External referents — age-, race-, sex-, calendar year- and cause-specific mortality
rates of United States population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality
ICD revision in effect at the date of death. Lymphomas in 4 groupings:
lymphosarcoma and reticulosarcoma, HD, leukemia and aleukemia, and other.
CATEGORY C: TCE-EXPOSURE CRITERIA
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
ICD revision in effect at the date of death. Lymphomas in 4 groupings:
lymphosarcoma and reticulosarcoma, HD, leukemia and aleukemia, and other.
Exposure assessment is lacking for all subjects except 70 deaths (14 esophageal and
56 others) who were included in a nested case-control study. Of the 362 jobs held by
these 70 subjects, 37% were identified as having potential for TCE exposure.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
w 1
vb 8
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
4.7% with unknown vital status.
Average 16 yr follow-up.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,804 deaths (12.8% of cohort), 453 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, race, sex, and calendar year.
SMR.
No.
SMR analysis, adequate; Published paper lacks documentation of nested case-control
study of esophageal cancer.
i
^§
H I
O
H
W
-------
1 B.3.1.2. Cancer Incidence Studies Using Biological Monitoring Databases
2 Finland and Denmark historically have maintained national databases of biological
3 monitoring data obtained from workers in industries where toxic exposures are a concern.
4 Legislation required that employers provide workers exposed to toxic hazards with regular health
5 examinations, which must include biological monitoring to assess the uptake of toxic chemicals,
6 including trichloroethylene. In Sweden, the only local producer of trichloroethylene operated a
7 free exposure-surveillance program for its customers, measuring U-TCA. These programs used
8 the linear relationship found for average inhaled trichloroethylene versus U-TCA:
9 trichloroethylene (mg/m3) = 1.96; U-TCA (mg/L) = 0.7 for exposures lower than 375 mg/m3
10 (69.8 ppm) (Ikeda et al., 1972). This relationship shows considerable variability among
11 individuals, which reflects variation in urinary output and activity of metabolic enzymes.
12 Therefore, the estimated inhalation exposures are only approximate for individuals but can
13 provide reasonable estimates of group exposures. There is evidence of nonlinear formation of
14 U-TCA above about 400 mg/m3 or 75 ppm of trichloroethylene. The half-life of U-TCA is about
15 100 hours. Therefore, the U-TCA value represents roughly the weekly average of exposure from
16 all sources, including skin absorption. The Ikeda et al. (1972) relationship can be used to convert
17 urinary values into approximate airborne concentration, which can lead to misclassification if
18 tetrachloroethylene and 1,1,1-trichloroethane are also being used because they also produce
19 U-TCA. In most cases, the Ikeda et al. relationship (1972) provides a rough upper boundary of
20 exposure to trichloroethylene.
21
22 B.3.1.2.1. Hansen et al. (2001).
23 B.3.1.2.1.1. Author's abstract.
24
25 Human evidence regarding the carcinogenicity of the animal carcinogen
26 trichloroethylene (TCE) is limited. We evaluated cancer occurrence among 803
27 Danish workers exposed to TCE, using historical files of individual air and
28 urinary measurements of TCE-exposure. The standardized incidence ratio (SIR)
29 for cancer overall was close to unity for both men and women who were exposed
30 to TCE. Men had significantly elevated SIRs for non-Hodgkin's lymphoma (SIR
31 = 3.5; n = 8) and cancer of the esophagus (SIR = 4.2; n = 6). Among women, the
32 SIR for cervical cancer was significantly increased (SIR = 3.8; n = 4). No clear
33 dose-response relationship appeared for any of these cancers. We found no
34 increased risk for kidney cancer. In summary, we found no overall increase in
35 cancer risk among TCE-exposed workers in Denmark. For those cancer sites
36 where excesses were noted, the small numbers of observed cases and the lack of
37 dose-related effects hinder etiological conclusions.
38
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-96 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.3.1.2.1.2. Study description and comment. This Danish study evaluated cancer incidence in
2 a small cohort of individuals (n = 803) who had been monitored for trichloroethylene exposures
3 in a national surveillance program between 1947 and 1989 for U-TCA or TCE in breath since
4 1974. In all, 2,397 samples were analyzed for U-TCA of workers at 275 companies and 472
5 breathing zone samples of TCE from workers at 81 companies. Individual workers could not be
6 identified for roughly one-third of the U-TCA measurements and 50% of breathing zone
7 measurements; many of the individuals most likely had died prior to 1968, the start of the
8 Central Population Registry from which workers were identified and follow-up for cancer
9 incidence. A cohort of 658 males and 145 females were identified from the remaining
10 1,519 U-TCA and 245 air-TCE measurements. Only two of 803 cohort subjects had both urine
11 and air measurements. Follow-up for cancer incidence ended as of 12-31-1996.
12 The retirement and measurement records contained general information about the type of
13 employer and the subject's job. The subjects in this study came predominantly from the iron and
14 metal industry with jobs such as metal-product cleaner. Each subject had 1 to 27 measurements
15 of U-TCA measurements, an average of 2.2 per subject, going back to 1947. Using the linear
16 relationship from Ikeda et al. (1972), the historic median exposures estimated from the U-TCA
17 concentrations were low: 9 ppm for 1947 to 1964, 5 ppm for 1965 to 1973, 4 ppm for 1974 to
18 1979, and 0.7 ppm for 1980 to 1989. However, the distributions were highly skewed.
19 Additionally, 5% of the cohort had urine or air samples below the limit of detection. Overall,
20 median exposure in this cohort was 4 ppm and suggests that, in general, workers in a wide
21 variety of industry and job groups and identified as "exposed" in this study had low TCE
22 intensity exposures. Overall, the cohort in this study is small, drawn from a wide variety of
23 industries, predominantly degreasing and metal cleaning, and had generally low exposures (most
24 less than 20 ppm). The study has a lower power to examine TCE exposure and cancer for these
25 reasons.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-97 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
w 1
vb 8
oo o
Hansen J, Raaschou-Nielsen O, Christensen JM, Johansen I, McLaughlin JK, Lipworth L, Blot WJ, Olsen JH. 2001. Cancer
incidence among Danish workers exposed to trichloroethylene. J Occup Environ Med 43:133-139.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From introduction — A study of incidence was carried out to address shortcomings in
earlier TCE studies related to the lack of direct exposure information and to
assessment of mortality as opposed to incidence.
803 subjects identified from biological monitoring of urine TCA from 1947-1989
(1,519 measurements) or breathing zone TCE since 1974 (245 measurements) and
who were alive as of 1968, followed to 1996.
External referents — cancer incidence rates of Danish population (age-, sex-, calendar
years-, and site-specific).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
ICD, 7th revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Biological marker of TCE in urine or in breath used to assign TCE exposure to
cohort subject. Historic median exposures estimated from the U-TCA were low:
9 ppm for 1947 to 1964, 5 ppm for 1965 to 1973, 4 ppm for 1974 to 1979, and
0.7 ppm for 1980 to 1989. Overall, median TCE exposure to cohort was 4 ppm
(arithmetic mean, 12 ppm).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No.
Unable to determine given insufficient information in paper; however, text notes
follow-up for most subjects achieved a full latency.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
w 1
vb ^
CATEGORY E: INTERVIEW TYPE
<90% Face-to-Face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
128 incident cancers among 804 cohort subjects (15%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex and calendar year.
SIR, Life table analysis.
Yes, as dichotomous variable for mean exposure (<4 ppm, 4+ ppm) and for
cumulative exposure.
Adequate.
i
^§
SIR = standardized incidence ratio.
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.2.2. Anttila et al (1995).
2 B.3.1.2.2.1. Author's abstract.
O
4 Epidemiologic studies and long-term carcinogenicity studies in experimental
5 animals suggest that some halogenated hydrocarbons are carcinogenic. To
6 investigate whether exposure to trichloroethylene, tetrachloroethylene, or
7 1,1,1-trichloroethane increases carcinogenic risk, a cohort of 2050 male and 1924
8 female workers monitored for occupational exposure to these agents was followed
9 up for cancer incidence in 1967 to 1992. The overall cancer incidence within the
10 cohort was similar to that of the Finnish population. There was an excess of
11 cancers of the cervix uteri and lymphohematopoietic tissues, however. Excess of
12 pancreatic cancer and non-Hodgkin lymphoma was seen after 10 years from the
13 first personal measurement. Among those exposed to trichloroethylene, the
14 overall cancer incidence was increased for a follow-up period of more than 20
15 years. There was an excess of cancers of the stomach, liver, prostate, and
16 lymphohematopoietic tissues combined. Workers exposed to 1,1,1 -trichloroethane
17 had increased risk of multiple myeloma and cancer of the nervous system. The
18 study provides support to the hypothesis that trichloroethylene and other
19 halogenated hydrocarbons are carcinogenic for the liver and lymphohematopoietic
20 tissues, especially for non-Hodgkin lymphoma. The study also documents excess
21 of cancers of the stomach, pancreas, cervix uteri, prostate, and the nervous system
22 among workers exposed to solvents.
23
24 B.3.1.2.2.2. Study description and comment. This Finnish study evaluated cancer risk in a
25 small cohort of individuals (2,050 males and 1,924 females) who had been monitored between
26 1965 and 1982 for exposures to trichloroethylene by measuring their U-TCA. The main source
27 of exposure was identified as degreasing or cleaning metal surfaces. Some workplaces identified
28 rubber work, gluing, and dry-cleaning. There was an average of 2.7 measurements per person.
29 Using the Ikeda et al. (1972) conversion relationship, the exposure for trichloroethylene was
30 approximately 7 ppm in 1965, which declined to approximately 2 ppm in 1982; the 75th
31 percentiles for these dates were 14 and 7 ppm, respectively. The maximum values for males
32 were approximately 380 ppm during 1965 to 1974 and approximately 96 ppm during 1974 to
33 1982. Females showed a similar pattern over time but had somewhat higher exposures than
34 males before the 1970s. Median TCE exposure for females of 4 ppm compared to 3 ppm for
35 males; maximum values were similar for both sexes. Duration of exposure was counted from the
36 first measurement of U-TCA, which might underestimate the length of exposure. Without job
37 histories, the length of exposure is uncertain. Another concern is the sampling strategy; it was
38 not reported how the workers were chosen for monitoring. Therefore, it is not clear what biases
39 might be present, especially the possibility of under sampling highly exposed workers.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-100 DRAFT—DO NOT CITE OR QUOTE
-------
1 Overall, this TCE exposed cohort drawn from a wide variety of industries was twice the
2 size of other Nordic biomonitoring studies (Axelson et al., 1994; Hansen et al., 2001) with urine
3 TCA measurements from a more recent period, 1965 to 1982, compared to other Nordic studies
4 of Danish cohorts, 1947 to 1980s, or Swedish cohorts, 1955 to 1975 (Axelson et al., 1994;
5 Hansen et al., 2001; Raaschou-Nielsen et al., 2002). Exposures to trichloroethylene were
6 generally low, less than 14 ppm for the 75th percentile of all measurements, and median TCE
7 exposures decreasing from 7 ppm to 2 ppm over the 17-year period. The medians are similar to
8 estimated exposures to Danish workers with biological markers of U-TCA (Hansen et al., 2001;
9 Raaschou-Nielson et al., 2001). The duration of exposure was uncertain.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-101 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
Anttila A, Pukkala E, Sallmen M, Hernberg S, Hemminki K. 1995. Cancer incidence among Finnish workers exposed to
halogenated hydrocarbons. J Occup Environ Med 37:797-806.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes, study aim was to assess cancer incidence among workers biologically monitored
for exposure to TCE, PERC, and 1,1,1-trichloroethane.
3, 976 subjects identified from biological monitoring of urine TCA between 1965 to
1982; PERC in blood, 1974 to 1983; and, 1,1,1-trichloroethane in blood, 1975 to
1983 (a total of 10.743 measurements). 109 of cohort subjects with TCE poisoning
report between 1965 to 1976. Follow-up for mortality between 1965 to 1991 and for
cancer between 1967 to 1992.
TCE subcohort— 3,089 (1,698 males, 1,391 females).
External referents — age-, sex-, calendar year-, and site-specific cancer incidence rates
of the Finnish population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality and cancer incidence.
ICD, 7th revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Biological marker of TCE in urine used to assign TCE exposure for TCE subcohort.
There were on average 2.5 U-TCA measurements per individual. 6% of cohort had
measurements for 2 or all three solvents. The overall median of U-TCA for females
was 8.3 mg/L and 6.3 mg/L for males, and before 1970, 10 to 13 mg/L for females
and 13 to 15 mg/L for males. Using Ikeda et al. (1972) relationship for U-TCA
and TCE concentration, median TCE exposures over the period of study were
roughly <4-9 ppm (median, 4 ppm; arithmetic mean, 6 ppm).
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No.
Yes, 18 yr mean follow-up period.
CATEGORY E: INTERVIEW TYPE
<90% Face-to-Face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
208 cancers among 3,089 TCE-exposed subjects (7%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and calendar year.
SMR and SIR, Life table analysis.
Yes, U-TCA as dichotomous variable (<6 ppm, 6+ ppm).
Adequate for SIR analysis; details on SMR analysis of TCE subcohort are few.
o §
H I
PERC = perchloroethylene, SIR = standardized incidence ratio.
O
H
W
-------
1 B.3.1.2.3. Axehon et al (1994).
2 B.3.1.2.3.1. Author's abstract.
3
4 There is limited evidence for mutagenicity and carcinogenicity of
5 trichloroethylene (TRI) in experimental test systems. Whether TRI is a human
6 carcinogen is unclear, however. This paper presents an update and extension of a
7 previously reported cohort of workers exposed to TRI, in total 1670 persons.
8 Among men (n = 1421), the overall standardized mortality ratio (SMR) and
9 cancer morbidity ratio (SIR) were close to the expected, with SMR, 0.97; 95%
10 confidence interval (CI), 0.86 to 1.10; and SIR, 0.96; 95% CI, 0.80 to 1.16,
11 respectively. The cancer mortality was significantly lower than expected (SMR,
12 0.65; 95% CI, 0.47 to 0.89), whereas an increased mortality from circulatory
13 disorders (cardiovascular, cerebrovascular) was of borderline significance (SMR,
14 1.17; 95% CI, 1.00 to 1.37). No significant increase of cancer of any specific site
15 was observed, except for a doubled incidence of nonmelanocytic skin cancer
16 without correlation with the exposure categories. In the small female subcohort
17 (n = 249), a nonsignificant increase of cancer and circulatory deaths was observed
18 (SMR, 1.53 and 2.02, respectively). For both genders, however, excess risks were
19 largely confined to groups of workers with lower exposure levels or short duration
20 of exposure or both. It is concluded that this study provides no evidence that TRI
21 is a human carcinogen, i.e., when the exposure is as low as for this study
22 population.
23
24 B.3.1.2.3.2. Study description and comment. This Swedish study evaluated cancer risk in a
25 small cohort of individuals (1,421 males and 249 females), who were monitored for U-TCA as
26 part of a surveillance system by the trichloroethylene producer during 1955 to 1975. Both
27 mortality between 1955 and 1986 and cancer morbidity between 1958 and 1987 are assessed in
28 males only due to the small number of female subjects. Eighty-one percent of the male subjects
29 had low exposures (<50 mg/L), corresponding to an airborne concentration of trichloroethylene
30 of approximately 20 ppm. There was uncertainty about the beginning and end of exposure.
31 Exposure was assumed to begin with the first urine sample and to end in 1979 (the reason for this
32 date is unclear). Because the investigators did not have job histories, there is considerable
33 uncertainty about the duration of exposure. No information is, additionally, presented to
34 evaluate if a large proportion of the cohort had a full latency period for cancer development.
35 Most subjects appear to have had short durations of exposure, but these might have been
36 underestimated. Another concern is the sampling strategy. It was not reported how the workers
37 were chosen for monitoring. Therefore, it is not clear what biases could be present in the data,
38 especially the possibility of under sampling highly exposed workers.
39 Overall, this study had a small cohort drawn from a wide variety of industries,
40 predominantly from industries involving degreasing and metal cleaning. Exposure to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-104 DRAFT—DO NOT CITE OR QUOTE
-------
1 trichloroethylene was generally low (most less than 20 ppm). The duration of exposure was
2 uncertain and bias related to under sampling of higher exposed workers is possible but can not be
3 evaluated.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-105 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Axelson O, Selden A, Andersson K, Hogstedt C. 1994. Updated and expanded Swedish cohort study on trichloroethylene and
cancer risk. J Occup Environ 36:556-562.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes- "This paper present an update and extension of a previously reported cohort of
workers exposure to TCE."
1,670 subjects (1,421 males, 249 females) with records of biological monitoring of
urine TCA from 1955 and 1975.
Analysis restricted to 1,421 males.
External referents — age-, sex-, calendar year-, site-specific mortality or cancer
incidence rates of Swedish population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence from 1958 to 1987 and all-cause mortality from 1955 to 1986.
ICD, 7th revision.
ICD, 8th revision from 1975 onward for all lympho-hematopoietic system cancers.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Biological marker of TCE in urine used to assign TCE exposure to cohort subject.
No extrapolation of U-TCA data to air-TCE concentration. Roughly % of cohort
had U-TCA concentrations equivalent to <20 ppm TCE.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No
Insufficient to estimate for full cohort; however, 42% of person years in subjects
with 2+ exposure years also had 10+ yrs of latency.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
g
td
-§
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
229 deaths (16% of male subjects).
107 incident cancer cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and calendar year.
SMR — age, sex, and calendar-year.
SIR — analyses restricted to males — age and
calendar-year.
Yes, by 3 categories of U-TCA concentration.
Adequate.
•3
1
§
i
td
SIR = standardized incidence ratio.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.3. Studies in the Taoyuan Region of Taiwan
2 B.3.1.3.1. Sung et al (2008, 2007).
3 B.3.1.3.1.1. Suns et al (2008) abstract.
4
5 There is limited evidence on the hypothesis that maternal occupational exposure
6 near conception increases the risk of cancer in offspring. This study is to
7 investigate whether women employed in an electronics factory increases
8 childhood cancer among first live born singletons. We linked the databases of
9 Birth Registration and Labor Insurance, and National Cancer Registry, which
10 identified 40,647 female workers ever employed in this factory who gave 40,647
11 first live born singletons, and 47 of them developed cancers during 1979-2001.
12 Mothers employed in this factory during their periconceptional periods (3 months
13 before and after conception) were considered as exposed and compared with those
14 not employed during the same periods. Poisson regression model was constructed
15 to adjust for potential confounding by maternal age, education, sex, and year of
16 birth. Based on 11 exposed cases, the rate ratio of all malignant neoplasms was
17 increased to 2.26 [95% confidence interval (CI), 1.12-4.54] among children
18 whose mothers worked in this factory during periconceptional periods. The RRs
19 were associated with 6 years or less (RR=3.05; 95% CI, 1.20-7.74) and 7-9 years
20 (RR=2.49; 95% CI, 1.26-4.94) of education compared with 10 years or more. An
21 increased association was also found between childhood leukemia and exposed
22 pregnancies (RR=3.83; 95% CI, 1.17-12.55). Our study suggests that maternal
23 occupation with potential exposure to organic solvents during periconception
24 might increase risks of childhood cancers, especially for leukemia.
25
26 B.3.1.3.1.2. Suns et al (2007) abstract.
27
28 Background In 1994, a hazardous waste site, polluted by the dumping of
29 solvents from a former electronics factory, was discovered in Taoyuan, Taiwan.
30 This subsequently emerged as a serious case of contamination through chlorinated
31 hydrocarbons with suspected occupational cancer. The objective of this study was
32 to determine if there was any increased risk of breast cancer among female
33 workers in a 23-year follow-up period. Methods A total of 63,982 female
34 workers were retrospectively recruited from the database of the Bureau of Labor
35 Insurance (BLI) covering the period 1973-1997; the data were then linked with
36 data, up to 2001, from the National Cancer Registry at the Taiwanese Department
37 of Health, from which standardized incidence ratios (SIRs) for different types of
38 cancer were calculated as compared to the general population. Results There
39 were a total of 286 cases of breast cancer, and after adjustment for calendar year
40 and age, the SIR was close to 1. When stratified by the year 1974 (the year in
41 which the regulations on solvent use were promulgated), the SIR of the cohort of
42 workers who were first employed prior to 1974 increased to 1.38 (95%
43 confidence interval, 1.11-1.70). No such trend was discernible for workers
44 employed after 1974. When 10 years of employment was considered, there was a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-108 DRAFT—DO NOT CITE OR QUOTE
-------
1 further increase in the SIR for breast cancer, to 1.62. Those workers with breast
2 cancer who were first employed prior to 1974 were employed at a younger age
3 and for a longer period. Previous qualitative studies of interviews with the
4 workers, corroborated by inspection records, showed a short-term high exposure
5 to chlorinated alkanes and alkenes, particularly trichloroethylene before 1974.
6 There were no similar findings on other types of cancer. Conclusions Female
7 workers with exposure to trichloroethylene and/or mixture of solvents, first
8 employed prior to 1974, may have an excess risk of breast cancer.
9
10 B.3.1.3.1.3. Study description and comment. Sung et al. (2007) examine breast cancer
11 incidence among females in a cohort of electronic workers with employment at one factory in
12 Taoyuan, Taiwan between 1973 and 1992, date of factory closure and followed to 2001. Some
13 female subjects in Sung et al. (2007) overlap those in Chang et al. (2003, 2005) who included
14 workers from the same factory whose employment dates were between 1978 and 1997, the
15 closing date of the study a date of vital status ascertainment. A total of 64,000 females were
16 identified with 63,982 in the analysis after the exclusion of 15 women with less than one full day
17 of employment and three women with cancer diagnoses prior to the time of first employment;
18 approximately 6,000 fewer female subjects compared to Chang et al. (2005) (70,735 females).
19 Cancer incidence between 1979 and 2001 as identified using the National Cancer Registry which
20 contained 80% of all cancer cases in Taiwan (Parkin et al., 2002) is examined using life table
21 methods with exposure lag periods of 5-15 years, depending on the cancer site, and cancer rates
22 from the larger Taiwanese population as referent.
23 Company employment records were lacking and the cohort was constructed using the
24 Bureau of Labor Insurance database that contained computer records since 1978 and paper
25 records for the period 1973 to 1978. Duration of employment was calculated from the beginning
26 of coverage of labor insurance and is likely an underestimate. Labor insurance hospitalization
27 data and a United Labor Association list of names were used to verify cohort completeness.
28 While these sources may have been sufficient to identified current employees, their ability to
29 identify former employees may be limited, particularly from the hospitalization data if the
30 subject's current employer was listed.
31 This study assumes all employees in the factory were exposed to chlorinated organic
32 solvent vapors and the primary exposure index was duration of employment at the plant. Most
33 subjects had employment durations of <1 year (65%). Durations of exposure were likely
34 underestimated as dates of commencement and termination of insurance coverage were
35 incomplete, 7.5% and 6%, respectively. There is little to no information on chemical usage and
36 exposure assignment to individual cohort subjects. As reported in Chang et al. (2003, 2005),
37 records of the Department of Labor Inspection ad Bureau of International Trade, in addition, to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-109 DRAFT—DO NOT CITE OR QUOTE
-------
1 recall of former industrial hygienists were used to identify chemicals used after 1975 in the
2 plants. No information is available prior to this date.
3 Sung et al. (2008) presents an analysis of childhood cancer incidence (1979-2001)
4 among first liveborn singleton births (1978 and 2001) of female subjects employed at the plant
5 during a period 3 months before and after beginning of pregnancy, an estimate derived by Sung
6 et al. (2008) from the date of birth and estimated length of gestation plus 14 days. Sung et al.
7 (2007) used Poisson regression methods and cancer incidence among first liveborn births of all
8 other women in Taiwan in the same time to calculate relative risks associated with leukemia risk
9 among exposed offspring. Poisson models were adjusted for maternal age, maternal educational
10 level, child's sex, and year of birth. A total of 8,506 first born singleton births among
11 63,982 female subjects were identified from the Taiwan Birth Registry database, and 11 cancers,
12 including 6 leukemia cases and no brain/central nervous system (CNS) cases identified from the
13 National Cancer Registry database.
14 Overall, these studies do not provide substantial weight for determining whether
15 trichloroethylene may cause increased risk of disease. The lack of TCE-assessment to individual
16 cohort subjects; grouping cohort subjects with different exposure potential, both to different
17 solvents and different intensities; and deficiencies in the record system used to construct the
18 cohort introduce uncertainty.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-110 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
td g
Dl
Co
o §
Sung T-I, Chen P_C, Lee L J-H, Lin Y-P, Hsieh G-Y, Wang J-D. 2007. Increased standardized incidence ratio of breast
cancer in female electronics workers. BMC Public Health 7:102. http://www.biomedcentral.com/content/pdf/1471-2458-7-
102.pdf.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or hypothesis
Selection and characterization in cohort studies of
exposure and control groups and of cases and
controls in case-control studies is adequate
From abstract "This study is to investigate whether women employed in an
electronics factory increases childhood cancer among first live born singletons."
This study was not able to evaluate TCE exposures uniquely.
63,982 females, some who were also subjects were also in cohort of Chang et al.
(2003, 2005) with 70,735 females.
Cohort initially established using labor insurance records (computer records after
1978 and paper records from 1973 and 1978) in the absence of company records.
Cohort definition dates are not clearly identified. Cohort identified from records
covering period 1973 and 1997 with vital status ascertained as of 2001. Factory
closed in 1992.
External referents: age-, calendar-, and sex-specific incidence rates of the
Taiwanese general population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's lymphoma
Cancer incidence as ascertained from National (Taiwan) Cancer Registry (80% of
all cancers reported to Registry).
ICD-Oncology, a supplement to ICD-9.
H I
O >
HH Oq
H TO
O
H
W
-------
to
VO Lo'
I
I
§
***.
£3'
1
TO'
td
g
Dl
Co
o §
H I
O >
HH Oq
H TO
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including adoption
of JEM and quantitative exposure estimates
All employees assumed to be potentially exposed to chlorinated organic solvent
vapors; study does not assign potential chemical exposures to individual subjects.
No information on specific chemical exposures or intensity. Limited identification
of solvents used in manufacturing process from the period after 1975 inferred from
records of Department of Labor Inspection, Bureau of International Trade, and
former industrial hygienists recall. No information on solvent usage was available
before 1975.
Exposure index defined as duration of exposure which was likely underestimated.
21% of cohort with >10 yrs duration of employment and 53% with <1 yr duration.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No information on loss to follow-up. Subject was assumed disease free at end of
follow-up if lacking cancer diagnosis as recorded in the National Cancer Registry.
No, 57% of cohort employed after November 21, 1978.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies; numbers
of exposed cases and prevalence of exposure in
case-control studies
1,311 cancer cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age-, calendar-, and sex-specific incidence rates.
SIR, analyses include a lag period of 5, 10, or 15 yrs since first employment (as
indicated by labor insurance record).
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td g
Exposure-response analysis presented in published
paper
Documentation of results
Cancer incidence examined by duration of employment; however, employment
durations were likely underestimates as dates of commencement and termination
dates on of insurance coverage date were incomplete and misclassification bias is
likely present.
Inadequate — analyses that do not include a lag are not presented nor discussed in
published paper or in supplemental documentation.
^§
H I
O >
HH Oq
H TO
O
SIR = standardized incidence ratio.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Sung T-I, Wang J-D, Chen P_C. 2008. Increased risk of cancer in the offspring of female electronics workers. Reprod
Toxicol 25:115-119.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract "The study was designed to examine whether breast cancer risk in
females was increased, as had been observed in Chang et al. (2003, 2005) in a cohort
with earlier employment dates." This study was not able to evaluate TCE exposure.
1 1 cancers among 8,506 first born singleton births between 1978-2001 in
63,982 female subjects of Sung et al. (2007). Cancers identified from National
Cancer Registry and births identified from Taiwan Birth Registration database.
External referents: cancer incidence among all other first birth singleton births
among Taiwanese females over the same time period.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence as ascertained from National (Taiwan) Cancer Registry (80% of all
cancers reported to Registry).
ICD-Oncology, a supplement to ICD-9, specific leukemia subtypes not identified in
paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
All births were among subjects with employment at factory during a period 3 mos
before and after beginning of pregnancy. All mothers were assumed potentially
exposed to chlorinated organic solvent vapors; specific solvents are not identified nor
assigned to individual subjects. Limited identification of solvents used in
manufacturing process from the period after 1975 inferred from records of
Department of Labor Inspection, Bureau of International Trade, and former industrial
hygienists recall. No information on solvent usage was available before 1975.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
No information on loss to follow-up for females in Sung et al. (2007).
g
td
Dl
Co
o §
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td g
>50% cohort with full latency
66% of births would have been 16 yrs of age as of 2001, the date cancer incidence
ascertainment ended.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1 1 cancer cases among 8,506 first born singleton births.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Maternal age, maternal educational level, child's sex and child's year of birth.
Poisson regression using childhood cancer incidence among all other first live born
children in Taiwan during same time period.
No.
Yes.
^§
H I
O
H
W
-------
1 B.3.1.3.2. Chang et al (2005, 2003).
2 B.3.1.3.2.1. Chans et al. (2005) abstract.
3
4 A retrospective cohort morbidity study based on standardized incidence ratios
5 (SIRs) was conducted to investigate the possible association between exposure to
6 chlorinated organic solvents and various types of cancers in an electronic factory.
7 The cohort of the exposure group was retrieved from the Bureau of Labor
8 Insurance (BLI) computer database records dating for 1978 through December 31,
9 1997. Person-year accumulation began on the date of entry to the cohort, or
10 January 1, 1979 (whichever came later), and ended on the closing date of the
11 study (December 31, 1997), if alive with out contracting any type of cancers, or
12 the date of death, or the date of the cancer diagnosis. Vital status and cases of
13 cancer of study subjects were determined from January 1, 1979 to December 31,
14 1997 by linking cohort data with the National Cancer Registry Database. The
15 cancer incidence of the general population was used fro comparison. After
16 adjustment for age and calendar year, only SIR for breast cancer in the exposed
17 female employees were significantly elevated when compared with the Taiwanese
18 general population, based on the entire cohort without exclusion. The SIR of
19 female breast cancer also showed a significant trend of period effect, but no
20 significant dos-response relationship on duration of employment. Although the
21 total cancer as well as the cancer for the trachea, bronchus[,] and lung for the
22 entire female cohort was not significantly elevated, trend analysis by calendar-
23 year interval suggested an upward trend. However, when duration of employment
24 or latency was taken into consideration, no significantly elevated SIR was found
25 for any type of cancer in either male or female exposed workers. In particular, the
26 risk of female breast cancer was not indicated to be increased. No significant
27 dose-response relationship on duration of employment and secular trend was
28 found for the above-mentioned cancers. This study provides no evidence that
29 exposure to chlorinated organic solvents at the electronics factory was associated
30 with elevated human cancers.
31
32 B.3.1.3.2.2. Chans et al. (2003) abstract.
33
34 PURPOSE: A retrospective cohort mortality study based on standardized
35 mortality ratios (SMRs) was conducted to investigate the possible association
36 between exposure to chlorinated organic solvents and various types of cancer
37 deaths. METHODS: Vital status and causes of death of study subjects were
38 determined from January 1, 1985 to December 31, 1997, by linking cohort data
39 with the National Mortality Database. Person-year accumulation began on the
40 date of entry to the cohort, or January 1, 1985 (whichever came later), and ended
41 on the closing date of the study (December 31, 1997), if alive; or the date of
42 death. RESULTS: This retrospective cohort study examined cancer mortality
43 among 86,868 workers at an electronics factory in the northern Taiwan. Using
44 various durations of employment and latency and adjusting for age and calendar
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-116 DRAFT—DO NOT CITE OR QUOTE
-------
1 year, no significantly elevated SMR was found for any cancer in either male or
2 female exposed workers when compared with the general Taiwanese population.
3 In particular, the risk of female breast cancer was not found to be increased.
4 Although ovarian cancer suggested an upward trend when analyzed by length of
5 employment, ovarian cancer risk for the entire female cohort was not elevated.
6 CONCLUSIONS: It is concluded that this study provided no evidence that
7 exposure to chlorinated organic solvents was associated with human cancer risk.
8
9 B.3.1.3.2.3. Study description and comment. Both Chang et al. (2003) and Chang et al.
10 (2005) studied a cohort of 86,868 subjects employed at an electronics factory between 1985 and
11 1997, and both administrative and nonadministrative (blue-collar) workers were included in the
12 cohort. Cancer incidence between 1979 and 1997 was presented by Chang et al. (2005) and
13 cancer mortality from 1985 to 1997 in Chang et al. (2003). The cohort was predominately
14 composed of females. The factory operated between 1968 and 1992, and the inclusion in the
15 cohort of subjects after factory closure is questionable. Incidence was ascertained from the
16 Taiwan National Cancer Registry which contains 80% of all cancer cases in Taiwan (Parkin et
17 al., 2002). The factory could be divided into three plants by manufacturing process: manufacture
18 of television remote controls, manufacture of solid state and integrated circuit products, and
19 manufacture of printed circuit boards. Furthermore, a factory waste disposal site was found to
20 have contaminated the underground water supply of area communities with organic solvents,
21 however, Chang et al. (2005) does not provide information on possible exposure to factory
22 employees through ingestion. The analysis of communities adjacent to the factory is described
23 in Lee et al. (2003).
24 Company employment records were lacking and the cohort was constructed using the
25 Bureau of Labor Insurance database that contained computer records since 1978. Labor
26 insurance hospitalization data and a United Labor Association list of names were used to verify
27 cohort completeness. While these sources may have been sufficient to identified current
28 employees, their ability to identify former employees may be limited, particularly from the
29 hospitalization data if the subject's currently employer was listed.
30 All employees in the factory were assumed with potential exposure to chlorinated organic
31 solvent vapors with duration of employment at the factory as the exposure surrogate. Subjects
32 had varying exposure potentials and employment durations of <1 year (65% of cohort in Chang
33 et al., 2005). Durations of exposure were likely underestimated as dates of commencement and
34 termination of insurance coverage were incomplete, 7.5 and 6%, respectively. Three plants
35 comprised the factory and with different production processes. A wide variety of organic
36 solvents were used in each process including dichloromethane, toluene, and methyl ethyl
37 alcohol, used at all three plants, and perchloroethylene, propanol, and dichloroethylene which
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-117 DRAFT—DO NOT CITE OR QUOTE
-------
1 was used at one of the 3 plants Chang et al. (2005). Records of the Department of Labor
2 Inspection and Bureau of International Trade, in addition, to recall of former industrial hygienists
3 were used to identify chemicals used after 1975 in the plants. No information is available prior
4 to this date. These sources documented the lack of TCE use between 1975 and 1991 and
5 perchloroethylene was after 1981. No information was available on TCE and perchloroethylene
6 usage during other periods. Given the period of documented lack of TCE usage is before the
7 cohort start date of 1978 and factory closure, there is great uncertainty of TCE exposure to
8 cohort subjects.
9 Overall, both studies are not useful for determining whether trichloroethylene may cause
10 increased risk of disease. The lack of TCE-assessment to individual cohort subjects and
11 uncertainty of TCE usage in the factory; potential bias likely introduced through missing
12 employment dates; and, examination of incidence using broad organ-level categories, i.e.,
13 lymphatic and hematopoietic tissue cancer together, decrease the sensitivity of this study for
14 examining trichloroethylene and cancer. Furthermore, few cancers are expected, 1% of the
15 cohort expected with cancer, and results in low statistical power from the cohort's young average
16 age of 39 years.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-118 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Chang Y-M, Tai C-F, Yang S-C, Lin R, Sung F-C, Shin T-S, Liou S-H. 2005. Cancer Incidence among Workers Potentially
Exposed to Chlorinated Solvents in An Electronics Factory. J Occup Health 47:171-180.
Chang Y-M, Tai C-F, Yang S-C, Chan C-J, S Shin T-S, Lin RS, Liou S-H. 2003. A cohort mortality study of workers exposed
to chlorinated organic solvents in Taiwan. Ann Epidemiol 13:652-660.
Description
CATEGORY A: STUDY DESIGN
g
td
Clear articulation of study objectives or
hypothesis
The study was not designed to uniquely evaluate TCE exposure but rather
chlorinated solvents exposures. From abstract: "... to investigate the possible
association between chlorinated organic solvents and various types of cancer in an
electronics factory."
This study is quite limited to meet stated hypothesis by the inclusion of all factory
employees in the cohort and lack of exposure assessment on individual study
subjects to TCE, specifically, and to chlorinated solvents, generally.
Dl
Co
o §
H I
O >
HH Oq
H TO
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
n = 86,868 in cohort. Cohort initially established using labor insurance records in the
absence of company records.
Cohort definition dates are not clearly identified. Cohort identified from labor
insurance records covering period 1978 and 1997; yet, plant closed in 1992. All
subjects followed through 1997.
Paper states cohort was completely identified; however, former workers who were
eligible for cohort membership may not have been identified if validation sources did
not identify former employer. Duration of employment reconstructed from insurance
records: -40% of subjects had employment durations <3 mos, 9% employed >5 yrs,
0.7% employed >10 yrs.
External referents: Age-, calendar-, and sex-specific incidence rates of the Taiwanese
general population.
O
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
td
to
o
•3
1
§
i
^§
"*4-
o§
H I
O >
HH Oq
H TO
O
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence as ascertained from National (Taiwan) Cancer Registry (80% of all
cancers reported to Registry) (Chang et al., 2005).
Mortality. ICD revision is not identified other than that used in 1981 (Chang et al.,
2003).
ICD-Oncology, a supplement to ICD-9 (Chang et al., 2005).
ICD, 9th revision was in effect in 1981, but paper does not identify to which ICD
revision used to assign cause of death (Chang et al., 2003).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
All employees assumed to be potentially exposed to chlorinated organic solvent
vapors. No information on specific chemical exposures or intensity. Limited
identification of solvents used in manufacturing process from the period after 1975
inferred from records of Department of Labor Inspection, Bureau of International
Trade, and former industrial hygienists recall. No information on solvent usage was
available before 1975.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Other
No information on loss to follow-up. Subject was assumed disease free at end of
follow-up if lacking cancer diagnosis as recorded in the National Cancer Registry.
Average 16-yr follow-up (incidence) and 12 yrs (mortality).
Subject's age determined by subtracting year of birth from 1997; however, insurance
records did not contain DOB for 6% of subjects. Furthermore, commencement and
termination dates were incomplete on insurance records, 7 and 6%, respectively.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,031 cancer cases.
1,357 total deaths (1.6% of cohort), 316 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age-, calendar-, and sex-specific incidence rates (Chang et al., 2005) or age-,
calendar-, and sex-specific mortality rates (Chang et al., 2003).
SIR (Chang et al., 2005) and SMR (Chang et al., 2003).
Cancer incidence and mortality examined by duration of employment; however,
employment durations were likely underestimates as dates of commencement and
termination dates on of insurance coverage date were incomplete and calculated from
date on insurance records. Misclassification bias is likely present.
Adequate.
•3
I
td
SIR = standardized incidence ratio.
-i§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4. Studies of Other Cohorts
2 B.3.1.4.1. Clapp and Hoffman (2008).
3 B.3.1.4.1.1. Author's abstract.
4
5 BACKGROUND: In response to concerns expressed by workers at a public
6 meeting, we analyzed the mortality experience of workers who were employed at
7 the IBM plant in Endicott, New York and died between 1969 - 2001. An
8 epidemiologic feasibility assessment indicated potential worker exposure to
9 several known and suspected carcinogens at this plant. METHODS: We used the
10 mortality and work history files produced under a court order and used in a
11 previous mortality analysis. Using publicly available data for the state of New
12 York as a standard of comparison, we conducted proportional cancer mortality
13 (PCMR) analysis. RESULTS: The results showed significantly increased
14 mortality due to melanoma (PCMR = 367; 95% CI: 119, 856) and lymphoma
15 (PCMR = 220; 95% CI: 101, 419) in males and modestly increased mortality due
16 to kidney cancer (PCMR = 165; 95% CI: 45, 421) and brain cancer (PCMR =
17 190; 95% CI: 52, 485) in males and breast cancer (PCMR = 126; 95% CI: 34,
18 321) in females. CONCLUSION: These results are similar to results from a
19 previous IBM mortality study and support the need for a full cohort mortality
20 analysis such as the one being planned by the National Institute for Occupational
21 Safety and Health.
22
23 B.3.1.4.1.2. Study description and comment. This proportional cancer mortality ratio study of
24 deaths between 1969 and 2001 among employees at an IBM facility in Endicott, NY, who were
25 included on the IBM Corporate Mortality File compared the observed number of site-specific
26 cancer deaths are compared to the expected proportion, adjusted for age, using 10-year rather
27 than 5-year grouping, and sex, of site-specific cancer deaths among New York residents during
28 1979 to 1998. Of the 360 deaths identified of Endicott employees, 115 deaths were due to
29 cancer, 11 of these with unidentified site of cancer. Resultant proportional mortality ratios
30 estimates do not appear adjusted for race nor does the paper identify whether referent rates
31 excluded deaths among New York City residents or are for New York deaths. The IBM
32 Corporate Mortality File contained names of employees who had worker >5 years, who were
33 actively employed or receiving retirement or disability benefits at time of death, or whose family
34 had filed a claim with IBM for death benefits and Endicott plant employees were identified using
35 worker employment data from the IBM Corporate Employee Resource Information System.
36 Study investigators had previously obtained the IBM Corporate Mortality file through a court
37 order and litigation.
38 The Endicott plant began operations in 1991 and manufactured a variety of products
39 including calculating machines, typewriters, guns, printers, automated machines, and chip
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-122 DRAFT—DO NOT CITE OR QUOTE
-------
1 packaging. The most recent activities were the production of printed circuit boards. It was
2 estimated from a National Institute of Occupational Safety and Health (NIOSH) feasibility study
3 that a larger percentage of the plant's employee were potentially exposure to multiple chemicals,
4 including asbestos, benzene, cadmium, nickel compounds, vinyl chloride, tetrachloroethylene,
5 TCE , PCBs, and o-toluidine. Chlorinated solvents were used at the plant until the 1980s. The
6 study does not assign exposure potential to individual study subjects.
7 This study provides little information on cancer risk and TCE exposure given its lack of
8 worker exposure history information and absence of exposure assignment to individual subjects.
9 Other limitations in this study which reduces interpretation of the observations included
10 incomplete identification of deaths, the analysis limited to only vested employees or to those
11 receiving company death benefits, incomplete identification of all employees at the plant, the
12 inherent limitation of the PMR method and instability of the effect measure particularly in light
13 of bias resulting of excesses or deficits in deaths, and observed differences in demographic (race)
14 between subjects and the referent (New York) population.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-123 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Clapp RW, Hoffman K. 2008. Cancer mortality in IBM Endicott plant workers, 1969-2001: an update on a NY production
plant. Environ health 7:13.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract ". . .In response to concerns expressed by workers at a public meeting,
we analyzed the mortality experience of workers who were employed at the IBM
plant in Endicott, New York and died between 1969-2001."
Deaths among IBM workers identified in IBM Corporate Mortality File; workers
with >5 yrs employment, who were actively employed or receiving retirement or
disability benefits at time of death, or whose family had filed a claim with IBM for
death benefits. Expected number of site-specific cancer deaths calculated from
proportion of cancer deaths among New York residents. Paper does not identify if
referent included all New York residents or those living upstate.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD9.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
This study lacks exposure information. TCE and other chemicals were used at the
factory and inclusion on the employee list served as a surrogate for TCE exposure
unspecified intensity and duration.
of
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Other
Not able to evaluate given inability to identify complete cohort.
Not able to evaluate given lack of work history records.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
•3
I
i
td
to
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
360 deaths, 115 due to cancer, between 1969-2001.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and gender. No information was available on race and PMRs are unadjusted for
race.
Proportionate mortality ratio.
No.
Yes.
I
td
to
i
^§
H I
O
H
W
-------
1 B.3.1.4.2. Agency for Toxic Substances and Disease Registry (ATSDR, 2004).
2 B.3.1.4.2.1. Author's abstract.
3
4 The View-Master stereoscopic slide viewer has been a popular children's toy
5 since the 1950s. For nearly half a century, the sole U.S. manufacturing site for the
6 View-Master product was a factory located on Hall Boulevard in Beaverton,
7 Oregon. Throughout this period, an on-site supply well provided water for
8 industrial purposes and for human consumption. In March 1998, chemical
9 analysis of the View-Master factory supply well revealed the presence of the
10 degreasing solvent trichloroethylene (TCE) at concentrations as high as 1,670
11 micrograms per liter (fg/L)—the U. S. Environmental Protection Agency
12 maximum contaminant level is 5 fg/L. Soon after the contamination was
13 discovered, the View-Master supply well was shut down. Up to 25,000 people
14 worked at the plant and may have been exposed to the TCE contamination. In
15 September of 2001, the Oregon Department of Human Services (ODHS) entered
16 into a cooperative agreement with the Agency for Toxic Substances and Disease
17 Registry (ATSDR) to determine both the need for and the feasibility of an
18 epidemiological study of the View-Master site. In this report, ODHS compiles the
19 findings of the feasibility investigation of worker exposure to TCE at the View-
20 Master factory.
21 On the basis of the levels of TCE found in the supply well, the past use of the
22 well as a source of drinking water, and the potential for adverse health effects
23 resulting from past exposure to TCE, ODHS determined that the site posed a
24 public health hazard to people who worked at or visited the plant prior to the
25 discovery of the contamination. Because the use of the View-Master supply well
26 was discontinued when the contamination was discovered in March 1998, the
27 View-Master supply well does not pose a current public health hazard. No other
28 drinking water wells tap into the contaminated aquifer, and the long-term
29 remediation efforts appear to be containing the contamination.
30 ATSDR and ODHS obtained a list of 13,700 former plant workers from the
31 Mattel Corporation. In collaboration with ATSDR, ODHS conducted a
32 preliminary analysis of mortality and identified excesses in the proportions of
33 deaths due to kidney cancer and pancreatic cancer among the factory's former
34 employees. Although this analysis was limited by the lack of information about
35 the entire worker population and individual exposures to TCE, the preliminary
36 findings underscore the need to fully investigate the impact of TCE exposure on
37 the population of former View-Master workers.
38 The findings of this feasibility investigation are:
39 • TCE appears to have been the primary contaminant of the drinking water
40 at the plant;
41 • Contamination was likely present for a long period of time (estimated to
42 have been present in the groundwater since the mid-1960s);
43 • A large number were likely exposed to the contamination:
44 • The primary route of exposure (for the last 18 years the factory operated)
45 was through contaminated drinking water;
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-126 DRAFT—DO NOT CITE OR QUOTE
-------
1 • Levels of TCE contamination were 300 time the maximum contaminant
2 levels; and
3 • A significant portion of the former workers of their next of kin can indeed
4 be located and invited to participate in a public health evaluation of their
5 exposures.
6 Therefore, ODHS recommends further investigation to include the following:
7 1. A fate and transport assessment to better establish when TCE reached the
8 supply well, and to provide a historical understanding of the concentration of
9 TCE in the well, and
10 2. Epidemiological studies among former workers to determine their exposure
11 and whether they have experienced adverse health and reproductive outcomes
12 associated with TCE exposure at the plant, to determine the mortality
13 experience of the population, and to document the cancer incidence in this
14 population.
15
16 B.3.1.4.2.2. Study description and comment. This proportionate mortality ratio study of
17 deaths between 1995-2001 among 13,697 former employees at a View-Master toy factory in
18 Beaverton, Oregon contains no exposure information on individual study subjects. The PMR
19 analysis was conducted as a feasibility study for further epidemiologic investigations of these
20 subj ects by Oregon Department of Health on behalf of ATSDR, and findings have not been
21 published in the peer-reviewed literature. A former plant owner provided a listing of former
22 employees; however, employees were not identified using IRS records and the roster was known
23 to be incomplete. Additionally, work history records were not available and not information was
24 available on employment length or job title. The goal of the feasibility analysis was to evaluate
25 ability to identify completeness of death identification using several sources.
26 Monitoring of a water supply well in March 1998 showed detectable concentrations of
27 TCE, and this study assumes all subjects had exposure to TCE in drinking water. TCE had been
28 used in large quantities for metal degreasing at the factory between 1952 and 1980; this activity
29 mostly occurred in the paint shop located in one building. At the time metal degreasing ceased,
30 company records suggested historical use of TCE was up to 200 gallons per month. Historical
31 practices resulted in releases of hazardous substances at the factory site and former employees
32 reported waste TCE from the degreased was transported to other sites on the premises, and
33 discharged to the ground (ATSDR, 2004). Additionally, chemical spills allegedly occurred in
34 the paint shop and one report in 1964 of an inspection of the degreaser indicated atmospheric
35 TCE concentrations above occupational limits. TCE was detected at concentrations between
36 1,220-1,670 |ig/L in four water samples and the Oregon Department of Environmental Quality
37 estimated the well had been contaminated for over 20 years. Other volatile organic compounds
38 (VOCs) besides TCE detected in the supply well water in March 1998 included
39 cis-l,2-dichloroethylene at levels up to 33 |ig/L and perchloroethylene at concentrations up to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-127 DRAFT—DO NOT CITE OR QUOTE
-------
1 56-|ig/L. The 160-foot-deep supply well was on the property since original construction in 1950
2 and it supplied water for drinking, sanitation, fire fighting, and industrial use. Connection to
3 municipal water supply occurred in 1956; however, although municipal water was directed to
4 some parts of the plant, the supply well continued to serve the facility's needs, including most of
5 the drinking and sanitary water (AT SDR, 2003).
6 This study provides little information on cancer risk and TCE exposure given the absence
7 of monitoring data beyond a single time period, absence of estimated TCE concentrations in
8 drinking water, and exposure pathways other than ingestion. Other limitation in this study which
9 reduces interpretation of the observations included incomplete identification of employees with
10 the result of missing deaths likely, the inherent limitation of the PMR method and instability of
11 the effect measure particularly in light of bias resulting of excesses or deficits in deaths, and
12 observed differences in demographic (age and male/female ratio) between subjects and the
13 referent (Oregon) population.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-128 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
ATSDR (Agency for Toxic Substances and Disease Registry). 2004. Feasibility investigation of worker exposure to
trichloroethylene at the View-Master Factory in Beaverton, Oregon. Final Report. Submitted by Environmental and
Occupational Epidemiology, Oregon Department of Human Services. December 2004.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The goal of this feasibility investigation for a cohort epidemiologic study of former
employees at a plant manufacturing stereoscopic slide viewers examined the ability
to identify former employees and ascertain vital status.
Name of -13,000 former employee names were provided to ATSDR by the former
plant owner. The current list of employees was known to be incomplete. The
proportion of site-specific mortality among workers between 1989-2001 was
compared to the proportion expected using all death in Oregon for a similar time
period.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD9andICD 10.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
This study lacks actual exposure information; work history records were not
available. TCE was used at the factory and inclusion on the employee list served as a
surrogate for TCE exposure of unspecified intensity and duration.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Other
Not able to evaluate given inability to identify complete cohort.
Not able to evaluate given lack of work history records.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
I
td
vo
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
616 deaths between 1989-2001.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and gender. No information was available on race and PMRs are unadjusted for
race.
Proportionate mortality ratio.
No.
Yes.
•3
1
§
td
^§
H I
O
H
W
-------
1 B.3.1.4.3. Raaschou-Nielsen et al (2003).
2 B.3.1.4.3.1. Author's abstract.
3
4 Trichloroethylene is an animal carcinogen with limited evidence of
5 carcinogenicity in humans. Cancer incidence between 1968 and 1997 was
6 evaluated in a cohort of 40,049 blue-collar workers in 347 Danish companies with
7 documented trichloroethylene use. Standardized incidence ratios for total cancer
8 were 1.1 (95% confidence interval (CI): 1.04, 1.12) in men and 1.2 (95% CI: 1.14,
9 1.33) in women. For non-Hodgkin's lymphoma and renal cell carcinoma, the
10 overall standardized incidence ratios were 1.2 (95% CI: 1.0, 1.5) and 1.2 (95% CI:
11 0.9, 1.5), respectively; standardized incidence ratios increased with duration of
12 employment, and elevated standardized incidence ratios were limited to workers
13 first employed before 1980 for non-Hodgkin's lymphoma and before 1970 for
14 renal cell carcinoma. The standardized incidence ratio for esophageal
15 adenocarcinoma was 1.8 (95% CI: 1.2, 2.7); the standardized incidence ratio was
16 higher in companies with the highest probability of trichloroethylene exposure. In
17 a subcohort of 14,360 presumably highly exposed workers, the standardized
18 incidence ratios for non-Hodgkin's lymphoma, renal cell carcinoma, and
19 esophageal adenocarcinoma were 1.5 (95% CI: 1.2, 2.0), 1.4 (95% CI: 1.0, 1.8),
20 and 1.7 (95% CI: 0.9, 2.9), respectively. The present results and those of previous
21 studies suggest that occupational exposure to trichloroethylene at past higher
22 levels may be associated with elevated risk for non-Hodgkin's lymphoma.
23 Associations between trichloroethylene exposure and other cancers are less
24 consistent.
25
26 B.3.1.4.3.2. Study description and comment. Raaschous-Nielsen et al. (2003) examine cancer
27 incidence among a cohort of workers drawn from 347 companies with documented
28 trichloroethylene. Almost half of these companies were in the iron and metal industry. The
29 cohort was identified using the Danish Supplementary Pension Fund, which includes type of
30 industry of a company and a history of employees, for the years 1964 to 1997. Altogether,
31 152,726 workers were identified of whom 39,074 were white-collar and assumed not to have
32 TCE exposure, 56,970 workers were of unknown status, and 56,578 blue-collar workers, of
33 which 40,049 had been employed at the company for more than 3 months and are the basis of the
34 analysis. The cohort was relatively young, 56% were 38 to 57 years old at end of follow-up, and
35 29% of subjects were older than 57 years of age. Cancer rates typically increase with increasing
36 ages; thus, the lower age of this cohort likely limits the ability of this study to fully examine TCE
37 and cancer, particularly cancers that may be associated with aging. Observed number of
38 site-specific incident cancers are obtained from 4-1-1968 to the end of 1997 and compared to
39 expected numbers of site-specific cancers based on incidence rates of the Danish population.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-131 DRAFT—DO NOT CITE OR QUOTE
-------
1 A separate exposure assessment was conducted using regulatory agency data from 1947
2 to 1989 (Raaschou-Nielsen et al., 2002). This assessment identified three factors as increasing
3 potential for TCE exposure, duration of employment, year of first employment, and number of
4 employees, to increase the likelihood of cohort subjects as TCE exposed. The percentage of
5 exposed workers was found to decrease as company size increased: 81% for <50 workers, 51%
6 for 50-100 workers, 19% for 100-200 workers, and 10% for >200 workers. About 40% of the
7 workers in the cohort were exposed (working in a room where trichloroethylene was used).
8 Smaller companies had higher exposures. Median exposures to trichloroethylene were
9 40-60 ppm for the years before 1970, 10-20 ppm for 1970 to 1979, and approximately 4 ppm
10 for 1980 to 1989. Additionally, an assessment of TCA concentrations in urine of Danish
11 workers suggested a similar trend over time; mean concentrations of 58 mg/L for the period
12 between 1960 and 1964 and 14 mg/L in sample taken between 1980 and 1985
13 (Raaschou-Nielsen et al., 2001).
14 Only a small fraction of the cohort was exposed to trichloroethylene. The highest
15 exposures occurred before 1970 at period in which 21.2% of blue-collar workers had begun
16 employment in a TCE-using company. The iron and metal industry doing degreasing and
17 cleaning with trichloroethylene had the highest exposures, with a median concentration of
18 60 ppm and a range up to about 600 ppm. Overall, strengths of this study include its large
19 numbers of subjects; however, the younger age of the cohort and the small fraction expected with
20 TCE exposure limit the ability of the study to provide information on cancer risk and TCE
21 exposure. For these reasons, positive associations observed in this study are noteworthy.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-132 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
•3
td g
^ s
^ r^
Raaschou-Nielsen O, Hansen J, McLaughlin JK, Kolstad H, Christensen JM, Tarone RE, Olsen JH. 2003. Cancer risk among
workers at Danish companies using trichloroethylene: a cohort study. Am J Epidemiol 158:1182-1192.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study was designed to evaluate associations observed in Hansen et al. (2001)
with TCE exposure and NHL, esophageal adenocarcinoma, cervical cancer, and
liver-biliary tract cancer.
Cohort of 40,049 blue-collar workers employed in 1968 or after with >3 mo
employment duration identified by linking 347 companies, who were considered as
having a high likelihood for TCE exposure, with the Danish Supplementary Pension
Fund to identify employees and with Danish Central Population Registry.
External referents are age-, sex-, calendar year-, site-specific cancer incidence rates
of the Danish population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence between 4-1-1968 and 12-31-1997 as identified from records of
Danish Cancer Registry.
ICD, 7th revision.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td
•3
1
§
i
^§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Qualitative exposure assessment. A previous industrial hygiene survey of Danish
companies identified several characteristics increase likelihood of TCE
exposure-duration of employment, year of 1st employment, and number of employees
in company (Raaschou-Nielsen et al., 2002). Exposure index defined as duration of
employment.
Median exposures to trichloroethylene were 40-60 ppm for the years before
1970, 10-20 ppm for 1970 to 1979, and approximately 4 ppm for 1980 to 1989.
Additionally, an assessment of TCA concentrations in urine of Danish workers
suggested a similar trend over time; mean concentrations of 58 mg/L for the
period between 1960 and 1964 and 14 mg/L in sample taken between 1980 and
1985 (Raaschou-Nielsen et al., 2001).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Danish Cancer Registry is considered to have a high degree of reporting and accurate
cancer diagnoses.
Yes, average follow-up was 18 yrs.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
3.244 cancers (8% of cohort had developed a cancer over the period from 1968 to
1997). Although of a large number of subjects, this cohort is of a young age, 29% of
cohort was >57 years of age at end of follow-up.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age, sex, and calendar year.
SIR using life-table analysis.
-------
This docw
10/20/09
Exposure-response analysis presented in
published paper
Documentation of results
Yes, duration of employment.
Adequate.
SIR = standardized incidence ratio.
§
***.
S3'
1
TO'
•s
td g
^ S
^§
H I
O
H
W
-------
1 B.3.1.4.4. Ritz (1999a).
2 B.3.1.4.4.1. Author's abstract.
O
4 Data provided by the Comprehensive Epidemiology Data Resource allowed us to
5 study patterns of cancer mortality as experience by 3814 uranium-processing
6 workers employed at the Fernald Feed Materials Production Center in Fernald,
7 Ohio. Using risk-set analyses for cohorts, we estimated the effects of exposure to
8 trichloroethylene, cutting fluids, and kerosene on cancer mortality. Our results
9 suggest that workers who were exposed to trichloroethylene experienced an
10 increase in mortality from cancers of the liver. Cutting-fluid exposure was found
11 to be strongly associated with laryngeal cancers and, furthermore, with brain,
12 hemato- and lymphopoietic system, bladder, and kidney cancer mortality.
13 Kerosene exposure increased the rate of death from several digestive-tract cancers
14 (esophageal, stomach, pancreatic, colon, and rectal cancers) and from prostate
15 cancer. Effect estimates for these cancers increased with duration and level of
16 exposure and were stronger when exposure was lagged.
17
18 B.3.1.4.4.2. Study description and comment. This study of 3,814 white male uranium
19 processing workers employed for at least 3 months between 1-1-1951 and 12-31-1972 at the
20 Fernald Feed Materials Production Center in Fernald, Ohio, was of deaths as of 1-1-1990.
21 Subjects were part of a larger cohort study of Fernald workers with potential uranium and
22 products of uranium decay exposures that observed associations with lung cancer and
23 lymphatic/hematopoietic cancer (Ritz, 1999b). Average length of follow-up time was 31.5 years.
24 During this period, 1,045 deaths were observed with expected numbers of deaths based upon
25 age- and calendar-specific U.S. white male mortality rates and age- and calendar-specific white
26 male mortality rates from the NIOSH Computerized Occupational Referent Population System
27 (CORPS) (Zahm, 1992). Internal analyses based upon risk-set sampling and Cox proportional
28 hazards modeling compared workers with differing exposure intensity rankings (light and
29 moderate) and a category for no- TCE exposure/<2 year duration TCE exposure.
30 Fernald produced uranium metal products for defense programs (Hornung et al., 2008).
31 Subjects had potential exposures to uranium, mainly as insoluble compounds and varying from
32 depleted to slight enriched, small amounts of thorium, an alpha particle emitter, respiratory
33 irritants such as tributyl phosphate, ammonium hydroxide, sulfuric acid and hydrogen fluoride,
34 trichloroethylene, and cutting fluids (Ritz, 1999a, b). Exposure assessment for analysis of
35 chemical exposures utilized a job-exposure matrix (JEM) to assign intensity of TCE, cutting
36 fluids, and kerosene to individual jobs from the period 1952 to 1977. Industrial hygienists, a
37 plant foreman, and an engineer during the late 1970s and early 1980s determined the likelihood
38 of exposure to TCE, cutting fluids, and kerosene for each job title and plant area. Based on work
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-136 DRAFT—DO NOT CITE OR QUOTE
-------
1 records, the workforce appeared stable and 54% were employed >5 years and had held only one
2 job title during employment. Both intensity or exposure level and duration of exposure in years
3 were used to rank subjects into 4 categories of no exposure (level 0), light exposure (level 1),
4 moderate exposure (level 2), and heavy exposure (level 3). Seventy eight (78) percent of the
5 cohort was identified with some potential for TCE exposure, 2,792 subjects were identified with
6 low TCE exposure (94%), 179 with moderate exposure (6%), and no subjects were identified
7 with heavy TCE exposure. TCE exposure was highly correlated with other chemical exposures
8 and with alpha radiation (Ritz, 1999a, b; Hornung et al., 2008). Fernald subjects had higher
9 exposures to radiation compared to those of radiation-exposed Rocketdyne workers (Ritz, 1999b;
10 Ritz et al., 1999, 2000). Atmospheric monitoring information is lacking on TCE exposure
11 conditions as is information on changes in TCE usage over time. The cohort was identified from
12 company rosters and personnel records and it is not known whether these were sources for a
13 subject's job title information. Analysis of TCE exposure carried out using conditional logistic
14 regression adjusting for pay status, time since first hired, external and internal radiation dose and
15 previous chemical exposure. Relative risks for TCE exposure are also presented with a lag time
16 period of 15 years.
17 Overall, strengths of this study are the long follow-up time and a large percentage of the
18 cohort who had died by the end of follow-up. TCE exposure intensity is low in this cohort, 94%
19 of TCE exposed subjects were identified with "light" exposure intensity, and all subjects had
20 potential for radiation exposure, which was highly correlated with chemical exposures. No
21 information is presented on the definition of "light" exposure and monitoring data are lacking.
22 Only 179 subjects were identified with TCE exposure above "light" and the number of cancer
23 deaths not presented. The published paper reported limited information on site-specific cancer
24 and TCE exposure; risk estimates are reported for lymphatic and hematopoietic cancers,
25 esophageal and stomach cancer, liver cancer, prostate cancer and brain cancer. Risk estimates
26 for bladder and kidney cancer and TCE exposure are found in NRC (2006). Few deaths were
27 observed with moderate TCE exposure and exposure durations of longer than 2 years: 1 death
28 due to lymphatic and hematopoietic cancer, 0 deaths due to kidney or bladder cancer (as noted in
29 NRC, 2006), and 2 liver cancer deaths among these subjects. Low statistical power reflecting
30 few cases with moderate TCE exposure and multicolinearity of chemical and radiation exposures
31 greatly limits the support this study provides in an overall weight-of-evidence analysis.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-137 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Ritz B. 1999a. Cancer mortality among workers exposed to chemicals during uranium processing. J Occup Environ Med
41:556-566.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The hypothesis in this study was to examine the influence of chemical exposures in
the work environment of the Fernald Feed Materials Production Center (FFMPC) in
Fernald, Ohio, on cancer mortality with a focus on the effects of TCE, cutting fluids,
and a combination of kerosene exposure with carbon (graphite) and other solvents.
3,814 white male subjects identified from company rosters and personnel records,
hired between 1951 and 1972 and who were employed continuously for 3 mos and
monitored for radiation. 2,971 subjects identified as exposed to TCE at "light" and
"moderate" exposures. Subjects were identified in a previous study of cancer
mortality and radiation exposure and most subjects had radiation exposures above
10+mSV(Ritz, 1999b).
External analysis: U.S. white male mortality rates and NIOSH-Computerized
Occupational Referent Population System mortality rates.
Internal analysis: cohort subjects according to level and duration of chemical
exposure.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
Vital status searched through Social Security Administration records, before 1979,
and National Death Index for the period 1979-1989.
External analysis: ICDA, 8th revision.
Internal analysis: aggregation of several subsite causes of deaths into larger
categories based on ICD, 9th revision.
•3
I
td
*%
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
td
1
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Semi quantitative approach and development of job-exposure matrix. JEM developed
by expert assessment by plant employees to classify jobs into four levels of chemical
exposures for the period 1952 to 1977. Intensity using the four-level scale and
duration of exposure to TCE, cutting fluids and kerosene were assigned to individual
cohort subjects using JEM. 73% of cohort identified as TCE exposed (2,971 male
with TCE exposure in cohort of 3,814 subjects). Only 4% of TCE-exposed subjects
with exposure identified as "moderate" and no subjects with "high" exposure. High
correlation between TCE and other chemical exposure and radiation exposure.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
Other
All workers without death certificate assumed alive at end of follow-up.
Average follow-up time, 31.5 yrs.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,045 deaths (27% of cohort), 328 due to cancer. No information on number of all-
cancer deaths among TCE exposed subjects, although reported numbers for specific
sites reported by Ritz (1999a) or NRC (2006): >2 year exposure duration, hemato-
and lymphopoietic cancer (n = 18 with light exposure, 1 with moderate exposure),
esophageal and stomach cancer (n = 15 with light exposure, 0 with moderate
exposure), liver cancers (n = 3 with light exposure, 1 with moderate exposure),
kidney and bladder cancers, (n = 7 with light exposure, 0 with moderate exposure)
prostate cancers (n = 10 with light exposure, 1 with moderate exposure), and brain
cancers (n = 6 with light exposure, 1 with moderate exposure).
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
External analysis: age- and calendar-specific mortality rates for white males.
Internal analysis: pay status, time since first hired, and cumulative time-dependent
external- and internal-radiation doses (continuous); indirect assessment of smoking
through examination of smoking distribution by chemical exposure.
SMR (external analysis) and RR (internal analysis).
Yes, RR presented for exposure to TCE (level 1 and level 2, separately) by duration
of exposure.
Adequate.
TO'
RR = relative risk.
td
•3
1
§
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4.5. Henschler et al (1995).
2 B.3.1.4.5.1. Author's abstract.
3
4 A retrospective cohort study was carried out in a cardboard factory in Germany to
5 investigate the association between exposure to trichloroethene (TRI) and renal
6 cell cancer. The study group consisted of 169 men who had been exposed to TRI
7 for at least 1 year between 1956 and 1975. The average observation period was 34
8 years. By the closing day of the study (December 31, 1992) 50 members of the
9 cohort had died, 16 from malignant neoplasms. In 2 out of these 16 cases, kidney
10 cancer was the cause of death, which leads to a standard mortality ratio of 3.28
11 compared with the local population. Five workers had been diagnosed with
12 kidney cancer: four with renal cell cancers and one with an urothelial cancer of
13 the renal pelvis. The standardized incidence ratio compared with the data of the
14 Danish cancer registry was 7.97 (95% CI: 2.59-18.59). After the end of the
15 observation period, two additional kidney tumors (one renal cell and one
16 urothelial cancer) were diagnosed in the study group. The control group consisted
17 of 190 unexposed workers in the same plant. By the closing day of the study 52
18 members of this cohort had died, 16 from malignant neoplasms, but none from
19 kidney cancer. No case of kidney cancer was diagnosed in the control group. The
20 direct comparison of the incidence on renal cell cancer shows a statistically
21 significant increased risk in the cohort of exposed workers. Hence, in all types of
22 analysis the incidence of kidney cancer is statistically elevated among workers
23 exposed to TRI. Our data suggest that exposure to high concentrations of TRI
24 over prolonged periods of time may cause renal tumors in humans. A causal
25 relationship is supported by the identity of tumors produced in rats and a valid
26 mechanistic explanation on the molecular level.
27
28 B.3.1.4.5.2. Study description and comment. This was a cohort study of workers in a
29 cardboard factory in the area of Arnsberg, Germany. Trichloroethylene was used in this area
30 until 1975 for degreasing and solvent needs. Plant records indicated that 2,800-23,000 L/year
31 was used. Small amounts of tetrachloroethylene and 1,1,1 -trichloroethane were used
32 occasionally, but in much smaller quantities than trichloroethylene. Trichloroethylene was used
33 in three main areas: cardboard machine, locksmith's area, and electrical workshop. Cleaning the
34 felts and sieves and cleaning machine parts of grease were done regularly every 2 weeks, in a job
35 that required 4-5 hours, plus whatever additional cleaning was needed. Trichloroethylene was
36 available in open barrels and rags soaked in it were used for cleaning. The machines ran hot
37 (80-120°C) and the cardboard machine rooms were poorly ventilated and warm (about 50°C),
38 which would strongly enhance evaporation. This would lead to very high concentrations of
39 airborne trichloroethylene. Cherrie et al. (2001) estimated that the machine cleaning exposures
40 to trichloroethylene were greater than 2,000 ppm. Workers reported frequent strong odors and a
41 sweet taste in their mouths. The odor threshold for trichloroethylene is listed as 100 ppm
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-141 DRAFT—DO NOT CITE OR QUOTE
-------
1 (ATSDR, 1997). Workers often left the work area for short breaks "to get fresh air and to
2 recover from drowsiness and headaches." Based on reports of anesthetic effects, it is likely that
3 concentrations of trichloroethylene exceeded 200 ppm (Stopps and McLaughlin, 1967). Those
4 reports, the work setting description, and the large volume of trichloroethylene used are all
5 consistent with very high concentrations of airborne trichloroethylene. The workers in the
6 locksmith's area and the electrical workshop also had continuous exposures to trichloroethylene
7 associated with degreasing activities; parts were cleaned in cold dip baths and left on tables to
8 dry. Trichloroethylene was regularly used to clean floors, work clothes, and hands of grease, in
9 addition to the intense exposures during specific cleaning exercises, which would produce a
10 background concentration of trichloroethylene in the facility. Cherrie et al. (2001) estimated the
11 long-term exposure to trichloroethylene was approximately 100 ppm.
12 The subjects in this study clearly had substantial peak exposures to trichloroethylene that
13 exceeded 2,000 ppm and probably sustained long-term exposures greater than 100 ppm, which
14 are not confounded by concurrent exposures to other chlorinated organic solvents.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-142 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
Henschler D, Vamvakas S, Lammert M, Dekant W, Kraus B, Thomas B, Ulm K. 1995. Increased incidence of renal cell
tumors in a cohort of cardboard workers exposed to trichloroethene. Arch Toxicol 69:291-299.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract "... retrospective cohort study was carried out in a cardboard factory I
Germany to investigate the association between exposure to trichloroethene and renal
cell cancer."
Employee records were used to identify 183 males employed in a cardboard factory
for at least 1 yr between 1956 and 1975 and with presumed TCE exposure and a
control group of 190 male workers at same factory during the same period of time
but in jobs not involving possible TCE exposure.
Mortality rates from German population residing near factory used as referent in
mortality analysis.
Renal cancer incidence rates from Danish Cancer Registry used to calculate expected
number of incident cancer. The age-standardized rate in the late 1990s among men
in Denmark was 10.6 and in Germany it was 1.2 (Ferlay, 2004). If these differences
in rates apply when the study was carried out, this would imply that the expect
number of deaths would have been inflated by about 14% (and the rate ratio
underestimated by that amount).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Mortality and renal cell cancer incidence.
CATEGORY C: TCE-EXPOSURE CRITERIA
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
ICD-9 for deaths.
Hospital pathology records were used to verify diagnosis of renal cell carcinoma.
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td
•3
1
§
i
^§
H I
O >
HH Oq
H TO
O
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Walkthrough survey and interviews with long-term employees were used to identify
work areas and jobs with potential TCE exposure. The workers in the locksmith's
area and the electrical workshop also had continuous exposures to trichloroethylene
associated with degreasing activities; parts were cleaned in cold dip baths and left on
tables to dry Cherrie et al. (2001) estimated that the machine cleaning
exposures to trichloroethylene were greater than 2,000 ppm with average
long-term exposure as 10-225 ppm. Estimated average chronic exposure to
TCE was -100 ppm to subjects using TCE in cold degreasing processes.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
14 exposed subjects (8%) were excluded from life-table analysis and no information
is presented in paper on loss-to-follow-up among control subjects.
Median follow-up period was over 30 yrs for both exposed and control subjects.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
50 total deaths (30%) and 15 cancer death among exposed subjects.
52 deaths (27%) and 15 cancer deaths among control subjects.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age and calendar-year.
SMR and SIR. Analysis excludes person-years of subjects excluded from exposed
population with the number of person-years underestimated and an underestimate of
the expected numbers of deaths and incident renal carcinoma cases.
H
W
-------
to
O k^j
j*"
vo £3'
Exposure-response analysis presented in
published paper
Documentation of results
No.
Adequate.
SIR = standardized incidence ratio.
Co
1
TO'
•3
td g
^ S
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4.6. Greenland et al (1994).
2 B.3.1.4.6.1. Author's abstract.
3
4 To address earlier reports of excess cancer mortality associated with employment
5 at a large transformer manufacturing plant each plant operation was rated for
6 seven exposures: Pyranol (a mixture of poly chlorinated biphenyls and
7 trichlorobenzene), trichloroethylene, benzene, mixed solvents, asbestos, synthetic
8 resins, and machining fluids. Site-specific cancer deaths among active or retired
9 employees were cases; controls were selected from deaths (primarily
10 cardiovascular deaths) presumed to be unassociated with any of the study
11 exposures. Using job records, we then computed person-years of exposure for
12 each subject. All subjects were white males. The only unequivocal association
13 was that of resin systems with lung cancer (odds ratio = 2.2 at 16.6 years of
14 exposure, P = 0.0001, in a multiple logistic regression including asbestos, age,
15 year of death, and year of hire). Certain other odds ratios appeared larger, but no
16 other association was so robust and remained as distinct after considering the
17 multiplicity of comparisons. Study power was very limited for most associations,
18 and several biases may have affected our results. Nevertheless, further
19 investigation of synthetic resin systems of the type used in the study plant appears
20 warranted.
21
22 B.3.1.4.6.2. Study discussion and comment. This nested case-control study at General
23 Electric's Pittsfield, MA, plant was of deaths reported to the GE pension fund among employees
24 vested in the pension fund. The cohort from which cases and controls were identified was
25 defined as plant employees who worked at the facility before 1984; whose date of deaths was
26 between 1969, the date pension records became available, and 1984; and existence of a job
27 history record. The size of the underlying employee cohort was unknown because work history
28 records did not exist for a large fraction of former employees, especially in the earlier years of
29 deaths. All deaths were identified from records maintained by GE's pension office; other record
30 sources such as the Social Security Administration and National Death Index were not utilized.
31 Requirements for eligibility or "vestment" for a pension varied over time, but for most of the
32 study period, required 10 to 15 years employment with the company. The analysis was restricted
33 to white males because of few deaths among females and nonwhite males. A total of
34 1,911 deaths were identified from pension records and cases and controls, with 90 deaths
35 excluded as possible cases and controls due to several reasons. Cases were identified as
36 site-specific deaths and controls were selected from the remaining noncancer deaths due to
37 circulatory disease, respiratory disease, injury, and other causes. No information was available
38 on the number of controls selected per case. Controls were not matched to cases, were slightly
39 older than cases, and were from earlier birth cohorts which have a lower job history availability
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-146 DRAFT—DO NOT CITE OR QUOTE
-------
1 or greater frequency of missing exposure ratings in work history records (Salvan, 1990).
2 Statistical analysis of the data included covariates for age and year of death.
3 The company's job history record served as the source for exposure rating. The JEM
4 linked possible exposures to over 1,000 job title from 50 separate departments and 100 buildings.
5 A categorical ranking was developed for exposure to seven exposures (Pyranol, TCE, benzene,
6 other solvents, asbestos, resin systems, machining fluids) from 1901 to 1984 based upon on-site
7 interviews with 18 long-term employees and knowledge of one of the study investigators who
8 was an industrial hygienist. Two categories were used for potential TCE exposure: Level 1,
9 duration of indirect exposure (TCE in workplace but does not work directly with TCE) and
10 Level 2, duration of direct work with TCE, with the continuous exposure scores rescaled to the
11 97th percentile of controls (Salvan, 1990). Statistical analyses in Greenaland et al. (1994)
12 collapsed these two categories into a dichotomous ranking of no exposure or any exposure. In
13 many instances, exposure levels were inaccurately estimated and some exposures were highly
14 correlated (Salvan, 1990). Although of low correlation, TCE exposure was statistically
15 significantly correlated with exposure to other solvents (r = 0.11), benzene (r = 0.22) and
16 machining fluids (r = 0.28) (Salvan, 1990). Industrial hygiene monitoring data were not
17 available before 1978 and limited production and purchase records did not extend far back in
18 time (Salvan, 1990). TCE was used as a degreaser since the 1930s and discontinued between
19 1966 and 1975, depending on department. In all, fewer than 10% of jobs were identified as have
20 TCE exposure potential, primarily through indirect exposure and not directly working with TCE.
21 In fact, few subjects were identified with as working directly with TCE (Salvan, 1990). It is not
22 surprising that exposure score distributions were highly skewed towards zero (Salvan, 1990). No
23 details were provided on the protocol for processing the jobs in the work histories into job
24 classifications.
25 Job history information was missing for roughly 35% of the cases and controls,
26 particularly from subjects with earlier years of death. The highest percentage of missing
27 information among cases was for leukemia deaths (43% of deaths) and the lowest percentage for
28 rectal deaths (11%). Moreover, work history records did not exist for a large fraction of former
29 employees, especially in the earlier years of death. Bias resulting from exposure
30 misclassification is likely high due to the lack of industrial monitoring to support rankings and
31 the inability of the JEM to account for changes in TCE exposure concentrations over time.
32 This study had a number of weaknesses with the likely result of dampening observed
33 risks. Deaths were underestimated given nonpensioned employees are not included in the
34 analysis; possible differences in exposure potential between pensioned and nonpensioned
35 workers may introduce bias, particularly if a subject leaves work as a consequence of a
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-147 DRAFT—DO NOT CITE OR QUOTE
-------
1 precondition related to exposure, and would dampen observed associations (Robins and Blevins,
2 1987). Misclassification bias related to exposure is highly likely given missing job history
3 records for over one-third of deaths, mostly among deaths from the earlier study period, a period
4 when TCE was used. Salvan (1990) noted "exposure measurements should be regarded as
5 heavily nondifferentially misclassified relative to the true exposure does" and exposure
6 associations with outcomes will be underestimated. For TCE specifically, the development of
7 exposure assignments in this study was insensitivity to define TCE exposures of the
8 cohort-industrial hygiene data were not available for the time period of TCE use, exposure rates
9 applied to a job-building-operation time matrix and may not reflect individual variation, and
10 exposure ratings obtained by employee interview are subject to subjective assessment and
11 measurement error. NRC (2006) also noted a low likelihood of exposure potential to subjects in
12 this nested case-control study. Overall, the sensitivity of this study for evaluating cancer and
13 TCE exposure is quite limited. The inability of this study to detect associations for two known
14 human carcinogens, benzene and leukemia and asbestos and lung cancer, provides ancillary
15 support for the study's low sensitivity and statistical power.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-148 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
td g
Greenland S, Salvan A, Wegman DH, Hallock MF, Smith TH. 1994. A case-control study of cancer mortality at the
transformer-assembly facility. Int Arch Occup Environ Health 66:49-54.
Greenland S. 1992. A semi-Bayes approach to the analysis of correlated multiple associations with an application to an
occupational cancer-mortality study. Stat Med 11:219-230.
Salvan A. 1990. Occupational exposure and cancer mortality at an electrical manufacturing plant: A case-control study.
Ph.D. Dissertation, University of California, Los Angeles.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The study was carried out to reevaluate an earlier observation from a PMR study of
GE employment and excess leukemia and colorectal cancer risks.
Selection of cases and controls is not adequate because only deaths among pensioned
workers were included in the analysis. Also, the size of the underlying cohort was
not known and potential for selection bias is likely given cases and controls are
drawn from a select population.
Cases were identified from deaths among white males employed before 1984, who
had died between 1969 and 1984, and for whom a job history record was available.
Controls selected from noncancer deaths due to cardiovascular disease, circulatory
disease, respiratory disease, injury, or other causes. Controls are not matched to
cases on covariates such as age, or date of hire.
In total, 2,653 subjects were identified as meeting criteria for inclusion in subject,
either as a case or as a control. Job history records were available for 1,714 (512
cases, 1,202 controls) of these subjects (65%).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Mortality.
^§
H I
O ^
HH Oq
H TO
O
H
W
-------
to
'
I
I
§
Co
1
TO'
•3
I
i
i
TO
CATEGORY C: TCE-EXPOSURE CRITERIA
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
ICDA, 8th revision.
Dichotomous ranking, not exposed/exposed, for indirect and direct exposure
potential. Most subjects identified with indirect TCE exposure. The company's job
history record served as the source for exposure rating. The JEM linked possible
exposures to over 1,000 job title from 50 separate departments and 100 buildings.
Potential TCE exposure assigned to 10% of all job titles. The seven exposures were
highly correlated. NRC (2006) noted a low likelihood of TCE exposure potential to
subjects in this nested case-control study.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Record study.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
td
H S
§
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
220 of 732 cases and 1,202 or 1,921 possible controls had job history records;
history records are missing for 35% of all possible cases and controls.
Any potential TCE exposure prevalence among cases:
Laryngeal, pharyngeal cancer, 38%
Liver and biliary passages, 22%
Pancreas, 45%
Lung, 33%
Bladder, 30%
Kidney, 33%
Lymphoma, 27%
Leukemias, 36%
Brain, 31%
Control exposure prevalence, 34%.
job
Age and year of death. Other unidentified covariates are included if risk estimate is
altered by more than 20%.
Logistic regression with (1) dichotomous exposure (Greenland, 1994) (2) continuous
exposure (Salvan, 1990), (3) epoch analysis (Salvan, 1990), and (4) empirical bayes
models (Greenland, 1992).
No.
Adequate.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4.7. Sinks et al (1992).
2 B.3.1.4.7.1. Author's abstract.
O
4 A physician's alert prompted us to investigate workers' can cancer risk at a
5 paperboard printing manufacturer. We conducted a retrospective cohort mortality
6 study of all 2,050 persons who had worked at the facility for more than 1 day,
7 calculated standardized incidence ratios (SIRs) for bladder and renal cell cancer,
8 and conducted a nested case-control study for renal cell cancer. Standardized
9 mortality ratios (SMRs) from all causes [SMR = 1.0, 95% confidence interval
10 (CI) = 0.9 - 1.2] and all cancers (SMR = 0.6, 95% CI = 0.3 - 1.0) were not greater
11 than expected. One bladder cancer and one renal cell cancer were included in the
12 mortality analysis. Six incident renal cell cancers were observed, however,
13 compared with less than two renal cell cancers expected (SIR = 3.7, 95% CI = 1.4
14 - 8.1). Based on a nested case-control analysis, the risk of renal cell carcinoma
15 was associated with overall length of employment but was not limited to any
16 single department or work process. Although pigments containing congeners of
17 dichlorobenzidine and o-toluidine had been used at the plant, environmental
18 sampling could not confirm any current exposure. Several limitations and a
19 potential selection bias limit the inferences that can be drawn.
20
21 B.3.1.4.7.2. Study description and comment. Sinks et al. (1992) is the published report of
22 analyses examining morbidity and mortality among employees at a James River Corporation
23 plant in Newnan, GA. This plant manufactured paperboard (cardboard) packaging. The study
24 was carried out as a National Institute of Occupational Safety and Health, Health Hazard
25 Evaluation to investigate a possible cluster of urinary tract cancers and work in the plant's
26 Finishing Department (NIOSH, 1992). A cohort of 2,050 white and nonwhite, male and female,
27 subjects were identified from company personnel and death records, considered complete since
28 1-1-1957, and were follows for site-specific mortality and cancer morbidity to 6-30-1988.
29 Records of an additionally 36 subjects were missing hire dates or birth dates, indicated
30 employment duration of less than 1 day, and or employment outside the study period and these
31 subjects were excluded from the analysis. This study suffers from missing information. A large
32 percentage of personnel records did not identify a subject's race and these subjects were
33 considered as white in statistical analyses. Additionally, vital status was unknown for
34 approximately 10% of the cohort. Life-table analyses are based upon U.S. population age-,
35 race-, sex-, calendar- and cause-specific mortality rates. Expected numbers of incident bladder
36 and kidney cancers for white males were derived using white male age-specific bladder and renal
37 cell incidence rates from the Atlanta-Surveillance, Epidemiology, and End Results (SEER)
38 registry for the years 1973 to 1977.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-152 DRAFT—DO NOT CITE OR QUOTE
-------
1 A nested case-control analysis of the incident renal carcinoma cases was also undertaken.
2 This analysis is based on 6 renal cell carcinoma cases and 48 controls (1:8 matching) who were
3 selected by risk set sampling of all employees born within 5 years of the case, the same sex as
4 the case, and having attained the age at which the case was diagnosed or died if date of diagnosis
5 was not known. A diagnosis of renal carcinoma was confirmed for 4 of the 6 cases through
6 pathologic examination. Both the nested case-control analysis and the life-table analyses of
7 morbidity included a renal carcinoma case from the original cluster.
8 Exposures are poorly defined in this study assessing renal cancer among paper board
9 printing workers. Trichloroethylene was mentioned in material-safety data sheets for one or
10 more materials used by the process but no information was provided regarding TCE usage and
11 use by job title. It was not possible to assess the degree of contact with trichloroethylene or the
12 printing inks which were identified as containing benzidine. Furthermore, the lack of monitoring
13 data precludes evaluation of possible exposure intensity. This study is limited for assessing risks
14 associated with exposures to trichloroethylene due to the large percentage of missing information
15 and due to its exposure assessment approach.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-153 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Sinks T, Lushniak B, Haussler BJ, Sniezek J, Deng J-F, Roper P, Dill P, Coates R. 1992. Renal cell cancer among paperboard
printing workers. Epidemiol 3:483-489.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The purpose of the cohort and nested case-control investigations was to determine
whether an excess of bladder or renal cell cancer had occurred among workers in a
paperboard packaging plant and, if so, to determine whether it was associated with
any specific exposure or work-related process.
Cohort analysis: 2 050 males and females employed at the plant between 1-1-1957
and 6-30-1988. External referents for mortality analysis were age-, sex-, race-, and
calendar- cause specific mortality rates of the U.S. population. External referents for
morbidity analysis were age-specific bladder and renal-cell cancer rate for white
males from the Atlanta-SEER registry for the years 1973-1977.
Nested case-control analysis: Cases were all subjects with renal cell cancer;
8 nonrenal cell carcinoma controls chosen from a risk set of all employees matched
to case on date of birth (within 5 yrs), sex and attained age of cancer diagnosis or
death, if diagnosis date unknown.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD revision in effect at the time of death; incident cases of renal cell carcinoma
diagnoses confirmed with pathology reports for 4 of 6 cases.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Exposure in cohort analysis defined broadly at level of the plant and, in case-control
study, department worked as identified on company's personnel.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
Yes, 10% of cohort with unknown vital status (n = 204).
P-Y for these workers were censored at the date of last follow-up.
•3
1
§
i
^§
H I
O >
HH Oq
H TO
O
H
W
td
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
>50% cohort with full latency
18 yr average follow-up.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Department assignment based on company personnel records.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
141 total deaths (7% of cohort had died by end of follow-up), 16 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Mortality analysis: Age, race
Morbidity analysis limited to
Nested case-control analysis:
birth (within 5 yrs), sex, and
, sex, and calendar year.
white males: age.
Risk set sampling matching controls to cases on date of
attained age at diagnosis.
SIR.
Conditional logistic regression used for nested case-control analysis.
No.
Adequate.
I
td
H I
O >
HH Oq
H ^
O
SIR = standardized incidence ratio.
H
W
-------
1 B.3.1.4.8. Blair et al (1989).
2 B.3.1.4.8.1. Author's abstract.
O
4 Work history records and fitness reports were obtained for 1 767 marine
5 inspectors of the U.S. Coast Guard between 1942 and 1970 and for a comparison
6 group of 1 914 officers who had never been marine inspectors. Potential exposure
7 to chemicals was assessed by one of the authors (RP), who is knowledgeable
8 about marine inspection duties. Marine inspectors and noninspectors had a deficit
9 in overall mortality compared to that expected from the general U.S. population
10 (standardized mortality ratios [SMRs = 79 and 63, respectively]). Deficits
11 occurred for most major causes of death, including infectious and parasitic
12 diseases, digestive and urinary systems, and accidents. Marine inspectors had
13 excesses of cirrhosis of the liver (SMR = 136) and motor vehicle accidents (SMR
14 = 107, and cancers of the lymphatic and hematopoietic system (SMR = 157,
15 whereas noninspectors had deficits for these causes of death. Comparison of
16 mortality rates directly adjusted to the age distribution of the inspectors and
17 noninspectors combined also demonstrated that mortality for these causes of death
18 was greater among inspectors than noninspectors (directly adjusted ratio ratios of
19 190, 145, and 198) for cirrhosis of the liver, motor vehicle accidents, and
20 lymphatic and hematopoietic system cancer, respectively. The SMRs rose
21 with increasing probability of exposure to chemicals for motor vehicle accidents,
22 cirrhosis of the liver, liver cancer, and leukemia, which suggests that contact with
23 chemicals during inspection of merchant vessels may be involved in the
24 development of these diseases among marine inspectors, physician's alert
25 prompted us to investigate workers' can cancer risk at a paperboard printing
26 manufacturer. We conducted a retrospective cohort mortality study of all 2,050
27 persons who had worked at the facility for more than 1 day, calculated
28 standardized incidence ratios (SIRs) for bladder and renal cell cancer, and
29 conducted a nested case-control study for renal cell cancer. Standardized
30 mortality ratios (SMRs) from all causes [SMR = 1.0, 95% confidence interval
31 (CI) = 0.9-1.2] and all cancers (SMR = 0.6, 95% CI = 0.3-1.0) were not greater
32 than expected. One bladder cancer and one renal cell cancer were included in the
33 mortality analysis. Six incident renal cell cancers were observed, however,
34 compared with less than two renal cell cancers expected (SIR = 3.7, 95% CI = 1.4
35 - 8.1). Based on a nested case-control analysis, the risk of renal cell carcinoma
36 was associated with overall length of employment but was not limited to any
37 single department or work process. Although pigments containing congeners of
38 dichlorobenzidine and o-toluidine had been used at the plant, environmental
39 sampling could not confirm any current exposure. Several limitations and a
40 potential selection bias limit the inferences that can be drawn.
41
42 B.3.1.4.8.2. Study description and comment. This cohort of 1,767 U. S. Coast Guard male
43 officers and enlisted personnel performing marine inspection duties between 1942 and 1970 and
44 1,914 noninspectors matched to inspectors for registry, rank and year that rank was achieved
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-156 DRAFT—DO NOT CITE OR QUOTE
-------
1 examined mortality as of January 1, 1980. Standardized mortality ratios compared the observed
2 number of site-specific deaths among marine inspectors (n = 483, 27%) to that expected of the
3 total U. S. white male population and to standardized mortality ratios of noninspectors (n = 369,
4 19%). The cohort was predominantly white (91%), race was unknown for the remaining 8% of
5 subjects, considered in the statistical analysis as white, with a large percentage (69%) of the
6 marine inspectors having >20 year employment duration. The minimum latent period was 10
7 years, calculated from the end date of cohort identification to the date of vital status
8 ascertainment.
9 This study lacks exposure information on potential exposures of marine inspectors, who
10 enter cargo tanks, void spaces, cofferdams, and pump rooms during inspections. TCE is
11 identified in the paper as a possible exposure along with nine other agents. One authors
12 acquainted with Coast Guard processes estimated the level of exposure to general chemical
13 exposures during a marine inspection. A four-point rating scales was developed: nonexposed,
14 person generally held administrative position; low exposed, assigned to staff with duties that
15 occasionally required vessel inspections; moderate exposed, assign to inspection duties that did
16 not regularly include hull structures, and regular inspection of hull structures in geographic areas
17 where chemicals were not major items of cargo; and, high exposed, assigned to subjects who
18 performed hull inspections at ports were vessels transported chemicals. A cumulative exposure
19 score was calculated by summing the product of the four-point rating scale and the duration in
20 each job.
21 Overall, the exposure assessment in this study is insufficient for examining TCE
22 exposure and cancer mortality. Furthermore, the few site-specific deaths among marine
23 inspectors greatly limits statistical power.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-157 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
Blair A, Haas T, Prosser R, Morrissette M, Blackman, Grauman D, van Dusen P, Morgan F. 1989. Mortality among United
States Coast Guard marine Inspectors. Arch Environ Health 44:150-156.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The purpose of the cohort study was to examine mortality patterns among Coast
Guard marine inspectors. This study was not designed to examine specific
exposures, including TCE.
1,767 U. S. Coast Guard male officers and enlisted personnel performing marine
inspections between 1942 and 1970 and 1,914 noninspectors matched to inspectors
on registry, rank, and year that rank was achieved.
External referents: age-specific mortality rates of the U. S. white male population
and noninspectors.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICDA, 8th revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
TCE identified in paper as one often potential exposures; however, no exposure
assessment to TCE to individual subjects. Exposure in cohort analysis defined
broadly at level of the plant and, in case-control study, department worked as
identified on company's personnel. A cumulative exposure surrogate developed from
duration in each job and a four-point rating scale: nonexposed, person generally held
administrative position; low exposed, assigned to staff with duties that occasionally
required vessel inspections; moderate exposed, assign to inspection duties that did
not regularly include hull structures, and regular inspection of hull structures in
geographic areas where chemicals were not major items of cargo; and, high exposed,
assigned to subjects who performed hull inspections at ports were vessels transported
chemicals.
CATEGORY D: FOLLOW-UP (COHORT)
•3
I
td
00
i
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
More than 10% loss to follow-up
>50% cohort with full latency
No
Not reported; minimum latent period was 10 years.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
483 deaths among marine inspectors (27% of cohort), 103 cancer deaths.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Mortality analysis: Age, race, sex, and calendar year. Directly adjusted rate ratios
compared cause-specific SMR of marine inspectors to that of noninspectors.
SMR and RR.
Yes, using a ranked cumulative exposure surrogate.
Adequate.
I
td
* ^
H I
O ^-
h^ C^
RR = relative risk. SMR = standardized mortality ratio.
O
H
W
-------
1 B.3.1.4.9. Shannon et al (1988).
2 B.3.1.4.9.1. Author's abstract.
O
4 A historical prospective study of cancer in lamp manufacturing workers in one
5 plant was conducted. All men and women who worked for a total of at least 6
6 months and were employed at some time between 1960 and 1975 were included.
7 Work histories were abstracted and subjects were divided according to whether
8 they had worked in the coiling and wire drawing area (CWD). Cancer morbidity
9 from 1964 to 1982 was ascertained via the provincial registry, and was compared
10 with the site-specific incidence in Ontario, adjusting for age, sex and calendar
11 period. Of particular interest were primary breast and gynecological cancers in
12 women.
13 The cancers of a priori concern were significantly increased in women in CWD,
14 but not elsewhere in the plant. The excess was greatest in those with more than 5
15 yr exposure (in CWD) and more than 15 yr since first working in CWD, with
16 eight cases of breast and gynecological cancers observed in this category
17 compared with 2.67 expected. Only three cancers occurred in men in CWD.
18 Environmental measurements had not been made in the past and little information
19 was available on substances used in the 1940s and 1950s, the period when the
20 women with the highest excess began employment. It is known that methylene
21 chloride and trichloroethylene have been used, but not enough is known about the
22 dates and patterns.
23
24 B.3.1.4.9.2. Study description and comments. This cohort of 1,770 workers (1,044 females,
25 826 males) employed >6 months and working between 1960 and 1975 at a General Electric plant
26 in Ontario, Canada, in the lamp manufacturing department identified cancer incidence cases from
27 a regional cancer registry from 1964, the first date of high quality information, to 1982. Office
28 workers were included in the study population. The study was carried out in response to
29 previous reports of excess breast and gynecological cancer in women employed in the CWD
30 area. Standardized incidence ratios (SIR) compared the observed number of site-specific
31 incident cancers to that expected of the Ontario population and supplied by the regional cancer
32 registry. SIR estimates were calculated for all lamp department workers, and for two subgroups
33 defined by job title, workers in the coil and wire-drawing area (CWD) and workers in all other
34 areas. The cohort was successfully traced, with low rates of lost to follow-up (6% among CWD
35 workers, 7 all other workers). A total of 98 incident cancer cases were identified (58 in females,
36 40 in males) and over half of the incident cancers in females (n = 31) due to breast and
37 gynecological cancers. The number of incident cancers is likely underestimated given the 4-year
38 period between cohort identification and the first date of high quality information in the cancer
39 registry. Additionally, cancer cases among workers who moved from the province would not be
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-160 DRAFT—DO NOT CITE OR QUOTE
-------
1 found in the registry, leading to underascertainment of cases. This is likely a small number given
2 follow-up tracing identified 2% of workers had left the province.
3 This study lacks exposure information on individual study subjects. Exposures in CWD
4 were of concern given previous reports. The study lacks exposure monitoring data and potential
5 exposures in CWD area were identified using purchase records. A number of chemicals were
6 identified including methylene chloride from 1959 onward and trichloroethylene, which records
7 suggested may have been used beforehand.
8 Overall, the exposure assessment in this study is insufficient for examining TCE
9 exposure and cancer mortality. The inclusion of office workers, who likely have low potential
10 exposure, would introduce a downward bias. Furthermore, the few site-specific deaths among
11 CWD and all other workers greatly limits statistical power.
12
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-161 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Shannon HS, Haines T, Bernholz C, Julian JA, Verma DK, Jamieson E, Walsh C. 1988. Cancer morbidity in lamp
manufacturing workers. Am J Ind Med 14:281-290.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study was undertaken in response to previous report of apparent excess breast
and gynecological cancers in women employed in the coil and wire drawing area of a
lamp manufacturing plant.
Cohort analysis: 1,770 workers (1,044 females, 826 males)in the lamp manufacturing
department of a GE plant in Ontario Province, Canada.
External referents: Age-, sex- and race-specific site-specific cancer incidence rates
for Ontario Province population
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
Not reported.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
This study does not assign TCE exposure to individual subjects.
in the CWD area used to assign exposure potential and chemical
identified from purchase records. Methylene chloride used from
one report from 1955 indicating TCE used as degreasing solvent
Job title and work
usage in CWD
1959 onward, with
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No, follow-up was complete for 6% of CWD workers and 7% for all other workers.
Not reported
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
I
td
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
98 incident cancer cases
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, race, sex, and calendar year.
SIR.
No.
Adequate.
td
'•I
CWD = coil and wire drawing area. SIR = standardized incidence ratio.
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4.10. Shindett and Vlrich (1985).
2 B.3.1.4.10.1. Author's abstract.
O
4 A prospective study was conducted of 2,646 employees who worked three months
5 or more during the period January, 1957, through July, 1983, in a manufacturing
6 plant that used trichloroethylene as a degreasing agent throughout the study
7 period. Ninety-eight percent of the study cohort were traced; they accounted for
8 16,388 person-years of employment and 38,052 person-years of follow-up.
9 Mortality experience was found to be generally more favorable than that of the
10 comparable segment of the U.S. population over the same period of time. For the
11 white male cohort there were fewer deaths than expected from heart disease,
12 cancer, and trauma (standard mortality rate for all causes = 0.79, p less than .01).
13 Reports by current and former employees of health problems requiring medical
14 treatment showed that there were only one third as many persons with heart
15 disease or hypertension as were reported in a comparable reference population
16 studied over the past five years.
17
18 B.3.1.4.10.2. Study description and comment. This study of 2, 546 current and former office
19 and production employees at a manufacturing plant in northern Illinois compares broad
20 groupings of cause-specific mortality between 1957 and 1983 to expected number of deaths
21 based on U.S. population mortality rates for the period. The published paper lacks an assessment
22 of TCE exposure other than noting TCE was used as a degreasing agent at the plant. No
23 information is presented on quantity used, job titles with potential exposure, or likely exposure
24 concentrations Not all study subjects had the same potential for exposure and the inclusion of
25 office workers who had a very low exposure potential decreased the study's detection sensitivity.
26 Deaths were identified from company records or from direct or indirect contact with former
27 employees or next-of-kin for subjects not known to the company to be deceased instead of using
28 national-based registries such as Social Security listings or National Death Index for identifying
29 vital status. There were few deaths in this cohort, a total of 141 among male and female
30 subjects; vital status could not be ascertained for 52 subjects. The few numbers of cancer deaths
31 (21 total) precluded examination of cause-specific cancer mortality. Overall, this study provides
32 no information on TCE and cancer; it lacked exposure assessment to TCE and the few cancer
33 deaths observed greatly limited its detection sensitivity.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-164 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
S'
1
TO'
Shindell S, Ulrich S. 1985. A cohort study of employees of a manufacturing plant using trichloroethylene. J Occup Med
27:577-579.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study was designed to assess mortality patterns of office and production
employees at an Illinois manufacturing plant.
2,646 males and female workers employed from 1-1-1957 to 7-31-1983. Mortality
rates of U.S. population used as referent. The paper lacks information on source for
identifying cohort subjects and if company records were complete.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
Not identified.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
The paper does not identify TCE usage other than as a degreaser. Conditions
exposure and jobs potentially exposure are not identified in paper. This study
an assessment of TCE exposure.
of
lacks
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
2%.
No information provided in paper.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
I
-
§
td
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
This study does not use standard approaches to verify deaths and vital status. Deaths
are self-reported in response to contact by employer representative. 141 deaths (6%)
were reported to employer, 9 deaths lacked a death certificate.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Sex and race.
SMR.
No.
The paper lacks discussion of process used to contact former employees to verify
vital status and methods used to identify subjects.
I
td
ON
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.1.4.11. Wilcosky et al (1984).
2 B.3.1.4.11.1. Author's abstract.
O
4 Some evidence suggests that solvent exposures to rubber industry workers may be
5 associated with excess cancer mortality, but most studies of rubber workers lack
6 information about specific chemical exposure. In one large rubber and tire-
7 manufacturing plant, however, historical documents allowed a classification of
8 jobs based on potential exposures to all solvents that were authorized for use in
9 the plant. A case-control analysis of a 6,678 member cohort compared the solvent
10 exposure histories of a 20% age-stratified random sample of the cohort with those
11 of cohort members who died during 1964-1973 for stomach cancer, respiratory
12 system cancer, prostate cancer, lymphosarcoma, or lymphatic leukemia. Of these
13 cancers, only lymphosarcoma and lymphatic leukemia showed significant positive
14 associations with any other potential solvents exposures. Lymphatic leukemia
15 was especially strongly related to carbon tetrachloride (OR = 1.3, p< .0001) and
16 carbon disulfide (OR = 8.9, p = .0003). Lymphosarcoma showed similar, but
17 weaker, association with these two solvents. Benzene, a suspected carcinogen,
18 was not significantly associated with any of the cancers.
19
20 B.3.1.4.11.2. Study description and comment. Exposure was assessed in this nested
21 case-control study of four site-specific cancers among rubber workers at a plant in Akron, OH
22 through use of a JEM originally used to examine benzene specifically, but had the ability to
23 assess 24 other solvents, including TCE, or solvent classes. Exposure was inferred using
24 information on production operations and product specifications that indicated whether solvents
25 were authorized for use during tire production, and by process area and calendar year. A
26 subject's work history record was linked to the JEM to assign exposure potential to TCE.
27 Overall, a low prevalence of TCE exposure, ranging from 9 to 20% for specific cancers was
28 observed among cases.
29 The JEM was developed originally to assign exposure to benzene and other aromatic
30 solvents in a nested case-control study of lymphocytic leukemia (Arp et al., 1983). Details of
31 exposure potential to TCE are not described by either Arp et al. (1983) or Wilcosky et al. (1984).
32 No data were provided on the frequency of exposure-related tasks. Without more information, it
33 is not possible to determine the quality of some of the assignments. Similarly, the lack of
34 industrial hygiene monitoring data precluded validation of the JEM.
35 Cases of respiratory, stomach and prostate cancers; lymphosarcoma and reticulum cell
36 sarcoma; and lymphatic leukemia were identified from a previous study which had observed
37 associations with these site-specific cancers among a cohort of rubber workers employed at a
38 large tire manufacturing plant in Akron, OH. Statistical power is low in this study, particularly
39 for evaluation of lymphatic cancer for which there were 9 cases of lymphosarcoma and 10 cases
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-167 DRAFT—DO NOT CITE OR QUOTE
-------
1 of lymphatic leukemia. Controls were chosen from a 20% age-stratified random sample of the
2 cohort. The published paper does not identify if subjects with other diseases associated with
3 solvents or TCE were excluded as controls. If no exclusion criteria were adopted, a bias may
4 have been introduced which would dampen observed associations towards the null.
5 The few details provided in the paper on exposure assessment and JEM developments,
6 few details of control selection, the low prevalence of TCE exposure and the few lymphatic
7 cancer cases greatly limit the ability of this study for assessing risks associated with exposures to
8 trichloroethylene.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-168 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Wilcosky TC, Checkoway H, Marshall EG, Tyroler HA. 1984. Cancer mortality and solvent exposure in the rubber industry.
Am Ind Hyg Assoc J 45:809-811.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study was identified as "exploratory" to examine several site-specific cancer
and specific solvents, primarily benzene.
Underlying population at risk was a cohort of 6,678 male workers employed in the
rubber industry in 1964. Cases are deaths due to respiratory, stomach and prostate
cancers; lymphosarcoma; and lymphatic leukemia observed in the cohort analysis —
30 deaths due to stomach cancer, 333 deaths from prostate cancer, 9 deaths from
lymphosarcoma, and 10 deaths from lymphatic leukemia.
Controls were a 20% age-stratified random sample of the cohort (exclusion criteria
not identified in paper).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICDA, 8th revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Procedure to assign TCE and other solvent exposures based upon JEM developed
originally to assess benzene and other solvent exposures (Arp et al., 1983). The JEM
was linked to a detailed work history as identified from a subject's personnel record
to assign TCE exposure potential. Details of JEM for TCE not well-described in
Wilcosky et al. (1984). Multiple solvent exposures likely (McMichael et al., 1976).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
I
td
vo
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td g
i
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Record study with exposure assignment using JEM and personnel records.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
N/A
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
TCE exposure prevalence:
Stomach cancer, 5 exposed cases (17% exposure prevalence)
Prostate cancer, 3 exposed cases (9% exposure prevalence)
Lymphosarcoma, 3 exposed cases (33% exposure prevalence)
Lymphatic leukemia, 2 exposed cases (20% exposure prevalence).
No information presented in paper on exposure prevalence among control subjects.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age.
Not described in published paper.
No.
Methods and analyses not fully described in published paper.
H 3
§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2. Case-Control Studies
2 B.3.2.1. Bladder Cancer Case-Control Studies
3 B.3.2.1.1. Pesch et al (2000a).
4 B.3.2.1.1.1. Author's abstract.
5
6 BACKGROUND: This multicentre population-based case-control study was
7 conducted to estimate the urothelial cancer risk for occupational exposure to
8 aromatic amines, polycyclic aromatic hydrocarbons (PAH), and chlorinated
9 hydrocarbons besides other suspected risk factors. METHODS: In a population-
10 based multicentre study, 1035 incident urothelial cancer cases and 4298 controls
11 matched for region, sex, and age were interviewed between 1991 and 1995 for
12 their occupational history and lifestyle habits. Exposure to the agents under study
13 was self-assessed as well as expert-rated with two job-exposure matrices and a job
14 task-exposure matrix. Conditional logistic regression was used to calculate
15 smoking adjusted odds ratios (OR) and to control for study centre and age.
16 RESULTS: Urothelial cancer risk following exposure to aromatic amines was
17 only slightly elevated. Among males, substantial exposures to PAH as well as to
18 chlorinated solvents and their corresponding occupational settings were associated
19 with significantly elevated risks after adjustment for smoking (PAH exposure,
20 assessed with a job-exposure matrix: OR = 1.6, 95% CI: 1.1-2.3, exposure to
21 chlorinated solvents, assessed with a job task-exposure matrix: OR = 1.8, 95% CI:
22 1.2-2.6). Metal degreasing showed an elevated urothelial cancer risk among males
23 (OR = 2.3, 95% CI: 1.4-3.8). In females also, exposure to chlorinated solvents
24 indicated a urothelial cancer risk. Because of small numbers the risk evaluation
25 for females should be treated with caution. CONCLUSIONS: Occupational
26 exposure to aromatic amines could not be shown to be as strong a risk factor for
27 urothelial carcinomas as in the past. A possible explanation for this finding is the
28 reduction in exposure over the last 50 years. Our results strengthen the evidence
29 that PAH may have a carcinogenic potential for the urothelium. Furthermore, our
30 results indicate a urothelial cancer risk for the use of chlorinated solvents.
31
32 B.3.2.1.1.2. Study description and comment. This multicenter study of urothelial (bladder,
33 ureter, and renal pelvis) and renal cell carcinoma in Germany included the five regions (West
34 Berlin, Bremen, Leverkusen, Halle, Jena), identified two case series from participating hospitals,
35 1,035 urothelial cancer cases and 935 renal cell carcinoma cases with a single population control
36 series matched to cases by region, sex, and age (1:2 matching ratio to urothelial cancer cases and
37 1:4 matching ratio to renal cell carcinoma cases). Findings in Pesch et al. (2000a) are from
38 analyses of urothelial cancer analysis and Pesch et al. (2000b) from analyses of renal cell
39 carcinoma. In all, 1,035 (704 males, 331 females) urothelial carcinoma cases were interviewed
40 face-to-face using with a structured questionnaire in the hospital within 6 months of first
41 diagnosis and 4,298 randomly selected population controls were interviewed at home. Logistic
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-171 DRAFT—DO NOT CITE OR QUOTE
-------
1 regression models were fit separately to for males and females conditional on age (nine 5-year
2 groupings), study region, and smoking, to examine occupational chemical exposures and
3 urothelial carcinoma.
4 Two general JEMs, British and German, were used to assign exposures based on
5 subjects' job histories reported in an interview. This approach was the same as that described for
6 the renal cell carcinoma analysis of Pesch et al. (2000b). Researchers also asked about job tasks
7 associated with exposure, such as metal degreasing and cleaning, and use of specific agents
8 (organic solvents chlorinated solvents, including specific questions about carbon tetrachloride,
9 trichloroethylene, and tetrachloroethylene) to evaluate TCE potential using a ITEM. A category
10 of "any use of a solvent" mixes the large number with infrequent slight contact with the few
11 noted earlier who have high intensity and prolonged contact. Analyses examining
12 trichloroethylene exposure using either the JEM of ITEM assigned a cumulative TCE exposure
13 index of none to low, medium high and substantial, defined as the product of exposure
14 probability x intensity x duration with the following cutpoints: none to low, <30th percentile of
15 cumulative exposure scores; medium, 30th-<60th percentile; high, 60th-<90th percentile; and,
16 substantial, >90th percentile. The use of the German JEM identified approximately twice as
17 many cases with any potential TCE exposure (44%) compared to the ITEM (22%) and, in both
18 cases, few cases identified with substantial exposure, 7% by JEM and 5% by JTEM. Pesch et al.
19 (2000a) noted "exposure indices derived from an expert rating of job tasks can have a higher
20 agent-specificity than indices derived from job titles." For this reason, the JTEM approach with
21 consideration of job tasks is considered a more robust exposure metric for examining TCE
22 exposure and urothelial carcinoma due to likely reduced potential for exposure misclassification
23 compared to TCE assignment using only job history and title.
24 While this case-control study includes a region in the North Rhine-Westphalia region
25 where the Arnsberg area is also located, several other regions are included as well, where the
26 source of the trichloroethylene and chlorinated solvent exposures are expected as much less well
27 defined. Few cases were identified as having substantial exposure to TCE and, as a result, most
28 subjects identified as exposed to trichloroethylene probably had minimal contact, averaging
29 concentrations of about 10 ppm or less (NRC, 2006).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-172 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Pesch B, Haerting H, Ranft U, Klimpel A, Oelschlagel B, Schill W, and the MURC Study Group. 2000a. Occupational risk
factors for urothelial carcinoma: agent-specific results from a case-control study in Germany. Int J Epidemiol 29:238-247.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups
and of cases and controls in case-control
studies is adequate
Yes, this case-control study was conducted to estimate urothelial carcinoma risk for
exposure to occupational-related agents; chlorinated solvents including trichloroethylene
were identified as exposures of a priori interest.
1,035 urothelial (bladder, ureter, renal pelvis) carcinoma cases were identified from
hospitals in a five-region area in Germany between 1991 and 1995. Cases were
confirmed histologically. 4,298 population controls identified from local residency
registries in the five-region area were frequency matched to cases by region, sex and age
comprised the control series for both the urothelial carcinoma cases and the RCC cases,
published as Pesch et al. (2000a).
Participation rate: cases, 84%; controls, 71%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
No information in paper.
1
td
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
td g
-§
^§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative
exposure estimates
A trained interviewer interviewed subjects using a structured questionnaire which covered
occupational history and job title for all jobs held longer than one yr, medical history, and
personal information. Two general JEMs, British and German, were used to assign
exposures based on subjects' job histories reported in an interview. Researchers also
asked about job tasks associated with exposure, such as metal degreasing and cleaning,
and use of specific agents (organic solvents chlorinated solvents, including specific
questions about carbon tetrachloride, trichloroethylene, and tetrachloroethylene) and
chemical-specific exposure were assigned using a ITEM. Exposure index for each
subject is the sum over all jobs of duration x probability x intensity. A four category
grouping was used in statistical analyses defined by exposure index distribution of
controls: no-low; medium, 30th percentile; high, 60th percentile; substantial, 90th
percentile.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Interviewers carried out face-to-face interview with all cases and controls. All cases were
interviewed in the hospital within 6 mos of initial diagnosis. All controls had home
interviews.
No, by nature of interview location.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No, all cases and controls were alive at time of interview.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality
studies; numbers of total cancer incidence
studies; numbers of exposed cases and
prevalence of exposure in case-control
studies
JEM: 460 cases with TCE exposure index of medium or higher (44% exposure prevalence
among cases), 71 cases with substantial exposure (7% exposure prevalence).
JTEM: 157 cases with TCE exposure index of medium or higher (22% exposure
prevalence among cases), and 36 males assigned substantial exposure (5% exposure
prevalence).
No information is presented in paper on control exposure prevalence.
-------
to
- a
o §"
vo £3'
I
§
***.
£3'
1
CATEGORY H: ANALYSIS
Control for potential confounders in
statistical analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, study center, and smoking.
Conditional
logistic regression.
Yes.
Yes.
TO'
•s
td g
^ S
-------
1 B.3.2.1.2. Siemiatycki et al (1994), Siemiatycki (1991).
2 B.3.2.1.2.1. Author's abstract.
O
4 A population-based case-control study of the associations between various
5 cancers and occupational exposures was carried out in Montreal, Quebec, Canada.
6 Between 1979 and 1986, 484 persons with pathologically confirmed cases of
7 bladder cancer and 1,879 controls with cancers at other sites were interviewed, as
8 was a series of 533 population controls. The job histories of these subjects were
9 evaluated by a team of chemist/hygienists for evidence of exposure to a list of 294
10 workplace chemicals, and information on relevant non-occupational confounders
11 was obtained. On the basis of results of preliminary analyses and literature
12 review, 19 occupations, 11 industries, and 23 substances were selected for in-
13 depth multivariate analysis. Logistic regression analyses were carried out to
14 estimate the odds ratio between each of these occupational circumstances and
15 bladder cancer. There was weak evidence that the following substances may be
16 risk factors for bladder cancer: natural gas combustion products, aromatic amines,
17 cadmium compounds, photographic products, acrylic fibers, polyethylene,
18 titanium dioxide, and chlorine. Among the substances evaluated which showed no
19 evidence of an association were benzo(a)pyrene, leather dust, and formaldehyde.
20 Several occupations and industries were associated with bladder cancer, including
21 motor vehicle drivers and textile dyers.
22
23 B.3.2.1.2.2. Study description and comment. Siemiatycki et al. (1994) and Siemiatycki (1991)
24 reported data from a case-control study of occupational exposures and bladder cancer conducted
25 in Montreal, Quebec (Canada) and part of a larger study of 10 other site-specific cancers and
26 occupational exposures. The investigators identified 617 newly diagnosed cases of primary
27 bladder cancer, confirmed on the basis of histology reports, between 1979 and 1985; 484 of these
28 participated in the study interview (78% participation). One control group (n = 1,295) consisted
29 of patients with other forms of cancer (excluding lung and kidney cancer) recruited through the
30 same study procedures and time period as the bladder cancer cases. A population-based control
31 group (n = 533, 72% response), frequency matched by age strata, was drawn using electoral lists
32 and random digit dialing. Face-to-face interviews were carried out with 82% of all cancer cases
33 with telephone interview (10%) or mailed questionnaire (8%) for the remaining cases. Twenty
34 percent of all case interviews were provided by proxy respondents. The occupational assessment
35 consisted of a detailed description of each job held during the working lifetime, including the
36 company, products, nature of work at site, job activities, and any additional information that
37 could furnish clues about exposure from the interviews.
38 A team of industrial hygienists and chemists blinded to subject's disease status translated
39 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
40 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-176 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure dimensions was categorized into none, any, or substantial exposure. Siemiatycki et al.
2 (1994) presents observations of analyses examining job title, occupation, and some chemical-
3 specific exposures, but not TCE. Observations on TCE are found in the original report of
4 Siemiatycki (1991). Any exposure to TCE was 2% among cases (n = 8) but <1% for substantial
5 TCE exposure (n = 5); "substantial" is defined as >10 years of exposure for the period up to
6 5 years before diagnosis. Logistic regression models adjusted for age, ethnicity, socioeconomic
7 status, smoking, coffee consumption, and status of respondent (Siemiatycki et al., 1994) or
8 Mantel-Henszel $ stratified on age, family income, cigarette smoking, coffee, and respondent
9 status (Siemiatycki, 1991). Odds ratios for TCE exposure are presented in Siemiatycki (1991)
10 with 90% confidence intervals.
11 The strengths of this study were the large number of incident cases, specific information
12 about job duties for all jobs held, and a definitive diagnosis of bladder cancer. However, the use
13 of the general population (rather than a known cohort of exposed workers) reduced the likelihood
14 that subjects were exposed to TCE, resulting in relatively low statistical power for the analysis.
15 The j ob exposure matrix, applied to the j ob information, was very broad since it was used to
16 evaluate 294 chemicals.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-177 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
Siemiatycki J, Dewar R, Nadon L, Gerin M. 1994. Occupational risk factors for bladder cancer: results from a case-control
study in Montreal, Quebec, Canada. Am J Epidemiol 140:1061-1080.
Siemiatycki J. 1991. Risk Factors for Cancer in the Workplace. Baca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
617 bladder cancer cases were identified among male Montreal residents between
1979 and 1985 of which 484 were interviewed.
740 eligible male controls identified from the same source population using random
digit dialing or electoral lists; 533 were interviewed. A second control series
consisted of all other cancer controls excluding lung and kidney cancer cases.
Participation rate: cases, 78%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 188 (Malignant neoplasm of bladder).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semi quantitative scale developed for 300 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
•3
I
td
oo
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
HH Oq
H ^
O
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were conducted
either at home or in the hospital; all population control interviews were conducted at
home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
484 cases (78% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 2% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), <1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, income, index for cigarette smoking, coffee, and respondent status
(Siemiatycki, 1991).
Age, ethnicity, socioeconomic status, smoking, coffee consumption, and status of
respondent (Siemiatycki etal., 1994).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Siemiatycki et al., 1994).
No.
Yes.
H
W
-------
1 B.3.2.2. Central Nervous System Cancers Case-Control Studies
2 B.3.2.2.1. De Roos et al (2001).
3 B.3.2.2.1.1. Author's abstract.
4
5 To evaluate the effects of parental occupational chemical exposures on incidence
6 of neuroblastoma in offspring, the authors conducted a multicenter case-control
7 study, using detailed exposure information that allowed examination of specific
8 chemicals. Cases were 538 children aged 19 years who were newly diagnosed
9 with confirmed neuroblastoma in 1992-1994 and were registered at any of 139
10 participating hospitals in the United States and Canada. One age-matched control
11 for each of 504 cases was selected through random digit dialing. Self-reported
12 exposures were reviewed by an industrial hygienist, and improbable exposures
13 were reclassified. Effect estimates were calculated using unconditional logistic
14 regression, adjusting for child's age and maternal demographic factors. Maternal
15 exposures to most chemicals were not associated with neuroblastoma. Paternal
16 exposures to hydrocarbons such as diesel fuel (odds ratio (OR) = 1.5; 95%
17 confidence interval (CI): 0.8, 2.6), lacquer thinner (OR = 3.5; 95% CI: 1.6, 7.8),
18 and turpentine (OR = 10.4; 95% CI: 2.4, 44.8) were associated with an increased
19 incidence of neuroblastoma, as were exposures to wood dust (OR = 1.5; 95% CI:
20 0.8, 2.8) and solders (OR = 2.6; 95% CI: 0.9, 7.1). The detailed exposure
21 information available in this study has provided additional clues about the role of
22 parental occupation as a risk factor for neuroblastoma.
23
24 B.3.2.2.1.2. Study description and comment. De Roos et al. (2001), a large multicenter
25 case-control study of neuroblastoma in offspring and part of the pediatric collaborative clinical
26 trial groups, the Children's Cancer Group and the pediatric Oncology Group, examined parental
27 and maternal chemical exposures, focusing on solvent exposures, expanding the exposure
28 assessment approach of Olshan et al. (1999) who examined parental occupational title among
29 cases and controls. Neuroblastoma in patients under the age of 19 years was identified at one of
30 139 participating hospitals in the United States and Canada from 1992 to 1996. One population
31 control per case s was using a telephone random digit dialing procedure and matched to the case
32 on date of birth (+6 months for cases 3 years old or younger and +1 year for cases old than
33 3 years of age). A total of 741 cases and 708 controls were identified with direct interviews by
34 telephone obtained from 538 case mothers (73% participation), 405 case fathers, 504 control
35 mothers (71% participation), and 304 control fathers. Mothers served as proxy respondents for
36 paternal information for 67 cases (12%) and 141 controls (28%).
37 A strength of the study was its use of industrial hygienist review of self-reported
38 occupational exposure to increase specificity, reduce the number of false-positive information
39 from self-reported exposures, and to minimize exposure misclassification bias. A parent was
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-180 DRAFT—DO NOT CITE OR QUOTE
-------
1 coded as having been exposed to individual chemicals or chemical group (halogenated
2 hydrocarbons, paints, metals, etc.) if the industrial hygiene review determined probable exposure
3 in any job. Individual chemicals in the halogenated hydrocarbons grouping included carbon
4 tetrachloride, chloroform, Freon, methylene chloride, perchloroethylene and TCE. Typical of
5 population case-control studies, reported TCE exposure was uncommon among cases and
6 controls. Only 6 case and 8 control mothers were identified by industrial hygiene review of
7 occupational information to have probable exposure to halogenated hydrocarbons. The few
8 numbers prevented examination of specific chemical exposure. Of the 538 cases and
9 504 controls, paternal exposure to TCE was self-reported for 22 cases (5%) and 12 controls (4%)
10 were identified with paternal TCE exposure with fewer fathers with probable TCE exposure
11 confirmed from industrial hygiene expert review, 9 cases (2%) and 7 controls (2%).
12 Overall, this study has a low sensitivity and statistical power for evaluating parental TCE
13 exposure and neuroblastoma in offspring due to the low exposure prevalence to TCE. Although
14 study investigators took effort to reduce false positive reporting, exposure misclassification bias
15 may still be possible from false negative reporting of occupational information. As discussed by
16 study authors, job duty information reported by parents was best used to infer exposure to
17 chemical categories but was not detailed sufficiently to infer specific exposures. The study's
18 reported risk estimates for TCE exposure are imprecise and do not provide support for or against
19 an association.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-181 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
De Roos AJ, Olshan AF, Teschke K, Poole Ch, Savitz DA, Blatt J, Bondy ML, Pollock BH. 2001. Parental occupational
exposure to chemicals and incidence of neuroblastoma in offspring. Am J Epidemiol 154:106-114.
Olshan AF, De Roos AJ, Teschke K, Neglin JP, Stram DO, Pollock BH, Castleberry RP. 1999. Neuroblastoma and parental
occupation. Cancer Causes Control 10:539-549.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or hypothesis
Selection and characterization in cohort studies of
exposure and control groups and of cases and
controls in case-control studies is adequate
This multicenter population case-control study examined parental
chemical-specific occupational exposures using detailed exposure information.
538 cases of neuroblastoma in children <19 years of age and diagnosed between
1992 and 1994 at any of 139 United States or Canadian hospitals participating in
the Children's Cancer Group and Pediatric Oncology Group studies.
504 population controls were selected through random digit dialing and matched
(1 : 1) with cases on date of birth. Controls could not be located for 34 cases.
538 of 741 potentially eligible cases (73% participation rate).
504 of 681 potentially eligible controls (74% participation rate).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's lymphoma
Incidence.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including adoption
of JEM and quantitative exposure estimates
Self-reported exposure to any of 65 chemicals, compounds, or broad categories
was obtained from structured questionnaire. An industrial hygienist confirmed
each respondent's self-reported chemical exposure responses. Exposures were
not assigned using JEM.
TCE exposure examined in analysis as separate exposure and as one of several
chemicals in the broader category of "halogenated hydrocarbons."
I
«
§
td
00
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
td
00
I
«
§
H I
O >
HH Oq
H ^
W 3
O ^
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Telephone interview with mother and father of each case and control.
Not identified in paper.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No proxy information on maternal exposure; direct interview with mother was
obtained for 537 cases and 503 controls.
Analysis of paternal chemical exposures did not include information on paternal
exposure from proxy interviews.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies; numbers
of exposed cases and prevalence of exposure in
case-control studies
Self-reported TCE exposure: 22 cases (5% exposure prevalence) and 12 controls
(4% exposure prevalence).
IH-reviewed TCE exposure: 9 cases (2% exposure prevalence) and 7 controls
(2% exposure prevalence).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in published
paper
Documentation of results
Analyses of maternal and paternal occupational exposure each adjusted
child's age, maternal race, maternal age, and maternal education.
for
Separate analyses are conducted for maternal and paternal exposure using
logistic regression methods.
No.
Yes, results are well documented.
O
H
W
-------
1 B.3.2.2.2. Heineman et al (1994).
2 B.3.2.2.2.1. Author's abstract.
3
4 Chlorinated aliphatic hydrocarbons (CAHs) were evaluated as potential risk
5 factors for astrocytic brain tumors. Job-exposure matrices for six individual
6 CAHs and for the general class of organic solvents were applied to data from a
7 case-control study of brain cancer among white men. The matrices indicated
8 whether the CAHs were likely to have been used in each industry and occupation
9 by decade (1920-1980), and provided estimates of probably and intensity of
10 exposure for "exposed" industries and occupations. Cumulative exposure indices
11 were calculated for each subject.
12 Associations of astrocytic brain cancer were observed with likely exposure to
13 carbon tetrachloride, methylene chloride, tetrachloroethylene, and
14 trichloroethylene, but were strongest for methylene chloride. Exposure to
15 chloroform or methyl chloroform showed little indication of an association with
16 brain cancer. Risk of astrocytic brain tumors increase with probability and
17 average intensity of exposure, and with duration of employment in j obs
18 considered exposed to methylene chloride, but not with a cumulative exposure
19 score. These trends could not be explained by exposures to the other solvents.
20
21 B.3.2.2.2.2. Study description and comment. Heineman et al. (1994) studied the association
22 between astrocytic brain cancer (ICD-9 codes 191, 192, 225, and 239.7) and occupational
23 exposure to chlorinated aliphatic hydrocarbons. Cases were identified using death certificates
24 from southern Louisiana, northern New Jersey, and the Philadelphia area. This analysis was
25 limited to white males who died between 1978 and 1981. Controls were randomly selected from
26 the death certificates of white males who died of causes other than brain tumors, cerebrovascular
27 disease, epilepsy, suicide, and homicide. The controls were frequency matched to cases by age,
28 year of death, and study area.
29 Next-of-kin were successfully located for interview for 654 cases and 612 controls,
30 which represents 88 and 83% of the identified cases and controls, respectively. Interviews were
31 completed for 483 cases (74%) and 386 controls (63%). There were 300 cases of astrocytic
32 brain cancer (including astrocytoma, glioblastoma, mixed glioma with astrocytic cells). The
33 ascertainment of type of cancer was based on review of hospital records which included
34 pathology reports for 229 cases and computerized tomography reports for 71 cases. After
35 excluding 66 controls with a possible association between occupational exposure to chlorinated
36 aliphatic hydrocarbons and cause of death (some types of cancer, cirrhosis of the liver), the final
37 analytic sample consisted of 300 cases and 320 controls.
38 In the next-of-kin interviews, the work history included information about each job held
39 since the case (or control) was 15 years old (job title, description of tasks, name and location of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-184 DRAFT—DO NOT CITE OR QUOTE
-------
1 company, kinds of products, employment dates, and hours worked per week). Occupation and
2 industry were coded based on four digit Standard Industrial Classification and Standard
3 Occupational Classification (Department of Commerce) codes. The investigators developed
4 matrices linked to jobs with likely exposure to six chlorinated aliphatic hydrocarbons (carbon
5 tetrachloride, chloroform, methyl chloroform, methylene dichloride, tetrachloroethylene, and
6 trichloroethylene), and to organic solvents (Gomez et al., 1994). This assessment was done
7 blinded to case-control status. Exposure was defined as the probability of exposure to a
8 substance (the highest probability score for that substance among all jobs), duration of
9 employment in the exposed occupation and industry, specific exposure intensity categories,
10 average intensity score (the three-level semiquantitative exposure concentration assigned to each
11 job multiplied by duration of employment in the job, summed across all jobs), and cumulative
12 exposure score (weighted sum of years in all exposed jobs with weights based on the square of
13 exposure intensity [1, 2, 3] assigned to each job). Secular trends in the use of specific chemicals
14 were considered in the assignment of exposure potential. Exposures were lagged 10 or 20 years
15 to account for latency. Thus, this exposure assessment procedure was quite detailed.
16 The strengths of this case-control study include a large sample size, detailed work
17 histories including information not just about usual or most recent industry and occupation, but
18 also about tasks and products for all jobs held since age 15, and comprehensive exposure
19 assessment and analysis along several different dimensions of exposure. The major limitation
20 was the lack of direct exposure information and potential inaccuracy of the description of work
21 histories that was obtained from next-of-kin interviews. The authors acknowledge this limitation
22 in the report, and in response to a letter by Norman (1996) criticizing the methodology and
23 interpretation of the study with respect to the observed association with methylene chloride,
24 Heineman et al. (1994) noted that while the lack of direct exposure information must be
25 interpreted cautiously, it does not invalidate the results. Differential recall bias between cases
26 and controls was unlikely because work histories came from next-of-kin for both groups and, the
27 industrial hygienists made their judgments blinded to disease status. Nondifferential
28 misclassification is possible due to underreporting of job information by next of kin and would,
29 on average, attenuate true associations.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-185 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Heineman EF, Cocco P, Gomez MR, Dosemeci M, Stewart PA, Hayes KB, Zahm SH, Thomas TL, Blair A. 1994.
Occupational exposure to chlorinated aliphatic hydrocarbons and risk of astrocytic brain cancer. Am J Ind Med 26:155-169.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes, study further examines six specific solvents including trichloroethylene in a
previous study of brain cancer which reported association with electrical equipment
production and repair.
Brain cancer deaths among white males in southern Louisiana, northern New Jersey
and Philadelphia, Pennsylvania, were identified using death certificates (n = 741).
Controls were randomly selected (source not identified in paper) among other
cause-specific deaths among white male residents of these areas and matched to
cases by age, year of death and study area (n = 741).
Participation rate, 483 of 741 (65% of cases with brain cancer); 386 of 741 controls
(52%). Of the 483, 300 deaths were due to astrocytic brain cancer.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD, 9th revision, Codes 191, 192, 225, 239.7.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
The job-exposure-matrix of Gomez et al. (1994) was used to assign potential
exposure to 6 solvents including trichloroethylene.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
td
00 «
ON §
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
•3
wl
>3
CATEGORY E: INTERVIEW TYPE
<90% Face-to-Face
Blinded interviewers
Interview with next-of-kin but paper
face.
does not identify whether telephone or face-to-
Interviewer was blinded as to case and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Proxy information was obtained from 100% of cases and controls.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
TCE exposure prevalence: 128 cases
(43%) and 125 controls (39%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Stratified analysis controlled for age, year of death and study area; employment in
electronics-related occupations was included in addition in logistic regression
analyses.
Stratified analysis using 2x2 tables
and logistic regression.
Yes.
Yes.
oo
^§
"*4-
o§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.3. Colon and Rectal Cancers Case-Control Studies
2 B.3.2.3.1. Goldberg et al (2001), Simiatycki (1991).
3 B.3.2.3.1.1. Author's abstract.
4
5 BACKGROUND: We conducted a population-based case-control study in
6 Montreal, Canada, to explore associations between hundreds of occupational
7 circumstances and several cancer sites, including colon. METHODS: We
8 interviewed 497 male patients with a pathologically confirmed diagnosis of colon
9 cancer, 1514 controls with cancers at other sites, and 533 population-based
10 controls. Detailed job histories and relevant potential confounding variables were
11 obtained, and the j ob histories were translated by a team of chemists and
12 industrial hygienists into a history of occupational exposures. RESULTS: We
13 found that there was reasonable evidence of associations for men employed in
14 nine industry groups (adjusted odds ranging from 1.1 to 1.6 per a 10-year increase
15 in duration of employment), and in 12 job groups (OR varying from 1.1 to 1.7). In
16 addition, we found evidence of increased risks by increasing level of exposures to
17 21 occupational agents, including polystyrene (OR for "substantial" exposure
18 (OR(subst) =10.7), polyurethanes (OR(subst) = 8.4), coke dust (OR(subst) = 5.6),
19 mineral oils (OR(subst) = 3.3), polyacrylates (OR(subst) = 2.8), cellulose nitrate
20 (OR(subst) = 2.6), alkyds (OR(subst) = 2.5), inorganic insulation dust (OR(subst)
21 = 2.3), plastic dusts (OR(subst) = 2.3), asbestos (OR(subst) = 2.1), mineral wool
22 fibers (OR(subst) = 2.1), glass fibers (OR(subst) = 2.0), iron oxides (OR(subst) =
23 1.9), aliphatic ketones (OR(subst) = 1.9), benzene (OR(subst) =1.9), xylene
24 (OR(subst) = 1.9), inorganic acid solutions (OR(subst) = 1.8), waxes, polishes
25 (OR(subst) = 1.8), mononuclear aromatic hydrocarbons (OR(subst) = 1.6),
26 toluene (OR(subst) = 1.6), and diesel engine emissions (OR(subst) = 1.5). Not all
27 of these effects are independent because some exposures occurred
28 contemporaneously with others or because they referred to a group of substances.
29 CONCLUSIONS: We have uncovered a number of occupational associations
30 with colon cancer. For most of these agents, there are no published data to support
31 or refute our observations. As there are few accepted risk factors for colon cancer,
32 we suggest that new occupational and toxicologic studies be undertaken focusing
33 on the more prevalent substances reported herein.
34
35 B.3.2.3.1.2. Study description and comment. Goldberg et al. (2001) and Siemiatycki (1991)
36 reported data from a case-control study of occupational exposures and colon cancer conducted in
37 Montreal, Quebec (Canada) and part of a larger study of 10 other site-specific cancers and
38 occupational exposures. The investigators identified 607 newly diagnosed cases of primary
39 colon cancer (ICD9, 153), confirmed on the basis of histology reports, between 1979 and 1985;
40 497 of these participated in the study interview (81.9% participation). One control group
41 (n = 1,514) consisted of patients with other forms of cancer (excluding cancers of the lung,
42 peritoneum, esophagus, stomach, small intestine, rectum, liver and intrahepatic bile ducts,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-188 DRAFT—DO NOT CITE OR QUOTE
-------
1 gallbladder and extrahepatic bile ducts and pancreas) recruited through the same study
2 procedures and time period as the colon cancer cases. A population-based control group
3 (n = 533, 72% response), frequency matched by age strata, was drawn using electoral lists and
4 random digit dialing. Face-to-face interviews were carried out with 82% of all cancer cases with
5 telephone interview (10%) or mailed questionnaire (8%) for the remaining cases. Twenty
6 percent of all case interviews were provided by proxy respondents. The occupational assessment
7 consisted of a detailed description of each j ob held during the working lifetime, including the
8 company, products, nature of work at site, job activities, and any additional information that
9 could furnish clues about exposure from the interviews.
10 A team of industrial hygienists and chemists blinded to subject's disease status translated
11 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
12 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
13 exposure dimensions was categorized into none, any, or substantial exposure. Goldberg et al.
14 (2001) presents observations of analyses examining industries, occupation, and some
15 chemical-specific exposures, but not TCE. Observations on TCE are found in the original report
16 of Siemiatycki (1991). Any exposure to TCE was 2% among cases (n = 12) and 1% for
17 substantial TCE exposure (n = 7); "substantial" is defined as >10 years of exposure for the
18 period up to 5 years before diagnosis.
19 Logistic regression models adjusted for a number of nonoccupational variables including
20 age, ethnicity, birthplace, education, income, parent's occupation, smoking, alcohol
21 consumption, tea consumption, respondent status, heating source and cooking source in
22 childhood home, consumption of nonpublic water supply, and body mass index (Goldberg et al.,
23 2001) or Mantel-Haenszel £ stratified on age, family income, cigarette smoking, coffee, ethnic
24 origin, and beer consumption (Siemiatycki, 1991). Odds ratios for TCE exposure are presented
25 in Siemiatycki (1991) with 90% confidence intervals.
26 The strengths of this study were the large number of incident cases, specific information
27 about job duties for all jobs held, and a definitive diagnosis of colon cancer. However, the use of
28 the general population (rather than a known cohort of exposed workers) reduced the likelihood
29 that subjects were exposed to TCE, resulting in relatively low statistical power for the analysis.
30 The job exposure matrix, applied to the job information, was very broad since it was used to
31 evaluate 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-189 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Goldberg MS, Parent M-E, Siemiatycki J, Desy M, Nadon L, Richardson L, Lakhani R, Lateille B, Valois M-F. 2001. A case-
control study of the relationship between the risk of colon cancer in men and exposure to occupational agents. Am J Ind Med
39:5310-546.
Siemiatycki J. 1991. Risk Factors for Cancer in the Workplace. Baca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort studies of
exposure and control groups and of cases and
controls in case-control studies is adequate
This population case-control study was designed to generate hypotheses on
possible association between 1 1 site-specific cancers and occupational title or
chemical exposures.
607 colon cancer cases were identified among male Montreal residents between
1979 and 1985 of which 497 were interviewed.
740 eligible male controls identified from the same source population using
random digit dialing or electoral lists; 533 were interviewed. A second control
series consisted of all other cancer controls excluding lung peritoneum and other
digestive cancers.
Participation rate: cases, 81.9%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's lymphoma
Incidence.
ICD-9, 153 (Malignant neoplasm of colon).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job
history with supplemental questionnaire for jobs of a priori interest (e.g.,
machinists, painters). Team of chemist and industrial hygienist assigned exposure
using job title with a semi quantitative scale developed for 294 exposures, including
TCE. For each exposure, a 3 -level ranking was used for concentration (low or
background, medium, high) and frequency (percent of working time: low, 1 to 5%;
medium, >5 to 30%; and high, >30%).
•3
1
§
i
td
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
•3
td g
^ S
^§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were
conducted either at home or in the hospital; all population control interviews were
conducted at home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
497 cases (81.9% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 2% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), 1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in published
paper
Documentation of results
Age, ethnicity, birthplace, education, income, parent's occupation, smoking,
alcohol consumption, tea consumption, respondent status, heating source and
cooking source in childhood home, consumption of nonpublic water supply, and
body mass index (Goldberg et al., 2001).
Age, family income, cigarette smoking, coffee, ethnic origin, and beer consumption
(Siemiatycki, 1991).
Mantel-Haenszel (Siemiatycki, 1991).
Logistic regression (Goldberg et al., 2001).
No.
Yes.
-------
1 B.3.2.3.2. Dumas et al (2000), Siemiatycki (1991).
2 B.3.2.3.2.1. Author's abstract.
3
4 In 1979, a hypothesis-generating, population-based case-control study was
5 undertaken in Montreal, Canada, to explore the association between occupational
6 exposure to 294 substances, 130 occupations and industries, and various cancers.
7 Interviews were carried out with 3,630 histologically confirmed cancer cases, of
8 whom 257 had rectal cancer, and with 533 population controls, to obtain detailed
9 job history and data on potential confounders. The job history of each subject was
10 evaluated by a team of chemists and hygienists and translated into occupational
11 exposures. Logistic regression analyses adjusted for age, education, cigarette
12 smoking, beer consumption, body mass index, and respondent status were
13 performed using population controls and cancer controls, e.g., 1,295 subjects with
14 cancers at sites other than the rectum, lung, colon, rectosigmoid junction, small
15 intestine, and peritoneum. We present here the results based on cancer controls.
16 The following substances showed some association with rectal cancer: rubber
17 dust, rubber pyrolysis products, cotton dust, wool fibers, rayon fibers, a group of
18 solvents (carbon tetrachloride, methylene chloride, trichloroethylene, acetone,
19 aliphatic ketones, aliphatic esters, toluene, styrene), polychloroprene, glass fibers,
20 formaldehyde, extenders, and ionizing radiation. The independent effect of many
21 of these substances could not be disentangled as many were highly correlated with
22 each other.
23
24 B.3.2.3.2.2. Study description and comment. Dumas et al. (2000) and Siemiatycki (1991)
25 reported data from a case-control study of occupational exposures and rectal cancer conducted in
26 Montreal, Quebec (Canada) and part of a larger study of 10 other site-specific cancers and
27 occupational exposures. The investigators identified 304 newly diagnosed cases of primary
28 rectal cancers, confirmed on the basis of histology reports, between 1979 and 1985; 257 of these
29 participated in the study interview (84.5% response). One control group (n = 1,295) consisted of
30 patients with other forms of cancer (excluding lung cancer and other intestinal cancers) recruited
31 through the same study procedures and time period as the rectal cancer cases. A population-
32 based control group (n = 533), frequency matched by age strata, was drawn using electoral lists
33 and random digit dialing (72% response). The occupational assessment consisted of a detailed
34 description of each job held during the working lifetime, including the company, products, nature
35 of work at site, job activities, and any additional information that could furnish clues about
36 exposure from the interviews. The percentage of proxy respondents was 15.2% for cases, 19.7%
37 for other cancer controls, and 12.6% for the population controls.
38 A team of industrial hygienists and chemists blinded to subject's disease status translated
39 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
40 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-192 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure dimensions was categorized into none, any, or substantial exposure. Any exposure to
2 TCE was 5% among cases (n = 12) and 1% for substantial TCE exposure (n = 3); "substantial" is
3 defined as >10 years of exposure for the period up to 5 years before diagnosis.
4 Logistic regression models adjusted for age, education, respondent status, cigarette
5 smoking, beer consumption and body mass index (Dumas et al., 2000) or Mantel-Haenszel %2
6 stratified on age, family income, cigarette smoking, coffee, ethnic origin, and beer consumption
7 (Siemiatycki, 1991). Dumas et al. (2000) presents observations of analyses examining
8 industries, occupation, and some chemical-specific exposures, including TCE. Observations on
9 TCE from Mantel-Haenszel analyses are found in the original report of Siemiatycki (1991).
10 Odds ratios for TCE exposure are presented in Siemiatycki (1991) with 90% confidence intervals
11 and 95% confidence intervals in Dumas et al. (2000).
12 The strengths of this study were the large number of incident cases, specific information
13 about job duties for all jobs held, and a definitive diagnosis of rectal cancer. However, the use of
14 the general population (rather than a known cohort of exposed workers) reduced the likelihood
15 that subjects were exposed to TCE, resulting in relatively low statistical power for the analysis.
16 The job exposure matrix, applied to the job information, was very broad since it was used to
17 evaluate 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-193 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
Dumas S, Parent M-E, Siemiatycki J, Brisson J. 2000. Rectal cancer and occupational risk factors: a hypothesis-generating,
exposure-based case-control study. Int J Cancer 87:874-879.
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. Boca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
304 rectal cancer cases were identified among male Montreal residents between 1979
and 1985 of which 294 were interviewed.
740 eligible male controls identified from the same source population using random
digit dialing or electoral lists; 533 were interviewed. A second control series
consisted of all other cancer controls excluding lung and other intestinal cancer cases.
Participation rate: cases, 84.5%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 154 (Malignant neoplasm of rectum, rectosigmoid junction and anus).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semiquantitative scale developed for 294 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
•3
1
§
i
td
^§
"*4-
o§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
td
I
H I
O >
HH Oq
H ^
O
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face to face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were conducted
either at home or in the hospital; all population control interviews were conducted at
home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
294 cases (78% response), 533 population controls (72% response).
Exposure prevalence: Any TCE exposure, 5% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), 1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, education, respondent status, cigarette smoking, beer consumption and body
mass index (Dumas et al., 2000).
Age, family income, cigarette smoking, coffee, ethnic origin, and beer consumption
(Siemiatycki, 1991).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Dumas et al., 2000).
No.
Yes.
H
W
-------
1 B.3.2.3.3. Fredriksson et al (1989).
2 B.3.2.3.3.1. Author's abstract.
O
4 A case-control study on colon cancer was conducted encompassing 329 cases and
5 658 controls. Occupations and various exposures were assessed by questionnaires.
6 A decreased risk was found in persons with physically active occupations. This
7 effect was most pronounced in colon descendens and sigmoideum with an odds
8 ratio (OR) of 0.49 whereas no reduced risk was found for right-sided colon
9 cancer. Regarding specific jobs, reduced ORs were found for agricultural,
10 forestry, and saw mill workers and increased OR for railway employees. High-
11 grade exposure to asbestos or to organic solvents gave a two-fold increased risk.
12 Regarding exposure to trichloroethylene in general, a slightly increased risk was
13 found whereas such exposure among dry cleaners gave a 7-fold increase of the
14 risk.
15
16 B.3.2.3.3.2. Study description and comment. Fredriksson et al. (1989) reported data from a
17 population case-control study of occupational and nonoccupational exposures and rectal cancer
18 conducted in Urea, Sweden. The investigators identified 329 diagnosed cases of rectal cancers
19 (ICD 8, 153), between 1980 and 1983, confirmed on the basis of histology reports and alive at
20 the time of data collect between 1984 and 1986; 302 (165 males and 165 females) of these
21 participated in the study interview (92% response). A population-based control group (n = 658),
22 matched by a 1:2 ratio to cases on age sex and county residence, was drawn using the Swedish
23 National Population Register list; 623 (306 males and 317 females) returned mailed
24 questionnaires and participated in the study (95% response).
25 The occupational assessment consisted of a detailed description of each job held during
26 the working lifetime, including details on specific occupations and exposures. Occupation
27 information was provided directly from each case and control given the study's eligibility
28 requirement of being alive at the time of data collection. A team of experts independently
29 classified three exposures of interest (asbestos, organic solvents, and impregnating agents) into
30 two categories, low grade exposure and high grade exposure and other chemical-specific
31 exposures, including TCE, as either "exposed" or "unexposed." Fredriksson et al. (1989) do not
32 define these categories nor do they provide information on exposure potential, frequency of
33 exposure, or concentration of exposure. No information is provided whether experts were
34 blinded as to disease status.
35 Statistical analysis examining occupation and agent-specific exposures was carried out
36 using Mantel-Haenszel ^ stratified on age, sex, and an index of physical activity. Odds ratios
37 associated with specific chemical exposure are presented with their 95% confidence intervals.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-196 DRAFT—DO NOT CITE OR QUOTE
-------
1 The strengths of this study were its specific information about job duties for all jobs held
2 and a definitive diagnosis of rectal cancer. However, the study's assignment of exposure
3 potential from information using mailed questionnaires is considered inferior to information
4 obtained directly from trained interviewers and expert assessment because of greater uncertainty
5 and misclassification (Fritschi et al., 1996). The degree of potential exposure misclassification
6 bias in this population case-control study of colon cancer is not known. Furthermore, exposure
7 prevalence to TCE appears low, as judged by the wide confidence interval around the odds ratio.
8 This study is considered as having decreased sensitivity for examining colon cancer and TCE
9 given the apparent lower exposure prevalence and likely exposure misclassification bias
10 associated with mailed questionnaire information.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-197 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
Fredriksson M, Bengtsson N-O, Hardell L, Axelson O. 1989. Colon cancer, physical activity, and occupational exposure. A
case-control study. Cancer 63:1838-1842.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Abstract — to evaluate occupational and nonoccupational exposures as risk factors for
colon cancer.
302 (165 males and 165 females) cases participated in study out of 329 eligible cases
reported to the Swedish Cancer Registry between 1980 and 1983, among resident of
Umea, Sweden, alive at time of data collection 1984 and 1986, and with
histological-confirmed diagnosis of colon cancer.
623 (306 males and 317 females) identified from Swedish Population Registry and
matched for age, sex, and county of residence.
Participation rate: cases, 92%; population controls, 95%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-8, 153 (Malignant neoplasm of large intestine, except rectum).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Self-reported information on occupational exposure as obtained from a mailed
questionnaire to study participants. Questionnaire sought information on complete
working history, other exposures, and dietary habits. Procedure for assigning
chemical exposures from job title information not described in paper.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
•3
I
td
oo
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Mailed questionnaire.
No information in published paper.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No proxy respondents, all cases and controls alive at time of data collection.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
302 cases (92% response), 623 population controls (95% response).
Exposure prevalence not calculated, published paper lacks number of TCE
cases and controls.
exposed
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Yes, age, sex, and index of physical activity.
Mantel -Haenszel.
No.
Yes.
I
td
^§
H I
O
H
W
-------
1 B.3.2.4. Esophageal Cancer Case-Control Studies
2 B.3.2.4.1. Parent et al (2000a), Siemiatycki (1991).
3 B.3.2.4.1.1. Parent et al (2000a) abstract.
4
5 OBJECTIVES: To describe the relation between oesophageal cancer and many
6 occupational circumstances with data from a population based case-control study.
7 METHODS: Cases were 99 histologically confirmed incident cases of cancer of
8 the oesophagus, 63 of which were squamous cell carcinomas. Various control
9 groups were available; for the present analysis a group was used that comprised
10 533 population controls and 533 patients with other types of cancer. Detailed job
11 histories were elicited from all subjects and were translated by a team of chemists
12 and hygienists for evidence of exposure to 294 occupational agents. Based on
13 preliminary results and a review of literature, a set of 35 occupational agents and
14 19 occupations and industry titles were selected for this analysis. Logistic
15 regression analyses were adjusted for age, birthplace, education, respondent (self
16 or proxy), smoking, alcohol, and beta-carotene intake. RESULTS: Sulphuric acid
17 and carbon black showed the strongest evidence of an association with
18 oesophageal cancer, particularly squamous cell carcinoma. Other substances
19 showed excess risks, but the evidence was more equivocal-namely chrysotile
20 asbestos, alumina, mineral spirits, toluene, synthetic adhesives, other paints and
21 varnishes, iron compounds, and mild steel dust. There was considerable overlap
22 in occupational exposure patterns and results for some of these substances may be
23 mutually confounded. None of the occupations or industry titles showed a clear
24 excess risk; the strongest hints were for warehouse workers, food services
25 workers, and workers from the miscellaneous food industry. CONCLUSIONS:
26 The data provide some support for an association between oesophageal cancer
27 and a handful of occupational exposures, particularly sulphuric acid and carbon
28 black. Many of the associations found have never been examined before and
29 warrant further investigation.
30
31 B.3.2.4.1.2. Study description and comment. Parent et al. (2000a) and Siemiatvcki (1991)
32 reported data from a case-control study of occupational exposures and esophageal cancer
33 conducted in Montreal, Quebec (Canada) and part of a larger study of 10 other site-specific
34 cancers and occupational exposures. The investigators identified 129 newly diagnosed cases of
35 primary esophageal cancers, confirmed on the basis of histology reports, between 1979 and
36 1985; 99 of these participated in the study interview (76.7% response). One control group
37 consisted of patients with other forms of cancer recruited through the same study procedures and
38 time period as the esophageal cancer cases. A population-based control group (n = 533),
39 frequency matched by age strata, was drawn using electoral lists and random digit dialing (72%
40 response). Face-to-face interviews were carried out with 82% of all cancer cases with telephone
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-200 DRAFT—DO NOT CITE OR QUOTE
-------
1 interview (10%) or mailed questionnaire (8%) for the remaining cases. Twenty percent of all
2 case interviews were provided by proxy respondents.
3 The occupational assessment consisted of a detailed description of each job held during
4 the working lifetime, including the company, products, nature of work at site, job activities, and
5 any additional information that could furnish clues about exposure from the interviews. A team
6 of industrial hygienists and chemists blinded to subject's disease status translated jobs into
7 potential exposure to 294 substances with three dimensions (degree of confidence that exposure
8 occurred, frequency of exposure, and concentration of exposure). Each of these exposure
9 dimensions was categorized into none, any, or substantial exposure. Any exposure to TCE was
10 1% among cases (n = 1) and 1% for substantial TCE exposure (n = 1); "substantial" is defined as
11 >10 years of exposure for the period up to 5 years before diagnosis.
12 Logistic regression models adjusted for age, education, respondent status, birthplace,
13 cigarette smoking, beer consumption spirits consumption and beta-carotene intake (Parent et al.,
14 2000a) or Mantel-Haenszel ^ stratified on age, family income, cigarette smoking, coffee, and an
15 index for alcohol consumption (Siemiatycki, 1991). Parent et al. (2000a) presents observations
16 of analyses examining industries, occupation, and some chemical-specific exposures, including
17 solvents, but not TCE. Observations on TCE from Mantel-Haenszel analyses are found in the
18 original report of Siemiatycki (1991). Odds ratios for TCE exposure are presented in
19 Siemiatycki (1991) with 90% confidence intervals and 95% confidence intervals in Parent et al.
20 (2000a).
21 The strengths of this study were the large number of incident cases, specific information
22 about job duties for all jobs held, and a definitive diagnosis of esophageal cancer. However, the
23 use of the general population (rather than a known cohort of exposed workers) reduced the
24 likelihood that subjects were exposed to TCE, resulting in relatively low statistical power for the
25 analysis. The job exposure matrix, applied to the job information, was very broad since it was
26 used to evaluate 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-201 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Parent M-E, Siemiatycki J, Fritschi L. 2000a. Workplace exposures and oesophageal cancer. Occup Environ Med
57:325-334.
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. Boca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
129 esophageal cancer cases were identified among male Montreal residents between
1979 and 1985 of which 99 were interviewed.
740 eligible male controls identified from the same source population using random
digit dialing or electoral lists; 533 were interviewed. A second control series
consisted of all other cancer controls.
Participation rate: cases, 76.7%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 150 (Malignant neoplasm of esophagus).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semiquantitative scale developed for 294 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
td
to
H I
O >
HH Oq
H ^
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
O
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were conducted
either at home or in the hospital; all population control interviews were conducted at
home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
99 cases (76.7% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 1% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), 1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, education, respondent status, birthplace, cigarette smoking, beer consumption
spirits consumption and beta-carotene intake (Parent et al., 2000a).
Age, family income, cigarette smoking, and index for alcohol consumption
(Siemiatycki, 1991).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Parent et al., 2000a).
No.
Yes.
H
W
-------
1 B.3.2.5. Liver Cancer Case-Control Studies
2 B.3.2.5.1. Lee et al (2003).
3 B.3.2.5.1.1. Author's abstract.
4
5 Aims: To investigate the association between cancer mortality risk and exposure
6 to chlorinated hydrocarbons in groundwater of a downstream community near a
7 contaminated site. Methods: Death certificates inclusive for the years 1966-97
8 were collected from two villages in the vicinity of an electronics factory operated
9 between 1970 and 1992. These two villages were classified into the downstream
10 (exposed) village and the upstream (unexposed) according to groundwater flow
11 direction. Exposure classification was validated by the contaminant levels in 49
12 residential wells measured with gas chromatography/mass spectrometry.
13 Mortality odds ratios (MORs) for cancer were calculated with cardiovascular-
14 cerebrovascular diseases as the reference diseases. Multiple logistic regressions
15 were performed to estimate the effects of exposure and period after adjustment for
16 age. Results: Increased MORs were observed among males for all cancer, and
17 liver cancer for the periods after 10 years of latency, namely, 1980-89, and 1990-
18 97. Adjusted MOR for male liver cancer was 2.57 (95% confidence interval 1.21
19 to 5.46) with a significant linear trend for the period effect. Conclusion: The
20 results suggest a link between exposure to chlorinated hydrocarbons and male
21 liver cancer risk. However, the conclusion is limited by lack of individual
22 information on groundwater exposure and potential confounding factors.
23
24 B.3.2.5.1.2. Study description and comment. Exposure potential to chlorinated hydrocarbons
25 was assigned in this community case-control study of liver cancer in males >30 years of age
26 using residency as coded on death certificates obtained from local household registration offices.
27 No information is available to assess the completeness of death reporting to the local registration
28 office. Of the 1,333 deaths between 1966 and 1997 in two villages surrounding a hazardous
29 waste site, an electronics factory operating between 1970 and 1992 in Taoyuan, Taiwan,1
30 266 cancer deaths were identified; 53 liver cancer deaths, 39 stomach cancer deaths,
31 26 colorectal deaths, and 41 lung cancer deaths. Controls were identified from 344 deaths due to
32 cardiovascular and cerebrovascular diseases, without arrhythmia; 286 were included in the
33 statistical analysis. Residents from a village north and northeast of the plant were considered
34 exposed and residents living south considered unexposed to chlorinated hydrocarbons.
35 Statistical analyses are limited to Mantel-Haenszel chi-square approaches stratified by sex and
36 age and, for male cases and controls, logistic regression with age as a covariate. Socioeconomic
37 characteristics were similar between residents of the two villages (Wang, 2004). The study does
1 The factory's workers were subjects in the cohort studies of Chang et al. (2003, 2005) and Sung et al. (2007,
2008).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-204 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
not include control for potential confounding from hepatitis virus; high rates of hepatitis B and C
are endemic to Taiwan and northern Taiwan, the location of this study, has a high prevalence of
hepatitis C virus infection (Lee et al., 2003). Confounding would be introduced if the prevalence
of hepatitis C differed between the two villages.
Exposure assessment is quite limited and misclassification bias likely high using
residence address as recorded on the death certificate as a surrogate for consumption of
contaminated drinking water. The paper not only lacks information on intensity and duration of
hydrocarbon exposures to individual cases and controls, but no information is available on an
estimate of the amount of TCE ingested. Information on residence length, population mobility,
and chemical usage at the plant are lacking. Similarly, well water monitoring is sparse, based on
seven chlorinated hydrocarbons monitored over a 7 month period between 1999-2000 in
69 groundwater samples from 44 wells to the north and northeast, or downstream from the
factory, and in 5 groundwater samples from 2 wells to the south or upstream from the factory.
Monitoring from other time periods is lacking with no information available to judge if current
monitoring are representative of past concentrations. Median concentrations (ug/L or ppb) and
ranges (ug/L or ppb) for these seven chemicals are below. Highest concentration of
contaminants was from wells closest to the factory boundary with concentrations detected at or
close to maximum contaminant levels in wells located 0.5 mile (1,000 meters) away. A
municipal system supplied water to upstream village residents (start date no identified); however,
wells served as source for water to of the north or downstream village residents. The exposure
assessment does not consider potential occupational exposure.
Chemical
Tri chl oroethy lene
Perchl oroethy 1 ene
cis- 1 ,2-dichloroethylene
1, 1-dichloroethane
1, 1-dichloroethylene
Vinyl chloride
Downstream
Median
28
3
O
2
1
0.003
Range
N.D.-1,791
N.D.-5,228
N.D.-1,376
N.D.-228
N.D.-1,240
N.D.-72
Upstream
Median
0.1
0.05
N.D.
0.05
N.D.
N.D.
Range
0.1-0.1
N.D.-O.l
N.D.
N.D.-O.l
N.D.
N.D.
23
24
N.D. = not detected
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-205 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
to
-
Lee L J-H, Chung C-W, Ma Y-C, Wang G-S, Chen P-C, Hwang Y-H, Wang J-D. 2003. Increased mortality odds ratio of
male liver cancer in a community contaminated by chlorinated hydrocarbons in groundwater. Occup Environ Med
60:364-369.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Study hypothesis of investigating cancer mortality risk and exposure to chlorinated
hydrocarbons in groundwater.
Deaths in 1966-1997 identified from local housing registration offices among
residents in two villages were the source for case and control series. The two villages
were north (contaminated community) and south (unexposed) of an electronics
factory declared as a hazardous waste site. No information if all death among
residents were reported to registration office.
Cases: 53 liver cancer deaths in males and females, 51 included in statistical analysis
(96%); stomach cancer deaths (n = 39), colon and rectum deaths (n = 26), and lung
cancer deaths (n = 41). Paper does not present numbers of stomach, colo-rectal and
lung cancer deaths used in statistical analyses.
Controls: 344 cardiovascular-cerebrovascular CV-CB disease deaths, 286 CV-CB
deaths without arrhythmia included in statistical analysis (83%).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD-9.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Exposure potential to chlorinated hydrocarbons in drinking water was inferred from
residence address on deaths certificate.
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
NA, Record based information.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
NA
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Liver cancer case exposure prevalence [downstream village resident], 53% (n =
males, n = 4 females).
Control exposure prevalence [upstream village resident], 30% (n = 44 males, n =
females).
24
= 41
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Sex and age (categorical). No control for potential confounding due to hepatitis
(for liver cancer) or smoking (for lung cancer analyses).
virus
Mantel-Haenszel Chi square.
Multiple logistic regressions (males deaths only).
No, MORs presented by time period.
Inadequate, the paper does not discuss mobility patterns of residents, percentage of
population who may have moved from area, pr completeness of death ascertainment
using certificates obtained from local housing registration offices.
td
to
o
HH Oq
H ^
MOR = mortality odds ratio.
O
H
W
-------
1 B.3.2.6. Lymphoma Case-Control Studies
2 B.3.2.6.1. Wang et al (2008).
3 B.3.2.6.1.1. Author's abstract.
4
5 A population-based case-control study involving 601 incident cases of non-
6 Hodgkin lymphoma (NHL) and 717 controls was conducted in 1996-2000 among
7 Connecticut women to examine associations with exposure to organic solvents. A
8 job-exposure matrix was used to assess occupational exposures. Increased risk of
9 NHL was associated with occupational exposure to chlorinated solvents (odds
10 ratio (OR) = 1.4, 95% confidence interval (CI): 1.1, 1.8) and carbon tetrachloride
11 (OR = 2.3, 95% CI: 1.3, 4.0). Those ever exposed to any organic solvent in work
12 settings had a borderline increased risk of NHL (OR =1.3, 95% CI: 1.0, 1.6);
13 moreover, a significantly increased risk was observed for those with average
14 probability of exposure to any organic solvent at medium-high level (OR =1.5,
15 95% CI: 1.1, 1.9). A borderline increased risk was also found for ever exposure to
16 formaldehyde (OR = 1.3, 95% CI: 1.0, 1.7) in work settings. Risk of NHL
17 increased with increasing average intensity (P = 0.01), average probability (p<
18 0.01), cumulative intensity (P = 0.01), and cumulative probability (p < 0.01) level
19 of organic solvent and with average probability level (P = 0.02) and cumulative
20 intensity level of chlorinated solvent (P = 0.02). Analyses by NHL subtype
21 showed a risk pattern for diffuse large B-cell lymphoma similar to that for overall
22 NHL, with stronger evidence of an association with benzene exposure. Results
23 suggest an increased risk of NHL associated with occupational exposure to
24 organic solvents for women.
25
26 B.3.2.6.1.2. Study description and comment. This population case-control study of
27 non-Hodgkin's lymphoma in Connecticut women was designed to examine possible personal
28 and occupational risk factors for NHL. The publication of Wang et al. (2008) examined solvent
29 exposure and adopted a job-exposure matrix to assign exposure potential to nine chemicals—
30 benzene, formaldehyde, chlorinated solvents, chloroform, carbon tetrachloride, dichloromethane,
31 methyl chloride and trichloroethylene. Histologically-confirmed incident cases of NHL in
32 women aged between 21 and 84 years of age and diagnosed in Connecticut between 1996 and
33 2000 were identified from the Connecticut Cancer Registry, a SEER reporting site, with
34 population controls having Connecticut address identified from random digit dialing for women
35 <65 years of age, or by random selection from Centers for Medicare and Medicaid Service files
36 for women aged 65 year or older. Controls were frequency matched to cases within 5-year age
37 groups. Face-to-face interviews were completed for 601 (72%) cases and 717 controls (69% of
38 those identified from random digit dialing and 47% identified using Health Care Financing
39 Administration files).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-208 DRAFT—DO NOT CITE OR QUOTE
-------
1 Trained interviewers administered a structured questionnaire through in-person
2 interviews with cases and controls to collect information on diet, nutrition, and alcohol intake;
3 reproductive factors; hair dye use; and lifetime occupational history of all jobs held >1 year.
4 Jobs were coded to standardized occupational classification and standardized industry
5 classification titles and assigned probability and intensity of exposure to formaldehyde and nine
6 other solvents (benzene, any chlorinated solvents, dichloroethylene, chloroform, methylene
7 chloride, dichloroethane, methyl chloride, TCE and carbon tetrachloride) using a job-exposure
8 matrix developed by the National Cancer Institutes (Gomez et al., 1994; Dosemeci et al., 1994).
9 All jobs held up to a year before cancer diagnosis were assigned blinded as to disease status
10 potential exposure to each exposure of interest. Lifetime exposure potential for cases and
11 controls was based on exposure duration and a weighted score for exposure intensity and
12 probability of each occupational and industry and defined as a cumulative exposure metric,
13 average metric, or ever/never metric. Of the 601 cases, 77 (13%) were assigned with potential
14 TCE exposure over their lifetime; eight cases were assigned potential for high intensity exposure,
15 but with low probability and the 31 cases identified with medium and high probability of
16 exposure were considered as having low intensity exposure potential. The low exposure
17 prevalence to TCE, overall, and few subjects identified with confidence with high TCE exposure
18 intensity or probability implies exposure misclassification bias is likely, and likely
19 nondifferential, notably for high exposure categories (Dosemeci et al., 1990).
20 Association between NHL and individual occupational solvent exposure was assessed
21 using unconditional logistic regression model which adjusted for age, family history of
22 hematopoietic cancer, alcohol consumption and race. Statistical analyses treated exposure
23 defined as a categorical variable, divided into tertiles based on the distribution of controls, in
24 logistic regression analyses and as a continuous variable, whenever possible, to test for linear
25 trend. Polytomous logistic regress was used to evaluate the association between histologic
26 subtypes of NHL (DLBCL, follicular lymphoma, or chronic lymphocytic leukemia/small
27 lymphocytic lymphoma) and exposure. The largest number of cases was of the cell type
28 DLBCL.
29 Strength of this study is assignment of TCE exposure potential to individual subjects
30 using a validated job-exposure matrix, although uncertainty accompanied exposure assignment
31 and TCE exposure was largely of low intensity/low probability, and no cases with medium to
32 high intensity/probability. Resultant misclassification bias would dampen observed associations
33 for high exposure potential categories. Low prevalence of high intensity TCE exposure would
34 reduce the study's statistical power.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-209 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Wang R, Zhang Y, Lan Q, Holford TR, Leaderer B, Zahm SH, Boyle P, Dosemeci M, Rothman N, Zhu Y, Qin Q, Zheng T.
2009. Occupational exposure to solvents and risk of non-Hodgkin lymphoma in Connecticut women.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study evaluated multiple potential risk factors of NHL in a population-based
case-control study of Connecticut women. Occupational exposure to TCE was not an
a priori hypothesis.
601 (832 eligible) cases of NHL, diagnosed between 1996 and 2000 among women,
age 20 to 84 yrs and residents of Connecticut and histologically-confirmed, were
identified from the Yale Comprehensive Cancer Center's Rapid Case Ascertainment
Shared Resource, a component of the Connecticut Tumor Registry; 717 (number of
eligible controls not identified) population controls were randomly identified using
random digit dialing, if age <65 yrs, or from Medicare and Medicaid Service files,
for women aged 65 yrs or older and stratified by sex and 5-yr age groups.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
NHL and chronic lymphatic leukemia incidence.
ICD-O-2 [Codes, M-9590-9642, 9690-9701, 9740-9750].
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
All jobs held for >1 yr were assigned to standardized occupation and industry
classifications. Using job exposure matrix of NCI (Gomez et al., 1994; Dosemeci et
al., 1994), probability of exposure level (low, medium and high) and intensity (very
low, low, medium and high) to TCE and other solvents (benzene, any chlorinated
solvents, dichloroethylene, chloroform, methylene chloride, dichloroethane, methyl
chloride, carbon tetrachloride, and formaldehyde) was assigned blinded as to case or
control status.
to
Dl
Co
o §
H I
O >
HH Oq
H TO
O
H
W
-------
to
VO Lo'
I
I
§
***.
£3'
1
TO'
to
Dl
Co
o §
H I
O >
HH Oq
H TO
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Face-to-face interview with questionnaire for detailed information about medical
history, lifestyle factors, education, lifetime occupational history (all jobs held >1 yr).
Unblinded interviews.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
None.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
601 cases (72% participation) and 717 controls (69% participation for random digit
dialing controls and 47% participation for HCFA controls).
Exposure prevalence, ever exposed to TCE, 77 (13%) NHL cases; medium to high
TCE intensity, 13 NHL cases (2%); medium to high TCE probability, 34 cases (6%).
All 34 cases with medium to high TCE probability assigned low intensity exposure.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, family history of hematopoietic cancer, alcohol consumption and race.
Unconditional logistic regression.
Yes, by exposure intensity and by exposure probability.
Yes.
O
H
W
-------
1 B.3.2.6.2. Costantini et al (2008), Miligi et al (2006).
2 B.3.2.6.2.1. Costantini et al (2008) abstract.
O
4 Background While there is a general consensus about the ability of benzene to
5 induce acute myeloid leukemia (AML), its effects on chronic lymphoid leukemia
6 and multiple myeloma (MM) are still under debate. We conducted a population-
7 based case-control study to evaluate the association between exposure to organic
8 solvents and risk of myeloid and lymphoid leukemia and MM.
9 Methods Five hundred eighty-six cases of leukemia (and 1,278 population
10 controls), 263 cases of MM (and 1,100 population controls) were collected.
11 Experts assessed exposure at individual level to a range of chemicals.
12 Results We found no association between exposure to any solvent and AML.
13 There were elevated point estimates for the associations between medium/high
14 benzene exposure and chronic lymphatic leukemia (OR: 1.8, 95% CF/40.9-3.9)
15 and MM (OR: 1.9, 95% CI: 0.9-3.9). Risks of chronic lymphatic leukemia were
16 somewhat elevated, albeit with wide confidence intervals, from medium/high
17 exposure to xylene and toluene as well.
18 Conclusions We did not confirm the known association between benzene and
19 AML, though this is likely explained by the strict regulation of benzene in Italy
20 nearly three decades prior to study initiation. Our results support the association
21 between benzene, xylene, and toluene and chronic lymphatic leukemia and
22 between benzene and MM with longer latencies than have been observed for
23 AML in other studies.
24
25 B.3.2.6.2.2. Milisi et al. (2006) abstract.
26
27 BACKGROUND: A number of studies have shown possible associations between
28 occupational exposures, particularly solvents, and lymphomas. The present
29 investigation aimed to evaluate the association between exposure to solvents and
30 lymphomas (Hodgkin and non-Hodgkin) in a large population-based, multicenter,
31 case-control study in Italy. METHODS: All newly diagnosed cases of malignant
32 lymphoma in men and women age 20 to 74 years in 1991-1993 were identified in
33 8 areas in Italy. The control group was formed by a random sample of the general
34 population in the areas under study stratified by sex and 5-year age groups. We
35 interviewed 1428 non-Hodgkin lymphoma cases, 304 Hodgkin disease cases, and
36 1530 controls. Experts examined the questionnaire data and assessed a level of
37 probability and intensity of exposure to a range of chemicals. RESULTS: Those
38 in the medium/high level of exposure had an increased risk of non-Hodgkin
39 lymphoma with exposure to toluene (odds ratio =1.8; 95% confidence interval =
40 1.1-2.8), xylene 1.7 (1.0-2.6), and benzene 1.6 (1.0-2.4). Subjects exposed to all 3
41 aromatic hydrocarbons (benzene, toluene, and xylene; medium/high intensity
42 compared with none) had an odds ratio of 2.1 (1.1-4.3). We observed an increased
43 risk for Hodgkin disease for those exposed to technical solvents (2.7; 1.2-6.5) and
44 aliphatic solvents (2.7; 1.2-5.7). CONCLUSION: This study suggests that
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-212 DRAFT—DO NOT CITE OR QUOTE
-------
1 aromatic and chlorinated hydrocarbons are a risk factor for non-Hodgkin
2 lymphomas, and provides preliminary evidence for an association between
3 solvents and Hodgkin disease.
4
5 B.3.2.6.2.3. Study description and comment. This series of papers of a population
6 case-control study of lymphomas in 11 areas in Italy (Costantini et al., 2001) and occupation
7 examines author's assigned exposure to TCE and other solvents using job-specific or
8 industry-specific questionnaires and expert rating to cases and controls. Miligi et al. (2006)
9 reported findings for non-Hodgkin lymphoma, a category which included chronic lymphocytic
10 leukemia, NHL subtypes, and Hodgkin lymphoma in 8 regions and Constantini et al. (2008)
11 presented observations for specific leukemia subtypes and multiple myeloma in 7 regions
12 (8 regions for chronic lymphocytic leukemia). Exclusion of the regions in the original study
13 does not appear to greatly reduce study power or to introduce a selection bias. For example,
14 Miligi et al. (2006) included 1,428 of the 1,450 total NHL cases, the largest percentage of all
15 lymphoma subtypes. The number of other lymphoma subtypes was much smaller compared to
16 NHL; 304 cases of Hodgkin disease, 586 cases of leukemia, and 263 cases of multiple myeloma.
17 All cases were identified from participating study centers and controls were randomly selected
18 from the each area's population using stratified sampling for sex and age.
19 A face-to-face unblinded interview was conducted primarily at the interviewee's home
20 with a high proportion of proxy responses among cases (19%) but not controls (5%). Bias is
21 likely introduced by the lack of blinding of interviewers and from the high proportion of proxy
22 interviews. A questionnaire was used to obtain information on medical history, lifestyle factors,
23 occupational exposure and nonoccupational solvent exposures. Industrial hygiene professionals
24 assessed the probability and intensity of exposure to individual and classes of solvents using
25 information provided by questionnaire. Probability was classified into 3 levels (low, medium,
26 and high) with a 4-category scale for intensity (very low, low, medium, and high). These
27 qualitative scales lacked information on exposure concentrations and likely introduces
28 misclassification bias that can either dampen or inflate observed risks given the study's use of
29 multiple exposure groupings. "Very low level" was used for subjects with occupational
30 exposure intensities judged to be comparable to the upper end of the normal range for the general
31 population; "low-level intensity" when workplace exposure was judged to be low because of
32 control measures but higher than background; "medium exposure" for occupational
33 environments with moderate or poor control measures; and "high exposure" for workplaces
34 lacking any control measures. Groupings of "very low/low" and "medium/high" exposure was
35 used to examine association with NHL. Prevalence of medium to high TCE exposure among
36 NHL cases was low, 3% for NHL cases and 2% for all leukemia subtypes. Whether temporal
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-213 DRAFT—DO NOT CITE OR QUOTE
-------
1 changes in TCE exposure concentrations were considered in assigning level and intensity is not
2 known. Overall, this study has low sensitivity for examining TCE and lymphoma given the low
3 prevalence of exposure, particularly to medium to high TCE intensity, the high proportion of
4 proxy interviews among cases, particularly NHL cases (15%), and qualitative exposure
5 assessment approach.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-214 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
Costantini AS, Benvenuti A, Vineis P, Kriebel D, Tumino R, Ramazzotti V, Rodella S, et al., 2008. Risk of leukemia and
multiple myeloma associated with exposure to benzene and other organic solvents: evidence from the Italian multicenter case-
control study. Am J Ind Med 51:803-811.
Miligi L, Costantini AS, Benvenuti A, Kreibel D, Bolejack V, et al. 2006. Occupational exposure to solvents and the risk of
lymphomas. Epidemiol 17:552-561.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study evaluated TCE and other solvent exposures and lymphoma in a large
population-based, multicenter, case-control study.
1,732 (2,066 eligible) cases of NHL, chronic lymphatic leukemia, and Hodgkin
lymphoma, diagnosed between 1991 and 1993 among men and women, age 20 to
74 yrs and residents of 8 regions in Italy, were identified from; 1,530 (2,086 eligible)
population controls were randomly selected from demographic files or from
sampling of National Health Service files and stratified by sex and 5-yr age groups.
586 leukemia and 263 multiple myeloma among men and women, age 20 to 74 in the
period 1991-1993, from 7 regions (8 regions for chronic lymphocytic leukemia) in
Italy, were identified from hospital or pathology department records or a regional
cancer registry; and 1,100 population controls selected from demographic files or
from sampling of National Health Service files and stratified by sex and 5-yr age
groups.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
NHL and Hodgkin' s lymphoma incidence (Miligi et al., 2006).
Leukemia and multiple myeloma (Costantini et al., 2008).
All NHL cases were defined following NCI Working Formulation Workgroup
classification and Hodgkin lymphomas defined following the Rye classification.
NHL diagnosis confirmed for 334 of 1,428 cases (23%).
to
^§
"*4-
o§
H
O
HH
H
O
H
W
-------
to
VO Lo'
I
I
§
***.
£3'
1
TO'
to
§
Dl
Co
o §
H I
O >
HH Oq
H TO
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
IH experts from each region using information collected on questionnaires assigned
the probability of exposure level (low, medium and high) and intensity (very low,
low, medium and high) to TCE and other solvents. Exposure was assigned blinded
as to case or control status.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Face-to-face interview with questionnaire for detailed information about medical
history, lifestyle factors, education, occupational history (period is not identified in
published paper), and nonoccupational exposures including solvent exposure.
Unblinded interviews.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
19% of all lymphoma cases and 5% of controls were with proxy respondents
(Costantinietal., 2001).
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,732 cases (83% participation) and 1,530 controls (73% participation) (Miligi et al.,
2006); no information on participation rate for leukemia or multiple myeloma cases
or their controls in Costantini et al. (2008).
Exposure prevalence, medium to high TCE intensity, 35 NHL cases (3%) (Miligi et
al., 2006); 1 1 leukemia cases (2%) and 5 multiple myeloma cases (2%) (Costantini et
al., 2008).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age, sex, region, education, and region.
Multiple logistic regressions.
H
W
-------
to
Exposure-response analysis presented in
published paper
Documentation of results
Yes, by
exposure intensity and by duration (years) of exposure.
Yes.
'
I
Co
1
TO'
to
§
i
^§
***.
o§
H I
O
H
W
-------
1 B.3.2.6.3. Seidler et al (2007).
2 B.3.2.6.3.1. Author's abstract.
O
4 AIMS: To analyze the relationship between exposure to chlorinated and aromatic
5 organic solvents and malignant lymphoma in a multi-centre, population-based
6 case-control study. METHODS: Male and female patients with malignant
7 lymphoma (n = 710) between 18 and 80 years of age were prospectively recruited
8 in six study regions in Germany (Ludwigshafen/Upper Palatinate,
9 Heidelberg/Rhine-Neckar-County, Wiirzburg/Lower Frankonia, Hamburg,
10 Bielefeld/Giitersloh, and Munich). For each newly recruited lymphoma case, a
11 gender, region and age-matched (+/-1 year of birth) population control was drawn
12 from the population registers. In a structured personal interview, we elicited a
13 complete occupational history, including every occupational period that lasted at
14 least one year. On the basis of job task-specific supplementary questionnaires, a
15 trained occupational physician assessed the exposure to chlorinated hydrocarbons
16 (trichloroethylene, tetrachloroethylene, dichloromethane, carbon tetrachloride)
17 and aromatic hydrocarbons (benzene, toluene, xylene, styrene). Odds ratios (OR)
18 and 95% confidence intervals (CI) were calculated using conditional logistic
19 regression analysis, adjusted for smoking (in pack years) and alcohol
20 consumption. To increase the statistical power, patients with specific lymphoma
21 subentities were additionally compared with the entire control group using
22 unconditional logistic regression analysis. RESULTS: We observed a statistically
23 significant association between high exposure to chlorinated hydrocarbons and
24 malignant lymphoma (Odds ratio = 2.1; 95% confidence interval 1.1-4.3). In the
25 analysis of lymphoma subentities, a pronounced risk elevation was found for
26 follicular lymphoma and marginal zone lymphoma. When specific substances
27 were considered, the association between trichloroethylene and malignant
28 lymphoma was of borderline statistical significance. Aromatic hydrocarbons were
29 not significantly associated with the lymphoma diagnosis. CONCLUSION: In
30 accordance with the literature, this data point to a potential etiologic role of
31 chlorinated hydrocarbons (particularly trichloroethylene) and malignant
32 lymphoma. Chlorinated hydrocarbons might affect specific lymphoma subentities
33 differentially. Our study does not support a strong association between aromatic
34 hydrocarbons (benzene, toluene, xylene, or styrene) and the diagnosis of a
35 malignant lymphoma.
36
37 B.3.2.6.3.2. Study description and comment. This population case-control study of NHL and
38 Hodgkin's lymphoma patients in six Germany regions is part of a larger multiple-center and
39 -country case-control study of lymphoma and environmental exposures, the EPILYMPH study.
40 A total of 710 cases and 710 controls that were matched to cases on age, sex, and region,
41 participated in this study. Participation rates were 88% for cases and 44% for controls. Potential
42 for selection bias may exist given the low control response rate. Strength of this study is the use
43 of WHO classification scheme for classifying lymphomas and the high percentage of cases with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-218 DRAFT—DO NOT CITE OR QUOTE
-------
1 histologically-confirmed diagnoses. An industrial physician blinded to case and control status
2 assigned exposure to specific solvents (i.e., TCE, perchloroethylene, carbon tetrachloride, etc.)
3 using a JEM developed for the EPILYMPH investigators, a modification of Bolm-Audorff et al.
4 (1988). Exposure prevalence to TCE among cases was 13%. A cumulative exposure score was
5 calculated and was the sum for every job held of intensity of solvent exposure, frequency of
6 exposure, and duration of exposure. High exposure to TCE was defined as >35 ppm-years; 3%
7 of cases had high cumulative exposure to TCE. Intensity of TCE exposure was assessed on a
8 semi quantitative scale with the following categories: low intensity, 2.5 ppm (0.5 to 5); medium
9 intensity, 25 ppm (>5 to 50), high intensity, 100 ppm (>50). The frequency of exposure was the
10 percentage of working time during which the exposure occurred based upon a 40-hour week. A
11 semi quantitative scale was adopted for frequency of exposure with the following categories: low
12 frequency, 3% of working time (range, 1 to 5%), medium frequency, 17.5 % (range, >5 to 30%),
13 high frequency, 65% of working time (>30%). A cumulative Prevalence of TCE exposure
14 among cases was 13% overall with 3% of cases identified with cumulative exposure
15 >35 ppm-years.
16 Overall, the use of expert assessment for exposure and WHO classification for disease
17 coding likely reduce misclassification bias in this study. This population case-control study, like
18 other population case-control studies of lymphoma and TCE, has a low prevalence of TCE
19 exposure and limits statistical power to detect risk factors.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-219 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Seidler A, Mohner M, Berger J, Mester B, Deeg E, Eisner G, Neiters A, Becker N. 2007. Solvent exposure and malignant
lymphoma: a population-based case-control study in Germany. J Occup Med Toxicol 2:2. Accessed August 27, 2007,
http://www.occup-med.eom/content/2/l/2.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This case-control study of NHL and Hodgkin lymphomas was designed to investigate
association between specific exposure and distinct lymphoma classifications which
are defined by REAL and WHO classifications.
812 male and female lymphoma patients between the ages of 18 and 80 yrs were
identified from a six German study regions from 1999 to 2003. 1,602 controls were
identified from population registers and matched (1 : 1) to cases on sex, region and
age. 710 cases and 710 controls were interviewed.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
NHL and Hodgkin' s lymphoma incidence.
WHO classification. Diagnosis confirmed by pathological report for 691 cases.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Blinded assignment of intensity and frequency of exposure to specific chlorinated
hydrocarbons (includes TCE) and to aromatic hydrocarbons based upon
questionnaire information on complete occupational history for all jobs of >1 yr
duration. Exposure assessment approach based on a modification of Bolm-Audorff
etal. (1988).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
td
to
o
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
td
to
to
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Face-to-face interview with questionnaire for detailed information about medical
history, lifestyle factors, and occupation. Job-task-specific supplementary
questionnaire administered to subjects having held jobs of interest; e.g., painters,
metal workers and welders, dry cleaners, chemical workers, shoemakers and leather
workers, and textile workers.
Unblinded interviews.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No information provided in paper.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
710 cases (87.4%) and 710 controls (44.3%).
Exposure prevalence: Any TCE exposure, Cases, 13%, Controls, 15%.
High cumulative exposure (>35 ppm-yr), Cases, 3%, Controls, 1%.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, region, pack years of smoking, and # grams of alcohol consumed per day.
Conditional logistic regression.
Yes, by ppm-yr as continuous variable.
Yes.
H I
O >
HH Oq
H ^
O
H
W
-------
1 B.3.2.6.4. Persson and Fredrikson (1999), Persson et al (1993,1989).
2 B.3.2.6.4.1. Author's abstract.
O
4 Non-Hodgkin's lymphoma (NHL) has been subject to several epidemiological
5 studies and various occupational and non-occupational exposures have been
6 identified as determinants. The present study is a pooled analysis of two earlier
7 methodologically similar case-referent studies encompassing 199 cases of NHL
8 and 479 referents, all alive. Exposure information, mainly on occupational agents,
9 was obtained by mailed questionnaires to the subjects. Exposure to white spirits,
10 thinner, and aviation gasoline as well as work as a painter was connected with
11 increased odds ratios, whereas no increased risk was noted for benzene. Farming
12 was associated with a decreased odds ratio and exposure to phenoxy herbicides,
13 wood preservatives, and work as a lumberjack showed increased odds ratios.
14 Moreover, exposure to plastic and rubber chemicals and also contact with some
15 kinds of pets appeared with increased odds ratios. Office employment and
16 housework showed decreased odds ratios. This study indicates the importance of
17 investigating exposures not occurring very frequently in the general population.
18 Solvents were studied as a group of compounds but were also separated into
19 various specific compounds. The present findings suggest that the carcinogenic
20 property of solvents is not only related to the aromatic ones or to the occurrence
21 of benzene contamination, but also to other types of compounds.
22
23 B.3.2.6.4.2. Study description and comment. The exposure assessment approach of Persson
24 and Fredriksson (1999), a pooled analysis of NHL cases and referents in Persson et al. (1993)
25 and Persson et al. (1989), was based upon self-reported information obtain from a mailed
26 questionnaire to cases and controls. Ten of 17 main questions of the detailed multiple-page
27 questionnaire concerned occupational exposure, with additional questions on specific job and
28 exposure details. These studies of the Swedish population considered exposure durations of 1 or
29 more years and those received 5 to 45 years before NHL diagnosis for cases and before the point
30 in time of selection for controls. The period of TCE exposure assessed in the between 1964 and
31 1986, a time period similar to that of Axelson et al. (1994). Semiqualitative information about
32 solvent exposure was obtained directly from the questionnaires. Assignment of exposure
33 potential to individual solvents such as TCE and white spirit is not described nor does the paper
34 describe whether assignment was done blinded as to case or control status. A five-category
35 classification for intensity was developed although statistical analyses grouped the TCE
36 categories as intensity scores of >2 compared to 0/1. TCE exposure prevalence among cases was
37 8% (16 of 199) and 7% among referents (32 of 479).
38 This small study of 199 NHL cases diagnosed between 1964 and 1986 at a regional
39 Swedish hospital (Orebro) and alive at the time of data acquisition in 1986 was similar in design
40 to other lymphoma (chronic lymphocytic leukemia, multiple myeloma) and occupation studies
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-222 DRAFT—DO NOT CITE OR QUOTE
-------
1 from these investigators (Flodin et al., 1987, 1988). A series of 479 referents from the same
2 catchment area and from the same time period, identified previously from the multiple myeloma
3 and chronic lymphocytic leukemia studies, served as the source for controls in Persson and
4 Fredrikson (1999) for the NHL analysis and in Persson et al. (1989, 1993) for the Hodgkin's
5 lymphoma analysis. Given the study's entrance date as 1964, with interviews carried out in the
6 1980s, some cases were deceased with information likely provided by proxy respondents. The
7 paper does not identify the percentage of deceased cases and the magnitude of potential bias
8 associated with proxy respondents can not be determined. Little information is provided in the
9 published paper on controls; however, the paper notes 17% of eligible controls were not able or
10 unwilling to respond to the questionnaire. Case and control series appear to differ given only
11 subjects 40 to 80 years of age were included in the statistical analysis. Cases in Perrson et al.
12 (1993) were histologically confirmed diagnosis of NHL; this was not so for Persson et al. (1989).
13 Misclassification associated with misdiagnosis is not expected to be large given observation in
14 Perrson et al. (1993) of 2% of lymphoma cases were misclassified.
15 Overall, the study's 20-year period between initial case and control identification and
16 interview suggests some subjects were either survivors or information was obtained from proxy
17 respondents. In both instances, misclassification bias is likely. No information is provided on
18 job titles or the nature of TCE exposure, which was defined in the exposure assessment as
19 "exposed or unexposed." Exposure prevalence to TCE in this study is higher than that found in
20 community population studies of Miligi et al. (2006), Seidler et al. (2007), and Costantini et al.
21 (2008).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-223 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Persson B, Fredrikson M. 1999. Some risk factors for non-Hodgkin's lymphoma. Int J Occup Med Environ Health
12:135-142.
Persson B, Fredriksson M, Olsen K, Boeryd B, Axelson O. 1993. Some occupational exposure as risk factors for malignant
lymphomas. Cancer 72:1773-1778.
Persson B, Dahlander A-M, Fredriksson M, Brage HN, Ohlson C-G, Axelson O. 1989. Malignant lymphomas and
occupational exposures. Br J Ind Med 46:516-520.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
These studies of Hodgkin's Lymphoma and NHL investigated occupational
associations. Examination of TCE is not stated as a priori hypothesis.
Incident NHL and Hodgkin's lymphoma cases reported to a regional cancer registry
between 1975 and 1984, n = 148 (Persson et al., 1993), or identified from hospital
records (Orebro Medical Center Hospital) for the period 1964 and 1986, n = 175
(Persson et al., 1989). Population controls from the same geographical area as cases
were identified from previous case-control studies of leukemia and multiple myeloma
and matched on age and sex. Analysis of NHL and Hodgkin's lymphoma each used
the same set of controls.
Persson and Fredrikson (1999) — 199 cases of NHL, 479 controls.
Persson et al., 1993 — 93 NHL and 31 Hodgkin's lymphoma (90% participation);
204 controls.
Persson et al., 1989—106 NHL and 54 Hodgkin's lymphoma (91%); 275 controls.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
Classification system not identified in papers.
td
to
to
J^.
H I
O >
HH Oq
H ^
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
-
§
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Self-reported occupational exposures as
obtained from a mailed questionnaire.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Mailed questionnaire, only.
N/A
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No information provided in paper.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Exposure prevalence to TCE —
Persson and Fredrikson (1999)— 16 NHL cases (8%) and 32 controls (7%).
Persson et al. (1993) — 8 NHL cases (8%) and 5 Hodgkin's lymphoma cases (16%);
18 controls (9%).
Persson et al. (1989)— 8 NHL cases (8%) and 7 Hodgkin's lymphoma cases (13%);
14 controls (5%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Cases and controls are matched on age and sex. Statistical analyses do not control
for other possible confounders.
Only crude odds ratios are presented for TCE exposure, although logistic regression
was used to examine other occupational exposure and NHL/Hodgkin's lymphoma.
No.
Poor, unable to determine response rate in control population, if controls were similar
to cases on demographic variables such as sex and age, and whether controls were
identified from same time period as cases.
-------
1 B.3.2.6.5. Nordstrom et al (1998).
2 B.3.2.6.5.1. Author's abstract.
3
4 To evaluate occupational exposures as risk factors for hairy cell leukemia (HCL),
5 a population-based case-control study on 121 male HCL patients and 484 controls
6 matched for age and sex was conducted. Elevated odds ratio (OR) was found for
7 exposure to farm animals in general: OR 2.0, 95% confidence interval (CI) 1.2-
8 3.2. The ORs were elevated for exposure to cattle, horse, hog, poultry and sheep.
9 Exposure to herbicides (OR 2.9, CI 1.4-5.9), insecticides (OR 2.0, CI 1.1-3.5),
10 fungicides (OR 3.8, CI 1.4-9.9) and impregnating agents (OR 2.4, CI 1.3-4.6) also
11 showed increased risk. Certain findings suggested that recall bias may have
12 affected the results for farm animals, herbicides and insecticides. Exposure to
13 organic solvents yielded elevated risk (OR 1.5, CI 0.99-2.3), as did exposure to
14 exhaust fumes (OR 2.1, CI 1.3-3.3). In an additional multivariate model, the ORs
15 remained elevated for all these exposures with the exception of insecticides. We
16 found a reduced risk for smokers with OR 0.6 (CI 0.4-1.1) because of an effect
17 among non-farmers.
18
19 B.3.2.6.5.2. Study description and comment. This population case-control of hairy cell
20 leukemia, a B-cell lymphoid neoplasm and NHL, examined occupational organic solvent and
21 pesticide exposures among male cases reported to the Swedish Cancer Registry between 1987
22 and 1992. A total of 121 cases, including 1 case one case, originally thought to have a diagnosis
23 within the study's window, but latter learned as in 1993, and four controls per case matched on
24 age and county of residence from the Swedish Population Registry. Occupational exposure was
25 assessed based upon self-reported information provided in a mailed questionnaire with telephone
26 follow-up by trained interviewer blinded to case or control status. Chemical-specific exposures
27 of at least 1 day duration and occurring one year prior to case diagnosis were assigned to study
28 subjects; however, the procedure for doing this was not described in the paper. Potential for
29 organic solvents exposure included exposure received during leisure activities and work-related
30 activities. Exposure prevalence to TCE among cases is 8 and 7% among controls. The low
31 exposure prevalence and study size limit the statistical power of this study for detecting relative
32 risks smaller than 2.0.
33 Odds ratios and 95% confidence intervals are presented for chemical-specific exposures,
34 including TCE, from logistic regression models in two separate analyses, univariate analysis and
35 multivariate analysis adjusting for age. The odds ratio for TCE exposure is presented only from
36 univariate analysis. Age may not greatly confound or bias the observed association; an
37 examination of risk estimates from univariate and multivariate analyses of the aggregated
38 exposure category for organic solvents showed similar odds ratios, indicating age was not a
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-226 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant source of bias in the statistical analyses because age was controlled in the study's
2 design, a control was matching to a case on age.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-227 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
***.
£3'
1
TO'
Nordstrom M, Hardell L, Hagberg H, Rask-Andersen A. 1998. Occupational exposures, animal exposure and smoking as risk
factors for hairy cell leukemia evaluated in a case-control study. Br J Cancer 77:2048-2052.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Abstract — To evaluate occupational exposure as risk factors for hairy cell leukemia.
121 cases of HCL in males reported to the Swedish Cancer Registry between 1987
and 1992.
484 controls (1:4 matching) identified from Swedish Population Registry and
matched for age and county of residence.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
Not identified in paper, likely ICD-9 (http://www.socialstyrelsen.se/, accessed
February 6, 2009).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Self-reported information on occupational exposure as obtained from a mailed
questionnaire to study participants. Questionnaire sought information on complete
working history, other exposures, and leisure time activities with telephone interview
in cases of incomplete information. Paper does not describe the procedure for
assigning chemical exposures from job title information.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Mailed questionnaire.
Follow-up telephone interview and job/exposure coding were done blinded as to case
and control status.
H I
O >
HH Oq
H ^
O
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Proxy responses: 4%, cases; 1% controls.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1 1 1 HCL cases, 400 controls.
Response rate: 91% cases and 83% controls.
Exposure prevalence among cases is 8 and 7% among controls.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Cases and controls are matched for age, sex, and county of residence. Effect measure
for TCE exposure from univariate analysis presented in paper; other possible
confounders or covariates not included in statistical analysis.
Logistic regression.
No.
Yes.
td
to
VO
^§
H I
O >
HH Oq
H TO
O
HCL = hairy cell leukemia.
H
W
-------
1 B.3.2.6.6. Fritschi and Siemiatycki (1996a), Siemiatycki (1991).
2 B.3.2.6.6.1. Author's abstract.
O
4 The known risk factors for lymphoma and myeloma cannot account for the
5 current incidence rates of these cancers, and there is increasing interest in
6 exploring occupational causes. We present results regarding lymphoma and
7 myeloma from a large case-control study of hundreds of occupational exposures
8 and 19 cancer sites. We examine in more detail those exposures previously
9 considered to be related to these cancers, as well as exposures which were
10 strongly related in our initial analyses. Lymphoma was not associated in our data
11 with exposure to solvents or pesticides, or employment in agriculture or wood-
12 related occupations, although numbers of exposed cases were sometimes small.
13 Hodgkin's lymphoma was associated with exposure to fabric dust, and non-
14 Hodgkin's lymphoma was associated with exposure to copper dust, ammonia and
15 a number of fabric and textile-related occupations and exposures. Employment as
16 a sheet metal worker was associated with development of myeloma.
17
18 B.3.2.6.6.2. Study description and comment. This population study of several cancer sites
19 included histologically-confirmed cases of NHL, Hodgkin's lymphoma and myeloma ascertained
20 from 16 Montreal-area hospitals between 1979 and 1985 and part of a larger study of 10 other
21 cancer sites. This study relies on the use of expert assessment of occupational information on a
22 detailed questionnaire and face-to-face interview. Fritschi and Siemiatycki (1996a) present
23 observations of analyses examining industries, occupation, and some chemical-specific
24 exposures, including solvents, but not TCE. Observations on TCE are found in the original
25 report of Siemiatycki (1991).
26 A total of 215 NHL cases (83% response) were identified from 19 Montreal-area
27 hospitals and while this case group is larger than that in Swedish lymphoma case-control studies,
28 there are fewer NHL cases than other multicenter studies published since 2000. The
29 533 population controls (72% response), identified through the use of random digit dialing, and
30 were used for each site-specific cancer case analyses. All controls were interviewed using
31 face-to-face methods; however, 20% of the NHL cases were either too ill to interview or had
32 died and, for these cases, occupational information was provided by a proxy respondent. The
33 quality of interview conducted with proxy respondents was much lower, increasing the potential
34 for misclassification bias, than that with the subject. The direction of this bias would diminish
35 observed risk towards the null. Interviewers were unblinded, although exposure assignment was
36 carried out blinded as to case and control status. The questionnaire sought information on the
37 subject's complete job history and included questions about the specific job of the employee and
38 work environment. Occupations considered with possible TCE exposure included machinists,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-230 DRAFT—DO NOT CITE OR QUOTE
-------
1 aircraft mechanics, and industrial equipment mechanics. An additional specialized questionnaire
2 was developed for certain job title ofa prior interest that sought more detailed information on
3 tasks and possible exposures. For example, the supplemental questionnaire for machinists
4 included a question on TCE usage.
5 A team of industrial hygienists and chemicals assigned exposures blinded based on job
6 title and other information obtained by questionnaire. A semiquantitative scale was developed
7 for 294 exposures and included TCE (any, substantial). Any exposure to TCE was 3% among
8 cases but <1% for substantial TCE exposure; "substantial" is defined as >10 years of exposure
9 for the period up to 5 years before diagnosis. The TCE exposure frequencies in this study are
10 lower than those in more recent NHL case-control studies examining TCE. The expert
11 assessment method is considered a valid and reliable approach for assessing occupational
12 exposure in community-base studies and likely less biased from exposure misclassification than
13 exposure assessment based solely on self-reported information (IOM, 2003; Fritschi et al., 2003;
14 Siemiatycki et al., 1997).
15 Logistic regression models adjusted for age, ethnicity, income, and respondent status
16 (Fritschi and Siemiatycki, 1996a) or Mantel-Haenszel $ stratified on age, family income, and
17 cigarette smoking (Siemiatycki, 1991). Odds ratios for TCE exposure are presented with 90%
18 confidence intervals in Siemiatycki (1991) and with 95% confidence intervals in Fritschi and
19 Siemiatycki (1996a).
20 The strengths of this study were the large number of incident cases, specific information
21 about job duties for all jobs held, and a definitive diagnosis of NHL. However, the use of the
22 general population (rather than a known cohort of exposed workers) reduced the likelihood that
23 subjects were exposed to TCE, resulting in relatively low statistical power for the analysis. The
24 job exposure matrix, applied to the job information, was very broad since it was used to evaluate
25 294 chemicals. Overall, a reasonably good exposure assessment is found in this analysis;
26 however, examination of NHL and TCE exposure is limited by statistical power considerations
27 related to low exposure prevalence, particularly for "substantial" exposure. For the exposure
28 prevalence found in this study to TCE and for NHL, the minimum detectable odds ratio was 3.0
29 when p = 0.02 and a = 0.05 (one-sided). The low statistical power to detect a doubling of risk
30 and an increased possibility of misclassification bias associated with case occupational histories
31 resulting from proxy respondents suggests this study is less sensitive than other NHL case-
32 controls published since 2000 for examining NHL and TCE.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-231 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Fritschi L, Siemiatycki J. 1996a. Lymphoma, myeloma and occupation: Results of a case-control study. Int J Cancer 67:
498-503.
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. J Siemiatycki, Ed. Baca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study of NHL was designed to investigate association
between specific exposure and cancers at 20 sites using expert assessment method for
exposure assignment.
258 histologically-confirmed NHL cases were identified among Montreal area males,
aged 35 to 70 yrs, diagnosed in 16 Montreal hospitals between 1979 and 1985.
740 male population controls were identified from the same source population using
random digit dialing methods.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
NHL.
ICDO-0, 200 and 202 (International Statistical Classification of Diseases for
Oncology, WHO, 1997).
ICDO-0 is based upon rubrics of ICD, 9th Revision.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semi quantitative scale developed for 300 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Yes, 82% of case interviews were face-to-face; 100% of control interviews were with
subject.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, -20% of cases had proxy respondents. Interviews were completed with all
control subjects.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
215 cases (83% response), 533 population controls (71%).
Exposure prevalence: Any TCE exposure, 3% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), <1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, income, index for cigarette smoking (Siemiatycki, 1991).
Age, proxy status, income, ethnicity (Fritschi and Siemiatycki, 1996a).
Mantel -Haenszel (Siemiatycki, 1991).
Unconditional logistic regression (Fritschi and Siemiatycki, 1996a).
No.
Yes.
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.6.7. Hardett et al (1994,1981).
2 B.3.2.6.7.1. Author's abstract.
O
4 Results on 105 cases with histopathologically confirmed non-Hodgkin's
5 lymphoma (NHL) and 335 controls from a previously published case-control
6 study on malignant lymphoma are presented together with some extended
7 analyses. No occupation was a risk factor for NHL. Exposure to phenoxyacetic
8 acids yielded, in the univariate analysis, an odds ratio of 5.5 with a 95%
9 confidence interval of 2.7-11. Most cases and controls were exposed to a
10 commercial mixture of 2, 4-dichlorophenoxyacetic acid and 2, 4, 5-
11 trichlorophenoxyacetic acid. Exposure to chlorophenols gave an odds ratio of 4.8
12 (2.7-8.8) with pentachlorophenol being the most common type. Exposure to
13 organic solvents yielded an odds ratio of 2.4 (1.4-3.9). These results were not
14 significantly changed in the multivariate analysis.
15 Dichlorodiphenyltrichloroethane, asbestos, smoking, and oral snuff were not
16 associated with an increased risk for NHL. The results regarding increased risk
17 for NHL following exposure to phenoxyacetic acids, chlorophenols, or organic
18 solvents were not affected by histopathological type, disease stage, or anatomical
19 site of disease presentation. Median survival was somewhat longer in cases
20 exposed to organic solvents than the rest. This was explained by more prevalent
21 exposure to organic solvents in the group of cases with good prognosis NHL
22 histopathology.
23 A number of men with malignant lymphoma of the histiocytic type and
24 previous exposure to phenoxy acids or chlorophenols were observed and reported
25 in 1979. A matched case-control study has therefore been performed with cases of
26 malignant lymphoma (Hodgkin's disease and non-Hodgkin lymphoma). This
27 study included 169 cases and 338 controls. The results indicate that exposure to
28 phenoxy acids, chlorophenols, and organic solvents may be a causative factor in
29 malignant lymphoma. Combined exposure of these chemicals seemed to increase
30 the risk. Exposure to various other agents was not obviously different in cases and
31 in controls.
32
33 B.3.2.6.7.2. Study description and comment. Exposure in these case-control studies of
34 histologically-confirmed lymphoma (NHL and Hodgkin's lymphoma) (Hardell et al., 1981) or
35 only the NHL cases only (Hardell et al., 1994) over a 4-year period, 1974-1978, in Umea,
36 Sweden was assessed based upon information provided in a self-administered questionnaire.
37 The questionnaire obtained information on a complete working history over the life of the
38 subjects along with information on various other exposures and leisure time activities. Organic
39 solvent exposures were examined secondary to this study's primary hypothesis examining
40 phenoxy acid or chlorophenol exposures and lymphoma. The extent of recall bias related to
41 self-reported information can not be determined nor is information provided in the published
42 papers misclassification bias resulting from next-of-kin interviews. Occupations were
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-234 DRAFT—DO NOT CITE OR QUOTE
-------
1 classification according to the Nordic Working Classification system. Chemical specific
2 exposures assignment was not described but appears to have been carried out blinded as to case
3 or control status. A semiquantitative classification scheme based on intensity and duration of
4 exposure was used to categorize solvent exposure into two groupings: low grade—less than
5 1 week continuously or less than 1 month in total—and high grade for all other exposure
6 scenarios. TCE exposure prevalence is similar in both studies; 4% for cases and 1% for controls.
7 The low exposure prevalence and small numbers of cases with TCE exposure (n = 4) limits the
8 statistical power of these analyses and results in wide confidence intervals around the estimated
9 odds ratio for TCE exposure (95% Confidence Interval, 1.3-42).
10 The Rappaport Classification was used to identify non-Hodgkin' s and Hodgkin' s
11 Lymphoma cases. The Rappaport Classification was in widespread use until the 1970s and was
12 based on a cell's pathologic characteristics. Equivalence of non-Hodgkin's lymphoma groupings
13 according to Rappaport Classification system to ICDA-8 groupings, also in use during this time
14 period, is 200 "Lymphosarcoma and reticulum-cell sarcoma" and 202 "Other neoplasms of
15 lymphoid tissue."
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-23 5 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Hardell L, Eriksson M, Degerman A. 1994. Exposure to phenoxyacetic acids, chlorophenols, or organic solvents in relation to
histopathology, stage, and anatomical localization of non-Hodgkin's lymphoma. Cancer Res 54:2386-2389.
Hardell L, Eriksson M, Lenner P, Lundgren E. 1981. Malignant lymphoma and exposure to chemicals, especially organic
solvents, chlorophenols and phenoxy acids: a case-control study. Br J Cancer 43:169-176.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
NHL cases from a case-control study of lymphoma (NHL and Hodgkin's lymphoma)
are analyzed separately to evaluate herbicide and organic solvents exposure.
105 cases of histologically-confirmed NHL among males aged 25-85 yrs admitted to
local hospital's oncology department between 1974 and 1978.
A total of 335 male controls identified from the Swedish Population Registry, for
living cases, and from the Swedish Registry for Causes of Death, for dead cases.
Controls matched to cases by age, residence municipality, and year of death, for dead
cases.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
Rappaport Classification; equivalent to ICDA-8 Codes, 200, and 202.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Self-reported information on occupational exposure as obtained by questionnaire,
with a telephone interview for incomplete or unclear information. Questionnaire
sought information on complete working history, other exposures and leisure time
activities. Paper does not describe the procedure for assigning chemical exposures
from job title information.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
No information in paper.
Follow-up telephone interview was done blinded as to case and control
status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No information in paper.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
105 NHL cases, 332 controls.
Response rates could not be calculated given insufficient information in
Prevalence of TCE exposure, 4% cases, 1% controls.
paper.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Cases and controls matched on sex, age, place of residence and vital status. For
deceased controls are matched to deceased cases on year of death.
Mantel-Haenszel stratified by age and vital status.
No.
Yes.
to
OJ
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.7. Childhood Leukemia
2 B.3.2.7.1. Shu et al (2004,1999).
3 B.3.2.7.1.1. Author's abstract.
4
5 Ras proto-oncogene mutations have been implicated in the pathogenesis of many
6 malignancies, including leukemia. While both human and animal studies have
7 linked several chemical carcinogens to specific ras mutations, little data exist
8 regarding the association of ras mutations with parental exposures and risk of
9 childhood leukemia. Using data from a large case control study of childhood
10 acute lymphoblastic leukemia (ALL; age <15 years) conducted by the Children's
11 Cancer Group, we used a case-case comparison approach to examine whether
12 reported parental exposure to hydrocarbons at work or use of specific medications
13 are related to ras gene mutations in the leukemia cells of children with ALL. DNA
14 was extracted from archived bone marrow slides or cryopreserved marrow
15 samples for 837 ALL cases. We examined mutations in K-ras and N-ras genes at
16 codons 12, 13, and 61 by PCR and allele-specific oligonucleotide hybridization
17 and confirmed them by DNA sequencing. We interviewed mothers and, if
18 available, fathers by telephone to collect exposure information. Odds ratios (ORs)
19 and 95% confidence intervals (CIs) were derived from logistic regression to
20 examine the association of parental exposures with ras mutations. A total of 127
21 (15.2%) cases had ras mutations (K-ras 4.7% and N-ras 10.68%). Both maternal
22 (OR 3.2, 95% CI 1.7-6.1) and paternal (OR 2.0, 95% CI 1.1-3.7) reported use of
23 mind-altering drugs were associated with N-ras mutations. Paternal use of
24 amphetamines or diet pills was associated with N-ras mutations (OR 4.1, 95% CI
25 1.1-15.0); no association was observed with maternal use. Maternal exposure to
26 solvents (OR 3.1, 95% CI 1.0-9.7) and plastic materials (OR 6.9, 95% CI 1.2-
27 39.7) during pregnancy and plastic materials after pregnancy (OR 8.3, 95% CI
28 1.4-48.8) were related to K-ras mutation. Maternal ever exposure to oil and coal
29 products before case diagnosis (OR 2.3, 95% CI 1.1-4.8) and during the postnatal
30 period (OR 2.2, 95% CI 1.0-5.5) and paternal exposure to plastic materials before
31 index pregnancy (OR 2.4, 95% CI 1.1-5.1) and other hydrocarbons during the
32 postnatal period (OR 1.8, 95% CI 1.0-1.3) were associated with N-ras mutations.
33 This study suggests that parental exposure to specific chemicals may be
34 associated with distinct ras mutations in children who develop ALL.
35 Parental exposure to hydrocarbons at work has been suggested to increase the
36 risk of childhood leukemia. Evidence, however, is not entirely consistent. Very
37 few studies have evaluated the potential parental occupational hazards by
38 exposure time windows. The Children's Cancer Group recently completed a large-
39 scale case-control study involving 1842 acute lymphocytic leukemia (ALL) cases
40 and 1986 matched controls. The study examined the association of self-reported
41 occupational exposure to various hydrocarbons among parents with risk of
42 childhood ALL by exposure time window, immunophenotype of ALL, and age at
43 diagnosis. We found that maternal exposure to solvents [odds ratio (OR), 1.8;
44 95% confidence interval (CI), 1.3-2.5] and paints or thinners (OR, 1.6; 95% CI,
45 1.2-2.2) during the preconception period (OR, 1.6; 95% CI, 1.1-2.3) and during
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-23 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 pregnancy (OR, 1.7; 95% CI, 1.2-2.3) and to plastic materials during the postnatal
2 period (OR, 2.2; 95% CI, 1.0-4.7) were related to an increased risk of childhood
3 ALL. A positive association between ALL and paternal exposure to plastic
4 materials during the preconception period was also found (OR, 1.4; 95% CI, 1.0-
5 1.9). The ALL risk associated with parental exposures to hydrocarbons did not
6 vary greatly with immunophenotype of ALL. These results suggest that the effect
7 of parental occupational exposure to hydrocarbons on offspring may depend on
8 the type of hydrocarbon and the timing of the exposure.
9
10 B.3.2.7.1.2. Study description and comment. Parent hydrocarbon occupational exposure in
11 this case-control study of acute lymphatic leukemia in children less than 15 years of age was
12 assessed from telephone questionnaire to mothers and, whenever available, fathers of cases and
13 controls who were part of the large-scale incidence study by the Children's Cancer/Oncology
14 Group. A recent paper examines hydrocarbon exposures and relationship with the ras
15 proto-oncogene (Shu et al., 2004). Nearly 50% of childhood leukemia cases in the United States
16 were treated by a Children's Cancer Group hospital or institution and between January 1, 1989
17 and June 15, 1993, the study period, a total of 2,081 incident childhood leukemia cases were
18 identified with 1,914 interviews with mothers. Controls were randomly selected using a random
19 digit dialing procedure and matched to cases on age, race, and geographic location. Using
20 structured questionnaires, parents or a surrogate when unavailable were asked about job title,
21 industry, duties, starting and stopping date for all jobs held by the father for more than 6 months
22 beginning at age 18 years and by the mother for all jobs held at least 6 months in the period from
23 2 year prior to the index pregnancy to date of diagnosis of leukemia case or the reference date of
24 the controls. The questionnaire sought information on specific exposures to solvents (carbon
25 tetrachloride, TCE, benzene, toluene, and xylene), plastic materials, paints, pigments or thinners,
26 and oil or coal products. Exposure quantitative was not possible. Statistical analyses use
27 self-reported exposure to specific hydrocarbons as defined as a dichotomous variable (yes/no).
28 The potential for misclassification bias is greater with exposure assessment based upon self-
29 reports compared to that by expert assessment (Teschke et al., 2002). Exposure information was
30 linked to start and stop data of the relevant job to determine the timing of exposure related to
31 specific windows of possible susceptibility for acute lymphoblastic leukemia (ALL). The
32 author's do not describe jobs associated with possible TCE exposure.
33 The father's questionnaire was completed for 1,801 of the 2,081 eligible cases and 1,813
34 of the 2,597 eligible controls. Of the 1,618 matched sets, direct interview with fathers were
35 obtained for 83% of cases and 68% of controls. Maternal interview were completed for 1,914 of
36 the 2,081 eligible cases (92%). The low prevalence of any exposure to TCE, 1% for mothers
37 (15 cases of 1,842 matched pairs with maternal exposure information) and 8% for fathers
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-239 DRAFT—DO NOT CITE OR QUOTE
-------
1 (136 cases out 1,618 matched pairs), limits the statistical power of this study to detect low to
2 moderate risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-240 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Shu Xo, Perentesis JP, Wen W, Buckley JD, Boyle E, Ross, JA, Robison LL. 2004. Parental exposure to medications and
hydrocarbons and ras mutations in children with acute lymphoblastic leukemia: A report from the Children's Oncology
Group. Cancer Epidemiol Biomarkers Prev 13:1230-1235.
Shu XO, Stewart P, Wen W-Q, Han D, Potter JD, Buckley JD, Heineman E, Robison LL. 1999. Parental occupational
exposure to hydrocarbons and risk of acute lymphocytic leukemia in offspring. Cancer Epidemiol Markers Prev 8:783-291.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Shu et al. (1999, 2004) examine possible association with a number of maternal and
paternal exposures among cases and controls identified from the Children's
Cancer/Oncology Group. The Children's Cancer/Oncology Group is an association
of more than 120 centers in the United States, Canada, and Australia who
collaboratively carry out research on risk factors and treatment of childhood cancers.
848 children with acute lymphatic leukemia of ages 0-9 yrs of age at diagnosis from
1980-1993 and <14 yrs old at diagnosis between 1994 and 2000 were identified from
cancer care centers in Quebec, Canada.
Controls are concurrently identified from population, from 1980-1993, from family
allowance files and from 1994-2000, from universal health insurance files; and,
matched (1 : 1 matching ratio) to cases on sex and age at the time of diagnosis
(calendar date).
Participation rates- 93.1% cases (790 of 849 eligible cases); 86.2% controls (790 of
916 eligible controls).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Childhood leukemia incidence.
ICD, 9th revision, Code 204.0.
td
to
H I
O >
HH Oq
H ^
O
H
W
-------
vo £;•
I
I
§
***.
£3'
1
TO'
i
H I
O >
HH Oq
H TO
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Telephone interviews of mothers of cases and controls using structured questionnaire
were administered to obtain information on general risk factors and potential
confounders. Questionnaire also sought information on a complete job history, for
the mother from 18 years of age to the end of pregnancy and included for each job,
job title, dates of employment, type of industry, and location of employer. Statistical
analyses based on self-reported occupational exposure to hydrocarbons as defined by
broad groups and individual hydrocarbons.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Telephone interview, >99% response.
Telephone interviews were not blinded, but exposure assignment and coding was
carried out blinded to case and control status by chemists and industrial hygienists.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
100% of cases and controls had maternal history provided by direct interview with
mothers.
13% of cases and 30% of controls had paternal information provided by proxy
respondent (e.g., through maternal interview).
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
15 cases (2% exposure prevalence) and 9 controls (1% exposure prevalence) with
maternal TCE exposure.
136 cases (8% exposure prevalence) and 104 controls (13% exposure prevalence)
with paternal TCE exposure.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Child's age at time of diagnosis , sex, and calendar year of diagnosis, maternal age
and level of schooling.
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Conditional logistic regression —
By two time periods; 2 yrs before pregnancy up to birth, during specific
pregnancy period.
By level of exposure; Level 1 (some exposure) compared to no exposure, and
Level 2 (greater exposure potential) compared to no exposure.
Yes.
Yes.
TO'
td
to
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.7.2. Costas et al (2002), MADPH (1997).
2 B.3.2.7.2.1. Author's abstract.
3
4 A 1981 Massachusetts Department of Public Health study confirmed a childhood
5 leukemia cluster in Woburn, Massachusetts. Our follow-up investigation attempts
6 to identify factors potentially responsible for the cluster. Woburn has a 130-year
7 industrial history that resulted in significant local deposition of tannery and
8 chemical manufacturing waste. In 1979, two of the city's eight municipal drinking
9 water wells were closed when tests identified contamination with solvents
10 including trichloroethylene. By 1986, 21 childhood leukemia cases had been
11 observed (5.52 expected during the seventeen year period) and the case-control
12 investigation discussed herein was begun. Nineteen cases and 37 matched
13 controls comprised the study population. A water distribution model provided
14 contaminated public water exposure estimates for subject residences. Results
15 identified a non-significant association between potential for exposure to
16 contaminated water during maternal pregnancy and leukemia diagnosis, (odds
17 RATIO8.33, 95% CI 0.73-94.67). However, a significant dose-response
18 relationship (P<0.05) was identified for this exposure period. In contrast, the
19 child's potential for exposure from birth to diagnosis showed no association with
20 leukemia risk. Wide confidence intervals suggest cautious interpretation of
21 association magnitudes. Since 1986, expected incidence has been observed in
22 Woburn including 8 consecutive years with no new childhood leukemia
23 diagnoses.
24
25 B.3.2.7.2.2. Study description and comment. Exposure in this case-control study of childhood
26 leukemia over a 20-year period in Woburn, MA was assessed based upon the potential for a
27 residence at the time of diagnosis to receive water from wells G and H, wells with a hydraulic
28 mixing model of Murphy (1991) which described the town's water distribution system.
29 Monitoring of wells G and H in 1979 showed the presence of several VOCs; TCE and
30 perchloroethylene (PERC) were found to exceed drinking water guidelines, at 267 ppb and
31 21 ppb, respectively. Low levels of other contaminates were detected including chloroform,
32 1,2-dichloroethylene methyl chloroform, trichlorotrifluoroethane, and inorganic arsenic. The
33 Murphy model described the water flow through Woburn during the lifetime of wells G and H.
34 The model uses data describing the physical layout of Woburn's municipal water system and
35 information regarding the pumping cycles of wells G and H and other active uncontaminated
36 wells that supplied the municipal water system. Model accuracy showed distribution of water
37 from wells G and H to a block area with predicted mixture concentrations with an average error
38 within 10% of the know concentration. Nearly 70% of the model predictions were within 20%
39 of the know validation concentrations. An exposure value for cases and controls by exposure
40 period was the sum of the model-predicted water concentration for each residence in Woburn as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-244 DRAFT—DO NOT CITE OR QUOTE
-------
1 assigned to a hydrologically-distinct area along the water distribution network. Both cumulative
2 and average exposure estimates were derived using the model.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-245 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
td g
l f^
tO Co
^ 2
i
Costas K, Knorr RS, Condon SK. 2002. A case-control study of childhood leukemia in Woburn, Massachusetts: the
relationship between leukemia incidence and exposure to public drinking water. Sci Total Environ 300:23-25.
Massachusetts Department of Public Health (MADPH). 1997. Woburn Childhood Leukemia Follow-up Study. Volumes I
and II. Final Report.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes, "this follow-up investigation attempts to identify factors potentially responsible
for the leukemia cluster in Woburn, MA" and the primary exposure of concern for
investigation is "the potential consumption of contaminated water from Wells G and
H by Woburn residents."
21 cases of leukemia diagnosed in children <19 yrs between 1969 and 1989 who
were residents of Woburn MA. Cases diagnosed from 1982 and latter were provided
by the Massachusetts Cancer Registry. Cases diagnosed prior to 1982 were
identified from local pediatric health professionals and by contacting all
greater-Boston childhood oncology centers that treated children with leukemia.
Two controls for each case were randomly selected from Woburn Public School
records on a geographically basis and matched to cases on race, sex and date of birth
(+3 mos).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Childhood leukemia incidence.
ICD-O (Acute Lymphatic Leukemia, Acute Myelogenous Leukemia, and Chronic
Myelogenous Leukemia).
H I
O >
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
In-person interviewers with mothers and fathers of cases and controls using
questionnaire to gather information regarding demographics, residential information
for the mother and child, occupational history, maternal medical and reproductive
history, child's medical history, and life-style questions. The father's questionnaire
contained questions concerning military and occupational history and also included
duplicate questions on maternal occupational history, child's medical history and
life-style habits.
A hydraulic mixing computer model describing Woburn's water distribution system
was utilized to assign an exposure index expressed as cumulative number of months
a household received contaminated drinking water from Wells G and H.
Exposure Index = fraction of time during month when water from Wells G and H
reached the user area + fraction of water from Wells G and H supplied to user area.
No quantitative measures of TCE and other volatile organic solvents concentrations
were included in hydraulic mixing model.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Personal interviews with cases and controls; 19 of 21 cases (91%) and 38 of possible
54 controls (70%) were interviewed.
Interviewers were not blinded as to case and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
One parent interviewed for 21% of cases and 1 1% of controls.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Participation rates- 93.1% cases (790 of 849 eligible cases); 86.2% controls (790 of
916 eligible controls).
H
W
-------
to
O k^j
o S"
VO Co'
I
I
§
^.
S3'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Composite covariates used to control for socioeconomic status
during pregnancy, maternal age at birth of child, and maternal
during pregnancy.
, maternal smoking
alcohol consumption
Conditional logistic regression.
Yes.
Yes and includes information in MADPH Final
Report (1997)
TO'
td
to
oo
li
>• TO
^ ^
H S
§
H I
O >
HH Oq
H ^
O
H
W
-------
1 B.3.2.7.3. McKinney et al (1991).
2 B.3.2.7.3.1. Author's abstract.
O
4 OBJECTIVE~To determine whether parental occupations and chemical and other
5 specific exposures are risk factors for childhood leukemia. DESIGN—Case-
6 control study. Information on parents was obtained by home interview.
7 SETTING—Three areas in north England: Copeland and South Lakeland (west
8 Cumbria); Kingston upon Hull, Beverley, East Yorkshire, and Holderness (north
9 Humberside), and Gateshead. SUBJECTS-109 children aged 0-14 born and
10 diagnosed as having leukemia or non-Hodgkin's lymphoma in study areas during
11 1974-88. Two controls matched for sex and date and district of birth were
12 obtained for each child. MAIN OUTCOME MEASURES-Occupations of
13 parents and specific exposure of parents before the children's conception, during
14 gestation, and after birth. Other adults living with the children were included in
15 the postnatal analysis. RESULTS—Few risk factors were identified for mothers,
16 although preconceptional association with the food industry was significantly
17 increased in case mothers (odds ratio 2.56; 95% confidence interval 1.32 to 5.00).
18 Significant associations were found between childhood leukemia and reported
19 preconceptional exposure of fathers to wood dust (2.73, 1.44 to 5.16), radiation
20 (3.23, 1.36 to 7.72), and benzene (5.81, 1.67 to 26.44); ionizing radiation alone
21 gave an odds ratio of 2.35 (0.92 to 6.22). Raised odds ratios were found for
22 paternal exposure during gestation, but no independent postnatal effect was
23 evident. CONCLUSION—These results should be interpreted cautiously because
24 of the small numbers, overlap with another study, and multiple exposure of some
25 parents. It is important to distinguish periods of parental exposures; identified risk
26 factors were almost exclusively restricted to the time before the child's birth.
27
28 B.3.2.7.3.2. Study description and comment. A population case-control study of ALL and
29 NHL in children of <14 years of age and residing in three areas in the United Kingdom was
30 carried out to identify possible risk factors for the region's observed increased background
31 childhood leukemia rates. The Sellafield nuclear reprocessing plant was located in one of the
32 areas and one hypothesis was an examination of parental radiation exposure and childhood
33 lymphoma. Un-blinded face-to-face interviews with cases, identified from regional tumor
34 registries, and controls, identified using regional birth registers, used a structured questionnaire
35 to ascertain a complete history of employment and exposure to specific substances and radiation
36 from both child's biological parents, preferred, although, in the absence of one parent, surrogate
37 information by the other parent was obtained from the date of first employment to end of the
38 study period or, if earlier, the date the parent ceased seeing the child. The questionnaire
39 additionally sought information on maternal and paternal exposure to 22 known chemical
40 carcinogens. McKinney et al. (1991) noted that exposures were highly correlated. Information
41 on job title and industry as reported in the questionnaire was coded independently by experts to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-249 DRAFT—DO NOT CITE OR QUOTE
-------
1 occupational groupings and titles using a national classification scheme from the Office of
2 Population Census and Surveys and is a strength of this study. The category of metal refining
3 industry and occupations was one of nine occupational groups identified a priori for hypothesis
4 testing. Statistical analyses are based on exposure as defined by industry, occupational title, or
5 chemical-specific exposure.
6 Interviewers with one or both parents were carried out for 109 of 151 eligible cases
7 (72%) and with 206 of 269 eligible controls (77%), and the low exposure prevalence; no
8 information was presented on the number of surrogate interviews, or, where only one parent
9 responded for both parents. The low prevalence of TCE exposure, 5 discordant pairs (one
10 subject with exposure and the matched subject without exposure) identified with maternal TCE
11 exposure and 16 discordant pairs with paternal preconceptional TCE exposure, greatly limited
12 the statistical power of this study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-250 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
McKinney PA, Alexander FE, Cartwright RA, Parker L. 1991. Parental occupations of children with leukemia in west
Cumbria, north Humberside, and Gateshead. BMJ 302:681-687.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study examines a number of risk factors (specific chemicals and occupational
groups) as possibly associated with the high background rate of acute lymphatic
leukemia and non-Hodgkin's lymphoma in children <14 yrs in the three regions.
22 individual chemicals and 7 occupational groups for a priori hypothesis testing.
151 case children identified from two tumor registries (Yorkshire and Northern
Region). No information provided in paper on reporting accuracy of these registries.
269 population controls identified from District health authority birth registers and
matched to cases on age, sex, and region of residency at time of case diagnosis.
Participation rates- 72% of cases (n = 109) and 77% of controls (n = 206).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Childhood leukemia incidence.
No information provided in published paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Face-to-face interviews of mothers of cases and controls using structured
questionnaire were administered to obtain information on general risk factors and
potential confounders. Questionnaire also sought information on a maternal and
paternal complete job history, from first employment to end of study and included for
job title, dates of employment, and industry. Questionnaire administered to both
parents, and, if one parent was unavailable, information was provided by proxy.
Questionnaire also sought information on 22 specific chemicals. Expert assignment
of occupation based upon National classification system. Statistical analyses
industry of employment, job or occupation, and specific exposures.
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
No, face-to-face interview with 72% of case parents and 77% of control parents.
Face-to-face interviews were not blinded. Expert assignment of occupation was
carried out blinded.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No information provided in paper on percentage of proxy interviews.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Exposure prevalence to TCE — maternal exposure, 2 cases (2%) and 3 controls (2%);
paternal exposure, 9 cases (9%) and 7 controls (4%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Cases and control matched on age, sex, and region of residency at time of case
diagnosis.
Discordant pair analysis.
No.
Limited reporting of odds ratios for job title and occupations.
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.7.4. Lowengart et al (1987).
2 B.3.2.7.4.1. Author's abstract.
O
4 A case-control study of children of ages 10 years and under in Los Angeles
5 County was conducted to investigate the causes of leukemia. The mothers and
6 fathers of acute leukemia cases and their individually matched controls were
7 interviewed regarding specific occupational and home exposures as well as other
8 potential risk factors associated with leukemia. Analysis of the information from
9 the 123 matched pairs showed an increased risk of leukemia for children whose
10 fathers had occupational exposure after the birth of the child to chlorinated
11 solvents [odds ratio (OR) = 3.5, P = .01], spray paint (OR = 2.0, P = .02), dyes or
12 pigments (OR = 4.5, P = .03), methyl ethyl ketone (CAS: 78-93-3; OR = 3.0, P =
13 .05), and cutting oil (OR = 1.7, P = .05) or whose fathers were exposed during the
14 mother's pregnancy with the child to spray paint (OR = 2.2, P = .03). For all of
15 these, the risk associated with frequent use was greater than for infrequent use.
16 There was an increased risk of leukemia for the child if the father worked in
17 industries manufacturing transportation equipment (mostly aircraft) (OR = 2.5, P
18 = .03) or machinery (OR = 3.0, P = .02). An increased risk was found for children
19 whose parents used pesticides in the home (OR = 3.8, P = .004) or garden (OR =
20 6.5, P = .007) or who burned incense in the home (OR = 2.7, P = .007). The risk
21 was greater for frequent use. Risk of leukemia was related to mothers'
22 employment in personal service industries (OR = 2.7, P = .04) but not to specified
23 occupational exposures. Risk related to fathers' exposure to chlorinated solvents,
24 employment in the transportation equipment-manufacturing industry, and parents'
25 exposure to household or garden pesticides and incense remains statistically
26 significant after adjusting for the other significant findings.
27
28 B.3.2.7.4.2. Study description and comment. Self-assessed parental exposure to chemical
29 classes and to individual chlorinated solvents was assigned in this case-control study of leukemia
30 in children 10 years or younger using information obtained through telephone interviews with
31 mothers and fathers of cases and controls. Interviews were carried out for 79% of case mothers
32 (159 or 202 cases) and 81% (124 of 154) case fathers. The number of potential controls was not
33 identified in the paper, although it was reported that interviews were carried out for 136 referent
34 mothers and 87 referent fathers. Mothers served as proxy respondents for paternal exposures in
35 roughly 20% of cases and 30% of controls. The complete occupational history was sought for
36 the period 1 year before the case diagnosis date, if the case was older than 2 years, 6 months
37 before the diagnosis date, if the case was between the ages of 1 and 2 years, and the same as the
38 date of diagnosis of the case was <1 year old. Questions on specific occupational exposures such
39 as solvents or degreasers, metals, and other categories were included on the questionnaire, with
40 self-reported information used to assign exposure potential. Exposure is defined only as a
41 dichotomous variable (yes/no). In this study using a matched-pair design in the statistical
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-253 DRAFT—DO NOT CITE OR QUOTE
-------
1 analyses, there were six case-control pairs of paternal cases but not controls and 3 case-control
2 pairs with paternal controls but not cases with TCE exposure before pregnancy or during
3 pregnancy. Few mothers reported exposure to chlorinated solvents. A strength of the study is
4 the ability to examine exposure at a number of developmental periods, preconception, during
5 pregnancy, and postnatal. Misclassification bias is likely strong in this study, introduced through
6 the large number of proxy respondents and exposure assessment based upon self-reported
7 information. Misclassification resulting from proxy information will dampen observed risks,
8 where as, misclassification of self-reported exposures may bias observed risks in either direction.
9 For this reason and because of the low prevalence of exposure nature of exposure assessment
10 approach, this study provides little information on childhood leukemia risks and TCE exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-254 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Lowengart RA, Peters JM, Cicioni C, Buckley J, Bernstein L, Preston-Martin S, Rappaport E. 1987. Childhood leukemia
and parents' occupational and home exposures. JNCI 79:39-46.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This case-control study of children <10 yrs of age was conducted to identify possible
risk factors of childhood leukemia. TCE exposure was one of many occupational
exposures assessed in this study.
202 cases of acute lymphatic leukemia in children <10 yrs of age at time of diagnosis
from 1980 through 1984 were identified from the Los Angeles County Cancer
Surveillance Program, a population-based cancer registry. Controls were identified
from among friends of cases with additional controls selected using random digit
dialing from the same population as cases and were matched to cases on age, sex,
race, and Hispanic origin.
123 cases (61% response rate) and 123 controls (not able to calculate response rate
since number of possible controls not identified in paper).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Telephone questionnaire sought information on maternal and paternal preconception,
pregnancy, and postnatal (up to 1 yr before case diagnosis) exposures, including a
full occupational history (job title, employers, and dates of employments) and on the
child's exposure from birth to 1 yr before case diagnosis. Parents also provide self-
reported information on specific exposures or occupational activities. Occupations
grouped according to hydrocarbon exposure potential using definition of Zack et al.
(1980).
i
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
to
(~f>
ON
^§
H I
O >
HH Oq
H TO
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Telephone interview with 159 of 202 (79%) case mothers and 124 of 202 case fathers
(61%). Of controls, interviews were obtained from 136 mothers (65 friends of cases,
71 population controls) and 87 fathers.
Unblinded interviews.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 19% of paternal exposure information on cases was provided by the mother. 43
of 130 control mothers provided information on paternal exposures (33%).
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Paternal TCE exposure —
1 yr before pregnancy, 1/0 discordant pairs
During pregnancy, 6/3 discordant pairs
After delivery 8/3 discordant pairs.
No information is provided in paper on maternal TCE exposure.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, race, and Hispanic origin.
Discordant pair analysis.
No.
Yes.
O
H
W
-------
1 B.3.2.8. Melanoma Case-Control Studies
2 B.3.2.8.1. Fritschi and Siemiatycki (1996b), Siemiatycki (1991).
3 B.3.2.8.1.1. Author's abstract.
4
5 OBJECTIVES: Associations between occupational exposures and the occurrence
6 of cutaneous melanoma were examined as part of a large population based case-
7 control study of 19 cancer sites. METHODS: Cases were men aged 35 to 70 years
8 old, resident in Montreal, Canada, with a new histologically confirmed cutaneous
9 melanoma (n = 103). There were two control groups, a randomly selected
10 population control group (n = 533), and a cancer control group (n = 533)
11 randomly selected from among subjects with other types of cancer in the large
12 study. Odds ratios for the occurrence of melanoma were calculated for each
13 exposure circumstance for which there were more than four exposed cases (85
14 substances, 13 occupations, and 20 industries) adjusting for age, ethnicity, and
15 number of years of schooling. RESULTS: Significantly increased risk of
16 melanoma was found for exposure to four substances (fabric dust, plastic dust,
17 trichloroethylene, and a group containing paints used on surfaces other than metal
18 and varnishes used on surfaces other than wood), three occupations (warehouse
19 clerks, salesmen, and miners and quarrymen), and two industries (clothing and
20 non-metallic mineral products). CONCLUSIONS: Most of the occupational
21 circumstances examined were not associated with melanoma, nor is there any
22 strong evidence from previous research that any of those are risk factors. For the
23 few occupational circumstances which were associated in our data with
24 melanoma, the statistical evidence was weak, and there is little or no supporting
25 evidence in the scientific literature. On the whole, there is no persuasive evidence
26 of occupational risk factors for melanoma, but the studies have been too small or
27 have involved too much misclassification of exposure for this conclusion to be
28 definitive.
29
30 B.3.2.8.1.2. Study description and comment. Fritschi and Siemiatycki (1996b) and
31 Siemiatycki (1991) reported data from a case-control study of occupational exposures and
32 melanoma conducted in Montreal, Quebec (Canada) and part of a larger study of 10 other
33 site-specific cancers and occupational exposures. The investigators identified 124 newly
34 diagnosed cases of melanoma (ICD-O, 172), confirmed on the basis of histology reports,
35 between 1979 and 1985; 103 of these participated in the study interview (83.1% participation).
36 One control group (n = 533) consisted of patients with other forms of cancer recruited through
37 the same study procedures and time period as the melanoma cancer cases. A population-based
38 control group (n = 533, 72% response), frequency matched by age strata, was drawn using
39 electoral lists and random digit dialing. Face-to-face interviews were carried out with 82% of all
40 cancer cases with telephone interview (10%) or mailed questionnaire (8%) for the remaining
41 cases. Twenty percent of all case interviews were provided by proxy respondents. The
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-257 DRAFT—DO NOT CITE OR QUOTE
-------
1 occupational assessment consisted of a detailed description of each job held during the working
2 lifetime, including the company, products, nature of work at site, job activities, and any
3 additional information that could furnish clues about exposure from the interviews.
4 A team of industrial hygienists and chemists blinded to subject's disease status translated
5 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
6 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
7 exposure dimensions was categorized into none, any, or substantial exposure. Fritschi and
8 Siemiatycki (1996b) present observations of logistic regression analyses examining industries,
9 occupation, and some chemical-specific exposures, but not TCE. Observations on TCE from
10 Mantel-Haenszel analyses are found in the original report of Siemiatycki (1991). Any exposure
11 to TCE was 6% among cases (n = 8) and 4% for substantial TCE exposure (n = 4); "substantial"
12 is defined as >10 years of exposure for the period up to 5 years before diagnosis.
13 Logistic regression models adjusted for age, ethnic origin, socioeconomic status, Quetlet
14 as an index of body mass, and respondent status (Fritschi and Siemiatycki, 1996b) or
15 Mantel-Haenszel ^ stratified on age, family income, cigarette smoking, Quetlet, ethnic origin,
16 and respondent status (Siemiatycki, 1991). Odds ratios for TCE exposure are presented with
17 90% confidence intervals in Siemiatycki (1991) and 95% confidence intervals in Fritschi and
18 Siemiatycki (1996b).
19 The strengths of this study were the large number of incident cases, specific information
20 about job duties for all jobs held, and a definitive diagnosis of melanoma. However, the use of
21 the general population (rather than a known cohort of exposed workers) reduced the likelihood
22 that subjects were exposed to TCE, resulting in relatively low statistical power for the analysis.
23 The job exposure matrix, applied to the job information, was very broad since it was used to
24 evaluate 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-258 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Fritschi L, Siemiatycki J. 1996b. Melanoma and occupation: Results of a case-control study. 1996. Occup Environ Med
53:168-173.
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. J Siemiatycki, Ed. Boca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
124 melanoma cases were identified among male Montreal residents between 1979
and 1985 of which 103 were interviewed.
740 eligible male controls identified from the same source population using random
digit dialing or electoral lists; 533 were interviewed. A second control series
consisted of other cancer cases identified in the larger study (n = 533).
Participation rate: cases, 83.1%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 172 (Malignant neoplasm of skin).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semiquantitative scale developed for 294 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were conducted
either at home or in the hospital; all population control interviews were conducted at
home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
99 cases (76.7% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 8% cases (n = 8); Substantial TCE
exposure (Exposure for >10 yrs and up to 5 yrs before disease onset), 4% cases
(w = 4).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, education, and ethnic origin (Fritschi and Siemiatycki, 1996b).
Age, family income, cigarette smoking, and ethnic origin (Siemiatycki, 1991).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Fritschi and Siemiatycki, 1996b).
No.
Yes.
O
H
W
-------
1 B.3.2.9. Pancreatic Cancer Case-Control Studies
2 B.3.2.9.1. Kernan et al (1999).
3 B.3.2.9.1.1. Author's abstract.
4
5 Background The relation between occupational exposure and pancreatic cancer is
6 not well established. A population-based case-control study based on death
7 certificates from 24 U.S. states was conducted to determine if occupations/
8 industries or work-related exposures to solvents were associated with pancreatic
9 cancer death.
10 Methods The cases were 63,097 persons who died from pancreatic cancer
11 occurring in the period 1984±1993. The controls were 252,386 persons who died
12 from causes other than cancer in the same time period.
13 Results Industries associated with significantly increased risk of pancreatic cancer
14 included printing and paper manufacturing; chemical, petroleum, and related
15 processing; transport, communication, and public service; wholesale and retail
16 trades; and medical and other health-related services. Occupations associated with
17 significantly increased risk included managerial, administrative, and other
18 professional occupations; technical occupations; and sales, clerical, and other
19 administrative support occupations.
20 Potential exposures to formaldehyde and other solvents were assessed by using a
21 job exposure matrix developed for this study. Occupational exposure to
22 formaldehyde was associated with a moderately increased risk of pancreatic
23 cancer, with ORs of 1.2, 1.2, 1.4 for subjects with low, medium, and high
24 probabilities of exposure and 1.2, 1.2, and 1.1 for subjects with low, medium, and
25 high intensity of exposure, respectively.
26 Conclusions The findings of this study did not suggest that industrial or
27 occupational exposure is a major contributor to the etiology of pancreatic cancer.
28 Further study may be needed to confirm the positive association between
29 formaldehyde exposure and pancreatic cancer.
30
31 B.3.2.9.1.2. Study description and comment. Kernan et al. (1999) reported data from a case-
32 control study of occupational exposures and pancreatic cancer, coding usual occupation as noted
33 on death certificates to assign potential TCE exposure to cases and controls. Deaths from
34 pancreatic cancer from 1984-1993 were identified from 24 U. S. state and frequency-matched to
35 nonpancreatitis or other pancreatic disease deaths by state, race, sex, and age (5-year groups);
36 63,097 pancreatic cancer deaths (case series) and 252,386 controls were selected for analysis.
37 Exposure assessment in this study group occupational (n = 509) and industry (n = 231)
38 codes into 16 broad occupational and 20 industrial categories. Additionally, a job exposure
39 matrix (JEM) of Gomez et al. (1994) was applied to develop exposure surrogates for
40 11 chlorinated hydrocarbons, including TCE, and two larger groupings, all chlorinated
41 hydrocarbons and organic solvents. A qualitative surrogate (ever exposed/never exposed) for
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-261 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCE exposure is developed and no information is provided on death certifications on
2 employment duration to examine exposure-response patterns. Kernan et al. (1999) report
3 mortality odds ratios from logistic regression for TCE exposure intensity and probability of
4 exposure.
5 Overall, this is a large study that examined specific exposures using a generic JEM.
6 Errors resulting from exposure misclassification are likely, not only introduced by the generic
7 JEM, but through the use of usual occupation as coded on death certificates, which may not fully
8 represent an entire occupational history.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-262 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Kernan GJ, Ji B-T, Dosemeci M, Silverman DT, Balbus J, Zahm SH. 1999. Occupational risk factors for pancreatic cancer:
A case-control study based on death certificates from 24 U. S. states. Am J Ind Med 36:260-270.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between pancreatic cancers and occupational title or chemical exposures.
63,097 pancreatic cancer cases were identified using death certificates from 24 U. S.
states between 1984 and 1993.
63,097 noncancer, nonpancreatitis or other pancreatic disease deaths (controls)
identified from the same source population and frequency-matched to cases by state,
race, sex, and age (1:4 matching).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality.
ICD-9, 157 (Malignant neoplasm of pancreas).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Usual occupation coded on death certificate coded to 1980 U. S. census classification
system for occupation and industry. 509 occupation codes and 231 industry codes
grouped into 16 broad occupational and 20 industrial categories based on similarity
of occupational exposures. Job exposure matrix of Gomez et al. (1994) used to
assign exposure surrogates for 1 1 chlorinated hydrocarbons, including TCE, and 2
broad categories, chlorinated hydrocarbons and organic solvents.
td
to
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
td
to
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
This study did not use interviews, information reported on death certificate used to
infer potential exposure.
No interviews were conducted in this study.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
Exposure prevalence: Any TCE exposure (Low intensity exposure or higher), 14%
cases (n = 9,068); High TCE exposure, 2% cases (n = 1,271).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, metropolitan status, region of residence, and martial status.
Logistic regression.
No.
Yes.
H I
O >
HH Oq
H ^
O
H
W
-------
1 B.3.2.10. Prostatic Cancer Case-Control Studies
2 B.3.2.10.1. Aronson et al (1996), Siemiatycki (1991).
3 B.3.2.10.1.1. Author's abstract.
4
5 A population-based case-control study of cancer and occupation was carried out
6 in Montreal, Canada. Between 1979 and 1986, 449 pathologically confirmed
7 cases of prostate cancer were interviewed, as well as 1,550 cancer controls and
8 533 population controls. Job histories were evaluated by a team of
9 chemist/hygienists using a checklist of 294 workplace chemicals. After
10 preliminary evaluation, 17 occupations, 11 industries, and 27 substances were
11 selected for multivariate logistic regression analyses to estimate the odds ratio
12 between each occupational circumstance and prostate cancer with control for
13 potential confounders. There was moderate support for risk due to the following
14 occupations: electrical power workers, water transport workers, aircraft
15 fabricators, metal product fabricators, structural metal erectors, and railway
16 transport workers. The following substances exhibited moderately strong
17 associations: metallic dust, liquid fuel combustion products, lubricating oils and
18 greases, and polyaromatic hydrocarbons from coal. While the population
19 attributable risk, estimated at between 12% and 21% for these occupational
20 exposures, may be an overestimate due to our method of analysis, even if the true
21 attributable fraction were in the range of 5-10%, this represents an important
22 public health issue.
23
24 B.3.2.10.1.2. Study description and comment. Aronson et al. (1996) and Siemiatycki (1991)
25 reported data from a case-control study of occupational exposures and prostate cancer conducted
26 in Montreal, Quebec (Canada) and was part of a larger study of 10 other site-specific cancers and
27 occupational exposures. The investigators identified 557 newly diagnosed cases of prostate
28 cancer (ICD-O, 185), confirmed on the basis of histology reports, between 1979 and 1985; 449
29 of these participated in the study interview (80.6% participation). One control group consisted of
30 patients with other forms of cancer recruited through the same study procedures and time period
31 as the prostate cancer cases. A population-based control group (n = 533, 72% response),
32 frequency matched by age strata, was drawn using electoral lists and random digit dialing.
33 Face-to-face interviews were carried out with 82% of all cancer cases with telephone interview
34 (10%) or mailed questionnaire (8%) for the remaining cases. Twenty percent of all case
35 interviews were provided by proxy respondents. The occupational assessment consisted of a
36 detailed description of each job held during the working lifetime, including the company,
37 products, nature of work at site, job activities, and any additional information that could furnish
38 clues about exposure from the interviews.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-265 DRAFT—DO NOT CITE OR QUOTE
-------
1 A team of industrial hygienists and chemists blinded to subject's disease status translated
2 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
3 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
4 exposure dimensions was categorized into none, any, or substantial exposure. Aronson et al.
5 (1996) presents observations of logistic regression analyses examining industries, occupation,
6 and some chemical-specific exposures, but not TCE. Observations on TCE from Mantel-
7 Haenszel analyses are found in the original report of Siemiatycki (1991). Any exposure to TCE
8 was 2% among cases (n= 11) and <2% for substantial TCE exposure (n = 7); "substantial" is
9 defined as >10 years of exposure for the period up to 5 years before diagnosis.
10 Logistic regression models adjusted for age, education, and ethnicity (Aronson et al.,
11 1996) or Mantel-Haenszel $ stratified on age, family income, cigarette smoking, coffee, and
12 ethnic origin (Siemiatycki, 1991). Odds ratios for TCE exposure are presented with 90%
13 confidence intervals in Siemiatycki (1991) and 95% confidence intervals in Aronson et al.
14 (1996).
15 The strengths of this study were the large number of incident cases, specific information
16 about job duties for all jobs held, and a definitive diagnosis of prostate cancer. However, the use
17 of the general population (rather than a known cohort of exposed workers) reduced the likelihood
18 that subjects were exposed to TCE, resulting in relatively low statistical power for the analysis.
19 The job exposure matrix, applied to the job information, was very broad since it was used to
20 evaluate 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-266 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Aronson KJ, Siemiatycki J, Dewar R, Gerin M. 1996. Occupational risk factors for prostate cancer: Results from a case-
control study in Montreal, Canada. Am J Epidemiol 143:363-373.
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. J Siemiatycki, Ed. Boca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
557 prostate cancer cases were identified among male Montreal residents between
1979 and 1985 of which 449 were interviewed.
740 eligible male controls identified from the same source population using random
digit dialing or electoral lists; 533 were interviewed. A second control series
consisted of other cancer cases identified in the larger study.
Participation rate: cases, 83.1%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 185 (Malignant neoplasm of prostate).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semiquantitative scale developed for 294 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
to
§
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
O
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10%
telephone interview, and 8% mailed questionnaire. Cases interviews were conducted
either at home or in the hospital; all population control interviews were conducted at
home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
449 cases (80.6% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 2% cases (n=\ 1); Substantial TCE
exposure (Exposure for >10 yrs and up to 5 yrs before disease onset), <2% cases
(w = 7).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, ethnic origin, socioeconomic status, Quetlet as an index of body mass, and
respondent status (Aronson et al., 1996).
Age, family income, cigarette smoking, ethnic origin, and respondent status
(Siemiatycki, 1991).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Aronson et al., 1996).
No.
Yes.
H
W
-------
1 B.3.2.11. Renal Cell Carcinoma Case-Control Studies—Arnsberg Region of Germany
2 A series of studies (including Henschler et al. [1995], discussed in cohort study section)
3 have been conducted in an area with a long history of trichloroethylene use in several industries.
4 The main importance of these studies is that there is considerable detail on the nature of
5 exposures, which made it possible to estimate the order of magnitude of exposure even though
6 there were no direct measurements.
7
8 B.3.2.11.1. Briining et al. (2003).
9 B.3.2.11.1.1. Author's abstract.
10
11 BACKGROUND: German studies of high exposure prevalence have been
12 debated on the renal carcinogenicity of trichloroethylene (TRI). METHODS: A
13 consecutive hospital-based case-control study with 134 renal cell cancer (RCC)
14 cases and 401 controls was conducted to reevaluate the risk of TRI in this region
15 which were estimated in a previous study. Exposure was self-assessed to compare
16 these studies. Additionally, the job history was analyzed, using expert-based
17 exposure information. RESULTS: The logistic regression results, adjusted for
18 age, gender, and smoking, confirmed a TRI-related RCC risk in this region. Using
19 the database CAREX for a comparison of industries with and without TRI
20 exposure, a significant excess risk was estimated for the longest held job in TRI-
21 exposing industries (odds ratio (OR) 1.80, 95% confidence interval (CI) 1.01-
22 3.20). Any exposure in "metal degreasing" was a RCC risk factor (OR 5.57, 95%
23 CI 2.33-13.32). Self-reported narcotic symptoms, indicative of peak exposures,
24 were associated with an excess risk (OR 3.71, 95% CI 1.80-7.54).
25 CONCLUSIONS: The study supports the human nephrocarcinogenicity of
26 trichloroethylene.
27
28 B.3.2.11.1.2. Study description and comment. This study is a second case-control follow-up of
29 renal cell cancer in the Arnsberg area of Germany, which was intended to deal with some of the
30 methodological issues present in the two earlier studies. The major advantage of studies in the
31 Arnsberg area is the high prevalence of exposure to trichloroethylene because of the large
32 number of companies doing the same kind of industrial work. An interview questionnaire
33 procedure for self-assessment of exposures similar to the one used by Vamvakas et al. (1998)
34 was used to obtain detailed information about solvents used, job tasks, and working conditions,
35 as well as the occurrence of neurological symptoms. The industry and job title information in
36 the subjects' job histories were also analyzed by two schemes of expert-rated exposure
37 assignments for broad groups of jobs. The CAREX database from the European Union, for
38 industry categories, and the British JEM developed by Pannett et al. (1985), for potential
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-269 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure to chemical classes or specific chemical, but not TCE, was adopted in an attempt to
2 obtain a potentially less biased assessment of exposures.
3 Exposure prevalences for employment in industries with potential TCE and
4 perchloroethylene exposures was high in both cases (87%) and controls (79%) using the CAREX
5 approach but much lower using the JEM approach for potential exposure to degreasing agents
6 (12% cases, 9% controls), self-reported exposure to TCE (18% cases, 10% controls), and TCE
7 exposure with any symptom occurrence (14% cases, 4% controls). Both the CAREX and British
8 JEM rating approaches are very broad and they have potentially high rates of misclassification of
9 exposure intensity in job groupings and industry groupings. In an attempt to avoid reporting
10 biases associated with the legal proceeding for compensation, analyses were conducted on
11 self-reported exposure to selected agents (yes or no). The regional use of trichloroethylene and
12 perchloroethylene (tetrachloroethylene) were so widespread that most individuals recognized the
13 local abbreviations. If individuals claimed to be exposed when they were not, it would reduce
14 the finding of a relationship if one existed. Similarly, subjects were grouped by frequency of
15 perceived symptoms (any, less than daily, daily) associated with TCE or perchloroethylene
16 exposure. Overreporting would also introduce misclassification and reduce evidence of any
17 relationship. Self-reporting of exposure to chemicals in case-control studies, generally, is
18 considered unreliable since, within the broad population, workers rarely know specific chemicals
19 to which they have potential exposure. However, in cohort studies and case-control studies in
20 which one industry dominates a local population such as in this study, this is less likely because
21 the numbers of possible industries and job titles are much smaller than in a broad population.
22 The Arnsberg area studies focused on a small area where one type of industry was very
23 prevalent, and that industry used primarily just two solvents: trichloroethylene and
24 tetrachloroethylene. As a result, it was common knowledge among the workers what solvent an
25 individual was using, and, for most, it was trichloroethylene. Self-reported TCE exposure is
26 considered to be less biased compared to possible misclassification bias associated with using the
27 CAREX exposure assessment approach which identified approximately 90% of all cases as
28 holding a job in an industry using TCE or perchloroethylene (see above discussion).
29 Some subjects in Briining et al. (2003) are drawn from the underlying Arnsberg
30 population as studied by Vamvakas et al. (1998) (reviewed below) and TCE exposures to these
31 subjects would be similar—substantial, sustained high exposures to TCE at 400-600 ppm during
32 hot dip cleaning and greater than 100 ppm overall. However, the larger ascertainment area
33 outside the Arnsberg region for case and control identification may have resulted in a lower
34 exposure prevalence compared to Vamvakas et al. (1998).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-270 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Briining T, Pesch B, Wiesenhiitter B, Rabstein S, Lammert M, Baumiiller A, Bolt H. 2003. Renal cell cancer risk and
occupational exposure to trichloroethylene: results of a consecutive case-control study in Arnsberg, Germany. Am J Ind Med
23:274-285.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
From abstract — study aim was to "reevaluate the risk of TRI in this region which
were estimated in a previous study."
162 renal cell carcinoma cases identified from September 1999 to April 2000 and
who had undergone nephrectomy between 1992 and 2000 [a time period preceding
that adopted in Vamvakas et al., 1998] from a regional hospital urology department
in Arnsberg, Germany; 134 of the recruited cases were interviewed. 401 hospital
controls were interviewed between 1999 and 2000 from local surgery departments or
geriatric departments and frequency matched to cases by sex and age.
134 of 162 (83%) cases; response rate among controls could not be calculated
lacking information on the number of eligible controls.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
N/A
to
^§
H I
O >
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
to
-
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Face-to-face interview with subjects or their next of kin using a structured
questionnaire with questions to obtain information on a complete job history by job
title, supplemental information on job tasks with suspected exposure to specific
agents, medical history, and personal habits. Questionnaires also sought
self-reported information on duration and frequency of exposure to TCE and
perchloroethylene, and, for these individuals, frequency of narcotic symptoms as a
marker of high peak exposure.
Jobs titles were coded according to a British classification of occupations and
industries with potential chemical-specific exposures identified for each occupation
using CAREX, a carcinogen exposure database or the British job-exposure matrix of
Pannett et al. (1985) for chemical groupings (e.g., degreasing agents, organic
solvents).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
100% of cases or their NOK and 100% controls with face-to-face interviews.
No information on whether interviewers were blinded.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 17% of case interviews with next-of-kin; all controls were alive at time of
interview.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
CAREX Job-exposure-matrix
117 cases with TCE exposure (87% exposure prevalence among cases).
316 cases with TCE exposure (79% exposure prevalence among controls).
Self-reported TCE exposure
25 cases with TCE exposure (18% exposure prevalence among cases).
38 cases with TCE exposure (9.5% exposure prevalence among controls).
-------
to
- a
o §"
vo £3'
I
§
^.
£5'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and tobacco smoking.
Conditional logistic regression.
Yes, duration of exposure as 4 categories (no, <10 yrs,
10-<20 years, and 20+ yrs.
Yes.
TO'
td g
I f^
to £
-i
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.11.2. Peschetal (2000b).
2 B.3.2.11.2.1. Author's abstract.
O
4 BACKGROUND: This case-control study was conducted to estimate the renal
5 cell cancer (RCC) risk for exposure to occupation-related agents, besides other
6 suspected risk factors. METHODS: In a population-based multicentre study, 935
7 incident RCC cases and 4298 controls matched for region, sex, and age were
8 interviewed between 1991 and 1995 for their occupational history and lifestyle
9 habits. Agent-specific exposure was expert-rated with two job-exposure matrices
10 and a job task-exposure matrix. Conditional logistic regression was used to
11 calculate smoking adjusted odds ratios (OR). RESULTS: Very long exposures in
12 the chemical, rubber, and printing industries were associated with risk for RCC.
13 Males considered as 'substantially exposed to organic solvents' showed a
14 significant excess risk (OR = 1.6, 95% CI: 1.1-2.3). In females substantial
15 exposure to solvents was also a significant risk factor (OR = 2.1, 95% CI: 1.0-
16 4.4). Excess risks were shown for high exposure to cadmium (OR = 1.4, 95% CI:
17 1.1-1.8, in men, OR = 2.5, 95% CI: 1.2-5.3 in women), for substantial exposure
18 to lead (OR =1.5, 95% CI: 1.0-2.3, in men, OR = 2.6, 95% CI: 1.2-5.5, in
19 women) and to solder fumes (OR = 1.5, 95% CI: 1.0-2.4, in men). In females, an
20 excess risk for the task 'soldering, welding, milling' was found (OR = 3.0, 95% CI
21 : 1.1-7.8). Exposure to paints, mineral oils, cutting fluids, benzene, polycyclic
22 aromatic hydrocarbons, and asbestos showed an association with RCC
23 development.
24 CONCLUSIONS: Our results indicate that substantial exposure to metals and
25 solvents may be nephrocarcinogenic. There is evidence for a gender-specific
26 susceptibility of the kidneys.
27
28 B.3.2.11.2.2. Study description and comment. This multicenter study of renal cell carcinoma
29 and bladder cancer and in Germany, which included the Arnsberg region plus four others,
30 identified two case series from participating hospitals, 1,035 urothelial cancer cases and
31 935 renal cell carcinoma cases with a single population control series matched to cases by
32 region, sex, and age (1:2 matching ratio to urothelial cancer cases and 1:4 matching ratio to renal
33 cell carcinoma cases). A strength of the study was the high percentage of interviews with renal
34 cell carcinoma cases within 2 months of diagnosis (88.5%), reducing bias associated with proxy
35 or next-of-kin interview, and few cases diagnoses confirmed by sonography only (5%). In all,
36 935 (570 males, 365 females) renal cell carcinoma cases were interviewed face-to-face with a
37 structured questionnaire.
38 Two general JEMs, British and German, were used to assign exposures based on
39 subjects' job histories reported in an interview. Researchers also asked about job tasks
40 associated with exposure, such as metal degreasing and cleaning, and use of specific agents
41 (organic solvents chlorinated solvents, including specific questions about carbon tetrachloride,
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-274 DRAFT—DO NOT CITE OR QUOTE
-------
1 trichloroethylene, and tetrachloroethylene) to evaluate TCE potential using a ITEM. A category
2 of "any use of a solvent" mixes the large number with infrequent slight contact with the few
3 noted earlier who have high intensity and prolonged contact. Analyses examining
4 trichloroethylene exposure using either the JEM of ITEM assigned a cumulative TCE exposure
5 index of none to low, medium high and substantial, defined as the product of exposure
6 probability x intensity x duration with the following cutpoints: none to low, <30th percentile of
7 cumulative exposure scores; medium, 30th-<60th percentile; high, 60th-<90th percentile; and,
8 substantial, >90th percentile. The use of the German JEM identified approximately twice as
9 many cases with any potential TCE exposure (42%) compared to the ITEM (17%) and, in both
10 cases, few cases identified with substantial exposure, 6% by JEM and 3% by JTEM. Pesch et al.
11 (2000b) noted "exposure indices derived from an expert rating of job tasks can have a higher
12 agent-specificity than indices derived from job titles." For this reason, the JTEM approach with
13 consideration of job tasks is considered as a more robust exposure metric for examining TCE
14 exposure and renal cell carcinoma due to likely reduced potential for exposure misclassification
15 compared to TCE assignment using only job history and title.
16 While this case-control study includes the Arnsberg area, several other regions are
17 included as well, where the source of the trichloroethylene and chlorinated solvent exposures are
18 much less well defined. Few cases were identified as having substantial exposure to TCE and, as
19 a result, most subjects identified as exposed to trichloroethylene probably had minimal contact,
20 averaging concentrations of about 10 ppm or less (NRC, 2006).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-275 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
Pesch B, Haerting J, Ranft U, Klimpet A, Oelschagel, Schill W, and the MURC Study Group. 2000b. Occupational risk
factors for renal cell carcinoma: agent-specific results from a case-control study in Germany. Int J Epidemiol 29:1014-1024.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This case-control study was conducted to estimate RCC risk for exposure to
occupational-related agents; chlorinated solvents including trichloroethylene were
identified as exposures of a priori interest.
935 RCC cases were identified from hospitals in a five-region area in Germany
between 1991 and 1995. Cases were confirmed histologically (95%) or by
sonography (5%) and selected without age restriction. 4,298 population controls
identified from local residency registries in the five-region area were frequency
matched to cases by region, sex, and age.
Participation rate: cases, 88%; controls, 71%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
N/A
td
^§
H I
O >
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
§
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
A trained interviewer interviewed subjects using a structured questionnaire which
covered occupational history and job title for all jobs held longer than 1 yr, medical
history, and personal information. Two general JEMs, British and German, were
used to assign exposures based on subjects' job histories reported in an interview.
Researchers also asked about job tasks associated with exposure, such as metal
degreasing and cleaning, and use of specific agents (organic solvents chlorinated
solvents, including specific questions about carbon tetrachloride, trichloroethylene,
and tetrachloroethylene) and chemical-specific exposure were assigned using a
ITEM. Exposure index for each subject is the sum over all jobs of duration x
probability x intensity. A four category grouping was used in statistical analyses
defined by exposure index distribution of controls: no-low; medium, 30th percentile;
high, 60th percentile; substantial, 90th percentile.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Interviewers carried out face-to-face interview with all cases and controls. All cases
were interviewed in the hospital; 88.5% of cases were interviewed within 2 mos after
diagnosis. All controls had home interviews.
No , by nature of interview location.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
JEM: 391 cases with TCE exposure index of medium or higher (42% exposure
prevalence among cases).
JTEM: 172 cases with TCE exposure index of medium or higher (18% exposure
prevalence among cases).
No information is presented in paper on control exposure prevalence.
H I
O >
HH Oq
H ^
O
H
W
-------
to
- a
o §"
vo £3'
§
^.
£5'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, study center, and
smoking.
Conditional logistic regression.
Yes.
Yes.
TO'
td
to
i
^§
H I
O
H
W
-------
1 B.3.2.11.3. Vamvakas et al (1998).
2 B.3.2.11.3.1. Author's abstract.
3
4 A previous cohort-study in a cardboard factory demonstrated that high and
5 prolonged occupational exposure to trichloroethene (C2HC13) is associated with
6 an increased incidence of renal cell cancer. The present hospital-based
7 case/control study investigates occupational exposure in 58 patients with renal
8 cell cancer with special emphasis on C2HC13 and the structurally and
9 lexicologically closely related compound tetrachloroethene (C2C14). A group of
10 84 patients from the accident wards of three general hospitals in the same area
11 served as controls. Of the 58 cases, 19 had histories of occupational C2HC13
12 exposure for at least 2 years and none had been exposed to C2C14; of the 84
13 controls, 5 had been occupationally exposed to C2HC13 and 2 to C2C14. After
14 adjustment for other risk factors, such as age, obesity, high blood pressure,
15 smoking and chronic intake of diuretics, the study demonstrates an association of
16 renal cell cancer with long-term exposure to C2HC13 (odds ratio 10.80; 95% CI:
17 3.36-34.75).
18
19 B.3.2.11.3.2. Study description and comment. In a follow-up to Henschler et al. (1995)
20 (discussed below), a case-control study was conducted in the Arnsberg region of Germany where
21 there has long been a high prevalence of small enterprises manufacturing small metal parts and
22 goods, such as nuts, lamps, screws, and bolts. Both cases and controls were identified from
23 hospital records; cases from of a large regional hospital in North Rhine Wetphalia during the
24 period 1987 and 1992 and controls who were admitted to accident wards during 1993 at three
25 other regional hospitals. Control selection was carried out independent of cases demographic
26 risk factors, i.e., controls were not matched to cases. Controls may not be fully representative of
27 the case series (NRC, 2006); they were selected from a time period after case selection which
28 may introduce bias if TCE use changes over time resulted in decreased potential for exposure
29 among controls, and use of accident ward patients may be representative of the target population.
30 Exposures to TCE resulted from dipping metal pieces into vats, with room temperatures
31 up to 60°C, and placing the wet parts on tables to dry. Some work rooms were noted to be small
32 and poorly ventilated. These conditions are likely to result in high inhalation exposure to
33 trichloroethylene (100-500 ppm). Cherrie et al. (2001) estimated the long-term exposures to be
34 approximately 100 ppm. Some of the cases included in this study were also pending legal
35 compensation. As a result, there had been considerable investigation of the exposure situation by
36 occupational hygienists from the Employer's Liability Insurance Association and occupational
37 physicians, including walk-through visits and interviews of long-term employees. The legal
38 action could introduce a bias, a tendency to overreport some of the subjective reports by the
39 subjects. However, the objective working conditions were assessed by knowledgeable
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-279 DRAFT—DO NOT CITE OR QUOTE
-------
1 professionals, who corroborated the presence of the poorly controlled hot dip tanks, extensive
2 use of trichloroethylene for all types of cleaning, and the process descriptions.
3 NRC (2006) discussed a number of criticisms in the literature on Vamvakas et al. (1998)
4 by Green and Lash (1999), Cherrie et al. (2001), and Mandel (2001) and noted the direction of
5 possible bias would be positive or negative depending on the specific criticism. Overall, cases in
6 this study substantial, sustained exposures to high concentrations of trichloroethylene at
7 400-600 ppm during hot dip cleaning and greater than 100 ppm overall and observations can
8 inform hazard identification although the magnitude of observed association is uncertain give
9 possible biases.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-280 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
Vamvakas S, Briining T, Thomasson B, Lammert M, Baumiiller A, Bolt HM, Dekant W, Birner G, Henschler D, Ulm K.
1998. Renal cell cancer risk and occupational exposure to trichloroethylene: results of a consecutive case-control study in
Arnsberg, Germany. Am J Ind Med 23:274-285.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort studies
of exposure and control groups and of cases
and controls in case-control studies is adequate
Yes. From introduction — study aim was designed to investigate further the role of
occupation exposure to TCE/perchloroethylene in the formation of renal cancer.
73 renal cell carcinoma cases that had undergone nephrectomy between December
1987 and May 1992 from a hospital urology department in Arnsberg, Germany were
contacted by mail; 58 of the recruited cases were. 1 12 controls identified from
accident wards of three area hospitals were interviewed during 1993. Controls
underwent abdominal sonography to exclude kidney cancer.
62 of 73 (85%) cases and 84 of 1 12 (75%) of controls participated in study.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
N/A
td
to
HH Oq
H TO
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
to
00 «
to
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Face-to-face interview with subjects or, if deceased, with their next of kin or former
colleagues using a structured questionnaire with questions to obtain information on job
tasks with selected exposure to specific agents and to self-reported selected exposures.
A supplemental questionnaire on job conditions was administered to subjects reporting
exposure to TCE and perchloroethylene. Subjects with TCE exposures were primarily
exposed through degreasing operations in small businesses. Self-reported TCE
exposure was ranked using a semi quantitative scale based upon total exposure time and
frequency /duration of self-reported acute prenarcotic symptoms. Cherrie et al. (2001)
estimated that the machine cleaning exposures to trichloroethylene were
-400-600 ppm, with long-term average TCE exposure as -100 ppm.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Personal physicians interviewed 100% of cases or their NOK/former colleague and
100% controls.
Interviewers were not blinded nor was developments of exposure assessment
semiquantitative scale.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No information provided in paper on number of cases with NOK interviews or
interviews with former colleagues; all controls were alive and interviewed by their
personal physician.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
19 cases with TCE or perchloroethylene exposure (33% exposure prevalence) and
1 control with perchloroethylene exposure.
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, obesity, high blood pressure, smoking, and diuretic use.
Mantel-Haenszel % '.
Yes, semi quantitative scale of 4 categories (no, +, ++, +++).
No information on number of eligible controls or number interviews with case NOK or
former colleagues.
TO'
td
to
-------
1 B.3.2.12. Renal Cell Carcinoma Case-Control Studies—Arve Valley Region of France
2 A case-control study was conducted in the Arve Valley to examine the a priori
3 hypothesis of an association with renal cell carcinoma and trichloroethylene exposure. The Arve
4 Valley, like the Arnsburg Region in Germany, has a long history of trichloroethylene use in the
5 screw-cutting industry. The Arve Valley, situated in the Rhone-Alpes region of eastern France is
6 a major metalworking sector with around 800 small and medium-sized firms specializing in
7 "screw-cutting" or the machining of small mechanical parts from bars, in small, medium, and
8 large series on conventional automatic lathes or by digital control. This industry evolved around
9 the time of World War I from the region's expertise in clock-making. A major point of this
10 study is that it was designed as a follow-up study to the German renal cell cancer case-control
11 studies but in a different population with similar exposure patterns and with high prevalence of
12 exposure to trichloroethylene. For this reason, there is considerable detail on the nature of
13 exposure, which made it possible to estimate the order of magnitude of exposure, even though
14 there were not direct measurements.
15
16 B.3.2.12.1. Charbotel et al (2009), Charbotel et al (2007) Charbotel et al (2006).
17 B.3.2.12.1.1. Charbotel et al. (2009) abstract.
18
19 Abstract Background- Several studies have investigated the association between
20 trichloroethylene (TCE) exposure and renal cell cancer (RCC) but findings were
21 inconsistent. The analysis of a case control study has shown an increased risk of
22 RCC among subjects exposed to high cumulative exposure. The aim of this
23 complementary analysis is to assess the relevance of current exposure limits
24 regarding a potential carcinogenic effect of TCE on kidney.
25 Methods- Eighty-six cases and 316 controls matched for age and gender were
26 included in the study. Successive jobs and working circumstances were described
27 using a detailed occupational questionnaire. An average level of exposure to TCE
28 was attributed to each job period in turn. The main occupational exposures
29 described in the literature as increasing the risk of RCC were assessed as well as
30 non-occupational factors. A conditional logistic regression was performed to test
31 the association between TCE and RCC risk. Three exposure levels were studied
32 (average exposure during the eight-hour shift): 35 ppm, 50 ppm and 75 ppm.
33 Potential confounding factors identified were taken into account at the threshold
34 limit of 10% ( p = 0.10) (body mass index [BMI], tobacco smoking, occupational
35 exposures to cutting fluids and to other oils).
36 Results- Adjusted for tobacco smoking and BMI, the odd-ratios associated with
37 exposure to TCE were respectively 1.62 [0.77-3.42], 2.80 [1.12-7.03] and 2.92
38 [0.85-10.09] at the thresholds of 35 ppm, 50 ppm and 75 ppm. Among subjects
39 exposed to cutting fluids and TCE over 50 ppm, the OR adjusted for BMI,
40 tobacco smoking and exposure to other oils was 2.70 [1.02-7.17].
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-284 DRAFT—DO NOT CITE OR QUOTE
-------
1 Conclusion- Results from the present study as well as those provided in the
2 international literature suggest that current French occupational exposure limits
3 for TCE are too high regarding a possible risk of RCC.
4
5 B.3.2.12.1.2. Charbotel et al (2007) abstract.
6
1 Background: We investigated the association between exposure to
8 trichloroethylene (TCE) and mutations in the von Hippel-Lindau (VHL) gene and
9 the subsequent risk for renal cell carcinoma (RCC).
10 Methods: Cases were recruited from a case-control study previously carried out in
11 France that suggested an association between exposures to high levels of TCE and
12 increased risk of RCC. From 87 cases of RCC recruited for the epidemiological
13 study, 69 were included in the present study. All samples were evaluated by a
14 pathologist in order to identify the histological subtype and then be able to focus
15 on clear cell RCC. The majority of the turn or samples were fixed either in
16 formalin or Bouin's solutions. The majority of the tumors were of the clear cell
17 RCC subtype (48 including 2 cystic RCC). Mutation screening of the 3 VHL
18 coding exons was carried out. A descriptive analysis was performed to compare
19 exposed and non exposed cases of clear cell RCC in terms of prevalence of
20 mutations in both groups.
21 Results: In the 48 cases of RCC, four VHL mutations were detected: within exon
22 1 (c.332G>A, p.Serl 11 Asn), at the exon 2 splice site (c.463+lG>C and
23 c.463+2T>C) and within exon 3 (c.506T>C, p.Leu!69Pro). No difference was
24 observed regarding the frequency of mutations in exposed versus unexposed
25 groups: among the clear cell RCC, 25 had been exposed to TCE and 23 had no
26 history of occupational exposure to TCE. Two patients with a mutation were
27 identified in each group.
28 Conclusion: This study does not confirm the association between the number and
29 type of VHL gene mutations and exposure to TCE previously described.
30
31 B.3.2.12.1.3. Charbotel et al (2006) abstract.
32
33 Background: We investigated the association between exposure to
34 trichloroethylene (TCE) and mutations in the von Hippel-Lindau (VHL) gene and
35 the subsequent risk for renal cell carcinoma (RCC).
36 Methods: Cases were recruited from a case-control study previously carried out in
37 France that suggested an association between exposures to high levels of TCE and
38 increased risk of RCC. From 87 cases of RCC recruited for the epidemiological
39 study, 69 were included in the present study. All samples were evaluated by a
40 pathologist in order to identify the histological subtype and then be able to focus
41 on clear cell RCC. The majority of the tumor samples were fixed either in
42 formalin or Bouin's solutions. The majority of the tumors were of the clear cell
43 RCC subtype (48 including 2 cystic RCC). Mutation screening of the 3 VHL
44 coding exons was carried out. A descriptive analysis was performed to compare
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-285 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposed and non-exposed cases of clear cell RCC in terms of prevalence of
2 mutations in both groups.
3 Results: In the 48 cases of RCC, four VHL mutations were detected: within exon
4 1 (c.332G>A, p.Serl 11 Asn), at the exon 2 splice site (c.463+lG>C and
5 c.463+2T>C) and within exon 3 (c.506T>C, p.Leu!69Pro). No difference was
6 observed regarding the frequency of mutations in exposed versus unexposed
7 groups: among the clear cell RCC, 25 had been exposed to TCE and 23 had no
8 history of occupational exposure to TCE. Two patients with a mutation were
9 identified in each group.
10 Conclusion: This study does not confirm the association between the number and
11 type of VHL gene mutations and exposure to TCE previously described.
12
13 To test the effect of the exposure to trichloroethylene (TCE) on renal cell cancer
14 (RCC) risk, a case-control study was performed in the Arve Valley (France), a
15 geographic area with a high frequency and a high degree of such exposure. Cases
16 and controls were selected from various sources: local general practitioners and
17 urologists practicing in the area and physicians (urologists and oncologists) from
18 other hospitals of the region who might treat patients from this area. Blinded
19 telephone interviews with cases and controls were administered by a single
20 trained interviewer using occupational and medical questionnaires. The analysis
21 concerned 86 cases and 316 controls matched for age and gender. Three
22 approaches were developed to assess the link between TCE exposure and RCC:
23 exposure to TCE for at least one job period (minimum 1 year), cumulative dose
24 number of ppm of TCE per j ob period multiplied by the number of years in the
25 job period) and the effect of exposure to peaks. Multivariate analysis was
26 performed taking into account potential confounding factors. Allowing for
27 tobacco smoking and Body Mass Index, a significantly 2-fold increased risk was
28 identified for high cumulative doses: odds ratio (OR) = 2.16 (1.02-4.60). A dose-
29 response relationship was identified, as was a peak effect; the adjusted OR for
30 highest class of exposure-plus-peak being 2.73 (1.06-7.07). After adjusting for
31 exposure to cutting fluids the ORs, although still high, were not significant
32 because of lack of power. This study suggests an association between exposures
33 to high levels of TCE and increased risk of RCC. Further epidemiological studies
34 are necessary to analyze the effect of lower levels of exposure.
35
36 B.3.2.12.1.4. Study description and comment. Cases in the population-based case-control study
37 were obtained retrospectively from regional medical practitioners or from teaching hospitals
38 from 1993 to 2002, and prospectively from 2002 to mid-2003. One case was excluded from
39 analysis because it was not possible to find a control subject. Controls were either selected from
40 the same urology practice as cases or, for cases selected from teaching hospitals, from among
41 patients of the case's general practitioner. Telephone interviews of 87 renal cell carcinoma cases
42 and 316 controls matched for age and sex by a trained interviewer were used to obtain
43 information on occupational and medical history for the case-control analysis of Charbotel et al.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-286 DRAFT—DO NOT CITE OR QUOTE
-------
1 (2006). Of the 87 RCC cases, 67 cases provided consent for mutational analysis of which
2 48 cases were diagnosed with clear cell RCC, suitable for mutational analysis of the von Hippel
3 Lindau (VHL) gene (Charbotel et al., 2007). Tissue samples were paraffin-embedded or frozen
4 tissues and ability to fully sequence the VHL gene depended on type of the fixative procedure;
5 only 26 clear cell RCC cases (34% of 73 clear cell RCC cases in the case-control study) could
6 full sequencing of the VHL gene occur.
7 Two occupational questionnaires were administered to both cases and controls, a
8 questionnaire developed specifically to evaluate jobs and exposure potential in the screw-cutting
9 industry and a more general one for any other jobs. Interviewers were essentially blinded to
10 subject status as case or control for the occupational questionnaires given the medical
11 questionnaire was administered afterwards (Fevotte et al., 2006). The medical questionnaire
12 included familial kidney disease and medical history, body mass index, and history of smoking.
13 A task/TCE-Exposure Matrix was designed using information obtained from questionnaires and
14 routine atmospheric monitoring of work shops or biological monitoring (U-TCA) of workers
15 carried out since the 1960s. Questionnaires were used to elicit from each subject the main tasks
16 associated with each job, working conditions, activities or jobs that might involve TCE
17 exposures and possible exposure to other occupational risk factors for renal cell carcinoma.
18 The JEM linked to corresponding TCE-exposure levels using available industrial hygiene
19 monitoring data on atmospheric TCE levels and from biological measurement on workers.
20 Estimates reflected task duration, use of protective equipment and distance from TCE source, as
21 well, as both dermal and inhalation exposure routes. Estimated TCE intensities for jobs
22 associated with open cold degreasing were 15-18 ppm, 120 ppm for jobs working near open hot
23 degreasing machines, with up to 300 ppm for work directly above tank and for job and intensities
24 of 300 to 600 ppm for emptying, cleaning and refilling degreasers. Eight local physicians with
25 knowledge of working conditions corroborated the working conditions for individual job periods
26 after 1980 in screw-cutting shops. Overall, there was good agreement (72%) between physician
27 and the JEM. Three exposure surrogates were assigned to each case and control: time-weighted-
28 average exposure (Charbotel et al., 2009), cumulative exposure (Charbotel et al., 2006), and
29 cumulative exposure with and without peak exposure (Charbotel et al., 2006).
30 An 8-hour time-weighted average (TWA) exposure concentration was developed for each
31 job period from 1924 to 2003 and was the product of the task-specific estimated TCE intensity
32 and duration of task. A subject's lifetime 8-hour TWA was the sum of each job period specific
33 estimated TWA. Exposure peak, daily exposure reaching >200 ppm for at least 15 minutes, was
34 assessed as an additive factor and was defined by frequency (seldom exposed, few times yearly
35 to frequently exposure, few time weekly).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-287 DRAFT—DO NOT CITE OR QUOTE
-------
1 Over the study period, 19% (295 of 1,486) job periods were assessed as having TCE
2 exposure with an 8-hour TWA of less than 35 ppm for 72% of exposed jobs and >75 ppm for 5%
3 of exposed jobs. Exposure prevalence to TCE peaked in the 1970s with roughly 20% of job
4 periods with TCE exposure and 8% of subjects identified with >75 ppm. By the 1990s, exposure
5 prevalence had not only decreased to 7% but also exposure intensity, only 5% of job periods
6 with >75 ppm.
7 Cumulative TCE exposure was the sum of 8-hour TWAs over all job periods with
8 statistical analysis using four categories: no, low, medium, and high. These were defined as low,
9 5-150 ppm-years; medium, 155-335 ppm-year; and high, >335 ppm-years (HSIA, 2005).
10 Analyses were also carried out examining peak exposure, classified as yes/no and without
11 assignment of quantitative level, as additional exposure to average TCE concentration;
12 33 subjects were exposed to peaks and very few to high peaks.
13 The high exposure prevalence and strong approach for exposure assessment provides
14 Charbotel et al. (2006, 2009) more statistical power and ability to assess association of renal cell
15 carcinoma and TCE exposure. However, the low participation rate, inability to fully sequence
16 the VHL gene in all clear cell RCC cases, the lower background prevalence of mutations (15% in
17 this study compared to roughly 50% in other series) in Charbotel et al. (2007) suggest a relative
18 insensitivity of assay used and lack of a positive control limits the mutational analysis. These
19 methodological limitations introduce bias with greater uncertainties for evaluating consistency of
20 findings with somatic VHL mutations observed in other TCE-exposed RCC cases (Brauch et al.,
21 1999; Briining et al., 1997). TCE exposure prevalence (>5 ppm-year) in Charbotel et al. (2006)
22 was 43% among cases and is higher than that observed in other population-based case-control
23 studies of renal cell carcinoma and TCE (e.g., Pesch et al., 2000a). While some subjects had
24 jobs with exposures to high concentrations of TCE during the 1970s and 1980s, a large
25 percentage of jobs were to TCE concentrations of less than 35 ppm (8-hour TWA). Jobs with
26 high TCE concentrations also were identified as having frequent exposure to peak TCE
27 concentrations, particularly before 1980. Peak TCE estimates in this study were judged to be
28 lower than those in German studies of the Arnsberg region (Henschler et al., 1995; Vamvakas et
29 al., 1998) but judged higher than those of Hill Air Force Base civilian workers (Blair et al., 1998;
30 Stewart et al., 1991) due to a lower frequency of degreasing tasks in Blair et al. (1998) cohort
31 and to slower technological changes in degreasing process in the French case-control study
32 (Fevotte et al., 2006).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-288 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
td g
t f^
tO Co
00 «
vo 2
Charbotel B, Fevotte J, martin JL, Bergeret A. 2009. Cancer du rein et expositions au trichloroethylene: les valeurs limites
d'exposition professionnelle fracaises en vigueur sont-elles adaptees. Rev Epidemiol Sante Publique 57:41-47.
Charbotel B, Fevotte J, Hours M, Martin J-L, Bergeret A. 2006. Case-control study on renal cell cancer and occupational
exposure to trichloroethylene. Part II: Epidemiological Aspects. Ann Occup Hyg 50:777-787.
Fevotte J, Charbotel B, Muller-Beaute P, Martin J-L, Hours, Bergeret A. 2006. Case-control study on renal cell cancer and
occupational exposure to trichloroethylene. Part I: Exposure assessment. Ann Occup Hyg 50:765-775.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes. From abstract — study aim was to "test the effect of TCE exposure on renal cell
cancer."
117 cases of renal cell carcinoma patients were identified retrospectively from 1993
to June 2002, and prospectively from June 2002 to June 2003 from patients of
urology practices and hospital urology and oncology departments in the region of
Arve Valley, France. 404 controls were identified from the same urology practice or
from the same general practitioner, for cases identified from hospital records and
matched on residency in the geographic study area at time of case diagnosis, sex, and
year of birth. Controls sought medical treatment for conditions other than kidney or
bladder cancer. Case definition included clear cell and other subtypes of renal cell
carcinoma including chromophil, chromophobe and collecting duct carcinomas.
87 or 1 17 (74%) cases and 3 16 of 404 (78%) controls participated in study.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
N/A
H I
O >
HH Oq
H ^
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Occupational questionnaires sought information for each study subject a complete
job history and was followed-up with either a questionnaire specific for jobs and
exposures in the screw-cutting industry or a General Occupational Questionnaire,
which ever was more applicable to subject. Questionnaires also sought self-reported
information on potential TCE exposures. A medical questionnaire seeking
information on medical history and familial kidney disease was administered after
occupational questionnaires.
Jobs titles were coded according to standardized classification of occupations and
1,486 job periods grouped into 3 categories (screw-cutting, nonscrew-cutting but job
with possible TCE exposure, and no TCE exposure). An estimated 8-hour TWA was
assigned to each job and job period using a job-task-exposure matrix.
RCC and TCE was examined using three exposure approaches: exposure to at least
5 ppm for at least one job period (minimum 1 yr), cumulative dose or £ (TCE ppm
per job x years) using quantitative ranking levels (no exposure, low, medium, and
high), and potential for peak defined as any exposure 200+ ppm. TCE
concentrations associated with quantitative ranking are low, 5-150 ppm-yrs;
medium, 155-335 ppm-yrs; high, >335 ppm-yrs.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Telephone interviews were conducted by a trained interviewer.
The paper notes interviewers were blinded "as far as possible" since medical
questionnaire was administered after the occupational questionnaires.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 22% of cases were dead at time of interview compared to 7% of controls.
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
td
to
VO
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
37 cases with TCE exposure (43% exposure prevalence), 110 controls with TCE
exposure (35% exposure prevalence).
16 cases with high level confidence TCE exposure (27% exposure prevalence),
37 controls with high level confidence TCE exposure (16%).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, tobacco smoking and body mass index (Charbotel et al., 2006).
Age, sex tobacco smoking, body mass index, and exposure to cutting or petroleum
oils (Charbotel et al., 2009).
Conditional logistic regression on matched pairs.
Yes, cumulative exposure as 4 categories (no, low, medium and high exposure) and
cumulative exposure plus peaks.
Yes.
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.2.13. Renal Cell Carcinoma Case-Control Studies in Other Regions
2 B.3.2.13.1. Parent et al (2000b), Siemiatycki (1991).
3 B.3.2.13.1.1. Author's abstract.
4
5 BACKGROUND: Little is known about the role of workplace exposures on the
6 risk of renal cell cancer. METHODS: A population-based case-control study was
7 undertaken in Montreal to assess the association between hundreds of
8 occupational circumstances and several cancer sites, including the kidney. A total
9 of 142 male patients with pathologically confirmed renal cell carcinoma, 1900
10 controls with cancer at other sites and 533 population-based controls were
11 interviewed. Detailed job histories and relevant data on potential confounders
12 were obtained. A group of chemists-hygienists evaluated each job reported and
13 translated them into a history of occupational exposures using a checklist of 294
14 substances. Multivariate logistic regression models using either population, cancer
15 controls, or a pool of both groups were used to estimate odds ratios. RESULTS:
16 There were some indications of excess risks among printers, nursery workers
17 (gardening), aircraft mechanics, farmers, and horticulturists, as well as in the
18 following industries: printing-related services, defense services, wholesale trade,
19 and retail trade. Notwithstanding the low precision of many of the odds ratio
20 estimates, the following workplace exposures showed some evidence of excess
21 risk: chromium compounds, chromium (VI) compounds, inorganic acid solutions,
22 styrene-butadiene rubber, ozone, hydrogen sulphide, ultraviolet radiation, hair
23 dust, felt dust, jet fuel engine emissions, jet fuel, aviation gasoline, phosphoric
24 acid and inks. CONCLUSIONS: For most of these associations there exist no, or
25 very little, previous data. Some associations provide suggestive evidence for
26 further studies.
27
28 B.3.2.13.1.2. Study description and comment. This population case-control study of
29 histologically-confirmed kidney cancer among males who resided in the Montreal Metropolitan
30 area relies on the use of expert assessment of occupational information on a detailed
31 questionnaire and face-to-face interview and was part of a larger study of 10 other site-specific
32 cancers and occupational exposures (Parent et al., 2000b; Siemiatycki, 1991). Interviewers were
33 unblinded, although exposure assignment was carried out blinded as to case and control status.
34 The questionnaire sought information on the subject's complete job history and included
35 questions about the specific job of the employee and work environment. Occupations considered
36 with possible TCE exposure included machinists, aircraft mechanics, and industrial equipment
37 mechanics. An additional specialized questionnaire was developed for certain job title of a prior
38 interest that sought more detailed information on tasks and possible exposures. For example, the
39 supplemental questionnaire for machinists included a question on TCE usage. A team of
40 industrial hygienists and chemicals assigned exposures blinded based on job title and other
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-292 DRAFT—DO NOT CITE OR QUOTE
-------
1 information obtained by questionnaire. A semiquantitative scale was developed for
2 300 exposures and included TCE (any, substantial). Parent et al. (2000b) presents observations
3 of analyses examining job title, occupation, and some chemical-specific exposures, but not TCE.
4 Observations on TCE are found in the original report of Siemiatycki (1991). Any exposure to
5 TCE was 3% among cases but <1% for substantial TCE exposure; "substantial" is defined as
6 >10 years of exposure for the period up to 5 years before diagnosis. The TCE exposure
7 frequencies in this study are lower than those in Briining et al. (2003) and Charbotel et al. (2006),
8 studies conducted in geographical areas with a high prevalence of industries using TCE. The
9 expert assessment method is considered a valid and reliable approach for assessing occupational
10 exposure in community-base studies and likely less biased from exposure misclassification than
11 exposure assessment based solely on self-reported information (IOM, 2003; Fritschi et al., 2003;
12 Siemiatycki et al., 1997). For example, Dewar et al. (1991) examine sensitivity of JEM of
13 Siemiatycki et al. (1987) to exposure assessment by chemists and industrial hygienists using
14 interview information and evaluation of job histories. Specific solvents are not examined,
15 although, a sensitive 84% and specificity of 97% was found for the JEM for general solvent
16 exposure.
17 This population study of several cancer sites included histologically-confirmed cases of
18 kidney cancer (ICD-O 189, malignant neoplasm of kidney and other and unspecified urinary
19 organs) ascertained from 16 Montreal-area hospitals between 1979 and 1985. A total of
20 227 eligible kidney cancer cases were identified were identified from 19 Montreal-area hospitals;
21 177 cases participated in the study (78% response). One control group (n = 1,295) consisted of
22 patients with other forms of cancer (excluding lung cancer and other intestinal cancers) recruited
23 through the same study procedures and time period as the rectal cancer cases. A
24 population-based control group (n = 533), frequency matched by age strata, was drawn using
25 electoral lists and random digit dialing. All controls were interviewed using face-to-face
26 methods; however, 20 % of the all cancer cases in the larger study were either too ill to interview
27 or had died and, for these cases, occupational information was provided by a proxy respondent.
28 The quality of interview conducted with proxy respondents was much lower, increasing the
29 potential for misclassification bias, than that with the subject. The direction of this bias would
30 diminish observed risk towards the null.
31 Statistical analysis are considered valid; logistic regression model which included terms
32 for respondent status, age, smoking and body mass index in Parent et al. (2000b) and
33 Mantel-Haenszel £ stratified on age, family income, cigarette smoking, and ethic origin in
34 Siemiatycki (1991). Odds ratios are presented with 90% confidence intervals in Siemiatycki
35 (1991) and 95% confidence intervals in Parent et al. (2000b).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-293 DRAFT—DO NOT CITE OR QUOTE
-------
1 Overall, exposure assessment in this study adopted a superior approach, using expert
2 knowledge and use of a job-exposure matrix. However, examination of NHL and TCE exposure
3 is limited by statistical power considerations related to low exposure prevalence, particularly for
4 "substantial" exposure. For the exposure prevalence found in this study to TCE and for kidney
5 cancer, the minimum detectable odds ratio was 3.0 when p = 0.02 and a = 0.05 (one-sided). The
6 low statistical power to detect a doubling of risk and an increased possibility of misclassification
7 bias associated with case occupational histories resulting from proxy respondents suggests a
8 decreased sensitivity in this study for examining kidney cancer and TCE.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-294 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
Parent M-E, Hua Y, Siemiatycki J. 2000b. Occupational risk factors for renal cell carcinoma in Montreal. Am J Ind Med
38:609-618.
Siemiatycki J. 1991. Risk Factors for Cancer in the Workplace. Baca Raton: CRC Press.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical
exposures.
277 kidney cancer cases were identified among male Montreal residents between
1979 and 1985 of which 177 (147 renal cell carcinomas) were interviewed.
740 male population controls were identified from the same source population using
random digit dialing; 533 were interviewed. A second control series consisted of all
other cancer controls excluding lung and bladder cancer cases.
Participation rate: cases, 78%; population controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD 189 (Malignant neoplasm of the kidney and other and unspecified urinary
organs) (Siemiatycki, 1991).
ICD 189.0, renal cell carcinoma (Parent et al., 2000b).
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history
with supplemental questionnaire for jobs ofapriori interest (e.g., machinists,
painters). Team of chemist and industrial hygienist assigned exposure using job title
with a semi quantitative scale developed for 300 exposures, including TCE. For each
exposure, a 3 -level ranking was used for concentration (low or background, medium,
high) and frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%;
and high, >30%).
td
to
H I
O >
HH Oq
H ^
O
H
W
-------
vo £3'
I
I
§
***.
£3'
1
TO'
H I
O >
HH Oq
H ^
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
100% of cases and controls were interviewed face-to-face by a trained interviewer.
Cases interviews were conducted either at home or in the hospital; all population
control interviews were conducted at home.
Interviews were unblinded but exposure coding was carried out blinded as to case
and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 16% of cases, 13% of population controls, and 22% of cancer controls
proxy respondents.
had
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
177 cases (78% response), 533 population controls (72%).
Exposure prevalence: Any TCE exposure, 2% cases; Substantial TCE exposure
(Exposure for >10 yrs and up to 5 yrs before disease onset), 1% cases.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, income, index for cigarette smoking (Siemiatycki, 1991).
Age, smoking, body mass index, and proxy status (Parent et al., 2000b).
Mantel -Haenszel (Siemiatycki, 1991).
Logistic regression (Parent et al., 2000b).
No.
Yes.
O
H
W
-------
1 B.3.2.13.2. Dosemeci et al (1999).
2 B.3.2.13.2.1. Author's abstract.
3
4 BACKGROUND: Organic solvents have been associated with renal cell cancer;
5 however, the risk by gender and type of solvents is nuclear. METHODS: We
6 evaluated the risk of renal cell carcinoma among men and women exposed to all
7 organic solvents-combined, all chlorinated aliphatic hydrocarbons (CAHC)-
8 combined, and nine individual CAHC using a priori job exposure matrices
9 developed by NCI in a population-based case-control study in Minnesota, U.S.
10 We interviewed 438 renal cell cancer cases (273 men and 165 women) and 687
11 controls (462 men and 225 women). RESULTS: Overall, 34% of male cases and
12 21% of female cases were exposed to organic solvents in general. The risk of
13 renal cell carcinoma was significantly elevated among women exposed to all
14 organic solvents combined (OR = 2.3; 95% CI = 1.3-4.2), to CAHC combined
15 (OR = 2.1; 95% CI = 1.1-3.9), and to trichloroethylene (TCE) (OR = 2.0; 95% CI
16 = 1.0-4.0). Among men, no significant excess risk was observed among men
17 exposed to any of these nine individual CAHCs, all CAHCs-combined, or all
18 organic solvents-combined. DISCUSSION: These observed gender differences in
19 risk of renal cell carcinoma in relation to exposure to organic solvents may be
20 explained by chance based on small numbers, or by the differences in body fat
21 content, metabolic activity, the rate of elimination of xenobiotics from the body,
22 or by differences in the level of exposure between men and women, even though
23 they have the same job title.
24
25 B.3.2.13.2.2. Study description and comment. Dosemeci et al. (1999) reported data from a
26 population-based case-control study of the association between occupation exposures and renal
27 cancer risk. The investigators identified newly diagnosed patients with histologically confirmed
28 renal cell carcinoma from the Minnesota Cancer Surveillance System from July 1, 1988 to
29 December 31, 1990. The study was limited to white cases, and age and gender-stratified controls
30 were ascertained using random digit dialing (for subjects ages 20-64) and from Medicare
31 records (for subjects 65-85 years). Of the 796 cases and 796 controls initially identified,
32 438 cases (273 men, 165 women) and 687 controls (462 men, 225 women) with complete
33 personal interviews were included in the occupational analysis.
34 Data were obtained using in-person interviews that included demographic variables,
35 residential history, diet, smoking habits, medical history, and drug use. The occupational history
36 included information about the most recent and usual industry and occupation (coded using the
37 standard industrial and occupation codes, Department of Commerce), job activities, hire and
38 termination dates, and full/part time status. A job exposure matrix developed by the National
39 Cancer Institute (Gomez et al., 1994) was used with the coded job data assign occupational
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-297 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure potential for 10 chlorinated aromatic hydrocarbons and organic solvents, and includes
2 trichloroethylene.
3 Dosemeci et al. (1999) adopted logistic regression methods to evaluate renal cancer and
4 occupational exposures. Odds ratios were adjusted for age, smoking, hypertension, and use of
5 drugs for hypertension, and body mass index.
6 Strengths of this study include the use of incident cases of renal cancer from a defined
7 population area, with confirmation of the diagnosis using histology reports. The occupation
8 history was based on usual and most recent job, in combination with a relatively focused job
9 exposure matrix. In contrast to the type of exposure assessment that can be conducted in cohort
10 studies within a specific workplace, however, exposure measurements, based on personal or
11 workplace measurement, were not used, and a full lifetime job history was not obtained.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-298 DRAFT—DO NOT CITE OR QUOTE
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
Dosemeci M, Cocco P, Chow W-H. 1999. Gender differences in risk of renal cell carcinoma and occupational exposures to
chlorinated aliphatic hydrocarbons. Am J Ind Med 36:54-59.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes. From abstract — study aim was to evaluate effect of organic solvents on RCC
risk using a priori job exposure matrices.
796 white males and females identified through the Minnesota Cancer Surveillance
System with histological confirmed RCC between July 1, 1988 and December 31,
1990. Interviews were obtained for 690 subjects of which 241 were with next-of-kin
and excluded; 438 cases (273 males and 165 females) were included in analysis.
707 white population controls identified through random digit dialing, and matched to
cases, aged 20-65 yrs old, by age and sex using a stratified random sample or, for
cases aged 65-85, from Health Care Financing Administration list. 687 controls
(462 males and 225 females) are included in the analysis.
Participation rate: cases, 87%; controls, 86%.
Occupational analysis: cases, 55%, controls 83%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence
N/A
td
to
H I
O >
HH Oq
H ^
O
H
W
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
o
o
i
o §
H I
O >
HH Oq
H ^
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
A trained interviewer blinded to case and control status interviewed subjects at home
using a questionnaire which covered occupational, residential, and medical histories;
demographic information; and personal information. Occupational history included
self-reporting of the most recent job and usual occupation and industry, employment
dates, and focused on 13 specific occupations or industries.
Occupation and industry were coded according to a standard occupational
classification or standard industrial classification with potential chemical-specific
exposures to TCE and eight other chlorinated hydrocarbons identified using the job
exposure matrix of Dosemeci et al. (1994) and Gomez et al. (1994).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
All cases and controls had face-to-face interviews.
Yes, interviewers were blinded as to case and control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
No, subjects with next-of-kin interviews were excluded from the analysis.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancers in incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
55 cases with TCE exposure (13% exposure prevalence among cases).
69 controls cases with TCE exposure (10% exposure prevalence among controls).
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age, sex, smoking, body mass index, and hypertension/ use of diuretics/use of
anti-hypertension drugs.
Logistic regression.
H
W
-------
This docu
10/20/09
Exposure-response analysis presented in
published paper
Documentation of results
No.
Yes.
Co
1
TO'
td
^§
H I
O
H
W
-------
1 B.3.2.14. Other Cancer Site Case-Control Studies
2 B.3.2.14.1. Siemiatycki (1991), Siemiatycki et al (198 7).
3 B.3.2.14.1.1. Author's abstract.
4
5 A multi-cancer site, multi-factor, case-referent study was undertaken to generate
6 hypotheses about possible occupational carcinogens. About 20 types of cancer
7 were included. Incident cases among men aged 35-70 years and diagnosed in any
8 of the major Montreal hospitals were eligible. Probing interviews were carried out
9 for 3,726 eligible cases. The interview was designed to obtain detailed lifetime
10 job histories and information on potential confounders. Each job history was
11 reviewed by a team of chemists who translated it into a history of occupational
12 exposures. These occupational exposures were then analyzed as potential risk
13 factors in relation to the sites of cancer included. For each site of cancer
14 analyzed, referents were selected from among the other sites in the study. The
15 analysis was carried out in stages. First a Mantel-Haenszel analysis was
16 undertaken of all cancer-substance associations, stratifying on a limited number of
17 covariates, and, then, for those associations which were noteworthy in the initial
18 analysis, a logistic regression analysis was made taking into account all potential
19 confounders. This report describes the fieldwork and analytical methods.
20
21 B.3.2.14.1.2. Study description and comment. Siemiatycki (1991) reported data from a
22 case-control study of occupational exposures and several site-specific cancers, including lung
23 and pancreas, conducted in Montreal, Quebec (Canada). Other cases included in this study were
24 cancers of the bladder, colon, rectum, esophagus prostate, and lymphatic system (NHL); a
25 description of the other case series are found in other sections in this appendix. The investigators
26 identified 1,082 newly diagnosed cases of lung cancer (ICD-O, 162) and 165 newly diagnosed
27 cases of pancreatic cancer (ICD-O, 157), confirmed on the basis of histology reports, between
28 1979 and 1985; 857 lung cancer (79.2% ) and 117 pancreatic cancer cases (70.7%) participated
29 in the study interview. One control group consisted of patients with other forms of cancer
30 recruited through the same study procedures and time period as the melanoma cancer cases. The
31 control series for lung cancer cases excluded other lung cancer cases; the control series for
32 pancreatic cancer cases excluded all lung cancer cases. Additionally, a population-based control
33 group (n = 533, 72% response), frequency matched by age strata, was drawn using electoral lists
34 and random digit dialing. Face-to-face interviews were carried out with 82% of all cancer cases
35 with telephone interview (10%) or mailed questionnaire (8%) for the remaining cases. Twenty
36 percent of all case interviews were provided by proxy respondents. The occupational assessment
37 consisted of a detailed description of each job held during the working lifetime, including the
38 company, products, nature of work at site, job activities, and any additional information that
39 could furnish clues about exposure from the interviews.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-302 DRAFT—DO NOT CITE OR QUOTE
-------
1 A team of industrial hygienists and chemists blinded to subject's disease status translated
2 jobs into potential exposure to 294 substances with three dimensions (degree of confidence that
3 exposure occurred, frequency of exposure, and concentration of exposure). Each of these
4 exposure dimensions was categorized into none, any, or substantial exposure. Any exposure to
5 TCE was 2% among cases (n = 21 lung cancer cases, 2 pancreatic cancer cases) and 1% for
6 substantial TCE exposure (n = 9 lung cancer cases); "substantial" is defined as >10 years of
7 exposure for the period up to 5 years before diagnosis. None of the pancreatic cancer cases was
8 identified with "substantial" exposure to TCE.
9 Mantel-Haenszel £ analyses examined occupation exposures and lung cancer stratified
10 on age, family income, cigarette smoking, ethnic origin, alcohol consumption, and respondent
11 status or pancreatic cancer stratified on age, income, cigarette smoking, and respondent status
12 (Siemiatycki, 1991). Odds ratios for TCE exposure in Siemiatycki (1991) are presented with
13 90% confidence intervals.
14 The strengths of this study were the large number of incident cases, specific information
15 about job duties for all jobs held, and a definitive diagnosis of cancer. However, the use of the
16 general population (rather than a known cohort of exposed workers) reduced the likelihood that
17 subjects were exposed to TCE, resulting in relatively low statistical power for the analysis. The
18 job exposure matrix, applied to the job information, was very broad since it was used to evaluate
19 294 chemicals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-303 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Siemitycki J. 1991. Risk Factors for Cancer in the Workplace. J Siemiatycki, Ed. Boca Raton: CRC Press.
Siemiatycki J, Wacholder S, Richardson L, Dewar R, Gerin M. 1987. Discovering carcinogens in the occupational
environment. Scand J Work Environ Health 13:486-492.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and
of cases and controls in case-control
studies is adequate
This population case-control study was designed to generate hypotheses on possible
association between 1 1 site-specific cancers and occupational title or chemical exposures.
1,082 lung cases were identified among male Montreal residents between 1979 and 1985
of which 857 were interviewed; 165 cases were identified among male Montreal residents
between 1979 and 1985 of which 117 were interviewed.
740 eligible male controls identified from the same source population using random digit
dialing or electoral lists; 533 were interviewed. A second control series consisted of other
cancer cases identified in the larger study.
Participation rate: lung cancer cases, 79.2 %, pancreatic cancer cases, 70.7%; population
controls, 72%.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Incidence.
ICD-O, 122 (Malignant neoplasm of trachea, bronchus and lung).
ICD-O, 157 Malignant neoplasm of pancreas.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Unblinded interview using questionnaire sought information on complete job history with
supplemental questionnaire for jobs of a priori interest (e.g., machinists, painters). Team
of chemist and industrial hygienist assigned exposure using job title with a
semi quantitative scale developed for 294 exposures, including TCE. For each exposure, a
3 -level ranking was used for concentration (low or background, medium, high) and
frequency (percent of working time: low, 1 to 5%; medium, >5 to 30%; and high, >30%).
H 3
§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
i
H 3
§
H I
O >
HH Oq
H TO
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
82% of all cancer cases interviewed face-to-face by a trained interviewer, 10% telephone
interview, and 8% mailed questionnaire. Cases interviews were conducted either at home
or in the hospital; all population control interviews were conducted at home.
Interviews were unblinded but exposure coding was carried out blinded as to case and
control status.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Yes, 20% of all cancer cases had proxy respondents.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality
studies; numbers of total cancer incidence
studies; numbers of exposed cases and
prevalence of exposure in case-control
studies
857 lung cancer cases (79.2% response), 117 pancreatic cancer cases (70.7% response);
533 population controls (72% response).
Exposure prevalence: Any TCE exposure, 2% cancer cases (n = 21 lung cancer cases and
2 pancreatic cancer cases); substantial TCE exposure (exposure for >10 yrs and up to
5 yrs before disease onset), 1% lung cancer cases (n = 9), no pancreatic cancer cases
assigned "substantial" TCE exposure.
CATEGORY H: ANALYSIS
Control for potential confounders in
statistical analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Lung cancer — age, family income, cigarette smoking, ethnic origin, alcohol consumption,
and respondent status.
Pancreatic cancer — age, income, cigarette smoking, and respondent status.
Mantel-Haenszel (Siemiatycki, 1991).
No.
Yes.
O
H
W
-------
1 B.3.3. Geographic-Based Studies
2 B.3.3.1. Coyleetal (2005)
3 B.3.3.1.1. Author's abstract.
4
5 Purpose. To investigate the role of environment in breast cancer development, we
6 conducted an ecological study to examine the association of releases for selected
7 industrial chemicals with breast cancer incidence in Texas.
8 Methods. During 1995-2000, 54,487 invasive breast cancer cases were reported
9 in Texas. We identified 12 toxicants released into the environment by industry
10 that: (1) were positively associated with breast cancer in epidemiological studies,
11 (2) were Environmental Protection Agency (EPA) Toxics Release Inventory
12 (TRI) chemicals designated as carcinogens or had estrogenic effects associated
13 with breast cancer risk, and (3) had releases consistently reported to EPA TRI for
14 multiple Texas counties during 1988-2000. We performed univariate, and
15 multivariate analyses adjusted for race and ethnicity to examine the association of
16 releases for these toxicants during 1988-2000 with the average annual age-
17 adjusted breast cancer rate at the county level.
18 Results. Univariate analysis indicated that formaldehyde, methylene chloride,
19 styrene, tetrachloroethylene, trichloroethylene, chromium, cobalt, copper, and
20 nickel were positively associated with the breast cancer rate. Multivariate
21 analyses indicated that styrene was positively associated with the breast cancer
22 rate in women and men (b = 0.219, p =0.004), women (b = 0.191, p=0.002), and
23 women J 50 years old (b = 0.187, p=0.002).
24 Conclusion. Styrene was the most important environmental toxicant positively
25 associated with invasive breast cancer incidence in Texas, likely involving
26 women and men of all ages. Styrene may be an important breast carcinogen due
27 to its widespread use for food storage and preparation, and its release from
28 building materials, tobacco smoke, and industry.
29
30 B.3.3.1.2. Study description and comment. Residential address in 254 Texas counties at time
31 of cancer diagnosis was the exposure surrogate in this ecologic study of invasive breast cancer in
32 over a 5-year period (1995-2000). Incident breast cancer cases in males and females were
33 identified from Texas Cancer Registry. During the 5-year period, 54,487 cases were diagnosed,
34 of which 53,910 were in females (99%). Median average annual age-adjusted breast cancer rates
35 for women and men, women, women <50 years old, and women >50 years old and 12 hazardous
36 air pollutants identified as exposures of interested were examined using nonparametric tests
37 (Mann-Whitney U test) and linear regression analyses. The 12 hazardous air pollutants (HAPs)
38 were: carbon tetrachloride, formaldehyde, methylene chloride, styrene, perchloroethylene, TCE,
39 arsenic, cadmium, chromium, cobalt, copper, and nickel. On-site atmospheric release data on
40 individual HAPs was identified from EPA's Toxics Release Inventory (TRI) for a 13-year
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-306 DRAFT—DO NOT CITE OR QUOTE
-------
1 period, 1998 to 2000 with an exposure surrogate as the annual total release in pounds/year for the
2 12HAPs.
3 Coyle et al. (2005) compared average annual age-adjusted breast cancer rate for counties
4 reporting a release to that rate for non-reporting counties using Mann-Whitney U test.
5 Additionally, multiple linear regression analyses was used to determine the association of the
6 average annual age-adjusted breast cancer rates with the 12 HAPs, adjusting for race and
7 ethnicity when associated with the study's outcome variable.
8 While this study provides insight on cancer rates in studied population, TCE and other
9 hazardous air pollutant exposures are poorly defined and the exposure surrogate unable to
10 distinguish subjects more with higher exposure potential from those with low or minimal
11 exposure potential. Some information may be provided through examination of inter-county
12 release rates; however, no information is provided by Coyle et al. (2005). Furthermore, the
13 ecologic design of the study does not address residential history or other information on an
14 individual-subject level and is subject to bias from "ecologic fallacy" or improper inference
15 about individual-level associations based on aggregate-level analysis. Overall, this study is not
16 able to identify risk factors (etiologic exposures), has low sensitivity for examining TCE, and
17 provides little weight in an overall weight of evidence evaluation of TCE and cancer.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-307 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Coyle YM, Hynan LS, Euhus DM, Minhajuddin ATM. 2005. An ecological study of the association of environmental
chemicals on breast cancer incidence in Texas. Breast Cancer Res Treat.92:107-114.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Hypothesis of this study
air pollutants (HAPs).
Cases are incident breast
(1995-2000) in subjects
was to evaluate breast risks in Texas
cancers in males and females over a
residing in Texas and reported to the
counties and hazardous
5-yr period
Texas Cancer Registry.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Residence in Texas county as time of diagnosis is exposure surrogate. Annual
release by county of 12 HAPs (carbon tetrachloride, formaldehyde, methylene
chloride, styrene, perchloroethylene, TCE, arsenic, cadmium, chromium, cobalt,
copper, and nickel) are obtained from EPA's TRI database.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £3'
I
I
§
^.
£5'
1
TO'
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
54,487 incident breast cancer cases in males and females.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and race/ethnicity.
Mann-Whitney U test (nonparametric) to compared average
breast cancer rate between counties reported HAP release to
counties.
Linear logistic regression
annual age-adjusted
that for non-reporting
No.
Yes.
td
EPA = Environmental Protection Agency. HAP = hazardous air pollutant. TRI = Toxic Release Inventory.
H I
O
H
W
-------
1 B.3.3.2. Morgan and Cassady (2002)
2 B.3.3.2.1. Author's abstract.
3
4 In response to concerns about cancer stemming from drinking water contaminated
5 with ammonium perchlorate and trichloroethylene, we assessed observed and
6 expected numbers of new cancer cases for all sites combined and 16 cancer types
7 in a California community (1988 to 1998). The numbers of observed cancer cases
8 divided by expected numbers defined standardized incidence ratios (SIRs) and
9 99% confidence intervals (CI). No significant differences between observed and
10 expected numbers were found for all cancers (SIR, 0.97; 99% CI, 0.93 to 1.02),
11 thyroid cancer (SIR, 1.00; 99% CI, 0.63 to 1.47), or 11 other cancer types.
12 Significantly fewer cases were observed than expected for cancer of the lung and
13 bronchus (SIR, 0.71; 99% CI, 0.61 to 0.81) and the colon and rectum (SIR, 0.86;
14 0.74 to 0.99), whereas more cases were observed for uterine cancer (SIR, 1.35;
15 99% CI, 1.06 to 1.70) and skin melanoma (SIR, 1.42; 99% CI, 1.13 to 1.77).
16 These findings did not identify a generalized cancer excess or thyroid cancer
17 excess in this community.
18
19 B.3.3.2.2. Study description and comment. Residential address in 13 census tracts in
20 Redlands (San Bernardino County, CA) at time of cancer diagnosis was the exposure surrogate
21 in this ecologic study of cancer incidence over a 10-year period (1988-1998). Seventeen cancers
22 in adults (all cancers, bladder, brain and other nervous system, breast [females only], cervix,
23 colon and rectum, Hodgkin lymphoma, kidney and renal pelvis, leukemia [all], liver and bile
24 duct, lung and bronchus, NHL, melanoma, ovary, prostate, thyroid and uterus) and 3 site-specific
25 incident cancers in children under 15 years of age (leukemia [all], brain/CNS, and thyroid) were
26 identified from the Desert Sierra Cancer Surveillance Program, a regional cancer registry
27 reporting to the California Cancer Registry, with expected numbers of site-specific cancer using
28 age-race annual site-specific cancer incidence rates between 1988 and 1992 to 1990
29 census-reported information on population size and demographics. The use of the Desert Sierra
30 Cancer Surveillance Program rates which include the studied population would inflate the
31 number of site-specific cancer expected; however, the potential magnitude of bias is likely
32 minimal given the Redlands populations was estimated as 2% of the total population of the
33 regional cancer registries ascertainment area (Morgan and Cassidy, 2002). This is a
34 record-based study and information on personal habits and potential risk factors other than race,
35 sex, and age are lacking for individual subjects.
36 Morgan and Cassidy (2002) identified TCE and perchlorate from drinking water as
37 exposures of interest. Limited monitoring data from the 1,980 identified TCE concentrations in
38 Redlands wells as between 0.09 and 97 ppb TCE and drinking water concentrations as below the
39 maximum contaminant level (MCL; 5 ppb) since 1991. The paper lacks information if water
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-310 DRAFT—DO NOT CITE OR QUOTE
-------
1 monitoring represented wells in the 13-census tract study area. Furthermore, the paper does not
2 include information on water treatment and distribution networks to provide an estimate of TCE
3 concentration in finished tap water to individual homes. These authors noted their inability to
4 identify higher or lower exposed subjects, as well, as minimally exposed subjects as a source of
5 uncertainty. No data are presented on perchlorate concentrations in well or drinking water. The
6 assumption of residence in 13 census tracts is insufficient as a surrogate of potential exposure to
7 TCE and perchlorate in the absence of exposure modeling and data on water distribution
8 patterns. Exposure misclassification bias is highly likely and of a nondifferential nature which
9 would dampen observed associations.
10 While this study provides insight on cancer rates in studied population, TCE exposure is
11 poorly defined and the exposure surrogate unable to distinguish subjects more with higher
12 exposure potential from those with low or minimal exposure potential. Furthermore, the
13 ecologic design of the study does not address residential history or other information on an
14 individual-subject level and is subject to bias from "ecologic fallacy" or improper inference
15 about individual-level associations based on aggregate-level analysis. Morgan and Cassidy
16 (2002) furthermore discuss the relatively high education and income levels in the Redlands
17 population compared with the average for the referent population may lead to lower tobacco use
18 and higher than average access to health care, biases that would dampen risks for lung and other
19 tobacco-related cancers, but may also increase risks for colon and cervical cancers. Overall, this
20 study is not able to identify risk factors (etiologic exposures), has low sensitivity for examining
21 TCE, and provides little weight in an overall weight of evidence evaluation of TCE and cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-311 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Co
Dl
TO
Co
o §
H I
O >
HH Oq
H ^
O
Morgan JW, Cassady RE. 2002. Community cancer assessment in response to long-time exposure to perchlorate and
trichloroethylene in drinking water. J Occup Environ Med 44:616-621.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Hypothesis of this study was to evaluate cancer risks in a California community, not
to evaluate TCE and cancer explicitly.
Cases are incident cancers over a 10-yr period (1988-1989) in subjects residing in
13 Redlands (CA) census tracts at time of diagnosis. 17 site-specific cancers are
identified in adults and 3 site-specific cancers in children less than 15 yrs old.
Cancer cases identified from Desert Sierra Cancer Surveillance Program (DSCSP), a
regional cancer registry.
Annual age-race-site specific cancer rates from DSCSP for 1988 and 1992 and
age-race-sex specific population estimates for 1990.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Residence in a 13-census tract area of Redlands, CA is exposure surrogate. No data
are presented on TCE or perchlorate concentrations in treated drinking water supplied
to residents.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
3,098 incident cancers, the largest number from 536 breast cancer
from Hodgkin disease.
and fewest number
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and race/ethnicity.
SIR with indirect standardization of estimated expected numbers of site-specific
cancers adjusted for population growth; 90% confidence intervals presented in tables.
No.
Yes.
i
^§
SIR = standardized incidence ratio.
H I
O
H
W
-------
1 B.3.3.3. Cohn et al (1994)
2 B.3.3.3.1. Author's abstract.
3
4 A study of drinking water contamination and leukemia and non-Hodgkin's
5 lymphoma (NHL) incidence (1979-1987) was conducted in a 75-town study area.
6 Comparing incidence in towns in the highest trichloroethylene (TCE) stratum (>5
7 microg/L) to towns without detectable TCE yielded an age-adjusted rate ratio
8 (RR) for total leukemia among females of 1.43 (95% CI 1.07-1.90). For females
9 under 20 years old, the RR for acute lymphocytic leukemia was 3.26 (95% CI
10 1.27-8.15). Elevated RRs were observed for chronic myelogenous leukemia
11 among females and for chronic lymphocytic leukemia among males and females.
12 NHL incidence among women was also associated with the highest TCE stratum
13 (RR = 1.36; 95% CI 1.08-1.70). For diffuse large cell NHL and non-Burkitt's
14 high-grade NHL among females, the RRs were 1.66 (95% CI 1.07-2.59) and 3.17
15 (95% CI 1.23-8.18), respectively, and 1.59 (95% CI 1.04-2.43) and 1.92 (95% CI
16 0.54-6.81), respectively, among males. Perchloroethylene (PCE) was associated
17 with incidence of non-Burkitt's high-grade NHL among females, but collinearity
18 with TCE made it difficult to assess relative influences. The results suggest a link
19 between TCE/PCE and leukemia/NHL incidence. However, the conclusions are
20 limited by potential misclassification of exposure due to lack of individual
21 information on long-term residence, water consumption, and inhalation of
22 volatilized compounds.
23
24 B.3.3.3.2. Study description and comment. This expanded study of a previous analysis of
25 TCE and perchloroethylene in drinking water in a 27-town study area (Fagliano et al., 1990)
26 examined leukemia and NHL incidence from 1979 to 1987 in residents and TCE and other
27 VOCs in drinking water delivered to 75 municipalities. Exposure estimates were developed
28 from data generated by a mandatory monitoring program for four trihalomethane chemicals and
29 14 other volatile organic chemicals in 1984-1985 for public water supplies and from historical
30 monitoring data conducted in 1978-1984 by the New Jersey Department of Environmental
31 Protection and Energy and the New Jersey Department of Health, which was the mean of
32 monthly averages for this period. The average and maximum concentration of TCE and other
33 chemicals were estimated by considering together, for the period prior to 1985, details of the
34 distribution system size, well or surface water use, patterns of water purchases among systems,
35 and significant changes in water supply, and for years after 1985, samples of finished water from
36 the plant and samples taken from the distribution system under the assumption of homogeneous
37 mixing. The number of distribution system samples for each supply varied from 2 to 50.
38 Additionally, a dilution factor assuming complete mixing was used to adjust for water purchased
39 from another source. A single summary average and maximum concentration for each
40 contaminate for a municipality was assigned to all cases residing in that municipality at the time
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-314 DRAFT—DO NOT CITE OR QUOTE
-------
1 of cancer diagnosis. Concentrations of TCE and perchloroethylene were highly correlated
2 (r = 0.63). A ranking of municipalities was the same when using average or maximum
3 concentration and the maximum concentration of TCE or perchloroethylene used in statistical
4 analyses was grouped into three strata: <0.1 ppb (referent group), 0.1-5 ppb, >5-20 ppb, and
5 >20 ppb.
6 Incident cases of NHL and forms of leukemia reported to the New Jersey State Cancer
7 Registry were identified from 1979 and 1987. Incidence rate ratios were estimated using Poisson
8 regression models fitted to age- and sex-specific numbers of cases by exposure strata and the
9 stratum-specific population. Statistical treatment considered exposure to other drinking water
10 contaminants, atmospheric emissions of hazardous air pollutants as reported to U.S. EPA's
11 Toxics Release Inventory (TRI) by municipality and two socioeconomic variables measured as
12 municipal—average annual household income and percentage of high school graduates. None of
13 the water trihalomethane or volatile organic contaminants other than perchloroethylene was
14 shown to be associated with childhood leukemia or adult lymphomas. Furthermore, neither
15 average income, education, nor TRI release data were associated with NHL or leukemia except
16 in one exception, TRI release was shown to modify the effects of TCE and high-grade
17 non-Burkett's lymphoma in females.
18 This ecological study is subject to known biases and confounding as introduced through
19 its study design (NRC, 1997). Exposure estimates are crude (averages), do not consider
20 individual differences in drinking water patterns, and assigns group exposure levels to all
21 subjects without consideration of residential history. Potential for misclassification bias is likely
22 great in this study as is the potential for bias. This study does attempt to examine three possible
23 confounding exposures, although these are crudely defined, and some potential for residual
24 confounding is possible given the study's use of aggregated data.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-315 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Co
§
Dl
Co
o §
H I
O >
HH Oq
H TO
o
Cohn P, Klotz J, Bove F, Berkowitz M, Fagliano J. 1994. Drinking water contamination and the incidence of leukemia and
non-Hodgkin's lymphoma. Environ Health Perspect 102:556-561.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This study was designed to further examine drinking water contaminates and
lymphoma; a previous study of TCE and perchloroethylene in drinking water found a
statistically significant association with leukemia among females residing in a
27-town study area (Fagliano et al., 1990).
Incident cases of various forms of leukemia (all leukemia, acute lymphocytic, chronic
lymphocytic, acute myelogenous, chronic myelogenous, other specified and
unspecified leukemia) and NHL (total, low-grade, intermediate-grade [total and
diffuse large cell a B-cell lymphoma], high-grade including non-Burkett's
lymphoma) from 1979-1987 are identified from New Jersey State Cancer Registry.
Subjects grouped in lowest exposure category are referents.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
§
i
H 3
§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Average and maximum concentration of TCE and other chemicals were estimated by
considering together, for the period prior to 1985, details of the distribution system
size, well or surface water use, patterns of water purchases among systems, and
significant changes in water supply, and for years after 1985, samples of finished
water from the plant and samples taken from the distribution system under the
assumption of homogeneous mixing. No difference in municipality ranking by
average or maximum concentration.
Three grouped categories of maximum concentration in statistical analysis are
<0.1 ppb (referent), 0.1-5 ppb, >5 ppb (U.S. EPA Maximum Contaminant Level for
TCE and perchloroethylene).
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1,190 leukemia cases (663 males, 527 females), 119 cases assigned >5.0 ppb TCE.
1,658 NHL cases (841 males, 817 females), 165 cases assigned >5.0 ppb TCE.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Age and sex.
Poisson regression fitted to the age-and sex-specific count of cases in towns grouped
by exposure strata and weighted by the logarithm of the strata-specific population.
-------
This docu
10/20/09
Exposure-response analysis presented in
published paper
Documentation of results
Yes.
Yes.
Co
1
TO'
i
^§
H I
O
H
W
-------
1 B.3.3.4. Vartiainen et al (1993)
2 B.3.3.4.1. Author's abstract.
O
4 Concentrations up to 212 ug/1 of trichloroethene (TCE) and 180 ug/1 of
5 tetrachloroethene (TeCE) were found in the drinking water from two villages in
6 Finland. To evaluate a possible exposure, urine sample fro m95 and 21
7 inhabitants in these villages and from two control groups of 45 and 15 volunteers
8 were collected. Dichloroacetic acid (DCA) and trichloroacetic acid (TCA), the
9 metabolites of TCE and TeCE, were also analyzed. The individuals using
10 contaminated water in one of the villages excreted TCE an average 19 ug/d (<1 -
11 110 ug/d) and in the other 7.9 ug/d (<1 - 50 ug/d), while the controls excreted an
12 average 2.0 ug/d (<1 - 6.4 ug/d) or 4.0 ug/d (<1 - 13 ug/d). No increased
13 incidence rates were found in the municipalities in question for total cancer, liver
14 cancer, non-Hodgkin's lymphomas, Hodgkin's disease, multiple myeloma, or
15 leukemia.
16
17 B.3.3.4.2. Study description and comment. This published study of two separate analyses,
18 (1) urinary biomonitoring of 106 subjects from two Finish municipalities, Hausjarvi and Hattula,
19 and, (2) calculation of total cancer and site-specific cancer incidence between 1953 and 1991 in
20 Hausjarvi and Hattula residents. Limited exposure monitoring data are presented in the paper.
21 TCE concentrations in drinking water from Oitti are lacking other than noting TCE and
22 perchloroethylene were 100-200 ug/L in 1992. TCE concentrations in drinking water from
23 Hattula were below 10 ug/L in December 1991; however, samples (number unknown) taken
24 6 months later contained 212 ug/L and 66 ug/L TCE. These two municipalities discontinued use
25 of these sources for drinking water in August 1992.
26 Cancer incidence for 6 sites (all cancers, liver cancer, NHL, Hodgkin's lymphoma,
27 multiple myeloma, and leukemia) between 1953-1991 in Hausjarvi and Hattula residents was
28 obtained from the Finnish Cancer Registry. A total of 1,934 cancers were observed during the
29 study period. Standardized incidence ratios for each municipality were calculated using
30 site-specific cancer incidence rates from the Finnish population for the entire time period and for
31 3 shorter periods, 1953-1971, 1972-1981, and 1982-1991. The paper does not identity the
32 source for or size of Hausjarvi and Hattula population estimates and if temporal changes in
33 population estimates were considered in the statistical analysis. This study using record systems
34 did not include information obtained directly from subjects and lacks information on personal
35 and lifestyle factors that may introduce bias or confounding.
36 This study provides little information in an overall weight-of-evidence analysis on cancer
37 risks and TCE exposure. A major limitation is its lack of exposure assessment to TCE and
38 perchloroethylene. While this study provides some information on cancer incidence in the two
39 towns over a 40-year period, this study is not able to identify potential risk factors and exposures.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-319 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
^§
"*4-
o§
H I
O >
HH Oq
H TO
O
H
W
Vartiainen T, Pukkala E, Rienoja T, Strandman T, Kaksonen K. 1993. Population exposure to tri- and tetrachloroethene and
cancer risk: two cases of drinking water pollution. Chemosphere 27:1171-1181.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Study aim was (1) to determine if residents of two villages in Finland had exposure
to TCE and perchloroethylene as indicated from urinary biomonitoring, (2) identify
biomarker for low-level exposure, and (3) to determine cancer incidence in Hausjarvi
and Hattula, two municipalities in Finland. This study could not identify potential
risk factors.
Cancer incidence cases identified from Finnish Cancer Registry.
Site-specific cancer rates for the Finnish population was used a referent.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Residence in two municipalities
paper lacks exposure assessment
Hausjarvi and Hattula.
is the exposure surrogate in this ecologic study. The
to TCE and perchloroethylene in drinking water in
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
3,846 cancer cases; 1,942
from Hausjarvi and 1,904 from Hattula.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and sex.
SIR with cancer incidence
rates in Finnish population as referent.
No.
Cancer incidence analysis
is not well documented.
SIR = standardized incidence ratio.
i
^§
H I
O
H
W
-------
1 B.3.3.5. Mattin (1990)
2 B.3.3.5.1. Author's abstract.
O
4 Cancer maps from 1950 through 1979 revealed areas of high mortality from
5 bladder cancer for both males and females in several northwestern Illinois
6 counties. In order to further explore this excess, a bladder cancer incidence study
7 was conducted in the eight counties comprising this region. Eligible cases were
8 those first diagnosed with bladder cancer between 1978 and 1985. Age adjusted
9 standardized incidence ratios were calculated for each county and for 97 zip codes
10 within these counties. County results revealed no excesses. Zip code results
11 indicated elevated risks in a few areas, but only two zip codes had significantly
12 elevated results. One of these zip codes had a significant excess in males
13 (standardized incidence ratio = 1.5) and females (standardized incidence ratio =
14 1.9). This excess was primarily confined to one town in this zip code, in which
15 standardized incidence ratios were significantly elevated in males (1.7) and
16 females (2.6). Further investigation revealed that one of four public drinking
17 water wells in this town had been closed due to contamination; two wells were
18 within a half mile (0.8 km) of a landfill site that had ceased operating in 1972.
19 Tests of these two wells revealed traces of trichloroethylene, tetrachloroethylene,
20 and other solvents. Further investigation of this cluster is discussed.
21
22 B.3.3.5.2. Study description and comment. This ecologic study of bladder cancer incidence
23 and mortality among white residents in nine Illinois counties between 1978-1985 was carried
24 out to further investigate a previous finding of elevated bladder cancer mortality rates in some
25 counties. The study lacks exposure assessment to subjects and potential sources of exposure was
26 examined in apost hoc manner in one case only, for a community with an observed elevated
27 bladder cancer incidence. The limited exposure examination focused on groundwater
28 contamination and proximity of Superfund sites to the community, lacked assignment of
29 exposure surrogates to individual study subjects, and findings are difficult to interpret given the
30 lack of exposure assessment for the other eight counties.
31 Histologically-confirmed incident bladder cancer cases were identified from hospital
32 records in eight of the nine counties. Since the 9-county area bordered on neighboring states of
33 Wisconsin and Iowa, incident bladder cancer cases were also ascertained from the Wisconsin
34 Cancer Reporting System and Iowa's State Health Registry. No information is provided in the
35 paper on completeness of ascertainment of bladder cancer cases among residents or on the source
36 for identifying bladder cancer deaths. Expected numbers of incident cancers calculated using
37 age-specific rates for white males and females from the SEER program (incidence) or the United
38 States population [mortality], and the census data on population estimates for the nine-county
39 area. Statistical analyses adopt indirect standardization methods to calculate SMR and
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-322 DRAFT—DO NOT CITE OR QUOTE
-------
1 standardized incidence ratios (SIRs) for a community and SIRs for individual postal zip codes.
2 The use of records and absence of information collected from subject personal interviews
3 precluded examination of possible confounders other than age and race.
4 This ecological study is subject to known biases and confounding as introduced through
5 its study design (NRC, 1997). Ecological studies like this study are subject to bias known as
6 "ecological fallacy" since variables of exposure and outcome measured on an aggregate level
7 may not represent association at the individual level. Consideration of this bias is important for
8 diseases with more than one risk factor, such as the site-specific cancers evaluated in this
9 assessment. Lack of information on smoking is another uncertainty. While this study provides
10 insight on bladder cancer rates in the studied communities, it does not provide any evidence on
11 cancer and TCE exposure. For this reason, this study provides little weight in an overall
12 weight-of-evidence analysis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-323 DRAFT—DO NOT CITE OR QUOTE
-------
Mallin K. 1990. Investigation of a bladder cancer cluster in Northwestern Illinois. Amer J Epidemiol 132:896-8106.
to
vo £;•
I
I
§
***.
£3'
1
TO'
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
The hypothesis of study was to "further exposure a previous finding of bladder
cancer excess in several northwestern Illinois counties." (from abstract).
Incident cancer cases diagnosed between 1978-1985 were identified in residents in
9 northwestern Illinois counties from the Illinois Cancer Registry, the Wisconsin
Cancer Reporting System or the Iowa State Health Registry. Source for deaths in
subjects residing at the time of death in the 9 counties was not identified in the
published paper.
Expected number of bladder cancer derived using (1) SEER age-race-sex specific
incidence rates and (2) age-race-sex specific mortality rates of the U.S. population for
1978-1981 and for 1982-1985 and census estimates of population for each county or
postal zip code area.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence and mortality.
Not identified in paper.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
This is a health survey and lacks exposure assessment to communities and to
individual subjects. Monitoring of volatile organic chemicals including
trichloroethylene in two municipal drinking water wells for 1982-1988 in a
community with elevated bladder cancer rates was identified in paper; TCE
concentrations were less than 15 ppb. It is not know whether monitoring data are
representative of exposure to study subjects.
-
i
H 3
§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
712 bladder cancer incident cases and 222 bladder cancer deaths among white males
and female residents in nine northwestern Illinois counties.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and sex .
SIR with cancer incidence rates from Surveillance, Epidemiology and End Results
program and mortality rates of U.S. population as referents.
No.
Yes.
to
i
^§
H I
O
H
W
-------
1 B.3.3.6. Issacson et al (1985)
2 B.3.3.6.1. Author's abstract.
O
4 With data from the Iowa Cancer Registry, age-adjusted sex-specific cancer
5 incidence rates for the years 1969-1981 were determined for towns with a
6 population of 1,000-10,000 and a public water supply from a single stable ground
7 source. These rates were related to levels of volatile organic compounds and
8 metals found in the finished drinking water of these towns in the spring of 1979.
9 Results showed association between 1,2 dichloroethane and cancers of the colon
10 and rectum and between nickel and cancers of the bladder and lung. The effects
11 were most clearly seen in males. These associations were independent of other
12 water quality and treatment variables and were not explained by occupational or
13 other sociodemographic features including smoking. Because of the low levels of
14 the metals and organics, the authors suggest that they are not causal factors, but
15 rather indicators of possible anthropogenic contamination of other types. The data
16 suggest that water quality variables other than chlorination and trihalomethanes
17 deserve further consideration as to their role in the development of human cancer.
18
19 B.3.3.6.2. Study description and comment. This ecologic study of cancer incidence at six
20 sites [bladder, breast, colon, lung, prostate, rectum] and chlorinated drinking water uses
21 monitoring data from finished public drinking water supplies to infer exposure to residents of
22 Iowa towns of 1,000-10,000 population sizes. Towns were included if they received water from
23 a single major source (surface water, wells of <150 feet depth, or wells >50 feet depth) prior to
24 1965. Water monitoring for VOCs, trace elements and heavy metals was carried in Spring,
25 1979, as part of a larger nation-wide collaborative study of bladder cancer and artificial
26 sweeteners (Hoover and Strasser, 1980), and samples analyzed using proton-induced x-ray
27 emission for trihalomethanes, TCE, perchloroethylene, 1,2-dichloroethane, 1,1,1-trichloroethane,
28 carbon tetrachloride, 1,2-dichloroethylene, and 43 inorganic elements. 1,1,1-trichloroethane was
29 the most frequently detected VOC in both surface and groundwater; TCE, perchloroethylene,
30 and 1,2-dichloroethane were more frequently detected in shallow wells than in deep (>150 feet)
31 wells.
32 Cancer incidence was obtained for the period 1969 and 1981 with age-adjusted
33 site-specific cancer incidence rates for males and females calculated separately for four VOCs
34 (1,2-dichloroethane, TCE, perchloroethylene, and 1,1,1-trichloroethane) in finished groundwater
35 supplies using the direct standardization method. Using the address at the time of diagnosis,
36 each cancer patient was classified into one of two groups: (1) residing within the city limits and,
37 thus, drinking the municipality's water, or (2) residing outside the city limits and consuming
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-326 DRAFT—DO NOT CITE OR QUOTE
-------
1 water from a private source. Age-adjusted incidence rates are reported by group study town into
2 two TCE water concentrations categories of <0.15 ug/L and >0.15 ug/L.
3 This ecological study on drinking water exposure and cancer provides little information
4 in a weight-of-evidence analysis of TCE and cancer. Exposure estimates are crude (averages),
5 do not consider individual differences in drinking water patterns or other sources of exposure,
6 and assigns group exposure levels to all subjects. Potential for misclassification bias is likely
7 great in this study, likely of a nondifferential nature, and dampen observations.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-327 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Isacson P, Bean JA, Splinter R, Olson DB, Kohler J. 1985. Drinking water and cancer incidence in Iowa. III. Association of
cancer with indices of contamination. Amer J Epidemiol 121:856-869.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This ecological study was designed to examine consistency with the hypothesis of an
association between cancer and chlorinated water through examination of other water
contaminants besides water chlorination by-products and trihalomethanes.
Subjects are incident cases of cancer of the bladder, breast, prostate, lung rectum, and
stomach reported to the Iowa Cancer Registry between 1969 and 1981 and, who
resided in towns with a 1970 population of 1,000-10,000 and a public drinking water
supply coming solely from a single major source (wells) prior to 1965.
Age-adjusted site-specific incidence rates are calculated using the direct method and
the 1970 Iowa population.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
Not identified in paper.
to
00
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
vo £;•
I
I
§
***.
£3'
1
TO'
i
H I
O >
HH Oq
H TO
O
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
As part of another epidemiologic study on water chlorination and bladder cancer,
finished drinking water samples from treatment plant were collected in Iowa
municipalities with populations of 1,000 or larger in Spring 1979 and analyzed using
proton induced x-ray emission for 4 trihalomethanes (chloroform,
chlorodibromomethane, bromoform, dibromochloromethane), 7 VOCs (TCE,
perchloroethylene, 1,1,1-trichloroethane, carbon tetrachloride, 1,2-dichloroethane,
and cis- and trans- 1,2-dichloroethylene) and 43 inorganic elements, including metals.
The predominant contaminant was 1,1,1-trichloroethane; detectable levels of TCE
were found in approximately 20% of sampled municipalities.
Study towns were ranked into two categories of TCE in finished water, <0.15 ug/L
and >0. 15 ug/L in the statistical analysis.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
1 1,091 cancer cases of which -20% of cases resided in municipality with finished
water TCE concentration of >0.15 ug/L.
Bladder, 852 cases
Breast (female), 1,866 cases
Colon, 2,032 cases
Lung 1,828 cases
Prostate, 1,823 cases
Rectum, 824 cases
-------
to
vo £;•
I
I
§
***.
£3'
1
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and sex.
Age-adjusted site-specific mortality
method and 1970 Iowa population.
rates calculated using direct standardization
No.
Yes.
TO'
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.3.7. Studies in the Endicott Area of New York
2 A series of health statistics reviews and exposure studies have been conducted in an area
3 with a history of VOCs, including trichloroethylene, detected in municipal wells used to supply
4 drinking water to residents of Endicott, Broome County, NY. These studies were carried out by
5 staff the New York State Department of Health (NYS DOH) with support from the ATSDR.
6 Early health surveys examined cancer incidence among Broome County residents between
7 1976-1980 or 1981-1990, with focused analyses of cancer incidence among residents of
8 Endicott Village and other nearby towns, childhood leukemia in the Town of Union and possible
9 etiologic factors, and adult leukemia deaths and employment in the shoe and boot manufacturing
10 industry (Forand, 2004; NYS DOH, 2008). Two recent studies focused on cancer incidence or
11 birth outcomes among Village of Endicott residents living in a geographically defined area with
12 VOC exposure potential as documented from indoor and soil vapor monitoring (ATSDR,
13 2006a, b, 2008).
14 The Village of Endicott is a mixed residential, commercial, and industrial community
15 with a rich industrial heritage and a number of VOCs were used at industrial locations in and
16 around Endicott, as well as, having been disposed at area landfills (ATSDR, 2006b). Three wells
17 provide drinking water to the Village of Endicott: Ranney, which supplied most of the water
18 used by the Endicott Municipal Water Works since it was first placed in service in 1950; and,
19 South Street, where two wells resided. The Endicott Municipal Water Supply operates on a
20 grid-water system, neighborhoods closest to the wells are usually supplied at a greater rate from
21 nearby wells as compared to wells farther away (ATSDR, 2006b).
22 Routine monitoring of the Ranney well in the early 1980s detected VOCs at levels above
23 New York State drinking water guidelines (ATSDR, 2006b). A groundwater contaminate plume
24 northwest of the Ranney Well was found in a lower aquifer from which the municipal drinking
25 supply is drawn. Several sources were initially recognized as contributing to contamination of
26 the wellfield with a supplemental remedial investigation concluding that the Endicott Village
27 Landfill was the source of the VOCs in the Endicott Wellfield water supply (ATSDR, 2006a).
28 Groundwater water samples collected from monitoring wells installed during previous
29 investigations, wells install as part of the supplemental remedial investigation, the Purge well,
30 and the Ranney well contained many VOCs. Remediation efforts starting in the 1980s have
31 reduced contamination in this well to current MCLs. Water monitoring of the South Street wells
32 (wells 5 and 28) has been carried out for VOCs since 1980 and 1981, respectively (ATSDR,
33 2006b). Detection limits for VOCs from the South Street wells varied from 0.5-1.0 ug/L;
34 1,1-dichloroethane had the highest detection frequency, in 44% of all samples and TCE was
35 detected in 3 of 116 samples obtained between 1980 and 2004 (ATSDR, 2006b).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-331 DRAFT—DO NOT CITE OR QUOTE
-------
1 An upper aquifer with a contaminant plume containing VOCs was also identified and
2 sampling data indicated there were multiple sources of vapor contamination including a former
3 IBM facility located in the Village (U.S. EPA, 2005; NYSDEC, 2007). This groundwater
4 contaminant plume flows directly beneath the center of the Village of Endicott and serves as a
5 source of soil vapor contamination. Findings of a 2002 investigation indicated vapor migration
6 had resulted in detectable levels of contaminants in indoor air structures, including locations in
7 the Village of Endicott and Town of Union. Of soil gas and indoor air monitoring at more than
8 300 properties in an area south of the IBM Endicott facility, TCE was the most commonly found
9 contaminant in indoor air, at levels ranging from 0.18 to 140 (NYSDEC, 2007). This area is
10 identified as the Eastern study area in the health statistics review of ATSDR (2006a, 2008).
11 Other contaminants besides TCE detected in soil gas and indoor air less frequently and at lower
12 levels included tetrachloroethylene, cis-l,2-dichloroethene, 1,1,1-trichloroethane,
13 1,1-dichloroethylene, 1,1-dichloroethane, and Freon 113. Vapor-intrusion contamination was
14 also identified in a neighborhood adjacent to the Eastern area, call the Western study in the
15 health statistic review, and perchloroethylene and its degradation by-products were detected by
16 vapor monitoring. Perchloroethylene levels generally ranged from 0.1 to 3.5 ug/m3 of air
17 (ATSDR, 2006a).
18
19 B.3.3.7.1. Agency for Toxic Substances and Disease Registry (ATSDR, 2006a, 2008).
20 B.3.3.7.1.1. Agency for Toxic Substances and Disease Registry fATSDR, 2006a) executive
21 summary.
22
23 Background The New York State Department of Health (NYS DOH) conducted
24 this Health Statistics Review because of concerns about health issues associated
25 with environmental contamination in the Endicott area. Residents in the Endicott
26 area may have been exposed to volatile organic compounds (VOCs) through a
27 pathway known as soil vapor intrusion. Groundwater in the Endicott area is
28 contaminated with VOCs as a result of leaks and spills associated with local
29 industry and commercial businesses. In some areas of Endicott, VOC
30 contamination from the groundwater has contaminated the adjacent soil vapor
31 which has migrated through the soil into structures through cracks in building
32 foundations (soil vapor intrusion). Trichloroethene (TCE), tetrachloroethene
33 (PCE) and several other VOCs have been found in the soil vapor and in the indoor
34 air of some structures.
35 Conclusions This health statistics review was conducted because of concerns that
36 exposure to VOCs through vapor intrusion may lead to adverse health effects.
37 Although this type of study cannot prove whether there is a causal relationship
38 between VOC exposure in the study area and the increased risk of several health
39 outcomes observed, it does serve as a first step in providing guidance for further
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-332 DRAFT—DO NOT CITE OR QUOTE
-------
1 health studies and interventions. The elevated rates of several cancers and birth
2 outcomes observed will be evaluated further to try to identify additional risk
3 factors which may have contributed to these adverse health outcomes.
4 Limitations in the current study included limited information about the levels
5 of VOCs in individual homes, the duration of the exposure, the amount of time
6 residents spent in the home each day and the multiple exposures and exposure
7 pathways that likely existed among long term residents of the Endicott area. In
8 addition, personal information such as medical history; dietary and lifestyle
9 choices such as smoking and drinking; and occupational exposures to chemicals
10 were not examined. Future evaluations of cancer and birth defects and VOC
11 exposures in the area should take these factors into account. The small population
12 size of the study area also limited the ability to detect meaningful elevations or
13 deficits in disease rates, especially for certain rare cancers and birth outcomes.
14 This study represents the first step in a step-wise approach to addressing
15 health outcome concerns related to environmental contamination in Endicott, NY.
16 Follow-up will consist of further reviewing of the cancer and birth outcome data
17 already collected. Additional efforts will include reviewing individual case
18 records of kidney and testicular cancers, heart defects, Down syndrome and term
19 low birth weight births. In addition, we will review spontaneous fetal deaths
20 among residents of the area. The information gained, along with the results of this
21 Health Statistics Review, will be used to assess if a follow up epidemiologic study
22 is feasible. Any follow-up study should be capable of accomplishing one of two
23 goals: either to advance the scientific knowledge about the relationship between
24 VOC exposure and health outcomes; or as part of a response plan to address
25 community concerns. While not mutually exclusive, the distinction between these
26 goals must be considered when developing a follow-up approach. Any plans for
27 additional study will need to address other risk factors for these health outcomes
28 such as smoking, occupation and additional information on environmental
29 exposures. As in the past, NYS DOH will solicit input from the community.
30
31 B.3.3.7.1.2. Agency for Toxic Substances and Disease Registry (ATSDR, 2008) executive
32 summary.
33
34 This follow-up investigation was conducted to address concerns and to provide
35 more information related to elevated cancers and adverse birth outcomes
36 identified in the initial health statistics review entitled "Health Statistics Review:
37 Cancer and Birth Outcome Analysis, Endicott Area, Town of Union, Broome
38 County, New York" (ATSDR; 2006a).
39 The initial health statistics review was carried out to address concerns about
40 health issues among residents in the Endicott area who may have been exposed to
41 volatile organic compounds (VOCs) through a pathway known as soil vapor
42 intrusion. The initial health statistics review reported a significantly elevated
43 incidence of kidney and testicular cancer among residents in the Endicott area. In
44 addition, elevated rates of heart defects and low birth weight births were
45 observed. The number of term low birth weight births, a subset of low birth
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-333 DRAFT—DO NOT CITE OR QUOTE
-------
1 weight births, and the number of small for gestational age (SGA) births were also
2 significantly higher than expected.
3 The purpose of this follow-up investigation was to gather more information
4 and conduct a qualitative examination of medical and other records of individuals
5 identified with adverse birth outcomes and cancers found to be significantly
6 elevated. Quantitative analyses were also carried out for two additional birth
7 outcomes, conotruncal heart defects (specific defects of the heart's outflow
8 region), and spontaneous fetal deaths (stillbirths), and for cancer incidence
9 accounting for race.
10
11 Cancer Incidence Adjusting for Race: Because a higher percentage of the
12 population in the study area was white compared to the comparison population,
13 we examined the incidence of cancer among whites in the study area compared to
14 the incidence in the white population of New York State, excluding New York
15 City. Cancer incidence among whites was evaluated for the years 1980-2001.
16 Results: Limiting the analysis of cancer to only white individuals had little effect
17 on overall cancer rates or standardized incidence ratios compared to those of the
18 entire study area population analyzed previously. The only difference was the
19 lung cancer which had been borderline non-significantly elevated was not
20 borderline significantly elevated.
21
22 Cancer Case Record Review: We reviewed medical and other records of
23 individuals with kidney and testicular cancers to try to determine smoking,
24 occupational and residential histories. A number of preexisting data sources were
25 used including: hospital medical records; cancer registry records; death
26 certificates; newspaper obituaries; Motor Vehicle records; and city and telephone
27 directories. Results: The case record review did not reveal any unusual patterns in
28 terms of age, gender, year of diagnosis, cell type, or mortality rate among
29 individuals with kidney or testicular cancer. There was some evidence of an
30 increased prevalence of smoking among those with kidney cancer and some
31 indication that several individuals diagnosed with testicular and kidney cancer
32 may have been recent arrivals to the study area.
33
34 Conclusions/Recommendations: The purpose of the additional analyses
35 reported in the draft for public comment follow-up report was to provide
36 information on certain cancers and reproductive outcomes which were elevated in
37 the initial health statistics review. Although these additional analyses could not
38 determine whether there was a causal relationship between VOC exposures in the
39 study area and the increased risk of several health outcomes that were observed,
40 they did provide more information to help guide additional follow-up. The March
41 2007 public comment report provided a list of follow-up options for consideration
42 and stated, "Although an analytical (case-control) epidemiologic study of cancer
43 or birth defects within this community is not recommended at this time, we
44 describe several follow up options for discussion with the Endicott community. A
45 case-control study would be the preferable method for progressing with this type
46 of investigation, but the potentially exposed population in the Endicott area is too
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-334 DRAFT—DO NOT CITE OR QUOTE
-------
1 small for conducting a study that would be likely to be able to draw strong
2 conclusions about potential health risks.
3 Alternative follow-up options were discussed at meetings with Endicott
4 stakeholders and were the subject of responses to comments on the draft report.
5 From these discussions and written responses, NYS DOH has noted community
6 interest in two possible options for future activities: a health statistics review
7 based on historic outdoor air emissions modeling, and a multi-site epidemiologic
8 study examining cancer outcomes in communities across the state with VOC
9 exposures similar to Endicott. NYS DOH has considered these comments and
10 examined whether these options would be able to accomplish one of two goals:
11 either to advance the scientific knowledge about the relationship between VOC
12 exposure and health outcomes or to be part of a response plan to address
13 community concerns.
14 An additional health statistics review using historic outdoor air emission
15 modeling results to identify and study a larger population of residents potentially
16 exposed to TCE is not likely to meet either of these goals at this time. Because of
17 the limitations of the health statistics review for drawing conclusions about cause
18 and effect, conducting an additional health statistics review is not likely to
19 increase our understanding of whether exposures in the Endicott area are linked to
20 health outcomes. Limitations with the available historic outdoor air data also
21 would make it difficult to accurately define the appropriate boundaries for the
22 exposure area. ATSDR historic outdoor air emissions modeling activity was
23 unable to model TCE due to a lack of available records.
24 A multi-site epidemiologic study of health outcomes in communities across
25 the state with VOC exposures similar to Endicott offers some promise of meeting
26 the goal of advancing the scientific knowledge about the relationship between
27 VOC exposures and health outcomes. The community has indicated its preference
28 that such a study focus on cancer outcomes. Given the complex issues involved in
29 conducting such a study (e.g., tracking down cases or their next of kin after many
30 years, participants' difficulty in accurately remembering possible risk factors from
31 many years ago, and the long time period between exposure to a carcinogen and
32 the onset of cancer), we do not consider a multisite case-control study of cancer as
33 the best option at this time. An occupational cancer study is a better option than a
34 community-based study because it can better incorporate information about past
35 workplace exposures and could use corporate records to assist in finding
36 individual employees many years after exposure.
37 Heart defects have been associated with TCE exposure in other studies. Given
38 the shorter latency period, and thus the shorter time period in which other risk
39 factors could come into play, a multi-site study of heart defects has some merit as
40 a possible option. Currently, NYS DEC and NYS DOH are investigating many
41 communities around New York State which could have VOC exposure patterns
42 similar to Endicott, and thus could be included in such a multi-site epidemiologic
43 study. However, in most of these communities exposure information sufficient to
44 identify a study population is not yet available. NYS DOH will continue to
45 evaluate these areas as additional exposure information becomes available, with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-335 DRAFT—DO NOT CITE OR QUOTE
-------
1 the goal of identifying other communities for possible inclusion in a multi-site
2 epidemiologic study of heart defects.
3 NYS DOH will continue to keep the Endicott community and stakeholders
4 informed about additional information regarding other communities with
5 exposures similar to those that occurred in the Endicott area. NYS DOH staff will
6 be available as needed to keep interested Endicott area residents up-to-date on the
7 feasibility of conducting a multi-site study that includes the Endicott area.
8
9 B.3.3.7.1.3. Study description and comment. Health statistics review conducted by NYS
10 DOH because of concerns about possible exposures to VOCs in Endicott area groundwater and
11 vapor intrusion into residences examined cancer incidence between 1980 and 2001 and birth
12 outcomes among residents living in a study area defined by soil vapor sampling and exposure
13 modeling. The reviews were supported by ATSDR and conclusions presented in final reports
14 (ATSDR, 2006a, 2008) have received external comment, but the studies have not been published
15 in the open peer-reviewed literature. Testing of soil gas and indoor air of more than 300
16 properties, including 176 residences [location not identified] for VOCs detected TCE levels
17 ranging from 0.18-140 ug/m3 ; other VOCs less commonly detected included perchloroethylene,
18 1,1-dichloroethane, 1,1-dichloroethylene, 1,2-dichloroethylene, vinyl chloride,
19 1,1,1-trichloroethane, methylene chloride, and Freon 113. A model was developed to predict
20 VOC presence in soil vapor based on measured results (Sanborn Head and Associates, 2003).
21 Subsequent sampling and data collection verified this model. Initial study area boundaries were
22 determined based on the extent of the probable soil vapor contamination greater than 10 ug/m3 of
23 VOCs as defined by the model. Contour lines of modeled VOC soil vapor contamination levels,
24 known as isopleths, were mapped using a geographic information system. This study area is
25 referred to as the Eastern study area in ATSDR (2006a, 2008). Additional sampling west of the
26 initial study area identified further contamination with the contaminant in this area primarily
27 identified as perchloroethylene at levels ranging from 0.1-3.5 ug/m3 in an area referred to as the
28 Western study area (ATSDR, 2006a, 2008). The source of perchloroethylene contamination was
29 not known. A digital map of the 2000 Census block boundaries was overlaid on these areas of
30 contamination. The study areas were then composed of a series of blocks combined to conform
31 as closely to the areas of soil vapor contamination as possible.
32 Incident cancer cases for 18 sites, including cancer in children 19 years or younger,
33 between 1980 and 2001 and obtained from the New York State Cancer Registry and addresses
34 were geocoded to identify cases residing in the study area. The observed numbers of site-
35 specific cancers were compared to that expected calculated using age-sex-year specific cancer
36 incidence rates for New York State exclusive of New York City and population estimates 1980,
37 1990 and 2000 Censuses. Expected numbers of site-specific cancer did not include adjustment
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-336 DRAFT—DO NOT CITE OR QUOTE
-------
1 for race in ATSDR (2006a); however, race was examined in the 2008 follow-up study which
2 compared cancer incidence among the white residents in the study area to that of whites in New
3 York State (ATSDR, 2008). Over the 22-year period, a total of 347 incident cancers were
4 observed among residents in the study area, 339 of these were in white residents. Less than
5 6 cases of cancers in children 19 years of age or younger were identified and ATSDR (2006a)
6 did not present a SIR for this grouping, similar to their treatment of other site-specific cancers
7 with less than six observed cases.
8 The follow-up analysis by ATSDR (2008) reviewed medical records of kidney and
9 testicular cancer cases for smoking, occupational and residential histories, and restricted the
10 statistical analysis to white residents, given the few numbers of observed cancers in the small
11 population of nonwhite residents. Limiting the analysis to only white individuals in the study
12 area had little effect on overall cancer rates or SIR estimates (ATSDR, 2006a). As observed in
13 ATSDR (2006a), statistically significant excess risks were observed for kidney cancer in both
14 sexes and testicular cancer in males. In addition, lung cancer estimate risks in males and in
15 males and females were of the same magnitude in both analyses, but confidence intervals
16 excluded a risk of 1.0 in the ATSDR (2008) analyses which adjusted for race. Review of
17 medical records for the 15 kidney cancer and six testicular cancer cases provided limited
18 information about personal exposures and potential risk factors because of incomplete reporting
19 in records. The record review did not reveal any unusually patterns in either kidney cancer or
20 testicular cancer in terms of age, year of diagnosis, anatomical site, cell type, or mortality rate.
21 Occupational history suggested possible workplace chemical exposure for roughly half of the
22 13 kidney cancer cases and none of the testicular cancer cases whose medical records included
23 occupational history. For smoking, half of the 9 kidney cancer cases and some (number not
24 identified) of the 3 testicular cancer cases with such information in medical records were current
25 or former smokers; smoking habits were not reported for the other cases. Last, examination of
26 city and phone directories revealed while half the kidney cancer cases as long term Endicott
27 residents, several cases of testicular cancer were among residents who recently moved into the
28 Endicott area.
29 These health surveys are descriptive; they provide evidence of cancer rates in a
30 geographical area with some documented exposures to several VOCs including trichloroethylene
31 but are unable to identify possible etiologic factors for the observed elevations in kidney,
32 testicular, or lung cancers. The largest deficiency is the lack of exposure assessment, notably
33 historical exposure, to individual subjects. Review of city and phone directories suggests some
34 kidney and testicular cancer cases were among recently-arrived residents, a finding inconsistent
35 with a cancer latent period; however, of greater importance is the finding of cancers among
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-337 DRAFT—DO NOT CITE OR QUOTE
-------
1 subjects with long residential history. On the other hand, the population in the study areas has
2 declined over the past 20 years (ATSDR, 2006a) and residents who may have moved from the
3 study area were not included, introducing potential bias if cancer risks differed in these
4 individuals. The medical history review suggests several risk factors including smoking and
5 occupational exposure as important to kidney and testicular cancer observations. Lacking
6 information for all subjects, there is uncertainty regarding the additive effect of other potential
7 risk factors such as smoking to residential exposures. For this reason, while excesses in several
8 incident cancers are observed in these reports, potential etiological risk factors are ill-defined,
9 and the weight these studies contribute in the overall weight-of-evidence analysis is limited.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-338 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
ATSDR (Agency for Toxic Substances and Disease Registry). 2006a. Health Consultation. Cancer and Birth Outcome
Analysis, Endicott Area, Town of Union, Broome County, New York. Health Statistics Review. Atlanta, GA: U.S.
Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry. May
26, 2006.
ATSDR (Agency for Toxic Substances and Disease Registry). 2008. Health Consultation. Cancer and Birth Outcome
Analysis, Endicott Area, Town of Union, Broome County, New York. Health Statistics Review Follow-Up. Atlanta, GA: U.S.
Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry. May
15, 2008.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
This health statistics review examined incidence for 18 types of cancer in residents
living in the Village of Endicott at the time of diagnosis. This study was not
designed to identify possible etiologic factors.
Subjects are incident cases of cancer of the 18 types of cancers including childhood
cancer (all cancers in children <19 yrs of age) reported to the New York Cancer
Registry between 1980 and 2001 among residents in two areas of the Village of
Endicott, NY.
The expected number of cancer cases for the period was calculated using cancer
incidence rates for New York State exclusion of New York City and population
estimates from 1980, 1990, and 2000 Censuses.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
ICD 9th Revision.
td
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
s§
i
^§
"*4-
o§
H I
O >
HH Oq
H TO
O
H
W
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
This geographic-based study does not develop quantitative estimates of exposure,
rather study boundaries are defined using soil gas and indoor air monitoring data and
computer modeling.
Testing of soil gas and indoor air of more than 300 properties, including
176 residences (location not identified) in the Eastern study area for VOCs detected
TCE levels ranging from 0.18-140 ug/m3; other VOCs less commonly detected
included perchloroethylene, 1,1-dichloroethane, 1,1-dichloroethylene,
1,2-dichloroethylene, vinyl chloride, 1,1,1-trichloroethane, methylene chloride, and
Freon 113. A model was developed to predict VOC presence in soil vapor based on
measured results (Sanborn Head and Associates, 2003). Subsequent sampling and
data collection verified this model. Initial study area boundaries were determined
based on the extent of the probable soil vapor contamination greater than 10 ug/m3 of
VOCs as defined by the model.
Additional sampling west of the initial study area identified further contamination
with the contaminant in this area primarily identified as perchloroethylene at levels
ranging from 0.1-3.5 ug/m3 in an area referred to as the Western study area.
The study areas were then composed of a series of blocks combined to conform as
closely to the areas of soil vapor contamination as possible.
Cancer incident cases in residents at the time of diagnosis in the two areas were
included in the study.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
No information.
No information.
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Record study.
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
td
i
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
Record study.
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
347 total cancers in males and females
3,540 (1980)-3,002 (2000).
among an estimated population size of
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age and sex (ATSDR, 2006a).
Age, sex, race (ATSDR, 2008).
Medical record review of 15 kidney and 6 testicular cancer cases provided limited
information on smoking, work history, and residential history for a small percentage
of these cases (ATSDR, 2008).
No.
Yes.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
1 B.3.3.8. Studies in Arizona
2 B.3.3.8.1. Studies of West Central Phoenix Area, Maricopa County, AZ.
3 B.3.3.8.1.1. Aickin et al (19921 Aickin (2004).
4 B.3.3.8.1.1.1. Aickin et al. (1992) author's abstract.
5
6 Reports of a suspected cluster of childhood leukemia cases in West Central
7 Phoenix have led to a number of epidemiological studies in the geographical area.
8 We report here on a death certificate-based mortality study, which indicated an
9 elevated rate ratio of 1.95 during 1966-1986, using the remainder of the Phoenix
10 standard metropolitan statistical area (SMSA) as a comparison region. In the
11 process of analyzing the data from this study, a methodology for dealing with
12 denominator variability in a standardized mortality ratio was developed using a
13 simple linear Poisson model. This new approach is seen as being of general use in
14 the analysis of standardized rate ratios (SRR), as well as being particularly
15 appropriate for cluster investigations.
16
17 B.3.3.8.1.1.2. Aickin (2004) author's abstract.
18
19 BACKGROUND AND OBJECTIVES: Classical statistical inference has attained
20 a dominant position in the expression and interpretation of empirical results in
21 biomedicine. Although there have been critics of the methods of hypothesis
22 testing, significance testing (P-values), and confidence intervals, these methods
23 are used to the exclusion of all others. METHODS: An alternative metaphor and
24 inferential computation based on credibility is offered here. RESULTS: It is
25 illustrated in three datasets involving incidence rates, and its advantages over both
26 classical frequentist inference and Bayesian inference, are detailed.
27 CONCLUSION: The message is that for those who are unsatisfied with classical
28 methods but cannot make the transition to Bayesianism, there is an alternative
29 path.
30
31 B.3.3.8.1.1.3. Study description and comment. This study by staff of Arizona Department of
32 Health Services of leukemia mortality or incidence rates among children <19 years old living at
33 the time a death in West Central Phoenix in Maricopa County assume residence in the defined
34 geographical area as a surrogate of undefined exposures. Aickin et al. (1994) adopted a classical
35 statistical approach, linear Poisson regression, to estimate age-, sex- and calendar year adjusted
36 relative risks for leukemia mortality between 1966 and 1986 among children 19 years of younger
37 living in the study area at the time of death. Leukemia mortality rates for the rest of Maricopa
38 County, excluding the study area and three additional geographic areas previously identified with
39 hazardous waste contamination, were selected as the referent (Aickin et al., 1992). Aickin
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-342 DRAFT—DO NOT CITE OR QUOTE
-------
1 (2004) adopt inferential or Bayesian approaches to test whether childhood leukemia incidence
2 between 1966 and 1986 would confirm the mortality analysis observation.
3 Both studies use residence at time of diagnosis or death in the study area, West Central
4 Phoenix, AZ, as the exposure surrogate; specific exposures such as drinking water contaminates
5 are not examined nor is information on parental factors considered in the analysis. Some
6 information on potential exposures in the community-at-large may be obtained from reports
7 prepared by the AZ DHS of epidemiologic investigations of cancer mortality rates among
8 residents of this area. Aickin et al. (1992) is the published finding on childhood leukemia. Past
9 exposure to the population of West Central Phoenix to environmental contaminants has been
10 difficult to quantify because of a paucity of environmental monitoring data (Flood et al., 1990).
11 Community concerns about the environment focused on TCE found in drinking water in the late
12 1981, air pollution, from benzene emission from a nearby major gasoline storage and distribution
13 facility, and pesticide residues. Two wells that occasionally supplemented the water supply in
14 West Central Phoenix were closed after TCE was detected at the wellhead. The levels of TCE
15 measured at the time contamination was detected were 8.9 ppb and 29.0 ppb (report does not
16 identify the number of samples nor concentration ranges). The period over which contaminant
17 water had been supplied from these wells was not known nor whether significant exposure to the
18 population occurred after mixing with surface water. Other compounds identified in the
19 contaminated plume besides TCE included 1,1-dichloroethylene, trans-1,2-dichloroethylene,
20 chloroform, and chromium. The exposure assessment in the AZ DHS reports is inadequate to
21 describe exposure potential to TCE to subjects of Aickin et al. (1992) and Aickin (2004).
22 Moreover, potential etiologic factors for the observed elevated estimated relative risk for
23 childhood leukemia bases are not examined. While these studies support an inference of
24 elevated childhood leukemia rates in residents of West Central Phoenix, these studies provide
25 little information on childhood leukemia and TCE exposure and contribute little weight in the
26 overall weight-of-evidence analysis of cancer and TCE.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-343 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Aickin M, Chapin CA, Flood TJ, Englender SJ, Caldwell GG. 1992. Assessment of the spatial occurrence of childhood
leukemia mortality using standardized rate ratios with a simple linear Poisson model. Int J Epidemiol 21:649-655.
Aickin M. 2004. Bayes without priors. J Clin Epidemiol 57:4-13.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Aickin et al. (1992) illustrated a methodologic approach to reduce variability in rate
ratios from small-sized populations. Childhood leukemia mortality in a
geographically-defined area in central Phoenix, AZ, was the case study adopted to
illustrate methodologic approach. The analysis was not designed to examine possible
etiologic factors.
The purpose of Aickin (2004) "was to determine whether a 1.95 standardized
mortality ratio [19] for leukemia in West Central Phoenix (compared to the
remainder of Maricopa County) would be confirmed in an incidence study" [p. 8].
Leukemia deaths among children <19 yrs of age between the years 1966 and 1986
and with addresses on death certificates in the geographically-defined study area
were identified from Arizona death tapes.
Referent group is childhood leukemia mortality rate of all other Maricopa residents
excluding the study area and 3 other areas with identified hazardous waste
contamination (Aickin et al., 1992).
Incident cases of childhood leukemia (<19 yrs) among residents living in study area
were identified from the Arizona Cancer Registry and from cancer registry and
medical record reviews at 13 area hospitals (Flood et al., 1990).
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Cancer mortality (Aickin et al., 1992).
Cancer incidence (Aickin, 2004).
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
^.
S3'
1
TO'
td
i
H I
O >
HH Oq
H TO
O
H
W
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Mortality— ICD 7, ICDA 8, ICD 9 (Flood and Chapin, 1988).
Incidence— ICD-O.
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Residence in geographical area is a surrogate of undefined exposures; possible
exposures are not identified in the paper.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Record study.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
38 childhood leukemia deaths over a period of 21 yrs.
49 childhood leukemia incident cases over a period of 21 yrs.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and year (1966-1969, 1979-1981, 1982-1986).
Poisson regression using 1970, 1980, and 1985 population estimates from U.S.
Bureau of the Census.
No.
Yes.
-------
1 B.3.3.8.2. Studies in Tucson, Pima County, AZ.
2 B.3.3.8.2.1. Arizona Department of Health Services (AZ DHS, 1990,1995)
3 B.3.3.8.2.1.1. Arizona Department of Health Services (AZ DHS, 1990) author's summary.
4
5 In 1986, responding to community concerns about possible past exposure to low
6 levels of trichloroethylene in drinking water, a committee appointed by the
7 Director of the Arizona Department of health Services recommended that the
8 incidence of childhood leukemia and testicular cancer be studied in the population
9 residing in the Tucson Airport Area (TAA). The study reported here was
10 designed to count all cancer cases occurring in 0-19 year-old Pima County
11 residents, and all testicular cancer cases in Pima County residents of all ages,
12 during the 1970-1986 time period. Based on the incidence rates in the remainder
13 of Pima County, approximately seven cases of childhood leukemia and
14 approximately eight cases of testicular cancer would have been expected in the
15 TAA. Eleven cases of leukemia (SIR = 1.50, 95% C.I. 0.76-2.70) and six cases of
16 testicular cancer (SIR = 0.78, 95% C.I. 0.32-1.59) were observed. Statistical
17 analyses showed that the incidence rates of these cancers were not significantly
18 elevated. Additionally, it was determined that the rates of other childhood cancers
19 in the TAA, grouped as lymphoma, brain/CNS and other, were not significantly
20 elevated. The childhood leukemia, childhood cancer, and testicular cancer rates
21 in Pima County were comparable to rates in other states and cities participating in
22 the National Cancer Institute's Surveillance Epidemiology and End Results
23 Program.
24
25 B.3.3.8.2.1.2. Arizona Department of Health Services (AZ DHS, 1995) author's summary.
26
27 In 1986, responding to community concerns about possible past exposure to low
28 levels of trichloroethylene in drinking water, a committee appointed by the
29 Director of the Arizona Department of health Services recommended that the
30 incidence of childhood leukemia and testicular cancer be studied in the population
31 residing in the Tucson Airport Area (TAA). The study reported here was
32 designed to count all cancer cases occurring in 0-19 year-old Pima County
33 residents, and all testicular cancer cases in Pima County residents of all ages,
34 during the 1986-1991 time period. Based on the incidence rates in the remainder
35 of Pima County, approximately 3 cases of childhood leukemia and 4 cases of
36 testicular cancer would have been expected in the TAA. Three cases of leukemia
37 (SIR = .80; 95% C.I. 0.31-2.05) and 4 cases of testicular cancer (SIR = .93; 95%
38 C.I. 0.37-2.35) were observed. Statistical analyses showed that the incidence
39 rates of these cancers were not significantly elevated. Additionally, results
40 indicate no statistically elevated incidence rates of childhood lymphoma,
41 brain/CNS, and other childhood cancers, for ages 0-19, in the TAA. No
42 consistent pattern of disease occurrence was observed when comparing the past
43 incidence and mortality studies conducted by ADHS in the TAA with this present
44 study regarding disease categories.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-346 DRAFT—DO NOT CITE OR QUOTE
-------
1 B.3.3.8.2.1.3. Study description and comment. These reports by staff of AZ DHS of cancer
2 incidence among children <19 years old and of testicular cancer incidence among males living at
3 the time a diagnosis in 1970-1986 or 1987-1991 in the Tucson International Airport Area
4 (TAA) of southwest Tucson (AZ DHS, 1990, 1995) compared to incidence rates for the rest of
5 Pima County were conducted in response to community concerns about cancer and possible past
6 exposure to low levels of TCE in drinking water. In contrast to studies in West Central Phoenix,
7 findings from the 1990 and 1995 AZ DHS studies in Tuscon have not been published in the
8 peer-reviewed literature. Childhood cancers included were leukemia, brain/CSN, lymphoma,
9 and a broad category of all other cancers diagnosed in children <19 years old. The Arizona
10 Cancer Registry and reviews of medical records of 10 Pima county hospitals served as sources
11 for identifying incident cases. The study area was defined as a geographical area overlaying a
12 plume of contaminated groundwater and was comprised of five census tracts. The approximate
13 areas boundaries are Ajo Way (north), Los Reales Road (south), Country Club Road (east), and
14 the Santa Cruz River (west). Adjacent census tracts in Pima County were aggregated into four
15 separate study areas and incident cancer rates during the 1970-1986 time period (AZ DHS,
16 1990) or 1987-1991 (AZ DHS, 1995) of the aggregated 4-area census tract, excluding the TAA
17 area., were used to calculate expected numbers of cancers using the indirect standardization
18 method and population estimates from 1960, 1970, 1975, 1980, and 1985 (AZ DHS, 1990) or
19 1990 (AZ DHS, 1995) of the U.S. Bureau of Census. A secondary analysis of AZ DHS (1990)
20 compared the incidence rate of childhood leukemia and testicular cancer among Pima County
21 residents to that reported to the SEER for a similar time period.
22 These studies assume residence in the defined geographical area as a surrogate of
23 undefined exposures. The reports do not identify specific exposures for the individual subjects
24 and some information on exposures in the community-at-large may be obtained from Public
25 Health Assessments of the Tucson International Airport Area Superfund Site prepared by the
26 AZ DHS for the ATSDR (2000, 2001). The TAA site includes one main contaminated
27 groundwater plume with smaller areas of groundwater contamination located east of the main
28 plume. Insufficient data existed to evaluate groundwater contamination prior to 1981. Studies
29 conducted by AZ DHS in 1981-1982 showed TCE concentrations of above 5 ppb, the maximum
30 contaminate level, in the main groundwater plume with TCE detected in some municipal
31 drinking water wells at concentrations of up to 239 ppb. An ATSDR health assessment
32 conducted in 1988 indicated that soil and groundwater in the Main Plume had been contaminated
33 by chromium and volatile organic compounds such as TCE and dichloroethylene (DCE)
34 (ATSDR, 2000). Sampling of private wells from 1981 through 1994 identified both drinking and
35 irrigation private wells in and near the TIAA with TCE concentrations ranging from nondetect to
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-347 DRAFT—DO NOT CITE OR QUOTE
-------
1 120 ppb. Concentrations of other VOCs and chromium from the 1980s are not presented in the
2 ATSDR reports. Besides groundwater, areas of contaminated soil and sediment have also been
3 identified as part of the site. The "Three Hangars" area of the airport was found to contain
4 polychlorinated biphenyls in drainage areas with migration off-site into residential
5 neighborhoods (ATSDR, 2001). The exposure assessment in these studies is inadequate to
6 describe exposure to TCE. The studies provide little information on cancer risks and TCE
7 exposure and carry little weight in the overall weight-of-evidence analysis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-348 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
AZ DHS (Arizona Department of Health Services). 1990. The incidence of childhood leukemia and testicular cancer in Pima
County: 1970-1986. Prepared by the Arizona Department of Health Services, Division of Disease Prevention, Office of Risk
Assessment and Investigation, Office of Chronic Disease Epidemiology. September 17,1990.
AZ DHS (Arizona Department of Health Services). 1995. Update of the incidence of childhood leukemia and testicular cancer
in Southwest Tucson, 1987-1991. Prepared by the Arizona Department of Health Services, Office of Risk Assessment and
Investigation, Disease Prevention Services. June 6,1995.
Description
CATEGORY A: STUDY DESIGN
Clear articulation of study objectives or
hypothesis
Selection and characterization in cohort
studies of exposure and control groups and of
cases and controls in case-control studies is
adequate
Yes, from AZ DHS (1990), "1) To determine whether there was an elevated
incidence of leukemia or other cancers among children residing in the Tucson Airport
Area (TAA) and 2) To determine whether there was an elevated incidence of
testicular cancer in males in the TAA."
From AZ DHS (1995), "The objective of this study is to determine whether the
incidence rates of childhood leukemia (ages 0-19) and testicular cancer in males of
all ages were significantly elevated in the TAA when compared to the rest of Pima
County for the years 1987 through 1991."
Cases are identified from the Arizona Cancer Registry and review of medical records
at 10 Pima County hospitals. The referent is incidence rates for the remaining
population of Pima County, excluding the study area.
CATEGORY B: ENDPOINT MEASURED
Levels of health outcome assessed
Changes in diagnostic coding systems for
lymphoma, particularly non-Hodgkin's
lymphoma
Cancer incidence.
ICD-O and ICD-9 or equivalent codes from ICDA-8, ICD-7, HICDA, or SNODO.
td
i
H I
O >
HH Oq
H TO
O
H
W
-------
to
vo £;•
I
I
§
***.
£3'
1
TO'
i
^§
***.
o§
H I
O >
HH Oq
H TO
CATEGORY C: TCE-EXPOSURE CRITERIA
Exposure assessment approach, including
adoption of JEM and quantitative exposure
estimates
Residence in geographical area is a surrogate of undefined exposures; possible
exposures are not identified in the paper.
CATEGORY D: FOLLOW-UP (COHORT)
More than 10% loss to follow-up
>50% cohort with full latency
CATEGORY E: INTERVIEW TYPE
<90% face-to-face
Blinded interviewers
Record study.
CATEGORY F: PROXY RESPONDENTS
>10% proxy respondents
CATEGORY G: SAMPLE SIZE
Number of deaths in cohort mortality studies;
numbers of total cancer incidence studies;
numbers of exposed cases and prevalence of
exposure in case-control studies
AZ DHS (1990), 31 childhood cancers — 11 leukemia cases, 2 lymphoma,
3 CNS/Brain, and 15 other, and 6 testicular cancers.
AZ DHS (1995), 11 childhood cancers — 3 leukemia, 1 lymphoma, 2 CNS/Brain, and
5 other, and 4 testicular cancers.
CATEGORY H: ANALYSIS
Control for potential confounders in statistical
analysis
Statistical methods
Exposure-response analysis presented in
published paper
Documentation of results
Age, sex, and year.
SIRs calculated using indirect standardization.
No.
Yes.
O
H
W
-------
1 B.4. REFERENCES
2 AickinM. 2004. Bayes without priors. J Clin Epidemiol 57:4-13.
3 Aickin M, Chapin CA, Flood TJ, Englender SJ, Caldwell GG. 1992. Assessment of the spatial occurrence of
4 childhood leukaemia mortality using standardized rate ratios with a simple linear poisson model. Inter J Epidemiol
5 21:649-655.
6 Alexander DD, Mink PJ, Adami H-O, Chang ET, Cole P, Mandel JS, Trichopoulos D. 2007. The non-Hodgkin
7 lymphomas: a review of the epidemiologic literature. IntJ Cancer 120:1-39.
8 Anttila A, Pukkala E, Sallmen M, Hernberg S, Hemminki K. 1995. Cancer incidence among Finnish workers
9 exposed to halogenated hydrocarbons. JOccup Environ Med 37:797-806.
10 Aronson KJ, Siemiatycki J, Dewar R, Gerin M. 1996. Occupational risk factors for prostate cancer: Results from a
11 case-control study in Montreal, Canada. Am J Epidemiol 143:363-373.
12 Arp EW, Wolf PH, Checkoway H. 1983. Lymphocytic leukemia and exposure to benzene and other solvents in the
13 rubber industry. J Occup Med 25: 598-602.
14 ATSDR (Agency for Toxic Substances and Disease Registry). 1997. Toxicological Profile for Trichloroethylene
15 (TCE). U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and
16 Disease Registry. September 1997.
17 ATSDR (Agency for Toxic Substances and Disease Registry). 2000. Public Health Assessment. Tucson
18 International Airport Area, Tucson, Pima County, Arizona. Prepared by the Arizona Department of Health Services,
19 Office of Environmental Health, Environmental Health Consultation Services under a Cooperative Agreement with
20 the Agency for Toxic Substances and Disease Registry. November 13, 2000.
21 ATSDR (Agency for Toxic Substances and Disease Registry). 2001. Public Health Assessment. Tucson
22 International Airport Area, Tucson, Pima County, Arizona. Prepared by the Arizona Department of Health Services,
23 Office of Environmental Health, Environmental Health Consultation Services under a Cooperative Agreement with
24 the Agency for Toxic Substances and Disease Registry. August 27, 2001.
25 ATSDR (Agency for Toxic Substances and Disease Registry). 2003. Health Consultation. View-Master Factory
26 Supply Well (a.k.a. Mattel Portland Operations) Beaverton, Washington County, Oregon. Prepared by Oregon
27 Department of Human Services under Cooperative Agreement with the Agency for Toxic Substances and Disease
28 Registry. October 20, 2003.
29 ATSDR (Agency for Toxic Substances and Disease Registry). 2004. Feasibility investigation of worker exposure
30 to trichloroethylene at the View-Master Factory in Beaverton, Oregon. Final Report. Submitted by Environmental
31 and Occupational Epidemiology, Oregon Department of Human Services. December 2004.
32 ATSDR (Agency for Toxic Substances and Disease Registry). 2006a. Health Consultation. Cancer and Birth
33 Outcome Analysis, Endicott Area, Town of Union, Broome County, New York. Health Statistics Review. Atlanta,
34 GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and
35 Disease Registry. May 26, 2006.
36 ATSDR (Agency for Toxic Substances and Disease Registry). 2006b. Health Consultation. Public Health
37 Implications of Exposures to Low-Level Volatile Organic Compounds in Public Drinking Water. Endicott Area
38 Investigation, Broome County, New York. Atlanta, GA: U.S. Department of Health and Human Services, Public
39 Health Service, Agency for Toxic Substances and Disease Registry. November 30, 2006.
40 ATSDR (Agency for Toxic Substances and Disease Registry). 2008. Health Consultation. Cancer and Birth
41 Outcome Analysis, Endicott Area, Town of Union, Broome County, New York. Health Statistics Review Follow-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-3 51 DRAFT—DO NOT CITE OR QUOTE
-------
1 Up. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic
2 Substances and Disease Registry. May 15, 2008.
3 Axelson O, Selden A, Andersson K, Hogstedt C. 1994. Updated and expanded Swedish cohort study on
4 trichloroethylene and cancer risk. J Occup Environ 36:556-562.
5 AZ DHS (Arizona Department of Health Services). 1990. The incidence of childhood leukemia and testicular
6 cancer in Pima County: 1970-1986. Prepared for the Arizona Department of Environmental Quality under the
7 Water Quality Assurance Revolving Fund and by the Arizona Department of Health Services, Division of Disease
8 Prevention, Office of Risk Assessment and Investigation, Office of Chronic Disease Epidemiology. September 17,
9 1990.
10 AZ DHS (Arizona Department of Health Services). 1995. Update of the incidence of childhood cancers and
11 testicular cancer in Southwest Tucson, 1987 - 1991. Prepared by the Arizona Department of Health Services,
12 Office of Risk Assessment and Investigations, Disease Prevention Services Arizona Department of Health Services.
13 June, 6, 1995.
14 Beaumont JJ, Breslow NE. 1981. Power considerations in epidemiologic studies of vinyl chloride workers. AmJ
15 Epidemiol 114:725-734.
16 Becker N, Deeg E, Nieters A. 2004. Population-based study of lymphoma in Germany: rationale, study design and
17 first results. Leuk Res 28:713-724.
18 Blair A, Haas T, Prosser R, Morrissette M, Blackman, Grauman D, van Dusen P, Morgan F. 1989. Mortality
19 among United States Coast Guard marine Inspectors. Arch Environ Health44:150-156.
20 Blair A, Hartge P, Stewart PA, McAdams M, Lubin J. 1998. Mortality and cancer incidence of aircraft
21 maintenance workers exposed to trichloroethylene and other organic solvents and chemicals: extended follow up.
22 Occup Environ Med 55:161-171.
23 Blatter BM, Roeleveld N, Zeielhuis GA, Verbeek. 1997. Assessment of occupational exposure in population based
24 case-control study: comparing postal questionnaires with personal interviews. Occup Environ Med 54:54-59.
25 Bloemen L, Tomenson J. 1995. Increased incidence of renal cell tumours in a cohort of cardboard workers exposed
26 to trichloroethylene [Letter]. Arch Toxicol 70:129-130.
27 Boice JD, Marano DE, Fryzek JP, Sadler CJ, McLaughlin JK. 1999. Mortality among aircraft manufacturing
28 workers. Occup Environ Med 56:581-597.
29 Boice JD, Marano DE, Cohen SS, Mumma MT, Blott WJ, Brill AB, Fryzek JP, Henderson BE, McLaughlin JK.
30 2006a. Mortality among Rocketdyne workers who tested rocket engines, 1948-1999. J Occup Environ Med
31 48:1070-1092.
32 Boice JD, Cohen SS, Mumma MT, Dupree Ellis E, Eckerman KF, Leggett RW, Boecker BB, Brill AB, Henderson
33 BE. 2006b. Mortality among radiation workers at Rocketdyne (Atomics International), 1948-1999. RadiatRes
34 166(1 Pt 1):98-115. Erratum in: Radiat Res. 2006 166(3):566.
35 Bolm-Audorff U, Ahrens W, J6ckelK-H, MolikB, GreiserE, Timm J, etal. 1988. Experience with supplementary
36 quesionnaires in al lung cancer case-referent study. In: Hemon D, Goldberg M, editors. Proceedings of the
3 7 Workshop on Methodology of Assessment of Occupational Exposures in the Context of Epidemiological Detection
38 of Cancer Risks, Paris, 18-19 February 1988. Luxembourg: Office of the Official Publications of the European
39 Communities; 1989, p. 117-168.
40 Boyle CA, Brann EA, and the Selected Cancers Cooperative Study Group. 1992. Proxy respondents and the
41 validity of occupational and other exposure data. Am J Epidemiol 136:712-721.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-3 52 DRAFT—DO NOT CITE OR QUOTE
-------
1 Brauch H, Weirich G, Hornauer MA, Storkel S, Wohl T, BriiningT. 1999. Trichloroethylene exposure and
2 specific somatic mutations in patients with renal cell carcinoma. J Natl Cancer Inst 91:854-861.
3 Brenner, H, Gefeller O. 1993. Use of the positive predictive value to correct for disease misclassification in
4 epidemiologic studies. Am. J. Epidemiol 138:1007-1015. (As cited in NRC, 2006.)
5 Bruning T, Weirich G, Hornauer MA, Hoffer H, Brauch H. 1997. Renal cell carcinomas in trichloroethylene (TRI)
6 exposed persons are associated with somatic mutations in the von Hippel-Lindau (VHL) tumour suppressor gene.
7 Arch Toxicol 71:332.335.
8 BriiningT, PeschB, Wiesenhtitter B, Rabstein S, LammertM, Baumtiller A, Bolt H. 2003. Renal cell cancer risk
9 and occupational exposure to trichloroethylene: results of a consecutive case-control study in Arnsberg, Germany.
10 Am JIndMed 23:274-285.
11 Chang Y-M, Tai C-F, Yang S-C, Chan C-J, S Shin T-S, Lin RS, Liou S-H. 2003. A cohort mortality study of
12 workers exposed to chlorinated organic solvents in Taiwan. AnnEpidemiol 13:652-660.
13 Chang Y-M, Tai C-F, Yang S-C, Lin R, Sung F-C, Shin T-S, Liou S-H. 2005. Cancer Incidence among Workers
14 Potentially Exposed to Chlorinated Solvents in An Electronics Factory. JOccupHealth47:171-180.
15 Charbotel B, Fevotte J, Hours M, Martin J-L, Bergeret A. 2006. Case-control study on renal cell cancer and
16 occupational expsoreu to trichloroethylene. Part II: epidemiological aspects. Ann OccupHyg 50:777-787.
17 Charbotel B, Gad S, Caiola D, Beroud C, Fevotte J, et al. 2007. Trichloroethylene exposure and somatic mutations
18 of the VHL gene in patients with renal cell carcinoma. J Occup Med Toxicol 2:2-13.
19 Charbotel B, Fevotte J, martin JL, Bergeret A. 2009. Cancer du rein et expositions au trichloroethylene: les valeurs
20 limites d'exposition professionnelle fracaises en vigueur sont-elles adaptees. Rev Epidemiol Sante Publique 57:41-
21 47.
22 Checkoway H, Pearch N, Crawford-Brown DJ. 1989. Research Methods in Occupational Epidemiology,
23 Monographs in Epidemiology and Biostatistics, Volume 13. New York: Oxford University Press.
24 Cherrie, J.W., H. Kromhout, and S. Semple. 2001. The importance of reliable exposure estimates in deciding
25 whether trichloroethylene can cause kidney cancer [letter]. J Cancer Res Clin Oncol 127:400-402.
26 Clapp RW, Hoffman K. 2008. Cancer mortality in IBM Endicott plant workers, 1969-2001: an update on a NY
27 production plant. Environ Health 7:13.
28 Cohn P, Klotz J, Bove F, Berkowitz M, Fagliano J. 1994. Drinking water contamination and the incidence of
29 leukemia and non-Hodgkin's lymphoma. Environ Health Perspect 102:556-561.
30 Costa G, Merletti F, Segnan N. 1989. A mortliaty cohort study in a north Italian aircraft factory. Br JIndMed
31 46:738-743
32 Costantini AS, Miligi L, Kriebel D, Ramazzotti V, Rodella S, Scarpi E, Stagnaro E, Tumino R et al. 2001. A
33 multicenter case-control study in Italy on hematolymphopoietic neoplasms and occupation. Epidemiol 12:78-87.
34 Costantini AS, Benvenuti A, Vineis P, Kriebel D, Tumino R, Ramazzotti V, Rodella S, et al., 2008. Risk of
3 5 leukemia and multiple myeloma associated with exposure to benzene and other organic solvents: evidence from the
36 Italian multicenter case-control study. Am JIndMed 51:803-811.
37 Costas K, Knorr RS, Condon SK. 2002. A case-control study of childhood leukemia in Woburn, Massachusetts: the
3 8 relationship between leukemia incidence and exposure to public drinking water. Sci Total Environ 300:23-35.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-353 DRAFT—DO NOT CITE OR QUOTE
-------
1 Coyle YM, HynanLS, EuhusDM, Minhajuddin ATM. 2005. An ecological study of the association of
2 environmental chemicals on breast cancer incidence in Texas. Breast Cancer Res Treat.92:107-114.
3 De Roos AJ, Olshan AF, Teschke K, Poole Ch, Savitz DA, Blatt J, Bondy ML, Pollock BH. 2001. Parental
4 occupational exposure to chemicals and incidence of neuroblastoma in offspring. Am J Epidemiol 154:106-114.
5 Dewar R, Siemiatycki J, Gerin M. 1991. Loss of statistical power associated with the use of a job-epxosure matrix
6 in occupational case-control studies. Appl Occup Environ Hyg 6:508-515.
7 Dosemeci M, Wacholder S, Lubin JH. 1990. Does nondifferential misclassification of exposure always bias a tru
8 effect oward the null value? Am J Epidemiol 132:746-748.
9 Dosemeci M, CoccoP, Gomez M, Stewart PA, HeinmanEF. 1994. Effects of three features of ajob-exposure
10 matrix on risk estimates. Epidemiol 5:124-127.
11 Dosemeci M, Cocco P, Chow W-H. 1999. Gender differences in risk of renal cell carcinoma and occupational
12 exposures to chlorinated aliphatic hydrocarbons. Am JIndMed 36:54-59.
13 DuCL, WangJD. 1998. Increased morbilidyt odds raio of primary liver cancer and cirrhosis of the liver among
14 vinyl chloride monomer workers. Occup Environ Med 55:528-532.
15 Dumas S, Parent M-E, Siemiatycki J, Brisson J. 2000. Rectal cancer and occupational risk factors: ahypothesis-
16 generating, exposure-based case-control study. Int J Cancer 87:874-879.
17 Environmental Health Strategies. 1997. Final Report. Written correspondence from Paul A. Cammer, Ph.D.,
18 Trichloroethylene Issues Group, to Cheryl Siegel Scott, US Environmental Protection Agency. December 22, 1997.
19 Fagliano J, Berry M, Bove F, Burke T. 1990. Drinking water contamination and the incidence of leukemia: an
20 ecologic stuy. Am J Public Health 80:1209-1212.
21 Ferlay J., Bray F, Pisani P, Parkin DM. 2004. GLOBOCAN 2002: Cancer Incidence, Mortality and Prevalence
22 Worldwide. IARC CancerBase No. 5, Version 2.0. Lyon: IARC
23
24 Fevotte J, Charbotel B, Muller-Beaute P, Martin J-L, Hours, Bergeret A. 2006. Case-control study on renal cell
25 cancer and occupational exposure to trichloroethylene. Part I: Exxposure assessment. Ann Occup Hyg
26 50:765-775.
27 Flodin U, Fredriksson M, Persson B. 1987. Multiple myeloma and engine exhausts, fresh wood, and creosote: a
28 case-referent study. Am J Ind Med 12:519-529.
29 Flodin U, Fredriksson M, Persson B, AxelsonO. 1988. Chronic lymphatic leukaemia and engine exhausts, fresh
30 wood, and DDT: a case-referent study. Br J Ind Med 45:33-38.
31 Flood TJ, CA Chapin. 1988. Report on mortality in Maricopa County 1966-1986. Arizona Department of Health
32 Services, Division of Disease Prevention, Office of Chronic Disease Epidemiology, Office of Risk Assessment and
33 Investigation, Phoenix, AZ. August 30, 1988.
34 Flood TJ, Aickin M, Lucier JL, Petersen NJ. 1990. Incidence study of childhood cancer in Maricopa County:
35 1965-1986. Final Report. Arizona Department of Health Services, Division of Disease Prevention, Office of
36 Chronic Disease Epidemiology, Office of Risk Assessment Investigation. April 11, 1990.
37 Forand SP. 2004. Leukaemia incidence among workers in the shoe and boot manufacturing industry: a case-control
38 study. Environ Health 3:7
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-3 54 DRAFT—DO NOT CITE OR QUOTE
-------
1 Fredriksson M, Bengtsson N-O, Hardell L, Axelson O. 1989. Colon cancer, physical activity, and occupational
2 exposure. A case-control study. Cancer 63:1838-1842.
3 Fritschi L, Siemiatycki J. 1996a. Lymphoma, myeloma and occupation: Results of a case-control study. Int J
4 Cancer 67: 498-503.
5 Fritschi L, Siemiatycki J. 1996b. Melanoma and occupation: Results of a case-control study. 1996. Occup
6 Environ Med 53:168-173.
7 Fritschi L, Siemiatycki J, Richardson L. 1996. Serf-assessed versus expert-assessed occupational exposures. Am J
8 Epidemiol 144:521-527.
9 Fritschi L, Nadon L, Benke G, Lakhani R, Latreille B, Parent M-E, Siemiatycki J. 2003. Validation of expert
10 assessment of occupational exposures. Am JInd Med 43:519-522.
11 GarabrantDH, Held J, LangholzB, Bernstein L. 1988. Mortality of Aircraft Manufacturing Workers in Southern
12 California. Am JInd Med 13:683-693.
13 Goldberg MS, Parent M-E, Siemiatycki J, Desy M, Nadon L, Richardson L, Lakhani R, Lateille B, Valois M-F.
14 2001. A case-control study of the relationship between the risk of colon cancer in men and exposure to occupational
15 agents. Am JInd Med 39:5310-546.
16 Gomez MR, Cocco P, Dosemeci M, Stewart PA. 1994. Occupational exposure to chlorinated aliphatic
17 hydrocarbons: job exposure matrix. Am J Ind Med 26:171-183.
18 Green LC, Lash TL. 1999. Re: "Renal cell cancer correlated with occupational exposure to trichloroethylene"
19 [Letter] J Cancer Res Clin Oncol 125:430-432.
20 Greenland S. 1992. A semi-Bayes approach to the analysis of correlated multiple associations with an application
21 to an occupational cancer-mortality study. Stat Med 11:219-230.
22 Greenland S, Salvan A, Wegman DH, Hallock MF, Smith TH. 1994. A case-control study of cancer mortality at
23 the transformer-assembly facility. Int Arch Occup Environ Health 66:49-54.
24 Hansen J, Raaschou-Nielsen O, Christensen JM, Johansen I, McLaughlin JK, Lipworth L, Blot WJ, Olsen JH. 2001.
25 Cancer incidence among Danish workers exposed to trichloroethylene. J Occup Environ Med 43:133-139.
26 Hardell L, Eriksson M, Lenner P, Lundgren E. 1981. Malignant lymphoma and exposure to chemicals, especially
27 organic solvents, chlorophenols and phenoxy acids: a case-control study. Br J Cancer 43:169-176.
28 Hardell L, Eriksson M, Degerman A. 1994. Exposure to phenoxyacetic acids, chlorophenols, or organic solvents in
29 relation to histopathology, stage, and anatomical localization of non-Hodgkin's lymphoma. Cancer Res
30 54:2386-2389.
31 Heineman EF, Cocco P, Gomez MR, Dosemeci M, Stewart PA, Hayes RB, Zahm SH, Thomas TL, Blair A. 1994.
32 Occupational exposure to chlorinated aliphatic hydrocarbons and risk of astrocytic brain cancer. Am J Ind Med
33 26:155-169.
34 HenschlerD, Vamvakas S, LammertM, DekantW, KrausB, Thomas B, UlmK. 1995. Increased incidence of
3 5 renal cell tumors in a cohort of cardboard workers exposed to trichloroethene. Arch Toxicol 69:291-299.
36 Hoover R, Strasser PH. 1980. Artificial sweeteners and human bladder cancer. Lancet 1:837-840.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-355 DRAFT—DO NOT CITE OR QUOTE
-------
1 Hornung RW, Pinney SM, Lodwick J, Killough GG, Brewer DE, Nasuta J. 2008. Estimate of radon exposure to
2 workers at the Fernald Feed Materials Production Center 1952-1988. J Expo Sci Environ Epidemiol 18:512-523.
3 HSIA (Halogenated Solvents Industry Alliance). 2005. Case-control study on renal cell cancer and occupational
4 trichloroethylene exposure in the Arve valley (France). Final Report. October 2005. Submitted to US EPA, 8EHQ-
5 1005-16030.
6 IEI (International Epidemiology Institute). 2005. Rocketdyne Worker Health Study. Final Report. Executive
7 Summary and Appendices. July 13, 2005.
8 http://www.boeing.com/aboutus/environment/santa_susana/healthstudy.html, accessed March 12, 2008.
9 Ikeda M, Ohtsuji H, Imamura T, Kokoike Y. 1972. Urinary excretion of total trichloro-compounds,
10 trichloroethanol, and trichloroacetic acid as a measure of exposure to trichloroethylene and tetrachloroethylene. Brit
11 J IndustrMed 29: 328-333.
12 IOM (Institute of Medicine). 2003. Gulf War and Health, Vol 2. Insecticides and Solvents. Washington, DC: The
13 National Academies Press.
14 IsacsonP, Bean JA, Splinter R, Olson DB, KohlerJ. 1985. Drinking water and cancer incidence in Iowa. III.
15 Association of cancer with indices of contamination. Am J Epidemiol 121:856-869.
16 Jaffe ES, Harris ML, Stein H, Vardiman JW, eds. 2001. World Health Organization Classification of Tumours.
17 Pathology and Genetics of Tumours of Haematopoietic and Lymphoid Tissues. Lyon, France: IARC Press.
18 Kernan GJ, Ji B-T, Dosemeci M, Silverman DT, Balbus J, Zahm SH. 1999. Occupational risk factors for pancreatic
19 cancer: A case-control study based on death certificates from 24 U. S. states. Am J Ind Med 36:260-270.
20 Krishnadasan A, Kennedy N, Zhao Y, Morgenstern H, Ritz B. 2007. Nested case-control study of occupational
21 chemical exposures and prostate cancer in aerospace and radiation workers. Am J Ind Med 50:383-390.
22 Langholz B, GoldsteinL. 1996. Risk Set Sampling inEpidemiologic Cohort Studies. Stat Sci 11:35-53.
23 Lee L J-H, Chung C-W, Ma Y-C, Wang G-S, Chen P-C, Hwang Y-H, Wang J-D. 2003. Increased mortality odds
24 ratio of male liver cancer in a community contaminated by chlorinated hydrocarbons in groundwater. Occup
25 Environ Med 60:364-369.
26 Lowengart RA, Peters JM, Cicioni C, Buckley J, Bernstein L, Preston-Martin S, Rappaport E. 1987. Childhood
27 leukemia and parents' occupational and home exposures. JNCI 79:39-46.
28 Lymphoma Information Network, http://www.lymphomainfo.net/nhl/classify-older.html. Accessed April 7, 2008.
29 MallinK. 1990. Investigation of a bladder cancer cluster in Northwestern Illinois. Amer J Epidemiol
30 132:896-8106.
31 Mandel JS. 2001. Renal cell cancer correlated with occupational exposure to trichloroethylene [Letter]. J Cancer
32 Res Clin Oncol 127:265-266.
3 3 Mandel JH, Kelsh MA, Mink PJ, Alexander DD, kalmes RM, Weingart M, Yost L, Goodman M. 2006.
34 occupational trichloroethylene exposure and non-Hodgkin's lymphoma: a meta-analysis and review. Occup
3 5 Environ Med 63:597-607.
36 Marano DE, Boice DJ, Fryzek JP, Morrison JA, Sadler CJ, McLaughlin JK. 2000. Exposure assessment for a large
37 epidemiological study of aircraft manufacturing workers. Appl Occup Environ Hyg 15:644-656.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-356 DRAFT—DO NOT CITE OR QUOTE
-------
1 Massachusetts Department of Public Health (MADPH). 1997. Woburn Childhood Leukemia Follow-up Study.
2 Volumes I and II. Final Report.
3 McKinney PA, Alexander FE, Cartwright RA, Parker L. 1991. Parental occupations of children with leukaemia in
4 west Cumbria, north Humberside, and Gateshead. BMJ 302:681-687.
5 McLaughlin JK, Blot WJ. 1997. A critical review of epidemiology studies of trichloroethylene and
6 perchloroethylene and risk of renal-cell cancer. Int Arch Occup Environ Health 70:222-31.
7 McMichael AJ, Spirtas R, Gamble JF, Tousey PM. 1976. Mortality among rubber workers: relationship to specific
8 jobs. J Occup Med 18:178-185.
9 MesterB, Nieters A, DeengE, Eisner G, Becker N, Seidler A. 2006. Occupation and malignant lymphoma: a
10 population based case control study in Germany. Occup Environ Med 63:17-26.
11 Miligi L, Costantini AS, Benvenuti A, Kreibel D, Bolejack V, et al. 2006. Occupational exposure to solvents and
12 the risk of lymphomas. Epidemiol 17:552-561.
13 Morgan JW, Cassady RE. 2002. Community cancer assessment in response to long-time exposure to perchlorate
14 and trichloroethylene in drinking water. J Occup Environ Med 44:616-621.
15 Morgan RW, Kelsh MA, Zhao K, Heringer S. 1998. Mortality of aerospace workers exposure to trichloroethylene.
16 Epidemiol 9:424-431
17 Morgan RW, Kelsh MA, Zhao K, Heringer S. 2000. Mortality of aerospace workers exposed to trichloroethylene.
18 Erratum. Epidemiology 9:424-431.
19 Morgenstern H, Froines J, Ritz B, Young B. 1997. Epidemiologic Study to Determine Possible Adverse Effects to
20 Rocketdyne/Atomics Internal Workers from Exposure to Ionizing Radiation. Final Report to the Public Health
21 Institute. Subcontract No. 324A-8701-S0163. June 1997.
22 Morgenstern H, Froines J, Ritz B, Young B. 1999. Addendum Report. Epidemiologic Study to Determine Possible
23 Adverse Effects to Rocketdyne/Atomics International Workers from Exposure to Selected Chemicals. Report to the
24 Public Health Institute (Formerly the California Public Health Foundation). Subcontract No. 324A-8701-S0163.
25 January 1999. http://www.ssflpanel.org/files/UCLA_chem.pdf, accessed March 12, 2008.
26 Morton LM, Turner JJ, Cerhan JR, Linet MS, Treseler PA, et al. 2007. Proposed classification of lymphoid
27 neoplasms for epidemiologic research from the International Lymphoma Epidemiology Consortium (InterLymph).
28 Blood 110:695-708.
29 Murphy PJ. Exposure to wells G and H in Woburn, Massachusetts. Boston, MA: Massachusetts Department of
30 Public Health, 1991.
31 NIOSH (National Institute of Occupational Safety and Health). 1992. Health Hazard Evaluation Report. HETA-86-
32 469-2189. Jame River Corporation, Newnan, Georgia.
33 Nordstrom M, Hardell L, Hagberg H, Rask-Andersen A. 1998. Occupational exposures, animal exposure and
34 smoking as risk factors for hairy cell leukaemia evaluated in a case-control study. Br J Cancer 77:2048-2052.
35 Norman WC. 1996. Flawed estimate of methylene chloride exposures. Am JInd Med 30:504-509.
36 NRC (National Research Council). 1997. Environmental Epidemiology. Use of the Gray Literature and Other Data
37 in Environmental Epidemiology. National Academy Press, Washington, DC.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-357 DRAFT—DO NOT CITE OR QUOTE
-------
1 NRC (National Research Council). 2006. Assessing the Human Health Risks of Trichloroethylene. Key Scientific
2 Issues. National Academy Press, Washington, DC.
3 NRC. (National Research Council). 2009. Contaminated Water Supplies at Camp lejeune - Assessing Potential
4 Health Effects. National Academy press, Washington, DC.
5 NYSDEC (New York State Department of Environmental Conservation). 2007. Project Information Summary.
6 Environmental Investigations Endicott, New York.
7 http://www.dec.state.ny.us/website/der/projects/endicott/endicottbkgrnd.html. Accessed March 19, 2007.
8 NYSDOH(New York State Department of Health). 2006. Trichloroethene Air Criteria Document. Final Report.
9 Center for Environmental Health, Bureau of Toxic Substance Assessment. October 2006.
10 NYS DOH (New York State Department of Health). 2008. IBM Endicott Site, Endicott Area Investigations -
11 Public Health Response Plan. Appendix C: Overview of Completed Health Studies, accessed December 4, 2008,
12 http://www.health.state.ny .us/environmental/investigations/broome/health_statistics_review.htm).
13 Olshan AF, De Roos AJ, Teschke K, Neglin JP, Stram DO, Pollock BH, Castleberry RP. 1999. Neuroblastoma and
14 parental occupation. Cancer Causes Control 10:539-549.
15 Pannett B, Coggon D, Acheson ED. 1985. A job-exposure matrix for use in population based studies in England
16 and Wales. BrJIndMed 42:777-783.
17 Parent M-E, Siemiatycki J, Fritschi L. 2000a. Workplace exposures and oesophageal cancer. Occup Environ Med
18 57:325-334.
19 Parent M-E, Hua Y, Siemiatycki J. 2000b. Occupational risk factors for renal cell carcinoma in Montreal. AmJ
20 Ind Med 38:609-618.
21 Parkin D, WhelanSL, Ferlay J, Teppo L, Thomas DB, editors. 2002. Cancer Incidence in Five Continents, Vol
22 VIII. Lyon, France: IARC Scientific Publication, IARC.
23 Percy C, RiesLG, Van HoltenVD. 1990. The accuracy of liver cancer as the underlying cause of death on death
24 certificates. Public Health Reports 105:361-367.
25 PerssonB, FredriksonM. 1999. Some risk factors for non-Hodgkin's lymphoma. Int J Occup Med Environ Health
26 12:135-142.
27 PerssonB, Dahlander A-M, FredrikssonM, Brage HN, Ohlson C-G, AxelsonO. 1989. Malignant lymphomas and
28 occupational exposures. Br J Ind Med 46:516-520.
29 PerssonB, Fredriksson M, OlsenK, BoerydB, AxelsonO. 1993. Some occupational exposure as risk factors for
30 malignant lymphomas. Cancer 72:1773-1778.
31 Pesch B, Haerting J, Ranft U, Klimpet A, Oelschagel, Schill W, and the MURC Study Group. 2000a. Occupational
32 risk factors for urothelial carcinoma: agent-specific results from a case-control study in Germany. Int J Epidemiol
33 29:238-247.
34 Pesch B, Haerting J, Ranft U, Klimpet A, Oelschagel, Schill W, and the MURC Study Group. 2000b. Occupational
3 5 risk factors for renal cell carcinoma: agent-specific results from a case-control study in Germany. Int J Epidemiol
36 29:1014-1024.
37 Raaschou-Nielsen O, Hansen J, Christensen JM, Blot WJ, McLaughlin JK, Olsen JH. 2001. Urinary concentrations
38 of trichloroacetic acid in Danish workers exposed to trichloroethylene, 1947-1985. Am J Ind Med 39:320-327.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-358 DRAFT—DO NOT CITE OR QUOTE
-------
1 Raaschou-Nielsen O, Hansen J, Thomsen BL, Johansen I, Lipworth L, McLaughlin JK, Olsen JH. 2002. Exposure
2 of Danish workers to trichloroethylene, 1947-1989. Appl Occup Environ Hyg 17:693-703.
3 Raaschou-Nielsen O, Hansen J, McLaughlin JK, Kolstad H, Christensen JM, Tarone RE, Olsen JH. 2003. Cancer
4 risk among workers at Danish companies using trichloroethylene: a cohort study. Am JEpidemiol 158:1182-1192.
5 RadicanL, Blair A, Stewart P, WartenbergD. 2008. Mortality of aircraft maintenance workers exposed to
6 trichloroethylene and other hydrocarbons and chemicals: extended follow-up. J Occup Environ Med 50:1306-1319.
7 Ries LAG, Melbert D, Krapcho M, Stinchcomb DG, Howlader N, Horner MJ, Marietta A, Miller BA, Feuer EJ,
8 Altekruse SF, Lewis DR, Clegg L, Eisner MP, Reichman M, Edwards BK (eds). SEER Cancer Statistics Review,
9 1975-2005, National Cancer Institute. Bethesda, MD, http://seer.cancer.gov/csr/1975_2005/, based on November
10 2007 SEER data submission, posted to the SEER web site, 2008, accessed April 22, 2008,
11 http://seer.cancer.gov/csr/1975_2005/.
12 Ritz B. 1999a. Cancer mortality among workers exposed to chemicals during uranium processing. J Occup Environ
13 Med 41:556-566.
14 Ritz B. 1999b. Radiation exposure and cancer mortality in uranium processing workers. Epidemiol 10:531-538.
15 Ritz B, Morgenstern H, Froines J, Young BB. 1999. Effects of exposure to external ionizing radiation on cancer
16 mortality in nuclear wokers monitored for radiation at Rocketdyne/Atomics International. Am J Ind Med 35:21-31.
17 Ritz B, Morgenstern H, Crawford_Brown D, Young B. 2000. The effects of internal radiation exposure on cancer
18 mortality in nuclear workers at Rocketdyne/Atomics International. Environ Health Perspect 108:743-751.
19 Robins JM, Blevins D. 1987. Analysis of proportionate mortality data using logistic regression models. Am J
20 Epidemiol 125:524-535.
21 Salvan A. 1990. Occupational exposure and cancer mortality at an electrical manufacturing plant: a case-control
22 study. Ph. D. Dissertation, University of California, Los Angeles.
23 Sandborn Head and Associaties. 2003. Groupwater Vapor Project, Endicott, New York, Summary of Findings,
24 Working Draft. Cited in ATSDR (2006), Health Statistics Reviews, Cancer and Birth Outcome Analysis, Endicott
25 Area, Town of Union, Broome county, New York. May 26, 2006.
26 Schlesselman JJ. 1982. Case-Control Studies. Design, Conduct, Analysis. New York: Oxford University Press.
27 Scott CS, Chiu WA. 2006. Trichloroethylene cancer epidemiology: a consideration of select issues. Environ
28 Health Perspect 114:1471-8.
29 Seidler A, MohnerM, BergerJ, MesterB, DeegE, Eisner G, Neiters A, Becker N. 2007. Solvent exposure and
30 malignant lymphoma: a population-based case-control study in Germany. J Occup Med Toxicol 2:2.
31 Shannon HS, Haines T, Bernholz C, Julian JA, VermaDK, JamiesonE, Walsh C. 1988. Cancer morbidity in lamp
32 manufacturing workers. Am J Ind Med 14:281-290.
33 Shindell S, Ulrich S. 1985. A cohort study of employees of a manufacturing plant using trichloroethylene. J Occup
34 Med 27:577-579.
3 5 Shu XO, Stewart P, Wen W-Q, Han D, Potter JD, Buckley JD, Heineman E, Robison LL. 1999. Parental
3 6 occupational exposure to hydrocarbons and risk of acute lymphocytic leukemia in offspring. Cancer Epidemiol
37 Markers Prev 8:783-291.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-359 DRAFT—DO NOT CITE OR QUOTE
-------
1 Shu Xo, parentesis JP, Wen W, Buckley JD, Boyle E, Ross, JA, Robison LL. 2004. Parental exposure to
2 medications and hydrocarbons and ras mutations in children with acute lymphblastic leukemia: A report from the
3 Children's Oncology Group. Cancer Epidemiol Biomarkers Prev 13:1230-1235.
4 Siemiatycki J. 1991. Risk Factors for Cancer in the Workplace. J Siemiatycki, Ed. Baca Raton: CRC Press.
5 Siemiatycki J, Wacholder S, Richardson L, Dewar R, Gerin M. 1987. Discovering carcinogens in the occupational
6 environment: methods of data collection and analysis of a large case-referent monitoring system. Scand J Work
7 Environ Health 13:486-492.
8 Siemiatycki J, Dewar R, Nadon L, Gerin M. 1994. Occupational risk factors for bladder cancer: results from a
9 case-control study in Montreal, Quebec, Canada. Am JEpidemiol 140:1061-1080.
10 Siemiatycki J, Fritschi L, Nadon L, Gerin M. 1997. Reliability of an expert rating procedure for retrospective
11 assessment of occupational exposures in community-based case-control studies. Am J Ind Med 31:280-286.
12 Sinks T, Lushniak B,Haussler BJ, Sniezek J, Deng J-F, Roper P, Dill P, Coates R. 1992. Renal cell cancer among
13 paperboard printing workers. Epidemiol 3:483-489.
14 Spirtas R, Stewart PA, Lee JS, Marano DE, Forbes CD, Grauman DJ, Pettigrew HM, Blair A, Hoover RN, Cohen
15 JL. 1991. Retrospective cohort mortality study of workers at an aircraft maintenance facility. I. Epidemiological
16 results. BrJ Ind Med 48:515-30.
17 Stewart PA, Lee JS, Marano DE, Spirtas R, Forbes CD, Blair A. 1991. Retrospective cohort mortality study of
18 workers at an aircraft maintenance facility. II. Exposures and their assessment. Br J Ind Med 48:531-7.
19 Stewart PA, Lees PS, Francis M. 1996. Quantification of historical exposures in occupational cohort studies.
20 Scand J Work Environ Health. 22:405-14.
21 Stopps GJ, McLaughlin M. 1967. Psychophysiological testing of human subjects exposed to solvent vapors. Am
22 Ind Hyg Assoc J. 28:43-50.
23 Sung TI, Chen PC, Lee L JH, Lin Y-P, Hsieh GY, Wang JD. 2007. Increased standardized incidence ratio of breast
24 cancer in female electronics workers. BMC Public Health 7:102. http://www.biomedcentral.com/content/pdf/1471-
25 2458-7-102.pdf.
26 Sung TI, Wang JD, Chen PC. 2008 Increased risk of cancer in the offspring of female electronics workers. Reprod
27 Toxicol 25:115-9.
28 Swaen GMH. 1995. Increased incidence of renal cell tumours in a cohrt of cardboard workers exposed to
29 trichloroethylene [Letter]. Arch Toxicol 70:127-128.
30 Teschke K, Olshan AF, Daniels AL, De Roos AJ, Parks CG, Schulz M, Vaughan TL. 2002. Occupational exposure
31 assessment in case-control studies: opportunities for improvement. Occup Environ Med 59:575-594.
32 U.S. EPA (Environmental Protection Agency). 2005. Documentation of Environmental Indicator Determination.
33 RCRA Corrective Action. http://www.epa.gov/Region2/waste/ibm_e725.pdf. Accessed March 19, 2006.
34 Vamvakas S, Briining T, Thomasson B, Lammert M, Baumuller A, Bolt HM, Dekant W, Birner G, Henschler D,
35 UlmK. 1998. Renal cell cancer risk and occupational exposure to trichloroethylene: results of a consecutive case-
36 control study in Arnsberg, Germany. Am J Ind Med 23:274-285.
37 Vartiainen T, Pukkala E, Rienoja T, Strandman T, Kaksonen K. 1993. Population exposure to tri- and
38 tetrachloroethene and cancer risk: two cases of drinking water pollution. Chemosphere 27:1171-1181.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-360 DRAFT—DO NOT CITE OR QUOTE
-------
1 WangJD. 2004. Increased Mortality Odds Ratio of Male Liver Cancer in a Community contaminated by
2 Chlorinated Hydrocarbons in Groundwater. Presentation at Symposium on New Scientific Research Related to the
3 Health Effects of Trichloroethylene, U. S. EPA (Environmental Protection Agency), February 26-27, 2004.
4 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=75934. Accessed, November 18, 2008.
5 Wang R, Zhang Y, Lan Q, Holford TR, Beaderer B, Zahm SH, Boyle P, Dosemeci M, Rothman N, Zhu Y, Qin Q,
6 Zheng T. 2009. Occupational exposure to solvents and risk of non-Hodgkin lymphoma in Connecticut women.
7 Am J Epidemiol DOI: 10.1093/aj3/kwn300. In Press.
8 Weistenhofer W, Blaszkewicz M, Bolt HM, Golka K. 2008. N-acetyltransferase-2 and medical history in bladder
9 cancer cases with a suspected occupational disease (BK 1301) in Germany. J Toxicol Environ Health (Part A)
10 71:906-910.
11 WHO (World Health Organization). 1997. Manual of the International Statistical Classification of Diseases,
12 Injuries, and Causes of Death (ICD). Geneva, Switzerland.
13 Wilcosky TC, Checkoway H, Marshall EG, Tyroler HA. 1984. Cancer mortality and solvent exposure in the rubber
14 industry. Am Ind Hyg Assoc 145:809-811.
15 Zack M, Cannon S, Loyd D, Heath CW, Falletta JM, Jones B, Housworth J, Crowley S. 1980. Cancer in children
16 of parents exposed to hydrocarbon-related industries and occupations. AmJEpidemiol 111:329-336.
17 ZahmSH. 1992. Computerized Occupational Referent Population System (CORPS): Study Documentation.
18 Rockville, MD: National Cancer Institute, National Institute for Occupational Safety and Health.
19 Zhao Y, Krishnadasan A, Kennedy N, Morgenstern H, RitzB. 2005. Estimated effects of solvents and mineral oils
20 on cancer incidence and Mortality in a cohort of aerospace workers. Am J Ind Med 48:249-258.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 B-361 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX C
Meta-Analysis of Cancer Results from
Epidemiological Studies
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix C: Meta-Analysis of Cancer Results from Epidemiological
Studies
LIST OF TABLES C-iii
LIST OF FIGURES C-iv
APPENDIX C. META-ANALYSIS OF CANCER RESULTS FROM
EPIDEMIOLOGICAL STUDIES C-l
C.I. METHODOLOGY C-l
C.2. META-ANALYSIS FOR LYMPHOMA C-4
C.2.1. Overall Effect of TCE Exposure C-4
C.2.1.1. Selection of RR Estimates C-4
C.2.1.2. Results of Meta-Analyses C-10
C.2.2. Lymphoma Effect in the Highest Exposure Groups C-15
C.2.2.1. Selection of RR Estimates C-15
C.2.2.2. Results of Meta-Analyses C-19
C.2.3. Discussion of Lymphoma Meta-Analysis Results C-21
C.3. META-ANALYSIS FOR KIDNEY CANCER C-24
C.3.1. Overall Effect of TCE Exposure C-24
C.3.1.1. Selection of RR Estimates C-24
C.3.1.2. Results of Meta-Analyses C-30
C.3.2. Kidney Cancer Effect in the Highest Exposure Groups C-33
C.3.2.1. Selection of RR Estimates C-33
C.3.2.2. Results of Meta-Analyses C-39
C.3.3. Discussion of Kidney Cancer Meta-Analysis Results C-43
C.4. META-ANALYSIS FOR LIVER CANCER C-46
C.4.1. Overall Effect of TCE Exposure C-46
CAM. Selection of RR Estimates C-46
C.4.1.2. Results of Meta-Analyses C-49
C.4.2. Liver Cancer Effect in the Highest Exposure Groups C-52
C.4.2.1. Selection of RR Estimates C-52
C.4.2.2. Results of Meta-Analyses C-56
C.4.3. Discussion of Liver Cancer Meta-Analysis Results C-58
C.5. DISCUSSION OF STRENGTHS, LIMITATIONS, AND UNCERTAINTIES
IN THE META-ANALYSES C-59
C.6. CONCLUSIONS C-60
C.7. REFERENCES C-62
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-ii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
C-l. Selected RR estimates for lymphoma associated with TCE exposure (overall
effect) from cohort studies C-6
C-2. Selected RR estimates for lymphoma associated with TCE exposure from case-
control studies C-8
C-3. Summary of some meta-analysis results for TCE (overall) and lymphoma C-ll
C-4. Selected RR estimates for lymphoma risk in highest TCE exposure groups C-l7
C-5. Summary of some meta-analysis results for TCE (highest exposure groups)
and lymphoma C-20
C-6. Selected RR estimates for kidney cancer associated with TCE exposure
(overall effect) from cohort studies C-26
C-7. Selected RR estimates for renal cell carcinoma associated with TCE exposure
from case-control studies C-27
C-8. Summary of some meta-analysis results for TCE (overall) and kidney cancer C-31
C-9. Selected RR estimates for kidney cancer risk in highest TCE exposure groups C-35
C-10. Summary of some meta-analysis results for TCE (highest exposure groups)
and kidney cancer C-41
C-ll. Selected RR estimates for liver cancer associated with TCE exposure (overall
effect) from cohort studies C-47
C-12. Summary of some meta-analysis results for TCE and liver cancer C-50
C-13. Selected RR estimates for liver cancer risk in highest TCE exposure groups C-54
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-iii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
C-l. Meta-analysis of lymphoma and overall TCE exposure C-12
C-2. Funnel plot of SE by log RR estimate for TCE and lymphoma studies C-14
C-3. Cumulative meta-analysis of TCE and lymphoma studies, progressively
including studies with increasing SEs C-15
C-4. Meta-analysis of lymphoma and TCE exposure—highest exposure groups C-21
C-5. Meta-analysis of kidney cancer and overall TCE exposure C-32
C-6. Funnel plot of SEby log RR estimate for TCE and kidney cancer studies C-34
C-7. Meta-analysis of kidney cancer and TCE exposure—highest exposure groups C-42
C-8. Meta-analysis of kidney cancer and TCE exposure—highest exposure groups,
with assumed null RR estimates for Anttila, Axelson, and Hansen (see text) C-43
C-9. Meta-analysis of liver cancer and TCE exposure C-51
C-10. Funnel plot of SEby log RR estimate for TCE and liver cancer studies C-53
C-11. Meta-analysis of liver cancer and TCE exposure—highest exposure groups,
with assumed null RR estimates for Hansen and Zhao (see text) C-57
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-iv DRAFT—DO NOT CITE OR QUOTE
-------
1 APPENDIX C. META-ANALYSIS OF CANCER RESULTS FROM
2 EPIDEMIOLOGICAL STUDIES
O
4
5 C.I. METHODOLOGY
6 An initial review of the epidemiological studies indicated some evidence for associations
7 between trichloroethylene (TCE) exposure and lymphomas and cancers of the kidney and liver.
8 To investigate further these possible associations, we performed meta-analyses of the
9 epidemiological study results for these three cancer types. Meta-analysis provides a systematic
10 way to combine study results for a given effect across multiple (sufficiently similar) studies. The
11 resulting summary (weighted average) estimate is a quantitatively objective way of reflecting
12 results from multiple studies, rather than relying on a single study, for instance. Combining the
13 results of smaller studies to obtain a summary estimate also increases the statistical power to
14 observe an effect, if one exists. Furthermore, meta-analyses typically are accompanied by other
15 analyses of the epidemiological studies, including analyses of publication bias and investigations
16 of possible factors responsible for any heterogeneity across studies.
17 Given the diverse nature of the epidemiological studies for TCE, random-effects models
18 were used for the primary analyses, and fixed-effect analyses were conducted for comparison.
19 Both approaches combine study results (in this case, relative risk [RR] estimates) weighted by
20 the inverse invariance; however, they differ in their underlying assumptions about what the study
21 results represent and how the variances are calculated. For a random-effects model, it is
22 assumed that there is true heterogeneity across studies and that both between-study and
23 within-study components of variation need to be taken into account; this was done using the
24 methodology of DerSimonian and Laird (1986). For a fixed-effect model, it is assumed that the
25 studies are all essentially measuring the same thing and all the variance is within-study variance;
26 thus, for the fixed-effect model, the RR estimate from each study is simply weighted by the
27 inverse of the (within-study) variance of the estimate.
28 Studies for the meta-analyses were selected as described in Appendix B, Section II-9.
29 The general approach for selecting RR estimates was to select the reported RR estimate that best
30 reflected an RR for TCE exposure vs. no TCE exposure (overall effect). When available, RR
31 estimates from internal analyses were selected over standardized incidence or mortality ratios
32 (SIRs, SMRs) and adjusted RR estimates were generally selected over crude estimates.
33 Incidence estimates would normally be preferred to mortality estimates; however, for the two
34 studies providing both incidence and mortality results, incidence ascertainment was for a
35 substantially shorter period of time than mortality follow-up, so the endpoint with the greater
36 number of cases was used to reflect the results that had better case ascertainment. For separate
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-1 DRAFT—DO NOT CITE OR QUOTE
-------
1 analyses, an RR estimate for the highest exposure group was selected from studies that presented
2 results for different exposure groups. Exposure groups based on some measure of cumulative
3 exposure were preferred, if available; however, often duration was the sole exposure metric used.
4 Specific selection choices are described in the following subsections detailing the actual
5 analyses.
6 The meta-analysis calculations are based on (natural) logarithm-transformed values.
7 Thus, each RR estimate was transformed to its natural logarithm (referred to here as "log RR,"
8 the conventional terminology in epidemiology), and either an estimate of the standard error (SE)
9 of the log RR was obtained, from which to estimate the variance for the weights, or an estimate
10 of the variance of the log RR was calculated directly. If the reported 95% confidence interval
11 limits were proportionally symmetric about the observed RR estimate (i.e., upper confidence
12 limit/RR ~ RR/lower confidence limit), then an estimate of the SE of the log RR estimate was
13 obtained using the formula
14
3.92
16
17 where UCL is the upper confidence limit and LCL is the lower confidence limit (for 90%
18 confidence intervals [CIs], the divisor is 3.29) (Rothman and Greenland, 1998). In all the TCE
19 cohort studies reporting SMRs or SIRs as the overall RR estimates, reported CIs were calculated
20 assuming the number of deaths (or cases) is approximately Poisson distributed. In such cases,
21 the CIs are not proportionally symmetric about the RR estimate (unless the number of deaths is
22 fairly large), and the SE of the log RR estimate was estimated as the inverse of the square root of
23 the observed number of deaths (or cases) (Breslow and Day, 1987). In some case-control
24 studies, no overall odds ratio (OR) was reported, so a crude OR estimate was calculated as
25 OR = (a/b)/(c/d), where a, b, c, and d are the cell frequencies in a 2 x 2 table of cancer cases vs.
26 TCE exposure, and the variance of the log OR was estimated using the formula
27
28 Var [log (OR}~] =- + - + - + -, (Eq. C-2)
L J a b c d
29
30 in accordance with the method proposed by Woolf (1955), as described by Breslow and Day
31 (1980).
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 The analyses that were performed for this assessment include
2
3 • meta-analyses to obtain overall summary estimates of RR
4 • heterogeneity analyses
5 • analyses of the influence of single studies on the summary estimates
6 • analyses of the sensitivity of the summary estimate to alternate study inclusion selections
7 or to alternate selections of RR estimates from a study
8 • publication bias analyses
9 • meta-analyses to obtain summary estimates for the highest exposure groups in studies
10 that provide data by exposure group, and
11 • consideration of some potential sources of heterogeneity across studies.
12
13 The analyses were conducted using Excel spreadsheets and the software package Comprehensive
14 Meta-Analysis, Version 2 (© 2006, Biostat, Inc.). Figures were generated using the
15 Comprehensive Meta-Analysis software. Note that for these figures, this software recalculates
16 CIs for the studies based on the SE inputs, and the resulting CIs are not always identical to those
17 reported in the original studies, in particular those based on Poisson distributions. However, the
18 recalculated CIs are merely outputs and are not the basis for any calculations in the software; SEs
19 were obtained as described above, and these SEs and the log RRs constitute the inputs for the
20 meta-analysis calculations.
21 The heterogeneity (or homogeneity) analysis tests the hypothesis that the study results are
22 homogeneous, i.e., that all the RR estimates are estimating the same population RR and the total
23 variance is no more than would be expected from within-study variance. Heterogeneity was
24 assessed using the statistic Q described by DerSimonian and Laird (1986). The ^-statistic
25 represents the sum of the weighted squared differences between the summary RR estimate
26 (obtained under the null hypothesis, i.e., using a fixed-effect model) and the RR estimate from
27 each study, and, under the null hypothesis, Q approximately follows a ^ distribution with
28 degrees of freedom equal to the number of studies minus one. However, this test can be under-
29 powered when the number of studies is small, and it is only a significance test, i.e., it is not very
30 informative about the extent of any heterogeneity. Therefore, the f value (Higgins et al., 2003)
31 was also considered, f = 100% x (Q - dfy/Q, where Q is the (^-statistic and dfis the degrees of
32 freedom, as described above. This value estimates the percentage of variation that is due to
33 study heterogeneity. Typically, I2 values of 25%, 50%, and 75% are considered low, moderate,
34 and high amounts of heterogeneity, respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 Subgroup analyses were sometimes conducted to examine whether or not the combined
2 RR estimate varied significantly between different types of studies (e.g., case-control vs. cohort
3 studies). In such subgroup analyses of categorical variables (e.g., study design), analysis of
4 variance was used to determine if there was significant heterogeneity between the subgroups.
5 Applying analysis of variance to meta-analyses with two subgroups (df = 1), (^between subgroups =
6 Coverall - (^subgroupi + 2subgrouP2) = z-valuQ2., where Coverall is the (^-statistic calculated across all the
7 studies and Qsubgroupi and 2subgrouP2 are the (^-statistics calculated within each subgroup.
8 Publication bias is a systematic error that occurs if statistically significant studies are
9 more likely to be submitted and published than nonsignificant studies. Studies are more likely to
10 be statistically significant if they have large effect sizes (in this case, RR estimates); thus, an
11 upward bias would result in a meta-analysis if the available published studies have higher effect
12 sizes than the full set of studies that was actually conducted. One feature of publication bias is
13 that smaller studies tend to have larger effect sizes than larger studies, since smaller studies need
14 larger effect sizes in order to be statistically significant. Thus, many of the techniques used to
15 analyze publication bias examine whether or not effect size is associated with study size.
16 Methods used to investigate potential publication bias for this assessment included funnel plots,
17 which plot effect size vs. study size (actually, SE vs. log RR here); the "trim and fill" procedure
18 of Duvall and Tweedie (2000), which imputes the "missing" studies in a funnel plot (i.e., the
19 studies needed to counterbalance an asymmetry in the funnel plot resulting from an ostensible
20 publication bias) and recalculates a summary effect size with these studies present; forest plots
21 (arrays of RRs and CIs by study) sorted by precision (i.e., SE) to see if effect size shifts with
22 study size; Begg and Mazumdar rank correlation test (Begg and Mazumdar, 1994), which
23 examines the correlation between effect size estimates and their variances after standardizing the
24 effect sizes to stabilize the variances; Egger's linear regression test (Egger et al., 1997), which
25 tests the significance of the bias reflected in the intercept of a regression of effect size/SE on
26 1/SE; and cumulative meta-analyses after sorting by precision to assess the impact on the
27 summary effect size estimate of progressively adding the smaller studies.
28
29 C.2. META-ANALYSIS FOR LYMPHOMA
30 C.2.1. Overall Effect of TCE Exposure
31 C.2.1.1. Selection ofRR Estimates
32 The selected RR estimates for lymphoma associated with TCE exposure from the
33 selected epidemiological studies are presented in Table C-l for cohort studies and in Table C-2
34 for case-control studies. A few of the more recent case-control studies classified lymphomas
35 along the lines of the recent WHO/REAL classification system (World Health
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 Organization/Revised European-American Classification of Lymphoid Neoplasms) (Harris et al.,
2 2000); however, most of the available TCE studies reported lymphoma results according to the
3 International Classification of Diseases (ICD), Revisions 7, 8, and 9, and focused on
4 non-Hodgkin lymphoma (NHL; ICD 200 + 202). For consistency of endpoint in the lymphoma
5 meta-analyses, RR estimates for ICD 200 + 202 were selected, wherever possible; otherwise,
6 estimates for the classification(s) best approximating NHL were selected. In addition, many of
7 the studies provided RR estimates only for males and females combined, and we are not aware of
8 any basis for a sex difference in the effects of TCE on lymphoma risk; thus, wherever possible,
9 RR estimates for males and females combined were used. The only study of much size (in terms
10 of number of lymphoma cancer cases) that provided results separately by sex was
11 Raaschou-Nielsen (2003). This study reports an insignificantly higher SIR for females (1.4,
12 95% CI: 0.73, 2.34) than for males (1.2, 95% CI: 0.98, 1.52).
13 Beyond selecting adjusted RR estimates for lymphoma classification and both sexes,
14 when multiple estimates were available, the preference was to select the RR estimate that
15 represented the largest population in a study, while trying to minimize the likelihood of TCE
16 exposure misclassification. Sensitivity analyses were generally done to investigate the impact of
17 these alternate selection choices, as well as to estimate the impacts of study findings that were
18 not reported.
19 Thus, for example, for Axelson et al. (1994), in which a small subcohort of females was
20 studied but only results for the larger male subcohort were reported, the reported male-only
21 results were used in the primary analysis; however, an attempt was made to estimate the female
22 contribution to an overall RR estimate for both sexes and its impact on the meta-analysis.
23 Axelson et al. (1994) reported that there were no cases of lymphoma observed in females, but the
24 expected number was not presented. To estimate the expected number, the expected number for
25 males was multiplied by the ratio of female-to-male person-years in the study and by the ratio of
26 female-to-male age-adjusted incidence rates for NHL.l The male results and the estimated
27 female contribution were then combined into an RR estimate for both sexes assuming a Poisson
28 distribution, and this alternate RR estimate for the Axelson et al. (1994) study was used in a
29 sensitivity analysis.
Person-years for men and women <79 years were obtained from Axelson etal. (1994): 23516.5 and 3691.5,
respectively. Lifetime age-adjusted incidence rates for NHL for men and women were obtained from the National
Cancer Institute's 2000-2004 SEER-17 (Surveillance Epidemiology and End Results from 17 geographical areas)
database (http://seer.cancer.gov/statfacts/html/nhl.html): 23.2/100,000 and 16.3/100,000, respectively. The
calculation for estimating the expected number of cases in females in the cohort assumes that the males and females
have similar TCE exposures and that the relative distributions of age-related incidence risk for the males and
females in the cohort are adequately represented by the ratios of person-years and lifetime incidence rates used in
the calculation.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-5 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-l. Selected RR estimates for lymphoma associated with TCE exposure (overall effect) from cohort
studies
Study
Anttila et a\.,
1995
Axelson et
al., 1994
Boice et al.,
1999
Greenland
etal., 1994
Hansen et
al.,2001
Morgan et
al., 1998
Raaschou-
Nielsen et
al., 2003
Radican et
al., 2008
RR
1.81
1.52
1.19
0.76
ZA
1.01
1.24
1.36
95%
LCL
0.78
0.49
0.65
0.24
1.3
0.46
1.01
0.77
95%
UCL
3.56
3.53
1.99
2.42
6.1
1.92
1.52
2.39
RR type
SIR
SIR
SMR
OR
SIR
SMR
SIR
Mortality
HR
log RR
0.593
0.419
0.174
-0.274
1.13
0.00995
0.215
0.307
SE(log RR)
0.354
0.447
0.267
0.590
0.354
0.333
0.104
0.289
Alternate RR
estimates
None
1.36(0.44,3.18)
with estimated
female
contribution to SIR
added (see text)
1.19(0.83, 1.65)
for any potential
exposure
None
None
1.36(0.35,5.21)
unpublished RR
for ICD 200 (see
text)
None
None
Comments
ICD-7 200 + 202.
ICD-7 200 and 202. Results reported
separately; combined assuming Poisson
distribution. Results reported for males only,
but there was a small female component to the
cohort.
ICD-9 200 + 202. For potential routine
exposure.
ICD-8 200-202. Nested case-control study.
ICD-7 200 + 202. Male and female results
reported separately; combined assuming
Poisson distribution.
ICD 200 + 202. Results reported by Mandel et
al. (2006). ICD Revision 7, 8, or 9, depending
on year of death.
ICD-7 200 + 202.
ICD-8,-9 200 + 202; ICD-10 C82-C85. Time
variable = age; covariates = sex and race.
Referent group is workers with no chemical
exposures.
Co
I
~p
TO'
^3
58
•
TO
o s
H fc
H
W
-------
o
to
§3
VO Co'
Table C-l. Selected RR estimates for lymphoma associated with TCE exposure (overall effect) from cohort
studies (continued)
Study
Zhao et a\.,
2005
RR
-\.44
95%
LCL
0.90
95%
UCL
2.30
RR type
Mortality
RR
log RR
0.363
SE(log RR)
0.239
Alternate RR
estimates
Incidence RR:
0.77 (0.42, 1.39)
Boice 2006 SMR
for ICD-9 200 +
202:0.21 (0.01,
1.18)
Comments
All lymphohematopoietic cancer (ICD-9 200-
208), not just 200 + 202. Males only; adjusted
for age, socioeconomic status (SES), time since
first employment. Mortality results reflect more
exposed cases (33) than do incidence results
(17). Overall RR estimated by combining
across exposure groups (see text). Boice 2006
cohort overlaps Zhao cohort; just 1 exposed
death for ICD 200 + 202; 9 for 200-208 (vs. 33
in Zhao).
Co
I
~p
TO'
^3
58
J
O TO
I Co
•
TO
o s
H fc
H
W
-------
o
to
§3
VO Co'
Table C-2. Selected RR estimates for lymphoma associated with TCE exposure from case-control studies"
Study
Hardell etal.,
1994
Miligi et al.,
2006
Nordstrom
etal., 1998
Persson and
Frederikson,
1999
Seidler et al.,
2007
Siemiatycki,
1991
Wang et al.,
2009
RR
7.2
0.93
1.5
1.2
1.0
1.1
1.2
95%
LCL
1.3
b
0.7
0.5
0.74
0.5
0.9
95%
UCL
42
b
3.3
2.4
1.4
2.5
1.8
log RR
1.97
-0.0726
0.405
0.182
-0.223
0.0953
0.182
SE(log
RR)
0.887
0.168
0.396
0.400
0.177
0.424
0.177
Lymphoma
type
NHL
NHL + CLL
HCL
NHL
B-cell and
T-cell NHL
NHL
"NHL";
various
lymphoma
subtypes +
mast cell
tumors
Comments
Rappaport classification system. Males only; controls
matched for age, place of residence, vital status.
NCI working formulation. Crude OR; overall adjusted OR
not presented.
HCL specifically. Males only; controls matched for age and
county; analysis controlled for age.
Classification system not specified. Controls selected from
same geographic areas; ORs stratified on age and sex.
WHO classification. Overall results for B-cell and T-cell
NHL from personal communication (see text). Adjusted for
smoking and alcohol consumption. Case-control pairs
matched on sex, region, and age.
ICD-9 200 + 202. SE and 95% Cl calculated from reported
90% CIs; males only; adjusted for age, income, and
cigarette smoking index.
ICD-0 M-9590-9642, 9690-9701, 9740-9750. Females
only; adjusted for age, family history of
lymphohematopoietic cancers, alcohol consumption, and
race.
Co
I
~p
TO'
^3
58
J
O TO
I Co
•
TO
o s
H fc
The RR estimates are all ORs for incident cases.
bNot calculated.
NHL: non-Hodgkin lymphoma; CLL: chronic lymphocytic leukemia; HCL: hairy cell leukemia (a subgroup of NHL).
H
W
-------
1 Most of the selections in Tables C-l and C-2 should be self-evident, but some are
2 discussed in more detail here, in the order the studies are presented in the tables. For Boice et al.
3 (1999), results for "potential routine exposure" were selected for the primary analysis, because
4 this exposure category was considered to have less exposure misclassification, and results for
5 "any potential exposure" were used in a sensitivity analysis. The Greenland et al. (1994) study is
6 a case-control study nested within a worker cohort, and we treat it here as a cohort study (see
7 Appendix B, Section II-9.1). For Morgan et al. (1998), the reported results did not allow for the
8 combination of ICD 200 and 202, so the SMR estimate for the combined 200 + 202 grouping
9 was taken from the meta-analysis paper of Mandel et al. (2006), who included one of the
10 investigators from the Morgan et al. (1998) study. RR estimates for overall TCE exposure from
11 internal analyses of the Morgan et al. (1998) cohort data were available from an unpublished
12 report (Environmental Health Strategies, 1997; the published paper only presented the internal
13 analyses results for exposure subgroups), but only for ICD 200; from these, the RR estimate
14 from the Cox model which included age and sex was selected, because those are the variables
15 deemed to be important in the published paper (Morgan et al., 1998). Although the results from
16 internal analyses are generally preferred, in this case the SMR estimate was used in the primary
17 analysis and the internal analysis RR estimate was used in a sensitivity analysis because the latter
18 estimate represented an appreciably smaller number of deaths (3, based on ICD 200 only) than
19 the SMR estimate (9, based on ICD 200 + 202). For Radican et al. (2008), the Cox model hazard
20 ratio (HR) from the 2000 follow-up was used. In the Radican et al. (2008) Cox regressions, age
21 was the time variable, and sex and race were covariates. It should also be noted that the referent
22 group is composed of workers with no chemical exposures, not just no exposure to TCE.
23 For Zhao et al. (2005), RR estimates were only reported for ICD-9 200-208 (all
24 lymphohematopoietic cancers), and not for 200 + 202 alone. Given that other studies have not
25 reported associations between leukemias and TCE exposure, combining all lymphohematopoietic
26 cancers would dilute any lymphoma effect, and the Zhao results are expected to be an
27 underestimate of any TCE effect on lymphoma alone. Another complication with the Zhao et al.
28 (2005) study is that no results for an overall TCE effect are reported. We were unable to obtain
29 any overall estimates from the study authors, so, as a best estimate, the results across the
30 "medium" and "high" exposure groups were combined, under assumptions of group
31 independence, even though the exposure groups are not independent (the "low" exposure group
32 was the referent group in both cases). Zhao et al. (2005) present RR estimates for both incidence
33 and mortality; however, the time frame for the incidence accrual is smaller than the time frame
34 for mortality accrual and fewer exposed incident cases (17) were obtained than deaths (33).
35 Thus, because better case ascertainment occurred for mortality than for incidence, the mortality
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-9 DRAFT—DO NOT CITE OR QUOTE
-------
1 results were used for the primary analysis, and the incidence results were used in a sensitivity
2 analysis. A sensitivity analysis was also done using results from Boice et al. (2006) in place of
3 the Zhao et al. (2005) RR estimate. The cohorts for these studies overlap, so they are not
4 independent studies and should not be included in the meta-analysis concurrently. Boice et al.
5 (2006) report an RR estimate for an overall TCE effect for lymphoma alone; however, it is based
6 on far fewer cases (1 death in ICD-9 200 + 202; 9 deaths for 200-208) and is an SMR rather
7 than an internal analysis RR estimate, so the Zhao et al. (2005) estimates are preferred for the
8 primary analysis.
9 For the case-control studies, the main issue was the lymphoma classifications.
10 Miligi et al. (2006) include chronic lymphocytic leukemias (CLLs) in their NHL results,
11 consistent with the current WHO/REAL classification. Also, Miligi et al. (2006) do not report an
12 overall adjusted RR estimate, so a crude estimate of the OR was calculated for the two TCE
13 exposure categories together vs. no TCE exposure. The Nordstrom et al. (1998) study was a
14 case-control study of hairy cell leukemias (HCLs), which are a subgroup of NHLs, so only
15 results for HCL were reported. For Seidler et al. (2007), an overall adjusted OR for B-cell and
16 T-cell NHL combined was kindly provided by Dr. Seidler (personal communication from
17 Andreas Seidler, Bundesanstalt fur Arbeitsschutz u. Arbeitsmedizin, to Cheryl Scott, U.S. EPA,
18 13 November 2007). Wang et al. (2009) refer to their cases as "NHL" cases; however, according
19 to the ICD-O classification system that they used, their cases are more specifically various
20 particular subtypes of malignant lymphoma (9590-9642, 9690-9701) and mast cell tumors (9740-
21 9750) (Morton et al., 2003). No alternate RR estimates were considered for any of the case-
22 control studies of lymphoma.
23
24 C.2.1.2. Results of Meta-Analyses
25 Results from some of the meta-analyses that were conducted on the epidemiological
26 studies of TCE and lymphoma are summarized in Table C-3. The summary estimate from the
27 primary random effects meta-analysis of the 16 studies was 1.23 (95% CI: 1.04, 1.44) (see
28 Figure C-l). No single study was overly influential; removal of individual studies resulted in
29 summary, or "pooled," RR (RRp) estimates that ranged from 1.16 (with the removal of Hansen)
30 to 1.28 (with the removal of Seidler) and were all statistically significant. Removal of Hardell,
31 whose RR estimate is a relative outlier (see Figure C-l), only decreased the RRp estimate to 1.20
32 (1.04, 1.39), since this study does not contribute a lot of weight to the meta-analysis. Removal of
33 studies other than Hansen or Hardell resulted in RRp estimates that were all greater than 1.20.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-10 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-3. Summary of some meta-analysis results for TCE (overall) and lymphoma
Analysis
A\\ studies
Cohort
Case-control
Alternate RR
selections3
Highest
exposure
groups
#of
studies
16
9
7
16
16
16
16
16
12
Model
Random
Fixed
Random
Fixed
Random
Fixed
Random
Random
Random
Random
Random
Random
Fixed
Summary
RR
estimate
(RRp)
1.23
1.19
1.35
1.33
1.07
1.03
1.19
1.21
1.22
1.22
1.24
1.57
1.57
95% LCL
1.04
1.06
1.13
1.14
0.84
0.86
1.00
1.01
1.04
1.05
1.05
1.27
1.27
95%
UCL
1.44
1.34
1.61
1.54
1.37
1.23
1.41
1.45
1.44
1.43
1.46
1.94
1.94
Heterogeneity
Not significant
(p = 0.10)
Not significant
(p = 0.35)
Not significant
(p = 0.17)
Not significant
(p = 0.07)
Not significant
(p = 0.053)
Not significant
(p = 0.10)
Not significant
(p = 0.10)
Not significant
(p = 0.10)
None
observable
(fixed =
random)
Comments
Statistical significance of RRp not dependent
on individual studies.
Not significant difference between CC and
cohort studies (p = 0.13).
Significant difference between CC and cohort
studies (p = 0.03).
With estimated Zhao overall RR for incidence
rather than mortality.
With Boice (2006) study rather than Zhao.
With estimated female contribution to Axelson.
With Boice (1999) any potential exposure
SMR.
With Morgan et al. (1998) unpublished RR.
Statistical significance not dependent on single
study.
See Table C-5 for results with alternate RR
selections.
Co
I
~p
TO'
^3
58
•
TO
o
H fc
""Changing the primary analysis by one alternate RR each time; more details on alternate RR estimates in text.
H
W
-------
TCE and Lymphoma
Study name
Statistics for each study
Rate Lower Upper
ratio limit limit p-Value
Rate ratio and 95% Cl
Anttila 1995
Axelson 1994
Boice 1999
Greenland 1994
Hansen 2001
Morgan 1998
Raaschou-Nielsen 2003
Radican 2008
Zhao 2005 mort
Hardell 1994
Miligi 2006
Nordstrom 1998
Persson&Fredrikson 19991.200
Seidler2007
Siemiatycki 1991
Wang 2008
1.810
1.520
1.190
0.760
3.100
1.010
1.240
1.360
1.437
7.200
0.933
1.500
11.200
0.800
1.100
1.200
1.228
0.905
0.633
0.705
0.239
1.550
0.526
1.011
0.772
0.899
1.267
0.671
0.691
0.548
0.566
0.479
0.849
1.044
3.619
3.652
2.009
2.413
6.199
1.941
1.521
2.396
2.297
40.923
1.298
3.257
2.629
1.131
2.525
1.697
1.444
0.093
0.349
0.515
0.642
0.001
0.976
0.039
0.287
0.130
0.026
0.682
0.305
0.649
0.207
0.822
0.302
0.013
0.1 0.2 0.5 1
random effects model
Figure C-l. Meta-analysis of lymphoma and overall TCE exposure. The
pooled estimate is in the bottom row. Symbol sizes reflect relative weights of the
studies. The horizontal midpoint of the bottom diamond represents the summary
RR estimate, and the horizontal extremes depict the 95% CI limits.
Similarly, the RRp estimate was not highly sensitive to alternate RR estimate selections.
Use of the five alternate selections, individually, resulted in RRp estimates that ranged from 1.19
to 1.24 (see Table C-3) and were all statistically significant except when the Zhao incidence
estimate (p = 0.050) was used instead of the Zhao mortality estimate. As discussed above, the
Zhao mortality estimate is preferred over the incidence estimate in this instance because it is
based on nearly twice as many cases (33 vs. 17).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-12 DRAFT—DO NOT CITE OR QUOTE
-------
There was some heterogeneity apparent across the 16 studies, although it was not
statistically significant (p = 0.10). The I2 value (see Section C. 1) was 33%, suggesting low-to-
moderate heterogeneity. Subgroup analyses were done examining the cohort and case-control
studies separately. With the random effects model (and tau-squared not pooled across
subgroups), the resulting RRp estimates were 1.35 (95% CI: 1.13, 1.61) for the cohort studies
and 1.07 (0.84, 1.37) for the case-control studies. There was residual heterogeneity in each of
the subgroups, but in neither case was it statistically. I2 values were 10% for the cohort studies,
suggesting low heterogeneity, and 33% for the case-control studies, suggesting low-to-moderate
heterogeneity. The difference between the RRp estimates for the cohort and case-control
subgroups was not statistically significant under the random effects model, although it was under
the fixed effect model (see Table C-3). Some thought was given to further analyses to
investigate the source(s) of the heterogeneity, such as qualitative tiering or subgroups based on
likelihood for correct exposure classification or on likelihood for higher vs. lower exposures
across the studies. Ultimately, these approaches were rejected because in many of the studies it
was difficult to judge (and weight) the extent of exposure misclassification or the degree of TCE
exposure with any precision. In other words, there was inadequate information to reliably assess
either the extent to which each study accurately classified exposure status or the relative TCE
exposure levels and prevalences of exposure to different levels across studies. See Section C.2.3
below for a qualitative discussion of some potential sources of heterogeneity.
As discussed in Section C.I, publication bias was examined in several different ways.
The funnel plot in Figure C-2 suggests some relationship between RR estimate and study size (if
there were no relationship, the studies would be symmetrically distributed around the pooled RR
estimate rather than veering towards higher RR estimates with increasing SEs), although the
observed asymmetry is highly influenced by the Hardell study, which is a relative outlier and
which contributes little weight to the overall meta-analysis, as discussed above. The Begg and
Mazumdar rank correlation test and Egger's linear regression test were not statistically
significant; it should be noted, however, that both of these tests have low power. Duval and
Tweedie's trim-and-fill procedure yielded a pooled RR estimate (under the random effects
model) of 1.13 (95% CI: 0.94, 1.35) when the 4 studies deemed missing from the funnel plot
were filled into the meta-analysis (these studies are filled in so as to counter-balance the apparent
asymmetry of the more extreme values in the funnel plot). Eliminating the Hardell study made
little difference to the results of the publication bias analyses. The results of a cumulative
meta-analysis, incorporating studies with increasing SE one at a time, are depicted in Figure C-3.
This procedure is a transparent way of examining the effects of including studies with increasing
SE. The figure shows that the pooled RR estimate is 1.05 after inclusion of the 4 largest (i.e.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-13 DRAFT—DO NOT CITE OR QUOTE
-------
most precise) studies, which constitute about 50% of the weight. The pooled RR estimate
increases to 1.12 with inclusion of the 8 most precise studies, which represent /^ of the total
number of studies and about 75% of the weight. The pooled RR estimate becomes fairly stable
after addition of the next 2 most precise study (RRp = 1.21), which adds another 9% of the
weight. Adding in the 6 least precise studies (16% of the weight) barely increases the pooled RR
estimate further. In summary, there is some evidence of potential publication bias in this data
set. It is uncertain, however, that this reflects actual publication bias rather than an association
between effect size and SE resulting for some other reason, e.g., a difference in study
populations or protocols in the smaller studies. Furthermore, if there is publication bias in this
data set, it does not appear to account completely for the findings of an increased lymphoma risk.
0.0
0.2
0.4
LJJ
•D
re
•a 0.6
3
0.8
1.0
Funnel Plot of Standard Error by Log rate ratio
O
n
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
Log rate ratio
Figure C-2. Funnel plot of SE by log RR estimate for TCE and lymphoma
studies.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-14 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Lymphoma
Study name
Cumulative statistics
Raaschou-Nielsen 2003
Miligi2006
Seidler2007
Wang 2008
Zhao 2005 mort
Boice 1999
Radican 2008
Morgan 1998
Anttila 1995
Hansen 2001
Nordstrom 1998
Persson&Fredrikson 19991.214
Siemiatycki 1991
Axelson 1994
Greenland 1994
Hardell 1994
Point
1.240
1.109
1.004
1.052
1.094
1.105
1.122
1.120
1.138
1.212
1.220
H.214
1.206
1.210
1.199
1.228
1.228
Lower
limit
1.011
0.845
0.763
0.855
0.904
0.936
0.965
0.979
0.989
1.006
1.022
1.027
1.029
1.040
1.035
1.044
1.044
Upper
limit
1.521
1.456
1.320
1.294
1.324
1.303
1.304
1.281
1.310
1.459
1.456
1.434
1.412
1.408
1.390
1.444
1.444
p-Value
0.039
0.455
0.980
0.631
0.356
0.238
0.135
0.099
0.071
0.043
0.028
0.023
0.020
0.014
0.016
0.013
0.013
Cumulative rate ratio (95% Cl)
0.5
random effects model; cumulative analysis, sorted by SE
Figure C-3. Cumulative meta-analysis of TCE and lymphoma studies,
progressively including studies with increasing SEs.
C.2.2. Lymphoma Effect in the Highest Exposure Groups
C.2.2.1. Selection of RR Estimates
The selected RR estimates for lymphoma in the highest TCE exposure categories, for
studies that provided such estimates, are presented in Table C-4. All 8 cohort studies (but not the
nested case-control study of Greenland et al. [1994]) and 4 of the 7 case-control studies did
report lymphoma risk estimates categorized by exposure level. As in Section C.2.1.1 for the
overall risk estimates, estimates to best correspond to NHL as represented by ICD-7, -8, and -
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-15 DRAFT—DO NOT CITE OR QUOTE
-------
9 200 and 202 were selected, and, wherever possible, RR estimates for males and females
combined were used.
As above for the overall TCE effect, for Axelson et al. (1994), in which a small subcohort
of females was studied but only results for the larger male subcohort were reported, the reported
male-only high-exposure group results were used in the primary analysis; however, an attempt
was made to estimate the female contribution to a high-exposure group RR estimate for both
sexes and its impact on the meta-analysis. To estimate the expected number in the highest
exposure group for females, the expected number in the highest exposure group for males was
multiplied by the ratio of total female-to-male person-years in the study and by the ratio of
female-to-male age-adjusted incidence rates for NHL. The RR estimate for both sexes was used
as an alternate RR estimate for the Axelson et al. (1994) study in a sensitivity analysis.
For Boice et al. (1999), only results for workers with "any potential exposure" (rather
than "potential routine exposure") were presented by exposure category, and the referent group is
workers not exposed to any solvent. For Hansen et al. (2001), exposure group data were
presented only for males. To estimate the female contribution to a highest-exposure group RR
estimate for both sexes, it was assumed that the expected number of cases in females had the
same overall-to-highest-exposure group ratio as in males. The RR estimate for both sexes was
then calculated assuming a Poisson distribution, and this estimate was used in the primary
analysis. Hansen et al. (2001) present results for three exposure metrics; the cumulative
exposure metric was preferred for the primary analysis, and results for the other two metrics
were used in sensitivity analyses. For Morgan et al. (1998), results did not allow for the
combination of ICD 200 and 202, so the highest-exposure group RR estimate for ICD 200 only
was used. The primary analysis used results for the cumulative exposure metric, and a
sensitivity analysis was done with the results for the peak exposure metric.
For Radican et al. (2008), it should be noted that the referent group is composed of
workers with no chemical exposures, not just no exposure to TCE. In addition, exposure group
results were reported separately for males and females and were combined for this assessment
using inverse-variance weighting, as in a fixed effect meta-analysis. Radican et al. (2008)
present only mortality HR estimates by exposure group; however, in an earlier follow-up of this
same cohort, Blair et al. (1998) present both incidence and mortality RR estimates by exposure
group. The mortality RR estimate based on the more recent follow-up of Radican et al. (2008)
(17 deaths in the highest exposure group) was used in the primary analysis, while the incidence
RR estimate based on similarly combined results from Blair et al. (1998) (9 cases) was used as
an alternate estimate in a sensitivity analysis.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-16 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-4. Selected RR estimates for lymphoma risk in highest TCE exposure groups
Study
Anttila et
a\., 1995
Axelson et
al., 1994
Boice et al.,
1999
Hansen et
al.,2001
Morgan et
al., 1998
Raaschou-
Nielsen et
al., 2003
RR
1.4
6.25
1.62
2.7
0.81
1.6
95%
LCL
0.17
0.16
0.82
0.56
0.1
1.1
95%
UCL
5.04
34.83
3.22
8.0
6.49
2.2
Exposure
category
100+ |jmol/L
U-TCA3
>2-yr
exposure
and 100+
mg/L U-TCA
>5-yr
exposure
>1080 mos x
mg/m3
High
cumulative
exp. score
>5 yrs in
subcohort
with
expected
higher exp.
levels
log RR
0.336
1.83
0.482
0.993
-0.211
0.470
SE(log
RR)
0.707
1.00
0.349
0.577
1.06
0.183
Alternate RR
estimates
none
5.62(0.14, 31.3)
with estimated
female
contribution
added (see text)
None
3.7(1.0, 9.5) for
>75 mos
exposure
duration
2.9 (0.79, 7.5) for
>19 mg/m3 mean
exposure
1.31 (0.28,6.08)
for med/high
peak vs. low/no
None
Comments
SIR. ICD200 + 202.
SIR. ICD 200 + 202. Results reported for
males only, but there was a small female
component to the cohort.
Mortality RR. ICD 200 + 202. For potential
routine or intermittent exposure. Adjusted for
date of birth, dates 1st and last employed,
race, and sex. Referent group is workers not
exposed to any solvent.
SIR. ICD 200 + 202. Exposure-group
results presented only for males. Female
results estimated and combined with male
results assuming Poisson distribution (see
text).
Mortality RR. ICD 200 only. Adjusted for
age and sex.
SIR. ICD 200 + 202.
Co
I
~p
TO'
^3
58
•
TO
o s
H fc
H
W
-------
o
to
§3
VO Co'
Table C-4. Selected RR estimates for lymphoma risk in highest TCE exposure groups (continued)
Study
Radican et
a\., 2008
Zhao et a\.,
2005
Miligi et al.,
2006
Seidleret
al., 2007
Siemiatycki
1991
Wang et al.,
2009
RR
1.41
1.30
1.2
2.3
0.8
2.2
95%
LCL
0.71
0.52
0.7
1.0
0.2
0.9
95%
UCL
2.81
3.23
2.0
5.2
3.3
5.4
Exposure
category
>25 unit-yr
High
exposure
score
Med/high
exposure
intensity
>35 ppm-yr
Substantial
Medium-high
intensity
log RR
0.337
0.262
0.182
0.833
-0.223
0.788
SE(log
RR)
0.350
0.466
0.268
0.421
0.719
0.457
Alternate RR
estimates
Blair et al.
(1998)0.97
(0.42, 2.2)
incidence RR
Incidence RR:
0.20 (0.03,
1.46)
1.0(0.5,2.6)
for med/high
intensity and
>15-yrexp.
None
None
None
Comments
Mortality HR. ICD200 + 202. Male and
female results presented separately and
combined (see text). Cox regression time
variable = age; covariate = race. Referent
group is workers with no chemical exposures.
Mortality RR. Results for all
lymphohematopoietic cancer (ICD-9 200-208),
not just 200 + 202. Males only; adjusted for
age, SES, time since first employment.
Mortality results reflect more exposed cases (6
in high-exposure group) than do incidence
results (1 in high-exposure group).
Incidence OR. NHL + CLL (see
Section C.2. 1.1).
Incidence OR. Results for B-cell and T-cell
NHL from personal communication (see
Section C.2.1 .1). Adjusted for smoking and
alcohol consumption. Case-control pairs
matched on sex, region, and age.
Incidence OR. NHL. SE and 95% Cl
calculated from reported 90% CIs. Males only;
adjusted for age, income, and cigarette
smoking index.
Incidence OR. "NHL" (various malignant
lymphoma subtypes and mast cell tumors).
Females only; adjusted for age, family history
of lymphohematopoietic cancers, alcohol
consumption, and race.
Co
I
~p
TO'
^3
58
oo
•
TO
o s
H fc
aMean personal trichloroacetic acid in urine. 1 umol/L = 0.1634 mg/L.
H
W
-------
1 For Zhao et al. (2005), RR estimates were only reported for ICD-9 200-208 (all
2 lymphohematopoietic cancers), and not for 200 + 202 alone. Given that other studies have not
3 reported associations between leukemias and TCE exposure, combining all lymphohematopoietic
4 cancers would dilute any lymphoma effect, and the Zhao results are expected to be an
5 underestimate of any TCE effect on lymphoma alone. Zhao et al. (2005) present RR estimates
6 for both incidence and mortality in the highest exposure group; however, the time frame for the
7 incidence accrual is smaller than the time frame for mortality accrual and fewer incident cases
8 (1) were obtained than deaths (6), so the mortality results were used for the primary analysis to
9 reflect the better case ascertainment in the mortality data, and the incidence results were used in
10 a sensitivity analysis.
11 Miligi et al. (2006) include CLLs in their NHL results, consistent with the current
12 WHO/REAL classification. Miligi et al. (2006) report RR estimates for medium and high
13 exposure intensity overall and by duration of exposure; however, there was incomplete
14 information for the duration breakdowns (e.g., a case missing), so the RR estimate for med/high
15 exposure intensity overall was used in the primary analysis, and the RR estimate for med/high
16 exposure for >15 years was used in a sensitivity analysis. For Seidler et al. (2007), an adjusted
17 OR for B-cell and T-cell NFIL combined for the >35 ppm-years exposure category was kindly
18 provided by Dr. Seidler (personal communication from Andreas Seidler, Bundesanstalt fur
19 Arbeitsschutz u. Arbeitsmedizin, to Cheryl Scott, U.S. EPA, 13 November 2007). Wang et al.
20 (2009) refer to their cases as "NHL" cases; however, according to the ICD-O classification
21 system that they used, their cases are more specifically various particular subtypes of malignant
22 lymphoma (9590-9642, 9690-9701) and mast cell tumors (9740-9750) (Morton et al., 2003).
23
24 C.2.2.2. Results of Meta-Analyses
25 Results from the meta-analyses that were conducted for lymphoma in the highest exposure
26 groups are summarized at the bottom of Table C-3 and reported in more detail in Table C-5. The
27 pooled RR estimate from the primary random effects meta-analysis of the 12 studies with results
28 presented for exposure groups was 1.57 (95% CI: 1.27, 1.94) (see Figure C-4). No single study
29 was overly influential; removal of individual studies resulted in RRp estimates that were all
30 statistically significant (all with/? < 0.001) and that ranged from 1.53 (with the removal of
31 Seidler) to 1.65 (with the removal of Miligi). Similarly, the RRp estimate was not highly
32 sensitive to alternate RR estimate selections. Use of the 7 alternate selections, individually,
33 resulted in RRp estimates that were all statistically significant (all with/? < 0.001) and all in the
34 narrow range from 1.54 to 1.60 (see Table C-5). There was no observable heterogeneity across
35 the 12 studies in either the primary analysis or any of the alternate RR analyses.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-19 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-5. Summary of some meta-analysis results for TCE (highest exposure groups) and lymphoma
Analysis
Primary
analysis
Alternate RR
selections3
Model
Random
Random
Random
Random
Random
Random
Random
Random
Combined
RR
estimate
1.57
1.54
1.55
1.57
1.57
1.58
1.60
1.60
95% LCL
1.27
1.24
1.24
1.27
1.27
1.28
1.28
1.30
95% UCL
1.94
1.91
1.92
1.94
1.95
1.96
2.00
1.98
Heterogeneity
None obs
(fixed =
random)
None obs
None obs
None obs
None obs
None obs
None obs
None obs
Comments
Statistical significance not dependent on single
study.
With Blair et al. (1998) incidence RR instead of
Radican mortality HR.
With Zhao incidence.
With estimated female contribution for Axelson.
With Morgan peak.
With Hansen mean exposure.
With Miligi with >1 5 yrs.
With Hansen duration.
Co
I
~p
TO'
°§
•
TO
""Changing the primary analysis by one alternate RR estimate each time.
obs = observable.
o s
H fc
H
W
-------
TCE and Lymphoma - highest exposure groups
Study name
Anttila 1995
Axelson 1994
Boice 1999
Hansen 2001 cum exp
Morgan 1998
Statistics for each study
Rate ratio and 95% Cl
Rate
ratio
1.400
6.250
1.620
2.700
0.810
Raaschou-Nielsen 2003 1.600
Radican 2008 mort
Zhao 2005 mort
Miligi 2006
Seidler2007
Siemiatycki 1991
Wang 2009
1.400
1.300
1.200
2.300
0.800
2.199
1.569
Lower
limit
0.350
0.880
0.818
0.871
0.101
1.119
0.705
0.522
0.709
1.008
0.195
0.898
1.267
Upper
limit
5.598
44.369
3.210
8.372
6.525
2.288
2.780
3.240
2.028
5.250
3.275
5.385
1.942
p-Value
0.634
0.067
0.167
0.085
0.843
0.010
0.336
0.573
0.497
0.048
0.756
0.085
0.000
0.1 0.2 0.5 1
5 10
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
random effects model; same for fixed
Figure C-4. Meta-analysis of lymphoma and TCE exposure—highest exposure
groups. (The pooled estimate is in the bottom row. Symbol sizes reflect relative
weights of the studies. The horizontal midpoint of the bottom diamond represents
the pooled RR estimate, and the horizontal extremes depict the 95% CI limits.)
C.2.3. Discussion of Lymphoma Meta-Analysis Results
For the most part, the meta-analyses of the overall effect of TCE exposure on lymphoma
suggest a small, statistically significant increase in risk. The pooled estimate from the primary
random effects meta-analysis of the 16 studies was 1.23 (95% CI: 1.04, 1.44). This result was
not overly influenced by any single study, nor was it overly sensitive to individual RR estimate
selections. In terms of the statistical significance of the RRp estimate, the only alternate analysis
(involving either a study removal or an alternate RR estimate) that did not yield a statistically
significant RRp was the analysis in which the Zhao mortality RR estimate was substituted with
the incidence estimate, resulting in an RRp estimate of 1.19 (1.00, 1.41); although, as noted
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-21 DRAFT—DO NOT CITE OR QUOTE
-------
1 above, this substitution is considered clearly inferior to the Zhao mortality estimate that was used
2 in the primary analysis. Thus, the finding of an increased risk of lymphoma associated with TCE
3 exposure, though the increased risk is not large in magnitude, is fairly robust.
4 There is some evidence of potential publication bias in this data set; however, it is
5 uncertain that this is actually publication bias rather than an association between SE and effect
6 size resulting for some other reason, e.g., a difference in study populations or protocols in the
7 smaller studies. Furthermore, if there is publication bias in this data set, it does not appear to
8 account completely for the finding of an increased lymphoma risk.
9 Although there was some heterogeneity across the 16 studies, it was not statistically
10 significant (p = 0.10). The / value was 33%, suggesting low-to-moderate heterogeneity.
11 Similarly, when subgroup analyses were done of cohort and case-control studies separately, there
12 was some observable heterogeneity in each of the subgroups, but it was not statistically
13 significant in either case. I2 values were 10% for the cohort studies, suggesting low
14 heterogeneity, and 33% for the case-control studies, suggesting low-to-moderate heterogeneity.
15 In the subgroup analyses, the increased risk of lymphoma was strengthened in the cohort study
16 analysis and virtually eliminated in the case-control study analysis, although the subgroup RRp
17 estimates were not statistically significantly different under the random effects model. Study
18 design itself is unlikely to be an underlying cause of heterogeneity and, to the extent that it may
19 explain some of the differences across studies, is more probably a surrogate for some other
20 difference(s) across studies that may be associated with study design. Furthermore, other
21 potential sources of heterogeneity may be masked by the broad study design subgroupings. The
22 true source(s) of heterogeneity across these studies is an uncertainty. As discussed above, further
23 quantitative investigations of heterogeneity were ruled out because of database limitations. A
24 qualitative discussion of some potential sources of heterogeneity follows.
25 Study differences in exposure assessment approach, exposure prevalence, average
26 exposure intensity, and lymphoma classification are possible sources of heterogeneity. Many
27 studies included TCE assignment from information on job and task exposures, e.g., a
28 job-exposure matrix (JEM) (Siemiatycki, 1991; Morgan et al., 1998; Boice et al., 1999, 2006;
29 Zhao et al., 2005; Miligi et al., 2006; Seidler et al., 2007; Radican et al., 2008; Wang et al.,
30 2009), or from an exposure biomarker in either breath or urine (Axelson et al., 1994; Anttila et
31 al., 1995; Hansen et al., 2001). Three case-control studies relied on self-reported exposure to
32 TCE (Hardell et al., 1994; Nordstrom et al., 1998; Persson and Fredrikson, 1999).
33 Misclassification is possible with all exposure assessment approaches. No information is
34 available to judge the degree of possible misclassification bias associated with a particular
35 exposure assessment approach; it is quite possible that in some cohort studies, in which past
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure is inferred from various data sources, exposure misclassification may be as great as in
2 population-based or hospital-based case-control studies. Approaches based upon JEMs can
3 provide order-of-magnitude estimates that are useful for distinguishing groups of workers with
4 large differences in exposure; however, smaller differences usually cannot be reliably
5 distinguished (NRC, 2006). Biomonitoring can provide information on potential TCE exposure
6 in an individual, but the biomarkers used aren't necessarily specific for TCE and they reflect only
7 recent exposures. The lack of heterogeneity in the analysis of the highest exposure groups
8 provides some evidence of exposure misclassification as a source of heterogeneity in the overall
9 analysis.
10 General population studies have special problems in evaluating exposure, because the
11 subjects could have worked in any job or setting that is present within the population (Copeland
12 et al., 1977; Nelson et al., 1994; McGuire et al., 1998; 't Mannetje et al., 2002; NRC, 2006).
13 Low exposure prevalence in the four population case-control studies (Siemiatycki, 1991;
14 Miligi et al., 2006; Seidler et al., 2007; Wang et al., 2009) may be another source of
15 heterogeneity. Prevalence of TCE exposure among cases in the case-control studies was low,
16 ranging from 3% in Siemiatycki (1991) to 13% in Seidler et al. (2007) and Wang et al. (2009).
17 However, prevalence of high TCE exposure in these case-control studies was even rarer—3% of
18 all cases in Miligi et al. (2006) and Seidler et al. (2007), 2% in Wang et al. (2009), and less than
19 1% in Siemiatycki (1991). Low exposure prevalence, especially in the relatively large Miligi et
20 al. (2006) and Seidler et al. (2007) case-control studies (see Figure C-l), may be one of the
21 underlying characteristics differentiating the case-control and cohort studies and explaining some
22 of the heterogeneity across the studies.
23 Study differences in lymphoma groupings and in lymphoma classification schemes are
24 another potential source of heterogeneity in the meta-analysis. All studies included a broad but
25 sometimes slightly different group of lymphosarcoma, reticulum-cell sarcoma, and other
26 lymphoid tissue neoplasms, with the exception of the Nordstrom et al. (1998) case-control study,
27 which examined hairy cell leukemia, now considered a lymphoma, and the Zhao et al. (2005)
28 cohort study, which reported only results for all lymphohematopoietic cancers, including
29 nonlymphoid types. Persson and Fredrikson (1999) do not identify the classification system for
30 defining NHL, and Hardell et al. (1994) define NHL using the Rappaport classification system.
31 Miligi et al. (2006) used an NCI classification system and considered chronic lymphocytic
32 leukemias and NHLs together as lymphomas, while Seidler et al. (2007) used the REAL
33 classification system, which reclassifies lymphocytic leukemias and NHLs as lymphomas of
34 B-cell or T-cell origin. The cohort studies (except for Zhao et al.) and the case-control study of
35 Siemiatycki (1991) have some consistency in coding NHL, with NHL defined as lymphosarcoma
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-23 DRAFT—DO NOT CITE OR QUOTE
-------
1 and reticulum-cell sarcoma (ICD code 200) and other lymphoid tissue neoplasms (ICD 202)
2 using the ICD Revisions 7, 8, or 9. Revisions 7 and 8 are essentially the same with respect to
3 NHL; under Revision 9, the definition of NHL was broadened to include some neoplasms
4 previously classified as Hodgkin's lymphomas (Banks, 1992). Wang et al. (2009) refer to their
5 cases as "NHL" cases; however, according to the ICD-O classification system that they used,
6 their cases are more specifically various particular subtypes of malignant lymphoma (9590-9642,
7 9690-9701) and mast cell tumors (9740-9750) (Morton et al., 2003).
8 Twelve of the 16 studies categorized results by exposure level. Different exposure
9 metrics were used, and the purpose of combining results across the different highest exposure
10 groups was not to estimate an RRp associated with some level of exposure, but rather to see the
11 impacts of combining RR estimates that should be less affected by exposure misclassification.
12 In other words, the highest exposure category is more likely to represent a greater differential
13 TCE exposure compared to people in the referent group than the exposure differential for the
14 overall (typically any vs. none) exposure comparison. Thus, if TCE exposure increases the risk
15 of lymphoma, the effects should be more apparent in the highest exposure groups. Indeed, the
16 RRp estimate from the primary meta-analysis of the highest exposure group results was 1.57
17 (95% CI: 1.27, 1.94), which is greater than the RRp estimate of 1.23 (95% CI: 1.04, 1.44) from
18 the overall exposure analysis. This result for the highest exposure groups was not overly
19 influenced by any single study, nor was it overly sensitive to individual RR estimate selections.
20 Heterogeneity was not observed in any of the relevant analyses. The robustness of this finding
21 lends substantial support to a conclusion that TCE exposure increases the risk of lymphoma.
22
23 C.3. META-ANALYSIS FOR KIDNEY CANCER
24 C.3.1. Overall Effect of TCE Exposure
25 C.3.1.1. Selection of RR Estimates
26 The selected RR estimates for kidney cancer associated with TCE exposure from the
27 epidemiological studies are presented in Table C-6 for cohort studies and in Table C-7 for
28 case-control studies. The majority of the cohort studies reported results for all kidney cancers,
29 including cancers of the renal pelvis and ureter (i.e., ICD-7 180; ICD-8 and -9 189.0-189.2;
30 ICD-10 C64-C66); whereas the majority of the case-control studies focused on renal cell
31 carcinoma (RCC), which comprises roughly 85% of kidney cancers. Where both all kidney
32 cancer and RCC were reported, the primary analysis used the results for RCC, because RCC and
33 the other forms of kidney cancer are very different cancer types and it seemed preferable not to
34 combine them; the results for all kidney cancers were then used in a sensitivity analysis. The
35 preference for the RRC results alone is supported by the results in rodent cancer bioassays,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 where TCE-associated rat kidney tumors are observed in the renal tubular cells (Section 4.3.5),
2 and in metabolism studies, where the focus of studies for the GSH conjugation pathway
3 (considered the primary metabolic pathway for kidney toxicity) is in renal cortical and tubular
4 cells (Sections 3.3.3.2 and 4.3.6).
5 As for lymphoma, many of the studies provided RR estimates only for males and females
6 combined, and we are not aware of any basis for a sex difference in the effects of TCE on kidney
7 cancer risk; thus, wherever possible, RR estimates for males and females combined were used.
8 Of the three larger (in terms of number of cases) studies that did provide results separately by
9 sex, Dosemeci et al. (1999) suggest that there may be a sex difference for TCE exposure and
10 RCC (OR= 1.04 [95% CI: 0.6, 1.7] in males and 1.96 [1.0, 4.0] in females), while
11 Raaschou-Nielsen et al. (2003) report the same SIR (1.2) for both sexes and crude ORs
12 calculated from data from the Pesch et al. (2000) study (provided in a personal communication
13 from Baeta Pesch, Forschungsinstitut fur Arbeitsmedizin (BGFA), to Cheryl Scott, U.S. EPA,
14 21 February 2008) are 1.28 for males and 1.23 for females. Radican et al. (2008) and Hansen et
15 al. (2001) also present some results by sex, but both of these studies have too few cases to be
16 informative about a sex difference for kidney cancer.
17 Most of the selections in Tables C-6 and C-7 should be self-evident, but some are
18 discussed in more detail here, in the order the studies are presented in the tables. For Axelson et
19 al. (1994), in which a small subcohort of females was studied but only results for the larger male
20 subcohort were reported, the reported male-only results were used in the primary analysis;
21 however, as for lymphoma, an attempt was made to estimate the female contribution to an
22 overall RR estimate for both sexes and its impact on the meta-analysis. Axelson et al. (1994)
23 reported neither the observed nor the expected number of kidney cancer cases for females. It
24 was assumed that none were observed. To estimate the expected number, the expected number
25 for males was multiplied by the ratio of female-to-male person-years in the study and by the ratio
26 of female-to-male age-adjusted incidence rates for kidney cancer.2 The male results and the
27 estimated female contribution were then combined into an RR estimate for both sexes assuming
28 a Poisson distribution, and this alternate RR estimate for the Axelson et al. (1994) study was
29 used in a sensitivity analysis.
2Person-years for men and women <79 years were obtained from Axelson et al. (1994): 23516.5 and 3691.5,
respectively. Lifetime age-adjusted incidence rates for cancer of the kidney and renal pelvis for men and women
were obtained from the National Cancer Institute's 2000-2004 SEER-17 (Surveillance Epidemiology and End
Results from 17 geographical locations) database (http://seer.cancer.gov/statfacts/html/kidrp.html): 17.8/100,000
and 8.8/100,000, respectively. The calculation for estimating the expected number of cases in females in the cohort
assumes that the males and females have similar TCE exposures and that the relative distributions of age-related
incidence risk for the males and females in the cohort are adequately represented by the ratios of person-years and
lifetime incidence rates used in the calculation.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-25 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-6. Selected RR estimates for kidney cancer associated with TCE exposure (overall effect) from
cohort studies
Study
Anttila et a\.,
1995
Axelson et
al., 1994
Boice et al.,
1999
Greenland et
al., 1994
Hansen et al.,
2001
Morgan et al.,
1998
Raaschou-
Nielsen et al.,
2003
Radican et
al., 2008
Zhao et al.,
2005
RR
0.87
1.16
0.99
0.99
1.1
1.14
1.20
1.18
1.7
95%
LCL
0.32
0.42
0.4
0.30
0.3
0.51
0.94
0.47
0.38
95%
UCL
1.89
2.52
2.04
3.32
2.8
2.58
1.50
2.94
7.9
RR type
SIR
SIR
SMR
OR
SIR
Mortality
RR
SIR
Mortality
HR
Mortality
RR
log RR
-0.139
0.148
-0.010
-0.010
0.095
0.134
0.182
0.166
0.542
SE(log
RR)
0.408
0.408
0.378
0.613
0.500
0.415
0.199
0.468
0.775
Alternate RR
estimates
none
1.07(0.39,2.33)
with estimated
female contribution
to SIR added (see
text)
None
None
None
Published SMR
1.32(0.57,2.6)
1.20(0.98, 1.46)
forlCD-7 180
None
Incidence RR: 2.0
(0.47, 8.2)
Mortality RR no
lag: 0.89(0.22, 3.6)
Incidence RR no
lag :2.1 (0.56, 8.1)
Boice (2006) SMR:
2.22 (0.89, 4.57)
Comments
ICD-7180.
ICD-7 180. Results reported for males only,
but there was a small female component to
the cohort.
ICD-9 189.0-189.2. For potential routine
exposure. Results for any potential exposure
not reported.
Nested case-control study. ICD-8 codes not
specified, presumably all of 189.
ICD-7 180. Male and female results reported
separately; combined assuming Poisson
distribution.
ICD-9 189.0-189.2. Unpublished RR,
adjusted for age and sex (see text).
RCC.
ICD-8, -9 189.0, ICD-10 C64. Time variable
= age; covariates = sex and race.
Referent group is workers with no chemical
exposures.
ICD-9 189. Males only. Adjusted for age,
SES, time since first employment, exposure to
other carcinogens. 20-yr lag. Mortality results
reflect same number exposed cases (10 with
no lag) as do incidence results, so no reason
to prefer mortality results, but they are used in
primary analysis to avoid appearance of
"cherry-picking." Overall RR estimated by
combining across exposure groups (see text).
Boice (2006) cohort overlaps Zhao cohort; just
7 exposed deaths.
Co
I
~p
TO'
^3
58
ol
V TO
to
O
3
•
TO
o §
o s
H fc
H
W
-------
o
to
§3
VO Co'
Table C-7. Selected RR estimates for renal cell carcinoma associated with TCE exposure from case-control
studies"
Study
Bruning et
a\., 2003
Charbotel et
a\., 2006
Dosemeci et
a\., 1999
Pesch et a\.,
2000
Siemiatycki
1991
RR
estimate
2.47
1.88
1.30
1.24
0.8
95% LCL
1.36
0.89
0.9
D
0.3
95%
UCL
4.49
3.98
1.9
D
2.2
log RR
0.904
0.631
0.262
0.215
-0.223
SE(log
RR)
0.305
0.382
0.191
0.094
0.524
Alternate
RR
estimates
1.80(1.01,
3.20) for
longest job
held in
industry
with TCE
exposure
1.64(0.95,
2.84) for
full study
1.1 3 with
German
JEM
Comments
Self-assessed exposure. Adjusted for age, sex,
and smoking.
Subgroup with good level of confidence about
exp assessment. Matched on sex, age.
Adjusted for smoking, body mass index.
Adjusted forage, sex, smoking, hypertension
and/or use of diuretics and/or anti-hypertension
drugs, body mass index.
With JTEM (job task exposure matrix). Crude
OR calculated from data provided in personal
communication (see text).
"Kidney cancer." SE and 95% Cl calculated from
reported 90% CIs. Males only; adjusted for age,
income, and cigarette smoking index.
Co
I
~p
TO'
•
TO
aThe RR estimates are all ORs for incident cases.
bNot calculated.
o s
H fc
H
W
-------
1 For Boice et al. (1999), only results for "potential routine exposure" were reported for
2 kidney cancer. This is our preferred TCE exposure definition for the Boice study, because it was
3 considered to have less exposure misclassification than "any potential exposure;" however, since
4 the results for the latter definition were not presented, they could not be used in a sensitivity
5 analysis, as was done for lymphoma. Boice et al. (1999) report in general that the SMRs for
6 workers with any potential exposure "were similar to those for workers with daily potential
7 exposure." In their published paper, Morgan et al. (1998) present only SMRs for overall TCE
8 exposure, although the results from internal analyses are presented for exposure subgroups. RR
9 estimates for overall TCE exposure from the internal analyses of the Morgan et al. (1998) cohort
10 data were available from an unpublished report (Environmental Health Strategies, 1997); from
11 these, the RR estimate from the Cox model which included age and sex was selected, because
12 those are the variables deemed to be important in the published paper. The internal analysis RR
13 estimate was preferred for the primary analysis, and the published SMR result was used in a
14 sensitivity analysis. Raaschou-Nielsen et al. (2003) reported results for RCC and renal
15 pelvis/ureter separately. As discussed above, RCC estimates were used in the primary analysis,
16 and the results for both kidney cancer categories were combined (across sexes as well), assuming
17 a Poisson distribution, and used in a sensitivity analysis. For Radican et al. (2008), the Cox
18 model hazard ratio (HR) from the 2000 follow-up was used. In the Radican et al. (2008) Cox
19 regressions, age was the time variable, and sex and race were covariates. It should also be noted
20 that the referent group is composed of workers with no chemical exposures, not just no exposure
21 to TCE.
22 For Zhao et al. (2005), no results for an overall TCE effect are reported. We were unable
23 to obtain any overall estimates from the study authors, so, as a best estimate, as was done for
24 lymphoma, the results across the "medium" and "high" exposure groups were combined, under
25 assumptions of group independence, even though the exposure groups are not independent (the
26 "low" exposure group was the referent group in both cases). Unlike for lymphoma, adjustment
27 for exposure to other carcinogens made a considerable difference, so Zhao et al. (2005) also
28 present kidney results with this additional adjustment, with and without a 20-year lag. Estimates
29 of RR with this additional adjustment were selected over those without. In addition, a 20-year
30 lag seemed reasonable for kidney cancer, so the lagged estimates were preferred to the unlagged;
31 unlagged estimates were used in sensitivity analyses. Zhao et al. (2005) present RR estimates for
32 both incidence and mortality. Unlike for lymphoma, the number of exposed incident cases (10
33 with no lag) was identical to the number of deaths, so there was no reason to prefer the mortality
34 results over the incidence results. (In fact, there were more exposed incident cases [10 vs. 7]
35 after lagging.) However, the mortality results, which yield a lower RR estimate, were selected
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 for the primary analysis to avoid any appearance of "cherry-picking," and incidence RR
2 estimates were used in sensitivity analyses. A sensitivity analysis was also done using results
3 from Boice et al. (2006) in place of the Zhao et al. (2005) RR estimate. The cohorts for these
4 studies overlap, so they are not independent studies and should not be included in the
5 meta-analysis concurrently. Boice et al. (2006) report results for an overall TCE effect for
6 kidney cancer; however, the results are SMR estimates rather than internal comparisons and are
7 based on fewer exposed deaths (7), so either Zhao et al. (2005) estimate is preferred over the
8 Boice et al. (2006) estimate.
9 Regarding the case-control studies, for Briining et al. (2003), the results based on
10 self-assessed exposure were preferred because, although TCE exposure was probably under
11 ascertained with this measure, there were greater concerns about the result based on the alternate
12 measure reported—longest-held job in an industry with TCE exposure. Even though this study
13 was conducted in the Arnsberg region of Germany, an area with high prevalence of exposure to
14 TCE, the exposure prevalence in both cases (87%) and controls (79%) seemed inordinately high,
15 and this for not just any job in an industry with TCE exposure, but for the longest-held job.
16 Furthermore, Table V of Briining et al., which presents this result, states that the result is for
17 longest-held job in industries with TCE or tetrachloroethylene exposure. Additionally, some of
18 the industries with exposure to TCE presented in Table V have many jobs that would not entail
19 TCE exposure (e.g., white-collar workers), so the assessment based on industry alone likely has
20 substantial misclassification. Both of these—inclusion of tetrachloroethylene and exposure
21 assessment by industry—could result in overstating TCE exposure prevalence. Results based on
22 the longest-held-job measure were used in a sensitivity analysis.
23 For Charbotel et al. (2006), results from the analysis that considered "only job periods
24 with a good level of confidence for TCE exposure assessment" (Table 7 of Charbotel et al.,
25 2006) were preferred, as these estimates would presumably be less influenced by exposure
26 misclassification. Estimates from the full study analysis were used in a sensitivity analysis. For
27 Pesch et al. (2000), TCE results were presented for 2 different exposure assessments. Estimates
28 using the job-task-exposure-matrix (ITEM) approach were preferred because they seemed to
29 represent a more comprehensive exposure assessment (see Appendix B, Section II-4); estimates
30 based on the JEM approach were used in a sensitivity analysis. Furthermore, results were
31 presented only by exposure category, with no overall RR estimate reported. Case and control
32 numbers for the different exposure categories were kindly provided by Dr. Pesch (personal
33 communication from Baete Pesch, BGFA, to Cheryl Scott, U.S. EPA, 21 February 2008), and we
34 calculated crude overall ORs for males and females combined for each exposure assessment
35 approach.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 C.3.1.2. Results of Meta-Analyses
2 Results from some of the meta-analyses that were conducted on the epidemiological
3 studies of TCE and kidney cancer are summarized in Table C-8. The pooled estimate from the
4 primary random effects meta-analysis of the 14 studies was 1.25 (95% CI: 1.11, 1.41) (see
5 Figure C-5). As shown in Figure C-5, the analysis was dominated by 2 (contributing almost 70%
6 of the weight) or 3 (almost 80% of the weight) large studies. No single study was overly
7 influential; removal of individual studies resulted in RRp estimates that were all statistically
8 significant (all with/? < 0.005) and that ranged from 1.22 (with the removal of Briining) to 1.27
9 (with the removal of Raaschou-Nielsen).
10 Similarly, the RRp estimate was not highly sensitive to alternate RR estimate selections.
11 Use of the 10 alternate selections, individually, resulted in RRp estimates that were all
12 statistically significant (all with/? < 0.002) and that ranged from 1.19 to 1.27 (see Table C-8). In
13 fact, as can be seen in Table C-8, all but one of the alternates had negligible impact. The Zhao,
14 Axelson, Briining, and Charbotel original values and alternate selections were associated with
15 very little weight and, thus, have little influence in the RRp. The Raaschou-Nielsen value carried
16 more weight, but the alternate RR estimate was identical to the original, although with a
17 narrower CI, and so did not alter the RRp. Only the Pesch alternate (with the JEM exposure
18 assessment approach instead of the ITEM approach) had much impact, resulting in an RRp
19 estimate of 1.19 (95% CI: 1.07, 1.32). As noted above, the ITEM approach is preferred. The
20 JEM approach takes jobs into account but not tasks; thus, it is expected to have greater potential
21 for exposure misclassification. Indeed, a comparison of exposure prevalences for the
22 two approaches suggests that the JEM approach is less discriminating about exposure; 42% of
23 cases were defined as TCE-exposed under the JEM approach, but only 18% of cases were
24 exposed under the JTEM approach.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-30 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-8. Summary of some meta-analysis results for TCE (overall) and kidney cancer
Analysis
All studies
Cohort
Case-control
Alternate RR
selections3
Highest
exposure
groups
#of
studies
14
9
5
14
14
14
14
14
14
14
14
9
12
Model
Random
Fixed
Random
Fixed
Random
Fixed
Random
Random
Random
Random
Random
Random
Random
Random
Random
Random
Combined
RR estimate
1.25
1.25
1.16
1.16
1.41
1.32
1.25
1.27
1.25
1.26
1.25
1.24
1.25
1.19
1.59
1.53
95% LCL
1.11
1.11
0.96
0.96
1.08
1.13
1.11
1.13
1.11
1.11
1.11
1.10
1.11
1.07
1.26
1.23
95% UCL
1.41
1.41
1.40
1.40
1.83
1.54
1.40-1.41
1.43
1.41
1.41
1.40
1.39
1.41
1.32
2.01
1.91
Heterogeneity
None obs
None obs
Not significant
(P = 0.17)
None obs
None obs
None obs
None obs
None obs
None obs
None obs
None obs
None obs
None obs
Comments
Statistical significance not dependent on
single study. No apparent publication bias.
Not significant difference between CC and
cohort studies (p = 0.23).
Not significant difference between CC and
cohort studies (p = 0.29).
With 3 different alternates from Zhao (see
Table C-6).
With Boice (2006) study rather than Zhao
With estimated female contribution to
Axelson.
With Morgan published SMR.
With Raaschou-Nielsen all kidney cancer.
With Bruning longest job held in industry
with TCE.
With Charbotel full study
With Pesch JEM.
Using RR = 1 for Anttila, Axelson, and
Hansen (see text).
See Table C-10 for alternate RR selection
results.
Co
I
~p
TO'
•
TO
o s
H fc
"Changing the primary analysis by one alternate RR each time.
H
W
obs = observable.
-------
TCE and Kidney Cancer
Study name
Anttila 1995
Axelson 1994
Boice 1999
Greenland 1994
Hansen 2001
Morgan 1998 unpub RR
Raaschou-Nielsen 2003 RCC
Radican 2008
Zhao 2005 mort 20 y lag
Bruning 2003
Charbotel 2007- high conf re:exp
Dosemeci 1999
Pesch 2000 JTEM
Siemiatycki 1991
Statistics for each study
Risk ratio and 95% Cl
Risk
ratio
0.870
1.160
0.990
0.990
1.100
1.143
1.200
1.180
1.720
2.470
1.880
1.300
1.240
0.800
1.251
Lower
limit
0.391
0.521
0.472
0.298
0.413
0.507
0.950
0.472
0.377
1.359
0.889
0.895
1.030
0.287
1.110
Upper
limit
1.937
2.582
2.077
3.293
2.931
2.576
1.516
2.951
7.853
4.488
3.976
1.889
1.492
2.233
1.410
p-Value
0.7330
0.7162
0.9788
0.9869
0.8488
0.7472
0.1262
0.7234
0.4840
0.0030
0.0985
0.1687
0.0227
0.6700
0.0002
.-
0.1 0.2 0.5 1
10
random effects model; same for fixed
1
2
3
4
5
6
7
8
9
10
11
12
13
Figure C-5. Meta-analysis of kidney cancer and overall TCE exposure.
The pooled estimate is in the bottom row. Symbol sizes reflect relative weights
of the studies. The horizontal midpoint of the bottom diamond represents the
pooled RR estimate and the horizontal extremes depict the 95% CI limits.
There was no apparent heterogeneity across the 14 studies, i.e., the random effects model
and the fixed effect model gave the same results. Nonetheless, subgroup analyses were done
examining the cohort and case-control studies separately. With the random effects model (and
tau-squared not pooled across subgroups), the resulting RRp estimates were 1.16 (95% CI: 0.96,
1.40) for the cohort studies and 1.41 (1.08, 1.83) for the case-control studies. There was
heterogeneity in the case-control subgroup, but it was not statistically significant and the I2 value
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-32 DRAFT—DO NOT CITE OR QUOTE
-------
1 of 38% suggests that the extent of the heterogeneity in this subgroup was low-to-moderate. Nor
2 was the difference between the RRp estimates for the cohort and case-control subgroups
3 statistically significant under either the random effects model or the fixed effect model. Further
4 quantitative investigations of heterogeneity were not pursued because of database limitations
5 and, in any event, there is no evidence for heterogeneity of study results in this database. A
6 qualitative discussion of some potential sources of heterogeneity across studies is nonetheless
7 included in Section C.3.3.
8 As discussed in Section C.I, publication bias was examined in several different ways.
9 The funnel plot in Figure C-6 shows little relationship between RR estimate and study size, and,
10 indeed, none of the other tests performed found any evidence of publication bias. Duval and
11 Tweedie's trim-and-fill procedure, for example, determined that no studies were missing from
12 the funnel plot, i.e., there was no asymmetry to counterbalance. Similarly, the results of a
13 cumulative meta-analysis, incorporating studies with increasing SE one at a time, shows no
14 evidence of a trend of increasing effect size with addition of the less precise studies. Including
15 the 3 most precise studies, reflecting 78% of the weight, the RRp goes from 1.24 to 1.22 to 1.23.
16 The addition of the Briining study brings the RRp to 1.32 and the weight to 82%. After the
17 addition of the next 5 studies, the RRp stabilizes at about 1.26, and further addition of the 5 least
18 precise studies has little impact.
19
20 C.3.2. Kidney Cancer Effect in the Highest Exposure Groups
21 C.3.2.1. Selection ofRR Estimates
22 The selected RR estimates for kidney cancer in the highest TCE exposure categories, for
23 studies that provided such estimates, are presented in Table C-9. Five of the 9 cohort studies and
24 4 of the 5 case-control studies reported kidney cancer risk estimates categorized by exposure
25 level. As in Section C.3.1.1 for the overall risk estimates, estimates for RCC were preferentially
26 selected when presented, and, wherever possible, RR estimates for males and females combined
27 were used.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-33 DRAFT—DO NOT CITE OR QUOTE
-------
Funnel Plot of Standard Error by Log risk ratio
o.o
0.2
LJJ
•5
re
•O
W
0.4
0.6
0.8
O
o
-2.0
-1.5
-1.0 -0.5
0.0
0.5
1.0
1.5
2.0
1
2
Log risk ratio
Figure C-6. Funnel plot of SE by log RR estimate for TCE and kidney
cancer studies
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-34 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-9. Selected RR estimates for kidney cancer risk in highest TCE exposure groups
Study
Anttila et
al., 1995
Axelson et
al., 1994
Boice et al.,
1999
Hansen et
al.,2001
Morgan et
al., 1998
Raaschou-
Nielsen et
al., 2003
Radican et
al., 2008
RR
0.69
1.59
1.7
1.11
95%
LCL
0.22
0.68
1.1
0.35
95%
UCL
2.12
3.71
2.4
3.49
Exposure
category
100+ |jmol/L
U-TCA a
>2 yr exposure
and 100+ mg/L
U-TCA
>5 yr exp
>1080 mos x
mg/m3
High cumulative
exposure score
>5 yrs in
subcohort with
expected higher
exposure levels
>25 unit-yr
log RR
-0.371
0.464
0.531
0.104
SE(log
RR)
0.578
0.433
0.183
0.582
Alternate RR
estimates
1 .0 assumed
1 .0 assumed
None
1 .0 assumed
1.89(0.85,
4.23) for
med/high
peak vs.
low/no
1.4(0.99, 1.9)
ICD-7180
>5 yrs in total
cohort
Blair et al.
(1998)
incidence RR
0.9 (0.3, 3.2)
Comments
Reported high exposure group results for
some cancer sites but not kidney.
Reported high exposure group results for
some cancer sites but not kidney.
Mortality RR. ICD-9 189.0-189.2. For
potential routine or intermittent exposure.
adjusted for date of birth, dates 1 and
last employed, race, and sex. Referent
group is workers not exposed to any
solvent.
Reported high exposure group results for
some cancer sites but not kidney.
Mortality RR. ICD-9 189.0-189.2.
Adjusted for age and sex.
SIR. RCC.
Mortality HR. ICD-8, -9 189.0, ICD-10
C64. Male and female results presented
separately and combined (see text).
Referent group is workers with no
chemical exposures.
Co
I
~p
TO'
^3
58
ol
V TO
•
TO
o §
o s
H fc
H
W
-------
o
to
Table C-9. Selected RR estimates for kidney cancer risk in highest TCE exposure groups (continued)
Study
Zhao et a\.,
2005
Bruning et
a\., 2003
Charbotel et
a\., 2006
RR
7.40
2.69
3.34
95%
LCL
0.47
0.84
1.27
95%
UCL
116
8.66
8.74
Exposure
category
High exposure
score
>20 yrs
self-assessed
exposure
High cumulative
dose
log RR
2.00
0.990
1.21
SE(log
RR)
1.41
0.595
0.492
Alternate RR
estimates
Mortality RR:
1.82(0.09,
38.6)
Incidence RR
no lag: 7.71
(0.65,91.4)
Mortality RR
no lag: 0.96
(0.09, 9.91)
Boice 2006
mortality RR:
2.12(0.63,
7. 11) for
>5 yrs as test
stand
mechanic;
3.13
(0.74, 13.2) for
>4 test-yr
engine flush
None
3.80(1.27,
1 1 .40) for high
cum + peaks
1.96(0.71,
5.37) for high
cum + peaks
in full study
2.63 (0.79,
8. 83) for high
cum in full
study
Comments
Incidence RR. ICD-9 189. Males only.
Adjusted for age, SES, time since first
employment, exposure to other
carcinogens. 20-yrlag. Incidence results
reflect more exposed cases (4 with no
lag) than do mortality results (3), so they
are used in primary analysis.
Incidence OR. RCC. Adjusted forage,
sex, and smoking.
Incidence OR. RCC. In subgroup with
good level of confidence for TCE
exposure. Adjusted for smoking and
body mass index. Matched on sex and
age. Alternate full study estimates were
additionally adjusted for exposure to
cutting fluids and other petroleum oils.
VO Co'
Co
I
~p
TO'
^3
58
ol
V TO
U> ^
ON O
3
•
TO
o §
o s
H fc
H
W
-------
o
to
Table C-9. Selected RR estimates for kidney cancer risk in highest TCE exposure groups (continued)
Co'
§
^.
Co'
1
Study
Pesch et a\.,
2000
Siemiatycki
1991
RR
1.4
0.8
95%
LCL
0.9
0.2
95%
UCL
2.1
3.4
Exposure
category
Substantial
Substantial
log RR
0.336
-0.233
SE(log
RR)
0.219
0.736
Alternate RR
estimates
1.2(0.9, 1.7)
for JEM
none
Comments
Incidence OR. RCC. JTEM approach.
Adjusted for age, study center, and
smoking. Sexes combined.
Incidence OR. Kidney cancer. SE and
95% Cl calculated from reported 90%
CIs. Males only; adjusted for age,
income, and cigarette smoking index.
TO'
aMean personal trichloroacetic acid in urine. 1 umol/L = 0.1634 mg/L.
•
TO
o s
H fc
H
W
-------
1 Three of the 9 cohort studies (Anttila et al., 1995; Axelson et al., 1994; Hansen et al.,
2 2001) did not report kidney cancer risk estimates categorized by exposure level even though
3 these same studies reported such estimates for selected other cancer sites. To address this
4 reporting bias, attempts were made to obtain the results from the primary investigators, and,
5 failing that, an alternate analysis was performed in which null estimates (RR =1.0) were
6 included for all 3 studies. This alternate analysis was then used as the main analysis, e.g., the
7 basis of comparison for the sensitivity analyses. For the SE (of the logRR) estimates for these
8 null estimates, SE estimates from other sites for which highest-exposure-group results were
9 available were used. For Anttila et al. (1995), the SE estimate for liver cancer in the highest
10 exposure group was used, because liver cancer and kidney cancer had similar numbers of cases
11 in the overall study (5 and 6, respectively). For Axelson et al. (1994), the SE estimate for NHL
12 in the highest exposure group was used, because NHL and kidney cancer had similar numbers of
13 cases in the overall study (5 and 6, respectively). For Hansen et al. (2001), the SE estimate for
14 NHL in the highest exposure group was used, because NHL was the only cancer site of interest
15 in this assessment for which highest-exposure-group results were available.
16 For Boice et al. (1999), only results for workers with "any potential exposure" (rather
17 than "potential routine exposure") were presented by exposure category, and the referent group is
18 workers not exposed to any solvent. For Morgan et al. (1998), the primary analysis used results
19 for the cumulative exposure metric, and a sensitivity analysis was done with the results for the
20 peak exposure metric.
21 For Radican et al. (2008), it should be noted that the referent group is workers with no
22 chemical exposures, not just no TCE exposure. In addition, exposure group results were
23 reported separately for males and females and were combined for this assessment using
24 inverse-variance weighting, as in a fixed effect meta-analysis. Radican et al. (2008) present only
25 mortality HR estimates by exposure group; however, in an earlier follow-up of this same cohort,
26 Blair et al. (1998) present both incidence and mortality RR estimates by exposure group. The
27 mortality RR estimate based on the more recent follow-up of Radican et al. (2008) (6 deaths in
28 the highest exposure group) was used in the primary analysis, while the incidence RR estimate
29 based on similarly combined results from Blair et al. (1998) (4 cases) was used as an alternate
30 estimate in a sensitivity analysis.
31 Zhao et al. (2005) present kidney cancer RR estimates adjusted for exposure to other
32 carcinogens, because, unlike for lymphoma, this adjustment made a considerable difference.
33 Estimates of RR with this additional adjustment were selected over those without. Furthermore,
34 the kidney results were presented with and without a 20-year lag. A 20-year lag seemed
35 reasonable for kidney cancer, so the lagged estimates were preferred to the unlagged; unlagged
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimates were used in sensitivity analyses. In addition, the incidence results reflect more cases
2 (4 with no lag) in the highest exposure group than do the mortality results (3), so the incidence
3 result (with the 20-year lag) was used for the primary analysis, and the unlagged incidence result
4 and the mortality results were used in a sensitivity analysis. Sensitivity analyses were also done
5 using results from Boice et al. (2006) in place of the Zhao et al. (2005) RR estimate. The cohorts
6 for these studies overlap, so they are not independent studies. Boice et al. (2006) report
7 mortality RR estimates for kidney cancer by years worked as a test stand mechanic, a job with
8 potential TCE exposure, and by a measure that weighted years with potential exposure from
9 engine flushing by the number of flushes each year. No results were presented for a third metric,
10 years worked with potential exposure to any TCE, because the Cox proportional hazards model
11 did not converge. The Boice et al. (2006) estimates are adjusted for years of birth and hire and
12 for hydrazine exposure.
13 For Charbotel et al. (2006), results from the analysis that considered "only job periods
14 with a good level of confidence for TCE exposure assessment" (Table 7 of Charbotel et al.,
15 2006) were preferred, as these estimates would presumably be less influenced by exposure
16 misclassification. Estimates from the full study analysis, additionally adjusted for exposure to
17 cutting fluids and other petroleum oils, were used in a sensitivity analysis. Additionally, the high
18 cumulative dose results were preferred, but the results for high cumulative dose + peaks were
19 included in sensitivity analyses. For Pesch et al. (2000), TCE results were presented for
20 two different exposure assessments. As discussed above, estimates using the ITEM approach
21 were preferred because they seemed to represent a more comprehensive exposure assessment;
22 estimates based on the JEM approach were used in a sensitivity analysis.
23
24 C.3.2.2. Results of Meta-Analyses
25 Results from the meta-analyses that were conducted for kidney cancer in the highest
26 exposure groups are summarized at the bottom of Table C-8 and reported in more detail in
27 Table C-10. The pooled RR estimate from the random effects meta-analysis of the 9 studies with
28 results presented for exposure groups was 1.59 (95% CI: 1.26, 2.01) (see Figure C-7). The RRp
29 estimate from the primary random effects meta-analysis with null RR estimates (i.e., 1.0)
30 included for Anttila, Axelson, and Hansen to address reporting bias (see above) was 1.53
31 (1.23, 1.91) (see Figure C-8). The inclusion of these 3 additional studies contributed just under
32 8% of the total weight. As with the overall kidney cancer meta-analyses, the meta-analyses of
33 the highest-exposure groups were dominated by 2 studies (Raaschou-Nielsen and Pesch), which
34 provided about 66% of the weight. No single study was overly influential; removal of individual
35 studies resulted in RRp estimates that were all statistically significant (all with/? < 0.02) and that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 ranged from 1.43 (with the removal of Raaschou-Nielsen) to 1.58 (with the removal of Boice
2 [1999]orPesch).
3 Similarly, the RRp estimate was not highly sensitive to alternate RR estimate selections.
4 Use of the 12 alternate selections, individually, resulted in RRp estimates that were all
5 statistically significant (all with/? < 0.002) and that ranged from 1.42 to 1.55, with all but 2 of
6 the alternate selections yielding RRp estimates in the narrow range of 1.49-1.55 (see
7 Table C-10). The lowest RRp estimates, 1.42 in both cases, were obtained when the alternate
8 selections involved the 2 large studies. One of the alternate selections was for Raaschou-
9 Nielsen, with a highest-exposure group estimate for all kidney cancer in the total cohort, rather
10 than RCC in the subcohort expected to have higher exposure levels. The latter value is strongly
11 preferred because, as discussed above, the subcohort is likely to have less exposure
12 misclassification. Furthermore, RCC is very different from other types of kidney cancer, and
13 TCE, if an etiological factor, may not be etiologically associated with all kidney cancers, so
14 using the broad category may dilute a true association with RCC, if one exists. The other
15 alternate selection with a considerable impact on the RRp estimate was for Pesch, with the
16 highest exposure group result based on the JEM exposure assessment approach, rather than the
17 ITEM approach. As discussed above, the ITEM approach is preferred because it seemed to be a
18 more comprehensive and discriminating approach, taking actual job tasks into account, rather
19 than just larger job categories. Thus, although results with these alternate selections are
20 presented for comprehensiveness and transparency, the primary analysis is believed to reflect
21 better the potential association between kidney cancer (in particular, RCC) and TCE exposure.
22 There was no observable heterogeneity across the studies for any of the meta-analyses
23 conducted with the highest-exposure groups, including those in which RR values for Anttila,
24 Axelson, and Hansen were assumed. No subgroup analyses (e.g., cohort vs. case-control studies)
25 were done with the highest exposure group results.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-40 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-10. Summary of some meta-analysis results for TCE (highest exposure groups) and kidney cancer
Analysis
Analysis
based on
reported
results
Primary
analysis
Alternate RR
selections3
Model
Random
Random
Random
Random
Random
Random
Random
Random
Random
Combined
RR estimate
1.59
1.53
1.52
1.55
1.42
1.51-1.54
1.53-1.54
1.49-1.52
1.42
95% LCL
1.26
1.23
1.22
1.24
1.15
1.21-1.23
1.23-1.24
1.19-1.22
1.16
95% UCL
2.01
1.91
1.90
1.94
1.75
1.89-1.92
1.91-1.92
1.86-1.91
1.74
Heterogeneity
None obs
(fixed =
random)
None obs
None obs
None obs
None obs
None obs
None obs
None obs
None obs
Comments
Includes assumed values for Anttila, Axelson,
and Hansen (see text).
Statistical significance not dependent on single
study.
With Blair et al. (1998) incidence RR instead of
Radican mortality HR.
With Morgan peak metric.
With Raaschou-Nielsen for all kidney cancer
>5 yrs in total cohort.
With Zhao incidence unlagged and mortality
with and without lag.
With Boice (2006) alternates for Zhao (see text).
With Charbotel high cumulative dose + peaks in
subgroup; and high cumulative dose and high
cumulative dose + peaks in full study
additionally adjusted for exposure to cutting
fluids and other petroleum oils..
With Pesch JEM.
Co
I
~p
TO'
•
TO
o s
H fc
""Changing the primary analysis by one alternate RR each time.
obs = observable.
H
W
-------
TCE and Kidney Cancer - highest exposure groups
Study name Statistics for each study
Risk Lower Upper
ratio limit limit p-Value
Boice1999 0.690 0.222 2.142 0.5208
Morgan 1998 1.590 0.681 3.714 0.2840
Raaschou-Nielsen 2003 1.700 1.189 2.431 0.0037
Radican2008 1.110 0.355 3.470 0.8576
Zhao 2005 inc 20y lag 7.400 0.471 116.249 0.1544
Bruning2003 2.690 0.838 8.634 0.0963
Charbotel 2007 good conf re: exp 3.340 1.273 8.761 0.0142
Pesch 2000 - JTEM 1.400 0.911 2.151 0.1244
Siemiatycki 1991 0.800 0.189 3.385 0.7618
1.586 1.255 2.006 0.0001
Risk ratio and 95% Cl
•
i
-4
-m-
+
i
0.1 0.2 0.5 1
10
random effects model
1
2
3
4
5
6
1
Figure C-7. Meta-analysis of kidney cancer and TCE exposure—highest
exposure groups. The pooled estimate is in the bottom row. Symbol sizes
reflect relative weights of the studies. The horizontal midpoint of the bottom
diamond represents the pooled RR estimate and the horizontal extremes depict the
95% CI limits.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-42 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Kidney Cancer - highest exposure groups
Study name Statistics for each study
Risk Lower Upper
ratio limit limit p-Value
Boice1999 0.690 0.222 2.142 0.5208
Morgan 1998 1.590 0.681 3.714 0.2840
Raaschou-Nielsen 2003 1.700 1.189 2.431 0.0037
Radican2008 1.110 0.355 3.470 0.8576
Zhao 2005 inc 20y lag 7.400 0.471 116.249 0.1544
Bruning2003 2.690 0.838 8.634 0.0963
Charbotel 2007 good confre:exp 3.340 1.273 8.761 0.0142
Pesch 2000 - JTEM 1.400 0.911 2.151 0.1244
Siemiatycki 1991 0.800 0.189 3.385 0.7618
Antilla 1.000 0.250 3.998 1.0000
Axelson 1.000 0.141 7.099 1.0000
Hansen 1.000 0.323 3.098 1.0000
1.531 1.225 1.913 0.0002
Risk ratio and 95% Cl
•
H
-4
-•-
^
•
0.1 0.2 0.5 1
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
random effects model; same for fixed
Figure C-8. Meta-analysis of kidney cancer and TCE exposure—highest
exposure groups, with assumed null RR estimates for Anttila, Axelson, and
Hansen (see text).
C.3.3. Discussion of Kidney Cancer Meta-Analysis Results
For the most part, the meta-analyses of the overall effect of TCE exposure on kidney
cancer suggest a small, statistically significant increase in risk. The pooled estimate from the
primary random effects meta-analysis of the 14 studies was 1.25 (95% CI: 1.11, 1.41). Although
the analysis was dominated by 2-3 large studies that contribute 70-80% of the weight, the
pooled estimate was not overly influenced by any single study, nor was it overly sensitive to
individual RR estimate selections. The largest downward impacts were from the removal of the
Briining study, resulting in an RRp estimate of 1.22 (95% CI: 1.08, 1.37), and from the
substitution of the Pesch JTEM RR estimate with the RR estimate based on the JEM approach,
resulting in an RRp estimate of 1.19 (1.07, 1.32). Thus, the finding of an increased risk of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-43 DRAFT—DO NOT CITE OR QUOTE
-------
1 kidney cancer associated with TCE exposure is robust. Furthermore, there is no evidence of
2 publication bias in this data set.
3 In addition, there was no heterogeneity observed across the results of the 14 studies.
4 When subgroup analyses were done of cohort and case-control studies separately, there was
5 some observable heterogeneity among the case-control studies, but it was not statistically
6 significant (p = 0.17) and the I2 value of 3 8% suggested the extent of the heterogeneity was low-
7 to-moderate. The increased risk of kidney cancer was strengthened in the case-control study
8 analysis and weakened in the cohort study analysis, but the difference between the 2 RRp
9 estimates was not statistically significant. One difference between the case-control and cohort
10 studies is that the case-control studies were of RCC and almost all of the cohort studies were of
11 all kidney cancers, including renal pelvis. As discussed above, RCC is very different from other
12 types of kidney cancer, and TCE, if an etiological factor, may not be etiologically associated
13 with all kidney cancers, so using the broad category may dilute a true association with RCC, if
14 one exists.
15 With respect to the nonsignificant heterogeneity in the 5 case-control studies, these
16 studies differ in TCE exposure potential to the underlying population from which case and
17 control subjects were identified, and this may be a source of some heterogeneity. Prevalence of
18 exposure to TCE among cases in these studies was 27% in Charbotel et al. (2006) (for
19 high-level-of-confidence jobs), 18% in Briining et al. (2003) (for self-assessed exposure), 18% in
20 Pesch et al. (2000), 13% in Dosemeci et al. (1999) and 1% in Siemiatycki (1991). Both Briining
21 et al. (2003) and Charbotel et al. (2006) are studies designed specifically to assess RCC and TCE
22 exposure. These studies were carried out in geographical areas with both a high prevalence and
23 a high degree of TCE exposure. Some information is provided in these and accompanying
24 papers to describe the nature of exposure, making it possible to estimate the order of magnitude
25 of exposure, even though there were no direct measurements (Cherrie et al., 2001; Briining et al.,
26 2003; Fevotte et al., 2006). The Charbotel et al. (2006) study was carried out in the Arve Valley
27 region in France, where TCE exposure was through metal-degreasing activity in small shops
28 involved in the manufacturing of screws and precision metal parts (Fevotte et al., 2006).
29 Industrial hygiene data from shops in this area indicated high intensity TCE exposures of
30 100 ppm or higher, particularly from exposures from hot degreasing processes. Considering
31 exposure only from the j obs with a high level of confidence about exposure, 18% of exposed
32 cases were identified with high cumulative exposure to TCE. The source population in the
33 Briining et al. (2003) study includes the Arnsberg region in Germany, which also has a high
34 prevalence of TCE exposure. A large number of small companies used TCE in metal degreasing
35 in small workrooms. Subjects in this study also described neurological symptoms previously
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-44 DRAFT—DO NOT CITE OR QUOTE
-------
1 associated with higher TCE intensities. While subjects in the Briining et al. (2003) study had
2 potential high TCE exposure intensity, average TCE exposure in this study is considered lower
3 than that in the Charbotel et al. (2006) study because the base population was enlarged beyond
4 the Arnsberg region to areas which did not have the same focus of industry.
5 Siemiatycki (1991), Dosemeci et al. (1999), and Pesch et al. (2000) are population-based
6 studies. Pesch et al. (2000) includes the Arnsberg area and 4 other regions. Sources of exposure
7 to TCE and other chlorinated solvents are much less well defined, and most subjects identified
8 with TCE exposure probably had minimal contact; estimated average concentrations to exposed
9 subjects were of about 10 ppm or less (NRC, 2006). Neither Dosemeci et al. (1999) nor
10 Siemiatycki (1991) describe the nature of the TCE exposure. TCE exposure potential in these
11 studies is likely lower than in the three other studies and closer to background. Furthermore, the
12 use of generic job-exposure-matrices for exposure assessment in these studies may result in a
13 greater potential for exposure misclassification bias.
14 Nine of the 14 studies categorized results by exposure level. Three other studies reported
15 results for other cancer sites by exposure level, but not kidney cancer; thus, to address this
16 reporting bias, null values (i.e., RR estimates of 1.0) were used for these studies. Different
17 exposure metrics were used in the various studies, and the purpose of combining results across
18 the different highest exposure groups was not to estimate an RRp associated with some level of
19 exposure, but rather to see the impacts of combining RR estimates that should be less affected by
20 exposure misclassification. In other words, the highest exposure category is more likely to
21 represent a greater differential TCE exposure compared to people in the referent group than the
22 exposure differential for the overall (typically any vs. none) exposure comparison. Thus, if TCE
23 exposure increases the risk of kidney cancer, the effects should be more apparent in the highest
24 exposure groups. Indeed, the RRp estimate from the primary meta-analysis of the highest
25 exposure group results was 1.53 (95% CI: 1.23, 1.91), which is greater than the RRp estimate of
26 1.25 (95% CI: 1.11, 1.41) from the overall exposure analysis. This result for the highest
27 exposure groups was not overly influenced by any single study, nor was it overly sensitive to
28 individual RR estimate selections. Heterogeneity was not observed in any of the analyses. The
29 robustness of this finding lends substantial support to a conclusion that TCE exposure increases
30 the risk of kidney cancer.
31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-45 DRAFT—DO NOT CITE OR QUOTE
-------
1 C.4. META-ANALYSIS FOR LIVER CANCER
2 C.4.1. Overall Effect of TCE Exposure
3 C'.4.1.1. Selection of RR Estimates
4 The selected RR estimates for liver cancer associated with TCE exposure from the
5 epidemiological studies are presented in Table C-l 1. There were no case-control studies for
6 liver cancer and TCE exposure that were selected for inclusion in the meta-analysis (see
7 Appendix B, Section II-9), so all of the relevant studies are cohort studies. All of the studies
8 reported results for liver cancers plus cancers of the gall bladder and extrahepatic biliary
9 passages (i.e., ICD-7 155.0 + 155.2; ICD-8 and -9 155 + 156). Three of the studies also report
10 results for liver cancer alone (ICD-7 155.0; ICD-8 and -9 155). For the primary analysis, results
11 for cancers of the liver, gall bladder, and biliary passages combined were selected, for the sake of
12 consistency, since these were reported in all the studies. An alternate analysis was also done
13 using results for liver cancer alone for the 3 studies that reported them and the combined liver
14 cancer results for the remainder of the studies.
15 As for lymphoma and kidney cancer, many of the studies provided RR estimates only for
16 males and females combined, and we are not aware of any basis for a sex difference in the
17 effects of TCE on liver cancer risk; thus, wherever possible, RR estimates for males and females
18 combined were used. The only study of much size (in terms of number of liver cancer cases)
19 that provided results separately by sex was Raaschou-Nielsen (2003). The results of this study
20 suggest that liver cancer risk in females might be slightly higher than the risk in males, but the
21 number of female cases is small (primary liver cancer SIR: males 1.1 [95% CI: 0.74, 1.64;
22 27 cases], females 2.8 [1.13, 5.80; 7 cases]; gallbladder and biliary passage cancers SIR:
23 males 1.1 [0.61, 1.87; 14 cases]; females 2.8 [1.28, 5.34; 9 cases]). Radican et al. (2008) report
24 HRs for liver/biliary passage cancers combined of 1.36 (95% CI: 0.59, 3.11; 28 deaths) for males
25 and 0.74 (95% CI: 0.18, 2.97; 3 deaths) for females, but these results are based on fewer cases,
26 especially in females.
27
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-46 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-ll. Selected RR estimates for liver cancer associated with TCE exposure (overall effect) from cohort
studies
Study
Anttila et a\.,
1995
Axelson et
al., 1994
Boice et al.,
1999
Greenland et
al., 1994
Hansen et al.,
2001
Morgan et al.,
1998
Raaschou-
Nielsen et al.,
2003
Radican et
al., 2008
Boice et al.,
2006
RR
1.89
1.41
0.54
0.54
2.1
1.48
1.35
1.12
1.28
95%
LCL
0.86
0.38
0.15
0.11
0.7
0.56
1.03
0.57
0.35
95%
UCL
3.59
3.60
1.38
2.63
5.0
3.91
1.77
2.19
3.27
RR type
SIR
SIR
SMR
OR
SIR
SMR
SIR
Mortality
HR
SMR
log RR
0.637
0.344
-0.616
-0.616
0.742
0.393
0.300
0.113
0.247
SE(log
RR)
0.333
0.5
0.5
0.810
0.447
0.495
0.138
0.343
0.5
Alternate RR
estimates
2.27 (0.74, 5.29)
for 155.0 alone
1.34(0.36, 3.42)
with estimated
female
contribution to
SIR added (see
text)
0.81 (0.45, 1.33)
for any potential
exposure
None
None
Published SMR
0.98(0.36,2.13)
1.28(0.89, 1.80)
for ICD-7 155.0
1.25(0.31,4.97)
for ICD-8, -9
155.0
1.0 assumed for
Zhao et al.
(2005)
Comments
ICD-7 155.0 + 155.1; combined assuming Poisson
distribution.
ICD-7 155. Results reported for males only, but
there was a small female component to the cohort.
ICD-9 155 + 156. For potential routine exposure.
ICD-8 155 + 156. Nested case-control study.
ICD-7 155. Male and female results reported
separately; combined assuming Poisson
distribution.
ICD-9 155 + 156. Unpublished RR, adjusted for
age and sex (see text).
ICD-7 155.0 + 155.1. Results for males and
females and different liver cancer types reported
separately; combined assuming Poisson
distribution.
ICD-8, -9 155 + 156, ICD-10 C22-C24. Time
variable = age; covariates = sex, race. Referent
group is workers with no chemical exposures.
ICD-9 155 + 156. Boice et al. (2006) used in lieu
of Zhao et al. (2005) because Zhao et al. (2005)
do not report liver cancer results. Boice (2006)
cohort overlaps Zhao cohort.
Co
I
~p
TO'
•
TO
o s
H fc
H
W
-------
1 Most of the selections in Table C-l 1 should be self-evident, but some are discussed in
2 more detail here, in the order the studies are presented in the table. For Axelson et al. (1994), in
3 which a small subcohort of females was studied but only results for the larger male subcohort
4 were reported, the reported male-only results were used in the primary analysis; however, as for
5 lymphoma and kidney cancer, an attempt was made to estimate the female contribution to an
6 overall RR estimate for both sexes and its impact on the meta-analysis. Axelson et al. (1994)
7 reported that there were no cases of liver cancer observed in females, but the expected number
8 was not presented. To estimate the expected number, the expected number for males was
9 multiplied by the ratio of female-to-male person-years in the study and by the ratio of female-to-
10 male age-adjusted incidence rates for liver cancer. The male results and the estimated female
11 contribution were then combined into an RR estimate for both sexes assuming a Poisson
12 distribution, and this alternate RR estimate for the Axelson et al. (1994) study was used in a
13 sensitivity analysis.
14 For Boice et al. (1999), results for "potential routine exposure" were selected for the
15 primary analysis, because this exposure category was considered to have less exposure
16 misclassification, and results for "any potential exposure" were used in a sensitivity analysis. To
17 estimate the SE(logRR) for the alternate RR selection, it was assumed that the number of
18 exposed cases (deaths) was 15. The actual number was not presented, but 15 was the number
19 that allowed us to reproduce the reported CIs. The number suggested by exposure level in Boice
20 et al. (1999) Table 9 is 13; however, it may be that exposure level data were not available for all
21 the cases. In their published paper, Morgan et al. (1998) present only SMRs for overall TCE
22 exposure, although the results from internal analyses are presented for exposure subgroups. RR
23 estimates for overall TCE exposure from the internal analyses of the Morgan et al. (1998) cohort
24 data were available from an unpublished report (Environmental Health Strategies, 1997); from
25 these, the RR estimate from the Cox model which included age and sex was selected, because
26 those are the variables deemed to be important in the published paper. The internal analysis RR
27 estimate was preferred for the primary analysis, and the published SMR result was used in a
28 sensitivity analysis.
29 Raaschou-Nielsen et al. (2003) reported results for primary liver cancer (ICD-7 155.0),
30 gallbladder and biliary passage cancers (ICD-7 155.1), and unspecified liver cancers (ICD-7 156)
31 separately. As discussed above, RR estimates for cancers of the liver, gall bladder, and biliary
32 passages combined were preferred for the primary analysis; thus, the results for primary liver
33 cancer and gallbladder/biliary passage cancers were combined (across sexes as well), assuming a
34 Poisson distribution. The results for primary liver cancer only (similarly combined across sexes)
35 were used in an alternate analysis. The results for unspecified liver cancers (ICD-7 156) were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-48 DRAFT—DO NOT CITE OR QUOTE
-------
1 not included in any analyses because, under the ICD-7 coding, 156 can include secondary liver
2 cancers. For Radican et al. (2008), the Cox model hazard ratio (HR) from the 2000 follow-up
3 was used. In the Radican et al. (2008) Cox regressions, age was the time variable, and sex and
4 race were covariates. It should also be noted that the referent group is composed of workers with
5 no chemical exposures, not just no exposure to TCE.
6 Zhao et al. (2005) did not present RR estimates for liver cancer; thus, results from Boice
7 et al. (2006) were used in the primary analysis. The cohorts for these studies overlap, so they are
8 not independent studies. Zhao et al. (2005), however, was our preferred study for lymphoma and
9 kidney cancer results; thus, in a sensitivity analysis, a null value (RR = 1.0) was assumed for
10 Zhao et al. (2005) to address the potential reporting bias. The SE estimate for kidney cancer
11 (incidence with 0 lag) was used as the SE for the liver cancer. (It is not certain that there was a
12 reporting bias in this case. In the "Methods" section of their paper, Zhao et al. [2005] list the
13 cancer sites examined in the cohort, and liver was not listed; it is not clear if the list of sites was
14 determined a priori or post hoc.) Also, on the issue of potential reporting bias, the Siemiatycki
15 (1991) study should be mentioned. This study was a case-control study for multiple cancer sites,
16 but only the more common sites, in order to have greater statistical power. Thus, NHL and
17 kidney cancer results were available, but not liver cancer results. Because no liver results were
18 presented for any of the chemicals, this is not a case of reporting bias.
19
20 C.4.1.2. Results of Meta-Analyses
21 Results from some of the meta-analyses that were conducted on the epidemiological
22 studies of TCE and liver cancer are summarized in Table C-12. The pooled estimate from the
23 primary random effects meta-analysis of the 9 studies was 1.33 (95% CI: 1.09, 1.64) (see
24 Figure C-9). As shown in Figure C-9, the analysis was dominated by one large study
25 (contributing about 57% of the weight). That large study was critical in terms of statistical
26 significance of the RRp estimate. Without the large Raaschou-Nielsen study, the RRp estimate
27 does not change noticeably, but it is no longer statistically significant (RRp = 1.31; 95% CI:
28 0.96, 1.79). No other single study was overly influential; removal of any of the other individual
29 studies resulted in RRp estimates that were all statistically significant and that ranged from 1.29
30 (with the removal of Anttila) to 1.39 (with the removal of Boice [1999]).
31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-49 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-12. Summary of some meta-analysis results for TCE and liver cancer
Analysis
All studies
(all cohort
studies)
All studies;
liver cancer
only, when
available
Alternate RR
selections3
Highest
exposure
groups
#of
studies
9
9
9
9
9
9
6
8
7-8
Model
Random
Fixed
Random
Random
Random
Random
Random
Random
Random
Random
Combined
RR estimate
1.33
1.33
1.31
1.33
1.29
1.33
1.30
1.32
1.28
1.24-1.26
95% LCL
1.09
1.09
1.02
1.08
1.06
1.09
1.07
0.93
0.93
0.88-0.91
95% UCL
1.64
1.64
1.67
1.63
1.56
1.63
1.59
1.86
1.77
1.73-1.82
Heterogeneity
None obs
(fixed =
random)
None obs
None obs
None obs
None obs
None obs
None obs
None obs
None obs
Comments
Statistical significance not dependent
on single study, except for
Raaschou-Nielsen, without which
p = 0.08. No apparent publication
bias.
Used RR estimates for liver cancer
alone for the 3 studies that
presented these; remaining RR
estimates are for liver and gall
bladder/biliary passage cancers.
With 1 .0 assumed for Zhao in lieu of
Boice (2006) (see text).
With Boice (1999) any potential
exposure rather than potential
routine exposure.
With estimated female contribution to
Axelson.
With Morgan published SMR.
Primary analysis. Using RR = 1 for
Hansen and Zhao (see text).
Using alternate selections for
Morgan and Raaschou-Nielsen and
excluding Axelson.3
Co
I
~p
TO'
°§
•
TO
o s
H fc
"Changing the primary analysis by one alternate RR each time.
obs = observable.
H
W
-------
TCE and Liver Cancer
Study name Statistics for each study Risk ratio and 95% Cl
Risk Lower Upper
ratio limit limit p-Value
Anttila1995 1.890 0.983 3.632 0.056
Axelson1994 1.410 0.529 3.757 0.492
Boice1999 0.540 0.203 1.439 0.218
Boice2006 1.280 0.480 3.410 0.622
Greenland 1994 0.540 0.110 2.640 0.447
Hansen2001 2.100 0.874 5.045 0.097
Morgan 1998 unpub RR 1.481 0.561 3.909 0.428
Raaschou-Nielsen 2003 1.350 1.030 1.770 0.030
Radican2008 1.120 0.571 2.195 0.741
1.334 1.088 1.636 0.006
*
0.1 0.2 0.5 1
5 10
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
random effects model; same for fixed
Figure C-9. Meta-analysis of liver cancer and TCE exposure. The pooled
estimate is in the bottom row. Symbol sizes reflect relative weights of the studies.
The horizontal midpoint of the bottom diamond represents the pooled RR estimate
and the horizontal extremes depict the 95% CI limits.
As discussed in Section C.4.1.1, all of the 9 studies presented results for liver and gall
bladder/biliary passage cancers combined, and these results were the basis for the primary
analysis discussed above. An alternate analysis was performed substituting, simultaneously,
results for liver cancer alone for the 3 studies for which these were available. The RRp estimate
from this analysis was slightly lower than the one based entirely on results from the combined
cancer categories (1.31; 95% CI: 1.02, 1.67). This result was driven by the fact that the RR
estimate from the large Raaschou-Nielsen et al. (2003) study decreased from 1.35 for liver and
gall bladder/biliary passage cancers combined to 1.28 for liver cancer alone.
Similarly, the RRp estimate was not highly sensitive to other alternate RR estimate
selections. Use of the 4 other alternate selections, individually, resulted in RRp estimates that
were all statistically significant (all with/? < 0.02) and that ranged from 1.29 to 1.33 (see
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-51 DRAFT—DO NOT CITE OR QUOTE
-------
1 Table C-12). In fact, as can be seen in Table C-12, only one of the alternates had notable impact.
2 The Boice (2006), Zhao, and Axelson original values and alternate selections were associated
3 with very little weight and, thus, have little influence in the RRp. Using the Boice (1999)
4 alternate RR estimate based on any potential exposure rather than potential routine exposure
5 decreased the RRp slightly from 1.33 to 1.29. The alternate Boice (1999) RR estimate is actually
6 larger than the original value (0.81 vs. 0.54); however, use of the less discriminating exposure
7 metric captures more liver cancer deaths, causing the weight of that study to increase from about
8 4.3% to almost 15%.
9 There was no apparent heterogeneity across the nine studies, i.e., the random effects
10 model and the fixed effect model gave the same results. Furthermore, all of the liver cancer
11 studies were cohort studies, so no subgroup analyses examining cohort and case-control studies
12 separately, as was done for lymphoma and kidney cancer, were conducted. No alternate
13 quantitative investigations of heterogeneity were pursued because of database limitations and, in
14 any event, there is no evidence for heterogeneity of study results in this database.
15 As discussed in Section C. 1, publication bias was examined in several different ways.
16 The funnel plot in Figure C-10 shows little relationship between RR estimate and study size, and,
17 indeed, none of the other tests performed found any evidence of publication bias. Duval and
18 Tweedie's trim-and-fill procedure, for example, suggested that no studies were missing from the
19 funnel plot, i.e., there was no asymmetry to counterbalance. Similarly, the results of a
20 cumulative meta-analysis, incorporating studies with increasing SE one at a time, shows no
21 evidence of a trend of increasing effect size with addition of the less precise studies. The
22 Raaschou-Nielsen study contributes about 57% of the weight. Including the 2 next most precise
23 studies, the RRp goes from 1.35 to 1.42 to 1.38 and the weight to 76%. With the addition of
24 each of the next 3 most precise studies, the RRp estimate is 1.42. Further addition of the 3 least
25 precise studies gradually brings the RRp back down to 1.33. Thus, if anything, the evidence is
26 somewhat suggestive of an inverse relationship between SE and effect size, contrary to what
27 would be expected if publication bias were occurring.
28 C.4.2. Liver Cancer Effect in the Highest Exposure Groups
29 C.4.2.1. Selection of RR Estimates
30 The selected RR estimates for liver cancer in the highest TCE exposure categories, for
31 studies that provided such estimates, are presented in Table C-13. Six of the 9 cohort studies
32 reported liver cancer risk estimates categorized by exposure level. As in Section C.4.1.1 for the
33 overall risk estimates, estimates for cancers of the liver and gall bladder/biliary passages
34 combined were preferentially selected, when presented, for the sake of consistency, and,
35 wherever possible, RR estimates for males and females combined were used.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-52 DRAFT—DO NOT CITE OR QUOTE
-------
Funnel Plot of Standard Error by Log risk ratio
HI
•5.
ro
•c
c
$
V)
0.0
0.2
0.4
0.6
0.8
1.0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
1
2
3
4
5
Log risk ratio
Figure C-10. Funnel plot of SE by log RR estimate for TCE and liver cancer
studies.
772/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-53 DRAFT—DO NOT CITE OR QUOTE
-------
o
to
§3
VO Co'
Table C-13. Selected RR estimates for liver cancer risk in highest TCE exposure groups
Study
Anttila et
a\., 1995
Axelson et
al., 1994
Boice et al.,
1999
Hansen et
al.,2001
Morgan et
al., 1998
Raaschou-
Nielsen et
al., 2003
Radican et
al., 2008
Zhao et al.,
2005
RR
2.74
3.7
0.94
1.19
1.2
1.49
95%
LCL
0.33
0.09
0.3Q
0.34
0.7
0.67
95%
UCL
9.88
21
2.46
4.16
1.9
3.34
Exposure
category
100+ |jmol/L
U-TCA a
100+ mg/L
U-TCA
> 5 yr exposure
> 1080 mos x
mg/m3
High cumulative
exposure score
> 5 yrs
> 25 unit-yr
High exposure
score
log RR
1.008
1.308
-0.062
0.174
0.182
0.399
SE(log
RR)
0.707
1.000
0.490
0.639
0.243
0.411
Alternate RR
estimates
Exclude study
None
1.0 assumed
0.98 (0.29,
3.35) for
med/high
peak vs.
low/no
1.1 (0.5,2.1)
ICD-7 155.0
(liver only)
None (see
text)
1.0 assumed
Comments
SIR. ICD-7 155.0 (liver only).
SIR. ICD-7 155. 0 cases observed in
highest exposure group (i.e., >2 y and
100+ U-TCA), so combined with <2 y and
100+ subgroup and females, estimating
the expected numbers (see text).
Mortality RR. ICD-9 155 + 156. For
potential routine or intermittent exposure.
Adjusted for date of birth, dates 1 st and
last employed, race, and sex. Referent
group is workers not exposed to any
solvent.
Reported high exposure group results for
some cancer sites but not liver.
Mortality RR. ICD-9 155 + 156. Adjusted
for age and sex.
SIR. ICD-7 155.0 + 155.1. Male and
female results presented separately and
combined assuming a Poisson
distribution.
Mortality HR. ICD-8, -9 155 + 156, ICD-
10C22-C24. Male and female results
presented separately and combined (see
text). Time variable = age, covariate =
race. Referent group is workers with no
chemical exposures.
No liver results reported.
Co
I
~p
TO'
•
TO
o s
H fc
H
W
"Mean personal trichloroacetic acid in urine. 1 umol/L = 0.1634 mg/L.
-------
1 Two of the 9 cohort studies (Hansen et al., 2001; Zhao et al., 2005) did not report liver
2 cancer risk estimates categorized by exposure level even though these same studies reported such
3 estimates for selected other cancer sites. To address this reporting bias (as discussed above,
4 Zhao et al. [2005] did not present any liver results, and it is not clear if this was actual reporting
5 bias or an a priori decision not to examine liver cancer in the cohort.), attempts were made to
6 obtain the results from the primary investigators, and, failing that, alternate analyses were
7 performed in which null estimates (RR =1.0) were included for both studies. This alternate
8 analysis was then used as the main analysis, e.g., the basis of comparison for the sensitivity
9 analyses. For the SE (of the logRR) estimates for the null estimates, SE estimates from other
10 sites for which highest-exposure-group results were available were used. For Hansen et al.
11 (2001), the SE estimate for NHL in the highest exposure group was used, because NHL was the
12 only cancer site of interest in this assessment for which highest-exposure-group results were
13 available. For Zhao et al. (2005), the SE estimate for kidney cancer in the highest-exposure
14 group (incidence with 0 lag) was used. (Note that Boice et al. [2006], who studied a cohort that
15 overlapped that of Zhao et al. [2005], also did not present liver cancer results by exposure level.)
16 For Axelson et al. (1994), there were no liver cancer cases in the highest exposure group
17 (>2 years and 100+ mean urinary-trichloroacetic acid [U-TCA] level), so no log RR and
18 SE(log RR) estimates were available for the meta-analysis. Instead, the <2 years and >2 years
19 results were combined, assuming expected numbers of cases were proportional to person-years,
20 and 100+ U-TCA (with any exposure duration) was used as the highest exposure category. The
21 female contribution to the expected number was also estimated, again assuming proportionality
22 to person-years, and adjusting for the difference between female and male age-adjusted liver
23 cancer incidence rates. The estimated RR and SE values for the combined exposure times and
24 sexes were used in the primary analysis. In an alternate analysis, the Axelson et al. (1994) study
25 was excluded altogether, because we estimated that less than 0.2 cases were expected in the
26 highest-exposure category, suggesting that the study had low power to detect an effect in the
27 highest-exposure group and would contribute little weight to the meta-analysis.
28 For Boice et al. (1999), only results for workers with "any potential exposure" (rather
29 than "potential routine exposure") were presented by exposure category, and the referent group is
30 workers not exposed to any solvent. For Morgan et al. (1998), the primary analysis used results
31 for the cumulative exposure metric, and a sensitivity analysis was done with the results for the
32 peak exposure metric. For Raaschou-Nielsen et al. (2003), unlike for NHL and RCC, liver
33 cancer results for the subcohort with expected higher exposure levels were not presented, so the
34 only highest-exposure group results were for duration of employment in the total cohort. Results
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-55 DRAFT—DO NOT CITE OR QUOTE
-------
1 for cancers of the liver and gall bladder/biliary passages combined were used for the primary
2 analysis and results for liver cancer alone in a sensitivity analysis.
3 For Radican et al. (2008), it should be noted that the referent group is workers with no
4 chemical exposures, not just no TCE exposure. Furthermore, exposure group results were
5 reported separately for males and females and were combined for this assessment using
6 inverse-variance weighting, as in a fixed effect meta-analysis. In addition to results for biliary
7 passage and liver cancer combined, Radican et al. (2008) present results for liver only by
8 exposure group; however, there were no liver cancer deaths in females and the number expected
9 was not reported, so no alternate analysis for the highest-exposure groups with an RR estimate
10 from Radican et al. (2008) for liver cancer only was conducted. Radican et al. (2008) present
11 only mortality FIR estimates by exposure group; however, in an earlier follow-up of this same
12 cohort, Blair et al. (1998) present both incidence and mortality RR estimates by exposure group.
13 As with the Radican et al. (2008) liver cancer only results, however, there were no incident cases
14 for females in the highest-exposure group in Blair et al. (1998) (and the expected number was
15 not reported). Additionally, there were more biliary passage/liver cancer deaths (31) in Radican
16 et al. (2008) than incident cases (13) in Blair et al. (1998) overall and in the highest-exposure
17 group (14 vs. 4). Thus, we elected to use only the Radican et al. (2008) mortality results from
18 this cohort and not to include an alternate analysis based on incidence results from the earlier
19 follow-up.
20
21 C.4.2.2. Results of Meta-Analyses
22 Results from the meta-analyses that were conducted for liver cancer in the highest
23 exposure groups are summarized at the bottom of Table C-12. The pooled RR estimate from the
24 random effects meta-analysis of the 6 studies with results presented for exposure groups was
25 1.32 (95% CI: 0.93, 1.86). As with the overall liver cancer meta-analyses, the meta-analyses of
26 the highest-exposure groups were dominated by one study (Raaschou-Nielsen), which provided
27 about 52% of the weight. The RRp estimate from the primary random effects meta-analysis with
28 null RR estimates (i.e., 1.0) included for Hansen and Zhao to address (potential) reporting bias
29 (see above) was 1.28 (95% CI: 0.93, 1.77) (see Figure C-l 1). The inclusion of these 2 additional
30 studies contributed about 10% of the total weight. No single study was overly influential
31 (removal of individual studies resulted in RRp estimates that ranged from 1.23 to 1.36) and the
32 RRp estimate was not highly sensitive to alternate RR estimate selections (RRp estimates with
33 alternate selections ranged from 1.24 to 1.26; see Table C-12). In addition, there was no
34 observable heterogeneity across the studies for any of the meta-analyses conducted with the
35 highest-exposure groups. However, none of the RRp estimates was statistically significant.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-56 DRAFT—DO NOT CITE OR QUOTE
-------
TCE and Liver Cancer - highest exposure groups
i
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Study name Statistics for each study Risk ratio and 95% Cl
Risk Lower Upper
ratio limit limit p-Value
Anttila1995 2.740 0.685 10.956 0.154
Axelson 1994 est 3.700 0.521 26.267 0.191
Boice1999 0.940 0.360 2.457 0.900
Morgan 1998 1.190 0.340 4.162 0.785
Raaschou-Nielsen 2003 1.200 0.746 1.930 0.452
Radican2008 1.490 0.666 3.332 0.331
Hansen2001 1.000 0.323 3.098 1.000
Zhao 2005 1.000 0.084 11.857 1.000
1.281 0.925 1.774 0.136
0
1 0
2 0
mv
,
*
t
I
5 1
2 5 10
random effects model; same for fixed
Figure C-ll. Meta-analysis of liver cancer and TCE exposure—highest
exposure groups, with assumed null RR estimates for Hansen and Zhao (see
text).
Furthermore, the RRp estimates for the highest-exposure groups were all less than the
significant RRp estimate for an overall effect on liver cancer (1.33; 95% CI: 1.09, 1.64; see
Section C.4.2.2 and Table C-12). This contradictory result is driven by the fact that the RR
estimate for the highest-exposure group was less than the overall RR estimate for Raaschou-
Nielsen, which contributes the majority of the weight to the meta-analyses. The liver cancer
results are relatively underpowered with respect to numbers of studies and number of cases, and
the Raaschou-Nielsen study, which dominates the analysis, uses duration of employment as an
exposure-level surrogate for liver cancer, and duration of employment is a notoriously weak
exposure metric. Thus, the contradictory finding that the RRp estimates for the highest-exposure
groups were all less than the RRp estimate for an overall effect does not rule out an effect of
TCE on liver cancer; however, it certainly does not provide additional support for such an effect.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-57 DRAFT—DO NOT CITE OR QUOTE
-------
1 C.4.3. Discussion of Liver Cancer Meta-Analysis Results
2 For the most part, the meta-analyses of the overall effect of TCE exposure on liver (and
3 gall bladder/biliary passages) cancer suggest a small, statistically significant increase in risk.
4 The pooled estimate from the primary random effects meta-analysis of the 9 (all cohort) studies
5 was 1.33 (95% CI: 1.09, 1.64). The analysis was dominated by one large study that contributed
6 about 57% of the weight. When this study was removed, the RRp estimate did not change much,
7 but it was no longer statistically significant (RRp = 1.31; 95% CI: 0.96, 1.79). The pooled
8 estimate was not overly influenced by any other single study, nor was it overly sensitive to
9 individual RR estimate selections. The largest downward impacts were from the removal of the
10 Anttila study, resulting in an RRp estimate of 1.29 (95% CI: 1.04, 1.59), and from the
11 substitution of the Boice (1999) RR estimate for potential routine exposure with that for any
12 potential exposure, resulting in an RRp estimate of 1.29 (1.06, 1.56). Substituting the RR
13 estimates for liver/gall bladder/biliary passage cancers with those of liver cancer alone for the
14 3 studies that provided these results yielded an RRp estimate of 1.31 (1.02, 1.67). There was no
15 evidence of publication bias in this data set, and there was no observable heterogeneity across the
16 study results.
17 Six of the 9 studies provided liver cancer results by exposure level. Two other studies
18 reported results for other cancer sites by exposure level, but not liver cancer; thus, to address this
19 reporting bias, null values (i.e., RR estimates of 1.0) were used for these studies. Different
20 exposure metrics were used in the various studies, and the purpose of combining results across
21 the different highest exposure groups was not to estimate an RRp associated with some level of
22 exposure, but rather to see the impacts of combining RR estimates that should be less affected by
23 exposure misclassification. In other words, the highest exposure category is more likely to
24 represent a greater differential TCE exposure compared to people in the referent group than the
25 exposure differential for the overall (typically any vs. none) exposure comparison. Thus, if TCE
26 exposure increases the risk of liver cancer, the effects should be more apparent in the highest
27 exposure groups. However, the RRp estimate from the meta-analyses of the highest exposure
28 group results were less than the RRp estimate from the overall exposure analysis. This
29 anomalous result is driven by the fact that, for Raaschou-Nielsen, which contributes the majority
30 of the weight to the meta-analyses, the RR estimate for the highest-exposure group, although
31 greater than 1.0, was less than the overall RR estimate.
32 Thus, while there is the suggestion of an increased risk for liver cancer associated with
33 TCE exposure, the statistical significance of the pooled estimates is dependent on one study,
34 which provides the majority of the weight in the meta-analyses. Removal of this study does not
35 change the RRp estimate; however, it becomes nonsignificant (p = 0.08). Furthermore, meta-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-58 DRAFT—DO NOT CITE OR QUOTE
-------
1 analysis results for the highest-exposure groups yielded lower RRp estimates than for an overall
2 effect. These results do not rule out an effect of TCE on liver cancer, because the liver cancer
3 results are relatively underpowered with respect to numbers of studies and number of cases and
4 the overwhelming study in terms of weight uses the weak exposure surrogate of duration of
5 employment for categorizing exposure level; however, at present, there is only modest support
6 for such an effect.
7
8 C.5. DISCUSSION OF STRENGTHS, LIMITATIONS, AND UNCERTAINTIES IN
9 THE META-ANALYSES
10 Meta-analysis provides a systematic way of objectively and quantitatively combining the
11 results of multiple studies to obtain a summary effect estimate. Use of meta-analysis can help
12 risk assessors avoid some of the potential pitfalls in overly relying on a single study or in making
13 more subjective qualitative judgments about the apparent weight of evidence across studies.
14 Combining the results of smaller studies also increases the statistical power to observe an effect,
15 if one exists. In addition, meta-analysis techniques assist in systematically investigating issues
16 such as potential publication bias and heterogeneity in a database.
17 While meta-analysis can be a useful tool for analyzing a database of epidemiological
18 studies, the analysis is limited by the quality of the input data. If the individual studies are
19 deficient in their abilities to observe an effect (in ways other than low statistical power, which
20 meta-analysis can help ameliorate), the meta-analysis will be similarly deficient. A critical step
21 in the conduct of a meta-analysis is to establish eligibility criteria and clearly and transparently
22 identify all relevant studies for inclusion in the meta-analysis. For the TCE database, a
23 comprehensive qualitative review of available studies was conducted and eligible studies were
24 identified, as described in Appendix B, Section II-9.
25 Identifying all relevant studies may be hampered if publication bias has occurred.
26 Publication bias is a systematic error that can arise if statistically significant studies are more
27 likely to be published than nonsignificant studies. This can result in an upward bias on the effect
28 size measure, i.e., the relative risk estimate. To address this concern, potential publication bias
29 was investigated for the databases for which meta-analyses were undertaken. For the studies of
30 kidney cancer and liver cancer, there was no evidence of publication bias. For the studies of
31 lymphoma, there was some evidence of potential publication bias. It is uncertain whether this
32 reflects actual publication bias or rather an association between SE and effect size (as discussed
33 in Section C. 1, a feature of publication bias is that smaller studies tend to have larger effect
34 sizes) resulting for some other reason, e.g., a difference in study populations or protocols in the
35 smaller studies. Furthermore, if there is publication bias in this data set, it may be creating an
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-59 DRAFT—DO NOT CITE OR QUOTE
-------
1 upward bias on the relative risk estimate, but this bias does not appear to account completely for
2 the finding of an increased lymphoma risk (see Section C.2.1.2).
3 Another concern in meta-analyses is heterogeneity across studies. Random-effects
4 models were used for the primary meta-analyses in this assessment because of the diverse nature
5 of the individual studies. When there is no heterogeneity across the study results, the
6 random-effects model will give the same result as a fixed-effect model. When there is
7 heterogeneity, the random-effects model estimates the between-study variance. Thus, when
8 there is heterogeneity, the random-effects model will generate wider confidence intervals and be
9 more "conservative" than a fixed-effect model. However, if there is substantial heterogeneity, it
10 may be inappropriate to combine the studies at all. In cases of significant heterogeneity, it is
11 important to try to investigate the potential sources of the heterogeneity.
12 For the studies of kidney cancer and liver cancer, there was no apparent heterogeneity
13 across the study results, i.e., random- and fixed-effects models gave identical summary
14 estimates. For the lymphoma studies, there was heterogeneity, but it was not statistically
15 significant (p = 0.10). The/2 value was 33%, suggesting low-to-moderate heterogeneity. When
16 subgroup analyses were done for the cohort and case-control studies separately, there was some
17 heterogeneity in both groups, but in neither case was it statistically significant. Further attempts
18 to quantitatively investigate the heterogeneity were not pursued because of limitations in the
19 database. The sources of heterogeneity are an uncertainty in the database of studies of TCE and
20 lymphoma. Some potential sources of heterogeneity, which are discussed qualitatively in
21 Section C.2.3, include differences in exposure assessment or in the intensity or prevalence of
22 TCE exposures in the study population and differences in lymphoma classification.
23 The joint occurrence of heterogeneity and potential publication bias in the database of
24 studies of TCE and lymphoma raises special concerns. Because of the heterogeneity, a
25 random-effects model should be used if these studies are to be combined; yet, the random-effects
26 model gives relatively large weight to small studies, which could exacerbate the potential
27 impacts of publication bias. For the lymphoma studies, the summary relative risk estimates from
28 the random-effects and fixed-effect models are not very different (RRp = 1.23 [95% CI: 1.04,
29 1.44] and 1.19 [1.06, 1.34], respectively); however, the confidence interval for the fixed-effect
30 estimate does not reflect the between-study variance and is, thus, overly narrow.
31
32 C.6. CONCLUSIONS
33 The strongest finding from the meta-analyses was for TCE and kidney cancer. The
34 summary estimate from the primary random-effects meta-analysis of the 14 studies was
35 RRp = 1.25 (95% CI: 1.11, 1.41). There was no apparent heterogeneity across the study results
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-60 DRAFT—DO NOT CITE OR QUOTE
-------
1 (i.e., fixed-effect model gave same summary estimate), and there was no evidence of potential
2 publication bias. The summary estimate was robust across influence and sensitivity analyses; the
3 estimate was not markedly influenced by any single study, not was it overly sensitive to
4 individual RR estimate selections. The findings from the meta-analyses of the highest exposure
5 groups for the studies that provided results categorized by exposure level were similarly robust.
6 The summary estimate was RRp = 1.53 (95% CI: 1.23, 1.91) for the 12 studies included in the
7 analysis. There was no apparent heterogeneity in the highest-exposure group results, and the
8 estimate was not markedly influenced by any single study, nor was it overly sensitive to
9 individual RR estimate selections. In sum, these robust results support a conclusion that TCE
10 exposure increases the risk of kidney cancer.
11 For the most part, the meta-analyses of the overall effect of TCE exposure on lymphoma
12 also suggest a small, statistically significant increase in risk. The summary estimate from the
13 primary random-effects meta-analysis of the 16 studies was 1.23 (95% CI: 1.04, 1.44). This
14 result was not overly influenced by any single study, nor was it overly sensitive to individual RR
15 estimate selections, although use of one alternate RR estimate considered clearly inferior
16 narrowly eliminated statistical significance of the summary estimate (p = 0.050). There is some
17 evidence of potential publication bias in the lymphoma study data set; however, it is uncertain
18 that this is actually publication bias rather than an association between SE and effect size
19 resulting for some other reason, e.g., a difference in study populations or protocols in the smaller
20 studies. Furthermore, if there is publication bias, it does not appear to account completely for the
21 findings of an increased lymphoma risk. There was some heterogeneity across the results of the
22 16 studies, but it was not statistically significant (p = 0.10). The / value was 33%, suggesting
23 low-to-moderate heterogeneity. The source(s) of this heterogeneity remains an uncertainty. The
24 summary estimate from the meta-analysis of the highest exposure groups for the 12 studies
25 which provided results categorized by exposure level was RRp = 1.57 (95% CI: 1.27, 1.94).
26 This result for the highest exposure groups was not overly influenced by any single study, nor
27 was it overly sensitive to individual RR estimate selections, and heterogeneity was not observed
28 in any of the relevant analyses. The robustness of the finding of an increased lymphoma risk for
29 the highest exposure groups strengthens the more moderate evidence from the meta-analyses for
30 overall effect.
31 The meta-analyses of the overall effect of TCE exposure on liver (and gall bladder/biliary
32 passages) cancer also suggest a small, statistically significant increase in risk, but the study
33 database is more limited. The pooled estimate from the primary random-effects meta-analysis of
34 the 9 (all cohort) studies was 1.33 (95% CI: 1.09, 1.64). The analysis was dominated by one
35 large study that contributed about 57% of the weight. When this study was removed, the RRp
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-61 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimate did not change much, but it was less precise (RRp = 1.31; 95% CI: 0.96, 1.79). The
2 pooled estimate was not overly influenced by any other single study, nor was it overly sensitive
3 to individual RR estimate selections. There was no evidence of publication bias in this data set,
4 and there was no observable heterogeneity across the study results. However, the findings from
5 the meta-analyses of the highest-exposure groups for the studies that provided results categorized
6 by exposure level do not add support to the overall effect findings. The summary estimate was
7 RRp = 1.28 (95% CI: 0.93, 1.77) for the 8 studies included in the analysis, which is lower than
8 the summary estimate for the overall effect. This contradictory result is driven by the fact that
9 the RR estimate for the highest-exposure group in the individual study which contributes the
10 majority of the weight to the meta-analyses, although greater than 1.0, was less than the overall
11 RR estimate for the same study. In sum, these results do not rule out an effect of TCE on liver
12 cancer, because the liver cancer results are relatively underpowered with respect to numbers of
13 studies and number of cases and the overwhelming study in terms of weight uses the weak
14 exposure surrogate of duration of employment for categorizing exposure level; however, at
15 present, there is only modest support for an increased risk of liver cancer.
16
17 C.7. REFERENCES
18 Anttila, A; Pukkala, E; Sallmen, M; et al. (1995) Cancer incidence among Finnish workers exposed to halogenated
19 hydrocarbons. J Occup Environ Med 37:797-806.
20 Axelson, O; Selden, A; Andersson, K; et al. (1994) Updated and expanded 1 Swedish cohort study on
21 trichloroethylene and cancer risk. J Occup Med 36:556-562.
22 Banks, PM. (1992) Changes in diagnosis of non-Hodgkin's lymphomas over time. Cancer Res 52:5453s-5455s.
23 Begg, CB; Mazumdar, M. (1994) Operating characteristics of a rank correlation test for publication bias. Biometrics
24 50:1088-1101.
25 Blair, A; Hartge, P; Stewart, PA; et al. (1998) Mortality and cancer incidence of aircraft maintenance workers
26 exposed to trichloroethylene and other organic solvents and chemicals: extended follow up. Occup Environ Med
27 55:161-171.
28 Boice, JD; Marano, DE; Fryzek, JP; et al. (1999) Mortality among aircraft manufacturing workers. Occup Environ
29 Med 56:581-597.
30 Boice, JD; Marano, DE; Cohen, SS; et al. (2006) Mortality among Rocketdyne worker who tested rocket engines,
31 1948-1999. J Occup Environ Med 48:1070-1092.
32 Breslow, NE; Day, NE. (1980) Statistical methods in cancer research. Vol. I. The Analysis of Case-Control Studies.
33 IARC Scientific Publications No. 32. Lyon, France: International Agency for Research on Cancer.
34 Breslow, NE; Day, NE. (1987) Statistical methods in cancer research, Vol. II. The Design and Analysis of Cohort
3 5 Studies. IARC Scientific Publications No. 82. Lyon, France: International Agency for Research on Cancer.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 Briining, T; Pesch, B; Wiesenhtitter, B; et al. (2003) Renal cell cancer risk and occupational exposure to
2 trichloroethylene: results of a consecutive case-control study in Arnsberg, Germany. Am J Ind Med 43:274-285.
3 Charbotel, B; Fevotte, J; Hours, M; et al. (2006) Case-control study on renal cell cancer and occupational exposure
4 to trichloroethylene. Part II: Epidemiological aspects. Ann Occup Hyg 50(8):777-787.
5 Cherrie, JW; Kromhout; H; Semple, S. (2001). The importance of reliable exposure estimates in deciding whether
6 trichloroethylene can cause kidney cancer [letter]. J Cancer Res Clin 127:400-402.
7 Copeland, KT; Checkoway, H; McMichael, AJ; et al. (1977) Bias due to misclassification in the estimation of
8 relative risk. Am JEpidemiol 105:488-495.
9 DerSimonian, R; Laird, N. (1986) Meta-analysis in clinical trials. Controlled Clinical Trials 7:177-188.
10 Dosemeci, M; Cocco, P; Chow, WH. (1999) Gender differences in risk of renal cell carcinoma and occupational
11 exposures to chlorinated aliphatic hydrocarbons. Am JIndMed36(l):54-59.
12 Duval, S; Tweedie, R. (2000) A nonparametric "trim and fill" method of accounting for publication bias in meta-
13 analysis. J Am StatAssoc 95:89-98.
14 Egger, M; Smith, GD; Schneider, M; et al. (1997) Bias in meta-analysis detected by a simple, graphical test. Brit
15 Med 1315:629-634.
16 Environmental Health Strategies. (1997) Final Report. Written correspondence from Paul A. Cammer, Ph.D.,
17 Trichloroethylene Issues Group, to Cheryl Siegel Scott, US Environmental Protection Agency dated December 22,
18 1997.
19 Fevotte, J; Charbotel, B; Muller-Beaute, P; et al. (2006) Case-control study on renal cell cancer and occupational
20 exposure to trichloroethylene. Part I: Exposure assessment. Ann Occup Hyg 50:765-775.
21 Greenland, S; Salvan, A; Wegman, DH; et al. (1994) A case-control study of cancer mortality at the transformer-
22 assembly facility. Int Arch Occ Env Heal 66:49-54.
23 Hansen, J; Raaschou-Nielsen, O; Christensen, JM; et al. (2001) Cancer incidence among Danish workers exposed to
24 trichloroethylene. J Occup Environ Med 43:133-139.
25 Hardell, L; Eriksson, M; Degerman, A. (1994) Exposure to phenoxyacetic acids, chlorophenols, or organic solvents
26 in relation to histopathology, stage, and anatomical localization of non-Hodgkin's lymphoma. Cancer Res
27 54:2386-2389.
28 Harris, NL; Jaffe, ES; Diebold, J; et al. (2000) Lymphoma classification - from controversy to consensus: the
29 R.E.A.L. and WHO classification of lymphoid neoplasms. AnnOncol 11 Suppl 1:3-10.
30 Higgins JPT, Thompson SG, Deeks JJ, Altman DG. 2003. Measuring inconsistency in meta-analyses. Brit Med J
31 327:557-560.
32 Mandel, JH; Kelsh, MA; Mink, PJ; et al. (2006) Occupational trichloroethylene exposure and non-Hodgkin's
33 lymphoma: a meta-analysis and review. Occup EnvironMed 63:597-607.
34 Mannetje, A.; Steenland, K; Attfield, M; et al. (2002) Exposure-response analysis and risk assessment for silica and
3 5 silicosis mortality in a pooled analysis of six cohorts. Occup. Environ. Med. 59:723-728.
36 McGuire, V; Nelson, LM; Koepsell, TD; et al. (1998) Assessment of occupational exposures in community-based
37 case-control studies. Annu. Rev. Publ Health 19:35-53.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-63 DRAFT—DO NOT CITE OR QUOTE
-------
1 Miligi, L; Costantini, AS; Benvenuti, A; et al. (2006) Occupational exposure to solvents and the risk of lymphomas.
2 Epidemiology 17:552-561.
3 Morgan, RW; Kelsh, MA; Zhao, K; et al. (1998) Mortality of aerospace workers exposure to trichloroethylene.
4 Epidemiology 9:424-431
5 Morton LM, Holford TR, Leaderer B, Zhang Y, Zahm SH, et al. 2003. Alcohol use and risk of non-Hodgkin's
6 lymphoma among Connecticut somen (United States). Cancer Causes Control 14:687-694.
7 Nelson, LM; Longstreth, WT, Jr,; Koepsell, TD; et al. (1994) Completeness and accuracy of interview data from
8 proxy respondents: Demographic, medical, and life-style factors. Epidemiology 5:204-217.
9 Nordstrom, M; Hardell, L; Hagberg, H; et al. (1998) Occupational exposures, animal exposure and smoking as risk
10 factors for hairy cell leukaemia evaluated in a case-control study. Brit J Cancer 77:2048-2052.
11 NRC (National Research Council). (2006) Assessing the Human Health Risks of Trichloroethylene. Key Scientific
12 Issues. Washington, DC: National Academy Press.
13 PerssonB, FredriksonM. 1999. Some risk factors for non-Hodgkin's lymphoma. Int J Occup Med Environ Health
14 12:135-142.
15 Pesch, B; Haerting, J; Ranft, U; et al. (2000) Occupational risk factors for renal cell carcinoma: Agent-specific
16 results from a case-control study in Germany. Int J Epidemiol 29:1014-1024.
17 Raaschou-Nielsen, O; Hansen, J; McLaughlin, JK; et al. (2003) Cancer risk among workers at Danish companies
18 using trichloroethylene: A cohort study. Am J Epidemiol 158:1182-1192.
19 Radican L, Blair A, Stewart P, Wartenberg D. 2008. Mortality of aircraft maintenance workers exposed to
20 trichloroethylene and other hydrocarbons and chemicals: extended follow-up. J Occup Environ Med 50:1306-1319.
21 Rothman, KJ; Greenland, S. (1998) Modern Epidemiology, 2nd Edition. Philadelphia, PA: Lippincott Williams &
22 Wilkins.
23 Seidler, A; Mohner, M; Berger, J; et al. (2007) Solvent exposure and malignant lymphoma: A population-based
24 case-control study in Germany. J Occup Med Toxicol 2:2.
25 Siemiatycki, J. (1991) Risk Factors for Cancer in the Workplace. J Siemiatycki, ed. Baca Raton, FL: CRC Press.
26 Wang R, Zhang Y, Lan Q, Holford TR, Beaderer B, Zahm SH, Boyle P, Dosemeci M, Rothman N, Zhu Y, Qin Q,
27 Zheng T. 2009. Occupational exposure to solvents and risk of non-Hodgkin lymphoma in Connecticut women.
28 Am J Epidemiol 169:176-185.
29 Woolf, B. (1955) On estimating the relationship between blood group and disease. Ann Hum Genet 19:251-253.
30 Zhao, Y; Krishnadasan, A; Kennedy, N; et al. (2005) Estimated effects of solvents and mineral oils on cancer
31 incidence and mortality in a cohort of aerospace workers. Am J Ind Med 48:249-258.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 C-64 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX D
Neurological Effects of Trichloroethylene
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix D: Neurological Effects of Trichloroethylene
LIST OF TABLES D-iii
APPENDIX D. NEUROLOGICAL EFFECTS OF TRICHLOROETHYLENE D-l
D. 1. HUMAN STUDIES ON THE NEUROLOGICAL EFFECTS OF
TRICHLOROETHYLENE (TCE) D-l
D.I.I. Changes in Nerve Conduction D-l
D.I. 1.1. Blink Reflex and Masseter Reflex Studies—Trigeminal Nerve D-l
D. 1.1.2. Trigeminal Somatosensory Evoked Potential (TSEP)
Studies—Trigeminal Nerve D-6
D.I. 1.3. Nerve Conduction Velocity Studies D-9
D.1.2. Auditory Effects D-10
D.1.3. VestibularEffects D-13
D.1.4. Visual Effects D-16
D.I.5. Cognition D-18
D.I.6. PsychomotorEffects D-22
D.I.6.1. Reaction Time D-22
D.I.6.2. MuscularDyscoordination D-26
D.1.7. Summary Tables D-28
D.2. CENTRAL NERVOUS TOXICITY IN ANIMAL STUDIES FOLLOWING
TRICHLOROETHYLENE (TCE) EXPOSURE D-90
D.2.1. Alterations in Nerve Conduction D-90
D.2.2. Auditory Effects D-92
D.2.2.1. Inhalation D-92
D.2.2.2. Oral and Injection Studies D-95
D.2.3. Vestibular System Studies D-95
D.2.4. Visual Effects D-96
D.2.5. Cognitive Function D-98
D.2.6. Psychomotor Effects D-99
D.2.6.1. Loss of Righting Reflex D-99
D.2.6.2. Functional Observational Battery (FOB) and Locomotor
Activity Studies D-100
D.2.6.3. Locomotor Activity D-102
D.2.7. Sleep and Mood Disorders D-104
D.2.7.1. Effects on Mood: Laboratory Animal Findings D-104
D.2.7.2. Sleep Disturbances D-104
D.2.8. Mechanistic Studies D-104
D.2.8.1. Dopaminergic (DA) Neurons D-104
D.2.8.2. Gamma-Amino Butyric Acid (GABA) and Glutamatergic
Neurons D-l 05
D.2.8.3. Demyelination Following Trichloroethylene (TCE)
Exposure D-l 07
D.2.9. Summary Tables D-109
D.3. REFERENCES D-122
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-ii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
D-l. Epidemiological studies: Neurological effects of trichloroethylene D-30
D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed
solvents D-68
D-3. Literature review of studies of TCE and domains assessed with
neurobehavioral/neurological methods D-88
D-4. Summary of mammalian in vivo trigeminal nerve studies D-l 10
D-5. Summary of mammalian in vivo otoxicity studies D-l 11
D-6. Summary of mammalian sensory studies—vestibular and visual systems D-l 13
D-7. Summary of mammalian cognition studies D-l 14
D-8. Summary of mammalian psychomotor function, locomotor activity, and
reaction time studies D-l 15
D-9. Summary of mammalian in vivo dopamine neuronal studies D-l 17
D-10. Summary of neurochemical effects with TCE exposure D-l 18
D-ll. Summary of in vitro ion channel effects with TCE exposure D-120
D-12. Summary of mammalian in vivo developmental neurotoxicity studies—oral
exposures D-121
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-iii DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX D. NEUROLOGICAL EFFECTS OF TRICHLOROETHYLENE
D.I. HUMAN STUDIES ON THE NEUROLOGICAL EFFECTS OF
TRICHLOROETHYLENE (TCE)
There is an extensive body of evidence in the literature on the neurological effects caused
by exposure to trichloroethylene (TCE) in humans. The primary functional domains that have
been studied and reported are trigeminal nerve function and nerve conductivity (latency),
psychomotor effects, including reaction times (simple and choice), visual and auditory effects,
cognition, memory, and subjective neurological symptoms, such as headache and dizziness. This
section discusses the primary studies presented for each of these effects. Summary tables for all
the human TCE studies are at the end of this section.
D.I.I. Changes in Nerve Conduction
There is strong evidence in the literature that exposure to TCE results in impairment of
trigeminal nerve function in humans exposed occupationally, by inhalation, or environmentally,
by ingestion. Functional measures such as the blink reflex and masseter reflex tests were used to
determine if physiological functions mediated by the trigeminal nerve were significantly
impacted. Additionally, trigeminal somatosensory evoked potentials were also measured in
some studies to ascertain if nerve activity was directly affected by TCE exposure.
D.I.1.1. Blink Reflex and Masseter Reflex Studies—Trigeminal Nerve
Barret et al. (1984) conducted a study on 188 workers exposed to TCE occupationally
from small and large factories in France (type of factories not disclosed). The average age of the
workers was 41 (standard deviation [SD] not provided, but authors noted 14% <30 years and
25% >50 years) and the average exposure duration was 7 hours/day for 7 years. The
188 workers were divided into high and low exposure groups for both TCE exposure measured
using detector tubes and trichloroacetic acid (TCA) levels measured in urine. There was no
unexposed control population, but responses in the high-exposure group were compared response
in the low-exposure group. TCE exposure groups were divided into a low exposure group
(<150 ppm; n = 134) and a high exposure group (>150 ppm; n = 54). The same workers
(n = 188) were also grouped by TCA urine measurements such that a high exposure was
>100 mg TCA/g creatinine. Personal factors including age, tobacco use and alcohol intake were
also analyzed. No mention was made regarding whether or not the examiners were blind to the
subjects' exposure status. Complete physical examination including testing visual performance
(acuity and color perception), evoked trigeminal potential latencies and audiometry, facial
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l DRAFT—DO NOT CITE OR QUOTE
-------
sensitivity, reflexes, and motoricity of the masseter muscles. Chi squared analysis was used to
examine distribution of the different groups for comparing high and low exposed workers
followed by one way analysis of variance. Overall, 22 out of 188 workers (11.7%) experienced
trigeminal nerve impairment (p < 0.01) as measured by facial sensitivity, reflexes (e.g., jaw,
corneal, blink) and movement of the masseter muscles. When grouped by TCE exposure, 12 out
of 54 workers (22.2%) in the high exposure group (>150 ppm) and 10 out of 134 workers (7.4%)
in the low exposure group had impaired trigeminal nerve mediated responses. When grouped by
the presence of TCA in the urine, 41 workers were now in the high TCA group and 10 out of 41
workers (24.4%) experienced trigeminal nerve impairment in comparison to the 12 out of 147
(8.2%) in the low TCA (<100 mg TCA/g creatinine) group. Statistically significant results were
also presented for the following symptoms based on TCE and TCA levels: trigeminal nerve
impairment (p < 0.01), asthenia (p < 0.01), optic nerve impairment (p < 0.001), and dizziness
(0.05
-------
supraorbital nerve (above the eyelid) with a shock (0.05 ms in duration) resulting in a response
and the response was measured using a recording electrode over the orbicularis oculi muscle (the
muscle responsible for closing the eyelid and innervated by the trigeminal nerve). The BR
generated an Rl and an R2 component from each individual. BRs were recorded and the
supraorbital nerve was stimulated with single electrical shocks of increasing intensity until nearly
stable Rl and R2 ipsilateral and R2 contralateral responses were obtained. The student's t-test
was used for testing the difference between the group means for the blink reflex component
latencies. Because of the variability of R2 responses, this study focused primarily on the Rl
response latencies. Highly significant differences in the conduction latency means of the BR
components for the TCE exposed population versus control population were observed when
comparing means for the right and left side Rl to the controls. The mean Rl BR component
latency for the exposed group was 11.35 ms, SD = 0.74 ms, 95% confidence interval (CI):
11.03-11.66. The mean for the controls was 10.21 ms, SD = 0.78 ms, 95% CI: 9.92-10.51;
(p < 0.001). The study was well conducted with consistency of methods, and statistically
significant findings for trigeminal nerve function impairment resulting from environmental
exposures to TCE. However, the presence of other solvents in the well water, self selection of
subjects involved in litigation, and incomplete characterization of exposure present problems in
drawing a clear conclusion of TCE causality or dose-response relationship.
Kilburn and Warshaw (1993) conducted an environmental study on 544 Arizona
residents exposed to TCE in well-water. TCE concentrations were from 6 to 500 ppb and
exposure ranged from 1 to 25 years. Subjects were recruited and categorized in 3 groups.
Exposed group 1 consisted of 196 family members with cancer or birth defects. Exposed group
2 consisted of 178 individuals from families without cancer or birth defects; and exposed group 3
included 170 parents whose children had birth defects and rheumatic disorders. Well-water was
measured from 1957 to 1981 by several governmental agencies and average annual TCE
exposures were calculated and then multiplied by each individual's years of residence for
170 subjects. A referent group of histology technicians (n= 113) was used as a comparison for
the BR test. For this test, recording electrodes were placed over the orbicularis oculi muscles
(upper and lower) and the BR was elicited by gently tapping the glabeela (located on the mid-
frontal bone at the space between the eyebrows and above the nose). A two-sided Student's
t-test and linear regression were used for statistical analysis. Significant increases in the Rl
component of the BR response was observed in the exposed population as compared to the
referent group. The Rl component measured from the right eye appeared within 10.9 ms in
TCE-exposed subjects whereas in referents, this component appeared 10.2 ms after the stimulus
was elicited indicating a significant delay (p < 0.008) in the reflex response. Similarly, delays in
the latency of appearance for the Rl component were also noted for the left eye but the effect
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-3 DRAFT—DO NOT CITE OR QUOTE
-------
was not statistically significant (p = 0.0754). This study shows statistically significant
differences in trigeminal nerve function between subjects environmentally exposed and
nonexposed to TCE. This is an ecological study with TCE exposure inferred to subjects by
residence in a geographic area. Estimates of TCE concentrations in drinking water to individual
subjects are lacking. Additionally, litigation is suggested and may introduce a bias, particularly
if no validity tests were used.
Kilburn (2002a) studied 236 residents (age range: 18-83 years old) lived nearby
manufacturing plants (e.g., microchip plants) in Phoenix, AZ. Analysis of the groundwater in
the residential area revealed contamination with many volatile organic compounds including
TCE. Concentrations of TCE in the well water ranged from 0.2 ppb to more than 10,000 ppb
and the exposure duration varied between 2 to 37 years. Additional associated solvents included
dichloroethane (DCE), perchloroethylene, and vinyl chloride. A group-match design was used to
compare the 236 TCE-exposed residents to 161 unexposed regional referents and 67 referents in
NE Phoenix in the BR test. The BR response was recorded from surface electrodes placed over
the location of the orbicularis oculi muscles. The reflex response was elicited by gently tapping
the left and right supraorbital notches with a small hammer. The Rl component of the BR
response was measured for both the left and right eye. Statistically significant increases in
latency time for the Rl component was observed for residents exposed to TCE in comparison to
the control groups. In unexposed individuals, the Rl component occurred within 13.4 ms from
the right eye and 13.5 ms from the left eye. In comparison, the residents near the manufacturing
plant had latency times of 14.2 ms (p < 0.0001) for the right eye and 13.9 ms (p < 0.008) for the
left eye. This study shows statistically significant differences between environmentally exposed
and unexposed populations for trigeminal nerve function, as a result of exposures to TCE. This
is an ecological study with TCE exposure potential to subjects inferred by residence in a
geographic area. Estimates of TCE concentrations in drinking water to individuals are lacking.
Additionally, litigation is suggested and may introduce a bias, particularly if no validity tests
were used.
Feldman et al. (1992) evaluated the BR reflex in 18 subjects occupationally exposed to
neurotoxic chemicals (e.g., degreasers, mechanics, and pesticide sprayers among many others).
Eight of the subjects were either extensively (n = 4) or occupationally (n = 4) exposed to TCE.
The remaining subjects (n = 10) were exposed to other neurotoxic chemicals, but not TCE.
Quantitative exposure concentration data were not reported in the study, but TCE exposure was
characterized as either "extensive" or "occupational." Subjects in the "extensive" exposure
group were chronically exposed (>1 year) to TCE at least 5 days a week and for at greater than
50% of the workday (n = 3) or experienced a direct, acute exposure to TCE for greater than
15 minutes (n = 1). Subjects in the "occupational" group were chronically exposed (>1 year) to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-4 DRAFT—DO NOT CITE OR QUOTE
-------
TCE for 1-3 days/week and for greater than 50% of the workday. The BR responses from the
TCE-exposed subjects were compared to a control group consisting of 30 nonexposed subjects
with no noted neurological disorders. BR responses were measured using surface electrodes
over the lower lateral portion of the orbicularis oculi muscle. Electrical shocks with durations of
0.05 ms were applied to the supraorbital nerve to generate the Rl and R2 responses. All of the
subjects that were extensively exposed to TCE had significantly increased latency times in the
appearance of the Rl component (no/?-value listed) and for 3 subjects this increased latency time
persisted for at least 1 month and up to 20 years postexposure. However, none of the subjects
occupationally exposed to TCE had changes in the BR response in comparison to the control
group. In comparing the remaining neurotoxicant exposed subjects to the TCE-exposed
individuals, the sensitivity, or the ability of a positive blink reflex test to identify correctly those
who had TCE exposure was 50%. However, in workers with no exposure to TCE, 90%
demonstrated a normal Rl latency.
Mixed results were obtained in a study by Ruitjen et al. (1991) on 31 male printing
workers exposed to TCE. The mean age was 44; mean exposure duration was 16 years and had
at least 6 years of TCE exposure. The control group consisted of 28 workers with a mean age
45 years. Workers in the control group were employed at least 6 years in print factories (similar
to TCE-exposed), had no exposure to TCE, but were exposed to "turpentine-like organic
solvents." TCE exposure potential was inferred from historical monitoring of TCE at the plant
using gas detection tubes. These data indicated TCE concentrations in the 1960s of around
80 ppm, mean concentration of 70 ppm in the next decade, with measurements from 1976 and
1981 showing a mean concentration of 35 ppm. The most recent estimate of TCE concentrations
in the factory was 17 ppm (stable for 3 years) at the time of the report. The authors calculated
that mean cumulative TCE exposure would be 704 ppm x years worked in factory. The masseter
and blink reflexes were measured to evaluate trigeminal nerve function in TCE-exposed and
control workers. For measurement of the masseter reflex, surface electrodes were attached over
the right masseter muscle (over the cheek area). A gentle tap on a roller placed under the
subject's chin was used to elicit the masseter reflex. For measurement of the blink reflex,
surface electrodes were placed on the muscle near the upper eyelid. Electrical stimulation of the
right supraorbital nerve was used to generate the blink reflex. There was a significant increase in
the latency of the masseter reflex to appear for the TCE-exposed workers (p < 0.05). However,
there was no significant change in the blink reflex measure between TCE-exposed workers and
control. Although no change in the blink reflex measures were observed between the two
groups, it should be noted that the control group was exposed to other volatile organic solvents
(not specified) and this volatile organic compound exposure could be a possible confounder for
determination of TCE-induced effects.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-5 DRAFT—DO NOT CITE OR QUOTE
-------
There are two studies that reported no effect of TCE exposure on trigeminal nerve
function (El-Ghawabi et al., 1973; Rasmussen et al., 1993c). El-Ghawabi et al. (1973) conducted
a study on 30 money printing shop workers occupationally exposed to TCE. Metabolites of total
trichloroacetic acid and trichloroethanol were found to be proportional to TCE concentrations up
to 100 ppm (550 mg/m3). Controls were 20 age- and socio-economic status (SES)-matched
nonexposed males and 10 control workers not exposed to TCE. Trigeminal nerve involvement
was not detected, but the authors failed to provide details as to how this assessment was made. It
is mentioned that each subject was clinically evaluated and trigeminal nerve involvement may
have been assessed through a clinical evaluation. As a result, the conclusions of this study are
tempered since the authors did not provide details as to how trigeminal nerve function was
evaluated in this study.
Rasmussen et al. (1993c) conducted an historical cohort study on 99 metal degreasers.
Subjects were selected from a population of 240 workers from 72 factories in Denmark. The
participants were divided into three groups based on solvent exposure durations where low
exposure was up to 0.5 years, medium was 2.1 years and high was 11.0 years (mean exposure
duration). Most of the workers (70 out of 99) were primarily exposed to TCE with an average
exposure duration of 7.1 years for 35 hours/week. TCA and trichloroethanol (TCOH) levels
were measured in the urine samples provided by the workers and mean TCA levels in the high
group was 7.7 mg/L and was as high as 26.1 mg/L. Experimental details of trigeminal nerve
evaluation were not provided by the authors. It was reported that 1 out of 21 people (5%) in the
low exposure, 2 out of 37 (5%) in the medium exposure and 4 out of 41 (10%) in the high
exposure group experienced abnormalities in trigeminal nerve sensory function. No linear
association was seen on trigeminal nerve function (Mantel-Haenzel test for linear association,
p = 0.42). However, the trigeminal nerve function findings were not compared to a control (no
TCE exposure) group and it should be noted that some of the workers (29 out of 99) were not
exposed to TCE.
D.I. 1.2. Trigeminal Somatosensory Evoked Potential (TSEP) Studies—Trigeminal Nerve
In a preliminary study, Barret et al. (1982) measured trigeminal sensory evoked potentials
(TSEPs) in eleven workers that were chronically exposed to TCE. Nine of these workers were
suffering effects from TCE intoxication (changes in facial sensitivity and clinical changes in
trigeminal nerve reflexes), and two were TCE-exposed without exhibiting any clinical
manifestations from exposure. A control group of 20 nonexposed subjects of varying ages were
used to establish the normal response curve for the trigeminal nerve function. In order to
generate a TSEP, a surface electrode was placed over the lip and a voltage of 0.05 ms in duration
was applied. The area was stimulated 500 times at a rate of two times per second. TSEPs were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-6 DRAFT—DO NOT CITE OR QUOTE
-------
recorded from a subcutaneous electrode placed between the international CZ point (central
midline portion of the head) and the ear. In eight of the eleven workers, an increased voltage
ranging from a 25 to a 45 volt increase was needed to generate a normal TSEP. Two of the
11 workers had an increased latency of appearance for the TSEP and three workers had increases
in TSEP amplitudes. The preliminary findings indicate that TCE exposure results in
abnormalities in trigeminal nerve function. However, the study does not provide any exposure
data and lacks information with regards to the statistical treatment of the observations.
Barret et al. (1987) conducted a study on 104 degreaser machine operators in France
(average age = 41.6 years; range = 18-62 years) who were highly exposed to TCE with an
average exposure of 7 hours/day for 8.23 years. Although TCE exposure concentrations were
not available, urinary concentrations of TCOH and TCA were measured for each worker. A
control group consisting of 52 subjects without any previous solvent exposure and neurological
deficits was included in the study. Trigeminal nerve symptoms and TSEPs were collected for
each worker. Trigeminal nerve symptoms were clinically assessed by examining facial
sensitivity and reflexes dependent on this nerve such as the jaw and blink reflex. TSEPs were
elicited by electrical stimulation (70-75 V for 0.05 ms) of the nerve using an electrode on the lip
commissure. Eighteen out of 104 TCE-exposed machine operators (17.3%) had trigeminal nerve
symptoms. The subjects that experienced trigeminal nerve symptoms were significantly older
(47.8 years vs. 40.5; p < 0.001). Both groups had a similar duration of exposure with a mean of
9.2 years in the sensitive group and 7.8 years in the nonsensitive group. Urinary concentrations
of TCOH and TCA were also statistically similar although the levels were slightly higher in the
sensitive group (245 mg/g creatinine vs. 162 mg/g creatinine for TCOH; 131 mg/g creatinine vs.
93 mg/g creatinine for TCA). However, in the same group, 40 out of 104 subjects (38.4%) had
an abnormal TSEP. Abnormal TSEPs were characterized as potentials that exhibited changes in
latency and/or amplitude that were at least 2.5 times the standard deviation of the normal TSEPs
obtained from the control group. Individuals with abnormal TSEP were significantly older
(45 years vs. 40.1 years;/? < 0.05) and were exposed to TCE longer (9.9 years vs. 5.6 years;
p < 0.01). Urinary concentrations TCOH and TCA were similar between the groups with
sensitive individuals having average metabolite levels of 195 mg TCOH/g creatinine and
98.3 mg TCA/g creatinine in comparison to 170 mg TCOH/g creatinine and 96 mg TCA/g
creatinine in nonsensitive individuals. When a comparison was made between workers that had
normal TSEP and no trigeminal symptoms and workers that had an abnormal TSEP and
experienced trigeminal symptoms, it was found that in the sensitive individuals (abnormal TSEP
and trigeminal symptoms) there was a significant increase in age (48.5 vs. 39.5 years old,
p < 0.01), duration of exposure (11 vs. 7.5 years,/? < 0.05) and an increase in urinary TCA (313
vs. 181 mg TCA/g creatinine). No significant changes were noted in urinary TCOH, but the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-7 DRAFT—DO NOT CITE OR QUOTE
-------
levels were slightly higher in sensitive individuals (167 vs. 109 mg TCOH/g creatinine).
Overall, it was concluded that abnormal TSEPs were recorded in workers who were exposed to
TCE for a longer period (average duration 9.9 years). This appears to be a well designed study
with statistically significant results reported for abnormal trigeminal nerve response in TCE
exposed workers. Exposure assessment to TCE is by exposure duration and mean urinary TCOH
and TCA concentrations. TCE concentrations to exposed subjects as measured by atmospheric
or personal monitoring are lacking.
Mhiri et al. (2004) measured TSEPs from 23 phosphate industry workers exposed to TCE
for 6 hours/day for at least two years while cleaning tanks. Exposure assessment was based on
measurement of urinary metabolites of TCE, which were performed 3 times/worker, and air
measurements. Blood tests and hepatic enzymes were also collected. The mean exposure
duration was 12.4 ± 8.3 years (exposure duration range = 2-27 years). Although TCE exposures
were not provided, mean urinary concentrations of TCOH, TCA, and total trichlorides were
79.3 ± 42, 32.6 ± 22, and 111.9 ± 55 mg/g urinary creatinine, respectively. The control group
consisted of 23 unexposed workers who worked in the same factory without being exposed to
any solvents. TSEPs were generated from a square wave pulses (0.1 ms in duration) delivered
through a surface electrode that was placed 1 cm under the corner of the mouth. The responses
to the stimuli (TSEPs) were recorded from another surface electrode that was placed over the
contralateral parietal area of the brain. The measured TSEP was divided into several
components and labeled according to whether it was (1) a positive (P) or negative (N) potential
and (2) the placement of the potential in reference to the entire TSEP (e.g., PI is the first positive
potential in the TSEP). TSEPs generated from the phosphate workers that were ±2.5 times the
standard deviation from the TSEPs obtained from the control group were considered abnormal.
Abnormal TSEP were observed in 6 workers with clinical evidence of trigeminal involvement
and in 9 asymptomatic workers. Significant increases in latency were noted for all TSEP
potentials (Nl, PI, N2, P2, N3,/> < 0.01) measured from the phosphate workers. Additionally,
significant decreases in the PI (p < 0.02) and N2 (p < 0.05) amplitudes were observed. A
significant positive correlation was demonstrated between duration of exposure and the N2
latency (p < 0.01) and P2 latency (p < 0.02). Only one subject had urinary TCE metabolite
levels over tolerated limits. TCE air contents were over tolerated levels, ranging from
50-150 ppm (275-825 mg/m3). The study is well presented with statistically significant results
for trigeminal nerve impairment resulting from occupational exposures to TCE. Exposure
potential to TCE is defined by urinary biomarkers, TCA, total trichloro-compounds, and TCOH.
The study lacks information on atmospheric monitoring of TCE in this occupational setting.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-8 DRAFT—DO NOT CITE OR QUOTE
-------
D.I. 1.3. Nerve Conduction Velocity Studies
Nerve conduction latencies were also studied in two occupational studies by Triebig et al.
(1982, 1983) using methods for measurement of nerve conduction which differ from most
published studies, but the results indicate a potential impact on nerve conduction following
occupational TCE exposure. There was no impact seen on latencies in the 1982 study, but a
statistically significant response was observed in the latter study. The latter study, however, is
confounded by multiple solvent exposures.
In Triebig et al. (1982), 24 healthy workers (20 males, 4 females) were exposed to TCE
occupationally at three different plants. The ages ranged from 17-56, and length of exposure
ranged from 1 month to 258 months (mean 83 months). TCE concentrations measured in air at
work places ranged from 5-70 ppm (27-385 mg/m3). A control group of 144 healthy,
complaint-free individuals were used to establish 'normal' responses on the nerve conduction
studies. The matched control group consisted of twenty-four healthy nonexposed individuals
(20 males, 4 females), chosen to match the subjects for age and sex. TCA, TCE, and
trichloroethanol were measured in blood, and TCE and TCA were measured in urine. Nerve
conduction velocities were measured for sensory and motor nerve fibers using the following
tests: MCVMAx (U): Maximum NLG of the motor fibers of the N. ulnaris between the wrist joint
and the elbow; dSCV (U): Distal NLG of mixed fibers of the N. ulnaris between finger V and the
wrist joint; pSCV (U): Proximal NLG of sensory fibers of the N. medianus between finger V and
Sulcus ulnaris; and dSCV (M): Distal NLG of sensory fibers of the N. medianus between finger
III and the wrist joint. Data were analyzed using parametric and nonparametric tests, rank
correlation, linear regression, with 5% error probability. Results show no statistically significant
difference in nerve conduction velocities between the exposed and unexposed groups. This
study has measured exposure data, but exposures/responses are not reported by dose levels.
Triebig et al. (1983) has a similar study design to the previous study (Triebig et al., 1982)
in the tests used for measurement of nerve conduction velocities, and in the analysis of blood and
urinary metabolites of TCE. However, in this study, subjects were exposed to a mixture of
solvents, including TCE, specifically "ethanol, ethyl acetate, aliphatic hydrocarbons (gasoline),
methyl ethyl ketone (MEK), toluene, and trichloroethene." The exposed group consists of
66 healthy workers selected from a population of 112 workers. Workers were excluded based on
polyneuropathy (n = 46) and alcohol consumption (n = 28). The control group consisted of
66 healthy workers with no exposures to solvents. Subjects were divided into three exposure
groups based on length of exposure, as follows: 20 employees with "short-term exposure"
(7-24 months); 24 employees with "medium-term exposure" (25-60 months); 22 employees
with "long-term exposure" (over 60 months). TCA, TCE, and trichloroethanol were measured in
blood, and TCE and TCA were measured in urine. Subjects were divided into exposure groups
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-9 DRAFT—DO NOT CITE OR QUOTE
-------
based on length of exposures, and results were compared for each exposure group to the control
group. In this study, there was a dose-response relationship observed between length of
exposure to mixed solvents and statistically significant reduction in nerve conduction velocities
observed for the medium and long-term exposure groups for the ulnar nerve (NCV).
Interpretation of this study is limited by the mixture of solvent exposure, with no results reported
for TCE alone.
D.I.2. Auditory Effects
There are three large environmental studies reported which assessed the potential impact
of TCE exposures through groundwater ingestion on auditory functioning. They present mixed
results. All three studies were conducted on the population in the TCE Subregistry from the
National Exposure Registry (NER) developed by the Agency for Toxic Substances Disease
Registry (ATSDR). The two studies conducted by Burg et al. (1995; Burg and Gist, 1999) report
an increase in auditory effects associated with TCE exposure, but the auditory endpoints were
self reported by the population, as opposed to testing of measurable auditory effects in the
subject population. The third of these studies, reported by ATSDR (2003) conducted
measurements of auditory function on the subject population, but failed to demonstrate a positive
relationship between TCE exposure and auditory effects. Results from these studies strongly
suggest that children <9 years are more susceptible to hearing impairments from TCE exposure
than the rest of the general population. These studies are described below.
Burg et al. (1995) conducted a study on registrants in the National Health Interview
Survey (NHIS) TCE subregistry of 4,281 (4,041 living and 240 deceased) residents
environmentally exposed to TCE via well water in Indiana, Illinois, and Michigan. Morbidity
baseline data were examined from the TCE Subregistry from the NER developed by the ATSDR.
Participants were interviewed in the NHIS, which consists of 25 questions about health
conditions. Data were self reported via face-to-face interviews. Neurological endpoints were
hearing and speech impairments. This study assessed the long-term health consequences of
long-term, low-level exposures to TCE in the environment. The collected data were compared to
the NHIS, and the National Household Survey on Drug Abuse. Poisson Regression analysis
model was used for registrants 19 and older. The statistical analyses performed treated the NHIS
population as a standard population and applied the age- and sex-specific period prevalence and
prevalence rates obtained from the NHIS data to the corresponding age- and sex-specific
denominators in the TCE Subregistry. This one-sample approach ignored sampling variability in
the NHIS data because of the large size of the NHIS database when compared to the TCE
Subregistry data file. A binomial distribution was assumed in estimating standard errors for the
TCE Subregistry data. Weighted age- and sex-specific period prevalence and prevalence rates
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-10 DRAFT—DO NOT CITE OR QUOTE
-------
by using the person-weights were derived for the TCE subregistry. These "standard" rates were
applied to the corresponding TCE Subregistry denominators to obtain expected counts in each
age and sex combination. In the NHIS sample, 18% of the subjects were nonwhite. In the TCE
Subregistry sample, 3% of the subjects were nonwhite. Given this discrepancy in the proportion
of nonwhites and the diversity of races reported among the nonwhites in the TCE Subregistry,
the statistical analyses included 3,914 exposed white TCE registrants who were alive at baseline.
TCE registrants that were 9 years old or younger had a statistically significant increase in hearing
impairment as reported by the subjects. The relative risk (RR) in this age group for hearing
impairments was 2.13. The RR decreased to 1.12 for registrants aged 10-17 years and to 0.32 or
less for all other age groups. As a result, the effect magnitude was lower for children
10-17 years and for all other age groups. The study reports a dose-response relationship, but the
hearing effects are self-reported, and exposure data are modeled estimates.
Burg and Gist (1999) reported a study conducted on the same subregistry population
described for Burg et al. (1995). It investigated intrasubregistry differences among 3,915 living
members of the National Exposure Registry's Trichloroethylene Subregistry (4,041 total living
members). The participants' mean age was 34 years (SD = 19.9 years), and included children in
the registry. All registrants had been exposed to TCE through domestic use of contaminated well
water. All were Caucasian. All registrants had been exposed to TCE though domestic use of
contaminated well water; there were four exposure Subgroups, each divided into quartiles:
(1) Maximum TCE measured in well water, exposure subgroups include 2-12 ppb; 12-60 ppb;
60-800 ppb; (2) Cumulative TCE exposure subgroups include <50 ppb, 50-500 ppb,
500-5,000 ppb, >5,000 ppb; (3) Cumulative chemical exposure subgroups include TCA, DCE,
dichloroacetic acid (DCA), in conjunction with TCE, with the same exposure Categories as in #
2; and (4) Duration of exposure subgroups include <2 years, 2-5 years, 5-10 years, >10 years;
2,867 had TCE exposure of <50 ppb; 870 had TCE exposure of 51-500 ppb; 190 had TCE
exposure of 501-5,000 ppb; 35 had TCE exposure >5,000 ppb. The lowest quartile was used as
a control group. Interviews included occupational, environmental, demographic, and health
information. A large number of health outcomes were analyzed, including speech impairment
and hearing impairment. Statistical methods used include Logistic Regression and Odds Ratios.
The primary purpose was to evaluate the rate of reporting health-outcome variables across
exposure categories. The data were evaluated for an elevation of the risk estimates across the
highest exposure categories or for a dose-response effect, while controlling for potential
confounders. Estimated prevalence odds ratios for the health outcomes, adjusted for the
potential confounders, were calculated by exponentiating the [^-coefficients from the exposure
variables in the regression equations. The standard error of the estimate was used to calculate
95% confidence intervals (CIs). The referent group used in the logistic regression models was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l 1 DRAFT—DO NOT CITE OR QUOTE
-------
the lowest exposure group. The results variables were modeled as dichotomous, binary
dependent variables in the regression models. Nominal, independent variables were modeled,
using dummy variables. The covariables used were sex, age, occupational exposure, education
level, smoking history, and the sets of environmental subgroups. The analyses were restricted to
persons 19 years of age or older when the variables of occupational history, smoking history, and
education level were included. When the registrants were grouped by duration of exposure to
TCE, a statistically significant association (adjusted for age and sex) between duration of
exposure and reported hearing impairment was found. The prevalence odds ratios were 2.32
(95% Cl: 1.18, 4.56) (>2 to <5 years); 1.17 (95% Cl: 0.55, 2.49) (>5 to <10 years); and 2.46
(95% Cl = 1.30, 5.02) (>10 years). Higher rates of speech impairment (although not statistically
significant) were associated with maximum and cumulative TCE exposure, and duration of
exposure. The study reports dose-response relationships, but the effects are self reported, and
exposure data are estimates. No information was reported on presence or absence of additional
solvents in drinking water.
ATSDR (2003) conducted a follow-up study to the TCE subregistry findings (Burg et al.,
1995, 1999) and focused on the subregistry children. Of the 390 subregistry children (<10 years
at time of original study), 116 agreed to participate. TCE exposure ranged from 0.4 to 5,000 ppb
from the drinking water. The median TCE exposure for this subgroup was estimated to be
23 ppb per year of exposure. To further the hearing impairments reported in Burg et al. (1995,
1999), comprehensive auditory tests were conducted with the 116 children and compared to a
control group of 182 children that was age-matched. The auditory tests consisted of a hearing
screening (typanometry, pure tone and distortion product otoacoustic emissions [DPOAE]) and a
more in-depth hearing evaluation for children that failed the initial screening. Ninety percent of
the TCE-exposed children passed the typanometry and pure tone tests, and there were no
significant differences between control and TCE-exposed groups. Central auditory processing
tests were also conducted and consisted of a test for acoustic reflexes and a screening test for
auditory processing disorders (SCAN). The acoustic reflex tested the ipsilateral and contralateral
auditory pathway at 1,000 Hz for each ear. In this test, each subject hears the sound frequency
and determines if the sound causes the stapedius muscle to tighten the stapes (normal reflex to
noise). Approximately 20% of the children in the TCE subregistry and 5-7% in the controls
exhibited an abnormal acoustic reflex, and this increased abnormality in the test was a significant
effect (p = 0.003). No significant effects were noted in the SCAN tests. The authors concluded
that the significant decrease in the acoustic reflex for the TCE subregistry children is reflective
of potential abnormalities in the middle ear, which may reflect abnormalities in lower brainstem
auditory pathway function. Lack of effects with the pure tone and typanometry tests suggests
that the cochlea is not affected by TCE exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-12 DRAFT—DO NOT CITE OR QUOTE
-------
Although auditory function was not directly measured, Rasmussen et al. (1993b) used a
psychometric test to measure potential auditory effects of TCE exposure in an environmental
study. Results from 96 workers exposed to TCE and other solvents were presented in this study.
The workers were divided into three exposure groups: low, medium, and high. Details of the
exposure groups and exposure levels are provided in Table 4-21 (under study description of
Rasmussen et al., 1993b). Three auditory-containing tasks were included in this study, but only
the acoustic motor function test could be used for evaluation of auditory function. In the
acoustic motor function test, high and low frequency tones were generated and heard through a
set of earphones. Each individual then had to imitate the tones by knocking on the table using
the flat hand for a low frequency and using a fist for a high frequency. A maximal score of 8
could be achieved through this test. The tones were provided in either a set of 1 or 3 groups. In
the one group acoustic motor function test, the average score for the low exposure group was 4.8
in comparison to 2.3 in the high exposure group. Similar decrements were noted in the 3 group
acoustic motor function test. A significant association was reported for TCE exposure and
performance on the one group acoustic motor function test (p < 0.05) after controlling for
confounding variables.
D.I.3. Vestibular Effects
The data linking acute TCE exposure with transient impairment of vestibular function are
quite strong based on human chamber studies, occupational exposure studies, and laboratory
animal investigations. It is clear from the human literature that these effects can be caused by
exposures to TCE, as they have been reported extensively in the literature.
The earliest reports of neurological effects resulting from TCE exposures focused on
subjective symptoms, such as headaches, dizziness, and nausea. These symptoms are subjective
and self-reported, and, therefore, offer no quantitative measurement of cause and effect.
However, there is little doubt that these effects can be caused by exposures to TCE, as they have
been reported extensively in the literature, resulting from occupational exposures (Grandjean et
al., 1955; Liu et al., 1988; Rasmussen and Sabroe, 1986; Smith, 1970), environmental exposures
(Hirsch et al., 1996), and in chamber studies (Stewart et al., 1970; Kylin et al., 1967). These
studies are described below in more detail.
Grandjean et al. (1955) reported on 80 workers exposed to TCE from 10 different
factories of the Swiss mechanical engineering industry. TCE air concentrations varied from
6-1,120 ppm (33-6,200 mg/m3) depending on time of day and proximity to tanks, but mainly
averaged between 20-40 ppm (100-200 mg/m3). Urinalysis (TCA) varied from 30 mg/L to
300 mg/L. This study does not include an unexposed referent group, although prevalences of
self-reported symptoms or neurological changes among the higher-exposure group are compared
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-13 DRAFT—DO NOT CITE OR QUOTE
-------
to the lower-exposure group. Workers were classified based on their exposures to TCE and there
were significant differences (p = 0.05) in the incidence of neurological disorder between
Groups I (10-20 ppm), II (20-40 ppm; 110-220 mg/m3) and III (>40 ppm; 220 mg/m3).
Thirty-four percent of the workers had slight or moderate psycho-organic syndrome; 28% had
neurological changes. Approximately 50% of the workers reported incidences of vertigo and
30% reported headaches (primarily an occasional and/or minimal disorder). Based on TCA
eliminated in the urine, results show that subjective, vegetative, and neurological disorders were
more frequent in Groups II (40-100 mg/L) and III (101-250 mg/L) than in Group I
(10-39 mg/L). Statistics do support a dose-effect relationship between neurological effects and
TCE exposure, but exposure data are questionable.
Liu et al. (1988) evaluated the effects of occupational TCE exposure on 103 factory
workers in Northern China. The workers (79 men, 24 women) were exposed to TCE during
vapor degreasing production or operation. An unexposed control group of 85 men and
26 women was included for comparison. Average TCE exposure was mostly at less than 50 ppm
(275 mg/m3). The concentration of breathing zone air during entire shift was measured by
diffusive samplers placed on the chest of each worker. Subjects were divided into three exposure
groups; 1-10 ppm (5.5-55 mg/m3), 11-50 ppm (60-275 mg/m3) and 51-100 ppm
(280-550 mg/m3). Results were based on a self-reported subjective symptom questionnaire.
The frequency of subjective symptoms, such as nausea, drunken feeling, light-headedness,
floating sensation, heavy feeling of the head, forgetfulness, tremors and/or cramps in extremities,
body weight loss, changes in perspiration pattern, joint pain, and dry mouth (all >3 times more
common in exposed workers); reported as 'prevalence of affirmative answers', was significantly
greater in exposed workers than in unexposed (p < 0.01). "Bloody strawberry jam-like feces"
was borderline significant in the exposed group and "frequentflatus" was statistically
significant. Dose-response relationships were established (but not statistically significant) for
symptoms. Most workers were exposed below 10 ppm, and some at 11-50 ppm. The
differences in exposure intensity between men and women was of borderline significance
(0.05
-------
petroleum, gasoline, toluene, xylene), (3) previously (1-5 years.) worked with chlorinated or
other solvents (n = 66), (4) never worked with organic solvents (n = 94). A dose-response
relationship was observed between exposure to chlorinated solvents and chronic
neuropsychological symptoms including vestibular system effects such as dizziness (p < 0.005),
and headache (p < 0.01). The authors indicated that TCE exposure resulted in the most overall
symptoms. Significant associations were seen between previous exposure and consumption of
alcohol with chronic neuropsychological symptoms. Results are confounded by exposures to
additional solvents.
Smith (1970) conducted an occupational study on 130 workers (108 males, 22 females)
exposed to TCE (industry not reported). The control group consisted of 63 unexposed men
working at the same factories matched by age, marital status and other nonspecified criteria. A
referent group was included and consisted of 112 men and women exposed to low concentration
of lead and matched to the TCE exposed group in age and sex distribution. Seventy-three out of
130 workers (56.2%) reported dizziness and 23 workers reported having headaches (17.7%).
The number of complaints reported by subjects was greater for those with 60 mg/L or greater
TCA than for those with less than 60 mg/L TCA. There was no difference in the number of
symptoms reported between those with shorter durations of exposure and those with longer
durations of exposure. No statistics were reported.
Hirsh et al. (1996) evaluated the vestibular effects of an environmental exposure to TCE
in Roscoe, IL residents. A medical questionnaire was mailed to 103 residents of Roscoe with
100% response. These 103 and an additional 15 residents, not previously surveyed, brought the
subject population to 118 residents. During the course of testing, 12 subjects (young children
and uncooperative patients) were excluded bringing the total number of subjects to 106 all of
whom were in the process of taking legal action against the company whose industrial waste was
assumed to be the source of the polluting TCE. This was a case series report with no controls.
Random testing of the wells between 1983-84 revealed groundwater in wells to have levels of
TCE between 0 to 2,441 ppb. The distance of residence from contaminated well was used to
estimate exposure level. Sixty-six subjects (62%) complained of headaches at the time of
evaluation. Diagnosis of TCE-induced cephalagia was considered credible for 57 patients
(54%). Forty-seven of these had a family history of headaches. Retrospective TCE level of well
water or well's distance from the industrial site analysis did not correlate with the occurrence of
possibly-TCE induced headaches. This study shows a general association between headaches
and exposure to TCE in drinking water wells. There were no statistics to support a
dose-response relationship. All subjects were involved in litigation.
Stewart et al. (1970) evaluated vestibular effects in 13 subjects who were exposed to TCE
vapor 100 ppm (550 mg/m3) and 200 ppm (1,100 mg/m3) for periods of 1 hour to a 5-day work
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-15 DRAFT—DO NOT CITE OR QUOTE
-------
week. Experiments 1-7 were for a duration of 7 hours with a mean TCE concentration of
3
198-200 ppm (1,090-1,100 mg/m ). Experiments 8 and 9 exposed subjects to 190-202 ppm
(1,045-1,110 mg/m3) TCE for a duration of 3.5 and 1 hour, respectively. Experiment 10
exposed subjects to 100 ppm (550 mg/m3) TCE for 4 hours. Experiments 2-6 were carried out
with the same subjects over 5 consecutive days. Gas chromatography of expired air was
measured. There were no self controls. Subjects reported symptoms of lightheadedness,
headache, eye, nose, and throat irritation. Prominent fatigue and sleepiness by all were reported
above 200 ppm (1,100 mg/m3). There were no quantitative data or statistics presented regarding
dose and effects of neurological symptoms.
Kylin et al. (1967) exposed 12 volunteers to 1,000 ppm (5,500 mg/m3) TCE for 2 hours
in a 1.5 x 2 x 2 meters chamber. Volunteers served as their own controls since 7 of the 12 were
pretested prior to exposure and the remaining 5 were post-tested days after exposure. Subjects
were tested for optokinetic nystagmus, which was recorded by electronystogmography, that is,
"the potential difference produced by eye movements between electrodes placed in lateral angles
between the eyes." Venous blood was also taken from the volunteers to measure blood TCE
levels during the vestibular task. The authors concluded that there was an overall reduction in
the limit ("fusion limit") to reach optokinetic nystagmus when individuals were exposed to TCE.
Reduction of the "fusion limit" persisted for up to 2 hours after the TCE exposure was stopped
and the blood TCE concentration was 0.2 mg/100 mL.
D.I.4. Visual Effects
Kilburn (2002a) conducted an environmental study on 236 people exposed to TCE in
groundwater in Phoenix, AZ. Details of the TCE exposure and population are described earlier
in Section D. 1.1.1 (see Kilburn [2002a]). Among other neurological tests, the population and
161 nonexposed controls was tested for color discrimination using the desaturated Lanthony
15-hue test, which can detect subtle changes in color vision deficiencies. Color discrimination
errors were significantly increased in the TCE exposed population (p < 0.05) with errors scores
averaging 12.6 in the TCE exposed in comparison to 11.9 in the control group. This study shows
statistically significant differences in visual response between exposed and nonexposed subjects
exposed environmentally. Estimates of TCE concentrations in drinking water to individual
subjects are lacking.
Reif et al. (2003) conducted a cross sectional environmental study on 143 residents of the
Rocky Mountain Arsenal community of Denver whose water was contaminated with TCE and
related chemicals from nearby hazardous waste sites between 1981 and 1986. The residents
were divided into three groups based on TCE exposure with the lowest exposure group at
<5 ppb, the medium exposure group at 5 to 15 ppb and the high exposure group defined as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-16 DRAFT—DO NOT CITE OR QUOTE
-------
>15 ppb TCE. Visual performance was measured by two different contrast sensitivity tests
(C and D) and the Benton visual retention test. In the two contrast sensitivity tests, there was a
20 to 22% decrease in performance between the low and high TCE exposure groups and
approached statistical significance (p = 0.06 or 0.07). In the Benton visual retention test, which
measures visual perception and visual memory, scores, dropped by 10% from the lowest
exposure to the highest TCE exposure group and was not statistically significant. It should be
noted that the residents were potentially exposed to multiple solvents including TCE and a
nonexposed TCE group was not included in the study. Additionally, modeled exposure data are
only a rough estimate of actual exposures, and possible misclassification bias associated with
exposure estimation may limit the sensitivity of the study.
Rasmussen et al. (1993b) conducted a cross-sectional study on 96 metal workers,
working in degreasing at various factories in Denmark (industries not specified) with chlorinated
solvents. These subjects were identified from a larger cohort of 240 workers. Details of the
exposure groups and TCE exposure levels are presented in Section D. 1.1.1 (under Rasmussen et
al., 1993c). Neuropsychological tests including the visual gestalts (test of visual perception and
retention) and the stone pictures test (test of visual learning and retention) were administered to
the metal workers. In the visual gestalts test, cards with a geometrical figure containing four
items were presented and workers had to redraw the figure from memory immediately (learning
phase) after presentation and after 1 hour (retention phase). In the learning phase, the figures
were redrawn until the worker correctly drew the figure. The number of total errors significantly
increased from the low group (3.4 errors) to the high exposure group (6.5 errors;/? = 0.01) during
the learning phase (immediate presentation). Similarly, during the retention phase of this task
(measuring visual memory), errors significantly increased from an average of 3.2 in the low
group to 5.9 in the high group (p < 0.001). In the stone pictures test, slides of 10 stones
(different shapes and sizes) were shown and the workers had to identify the 10 stones out of a
lineup of 25 stones. There were no significant changes in this task, but the errors increased from
4.6 in the low exposure group to 6.3 in the high exposure group during the learning phase of this
task. Although this study identifies visual performance deficits, a control group (no TCE
exposure) was not included in this study and the presented results may actually underestimate
visual deficits from TCE exposure.
Troster and Ruff (1990) presented case studies conducted on two occupationally exposed
workers to TCE and included a third case study on an individual exposed to
1,1,1-trichloroethane. Case #1 was exposed to TCE (concentration unknown) for 8 months and
Case #2 was exposed to TCE over a 3-month period. Each patient was presented with a
visual-spatial task (Ruff-Light Trail Learning test as referenced by the authors). Both of the
individuals exposed to TCE were unable to complete the visual-spatial task and took the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-17 DRAFT—DO NOT CITE OR QUOTE
-------
maximum number of trials (10) to attempt to complete the visual task. A control group of
30 individuals and the person exposed to 1,1,1-trichloroethane were able to complete this task
accordingly. The lack of quantitative exposure data and a small sample size severely limits the
study and does not allow for statistical comparisons.
Vernon and Ferguson (1969) exposed eight male volunteers (ages 21-30) to 0, 100, 300,
and 1,000-ppm TCE for 2 hours. Each individual was exposed to all TCE concentrations and a
span of at least 3 days was given between exposures. The volunteers were presented with six
visuo-motor tests during the exposure sessions. When the individuals were exposed to
1,000-ppm TCE (5,500 mg/m3), significant abnormalities were noted in depth perception as
measured by the Howard-Dolman test (p < 0.01), but no effects on the flicker fusion frequency
test (threshold frequency at which the individual sees a flicker as a single beam of light) or on the
form perception illusion test (volunteers presented with an illusion diagram). This is one of the
earliest chamber studies of TCE. This study included only healthy young males, is of a small
size, limiting statistical power, and reports mixed results on visual testing following TCE
exposure.
D.1.5. Cognition
There is a single environmental study in the literature that presents evidence of a negative
impact on intelligence resulting from TCE exposure. Kilburn and Warshaw (1993—study
details in Section D.I. 1.1) evaluated the effects on cognition for 544 Arizona residents exposed
to TCE in well-water. Subjects were recruited and categorized into three groups. Exposed
Group 1 consisted of 196 family members with cancer or birth defects. Exposed Group 2
consisted of 178 individuals from families without cancer or birth defects; and exposed Group 3
included 170 parents whose children had birth defects and rheumatic disorders. Sixty-eight
referents were used as a comparison group for the clinical memory tests. Several cognitive tests
were administered to these residents in order to test memory recall skills and determine if TCE
exposure resulted in memory impairment. Working or short-term memory skills were tested by
asking each individual to recall two stories immediately after presentation (verbal recall) and
also draw three diagrams immediately after seeing the figures (visual recall). Additionally, a
digit span test where increasing numbers of digits were presented and then the subject had to
recall the digits was conducted to the extent of the short-term memory. Exposed subjects had
lower intelligence scores and there were significant impairments in verbal recall (p = 0.001),
visual recall (p = 0.03) and with the digit span test (p = 0.07). Significant impairment in
short-term memory as measured by three different cognitive test was correlated with TCE
exposure. Lower intelligence scores (p = 0.0001) as measured by the Culture Fair IQ test may be
a possible confounder in these findings. Additionally, the large range of TCE concentrations
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-18 DRAFT—DO NOT CITE OR QUOTE
-------
(6-500 ppb) and exposure durations (1 to 25 years) and overall poor exposure characterization
precludes a no-observed-adverse-effect level (NOAEL)/lowest-observed-adverse-effect level
(LOAEL) from being estimated from this study on cognitive function.
Rasmussen et al. (1993 a, b) and Troster and Ruff (1990) present results of positive
findings in occupational studies for cognitive effects of TCE. Rasmussen et al. (1993a) reported
an historical cohort study conducted on 96 metal degreasers, identified 2 years previously and
were selected from a population of 240 workers from 72 factories in Denmark. They reported
psychoorganic syndrome, a mild syndrome of dementia characterized by cognitive impairment,
personality changes, and reduced motivation, vigilance, and initiative, was increased in the three
exposure groups. The medium and high exposure groups were compared with the low exposure
group. Neuropsychological tests included WAIS (original version, Vocabulary, Digit Symbol,
Digit Span), Simple Reaction Time, Acoustic-motor function (Luria), Discriminatory attention
(Luria), Sentence Repetition, Paced Auditory Serial Addition Test (PASAT), Text Repetition,
Rey's Auditory Verbal Learning, Visual Gestalts, Stone Pictures (developed for this study,
nonvalidated), revised Santa Ana, Luria motor function, and Mira. The prevalence of
psychoorganic syndrome was 10.5% in low exposure group; 38.9% in medium exposure group;
63.4% in high exposure group, (x2 trend analysis: low vs. medium exposure x2 = 11.0,p value
<0.001; low vs. high exposure x2 = 19.6,/>-value <0.001.) Psychoorganic syndrome increased
with age (p < 0.01). Age was strongly correlated with exposure.
Rasmussen et al. (1993b) used a series of cognitive tests to measure effects of
occupational TCE exposure. Short-term memory and retention following an latency period of
one hour was evaluated in several tests including a verbal recall (auditory verbal learning test),
visual gestalts, visual recall (stone pictures), and the digit span test. Significant cognitive
performance decreases were noted in both short-term memory and memory retention. In the
verbal recall test immediate memory and learning were significantly decreased (p = 0.03 and
0.04, respectively). No significant effects were noted for retention following a one hour latency
period was noted. Significant increases in errors were noted in both the learning (p = 0.01) and
memory (p < 0.001) phases for the visual gestalts test. No significant effects were found in the
visual recall test in either the learning or memory phases or in the digit span test. As a result,
there were some cognitive deficits noted in TCE-exposed individuals as measured through
neuropsychological tests.
Troster and Ruff (1990) provides additional supporting evidence in an occupational study
for cognitive impairment, although the results reported in a qualitative fashion are limited in their
validity. In the two case studies that were exposed to TCE, there were decrements (no statistical
analysis performed) in cognitive performance as measured in verbal and visual recall tests that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-19 DRAFT—DO NOT CITE OR QUOTE
-------
were conducted immediately after presentation (learning phase) and one hour after original
presentation (retention/memory phase).
Triebig et al. (1977b) presents findings of no impairment of cognitive ability resulting
from TCE exposure in an occupational setting. This study was conducted on 8 subjects
occupationally exposed to TCE. Subjects were 7 men and 1 woman with an age range from
23-38 years. Measured TCE in air averaged 50 ppm (260 mg/m3). Length of occupational
exposure was not reported. There was no control group. Results were compared after exposure
periods, and compared to results obtained after periods removed from exposure. TCA and TCE
metabolites in urine and blood were measured. The testing consisted of the Syndrome Short
Test, which consists of nine subtests through which amnesic and simple perceptive and cognitive
functional deficits are detected; the "Attention Load Test" or "d2 Test" from Brickenkamp is a
procedure that measures attention, concentration, and stamina. Number recall test, letter recall
test, the "Letter Reading Test," "Word Reading Test." Data were assessed using Wilcoxon and
Willcox nonparametric tests. Due to the small sample size a significance level of 1% was used.
The concentrations of TCE, trichloroethanol, and TCA in the blood and total TCE and total TCA
elimination in the urine were used to assess exposure in each subject. The mean values observed
were 330 mg trichloroethanol and 319 mg TCA/g creatinine, respectively, at the end of a work
shift. The psychological tests showed no statistically significant difference in the results before
or after the exposure-free time period. The small sample size may limit the sensitivity of the
study.
Salvini et al. (1971), Gamberale et al. (1976), and Stewart et al. (1970) reported positive
findings for the impairment of cognitive function following TCE exposures in chamber studies.
Salvini et al. (1971) reported a controlled exposure study conducted on six male university
students. TCE concentration was 110 ppm (550 mg/m3) for 4-hour intervals, twice per day.
Each subject was examined on two different days, once under TCE exposure, and once as self
controls, with no exposure. Two sets of tests were performed for each subject corresponding to
exposure and control conditions. The test battery included a perception test with tachistoscopic
presentation, the Wechsler memory scale test, a complex reaction time test, and a manual
dexterity test. Statistically significant results were observed for perception tests learning
(p < 0.001), mental fatigue (p < 0.01), subjects (p < 0.05); and CRT learning (p < 0.01), mental
fatigue (p < 0.01), subjects (p < 0.05). This is controlled exposure study with measured dose
(110 ppm; 600 mg/m3) and clear, statistically significant impact on neurological functional
domains. However, it only assesses acute exposures.
Gamberale et al. (1976) reported a controlled exposure study conducted on 15 healthy
men aged 20-31 yrs old, employed by the Department of Occupational Medicine in Stockholm,
Sweden. Controls were within subjects (15 self-controls), described above. Test used included
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-20 DRAFT—DO NOT CITE OR QUOTE
-------
reaction time (RT) Addition and short term memory using an electronic panel. Subjects also
assessed their own conditions on a 7-point scale. Researchers used a repeated measures analysis
of variance (ANOVA) for the 4 performance tests based on a 3 x 3 Latin square design. In the
short-term memory test (version of the digit span test), a series of numbers lasting for one second
was presented to the subject. The volunteer then had to reproduce the numerical sequence after a
latency period (not specified). No significant effect on the short-term memory test was observed
with TCE exposure in comparison to air exposure. Potential confounders from this study include
repetition of the same task for all exposure conditions, volunteers served as their own controls,
and TCE exposure preceded air exposure in two of the three exposure experimental designs.
This is a well controlled study of short term exposures with measured TCE concentrations and
significant response observed for cognitive impairment.
Additional qualitative support for cognitive impairment is provided by Stewart et al.
(1970). This was a controlled exposure study conducted on 13 subjects in 10 experiments, which
consisted often chamber exposures to TCE vapor of 100 ppm (550 mg/m3) and 200 ppm
(1,100 mg/m3) for periods of 1 hour to a 5-day work week. Experiments 1-7 were for 7 hours
with a mean TCE concentration of 198-200 ppm (1,090-1,100 mg/m3). Experiments 8 and
9 exposed subjects to 190-202 ppm (1,045-1,110 mg/m3) TCE for a duration of 3.5 and 1 hour,
respectively. Experiment 10 exposed subjects to 100 ppm (550 mg/m3) TCE for 4 hours.
Experiments 2-6 were carried out with the same subjects over 5 consecutive days. Gas
chromatography of expired air was measured. There were no self controls. All had normal
neurological tests during exposure, but 50% reported greater mental effort was required to
perform a normal modified Romberg test on more than one occasion. There were no quantitative
data or statistics presented regarding dose and effects of neurological symptoms.
Two chamber studies conducted by Triebig et al. (1976, 1977a) report no impact of TCE
exposure on cognitive function. Triebig et al. (1976) was a controlled exposure study conducted
on 7 healthy male and female students (4 females, 3 males) exposed for 6 hours/day for 5 days to
100 ppm (550 mg/m3 TCE). The control group was 7 healthy students (4 females, 3 males)
exposed to hair care products. This was assumed as a zero exposure, but details of chemical
composition were not provided. Biochemical and psychological testing was conducted at the
beginning and end of each day. Biochemical tests included TCE, TCA, and trichloroethanol in
blood. Psychological tests included the d2 test, which was an attention load test; the short test
(as characterized in the translated version of Treibig, 1976) is used to record patient performance
with respect to memory and attention; daily Fluctuation Questionnaire measured the difference
between mental states at the start of exposure and after the end of exposure is recorded; The
MWT-A is a repeatable short intelligence test; Culture Fair Intelligence Test (CFT-3) is a
nonverbal intelligence test that records the rather "fluid" part of intelligence, that is, finding
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-21 DRAFT—DO NOT CITE OR QUOTE
-------
solution strategies; Erlanger Depression Scale. Results were not randomly distributed. The
median was used to describe the mean value. Regression analyses were conducted. In this study
the TCE concentrations in blood reported ranged from 4 to 14 ug/mL. A range of 20 to
60 ug/mL was obtained for TCA in the blood. There was no correlation seen between exposed
and unexposed subjects for any measured psychological test results. The biochemical data did
demonstrate subjects' exposures. This is a well controlled study with excellent exposure data,
although the small sample size may have limited sensitivity.
Triebig et al. (1977a) is an additional report on the seven exposed subjects and seven
controls evaluated in Triebig et al. (1976). Additional psychological testing was reported. The
testing included the Syndrome Short Test, which consists of nine subtests, described above.
Statistics were conducted using Whitney Mann. Results indicated the anxiety values of the
placebo random sample group dropped significantly more during the course of testing (p < 0.05)
than those of the active random sample group. No significantly different changes were obtained
with any of the other variables. Both these studies were well controlled with excellent exposure
data, which may provide some good data for establishing a short term NOAEL. The small
sample size may have limited the sensitivity of the study.
Additional reports on the impairment of memory function as a result of TCE exposures
have been reported, and provide additional evidence of cognitive impairment. The studies by
Chalupa et al. (1960), Rasmussen et al. (1986, 1993b), and Troster and Ruff (1990) report
impairment of memory resulting from occupational exposures to TCE. Kilburn and Warshaw
(1993) and Kilburn (2002a) report impairment of memory following environmental exposures to
TCE. Salvini et al. (1971) reports impairment of memory in a chamber study, although Triebig
et al. (1976) reports no impact on memory following TCE exposure in a chamber study.
D.1.6. Psychomotor Effects
There is evidence in the literature that TCE can have adverse psychomotor effects in
humans. The effects of TCE exposure on psychomotor response have been studied primarily as
the impact on RTs, which provide a quantitative measure of the impact TCE exposure has on
motor skills. Studies on motor dyscoordination resulting from TCE exposure are more
subjective, but provide additional evidence that TCE may cause adverse psychomotor effects.
These studies are described below.
D.l.6.1. Reaction Time
There are several reports in the literature that report an increase in reaction times
following exposures to TCE. The best evidence for TCE exposures causing an increase in choice
reaction times comes from environmental studies by Kilburn (2002a), Kilburn and Warshaw
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-22 DRAFT—DO NOT CITE OR QUOTE
-------
(1993), Reif et al. (2003), and Kilburn and Thornton (1996), which were all conducted on
populations which were exposed to TCE through groundwater contaminated as the result of
environmental spills. Kilburn (2002a—study details described in Section D.I.I) evaluated
reaction times in a Phoenix, Arizona population exposed to TCE through groundwater.
Volunteers were tested for response rates in the simple reaction time (SRT) and 2 choice reaction
time (CRT) tests. Various descriptive statistics were used, as well as analysis of covariance
(ANCOVA) and a step-wise adjustment of demographics. The principal comparison, between
the 236 exposed persons and the 161 unexposed regional controls, revealed significant
differences (p < 0.05) indicating that SRTs and CRTs were delayed. Balance was also abnormal
with excessive sway speed (eyes closed), but this was not true when both eyes were open. This
study shows statistically significant differences in psychomotor responses between exposed and
nonexposed subjects exposed environmentally. However, it is limited by poor exposure
characterization.
Kilburn and Warshaw (1993; study details described in Section D.I. 1.1) evaluated
reaction times in 170 Arizona residents exposed to TCE in well water. A referent group of
68 people was used for comparison. TCE concentration was from 6 to 500 ppb and exposure
ranged from 1 to 25 years. SRT was determined by presenting the subject a letter on a computer
screen and measuring the time (in milliseconds [msec]) it took for the person to type that letter.
SRT significantly increased from 281 ± 55 msec to 348 ± 96 msec in TCE-exposed individuals
(p < 0.0001). Similar increases were reported for CRT where subjects were presented with two
different letters and required to make a decision as to which letter key to press. CRT of the
exposed subjects was 93 msec longer in the third trial (p < 0.0001) than referents. It was also
longer in all trials, and remained significantly different after age adjustment. This study shows
statistically significant differences for neurological test results between subjects environmentally
exposed and nonexposed to TCE, but is limited by poor exposure data on individual subjects
given the ecological design of this study. Additionally, litigation is suggested and may introduce
a bias, particularly if no validity tests were used.
Kilburn and Thornton (1996) conducted an environmental study that attempts to use
reference values from two control groups in assessing neurological responses for chemically
exposed subjects using neurophysiological and neuropsychological testing on three groups.
Group A included randomly selected registered voters from Arizona and Louisiana with no
exposure to TCE: n = 264 unexposed volunteers aged 18-83. Group B included volunteers from
California n = 29 (17 males and 12 females) that were used to validate the equations; Group C
included those exposed to TCE and other chemicals residentially for 5 years or more n = 237.
Group (A), was used to develop the regression equations for SRT and choice reaction time
(CRT). A similarly selected comparison group B was used to validate the equations. Group C,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-23 DRAFT—DO NOT CITE OR QUOTE
-------
the exposed population, was submitted to SRT and CRT tests (n = 237) and compared to the
control groups. All subjects were screened by a questionnaire. Reaction speeds were measured
using a timed computer visual-stimulus generator. No exposure data were presented. The Box-
Cox transformation was used for dependent variables and independent variables. They evaluated
graphical methods to study residual plots. Cook's distance statistic was used as a measure of
influence to exclude outliers with undue influence and none of the data were excluded. Lack-of-
fit test was performed on Final model and F statistic was used to compare estimated error to
lack-of-fit component of the model's residual sum of squared error. Final models were validated
using group B data and paired t-test to compare observed values for SRT and CRT. F statistic
was used to test the hypothesis that parameter estimates obtained with group B were equal to
those of Group A, the model. The results are as follows: Group A: SRT = 282 ms;
CRT = 532 ms. Group B: SRT = 269 ms; CRT = 531 ms. Group C: SRT = 334 ms;
CRT = 619 ms. TCE exposure produced a step increase in reaction times (SRT and CRT). The
coefficients from Group A were valid for group B. The predicted value for SRT and for CRT,
plus 1.5 SDs selected 8% of the model group as abnormal. The model produced consistent
measurement ranges with small numerical variation. This study is limited by lack of any
exposure data, and does not provide statistics to demonstrate dose-response effects.
Kilburn (2002a) conducted an environmental study on 236 residents chronically exposed
to TCE-associated solvents in the groundwater resulting from a spill from a microchip plant in
Phoenix, AZ. Details of the TCE exposure and population are described earlier in
Section D. 1.1.1 (see Kilburn [2002a]). The principal comparison, between the 236 exposed
persons and the 161 unexposed regional controls, revealed significant differences indicating that
SRTs and choice reaction times (CRTs) were increased. SRTs significantly increased from
283 ± 63 msec in controls to 334 ±118 msec in TCE exposed individuals (p < 0.0001).
Similarly, CRTs also increased from 510 ± 87 msec to 619 ± 153 msec with exposure to TCE
(p < 0.0001). This study shows statistically significant differences in psychomotor responses as
measured by reaction times between TCE-exposed and nonexposed subjects. Estimates of TCE
concentrations in drinking water to individual subjects were not reported in the paper. Since the
TCE exposure ranged from 0.2 to over 10,000 ppb in well water, it is not possible to determine a
NOAEL for increased reaction times through this study. Additionally, litigation is suggested and
may introduce a bias, particularly if no validity tests were used.
Reif et al. (2003) conducted a cross sectional study on 143 residents of the Rocky
Mountain Arsenal (RMA) community of Denver exposed environmentally to drinking water
contaminated with TCE and related chemicals from nearby hazardous waste sites between 1981
and 1986. The referent group was at the lowest estimated exposure concentration (<5 ppb). The
socioeconomic profile of the participants closely resembled those of the community in general.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-24 DRAFT—DO NOT CITE OR QUOTE
-------
"A total of 3393 persons was identified through the census, from which an age- and
gender-stratified sample of 1267 eligible individuals who had lived at their current residence for
at least 2 years was drawn. Random selection was then used to identify 585 persons from within
the age-gender strata, of whom 472 persons aged 2-86 provided samples for biomonitoring.
Neurobehavioral testing was conducted on 204 adults who lived in the RMA exposure area for a
minimum of 2 years. Among the 204 persons who were tested, 184 (90.2%) lived within the
boundaries of the LWD and were originally considered eligible for the current analysis.
Therefore, participants who reported moving into the LWD after 1985 were excluded from the
total of 184, leaving 143 persons available for study." An elaborate hydraulic simulation model
(not validated) was used in conjunction with a geographic information system (GIS) to model
estimates of residential exposures to TCE. The TCE concentration measured in community
wells exceeded the MCL of 5 ppb in 80% of cases. Approximately 14% of measured values
exceeded 15 ppb. Measured values were used to model actual exposure estimates based on
distance of residences from sampled wells. The estimated exposure for the high exposure group
was >15 ppb; the estimate for the low exposure referent group was <5 ppb. The medium
exposure group was estimated at exposures 5< x <15 ppb TCE. The test battery consisted of the
Neurobehavioral Core Test Battery (NCTB), which consists of 7 neurobehavioral tests including
simple reaction time. Results were assessed using the Multivariate Model. Results were
statistically significant (p < 0.04) for the simple reaction time tests. The results are confounded
by exposures to additional solvents and modeled exposure data, which while highly technical,
are still only a rough estimate of actual exposures, and may limit the sensitivity of the study.
Gamberale et al. (1976) conducted a controlled exposure (chamber) study on 15 healthy
men aged 20-31 yrs old, employed by the Department of Occupational Medicine in Stockholm,
Sweden. Controls were within subjects (15 self-controls). Subjects were exposed to TCE for
70 minutes via a breathing valve to 540 mg/m3 (97 ppm), 1,080 mg/m3 (194 ppm), and to
ordinary atmospheric air (0 ppm). Sequence was counterbalanced between the 3 groups, days,
and exposure levels. Concentration was measured with a gas chromatographic technique every
third minute for the first 50 minutes, then between tests thereafter. Test used were RT addition,
simple RT, choice RT and short term memory using an electronic panel. Subjects also assessed
their own conditions on a 7 point scale. The researchers performed Friedman two-way analysis
by ranks to evaluate differences between the 3 conditions. The results were nonsignificant when
tested individually, but significant when tested on the basis of six variables. Nearly half of the
subjects could distinguish exposure/nonexposure. Researchers performed ANOVA for the four
performance tests based on a 3 x 3 Latin square design with repeated measures. In the RT-
Addition test the level of performance varied significantly between the different exposure
conditions (F[2.24] = 4.35;p < 0.05) and between successive measurement occasions
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-25 DRAFT—DO NOT CITE OR QUOTE
-------
(F[2.24] = 19.25;p < 0.001). The level of performance declined with increased exposure to
TCE, whereas repetition of the testing led to a pronounced improvement in performance as a
result of the training effect. No significant interaction effects were observed between exposure
to TCE and training. This is a good study of short term exposures with measured TCE
concentrations and significant response observed for reaction time.
Gun et al. (1978) conducted an occupational study on 8 TCE-exposed workers who
operated degreasing baths in two different plants. Four female workers were exposed to TCE
only in one plant and four female workers were exposed to TCE and nonhalogenated
hydrocarbon solvents in the second plant. The control group (n = 8) consisted of 4 female
workers from each plant who did not work near TCE. Each worker worked 2 separate 4-hour
shifts daily, with one shift exposed to TCE and the second 4-hour shift not exposed. Personal air
samples were taken continuously over separate 10-minute sessions. Readings were taken every
30 seconds. Eight-choice reaction times were carried out in four sessions; at the beginning and
end of each exposure to TCE or TCE + solvents; a total of 40 reaction time trials were
completed. TCE concentrations in the TCE only plant 1 (148-418 ppm [800-2,300 mg/m3])
were higher than in the TCE + solvent plant 2 (3-87 ppm (16-480 mg/m3). Changes in choice
reaction times (CRT) were compared to level of exposure. The TCE only group showed a mean
increase in reaction time, with a probable cumulative effect. In the TCE + solvent group, mean
reaction time shortened in Session 2, then increased to be greater than at the start. Both control
groups showed a shortening in mean choice reaction time in Session 2, which was sustained in
Sessions 3 and 4 consistent with a practice effect. This is a study with well-defined exposures
and reports of cause and effect (TCE exposure on reaction time); however, no statistics were
presented to support the conclusions or the significance of the findings, and the small sample size
is a limitation of the study.
D.I.6.2. Muscular Dyscoordination
Effects on motor dyscoordination resulting from TCE exposure have been reported in the
literature. These impacts are subjective, but may provide additional evidence that TCE can cause
adverse psychomotor effects. There are three reports summarized below which suggest that
muscular dyscoordination resulted from TCE exposure, although all three have significant
limitations due to confounding factors. Rasmussen et al. (1993c) presented findings on muscular
dyscoordination as it relates to TCE exposure. This was a historical cohort study conducted on
96 metal degreasers, identified 2 years previously. Subjects were selected from a population of
240 workers from 72 factories in Denmark. Although the papers report a population of
99 participants, tabulated results were presented for a total of only 96. No explanation was
provided for this discrepancy. These workers had chronic exposure to fluorocarbon (CFC 113)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-26 DRAFT—DO NOT CITE OR QUOTE
-------
(n = 25) and mostly TCE (n = 70; average duration: 7.1 years.). There were no external controls.
The range of working full-time degreasing was 1 month to 36 years. Researchers collected data
regarding the workers' occupational history, blood and urine tests, as well as biological
monitoring for TCE and TCE metabolites. A chronic exposure index (CEI) was calculated based
on number of hours per week worked with solvents multiplied by years of exposure multiplied
by 45 weeks per year. No TCE air concentrations were reported. Participants were categorized
into three groups: (1) "Low exposure:" n = 19, average full-time exposure = 0.5 years.
(2) "Medium exposure:" n = 36, average full-time exposure = 2.1 years. (3) "High exposure:"
n = 41, average full-time exposure =11 years. The mean TCA level in the "high" exposure
group was 7.7 mg/L (max = 26.1 mg/L). Time-weighted average (TWA) measurements of
CFC 113 levels were 260-420 ppm (U.S. and Danish TLV was 500 ppm). A significant trend of
dyscoordination from low to high solvent exposure was observed (p = 0.003). This study
provides evidence of causality for muscular dyscoordination resulting from exposure to TCE, but
no measured exposure data were reported.
Additional evidence of the psychomotor effects caused by exposure to TCE are presented
in Gash et al. (2007) and Troster and Ruff (1990). There are, however, significant limitations
with each of these studies. In Gash et al. (2007), the researchers evaluated the clinical features of
1 Parkinson's disease (PD) patient, identified in a Phase 1 clinical trial study, index case, and an
additional 29 coworkers of the patient, all with chronic occupational exposures to TCE. An
additional 2 subjects with Parkinson's Disease were included, making the total of 3 Parkinson's
disease patients, and 27 non-Parkinson's coworkers making up the study population. Coworkers
for the study were identified using a mailed questionnaire to 134 former coworkers. No details
are provided in the paper on selection criteria for the 134 former coworkers. Of the 134 former
workers sent questionnaires, 65 responded. Twenty-one self-reported no symptoms, 23 endorsed
1-2 symptoms, and 21 endorsed 3 or more signs of Parkinsonism. Fourteen of the 21 with 3 or
more signs and 13 of the 21 without any signs agreed to a clinical exam; this group comprises the
27 additional workers examined for Parkinsonian symptoms. No details were provided on
nonresponders. All subjects were involved in degreasing with long-term chronic exposure to
TCE through inhalation and dermal exposure (14 symptomatic: age range = 31-66, duration of
employment range: 11-35 yrs) (13 asymptomatic: age range = 46-63, duration of employment
range: 8-33 yrs). The data were compared between groups and with data from 110 age- matched
controls. Exposure to TCE is self-reported and based on job proximity to degreasing operations.
The paper lacks any description of degreasing processes including TCE usage and quantity.
Mapping of work areas indicated that workers with PD worked next to the TCE container, and
all symptomatic workers worked close to the TCE container. Subjects underwent a general
physical exam, neurological exam and Unified Parkinson's Disease Rating Scale (UPDRS),
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-27 DRAFT—DO NOT CITE OR QUOTE
-------
timed motor tests, occupational history survey, and mitochondrial neurotoxicity. ANOVA
analysis was conducted, comparing symptomatic versus nonsymptomatic workers, and
comparing symptomatic workers to age-matched nonexposed controls. No description of the
control population (n =110), nor how data were obtained for this group, was presented. The
symptomatic non-Parkinson's group was significantly slower in fine motor hand movements
than age-matched nonsymptomatic group (p < 0.001). The symptomatic group was significantly
slower (p < 0.0001) than age-matched unexposed controls as measured in fine motor hand
movements on the Movement Analysis Panel. All symptomatic workers had positive responses
to 1 or more questions on UPDRS Part II (diminished activities of daily life), and/or
deteriorization of motor functions on Part III. The fine motor hand movement times of the
asymptomatic TCE-exposed group were significantly slower (p < 0.0001) than age-matched
nonexposed controls. Also, in TCE-exposed individuals, the asymptomatic group's fine motor
hand movements were slightly faster (p < 0.01) than those of the symptomatic group. One
symptomatic worker had been tested 1 year prior and his UPDRS score had progressed from
9 to 23. Exposures are based on self-reported information, and no information on the control
group is presented. One of the PD patients predeceased the study and had a family history
ofPD.
Troster and Ruff (1990) reported a case study conducted on two occupationally exposed
workers to TCE. Patients were exposed to low levels of TCE. There were 2 groups of n = 30
matched controls (all age and education matched) whose results were compared to the
performance of the exposed subjects. Exposure was described as "Unknown amount of TCE for
8 months." Assessment consisted of the San Diego Neuropsychological Test Battery (SDNTB)
and "1 or more of Thematoc Apperception Test (TAT), Minnesota Multiphasic Personal
Inventory (MMPI), and Rorschach. Medical examinations were conducted, including
neurological, CT scan, and/or chemo-pathological tests, and occupational history was taken, but
not described. There were no statistical results reported. Results were reported for each test, but
no tests of significance were included, therefore, the authors presented their conclusions for each
"case" in qualitative terms, as such: Case 1: Intelligence "deemed" to drop from premorbid
function at 1 year 10 months after exposure. Impaired functions improved for all but reading
comprehension, visuospatial learning and categorization (abstraction). Case 2: Mild deficits in
motor speed, but symptoms subsided after removal from exposure.
D.1.7. Summary Tables
The following Tables (D-l through D-3) provide a detailed summary of all the
neurological studies conducted with TCE in humans. Tables D-l and D-2 summarize each
individual human study where there was TCE exposure. Table D-l consists of studies where
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-28 DRAFT—DO NOT CITE OR QUOTE
-------
humans were primarily or soley exposed to TCE. Table D-2 contains human studies where there
was a mixed solvent exposure and TCE was one of the solvents in the mixture. For each study
summary, the study population, exposure assessment, methods, statistics, and results are
provided. Table D-3 indicates the neurological domains that were tested from selected
references (primarily from Table D-l).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-29 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-l. Epidemiological studies: Neurological effects of trichloroethylene
to
o
vo
Reference
Barret etal.,
1984
Study population
188 workers exposed
to TCE occupationally
from small and large
factories in France
(type of factories not
disclosed); average
age = 41; 6yrs
average exposure
time.
The workers were
divided into high and
low exposure groups
for both TCE and
urinary TCA. No
control group was
mentioned.
Exposure assessment
and biomarkers
Review of medical
records and analysis of
TCE atmospheric levels
(detector tubes) and
level of urinary
metabolites
measurement (TCA).
TCE exposure groups
included high exposure
group (>150 ppm;
n = 54) and low
exposure group
<150 ppm; n = 134).
Personal factors
including age, tobacco
use, and alcohol intake
were also analyzed;
Exposure duration =
7 h/d for 7 yrs; no
mention was made
regarding whether or
not the examiners were
blind to the subjects'
exposure status.
Tests used
Complete physical
examination including
testing visual
performance (acuity
and color perception),
evoked trigeminal
potential latencies and
audiometry, facial
sensitivity, reflexes,
and motoricity of the
masseter muscles.
Statistics
X2 examined
distribution of the
different groups
for comparing high
and low exposed
workers, one way
analysis of
variance, Mann
Whitney U and
t-test for analyzing
personal factors.
Results
Symptoms for which TCE role is statistically
significant include the following: Trigeminal
nerve impairment was reported in 22.2% (n = 12)
of workers in the high-exposure group for TCE,
7.4% (n = 10) in the low-exposure group for TCE,
24.4% (n = 10) in the high-exposure group for
TCA and 8.2% (n = 12) in the low-exposure
group for TCA.
TCE Results
Trigeminal
nerve
Impairment
asthenia
Optic nerve
impairment
Headache
Dizziness
High
dose%
22.2
18.5
14.8
20.3
13
Low
dose%
7.4
4.5
0.75
19.4
4.5
P
0.01
<0.01
0.001
NS
0.05
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
^§
H I
O ^
HH Oq
H TO
O
H
W
Barret etal.,
1987
104 occupationally
exposed workers
highly exposed to TCE
during work as
degreaser machine
operators in France.
Controls: 52 healthy,
nonexposed controls of
various ages who were
free from neurological
problems.
Urinary analysis
determined TCE and
TCA rates. The average
of the last
5 measurements were
considered indicative of
the average level of past
exposure. Mean
exposure 8.2 yrs,
average daily exposure
7 hrs/d. Mean age
41.6 yrs.
Evoked trigeminal
potentials were studied
while eyes closed and
fully relaxed. Also,
physical exams with
emphasis on nervous
system, a clinical study
of facial sensitivity, and
of the reflexes
depending on the
trigeminal nerve were
systematically
performed. Normal
latency and amplitude
values for TSEP
obtained from data from
control population.
Normal response
characterized from
4 main peaks,
alternating from
negative to positive,
respective latency of
12.8 ms (SD = 0.6),
19.5 ms (SD = 1.3),
27.6 ms(SD = 1.6), and
36.8 ms (SD = 2.2),
mean amplitude of
response is 2.5 uv
(SD = 0.5 uv).
Pathological responses
were results 2 l/i SDs
over the normal value.
Student's t-test and
one-way ANOVA
used as well as
nonparametric tests
Mann-Whitney U
test and Kruskal-
Wallis test. Also
decision matrix and
the analysis of the
receiver operating
curve to appreciate
the accuracy of the
TSEP method. The
distribution of the
different
populations was
compared by a chi
square test.
Dizziness (71.4%), headache (55.1%), asthenia
(46.9%), insomnia (24.4%), mood perturbation
(20.4%), and sexual problems (12.2%) were found.
Symptomatic patients had significantly longer
exposure periods and were older than
asymptomatic patients. 17.3% of patients had
trigeminal nerve symptoms. Bilateral
hypoesthesia with reflex alterations in 9 cases.
Hypoesthesia was global and predominant in the
mandibular and maxillary nerve areas. Several
reflex abolitions were found without facial palsy
and without convincing hypoesthesia in 9 cases.
Cornea! reflexes were bilaterally abolished in
5 cases as were naso-palpebral reflexes in 6 cases;
length of exposure positively correlated with
functional manifestations (p < 0.01); correlation
between symptoms and exposure levels were
nonsignificant; 40 (38.4%) subjects had
pathological response to TSEP with increased
latencies, amplitude or both; of these 28 had
normal clinical trigeminal exam and 12 had
abnormal exam; TSEP was positively correlated
with length of exposure (p < 0.01); and with age
(p < 0.05), but not with exposure concentration;
trigeminal nerve symptoms (n= 18) were
positively correlated with older age (p < 0.001).
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
•••2
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Barret etal.,
1982
Burg et al.,
1995
Study population
Eleven workers with
chronic TCE exposure;
9 were suffering
effects of solvent
intoxication; 2 were
work place controls.
Control group was
20 unexposed subjects
of all ages.
FromanNHISTCE
subregistry of 4,281
(4,041 living and
240 deceased)
residents
environmentally
exposed to TCE via
well water in Indiana,
Illinois, and Michigan;
compared to NHIS
registrants.
Exposure assessment
and biomarkers
Selected following
clinical evaluations of
their facial sensitivity
and trigeminal nerve
reflexes; exposures
verified by urinalysis.
Presence of TCE and
TCA found. (Exposure
rates not reported).
Morbidity baseline data
were examined from the
TCE Subregistry from
the NER developed by
the ATSDR; were
interviewed in the
NHIS.
Tests used
Somatosensory evoked
potential (SEP)
following stimulation of
the trigeminal nerve
through the lip
alternating right and left
by a bipolar surface
electrode utilizing
voltage, usually 75 to
80 V, just below what is
necessary to stimulate
the orbicularis oris
muscle. Duration was
approx. 0.05 ms
stimulated 500 times
(2x/sec).
Self report via face-to-
face interviews —
25 questions about
health conditions; were
compared to data from
the entire NHIS
population;
neurological endpoints
were hearing and
speech impairments.
Statistics
SEP recordings
illustrated from
trigeminal nerve
graphs.
Poisson Regression
analysis model used
for registrants 19
and older.
Maximum
likelihood
estimation and
likelihood ratio
statistics and Wald
CI; TCE
subregistry
population was
compared to larger
NHIS registry
population.
Results
3 pathological abnormalities present in exposed
(TCE intoxicated) workers: (1) in 8 workers higher
voltage required to obtain normal response, (2)
excessive delay in response observed twice, (3)
excessive graph amplitude noted in 3 cases. One
subject exhibited all 3 abnormalities. Correlation
was reported between clinical observation and test
results. Most severe SEP alternations observed in
subjects with the longest exposure to TCE
(although exposure levels or exposure durations
are not reported). No statistics presented.
Speech impairments showed statistically
significant variability in age-specific risk ratios
with increased reporting for children <9 yrs
(RR: 2.45, 99% CI: 1.31, 4.58) and for registrants
>35 yrs (data broken down by 10-yr ranges).
Analyses suggest a statistically significant increase
in reported hearing impairments for children <9
yrs (RR: 2.13, 99% CI: 1.12, 4.06). It was lower
for children 10-17 yrs (RR: 1.12, 99% CI: 0.52,
2.44) and <0.32 for all other age groups.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
vo S3'
&
a
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
TO'
OJ o
S
i
Burg and
Gist, 1999
4,041 living members
of the National
Exposure Registry's
Trichloroethylene
Subregistry; 97%
white; mean age 34 yrs
(SD= 19.9 yrs.);
divided in 4 groups
based on type and
duration of exposure;
analysis reported only
for 3,915 white
registrants; lowest
quartile used as control
group.
All registrants exposed
to TCE though domestic
use of contaminated
well water; 4 exposure
Subgroups, each divided
into quartiles:
(1) Maximum TCE
measured in well water,
exposure subgroups:
2-12 ppb; 12-60 ppb;
60-800 ppb;
(2) Cumulative TCE
exposure subgroups:
<50 ppb, 50-500 ppb,
500-5,000 ppb,
>5,000 ppb;
(3) Cumulative
chemical exposure
subgroups: include
TCA, DCE, DCA, in
conjunction with TCE,
with the same exposure
Categories as in # 2;
(4) Duration of exposure
subgroups: <2 yrs,
2-5 yrs, 5-10 yrs.,
>10 yrs.; 2,867 had TCE
exposure of <50 ppb;
870 had TCE exposure
of 51-500 ppb; 190 had
TCE exposure of
501-5,000 ppb; 35 had
TCE exposure
>5,000 ppb.
Interviews
(occupational,
environmental,
demographic, and
health information); A
large number of health
outcomes analyzed,
including speech
impairment and hearing
impairment.
Logistic
Regression, Odds
Ratios; lowest
quartile used as
reference
population.
When the registrants were grouped by duration of
exposure to TCE, a statistically significant
association (adjusted for age and sex) between
duration of exposure and reported hearing
impairment was found. The prevalence odds ratios
were 2.32 (95% Cl: 1.18, 4.56) (>2 to <5 yrs); 1.17
(95% Cl: 0.55, 2.49) (>5 to <10 yrs); and 2.46
(95% Cl:1.30, 5.02) (>10 yrs); Higher rates of
speech impairment (not statistically significant)
associated with maximum and cumulative TCE
exposure, and duration of exposure.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
Buxton and
Hayward,
1967
This was a case study
on 4 workers exposed
to very high
concentrations of TCE,
which resulted from an
industrial accident. No
controls were
evaluated.
Case 1 was a 44-yr old
man exposed for
10 min; Case 2 was a
39-yr old man exposed
for 30 min; Case 3 was
a 43-yr old man exposed
for 2.5 h; Case 4 was a
39-yr old man exposed
for4h. TCE
concentrations were not
reported.
Clinical evaluations
were conducted by a
physician when patients
presented with
symptoms; numbness of
face, ocular pain,
enlarged right blind
spot, nausea, loss of
taste, headache,
dizziness, unsteadiness,
facial diplesia, loss of
gag and swallowing
reflex, absence of
corneal reflex, and
reduction of trigeminal
response.
There was no
statistical
assessment of
results presented.
Case 1 exhibited headaches and nausea for 48 h,
but had a full recovery. Case 2 exhibited nausea
and numbness of face, but had a full recovery.
Case 3 was seen and treated at a hospital with
numbness efface, insensitivity to pin prick over
the trigeminal distribution, ocular pain, enlarged
right blind spot, nausea, and loss of taste. No loss
of mental faculty was observed. Case 4 was seen
and treated for headache, nausea, dizziness,
unsteadiness, facial diplesia, loss of gag and
swallowing reflex, facial analgesia, absence of
corneal reflex, and reduction of trigeminal
response. The patient died and was examined
postmortem. There was demyelination of the 5th
cranial nerve evident.
^§
H I
O ^
HH Oq
H TO
Chalupa et
al., 1960
This was a case study
conducted on 22
patients with acute
poisoning caused by
carbon monoxide and
industrial solvents.
Six subjects were
exposed to TCE (doses
not known). Average
age 38.
No exposure data were
reported.
Medical and
psychological exams
were given to all
subjects. These
included EEGs,
measuring middle
voltage theta activity of
5-6 sec duration.
Subjects were tested for
memory disturbances.
No statistics were
performed.
80% of those with pathological EEG displayed
memory loss; 30% of those with normal EEGs
displayed memory loss. Pathology and memory
loss were most pronounced in subjects exposed to
carbon monoxide.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
El Ghawabi
etal., 1973
Feldman et
al, 1988
Study population
30 money printing
shop workers
occupationally
exposed to TCE;
Controls: 20 age and
SES matched
nonexposed males and
10 control workers not
exposed to TCE but
exposed to inks used in
printing.
21 Massachusetts
residents with alleged
chronic exposure to
TCE in drinking water;
27 laboratory controls.
Exposure assessment
and biomarkers
Air samples on
30 workers. Mean TCE
air concentrations
ranged from 41 to
163 ppm throughout the
Intalgio process
Colorimetric
determination of both
TCA and total trichloro-
compounds in urine
with Fujiware reaction.
TCE in residential well
water was 30-80 times
greater than U.S. EPA
MCL; maximum
reported concentration
was 267 ppb; other
solvents also present.
Tests used
Inquiries about
occupational, past and
present medical
histories, and family
histories in addition to
age and smoking habits.
EKGs were performed
on 25 of the workers.
Lab investigations
included complete
blood and urine
analysis, and routine
liver function tests.
BR used as an objective
indicator of neurotoxic
effects of TCE; clinical
neurological exam,
EMGs to evaluate blink
reflex, nerve conduction
studies, and extensive
neuropsychological
testing.
Statistics
Descriptive
statistics and
central tendency
evaluation for
metabolites; no
stats reported for
neurological
symptoms.
Student' st-test
used for testing the
difference between
the group means for
the Blink reflex
component
latencies.
Results
Most frequent symptoms: prenarcotic headache
(86% vs. 30% for controls), dizziness (67% vs.
6.7% for controls), and sleepiness (53% vs. 6% for
controls) main presenting symptoms in addition to
suppression of libido. Trigeminal nerve
involvement was not detected. The concentration
of total trichloro-compounds increased toward
mid-week and was stationary during the last 2
working days. Metabolites of total trichloroacetic
acid and trichloroethanol are only proportional to
TCE concentrations up to 100 ppm.
Highly significant differences in the conduction
latency means of the BR components for the TCE
exposed population vs. control population, when
comparing means for the right and left side Rl to
the controls (p < 0.001). The mean Rl BR
component latency for the exposed group was
11.35 ms, SD = 0.74 ms, 95% Cl: 11.03-11.66.
The mean for the controls was 10.21 ms, SD =
0.78 ms, 95% Cl: 9.92-10.51; /?< 0.001. Suggests
a subclinical alteration of the trigeminal nerve
function due to chronic, environmental exposure to
TCE.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
j*"
vo £;•
s^
TO
•§
1
S3'
a
^
^
C^
^
TO
TO'
^
1e
^3
M S
UJ ^
ON 0
*5
^
Reference
Feldman et
al., 1992
Study population
18 workers
occupationally
exposed to TCE; 30
laboratory controls.
Exposure assessment
and biomarkers
Reviewed exposure
histories of each worker
(job type, length of
work) and audited
medical records to
categorize into three
exposure categories:
"extensive,"
"occasional," and
"chemical other than
TCE".
Tests used
Blink reflexes using
TECA 4 EMG.
Statistics
Non-Gaussian
distribution and
high coefficient of
variance data were
log-transformed
and then compared
to the log-
transformed control
mean values. MRV
was calculated by
subtracting the
subjects value (x)
from the control
group mean (M),
and the difference
is divided by the
control group
standard deviation.
Results
The "extensive" group revealed latencies greater
than 3 SD above the nonexposed group mean on
Rl component of blink reflex; none of the
"occasional" group exhibited such latencies,
however, two of them demonstrated evidence of
demyelinating neuropathy on conduction velocity
studies; the sensitivity, or the ability of a positive
blink reflex test to correctly identify those who had
TCE exposure, was 50%. However, the specificity
was 90%, which means that of those workers with
no exposure to TCE, 90% demonstrated a normal
Kl latency. Subclinical alteration of the Vth
cranial nerve due to chronic occupational exposure
to TCE is suggested.
PI
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
§
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Gash et al.,
2007
Study population
30 Parkinson's Disease
patients and
27 non-Parkinson
coworkers exposed to
TCE; No unexposed
controls.
Exposure assessment
and biomarkers
Mapping of work areas.
Tests used
General physical exam,
neurological exam and
UPDRS, timed motor
tests, and occupational
history survey;
mitochondrial
neurotoxicity;
Questionnaire mailed to
134 former
non-Parkinson' s
workers,
(14 symptomatic of
parkinsonism: age
range = 31-66, duration
of employment range:
ll-35yrs)
(13 asymptomatic: age
range = 46-63, duration
of employment range:
8-33 yrs);.
Statistics
Workers' raw
scores given;
ANOVA
comparing
symptomatic vs.
nonsymptomatic
workers.
Results
Symptomatic non-Parkinson's group was
significantly slower in fine motor hand movements
than age-matched nonsymptomatic group
(p < 0.001); All symptomatic workers had positive
responses to 1 or more questions on UPDRS Part I
and Part 11, and/or had signs of parkinsonism on
Part III; One symptomatic worker had been tested
1 yr. prior and his UPDRS score had progressed
from 9 to 23.
H I
O
Oq
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
00 O
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
^§
H I
O >
HH Oq
H TO
Grandjean et
al,1955
80 workers employed
in 10 different
factories of the Swiss
mechanical
engineering industry
exposed to TCE, seven
of whom stopped
working with TCE
from 3 wks to 6 yrs
prior; no unexposed
control group.
Vapors were collected
in ethylic alcohol 95%.
Volume of air was
checked using a
flowmeter, and
quantitatively measured
according to the method
ofTruhaut(1951),
which is based on a
colored reaction
between TCE and the
pyridine in an alkaline
medium (with
modifications). Urine
analysis of TCA levels;
TCE air concentrations
varied from
6-1,120 ppm depending
on time of day and
proximity to tanks, but
mainly averaged
between 20-40 ppm.
Urinalysis varied from
30 mg/L to 300 mg/L;
Could not establish a
relationship between
TCE eliminated through
urine and TCE air
levels. Four exposure
groups estimated based
on air sampling data.
Medical exam,
including histories;
Blood and biochem.
tests, and psychiatric
exam. Psychological
exam; Meggendorf,
Bourdon, Rorschach,
Jung, Knoepfel's
"thirteen mistakes" test,
and Bleuler's test.
Coefficient of
determination,
Regression
coefficient.
Men working all day with TCE showed on average
larger amounts of TCA than those who worked
part time with TCE. Relatively high frequency of
subjective complaints, of alterations of the
vegetative nervous system, and of neurological
and psychiatric symptoms. 34% had slight or
moderate psycho-organic syndrome; 28% had
neurological changes; There is a relationship
between the frequency of those alterations and the
degree of exposure to TCE. There were
significant differences (p = 0.05) in the incidence
of neurological disorder between Groups I and III,
while between Groups II and III there were
significant differences (p = 0.05) in vegetative and
neurological disorders. Based on TCA eliminated
in the urine, results show that subjective,
vegetative, and neurological disorders were more
frequent in Groups II and III than in Group I.
Statistical analysis revealed the following
significant differences (p < 0.01): subjective
disorders between I and II; vegetative disorders
between I and II and between I and III;
neurological disorders between I and (II and III).
Vegetative, neurological, and psychological
symptoms increased with the length of exposure to
TCE. The following definite differences were
shown by statistical analysis (p < 0.03) : vegetative
disorders between I and IV ; neurological
disorders between I and II and between I and IV;
psychological disorders between I and III and
between I and IV.
O
H
W
-------
to
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
o ^
S.
R
s
3
1
£3'
1
<•£
C^
^
TO
TO'
*
S.
>9
Reference
Gun, el al.,
1978
Study population
8 exposed: 4 female
workers from one
plant exposed to TCE
and 4 female workers
from another plant
exposed to TCE +
nonhalogenated
hydrocarbon solvent
used in degreasing;
control group (n = 8)
consisted of 4 female
workers from each
plant who did not work
near TCE.
Exposure assessment
and biomarkers
Air sampled
continuously over
separate 10 min
durations drawn into a
Davis Halide Meter.
Readings taken every 30
sec.; ranged from 3-419
ppm.
Tests used
Eight-Choice reaction
times carried out in four
sessions; 40 reaction
time trials completed.
Statistics
Variations in RT by
level of exposure;
ambient air
exposure TCE
concentrations and
mean air TCE
values.
Results
TCE only group had consistently high mean
ambient air TCE levels (which exceeded the 1978
TLV of 100 ppm) and showed a mean increase in
reaction time, with a probable cumulative effect.
In TCE + solvent group, ambient TCE was lower
(did not exceed 100 ppm) and mean reaction time
shortened in Session 2, then rose subsequently to
be greater than at the start. Both control groups
showed a shortening in mean choice reaction time
in Session 2 which was sustained in Sessions 3 and
4 consistent with a practice effect; No stats
provided.
\o o
a
s
0^
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Hirschetal.,
1996
Study population
106 residents of
Roscoe, a community
in Illinois on the Rock
River, in direct
proximity to an
industrial plant that
released an unknown
amount of TCE into
the River. All
involved in litigation.
Case series report; No
unexposed controls.
Exposure assessment
and biomarkers
Random testing of the
wells between 1983-84
revealed groundwater in
wells to have levels of
TCE between 0 to 2,441
ppb; distance of
residence from well
used to estimate
exposure level.
Tests used
Medical, neurologic,
and psychiatric exams
and histories. For those
who complained of
headaches, a detailed
headache history was
taken, and an extensive
exam of nerve-threshold
measurements of toes,
fingers, face, olfactory
threshold tests for
phenylethyl methylethyl
carbinol, brain map,
Fast Fourier Transform
(FFT), P300 Cognitive
auditory evoked
response, EEG, Visual
Evoked Response
(VER), Somato sensory
Evoked Potential
(SSER), Brainstem
Auditory Evoked
Response (BAER),
MMPI-II, MCMI-II,
and Beck Depression
Inventory were also
given.
Statistics
Student t-test, Chi
square analysis,
nonparametric t-test
and ANOVA,
correlating all
history, physical
exam findings, test
data, TCE levels in
wells, and distance
from plant.
Results
66 subjects (62%) complained of headaches,
Diagnosis of TCE-induced cephalagia was
considered credible for 57 patients (54%).
Retrospective TCE level of well water or well's
distance from the industrial site analysis did not
correlate with the occurrence of possibly -TCE
induced headaches. Studies that were not
statistically significant with regard to possible
TCE-cephalalgia included P300, FFT, VER,
BAER, MMPI, MCMI, Beck Depression
Inventory, SSER, and nerve threshold
measurements. Headache might be associated
with exposure to TCE at lower levels than
previously reported. Headaches mainly occurred
without sex predominance, gradual onset,
bifrontal, throbbing, without associated features;
No quantitative data presented to support
statement of headache in relation to TCE exposure
levels, except for incidences of headache reporting
and measured TCE levels in wells.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
o
vo
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Kilburn and
Thornton,
1996
Study population
Group A: Randomly
selected registered
voters from Arizona
and Louisiana with no
exposure to TCE:
n = 264 unexposed
volunteers aged
18-83: Group B
volunteers from
California n = 29 17
males and 12 females
to validate the
equations; Group C
exposed to TCE and
other chemicals
residentially for 5 yrs
or more n = 237.
Exposure assessment
and biomarkers
No exposure or
groundwater analyses
reported.
Tests used
Reaction speed using a
timed computer visual-
stimulus generator;
Compared groups to
plotted measured SRT
and CRT Questionnaire
to eliminate those
exposed to possibly
confounding chemicals.
Statistics
Box-Cox
transformation for
dependent and
independent
variables.
Evaluated graphical
methods to study
residual plots.
Cooks distance
statistic measured
influence of outliers
examined. Lack-
of-fit test
performed on Final
model and
F statistic to
compare estimated
error to lack-of-fit
component of the
model's residual
sum of squared
error. Final models
were validated
using Group B data
and paired t-test to
compare observed
values for SRT and
CRT. F statistic to
test hypothesis that
parameter estimates
obtained with
Group B were
equal to those of
the model.
Results
Group A: SRT = 282 ms CRT = 532 ms
Group B: SRT = 269 ms CRT = 53 1 ms
Group C: SRT = 334 ms CRT = 619 ms
Lg(SRT) = 5.620, SD = 0.198
Regression equation for Lg(CRT) = 6.094389 +
0.0037964 x age. TCE exposure produced a step
increase in SRT and CRT, but no divergent lines.
Coefficients from Group A were valid for Group
B. Predicted value for SRT and for CRT, plus 1.5
SDs. selected 8% of the model group as abnormal.
'
I
I
§
Co
1
TO'
•3
, o
§
H I
O
Oq
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to o
•••2
i
^§
H I
O >
HH Oq
H TO
O
H
W
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Kilburn and
Warshaw,
1993
Kilburn,
2002b
Study population
Well-water exposed
subjects to 6 to 500
ppbofTCEforlto
25 yrs; 544 recruited
test subjects; Group 1
= 196 exposed family
members of subjects
with cancer or birth
defects; Group 2 = 178
from exposed families
without cancer or birth
defects; Group 3
= 170 exposed parents
whose children had
birth defects and
rheumatic disorders;
Controls: 68 referents
and 113 histology
technicians (HTs)
without environmental
exposure to TCE.
236 residents
chronically exposed to
TCE and associated
solvents, including
DCE,PCE, and vinyl
chloride, in the
environment from a
Exposure assessment
and biomarkers
Well-water was
measured from 1957 to
1981 by several
governmental agencies,
and average annual TCE
exposures were
calculated and then
multiplied by each
individual's years of
residence for
170 subjects.
Exposure estimate based
on groundwater plume
based on contour
mapping; concentrations
between 0.2-10,000 ppb
of TCE over a 64 km2
area; additional
Tests used
Neurobehavioral testing
- augmented NET; Eye
Closure and Blink using
EMG;
neuropsychological
(NFS) test - Portions of
Wechsler's Memory
Scale, and WA15 and
embedded figures test,
grooved pegboard, Trail
Making A and B,
POMS, and Culture Fair
Test;
neurophysiological
(NPH) testing - Simple
visual reaction time,
body balance apparatus,
cerebellar function,
proprioception, visual,
associative links and
motor effector function.
Simple reaction time,
choice reaction time,
Balance sway speed
(with eyes open and
eyes closed), color
errors, blink reflex
latency, Supra orbital
Statistics
Two sided student
t-test with a
/?<0.05.
Linear regression
coefficients to test
how demographic
variables or other
factors may
contribute.
Descriptive
statistics;
ANCOVA; step-
wise adjustment of
demographics.
Results
Exposed subjects had lower intelligence scores and
more mood disorders.
NPH: Significant impairments in sway speed with
eyes open and closed, blink reflex latency (R-l),
eye closure speed, and two choice visual reaction
time.
NPS: Significant impairments in Culture Fair
(intelligence) scores, recall of stories, visual recall,
digit span, block design, recognition of fingertip
numbers, grooved pegboard, and Trail Making A
andB.
POMS: all subtests, but the fatigue, were elevated
Mean speeds of sway were greater with eyes open
at/? < 0.0001) and with eyes closed/? < 0.05) in
the exposed group compared to the combined
referents. The exposed group mean simple
reaction time was 67 msec longer than the referent
group/? < 0.0001). Choice reaction time (CRT) of
the exposed subjects was 93 msec longer in the
third trial (p < 0.0001) than referents. It was also
longer in all trials, and remained significantly
different after age adjustment. Eye closure latency
was slower for both eyes in the exposed and
significantly different (p< 0.0014) on the right
compared to the HT referent group.
The principal comparison, that was between the
236 exposed persons and the 161 unexposed
regional controls, revealed 13 significant
differences (p < 0.05). SRTs and CRTs were
delayed. Balance was abnormal with excessive
sway speed (eyes closed), but this was not true
when both eyes were open. Color discrimination
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
•3
, o
i
^§
H I
O >
HH Oq
H TO
O
H
W
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Kilburn,
2002b
(continued)
Study population
nearby microchip
plant, some involved
in litigation, prior to
1983 and those who
lived in the area
between 1983 and
1993 during which
time dumping of
chlorinated solvents
had supposedly ceased
and clean-up activities
had been enacted;
Controls: 67 referents
from northeast
Phoenix, who had
never resided near the
2 plants (mean
distance = 2,000 m,
range
= 1,400-3 ,600m from
plants) and
161 regional referents
from Wickenburg, AZ
up-wind of Phoenix,
recruited via random
calls made to numbers
on voter registration
rolls, matched to
exposed subjects by
age and years of
education, records
showed no current or
past water
contamination in the
areas.
Exposure assessment
and biomarkers
associated solvents,
including DCE, PCE,
and vinyl chloride, No
air sampling.
Tests used
tap (left and right),
Culture Fair A,
Vocabulary, Pegboard,
Trail Making A and B,
Immediate verbal recall,
POMS; Pulmonary
Function;
The same examiners
who were blinded to the
subjects' exposure
status examined the
Phoenix group, but the
Wickenburg referents'
status was known to the
examiners. Exact order
or timing of testing not
stated.
Statistics
Results
errors were increased. Both right and left blink
reflex latencies (R-l) were prolonged. Scores on
Culture Fair 2A, vocabulary, grooved pegboard
(dominant hand), trail making A and B, and verbal
recall (i.e., memory) were decreased in the
exposed subjects.
Litigation is suggested but not stated and study
paid by lawyers.
Litigation status may introduce a bias, particularly
if no validity tests were used.
-------
to
O k^j
o J5"
vo £;•
a
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
TO'
|
&•
Kilburn,
2002b
236 residents exposed
environmentally from
a nearby microchip
plant (exact number of
litigants not stated);
156 individuals
exposed for > 10 y
compared to 80
individuals <10 y of
exposure; Controls: 58
nonclaimants in 3
areas within exposure
zone (Zones A, B,
andC).
No discussion of
exposure assessment
methods and results.
Solvents included TCE,
DCE, PCE, and vinyl
chloride; concluded
exposure is primarily
due to groundwater
plume rather than air
releases.
Simple reaction time,
choice reaction time,
Balance sway speed
(with eyes open and
eyes closed), color
errors, blink reflex
latency, Supra orbital
tap (left and right),
Culture Fair A,
Vocabulary, Pegboard,
Trail Making A and B,
Immediate verbal recall,
POMS.
Descriptive
statistics,
Regression
analysis; Similar
study to the one
reported above with
the exception of
looking at the
effects of duration
of residence,
proximity to the
microchip plant,
and being involved
in litigation.
Insignificant effects of longer duration of
residence. No effect of proximity and litigation.
Effects of longer duration of residence modest and
insignificant. No effect of proximity. No
litigation effect. Zone A-100 clients were not
different from the 9 nonclients.
Zone B, nonclients were more abnormal in color
different than clients and right-sided blink was less
abnormal in nonclients.
Zone C, 9 of the 13 measurements were not
significantly different.
26 of the original 236 subjects re-tested in 1999:
maintained impaired levels of functioning and
mood; No tests of effort and malingering used,
limiting interpretations.
Again, no tests of effort and malingering were
used, thus, limiting interpretation.
Litigation is suggested but not stated and study
paid by lawyers.
Litigation status may introduce a bias, particularly
if no validity tests were used.
o §
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
•3
, o
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Landrigan et
al., 1987
Study population
13 Pennsylvania
residents exposed
through drinking and
bathing water
contaminated by
approximately 1,900
gallon TCE spill; Feb
1980: 9 workers
exposed to TCE while
degreasing metal in
pipe manufacturing
plant and 9 unexposed
controls (mean ages
were 42.7 exposed and
46.4-y old unexposed;
mean durations of
employment = 4.4,
exposed, and 9.4 y,
unexposed.;
May 1980: 10 exposed
workers and same 9
unexposed worker
controls from Feb
monitoring.
Exposure assessment
and biomarkers
Community Evaluation:
Nov 1979-
Questionnaires on TCE
and other chemical
exposures, and
occurrence of signs and
symptoms of exposure
to TCE, morning urine
samples, urine samples
analyzed
coloreimetrically for
total trichloro-
compounds.
Occupational
Evaluations (in
workers): breathing-
zone air samples( mean
205 mg/m3; 37 ppm);
medical evaluations, pre
and post shift spot urine
samples in Feb and
again in May, mid and
post shift venous blood
samples during the May
survey,
Tests used
Community evaluation,
occupational
evaluations; urine
evaluations for TCE
metabolites;
Questionnaires to
evaluate neurologic
effects and symptoms;
ISO concentrations,
Map of TCE in
groundwater.
Statistics
Descriptive
statistics
Results
Community Evaluation: No urinary TCA detected
in community population except for 1 resident also
working at plant and 1 resident with no exposure;
Occupational Evaluation: Range 117-357 mg/m3-
(2 1-64 ppm).
Feb: airborne exposures exceeded NIOSH limit by
up to 222 mg/m3 (40 ppm)(NIOSH TWA <135
mg/m3).
(24 ppm). Short term exposure exceeded NIOSH
values of 535 mg/m3 (96 ppm) by up to 1,465
mg/m3 (264 ppm).
Personal breathing zone of other workers within
recommended limits (0.5-125 mg/m3) (0.1-23
ppm).
7 exposed workers reported acute symptoms,
including fatigue, light-headedness, sleepiness,
nausea, headache, consistent with TCE exposure;
No control workers reported such symptoms;
Prevalence of 1 or more symptoms 78% in
exposed worker group, 0% in control worker
group; Symptoms decreased after
recommendations were in place for 3 mos (may
testing) for reduced exposures.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o J5"
vo £;•
I
I
§
***.
&
1
TO'
^
, o
§
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Liuetal.,
1988
Study population
103 workers from
factories in Northern
China, exposed to TCE
(79 men, 24 women),
during vapor
degreasing production
or operation. The
unexposed control
group included 85 men
and 26 women.
Exposure assessment
and biomarkers
Exposed to TCE, mostly
at less than 50 ppm;
concentration of
breathing zone air
during entire shift
measured by diffusive
samplers placed on the
chest of each worker;
divided into three
exposure groups;
1-10 ppm, 11-50 ppm
and 51-100 ppm; Also,
hematology, serum
biochemistry, sugar,
protein, and occult
blood in urine were
collected.
Tests used
Serf-reported subjective
symptom questionnaire.
Statistics
Prevalence of
affirmative
answers = total
number of
affirmative answers
divided by (number
of respondents x
number of
questions); X2.
Results
Dose-response relationship established in
symptoms such as nausea, drunken feeling, light-
headedness, floating sensation, heavy feeling of
the head, forgetfulness, tremors and/or cramps in
extremities, body weight loss, changes in
perspiration pattern, joint pain, and dry mouth (all
>3 times more common in exposed workers);
"bloody strawberry jam-like feces" was borderline
significant in the exposed group and "frequent
flatus " was statistically significant. Exposure
ranged up to 100 ppm, however, most workers
were exposed below 10 ppm, and some at 1 1-50
ppm. Contrary to expectations, production plant
men had significantly higher levels of exposure
(24 had levels of 1-10 ppm, 15 had levels of
1 1-50 ppm, 4 had levels of 5 1-100 ppm) than
degreasing plant men (31 had levels of 1-10 ppm,
2 had levels of 1 1-50 ppm, 0 had levels of 5 1-100
ppm); p < 0.05 by chi-square test. No significant
difference (p > 0. 10) was found in women
workers. The differences in exposure intensity
between men and women was of borderline
significance (0.05
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
•3
, o
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
^§
H I
O >
HH Oq
H TO
McCunney,
1988
This is a case study
conducted on 3 young
white male workers
exposed to TCE in
degreasing operations.
There were no controls
included. Case 1 was
a 25-yr old male, Case
2 was a 28-yr old
white male, Case 3
was a 45-yr old white
male.
Case 1: TCE in air at the
work place was
measured at 25 ppm, but
his TCA in urine was
measured at 210 mg/L.
This is likely due to
dermal exposure while
cleaning metal rods in
TCE. Case 2: no TCE
exposure data presented,
TCA at 9 mg/L after
6 mos; Case 3: no TCE
exposure data presented.
Clinical evaluation of
loss of balance, light
headedness, resting
tremor, blurred vision,
and
dysdiadochokinesia,
change in demeanor and
loss of coordination,
cognitive changes were
noted, as well as
depression; CT scan,
EEG, nerve
conductivity, and visual
and somatosensory
evoked response.
Neurological exams
included sensitivity to
pinprick over the face;
Ophthalmic evaluation.
There were no
statistical analyses
of results presented.
Case 1 was a 25-yr old male, who presented with a
loss of balance, light headedness, resting tremor,
blurred vision, and dysdiadochokinesia. The
subject had been in a car accident and suffered
head injuries. He later returned with a change in
demeanor and loss of coordination. He showed a
normal CT scan, EEG, nerve conductivity, and
visual and somatosensory evoked response.
Neurological exams revealed reduced sensitivity to
pinprick over the face, deep tendon reflexes were
reduced, mild to moderate cognitive changes were
noted, as well as depression. Ophthalmic
evaluation was normal. He was removed form the
TCE exposure and appeared to recover.
Case 2 was a 28-yr old white male who presented
with numbness and shooting pains in fingers. He
exhibited anorexia, tiredness. He worked in a
degreasing operation for a jeweler using open
containers filled with TCE in a small, unventilated
room. There were no exposure data provided, but
his TCA was 9 mg/L at 6 mos after exposure. He
had been hospitalized with hepatitis previously.
No neurological tests were administered.
Case 3 was a 45-yr old white male who presented
with numbness in hands and an inability to sleep.
He exhibited slurred speech. He was positive for
blood in stool, but had a history of duodenal
ulcers.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
•3
, o
00 O
•••2
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Mhiri et al.,
2004
Mitchell and
Parsons-
Smith, 1969
Study population
23 phosphate industry
workers exposed to
TCEfor6h/dforat
least 2 yrs while
cleaning walls to be
painted; Controls:
23 unexposed workers
from the department of
neurology.
This was a case study
of 1 male patient, age
33, occupational
exposed to TCE during
degreasing. There
were no controls.
Exposure assessment
and biomarkers
Measurement of urinary
metabolites of TCE
were performed 3
times/worker. Blood
tests and hepatic
enzymes were also
collected.
No exposure data are
presented.
Tests used
Trigeminal
somatosensory evoked
potentials recorded
using Nihon-Kohden
EMG- evoked potential
system; baseline clinical
evaluations regarding
facial burn or
numbness, visual
disturbances,
restlessness,
concentration difficulty,
fatigue, mood changes,
assessment of cranial
nerves, quality of life;
biological tests
described under
biomarkers.
Trigeminal nerve, loss
of taste, X-rays of the
skull, EEC,
hemoglobin, and
Wassermann reaction.
Statistics
Paired or unpaired
Student's t-test as
appropriate.
p-value set at
O.05. Spearman
rank-correlation
procedure was used
for correlation
analysis.
No statistics
provided.
Results
Abnormal TSEP were observed in 6 workers with
clinical evidence of Trigeminal involvement and in
9 asymptomatic workers. A significant positive
correlation between duration of exposure and the
N2 latency (p < 0.01) and P2 latency (p < 0.02)
was observed. Only one subject had urinary TCE
metabolite levels over tolerated limits. TCE air
contents were over tolerated levels, ranging from
50-150 ppm.
The patient had complete analgesia in the right
trigeminal nerve and complete loss of taste, patient
complained of loss of sensation on right side of
face, and uncomfortable right eye, as well as
vertigo and depression. X-rays of the skull, EEG,
hemoglobin, and Wassermann reaction were all
normal.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
vo S3'
&
a
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
Nagaya et
al., 1990
84 male workers ages
18-61 (mean 36.2)
constantly using TCE
in their jobs. Duration
of employment (i.e.,
exposure)
0.1-34.0yrs, (mean
6.1yrs;SD = 5.9).
Controls:
83 age-matched office
workers and students
with no exposure.
Workers exposed to
about 22-ppm TCE in
air. Serum dopamine-p-
hydroxylase (DBH)
activity levels measured
from blood. Urinary
total trichloro-
compounds (U-TTC)
also measured.
Blood drawn during
working time and DBH
activities were
analyzed; Spot urine
collected at time of
blood sampling and
U-TTC determined by
alkaline-pyridine
method.
Student's t-test and
linear correlation
coefficient. Results
of U-TTC
presented by age
groups: <25;
26-40; >41.
A slight decrease in serum DBH activity with age
was noted in both groups. Significant inverse
correlation of DBH activity and age was found in
workers (r = -0.278, O.OK/? < 0.02), but not in
controls (r = -0.182, 0.05 TO
£ **
1
H
O,
Reifetal.,
2003
143 residents of the
Rocky Mountain
Arsenal community of
Denver whose water
was contaminated with
TCE and related
chemicals from nearby
hazardous waste sites
between 1981 and
1986; Referent group
at lowest concentration
(<5 ppb).
Hydraulic simulation
model used in
conjunction with a GIS
estimated residential
exposures to TCE;
Approximately 80% of
the sample exposed to
TCE exceeding MCL of
5 ppb and
approximately 14%
exceeded 15 ppb. High
exposure group
>15 ppb, low exposure
referent group <5 ppb,
medium exposure group
5
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Rasmussen
and Sabroe,
1986
Study population
368 metal workers
working in degreasing
at various factories in
Denmark (industries
not specified) with
chlorinated solvents;
94 controls randomly
selected semiskilled
metal workers from
same area; mean age:
37.7 (range: 17-65+).
Total 443 men; 19
women.
Exposure assessment
and biomarkers
Questionnaire:
categorized in 4 groups;
3 exposure groups plus
control: (1) currently
working with
chlorinated solvents
(n = 171; average.
duration: 7.3 yrs,
16.5 h/wk; 57% TCE
and 37%
1,1,1 -trichloroethane),
(2) currently working
with other solvents
(n = 131; petroleum,
gasoline, toluene,
xylene), (3) previously
(1-5 yrs) worked with
chlorinated or other
solvents (n = 66) (4)
never worked with
organic solvents
(n = 94).
Tests used
Questionnaire: 74 items
about
neuropsychological
symptoms (memory,
concentration,
irritability, alcohol
intolerance, sleep
disturbance, fatigue).
Statistics
Chi-square; Odds
ratios; t-test;
logistic regression.
Results
Neuropsychological symptoms significantly more
prevalent in the chlorinated solvents-exposed
group; TCE caused the most "inconveniences and
symptoms;" dose response between exposure to
chlorinated solvents and chronic
neuropsychological symptoms (memory
\p < 0.001], concentration \p < 0.02], irritability
\p < 0.004], alcohol intolerance \p < 0.004],
forgetfulness [p < 0.001], dizziness \p < 0.005],
and headache \p < 0.01]); Significant associations
between previous exposure and consumption of
alcohol with chronic neuropsychological
symptom, s
^§
"*4-
o§
H I
O
Oq
O
H
W
-------
5 Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
o
o S3'
sx
|
3
1
S3'
a
^
^
C^
^
TO
TO'
^
1
^3
-) S
yi £
~" §
^
§
^J ^^
eg-
> ^
H ^
^
2 §
? ^
5 !'
H ^
3 c^
H ^
•H *^J
TJ £*
0^3
^ **•'•
-H O
Reference
Rasmussen
etal., 1993a
Study population
96 Danish workers
involved in metal
degreasing with
chlorinated solvents,
mostly TCE (n = 70);
(industries not
specified), age range:
19-68; no external
controls.
Exposure assessment
and biomarkers
Chronic exposure to
TCE (n = 70); CFC
(n = 25); HC (n = I);
average duration: 7.1
yrs; range of full-time
degreasing: 1 mo to
36 yrs; occupational
history, blood and
urinary metabolites
(TCA); biological
monitoring for TCE and
TCE metabolites; CEI
calculated based on
number of h/wk worked
with solvents x yr of
exposure x 45 wk per
yr; 3 groups: (1) low
exposure: n = 19,
average full-time
exposure 0.5 yr; (2)
medium exposure:
n = 36, average full-time
exposure 2.1 yrs.; (3)
high exposure: n = 4l,
average full-time
exposure 1 1 yrs; Mean
TCA in high exposure
group = 7.7 mg/L (max
= 26.1mg/L);TWA
measurements of CFC
113 levels:
260-420 ppm (U.S. and
Danish TLV is 500
ppm).
Tests used
Medical interview,
neurological exam,
neuropsychological
exam; Tests: WAIS:
Vocabulary, Digit
Symbol; Simple
Reaction Time,
acoustic -motor
function, discriminatory
attention, Sentence
Repetition, Paced
Auditory Serial
Addition Test, Text
Repetition, Rey's
Auditory Verbal
Learning, visual gestalt,
Stone Pictures
(developed for this
study, nonvalidated),
revised Santa Ana,
Luria motor function,
Mira; Blind study.
Statistics
Fisher's exact test,
Chi-square trend
test, t-test,
ANOVA, logistic
regression, odds
ratios, Chi-square
goodness-of-fit test;
Confounders
examined: age,
primary intellectual
level,
arteriosclerosis,
neurological/psychi
atric disease,
alcohol abuse, and
present solvent
exposure.
Results
After adjusting for confounders, the high exposure
group has significantly increased risk for
psychoorganic syndrome following exposure (OR:
11.2); OR for medium exposed group = 5.6;
Significant increase in risk with age and with
decrease in WAIS Vocabulary scores; Prevalence
of psychoorganic syndrome: 10.5% in low
exposure group, 38.9 in medium exposure group,
63.4% in high exposure group; no significant
interaction between age and solvent exposure.
D
n
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
•••2
i
^§
H I
O >
HH Oq
H TO
O
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Rasmussen
et al., 1993b
Study population
96 Danish workers
involved in metal
degreasing with
chlorinated solvents
(industries not
specified), age range:
19-68; No external
controls.
Exposure assessment
and biomarkers
Chronic exposure to
TCE (n = 70); CFC
(n = 25); HC (n = 1);
average duration: 7.1
yrs); range of full-time
degreasing: 1 mo to 36
yrs; occupational
history, blood and
urinary metabolites
(TCA); biological
monitoring for TCE and
TCE metabolites; CEl
calculated based on
number of h/wk worked
with solvents x yr of
exposure x 45 wks per
yr; 3 groups: (1) low
exposure: n = 19,
average full-time expo
0.5 yr; (2) medium
exposure: n = 36,
average full-time
exposure 2.1 yrs; (3)
high exposure: n = 41,
average full-time
exposure 1 1 yrs; Mean
TCA in high exposure
group = 7.7 mg/L (max
= 26.1mg/L);TWA
measurements of CFC
113 levels: 260-420
ppm (U.S. and Danish
TLV is 500 ppm).
Tests used
WAIS (original
version): Vocabulary,
Digit Symbol, Digit
Span; Simple Reaction
Time, Acoustic-motor
function (Luria),
Discriminatory
attention (Luria),
Sentence Repetition,
PASAT, Text
Repetition, Rey's
Auditory Verbal
Learning, Visual
Gestalts, Stone Pictures
(developed for this
study, nonvalidated),
revised Santa Ana,
Luria motor function,
Mira; Blind study.
Statistics
Linear regression
analysis;
Confounding
variables analyzed:
age, primary
intellectual
function, word
blindness,
education,
arteriosclerosis,
neurological/psychi
atric disease,
alcohol use, present
solvent exposure.
Results
Dose response with 9 of 15 tests; Controlling for
confounds, significant relationship of exposure
was found with Acoustic-motor function
(p < 0.001), PASAT (p < 0.001), Rey AVLT
(p < 0.001), vocabulary (p < 0.001), and visual
gestalts (p < 0.001); significant age effects.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Rasmussen
etal., 1993c
Study population
96 Danish workers
involved in metal
degreasing with
chlorinated solvents
(industries not
specified), age range:
19-68; No external
controls.
Exposure assessment
and biomarkers
Chronic exposure to
TCE (n = 70); CFC
(n = 25); HC (n = 1);
average duration: 7.1
yrs); range of full-time
degreasing: 1 mo to 36
yrs; occupational
history, blood and
urinary metabolites;
biological monitoring
for TCE and TCE
metabolites; CEl
calculated based on
number of h/wk worked
with solvents x yr of
exposure x 45 wk per
yr; 3 groups: (1) low
exposure: n = 19,
average full-time expo
0.5 yr; (2) medium
exposure: n = 36,
average full-time
exposure 2.1 yrs; (3)
high exposure: n = 41,
average full-time
exposure 1 1 yrs; Mean
TCA in high exposure
group = 7.7 mg/L (max
= 26.1mg/L);TWA
measurements of CFC
113 levels:
260-420 ppm (U.S. and
Danish TLV is 500
ppm).
Tests used
Medical interview,
clinical neurological
exam,
neuropsychological
exam.
Statistics
Multiple
regression; Fisher's
exact test; Mantel-
Haenzel test for
linear association.
Results
Significant dose response between exposure and
motor dyscoordination remained after controlling
for confounders; Bivariate analysis showed
increased vibration threshold with increased
exposure, but with multivariate analysis, age was a
significant factor for the increase.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
vo S3'
a
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
TO'
i
o
H
W
Ruijten et
al., 1991
31 male printing
workers exposed to
TCE. Mean age 44;
Mean duration 16 yrs;
Controls: 28; mean age
45 yrs.
Relied on exposure data
from past monitoring
activities conducted by
plant personnel using
gas detection tubes.
Estimated 17 ppm for
past 3 yrs, 35 ppm for
preceding 8 yrs and
70 ppm before that.
Individual cumulative
exposure was calculated
as time spent in
different exposure
periods and the
estimated exposure in
those periods. Mean
cumulative exposure
= 704 ppm x yrs (SD
583, range:
160-2,150 ppm x yrs.
General questionnaire,
cardiotachogram
recorded on ink writer
to measure Autonomic
nerve function,
including forced
respiratory sinus
arrhythmia (FRSA),
muscle heart reflex
(MHR), resting
arrhythmia; Trigeminal
nerve function
measured using
masseter reflex and
blink reflex;
electrophysiological
testing of peripheral
nerve functioning using
motor nerve conduction
velocity of the peroneal
nerve.
Combined Z score
= individual Z
scores of the FRSA
and MHR;
ANCOVA to
calculate difference
between
exposed/nonexpose
d workers;
Cumulative
exposure effect
calculated by
multiple linear
regression analysis.
Controlled for age,
alcohol
consumption, and
nationality by
including them as
covariables.
Quetelet-index
included for
autonomic nerve
parameters; Body
length and skin
temperature used
for all peripheral
nerve functions;
one-sided
significance level
of 5% used. Non-
normal
distributions were
log or square root
transformed.
Slight reduction in Sural nerve conduction velocity
was found and a prolongation of the Sural
refractory period. Latency of the masseter reflex
had increased. No prolongation of the blink reflex
was found; no impairment of autonomic or motor
nerve function were found. Long term exposure to
TCE at threshold limit values (approximately
35 ppm) may slightly affect the trigeminal and
sural nerves.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Smith, 1970
Triebig et
al., 1977b
Study population
130 (108 males,
22 females); Controls:
63 unexposed men
working at the same
factory matched by
age, marital status.
This study was
conducted on
8 subjects
occupationally
exposed to TCE.
Subjects were 7 men
and 1 woman with an
age range from 23-38
yrs. There was no
control group.
Exposure assessment
and biomarkers
TCA metabolite levels
in urine were measured:
60.8% had levels up to
20 mg/L, and 82.1% had
levels up to 60 mg/L.
Measured TCE in air
averaged 50 ppm
(260mg/m3). Length of
occupational exposure
was not reported.
Tests used
Cornell Medical Index
Questionnaire
(Psychiatric section),
Heron's Personality
Questionnaire, Fluency
Test, 13 -Mistake Test,
Serial Sevens, Digit
Span, General
Knowledge Test, tests
of memory.
Results were compared
after exposure periods,
and compared to results
obtained after periods
removed from
exposure. TCA and
TCE metabolites in
urine and blood were
measured.
Psychological tests
included d2, MWT-A,
and short test.
Statistics
Descriptive
Statistics.
Wilcoxon and
Willcox
nonparametric tests.
Due to the small
sample size a
significance level
of 1% was used.
Results
Of the 130 subjects exposed 27% had no
complaints of symptoms, 74.5% experienced
fatigue, 56.2% dizziness, 17.7% headache, 25.4%
gastro-intestinal problems, 7.7% autonomic
effects, and 24.9% had other symptoms. The
number of complaints reported by subjects were
statistically significant between those with 20
mg/L or less TCA (M = 1.8 complaints) and those
60 mg/L or more (M = 2.7). Each group, however,
had a similar proportion of subjects who reported
having only 'slight' symptoms. The total time of
continuous exposure to TCE (ranging from less
than 1 yr to more than 10 yrs) appeared to have
little influence on frequency of symptoms. No
results of the tests are reported; Author postulates
that symptom assessment raises the possibility of
"errors of subjective judgment."
Mean values observed were 330-mg
trichloroethanol and 3 19-mg TCA/g creatinine,
respectively, at the end of a work shift. The
psychological tests showed no statistically
significant difference in the results before or after
the exposure-free time period.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
ON O
•••2
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Triebig,
1982
Study population
This study was
conducted on
24 healthy workers (20
males, 4 females)
exposed to TCE
occupationally at three
different plants. The
ages 17-56; length of
exposure ranged from
1 to 258 mos (mean
83 mos). A control
group of 144 controls
used to establish
'normal' responses on
the nerve conduction
studies. The matched
control group
consisted of 24 healthy
nonexposed
individuals (20 males,
4 females), chosen to
match the subjects for
age and sex.
Exposure assessment
and biomarkers
Length of exposure
ranged from 1 to
258 mos (mean 83 mos).
TCE concentrations
measured in air at work
places ranged from
5-70 ppm. TCA, TCE,
and trichloroethanol
were measured in blood,
and TCE and TCA were
measured in urine.
Tests used
Nerve conduction
velocities were
measured for sensory
and motor nerve fibers
using the following
tests: MCVMAx(U):
Maximum NLG of the
motor fibers of the N.
ulnaris between the
wrist joint and the
elbow; dSCV (U),
pSCV (U), and dSCV
(M).
Statistics
Data were analyzed
using parametric
and nonparametric
tests, rank
correlation, linear
regression, with 5%
error probability.
Results
Results show no statistically significant difference
in nerve conduction velocities between the
exposed and unexposed groups. This study has
measured exposure data, but exposures/responses
are not reported by dose levels.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Triebig,
1983
Study population
The exposed group
consists of 66 healthy
workers selected from
a population of
112 workers. Workers
were excluded based
on polyneuropathy
(n = 46) and alcohol
consumption (n = 28).
The control group
consisted of 66 healthy
workers with no
exposures to solvents.
Exposure assessment
and biomarkers
Subjects were exposed
to a mixture of solvents,
including TCE,
specifically "ethanol,
ethyl acetate, aliphatic
hydrocarbons
(gasoline), MEK,
toluene, and
trichloroethene."
Subjects were divided
into 3 exposure groups
based on length of
exposure, as follows: 20
employees with "short-
term exposure" (7-24
mos); 24 employees
with "medium-term
exposure" (25-60 mos);
22 employees with
"long-term exposure"
(over 60 mos). TCA,
TCE, and
trichloroethanol were
measured in blood, and
TCE and TCA were
measured in urine.
Tests used
Nerve conduction
velocities were
measured for sensory
and motor nerve fibers
using the following
tests: MCVMAx(U):
Maximum NLG of the
motor fibers of the N.
ulnaris between the
wrist joint and the
elbow; dSCV (U),
pSCV (U), and dSCV
(M).
Statistics
Data were analyzed
using parametric
and nonparametric
tests, rank
correlation, linear
regression, with 5%
error probability.
Results
There was a dose-response relationship observed
between length of exposure to mixed solvents and
statistically significant reduction in nerve
conduction velocities observed for the medium and
long-term exposure groups for the NCV.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
00 O
•••2
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Troster and
Ruff, 1990
Study population
3 occupationally
exposed workers to
TCE or TCA:
2 patients acutely
exposed to low levels
of TCE and 1 patient
exposed to TCA;
Controls: 2 groups of
n = 30 matched
controls; (all age and
education matched).
Exposure assessment
and biomarkers
"Unknown amount of
TCE for 8 months."
Tests used
SDNTB, "1 or more
of:" TAT, MMPl,
Rorschach, and
Interviewing
questionnaire, Medical
examinations (including
neurological, CT scan,
and/or Chemo-
pathological tests and
occupational history).
Statistics
Not reported.
Results
Case 1 : Intelligence "deemed" to drop from
premorbid function at 1 y 10 mos after exposure.
Impaired functions improved for all but reading
comprehension, visuospatial learning and
categorization (abstraction). Case 2: Mild deficits
in motor speed, verbal learning, and memory;
"marked" deficits in visuospatial learning; good
attention; diagnosis of mild depression and
adjustment disorder, but symptoms subsided after
removal from exposure. Case 3 : Manual dexterity
and logical thinking borderline impaired; no
emotional changes, cognitive function spared,
diagnosis of somatoform disorder.
i
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
2
•••2
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
White etal.,
1997
Study population
Group 1:
28 individuals in
Massachusetts exposed
to contaminated well
water; source: tanning
factory and chemical
plant; age range: 9-55.
Group 2:
12 individuals in Ohio
exposed to
contaminated well
water; source:
degreasing; age range:
12-68 Group 3:
20 individuals in
Minnesota exposed to
contaminated well
water; n = 14 for nerve
conduction studies and
n = 6 for
neuropsychological
testing; source:
ammunition plant; age
range: 8-62. No
controls.
Exposure assessment
and biomarkers
Group 1 : 2 wells tested
in 1979: 267 ppb TCE,
21ppb
Tetrachloroethylene,
12 ppb chloroform,
29 ppb dichloro-
ethylene, 23 ppb
Trichlorotrifluoroethane
; 2 yrs average TCE
256 ppb for well G, and
111 ppb for well H.
Group 2:13 wells with
1,1,1 -trichloroethane
(up to 2,569 ppb) and
TCE (up to 760 ppb);
blood analysis of
individuals 2 yrs after
end of exposure and
soon after exposure
showed normal or mild
elevations of TCE,
elevations of
1,1,1 -trichloroethane,
ethylbenzene, and
xylenes. Group 3 : mean
TCE for one well
261 ppb;
1, 1-dichloroethylene
9.0 ppb;
1 ,2-dichloroethylene
107 ppb.
Tests used
Occupational and
environmental
questionnaire,
neurological exam,
neuropsychological
exam: WAIS-R,
WISC-R, WMS,
WMS-R, Wisconsin
Card Sorting, COWAT,
Boston Naming, Boston
Visuospatial
Quantitative Battery,
Milner Facial
Recognition Test,
Sticks Visuospatial
Orientation Task, Word
triads, Benton Visual
Retention Test, Santa
Ana, Albert's Famous
Faces, Peabody Picture
Vocabulary Test,
WRAT, POMS, MMPI,
Trail-making,
Fingertapping, Delayed
Recognition Span Test;
Neurophysiological
exam: eyeblink, evoked
potentials, nerve
conduction; Other:
EKG, EEC, medical
tests.
Statistics
Data shown in
proportion in 3
communities,
clinical diagnostic
categories, analysis
of central
tendencies, and
descriptive
statistics.
Results
Group 1 : Some individuals with subclinical
peripheral neuropathy; 92.8% with reflex
abnormalities; 75% total diagnosed with peripheral
neuropathy; 88.9% with impairment in at least 1
memory test; Impairments: attention and executive
function in 67.9%; motor function in 60.71%,
Visuospatial in 60.71%, mild to moderate
encephalopathy in 85.7%.
Group 2: 25% with abnormal nerve conduction;
Impairments: attention and executive function in
83.33%, memory in 58.33%, language/verbal in
50%.
Group 3: 35.7% with peripheral neuropathy;
neuropsychological: all 6 tested had memory
impairment, attention and executive function
impairment, 3 had manual motor slowing.
Participants younger at time of exposure with
wider range of deficits; Language deficits in
younger, but not in older participants.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Winneke,
1982
Study population
This is a review article
presenting multiple
studies that evaluated
neurological effects of
TCE, and other
solvents. Only the
TCE results are
summarized herein.
Experiment 1: 18
subjects (results taken
from Schlipkoter et al.,
1974 and summary is
based on informations
from Winneke, 1982)
Experiment 2: 12
subjects (results taken
from Winneke et al.,
1974, 1976, 1978 and
summary is based on
information from
Winneke, 1982)
Exposure assessment
and biomarkers
Experiment 1: Subjects
were exposed to 50 ppm
TCE for 3. 5 hours
Experiment 2:
Comparative study of
effects from (a) 50 ppm
TCE for 3.5 hours and
(b) 0.76 ml/kg ethanol.
Tests used
For both experiments 1
and 2: critical flicker
fusion, sustained
attention task, auditory
evoked potentials
Statistics
No statistical
details were
reported.
Results
Significant decrease (p < 0.05) in auditory evoked
potentials in individuals (experiments 1 and 2)
exposed to 50 ppm TCE. No significant effects
were noted in the critical flicker fusion or the
sustained attention tasks.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
ATSDR,
2003
Study population
116 children from
registry of
14 hazardous waste
sites with TCE in
groundwater; under 10
yrs of age at time of
registry; Control
population (n = 111)',
communities with no
evidence of TCE in
groundwater
(measured below
MCL); matched by age
and race; there were
other chlorinated
solvents present in the
exposed group wells.
Exposure assessment
and biomarkers
Exposures were
modeled using tap water
TCE concentrations and
GIS for spatial
interpolation, and
LaGrange for temporal
interpolation to estimate
exposures from
gestation to 1990 across
the area of subject
residences, modeled
data were used to
estimate lifetime
exposures (ppb-yrs) to
TCE in residential
wells; 3 exposure level
groups; control = 0 ppb;
low exposure group = 0
<23 ppb-yrs; and high
exposure group =
>23 ppb-yrs;
confounding exposure
was a concern.
Tests used
Fisher Logemann test;
OSME-R; CSP;
D-COME-T; hearing
screening; DPOAE;
SCAN.
Statistics
Screening results as
binary variables
using logistic
regression within
SAS; independent
variables included
exposure measures,
age, gender, case
history; chi-square
test, Fisher's exact
test, t-tests, linear
models.
Results
Exposed children had higher abnormalities for D-
COME-T (p < 0.002), CSP (p < 0.008),
velopharyngeal function (p < 0.04), high palatal
arch (p < 0.04), abnormal outer ear cochlear
function; No difference observed in exposed and
nonexposed populations for speech or hearing
function; No difference found in OSH function.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
i
^§
H I
O ^
HH Oq
H TO
O
H
W
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
Epidemiological Studies: Controlled Exposure Studies; Neurological Effects of Trichloroethylene
Gamberale
etal., 1976
15 healthy men aged
20-31-yr old
employed by the
Department of
Occupational
Medicine in
Stockholm, Sweden;
Controls: Within
Subjects
(15 serf-controls).
Exposed for TCE 70
mins via a breathing
valve to 540 mg/m3 (97
ppm), 1,080 mg/m3 (194
ppm), and during
ordinary atmospheric
air. Sequence was
counterbalanced
between the 3 groups,
days, and exposure
levels. Concentration
was measured with a
gas chromatographic
technique every third
min for the 1st 50 mins,
then between tests
thereafter.
RT addition, simple RT,
choice RT and short
term memory using an
electronic panel.
Subjects also assessed
their own conditions on
a 7-pt scale.
Friedman two-way
analysis by ranks to
evaluate difference
between 3
conditions,
nonsignificant
when tested
individually, but
significant when
tested on the basis
of 6 variables.
Nearly half of the
subjects could
distinguish
exposure/nonexpos
ure. ANOVAfor
4 performance tests
based on a 3 x 3
Latin square design
with repeated
measures.
In the RT-Addition test the level of performance
varied significantly between the different exposure
conditions (F[2.24] = 4.35;;? < 0.051) and between
successive measurement occasions (tF[2.24] =
19.25;;? < 0.001); The level of performance
declined with increased exposure to TCE, whereas
repetition of the testing led to a pronounced
improvement in performance as a result of the
training effect; No significant interaction effects
between exposure to TCE and training.
Konietzko et
al., 1975
This is a controlled
exposure study
conducted on
20 healthy male
students and scientific
assistants with a mean
age of 27.2yrs.
Subjects were exposed
to a constant TCE
concentration of 95.3
ppm (520 mg/m3) for up
to 12 h, and Blood
concentrations of TCE
were also analyzed at
hourly intervals.
Evaluated for changes
in alpha waves
(< 14 Hz) in the EEC
recordings; EEG
recordings were
performed hourly for a
period of 1 min with the
eyes closed. This was
used as a potential
measure of
psychomotor
disturbance.
The alpha segment increased over time of
exposure (from 0800 to 0900 and 1000 h [military
time]) (P = 0.05). There were no significant
differences for the other time spans or for other
parameters. Subjects with highest and lowest TCE
blood levels <2 ug/mL and >5 ug/mL were
compared to determine if they showed different
responses, but no case were the differences
statistically different.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Kylinetal.,
1967
Salvini,
1971
Study population
12 subjects exposed to
l,OOOppmTCEfor2h
in a 1.5 x 2 x 2 meters
chamber; 2 subjects
were given alcohol
(0.7 gm of body
weight); Controls: 7 of
the 12 were tested
some days prior to
exposure and 5 of the
12 were tested some
days after exposure.
This is a controlled
exposure study
conducted on 6 male
university students.
Each subject was
examined on 2
different days, once
under TCE exposure,
and once as serf
controls, with no
exposure.
Exposure assessment
and biomarkers
l,OOOppmofTCEwas
blown into a chamber
via an infusion unit and
vaporizing system.
Ostwald's distribution
factor for TCE— the
quotient of the amount
of solvent in the blood
by the amount of
alveolar air.
TCE concentration was
110ppmfor4-h
intervals, twice per day.
0-ppm control exposure
for all as serf controls.
Tests used
Optokinetic Nystagmus;
Venus blood and
alveolar air specimens
were taken at various
times after exposure
and analyzed in a gas
chromatograph with a
flame ionization
detector.
Two sets of tests were
performed for each
subject corresponding
to exposure and control
conditions. Perception
test with tachistoscopic
presentation, Wechsler
memory scale, complex
reaction time test
(CRT), and manual
dexterity test.
Statistics
Ostwald's
distribution factor
for TCE (the
quotient of the
amount of solvent
in the blood in
mg/L by the
amount of the
alveolar air in
mg/L) = 9.7;
Significant
relationship
between TCE in air
and blood (0.88).
ANOVA
Results
"A number" of subjects showed reduction in
Fusion limit although more pronounced in the 2
subjects who consumed alcohol. "Others,"
however, showed little if any effect. No stats.
A decrease in function for all measured effects was
observed. Statistically significant results were
observed for perception tests learning (p < 0.001),
mental fatigue (p < 0.01), subjects (p < 0.05); and
CRT learning (p < 0.01), mental fatigue (p < 0.01),
subjects (p < 0.05).
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Stewart et
al., 1970
Study population
13 subjects in
10 experiments
Exposure assessment
and biomarkers
Ten chamber exposures
to TCE vapor (100 ppm
and 200 ppm) for
periods of 1 h to a 5-day
work week.
Experiments 1-7 were
for a duration of 7 h
with a mean TCE
concentration of
198-200 ppm.
Experiments 8 and
9 exposed subjects to
202 ppm TCE for a
duration of 3.5 and 1 h,
respectively.
Experiment 10 exposed
subjects to 100 ppm
TCE for 4 h.
Experiments 2-6 were
carried out with the
same subjects over
5 consecutive days; Gas
chromatography of
expired air; No self
controls.
Tests used
Physical examination
1 h prior to exposure.
Blood analysis for
complete blood cell
count (CBC),
sedimentation rate, total
serum lipid, total serum
protein, serum
electrophoresis, serum
glutamic oxaloacetic
transaminase (SCOT)
and serum glutamic
pyruvic transaminase.
24-h urine collection for
urobilinogen, TCA and
TCE. Also a
preexposure
expirogram, tidal
volume measurement,
and an alveolar breath
sample for TCE; Short
neurological exam
including modified
Romberg test, heel-to-
toe test, finger-to-nose
test.
Statistics
Descriptive
statistics.
Results
Ability to perceive TCE odor diminished as
duration of expo increased; 40% had dry throat
after 30 min. exposure; 20% reported eye
irritation; Urine specimens showed progressive
increase in amounts of TCE metabolites over the
5 consecutive exposures. Concentrations of TCA
and TCE decreased exponentially after last
exposure, but still present in abnormal amounts in
urine specimens 12 d after exposure. Loss of
smelling TCE: >1 h = 33%; >2 h = 80%; >6.5 h
= 100%; Symptoms of lightheadedness, headache,
eye, nose and throat irritation. Prominent fatigue
and sleepiness by all after 200 ppm. These
symptoms may be of clinical significance. All had
normal neurological tests during exposure, but
50% reported greater mental effort was required to
perform a normal modified Romberg test on more
than one occasion.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Triebig,
1976
Study population
This was a controlled
exposure study
conducted on 7 healthy
male and female
students (4 females, 3
males). The control
group was 7 healthy
students (4 females,
3 males).
Exposure assessment
and biomarkers
Subjects exposed for 6
h/d for 5 d to 100 ppm
(550 mg/m3 TCE).
Controls were exposed
in chamber to zero TCE.
Biochemical tests
included TCE, TCE, and
trichloroethanol in
blood. In this study the
TCE concentrations in
blood reported ranged
from 4 to 14 ug/mL. A
range of 20 to 60 ug/mL
was obtained for TCA
in the blood.
Tests used
Psychological tests
were: the d2 test was an
attention load test; the
short test is used to
record patient
performance with
respect to memory and
attention; daily
Fluctuation
Questionnaire measured
the difference between
mental states at the start
of exposure and after
the end of exposure is
recorded; The MWT-A
is a repeatable short
intelligence test; the
Freiburg Personality
Inventory is a test for
12 independent
personality traits;
CFT-3 is a nonverbal
intelligence test;
Erlanger Depression
Scale.
Statistics
Regression
analyses were
conducted.
Results
There was no correlation seen between exposed
and unexposed subjects for any measured
psychological test results. The biochemical data
did demonstrate that exposed subjects' exposures.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
5 Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
o
o S3'
sx
|
•§
1
S3'
a
g^
^
C^
^
f^
^
^
1e
^
1 $>
^ %
zL
^
i
3 ~-
H W^
£ S
^ O
_, ^
^ o
D s
2, Q
5 !'
H ^
-H Oq
rt §
D^
Reference
Triebig et
al., 1977a
Vernon and
Ferguson,
1969
Study population
This was a controlled
exposure study
conducted on 7 healthy
male and female
students (4 females, 3
males) The control
group was 7 healthy
students (4 females,
3 males).
8 male volunteers age
range 2 1-30; serf
controls: 0 dose.
Exposure assessment
and biomarkers
Subjects exposed for 6
h/d for 5 days to 100
ppm (550 mg/m3 TCE).
Controls were exposed
in chamber to zero TCE.
Biochemical tests
included TCE, TCA and
trichloroethanol in
blood. In this study the
TCE concentrations in
blood reported ranged
from 4 to 14 ug/mL. A
range of 20 to 60 ug/mL
was obtained for TCA
in the blood.
TCE administered as
Trilene air-vapor
mixtures through
spirometers
administered at random
concentrations of 0, 100,
300, or 1,000 ppm of
TCE for 2 h at a time,
during which testing
took place.
Concentrations were
measured with a halide
meter. Medical history,
exam including CBC,
urinalysis, BUN, and
SCOT.
Tests used
The testing consisted
of: the Syndrome Short
Test; the "Attention
Load Test" or "d2
Test;" Number recall
test, letter recall test,
The "Letter Reading
Test," "Word Reading
Test," Erlanger
Depression Scale.
Scale for Autonomic
Dysfunction, Anxiety
Scale, Pain Short Scale,
and Information on
Daily Fluctuations.
Flicker Fusion with
Krasno-Ivy Flicker
Photometer,
Howard-Dolman depth
perception apparatus,
Muller-Lyer
two-dimensional
illusion, groove-type
steadiness test, Purdue
Pegboard, Written
"code substitution,"
blood studies.
Statistics
Statistics were
conducted using
Whitney Mann.
ANOVAs,
Dunnett's test.
Results
Results indicated the anxiety values of the placebo
random sample group dropped significantly more
during the course of testing (p < 0.05) than those
of the active random sample group. No
significantly different changes were obtained with
any of the other variables.
TCE did not produce any appreciable effects at
lower concentrations. Compared to controls,
participants exposed to 1,000 ppm of TCE had
adverse effects on the Howard-Dolman,
steadiness, and part of the pegboard, but no effects
on Flicker Fusion, from perception or code
substitution. No appreciable changes in CBC,
urinalysis, SCOT, or BUN.
O
H
W
-------
to
O k^j
o J5"
vo £;•
a
Table D-l. Epidemiological studies: neurological effects of trichloroethylene (continued)
Reference
Study population
Exposure assessment
and biomarkers
Tests used
Statistics
Results
TO'
I
§1
Windemuller
and Ettema,
1978
Pilot study: 24 healthy
male volunteers; age
range = 19-26yr, 4
groups with 6
volunteers in each: (1)
control, (2) exposed to
TCE, (3) exposed to
alcohol, (4) exposed to
TCE and alcohol;
Final study: 15 other
volunteers, each
exposed to all
4 conditions.
Chamber study; Group 1
no exposure; Group 2
TCE exposure: 2.5 h
with 200 ppm; Group 3
alcohol exposure: 0.35
g/kg body weight;
Group 4 TCE and
alcohol: same as above
levels; Blood alcohol
levels taken with
breathalyzer; exhaled air
sampled for levels of
TCE and
trichloroethanol; TCE
exposure: average
measured TCE in
exhaled air = 29 ug/L
(SD = 3); TCE and
alcohol expo: average
measured TCE in
exhaled air = 63 ug/L
(SD = 12).
Binary Choice Task
(Visual); Pursuit Rotor;
Recording of heart rate,
sinus arrhythmia,
breathing rate;
Questionnaire (15 items
on subjective feelings).
K-sample trend
test; two-tailed
Wilcoxon test.
Pilot study: no systematic effect of exposure on
test perform. Alcohol group had higher heart rate
than TCE group, and TCE and alcohol group;
minimal effect of mental load on heart rate; sinus
arrhythmia suppressed as mental load increased
with higher suppression in exposed groups (all 3)
compared to controls (differences possibly due to
existing group differences); Final Study: pursuit-
rotor task "somewhat impaired by exposure
condition;" authors acknowledge possibility of
sequence effects; no significant difference between
conditions on questionnaire responses; performing
mental tasks resulted in higher heart rate in the
TCE + alcohol condition than in Alcohol alone
condition; Mental load suppressed sinus
arrhythmia, especially in TCE + alcohol condition;
Conclusion: TCE and alcohol together impair
mental capacity more than each one alone.
BUN = blood urea nitrogen, EEG = electroencephalograph, GI = gastrointestinal, NIOSH = National Instutute of Occupational Safety and Health,
OR = odds ratio, PCE = perchloroethylene.
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents
o S3'
a.
s
1
§
S3'
a
^
"§>
^
^
TO
TO'
KQ
5-
O
3 TO
» O
£
-.8.
3|
t> TO
n <»
^ S.
3 8
D g
2 »•
— < 0*4,
•N ^
H TO"
Reference
Albers etal.,
1999
Antti-Poika,
1982
Study population
30 railroad workers with
toxic encephalopathy;
involved in litigation;
long-term exposure to
solvents (n = 20 yrs.;
range = 10-29 yrs.);
Historical controls
matched by gender, age,
and body mass.
87 patients (painters,
paint and furniture
factory workers, carpet
and laundry workers)
diagnosed 3-9 yrs prior
with chronic solvent
exposure (mean age
38.6 yrs)
Control: 29 patients with
occupational asthma.
Exposure assessment and
biomarkers
Most common solvents
included trichloroethylene,
trichloroethane,
perchloroethy lene ;
respirator not typically
used.
Mean duration of exposure
10.4 yrs; solvents:
trichloroethylene,
perchloroethylene, solvent
mixture; based on patients'
and/or employers' reports;
9 worksites visited for
environmental measures;
biological measures at
1 worksite; exposure
classified as low, moderate,
or high.
Tests used
Neurologic exams (cranial
nerves, motor function,
alternate motion range,
subjective sensory
function, Romberg test,
reflexes), occupational
history, medical history,
sensory and motor nerve
conduction studies (NCS).
Interview, Neurologic
exam, EEG,
electroneuromyographs,
psychological examination
(intellectual, short-term
memory, sensory and
motor functions).
Statistics
Log
transformations of
amplitude data;
Mann-Whitney
U Test for NCS;
t-test; simple linear
regression and
stepwise regression
for dose response.
Correlation
coefficients for
prognosis and
factors influencing
diagnosis.
Results
3 workers met clinical
polyneuropathy criteria; NCS values
not influenced by exposure duration
or job title; no significant difference
in NCS between presence or
absence of polyneuropathy
symptoms, disability status, severity
or type of encephalopathy, or prior
polyneuropathy diagnosis.
Reported symptoms: fatigue,
headaches, memory disturbances,
pain, numbness, paresthesias;
1st exam: 87 patients with objective
and subjective neurological signs,
61 with psychological disturbance,
58 abnormal EEG, 25 clinical
abnormalities, 57 PNS symptoms;
69 patients had neurophysiological
or psychological disturbances
identified by neurologist in only 4
patients; 2nd exam: 42 with clinical
neurological signs, ; 21 patients
deteriorated, 23 improved, 43 same;
poor correlation between prognosis
of examinations; no significant
correlation between prognosis and
age, sex, exposure duration and
level, alcohol use, or other diseases.
H TO1
W S
O
H
W
-------
o
to
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
o S3'
§-
ri
1
S-
S3'
a
^s
^=s>
s
TO
TO'
^3
1
g
° !
Si
_, ^
eg-
>• ^
~
_j "*^
S.
3 o
0 ^
2 a-
Reference
Aratanietal.,
1993
Binaschi and
Cantu, 1982
Study population
437 exposed workers
from various industries
(not specified);
394 males, 43 females
and 1,030 male clerical
workers as controls; age
range: 16-72.
35 patients with
occupational exposure to
organic solvents;
Industry not specified; no
controls.
Exposure assessment and
biomarkers
Exposed to Thinner,
G/5100, TCE, xylene,
toluene, methylchloride,
gasoline.
Occupational history
provided by patients;
Descriptions of jobs and
conditions provided by
employer; Workplace
observations; Some
available measurements of
solvents in air; 9 patients
exposed to
trichloroethylene;
1 1 exposed to toluene and
xylene; 15 exposed to
mixtures of solvents; all
exposures described to be
under TLV-TWA, but short
exposure might have
exceeded ACGIF limit for
short time.
Tests used
Vibrometer (VPT);
Urinary Metabolites.
Examination of provoked
and spontaneous vestibular
symptoms; Pure tone
threshold measurement;
EEG; psychiatric
interviews and psychiatric
history; Prevalence of
37 psychiatric symptoms.
Statistics
Spearman
correlation.
Not stated.
Results
Positive correlations between age
and VPT 7; between job experience
and VPT; Urinary metabolites not
significantly correlated with VPT;
no dose-effect for subjective
symptoms and neurological signs.
All patients had subjective
symptoms (fatigue, psychic
disturbances, dizziness, vegetative
symptoms, vertigo); Vestibular
system affected in most cases, with
lesions in nucleo-reticular substance
and brain stem; EEG change with
diffuse and focal slowing; 71% of
patients had mild neurasthenic
symptoms (fatigue, emotional
instability, memory and
concentration difficulties).
H ^
H TO1
W S
O
H
W
-------
o
to
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
o S3'
§"
TO
S
TO
s
S3'
a
^
^
^
^
TO
TO'
^5
1
'O
O
5
D O
^
£
^
o a.
> TO
TJ
T ""S4
~* 0
^s
3 g
2 £•
H ^
-} ^
H ^
ii jsj
id ^a
Reference
Bowler et al.,
1991
Colvin et al.,
1993
Study population
67 former
microelectronics workers
exposed to multiple
organic solvents;
Controls (n = 157) were
recruited from the same
region; 67 pairs were
matched on the basis of
age, sex, ethnicity,
educational level, sex,
and number of children.
Final sample: 67 workers
(43 exposed;
24 unexposed) in a paint
manufacturing plant
employed there for at
leasts yrs.; all black
males; exclusion criteria:
encephalopathy, head
injury with 24 + h
unconsciousness,
psychotropic medication,
alcohol/drug dependence
history, epilepsy, mental
illness.
Exposure assessment and
biomarkers
Self-report and work history
from microelectronics
workers. Exposures and
risks were estimated.
Solvents include TCE,
TCA, benzene, toluene,
methylene chloride,
n-hexane.
Chronic exposure was
assessed through
self-reported detailed work
history for each worker;
past and current industrial
hygiene measurements of
solvent levels in air; "total
cumulative expo" in the
factory and "average
lifetime exposures" were
calculated; visitations to
establish areas with
"homogeneous exposure;"
All exposures below the
ACGIH limit. Solvents
include MEK, benzene,
TCE, MIBK, toluene, butyl
acetate, xylene, cellosolve
acetate, isophorone, and
white spirits.
Tests used
California
Neuropsychological
Screening Battery.
Work and personal history
interview; brief
neurological evaluation,
WHO Neurobehavioral
Core Test Battery (all tests
except POMS); Computer-
administered tests:
Reaction time,
Fingertapping, Continuous
Performance Test,
Switching attention,
Pattern Recognition Test,
Pattern Memory; UNISA
Neuropsychological
Assessment Procedure:
Four word memory test,
Paragraph memory,
Geometric Shape drawing;
symptom and health
questionnaires.
Statistics
t-test for matched
pairs; Wilcoxon
Signed Rank test.
Division into
exposed and
unexposed;
Student's t-test;
Multiple linear
regression.
Results
Exposed workers performed
significantly worse on tests of
attention, verbal ability, memory,
visuospatial, visuomotor speed,
cognitive flexibility, psychomotor
speed, and reaction time; no
significant differences in mental
status, visual recall, learning, and
tactile function.
Exposed group performed worse
than unexposed on 27 out of 33 test
results; only significant difference
was on latency times of two
switching attention tests; no
difference in subjects' symptom
reporting between groups when
questions analyzed separately or
analyzed as a group; Average
lifetime exposure was a significant
predictor for Continuous
performance latency time,
Switching attention latency time,
Mean reaction time, Pattern
Memory; fine visuomotor tracking
speed significantly associated with
cumulative exposure; effects of
exposure concluded to be "relatively
mild" and subclinical.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Daniell et al.,
1999
89 retired male workers
(62-74-yr old) with prior
long-term exposure to
solvents including
67 retired painters and
22 aerospace
manufacturing workers;
Controls: 126 retired
carpenters with minimal
solvent exposure.
Chronic occupational
exposure; Structured
clinical interview about past
and present exposure to
solvents; Cumulative
Exposure Index was
constructed. Solvents not
specified.
Psychiatric interview;
questionnaires; physical
exam; blood cell counts,
chemistry panel, blood
lead levels,
Neuropsychological: BDI,
verbal fluency test, WAIS-
R: Vocabulary,
Similarities, Block Design,
Digit Span, Digit Symbol;
Wisconsin Card Sorting;
verbal aphasia screening
test, Trails A and B,
Fingertapping; WMS-R:
logical memory and visual
subtests; Rey Auditory
Verbal Learning; Benton
Visual Retention test; d2
test; Stroop; Grooved
pegboard; simple reaction
time.
Odds ratio,
logarithmic
transformation of
non-Gaussian data,
standardization of
test scores,
ANCOVA,
Multiple Linear
regression; Kruskal
Wallis test for
differences in
blood lead
concentration.
CEI was similar for painters and
aerospace workers; Painters
reported greater alcohol use than
carpenters; painters also had lower
scores on WAIS-R Vocabulary
subtest; Controlling for age,
education, alcohol use, and
vocabulary score, painters
performed worse on motor,
memory, and reasoning ability tests;
painters reported more symptoms of
depression and neurological
symptoms; painters more likely to
have more abnormal test scores
(odds ratio: 3.1) as did aerospace
workers (odds ratio: 5.6); no dose
effect with increasing exposure and
neuropsychological tests.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
i
^§
H I
O ^
HH Oq
H TO
O
H
W
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Donoghue et
al., 1995
16 patients diagnosed
with organic-solvent-
induced toxic
encephalopathy with
various occupations
compared to
age-stratified normal
groups (n = 38); average
age: 43 y (range
= 31-58); Exclusion
criteria: diabetes mellitus,
ocular disease impairing
vision, visual acuity with
existing refractive
correction of less than
4/6, abnormal direct
ophthalmoscopic exam.
Average exposure duration
was 19 yrs (range = 5-36
yrs); Solvents include TCE,
MEK, toluene, thinners,
unidentified hydrocarbons.
Visual acuity measured
with a 4-m optotype chart;
Contrast sensitivity
measured with Vistech
VCTS 6500 chart;
monocular thresholds,
pupil diameter.
Chi-square test.
6 participants (37.5%) with
abnormal contrast sensitivity; 2 of
the 6 (33%) had monocular
abnormalities; abnormalities
occurred at all tested spatial
frequencies; significant difference
between groups at 3 cpd, 6 cpd,
12 cpd frequencies.
Elofsson et
al., 1980
Epidemiologic study of
car or industrial spray
painters (male) exposed
long-term to low levels
of organic solvents
(n = 80); 2 groups of
matched controls;
80 nonexposed male
industrial workers in each
control group.
Long term, low level expo
to multiple solvents;
Assessed by interviews, on-
the-job measurements, and
a 1955 workshop model;
Blood analysis: mean
values were within normal
limits for both groups;
Exposed group had
significantly higher values
for alkaline phosphates,
hemoglobin, hematocrit,
and erythrocytes; early
exposure TLVs in Sweden
were significantly lower;
solvents include TCE, TCA,
methylene chloride, and
others.
Serf-administered
psychiatric questionnaires,
Eysenck's Personality
Inventory, psychosocial
structured interview,
Comprehensive
Psychopathological Rating
Scale; Visual Evoked
Responses; EEG;
Electroneurography;
Vibration Sense Threshold
estimations; Neurological
exam.
Calculation of
z values; Pearson
correlation;
Multiple
Regression
Analysis.
Significant differences between
controls and exposed in symptoms
of neurasthenic syndrome, in
reaction time, manual dexterity,
perceptual speed, and short-term
memory; no significant differences
on verbal, spatial, and reasoning
ability; Some differences on EEG,
VER, ophthalmologic, and CT.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Gregersen,
1988
Workers exposed to
organic solvents (paint,
lacquer, photogravure,
and polyester boat
industries); Controls:
warehousemen
electricians; 1st follow-up
5.5yrs after initial
evaluation (59 exposed,
30 unexposed);
2nd follow-up: 10.6 yrs
after initial evaluation
(53 exposed,
30 unexposed controls).
1st follow-up: data about
working conditions,
materials and exposure in
prior 5 yrs used for
exposure index; 2nd follow-
up: 9 questions asking about
exposure to solvents in the
prior 5 yrs; TCE, toluene,
styrene, white spirits.
1st follow-up: structured
interviews on
occupational, social,
medical history; clinical
exam, neurological exam;
2nd follow-up: mailed
questionnaire (49 follow-
up issues to 1st follow-up).
Wilcoxon-Mann-
Whittney tests;
Kruskal-Wallis
test; Chi-square;
Spearman Rank
Partial Correlation
Coefficient.
More acute neurotoxic symptoms in
exposed group at both follow-ups,
but fewer symptoms at
2nd follow-up than at 1st follow-up;
at both follow-ups exposed
participants had more
encephalopathy symptoms,
especially memory and
concentration; no encephalopathy
symptoms in control group;
symptoms and signs of peripheral,
sensory, and motor neuropathy
significantly worse in participants
still exposed; Exposure index
showed dose-effect with memory
and concentration; Both follow-ups:
improvement in acute symptoms;
aggravation in CNS; more
symptoms of peripheral nervous
system and social consequences.
^§
"*4-
o§
H I
O ^
HH Oq
H TO
O
Juntunen et
al., 1980
37 patients with
suspected organic solvent
poisoning (mean age
= 40.1 yrs.); selection
based on
pneumoencephalography;
no controls.
Patients were exposed to
Carbon disulphide (n = 6),
trichloroethylene (5),
styrene (1), thinner (2),
toluene (1), methanol (1),
and carbon tetrachloride (2),
mixtures (19); Exposure
was assessed by patients'
and employers' reports and
measurements of air
concentrations when
available.
Neurologic examination,
pneumoencephalographic
exam, EEG, tests assessing
intelligence, memory and
learning, motor function,
and personality.
Descriptive
Statistics.
Clinical neurological findings of
slight psychoorganic alterations,
cerebellar dysfunction, and
peripheral neuropathy; 63% had
indication of brain atrophy; 23 of
the 28 patients examined with
electroneuromyography showed
signs of peripheral neuropathy; 94%
had personality changes, 80% had
psychomotor deficits, 69% had
impaired memory, and 57% had
intelligence findings; No dose-effect
found.
H
W
-------
o
to
vo S3'
TO
s
s
a
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
TO
TO'
Juntunen et
al., 1982
80 (41 women, 39 men)
Finnish patients
diagnosed 3-9 yrs prior
with chronic solvent
exposure (mean age
= 38.6 yrs); 31 had slight
neurological signs; no
controls.
Assessed by patients'
occupational history,
employers' workplace
description, observations
and data collected at
workplace, environmental
measurements, biological
tests; TCE, PCE, or mixed
solvent exposures.
Neurologic examination;
EEC and ENMG; tests of
intellectual function,
memory, learning,
personality and
psychomotor performance.
Chi-square,
Maxwell-Stuart,
Correlation and
multiple linear
regression
analyses.
Significant correlations between
prognosis of disturbances in gait
(p < 0.05) and station and length of
follow-up, duration and level of
exposure and multiplying the two;
no gender effects; Common
subjective symptoms; headaches,
fatigue, and memory problems;
Impairment in fine motor skills,
gait, and cerebellar functions;
Subjective symptoms decreased
during follow-up, but clinical signs
increased.
TO
H!
o §
o
H
W
Laslo-Baker
et al., 2004
32 mothers with
occupational exposure to
organic solvents during
pregnancy and their
children (3-9 yrs of age);
included if exposure
started in 1st trimester
and lasted for at least
8 wks of pregnancy
(32 mother-child pairs);
Controls: 32 unexposed
control mothers matched
on age, child age, child
sex, SES, and reported
cigarette use and their
children (32 mother-child
pairs).
Exposure information
collected at 3 times:
(1) during pregnancy,
(2) when contacted for
study participation later in
pregnancy, (3) at time of
assessment; Information
collected included types of
solvent, types of setting,
duration of exposure during
pregnancy, use of
protection, symptoms,
ventilation; Solvents
include toluene (n = 12
women), xylene (10),
ethanol (7), acetone (6),
methanol (5), TCE (3), etc.
(a total of 78 solvents were
reported).
Children: Wechsler
Preschool and Primary
Scale of Intelligence,
WISC, Preschool
Language Scale, Clinical
Evaluations of Language
Fundamentals, Beery-
Buktenica Developmental
test of Visuo-Motor
Integration, Grooved
Pegboard Test, Child
Behavior Checklist (Parent
Version), Connor's Rating
Scale-Revised (Parent
Version), Behavioral Style
Questionnaire; Mothers:
WASI.
Power analysis,
Multiple linear
regression.
Verbal IQ was lower (104) in
children exposed in Mtero vs.
unexposed children controls (110);
Children did not differ between
groups in birth weight, gestational
age, or developmental milestones;
Children in the exposed group had
significantly lower VIQ (108) and
Full IQ (108) than controls (VIQ
= 116 and Full IQ = 114; No
significant difference in PIQ;
Performance on expressive
language, total language, and
receptive language was significantly
worse in children from exposed
group.
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Lee et al.,
1998
Study population
40 Korean female shoe
factory workers
employed there for at
least 5 yrs.; cases with
head injury, neurological
or psychological
disorder, or hearing or
visual impairment were
excluded; Controls:
28 (housekeepers); no
in-plant controls
available.
Exposure assessment and
biomarkers
4 workers wore passive
personal air samplers for a
full 8-h shift; Detected
solvents: toluene, methyl
ethyl ketone, w-hexane,
c-hexane, cyclohexane,
dichloroethylene,
trichloroethylene, benzene,
and xylene; In frame-
making air concentration of
solvents was 0.46-0.71; In
adhesive process solvent air
concentrations were
1.83-2.39; three exposure
indices were calculated:
current exposures, exposure
duration (yrs), and
Cumulative Exposure
Estimate (CEE) (yrs
x average exposures).
Tests used
Questionnaire;
Neurobehavioral Core Test
Battery (includes POMS,
Simple Reaction Time,
Santa Ana Dexterity test,
Digit Span, Benton Visual
Retention Test, Pursuit
aiming motor steadiness
test); POMS was excluded
because of cultural
inapplicability.
Statistics
Multivariate
ANOVA for tests
with 2 outcomes;
ANOVA for tests
with 1 outcome;
education was
adjusted in
analyses.
Results
Significant differences between
groups based on exposure index;
Differences in performance between
controls and participants on Santa
Ana were found only in the CEE
(participants performed worse);
CEE is a more sensitive measure of
exposure to organic solvents.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
si
ON O
•••2
i
^§
H I
O >
HH Oq
H TO
O
H
W
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Lindstrom,
1973
Lindstrom,
1980
Study population
168 male workers with
suspected occupational
exposure to solvents
Group I with solvent
poisoning (n = 42);
Group II with solvent
exposure, undergoing
mandatory periodic
health check (n = 126);
Control-50 healthy
nonexposed male
volunteers working in a
viscose factory; Group
IV 50 male workers with
carbon disulfide
poisoning.
56 male workers
diagnosed with
occupational disease
caused by solvents;
Controls:
98 styrene-exposed
workers; 43 nonexposed
construction workers.
Exposure assessment and
biomarkers
44 exposed to TCE, 8 to
tetrachloroethylene, 26 to
toluene, 25 to toluene and
xylene, 44 to thinners, 21 to
"miscellaneous;" Solvent-
exposed group had an
average of 6 y of expo; CS2
group had average of 9 yrs
of exposure.
Chronic "excessive"
exposure: Mean duration of
exposure = 9.1 yrs (SD
= 8.3); Exposed to;
halogenated and aromatic
hydrocarbons, paint
solvents, alcohols, and
aliphatic hydrocarbons
(TCE n= 14); Individual
exposure levels estimated as
time-weighted averages,
based on information
provided by subjects,
employer, or workplace
measurements, were
categorized as low (3
patients), intermediate (26
patients), and high (27
patients).
Tests used
WAIS: Similarities,
Picture Completion, Digit
Symbol; Bourdon-
Wiersma vigilance test,
Santa Ana, Rorschach
Inkblot test, Mira test.
WAIS subtests:
Similarities, Digit Span,
Digit Symbol, Picture
Completion, Block Design;
WMS subtests: Visual
Reproduction; Benton
Visual Retention test;
Symmetry Drawing; Santa
Ana Dexterity test; Mira
test.
Statistics
Student' st-test.
Factor analysis;
Student's t-test;
Multivariate
Discriminant
analysis.
Results
The solvent-exposed group and CS2
group had significantly worse
"psychological performances" than
controls; Greatest differences in
sensorimotor speed and
psychomotor function;
solvent-exposed and CS2 groups had
deteriorated visual accuracy.
Significant decline in visuomotor
performance and freedom from
distractibility (attention) in the
solvent-exposed participants;
significant relationship between
duration of solvent exposure and
visuomotor performance; solvent
exposure level was not significant;
psychological test performance of
styrene-exposed control was only
slightly different from nonexposed
controls.
-------
to
O k^j
o J5"
vo £;•
TO
s
a
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
TO
TO'
O
o
Lindstrom et
al., 1982
86 Patients with prior
diagnosis of solvent
intoxication (mean age
38.6 yrs.); 40 male, 46
female; 52 exposed to
mixed solvents; 21
exposed to TCE or PCE;
13 exposed to both;
results at follow-up
compared to those at
initial diagnosis.
Mean duration of exposure
10.4 yrs; solvents:
trichloroethylene,
perchloroethylene, solvent
mixture; based on patients'
and/or employers' reports.
Intellectual Function: from
WAIS - Similarities,
Block Design, Picture
Completion; Short Term
Memory: from WMS -
Digit Span, Logical
Memory, Visual
Reproduction; Benton
Visual Retention test;
Sensory and Motor
Functions: Bourdon
Wiersma Vigilance Test,
Symmetry Drawing, Santa
Ana Dexterity test, Mira
test.
Frequency
distributions,
Student'st-test for
paired data,
stepwise linear
regression.
All patients grouped together
regardless of types of past solvent
exposure; on follow-up, significant
learning effects for Similarities
when compared to results at initial
diagnosis; group mean for
intellectual functioning increased;
no significant change in memory
test results; group means for sensory
and motor tasks were lower;
prognosis was better for longer
follow-up and younger age and
poorer for users of medicines with
neurological effects.
o §
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
o S3'
§"
TO
TO
S
S3'
a
^
^
^
^
TO
TO'
^
s
o
-J
» o
it
^
i
^ ^
° o
n <»
^ 0
^
3 g
2 £•
H ^
-} ^
H TO
r| S
Reference
Marshall et
al., 1997
McCarthy
and Jones,
1983
Study population
All singleton births in
1983-1986 in 188 New
York State counties (total
number not specified);
473 CNS-defect births
and
3,305 musculoskeletal-
defect births; Controls:
12,436 normal births;
Exclusion criteria:
Trisomy 13, 18, or 21,
birth weight of less than
1,000 g, sole diagnosis of
hydrocephaly or
microencephalopathy,
hip subluxation.
384 industrial workers
with solvent poisoning;
103 operated degreasing
baths, 62 maintained
degreasing baths, 37 used
TCE in portable form,
37 misc; no controls.
Exposure assessment and
biomarkers
Information on inactive
waste sites was examined,
including air vapor, air
particulates, groundwater
exposure via wells, and
groundwater exposure, via
basements; exposure was
categorized as "high,"
"medium," "low," or
unknown based on
probability of exposure;
proximity to waste sites was
also considered; Most
common solvents: TCE,
toluene, xylenes,
tetrachloroethene,
1,1,1-trichloroethane; Most
common metals found lead,
mercury, cadmium,
chromium, arsenic, and
nickel.
Individuals poisoned with
trichloroethylene,
perchloroethylene, and
methylchloroform were
examined retrospectively;
Medical record review; 288
exposed to TCE, 44 to
perchloroethylene, 52 to
1,1,1-trichloroethane.
Tests used
Symptoms reported in
occupational/medical
records from industrial
poisoning incidents; data
from 1961 to 1980 on
demographics, occupation,
work process, type of
industry, if incident caused
fatality.
Statistics
Odds ratios (OR),
Fisher's exact test,
Chi-square,
unconditional
logistic regression.
Results
13 CNS cases and 351 controls with
potential exposures; crude OR: 0.98;
When controlling for mother's
education, prenatal care, and
exposure to a TCE facility, OR was
0.84; CNS and solvents OR: 0.8;
CNS and metals OR: 1.0,
musculoskeletal defects and
solvents OR: 0.9, musculoskeletal
defects and pesticides OR: 0.8;
higher risk for CNS defects when
living close to solvent-emitting
facilities.
17 fatality cases, with 10 in
confined spaces; Most common
symptoms include effects on CNS;
Gastrointestinal and Respiratory
symptoms; no strong evidence for
cardiac and hepatic toxicity; no
change in affected number of
workers in 1961 to 1980; greatest
effect due to narcotic properties.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Mergler et
al.,1991
54 matched pairs;
Matching on the basis of
age, sex, ethnicity,
educational level, sex,
and number of children
taken fromlSO former
microelectronics workers
exposed to multiple
organic solvents and
control population of
157 recruited from the
same region.
Average duration of
employment: 6.1 yrs (range:
1-15 yrs); information
about products used and
chemical make-up from
employer; chemicals:
chlorofluorocarbons,
chlorinated hydrocarbons,
glycol ethers, isopropanol,
acetone, toluene, xylene,
and ethyl alcohol.
Sociodemographic
questionnaire; Monocular
examination of visual
function: Far visual acuity
using a Snellen chart, near
visual acuity using a
National Optical Visual
Chart, color vision using
Lanthony D-15, near
contrast sensitivity using
Vistech grating charts.
Signed-rank
Wilcoxon test;
Mann-Whitney;
Chi-square test for
matched pairs;
Multiple
Regression;
Stepwise
regression.
Significant difference in near
contrast sensitivity: 75% of exposed
workers with poorer contrast
sensitivity at most frequencies than
the matched controls (no difference
in results based on smoking, alcohol
use, and near visual acuity loss);
Significant differences on near
visual acuity, color vision, and rates
of acquired dyschromatopsia for one
eye only; No difference between
groups in near or far visual acuity.
^§
H I
O ^
HH Oq
H TO
O
Morrow et
al., 1989
22 male patients with
exposure to multiple
organic solvents;
4 involved in litigation;
Exclusion: neurologic or
psychiatric disorder prior
to assessment, alcohol
consumption more than
2 drinks/day; Average yrs
education 12 (range:
10-16 yrs); average age
38 yrs (range: 27-61);
compared to responses of
WWII prisoner of war
(POW) population with
posttraumatic stress
disorder (PTSD).
Exposure assessed with
questionnaire (duration,
type of solvents, weeks
since last exposure, cases of
excessive exposure);
Average exposure duration
= 7.3 yrs (range: 2 mos-19
yrs); average weeks since
last exposure was 19.8
(range: 1-84 wks); 28% had
at least one instance of
excessive exposure.
Exposure questionnaire,
Group form of the MMPI.
Stepwise multiple
regression.
All profiles valid; 90% with at least
2 elevated scales above T score of
70 (clinically significant); Highest
elevations on scales 1, 2, 3, and 8;
only 1 case within normal limits;
when compared to a group of
nonpsychiatric patients, exposed
patients had more elevations,
although both groups have physical
complaints; When compared with
WWII POW (1/2 diagnosed with
PTSD) with similar SES and
education, both groups have similar
profiles; no age effects found;
significant positive correlation
between scale 8 and duration of
exposure; no significant difference
based on time since last exposure or
on experiencing excessive exposure.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
^§
H I
O ^
HH Oq
H TO
O
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Morrow et
al., 1992
9 men and 3 women
occupationally exposed
to multiple organic
solvents with CNS
complaints; all met
criteria for mild toxic
encephalopathy; exposed
group average age was
47 y; Controls:
19 (healthy male
volunteers);
26 psychiatric controls
(male patients with
chronic schizophrenia)
average age unexposed
controls: 34 yrs; average
age schizophrenic
patients.: 36yrs.
Exposure assessed with
occupational and
environmental exposure
questionnaire; mean
duration of expo = 3 y
(range = <1 d-30 y);
average time between last
exposure and assessment
was 2 y (range;
2 mos-10 y); solvents
toluene, TCE.
Auditory event-related
potentials under the
oddball paradigm:
counting and choice
reaction time tasks.
Repeated measures
ANOVA.
Exposed patients had significant
delays in N250 and P300 compared
to normal controls and in P300
compared to psychiatric controls;
Exposed patients had higher
amplitudes for N100, P200, and
N250; no difference inP300
amplitude between groups; for the
exposed group, P300 positively
correlated with exposure duration;
findings indicate that solvent
exposure affects neural networks.
Seppalainen
and
Antti-Poika,
1983
87 patients with solvent
poisoning (40 male and
47 female) with
occupational exposure to
solvents; Follow-up
3-9 yrs after initial
diagnosis; Mean age at
diagnosis 38.6 (range:
20-59 yrs); no control
population.
Chronic exposure with
average duration of 10.7 yrs
(range: 1-33); patients were
exposed to TCE (n = 21),
perchloroethylene (n = 12),
mixtures of solvents
(n = 53), mixtures and TCE
or perchloroethylene
(n = 13); Exposure of 54
patients stopped after
diagnosis, 33 continued to
be exposed; at follow-up,
only 5 working with
potential of some exposure.
EEC using 10/20 system
with 25-30 mins of
recording, 3 mins
hyperventilation and
intermittent photic
stimulation; ENMG.
Chi-square,
Hypergeometric
distribution,
McNemar test.
Significantly more ENMG
abnormalities at follow-up than at
initial diagnosis; Most common
finding: slight polyneuropathy; 43%
showed improved ENMG, 33% had
deteriorated, and 18 pts. with similar
ENMG findings (6 normal at both
exams); at follow-up, slow-wave
abnormalities decreased and
paroxysmal abnormalities increased;
41 with improved EEC, 28 with
similar EEG (19 had normal EEG at
diagnosis), and 18 with deteriorated
EEG; EEG pattern of change
compared to external head injuries.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
^§
H I
O ^
HH Oq
H TO
O
Shlomo et
al, 2002
Male industrial workers;
Mercury exposure group
(n = 40); average age
49.7 (±6.4) yrs;
chlorinated hydrocarbons
(CHs) exposure group
(n = 37) average age 46.0
(±4.73); Controls,
unexposed (n = 36)
average age 49.8 (±5.8),
matched by age;
(industries not specified).
Interview and record
review; Urine samples
collected at end of work
shift prior to testing and
tested for mercury and TCA
; chlorinated hydrocarbons:
TCE (n = 7), PCE (n = 8),
trichloroethane (n = 22);
Mean duration of chloral
hydrate (CH) exposure 15.8
(±7.2) yrs; Mean duration
of mercury exposure 15.5
(±6.4) yrs; Air sampling:
mercury: 0.008 mg/m3
(TLV = 0.025); TCE: 98
ppm(TLV = 350);PCE:
12.7 ppm (TLV = 25);
trichloroethane: 14.4 ppm
(TLV = 200); Blood levels:
mercury (B-hg) 0.5 gr%
(±0.3); TCA urine levels:
1-80% of Biologic
Exposure Index (BEI); CH
urine levels: 0.11-0.2 of
BEI.
Medical history,
Neurological tests
assessing cranial nerves
and cerebellar function;
Otoscopy, review of
archival data from pure-
tone audiometric tests;
Auditory brain stem
responses (ABR).
Student'st-test,
proportions test.
Significant differences between
exposed and controls: 33.8% of CH
exposed workers with abnormal
IPLI-III; 18% of controls; Authors
suggest ABRs are sensitive for
detecting subclinical CNS effects of
CH and mercury.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
to O
•••2
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Tilletal.,
2001
Study population
The children of mothers
who had contacted a
Canadian pregnancy risk
counseling program
during pregnancy and
reported occupational
exposure to solvents
(n = 33); children age
range: 3-7; Mothers'
occupations: lab
technicians, factory
workers, graphic
designers, artists, and dry
cleaning; Controls:
28 matched on age,
gender, parental SES, and
ethnicity; children of
mothers exposed to
nonteratogenic agents.
Exposure assessment and
biomarkers
Structured questionnaire
about exposure; Method:
weight assigned to each
exposure Parameter (length
of exposure, frequency of
exposure, symptoms); sum
of scores for each parameter
used as exposure index;
median split used to
categorize in low (n = 19)
and high (n= 14)
exposures; solvents include
benzene, toluene, methane,
ethane, TCE, methyl
chloride, etc.
Tests used
NEPSY: Visual Attention,
Statue, Tower, Body Part
Naming, Verbal Fluency,
Speeded Naming,
Visuomotor Precision,
Imitating Hand Positions,
Block Construction,
Design Copying, Arrows;
Peabody Picture
Vocabulary Test;
WRAVMA Pegboard test;
Child Behavior Checklist
(Parent form); Continuous
Performance Test.
Statistics
Mantel Haenszel
test, t-test,
ANCOVA,
Hierarchical
multiple linear
regression.
Results
Lower composite neurobehavioral
scores as exposure increased after
adjusting for demographics in
Receptive language, expressive
language, graphomotor ability;
Significantly more exposed children
rated with mild-severe problems;
No significant difference between
groups in attention, visuo-spatial
ability, and fine-motor skills; Mean
difference on broad- and
narrow-band scales of Child
Behavior Checklist scores not
significant.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Tilletal.,
2001
Study population
Children of mothers who
had contacted a Canadian
pregnancy risk
counseling program
during pregnancy and
reported occupational
exposure to solvents
(n = 32); children age
range: 3-7; Mothers'
occupations: lab
technicians, factory
workers, graphic
designers, artists, and dry
cleaning; Controls:
27 matched on age,
gender, parental SES, and
ethnicity; children of
mothers exposed to
nonteratogenic agents.
Exposure assessment and
biomarkers
Structured questionnaire
about exposure; Method:
weight assigned to each
exposure parameter (length
of exposure, frequency of
exposure, symptoms); sum
of scores for each parameter
used as exposure index;
median split used to
categorize in low (n = 19)
and high (n = 14)
exposures; solvents include
benzene, toluene, methane,
ethane, TCE, methyl
chloride, etc.
Tests used
Minimalist test to assess
color vision; Cardiff Cards
to assess visual acuity.
Statistics
Independent
samples t-tests,
Mantel Haenszel
Chi test;
Wilcoxon-Mann-
Whitney test;
Kruskal-Wallis Chi
square.
Results
Significantly higher number of
errors on red-green and blue-yellow
discrimination in exposed children
compared to controls; exposed
children had poorer visual acuity
than controls; No significant
dose-response relationship between
exposure index and color
discrimination and visual acuity.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Tilletal.,
2005
Study population
21 infants (9 male,
12 female)of mothers
who contacted a
Canadian pregnancy risk
counseling program and
reported occupational
exposure to solvents
(occupations: factory,
lab., dry cleaning;
Controls: 27 age-matched
infants (17 male,
10 female) of mothers
contacted the program
due to exposure during
pregnancy to
nonteratogenic
substances).
Exposure assessment and
biomarkers
Structured questionnaire
about exposure; Method:
weight assigned to each
exposure parameter (length
of exposure, frequency of
exposure, symptoms); sum
of scores for each parameter
used as exposure index;
median split used to
categorize in low and high
exposures; exposure groups:
(1) aliphatic and/or
aromatic hydrocarbons
(n = 9), (2) alcohols (n = 3),
(3) multiple solvents
(n = 6), (4) PCE, (n = 3);
mean duration of exposure
during pregnancy 27.2 wks.
(SD 7.93, range = 12-40);
solvents include benzene,
toluene, methane, ethane,
TCE, methyl chloride, etc.
Tests used
1st visit: Sweep visual
evoked potentials (VEP) to
assess contrast sensitivity
and grating acuity; 2nd visit
(2 wks after 1st): Transient
VEPs to assess chromatic
and achromatic
mechanisms;
ophthalmological exam,
physical and neurological
exam; testers masked to
exposure status of infant.
Statistics
Median split;
Multiple Linear
Regression;
Chi-square, t-test,
Mann-Whitney
U test, Multivariate
ANCOVA,
Pearson
correlation,
Logistic
Regression.
Results
Significant decline of contrast
sensitivity in low and intermediate
spatial frequencies in exposed
infants when compared with
controls; Significant effect of
exposure level on grating acuity,
26.3% of exposed (but 0% of
controls) with abnormal VEP to
red-green onset stimulus; No
differences between groups in
latency and amplitude of chromatic
and achromatic response.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
o
to
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
JS"
S3'
§"
TO
S
TO
s
S3'
a
^
•=*>
^
^
TO
TO'
Xs
O
o
«L
xl
i
a.
TO
a
o
|
£•
§.
&x
^
JSJ
x;
T^
O
Reference
Valic et al.,
1997
Windham et
al., 2006
Study population
138 occupationally
exposed and
100 unexposed controls;
Exclusion criteria:
congenital color vision
loss, severe ocular
disease, significant vision
impairment, tainted
glasses or contact lenses,
diabetes mellitus,
neurological disease,
prior severe head or eye
injuries, alcohol abuse,
medication impairing
color vision.
Children born in 1994 in
San Francisco Bay Area
with Autism Spectrum
Disorders
(ASDs)(w = 284) and
controls (» = 657),
matched on basis of
gender and month of
birth.
Exposure assessment and
biomarkers
Solvents: TCE, PCE,
toluene, xylene; Historical
data on duration of
exposure protective
equipment use, subjective
evaluation of exposure,
nonoccupational solvent
exposure, solvent-related
symptoms at work, alcohol
and smoking, drug intake;
Mean urinary levels of
trichloroacetic acid: 1.55
(±1.75) mg/L.
Birth addresses were
geocoded and linked to
hazardous air pollutant
database; Exposure levels
assigned for 19 chemicals;
chemicals were grouped
based on mechanistic and
structural properties;
Summary index scores were
calculated; risk of ASD
calculated in upper quartiles
of groups or individual
chemical concentrations;
Adjustment for
demographic factors.
Tests used
Lanthony D15.
Archival data.
Statistics
Polytomous
logistic regression.
Pearson
correlation,
Logistic
Regression.
Results
Significant effect of age in exposed
group; With alcohol of <250 g/wk
no significant correlation between
color confusion and solvent
exposure; Significant interaction
between solvent exposure and
alcohol intake; Color Confusion
Index significantly higher in
exposed group with alcohol use of
>250 g/wk.
Elevated adjusted odds ratios for
ASD (by 50%) in top quartile of
chlorinated solvents, but not for
aromatic solvents; AOR for TCE in
4th quartile = 1.47; lessened when
adjusted for metals; correlation
between hydrocarbon and metals
exposures; when adjusted, increased
risk for metals (in 3rd quartile
= 1.95; in 4th quartile =1.7).
Contributing compounds: mercury,
cadmium, nickel, TCE, vinyl
chloride; Results interpreted to
suggest relationship between autism
and estimated metal and solvent
concentrations in air around place of
birth residence.
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
Epidemiological Studies: Controlled Exposure Studies; Neurological Effects of Trichloroethylene/Mixed Solvents
Levy et al.,
1981
9 participants (8 males
and 1 female) recruited
through newspaper ad;
8 h fasting before testing;
no control.
Experiment 1: alcohol
consumption (3 doses)—
blood alcohol levels were
measured with breath
analyzer pre (multiple
baselines) and post test
(multiple).
Experiment 2: Chloral
hydrate administered orally
over 2 mins in either 500
mg or 1,500 mg dose;
multiple baseline smooth
pursuit eye movement
(SPEM) tests and multiple
posttests after exposure; No
control dose administered.
SPEM tests of following a
sinusoidally oscillated
target at 0.4 Hz; eye
movements were recorded
through electrodes at each
eye.
t-tests; ANOVA.
Experiment 1: prealcohol all
subjects had intact SPEM; no
significant effect for 1.5 mL/kg of
alcohol; significant decline in
SPEM at 2.0 and 3.0 mL/kg alcohol;
significant dose-effect.
Experiment 2: at 500 mg. chloral
hydrate, no significant change in
pursuit was noted; at 1,500 mg
chloral hydrate, qualitative
disruptions in pursuit in all
participants (4); at 500 mg
participants observed to be drowsy;
When number reading was added
SPEM impairment was 'attenuated'
in both alcohol and chloral hydrate
conditions.
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table D-2. Epidemiological studies: Neurological effects of trichloroethylene/mixed solvents (continued)
Reference
Study population
Exposure assessment and
biomarkers
Tests used
Statistics
Results
^§
H I
O ^
HH Oq
H TO
O
Stopps and
McLaughlin,
1967
Chamber study using
2 healthy male
volunteers exposed to
Freon-113; 1 volunteer
exposed to TCE; No
control.
Exposure booth was
constructed; TCE in air:
TCE concentrations: 100,
200, 300, 400 ppm (1965
TLV: 100 ppm for 8-h
exposure) in ascending and
descending order; total time
in chamber: 2.75 h; Freon-
113 concentrations: 1,500,
2,500, 3,500, 4,500 ppm
(1965 TLV: 1,000 ppm for
8-h exposure), duration 1.5
h;
TCE: (1) reduction of
weight of compound during
exposure was calculated, (2)
continuous air sampling in
the chamber; Freon-113 in
air: (1) and (2) same; (3) gas
chromatography on air
captured in bottles sealed in
the chamber; no control
dose given.
Crawford Small Parts
Dexterity Test, Necker
Cube Test, Card Sorting,
Card Sorting with an
Auxiliary Task, Dial
Display (TCE participant
only); Short Employment
Test-Clerical (Freon-113
participants only).
Descriptive
statistics for air
measurement plots
by % of TCE
change in groups.
No TCE effect at 100 ppm, but test
performance deteriorated with
increase of TCE concentration; No
effect of Freon-113 on psychomotor
function at 1,500 ppm, deterioration
at 2,500 ppm, as concentration
increased, performance deteriorated.
CNS = central nervous system, EEG = electroencephalograph, PCE = perchloroethylene, WHO = World Health Organization.
H
W
-------
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
00 O
•••2
i
H S
§
H I
O >
HH Oq
H TO
O
Table D-3. Literature review of studies of TCE and domains assessed with neurobehavioral/neurological methods
Authors
ATSDR
Barret et al.
Barret et al.
Barrett, et al.
Burg, et al.
Burg and Gist
El Ghawabi et al.
Feldman et al.
Feldman et al.
Gamberale, et al.
Gash et al.
Grandjean et al.
Gun, et al.
Hirsch, et al.
Kilbum and
Thornton
Kilbum and
Warshaw
Kilbum
Kilbum
Konietzko, et al.
Kylin,etal. 1967
Landrigan, et al.
Liu, et al.
Mhiri et al.
Nagaya et al.
Year
2003
1984
1987
1982
1995
1999
1973
1988
1992
1976
2007
1955
1978
1996
1996
1993
2002a
2002b
1975
1967
1987
1988
2004
1990
Study
type
E
O
0
O
E
E
0
E
O
C
O
0
o
E
E
E
E
E
C
C
o
0
0
o
Participants no.
(N = exposed
C = nonexposed)
N= 116, C= 177
N= 188
N=104,C = 52
N= 11,C = 2
N = 4,281
N=3915
N=30,C = 30
N = 21,C = 27
N= 18, C = 30
N=15
N = 30
N = 80
N = 8, C = 8
N=106
N = 237, C = 264
N=544,C=181
N = 236, C = 228
N = 236,C = 58
N = 20
N=12
Residents and 12 W
N=103,C=111
N = 23, C = 23
N = 84, C = 83
Dur
C
C
C
C
C
C
C
C
A,C
A
C
C
C
C
C
C
C
C
A
A
A,C
C
A
C
PM/RT
ne
ne
ne
ne
ne
ne
ne
ne
ne
^
^
ne
^
ne
^
^
ne
(-)
ne
^
ne
ne
ne
ne
VM
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
^
ne
ne
ne
ne
ne
ne
ne
ne
Cogn
ne
ne
ne
ne
ne
ne
ne
ne
ne
^
ne
ne
^
ne
^
^
^
ne
ne
ne
^
ne
ne
ne
M&L
ne
ne
ne
ne
ne
ne
ne
ne
ne
(-)
ne
ne
ne
ne
ne
^
ne
ne
ne
ne
ne
^
ne
ne
M&P
ne
ne
^
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
^
ne
(-)
ne
ne
ne
ne
ne
ne
Sympf
ne
H,D
H, D, S, I
ne
ne
ne
H, S
ne
ne
ne
M,N
ne
ne
H
ne
M
M
ne
M
ne
H,D
D,N
ne
ne
Sentt
A
T,N, V
T,N
T
A,N
A,N
(-)
T
T,N
ne
N
N
ne
ne
T,N
B
ne
N
N
ne
N
T
N
Resp
ne
ne
ne
ne
V
V
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
Dose effect
vv
urinary
metabolitesV
ne
V
V
V
V
vv
V
ne
ne
ne
ne
v,vv
ne
ne
ne
ne
ne
ne
V
ne
vv
vv
v,w
V
TCE levels
0 — » 23 ppb in
dg water
150 ppm
ne
ne
ne
4gps:
2-75,000 ppb
165 ppm
ne
ne
540-1, 080 mg3
ne
6- 1,120 ppm
3-4 18 ppm
0-2,441 ppb
ne
6-500 ppb
6-500 ppb
0.2- 1,000 ppb
953 ppm
1,000 ppm
>1 83,000 ppb
1-100 ppm
ne
22 ppm
H
W
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
Table D-3. Literature review of studies of TCE and domains assessed with neurobehavioral/neurological methods
(continued)
Authors
Rasmussen and
Sabroe
Rasmussen et al.
Rasmussen et al.
Rasmussen et al.
Reifetal.
Ruijten, et al.
Smith
Stewart et al
Triebig, et al.
Triebig, et al.
Triebig, et al.
Triebig, et al.
Triebig, et al.
Troster and Ruff
Vemon and
Ferguson
Windemuller and
Ettema
Winneke
Year
1986
1993
1993
1993
2003
1991
1970
1970
1976
1977
1977
1982
1983
1990
1969
1978
1982
Study
type
O
0
O
0
E
O
0
c
c
c
o
0
o
0
c
c
0
Participants no. (N =
exposed
C = nonexposed)
N = 240, C = 350
N=96
N=96
N=99
N= 143
N=31,C = 28
N=130,C = 63
N=13
N = 7, C = 7
N = 7, C = 7
N=8
N = 24, C = 24
N = 66, C = 66
N = 3, C = 60
N = 8
N = 39
Not reported
Dur
C
C
C
c
c
c
c
A
A
A
A,C
C
C
A
A
A
ne
PM/RT
ne
ne
ne
V
V
V
ne
ne
ne
ne
ne
ne
ne
V
V
V
(-)
VM
ne
ne
V
ne
V
ne
ne
ne
ne
ne
V
ne
ne
V
V
ne
(-)
Cogn
ne
V
V
ne
ne
ne
ne
V
V
V
V
ne
ne
V
ne
ne
ne
M&L
ne
ne
ne
ne
ne
ne
ne
V
V
V
ne
ne
V
ne
ne
ne
M&P
V
ne
ne
ne
V
ne
ne
ne
V
V
ne
ne
ne
V
ne
ne
ne
Sympf
H,D, I, M
ne
ne
ne
M
ne
H,D
H
(-)
M
ne
ne
N,H
ne
ne
ne
ne
Sentt
ne
ne
ne
N
M
ne
N
ne
ne
(-)
ne
N
N
N
N
ne
ne
Resp
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
ne
Dose effect
vv
urinary
metabolitesV
ne
\W
\ \
vv
\w
V V
vv
ne
v,w
V
v,vv
V, W
V
v,w
V
ne
vv
ne
ne
TCE levels
ne
ne
ne
ne
5-15 ppb
17-70 ppm
ne
100-202 ppm
0-1 00 ppm
0-1 00 ppm
50 ppm
5-70 ppm
10-600 mg/m3
ne
0-1 000 ppm
200 ppm
50 ppm
2
•••2
i
^§
H I
O >
HH Oq
H TO
O
H
W
fH = Headaches; D = Dizziness; \ = Insomnia; S = Sex Probls; M = Mood; N = Neurological.
•f"f A = Audition; B = Balance; V = Vision; T = Trigeminal nerve; N = Other Neurological.
Study: C = Chamber; E = Environmental; O = Occupational.
Duration: A = Acute, C = Chronic.
V = positive findings; (-) = findings not significant; ne = not examined or reported; Dur = duration; PM/RT = psychomotor/reaction time; VM = visuo-motor; Cogn = cognitive;
M&L = memory and learning; M&P = mood and personality; Symp = symptoms; Sen = sensory; Resp = respiratory.
-------
D.2. CENTRAL NERVOUS TOXICITY IN ANIMAL STUDIES FOLLOWING
TRICHLOROETHYLENE (TCE) EXPOSURE
In vivo studies in animals and in vitro models have convincingly demonstrated that TCE
produces functional and physiological neurological changes. Overall, these effects collectively
indicate that TCE has central nervous system (CNS) depressant-like effects at lower exposures
and causes anesthetic-like effects at high exposures. Studies of TCE toxicity in animals have
generally not evaluated whether or not adverse effects seen acutely persist following exposure or
whether there are permanent effects of exposure. Exceptions to the focus on acute impairment
while under TCE intoxication include studies of hearing impairment and histopathological
investigations focused primarily on specific neurochemical pathways, hippocampal development,
and demyelination. These persistent TCE effects are discussed initially followed by the results
of studies that examined the acute effects of this agent. Summary tables for all the animal
studies are at the end of this section.
D.2.1. Alterations in Nerve Conduction
There is little evidence that TCE disrupts trigeminal nerve function in animal studies.
Two studies demonstrated TCE produces morphological changes in the trigeminal nerve at a
dose of 2,500 mg/kg-day for 10 weeks (Barret et al., 1991, 1992). However, dichloroacetylene, a
degradation product formed during the volatilization of TCE was found to produce more severe
morphological changes in the trigeminal nerve and at a lower dose of 17 mg/kg-day (Barret et
al., 1991, 1992). Only one study (Albee et al., 2006) has evaluated the effects of TCE on
trigeminal nerve function, and a subchronic inhalation exposure did not result in any significant
functional changes. A summary of these studies is provided in Table D-4.
Barret et al. (1991, 1992) conducted two studies evaluating the effects of both TCE and
dichloroacetylene on trigeminal nerve fiber diameter and internodal length as well as several
markers for fiber myelination. Female Sprague Dawley rats (n = 7/group) were dosed with
2,500 mg/kg TCE or 17 mg/kg-day dichloroacetylene by gavage for 5 days/week for 10 weeks.
These doses were selected based upon the ratio of the LD50s (dose at which there is 50%
lethality) for these two agents. Two days after administration of the last dose, a morphometric
approach was used to study the diameter of teased fibers from the trigeminal nerve. The fibers
were classified as Class A or Class B and evaluated for internode length and fiber diameter.
TCE-dosed animals only exhibited changes in the smaller Class A fibers where internode length
increased marginally (<2%) and fiber diameter increased by 6%. Conversely, dichloroacetylene-
treated rats exhibited significant and more robust decreases in internode length and fiber
diameter in both fiber classes A and B. Internode length decreased 8% in Class A fibers and 4%
in Class B fibers. Fiber diameter decreased 10% in Class A fibers and 6% in Class B fibers.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-90 DRAFT—DO NOT CITE OR QUOTE
-------
Biochemical data are presented for fatty acid composition from total lipid extractions from the
trigeminal nerve. These two studies identify a clear effect of dichloroacetylene on trigeminal
nerve fibers, but the effect by TCE is quite limited.
Albee et al. (2006) evaluated the effects of a subchronic inhalation TCE exposure in
Fischer 344 rats (10/sex/group). Rats were exposed to 0-, 250-, 800-, and 2,500-ppm TCE for
6 hours/day, 5 days/week for 13 weeks. At the eleventh week of exposure, rats were surgically
implanted with epidural electrodes over the somatosensory and cerebellar regions, and TSEPs
were collected 2-3 days following the last exposure. TSEPs were generated using subcutaneous
needle electrodes to stimulate the vibrissal pad (area above the nose). The resulting TSEP was
measured with electrode previously implanted over the somatosensory region. The TCE
exposures were adequate to produce permanent auditory impairment even though TSEPs were
unaffected. While TCE appears to be negative in disrupting the trigeminal nerve, the TCE
breakdown product, dichloroacetylene, does impair trigeminal nerve function.
Albee et al. (1997) reported that dichloroacetylene disrupted trigeminal nerve
somatosensory evoked potentials in Fischer 344 male rats. The subjects were exposed to a
mixture of 300-ppm dichloroacetylene, 900-ppm acetylene, and 170-ppm TCE for a single
2.25-hour period. This dichloroacetylene was generated by decomposing TCE in the presence of
potassium hydroxide and stabilizing with acetylene. A second treatment group was exposed to a
175-ppm TCE/l,030-ppm acetylene mix with no potassium hydroxide present. Therefore, no
dichloroacetylene was present in the second treatment group, providing an opportunity to
determine the effects on the trigeminal nerve somatosensory evoked potential in the absence of
dichloroacetylene. Evoked potentials from the dichloroacetylene/TCE/acetylene-exposed rats
were about 17% smaller measured between peaks I and II and 0.13 msec slower in comparison to
the preexposure measurements. Neither latency nor amplitude of this potential changed
significantly between the preexposure and postexposure test in the air-exposed animals (control).
The dichloroacetylene-mediated evoked potential changes persisted at least until Day 4
postexposure. No changes in evoked potentials were observed in the 175-ppm TCE/l,030-ppm
acetylene mix group. It is noteworthy that dichloroacetylene treatment produced broader
evidence of toxicity as witnessed by a persistent drop in body weight among subjects over the
7-day postexposure measuring period. In light of the differences observed between the effects of
TCE and dichloroacetylene on the trigeminal nerve, it would be instructive to calculate the dose
of TCE that would be necessary to produce comparable tissue levels of dichloroacetylene
produced in the Albee et al. (1997) study.
Kulig (1987) also measured peripheral (caudal nerve) nerve conduction time in male
Wistar rats and failed to show an effect of TCE with exposures as high as 1,500 ppm for
16 hours/day, 5 days/week for 18 weeks.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-91 DRAFT—DO NOT CITE OR QUOTE
-------
D.2.2. Auditory Effects
D.2.2.1. Inhalation
The ability of TCE to disrupt auditory function and produce inner ear histopathology
abnormalities has been demonstrated in several studies using a variety of test methods. Two
different laboratories have identified NOAELs for auditory function of 1,600 ppm following
inhalation exposure for 12 hours/day for 13 weeks in Long Evans rats (n = 6-10) (Rebert et al.,
1991) and 1,500 ppm in Wistar-derived rats (n = 12) exposed by inhalation for 18 hours/day,
5 days/week for 3 weeks (Jaspers et al., 1993). The LOAELs identified in these and similar
studies are 2,500-4,000-ppm TCE for periods of exposure ranging from 4 hours/day for 5 days
to 12 hours/day for 13 weeks (e.g., Muijser et al., 2000; Rebert et al., 1995, 1993; Crofton et al.,
1994; Crofton and Zhao, 1997; Fechter et al., 1998; Boyes et al., 2000; Albee et al., 2006).
Rebert et al. (1993) estimated acute blood TCE levels associated with permanent hearing
impairment at 125 ug/mL by methods that probably underestimated blood TCE values (rats were
anaesthetized using 60% carbon dioxide). A summary of these studies is presented in Table D-5.
Rebert et al. (1991) evaluated auditory function in male Long Evans rats (n = 10) and
F344 rats (n = 4-5) by measuring brainstem auditory-evoked responses (BAERs) following
stimulation with 4-, 8-, and 16-kHz sounds. The Long-Evans rats were exposed to 0-, 1,600-, or
3,200-ppm TCE, 12 hour/day for 12 weeks and the F344 rats were exposed to 0-, 2,000-, or
3,200-ppm TCE, 12 hours/day for 3 weeks. BAERs were measured every 3 weeks during the
exposure and then for an additional 6 weeks following the end of exposure. For the F344 rats,
both TCE exposure (2,000 and 3,200 ppm) significantly decreased BAER amplitudes at all
frequencies tested. In comparison, Long Evans rats exposed to 3,200-ppm TCE also had
significantly decreased BAER amplitude, but exposure to 1,600 ppm did not significantly affect
BAERs at any stimulus frequency. These data suggest a LOAEL at 2,000 ppm for the F344 rats
and a NOAEL at 1,600 ppm for the Long Evans rats. In subsequent studies, Rebert et al. (1993,
1995) again demonstrated TCE significantly decreases BAER amplitudes and significantly
increases the latency of the initial peak (identified as PI).
Jaspers et al. (1993) exposed Wi star-derived WAG-Rii/MBL rats (n = 12) to 0, 1,500 and
3,000-ppm TCE exposure for 18 hours/day, 5 days/week for 3 weeks. Auditory function for
each frequency was assessed by reflex modification (recording the decibel threshold required to
generate a startle response from the rat). Three tones (5, 20, and 35 kHz) were used to test
auditory function. The startle measurements were made prior to exposure and at 1, 3, 5, and
6 weeks after exposure. A selective impairment of auditory threshold for animals exposed to
3,000-ppm TCE was observed at all postexposure times at 20 kHz only. No significant effects
were noted in rats exposed to 1,500-ppm TCE. This auditory impairment was persistent up
through 6 weeks after exposure, which was the last time point presented. There was no
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-92 DRAFT—DO NOT CITE OR QUOTE
-------
impairment of hearing at either 5 or 25 kHz for animals exposed to 1,500- or 3,000-ppm TCE.
This study indicates TCE selectively produces a persistent mid-frequency hearing loss and
identifies a NOAEL of 1,500 ppm. Similarly, Crofton et al. (1994) exposed male Long Evans
rats (n = 7-8) to 3,500-ppm TCE, 8 hours/day for 5 days. Auditory thresholds were determined
by reflex modification audiometry 5-8 weeks after exposure. TCE produced a selective
impairment of auditory threshold for mid frequency tones, 8 and 16 kHz.
Muijser et al. (2000) evaluated the ability of TCE to potentiate the damaging effect of
noise on hearing. Wistar rats (n = 8 per group) were exposed by inhalation to 0 or 3,000-ppm
TCE alone for 18 hours/day, 5 days/week for 3 weeks (no noise) or in conjunction with 95-dB
broad band noise. The duration of noise exposure is not specified, but presumably was also
18 hours/day, 5 days/week for 3 weeks. Pure tone auditory thresholds were determined using
reflex modification audiometry 1 and 2 weeks following the exposures. Significant losses in
auditory sensitivity were observed for rats exposed to noise alone at 8, 16, and 20 kHz, for rats
exposed to TCE alone at 4, 8, 16, and 20 kHz and for combined exposure subjects at 4, 8, 16, 20,
and 24 kHz. The loss of hearing sensitivity at 4 kHz is particularly striking for the combined
exposure rats, suggesting a potentiation effect at this frequency. Impairment on this auditory test
suggests toxicity at the level of the cochlea or brainstem.
Fechter et al. (1998) exposed Long Evans rats inhalationally to 0 or 4,000-ppm TCE
6 hours/day for 5 days. Three weeks later auditory thresholds were assessed by reflex
modification audiometry (n = 12), and then 5-7 weeks later, cochlear function was assessed by
measuring compound action potentials (CAPs) and the cochlear microphonic response
(n = 3-10). Cochlear histopathology was assessed at 5-7 weeks (n = 4) using light microscopy.
Reflex modification thresholds were significantly elevated at 8 and 18 kHz, as were CAP
thresholds. The growth of the Nl evoked potential was reduced in the TCE group, and they
failed to show normal Nl amplitudes even at supra-threshold tone levels. There was no effect on
the sound level required to elicit a cochlear microphonic response of 1 uV. Histological data
suggest that TCE produces a loss of spiral ganglion cells.
Albee et al. (2006) exposed male and female F344 rats to TCE at 250, 800, or 2,500 ppm
for 6 hours/day, 5 days/week, for 13 weeks. At 2,500-ppm TCE, mild frequency-specific
hearing deficits were observed, including elevated tone-pip auditory brainstem response
thresholds. Focal loss of hair cells in the upper basal turn of the cochlea was observed in
2,500-ppm-exposed rats; this was apparently based upon midmodiolar sections, which lack
power in quantification of hair cell death. Except for the cochleas of 2,500-ppm-exposed rats, no
treatment-related lesions were noted during the neuro-histopathologic examination. The
NOAEL for this study was 800 ppm based on ototoxicity at 2,500 ppm.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-93 DRAFT—DO NOT CITE OR QUOTE
-------
The relationship between dose and duration of exposure with respect to producing
permanent auditory impairment was presented in Crofton and Zhao (1997) and again in Boyes et
al. (2000). The LOAELs identified in Long Evans rats (n = 10-12) were 6,000 ppm for a 1-day
exposure, 3,200 ppm per day for both the 1- and 4-week exposures, and 2,400 ppm per day for
the 13-week exposure. It was estimated from these data that the LOAEL for a 2-year long
exposure would be 2,100 ppm. Auditory thresholds were determined for a 16-kHz tone
3-5 weeks after exposure using reflex modification audiometry. Results replicated previous
findings of a hearing loss at 16 kHz for all exposure durations. One other conclusion reached by
this study is that TCE concentration and not concentration x duration of exposure is a better
predictor of auditory toxicity. That is, the notion that total exposure represented by the function,
concentration (C) x time (t), or Haber's law, is not supported. Therefore, higher exposure
concentrations for short durations are more likely to produce auditory impairment than are lower
concentrations for more protracted durations when total dosage is equated. Thus, consideration
needs to be given not only to total C x t, but also to peak TCE concentration.
Crofton and Zhao (1997) also presented a benchmark dose for which the calculated dose
of TCE would yield a 15-dB loss in auditory threshold. This benchmark response was selected
because a 15-dB threshold shift represents a significant loss in threshold sensitivity for humans.
The benchmark concentrations for a 15-dB threshold shift are 5,223 ppm for 1 day, 2,108 ppm
for 5 days, 1,418 ppm for 20 days, and 1,707 ppm for 65 days of exposure. While more sensitive
test methods might be used and other definitions of a benchmark effect chosen with a strong
rationale, these data provide useful guidance for exposure concentrations that do yield hearing
loss in rats.
These data demonstrate that the ototoxicity of TCE was less than that predicted by a strict
concentration x time relationship. These data also demonstrate that simple models of
extrapolation (i.e., C x t = k, Haber's Law) overestimate the potency of TCE when extrapolating
from short-duration to longer-duration exposures. Furthermore, these data suggest that, relative
to ambient or occupational exposures, the ototoxicity of TCE in the rat is a high-concentration
effect; however, the selection of a 15-dB threshold for detecting auditory impairment along with
tests at a single auditory frequency may not capture the most sensitive reliable measure of
hearing impairment.
With the exception of a single study performed in the Hartley guinea pig (n = 9-10;
Yamamura et al., 1983), there are no data in other laboratory animals related to TCE-induced
ototoxicity. Yamamura et al. (1983) exposed Hartley guinea pigs to TCE at doses of 6,000,
12,000, and 17,000 ppm for 4 hours/day for 5 days and failed to show an acute impairment of
auditory function. However, despite the negative finding in this study, it should be considered
that auditory testing was performed in the middle of a laboratory and not in an audiometric sound
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-94 DRAFT—DO NOT CITE OR QUOTE
-------
attenuating chamber. The influence of extraneous and uncontrolled noise on cochlear
electrophysiology is marked and assesses auditory detection thresholds in such an environment
unrealistic. Although the study has deficiencies, it is important to note that the guinea pig has
been reported to be far less sensitive than the rat to the effects of ototoxic aromatic hydrocarbons
such as toluene.
It may be helpful to recognize that the effects of TCE on auditory function in rats are
quite comparable to the effects of styrene (e.g., Pryor et al., 1987; Crofton et al., 1994; Campo et
al., 2006), toluene (e.g., Pryor et al., 1983; Campo et al., 1999) ethylbenzene (e.g., Cappaert et
al., 1999, 2000; Fechter et al., 2007), and/7-xylene (e.g., Pryor et al., 1987; Gagnaire et al.,
2001). All of these aromatic hydrocarbons produce reliable impairment at the peripheral
auditory apparatus (inner ear), and this impairment is associated with death of sensory receptor
cells, the outer hair cells. In comparing potency of these various agents to produce hearing loss,
it appears that TCE is approximately equipotent to toluene and less potent than, in order,
ethylbenzene, />-xylene, and styrene. Occupational epidemiological studies do appear to identify
auditory impairments in workers who are exposed to styrene (Sliwinska-Kowalska et al., 1999;
Morioka et al., 2000; Morata et al., 2002) and those exposed to toluene (Abbate et al., 1993;
Morata et al., 1997), particularly when noise is also present.
D.2.2.2. Oral and Injection Studies
No experiments were identified in which auditory function was assessed following TCE
administration by either oral or injection routes.
D.2.3. Vestibular System Studies
The effect of TCE on vestibular function was evaluated by either (1) promoting
nystagmus (vestibular system dysfunction) and comparing the level of effort required to achieve
nystagmus in the presence and absence of TCE or (2) using an elevated beam apparatus and
measuring the balance. Overall, it was found that TCE disrupts vestibular function as presented
below. Summary of these studies is found in Table D-6.
Tham et al. (1979, 1984) demonstrated disruption in the stimulated vestibular system in
rabbits and Sprague Dawley rats during intravenous (i.v.) infusion with TCE. It is difficult to
determine the dosage of TCE necessary to yield acute impairment of vestibular function since
testing was performed under continuing infusion of a lipid emulsion containing TCE, and
therefore, blood TCE levels were increasing during the course of the study. Tham et al. (1979),
for example, infused TCE at doses of 1-5 mg/kg/min reaching arterial blood concentrations as
high as 100 ppm. They noted increasing numbers of rabbits experiencing positional nystagmus
as blood TCE levels increased. The most sensitive rabbit showed nystagmus at a blood TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-95 DRAFT—DO NOT CITE OR QUOTE
-------
concentration of about 25 ppm. Similarly, the Sprague Dawley rats also experienced increased
nystagmus with a threshold effect level of 120 ppm as measured in arterial blood (Tham et al.,
1984). Animals demonstrated a complete recovery in vestibular function when evaluated for
nystagmus within 5-10 minutes after the i.v. infusion was stopped.
Niklasson et al. (1993) showed acute impairment of vestibular function in male and
female pigmented rats during acute inhalation exposure to TCE (2,700-7,200 ppm) and to
trichloroethane (500-2,000 ppm). Both of these agents were able to promote nystagmus during
optokinetic stimulation in a dose related manner. While there were no tests performed to assess
persistence of these effects, Tham et al. (1979, 1984) did find complete recovery of vestibular
function in rabbits (n = 19) and female Sprague-Dawley rats (n = 11) within minutes of
terminating a direct arterial infusion with TCE solution.
The finding that trichloroethylene can yield transient abnormalities in vestibular function
is not unique. Similar impairments have been shown for toluene, styrene, along with
trichloroethane (Niklasson et al., 1993) and by Tham et al. (1984) for a broad range of aromatic
hydrocarbons. The concentration of TCE in blood at which effects were observed for TCE
(0.9 mM/L) was quite close to that observed for most of these other vestibulo-active solvents.
D.2.4. Visual Effects
Changes in visual function have also been demonstrated in animal studies following acute
(Boyes et al., 2003, 2005) and subchronic exposure (Blain et al., 1994). Summary of all TCE
studies evaluating visual effects in animals can be found in Table D-6. In these studies, the
effect of TCE on visual-evoked responses to patterns (Boyes et al., 2003, 2005; Rebert et al.,
1991) or a flash stimulus (Rebert et al., 1991; Blain et al., 1994) were evaluated. Overall, the
studies demonstrated that exposure to TCE results in significant changes in the visual evoked
response, which is reversible once TCE exposure is stopped. Only one study (Rebert et al.,
1991) did not demonstrate changes in visual system function with a subchronic TCE exposure,
but visual testing was conducted 10 hours after each exposure.
Boyes et al. (2003, 2005) found significant reduction in the visual evoked potential
acutely while Long Evans male rats were being exposed to TCE concentrations of 500, 1,000,
2,000, 3,000, 4,000, and 5,000 ppm for intervals ranging from 4 to 0.5 hours, respectively. In
both instances, the degree of effect correlated more with brain TCE concentrations than with
duration of exposure.
Boyes et al. (2003) exposed adult, male Long-Evans rats to TCE in a head-only exposure
chamber while pattern onset/offset visual evoked potentials (VEPs) were recorded. Exposure
conditions were designed to provide C x t products of 0 ppm/hour (0 ppm for 4 hours) or
4,000 ppm/hour created through four exposure scenarios: 1,000 ppm for 4 hours; 2,000 ppm for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-96 DRAFT—DO NOT CITE OR QUOTE
-------
2 hours; 3,000 ppm for 1.3 hours; or 4,000 ppm for 1 hour (n = 9-10/concentration). Blood TCE
concentrations were assessed by GC with BCD, and brain TCE concentrations were estimated
using a physiologically based pharmacokinetic (PBPK) model. The amplitude of the VEP
frequency double component (F2) was decreased significantly (p < 0.05) by exposure. The mean
amplitude (+SEM in uV) of the F2 component in the control and treatment groups measured
4.4 + 0.5 (0 ppm/4 hours), 3.1 + 0.5 (1,000 ppm/4 hours), 3.1 + 0.4 (2,000 ppm/2 hours),
2.3 + 0.3 (3,000 ppm/1.3 hours), and 1.9 + 0.4 (4,000 ppm/1 hour). A PBPK model was used to
estimate the concentrations of TCE in the brain achieved during each exposure condition. The
F2 amplitude of the VEP decreased monotonically as a function of the estimated peak brain
concentration but was not related to the area under the curve of the brain TCE concentration.
These results indicate that an estimate of the brain TCE concentration at the time of VEP testing
predicted the effects of TCE across exposure concentrations and duration.
In a follow-up study, Boyes et al. (2005) exposed Long Evans male rats
(n = 8-10/concentration) to TCE exposures of 500 ppm for 4 hours, 1,000 ppm for 4 hours,
2,000 ppm for 2 hours, 3,000 ppm for 1.3 hours, 4,000 ppm for 1 hour and 5,000 ppm for
0.8 hour. VEP recordings were made at multiple time points, and their amplitudes were adjusted
in proportion to baseline VEP data for each subject. VEP amplitudes were depressed by TCE
exposure during the course of TCE exposure. The degree of VEP depression showed a high
correlation with the estimated brain TCE concentration for all levels of atmospheric TCE
exposure.
This transient effect of TCE on the peripheral visual system has also been reported by
Blain (1994) in which New Zealand albino rabbits were exposed by inhalation to 350- and
700-ppm TCE 4 hours/day, 4 days/week for 12 weeks. Electroretinograms (ERGs) and
oscillatory potentials (OPs) were recorded weekly under mesopic conditions. Recordings from
the 350- and 700-ppm exposed groups showed a significant increase in the amplitude of the a-
and b-waves (ERG). The increase in the a-wave was dose related increasing 30% at the low dose
and 84% in the high dose. For the b-wave, the lower exposure dose yielded a larger change from
baseline (52%) than did the high dose (33%). The amplitude of the OPs was significantly
decreased at 350 ppm (57%) and increased at 700 ppm (117%). The decrease in the oscillatory
potentials (OPs) shown in the low-dose group appears to be approximately 25% from
9-12 weeks of exposure. These electroretinal changes were reversed to the baseline value within
6 weeks after the inhalation stopped.
Rebert et al. (1991) evaluated visual evoked potentials (flash evoked potentials and
pattern reversal evoked potentials) in male Long Evans rats that received 1,600- or 3,200-ppm
TCE for 3 weeks 12 hours/day. No significant changes in flash evoked potential measurements
were reported following this exposure paradigm. Limited shifts in pattern reversal visual evoked
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-97 DRAFT—DO NOT CITE OR QUOTE
-------
potentials were reported during subchronic exposure, namely a reduction in the N1-P1 response
amplitude that reached statistical significance following 8, 11, and 14 weeks of exposure. The
drop in response amplitude ranged from approximately 20% after 8 weeks to nearly 50% at
Week 14. However, this potential recovered completely during the recovery period.
D.2.5. Cognitive Function
There have been a number of reports (e.g., Kjellstrand et al.,1980; Kulig, 1987; Kishi et
al., 1993) showing alteration in performance in learning tasks such as a change in speed to
complete the task, but little evidence that learning and memory function are themselves impaired
by exposure. Table D-7 presents the study summaries for animal studies evaluating cognitive
effects following TCE exposure. Such data are important in efforts to evaluate the functional
significance of decreases in myelinated fibers in the hippocampus reported by Isaacson et al.
(1990) and disruption of long-term potentiation discovered through in vitro testing (Ohta et al.,
2001) since the hippocampus has been closely tied to memory formation.
Kjellstrand et al. (1980) exposed Mongolian gerbils (n = 12/sex) to 900-ppm TCE by
inhalation for 9 months. Inhalation was continuous except for 1-2 hours/week for cage cleaning.
Spatial memory was tested using the radial arm maze task. In this task, the gerbils had to visit
each arm of the maze and remember which arm was visited and unvisited in selecting an arm to
visit. The gerbils received training and testing in a radial arm maze starting after 2 months of
TCE exposure. There was no effect of TCE on learning or performance on the radial arm maze
task.
Kishi et al. (1993) acutely exposed Wistar rats to TCE at concentrations of 250, 500,
1,000, 2,000, and 4,000 ppm for 4 hours. Rats were tested on an active (light) signaled shock
avoidance operant response. Rats exposed to 250-ppm TCE showed a significant decrease both
in the total number of lever presses and in avoidance responses at 140 minutes of exposure
compared with controls. The rats did not recover their pre-exposure performance until
140 minutes after the exhaustion of TCE vapor. Exposures in the range 250- to 2,000-ppm TCE
for 4 hours produced concentration related decreases in the avoidance response rate. No
apparent acceleration of the reaction time was seen during exposure to 1,000- or 2,000-ppm
TCE. The latency to a light signal was somewhat prolonged during the exposure to 2,000- to
4,000-ppm TCE. It is estimated that there was depression of the central nervous system with
slight performance decrements and the corresponding blood concentration was 40 ug/mL during
exposure. Depression of the central nervous system with anesthetic performance decrements
was produced by a blood TCE concentration of about 100 ug/mL. In general, they observed
dose related reductions in total number of lever presses, but these changes may be more
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-98 DRAFT—DO NOT CITE OR QUOTE
-------
indicative of impaired motor performance than of cognitive impairment. In any event, recovery
occurred rapidly once TCE exposure ceased.
Isaacson et al. (1990) studied the effects of oral TCE exposure in weanling rats at
exposure doses of 5.5 mg/day for 4 weeks, followed by an additional 2 weeks of exposure at
8.5 mg/day. No significant changes were observed in locomotor activity in comparison to the
control animals. This group actually reported improved performance on a Morris swim test of
spatial learning as reflected in a decrease in latency to find the platform from 14 seconds in
control subjects to 12 seconds in the lower dose TCE group to a latency of 9 seconds in the
higher TCE group. The high dose TCE group differed significantly from the control and low
TCE dose groups while these latter two groups did not differ significantly from each other. This
improvement relative to the control subjects occurred despite a loss in hippocampal myelination,
which approached 8% and was shown to be significant using Duncan's multiple range test.
Likewise, Umezu et al. (1997) exposed ICR strain male mice acutely to doses of TCE
ranging from 62.5-1,000 mg/kg depending upon the task. They reported a depressed rate of
operant responding in a conditioned avoidance task that reached significance with intraperitoneal
(i.p.) injections of 1,000 mg/kg. Increased responding during the signaled avoidance period at
lower doses (250 and 500 mg/kg) suggests an impairment in ability to inhibit responding or
failure to attend to the signal. However, all testing was performed under TCE intoxication.
D.2.6. Psychomotor Effects
Changes in psychomotor activity such as loss of righting reflex, functional observational
battery changes, and locomotor activity have been demonstrated in animals following exposure
to TCE. Summaries for some of these studies can be found below and are presented in detail in
Table D-8.
D.2.6.1. Loss of Righting Reflex
Kishi et al. (1993) evaluated the activity and performance of male Wistar rats in a series
of tasks following an acute 4-hour exposure to 250, 500, 1,000, 2,000, and 4,000 ppm. They
reported disruption in performance at the highest test levels with CNS depression and anesthetic
performance decrements. Blood TCE concentrations were about 100 ug/mL in Wistar rats (such
blood TCE concentrations were obtained at inhalation exposure levels of 2,000 ppm).
Umezu et al. (1997) studied disruption of the righting reflex following acute injection of
250, 500, 1,000, 2,000, 4000, and 5,000 mg/kg TCE in male ICR mice. At 2,000 mg/kg, loss of
righting reflex (LORR) was observed in only 2/10 animals injected. At 4,000 mg/kg,
9/10 animals experienced LORR, and 100% of the animals experienced LORR at 5,000 mg/kg.
Shih et al. (2001) reported impaired righting reflexes at exposure doses of 5,000 mg/kg in male
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-99 DRAFT—DO NOT CITE OR QUOTE
-------
Mfl mic although lower exposure doses were not included. They showed, in addition, that
pretreatment prior to TCE with DMSO or disulfiram (which is a CYP2E1 inhibitor) in DMSO
could delay loss of the righting reflex in a dose related manner. By contrast, the alcohol
dehydrogenase inhibitor, 4-metylpyradine did not delay loss of the righting reflex that resulted
from 5,000 mg/kg TCE. These data suggest that the anesthetic properties of TCE involve its
oxidation via CYP2E1 to an active metabolite, a finding that is consistent with the anesthetic
properties of chloral hydrate.
D.2.6.2. Functional Observational Battery (FOB) and Locomotor Activity Studies
D.2.6.2.1. Functional observational battery (FOB) and locomotor activity studies with
trichloroethylene (TCE). A number of papers have measured locomotor activity and used
functional observational batteries (FOBs) in order to obtain a more fine grained analysis of the
motor behaviors that are impaired by TCE exposure. While exposure to TCE has been shown
repeatedly to yield impairments in neuromuscular function acutely, there is very little evidence
that the effects persist beyond termination of exposure.
One of the most extensive evaluations of TCE on innate neurobehavior was conducted by
Moser et al. (1995, 2003) using FOB testing procedures. Moser et al. (1995) evaluated the
effects of acute and subacute (14-day) oral gavage administration of TCE in adult female Fischer
344 rats. Testing was performed both 4 hours post TCE administration and 24 hours after TCE
exposure, and a comparison of these two time points along with comparison between the first
day and the last day of exposure provides insight into the persistence of effects observed.
Various outcome measures were grouped into five domains: autonomic, activity, excitability,
neuromuscular, and sensorimotor. Examples of tests included in each of these groupings are as
follows: Autonomic—lacrimation, salivation, palpebral closure, pupil response, urination, and
defecation; Activity—rearing, motor activity counts home cage position. Excitability—ease of
removal, handling reactivity, arousal, clonic, and tonic movements; and Neuromuscular—gait
score, righting reflex, fore and hindlimb grip strength, and landing foot splay. Sensorimotor-tail-
pinch response, click response, touch response, and approach response. Scoring was performed
on a 4-point scale ranging from "1" (normal) to "4" (rare occurrence for control subjects). In the
acute exposure, the exposure doses utilized were 150, 500, 1,500, and 5,000 mg/kg TCE in corn
oil. These doses represent 3, 10, 30, and 56% of the limit dose. For the 14-day subacute
exposure, the doses used were 50, 150, 500, and 1,500 mg/kg. Such doses represent 1, 3, 10, and
30% of the limit dose for TCE.
The main finding for acute TCE administration is that a significant reduction in activity
level occurred after the highest dose of TCE (5,000 mg/kg) only. This effect showed substantial
recovery 24 hours after exposure though residual decrements in activity were noted.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-100 DRAFT—DO NOT CITE OR QUOTE
-------
Neuromuscular function as reflected in the gait score was also severely affected only at
5,000-mg/kg dose and only at the 4-hour test period. Sensorimotor function reflected in response
to a sudden click, was abnormal at both 1,500 and 5,000 mg/kg with a slight difference observed
at 1,500 mg/kg and a robust difference apparent at 5,000 mg/kg. Additional effects noted, but
not shown quantitatively were abnormal home-cage posture, increased landing foot splay,
impaired righting and decreased fore and hind limb grip strength. It is uncertain at which doses
such effects were observed.
With the exception of sensorimotor function, these same categories were also disrupted in
the subacute TCE administration portion of the study. The lack of effect of TCE on
sensorimotor function with repeated TCE dosing might reflect either habituation, tolerance, or an
unreliable measurement at one of the time points. Given the absence of effect at a range of
exposure doses, a true dose-response relationship cannot be developed from these data.
In the subacute study, there are no clearly reliable dose-related differences observed
between treated and control subjects. Rearing, a contributor to the activity domain, was elevated
in the 500-mg/kg dose group, but was normal in the 1,500-mg/kg group. The neuromuscular
domain was noted as significantly affected at 15 days, but it is not clear which subtest was
abnormal. It appears that the limited group differences may be random among subjects unrelated
to exposure condition.
In a follow-up study, Moser et al. (2003) treated female Fischer 344 rats with TCE by
oral gavage for periods of 10 days at doses of 0, 40, 200, 800, and 1,200 mg/kg/d, and testing
was undertaken either 4 hours following the first or 10th dose as well as 24 hours after these two
time points. The authors identified several significant effects produced by TCE administration
including a decrease in motor activity, tail pinch responsiveness, reactivity to handling, hind limb
grip strength, and body weight. Rats administered TCE also showed significantly more
piloerection, higher gait scores, lethality, body weight loss, and lacrimation compared to
controls. Only effects observed 4 hours after the 10th exposure dose were presented by the
authors, and no quantitative information of these measurements is provided.
Albee et al. (2006) exposed male and female Fischer 344 rats to 250-, 800-, and
2,500-ppm TCE for 6 hours/day, 5 days/week for 13 weeks. FOB was performed 4 days prior to
exposure and then monthly. Auditory impairments found by others (e.g., Muijser et al., 2000;
Rebert et al., 1995; Crofton et al., 1994; Crofton and Zhao, 1997; Fechter et al., 1998; Boyes et
al., 2000) were replicated at the highest exposure dose, but treatment related differences in grip
strength or landing foot splay were not demonstrated. The authors report slight increases in
handling reactivity among female rats and slightly more activity than in controls at an
intermediate time point, but apparently did not conduct systematic statistical analyses of these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-101 DRAFT—DO NOT CITE OR QUOTE
-------
observations. In any event, there were no statistically significant effects on activity or reactivity
by the end of exposure.
Kulig (1987) also failed to show significant effects of TCE inhalation exposure on
markers of motor behavior. Wistar rats exposed to 500, 1,000, and 1,500 ppm for 16 hours/day,
5 days/week for 18 weeks failed to show changes in spontaneous activity, grip strength, or
coordinated hind limb movement. Measurements were made every three weeks during the
exposure period and occurred between 45 minutes and 180 minutes following the previous TCE
inhalation exposure. This study establishes a NOAEL of 1,500-ppm TCE with an exposure
duration of 16 hours/day.
D.2.6.2.2. Acute and subacute oral exposure to dichloroacetic acid on functional
observational batteries (FOB). Moser et al. (1999) conducted a series of experiments on DC A
ranging from acute to chronic exposures. The exposure doses used in the acute experiment were
100, 300, 1,000, and 2,000 mg/kg. In the repeated exposure studies (8 weeks-24 months), doses
varied between 16 and 1,000 mg/kg/d. The authors showed pronounced neuromuscular changes
in Long Evans and F344 rats dosed orally with the TCE metabolite, DC A, over a period ranging
from 9 weeks to 24 months at different exposure doses. Using a multitude of exposure protocols
which most commonly entailed daily exposures to DCA either by gavage or drinking water the
authors identify effects that were "mostly limited" to the neuromuscular domain. These included
disorders of gait, grip strength, foot splay and righting reflex that are dose and duration
dependent. Data on gait abnormality and grip strength are presented in greatest detail. In adults
exposed to DCA by gavage, gait scores were "somewhat abnormal" at the 7-week test in both the
adult Long Evans rats receiving 300 and those receiving 1,000 mg/kg/d. There was no adverse
effect in the rats receiving 100 mg/kg/d. In the chronic study, which entailed intake of DCA via
drinking water yielding an estimated daily dose of 137 and 235 mg/kg/d "moderately to severely
abnormal" gait was observed within 2 months of exposure and dosing was either reduced or
discontinued because of the severity of toxicity. For the higher DCA dose, gait scores remained
"severely abnormal" at the 24-month test time even though the DCA had been discontinued at
the 6-month test time. Hindlimb grip strength was reduced to about Va the control value in both
exposure doses and remained reduced throughout the 24 months of testing even though DCA
administration ceased at 6 months for the 235 mg/kg/d group. Forelimb grip strength showed a
smaller and apparently reversible effect among DCA treated rats.
D.2.6.3. Locomotor Activity
Wolff and Siegmund (1978) administered 182 mg/kg TCE (i.p.) in AB mice and
observed a decrease in spontaneous locomotor activity. In this study, AB mice were injected
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-102 DRAFT—DO NOT CITE OR QUOTE
-------
with TCE 30 minutes prior to testing for spontaneous activity at one of 4 time points during a
24 hours/day (0600, 1200, 1800, and 2400 hours). Marked decreases (estimated 60-80% lower
than control mice) in locomotor activity were reported in 15-minute test periods. The reduction
in locomotion was particularly profound at all time intervals save for the onset of light (0600).
Nevertheless, even at this early morning time point, activity was markedly reduced from control
levels (60% lower than controls as approximated from a graph).
Moser et al. (1995, 2003) included locomotor activity as one of their measures of
neurobehavioral effects of TCE given by gavage over a 10-14 day period. In the 1995 paper,
female Fischer 344 rats were dosed either acutely with 150, 500, 1,500 or 5,000 mg/kg TCE or
for 14 days with 50, 150, 500 or 1,500 mg/kg. In terms of the locomotor effects, they report that
acute exposure produced impaired locomotor scores only at 5,000 mg/kg while in the subacute
study, locomotion was impaired at the 500 mg/kg dose, but not at the 1,500 mg/kg dose. In the
Moser (2003) study, it appears that 200 mg/kg TCE may actually have increased locomotor
activity while the higher test doses (800 and 1,200 mg/kg) decreased activity in a dose related
manner. What is common to both studies, however, is a depression in motor activity that occurs
acutely following TCE administration and which may speak to the anesthetic if not central
nervous system depressive effects of this solvent.
There are also a number of reports (Waseem et al., 2001; Fredriksson et al., 1993; Kulig,
1987) that failed to demonstrate impairment of motor activity or ability following TCE exposure.
Waseem et al. (2001) failed to show effects of TCE given in the drinking water of Wistar rats
over the course of a 90 day trial. While nominal solvent levels were 350, 700, and 1,400 ppm in
the water, no estimate is provided of daily TCE intake or of the stability of the TCE solution over
time. However, assuming a daily water intake of 25 mL/day and body weight of 330 g, these
exposures would be estimated to be approximately 26, 52, and 105 mg/kg. These doses are far
lower than those studied by Moser and colleagues.
Fredriksson et al. (1993) studied the effects of TCE given by oral gavage to male NMRI
mice at doses of 50 and 290 mg/kg/d from postnatal Day 10-16 on locomotion assessed either
on the day following exposure or at age 60 days. They found no significant effect of TCE on
locomotor activity and no consistent effects on other motor behaviors (e.g., rearing).
Waseem et al. (2001) studied locomotor activity in Wistar rats exposed for up to 180 days
to 376-ppm TCE by inhalation for 4 hours/day, 5 days/week and acutely intoxicated with TCE.
Here the authors report seemingly inconsistent effects of TCE on locomotion. After 30 days of
exposure, the treated rats show an increase in locomotor activity relative to control subjects.
However, after 60 days of exposure they note a significant increase in distance traveled found
among experimental subjects, but a decrease in horizontal activity in this experimental group.
Moreover, the control subjects vary substantially in horizontal counts among the different time
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-103 DRAFT—DO NOT CITE OR QUOTE
-------
periods. No differences between the treatment groups are found after 180 days of exposure. It is
difficult to understand the apparent discrepancy in results reported at 60 days of exposure.
D.2.7. Sleep and Mood Disorders
D.2.7.1. Effects on Mood: Laboratory Animal Findings
It is difficult to obtain comparable data of emotionality in laboratory studies. However,
Moser et al. (2003) and Albee et al. (2006) both report increases in handling reactivity among
rats exposed to TCE. In the Moser study, female Fischer 344 rats received TCE by oral gavage
for periods of 10 days at doses of 0, 40, 200, 800, and 1,200 mg/kg/d while Albee et al. (2006)
exposed Fischer 344 rats to TCE by inhalation at exposure doses of 250, 800, and 2,500 ppm for
6 hours/day, 5 days/week for 13 weeks.
D.2.7.2. Sleep Disturbances
Arito et al. (1994) exposed male Wistar rats to 50-, 100-, and 300-ppm TCE for
8 hours/day, 5 days/week for 6 weeks and measured electroencephalographic (EEG) responses.
EEG responses were used as a measure to determine the number of awake (wakefulness hours)
and sleep hours. Exposure to all the TCE levels significantly decreased amount of time spent in
wakefulness during the exposure period. Some carry over was observed in the 22-hour
postexposure period with significant decreases in wakefulness seen at 100-ppm TCE.
Significant changes in wakefulness-sleep elicited by the long-term exposure appeared at lower
exposure levels. These data seem to identify a low dose of TCE that has anesthetic properties
and established a LOAEL of 50 ppm for sleep changes.
D.2.8. Mechanistic Studies
D.2.8.1. Dopaminergic (DA) Neurons
In two separate animal studies, subchronic administration of TCE has resulted in a decrease
of dopaminergic (DA) cells in both rats and mice. Although the mechanism for DA neurons
resulting from TCE exposure is not elucidated, disruption of DA-containing neurons has been
extensively studied with respect to Parkinson's Disease and parkinsonism. In addition to
Parkinson's Disease, significant study of MPTP and of high-dose manganese toxicity provides
strong evidence for extrapyramidal motor dysfunction accompanying loss of dopamine neurons in
the substantia nigra. These databases may provide useful comparisons to the highly limited
database with regard to TCE and dopamine neuron effects. The studies are presented in Table D-9.
Gash et al. (2007) assessed the effects of subchronic TCE administration on
dopaminergic neurons in the central nervous system. Fischer 344 male rats were orally
administered by gavage 1,000 mg/kg TCE in olive oil, 5 days/week for 6 weeks. Degenerative
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-104 DRAFT—DO NOT CITE OR QUOTE
-------
changes in DA containing neurons in the substantia nigra were reported as indexed by a 45%
decrease in the number of tyrosine hydroxylase positive cells. Additionally, there was a decrease
in the ratio of 3,4-dihydroxyphenylacetic acid, a metabolite of DA, to DA levels in the striatum.
This shift in ratio, on the order of 35%, was significant by Student's t-test, suggesting a decrease
in release and utilization of this neurotransmitter. While it is possible that long-term adaptation
might occur with regard to release rates for DA, the loss of DA cells in the substantia nigra is
viewed as a permanent toxic effect. The exposure level used in this study was limited to one
high dose and more confidence in the outcome will depend upon replication and development of
a dose-response relationship. If the results are replicated, they might be important in
understanding mechanisms by which TCE produces neurotoxicity in the central nervous system.
The functional significance of such cellular loss has not yet been determined through behavioral
testing.
Guehl (1999) also reported persistent effects of TCE exposure on DA neurons. In this
study, OF1 male mice (n = 10) were injected i.p. daily for 5 days/week for 4 weeks with TCE
(400 mg/kg/d). Following a 7 day period when the subjects did not receive TCE, the mice were
euthanized and tyrosine hydroxylase immunoreactivity was used to measure neuronal death in
the substantia nigra pars compacta. Treated mice presented significant dopaminergic neuronal
death (50%) in comparison with control mice based upon total cell counts conducted by an
examiner blinded as to treatment group in six samples per subject. The statistical comparison
appears to be by Student's t-test (only means, standard deviations, and a probability ofp < 0.001
are reported). While this study appears to be consistent with that of Gash et al. (2007) there are
some limitations of this study. Specifically, no photomicrographs are provided to assess
adequacy of the histopathological material. Additionally, no dose-response data are available to
characterize dose-response relationships or identify either a benchmark dose or NOAEL.
Behavioral assessment aimed at determining functional significance was not determined.
The importance of these two studies suggesting death of dopaminergic neurons following
TCE exposure may be addressable by human health studies because they suggest the potential
for TCE to produce a parkinsonian syndrome.
D.2.8.2. Gamma-Amino Butyric Acid (GABA) and Glutamatergic Neurons
Disruption of GABAergic and glutamatergic neurons by toxicants can represent serious
impairment as gamma-amino butyric acid (GABA) serves as a key inhibitory neurotransmitter
while glutamate is equally important as an excitatory neurotoxicant. Moreover, elevations in
glutamatergic release have been identified as an important process by which more general
neurotoxicity can occur through a process identified as excitotoxicity. The data with regard to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-105 DRAFT—DO NOT CITE OR QUOTE
-------
TCE exposure and alteration in GABA and glutamate function is limited. The studies are
presented in Table D-10.
Briving et al. (1986) conducted a chronic inhalation exposure in Mongolian gerbils to
50-and 150-ppm TCE continuously for 12 months and reported the changes in amino acids levels
in the hippocampus and cerebellar vermis and on high affinity uptake of GABA and glutamate in
those same structures. A dose related elevation of glutamine in the hippocampus of
approximately 20% at 150 ppm was reported, but no other reliable changes in amino acids in
either of these two structures. With regard to high affinity uptake of glutamate and GABA, there
were no differences in the hippocampal uptake between control and treated gerbils although in
the cerebellar vermis there was a dose related elevation in the high affinity uptake for both of
these neurotransmitter. Glutamate uptake was increased about 50% at 50 ppm and 100% at
150 ppm. The corresponding increases for GABA were 69% and 74%. Since control tissue
uptake is identified as being 100% rather than as an absolute rate, the ability to assess quality of
the control data are limited. It is unclear if this finding in cerebellar vermis is also present in
other brain tissues and should be studied further. If these findings are reliable, the changes in
high affinity uptake in cerebellum for GABA and glutamate might represent alterations that
could have functional outcomes. For example, alteration in GABA release and reuptake from the
cerebellum might be consistent with acute alteration in vestibular function described below.
However, there are presently no compelling data to support such a relationship.
The change in hippocampal glutamine levels is not readily interpretable. What is not
clear from this paper is whether the alterations observed were acute effects observable only while
subjects were intoxicated with TCE or whether they would persist once TCE had been removed
from the neural tissue. This study used inhalation doses that were at least 1 order of magnitude
lower than those required to produce auditory impairment.
A study by Shih et al. (2001) provides indirect evidence in male Mfl mice that TCE
exposure by injection might alter GABAergic function. The mice were injected i.p. with 250,
500, 1,000 and 2,000 mg/kg TCE in corn oil and the effect of these treatments on susceptibility
to seizure induced by a variety of drugs was observed. Shih et al. report that doses of TCE as
low as 250 mg/kg could reduce signs of seizure induced by picrotoxin, bicuculline, and
pentylenetetrazol. These drugs are all GABAergic antagonists. TCE treatment had a more
limited effect on seizure threshold induced by non-GABAergic convulsant drugs such as
strychnine (glycine receptor antagonist), 4-aminopyridine (alcohol dehydrogenase inhibitor) and
N-methyl-d-aspartate (glutamatergic agonist) than was observed with the GABAergic
antagonists. While these data suggest the possibility that TCE could act at least acutely on
GABAergic neurons, there are no direct measurements of such an effect. Moreover, there is no
obvious relationship between these findings and those of Briving et al. (1986) with regard to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-106 DRAFT—DO NOT CITE OR QUOTE
-------
increased high affinity uptake of glutamate and GAB A in cerebellum. Beyond that fact, this
study does not provide information regarding persistent effects of TCE on either seizure
susceptibility or GAB Aergic function as all measurements were made acutely shortly following a
single injection of TCE.
D.2.8.3. Demyelination Following Trichloroethylene (TCE) Exposure
Because of its anesthetic properties and lipophilicity, it is hypothesized that TCE may
disrupt the lipid-rich sheaths that cover many central and peripheral nerves. This issue has also
been studied both in specific cranial nerves known to be targets of TCE neurotoxicity (namely
the trigeminal nerve) and in the central nervous system including the cerebral cortex,
hippocampus and cerebellum in particular. For peripheral and cranial nerves, there are limited
nerve conduction velocity studies that are relevant as a functional measure. For central
pathways, the most common outcomes studied include histological endpoints and lipid profiles.
A significant difficulty in assessing these studies concerns the permanence or persistence
of effect. There is a very large literature unrelated to TCE, which demonstrates the potential for
repair of the myelin sheath and at least partial if not full recovery of function. In the studies
where nerve myelin markers are assessed, it is not possible to determine if the effects are
transient or persistent.
There are two published manuscripts (Isaacson and Taylor, 1989; Isaacson et al., 1990)
that document selective hippocampal histopathology when Sprague Dawley rats are exposed to
TCE within a developmental model. Both of these studies employed oral TCE administration
via the drinking water. In Isaacson and Taylor, (1989), a combined prenatal and neonatal
exposure was used while Isaacson's et al. (1990) report focused on a neonatal exposure. In
addition, Ohta et al. (2001) presented evidence of altered hippocampal function in an in vitro
preparation following acute in vivo TCE intoxication. The latter most manuscript details a shift
in long term potentiation elicited by tetanic shocks to hippocampal slices in vitro. In the two
developmental studies the exposure doses are expressed in terms of the concentration of TCE
placed in the drinking water and the total daily dose is then estimated based upon average water
intake by the subjects. However, since the subjects' body weight is not provided, it is not
possible to estimate dosage on a mg/kg body weight basis.
Isaacson and Taylor (1989) examined the development of the hippocampus in neonatal
rats that were exposed in utero and in the preweaning period to TCE via their dam. TCE was
added to the drinking water of the dam and daily maternal doses are estimated based upon water
intake of the dam as being 4 and 8.1 mg/day. Based upon body weight norms for 70-day old
female Sprague Dawley rats, which would predict body weights of about 250 g at that age, such
a dose might approach 16-32 mg/kg/d initially during pregnancy. Even if these assumptions
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-107 DRAFT—DO NOT CITE OR QUOTE
-------
hold true, it is not possible to determine how much TCE was received by the pups although the
authors do provide an estimate of fetal exposure expressed as ug/mL of TCE, trichloroethanol,
and trichloroacetic acid. The authors reported a 40% decline in myelinated fibers in the CA1
region of the hippocampus of the weanling rats. There was no effect of TCE treatment on
myelination in several other brain regions including the internal capsule, optic tract or fornix and
this effect appears to be restricted to the CA1 region of the hippocampus at the tested exposures.
In a second manuscript by that group (Isaacson et al., 1990), weanling rats were exposed
to TCE via their drinking water at doses of 5.5 mg/day for 4 weeks or 5.5 mg/day for 4 weeks, a
2 week period with no TCE and then a final 2 weeks of exposure to 8.5 mg/day TCE. Spatial
learning was studied using the Morris water maze and hippocampal myelination was examined
histologically starting 1 day postexposure. The authors report that the subjects receiving a total
of 6 weeks exposure to TCE showed better performance in the Morris swim test (p < 0.05) than
did controls while the 4 week exposed subjects performed at the same level as did controls.
Despite this apparent improvement in performance, histological examination of the hippocampus
demonstrated a dose dependent relationship with hippocampal myelin being significantly
reduced in the TCE exposed groups while normal myelin patterns were found in the internal
capsule, optic tract and fornix. The authors did not evaluate the signs of gross toxicity in treated
animals such as growth rate, which might have influenced hippocampal development.
Ohta et al. (2001) administered 300 or 1,000 mg/kg TCE, i.p., to male ddY mice.
Twenty-four hours after TCE administration, the mice were sacrificed and hippocampal sections
were prepared from the excised brains and long term potentiation was measured in the slices. A
dose related reduction in the population spike was observed following a tetanic stimulation
relative to the size of the population spike elicited in the TCE mice prior to tetany. The spike
amplitude was reduced 14% in the 300 mg/kg TCE group and 26% in the 1,000 mg/kg group.
Precisely how such a shift in excitability of hippocampal CA1 neurons relates to altered
hippocampal function is not certain, but it does demonstrate that injection with 300 mg/kg TCE
can have lingering consequences on the hippocampus at least 24 hours following i.p.
administration.
A critical area for future study is the potential that TCE might have to produce
demyelination in the central nervous system. While it is realistic to imagine that an anesthetic
and lipophilic agent such as TCE might interact with lipid membranes and produce alterations,
for example, in membrane fluidity at least at anesthetic levels, the data collected by Kyrklund
and colleagues suggest that low doses of TCE (50 and 150 ppm chronically for 12 months,
320 ppm for 90 days, 510 ppm 8 hours/day for 5 months) might alter fatty acid metabolism in
Sprague Dawley rats and Mongolian gerbils. Because they have not included high doses in their
studies and because the low doses produce only sporadic significant effects and these tend to be
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-108 DRAFT—DO NOT CITE OR QUOTE
-------
of very small magnitude (5-10%) it is not certain that they are truly observing events with
biological significance or whether they are observing random effects. A key problem in
determining whether the effects under study are spurious or are due to ongoing exposure is that
the magnitude and direction of the effect does not grow larger as exposure continues. It could be
hypothesized that the alterations in fatty acid metabolism could be an underlying mechanism for
demyelination. However, there is not enough evidence to determine if the changes in the lipid
profiles lead to demyelination or if the observed effects are purely due to chance. Similarly, the
size of statistically significant effects (5-12%) is generally modest. A broad dose-response
analysis or the addition of a positive control group that is treated with an agent well-known to
produce central demyelination would be important in order to characterize the potency of TCE as
an agent that disrupts central nervous system lipid profiles.
Kyrklund and colleagues (e.g., 1986) have generally evaluated the hippocampus, cerebral
cortex, cerebellum, and in some instances brainstem in adult gerbil. It is not apparent that one
brain region is more vulnerable to the effects of TCE than is another region. While this group
does not report significant changes in levels of cholesterol, neutral and acidic phospholipids or
total lipid phospholipids, they do suggest a shift in lipid profiles between treated and untreated
subjects. Similarly, inhalation exposure to trichloroethane at 1,200 ppm for 30 days (Kyrklund
and Haglid, 1991) leads to sporadic changes in fatty acid profiles in Sprague Dawley rats.
However, these changes are small and are not always in the same direction as the changes
observed following trichloroethylene exposure. In the case of trichloroethane, a NOAEL of
320 ppm for 30 days 24 hours/day was observed and no other doses were evaluated (Kyrklund et
al., 1988).
D.2.9. Summary Tables
Tables D-4 through D-8 summarize the animal studies by neurological domains
(Table D-4—trigeminal nerve; Table D-5—ototoxicity; Table D-6—vestibular and visual
systems; Table D-7—cognition; and Table D-8—psychomotor function and locomotor activity).
For each table, the reference, exposure route, species, dose level, effects and NOAEL/LOAEL
are provided. Tables D-9 through D-l 1 summarize mechanistic (Tables D-9 and D-l 1) and
neurochemical studies (Table D-10). Brief summaries of developmental neurotoxicity studies
are provided in Table D-12.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-109 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-4. Summary of mammalian in vivo trigeminal nerve studies
Reference
Barret et
al., 1991
Barret et
al., 1992
Albee et
al., 2006
Exposure route
Direct Gastric
Administration
Direct Gastric
Administration
Inhalation
Species, strain,
sex, number
Rat,
Sprague-Dawley,
female, 21
Rat,
Sprague-Dawley,
female, 18
Rat, Fischer 344,
male and female,
10/sex/group
Dose level/
exposure
duration
0, 2.5 g/kg,
acute
administration
0, 2.5 g/kg;
1 dose/d,
5 d/wk, 10 wks
0, 250, 800,
2,500 ppm
NOAEL:
LOAEL
LOAEL:
2.5 g/kg
LOAEL:
2.5 g/kg
NOAEL:
2,500 ppm
Effects
Morphometric analysis was
used for analyzing the
trigeminal nerve. Increase in
external and internal fiber
diameter as well as myelin
thickness was observed in the
trigeminal nerve after TCE
treatment.
Trigeminal nerve analyzed
using morphometric analysis.
Increased internode length and
fiber diameter in class A fibers
of the trigeminal nerve
observed with TCE treatment.
Changes in fatty acid
composition also noted.
No effect on trigeminal nerve
function was noted at any
exposure level.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l 10 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-5. Summary of mammalian in vivo ototoxicity studies
Reference
Rebert et
al., 1991
Rebert et
al., 1993
Rebert et
al., 1995
Crofton et
al., 1994
Crofton
and Zhao,
1997;
Boyes et
al., 2000
Exposure
route
Inhalation
Inhalation
Species, strain,
sex, number
Rat, Long Evans,
male, 10/group
Rat, F344, male,
4-5/group
Rat, Long Evans,
male, 9/group
Rat, Long Evans,
male, 9/group
Rat, Long Evans,
male, 7-8/group
Rat, Long Evans,
male, 9-12/group
Rat, Long Evans,
male, 8-10/group
Rat, Long Evans,
male, 8-10/group
Rat, Long Evans,
male, 8-10/group
Dose level/
exposure
duration
Long Evans: 0,
1,600,
3,200 ppm;
12 h/d, 12 wk
F344: 0, 2000,
3200 ppm;
12 h/d, 3 wk
0, 2,500, 3,000,
3,500 ppm; 8 h/d,
5d
0, 2,800 ppm;
8 h/d, 5 d
0, 3,500 ppm
TCE; 8 h/d, 5 d
0, 4,000, 6,000,
8,000 ppm; 6 h
0, 1,600, 2,400,
3,200 ppm; 6 h/d,
5d
0, 800, 1,600,
2,400, 3,200
ppm; 6 h/d,
5 d/wk, 4 wk
0, 800, 1,600,
2,400, 3,200
ppm; 6 h/d,
5 d/wk, 13 wk
NOAEL;
LOAEL
Long Evans:
NOAEL:
1,600 ppm;
LOAEL:
3,200 ppm
F344:
LOAEL:
2,000 ppm
NOAEL:
2,500 ppm
LOAEL:
3,000 ppm
LOAEL:
2,800 ppm
LOAEL:
3,500 ppm
NOAEL:
6,000 ppm
LOAEL:
8,000 ppm
NOAEL:
2,400 ppm
LOAEL:
3,200 ppm
NOAEL:
2,400 ppm
LOAEL:
3,200 ppm
NOAEL:
1,600 ppm
LOAEL:
2,400 ppm
Effects
BAERs were measured.
Significant decreases in B AER
amplitude and an increase in
latency of appearance of the
initial peak (PI).
BAERs were measured 1-2 wk
postexposure to assess auditory
function. Significant decreases
in BAERs were noted with TCE
exposure.
BAER measured 2-14 d
postexposure at a 16-kHz tone.
Hearing loss ranged from
55-85 dB.
BAER measured and auditory
thresholds determined 5-8 wk
postexposure. Selective
impairment of auditory function
for mid-frequency tones (8 and
16 kHz).
Auditory thresholds as measured
by BAERs for the 16-kHz tone
increased with TCE exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 11 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-5. Summary of mammalian in vivo ototoxicity studies (continued)
Reference
Fechter et
al., 1998
Jaspers et
al., 1993
Muijseret
al., 2000
Albee et
al., 2006
Yamamura
etal., 1983
Exposure
route
Inhalation
Inhalation
Inhalation
Inhalation
Inhalation
Species, strain,
sex, number
Rat, Long Evans,
male, 12/group
Rat, Wistar derived
WAG-Rii/MBL,
male, 12/group
Rat, Wistar derived
WAG-Rii/MBL,
male, 8
Rat, Fischer 344,
male and female,
10/sex/group
Guinea Pig, albino
Hartley, male,
7-10/group
Dose level/
exposure
duration
0, 4,000 ppm;
6 h/d, 5 d
0, 1,500, 3,000
ppm; 18 h/d,
5 d/wk, 3 wk
0, 3,000 ppm
0, 250, 800, 2,500
ppm
0, 6,000, 12,000,
17,000 ppm;
4 h/d, 5 d
NOAEL;
LOAEL*
LOAEL:
4,000 ppm
LOAEL:
1,500 ppm
LOAEL:
3,000 ppm
NOAEL:
800 ppm
LOAEL:
2,500 ppm
NOAEL:
17,000 ppm
Effects
Cochlear function measured
5-7 wk after exposure. Loss of
spiral ganglion cells noted.
Auditory function was
significantly decreased as
measured by compound action
potentials.
Auditory function assessed
repeatedly 1-5 wk postexposure
for 5-, 20-, and 35-kHz tones; No
effect at 5 or 35 kHz; Decreased
auditory sensitivity at 20 kHz.
Auditory sensitivity decreased
with TCE exposure at 4-, 8-, 16-,
and 20-kHz tones.
Mild frequency specific hearing
deficits; Focal loss of hair cells
and cochlear lesions.
No change in auditory sensitivity
at any exposure level as
measured by cochlear action
potentials and microphonics.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l 12 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-6. Summary of mammalian sensory studies—vestibular and visual
systems
Reference
Exposure route
Species, strain,
sex, number
Dose level/
exposure duration
NOAEL;
LOAEL
Effects
Vestibular system studies
Tham et
al., 1979
Tham et
al., 1984
Niklasson
etal., 1993
Umezu et
al., 1997
Intravenous
Intravenous
Inhalation
Intraperitoneal
Rabbit, strain
unknown, sex
unspecified, 19
Rat, Sprague-
Dawley, female,
11
Rat, strain
unknown, male
and female, 28
Mouse, ICR,
male, 116
1-5 mg/kg/min
80 ug/kg/min
0, 2,700, 4,200,
6,000, 7,200 ppm;
Ih
0, 250, 500,
1,000 mg/kg, single
dose and evaluated
30 min
postadministration
LOAEL:
2,700 ppm
NOAEL:
250 mg/kg
LOAEL:
500 mg/kg
Positional nystagmus
developed once blood
levels reached 30 ppm.
Excitatory effects on the
vestibule-oculomotor
reflex. Threshold effect at
blood (TCE) of 120 ppm or
0.9mM/L.
Increased ability to produce
nystagmus.
Decreased equilibrium and
coordination as measured by
the Bridge test (staying time
on an elevated balance
beam).
Visual system studies
Rebert et
al., 1991
Boyes et
al., 2003
Boyes et
al., 2005
Blainetal.,
1994
Inhalation
Inhalation
Inhalation
Inhalation
Rat, Long Evans,
male, 10/group
Rat, F344, male,
4-5/group
Rat, Long Evans,
male,
9- 10/group
Rat, Long Evans,
male,
8- 10/group
Rabbit, New
Zealand albino,
male, 6-8/group
0, 1,600, 3,200 ppm;
12 h/d, 12 weeks
0, 2,000, 3,200 ppm;
12 h/d, 3 wk
0 ppm, 4 h;
1,000 ppm, 4 h; 2,000
ppm, 2 h; 3,000 ppm,
1. 3 h; 4,000 ppm, Ih
0 ppm, 4 h; 500 ppm,
4 h; 1,000 ppm, 4 h;
2,000 ppm, 2 h;
3,000 ppm, 1.3 h;
4,000 ppm, 1 h;
5,000 ppm, 0.8 h
0, 350, 700 ppm;
4 h/d, 4 d/wk, 12 wk
NOAEL:
3,200 ppm
NOAEL:
3,200 ppm
LOAEL:
1,000 ppm,
4h
LOAEL:
500 ppm,
4h
LOAEL:
350 ppm
No effect on visual function
as measured by visual
evoked potential changes.
Visual function
significantly affected as
measured by decreased
amplitude (F2) in
Fourier-transformed visual
evoked potentials.
Visual function
significantly affected as
measured by decreased
amplitude (F2) in
Fourier-transformed visual
evoked potentials.
Significant effects noted in
visual function as measured
by ERG and OPs
immediately after exposure.
No differences in ERG or
OP measurements were
noted at 6 wk post-TCE
exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 13 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-7. Summary of mammalian cognition studies
Reference
Kjellstrand
etal., 1980
Kuligetal.,
1987
Isaacson et
al, 1990
Kishietal.,
1993
Umezu et
al., 1997
Ohtaetal.,
2001
Oshiro et
al., 2004
Exposure route
Inhalation
Inhalation
Oral, drinking
water
Inhalation
Intraperitoneal
Intraperitoneal
Inhalation
Species, strain,
sex, number
Gerbil,
Mongolian,
males and
females,
12/sex/dose
Rat, Wistar,
male, 8/dose
Rat, Sprague
Dawley, male,
12/dose
Rats, Wistar,
male, number
not specified
Mouse, ICR,
male,
6 exposed to all
treatments
Mouse, ddY,
male, 5/group
Rat, Long
Evans, male, 24
Dose level/
exposure duration
0, 320 ppm; 9 mos,
continuous (24 h/d)
except 1-2 h/wk for cage
cleaning
0, 500, 1,000, 1,500 ppm;
16 h/d, 5 d/wk, 18 wk
(1) 0 mg/kg/d, 8 wk
(2) 5.5 mg/d
(47 mg/kg/d*), 4 wk
+ 0 mg/kg/d, 4 wk
(3) 5.5 mg/d, 4 wk
(47 mg/kg/d*) +
0 mg/kg/d, 2 wk
+ 8.5 mg/d
(24 mg/kg/d),* 2 wk
0, 250,500, 1,000, 2,000,
4,000 ppm, 4 h
0, 125, 250, 500,
1,000 mg/kg, single dose
and evaluated 30 min
postadministration
0, 300, 1,000 mg/kg,
sacrificed 24 h after
injection
0, 1,600, 2,400 ppm;
6 h/d, 5 d/wk, 4 wk
NOAEL;
LOAEL
NOAEL:
320 ppm
NOAEL:
500 ppm
LOAEL:
1,000 ppm
NOAEL: 5.5
mg/d, 4 wk
spatial learning
LOAEL:
5.5 mg/d
hippocampal
demyelination
LOAEL:
250 ppm
NOAEL:
500 mg/kg
LOAEL:
1,000 mg/kg
LOAEL:
300 mg/kg
NOAEL:
2,400 ppm
Effects
No significant effect
on spatial memory
(radial arm maze).
Increased latency
time in the
two-choice visual
discrimination task
(cognitive disruption
and/or motor activity
related effect).
Decreased latency to
find platform in the
Morris water maze
(Group #3);
Hippocampal
demyelination
observed in all
TCE-treated groups.
Decreased lever
presses and
avoidance responses
in a shock avoidance
task.
Decreased response
rate in an operant
response-cognitive
task.
Decreased response
(LTP response) to
tetanic stimulation in
the hippocampus.
No change in
reaction time in
signal detection task
and when challenged
with amphetamine,
no change in
response from
control.
*mg/kg/d conversion estimated from average male Sprague-Dawley rat body weight from ages 21-49 days (118 g)
for the 5.5 mg dosing period and ages 63-78 days (354 g) for the 8.5 mg dosing period.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 14 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-8. Summary of mammalian psychomotor function, locomotor
activity, and reaction time studies
Reference
Savolainen et
al, 1977
Wolff and
Siegmund,
1978
Kuligetal.,
1987
Motohashi
and
Miyazaki,
1990
Fredericksson
etal., 1993
Moser et al.,
1995
Bushnell,
1997
Exposure
route
Inhalation
Intraperitoneal
Inhalation
Intraperitoneal
Oral
Oral
Inhalation
Species/strain/
sex/number
Rat, Sprague
Dawley, male,
10
Mouse, AB,
male, 144
Rat, Wistar,
male, 8/dose
Rat, Wistar,
male, 44
Mouse, NMRI,
male, 12 (3-4
litters)
Rat, Fischer
344, female,
8/dose
Rat, Long
Evans, male,
12
Dose level/
exposure
duration
0, 200 ppm; 6
h/d, 4d
0, 182 mg/kg,
tested 30 min
after injection
0, 500, 1,000,
1,500 ppm; 16
h/d, 5 d/wk, 18
wk
0, 1.2g/kg,
tested 30 min
after injection
0, 1.2g/kg/d, 3d
0, 50, 290
mg/kg/d, at Days
10-16
0, 150, 500,
1,500, 5,000
mg/kg, 1 dose
0, 50, 150, 500,
1,500 mg/kg/d,
14 d
0, 400, 800,
1,200, 1,600,
2,000, 2,400
ppm, 1-h/test
day, 4
consecutive test
days, 2 wk
NOAEL;
LOAEL
LOAEL: 200
ppm
LOAEL: 182
mg/kg
NOAEL: 1,500
ppm
LOAEL: 1.2
g/kg
LOAEL: 1.2
g/kg
...
NOAEL: 500
mg/kg
LOAEL: 1,500
mg/kg
NOAEL: 150
mg/kg/d
LOAEL: 500
mg/kg/d
NOAEL: 800
ppm
LOAEL: 1,200
ppm
Effects
Increased frequency of
preening, rearing, and
ambulation. Increased
preening time.
Decreased spontaneous
motor activity.
No change in spontaneous
activity, grip strength or
hindlimb movement.
Increased incidence of rats
slipping in the inclined
plane test.
Decreased spontaneous
motor activity.
Decreased rearing; No
evidence of dose response.
Decreased motor activity;
Neuro-muscular and
sensorimotor impairment.
Increased rearing activity.
Decreased sensitivity and
increased response time in
the signal detection task.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 15 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-8. Summary of mammalian psychomotor function, locomotor
activity, and reaction time studies (continued)
Reference
Umezu et al.,
1997
Bushnell and
Oshiro, 2000
Nunes et al.,
2001
Waseem et
al., 2001
Moser et al.,
2003
Albee et al.,
2006
Exposure
route
Intraperitoneal
Inhalation
Oral
Oral
Inhalation
Oral
Inhalation
Species/strain/
sex/number
Mouse, ICR,
male, 6
exposed to all
treatments
Rat, Long
Evans, male,
32
Rat, Sprague
Dawley, male,
10/group
Rat, Wistar,
male, 8/group
Rat, Wistar,
male, 6/group
Rat, Fischer
344, female,
10/group
Rat, Fischer
344, male and
female,
10/sex/group
Dose level/
exposure
duration
0, 2,000, 4,000,
5,000 mg/kg -
loss of righting
reflex measure
0, 62.5, 125,
250, 500, 1,000
mg/kg, single
dose and
evaluated 30 min
postadministrati
on
0, 2,000, 2,400
ppm; 70 min/d, 9
d
0, 2,000
mg/kg/d, 7 d
0, 350, 700,
1,400 ppm in
drinking water
for 90 d
0, 376 ppm for
up to 180 d
0, 40, 200, 800,
1,200 mg/kg/d,
10 d
0, 250, 800,
2,500 ppm
NOAEL;
LOAEL*
LOAEL: 2,000
mg/kg - loss of
righting reflex
NOAEL: 500
mg/kg
LOAEL: 1,000
mg/kg - operant
behavior
NOAEL: 125
mg/kg
LOAEL: 250
mg/kg -
punished
responding
LOAEL: 2,000
ppm
LOAEL: 2,000
mg/kg/d
NOAEL: 1,400
ppm
LOAEL: 376
ppm
—
NOAEL: 2,500
ppm
Effects
Loss of righting reflex,
decreased operant
responses, increased
punished responding.
Decreased performance on
the signal detection task.
Increased response time
and decreased response
rate.
Increased foot splay. No
change in any other FOB
parameter (e.g.,
piloerection, activity,
reactivity to handling).
No significant effect on
spontaneous locomotor
activity.
Changes in locomotor
activity but not consistent
when measured over the
180-day period.
Decreased motor activity;
Decreased sensitivity;
Increased abnormality in
gait; Adverse changes in
several FOB parameters.
No change in any FOB
measured parameter.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 16 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-9. Summary of mammalian in vivo dopamine neuronal studies
Reference
Guehl et
al, 1999
Gash et
al., 2007
Exposure
route
Intraperitoneal
administration
Oral
Species/strain/
sex/number
Mouse, OF1,
male, 10
Rat, Fischer 344,
male, 17/group
Dose level/
exposure
duration
0, 400 mg/kg
0, 1,000 mg/kg
NOAEL;
LOAEL
LOAEL:
400 mg/kg
LOAEL:
1,000
mg/kg
Effects
Significant dopaminergic
neuronal death in substantia
nigra.
Degeneration of dopamine-
containing neurons in
substantia nigra.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l 17 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-10. Summary of neurochemical effects with TCE exposure
Reference
Exposure
route
Species/strain/
sex/number
Dose level/
exposure duration
NOAEL;
LOAEL
Effects
In vivo studies
Shih et al.,
2001
Briving et al.,
1986
Subramoniam
etal., 1989
Kjellstrand et
al., 1987
Intraperitoneal
Inhalation
Oral
Inhalation
Mouse, Mfl,
male, 6/group
Gerbils,
Mongolian,
male and
female, 6/group
Rat, Wistar,
female,
Mouse, NMRI,
male
Rat, Sprague-
Dawley, female
0, 250 500, 1,000,
2,000 mg/kg, 15
min; followed by
tail infusion of PTZ
(5 mg/mL),
picrotoxin (0.8
mg/mL),
bicuculline (0.06
mg/mL), strychnine
(0.05 mg/mL), 4-
AP (2 mg/mL), or
NMDA (8 mg/mL)
0, 50, 150 ppm,
continuous, 24 h/d,
12 mos
0, 1,000 mg/kg, 2
or20h
0, 1,000 mg/kg/d, 5
d/wk, 1 y
0, 150, 300 ppm, 24
h/d, 4 or 24 d
0, 300 ppm, 24 h/d,
4 or 24 d
—
NOAEL: 50
ppm; LOAEL:
150 ppm for
glutamate levels
in hippocampus
NOAEL: 150
ppm for
glutamate and
GABA uptake in
hippocampus
LOAEL: 50 ppm
for glutamate and
GABA uptake in
cerebellar vermis
—
LOAEL: 150
ppm, 4 and 24 d
NOAEL: 300
ppm, 4 d
LOAEL: 300
ppm, 24 d
Increased threshold for
seizure appearance
with TCE pretreatment
for all convulsants.
Effects strongest on the
GABAA antagonists,
PTZ, picrotoxin, and
bicuculline suggesting
GAB AA receptor
involvement. NMDA
and glycine Re
involvement also
suggested.
Increased glutamate
levels in the
hippocampus.
Increased glutamate
and GABA uptake in
the cerebellar vermis.
PI and PIP2 decreased
by 24 and 17% at 2 h.
PI and PIP2 increased
by 22 and 3 8% at 20 h.
PI, PIP, andPIP2
reduced by 52,23, and
45% in 1-yr study.
Sciatic nerve
regeneration was
inhibited in both mice
and rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-l 18 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-10. Summary of neurochemical effects with TCE exposure (continued)
Reference
Haglid et al.,
1981
Exposure
route
Inhalation
Species/strain/
sex/number
Gerbil,
Mogolian, male
and female,
6-7/group
Dose level/
exposure duration
0, 60, 320 ppm, 24
h/d, 7 d/wk, 3 mos
NOAEL;
LOAEL*
LOAEL: 60 ppm,
brain protein
changes
NOAEL: 60
ppm; LOAEL:
320 ppm, brain
DNA changes
Effects
(1) Decreases in total
brain soluble protein
whereas increase in
S 100 protein.
(2) Elevated DNA in
cerebellar vermis and
sensory motor cortex.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-l 19 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-ll. Summary of in vitro ion channel effects with TCE exposure
Reference
Cellular
system
Neuronal channel/
receptor
Concentrations
Effects
In vitro studies
Shafer et al.,
2005
Beckstead et
al., 2000
Lopreato et
al., 2003
Krasowski
and Harrison,
2000
PC12 cells
Xenopus
oocytes
Xenopus
oocytes
Human
embryonic
kidney 293
cells
Voltage sensitive
calcium channels
(VSCC)
Human recombinant
Glycine receptor
al, GABAA
receptors, alpl,
alp2y2L
Human recombinant
serotonin 3 A
receptor
Human recombinant
Glycine receptor al,
GABAA receptors
a2pl
0, 500, 1,000,
1,500, 2,000 uM
0, 390 uM
???
Not provided
Shift of VSCC activation to a more
hyperpolarizing potential.
Inhibition of VSCCs at a holding
potential of -70 mV.
50% potentiation of the GAB AA
receptors; 100% potentiation of the
glycine receptor.
Potentiation of serotonin receptor
function.
Potentiation of glycine receptor function
with an EC50 of 0.65 ± 0.05 mM.
Potentiation of GAB AA receptor
function with an EC50 of 0.85 ± 0.2.
EC™ = median effective concentration.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-120 DRAFT—DO NOT CITE OR QUOTE
-------
Table D-12. Summary of mammalian in vivo developmental neurotoxicity
studies—oral exposures
Reference
Fredriksson
etal., 1993
George et
al, 1986
Isaacson
and Taylor,
1989
Noland-
Gerbec et
al., 1986
Taylor et
al., 1985
Species/strain/
sex/number
Mouse, NMRI,
male pups, 12
pups from 3-4
different
litters/group
Rat, F334, male
and female, 20
pairs/treatment
group,
40 controls/sex
Rat, Sprague-
Dawley,
females, 6
dams/group
Rat, Sprague-
Dawley,
females, 9-11
dams/group
Rat, Sprague-
Dawley,
females, no.
dams/group not
reported
Dose level/
exposure duration
0, 50, or 290 mg/kg-d
PND 10-16
0,0.15, 0.30, or
0.60%
microencapsulated
TCE.
Breeders exposed 1
wk premating, then
for 13 wk; pregnant
$s throughout
pregnancy (i.e., 18-
wk total).
0,312, or 625 mg/L.
(0,4.0, or 8.1 mg/d)b
Dams (and pups)
exposed from 14 d
prior to mating until
end of lactation.
0,3 12 mg/L
(Avg. total intake of
dams: 825 mg TCE
over61d.)b
Dams (and pups)
exposed from 14 d
prior to mating until
end of lactation.
0,312, 625, and
1,250 mg/L
Dams (and pups)
exposed from 14 d
prior to mating until
end of lactation.
Route/vehicle
Gavage in a
20% fat
emulsion
prepared from
egg lecithin
and peanut oil
Dietary
Drinking
water
Drinking
water
Drinking
water
NOAEL;
LOAEL3
Dev.
LOAEL: 50
mg/kg/d
LOAEL:
0.15%
Dev.
LOAEL: 3 12
mg/L
Dev. LOEL:
3 12 mg/L
Dev.
LOAEL:
3 12 mg/L
Effects
Rearing activity sig. -i- at
both dose levels on PND
60.
Open field testing in pups:
a sig. dose-related trend
toward t time required for
male and female pups to
cross the first grid in the
test device.
Sig. -i- myelinated fibers in
the stratum lacunosum-
moleculare of pups.
Reduction in myelin in the
hippocampus.
Sig. ^ uptake of 3H-2-
DG in whole brains and
cerebella (no effect in
hippocampus) of exposed
pups at 7, 11, and 16 d,
but returned to control
levels by 21 d.
Exploratory behavior sig.
t in 60- and 90-d old male
rats at all treatment levels.
Locomotor activity was
higher in rats from dams
exposed to 1,250-ppm
TCE.
"NOAEL, LOAEL, and LOEL (lowest-observed-effect level) are based upon reported study findings.
bDose conversions provided by study author(s).
PND = postnatal day.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09
D-121 DRAFT—DO NOT CITE OR QUOTE
-------
D.3. REFERENCES
Abbate C, Giorgianni C, Munao F, Brecciaroli R (1993). Neurotoxicity induced by exposure to toluene. An
electrophysiologic study. Int Arch Occup Environ Health. 64(6):389-392.
Albee, RR; Nitschke, KD; Mattsson, JL; et al. (1997) Dichloroacetylene: Effects on the rat trigeminal nerve
somatosensory evoked potential. Neurotoxicology and Teratology 19(l):27-37.
Albee, RR; Spencer, PJ; Johnson, KA; et al. (2006) Lack of trigeminal nerve toxicity in rats exposed to
trichloroethylene vapor for 13 weeks. Int J Toxicol 25(6):53 1-540.
Albers JW, Wald JJ, Werner RA, Franzblau A, Berent S (1999). Absence of polyneuropathy among workers
previously diagnosed with solvent-induced toxic encephalopathy. J Occup Environ Med. 41(6):500-509.
Antti-Poika M (1982). Overall prognosis of patients with diagnosed chronic organic solvent intoxication. Int Arch
Occup Environ Health. 5 1(2): 127-138.
Aratani J, Suzuki H, Hashimoto K (1993). Measurement of vibratory perception threshold (VPT) in workers
exposed to organic solvents. Environ Res. 61(2):357-361.
Arito, H; Takahashi, M; Ishikawa, T. (1994) Effect of subchronic inhalation exposure to low-level trichloroethylene
on heart rate and wakefulness-sleep in freely moving rats. Jpn J Ind Health (Sangyo Igaku) 36(1): 1-8.
ATSDR (Agency for Toxic Substances and Disease Registry). (2003) Impact of trichloroethylene exposure on oral
motor, speech, and hearing in children. U.S. Department of Health and Human Services, Atlanta, GA.
Barret, L; Arsac, P; Vincent, M; et al. (1982) Evoked trigeminal nerve potential in chronic trichloroethylene
intoxication. J Toxicol Clin Toxicol 1 9(4) :4 19-423.
Barret, L; Faure, J; Guilland, B; et al. (1984) Trichloroethylene occupational exposure: elements for better
prevention. Int Arch Occup Environ Health 53(4):283-289.
Barret, L; Garrel, S; Danel, V; et al. (1987) Chronic trichloroethylene intoxication: a new approach by trigeminal-
evoked potentials? Arch Environ Health 42(5):297-302.
Barret, L; Torch, S; Leray, CL; Sarlieve, L; Saxod, R (1992) Morphometric and Biochemical Studies in Trigeminal
Nerves of Rat after Trichlorethylene or Dichloracetylene Oral Administration. Neurotoxicology 13:601-614.
Barret, L; Torch, S; Usson, Y Barret, L; Torch, S; Usson, Y; et al. (1991) A morphometric evaluation of the effects
of trichloroethylene and dichloroacetylene on the rat mental nerve. Preliminary results. Neurosci Lett
Beckstead MJ, Weiner JL, Eger El 2nd, Gong DH, Mihic SJ (2000). Glycine and gamma-aminobutyric acid(A)
receptor function is enhanced by inhaled drugs of abuse. Mol Pharmacol. 57(6): 1 199-1205.
Binaschi S, and Cantu L. (1982) Vestibular Function In Solvent Exposure: Clinical Criteria. Neurobehavioral
Methods in Occupational Health. 46: 205-210.
Blain, L; Lachapelle, P; Molotchnikoff, S. (1994) Electroretinal responses are modified by chronic exposure to
trichloroethylene. Neurotoxicology 15(3):627-631.
Bowler RM, Mergler D, Huel G, Harrison R, Cone J (1991). Neuropsychological impairment among former
microelectronics workers. Neurotoxicology. 12(1):87-103.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-122 DRAFT—DO NOT CITE OR QUOTE
-------
Boyes, WK; Bercegeay, M; All, JS; et al. (2003) Dose-based duration adjustments for the effects of inhaled
trichloroethylene on rat visual function. Toxicol Sci 76(1): 121-130.
Boyes, WK; Bercegeay, M; Krantz, T; et al. (2005) Momentary brain concentration of trichloroethylene predicts the
effects on rat visual function. Toxicol Sci 87(1): 187-196.
Boyes, WK; Bushnell, PJ; Crofton, KM; et al. (2000) Neurotoxic and pharmacokinetic responses to
trichloroethylene as a function of exposure scenario. Environ Health Perspect 108(Suppl 2):317-322.
Briving, C; Jacobson, I; Hamberger, A; et al. (1986) Chronic effects of perchloroethylene and trichloroethylene on
the gerbil brain amino acids and glutathione. Neurotoxicology 7(1):101-108.
Burg, JR; Gist, G; Allred, SL; et al. (1995) The national exposure registry - morbidity analyses of noncancer
outcomes from the trichloroethylene subregistry baseline data. Int J Occ Med Toxicol 4(2):237-257.
Burg, JR; Gist, GL. (1999) Health effects of environmental contaminant exposure: an intrafile comparison of the
Trichloroethylene Subregistry. Arch Environ Health 54(4):231-241.
Bushnell PJ (1997). Concentration-time relationships for the effects of inhaled trichloroethylene on signal detection
behavior in rats. Fundam Appl Toxicol. 36(l):30-38.
Bushnell PJ, Oshiro WM (2000). Behavioral components of tolerance to repeated inhalation of trichloroethylene
(TCE) in rats. Neurotoxicol Teratol. 22(2):221-229.
Buxton, P.H., and Hayward, M. (1967). Polyneuritis cranialis associated with industrial trichloroethylene poisoning.
J. Neurol. Neurosurg. Psych. 30, 511.
Campo P, Blancher V, Payan JP, Cossec B, Ducos P. Toluene toxicokinetics and metabolism parameters in the rat
and guinea pig. Environmental Toxicology and Pharmacology 21 (2006) 276-282.
Campo P, Loquet G, Blachere V, Roure M (1999). Toluene and styrene intoxication route in the rat cochlea.
Neurotoxicol Teratol. 21(4):427-434.
Cappaert ML, Klis SF, Baretta AB, Muijser H, Smoorenburg GF. Ethyl benzene-induced ototoxicity in rats: a dose-
dependent mid-frequency hearing loss. J Assoc Res Otolaryngol. 2000 Dec;l(4):292-9.
Cappaert ML, Klis SF, Muijser H, de Groot JC, Kulig BM, Smoorenburg GF. The ototoxic effects of ethyl benzene
in rats. Hear Res. 1999Nov;137(l-2):91-102.
Chalupa, B., J. Synkova, and M. Sevcik 1960. The Assessment of Electro-encephalographic changes in memory
disturbances in acute intoxications with industrial poisons, Brit. J. Ind. Med. 17, 238.
Colvin M, Myers J, Nell V, Rees D, Cronje R (1993). A cross-sectional survey of neurobehavioral effects of
chronic solvent exposure on workers in a paint manufacturing plant. Environ Res. 63(1): 122-132.
Crofton, KM Lassiter, TL and Rebert, CS. Solvent-induced ototoxicity in rats: an atypical selective mid-frequency
hearing deficit, Hear. Res., 80 (1994, 25-30.
Crofton, KM; Zhao, X. (1997) The ototoxicity of trichloroethylene: extrapolation and relevance of high-
concentration, short-duration animal exposure data. Fundam Appl Toxicol 38(1):101-106.
Daniell WE, Claypoole KH, Checkoway H, Smith-Weller T, Dager SR, Townes BD, Rosenstock L (1999).
Neuropsychological function in retired workers with previous long-term occupational exposure to solvents. Occup
Environ Med. 56(2):93-105.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-123 DRAFT—DO NOT CITE OR QUOTE
-------
Donoghue AM, Dryson EW, Wynn-Williams G (1995). Contrast sensitivity in organic-solvent-induced chronic
toxic encephalopathy. J Occup Environ Med. 37(12): 1357-1363
El Ghawabi, SM; Mansoor, MB; El Gamel, MS; et al. (1973) Chronic trichloroethylene exposure. J Egypt Med
Assoc 56(ll-12):715-724.
Elofsson SA, Gamberale F, Hindmarsh T, Iregren A, Isaksson A, Johnsson I, Knave B, Lydahl E, Mindus P, Persson
HE, Philipson B, Steby M, Struwe G, Soderman E, Wennberg A, Widen L (1980). Exposure to organic solvents. A
cross-sectional epidemiologic investigation on occupationally exposed care and industrial spray painters with special
reference to the nervous system. Scand J Work Environ Health. 6(4):239-273.
Fechter LD, Gearhart C, Fulton S, Campbell J, Fisher J, Na K, Cocker D, Nelson-Miller A, Moon P, Pouyatos B.
Promotion of noise-induced cochlear injury by toluene and ethylbenzene in the rat. Toxicol Sci. 2007
Aug;98(2):542-51.
Fechter, LD, Liu, Y., Herr, DW, Crofton, KM. Trichloroethylene ototoxicity: Evidence for a cochlear origin.,
Toxicol. Sci. 42, 28-35, 1998.
Feldman, RG; Chirico-Post, J; Proctor, SP. (1988) Blink reflex latency after exposure to trichloroethylene in well
water. Arch Environ Health 43(2): 143-148.
Feldman, RG; Niles, C; Proctor, SP; et al. (1992) Blink reflex measurement of effects of trichloroethylene exposure
on the trigeminal nerve. Muscle Nerve 15(4):490-495.
Fredriksson et al., tox lett. 66, 1993, 13-19. Altered behavior in adult mice orally exposed to tri- and
tetrachloroethylene as neonates.
Gagnaire F, Marignac B, Langlais C, Bonnet P (2001). Ototoxicity in rats exposed to ortho-, meta- and paraxylene
vapours for 13 weeks. Pharmacol Toxicol. 89(1):6-14.
Gamberale, F; Annwall, G; Olson, BA. (1976) Exposure to trichloroethylene III. Psychological functions. Scand J
Work Environ Health 2(4):220-224.
Gash, DM; Rutland, K; Hudson, NL; et al. (2007) Trichloroethylene: Parkinsonism and complex 1 mitochondria!
neurotoxicity. AnnNeurol.
George, JD; Reel, JR; Myers, CB; Lawton, AD; Lamb, JC. (1986) Trichloroethylene: reproduction and fertility
assessment in F344 rats when administered in the feed. RTI Project No. 310-2344, NTP-86-085. National Institute
of Environmental Health Sciences, National Toxicology Program, RTF, NC.
Grandjean, E; Munchinger, R; Turrian, V; et al. (1955) Investigations into the effects of exposure to
trichloroethylene in mechanical engineering. Br J Ind Med 12(2): 131-142.
Gregersen P. (1988) Neurotoxic effects of organic solvents in exposed workers: two controlled follow-up studies
after 5.5 and 10.6 years. Am J Ind Med. 14(6):681-701.
Guehl, D; Bezard, E; Dovero, S; et al. (1999) Trichloroethylene and parkinsonism: a human and experimental
observation. Eur J Neurol 6(5):609-611.
Gun, RT; Grygorcewicz, C; Nettelbeck, TJ. (1978) Choice reaction time in workers using trichloroethylene. Med J
Aust l(10):535-536.
Haglid KG, Briving C, Hansson HA, Rosengren L, Kjellstrand P, Stavron D, Swedin U, Wronski A (1981).
Trichloroethylene: long-lasting changes in the brain after rehabilitation. Neurotoxicology. 2(4):659-673.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-124 DRAFT—DO NOT CITE OR QUOTE
-------
Hirsch, AR; Rankin, KM; Panelli, PP. (1996) Trichloroethylene exposure and headache. Headache Quarterly
7(2):126-138.
Isaacson, LG; Spohler, SA; Taylor, DH. (1990) Trichloroethylene affects learning and decreases myelin in the rat
hippocampus. Neurotoxicol Teratol 12(4):375-381.
Isaacson, LG; Taylor, DH. (1989) Maternal exposure to 1,1,2-trichloroethylene affects myelin in the hippocampal
formation of the developing rat. Brain Res 488(1-2):403-407.
Jaspers, RMA, Muijser, H., et al., Mid-frequency hearing loss and reduction of acoustic startle responding in rats
following trichloroethylene exposure. Neurotoxicol. Teratol., 15, 407-412, 1993.
Juntunen J, Antti-Poika M, Tola S, Partanen T (1982). Clinical prognosis of patients with diagnosed chronic solvent
intoxication. ActaNeurol Scand. 65(5):488-503.
Juntunen J, Hupli V, Hernberg S, Luisto M (1980). Neurological picture of organic solvent poisoning in industry. A
retrospective clinical study of 37 patients. Int Arch Occup Environ Health. 46(3):219-331.
Kilburn, KH. (2002a): Is neurotoxicity associated with environmental trichloroethylene (TCE)? Arch Environ
Health 57(2): 113-120.
Kilburn, KH. (2002b): Do duration of exposure, proximity to electronic manufacturing plants, and involvement in a
lawsuit affect chlorinated solvent toxicity? Arch Environ Health 57(2): 121-126
Kilburn, KH; Thornton, JC. (1996) Prediction equations for simple and visual two-choice reactions times in
environmental neurotoxicology. Arch Environ Health 51(6):439-444.
Kilburn, KH; Warshaw, RH. (1993) Effects on neurobehavioral performance of chronic exposure to chemically
contaminated well water. Toxicol Ind Health 9(3):391-404.
Kishi, R; Harabuchi, I; Ikeda, T; et al. (1993) Acute effects of trichloroethylene on blood concentrations and
performance decrements in rats and their relevance to humans. Br J Ind Med 50(5):470-480.
Kjellstrand, P; Kanje, M; Bjerkemo, M. (1987) Regeneration of the sciatic nerve in mice and rats exposed to
trichloroethylene. Toxicol Lett 38(1-2): 187-191.
Kjellstrand, P; Lanke, J; Bjerkemo, M; et al. (1980) Irreversible effects of trichloroethylene exposure on the central
nervous system. Scand J Work Environ Health 6(l):40-47.
Konietzko, H; Elster, I; Sayer, H; et al.,(1975) EEG Changes Under Standardized Trichloroethylene Load; Int. Arch.
Occup. Environ. Hlth 35, 257-264 Zentralnervose schaden durch trichlorathylen. Staub, Reinhaltung
DerLuft35(6):240-241.
Krasowski MD, Harrison ML. 2000. The actions of ether, alcohol and alkane general anaesthetics on GABAA and
glycine receptors and the effects of TM2 andTM3 mutations. Br J Pharmacol. 129(4):731-43.
Kulig, BM. (1987) The effects of chronic trichloroethylene exposure on neurobehavioral functioning in the rat.
Neurotoxicol Teratol 9(2):171-178.
Kylin, B; Axell, K; Samuel, HE; et al. (1967) Effect Of Inhaled Trichloroethylene On The CNS As Measured By
Optokinetic Nystagmus. Archives of Environmental Health 15:48-52.
Kyrklund T, Haglid K (1991). Brain lipid composition in guinea pigs after intrauterine exposure to
perchloroethylene. Pharmacol Toxicol. 68(2): 146-148.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-125 DRAFT—DO NOT CITE OR QUOTE
-------
Kyrklund T, Kjellstrand P, Haglid KG (1986). Fatty acid changes in rat brain ethanolamine phosphoglycerides
during and following chronic exposure to trichloroethylene. Toxicol Appl Pharmacol. 85(2):145-153.
Kyrklund T, Kjellstrand P, Haglid KG (1988). Effects of exposure to Freon 11, 1,1,1-trichloroethane or
perchloroethylene on the lipid and fatty-acid composition of rat cerebral cortex. Scand J Work Environ Health.
14(2):91-94.
Landrigan, PJ; Kominsky, JR; Stein, GF; et al. (1987) Common-source community and industrial exposure to
trichloroethylene. Arch Environ Health 42(6):327-332.
Laslo-Baker D, Barrera M, Knittel-Keren D, Kozer E, Wolpin J, Khattak S, Hackman R, Rovet J, Koren G (2004).
Child neurodevelopmental outcome and maternal occupational exposure to solvents. Arch Pediatr Adolesc
Med.l58(10):956-961.
Lee DH, Park IG, Kim JH, Lee YH, Kim D, Kang SK (1998). Neurobehavioral changes in shoe manufacturing
workers. Neurotoxicol Teratol. 20(3):259-263.
Levy DL, Lipton RB, Holzman PS (1981). Smooth pursuit eye movements: effects of alcohol and chloral hydrate. J
Psychiatr Res. 16(1): 1-11.
Lindstrom K (1973). Psychological performances of workers exposed to various solvents. Work Environ Health.
10:150-155.
Lindstrom K (1980). Changes in psychological performances of solvent-poisoned and solvent-exposed workers.
AmJIndMed. l(l):69-84.
Lindstrom K, Antti-Poika M, Tola S, Hyytiainen A (1982). Psychological prognosis of diagnosed chronic organic
solvent intoxication. Neurobehav Toxicol Teratol. 4(5):581-588.
Liu, YT; Jin, C; Chen, Z; et al. (1988) Increased subjective symptom prevalence among workers exposed to
trichloroethylene at sub-OEL levels. Tohoku J Exp Med 155(2): 183-195.
Lopreato GF, Phelan R, Borghese CM, Beckstead MJ, Mihic SJ. (2003) Inhaled drugs of abuse enhance serotonin-3
receptor function. Drug Alcohol Depend. 70(1): 11-5.
Marshall EG, Gensburg LJ, Deres DA, Geary NS, Cayo MR (1997). Maternal residential exposure to hazardous
wastes and risk of central nervous system and musculoskeletal birth defects. Arch Environ Health. 52(6):416-425.
McCarthy TB, Jones RD (1983). Industrial gassing poisonings due to trichlorethylene, perchlorethylene, and 1-1-1
trichloroethane, 1961-80. Br JIndMed. 40(4):450-455.
McCunney, R.J., Diverse manifestations of trichloroethylene (1988). Br J Ind Med 45:122-126.
Mergler D, Huel G, Bowler R, Frenette B, Cone J (1991). Visual dysfunction among former microelectronics
assembly workers. Arch Environ Health. 46(6):326-334.
Mhiri, C; Choyakh, F; Ben, HM; et al. (2004) Trigeminal somatosensory evoked potentials in trichloroethylene-
exposed workers. Neurosciences 9(2): 102-107.
Mitchell, A. B. S., and Parsons-Smith, B. G., (1969) Trichloroethylene Neuropathy. Brit. med. J. 1, 422-423
Morata TC, Fiorini AC, Fischer FM, Colacioppo S, Wallingford KM, Krieg EF, Dunn DE, Gozzoli L, Padrao MA,
Cesar CL (1997). Toluene-induced hearing loss among rotogravure printing workers. Scand J Work Environ
Health. 23(4):289-298.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-126 DRAFT—DO NOT CITE OR QUOTE
-------
Morata TC, Johnson AC, Nylen P, Svensson EB, Cheng J, Krieg EF, Lindblad AC, Ernstgard L, Franks J (2002).
Audiometric findings in workers exposed to low levels of styrene and noise. J Occup Environ Med. 44(9):806-814.
Morioka I, Miyai N, Yamamoto H, Miyashita K (2000). Evaluation of combined effect of organic solvents and
noise by the upper limit of hearing. Ind Health. 38(2):252-257.
Morrow LA, Ryan CM, Goldstein G, Hodgson MJ (1989). A distinct pattern of personality disturbance following
exposure to mixtures of organic solvents. J Occup Med. 31(9):743-746.
Morrow LA, Steinhauer SR, Hodgson MJ (1992). Delay in P300 latency in patients with organic solvent exposure.
Arch Neural. 49(3):315-320.
Moser VC, Phillips PM, McDaniel KL, MacPhail RC (1999). Behavioral evaluation of the neurotoxicity produced
by dichloroacetic acid in rats. Neurotoxicol Teratol. 21(6):719-731.
Moser, Cheek and MacPhail. A multidisciplinary approach to lexicological screening III. Neurobehavioral toxicity.
J Toxicol. Environ. Kith., 1995, 45, 173-210.
Moser, VC; MacPhail, RC; Gennings, C. (2003) Neurobehavioral evaluations of mixtures of trichloroethylene,
heptachlor, and di(2-ethylhexyl)phthlate in a full-factorial design. Toxicology 188(2-3): 125-137.
Motohashi Y, Miyazaki Y (1990). Time-dependent effects of trichloroethylene on motor activity in rats.
Chronobiol Int. 7(3):193-201.
Muijser, H; Lammers, J; Kulig, BM. (2000) Effects of exposure to trichloroethylene and noise on hearing in rats.
Noise and Health 6:57-66.
Nagaya, T; Ishikawa, N; Hata, H. (1990) No Change in Serum Dopamine-beta-hydroxylase Activity in Workers
Exposed to Trichloroethylene. Toxicology Letters 54(2/3):221-227.
Niklasson, Tham, Larsby, and Eriksson, Effects of toluene, styrene, trichloroethylene and trichloroethane on the
vestibule and opto-oculo motor system in rats, Neurotoxicol. Teratol., 1993, 15, 327-334.
Noland-Gerbec, Pfohl, Taylor and Bull. 2-deopxyglucose uptake in the developing rat brain upon pre and postnatal
exposure to trichloroethylene, Neurotoxicology, 7,, 157-164, 1986.
Nunes J, Ehrich M, Robertson J (2001). Toxicosis associated with dual oral exposure of rats to lead and
trichloroethylene. Toxicol Pathol. 29(4):451-457.
Ohta, M; Saito, T; Saito, K; et al. (2001) Effect of trichloroethylene on spatiotemporal pattern of LTP in mouse
hippocampal slices. Int J Neurosci 111(3-4):257-271.
Oshiro WM, Krantz QT, Bushnell PJ (2004). A search for residual behavioral effects of trichloroethylene (TCE) in
rats exposed as young adults. Neurotoxicol Teratol. 26(2) :23 9-251.
Pryor GT, Dickinson J, Howd RA, Rebert CS (1983). Transient cognitive deficits and high-frequency hearing loss
in weanling rats exposed to toluene. Neurobehav Toxicol Teratol. 5(l):53-57.
Pryor GT, Rebert CS, Howd RA (1987). Hearing loss in rats caused by inhalation of mixed xylenes and styrene. J
Appl Toxicol. 7(1):55-61.
Rasmussen, K; Arlien-Soborg, P; Sabroe, S. (1993c) Clinical neurological findings among metal degreasers exposed
to chlorinated solvents. Acta Neurol Scand 87(3):200-204.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-127 DRAFT—DO NOT CITE OR QUOTE
-------
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (1993a) Solvent-Induced Chronic Toxic Encephalopathy. American
Journal of Industrial Medicine 23:779-792.
Rasmussen, K; Jeppesen, HJ; Sabroe, S. (1993b) Psychometric tests for assessment of brain function after solvent
exposure. Am J Ind Med 24(5):553-565.):
Rasmussen, K; Sabroe, S. (1986) Neuropsychological symptoms among metal workers exposed to halogenated
hydrocarbons. Scand J Soc Med 14(3) : 161-168.
Rebert CS, Boyes WK, Pryor GT, Svensgaard DJ, Kassay KM, Gordon GR, Shinsky N (1993). Combined effects of
solvents on the rat's auditory system: styrene and trichloroethylene. Int J Psychophysiol. 14(l):49-59
Rebert CS, Day VL, Matteucci MJ, Pryor GT (1991). Sensory-evoked potentials in rats chronically exposed to
trichloroethylene: predominant auditory dysfunction. Neurotoxicol Teratol. 13(1):83-90
Rebert CS, Schwartz RW, Svendsgaard DJ, Pryor GT, Boyes WK (1995). Combined effects of paired solvents on
the rat's auditory system. Toxicology. 105(2-3):345-54.
Reif, JS; Burch, JB; Nuckols, JR; et al. (2003) Neurobehavioral effects of exposure to trichloroethylene through a
municipal water supply. Environ Res 93(3):248-258.
Ruijten, M; Verberk, MM; Salle, HJ. (1991) Nerve function in workers with long term exposure to trichloroethene.
BrJIndMed48(2):87-92.
Salvini, M., Binasch, S., and Riva, M. (1971). Evaluation of the psychophysiological functions in humans exposed
to trichloroethylene. Br J Ind Med, 28:293-295.
Savolainen H, Pfaffli P, Tengen M, Vainio H. (1977). Trichloroethylene and 1,1,1-trichloroethane: effects on brain
and liver after five days intermittent inhalation. Arch Toxicol. 38(3):229-237. Available at
http://www.ncbi.nlm.nih.gov/pubmed/578726?ordinalpos=6&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_Resul
tsPanel.Pubmed_DefaultReportPanel.Pubmed_RVDocSum.
Seppalainen AM, Antti-Poika M (1983). Time course of electrophysiological findings for patients with solvent
poisoning. A descriptive study. Scand J Work Environ Health. 9(1): 15-24 (As cited in Winneke, 1982).
Shafer, TJ; Bushnell, PJ; Benignus, VA; et al. (2005) Perturbation of voltage-sensitive Ca2+ channel function by
volatile organic solvents. J Pharmacol Exp Ther 315(3): 1109-1118.
Shih, CL, Chen, HH, Chiu, TH. Acute exposure to trichloroethylene differentially alters the susceptibility to
chemoconvulsants in mice. Toxicology, 162, 2001, 35-42.
Shlomo M, AvrahamF, Moshe H, Mario S, Jacklin S, Mordechai H (2002). Effects of Occupational Exposure to
Mercury or Chlorinated Hydrocarbons on the Auditory Pathway. Noise and Health. 4(16): 71-77.
Sliwinska-Kowalska M, Jedlinska U, Rzadziriska A, Rajkowska E (1999). Hair cell regeneration in a chicken's
inner ear after damage due to exposure to industrial noise Otolaryngol Pol. 53(4):469-477.
Smith, GF. (1970) The investigation of the mental effects of trichloroethylene. Ergonomics 13(5):580-586.
Stewart, RD; Dodd, HC; Gay, HH; et al. (1970) Experimental human exposure to trichloroethylene. Arch Environ
Health 20(1):64-71.
Stopps GJ, McLaughlin M (1967). Psychophysiological testing of human subjects exposed to solvent vapors. Am
Ind Hyg Assoc J. 28(1):43-50.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-128 DRAFT—DO NOT CITE OR QUOTE
-------
Subramoniam A, Goel SK, Pandya KP, Seth PK (1989). Influence of trichloroethylene treatment on
phosphoinositides in rat brain. Toxicol Lett. 49(1):55-60.
Taylor DH, Lagory KE, Zaccaro DJ, Pfohl RJ, Laurie RD (1985). Effect of trichloroethylene on the exploratory and
locomotor activity of rats exposed during development. Sci Total Environ. 47:415-40.
Tham, R; BUNNFORS, I; Eriksson, B; et al. (1984) Vestibule ocular disturbances in rats exposed to organic
solvents. ActaPharmacologicaEt Toxicologica 54(l):58-63.
Tham, R; Larsby, B; Odkvist, LM; et al. (1979) The influence of trichloroethylene and related drugs on the
vestibular system. Acta Pharmacol Toxicol (Copenh) 44(5):336-342.
Till C, Westall CA, Koren G, Nulman I, Rovet JF (2005). Vision abnormalities in young children exposed
prenatally to organic solvents. Neurotoxicology. 26(4):599-613.
Till C, Westall CA, Rovet JF, Koren G (2001). Effects of maternal occupational exposure to organic solvents on
offspring visual functioning: a prospective controlled study. Teratology. 64(3): 134-141.
Triebig, G., S. Lehrl, W. Kinzel, H. Erzigkeit, J. V. GalsterandK. H. Schaller (1977a). Psycho-pathometric results
of follow-up studies of persons exposed to trichloroethylene. Zbl. Bakt. Hyg., I. Abt. Orig. B 164, 314-327 (1977).
(translated from German)
Triebig, G., W. Bestler, P. Baumeister, H. Valentin (1983). Investigations on the neurotoxicity of chemical
substances at the workplace, Part IV: Determination of the motor and sensory nerve conduction velocity in persons
occupationally exposed to a mixture of organic solvents. Int. Arch. Occup. Environ. Health, 52, 139-150.
(translated from German)
Triebig, G; Essing, HG; Schaller, KH; et al. (1976) Biochemical and psychological examinations of
trichloroethylene exposed volunteers. Zentralbl Bakteriol [OrigB] 163(5-6):383-416. (translated from German)
Triebig, G; Schaller, KH; Erzigkeit, H; et al. (1977b). Biochemical investigations and psychological studies of
persons chronically exposed to trichloroethylene with regard to non-exposure intervals. Int Arch Occup Environ
Health 38(3): 149-162. (translated from German)
Triebig, G; Trautner, P; Weltle, D; Saure, E; Valentin, H. (1982). Investigations on the Neurotoxicity of processing
materials III. Measurement of Motor and Sensory Nerve Conduction Velocity in Persons with Job-Related
Trichloroethylene Exposure. Int Arch Occup Environ Health (1982) 51: 25-34 (translated from German)
Troster, Al; Ruff, RM. (1990) Neuropsychological sequelae of exposure to the chlorinated hydrocarbon solvents
trichloroethylene and trichloroethane. Archives of Clinical Neuropsychology 5(l):31-47.
Truhaut, R. (1951). [Contribution to the study of toxicology of trichloroethylene; establishment of a method of
determination applicable to atmosphere and biological media.] Ann Pharm Fr. 9:175-87.
Umezu, T; Yonemoto, J; Soma, Y; et al. (1997) Behavioral effects of trichloroethylene and tetrachloroethylene in
mice. Pharmacol Biochem Behav 58(3):665-671.
Valic E, Waldhor T, Konnaris C, Michitsch A, Wolf C (1997). Acquired dyschromatopsia in combined exposure to
solvents and alcohol. Int Arch Occup 70(6):403-406.
Vernon RJ, Ferguson RK. (1969) Effects of trichloroethylene on visual-motor performance. Arch Environ Health
18(6): 894-900.
Waseem, M; Ali, M; Dogra, S; et al. (2001) Toxicity of trichloroethylene following inhalation and drinking
contaminated water. J Appl Toxicol 21(6):441-444.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-129 DRAFT—DO NOT CITE OR QUOTE
-------
White RF, Feldman RG, Eviator II, Jabre JF, Niles CA (1997). Hazardous waste and neurobehavioral effects: a
developmental perspective. Environ Res. 73(1-2): 113-124.
Windemuller, FJ; Ettema, JH. (1978) Effects of combined exposure to trichloroethylene and alcohol on mental
capacity. Int Arch Occup Environ Health 41(2):77-85.
Windham GC, Zhang L, Gunier R, Croen LA, Grether JK (2006). Autism spectrum disorders in relation to
distribution of hazardous air pollutants in the San Francisco bay area. Environ Health Perspect. 114(9): 1438-1444.
Winneke,G. (1982) Acute behavioral effects of exposure to some organic solvents-psychophysiological aspects,
ActaNeurol Scand, 32:117-129.
Winneke G, Kastka J, Fodor GG (1974). Psychophysiological effects of low-level exposure to trichloroethylene and
toluene. Presented at the 2nd Industrial and Environmental Neurology Congress held October 22-26, 1974 in Prague.
(As cited in Winneke, 1982)
Winneke G, Kramer U, Kastka J (1976), Zur Beeinflussung psychomotor Leistungen Durch Alkohol und durch
verschiedene Losungsmitteldampfe. Adverse Effects of Environmental Chemicals and Psychotropic Drugs, ed. M.
Horvath, 99-110, Elsevier, Amsterdam. (As cited in Winneke, 1982)
Winneke G, Fodor GG, Schlipkoter HW (1978). Carbon monoxide, trichloroethylene, and alcohol: Reliability and
validity of neurobehavioral effects. Multidisciplinary Perspectives in Event Related Brain Potential Research, ed.
D.A. Otto, 463-469, U.S. Environmental Protection Agency, Washington. (As cited in Winneke, 1982)
Wolff and Siegmund. (1978) The effect of trichloroethylene on the spontaneous locomotor activity of mice and of
tetrachloroethane on the mortality of mice as a function of the time of day. Biol. Zbl, 97, 345-351 (As cited in
Winneke, 1982).
Yamamura, K; Ikeda, I; Sadamoto, T; et al. (1983) Effects Of Trichloroethylene Exposure On Hearing. An
Investigation Of Cochlear Microphonics And Action Potential Of The Guinea Pig. European Journal of Applied
Physiology and Occupational Physiology 52:47-50.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 D-130 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX E
Analysis of Liver and Coexposure Issues for
the TCE Toxicological Review
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix E: Analysis of Liver and Coexposure Issues for the TCE
Toxicological Review
LIST OF TABLES E-vi
LIST OF FIGURES E-vii
FOREWORD E-viii
AUTHORS, CONTRIBUTORS, AND REVIEWERS E-ix
ACKNOWLEDGMENTS E-ix
APPENDIX E. ANALYSIS OF LIVER AND COEXPOSURE ISSUES FOR THE
TCE TOXICOLOGICAL REVIEW E-l
E. 1. BASIC PHYSIOLOGY AND FUNCTION OF THE LIVER—A STORY OF
HETEROGENEITY E-l
E. 1.1. Heterogeneity of Hepatocytes and Zonal Differences in Function and
Ploidy E-l
E. 1.2. Effects of Environment and Age: Variability of Response E-7
E.2. CHARACTERIZATION OF HAZARD FROM TRICHLOROETHYLENE
(TCE) STUDIES E-10
E.2.1. Acute Toxicity Studies E-ll
E.2.1.1. Sonietal., 1998 E-ll
E.2.1.2. Sonietal., 1999 E-14
E.2.1.3. Okinoetal., 1991 E-14
E.2.1.4. Nunesetal., 2001 E-16
E.2.1.5. Tao et al., 2000 E-16
E.2.1.6. Tucker etal., 1982 E-17
E.2.1.7. Goldsworthy and Popp, 1987 E-19
E.2.1.8. Elcombe et al., 1985 E-20
E.2.1.9. Dees and Travis, 1993 E-33
E.2.1.10. Nakajima etal.,2000 E-37
E.2.1.11.Bermanetal., 1995 E-40
E.2.1.12. Melnick etal., 1987 E-42
E.2.1.13. Laughter etal.,2004 E-45
E.2.1.14.Ramdhan etal., 2008 E-49
E.2.2. Subchronic and Chronic Studies of Trichloroethylene (TCE) E-56
E.2.2.1. Merrick et al., 1989 E-57
E.2.2.2. Goeletal., 1992 E-61
E.2.2.3. Kjellstrandetal., 1981 E-63
E.2.2.4. Woolhiser et al., 2006 E-66
E.2.2.5. Kjellstrandetal., 1983a E-68
E.2.2.6. Kjellstrandetal., 1983b E-72
E.2.2.7. Buben and O'Flaherty, 1985 E-76
E.2.2.8. Channel et al., 1998 E-79
E.2.2.9. Dorfmueller et al., 1979 E-83
E.2.2.10. Kumar etal., 2001 E-83
E.2.2.11.Kawamoto etal., 1988 E-84
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-ii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
E.2.2.12. National Toxicology Program (NTP), 1990 E-86
E.2.2.13.National Toxicology Program (NTP), 1988 E-90
E.2.2.14.Fukudaetal., 1983 E-92
E.2.2.15.Henschleretal., 1980 E-93
E.2.2.16.Maltonietal., 1986 E-95
E.2.2.17.Maltonietal., 1988 E-100
E.2.2.18.VanDuurenetal., 1979 E-100
E.2.2.19. National Cancer Institute (NCI), 1976 E-101
E.2.2.20.Herren-Freundetal., 1987 E-106
E.2.2.21.Annaetal., 1994 E-106
E.2.2.22. Bull etal., 2002 E-108
E.2.3. Mode of Action: Relative Contribution of Trichloroethylene (TCE)
Metabolites E-110
E.2.3.1. Acute studies of Dichloroacetic Acid (DCA)/Trichloroacetic
Acid(TCA) E-lll
E.2.3.2. Subchronic and Chronic Studies of Dichloroacetic Acid
(DCA) and Trichloroacetic Acid (TCA) E-137
E.2.4. Summaries and Comparisons Between Trichloroethylene (TCE),
Dichloroacetic Acid (DCA), and Trichloroacetic Acid (TCA) Studies E-197
E.2.4.1. Summary of Results For Short-term Effects of
Trichloroethylene (TCE) E-198
E.2.4.2. Summary of Results For Short-Term Effects of
Dichloroacetic Acid (DCA) and Trichloroacetic Acid
(TCA): Comparisons With Trichloroethylene (TCE) E-204
E.2.4.3. Summary Trichloroethylene (TCE) Subchronic and
Chronic Studies E-226
E.2.4.4. Summary of Results For Subchronic and Chronic Effects
of Dichloroacetic Acid (DCA) and Trichloroacetic Acid
(TCA): Comparisons With Trichloroethylene (TCE) E-238
E.2.5. Studies of Chloral Hydrate (CH) E-261
E.2.6. Serum Bile Acid Assays E-267
E.3. STATE OF SCIENCE OF LIVER CANCER MODES OF ACTION
(MOAs) E-269
E.3.1. State of Science for Cancer and Specifically Human Liver Cancer E-271
E.3.1.1. Epigenetics and Disease States (Transgenerational Effects,
Effects of Aging and Background Changes) E-271
E.3.1.2. Emerging Technologies, DNA and siRNA, miRNA
Microarrays—Promise and Limitations for Modes of
Action (MOAs) E-279
E.3.1.3. Etiology, Incidence and Risk Factors for Hepatocellular
Carcinoma (HCC) E-288
E.3.1.4. Issues Associated with Target Cell Identification E-291
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-iii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
E.3.1.5. Status of Mechanism of Action for Human Hepatocellular
Carcinoma (HCC) E-295
E.3.1.6. Pathway and Genetic Disruption Associated with
Hepatocellular Carcinoma (HCC) and Relationship to
Other Forms of Neoplasia E-298
E.3.1.7. Epigenetic Alterations in Hepatocellular Carcinoma (HCC) E-300
E.3.1.8. Heterogeneity of Preneoplastic and Hepatocellular
Carcinoma (HCC) Phenotypes E-302
E.3.2. Animal Models of Liver Cancer E-309
E.3.2.1. Similarities with Human and Animal Transgenic Models E-313
E.3.3. Hypothesized Key Events in HCC Using Animal Models E-317
E.3.3.1. Changes in Ploidy E-317
E.3.3.2. Hepatocellular Proliferation and Increased DNA Synthesis E-323
E.3.3.3. Nonparenchymal Cell Involvement in Disease States
Including Cancer E-326
E.3.3.4. Gender Influences on Susceptibility E-333
E.3.3.5. Epigenomic Modification E-335
E.3.4. Specific Hypothesis for Mode of Action (MOA) of Trichloroethylene
(TCE) Hepatocarcinogenicity in Rodents E-338
E.3.4.1. PPARa Agonism as the Mode of Action (MOA) for Liver
Tumor Induction—The State of the Hypothesis E-338
E.3.4.2. Other Trichloroethylene (TCE) Metabolite Effects That
May Contribute to its Hepatocarcinogenicity E-368
E.4. EFFECTS OF COEXPOSURES ON MODE OF ACTION (MOA)—
INTERNAL AND EXTERNAL EXPOSURES TO MIXTURES
INCLUDING ALCOHOL E-379
E.4.1. Internal Coexposures to Trichloroethylene (TCE) Metabolites:
Modulation of Toxicity and Implications for TCE Mode of Action
(MOA) E-381
E.4.2. Initiation Studies as Coexposures E-382
E.4.2.1. Herren-Freund et al., 1987 E-382
E.4.2.2. Parnell et al., 1986 E-383
E.4.2.3. Pereira and Phelps, 1996 E-384
E.4.2.4. Tao et al., 2000 E-389
E.4.2.5. Lantendresse and Pereira, 1997 E-390
E.4.2.6. Pereira et al., 1997 E-392
E.4.2.7. Taoetal., 1998 E-394
E.4.2.8. Stauber et al., 1998 E-395
E.4.3. Coexposures of Haloacetates and Other Solvents E-397
E.4.3.1. Carbon tetrachloride, Dichloroacetic Acid (DCA),
Trichloroacetic Acid (TCA): Implications for Mode of
Action (MOA) from Coexposures E-397
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-iv DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
E.4.3.2. Chloroform, Dichloroacetic Acid (DCA), and
Trichloroacetic Acid (TCA) Coexposures: Changes in
Methylation Status E-400
E.4.3.3. Coexposures to Brominated Haloacetates: Implications for
Common Modes of Action (MO As) and Background
Additivity to Toxicity E-402
E.4.3.4. Coexposures to Ethanol: Common Targets and Modes of
Action (MOAs) E-404
E.4.3.5. Coexposure Effects on Pharmacokinetics: Predictions
Using Physiologically Based Pharmacokinetic (PBPK)
Models E-406
E.5. POTENTIALLY SUSCEPTIBLE LIFE STAGES AND CONDITIONS
THAT MAY ALTER RISK OF LIVER TOXICITY AND CANCER E-409
E.6. UNCERTAINTY AND VARIABILITY E-410
E.7. REFERENCES E-410
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-v DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
E-l. Mice data for 13 weeks: mean body and liver weights E-87
E-2. Prevalence and Multiplicity data from DeAngelo et al. (1999) E-150
E-3. Difference in pathology by inclusion of unscheduled deaths from DeAngelo et
al. (1999) E-150
E-4. Comparison of data from Carter etal. (2003) and DeAngelo et al. (1999) E-156
E-5. Prevalence of foci and tumors in mice administered NaCl, DCA, or TCA from
Pereira(1996) E-160
E-6. Multiplicity of foci and tumors in mice administered NaCl, DCA, or TCA from
Pereira(1996) E-161
E-7. Phenotype of foci reported in mice exposed to NaCl, DCA, or TCA by Pereira
(1996) E-162
E-8. Phenotype of tumors reported in mice exposed NaCl, DCA, or TCA by Pereira
(1996) E-162
E-9. Multiplicity and incidence data (31 week treatment) from Pereira and Phelps
(1996) E-164
E-10. Comparison of descriptions of control data between George et al. (2000) and
DeAngelo et al. (2008) E-181
E-l 1. TCA-induced increases in liver tumor occurrence and other parameter over
control after 60 weeks E-l87
E-12. TCA-induced increases in liver tumor occurrence after 104 wks E-191
E-13. Comparison of liver effects from TCE, TCA, and DCA E-207
E-14. Liver weight induction as percent liver/body weight fold-of-control in male
B6C3F1 mice from DCA or TCA drinking water studies E-210
E-l5. Liver weight induction as percent liver/body weight fold-of-control in male
B6C3F1 or Swiss mice from TCE gavage studies E-211
E-16. B6C3F1 and Swiss (data sets combined) E-212
E-17. Power calculations for experimental design described in text, using Pereira
et al. as an example E-249
E-18. Comparison between results for Yang et al. (2007b) and Cheung et al. (2004) E-363
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-vi DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
E-l. Comparison of average fold-changes in relative liver weight to control and
exposure concentrations of 2 g/L or less in drinking water for TCA and DCA
inmaleB6C3Fl mice for 14-30 days E-215
E-2. Comparisons of fold-changes in average relative liver weight and gavage dose
of male B6C3F1 mice for 10-28 days of exposure and in male B6C3F1 and
Swiss mice E-217
E-3. Comparison of fold-changes in relative liver weight for data sets in male
B6C3F1, Swiss, and NRMI mice between TCE studies [duration 28-42 days]
and studies of direct oral TCA administration to B6C3 Fl mice [duration
14-28 days] E-220
E-4. Fold-changes in relative liver weight for data sets in male B6C3F1, Swiss,
and NRMI mice reported by TCE studies of duration 28-42 days using
internal dose metrics predicted by the PBPK model described in Section E.3.5 E-222
E-5. Comparison of Ito et al. and David et al. data for DEHP tumor induction from
Guytonetal. (2009) E-348
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-vii DRAFT—DO NOT CITE OR QUOTE
-------
FOREWORD
The purpose of this Appendix is to provide scientific support and rationale for the hazard
and dose-response sections of the Toxicological Review of Trichloroethylene (TCE) regarding
liver effects and those of coexposures. It is not intended to be a comprehensive treatise on the
chemical or toxicological nature of TCE. Please refer to the Toxicological Review of TCE for
characterization of EPA's overall confidence in the quantitative and qualitative aspects of hazard
and dose-response for TCE-induced liver effects. Matters considered in this appendix include
knowledge gaps, uncertainties, quality of data, and scientific controversies. This characterization
is presented in an effort to make apparent the scientific issues regarding the data and MOA
considerations for experimental animal data for liver effects in the TCE assessment.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-viii DRAFT—DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGER
Weihsueh A. Chiu
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
PRINCIPAL AUTHOR
Jane C. Caldwell
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
CONTRIBUTORS
Weihsueh A. Chiu
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
Marina V. Evans
National Health and Environmental Effects Research Laboratory
(on detail to National Center for Environmental Assessment—Washington Office)
U.S. Environmental Protection Agency
Research Triangle Park, NC
KathrynZ. Guyton
National Center for Environmental Assessment—Washington Office
U.S. Environmental Protection Agency
Washington, DC
ACKNOWLEDGMENTS
The author and contributors would like to thank the NCEA management team for their
comments and support, and Terri Konoza and the TSS for their extensive technical editing
support.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-ix DRAFT—DO NOT CITE OR QUOTE
-------
1 APPENDIX E. ANALYSIS OF LIVER AND COEXPOSURE ISSUES FOR THE
2 TCE TOXICOLOGICAL REVIEW
3
4
5 E.I. BASIC PHYSIOLOGY AND FUNCTION OF THE LIVER—A STORY OF
6 HETEROGENEITY
7 The liver is a complex organ whose normal function and heterogeneity are key to
8 understanding and putting into context perturbations by trichloroethylene (TCE), cancer biology,
9 and variations in response observed and anticipated for susceptible life stages and background
10 conditions.
11
12 E.I.I. Heterogeneity of Hepatocytes and Zonal Differences in Function and Ploidy
13 Malarkey et al. (2005) state that (1) the liver transcriptome (i.e., genes expressed as
14 measured by mRNA) is believed only second to the brain in its complexity and includes about
15 25-40% of the approximately 50,000 mammalian genes, (2) during disease states the
16 transcriptome can double or triple and its increased complexity is due not only to differential
17 gene expression (up- and down-regulation of genes) but also to the mRNA contributions from
18 the heterogeneous cell populations in the liver, and (3) when one considers that over a dozen cell
19 types comprise the liver in varying proportions, particularly in disease states, knowledge about
20 the cell types and cell-specific gene expression profiles help unravel the complex genomic and
21 protenomic data sets. Gradients of gene and protein activity varying from the periportal region
22 to the centrilobular region also exist for sinusoidal endothelial cells, Kuffper cells, hepatic
23 stellate cells, and the matrix in the space of Disse. Malarkey et al. (2005) also estimate that
24 hepatocytes constitute 60%, sinusoidal endothelial cells 20%, Kupffer cells 15%, and stellate
25 cells 5% of liver cells. Therefore, in experimental paradigms where liver homogenates are used
26 for the determination of "changes in liver," gene expression, or other parameters the individual
27 changes from cells residing in differing zones and by differing cell type is lost. Malarkey et al.
28 (2005) define the need to better characterize the histological cellular components of the tissues
29 from which mRNA and protein is extracted and referred to "phenotypic anchoring" and cite
30 acetaminophen as a "model hepatotoxicant under study to assess the strengths and weaknesses of
31 genomics and proteinomics technologies" as well as "a good example for understanding and
32 utilizing phenotypic anchoring to better understand genomics data." After acetaminophen
33 exposure "there is an unexplained and striking inter and intralobular variability in acute hepatic
34 necrosis with some regions having massive necrosis and adjacent areas within the same lobe or
35 other lobes showing no injury at all." Malarkey et al. (2005) go on to cite similar lobular
36 variability in response for "copper distribution, iron and phosphorous, chemical and spontaneous
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-1 DRAFT—DO NOT CITE OR QUOTE
-------
1 carcinogenesis, cirrhosis and regeneration" and suggest that although uncertain "factors such as
2 portal streamlining of blood to the liver, redistribution of blood to core of the liver secondary to
3 nerve stimulation, and exposures during fetal development and possibly lobular gradients are
4 important." Hepatic interlobe differences exist for initiating agents in terms of DNA alkylation
5 and cell replication. In the rat, diethylnitrosamine (DEN) alkylation has been reported to occur
6 preferentially in the left and right median lobes, while cell replication was higher in the right
7 median and right anterior lobes (Richardson et al., 1986). Richardson et al. (1986) reported that
8 exposure to DEN induced a 100% incidence of hepatocellular carcinoma (HCC) in the left,
9 caudate, left median and right median lobes of the liver by 20 weeks versus only 30% in the right
10 anterior and right posterior hepatic lobes. There was a reported interlobe difference in adduct
11 formation, cell proliferation, liver lobe weight gain, number and size of y-glutamyltranspeptidase
12 (GGT)+ foci, and carbon 14 labeling from a single dose of DEN. Richardson et al. (1986)
13 suggest that many growth-select!on studies utilizing the liver to evaluate the carcinogenic
14 potential of a chemical often focus on only one or two of the hepatic lobes, which is especially
15 true for partial hepatectomy, and that for DEN and possibly other chemicals this procedure
16 removes the lobes most likely to get tumors. Thus, the "distribution of toxic insult may not be
17 correctly assessed with random sampling of the liver tissue for microarray gene expression
18 analysis" (Malarkey et al., 2005) and certainly any such distributional differences are lost in
19 studies of whole-liver homogenates.
20 The liver is normally quiescent with few hepatocytes undergoing mitosis and, as
21 described below, normally occurring in the periportal areas of the liver. Mitosis is observed only
22 in approximately one in every 20,000 hepatocytes in adult liver (Columbano and
23 Ledda-Columbano, 2003). The studies of Schwartz-Arad et al. (1989), Zajicek et al. (1991),
24 Zajicek and Schwartz-Arad (1990), and Zajicek et al. (1989) have specifically examined the
25 birth, death, and relationship to zone of hepatocytes as the "hepatic streaming theory." They
26 report that hepatocytes and littoral cells continuously steam from the portal tract toward the
27 terminal hepatic vein and that the hepatocyte differentiates as it goes with biological age closely
28 related to cell differentiation. In other words, the acinus may be represented by a tube with two
29 orifices: for cell inflow situated at the portal tract rim and other for cell outflow, at the terminal
30 hepatic vein with hepatocytes streaming through the tube in an orderly fashion. In normal liver,
31 cell proliferation is suggested as the only driving force of this flow with each mitosis associated
32 with displacement of the cells by one cell location and the greater the cell production, the faster
33 the flow and visa versa (Zajicek et al., 1991). Thus, the microscopic section of the liver
34 "displays an instantaneous image of a tissue in flux" (Schwartz-Arad et al., 1989). Schwartz -
3 5 Arad et al. (1989) further suggest that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-2 DRAFT—DO NOT CITE OR QUOTE
-------
1 throughout its life the hepatocyte traverses three acinus zones; in each it is
2 engaged in different metabolic activity. When young it performs among other
3 functions gluconeogenesis, which is found in zone 1 hepatocytes (i.e. periportal),
4 and when old it turns into a zone 3 cell (i.e., pericentral), with a pronounced
5 glycolitic make up. The three zones thus represent differentiation stages of the
6 hepatocyte, and since they differ by their distance from the origin, e.g. zone 2
7 (i.e., midzonal) is more distant than zone 1, again, hepatocyte differentiation is
8 proportional to its distance.
9
10 Chen et al. (1995) report that
11
12 Hepatocytes are a heterogeneous population that are composed of cells expressing
13 different patterns of genes. For example, gamma-glutamyl transpeptidase and
14 genes related to gluconeogenesis are expressed preferential in periportal
15 hepatocytes, whereas enzymes related to glycolysis are more abundant in the
16 centrilobular area. Glutamine synthetase is expressed in a small number of
17 hepatocytes surrounding the central veins. Most cytochrome p450 enzymes are
18 expressed or induced preferentially in centrilobular hepatocytes relative to
19 periportal hepatocytes.
20
21 Along with changes in metabolic function, Vielhauer et al. (2001) reported that there is evidence
22 of zonal differences in carcinogen DNA effects and, also, chemical-specific differences for DNA
23 repair enzyme and that enhanced DNA repair is a general feature of many carcinogenic states
24 including the enzymes that repair alkylating agents but also oxidative repair. As part of this
25 process of differentiation and as livers age, the hepatocyte changes and increases its ploidy with
26 polyploid cells predominant in zone 2 of the acinus (Schwartz-Arad et al., 1989). The reported
27 decrease in DNA absorbance in zone 3 may be due to (1) a decline in chromatin affinity to the
28 dye, (2) cell death, and (3) DNA exit from intact cells and Zajicek and Schwartz-Arad (1990)
29 suggest that the fewer metabolic demands in Zone 3, under normal conditions, causes the cell to
30 "deamplify" its genes and for DNA excess to leak out cells adjacent to the terminal hepatic vein
31 or to be eliminated by apoptosis reflecting cell death. Thus, the three acinus zones represent
32 differentiation states of one and the same hepatocyte, which increase ploidy as functional
33 demands change. Zajicek and Schwartz-Arad (1990) also report that nuclear size is generally
34 proportional to DNA content and that as DNA accumulates, the nucleus enlarges. This has
35 import for histopathological descriptions of hepatocellular hypertrophy and attendant nuclear
36 changes after toxic insult as well.
37 The gene amplification associated with polyploidy is manifested by DNA accumulation
38 that involves the entire genome (Zajicek and Schwartz-Arad, 1990). Polyploidization is always
39 attended by the intensification of the transcription and translation and in rat liver the amino acid
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 DRAFT—DO NOT CITE OR QUOTE
-------
1 label and activity of many enzymes increases proportionately to their ploidy. "Individual
2 chromosomes of a tetraploid genome of a hepatocyte reduplicate in the same sequence as in a
3 diploid one. In this case the properties of the chromosomes evidently remain unchanged and
4 polyploidy only means doubling the indexes of the diploid genome" (Brodsky and Uryvaeva,
5 1977). Polyploidy will be manifested in the liver by either increases in the number of
6 chromosomes per nucleus in an individual cell or by the appearance of two nuclei in a single cell.
7 Most cell polyploidization occurs in youth with mitotic polyploidization occurring
8 predominantly from 2 to 3 weeks postnatally and increases with age in mice (Brodsky and
9 Uryvaeva, 1977). Hepatocytes progress through a modified or polyploidizing cell cycle which
10 contains gaps and S-phases, but proceeds without cytokinesis. The result is the formation of the
11 first polyploidy cell, which is binucleated with diploid nuclei and has increased cell ploidy but
12 not cell number. The subsequent proliferation of bi-nucleated hepatocytes occurs with a fusion
13 of mitotic nuclei during metaphase that gives rise to mononucleated cells with higher levels of
14 ploidy. Thus, during normal liver ontogenesis, a polyploidizing cell cycle without cytokinesis
15 alternates with a mitotic cycle of binucleated cells and results in progressive and irreversible
16 increases in either cell or nuclear ploidy (Brodsky and Uryvaeva, 1977).
17 Polyploidization of the liver occurs during maturation in rodents and therefore,
18 experimental paradigms that treat or examine rodent liver during that period should take into
19 consideration the normally changing baseline of polyploidy in the liver. The development of
20 polyploidy has been correlated in rodents to correspond with maturation. Brodsky and Uryvaeva
21 (1977) report it is cells with diploid nuclei that proliferate in young mice, but that among the
22 newly formed cells, the percentage of those with tetraploid nuclei is high. By 1 month, most
23 mice (CBA/C57BL mice) already have a polyploid parenchyma, but binucleate cells with diploid
24 nuclei predominate. In adult mice, the ploidy class with the highest percentage of hepatocytes
25 was the 4n X 2 class. The intensive proliferation of diploid hepatocytes occurs only in baby
26 mice during the first 2 weeks of life and then toward 1 month, the diploid cells cease to maintain
27 themselves and transform into polyploid cells. In aged animals, the parenchyma retains only
28 0.02 percent of the diploid cells of the newborn animal. While the weight of the liver increases
29 almost 30 times within 2 years, the number of cells increase much less than the weight or mean
30 ploidy. Hence, the postnatal growth of the liver parenchyma is due to cell polyploidization
31 (Brodsky and Uryvaeva, 1977). In male Wistar rats fetal hepatocytes (22 days gestation) were
32 reported to be 85.3% diploid (2n) and 7.4% polyploid (4n + 8n) cells with 7.3% of cells in
33 S-phase (SI and S2). By one month of age (25-day old suckling rats) there were 92.9% diploid
34 and 2.5% polyploid, at 2 months 47.5% diploid and 50.9% polyploid, at 6 months 29.1% diploid
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 and 69.6% polyploid, and by 8 months 11.1% diploid and 87.3% polyploidy (Sanz et al., 1996).
2 However, mouse and rat differ in their polyploidization.
3
4 In the mouse, which has a higher degree of polyploidy than the rats, the scheme of
5 polyploidization differs in that each cell class, including mononucleate cells,
6 forms from the preceding one without being supplemented by self-maintenance.
7 Each cell class is regarded as the cell clone and it is implied that the cells of each
8 class have the same mitotic history and originate from diploid initiator cells with
9 similar properties. In this model 1 reproduction would give a 2n x 2 cell, the
10 second reproduction a 4n cell, and third reproduction a 4n X 2 cell all coming
11 from an originator diploid cell (Brodsky and Uryvaeva, 1977).
12
13 The cell polyploidy is most extensive in mouse liver, but also common for rat and
14 humans livers. The livers of young and aged mice differ considerably in the ploidy of the
15 parenchymal cells, but still perform fundamentally the same functions. In some mammals, such
16 as the mouse, rats, dog and human, the liver is formed of polyploid hepatocytes. In others, for
17 example, guinea pig and cats, the same functions are performed by diploid cells (Brodsky and
18 Uryvaeva, 1977). One obvious consequence of polyploidization is enlargement of the cells. The
19 volume of the nucleus and cytoplasm usually increases proportionately to the increased in the
20 number of chromosome sets with polyploidy reducing the surface/volume ratio. The labeling of
21 tritium doubles with the doubling of the number of chromosomes in the hepatocyte nucleus
22 (Brodsky and Uryvaeva, 1977). Kudryavtsev et al. (1993) have reported that the average levels
23 of cell and nuclear ploidy are relatively lower in humans than in rodent but the pattern of
24 hepatocyte polyploidization is similar and at maturity and especially during aging, the rate of
25 hepatocyte polyploidization increases with elderly individuals having binucleated and polyploid
26 hepatocytes constituting about one-half of liver parenchyma. Gramantieri et al. (1996) report
27 that in adult human liver a certain degree of polyploidization is physiological; the polyploidy
28 compartment (average 33% of the total hepatocytes) includes both mononucleated (28%) and
29 binucleated (72%) cells and the average percentage of binucleated cells in the total hepatocyte
30 population is 24% (Melchiorri et al., 1994). Historically, aging in human liver has been
31 characterized by fewer and larger hepatocytes, increased nuclear polyploidy and a higher index
32 of binucleate hepatocytes (Popper, 1986) but Schmucker (2005) notes that data concerning the
33 effect of aging on hepatocyte volume in rodent and humans are in conflict with some showing
34 increases volume to be unchanged and to increase by 25% by age 60 by others in humans. The
35 irreversibility of hepatocyte polyploidy has been used in efforts to identify the origin of tumor
36 progenitor cells (diploid vs. polyploidy) (see Section E.3.1.8, below). The associations with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-5 DRAFT—DO NOT CITE OR QUOTE
-------
1 polyploidy and disease have been an active area of study in cancer mode-of-action (MOA)
2 studies (see SectionsE.3.1.4 and E.3.3.1, below).
3 Not only are polyploid cells most abundant in zone 2 of the liver acinus and increase in
4 number with age, but polyploid cells have been reported to be more abundant following a
5 number of toxic insults and exposure to chemical carcinogens. Wanson et al. (1980) reported
6 that one of the earliest lesions obtained in the liver after 7V-nitrosomorpholine treatment
7 development of hypertrophic parenchymal cells presenting a high degree of ploidy. Gupta
8 (2000) reports hepatic polyploidy is often encountered in the presence of liver disease and that
9 for animals and people, polyploidy is observed during advancement of liver injury due to
10 cirrhosis or other chronic liver disease (often described as large-cell dysplasia referring to
11 nuclear and cytoplasmic enlargement, nuclear pleomorphisms and multinucleation and probably
12 representing increased prevalence of polyploidy cells) and in old animals with toxic liver injury
13 and impaired recovery. Gorla et al. (2001) report that weaning and commencement of feeding,
14 compensatory liver hypertrophy following partial hepatectomy, toxin and drug-induced liver
15 disease, and administration of specific growth factors and hormones may induce hepatic
16 polyploidy. They go on to state that "although liver growth control has long been studied,
17 whether the replication potential of polyploidy hepatocytes is altered remains unresolved, in part,
18 owing to difficulties in distinguishing between cellular DNA synthesis and generation of
19 daughter cells." Following CCL4 intoxication, the liver ploidy rises and more cells become
20 binucleate (Zajicek et al., 1989). Minamishima et al. (2002) report that in 8-12 week old female
21 mice before partial hepatectomy there were 78.6% 2C, 19.1% 4C, and 2.3% 8C cells but 7 days
22 after there were 42.0% 2C, 49.1% 4C, and 9.0% 8C. Zajicek et al. (1991) describe how
23 hepatocyte streaming is affected after the rapid hepatocyte DNA synthesis that occurs after the
24 mitogenic stimulus of a partial hepatectomy. These data are of relevance to findings of increased
25 DNA synthesis and liver weight gain following toxic insults and disease states. Zajicek et al.
26 (1991) suggest that following a mitogenic stimulus, not all DNA synthesizing cells do divide but
27 accumulate newly formed DNA and turn polyploid (i.e., during the first 3 days after partial
28 hepatectomy in rats 50% of synthesized DNA was accumulated) and that since the acinus
29 increased 15% and cell density declined 10%, overall cell mass increased 5%. However, cell
30 influx rose 1,300%. "In order to accommodate all these cells, the 'acinus-tube' ought to swell
31 13-fold, while in reality it increased only 5%" and that on day 3 "the liver remnant did not even
32 double in its size." Zajicek et al. conclude that apparently "cells were eliminated very rapidly,
33 and may have even been sloughed off, since the number of apoptotic bodies was very low" and
34 therefore, "partial hepatectomy triggers two processes: an acute process lasting about a week
35 marked by massive and rapid cell turnover during which most newly formed hepatocytes are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-6 DRAFT—DO NOT CITE OR QUOTE
-------
1 eliminated, probably sloughed off into the sinusoids; and a second more protracted process
2 which served for liver mass restoration mainly by forming new acini." Thus, a mitogenic
3 stimulus may induce increased ploidy and increased cell number as a result of increased DNA
4 synthesis, and many of the rapidly expanding number of cells resulting from such stimulation are
5 purged and therefore, do not participate in subsequent disease states of the liver.
6 Zajicek et al. (1989) note that the accumulation of DNA rather than proliferation of
7 hepatocytes "should be considered when evaluating the labeling index of hepatocytes labeled
8 with tritiated thymidine" as the labeling index, defined as the proportion of labeled cells, can
9 serve as a proliferation estimate only if it is assumed that a synthesizing cell will ultimately
10 divide. In tissues, such as the liver, "where cells also accumulate DNA, proliferation estimates
11 based on this index may fail" (Zajicek et al., 1989). The tendency to accumulate DNA is also
12 accompanied by a decreasing probability of a cell to proliferate, since young hepatocytes
13 generally divide after synthesizing DNA while older cells prefer instead to accumulate DNA.
14 However, polyploidy per se does not preclude cells from dividing (Zajicek et al., 1989). The
15 ploidy level achieved by the cell, no matter how high, does not, in itself, prevent it from going
16 through the next mitotic cycle and the reproduction of hepatocytes in the ploidy classes of 8n and
17 8n X 2 is common phenomenon (Brodsky and Uryvaeva, 1977). However, along with a reduced
18 capacity to proliferate, Sigal et al. (1999) report that the onset of polyploidy increases the
19 probability of cell death. The proliferative potentials of hepatocytes not only depend on their
20 ploidy, but also on the age of the animals with liver restoration occurring more slowly in aged
21 animals after partial hepatectomy (Brodsky and Uryvaeva, 1977). Species differences in the
22 ability of hepatocytes to proliferate and respond to a mitogenic stimulus have also been
23 documented (see Section E.3.4.2, below). The importance of the issues of cellular proliferation
24 versus DNA accumulation and the differences in ability to respond to a mitogenic stimulus
25 becomes apparent as identification of the cellular targets of toxicity (i.e., diploid vs. polyploidy)
26 and the role of proliferation in proposed MO As are brought forth. Polyploidization, as discussed
27 above, has been associated with a number of types of toxic injury, disease states, and
28 carcinogenesis by a variety of agents.
29
30 E.1.2. Effects of Environment and Age: Variability of Response
31 The extent of polyploidization of the liver not only changes with age, but structural and
32 functional changes, as well as environmental factors (e.g., polypharmacy), affect the
33 vulnerability of the liver to toxic insult. In a recent review by Schmucker (2005), several of
34 these factors are discussed. Schmucker reports that approximately 13% of the population of the
35 United States is over the age of 65 years, that the number will increase substantially over the next
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-7 DRAFT—DO NOT CITE OR QUOTE
-------
1 50 years, and that increased age is associated with an overall decline in health and vitality
2 contributing to the consumption of nearly 40% of all drugs by the elderly. Schmucker estimates
3 that 65% of this population is medicated and many are on polypharmacy regimes with a major
4 consequence of a marked increase in the incidence of adverse drug reactions (ADRs) (i.e., males
5 and females exhibit 3- and 4-fold increases in ADRs, respectively, when 20- and 60-year-old
6 groups are compared). The percentage of deaths attributed to liver diseases dramatically
7 increases in humans beyond the age of 45 years with data from California demonstrating a 4-fold
8 increase in liver disease-related mortality in both men and women between the ages of 45 and
9 85 years (Seigel and Kasmin, 1997). Furthermore, Schmucker cites statistics from the United
10 Stated Department of Health and Human Services to illustrate a loss in potential lifespan prior to
11 75 years of age due to liver disease (i.e., liver disease reduced lifespan to a greater extent than
12 colorectal and prostatic cancers, to a similar extent as chronic obstructive pulmonary disease, and
13 nearly as much as HIV). Thus, the elderly are predisposed to liver disease.
14 As stated above, the presence of high polyploidy cell in normal adults, nuclear
15 polyploidization with age, and increase in the mean nuclear volume have been reported in
16 people. Wantanabe et al. (1978) reported the results from a cytophotometrical analysis of
17 35 cases of sudden death including 22 persons over 60 years of age that revealed that although
18 the nuclear size of most hepatocytes in a senile liver remains unchanged, there was an increase in
19 cells with larger nuclei. Variations in both cellular area and nucleocytoplasmic ratio were also
20 analyzed in the study, but the binuclearity of hepatocytes was not considered. No cases with a
21 clinical history of liver disease were included. Common changes in senile liver were reported to
22 include atrophy, fatty metamorphosis of hepatocytes, and occasional collapse of cellular cords in
23 the centrilobular area, slight cellular infiltration and proliferation of Kupffer cells in sinusoids,
24 and elongation of Glisson's triads with slight to moderate fibrosis in association with round cell
25 infiltration. Furthermore, cells with giant nuclei, with each containing two or more prominent
26 nucleoli, and binuclear cell. There was a decrease in diploid populations with age and an
27 increase in tetraploid population and a tendency of polyploidy cells with higher values than
28 hexaploids with age. Cells with greater nuclear size and cellular sizes were observed in livers
29 with greater degrees of atrophy.
30 Schmucker notes that one of the most documented age-related changes in the liver is a
31 decline in organ volume but also cites a decrease in functional hepatocytes and that other studies
32 have suggested that the size or volume of the liver lobule increases as a function of increasing
33 age. Data are cited for rats suggesting sinusoidal perfusion rate in the rat liver remains stable
34 throughout the lifespan (Vollmar et al., 2002) but evidence in humans shows age-related shifts in
35 the hepatic microcirculation attributable to changes in the sinusoidal endothelium (McLean et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-8 DRAFT—DO NOT CITE OR QUOTE
-------
1 2003) (i.e., a 60% thickening of the endothelial cell lining and an 80% decline in the number of
2 endothelial cell fenestrations, or pores, with increasing age in humans) that are similar in baboon
3 liver (Cogger et al., 2003). Such changes could impair sinusoidal blood flow and hepatic
4 perfusion, and the uptake of macromolecules such as lipoproteins from the blood. Schmucker
5 reports that there is a consensus that hepatic volume and blood flow decline with increasing age
6 in humans but that the effects of aging on hepatocyte structure are less clear. In rats, the volume
7 of individual hepatocytes was reported to increase by 60% during development and maturation,
8 but subsequently decline during senescence yielding hepatocytes of equivalent volumes in
9 senescent and very young animals (Schmucker, 2005). The smooth surfaced endoplasmic
10 reticulum (SER), which is the site of a variety of enzymes involved in steroid, xenobiotic, lipid
11 and carbohydrate metabolism, also demonstrated a marked age-related decline rat hepatocytes
12 (Schumucker et al., 1977, 1978). Schmucker also notes that several studies have reported that
13 the older rodents have less effective protection against oxidative injury in comparison to the
14 young animals, age-related decline in DNA base excision repair, and increases in the level of
15 oxidatively damaged DNA in the livers of senescent animals in comparison to young animals.
16 Age-related increases in the expression an activity of stress-induced transcription factors (i.e.,
17 increased NF-KB binding activity but not expression) were also noted, but that the importance of
18 changes in gene expression to the role of oxidative stress in the aging process remains unsolved.
19 An age-related decline in the proliferative response of rat hepatocytes to growth factors
20 following partial hepatectomy was noted, but despite a slower rate of hepatic regeneration, older
21 livers eventually achieved their original volume with the mechanism responsible for the age-
22 related decline in the posthepatectomy hepatocyte proliferative response unidentified. As with
23 other tissues, telomere length has been identified as a critical factor in cellular aging with the
24 sequential shortening of telomeres to be a normal process that occurs during cell replication (see
25 Sections E.3.1.1 and E.3.1.7, below). An association in telomere length and strain susceptibility
26 for carcinogenesis in mice has been raised. Herrera et al., (1999) examined susceptibility to
27 disease with telomere shortening in mice. However, this study only cites shorter telomeres for
28 C57BL6 mice in comparison to mixed C57BL6/129sv mice. The actual data are not in this paper
29 and no other strains are cited. Of the differing cell types examined, Takubo and Kaminishi
30 (2001) report that hepatocytes exhibited the next fastest rate of telomere shortening despite being
31 relatively long-lived cells raising the question of whether or not there are correlations between
32 age, hepatocyte telomere length and the incidence of liver disease (Schmucker, 2005). Aikata et
33 al. (2000) and Takubo et al. (2000) report that the mean telomere length in healthy livers is
34 approximately 10 kilobase pairs at 80 years of age and these hepatocytes retain their proliferative
35 capacity but that in diseased livers of elderly subjects was approximately 5 kb pairs. Thus, short
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-9 DRAFT—DO NOT CITE OR QUOTE
-------
1 telomere length may compromise hepatic regeneration and contribute to a poor prognosis in liver
2 disease or as a donor liver (Schmucker, 2005).
3 Schmucker (2005) reports that interindividual variability in Phase I drug metabolism was
4 so large in human liver microsomes, particularly among older subjects, that the determination of
5 any statistically significant age or gender-related differences were precluded. In fact Schmucker
6 (2001) notes that "the most remarkable characteristic of liver function in the elderly is the
7 increase in interindividual variability, a feature that may obscure age-related differences."
8 Schumer notes that The National Institute on Aging estimates that only 15% of individuals aged
9 over 65 years exhibit no disease or disability with this percentage diminishing to 11 and 5% for
10 men and women respectively over 80 years. Thus, the large variability in response and the
11 presence of age-related increases in pharmacological exposures and disease processes are
12 important considerations in predicting potential risk from environmental exposures.
13
14 E.2. CHARACTERIZATION OF HAZARD FROM TRICHLOROETHYLENE (TCE)
15 STUDIES
16 The 2001 Draft assessment of the health risk assessment of TCE (U.S. EPA, 2001)
17 extensively cited the review article by Bull (2000) to describe the liver toxicity associated with
18 TCE exposure in rodent models. Most of the attention has been paid to the study of TCE
19 metabolites, rather than the parent compound, and the review of the TCE studies by Bull (2000)
20 was cursory. In addition, gavage exposure to TCE has been associated with a significant
21 occurrence of gavage-related accidental deaths and vehicle effects, and TCE exposure through
22 drinking water has been reported to decrease palatability and drinking water consumption, and to
23 have significant loss of TCE through volatilization, thus, further limiting the TCE database. In
24 its review of the draft assessment, U.S. Environmental Protection Agency (U.S. EPA)'s Science
25 Advisory regarding this topic suggested that in its revision, the studies of TCE should be more
26 fully described and characterized, especially those studies considered to be key for the hazard
27 assessment of TCE. Although the database for studies of the parent compound is somewhat
28 limited, a careful review of the rodent studies involving TCE can bring to bring to light the
29 consistency of observations across these studies, and help inform many of the questions
30 regarding potential MO As of TCE toxicity in the liver. Such information can inform current
31 MOA hypothesis (e.g., such as peroxisome proliferator activated receptor alpha [PPARa]
32 activation) as well. Accordingly the primary acute, subchronic and chronic studies of TCE will
33 be described and examined in detail below and with comments on consistency, major
34 conclusions and the limitations and uncertainties that their design and conduct. Since all chronic
35 studies were conducted primarily with the goal of ascertaining carcinogenicity, their descriptions
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-10 DRAFT—DO NOT CITE OR QUOTE
-------
1 focus on that endpoint, however, any noncancer endpoints described by the studies are described
2 as well. For details regarding evidence of hepatotoxicity in humans and associations with
3 increased risk of hepatocellular carcinoma, please refer to Sections 4.5.1 and 4.5.2. Given that
4 some of the earlier studies with TCE were contaminated with epichlorhydrin, only the ones
5 without such contamination are examined below.
6
7 E.2.1. Acute Toxicity Studies
8 A number of acute studies have been undertaken to describe the early changes in the liver
9 after TCE administration with the majority using the oral gavage route of administration. Some
10 have been detailed examinations while others have reported primarily liver weight changes as a
11 marker of TCE-response. The matching and recording of age but especially initial and final
12 body weight for control and treatment groups is of particular importance for studies using liver
13 weight gain as a measure of TCE-response as difference in these parameter affect TCE-induced
14 liver weight gain. Most data are for exposures of at least 10 days.
15
16 E.2.1.1. Soni et al, 1998
17 Soni et al. (1998) administered TCE in corn oil to male Sprague-Dawley (S-D) rats
18 (200-250 g, 8-10 weeks old) intraperitoneally at exposure levels of 250, 500, 1,250, and
19 2,500 mg/kg. Groups (4-6 animals per group) were sacrificed at 0, 6, 12, 24, 36, 48, 72, and
20 96 hours after administration of TCE or corn oil. Using this paradigm only 50% of rats survived
21 the 2,400 mg/kg intraperitoneal (i.p.) TCE administration with all deaths occurring between days
22 1 and 3 after TCE administration. Tritiated thymidine was also administered i.p. to rats 2 hours
23 prior to euthanasia. Light microscopic sections of the central lobe in 3-4 sections examined for
24 each animal. The grading scheme reported by the authors was: 0, no necrosis; +1 minimal,
25 defined as only occasional necrotic cells in any lobule; +2, mild, defined as less than one-third of
26 the lobular structure affected; +3, moderate, defined as between one-third and two-thirds of the
27 lobular structure affected; +4 severe, defined as greater than two-thirds of the lobular structure
28 affected. At the 2,500 mg/kg dose histopathology data were obtained for the surviving rats
29 (50%). Lethality studies were done separately in groups of 10 rats. The survival in the groups of
30 rats administered TCE and sacrificed from 0 to 96 hours was given as 30% mortality at 48 hours
31 and 50% mortality by 72 hours.
32 The authors report that controls and 0-hour groups did not show sign of tissue injury or
33 abnormality. The authors only report a single number with one significant figure for each group
34 of animals with no means or standard deviations provided. In terms of the extent of necrosis
35 there is no difference between the 250 and 500 mg/kg/treated dose groups though 96 hours with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-11 DRAFT—DO NOT CITE OR QUOTE
-------
1 a single +1 given as the maximal amount of hepatocellular necrosis (minimal as defined by
2 occasional necrotic cells in any lobule). At the 1,250 mg/kg dose the maximal score was
3 achieved 24 hours after TCE administration and was reported as simply +2 (mild, defined as less
4 than one-third of lobular structure affected). The level of necrosis was reported to diminish to a
5 score of 0 72 hours after 250 mg/kg TCE with no decrease at 500 mg/kg. At 1,250 mg/kg, the
6 extent of necrosis was reported to diminish from +2 to +1 by 72 hours after administration. At
7 the 2,500 mg/kg dose (LD50 for this route) by 48 hours, the surviving rats were reported to have a
8 score of+4 (severe as defined by greater than two thirds of the lobular structure affected). The
9 authors report that
10
11 The necrosed cells were concentrated mostly in the midzonal areas and the cells
12 around central vein area were unaffected. Extensive necrosis was observed
13 between 24 and 48 hours for both 1250 and 2500 mg/kg groups. Injury was
14 maximal in the group receiving 2500 mg/kg between 36 and 48 hours as
15 evidenced by severe midzonal necrosis, vacuolization, and congestion.
16 Infiltration of polymorphonuclear cell was evident at this time as a mechanism for
17 cleaning dead cells and tissue debris from the lobules. At the highest dose, the
18 injury also started to spread toward the centrilobular areas. At highest dose, 30
19 and 50% lethality was observed at 48 and 72 h, respectively. After 48 h, the
20 number of necrotic cells decreased and the number of mitotic cells increased. The
21 groups receiving 500 and 1250 mg/kg TCE showed relatively higher mitotic
22 activity as evidenced by cells in metaphase compared to other groups.
23
24 The authors do not give a quantitative estimate or indication as to the magnitude of the number
25 of cells going through mitosis. Although there was variability in the number of animals dying at
26 1,250 mg/kg TCE exposure though this route of exposure, no indication of variability in response
27 within these treatment groups is given by the author in regard to extent of histopathological
28 changes. The authors do not comment on the manner of death using this paradigm or of the
29 effects of i.p. administration regarding potential peritonitis and inflammation.
30 TCE hepatotoxicity was "assessed by measuring plasma" sorbitol dehydrogenase (SDH)
31 and alanine aminotransferase (ALT) after TCE administration with vehicle treated control groups
32 reported to induce no increases in these enzymes. Plasma SDH levels were reported to increase
33 in a linear fashion after 250, 500, and 1,250 mg TCE/kg i.p. administration by 6 hours (i.e., ~3-,
34 10.5-, 22-, and 24.5-fold in comparison to controls from 250, 500, 1,250, and 2,500 mg/kg TCE,
35 respectively) with little difference between the 1,250 and 250 mg/kg dose. By 12 hours the 250,
36 500, and 1,250 levels has diminished to levels similar to that of the 250 mg/kg dose at 6 hours.
37 The 2,500 mg/kg levels was somewhat diminished from its 6 hour level. By 24 hours after TCE
38 administration by the i.p. route of administration all doses were similar to that of the 250-mg/kg-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-12 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCE 6-hour level. This pattern was reported to be similar for 5-, 36-, 48-, 72-, and 96-hour time
2 points as well. The results presented were the means and SE for four rats per group. The authors
3 did not indicate which rats were selected for these results from the 4-6 that were exposed in each
4 group. Thus, only SDH levels showed dose dependence in results at the 6 hour time point and
5 such increases did not parallel the patterns reported for hepatocellular necrosis from
6 histopathological examination of liver tissues.
7 For ALT, the pattern of plasma concentrations after i.p. TCE administration differed both
8 from that of SDH but also from liver histopathology. Plasma ALT levels were reported to
9 increase in a nonlinear fashion and to a much smaller extent that SDH (i.e., -2.7-, 1.9-, 2.1-, and
10 4.0-fold of controls from 250, 500, 1,250, and 2,500 mg/kg TCE, respectively). The patterns for
11 12, 24, 36, 48, 72, and 96 hours were similar to that of the 6-hour exposure and did not show a
12 dose-response. The authors injected carbon tetrachloride (2.5.mL/kg) into a separate group of
13 rats and then incubated the resulting plasma with unbuffered trichloroacetic acid (TCA; 0, 200,
14 600, or 600 nmol) and no decreases in enzyme activity in vitro at the two higher concentrations.
15 It is not clear whether in vitro unbuffered TCE concentrations of this magnitude, which could
16 precipitate proteins and render the enzymes inactive, are relevant to the patterns observed in the
17 in vivo data. The extent of extinguishing of SDH and ALT activity at the two highest TCA
18 levels in vitro were the same, suggestive of the generalized in vitro pH effect. However, the
19 enzyme activity levels after TCE exposure had different patterns, and thus, suggesting that in
20 vitro TCA results are not representative of the in vivo TCE results. Neither ALT nor SDH levels
21 corresponded to time course or dose-response reported for the histopathology of the liver
22 presented in this study.
23 Tritiated thymidine results from isolated nuclei in the liver did not show a pattern
24 consistent with either the histopathology or enzyme results. These results were for whole-liver
25 homogenates and not separated by nuclear size or cell origin. Tritiated thymidine incorporation
26 was assumed by the authors to represent liver regeneration. There was no difference between
27 treated and control animals at 6 hours after i.p. TCE exposure and only a decrease (-50%
28 decrease) in thymidine incorporation after 12 hours of the 2,500 mg/kg TCE exposure level. By
29 24 hours, there as 5.6- and 2.8-fold tritiated thymidine incorporation at the 500 and 1,250 mg/kg
30 TCE levels with the 250 and 2,500 mg/kg levels similar to controls. For 36, 48, and 72 hours
31 after i.p. TCE exposure there continued to be no dose-response and no consistent pattern with
32 enzyme or histopathological lesion patterns. The authors presented "area under the curve" data
33 for tritiated thymidine incorporation for 0 to 95 hours, which did not include control values.
34 There was a slight elevation at 500 mg/kg TCE and slight decrease at 2,500 mg/kg from the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-13 DRAFT—DO NOT CITE OR QUOTE
-------
1 250 mg/kg TCE levels. Again, these data did not fit either histopathology or enzyme patterns
2 and also can include the contribution of nonparenchymal cell nuclei as well as changes in ploidy.
3 The use of an i.p. route of administration is difficult to compare to oral and inhalation
4 routes of exposure given that peritonitis and direct contact with TCE and corn oil with liver
5 surfaces may alter results. Whereas Soni et al. (1998) report the LDso to be 2,500 mg/kg TCE
6 via i.p. administration, both Elcombe et al. (1985) and Melnick et al. (1987) do not report
7 lethality from TCE administered for 10 days at 1,500 mg/kg in corn oil, or up to 4,800 mg/kg/d
8 for 10-days in encapsulated feed. Also TCE administered via gavage or oral administration
9 through feed will enter the liver through the circulation with periportal areas of the liver the first
10 areas exposed with the entire liver exposed in a fashion dependent on blood concentrations
11 levels. However, with i.p. administration, the absorption and distribution pattern of TCE will
12 differ. The lack of concordance with measures of liver toxicity from this study and the lack
13 concordance of patterns and dose-response relationships of toxicity reported from other more
14 environmentally and physiologically relevant routes of exposure make the relevance of these
15 results questionable.
16
17 E.2.1.2. Soni et al, 1999
18 A similar paradigm and the same results were reported for Soni et al. (1999), in which
19 hepatocellular necrosis, tritiated thymidine incorporation, and in vitro inhibition of SDH and
20 ALT data were presented along with dose-response studies with ally alcohol and a mixture of
21 TCE, Thioacetamine, allyl alcohol, and chloroform. The same issues with interpretation present
22 for Soni et al. (1998) also apply to this study as well.
23
24 E.2.1.3. Okino et al, 1991
25 This study treated adult Wistar male rats (8 weeks of age) with TCE after being on a
26 liquid diet for 3 weeks and either untreated or pretreated with phenobarbital or ethanol. TCE
27 exposure was at 8,000 ppm for 2 hours, 2,000 or 8,000 ppm for 2 hours, and 500 or 2,000 ppm
28 for 8 hours. Each group contained 5 rats. Livers from rats that were not pretreated with either
29 ethanol or phenobarbital were reported to show only a few necrotic hepatocytes around the
30 central vein at 6 and 22 hours after 2 hours of 8,000-ppm TCE exposure. At increased lengths
31 and/or concentrations of TCE exposure, the frequencies of necrotic hepatocytes in the
32 centrilobular area were reported to be increased but the number of necrotic hepatocytes was still
33 relatively low (out of-150 hepatocytes the percentages of necrotic pericentral hepatocytes were
34 0.2% ± 0.4%, 0.3% ± 0.4%, 2.7% ± 1.0%, 0.2% ± 0.4%, and 3.5% ± 0.4% for control,
35 2,000 ppm TCE for 2 hours, 8,000 ppm TCE for 2 hours, 500 ppm TCE for 8 hours, and 2,000
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-14 DRAFT—DO NOT CITE OR QUOTE
-------
1 ppm TCE for 8 hours, respectively). "Ballooned" hepatocytes were reported to be zero for
2 controls and all TCE treatments with the exception of 0.3% ± 0.6% ballooned midzonal
3 hepatocytes after 8,000 ppm TCE for 2 hours exposure. Microsomal protein (mg/g/liver) was
4 increased with TCE exposure concentration and duration, but not reported to be statistically
5 significant (mg/g/liver microsomal protein was 21.2 ± 4.3, 22.0 ± 1.5, 25.9 ± 1.3, 23.3 ± 0.8, and
6 24.1 ± 1.0 for control, 2,000 ppm TCE for 2 hours, 8,000 ppm TCE for 2 hours, 500 ppm TCE
7 for 8 hours, and 2,000 ppm TCE for 8 hours, respectively). The metabolic rate of TCE was
8 reported to be increased after exposures over 2,000 ppm TCE (metabolic rate of TCE in
9 nmol/g/liver/min was 29.5 ± 5.7, 51.3 ± 6.0, 63.1 ± 16.0, 37.3 ± 3.3, and 69.5 ± 4.3 for control,
10 2,000 ppm TCE for 2 hours, 8,000 ppm TCE for 2 hours, 500 ppm TCE for 8 hours, and 2,000
11 ppm TCE for 8 hours, respectively). However, the cytochrome P450 content of the liver was not
12 reported to increase with TCE exposure concentration or duration. The liver/body weight ratios
13 were reported to increase with all TCE exposures except 500 ppm for 8 hours (the liver/body
14 weight ratio was 3.18% ± 0.15%, 3.35% ± 0.10%, 3.39% ± 0.20%, 3.15% ± 0.10%, and 3.57% ±
15 0.14% for control, 2,000 ppm TCE for 2 hours, 8,000 ppm TCE for 2 hours, 500 ppm TCE for 8
16 hours, and 2,000 ppm TCE for 8 hours, respectively). These values represent 1.05-, 0.99-, 1.06-,
17 and 1.12-fold of control in the 2,000 ppm TCE for 2 hours, 8,000 ppm TCE for 2 hours, 500 ppm
18 TCE for 8 hours, and 2,000 ppm TCE for 8 hours treatment groups, respectively, with a
19 statistically significant difference observed after 8 hours of 2,000-ppm TCE exposure. Initial
20 body weights and those 22 hours after cessation of exposure were not reported, which may have
21 affected liver weight gain. However, these data suggest that TCE-related increases in
22 metabolism and liver weight occurred as early as 22 hours after exposures of this magnitude
23 from 2 to 8 hours of TCE with little concurrent hepatic necrosis.
24 Ethanol and phenobarbital pretreatment were reported to enhance TCE toxicity. In
25 ethanol-treated rats a few necrotic hepatocytes were reported to be around the central vein along
26 with hepatocellular swelling without pyknotic nuclei at 6 hours after TCE exposure with no
27 pathological findings in the midzonal or periportal areas. At 22 hours centrilobular hepatocytes
28 were reported to have a few necrotic hepatocytes and cell infiltrations around the central vein but
29 midzonal areas were reported to have ballooned hepatocytes with pyknotic nuclei frequently
30 accompanied by cell infiltrations. In phenobarbital treated rats 6 hours after TCE exposure,
31 centrilobular hepatocytes showed prenecrotic changes with no pathological changes reported to
32 be observed in the periportal areas. By 22 hours, zonal necrosis was reported in centrilobular
33 areas or in the transition zone between centrilobular and periportal areas. Treatment with
34 phenobarbital or ethanol induced hepatocellular necrosis primarily in centrilobular areas with
35 phenobarbital having a greater effect (89.1% ± 8.5% centrilobular necrosis) at the higher dose
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-15 DRAFT—DO NOT CITE OR QUOTE
-------
1 and shorter exposure duration (8,000 ppm x 2 hours) and ethanol having a greater effect
2 (16.8% ± 5.3% centrilobular necrosis) at the lower concentration and longer duration of exposure
3 (2,000 ppm x 8 hours).
4
5 E.2.1.4. Nunes et al, 2001
6 This study was focused on the effects of TCE and lead coexposure but treated male
7 75-day old S-D rats with 2,000 mg/kg TCE for 7 days via corn-oil gavage (n = 10). The rats
8 ranged in weight from 293 to 330 g (-12%) at the beginning of treatment and were pretreated
9 with corn oil for 9 days prior to TCE exposure. TCE was reported to be 99.9% pure. Although
10 the methods section states that rats were exposed to TCE for 7 days, Table 1 of the study reports
11 that TCE exposure was for 9 days. The beginning body weights were not reported specifically
12 for control and treatment groups, but the body weights at the end of exposure were reported to be
13 342 ± 18 g for control rats and 323 ± 3 g for TCE exposed rats, and that difference (-6%) to be
14 statistically significant. Because beginning body weights were not reported, it is difficult to
15 distinguish whether differences in body weight after TCE treatment were treatment related or
16 reflected differences in initial body weights. The liver weights were reported to be 12.7 ± 1.0 g
17 in control rats and 14.0 ± 0.8 g for TCE treated rats with the percent liver/body weight ratios of
18 3.7 and 4.3%, respectively. The increase in percent liver/body weight ratio represents 1.16-fold
19 of control and was reported to be statistically significant. However, difference in initial body
20 weight could have affected the magnitude of difference in liver weight between control and
21 treatment groups. The authors report no gross pathological changes in rats gavaged with corn oil
22 or with corn oil plus TCE but observed that one animal in each group had slightly discolored
23 brown kidneys. Histological examinations of "selected tissues" were reported to show an
24 increased incidence of chronic inflammation in the arterial wall of lungs from TCE-dosed
25 animals. There were no descriptions of liver histology given in this report for TCE-exposed
26 animals or corn-oil controls.
27
28 E.2.1.5. Tao et al, 2000
29 The focus of this study was to assess the affects of methionine on methylation and
30 expression of c-Jun and C-Myc in mouse liver after 5 days of exposure to TCE (1,000 mg/kg in
31 corn oil) and its metabolites. Female 8-week old B6C3F1 mice (n = 4-6) were administered
32 TCE ("molecular biology or FIPLC grade") for 5 days with and without methionine (300 mg/kg
33 i.p.). Data regarding % liver/body weight was presented as a figure. Of note is the decrease in
34 liver/body weight ratio by methionine treatment alone (-4.6% liver/body weight for control and
35 -4.0% liver/body weight for control mice with methionine or -13% difference between these
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-16 DRAFT—DO NOT CITE OR QUOTE
-------
1 groups). Neither initial body weights nor body weights after exposure were reported by the
2 authors so that the reported effects of treatment could have reflected differences in initial body
3 weights of the mice. TCE exposure was reported to increase the percent liver/body weight ratio
4 to -5.8% without methionine and to increase percent liver/body weight ratio to -5.7% with
5 methionine treatment. These values represent 1.26-fold of control levels from TCE exposure
6 without methionine and 1.43-fold of control from TCE exposure with methionine. The number
7 of animals examined was reported to be 4-6 per group. The authors reported the differences
8 between TCE treated animals and their respective controls to be statistically significant but did
9 not examine the differences between controls with and without methionine. There were no
10 descriptions of liver histology given in this report for TCE-exposed animals or corn-oil controls.
11
12 E.2.1.6. Tucker et al, 1982
13 This study describes acute LDso, and 5- and 14-days studies of TCE in a 10% emulphor
14 solution administered by gavage. Screening level subchronic drinking water experiments with
15 TCE dissolved in 1% emulphor in mice were also conducted but with little detail reported. The
16 authors did describe the strains used (CD-I and ICR outbred albino) and that they are "weanling
17 mice," but the ages of the mice and their weights were not given. The TCE was described as
18 containing 0.004% diisopropylamine as the preservative and that the stabilizer had not been
19 found carcinogenic or overtly toxic. The authors report that "the highest concentration a mouse
20 would receive during these studies is 0.03 mg/kg/day." The main results are basically an LD50
21 study and a short term study with limited reporting for 4 and 6-month studies of TCE.
22 Importantly, the authors documented the loss of TCE from drinking water solutions (less than
23 20% of the TCE was lost during the 3 or 4 days in the water bottles at 1.0, 2.5, and 5.0 mg/mL
24 concentrations, but in the case of 0.1 mg/mL, up to 45% was lost over a 4-day period). The
25 authors also report that high doses of TCE in drinking water reduced palatability to such an
26 extent that water consumption by the mice was significantly decreased.
27 The LD50 with 95% confidence were reported to be 2,443 mg/kg (1,839 to 3,779) for
28 female mice and 2,402 mg/kg (2,065 to 2,771) for male mice. However, the number of mice
29 used in each dosing group was not given by the authors. The deaths occurred within 24 hours of
30 TCE administration and no animals recovering from the initial anesthetic effect of TCE died
31 during the 14-day observation period. The authors reported that the only gross pathology
32 observed was hyperermia of the stomach of mice dying form lethal doses of TCE, and that mice
33 killed at 14 days showed not gross pathology. In a separate experiment, male CD-I mice were
34 exposed to TCE by daily gavage for 14 days at 240 and 24 mg/kg. These two doses did not
35 cause treatment related deaths and body weight and "most" organ weights were reported by the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-17 DRAFT—DO NOT CITE OR QUOTE
-------
1 authors to not be significantly affected but the data was not shown. The only effect noted was an
2 increased liver weight, which appeared to be dose dependent but was reported to be significant
3 only at the higher dose. The only significant difference found in hematology was s 5% lower
4 hematocrit in the higher dose group. The number of animals tested in this experiment was not
5 give by the authors. Male CD-I mice (n= 11) were given TCE via gavage for 5 days (0.73 g/kg
6 TCE twice on Day 0, 1.46 g/kg twice on Day 1, 2.91 g/kg twice on Day 3, and 1.46 g/kg TCE on
7 Days 4 and 5) with only 4 of 11 mice treated with TCE surviving.
8 In a subchronic study, male and female CD-I mice received TCE in drinking water at
9 concentrations of 0, 0.1, 1.0, 2.5, and 5 mg/mL in 1% emulphor, and a naive group received
10 deionized water. There were 140 animals of each sex in the naive group and in each treatment
11 group, except for 260 mice in the vehicle groups. Thirty mice of each sex and treatment were
12 selected for recording body weights for 6 months. The method of "selection" was not given by
13 the authors. These mice were weighed twice weekly and fluid consumption was measured by
14 weighing the six corresponding water bottles. The authors reported that male mice at the two
15 highest doses of TCE consumed 41 and 66 mL/kg/day less fluid over the 6 months of the study
16 than mice consuming vehicle only and that this same decreased consumption was also seen in the
17 high dose (5 mg/mL) females. They report that weight gain was not affected except at the high
18 dose (5mg/mL) and even though the weight gain for both sexes was lower than the vehicle
19 control group, it was not statistically significant but these data were not shown. The authors
20 report that gross pathological examinations performed on mice killed at 4 and 6 months were
21 unremarkable and that a number of mice from all the dosing regimens had liver abnormalities,
22 such as pale, spotty, or granular livers. They report that 2 of 58 males at 4 months, and 11 of
23 59 mice at 6 months had granular livers and obvious fatty infiltration, and that mice of both sexes
24 were affected. Animals in the naive and vehicle groups were reported to infrequently have pale
25 or spotty livers, but exhibit no other observable abnormalities. No quantitation or more detailed
26 descriptions of the incidence of or severity of effects were given in this report.
27 The average body weight of male mice receiving the highest dose of TCE was reported to
28 be 10% lower at 4 months and 11% lower at 6 months with body weights of female mice at the
29 highest dose also significantly lower. Enlarged livers (as percentage of body weight) were
30 observed after both durations of exposure in males at the three highest doses, and in females at
31 the highest dose. In the 4-month study, brain weights of treated females were significantly
32 increased when compared to vehicle control. However, the authors state
33
34 this increase is apparently because the values for the vehicle group were low,
35 because the naive group was also significantly increased when compared to
36 vehicle control. A significant increase in kidney weight occurred at the highest
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-18 DRAFT—DO NOT CITE OR QUOTE
-------
1 dose in males at 6 months and in females, after both 4 and 6 months of TCE
2 exposure. Urinalysis indicated elevated protein and ketone levels in high-dose
3 females and the two highest dose males after 6 months of exposure (data not
4 shown).
5
6 The authors describe differences in hematology to include
7
8 a decreased erythrocyte count in the high dose males at 4 and 6 months (13% and
9 16%, respectively); decreased leukocyte counts, particularly in the females at 4
10 months and altered coagulation values consisting of increased fibrinogen in males
11 at both times and shortened prothrombin time in females at 6 months (data not
12 shown). No treatment-related effects were detected on the types of white cells in
13 peripheral blood.
14
15 It must be noted that effects reported from this study may have also been related to decreased
16 water consumption, this study did included any light microscopic evaluation, and that most of the
17 results described are for data not shown. However, this study does illustrate the difficulties
18 involved in trying to conduct studies of TCE in drinking water, that the LD50s for TCE are
19 relatively high, and that liver weight increases were observed with TCE exposure as early as few
20 weeks and increased liver weight were sustained through the 6-month study period.
21
22 E.2.1.7. Goldsworthy and Popp, 1987
23 The focus of this study was peroxisomal proliferation activity after exposure to a number
24 of chlorinated solvents. In this study 1,000 mg/kg TCE (99+ % epoxide stabilizer free) was
25 administered to male F-344 rats (170-200 g or -10% difference) and B6C3F1 (20-25 g or -20%
26 difference) mice for 10 days in corn oil via gavage. The ages of the animals were not given. The
27 TCE-exposed animals were studied in two experiments (Experiments #1 and #3). In experiment
28 #2 corn oil and methyl cellulose vehicles were compared. Animals were killed 24 hours after the
29 last exposure. The authors did not show data on body weight but stated that the administration of
30 test agents (except WY-14,643 to rats which demonstrated no body weight gain) to rats and mice
31 for 10 days "had little or no effect on body weight gain." Thus, differences in initial body weight
32 between treatment and control groups, which could have affected the magnitude of TCE-induced
33 liver weight gain, were not reported. The liver/body weight ratios in corn oil gavaged rats were
34 reported to be 3.68% ± 0.06% and 4.52% ± 0.08% after TCE treatment which represented
35 1.22-fold of control (n = 5). Cyanide-(CN-)insenstive palmitoyl CoA1 oxidation (PCO) was
36 reported to be 1.8-fold increased after TCE treatment in this same group. In B6C3F1 mice the
1 CoA = coenzyme A.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-19 DRAFT—DO NOT CITE OR QUOTE
-------
1 liver/body weight ratio in corn oil gavaged mice was reported to be 4.55% ± 0.13% and
2 6.83% ± 0.13% after TCE treatment which represented 1.50-fold of control (n = 7).
3 CN-insensitive PCO activity was reported to be 6.25-fold of control after TCE treatment in this
4 same group. The authors report no effect of vehicle on PCO activity but do not show the data
5 nor discuss any effects of vehicle on liver weight gain. Similarly the results for experiment #3
6 were not shown nor liver weight discussed with the exception of PCO activity reported to be
7 2.39-fold of control in rat liver and 6.25-fold of control for mouse liver after TCE exposure. The
8 number of animals examined in Experiment #3 was not given by the authors or the variation
9 between enzyme activities. However, there appeared to be a difference in PCO activity
10 Experiments #1 and #3 in rats. There were no descriptions of liver histology given in this report
11 for TCE-exposed animals or corn-oil controls.
12
13 E.2.1.8. Elcombe et al, 1985
14 In this study, preservative free TCE was given via gavage to rats and mice for
15 10 consecutive days with a focus on changes in liver weight, structure, and hepatocellular
16 proliferation induced by TCE. Male Alderly Park rats (Wistar derived) (180-230 g), male
17 Obsborne-Mendel rats (240-280 g), and male B6C3F1 or male Alderly Park Mice (Swiss)
18 weighing 30 to 35 g were administered 99.9% pure TCE dissolved in corn oil via gavage. The
19 ages of the animals were not given by the authors. The animals were exposed to 0, 500, 1,000,
20 or 1,500 mg/kg body wt TCE for 10 consecutive days. The number of mice and rats varied
21 widely between experiments and treatment groups and between various analyses. In some
22 experiments animals were injected with tritiated thymidine approximately 24 hours following the
23 final dose of TCE and killed one hour later. The number of hepatocytes undergoing mitosis was
24 identified in 25 random high-power fields (X40) for each animal with 5,000 hepatocyte per
25 animal examined. There was no indication by the authors that zonal differences in mitotic index
26 were analyzed. Sections of the liver were examined by light and electron microscopy by
27 conventional staining techniques. Tissues selected for electron microscopy included central vein
28 and portal tract so that zonal differences could be elucidated. Morphometric analysis of
29 peroxisomes was performed "according to general principles of Weibel et al (1964) on
30 electronphotomicrographs from pericentral hepatocytes." DNA content of samples and
31 peroxisomal enzyme activities were determined in homogenized liver (catalase and PCO
32 activity).
33 The authors reported that TCE treatment had no significant effect on body-weight gain
34 either strain of rat or mouse during the 10 days exposure period. However, marked increases (up
35 to 175% of control value) in the percent liver/body weight ratio were observed in TCE-treated
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-20 DRAFT—DO NOT CITE OR QUOTE
-------
1 mice. Smaller increases (up to 130% of control) in relative liver weight were observed in
2 TCE-treated rats. No significant effects of TCE on hepatic water content were seen so that the
3 liver weight did not represent increased water retention.
4 An interesting feature of this study was that it was conducted in treatment blocks at
5 separate times with separate control groups of mice for each experimental block. Therefore,
6 there were three control groups of B6C3F1 mice (n = 10 for each control group) and three
7 control groups for Alderly Park (n = 9 to 10 for each control group) mice that were studied
8 concurrently with each TCE treatment group. However, the percent liver/body weight ratios
9 were not the same between the respective control groups. There was no indication from the
10 authors as to how controls were selected or matched with their respective experimental groups.
11 The authors did not give liver weights for the animals so the actual changes in liver weights are
12 not given. The body weights of the control and treated animals were also not given by the
13 authors. Therefore, if there were differences in body weight between the control groups or
14 treatment groups, the liver/body weight ratios could also have been affected by such differences.
15 The percentage increase over control could also have been affected by what control group each
16 treatment group was compared to. There was a difference in the mean percent liver/body weight
17 ratio in the control groups, which ranged from 4.32 to 4.59% in the B6C3F1 mice (-6%
18 difference) and from 5.12 to 5.44% in the Alderly Park mice (-6% difference). The difference in
19 average percent liver/body weight ratio for untreated mice between the two strains was -16%.
20 Because the ages of the mice were not given, the apparent differences between strains may have
21 been due to both age or to strain. After TCE exposure, the mean percent liver/body weight ratios
22 were reported to be 5.53% for 500 mg/kg, 6.50% for 1,000 mg/kg, and 6.74% for 1,500 mg/kg
23 TCE-exposed B6C3F1 mice. This resulted in 1.20-, 1.50-, and 1.47-fold values of control in
24 percent liver weight/body weight for B6C3F1 mice. For Alderly Park mice, the percent
25 liver/body weight ratios were reported to be 7.31, 8.50, and 9.54% for 500, 1,000, and
26 1,500 mg/kg TCE treatment, respectively. This resulted in 1.43-, 1.56-, and 1.75-fold of control
27 values. Thus, there appeared to be more of a consistent dose-related increase in liver/body
28 weight ratios in the Alderly Park mice than the B6C3F1 mice after TCE treatment. However, the
29 variability in control values may have distorted the dose-response relationship in the B6C3F1
30 mice. The Standard deviations for liver/body weight ratio were as much as 0.52% for the treated
31 B6C3F1 mice and 0.91% for the Alderly Park treated mice. In regard to the correspondence of
32 the magnitude of the TCE-induced increases in percent liver/body weight with the magnitude of
33 difference in TCE exposure concentrations, in the B6C3F1 mice the increases were similar
34 (~2-fold) between the 500 mg/kg and 1,000 mg/k TCE exposure groups. For the Alderly Park
35 mice, the increases in TCE exposure concentrations were slightly less than the magnitude of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-21 DRAFT—DO NOT CITE OR QUOTE
-------
1 increases in percent liver/body ratios between all of the concentrations (i.e., ~1.3-fold of control
2 vs. 2-fold for 500 and 1,000 mg/kg TCE dose and 1.3-fold of control vs. 1.5-fold for the 1,000
3 and 1,500 mg/kg TCE dose).
4 The DNA content of the liver varied greatly between control animal groups. For B6C3F1
5 mice it ranged from 2.71 to 2.91 mg/g liver. For Alderly Park mice it ranged from 1.57 to
6 2.76 mg/g liver. The authors do not discuss this large variability in baseline levels of DNA
7 content. The DNA content in B6C3F1 mice was mildly depressed by TCE treatment in a
8 nondose dependent manner. DNA concentration decrease from control ranged from 20-25%
9 between all three TCE exposure levels in B6C3F1 mice. For Alderly Park mice there was also
10 nondose related decrease in DNA content from controls that ranged from 18% to 34%. Thus, the
11 extent of decrease in DNA content of the liver from TCE treatment in B6C3F1 mice was similar
12 to the variability between control groups. The lack of dose-response in apparent treatment
13 related effect in B6C3F1 mice and especially in the Alderly Park mice was confounded by the
14 large variability in the control animals. The changes in liver weight after TCE exposure for the
15 AP mice did not correlate with changes in DNA content further, raising doubt about the validity
16 of the DNA content measures. However, a small difference in DNA content due to TCE
17 treatment in all groups was reported for both strains and this is consistent with hepatocellular
18 hypertrophy.
19 The reported results for incorporation of tritiated thymidine in liver DNA showed large
20 variation in control groups and standard deviations that were especially evident in the Alderly
21 Park mice. For B6C3F1 mice, mean control levels were reported to range from 5,559 to
22 7,767 dpm/mg DNA with standard deviations ranging from 1,268 to 1,645 dpm/mg DNA. In
23 Alderly Park mice mean control levels were reported to range from 6,680 to 10,460 dpm/mg
24 DNA with standard deviations ranging from 308 to 5,235 dpm/mg DNA. For B6C3F1 mice,
25 TCE treatment was reported to induce an increase in tritiated thymidine incorporation with a
26 very large standard deviation, indicating large variation between animals. For 500 mg/kg TCE
27 treatment group the values were reported as 12,334 ± 4,038, for 1,000 mg/kg TCE treatment
28 group 21,909 ± 13,386, and for 1,500 mg/kg treatment TCE group 26,583 ± 10,797 dpm/mg
29 DNA. In Alderly Park mice TCE treatment was reported to give an increase in tritiated
30 thymidine incorporation also with a very large standard deviation. For 500 mg/kg TCE, the
31 values were reported as 19,315 ± 12,280, for 1,000 mg/kg TCE 21,197 ± 8,126 and for
32 1,500 mg/kg TCE 38,370 ± 13,961. As a percentage of concurrent control, the increase in
33 tritiated thymidine was reported to be 2.11-, 2.82-, and 4.78-fold of control in B6C3F1 mice, and
34 2.09-, 2.03-, and 5.74-fold of control in Alderly Park mice. Accordingly, the change in tritiated
35 thymidine incorporation did show a treatment related increase but not a dose-response. Similar
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-22 DRAFT—DO NOT CITE OR QUOTE
-------
1 to the DNA content of the liver, the large variability in measurements between control groups
2 and variability between animals limit quantitative interpretation of these data. The increase in
3 tritiated thymidine, seen most consistently only at the highest exposure level in both strains of
4 mice, could have resulted from either a change in ploidy of the hepatocytes or cell number.
5 However, the large change in volume in the liver (75%) in the Alderly Park mice, could not have
6 resulted from only a 4-fold of control in cell proliferation even if all tritiated thymidine
7 incorporation had resulted from changes in hepatocellular proliferation. As mentioned in Section
8 E. 1.1 above, the baseline level of hepatocellular proliferation in mature control mice is very low
9 and represents a very small percentage of hepatocytes.
10 In the experiments with male rats, the same issues discussed above, associated with the
11 experimental design, applied to the rat experiments with the additional concern that the numbers
12 of animals examined varied greatly (i.e., 6 to 10) between the treatment groups. In Obsborne-
13 Mendel rats, the control liver/body weight ratio was reported to vary from 4.26 to 4.36% with the
14 standard deviations varying between 0.22 to 0.27%. For the Alderly Park rats, the liver/body
15 weight ratios were reported to vary between 4.76 and 4.96% (in control groups) with standard
16 deviations varying between 0.24 to 0.47%. TCE treatment was reported to induce a dose-related
17 increase in liver/body weight ratio in Obsborne-Mendel rats with mean values of 5.16, 5.35, and
18 5.53% in 500, 1,000, and 1,500 mg/kg TCE treated groups, respectively. This resulted in 1.18-,
19 1.26-, and 1.30-fold values of control. In Alderly Park rats, TCE treatment was reported to result
20 in increased liver weights of 5.45, 5.83, and 5.65% for 500, 1,000, and 1,500 mg/kg TCE
21 respectively. This resulted in 1.14-, 1.17-, and 1.17-fold values of control. Again, the variability
22 in control values may have distorted the nature of the dose-response relationships in Alderly Park
23 rats. TCE treatment was reported to result in standard deviations that ranged from 0.31 to 0.48%
24 for OM rats and 0.24 to 0.38% for Alderly Park rats. What is clear from these experiments is
25 that TCE exposure was associated with increased liver/body weight in rats.
26 The reported mean hepatic DNA concentrations and standard deviations varied greatly in
27 control rat liver as it did in mice. The variation in DNA concentration in the liver varied more
28 between control groups than the changes induced by TCE treatment. For Obsborne-Mendel rats,
29 the mean control levels of mg DNA/g liver were reported to range from 1.99 to 2.63 mg
30 DNA/liver with standard deviations varying from 0.17 to 0.33 mg DNA/g. For Alderly Park
31 rats, the mean control levels of mg DNA/g liver were reported to be 2.12 to 3.16 mg DNA/g with
32 standard deviation ranging from 0.06 to 1.04 mg DNA/g. TCE treatment decreased the liver
33 DNA concentration in all treatment groups. For Obsborne-Mendel rats, the decrease ranged
34 from 8 to 13% from concurrent control values and for Alderly Park rats the decrease ranged from
35 8 to 17%. There was no apparent dose response in the decreases in DNA content with all TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-23 DRAFT—DO NOT CITE OR QUOTE
-------
1 treatment levels giving a similar decrease from controls and the same limitations discussed above
2 for the mouse data apply here. The magnitude of increases in liver/body ratios shown by TCE
3 treatment were not correlated with the changes in DNA content. However, as with the mouse
4 data, the small differences in DNA content due to TCE treatment in all groups and in both strains
5 was consistent with hepatocellular hypertrophy.
6 Incorporation of tritiated thymidine was reported to be even more variable between
7 control groups of rats than it was for mice and was reported to be especially variable between
8 control groups (i.e., 2.7-fold difference between control groups within strain) and differed
9 between the strains (average of 2.5-fold between strains). For Obsborne-Mendel rats the mean
10 control levels were reported to range from 13,315 to 33,125 dpm/mg DNA, while for Alderly
11 Park rats tritiated thymidine incorporation ranged from 26,613 to 69,331 dpm/mg DNA for
12 controls. The standard deviations were also very large (i.e., for control groups of Obsborne-
13 Mendel rats they were reported to range from 8,159 to 13,581 dpm/mg DNA, while for Alderly
14 Park rats they ranged from 9,992 to 45,789 dpm/mg DNA). TCE treatment was reported to
15 induce increases over controls of 110, 118, and 106% for 500, 1,000, and 1,500 mg/kg TCE-
16 exposed groups, respectively, in Obsborne-Mendel rats with large standard deviations for these
17 treatment groups as well. In Alderly Park rats, the increases over controls were reported to be
18 206, 140, and 105% for 500, 1,000, and 1,500 mg/kg TCE, respectively. In general, these data
19 do indicate that TCE treatment appeared to give a mild increase in tritiated thymidine
20 incorporation but the lack of dose-response can be attributable to the highly variable
21 measurements of tritiated thymidine incorporation in control animal groups. The variation in the
22 number of animals examined between groups and small numbers of animals examined
23 additionally decrease the likelihood of being able to discern the magnitude of difference between
24 species- or strain-related effects for this parameter. Again, given the very low level of
25 hepatocyte turnover in control rats, this does not represent a large population of cells in the liver
26 that may be undergoing proliferation and cannot be separated from changes in ploidy.
27 The authors report that the reversibility of these phenomena was examined after the
28 administration of TCE to Alderly Park mice for 10 consecutive days. Effects upon liver weight,
29 DNA concentration, and tritiated thymidine incorporation 24 and 48 hours after the last dose of
30 TCE were reported to be still apparent. However, 6 days following the last dose of TCE, all of
31 these parameters were reported to return to control values with the authors not showing the data
32 to support this assertion. Thus, cessation of TCE exposure would have resulted in a 75%
33 reduction in liver weight by one week in mice exposed to the highest TCE concentration.
34 Analyses of hepatic peroxisomal enzyme activities were reported for catalase and
35 p-oxidation (PCO activity) following administration of TCE to B6C3F1 mice and Alderly Park
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-24 DRAFT—DO NOT CITE OR QUOTE
-------
1 rats exposed to 1,000 mg/kg TCE for 10 days. The authors only used 5 control and 5 exposed
2 animals for these tests. An 8-fold of control value for PCO activity and a 1.5-fold of control
3 value for catalase activity were reported for B6C3F1 mice exposed to 1,000 mg/kg TCE. In the
4 Alderly Park rats no significant changed occurred. It is unclear which mice or rats were selected
5 from the previous experiments for these analyses and what role selection bias may have played
6 in these results. The reduced number of animals chosen for this analysis also reduces the power
7 of the analysis to detect a change. In rats, there was a reported 13% increase in PCO; however,
8 the variation between the TCE treated rats was more than double that of the control animals in
9 this group and the other limitations described above limit the ability to detect a response. There
10 was no discussion given by the authors as to why only one dose was tested in half of the animals
11 exposed to TCE or why the strain with the lowest liver weight change due to TCE exposure was
12 chosen as the strain to test for peroxisomal proliferative activity.
13 The authors provided a description of the histopathology at the light microscropy level in
14 B6C3F1 mice, Alderly Park mice, Osborne-Mendel rats, and Alderly Park rats, but did not
15 provide a quantitative analysis or specific information regarding the variability of response
16 between animals within groups. There appeared to be 20 animals examined in the 1,000 mg/kg
17 TCE exposed group of B6C3F1 mice but no explanation as to why there were only 10 animals
18 examined in analyses for liver weight changes, DNA concentration, and tritiated thymidine
19 incorporation. There was no indication by the authors regarding how many rats were examined
20 by light microscopy.
21 Apart from a few inflammatory foci in occasional animals, hematoxylin and eoxin (H&E)
22 section from B6C3F1 control mice were reported to show no abnormalities. The authors suggest
23 that this is a normal finding in the livers of mice kept under "non-SPF conditions." A stain for
24 neutral lipid was reported to not be included routinely in these studies, but subsequent electron
25 microscopic examination of lipid to show increases in the livers of corn-oil treated control
26 animals. The individual fat droplets were described as "generally extremely fine and are not
27 therefore detectable in conventionally process H&E stained sections, since both glycogen and
28 lipid are removed during this procedure." Thus, this study documents effects of using corn oil
29 gavage in background levels of lipid accumulation in the liver.
30 The finding of little evidence of gross hepatotoxicity in TCE-treated mice was reported,
31 even at a dose of 1,500 mg/kg. Specifically,
32
33 Of 19 animals examined receiving 1500 mg/kg body weight TCE, only 6 showed
34 any evidence of hepatocyte necrosis, and this pathology was restricted to single
35 small foci or isolated single cells, frequently occurring in a subcapsular location.
36 Examination of 20 animals receiving 1000 mg/kg body wt TCE demonstrated no
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-25 DRAFT—DO NOT CITE OR QUOTE
-------
1 hepatocyte necrosis. Of 20 animals examined receiving 500 mg/kg body wt TCE,
2 1 showed necrosis of single isolated hepatocytes; however, this change was not a
3 treatment-related finding.
4
5 TCE-treated mice were reported to show
6
7 a change in staining characteristic of the hepatocytes immediately adjacent to the
8 central vein of the hepatocyte lobules, giving rise to a marked 'patchiness' of the
9 liver sections. Often this change consisted of increased eosinophilia of the central
10 cells. There was some evidence of cell hypertrophy in the centrilobular regions.
11 These changes were evident in most of the TCE treated animals, but there was a
12 dose-related trend, relatively few of the 500 mg/kg animals being affected, while
13 the majority of the 1,500 mg/kg animals showed central change. No other
14 significant abnormalities were seen in the liver of TCE treated mice compared to
15 controls apart from occasional mitotic figures and the appearance of isolated
16 nuclei with an unusual chromatin pattern. This pattern generally consisted of a
17 course granular appearance with a prominent rim of chromatin around the
18 periphery of the nucleus. These nuclei may have been in the very early stages of
19 mitosis. Similar changes were not seen in control mice.
20
21 The authors briefly commented on the findings in the Alderly Park mice stating that
22
23 H& E sections from Alderly Park mice gave similar results as for B6C3F1 mice.
24 No evidence of hepatotoxicity was seen at a dose of 500 mg/kg body wt TCE.
25 However, a few animals at the higher doses showed some necrosis and other
26 degenerative changes. This change was very mild in nature, being restricted to
27 isolated necrotic cells or small foci, frequently in subcapsular position.
28 Hypertrophy and increased eosinophilia were also noticed in the centrilobular
29 regions at higher doses.
30
31 Thus, from the brief description given by the authors, the centrilobular region is identified as the
32 location of hepatocellular hypertrophy due to TCE exposure in mice, and for it to be dose-related
33 with little evidence of accompanying hepatotoxicity.
34 The description of histopathology for rats was even more abbreviated than for the mouse.
35 H& E sections from Osborne-Mendel rats showed that
36
37 livers from control rats contained large quantities of glycogen and isolated
38 inflammatory foci, but were otherwise normal. The majority of rats receiving
39 1,500 mg/kg body weight TCE showed slight changes in centrilobular
40 hepatocytes. The hepatocytes were more eosinophilic and contained little
41 glycogen. At lower doses, these effects were less marked and were restricted to
42 fewer animals. No evidence of treatment-related hepatotoxicity (as exemplified
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-26 DRAFT—DO NOT CITE OR QUOTE
-------
1 by single cell or focal necrosis) was seen in any rat receiving TCE. H& E
2 sections from Alderly Park Rats showed no signs of treatment-related
3 hepatotoxicity after administration of TCE. However, some signs of dose-related
4 increase in centrilobular eosinophilia were noted.
5
6 Thus, both mice and rats exhibited pericentral hypertrophy and eosinophilia as noted from the
7 histopathological examination.
8 The study did report a quantitative analysis of the effects of TCE on the number of
9 mitotic figures in livers of mice. Few if any control mice exhibited mitotic figures. But, the
10 authors report
11
12 a considerable increase in both the numbers of figures per section was noted after
13 administration of TCE." The numbers of animals examined for mitotic figures
14 ranged from 75 (all control groups were pooled for mice) to 9 in mice, and ranged
15 from 15 animals in control rat groups to as low as 5 animals in the TCE treatment
16 groups. The range of mitotic figures found in 25 high-power fields was reported
17 and is equivalent to the number of mitotic figures per 5,000 hepatocytes examined
18 in random fields.
19
20 Thus, the predominance of mitotic figures in any zone of the liver cannot be ascertained.
21 For B6C3F1 mice the number of animals with mitotic figures was reported to be 0/75,
22 3/20, 7/20, and 5/20 for control, 500, 1,000, and 1,500 mg/kg TCE exposed mice, respectively.
23 The range of the number of mitotic figures seen in 5,000 hepatocytes was reported to be 0, 0-1,
24 0-5, 0-5 for those same groups with group means of 0, 0.15 ± 0.36, 0.6 ± 1.1, and 0.5 ± 1.2.
25 These results demonstrate a very small and highly variable response due to TCE treatment in
26 B6C3F1 mice in regard to mitosis. Thus, the highest percentage of cells undergoing mitosis
27 within the window of observation would be on average 0.012% with a standard deviation twice
28 that value. The data presented for mitotic figures also indicated no differences in results between
29 1,000 and 1,500 mg/kg treated B6C3F1 mice in regard to mitotic figure detection. However, the
30 tritiated thymidine incorporation data indicated that thymidine incorporation was ~2-fold greater
31 at 1,500 than 1,000 mg/kg TCE in B6C3F1 mice. For Alderly Park mice, the number of animals
32 with mitotic figures was reported to be 1/15, 0/9, 4/9, and 2/9 for control, 500, 1,000, and
33 1,500 mg/kg TCE exposed mice. The range of the number of mitotic figures seen in 5,000
34 hepatocytes was 0-1, 0, 0-2, 0-1 for those same groups with group means of 0.06 ± 0.25,
35 0.7 ± 0.9, and 0.2 ± 0.4. These results reveal the detection of at the most 2 mitotic figure in
36 5,000 hepatocytes for any mouse an any treatment group and no dose-related increased after
37 TCE treatment in Alderly Park mice. Thus, the highest percentage of cells with a mitotic figure
38 would be on average 0.014% with a standard deviation twice that value. The small number of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 animals examined reduces the power of the experiment to draw any conclusions as to a dose-
2 response. Similar to the B6C3F1 mice, there did not appear to be concordance between mitotic
3 figure detection and thymidine incorporation for Alderly park mice. Thymidine incorporation
4 showed a 2-fold increase over control for 500 and 1,000 mg/kg TCE and a 5.7-fold increase for
5 1,500 mg/kg TCE treated animals. However, in regard to mitotic figure detection, there were
6 fewer mitotic figures in 500 mg/kg TCE treated mice than controls, and fewer animals with
7 mitotic figures and fewer numbers of figures in the 1,500 mg/kg dose than the 1,000 mg/kg
8 exposed group. The inconsistencies between mitotic index data and thymidine incorporation
9 data in both strains of mice suggests that either thymidine incorporation is representative of only
10 DNA synthesis and not mitosis, an indication of changes in ploidy rather than proliferation, or
11 that this experimental design is incapable of discerning the magnitude of these changes
12 accurately. Data from both mouse strains show very little if any hepatocyte proliferation due to
13 TCE exposure with the mitotic figure index data having that advantage of being specific for
14 hepatocytes and to not to also include nonparenchymal cells or inflammatory cells in the liver.
15 The results for rats were similar to those for mice and even more limited by the varying
16 and low number of animals examined. For Osborne-Mendal rats the number of animals with
17 mitotic figures were reported to be 8/15, 2/9, 0/7, and 0/6 for control, 500, 1,000, and 1,500
18 nig/kg TCE exposed rats groups, respectively, with the range of the number of mitotic figures
19 seen in 5,000 hepatocytes to be 0-8, 0-3, 0, and 0. The group mean was 1.5 ± 2.0, 0.4 ± 1.0, 0,
20 and 0 for these groups. It would appear from these results that there are fewer mitotic figures
21 after TCE treatment with the highest percentage of cells undergoing mitosis to be on average
22 0.03% in control rats. However, thymidine incorporation studies show a modest increase at all
23 treatment levels over controls in Osborne Mendel rats rather than a decrease from controls. For
24 Alderly Park rats the number of animals with mitotic figures was reported to be 13/15, 5/9, 9/9,
25 and 4/9 for control, 500, 1,000, and 1,500 mg/kg TCE exposed rat groups with the range of the
26 number of mitotic figures seen in 5,000 hepatocytes to be 0-26, 0-5, 1-7, and 0-9. The group
27 mean was 7.2 ± 4.7, 1.6 ± 4.3, 3.8 ± 3.4, and 1.8 ± 2.9 for these groups. It would appear that
28 there are fewer mitotic figures after TCE treatment with the highest percentage of cells to an
29 average of 0.14% in control rats. However, thymidine incorporation studies show 2-fold greater
30 level at 500 mg/kg TCE than for control animals and a 40 and 5% increase at 1,000 mg/kg and
31 1,500 mg/kg TCE exposure groups, respectively. Similar to the results reported in mice, results
32 in both rat strains show an inconsistency in mitotic index and thymidine incorporation. The
33 control rats appear to have a much greater mitotic index than any of the mouse groups (treated or
34 untreated) or the TCE-treatment groups. However, it is the mice that were exhibiting the largest
35 increased in liver weight after TCE exposure. By either thymidine incorporation or mitosis,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 these data do provide a consistent result that at 10 days of exposure very little sustained
2 hepatocellular proliferation is occurring in either mouse or rat and neither is correlated well with
3 the concurrent changes in liver weight observed from TCE exposure.
4 This study provided a qualitative discussion and quantitative analysis of structural
5 changes using electron microscopy. The qualitative discussion was limited and included
6 statements about increased observances without quantitative data shown other than the
7 morphometric analysis. The authors reported that
8
9 the ultrastructure of control mouse liver was essentially normal, although mild
10 dilatation of RER and SER was a frequent finding. Lipid droplets were also
11 usually present in the cell cytoplasm. The ultrastructural changes seen in mouse
12 liver following administration of up to 1,500 mg/kg body wt TCE for 10 days
13 were essentially similar in the B6C3F1 mouse and the Alderly Park mouse. The
14 most notable change in both strains of mouse was a dramatic increase in the
15 number of peroxisomes. This change was only apparent in the cells immediately
16 surrounding the central veins. Peroxisome proliferation was not noticeable in
17 periportal cells. The induced peroxisomes were generally small and very electron
18 dense and frequently lacked the characteristic nucleoid core found in peroxisomes
19 of control livers.
20
21 The authors conclude that
22
23 morphometric analysis showed evidence of a dose-related response, peroxisomal
24 induction appearing to reach a maximum at 1,000 mg/kg in B6C3F1 mice.. .Lipid
25 was increased in the livers of treated mice at all doses and was present both as
26 free droplets in the cytoplasm and as liposomes (small lipid droplets in ER
27 cisternae). The centrilobular cell, which showed the greatest increase in numbers
28 of peroxisomes, showed no evidence of this lipid accumulation: fatty change was
29 more prominent in those cells away from the central vein (i.e., zone 2 of the liver
30 acinus). Accumulation of lipid, particularly in liposomes, was less marked in
31 Alderly Park mouse than in B6C3F1 mouse. Mild proliferation of smooth
32 endoplasmic reticulum was seen in both strains and both rough and smooth
33 endoplasmic reticulum was generally more dilated than in control mice.
34
35 Electron microscopic results for rat liver were reported
36
37 to show similar changes in Osborne-Mendel and Alderly Park rat treated with
38 TCE.. .Rats receiving either 1,000 or 1,500 mg/kg TCE for 10 days generally
39 showed mild proliferation of SER in centrilobular hepatocytes. The cisternae of
40 RER were frequently dilated, giving rise to a rather disorganized appearance in
41 contrast to the parallel stacks seen in control livers, although no detachment of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-29 DRAFT—DO NOT CITE OR QUOTE
-------
1 ribosomes was evident. The SER was also dilated. In contrast to mice,
2 peroxisomes were only very slightly and not significantly, increased in the liver of
3 TCE -treated rats. Morphometric analysis confirmed this observation, with the
4 volume density of peroxisomes in the cytoplasm of centrilobular hepatocytes
5 being only slightly increased in rats of both strains receiving 1,000 or 1,500
6 rng/kg body wt TCE.. .Lipid droplets were occasionally increased in some livers
7 obtained from rats receiving TCE, but the degree of fatty change generally
8 appeared similar to that found in control rats receiving corn oil. There were no
9 changes in membrane -bound liposomes, other organelles, or Golgi condensing
10 vesicles. Centrilobular glycogen was somewhat depleted in male rats receiving
11 1,500 mg/kg TCE. Periportal cells were ultrastructually normal in all rats.
12
13 For the morphometric analysis, the number of mice examined ranged from 7 in the
14 control group to 8 in the 1,500 mg/kg TCE exposed group. The authors did not indicate which
15 control animals were used for the morphometric analysis from the 75 animals examined for
16 mitotic index, the 20 examined by light microscopy, or the 30 mice used as concurrent controls
17 in the liver weight, DNA concentration, and tritiated thymidine incorporation studies. The
18 authors stated that morphometry was performed on three randomly selected photomicrographs
19 from each of three randomly selected pericentral hepatocytes for each animal (i.e., nine
20 photomicrographs per animal). A mean value representing the exposure group was reported with
21 the variability between photomitographs per animal or the variation between animals unclear.
22 The morphometric analysis did not examine all treatment groups (e.g., only the control and
23 500 mg/kg TCE group were examined in Alderly Park mice). The percent cytoplasmic volume
24 of the peroxisomal compartment (mean ± standard deviation [SD]) was reported to be
25 0.6% ± 0.6% for controls, 4.8% ± 3.3% for 500 mg/kg TCE, 6.7% ± 1.9% for 1,000 mg/kg TCE,
26 and 6.4% ± 2.5% for 1,500 mg/kg TCE in B6C3F1 mice. In Alderly Park mice, only 12 control
27 and 12 500 mg/kg TCE exposed mice were examined and, similarly, their selection criteria was
28 not given. The percent cytoplasmic volume of the peroxisomal compartment was 1.2% ± 0.4%
29 for control and 4.7 ± 2.8% for 500 mg/kg TCE exposed mice. For Osborne-Mendel rats control
30 rats were reported to have a percent cytoplasmic volume of the peroxisomal compartment for
31 control rats (n = 9) of 1.8% ± 0.4%, 1,000 mg/kg TCE (n = 5) 2.3% ± 1.6%, and for 1,500 mg/kg
32 exposed rats (n = 7) 2.3% ± 2.0%. For Alderly Park rats only two groups were examined
33 (control and 1,000 mg/kg TCE exposure). The percent cytoplasmic volume of the peroxisomal
34 compartment for control rats (n = 15) was reported to be 1.8% ± 0.8% and for 1,000 mg/kg TCE
35 (n = 16) to be 2.4% ± 1.2%. The varying numbers of animals examined, the varying and
36 inconsistent number of treatment groups examined, the limited number of photomitographs per
37 animal, and the potential selection bias for animals examined make quantitative conclusions
38 regarding this analysis difficult. Although control levels differed by a factor of 2 between the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-30 DRAFT—DO NOT CITE OR QUOTE
-------
1 two strains of mice examined, as well as the number of control animals examined (7 vs. 12), it
2 appears that the 500-mg/kg TCE-exposed B6C3F1 and Alderly Park mice had similar
3 percentages of peroxisomal compartment in the pericentral cells examined (-4.8%). There also
4 appeared to be little difference between 1,000 mg/kg TCE treated Osborne-Mendel and Alderly
5 Park rats for this parameter (-2.4%). Although few animals were examined, there was little
6 difference reported between 500, 1,000, and 1,500 mg/kg TCE exposure groups in regard to
7 percentages of peroxisomal compartment in B6C3F1 mice (4.8-6.7%). For the few rats of the
8 Osborne-Mendel strain examined, there also did not appear to be a difference between 1,000 and
9 1,500 mg/kg TCE exposure for thi s parameter (2.3 %).
10 Based on peroxisome compartment volume data, one would expect there to be little
11 difference between TCE exposure groups in mice or rats in regard to enzyme activity or other
12 "associated events." However, such comparisons are difficult due to limited power to detect
13 differences and the possibility of bias in selection of animals in differing assays. For the
14 B6C3F1 mice, only 5 animals per group were examined for enzyme analysis, 7 to 8 for
15 morphometric analysis, 75 animals in control, and 20 animals in 1,000 mg/kg TCE-exposed
16 groups for mitotic figure identification, and 10 animals per group for thymidine incorporation.
17 Since only a few animals were tested for enzyme activity the comparison between peroxisomal
18 compartment volume and that parameter is very limited. There was a reported 47% increase in
19 catalase activity between control (n = 5) and 1,000 mg/kg TCE exposed B6C3F1 mice (n = 5)
20 and 7.8-fold increase in PCO activity. The percent peroxisome compartment was reported to be
21 10.6-fold greater (0.6 vs. 6.4%). However, the B6C3F1 control percent volume of peroxisomal
22 compartment was reported to be half that of the AP mouse control. An accurate determination of
23 the quantitative differences in peroxisomal proliferation would be dependent on an accurate and
24 stable control value. For Alderly Park rats there was an 8% decrease in catalase activity between
25 control (n = 5) and 1,000 mg/kg TCE exposed rats (n = 5), and a 13% increase in PCO activity.
26 The percent peroxisome compartment was reported to be 33% greater in the TCE-exposed than
27 control group. Thus, for the very limited data that was available to compare peroxisomal
28 compartment volume with enzyme activity, there was consistency in result.
29 However, were such increases in peroxisomes associated with other events reported in
30 this study? Mouse peroxisome proliferation associated enzyme activities in B6C3F1 mice at
31 1,000 mg/kg TCE were reported to be 8-fold over control values in mice after 10 days of
32 treatment. However, this increase in activity was not accompanied by a similar increase in
33 thymidine incorporation (2.8-fold of control) or concordant with increases in mitotic figures
34 (7/20 mice having any mitotic figures at all with a range of 0-5 and a mean of 0.014% of cells
35 undergoing mitosis for 1,000 mg/kg TCE vs. 0 for control). Although results reported in the rat
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 showed discordance between thymidine incorporation and detection of mitotic figures, there was
2 also discordance with these indices and those for peroxisomal proliferation. In comparison to
3 controls, there was a reported 13% increase in PCO activity in Alderly park rats exposed to
4 1,000 mg/kg TCE, a group mean of mitotic figures half that in the TCE treated animals versus
5 controls, and increase in thymidine incorporation of 40%. Thus, these results are not consistent
6 with TCE induction of peroxisome enzyme activity to be correlated with hepatocellular
7 proliferation by either mitotic index or thymidine incorporation. Thymidine incorporation in
8 liver DNA seen with TCE exposure also did not correlate with mitotic index activity in
9 hepatocytes and suggests that this parameter may be a reflection of polyploidization rather than
10 hepatocyte proliferation. More importantly, these data show that hepatocyte proliferation,
11 indicated by either measure, is confined to a very small population of cells in the liver after
12 10 days of TCE exposure. Hepatocellular hypertrophy in the centrilobular region appears to be
13 responsible for the liver weight gains seen in both rats and mice rather than increases in cell
14 number. These results at 10 days do not preclude the possibility that a greater level of
15 hepatocyte proliferation did not occur earlier and then had subsided by 10 days, as is
16 characteristic of many mitogens. Thymidine incorporation represents the status of the liver at
17 one time point rather than over a period of whole week and thus, would not capture the earlier
18 bouts of proliferation. However, there is no evidence of a sustained proliferative response, as
19 measured at the 10-day time period, in hepatocytes in response to TCE indicated from these data.
20 In regards to weight gain, although the volume of the peroxisomal compartment was
21 reported to be similar at 500 mg/kg TCE in B6C3F1 and Alderly Park mice (4.3%), the liver
22 weight./body weight gain in comparison to control was 20% higher in B6C3F1 mice versus 43%
23 higher in Alderly Park mice after 10 days of exposure. The liver/body weight ratio was 5.53% in
24 the B6C3F1 mice and 7.31% in the Alderly Park mice at 500 mg/kg TCE for 10 days. Similarly,
25 although the peroxisomal compartment was similar at 1,000 mg/kg TCE in Osborne-Mendel
26 (2.3%) and Alderly Park rats (2.4%), the liver weight/body weight gain was 26% in Osborne-
27 Mendel rats but 17% in Alderly Park rats at this level of TCE exposure. The liver/body weight
28 ratio was 5.35% in the Osborne-Mendel rats and 5.83% in the Alderly Park mice at 1,000 mg/kg
29 TCE for 10 days. Although there are several limitations regarding the quantitative interpretation
30 of the data, as discussed above, the data suggest that liver weight and weight gain after TCE
31 treatment was not just a function of peroxisome proliferation. This study does clearly
32 demonstrate TCE-induced changes at the lowest level tested in several parameters without
33 toxicity and without evidence of regenerative hyperplasia or sustained hepatocellular
34 proliferation. In regards to susceptibility to liver cancer induction in more susceptible (B6C3F1)
35 versus less susceptible (Alderly Park/Swiss) strains of mice (Maltoni et al., 1988), there was a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-32 DRAFT—DO NOT CITE OR QUOTE
-------
1 greater baseline level of liver weight/body weight ratio change, a greater baseline level of
2 thymidine incorporation as well as greater responses for those endpoints due to TCE exposure in
3 the "less susceptible" strain. However, both strains showed a hepatocarcinogenic response to
4 TCE induction and the limitations of being able to make quantitative conclusions regarding
5 species and strain susceptibility TCE toxicity from this study have been described in detail
6 above.
7
8 E.2.1.9. Dees and Travis, 1993
9 The focus of this study was to evaluate the nature of DNA synthesis induced by TCE
10 exposure in mice. The mitotic rate of liver cells was extrapolated using tritiated thymidine
11 uptake into DNA of male and female mice treated with HPLC grade (99 + pure) TCE. Male and
12 female hybrid B6C3F1 mice 8 weeks of age (male mice weighed 24-27 g (-12% difference) and
13 females weighing 18-21 g (-4% difference) were dosed orally by gavage for 10 days with 100,
14 250, 500, and 1,000 mg/kg body weight TCE in corn oil (n = 4 per treatment group). 16 hours
15 after the last daily dose of TCE, mice received tritiated thymidine and were sacrificed 6 hours
16 later. Hepatic DNA was extracted form whole liver and standard histopathology was also
17 performed. Hepatic DNA content and cellular distributions were also determined for thymidine
18 uptake using autoradiography of tissue sections. Tritiated thymidine incorporation into DNA
19 was determined by microscopic observations of autoradiography slides and reported as positive
20 cells per 100 (200x power) fields.
21 Changes in the treatment groups were reported to
22
23 include an increase in eosinophilic cytoplasmic staining of hepatocytes located
24 near central veins, accompanied by loss of cytoplasmic vacuolization.
25 Intermediate zones appeared normal and no changes were noted in portal triad
26 areas. Male and female mice given 1,000 mg/kg body weight TCE exhibited
27 apoptosis located near central veins. No evidence of cellular proliferation was
28 seen in the portal areas. No evidence of increased lipofuscin was seen in liver
29 sections from male and female mice treated with TCE. Evaluation of cell death in
30 male and female mice receiving TCE was performed by enumerating apoptoses.
31
32 The apoptoses "did not appear to be in proportion to the applied TCE dose given to male or
33 female mice." The mean number of apopotosis per 100 (400x) fields in each group of 4 animals
34 (male mice) was 0, 0, 0, 1, and 8 for control, 100, 250, 500, and 1,000 mg/kg TCE treated
35 groups, respectively. Variations in number of apoptoses between mice were not given by the
36 authors. Feulgen stain was <1 for all doses except for 9 at 1,000 mg/kg.
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 Mitotic figure were reported to be
2
3 frequently seen in liver sections from both male and female mice treated with
4 TCE. Dividing cells were most often found in the intermediate zone and
5 resembled mature hepatocytes. Incorporation of the radiolabel into cells located
6 near the portal triad areas was rare. In general, mitotic figures were very rare, but
7 when found they were usually located in the intermediate zone. Little or no
8 incorporation of label was seen in areas near the bile duct epithelia or in areas
9 close to the portal triad.
10
11 No quantitative description of mitotic index was reported by the authors but this description is
12 consistent with there being replication of mature hepatocytes induced by TCE.
13 The distribution of tritiated thymidine was given for specific cell types in the livers of
14 5 animals per treatment group and radiolabel was reported to be predominantly associated with
15 perisinusoidal cell in control mice. The authors state that the label was more often found in cells
16 resembling mature hepatocytes. The mean number of labeled cells in autoradiographs per 100
17 (20Qx power) fields was reported to be -125 and -150 labeled perisinusoidal cells in controls
18 male and female mice, respectively. The authors do not give any standard deviations for the
19 female perisinusoidal data except for the 1,000-mg/kg exposure group. For mature hepatocytes,
20 the mean baseline level of cell labeling for control male and female mice were reported to be -65
21 and -90 labeled cells, respectively. Although the baseline levels of hepatocyte labeling were
22 reported to differ between male and female mice, the mean peak level of labeling was similar at
23 -250 labeled cells for male and female mice treated with TCE. Thus, in male mouse liver, the
24 number of labeled cells increased -2-fold of control levels after 500 and 1,000 mg/kg TCE and
25 in female mouse liver increased -4-fold of control levels after 250, 500, and 1,000 mg/kg TCE in
26 female mouse liver hepatocytes over their respective control levels.
27 Incorporation of tritiated thymidine into DNA extracted from whole liver in male and
28 female mice was reported to be significantly elevated after TCE treatment but, unlike the
29 autoradiographic data, there was no difference between genders and the mean peak level of
30 tritiated thymidine incorporation occurred at 250 mg/kg TCE treatment and remained constant
31 for the 500 and 1,000 mg/kg treated groups. Increased thymidine incorporation into DNA
32 extracted from liver of male and female mice were reported to show a very large standard
33 deviation with TCE treatment (e.g., at 100 mg/kg TCE exposure, male mice had a mean of
34 -130 dpm tritiated thymidine/microgram DNA with the upper bound of the standard deviation to
35 be 225 dpm). The increased thymidine incorporation peaked at a level that was a little less than
36 2-fold of control level. Thus, for both male and female mice both autoradiographs and total
37 hepatic DNA were reported to show that male and female mice had similar peaks of increased
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-34 DRAFT—DO NOT CITE OR QUOTE
-------
1 thymidine incorporation after TCE exposure that reached a plateau at the 250 mg/kg TCE
2 exposure level and did not increase with increasing exposure concentration. These data also
3 indicate a very small population undergoing mitosis due to TCE exposure after 10 days of
4 exposure. If higher levels of hepatocyte replication had occurred earlier, such levels were not
5 sustained by 10 days of TCE exposure. More importantly, these data suggest that tritiated
6 thymidine levels were targeted to mature hepatocytes and in areas of the liver where greater
7 levels of polyploidization. The ages and weights of the mice were described by these authors,
8 unlike Elcombe et al, and a different strain was used. However, these results are consistent with
9 those of Elcombe in regard to the magnitude of thymidine incorporation induced by TCE
10 treatment and the lack of a dose response once a relative low level of exposure has been
11 exceeded.
12 The total liver DNA content of male and female mice treated with TCE were also
13 determined with the total micrograms DNA/g liver reported to be ~4 microgram/g for female
14 control mice and ~2 micrograms/g for male control mice. Although not statistically significant,
15 the total DNA concentration dropped from ~4 to ~3 at 100 mg/kg through 1,000 mg/kg exposure
16 to TCE in female mice. For male mice the total DNA rose slightly in the 250- and 500-mg/kg
17 groups to ~3 micrograms/gram and was similar to control levels at the 100 and 1,000 mg/kg TCE
18 treatment groups. The standard deviation in male mice was very large and the number of
19 animals small making quantitative judgments regarding this parameter difficult. The slight
20 decrease reported for female mice would be consistent with the results of Elcombe et al. (1985)
21 who describe a slight decrease in hepatic DNA in male mice. However, the reported slight
22 increase in hepatic DNA in male mice in this study is not consistent. Given the small number of
23 animals and the large deviations for female and male mice in the TCE treated groups, this study
24 may not have had the sensitivity to detect slight decreases reported by Elcombe et al.
25 In regard to clinical evaluation and weight analyses, both male and female mice given
26 TCE were reported "to appear clinically ill. These mice showed reduced activity and failed to
27 groom. Control mice showed no adverse effects. Female mice were markedly more affected by
28 TCE than their male counterparts. Several deaths of female mice occurred during the course of
29 the TCE treatment regimen." The authors do not give cause of deaths but state that two female
30 mice died in the group receiving 250 mg/kg TCE and one in the group receiving 1,000 mg/kg
31 during the gavage regimen of the female mice. This appears to be similar gavage error or
32 "accidental death" reported in National Toxicology Program (NTP) studies chronic studies of
33 TCE (see below).
34
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 The authors report
2
3 no significant difference in the absolute body weight of male and female mice
4 were noted in control groups. Body weight gain in female and males mice treated
5 with TCE was not significantly different from that of control mice. Liver weights
6 in male mice given 500 or 1,000 mg/kg and corrected for total body weight were
7 significantly elevated. The corrected liver weights of female mice increase
8 proportionally with the applied dose of TCE.
9
10 For male mice, liver weights were reported to be 1.40 ±0.16, 1.38± 1.23, 1.48 ±0.09,
11 1.61 ± 0.07, and 1.63 ± 0.11 g for control, 100, 250, 500, and 1,000 mg/kg TCE in male mice
12 (n = 5), respectively. Body weights were smaller for the 100 mg/kg TCE treatment group
13 although not statistically significant. The liver weights after treatment had a much larger
14 reported standard deviation (1.23 g for 100 mg/kg group vs. <0.16 for all other groups). The
15 percent liver/body weight ratios were reported to be 5.40, 5.41, 5.42, 5.71, and 6.34% for the
16 same groups in male mice. This represents 1.06- and 1.17-fold of control at the 500 and
17 1,000 mg/kg dose. The authors report a statistically significant increase in percent liver/body
18 weight ratio only for the 500 mg/kg (i.e., 1.06-fold of control) and 1,000 mg/kg (i.e., 1.17-fold of
19 control) TCE exposure groups. The results for female mice liver weights were reported in
20 Table III of the paper, which was mistakenly labeled as for male mice. The reported values for
21 liver weight were 1.03 ±0.07, 1.05 ±0.10, 1.15 ±0.98, 1.21 ± 0.18, and 1.34 ± 0.08 g for
22 control, 100, 250, 500, and 1,000 mg/kg TCE in female mice (n = 5, except for 250 mg/kg and
23 1,000 mg/kg groups), respectively. The percent liver/body weight ratios were 5.26, 5.44, 5.68,
24 6.24, and 6.57% for the same groups. These values represent 1.03-, 1.08-, 1.19-, and 1.25-fold
25 of controls in percent liver/body weight. The magnitude of increase in TCE-induced percent
26 liver/body weight ratio in female mice is reflective of the magnitude of the difference in dose up
27 to 1,000 mg/kg where it is slightly lower. The female mice were reported to have statistically
28 significant increases in percent liver/body ratios at the lowest dose tested (100 mg/kg TCE) after
29 10 days of TCE exposure that also increased proportionately with dose. Male mice were not
30 reported to have a significant increase in percent liver/body weight until 500 mg/kg TCE but a
31 statistically significant increase in liver weight at 250 mg/kg TCE. Male mice had a much larger
32 variation in initial body weight than did female mice (range of means of 24.86 to 27.84 g
33 between groups for males or ~11% difference and range of means of 19.48 to 20.27 g for females
34 or -4%) which may contribute to an apparent lack of effect for a parameter that is dependent on
35 body weight. Only 5 mice were used in each group so the power to detect a change was
36 relatively small.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-36 DRAFT—DO NOT CITE OR QUOTE
-------
1 The results from this experiment are consistent with those of Elcombe et al. (1985) in
2 showing a slight increase in thymidine incorporation (~2-fold of control) and mitotic figures that
3 are rare after TCE exposure. This study also records a lack of apoptosis with TCE treatment
4 except at the highest exposure level (i.e., 1,000 mg/kg). The increases in liver weight induced by
5 TCE were reported to be dose-related, especially in female mice where baseline body weights
6 were more consistent. However, the incorporation of tritiated thymidine reached a plateau at
7 250 mg/kg TCE in the DNA of both genders of mice. This study specifically identified where
8 thymidine incorporation and mitotic figures were occurring in TCE-treated livers and noted that
9 the mature hepatocyte that appeared to be primarily affected, as well as in the portion of the liver
10 where mature hepatocytes with higher ploidy are found. The authors note that the "lack of
11 thymidine incorporation in the periportal area, where the liver stem cells are reside," suggesting
12 that the mature hepatocyte is the target of TCE effects on DNA synthesis. This finding is
13 consistent with a change in ploidy accompanying hepatocellular hypertrophy and not just cell
14 proliferation after 10 days of TCE exposure. Like Elcombe et al. (1985), these data represent "a
15 snapshot in time" which does not show whether increased cell proliferation may have happened
16 at an earlier time point and then subsided by 10 days. However, like Elcombe et al. (1985) it
17 suggests that sustained proliferation is not a feature of TCE exposure and that the level of DNA
18 synthesis (which is very low in quiescent control liver) is increased in a small population of
19 hepatocytes due to TCE exposure that is not dose-dependent (only 2-fold increase over control in
20 animals exposed from 250 to 1,000 mg/kg TCE). In regards to toxicity, no evidence of increased
21 lipid peroxidation in TCE-treated animals was reported using histopathologic sections stained to
22 enhance observation of lipofuscin. No necrosis is noted by these authors and the deaths in
23 female mice are likely due to gavage error.
24
25 E.2.1.10. Nakajima et al, 2000
26 This study focused on the effect of TCE treatment on PPARa-null mice in terms of
27 peroxisome proliferation but also included information on differences in liver weight between
28 null and wild-type mice, as well as gender-related effects. SV129 wild-type and PPARa-null
29 mice (10 weeks of age) were treated with corn oil or 750 mg/kg TCE in corn oil daily for
30 2 weeks via gavage (n = 6 per group). A small portion of the liver was removed for
31 histopathological examination but the lobe used was not specified by the authors. Liver
32 peroxisome proliferation was reported to be evaluated morphologically using
33 3,3'-diaminobenzidine (DAB) staining of sections and electron photomicroscopy to detect the
34 volume density of peroxisomes (percent of cytoplasm) in 15 micrographs of the pericentral area
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-37 DRAFT—DO NOT CITE OR QUOTE
-------
1 per liver. A number of p-oxidation enzymes and P450s were analyzed by immunoblot of liver
2 homogenates.
3 The final body weights, liver weights and percent liver/body weight ratios were reported
4 for all treatment groups. For male mice, vehicle treated PPARa-null mice had slightly lower
5 mean body weights (24.5 ± 1.8 g vs. 25.4 ± 1.9 g [SD]), slightly larger liver weights
6 (1.14 ± 0.13 g vs. 1.05 ± 0.15 g or -9%), and slightly higher percent liver/body weight ratios
7 (4.12% ± 0.32% vs. 4.10% ± 0.37%) than wild-type mice. The mean values for final body
8 weights of the groups of mice in this study were reported and were similar which, as
9 demonstrated by the inhalation studies by Kjellstrand et al. (1983a) (see Section E.2.2.4), is
10 particularly important for determining the effects of TCE treatment on percent liver/body weight
11 ratios. For both groups of male mice, 2 weeks of TCE treatment significantly increased both
12 liver weight and percent liver/body weight ratios. For male wild-type mice the increase in
13 percent liver/body weight was 1.50-fold of vehicle control and for male PPARa-null mice the
14 increase was 1.26-fold of control after 2 weeks of TCE treatment. For female mice, vehicle
15 treated PPARa-null mice had slightly higher mean body weights (22.7 ± 2.1 g vs. 22.4 ± 2.0 g),
16 slightly larger liver weights (0.98 ± 0.15 g vs. 0.95 ± 0.14 g or -3%), and slightly higher percent
17 liver/body weight ratios (4.32% ± 0.35% vs. 4.24% ± 0.41%) than wild-type mice. For both
18 groups of female mice, 2 weeks of TCE treatment significantly increased percent liver/body
19 weight ratios. For liver weights there was a reporting error for PPARa-null female treated with
20 TCE so that liver weight changes due to TCE treatment cannot be determined for this group. For
21 female wild-type mice the increase in percent liver/body weight was 1.24-fold of vehicle control
22 and for female PPARa-null mice the increase was 1.26-fold of control after 2 weeks of TCE
23 treatment. Thus, for both wild-type and PPARa-null mice, TCE exposure resulted in increased
24 percent liver/body weight over controls that was statistically significant after 2 weeks of oral
25 gavage exposure using corn oil as the vehicle. For male mice there was a greater TCE-induced
26 increase in percent liver/body weight in wild-type than PPARa-null mice (1.50- vs. 1.26-fold of
27 control) that was statistically significant, but for female mice the induction of increased liver
28 weight was statistically increased but the same in wild-type and PPARa-null mice (i.e., both
29 were ~1.25-fold of control). These date indicate that TCE-induced increases in mouse liver
30 weight were not dependent on a functional PPARa receptor in female mice and suggest that
31 some portion may be in male mice.
32 In regard to light and electron microscopic results, the numbers of peroxisomes in
33 hepatocytes of wild-type mice were reported to be increased, especially in the pericentral area of
34 the hepatic lobule, to a similar extent in both males and females (15 micrographs, n = 4 mice).
35 TCE exposure was reported to increase the volume density of peroxisomes 2-fold of control in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 the pericentral area with no evident change in peroxisomes in the periportal areas, but data was
2 not shown for that area of the liver lobule. In contrast, no increase in peroxisomes was reported
3 to be observed in PPARa-null mice. Therefore, increases in liver weight observed in PPARa-
4 null mice after TCE treatment did not result from peroxisome proliferation. Similarly, the small
5 2-fold increase in peroxisome volume from 2 to 4% of cytoplasmic volume in the pericentral
6 area of the liver lobule in wild-type mice could not have been responsible for the 50% increase
7 liver weight observed in male wild-type mice.
8 Although no difference was reported between male and female wild-type mice in regard
9 to TCE-induced peroxisome proliferation in wild-type mice, the levels of hepatic enzymes
10 associated with peroxisomes (acyl-CoA [AOX], peroxisomal bifunctional protein [PH],
11 peroxisomal thiolase [PT], very long chain acyl-CoA synthetase, and D-type peroxisomal
12 bifunctional protein [DBF], cytosolic enzyme [cytosolic thioesterase II (CTEII)], mitochondrial
13 enzymes [mitochondrial trifunctional protein a subunits a and P(TPa and TPP)], and microsomal
14 enzymes [cytochrome P450 4A1 (CYP4A1)]) as measured by immunoblot analysis were
15 significantly elevated in male wild-type mice (n = 4) by a factor of -2-3, but except for a slight
16 elevation in PH and PT, were reported to not be elevated in female wild-type mice (n = 4). The
17 magnitude of increase in peroxisomal enzymes was similar to that of peroxisomal volume in
18 male mice. No TCE-induced increases in any of these enzymes were reported in male or female
19 PPARa-null mice by the authors. For CYP4A1, an enzyme reported to be induced by
20 peroxisomal proliferators, TCE exposure resulted in a much lower amount in female than male
21 wild-type mice (i.e., 2% of the level induced by TCE in males). However, the expression of
22 catalase was reported to be "nearly constant in all samples" (at most -30% change) which the
23 authors suggested resulted from induction by TCE that was independent of PPARa. The basis
24 for selection of 4 mice for this comparison out of the 6 studied per group was not given by the
25 authors. A comparison of control wild-type and PPARa-null mice showed that in males
26 background levels of the enzymes examined were generally similar except for DBF in which the
27 null mice had values -50% of the wild-type controls. A similar decrease was reported for female
28 PPARa-null mice. With regard to gender differences in wild-type mice, females had similar
29 values as males with the exceptions of TPa, TPP, and CYP2E1 which were in untreated female
30 wild-type mice at a 3.06-, 2.38-, and 1.63-fold for 1 TPa, TPp, and CYP2E1 levels over males,
31 respectively. Female PPARa-null mice had increases of 2.50-, 1.54-, and 2.07-fold over male
32 wild-type mice.
33 With regard to the induction of TCE metabolizing enzymes (CYP1A2, CYP2E1, and
34 ALDH), CYP1A2 was reported to be decreased by TCE treatment of both male and female wild-
35 type mice but liver CYP2E1 reported to be increased in male mice and constant in female mice
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-39 DRAFT—DO NOT CITE OR QUOTE
-------
1 which resulted in similar expression level in both genders after TCE treatment. There was no
2 gender difference in ALDH activity reported after TCE exposure and activity was reported to be
3 independent of PPARa. The authors concluded that TCE metabolizing abilities of the liver of
4 male and female mice were similar and therefore, poor induction of peroxisomal related enzymes
5 was not due to gender-related differences in TCE metabolism.
6 To investigate whether the a gender-related difference peroxisomal enzymes after TCE
7 exposure was due to a lower levels of PPARa and RXRa receptors, western blotting was
8 employed (n = 3). The level of PPARa protein was reported to be increased in both male wild-
9 type mice with less induction in females (control vs. TCE, 1.00 ± 0.20 vs. 2.17 ± 0.24 in males
10 and 0.95 ± 0.25 vs. 1.44 ± 0.09 in females) after TCE treatment. The hepatic level of RXRa was
11 also reported to be increased in the same manner as PPARa (control vs. TCE, 1.00 ± 0.33 vs.
12 1.92 ± 0.04 in males 0.81 ± 0.16 vs. 1.14 ± 0.10 in females). Northern blot analysis of hepatic
13 PPARa mRNA was reported to show greater TCE induction in male (2.6-fold of control) than in
14 female (1.5-fold of control) wild-type mice. Thus, males appeared to have higher induction of
15 the two receptor proteins as well as a greater response in peroxisomal enzymes and CYP4A1,
16 even though TCE-induced increases in peroxisomal volume was similar between male and
17 female mice. The increased response in males for induction of the two receptor proteins is
18 consistent with liver weight data that shows some portion of the induction of increased liver
19 weight response in male mice using this paradigm may be due to gender-specific differences in
20 PPARa response. However, as noted below (see Section E.2.2), corn oil vehicle has liver effects
21 alone, especially in the male liver, that have also been associated with PPARa responses.
22
23 E.2.1.11. Berman et al, 1995
24 This study included TCE in a suite of compounds used to compare endpoints for
25 toxicological screening methods. Female Fischer 344 rats of 77 days of age (n = 8 per group)
26 were administered TCE in corn oil for 1 day (0, 150, 500, 1,500, or 5,000 mg/kg/d) or for
27 14 days (0, 50, 150, 500, or 1,500 mg/kg/d). Blood samples were taken 24 hours after the last
28 dose and livers were weighed and H&E sections were examined for evidence of parenchymal
29 cell degeneration, necrosis, or hypertrophy. No details were provided by the authors for the
30 extent or severity of the liver affects by histopathological examination. The serum chemistry
31 analysis included lactate dehydrogenase (LDH), alkaline phosphatase, ALT, aspartate
32 aminotrasferase (AST), total bilirubin, creatine, and blood urea nitrogen. The starting and
33 ending body weights of the animals or the absolute liver weights were not reported by the
34 authors.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-40 DRAFT—DO NOT CITE OR QUOTE
-------
1 The results of a multivariate analysis were reported to show a lowest effective dose of
2 1,500 mg/kg after 1 day of TCE exposure and 150 mg/kg after 14 days of TCE exposure that was
3 statistically significant. Liver weight and liver weight changes were not reported by the authors
4 but the percent liver to body weight ratios were. For the two control groups there was a
5 difference in percent liver/body weight of-8% (3.43% ± 0.74% for the 1-day control group and
6 3.16% ± 0.41% for the 14-day control group, mean ± SEM). For the 1-day groups only the
7 5,000 mg/kg group was reported to show a statistically significant difference in percent
8 liver/body weight between control and TCE treatment (i.e., ~1.08-fold increase). Hepatocellular
9 necrosis was noted to occur in the 1,500 and 5,000 mg/kg groups in 6/7 and 6/8 female rats,
10 respectively but not to occur in lower doses. The extent of necrosis was not noted by the authors
11 for the two groups exhibiting a response after 1 day of exposure. Serum enzymes indicative of
12 liver necrosis were not presented and because only positive results were presented in the paper,
13 presumed to be negative. Therefore, the extent of necrosis was not of a magnitude to affect
14 serum enzyme markers of cellular leakage.
15 After 14 days of TCE exposure, there was a dose-related increase reported for percent
16 liver/body weight ratios that was statistically significant at all TCE dose levels although the
17 multivariate analysis indicated the lowest effective dose to be 150 mg/kg. The percent
18 liver/body weight ratio was 3.16% ± 0.41%, 3.38% ± 0.56%, 3.49% ± 0.69%, 3.82% ± 0.76%,
19 and 4.47% ± 0.66% for control, 50, 150, 500, and 1,500 mg/kg TCE exposure levels,
20 respectively after 14 days of exposure. No hepatocellular necrosis was reported at any dose and
21 hepatocellular hypertrophy was reported only at the 1,500 mg/kg dose and in all rats. These rat
22 liver weights are 1.07-, 1.10-, 1.21-, and 1.41-fold of controls for the 50, 150, 500, and
23 1,500 mg/kg TCE dose groups, respectively. The 7% increase in liver weight at the 50 mg/kg
24 dose is approximately the same difference between the two control groups for Days 1 and
25 14 treatments. Without the data for starting and final body weights and an examination of
26 whether the control animals had similar body weight, it is impossible to discern whether the
27 reported effects at the low dose of TCE was also reflected differences between the control
28 groups. No serum enzyme levels changes were reported after 14 days of exposure to TCE for
29 any group.
30 The authors note that their study provided evidence of liver effects at lower levels than
31 other studies citing Elcombe et al. (1985) and Goldsworthy and Popp (1987). They suggest that
32 the differences in sensitivity to TCE between their results and those of these two studies may
33 reflect differences in strain or gender of the rats examined. However, they did not study male
34 rats of this strain concurrently so that differences in gender may have reflected differences
35 between experiments. The increase in liver weight without reporting increases in hepatocellular
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-41 DRAFT—DO NOT CITE OR QUOTE
-------
1 hypertrophy as well as the lack of necrosis as low doses is consistent with the results of Melnick
2 et al. (1987) in male Fischer rats given TCE orally (see Section E.2.1.11, below).
3
4 E.2.1.12. Melnick et al, 1987
5 The focus of this study was to assess microencapsulation as a way to expose rodents to
6 substances such as TCE that have issues related to volatilization in drinking water or apparent
7 gavage-related deaths. In this study, liver weight changes, extent of focalized necrosis, and
8 indicators of peroxisome proliferation were reported as metrics of TCE toxicity. TCE (99+ %)
9 was encapsulated in gelatin-sorbitol microcapsules and was 44.1% TCE w/w. The TCE
10 microcapsules were administered to male Fischer 344 rats (6-week old and weighing between 89
11 and 92 g or -3% difference) in the diet (0, 0.55, 1.10, 2.21, and 4.42% TCE in the diet) for
12 14 days. The number of animals in each group was 10. A parallel group of animals was
13 administered TCE in corn oil gavage for 14 consecutive days (corn oil control, 0.6, 1.2, and
14 2.8 g/kg/day TCE). The dosage levels of TCE in the gavage study were reported to be "adjusted
15 5 times during the 14-day" treatment period to be similar to the dosage levels of TCE in the feed
16 study. The time-weighted average dosage levels of TCE in the feed study were reported to be
17 0.6, 1.3, 2.2, and 4.8 g/kg/day.
18 There was less food consumption reported in the 2.2 and 4.8 g/kg/day dose feed groups,
19 which the authors attribute to either palatability or toxicity. There were no deaths in any of the
20 groups treated with microencapsulated TCE while, similar to many other gavage studies of TCE
21 reported in the literature, there were 4 deaths in the high-dose gavage group. Mean body weight
22 gains of the two highest dose groups of the feed study and of the highest dose group of the
23 gavage study were reported to be significantly lower than the mean body weight gains of the
24 respective control groups (i.e., -22 and -35% reduction at 2.2 and 4.8 g/kg/day in the feed study,
25 respectively, and -33% reduction at 2.8 g/kg/day TCE in the gavage study). After 14 days of
26 treatment, liver weights were reported to be 8.1 ±0.8, 8.4 ±0.8, 9.5 ± 0.5, 10.1 ± 1.2, 8.9 ± 1.3,
27 and 7.4 ± 0.5 g for untreated control, placebo control, 0.6, 1.3, 2.2, and 4.8 g/kg TCE exposed
28 feed groups, respectively. The corresponding percent liver/body weight ratios were reported to
29 be 5.2% ± 0.3%, 5.3% ± 0.2%, 6.0% ± 0.3%, 6.5% ± 0.5%, 7.0% ± 0.9%, and 7.1% ± 0.5% for
30 untreated control, placebo control, 0.6, 1.3, 2.2, and 4.8 g/kg TCE exposed groups, respectively.
31 The increased percent liver/body weight ratio represents 1.13-, 1.23-, 1.32-, and 1.34-fold of
32 placebo controls, respectively. For the gavage experiment, after 14 days of treatment liver
33 weights were reported to be 7.1 ± 1.3, 9.3 ± 1.2, 9.1 ± 0.9, and 7.7 ± 0.4 g for corn oil control,
34 0.6, 1.2, and 2.8 g/kg TCE exposed groups, respectively. The corresponding percent liver/body
35 weight ratios were reported to be 5.0% ± 0.4%, 6.0% ± 0.4%, 6.1% ± 0.3%, and 7.3% ± 0.5% for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-42 DRAFT—DO NOT CITE OR QUOTE
-------
1 corn oil control, 0.6, 1.2, and 2.8 g/kg TCE exposed groups, respectively. The percent liver/body
2 weight ratios represent 1.20-, 1.22-, and 1.46-fold of corn oil controls, respectively. The 2.8
3 g/kg TCE gavage results are reflective of the 6 surviving animals in the group rather than 10
4 animals in the rest of the groups. There was no explanation given by the authors for the lower
5 liver weights in the control gavage group than the placebo control in the feed group (i.e., 20%
6 difference) although the initial and final body weights appeared to be similar. The decreased
7 body weights in the feed and gavage study are reflective if TCE systemic toxicity and appeared
8 to affect the TCE-induced liver weight increases in those groups.
9 The authors reported that the only treatment-related lesion observed microscopically in
10 rats from either dosed-feed or gavage groups was individual cell necrosis of the liver with the
11 frequency and severity of this lesion similar at each dosage levels of TCE administered
12 microencapsulated in the feed or in corn oil. Using a scale of minimal = 1-3 necrotic
13 hepatocytes/10 microscopic 200x fields, mild = 4-7 necrotic necrotic hepatocytes/10
14 microscopic 200* fields, and moderate = 8-12 necrotic hepatocytes/10 microscopic 200* fields,
15 the frequency of lesion was 0-1/10 for controls, 2/10 for 0.6 and 1.3 g/kg and 9/10 for 2.2 and
16 4.8 g/kg feed groups. The mean severity was reported to be 0.0-0.1 for controls, 0.3-0.4 for 0.6
17 and 1.3 g/kg, and 2.0-2.5 for 2.2 and 4.8 g/kg feed groups. For the corn oil gavage study, the
18 corn oil control and 0.6 g/kg groups were reported to have a frequency of 0 lesions/10 animals,
19 the 1.2 g/kg group a frequency of 1/10 animals, while the 2.8 g/kg group to have a frequency of
20 5/6 animals. The mean severity score was reported to be 0 for the control and 0.6 g/kg groups,
21 0.1 for the 1.2 g/kg groups, and 1.8 for the remaining 6 animals in the 2.8 g/kg group. The
22 individual cell necrosis was reported to be randomly distributed throughout the liver lobule with
23 the change to not be accompanied by an inflammatory response. The authors also report that
24 there was no histologic evidence of cellular hypertrophy or edema in hepatic parenchymal cells.
25 Thus, although there appeared to be TCE-treatment related increases in focal necrosis after
26 14 days of exposure, the extent was even at the highest doses mild and involved few hepatocytes.
27 Microsomal NADPH cytochrome c-reductase was reported to be elevated in the 2.2 and
28 4.8 g/kg feed groups and in the 1.2 and 2.8 g/kg gavage groups. Cytochrome P450 levels were
29 reported to be elevated only in the two highest dose groups of the feed study. The authors
30 reported a dose-related increase in peroxisome PCO and catalase activities in liver homogenates
31 from rats treated with TCE microcapsules or by gavage and that treatment with corn oil alone,
32 but not placebo capsules, caused a slight increase in PCO activity. After 14 days of treatment,
33 PCO activities were reported to be 270 ± 12, 242 ± 17, 298 ± 64, 424 ± 55, 651 ± 148, and
34 999 ± 266 nmol H202 produced/min/g liver for untreated control, placebo control, 0.6, 1.3, 2.2,
35 and 4.8 g/kg TCE exposed feed groups, respectively. This represents 1.23-, 1.75-, 2.69-, and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-43 DRAFT—DO NOT CITE OR QUOTE
-------
1 4.13-fold of placebo controls, respectively. After 14 days of treatment, catalase activities were
2 reported to be 8.49 ±0.81, 7.98 ± 1.62, 8.49 ± 1.92, 8.59 ± 1.31, 13.03 ± 2.01, and
3 15.76± 1.11 nmol H202 produced/min/g liver for untreated control, placebo control, 0.6, 1.3, 2.2,
4 and 4.8 g/kg TCE exposed groups, respectively. This represents 1.06-, 1.07-, 1.63-, and
5 1.97-fold of placebo controls, respectively. Thus, although reported to be dose related, only the
6 two highest exposure levels of TCE increased catalase activity and to a smaller extent than PCO
7 activity in microencapsulated TCE fed rats. For the gavage experiment, after 14 days of
8 treatment PCO activities were reported to be 318 ± 27, 369 ± 26, 413 ± 40, and
9 1,002 ± 271 nmol hydrogen peroxide (H2O2) produced/min/g liver for corn oil control, 0.6, 1.2,
10 and 2.8 g/kg TCE exposed groups, respectively. This represents 1.16-, 1.29-, and 3.15-fold of
11 corn oil controls. After 14 days of treatment, catalase activities were reported to be 8.59 ± 0.91,
12 10.10 ± 1.82, 12.83 ± 3.43, and 13.54 ± 2.32 nmol H2O2 produced/min/g liver for corn oil
13 control, 0.6, 1.2, and 2.8 g/kg TCE exposed groups, respectively. This represents 1.18-, 1.49-,
14 and 1.58-fold of corn oil controls. As stated by the authors the corn oil vehicle appeared to
15 elevate catalase activities and PCO activities.
16 In regard to dose-response, liver and body weight were affected by decreased body
17 weight gain in the higher dosed animals in this experiment (i.e., 2.2 g/kg/day TCE exposure and
18 above) and by gavage related deaths in the highest-dosed group. The lower liver weight in the
19 gavage control group also may have affected the determination of the magnitude of TCE-related
20 liver weight gain at that dose. At the 2 doses, below which body weight gain was affected, there
21 appeared to be an approximately 20% increase in percent liver/body weight ratio in the gavage
22 study and a 13 and 23% weight increase in the feed study. The extent of PCO activity appeared
23 to increase more steeply with dose in the feed study than did liver weight gain (i.e., a 1.23-fold of
24 liver/body weight ratio at 1.3 g/kg/day corresponded with a 1.75-fold PCO activity over control).
25 At the two highest doses in the feed study, the increase in PCO activity was 2.69- and 4.13-fold
26 of control but the increase in liver weight was not more than 34%. For the gavage study, there
27 was also a steeper increase in PCO activity than liver weight gain. For catalase activity, the
28 increase was slightly less than that of liver/body weight ratio percent for the two doses that did
29 not decrease body weight gain in the feed study. In the gavage study, they were about the same.
30 In regard to what the cause of liver weight gain was, the authors report that there was no
31 histologic evidence of cellular hypertrophy or edema in hepatic parenchymal cells and do not
32 describe indicators of hepatocellular proliferation or increased polyploidy. Accordingly, the
33 cause of liver weight gain after TCE exposure in this paradigm is not readily apparent.
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-44 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.1.13. Laughter et al, 2004
2 Although the focus of the study was an exploration of potential MO As for TCE effects
3 through macroarray transcript profiling (see Section E.3.1.2 for discussions of limitations of this
4 approach and especially the need for phenotypic anchoring, Section E.3.4.1.3 for use of PPARa
5 knockout mice, and Section E.3.4.2.2 for discussion of genetic profiling data for TCE),
6 information was reported regarding changes in the liver weight of PPARa-null mouse and their
7 background strains. SV129 wild-type and PPARa-null male mice (9 ± 1.5 weeks of age) were
8 treated with 3 daily doses of TCE in 0.1% methyl cellulose for either 3 days or 3 weeks
9 (n = 4-5/group). Thus, this paradigm does not use corn oil, which has been noted to affect
10 toxicity (see Section E.2.2 below), but is not comparable to other paradigms that administer the
11 total dose in one daily gavage administration rather than to give the same cumulative dose but in
12 3 daily doses of lower concentration. The initial or final body weights of the mice were not
13 reported. Thus, the effects of systemic toxicity from TCE exposure on body weight and the
14 influence of differences in initial body weight on percent liver/body weight determinations
15 cannot be made. For the 3-day study, mice were administered 1,500 mg/kg TCE or vehicle
16 control. For the 3-week study, mice were administered 0, 10, 50, 125, 500, 1,000, or
17 1,500 mg/kg TCE 5 days a week except for 4 day/week on the last week of the experiment. In a
18 separate study, mice were given TCA or dichloroacetic acid (DCA) at 0.25, 0.5, 1, or 2 g/L
19 (pH ~7) in the drinking water for 7 days. For each animal a block of the left, anterior right, and
20 median liver lobes was reported to be fixed in formalin with 5 sections stained for H&E and
21 examined by light microscopy. The remaining liver samples were combined and used as
22 homogenates for transcript arrays. In the 3-week study, bromodeoxyuridine (BrdU) was
23 administered via miniosmotic pump on day one of Week 3 and sections of the liver assessed for
24 BrdU incorporation in at least 1,000 cells per animal in 10-15 fields.
25 Although initial body weights, final body weights, and the liver weights were not
26 reported, the percent liver/body ratios were. In the 3-day study, control wild-type and PPARa-
27 null mice were reported to have similar percent liver/body weight ratios of-4.5%. These
28 animals were -10 weeks of age upon sacrifice. However, at the end of the 3-week experiment
29 the percent liver/body weight ratios were increased in the PPARa-null male mice and were 5.1%.
30 There was also a slight difference in the percent liver/body weight ratios in the 1-week study
31 (4.3% ± 0.4% vs. 4.6% ± 0.2% for wild-type and PPARa-null mice, respectively). These results
32 are consistent with an increasing baseline of hepatic steatosis with age in the PPARa-null mice
33 and increase in liver weight. In the 3-day study, the mean reported the percent liver/body ratio
34 was 1.4-fold of the animals tested with TCE in comparison to the control level. In the PPARa-
35 null mice, there was a 1.07-fold of control level reported by the authors to not be statistically
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-45 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant. However, given the low number of animals tested (the authors give only that
2 4-5 animals were tested per group without identification as to which groups has 4 animals and
3 which had 5), the ability of this study to discern a statistically significant difference is limited. In
4 the 3-week study, wild-type mice exposed to various concentrations of TCE had percent
5 liver/body weights that were within -2% of control values except for the 1,000 mg/kg and
6 1,500 mg/kg groups that were -1.18- and 1.30-fold of control levels, respectively. For the
7 PPARa-null mice exposed to TCE for 3 weeks, the variability in percent liver/body weight was
8 greater than that of the wild-type mice in most of the groups. The baseline level percent
9 liver/body weight was 1.16-fold in the PPARa-null mice in comparison to wild-type mice. At
10 the 1,500 mg/kg TCE exposure level percent liver/body weights were not recorded because of
11 the death of the null mice at this level. The authors reported that at the 1,500 mg/kg level all
12 PPARa-null mice were moribund and had to be removed from the study. However, at
13 1,000 mg/kg TCE exposure level there was a 1.10-fold of control percent liver/body weight
14 value that was reported to not be statistically significant. However, as noted above, the power of
15 the study was limited due to low numbers of animals and increased variability in the null mice
16 groups. The percent liver/body weight reported in this study was actually greater in the null
17 mice than the wild-type male mice at the 1,000 mg/kg TCE exposure level (5.6% ± 0.4% vs.
18 5.2% ± 0.5%, for null and wild-type mice, respectively). Thus, at 1-week and at 3-weeks, TCE
19 appeared to induce increases in liver weight in PPARa-null mice, although not reaching
20 statistical significance in this study, with concurrent background of increased liver weight
21 reported in the knockout mice. At 1,000 mg/kg TCE exposure for 3 weeks, percent liver/body
22 weight was reported to be 1.18-fold in wild-type and 1.10-fold in null mice of control values. As
23 discussed above, Nakajima et al. (2000) reported statistically significant increased liver weight in
24 both wild-type and PPARa-null mice after 2 weeks of exposure with less TCE-induced liver
25 weight increases in the knockout mice (see Section E.2.1.10). They also used more mice,
26 carefully matched to weights of their mice, and used a single dose of TCE each day with corn oil
27 gavage.
28 The authors noted that inspection of the livers and kidneys of the moribund null mice,
29 who were removed from the 3-week study, "did not reveal any overt signs of toxicity in this dose
30 group that would lead to morbidity" but did not show the data and did not indicate when the
31 animals were affected and removed. For the wild-type mice exposed to the same concentration
32 (1,500 mg/kg) but whose survival was not affected by TCE exposure, the authors reported that at
33 the 1,500 mg/kg dose these mice exhibited mild granuloma formation with calcification or mild
34 hepatocyte degeneration but gave not other details or quantitative information as to the extent of
35 the lesions or what parts of the liver lobule were affected. The authors noted that "wild-type
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-46 DRAFT—DO NOT CITE OR QUOTE
-------
1 mice administered 1000 and 1500 mg/kg exhibited centrilobular hypertrophy" and that "the mice
2 in the other groups did not exhibit any gross pathological changes after TCE exposure." Thus,
3 the hepatocellular hypertrophy reported in this study for TCE appeared to be correlated with
4 increases in percent liver/body weight in wild-type mice. In regard to the PPARa-null mice, the
5 authors stated that "differences in the liver to body weights in the control PPARa-null mice
6 between Study 1 and 2 the 3-day and 3-week studies] were noted and may be due to differences
7 in the degree of steatosis that commonly occurs in this strain." Further mention of the
8 background pathology due to knockout of the PPARa was not discussed. The increased percent
9 liver/body weight reported between control and 1,000 mg/kg TCE exposed mice (5.1 vs. 5.6%)
10 was not accompanied by any discussion of pathological changes that could have accounted for
11 the change.
12 Direct comparisons of the effects of TCE, DC A, and TCA cannot be made from this
13 study as they were not studied for similar durations of exposure. However, while TCE induced
14 increased in percent liver/body weight ratios after 3 days and 3 weeks of exposure in wild-type
15 mice at the highest dose levels, for TCA exposure percent liver/body weight after 1 week
16 exposure in drinking water was slightly elevated at all dose levels with no dose-response (-10%
17 increase), and for DC A exposure in drinking water a similar elevation in percent liver/body
18 weight was also reported for the 0.25, 0.5, and 1.0 g/L dose levels (~11%) and that was increased
19 at the 2.0 g/L level by -25% reaching statistical significance. The authors interpret these data to
20 show no TCA-related changes in wild-type mice but the limited power of the study makes
21 quantitative conclusions difficult. For PPARa-null mice all there was a slight decrease in
22 percent liver/body weight between control and TCA treated mice at the doses tested (-2%). For
23 DCA-treated mice, all treatment levels of DCA were reported to induce a higher percent
24 liver/body weight ratio of at least -5% with a 13% increase at the 2.0 g/L level. Again the
25 limited power of the study and the lack of data for TCE at similar durations of exposure as those
26 studied for TCA and DCA makes quantitative conclusions difficult and comparisons between the
27 chemicals difficult. However, the pattern of increased percent liver/body weight appears to be
28 more similar between TCE and DCA than TCA in both wild-type and PPARa-null mice. In
29 terms of histological description of effects, the authors note that "livers from the 2 g/L DCA-
30 treated wild-type and PPARa-null mice had hepatocyte cytoplasmic rarefication probably due to
31 an increase in glycogen accumulation." However, no special procedures are staining were
32 performed to validate the assumption in this experiment. No other pathological descriptions of
33 the DCA treatment groups were provided. In regard to TCA, the authors noted that "the livers
34 from wild-type but not PPARa-null mice exposed to 2.0g/L TCA exhibited centrilobular
35 hepatocyte hypertrophy." No quantitative estimate of this effect was given and although the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-47 DRAFT—DO NOT CITE OR QUOTE
-------
1 extent of increase of percent liver/body weight was similar for all dose levels of TCA, there is no
2 indication from the study that lower concentrations of TCA also increased hepatocellular
3 hypertrophy or why there was no concurrent increase in liver weight at the highest dose of TCA
4 in which hepatocellular hypertrophy was reported. Thus, reports of hepatocellular hypertrophy
5 for DC A and TCA in the 1-week study were not correlated with changes in percent liver/body
6 weight.
7 For control animals, BrdU incorporation in the last week of the 3-week study was
8 reported to be at a higher baseline level in PPARa-null mice than wild-type mice (~2.5-fold).
9 For wild-type mice the authors reported a statistically significant increase at 500 and
10 1,000 mg/kg TCE at levels of ~1 and -4.5% hepatocytes incorporating the label after 5 days of
11 BrdU incorporation. Whether this measure of DNA synthesis is representative of cellular
12 proliferation or of polyploidization was not examined by the authors. Even at 1,000 mg/kg TCE
13 the percent of cells that had incorporated BrdU was less than 5% of hepatocytes in wild-type
14 mice. The magnitude percent liver/body weight ratio change at this exposure level was 4-fold
15 greater than that of hepatocytes undergoing DNA synthesis (16% increase in percent liver/body
16 weight ratio vs. 4% increase in DNA synthesis). The -1% of hepatocytes undergoing DNA
17 synthesis at the 500 mg/kg TCE level, reported to be statistically significant by the authors, was
18 not correlated with a concurrent increase in percent liver/body weight ratio. Thus, TCE-induced
19 changes in liver weight were not correlated with increases in DNA synthesis in wild-type mice
20 after 3 weeks of TCE exposure. For PPARa-null mice, there was a ~3-fold of control value for
21 the percent of hepatocytes undergoing DNA synthesis at the 1,000 mg/kg TCE exposure level.
22 The higher baseline level in the null mouse, large variability in response at this exposure level,
23 and low power of this experimental design limited the ability to detect statistical significance of
24 this effect although the level was greater than that reported for the 500 mg/kg TCE exposure in
25 wild-type mice that was statistically significant. Thus, TCE appeared to induce an increase in
26 DNA synthesis in PPARa-null mice, albeit at a lower level than wild-type mice. However, the
27 -2% increase in percent of hepatocytes undergoing DNA synthesis during the 3rd week of a
28 3-week exposure to 1,000 mg/kg TCE in PPARa-null mice was insufficient to account for the
29 -10% observed increase in liver weight. For wild-type and PPARa-null mice, the magnitude of
30 TCE-induced increases in liver weight were 4-5-fold higher than that of increases in DNA-
31 synthesis under this paradigm and in both types of mice, a relatively small portion of hepatocytes
32 were undergoing DNA synthesis during the last week of a 3-week exposure duration. Whether
33 the increases in liver weight could have resulted from and early burst of DNA synthesis as well
34 as whether the DNA synthesis results reported here represents either proliferation or
35 polyploidization, cannot be determined from this experiment. Because of the differences in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-48 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure protocol (i.e., use of 3 daily doses in methylcellulose rather than one dose in corn oil)
2 the time course of the transient increase in DNA synthesis reported cannot be assumed to be the
3 same for this experiment and others.
4 Not only were PPARa-null mice different than wild-type mice in terms of background
5 levels of liver weights, and hepatic steatosis, but this study reported that background levels of
6 PCO activity to be highly variable and in some instances different between wild-type and null
7 mice. There was reported to be ~6-fold PCO activity in PPARa-null control mice in comparison
8 to wild-type control mice in the 1-week DCA/TCA experiment (-0.15 vs. 0.85 units of activity/g
9 protein). However, in the same figure a second set of data are reported for control mice for
10 comparison to WY-14,643 treatment in which PCO activity was slightly decreased in PPARa-
11 null control mice versus wild-type controls (-0.40 vs. 0.65 units of activity/g protein). In the
12 experimental design description of the paper, WY-14,643 treatment and a separate control were
13 not described as part of the 1-week DCA/TCA experiment. For the only experiment in which
14 PCO activity was compared between wild-type and PPARa-null mice exposed to TCE (i.e.,
15 3-day exposure study), there was a reported increased over the control value of-2.5-fold that
16 was reported to be statistically significant at 1,500 mg/kg TCE (1.5 vs. 0.60 units of activity/g
17 protein). For control mice in the 3-day TCE experiment, there was an increase in this activity in
18 PPARa-null mice in comparison to wild-type mice (-0.60 vs. 0.35 units of activity/g protein).
19 While not statistically significant, there appeared to be a slight increase in PCO activity after
20 1,500 mg/kg TCE exposure for 3 days in PPARa-null mice of-30%. However, as noted above
21 the background levels of this enzyme activity varied widely between the experiments with not
22 only values for control animals varying as much as 6-fold (i.e., for PPARa-null mice) but also
23 for WY-14,643 administration. There was a 6.6-fold difference in PCO results for WY-14,643
24 in PPARa-null mice at the same concentration of WY-14,643 in the 3-day and 1-week
25 experiment, and a 1.44-fold difference in results in wild-type mice in these two data sets.
26
27 E.2.1.14. Ramdhan et al, 2008
28 Ramdhan et al. (2008) examined the role of CYP2E1 in TCE-induced hepatotoxicity,
29 using CYP2E1 +/+ (wild-type) and CYP2E1 -/- (null) Sv/129 male mice (6/group) which were
30 exposed for 7 days to 0, 1,000, or 2,000-ppm TCE by inhalation for 8 hours/day (Ramdhan et al.,
31 2008). The exposure concentrations are noted by the authors to be much higher than
32 occupational exposures and to have increased liver toxicity after 8 hours of exposure as
33 measured by plasma AST levels. To put this exposure concentration into perspective, the
34 Kjellstrand et al. (1983a, b) inhalation studies for 30 days showed that these levels were well
35 above the 150-ppm exposure levels in male mice that induced systemic toxicity. Nunes also
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-49 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported hepatic necrosis up to 4% in rats at 2,000 ppm for just 8 hours not 7 days. AST and
2 ALT were measured at sacrifice. Histological changes were scored using a qualitative scale of
3 0 = no necrosis, 1 = minimal as defined as only occasional necrotic cells in any lobule, 2 = mild
4 as defined as less than one-third of the lobule structure affected, 3 = moderate as defined as
5 between one-third and two-thirds of the lobule structure affected and 4 = severe defined as
6 greater than two-thirds of the lobule structure affected. Real-time polymerase chain reaction
7 (PCR) was reported for mRNA encoding a number of receptors and proteins. Total RNA and
8 Western Blot analysis was obtained from whole-liver homogenates. The changes in mRNA
9 expression were reported as means for 6 mice per group after normalization to a level of p-actin
10 mRNA expression and were shown relative to the control level in the CYP2E1 wild-type mice.
11 The deletion of the CYP2E1 gene in the null mouse had profound effects on liver weight.
12 The body were was significantly increased in control CYP2E1 -/- mice in comparison to wild-
13 type controls (24.48 ± 1.44 g for null mice vs. 23.66 ± 2.44 g, m ± SD). This represents a 3.5%
14 increase over wild-type mice. However, the liver weight was reported in the CYP2E1 -/- mice to
15 be 1.32-fold of that of CYP2E1+/+mice (1.45 ±0.10 g vs. 1.10±0.14g). The percent
16 liver/body weight ratio was 5.47 versus 4.63% or 1.18-fold of wild-type control for the null
17 mice. The authors report that 1,000-ppm and 2,000-ppm TCE treatment did induce a statistically
18 significant change body weight for null or wild-type mice. However, there was an increase in
19 body weight in the wild-type mice (i.e., 23.66 ± 2.44, 24.52 ±1.17, and 24.99 ± 1.78 for control,
20 1,000 ppm, and 2,000-ppm groups, respectively) and an increase in the variability in response in
21 the null mice (i.e., 24.48 ± 1.44, 24.55 ± 2.26, and 24.99 ± 4.05, for control, 1,000 ppm, and
22 2,000 ppm exposure groups, respectively). The percent liver/body weight was 5.47% ± 0.23%,
23 5.51% ± 0.27%, and 5.58% ± 0.70% for control, 1,000 ppm and 2,000 ppm the CYP2E1 -/-
24 mice, respectively. The percent liver/body weight was 4.63% ± 0.13%, 6.62% ± 0.40%, and
25 7.24% ± 0.84% for control, 1,000 ppm, and 2,000 ppm wild-type mice, respectively. Therefore,
26 while there appeared to be little difference in the TCE and control exposures for percent
27 liver/body weights in the CYP2E1 -/- mice (2%) there was a 1.56-fold of control level after
28 2,000 ppm in the wild-type mice after 7 days of inhalation exposure.
29 The authors reported that "in general, the urinary TCE level in CYP2E1 -/- mice was less
30 than half that in CYP2E1 +/+ mice: urinary TCA levels in the former were about one-fourth
31 those in the latter." Of note is the large variability in urinary TCE detected in the 2,000-ppm
32 TCE exposed wild-type mice, especially after Day 4, and that in general the amount of TCE in
33 the urine appeared to be greatest after the 1st day of exposure and steadily declined between 1
34 and 7 days (i.e., -45% decline at 2,000 ppm and a -70% decline at 1,000 ppm) in the wild-type
35 mice. The amount of TCE in the urine was proportional to the difference in dose at days 1 and 5
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-50 DRAFT—DO NOT CITE OR QUOTE
-------
1 (i.e., a 2-fold difference in dose resulted in a 2-fold difference in TCE detected in the urine). As
2 the detection of TCE in the urine declined with time, the amount of TCA was reported to steadily
3 increase between days 1 and 7 (e.g., from ~3 mg TCA after the 1st day to -5.5 mg after 7 days
4 after 2,000 ppm exposure in wild-type mice). However, unlike TCE, there was a much smaller
5 differences in response between the two TCE exposure levels (i.e., a 12-44% or 1.12- to 1.44-
6 fold difference in TCA levels in the urine at days 1-7 for exposure concentrations that differ by a
7 factor of 2). This could be indicative of saturation in metabolism and TCA clearance into urine
8 at these high concentrations levels. The authors note that their results suggest that the
9 metabolism of TCE in both null and wild-type mice may have reached saturation at 1,000 ppm
10 TCE.
11 For ALT and AST activities in CYP2E1 -/- or CYP2E1 +/+ mice, both liver enzymes
12 were significantly elevated only at the 2,000 ppm level in CYP2E1 +/+ mice. Although the
13 increases in excreted TCA in the urine differed by only -33% between the 1,000 and 2,000 ppm
14 levels, liver enzyme levels in plasma differed by a much greater extent after 7 days exposure
15 between the 1,000 and 2,000-ppm groups of CYP2E1 +/+ mice (i.e., 1.26- and 1.83-fold of
16 control [ALT] and 1.40- and 2.20-fold of control [AST] for 1,000 ppm and 2,000 ppm TCE
17 exposure levels, respectively). The authors reported a correlation between plasma ALT and both
18 TCE (r = 0.7331) and TCA (r = 0.8169) levels but do not report details of what data were
19 included in the correlation (i.e., were data from CYP2E1 +/+ mice combined with those of the
20 CYP2E1 -/- mice and were control values included with treated values?).
21 The authors show photomicrograph of a section of liver from control CYP2E1 +/+ and
22 CYP2E1 -/- mice and describe the histological structure of the liver to appear normal. This
23 raises the question as to the cause of the hepatomegaly for the CYP2E1 mice in which the liver
24 weight was increased by a third. The qualitative scoring for each of the 6 animals per group
25 showed that none of the CYP2E1 -/- control or treated mice showed evidence of necrosis. For
26 the CYP2E1 +/+ mice there was no necrosis reported in the control mice and in 3/6 mice treated
27 with 1,000 ppm TCE. Of the 3 mice that were reported to have necrosis, the score was reported
28 as 1-2 for 2 mice and 1 for the third. It is not clear what a score of 1-2 represented given the
29 criteria for each score given by the authors, which defined a score of 1 as minimal and one of 2
30 as mild. For the 2,000 ppm TCE-exposed mice, all mice were reported to have at least minimal
31 necrosis (i.e., 4 mice were reported to have scores of 1-2, one mouse a score of 3 and one mouse
32 a score of 1). What is clear from the histopathology data are that there appeared to be great
33 heterogeneity of response between the 6 animals in each TCE-exposure group in CYP2E1 +/+
34 mice and that there was a greater necrotic response in the 2,000-ppm-exposed mice than the
35 1,000 ppm mice. These results are consistent with the liver enzyme data but not consistent with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-51 DRAFT—DO NOT CITE OR QUOTE
-------
1 the small difference between the 1,000 ppm and 2,000 ppm exposure groups for TCA content in
2 urine and by analogy, metabolism of TCE to TCA. A strength of this study is that it reports the
3 histological data for each animal so that the heterogeneity of liver response can be observed (e.g.,
4 the extent of liver necrosis was reported to range from only occasional necrotic cells in any
5 lobule to between one-third and two-thirds of the lobular structure affected after 2,000 ppm TCE
6 exposure for 7 days). Immunohistochemical analysis was reported to show that CYP2E1 was
7 expressed mainly around the centrilobular area in CYP2E1 +/+ mice where necrotic changes
8 were observed after TCE treatment.
9 Given the large variability in response within the liver after TCE exposure in CYP2E1
10 mice, phenotypic anchoring becomes especially important for the interpretation of mRNA
11 expression studies (see Sections E. 1.1 and E.3.1.2 for macroarray transcript profiling limitations
12 and the need for phenotypic anchoring). However, the data for mRNA expression of PPARa,
13 peroxisomal bifunctional protein (hydratase+3-hydroxyacyl-CoA dehydrogenase),very long
14 chain acyl-CoA dehydrogenase (VLCAD), CYP4A10, NFKB (p65, P50, P52), and iKBa was
15 reported at the means ± SD for 6 mice per group and represented total liver homogenates. A
16 strength of the study was that they did not pool their RNA and can show means and standard
17 deviations between treatment groups. The low numbers of animals tested however, limits the
18 ability to detect statistically significance of the response. By reporting the means, differences in
19 the responses within dose groups was limited and reflected differential response and involvement
20 for different portions of the liver lobule and for the responses of the heterogeneous group of liver
21 cells populating the liver. The authors reported that they normalized values to the level of
22 p-actin mRNA in same preparation with a value of 1 assigned as the mean from each control
23 group. The values for mRNA and protein expression reported in the figures appeared to have all
24 been normalized to the control values for the CYP2E1 -/- mice. Although all of the CYP2E1 -/-
25 control values were reported as a value of 1, the control values for the CYP2E1+/+ mice differed
26 with the greatest difference being presented for the CYP4A10-mRNA (i.e., the control level of
27 CYP4A10 mRNA was ~3-fold higher in the CYP2E1+/+ mice than the CYP2E1 -/- mice).
28 Further characterization of the CYP2E1 mouse model was not provided by the authors.
29 The mean expression of PPARa mRNA was reported slightly reduced after TCE
30 treatment in CYP2E1 -/- mice (i.e., 0.72- and 0.78-fold of control after 1,000 and 2,000 ppm
31 TCE exposure, respectively). The CYP2E1 -/- mice had a higher baseline of PPARa mRNA
32 expression than the CYP2E1+/+ mice (i.e., the control level of the CYP2E1 -/- mice was 1.5-fold
33 of the CYP2E1+/+ mice). After TCE exposure, the CYP2E1 +/+ had a similar increase in
34 PPARa mRNA (-2.3-fold) at both 1,000 ppm and 2,000 ppm TCE. Thus, without the presence
35 of CYP2E1 there did not appear to be increased PPARa mRNA expression. For PPARa protein
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-52 DRAFT—DO NOT CITE OR QUOTE
-------
1 expression, there was a similar pattern with ~1.6-fold of control levels of protein in the
2 CYP2E1 -/- mice after both 1,000 ppm and 2,000 ppm TCE exposures. In the CYP2E1 +/+ mice
3 the control level of PPARa protein was reported to be ~1.5-fold of the CYP2E1 -/- control level.
4 Thus, while the mRNA expression was less, the protein level was greater. After TCE treatment,
5 there was a 2.9-fold of control level of protein at 1,000 ppm TCE and a 3.1-fold of control level
6 of protein at 2,000 ppm. Thus, the magnitude of mRNA increase was similar to that of protein
7 expression for PPARa in CYP2E1 +/+ mice. The magnitude of both was 3-fold or less over
8 control after TCE exposure. This pattern was similar to that of TCA concentration formed in the
9 liver where there was very little difference between the 1,000 and 2,000 ppm exposure groups in
10 CYP2E1 +/+ mice. However, this pattern was not consistent with the liver enzyme and
11 histopathology of the liver that showed a much greater response after 2,000-ppm exposure than
12 1,000-ppm TCE. In addition, where the mean enzyme markers of liver injury and individual
13 animals displayed marked heterogeneity in response to TCE exposure, there was a much smaller
14 degree of variability in the mean mRNA expression and protein levels of PPARa.
15 For peroxisomal bifunctional protein there was a greater increase after 1,000 ppm TCE-
16 treated exposure than after 2,000 ppm TCE-treatment for both the CYP2E1 -/- and CYP2E1 +/+
17 mice (i.e., there was a 2:1 ratio of mRNA expression in the 1,000- vs. 2,000-ppm-exposed
18 groups). The CYP2E1 +/+ mice had a much greater response than the CYP2E1 -/- mice (i.e., the
19 CYP2E1 -/- mice had a 2-fold of control and the CYP2E1 +/+ mice had a 7.8-fold of control
20 level after 1,000 ppm TCE treatment). For peroxisomal bifunctional protein expression, the
21 magnitude of protein induction after TCE exposure was much greater than the magnitude of
22 increase in mRNA expression. In the CYP2E1 -/- mice 1,000 ppm TCE exposure resulted in a
23 6.9-fold of control level of protein while the 2,000 ppm TCE group had a 2.3-fold level.
24 CYP2E1 +/+ mice had a -50% higher control level than CYP2E1 mice and after TCE exposure
25 the level of peroxisomal bifunctional protein expression was 44-fold of control at 1,000 ppm
26 TCE and 40-fold of control at 2,000 ppm. Thus, CYP2E1 -/- mice were reported to have less
27 mRNA expression and peroxisomal bifunctional protein formed than CYP2E1 +/+ mice after
28 TCE exposure. However, there appeared to be more mRNA expression after 1,000 ppm than
29 2,000 ppm TCE in both groups and protein expression in the CYP2E1 -/- mice. After 2,000 ppm
30 TCE, there was similar peroxisomal bifunctional protein expression between the 1,000 ppm and
31 2,000 ppm TCE treated CYP2E1 +/+ mice. Again this pattern was more similar to that of TCA
32 detection in the urine—not that of liver injury.
33 For VLCAD the expression of mRNA was similar between control and treated
34 CYP2E1 -/- mice. For CYP2E1 +/+ mice the control level of VLCAD mRNA expression was
35 half that of the CYP2E1 -/- mice. After 1,000 ppm TCE the mRNA level was 3.7-fold of control
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-53 DRAFT—DO NOT CITE OR QUOTE
-------
1 and after 2,000 ppm TCE the mRNA level was 3.1-fold of control. For VLCAD protein
2 expression was 1.8-fold of control after 1,000 ppm and 1.6-fold of control after 2,000 ppm in
3 CYP2E1 -/- mice. The control level of VLCAD protein in CYP2E1 +/+ mice appeared to be
4 1.2-fold control CYP2E1 -/- mice. After 1,000-ppm TCE treatment the CYP2E1 -/- mice were
5 reported to have 3.8-fold of control VLCAD protein levels and after 2,000-ppm TCE treatment
6 to have 3.9-fold of control protein levels. Thus, although showing no increase in mRNA there
7 was an increase in VLCAD protein levels that was similar between the two TCE exposure
8 groups in CYP2E1 -/- mice. Both VLCAD mRNA and protein levels were greater in CYP2E1
9 +/+ mice than CYP2E1 -/- mice after TCE exposure. This was not the case for peroxisomal
10 bifunctional protein. The magnitudes of TCE-induced increases in mRNA and protein increases
11 were similar between the 1,000 and 2,000 ppm TCE exposure concentrations, a pattern more
12 similar to TCA detection in the urine but not that of liver injury.
13 Finally, for CYP4A10 mRNA expression, there was an increase in expression after TCE
14 treatment of 3-fold for 1,000 ppm and 5-fold after 2,000 ppm in CYP2E1 -/- mice. Thus,
15 although the enzyme assumed to be primarily responsible for TCE metabolism to TCA was
16 missing, there was still a response for the mRNA of this enzyme commonly associated with
17 PPARa activation. Of note is that urinary concentrations of TCA were not zero after TCE
18 exposure in CYP2E1 -/- mice. Both 1,000 and 2,000 ppm TCE exposure resulted in -0.44 mg
19 TCA after 1 day or about 15-22% of that observed in CYP2E1 +/+ mice. Thus, some
20 metabolism of TCE to TCA is taking place in the null mice, albeit at a reduced rate. For
21 CYP2E1 +/+ mice, 1,000 ppm TCE resulted in an 8.3-fold of control level of CYP4A10 mRNA
22 and 2,000 ppm TCE resulted in a 9.3-fold of control level. The authors did not perform an
23 analysis of CYP4A10 protein. The authors state that "in particular, the mRNA levels of
24 microsomal enzyme CYP4A10 significantly increased in CYP2E1+/+ mice after TCE exposure
25 in a dose-dependent manner." However, the 2-fold difference in TCE exposure concentrations
26 did not result in a similar difference in response as shown above. Both resulted in ~9-fold of
27 control response in CYP2E1 +/+ mice. As with PPARa, peroxisomal bifunctional protein, and
28 VLCAD, the response was more similar to that of TCA detection in the urine and not measured
29 of hepatic toxicity. These data are CYP2E1 metabolism of TCE is important in the manifestation
30 of TCE liver toxicity, however, it also suggests that effects other than TCA concentration and
31 indicators of PPARa are responsible for acute hepatotoxicity resulting from very high
32 concentrations of TCE.
33 The NFKB family and IicBa were also examined for mRNA and protein expression.
34 These cell signaling molecules are involved in inflammation and carcinogenesis and are
35 discussed in Section E.3.3.3.3 and E.3.4.1.4. Given that presence of hepatocellular necrosis in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-54 DRAFT—DO NOT CITE OR QUOTE
-------
1 some of the CYP2E1 +/+ mice to varying degrees, inflammatory cytokines and cell signaling
2 pathways would be expected to be activated. The authors reported that
3
4 overall, TCE exposure did not significantly increase the expression of p65 and
5 p50 mRNAs in either CYP2E1+/+ or CYP2E1 -/- mice... However, p52 mRNA
6 expression significantly increased in the 2,000 ppm group of CYP2E1+/+ mice,
7 and correlation analysis showed that a significant positive relationship existed
8 between the expression of NFicB p52 mRNA and plasma ALT activity.., while no
9 correlation was seen between NFicB p64 or p50 and ALT activity (data not
10 shown).
11
12 The authors also note that TCE treatments "did not increase the expression of TNFR1 and
13 TNFR2 mRNA in CYP2E1+/+ and CYP2E1 -/- mice (data not shown)."
14 A more detailed examination of the data reveals that there was a similar increases in p65,
15 p50, and p52 mRNA expression increases with TCE treatment in CYP2E1 +/+ mice at both TCE
16 exposure levels. However, only p52 levels for the 2,000 ppm-exposed mice were reported to be
17 statistically significant (see comment above about the statistical power of the experimental
18 design and variability between animals). For 1,000 ppm TCE exposure the levels of p65, p50,
19 and p52 mRNA expression were 1.5-, 1.8-, and 2.0-fold of control. For 2,000 ppm TCE the
20 levels of p65, p50, and p52 mRNA expression were 1.8-, 1.8-, and 2.1-fold of control. Thus,
21 there was generally a similar response in all of these indicators of NFicB mRNA expression in
22 CYP2E1 +/+ mice that was mild with little to no difference between the 1,000 ppm and
23 2,000 ppm TCE exposure levels. For iKBa mRNA expression there was not difference between
24 control and treatment groups for either type of mice. For CYP2E1 -/- mice there appeared to be
25 a -50% decrease in P52 mRNA expression in mice treated with both exposure concentrations of
26 TCE. The authors plotted the relationship between p52 mRNA and plasma ALT concentration
27 for both CYP2E1 -/- and CYP2E1 +/+ mice together and claimed the correlation coefficient
28 (r = 0.5075) was significant. However, of note is that none of the CYP2E1 -/- mice were
29 reported to have either hepatic necrosis or significant increases in ALT detection.
30 For protein expression, the authors showed results for p50 and p42 proteins. The control
31 CYP2E1 -/- mice appeared to have a slightly lower level of p50 protein expression (-30%) with
32 a much larger increase in p52 protein expression (i.e., 2.1-fold) than CYP2E1 +/+ mice. There
33 appeared to be a 2-fold increase in p50 protein expression after both 1,000-ppm and 2,000 ppm
34 TCE exposures in the CYP2E1 +/+ mice and a similar increase in p52 protein levels (i.e., 1.9-
35 and 2.5-fold of control for 1,000- and 2,000-ppm TCE exposures, respectively). Thus, the
36 magnitude of mRNA and protein levels were similar for p50 and p52 in CYP2E1 +/+ mice and
37 there was no difference between the 1,000- and 2,000-ppm treatments. For the CYP2E1 -/- mice
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-55 DRAFT—DO NOT CITE OR QUOTE
-------
1 there was a modest increase in p50 protein after TCE exposure (1.1- and 1.3-fold of control for
2 1,000 and 2,000 ppm respectively) and a slight decrease in p52 protein (0.76- and 0.79-fold of
3 control). There was little evidence that the patterns of either expression or protein production of
4 NFKB family and IicBa corresponded to the markers of hepatic toxicity or that they exhibited a
5 dose-response. The authors note that although he expression of p50 protein increased in
6 CYP2E1 +/+ mice, "the relationship between p50 protein and ALT levels was not significant
7 (data not shown)." For TNFR1 there appeared to be less protein expression in the CYP2E1 +/+
8 mice than the CYP2E1 -/- mice (i.e., the null mice levels were 1.8-fold of the wild-type mice
9 levels). Treatment with TCE resulted in mild decrease of protein levels in the CYP2E1 -/- mice
10 and a 1.4- and 1.7-fold of control level in the CYP2E1 +/+ mice for 1,000 ppm and 2,000 ppm
11 levels, respectively. For p65, although TCE treatment-related effects were reported, of note the
12 levels of protein were 2.4 higher in the CYP2E1 +/+ mice than the CYP2E1 -/- mice. Thus,
13 protein levels of the NFicB family appeared to have been altered in the knockout mice. Also, as
14 noted in Section E.3.4.1.4, the origin of the NF-KB is crucial as to its effect in the liver and the
15 results of this report are for whole liver homogenates that contain parenchymal as well as
16 nonparenchymal cell and have been drawn from liver that are heterogeneous in the magnitude of
17 hepatic necrosis. The authors suggest that "TCA may act as a defense against hepatotoxicity
18 cause by TCE-delivered reactive metabolite(s) via PPARa in CYP2E1+/+ mice." However, the
19 data from this do not support such an assertion.
20
21 E.2.2. Subchronic and Chronic Studies of Trichloroethylene (TCE)
22 For the purposes of this discussion, studies of duration of 4 weeks or more are considered
23 subchronic. Like those of shorter duration, there is variation in the depth of study of liver
24 changes induced by TCE with many of the longer duration studies focused on the induction of
25 liver cancer. Many subchronic studies were conducted a high doses of TCE that caused toxicity
26 with limited reporting of effects. Similar to acute studies some of the subchronic and chronic
27 studies have detailed examinations of the TCE-induced liver effects while others have reported
28 primarily liver weight changes as a marker of TCE-response. Similar issues also arise with the
29 impact of differences in initial and final body weights between control and treatment groups on
30 the interpretation of liver weight gain as a measure of TCE-response. For many of the
31 subchronic inhalation studies, issues associated with whole body exposures make determination
32 of dose levels difficult. For gavage experiments, death from gavage dosing, especially at higher
33 TCE exposures, is a recurring problem and, unlike inhalation exposures, the effects of vehicle
34 can also be at issue for background liver effects. Concerns regarding effects of oil vehicles,
35 especially corn oil, have been raised with Kim et al. (1990) noting that a large oil bolus will not
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-56 DRAFT—DO NOT CITE OR QUOTE
-------
1 only produce physiological effects, but alter the absorption, target organ dose, and toxicity of
2 volatile organic compounds (VOCs). Charbonneau et al. (1991) reported that corn oil potentiates
3 liver toxicity from acetone administration that is not related to differences in acetone
4 concentration. Several oral studies in particular document that use of corn oil gavage induces a
5 different pattern of toxicity, especially in male rodents (see Merrick et al., 1989, Section E.2.2.1
6 below). Several studies listed below report the effects of hepatocellular DNA synthesis and
7 indices of lipid peroxidation (i.e., Channel et al., 1998) are especially subject to background
8 vehicle effects. Rusyn et al. (1999) report that a single dose of dietary corn oil increases
9 hepatocyte DNA synthesis 24 hours after treatment by ~3.5-fold, activation of NF-KB to a
10 similar extent ~2 hours after treatment almost exclusively in Kupffer cells, a ~3-4-fold increase
11 in hepatocytes after 8 hours, and increased in TNFa mRNA between 8 and 24 hours after a
12 single dose in female rats. In regard to studies that have used the i.p. route of administration, as
13 noted by Kawamoto et al. (1988) (see Section E.2.2.10 below), injection of TCE may result in
14 paralytic ileus and peritonitis and that subcutaneous treatment paradigm will result in TCE not
15 immediately being metabolized but retained in the fatty tissue. Wang and Stacey (1990) state
16 that "intraperitoneal injection is not particularly relevant to humans" and that intestinal
17 interactions require consideration in responses such as increase serum bile acid (see Section
18 E.2.3.5below).
19
20 E.2.2.1. Merrick et al., 1989
21 The focus of this study was the examination of potential differences in toxicity or orally
22 gavaged TCE administered in corn oil an aqueous vehicle in B6C3F1 mice. As reported by
23 Melnick et al. (1987) above, corn oil administration appeared to have an effect on peroxisomal
24 enzyme induction. TCE (99.5% purity) was administered in corn oil or an aqueous solution of
25 20% Emulphor to 14-17 week old mice (n = 12/group) at 0, 600, 1,200 and 2,400 mg/kg/d
26 (males) and 0, 450, 900, and 1,800 mg/kg/d (females) 5 times a week for 4 weeks. The authors
27 state that due to "varying lethality in the study, 10 animals per dose group were randomly
28 selected (where possible) among survivors for histological analysis." Hepatocellular lesions
29 were characterized
30
31 as a collection of approximately 3-5 necrotic hepatocytes surrounded by
32 macrophages and polymorphonuclear cells and histopathological grading was
33 reported as based on the number of necrotic lesions observed in the tissue
34 sections: 0 = normal; 1 = isolated lesions scattered throughout the section; 2 = one
35 to five scattered clusters of necrotic lesions; 3 = more than five scattered clusters
36 of necrotic lesions; and 4 = clusters of necrotic lesions observed throughout the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-57 DRAFT—DO NOT CITE OR QUOTE
-------
1 entire section." The authors described lipid scoring of each histological section as
2 "0 = no Oil-Red O staining present; 1 = less than 10% staining; 2 = 10-25%
3 staining; 3 = 25-30% staining; and 4 = greater than 50% staining.
4
5 The authors reported dose-related increases in lethality in both males and females
6 exposed to TCE in Emulphor with all male animals dying at 2,400 mg/kg/d with 8/12 females
7 dying at 1,800 mg/kg/d. In both males and females, 2/12 animals also died at the next highest
8 dose as well with no unscheduled deaths in control or lowest dose animals. For corn oil gavaged
9 mice, there were 1-2 animals in each TCE treatment groups of male mice that died while there
10 were no unscheduled deaths in female mice. The authors state that lethality occurred within the
11 first week after chemical exposure. The authors present data for final body weight and
12 liver/body weight values for 4 weeks of exposure and list the number of animals per group to be
13 10-12 for corn oil gavaged animals and the reduced number of animals in the Emulphor gavaged
14 animals reflective of lethality and limiting the usefulness of this measure at the highest doses
15 (i.e., 1,800 mg/kg/d for female mice). In mice treated with TCE in Emulphor gavage, the final
16 body weight of control male animals appeared to be lower than those that were treated with TCE
17 while for female mice the final body weights were similar between treated and control groups.
18 For male mice treated with Emulphor, body weights were 22.8 ± 0.8, 25.3 ± 0.5, and 24.3 ± 0.4 g
19 for control, 600 mg/kg/d, and 1,200 mg/kg/d and for female mice body weights were 20.7 ± 0.4,
20 21.4 ± 0.3, and 20.5 ± 0.3 g for control, 450 mg/kg/d, and 900 mg/kg/d of TCE.
21 For percent liver/body weight ratios, male mice were reported to have 5.6% ± 0.2%,
22 6.6% ± 0.1%, and 7.2% ± 0.2% for control, 600, and 1,200 mg/kg/d and for female mice were
23 5.1% ± 0.1%, 5.8% ± 0.1%, and 6.5% ± 0.2% for control, 450 mg/kg/d, and 900 mg/kg/d of
24 TCE. These values represent 1.11- and 1.07-fold of control for final body weight in males
25 exposed to 600 and 1,200 mg/kg/d and 1.18- and 1.29-fold of control for percent liver/body
26 weight, respectively. For females, they represent 1.04- and 0.99-fold of control for final body
27 weights in female exposed to 450mg/kg/d and 900 mg/kg/d and 1.14- and 1.27-fold of control
28 for percent liver/body weight, respectively.
29 In mice treated with corn oil gavage the final body weight of control male mice was
30 similar to the TCE treatment groups and higher than the control value for male mice given
31 Emulphor vehicle (i.e., 22.8 ± 0.8 g for Emulphor control vs. 24.3 ± 0.6 g for corn oil gavage
32 controls or a difference of-7%). The final body weights of female mice were reported to be
33 similar between the vehicles and TCE treatment groups. The baseline percent liver/body weight
34 was also lower for the corn oil gavage control male mice (i.e., 5.6% for Emulphor vs. 4.7% for
35 corn oil gavage or a difference of-19% that was statistically significant). Although the final
36 body weights were similar in the female control groups, the percent liver/body weight was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-58 DRAFT—DO NOT CITE OR QUOTE
-------
1 greater in the Emulphor vehicle group (5.1% ± 0.1% in Emulphor vehicle group vs. 4.7% ± 0.1%
2 for corn oil gavage or a difference of-9% that was statistically significant). For male mice
3 treated with corn oil, final body weights were 24.3 ± 0.6, 24.3 ± 0.4, 25.2 ± 0.6, and 25.4 ± 0.5 g
4 for control, 600, 1,200, and 2,400 mg/kg/d, and for female mice body weights were 20.2 ± 0.3,
5 20.8 ± 0.5, 21.8 ± 0.3 g, and 22.6 ± 0.3 g for control, 450, 900, and 1,800 mg/kg/d of TCE. For
6 percent liver/body weight ratios, male mice were reported to have 4.7% ± 0.1%, 6.4% ± 0.1%,
7 7.7% ± 0.1%, and 8.5% ± 0.2% for control, 600, 1,200, and 2,400 mg/kg/d and for female mice
8 were 4.7% ± 0.1%, 5.5% ± 0.1%, 6.0% ± 0.2%, and 7.2% ± 0.1% for control, 450, 900, and
9 1,800 mg/kg/d of TCE. These values represent 1.0-, 1.04-, and 1.04-fold of control for final
10 body weight in males exposed to 600, 1,200, and 2,400 mg/kg/d TCE and 1.36-, 1.64-, and
11 1.81 -fold of control for percent liver/body weight, respectively. For females, they represent
12 1.03-, 1.08-, and 1.12-fold of control for body weight in female exposed to 450, 900, and 1,800
13 mg/kg/d and 1.17-, 1.28-, and 1.53-fold of control for percent liver/body weight, respectively.
14 Because of premature mortality, the difference in TCE treatment between the highest
15 doses that are vehicle-related cannot be determined. The decreased final body weight and
16 increased percent liver/body weight ratios in the Emulphor control animals make comparisons of
17 the exact magnitude of change in these parameters due to TCE exposure difficult to determine as
18 well as differences between the vehicles. The authors did not present data for age-matched
19 controls which did not receive vehicle so that the effects of the vehicles cannot be determined
20 (i.e., which vehicle control values were most similar to untreated controls given that there was a
21 difference between the vehicle controls). A comparison of the percent liver/body weight ratios at
22 comparable doses between the two vehicles shows little difference in TCE-induced liver weight
23 increases in female mice. However, the corn oil vehicle group was reported to have a greater
24 increase in comparison to controls for male mice treated with TCE at the two lower dosage
25 groups. Given that the control values were approximately 19% higher for the Emulphor group,
26 the apparent differences in TCE-dose response may have reflected the differences in the control
27 values rather than TCE exposure. Because controls without vehicle were not examined, it cannot
28 be determined whether the difference in control values was due to vehicle administration or
29 whether a smaller or younger group of animals was studied on one of the control groups. The
30 body weight of the animals was also not reported by the authors at the beginning of the study so
31 that the impact of initial differences between groups versus treatment cannot be accurately
32 determined.
33 Serum enzyme activities for ALT, AST and LDH (markers of liver toxicity) showed that
34 there was no difference between vehicle groups at comparable TCE exposure levels for male or
35 female mice. Enzyme levels appeared to be elevated in male mice at the higher doses (i.e., 1,200
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-59 DRAFT—DO NOT CITE OR QUOTE
-------
1 and 2,400 mg/kg/d for ALT and 2,400 mg/kg/d for AST) with corn oil gavage inducing similar
2 increases in LDH levels at 600, 1,200, and 2,400 mg/kg/d TCE. For ALT and AST there
3 appeared to be a dose-related increase in male mice with the 2,400 mg/kg treatment group having
4 much greater levels than the 1,200 mg/kg group. In Emulphor treatment groups there was a
5 similar increase in ALT levels in males treated with 1,200 mg/kg TCE as with those treated with
6 corn oil and those increases were significantly elevated over control levels. For LDH levels
7 there were similar increase at 1,200 mg/kg TCE for male mice treated using either Emulphor or
8 corn oil. The authors report that visible necrosis was observed in 30-40% of male mice
9 administered TCE in corn oil but not that there did not appear to be a dose-response (i.e., the
10 score for severity of necrosis was reported to be 0, 4, 3, and 4 for corn oil control, 600, 1,200,
11 and 2,400 mg/kg/d treatment groups from 10 male mice in each group). No information in
12 regard to variation between animals was given by the authors. For male mice given Emulphor
13 gavage the extent of necrosis was reported to be 0, 0, and 1 for 0, 600, and 1,200 mg/kg/d TCE
14 exposure, respectively. For female mice, the extent of necrosis was reported to be 0 for all
15 control and TCE treatment groups using either vehicle. Thus, except for LDH levels in male
16 mice exposed to TCE in corn oil there was not a correlation with the extent of necrosis and the
17 increases in ALT and AST enzyme levels. Similarly, there was an increase in ALT levels in
18 male mice treated with 1,200 mg/kg/d exposure to TCE in Emulphor that did not correspond to
19 increased necrosis. For Oil-Red O staining there was a score of 2 in the Emulphor treated
20 control male and female mice while 600 mg/kg/d TCE exposure in Emulphor gavaged male mice
21 and 900 mg/kg/d TCE in corn oil gavaged female mice had a score of 0, along with the corn oil
22 gavage controls in male mice. For female control mice treated with corn oil gavage, the staining
23 was reported to have a score of 3. Thus, there did not appear to be a dose-response in Oil-Red
24 oil staining although the authors claimed there appeared to be a dose-related increase with TCE
25 exposure. The authors described lesions produced by TCE exposure as
26
27 focal and were surrounded by normal parenchymal tissue. Necrotic areas were
28 not localized in any particular regions of the lobule. Lesions consisted of central
29 necrotic cells encompassed by hepatocytes with dark eosinophilic staining
30 cytoplasm, which progressed to normal-appearing cells. Areas of necrosis were
31 accompanied by localized inflammation consisting of macrophages and
32 polymorphonuclear cells.
33
34 No specific descriptions of histopathology of mice given Emulphor were provided in terms of
35 effects of the vehicle or TCE treatment. The scores for necrosis was reported to be only a 1 for
36 the 1,200 mg/kg concentration of TCE in male mice gavaged with Emulphor but 3 for male mice
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-60 DRAFT—DO NOT CITE OR QUOTE
-------
1 given the same concentration of TCE in corn oil. However, enzyme levels of ALT, AST, and
2 LDH were similarly elevated in both treatment groups.
3 These results do indicate that administration of TCE for 4 weeks via gavage using
4 Emulphor resulted in mortality of all of the male mice and most of the female mice at a dose in
5 corn oil that resulted in few deaths. Not only was there a difference in mortality, but vehicle also
6 affected the extent of necrosis and enzyme release in the liver (i.e., Emulphor vehicle caused
7 mortality as the highest dose of TCE in male and female mice that was not apparent from corn
8 oil gavage, but Emulphor and TCE exposure induced little if any focal necrosis in males at
9 concentrations of TCE in corn oil gavage that caused significant focal necrosis). In regard to
10 liver weight and body weight changes, TCE exposure in both vehicles at nonlethal doses induced
11 increased percent liver/body weight changes male and female mice that increased with TCE
12 exposure level. The difference in baseline control levels between the two vehicle groups
13 (especially in males) make a determination of the quantitative difference vehicle had on liver
14 weight gain problematic although the extent of liver weight increase appeared to be similar
15 between male and female mice given TCE via Emulphor and female mice given TCE via corn
16 oil. In general, enzymatic markers of liver toxicity and results for focal hepatocellular necrosis
17 were not consistent and did not reflect dose-responses in liver weight increases. The extent of
18 necrosis did not correlate with liver weight increases and was not elevated by TCE treatment in
19 female mice treated with TCE in either vehicle, or in male mice treated with Emulphor. There
20 was a reported difference in the extent of necrosis in male mice given TCE via corn oil and
21 female mice given TCE via corn oil but the necrosis did not appear to have a dose-response in
22 male mice. Female mice given corn oil and male and female mice given TCE in Emulphor had
23 no to negligible necrosis although they had increased liver weight from TCE exposure.
24
25 E.2.2.2. Gael et al, 1992
26 The focus of this study was the description of TCE exposure related changes in mice after
27 28 days of exposure with regard to TCE-induced pathological and liver weight change. Male
28 Swiss mice (20-22 g body weight or 9% difference) were exposed to 0, 500, 1,000 or 2,000
29 mg/kg/d TCE (BDH analytical grade) by gavage in groundnut oil (n = 6 per group) 5 days a
30 week for 28 days. The ages of the mice were not given by the authors. Livers were examined
31 for "free -SH contents," total proteins, catalase activity, acid phosphatase activity, and "protein
32 specific for peroxisomal origin of approx, 80 kd." The authors report no statistically significant
33 change in body weight with TCE treatment but a significant increase in liver weight. Body
34 weight (mean ± SE) was reported to be 32.67 ± 1.54, 31.67 ±0.61, 33.00± 1.48, and
35 27.80 ± 1.65 g from exposure to oil control, 500, 1,000, and 2,000 mg/kg/d TCE, respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-61 DRAFT—DO NOT CITE OR QUOTE
-------
1 There was a 15% decrease in body weight at the highest exposure concentration of TCE that was
2 not statistically significant, but the low number of animals examined limits the power to detect a
3 significant change. The percent relative liver/body weight was reported to be 5.29% ± 0.48%,
4 7.00% ± 0.36%, 7.40% ± 0.39%, and 7.30% ± 0.48% from exposure to oil control, 500, 1,000,
5 and 2,000 mg/kg/d TCE, respectively. This represents 1.32-, 1.41-, and 1.38-fold of control in
6 percent liver/body weight for 500, 1,000, and 2,000 mg/kg/d TCE, respectively. The "free -SH
7 content" in umol ~SH/g tissue was reported to be 5.47 ± 0.17, 7.46 ± 0.21, 7.84 ± 0.34, and
8 7.10 ± 0.34 from exposure to oil control, 500, 1,000, and 2,000 mg/kg/d TCE, respectively. This
9 represents 1.37-, 1.44-, and 1.30-fold of control in -SH/g tissue weight for 500, 1,000, and
10 2,000 mg/kg/d TCE, respectively. Total protein content in the liver in mg/g tissue was reported
11 to be 170 ± 3, 183 ± 5, 192 ± 7, and 188 ± 3 from exposure to oil control, 500, 1,000, and
12 2,000 mg/kg/d TCE, respectively. This represents 1.08-, 1.13-, and 1.11-fold of control in total
13 protein content for 500, 1,000, and 2,000 mg/kg/d TCE, respectively. Thus, the increases in liver
14 weight, "free -SH content" and increase protein content were generally parallel and all suggest
15 that liver weight increases had reached a plateau at the 1,000 mg/kg/d exposure concentration
16 perhaps reflecting toxicity at the highest dose as demonstrated by decreased body weight in this
17 study.
18 The enzyme activities of 5-ALA dehydrogenase ("a key enzyme in heme biosynthesis"),
19 catalase, and acid phosphatase were assayed in liver homogenates. Treatment with TCE
20 decreased 5-ALA dehydrogenase activity to a similar extent at all exposure levels (32-35%
21 reduction). For catalase the activity as units of catalase/mg protein was reported to be
22 25.01 ± 1.81, 32.46 ± 2.59, 41.11 ± 5.37, and 33.96 ± 3.00 from exposure to oil control, 500,
23 1,000, and 2,000 mg/kg/d TCE, respectively. This represents 1.30-, 1.64-, and 1.36-fold in
24 catalase activity for 500, 1,000, and 2,000 mg/kg/d TCE, respectively. The increasing variability
25 in response with TCE exposure concentration is readily apparent from these data as is the
26 decrease at the highest dose, perhaps reflective of toxicity. For acid phosphatase activity in the
27 liver there was a slight increase (5-11%) with TCE exposure that did not appear to be dose-
28 related.
29 The authors report that histologically, "the liver exhibits swelling, vacuolization,
30 widespread degeneration/necrosis of hepatocytes as well as marked proliferation of endothelial
31 cells of hepatic sinusoids at 1000 and 2000 mg/kg TCE doses." Only one figure is given at the
32 light microscopic level in which it is impossible to distinguish endothelial cells from Kupffer
33 cells and no quantitative measures or proliferation were examined or reported to support the
34 conclusion that endothelial cells are proliferating in response to TCE treatment. Similarly, no
35 quantitation regarding the extent or location of hepatocellular necrosis is given. The presence or
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-62 DRAFT—DO NOT CITE OR QUOTE
-------
1 absence of inflammatory cells is not noted by the authors as well. In terms of white blood cell
2 count, the authors note that it is slightly increased at 500 mg/kg/d but decreased at 1,000 and
3 2,000 mg/kg/d TCE, perhaps indicating macrophage recruitment from blood to liver and kidney,
4 which was also noted to have pathology at these concentrations of TCE.
5
6 E.2.2.3. Kjelhtmndetal.,1981
1 This study was conducted in mice, rats, and gerbils and focused on the effects of
8 150-ppm TCE exposure via inhalation on body and organ weight. No other endpoints other than
9 organ weights were examined in this study and the design of the study is such that quantitative
10 determinations of the magnitude of TCE response are very limited. NMRI mice (weighing -30 g
11 with age not given), S-D rats (weighing -200 g with age not given, and Mongolian gerbils
12 (weighing -60 g with age not given) were exposed to 150-ppm TCE continuously. Mice were
13 exposed for 2, 5, 9, 16, and 30 days with the number of exposed animals and controls in the 2, 5,
14 9, and 16 days groups being 10. For 30-day treatments there were two groups of mice containing
15 20 mice per group and one group containing 12 mice per group. In addition there was a group of
16 mice (n= 15) exposed to TCE for 30 days and then examined 5 days after cessation of exposure
17 and another group (n = 20) exposed to TCE for 30 days and then examined 30 days after
18 cessation of exposure. For rats there were three groups exposed to TCE for 30 days, which
19 contained 24, 12, and 10 animals per group. For gerbils there were three groups exposed to TCE
20 for 30 days, which contained 24, 8, and 8 animals per group. The groups were reported to
21 consist of equal numbers of males and female but for the mice exposed to TCE for 30 days and
22 then examined 5 days later, the number was 10 males and 5 females. Body weights were
23 reported to be recorded before and after the exposure period. However, the authors state "for
24 technical reasons the animals within a group were not individually identified, i.e., we did not
25 know which initial weight in the group corresponded to which final one." They authors state that
26 this design presented problems in assessing the precision of the estimate. They go on to state
27 that rats and gerbils were partially identifiable as the animals were housed 3 to a cage and cage
28 averages could be estimated. Not only were mice in one group housed together but
29
30 even worse: at the start of the experiment, the mice in M2 [group exposed for 2
31 days] and M9 [group exposed for 9 days] were housed together, and similarly M5
32 [group exposed for 5 days] and Ml6 [group exposed for 16 days]. Thus, we had,
33 e.g., 10 initial weights for exposed female mice in M2 and M9 where we could
34 not identify those 5 that were M2 weights. Owing to this bad design (forced upon
35 us by the lack of exposure units), we could not study weight gains for mice and so
36 we had to make do with an analysis of final weights.
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-63 DRAFT—DO NOT CITE OR QUOTE
-------
1 The problems with the design of this study are obvious from the description given by the authors
2 themselves. The authors state that they assumed that the larger the animal the larger the weight
3 of its organs so that all organ weights were converted into relative weights as percentage of body
4 weight. The fallacy of this assumption is obvious, especially if there was toxicity that decreased
5 body weight and body fat but at the same time caused increased liver weight as has been
6 observed in many studies at higher doses of TCE. In fact, Kjellstrand et al. (1983b) report that a
7 150-ppm TCE exposure for 30 days does significantly decreases body weight while elevating
8 liver weight in a group of 10 male NMRI mice. Thus, the body weight estimates from this study
9 are inappropriate for comparison to those in studies where body weights were actually measured.
10 The liver/body weight ratios that would be derived from such estimates of body weights would
11 be meaningless. The group averages for body weight reported for female mice at the beginning
12 of the 30-day exposure varied significantly and ranged from 23.2 to 30.2 g (-24%). For males,
13 the group averages ranged from 27.3 to 31.4 g (-14%). For male mice there was no weight
14 estimate for the animals that were exposed for 30 days and then examined 30 days after cessation
15 of exposure.
16 The authors only report relative organ weight at the end of the experiment rather than the
17 liver weights for individual animals. Thus, these values represent extrapolations based on to
18 what body weight may have been. For mice that were exposed to TCE for 30 days and the
19 examined after 30 days of exposure, male mice were reported to have "relative organ weight" for
20 liver of 4.70% ± 0.10% versus 4.27% ± 0.13% for controls. However, there were no initial body
21 weights reported for these male mice and the body weights are extrapolated values. Female mice
22 exposed for 30 days and then examined 30 days after cessation of exposure were reported to
23 have "relative organ weights" for liver of 4.42% ± 0.11% versus 3.62% ± 0.09%. The group
24 average of initial body weights for this group was reported by the authors. Although the initial
25 body weight for female control mice as a group average was reported to be similar between the
26 female group exposed to 30 days of TCE and sacrificed 30 days later and those exposed for
27 30 days and sacrificed 5 days later (30.0 g vs. 30.8 g), the liver/body weight ratio varied
28 significantly in these controls (4.25 ±0.19 vs. 3.62 ± 0.09) as did the number of animals studied
29 (5 female mice in the animals sacrificed after 5 days exposure versus 10 female mice in the
30 group sacrificed after 30 days exposure). In addition, although there were differences between
31 the 3 groups of mice exposed to TCE for 30 days and then sacrificed immediately, the authors
32 present the data for extrapolated liver/body weight as pooled results between the 3 groups. In
33 comparison to control values, the authors report 1.14-, 1.35-, 1.58-, 1.47-, and 1.75-fold of
34 control for percent liver/body weight using body weight extrapolated values in male mice at 2, 5,
35 9, 16, and 30 days of TCE exposure, respectively. For females, they report 1.27-, 1.28-, 1.49-,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-64 DRAFT—DO NOT CITE OR QUOTE
-------
1 1.41-, and 1.74-fold of control at 2, 5, 9, 16, and 30 days of TCE, respectively. Although the
2 authors combine female and male relative increases in liver weight in a figure, assign error bars
3 around these data point, and attempt to draw assign a time-response curve to it, it is clear from
4 these data, especially for female mice, do not display time-dependent increase in liver/body
5 weight from 5 to 16 days of exposure and that a comparison of results between 5 animals and 26
6 is very limited in interpretation. Of note is the wide variation in the control values for relative
7 liver/body weight. For male mice there did not seem to be a consistent pattern with increasing
8 duration of the experiment with values at 4.61, 5.15, 5.05, 4.93, and 4.04% for 2, 5, 9, 16, and
9 30-day exposure groups. This represented a difference of-27%. For female mice, the relative
10 liver/body weight was 4.14, 4.58, 4.61, 4.70, and 3.99% for 2, 5, 9, 16, and 30 day exposure
11 groups. Thus, it appears that the average relative liver/body weight percent was higher in the 5,
12 9, and 16 day treatment group for both genders than that to the 30 day group and was consistent
13 between these days. There is no apparent reason for there to be such large difference between 16
14 day and 30-day treatment groups due to increasing age of the animals. Of note is that for the
15 control groups pared with animals treated for 30 days and then examined 30 days later, the male
16 mice had increases in relative liver/body weight (4.27 vs. 4.04%) but that the females had a
17 decrease (3.62 vs. 3.99%). Such variation between controls does not appear to be age and size
18 related but to variations in measure or extrapolations, which can affect comparisons between
19 treated and untreated groups and add more uncertainty to the estimates.
20 The number of mice in the groups exposed to 2 though 16 days were only 5 animals for
21 each gender in each group while the number of animals reported in the 30-day exposure group
22 numbered 26 for each gender.
23 For animals exposed to 30 days and then examined after 5 or 30 days, male mice were
24 reported to have percent liver/body weight 1.26- and 1.10-fold of control after 5 and 30 days
25 cessation of exposure while female mice were reported to have values of 1.14- and 1.22-fold of
26 control after 5 and 30 days cessation of exposure, respectively. Again, the male mice exposed
27 for 30 days and then examined after 30 days of cessation of exposure did not have reported
28 initial body weights giving this value a great deal of uncertainty. Thus, while liver weights
29 appeared to increase during 30 days of exposure to TCE and decreased after cessation of
30 exposure in both genders of mice, the magnitudes of the increases and decrease cannot be
31 determined from this experimental design. Of note is that liver weights appeared to still be
32 elevated after 30 days of cessation exposure.
33 In regard to initial weights, the authors report that the initial weight of the rats were
34 different in the 3 experiments they conducted with them and state that "in those 2 where
35 differences were found in females, their initial weights were about 200 g and 220 g, respectively,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-65 DRAFT—DO NOT CITE OR QUOTE
-------
1 while the corresponding weights were only about 160 g in that experiment where no differences
2 were found." The differences in initial body weight of the rat groups were significant. In
3 females group averages were 198, 158, and 224 g, for groups 1, 2, and 3, respectively, and for
4 males group averages were 222, 166, and 248 g for groups 1,2, and 3 respectively. This
5 represents as much as a 50% difference in initial body weights between these TCE treatment
6 groups. Control values varied as well with group averages for controls ranging from 167 g for
7 group 2 to 246 g for group 3 at the start of exposure. For female rats control groups ranged from
8 158 to 219 g at the start of the experiment. The number of animals in each group varied greatly
9 as well making quantitative comparison even more difficult with the numbers varying between 5
10 and 12 for each gender in rats exposed for 30 days to TCE. The authors pooled the results for
11 these very disparate groups of rats in their reporting of relative organ weights. They reported
12 1.26- and 1.21-fold of control in male and female rat percent relative liver/body weight after
13 30 days of TCE exposure. However, as stated above, these estimates are limited in their ability
14 to provide a quantitative estimate of liver weight increase due to TCE.
15 There were evidently differences between the groups of gerbils in response to TCE with
16 one group reported to have larger weight gain than control and the other 2 groups reported to not
17 show a difference by the authors. Of the 3 groups of gerbils, group 1 contained 12 animals per
18 gender but groups 2 and 3 only 4 animals per gender. As with the rat experiments, the initial
19 average weights for the groups varied significantly (30% in females and males). The authors
20 pooled the results for these very disparate groups of gerbils in their reporting of relative organ
21 weights as well. They reported a nearly identical increase in relative liver/body weight increase
22 for gerbil (1.22-fold of control value in males and 1.25-fold in females) as for the rat after
23 30 days of TCE exposure. However, similar caveats should be applied in the confidence in this
24 experimental design to determine the magnitudes of response to TCE exposure.
25
26 E.2.2.4. Woolhiser et al, 2006
27 An unpublished report by Woolhiser et al. (2006) was received by the U.S. EPA to fill
28 the "priority data needed" for the immunotoxicity of TCE as identified by the Agency for Toxic
29 Substances and Disease Registry and designed to satisfy U.S. EPA OPPTS 870.7800
30 Immunotoxicity Test Guidelines. The study was conducted on behalf of the Halogenated
31 Solvents Industry Alliance and has been submitted to the U. S. EPA but not published. Although
32 conducted as an immunotoxicity study, it does contain information regarding liver weight
33 increases in female Sprague Dawley (S-D) female rats exposed to 0, 100, 300, and 1,000 ppm
34 TCE for 6 hours/day, 5 days/week for 4 weeks. The rats were 7 weeks of age at the start of the
35 study. The report gives data for body weight and food weight for 16 animals per exposure group
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-66 DRAFT—DO NOT CITE OR QUOTE
-------
1 and the mean body weights ranged between 181.8 to 185.5 g on the first day of the experiment.
2 Animals were weight pre-exposure, twice during the first week, and then "at least weekly
3 throughout the study." All rats were immunized with a single intravenous injection of sheep red
4 blood cells via the tail vein at Day 25. Liver weights were taken and samples of liver retained
5 "should histopathological examination have been deemed necessary." But, histopathological
6 analysis was not conducted on the liver.
7 The effect on body weight gain by TCE inhalation exposure was shown by 5 days and
8 continued for 10 days of exposure in the 300-ppm and 1,000-ppm-exposed groups. By Day 28,
9 the mean body weight for the control group was reported to be 245.7 g but 234.4 g, 232.4 g, and
10 232.4 g for the 100-ppm, 300-ppm, and 1,000-ppm exposure groups, respectively. Food
11 consumption was reported to be decreased in the day 1-5 measurement period for the 300-and
12 1,000-ppm exposure groups and in the 5-10 day measurement period for the 100-ppm group.
13 Although body weight and food consumption data are available for 16 animals per exposure
14 group, for organ and organ/body weight summary data, the report gives information for only
15 8 rats per group. The report gives individual animal data in its appendix so that the data for the
16 8 animals in each group examined for organ weight changes could be examined separately. The
17 final body weights were reported to be 217.2, 212.4, 203.9, and 206.9 g for the control, 100-,
18 300-, and 1,000-ppm exposure groups containing only 8 animals. For the 8-animal exposure
19 groups, the mean initial body weights were 186.6, 183.7, 181.6, and 181.9 g for the control, 100-,
20 300-, and 1,000-ppm exposure groups. Thus, there was a difference from the initial and final
21 body weight values given for the groups containing 16 rats and those containing 8 rats. The
22 ranges of initial body weights for the eight animals were 169.8-204.3, 162.0-191.2,
23 169.0-201.5, and 168.2-193.7 g for the control, 100-, 300 -, and 1,000-ppm groups. Thus, the
24 control group began with a larger mean value and large range of values (20% difference between
25 highest and lowest weight rat) than the other groups.
26 In terms of the percent liver/body weight ratios, an increase due to TCE exposure is
27 reported in female rats, although body weights were larger in the control group and the two
28 higher exposure groups did not gain body weight to the same extent as controls. The mean
29 percent liver/body weight ratios were 3.23, 3.39, 3.44, and 3.65%, respectively for the control,
30 100-ppm, 300-ppm, and 1,000-ppm exposure groups. This represented 1.05-, 1.07-, and
31 1.13-fold of control percent liver/body weight changes in the 100-, 300-, and 1,000-ppm
32 exposure groups. However, the small number of animals and the variation in initial animal
33 weight limit the ability of this study to determine statistically significant increases and the
34 authors report that only the 1,000-ppm group had statistically significant increased liver weight
35 increases.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-67 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.2.5. Kjelhtmndetal,1983a
2 This study examined seven strains of mice (wild, C57BL, DBA, B6CBA, A/sn, NZB, and
3 NMRI) after continuous inhalation exposure to 150-ppm TCE for 30 days. "Wild" mice were
4 reported to be composed of "three different strains: 1. Hairless (HR) from the original strain, 2.
5 Swiss (outbred), and 3. Furtype Black Pelage (of unknown strain)." The authors do not state the
6 age of the animals prior to TCE exposure but state that weight-matched controls were exposed to
7 air only chambers. The authors state that "the exposure methods" have been described earlier
8 (Kjellstrand et al., 1980) but only reference Kjellstrand et al. (1981). In both of this and the 1981
9 study, animals were continuously exposed with only a few hours of cessation of exposure noted a
10 week for change of food and bedding. Under this paradigm, there is the possibility of additional
11 oral exposure to TCE due to grooming and consumption of TCE on food in the chamber. The
12 study was reported to be composed of two independent experiments with the exception of strain
13 NMRI which had been studied in Kjellstrand et al. (1981, 1983b). The number of animals
14 examined in this study ranged from 3-6 in each treatment group. The authors reported
15 "significant difference between the animals intended for TCE exposure and the matched controls
16 intended for air-exposure were seen in four cases (Table 1.)," and stated that the grouping effects
17 developed during the 7-day adaptation period. Premature mortality was attributed to an accident
18 for one TCE-exposed DBA male and fighting to the deaths of two TCE-exposed NZB females
19 and one B6CBA male in each air exposed chamber. Given the small number of animals
20 examined in this study in each group, such losses significantly decrease the power of the study to
21 detect TCE-induced changes. The range of initial body weights between the groups of male
22 mice for all strains was between 18 g (as mean value for the A/sn strain) and 32 g (as mean value
23 for the B6CBA strain) or -44%. For females, the range of initial body weights between groups
24 for all strains was 15 g (as mean value for the A/sn strain) and 24 g (as mean value for the DBA
25 strain) or-38%.
26 Rather than reporting percent liver/body weight ratios or an extrapolated value, as was
27 done in Kjellstrand et al. (1981), this study only reported actual liver weights for treated and
28 exposed groups at the end of 30 days of exposure. The authors report final body weight changes
29 in comparison to matched control groups at the end of the exposure periods but not the changes
30 in body weight for individual animals. They report the results from statistical analyses of the
31 difference in values between TCE and air-exposed groups. A statistically significant decrease in
32 body weight was reported between TCE exposed and control mice in experiment 1 of the C57BL
33 male mice (-20% reduction in body weight due to TCE exposure). This group also had a slight
34 but statistically significant difference in body weight at the beginning of exposure with the
35 control group having a -5% difference in starting weight. There was also a statistically
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-68 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant decrease in body weight of 20% reported after TCE exposure in one group of male
2 B6CBA mice that did not have a difference in body weight at the beginning of the experiment
3 between treatment and control groups. One group of female and both groups of male A/sn mice
4 had statistically significant decreases in body weight after TCE exposure (10% for the females,
5 and 22 and 26% decreases in the two male groups) in comparison to untreated mice of the same
6 strain. The magnitude of body weight decrease in this strain after TCE treatment also reflects
7 differences in initial body weight as there were also differences in initial body weight between
8 the two groups of both treated and untreated A/sn males that were statistically significant, 17 and
9 10% respectively. One group of male NZB mice had a significant increase in body weight after
10 TCE exposure of 14% compared to untreated animals. A female group from the same strain
11 treated with TCE was reported to have a nonsignificant but 7% increase in final body weight in
12 comparison to its untreated group. The one group of male NMRI mice (n = 10) in this study was
13 reported to have a statistically significant 12% decrease in body weight compared to controls.
14 For the groups of animals with reported TCE exposure-related changes in final body
15 weight compared to untreated animals, such body weight changes may also have affected the
16 liver weights changes reported. The authors do not explicitly state that they did not record liver
17 and body weights specifically for each animal, and thus, would be unable to determine liver/body
18 weight ratios for each, however, they do state that he animals were housed 4-6 in each cage and
19 placed in exposure chambers together. The authors only present data for body and liver weights
20 as the means for a cage group in the reporting of their results. While this approach lends more
21 certainty in their measurements than the approach taken by Kjellstrand et al. (1981) as described
22 above, the relative liver/body weights cannot be determined for individual animals. It appears
23 that the authors have tried to carefully match the body weights of the control and exposed mice
24 at the beginning of the experiment to minimize the effects of initial body weight differences and
25 distinguish the effects of treatment on body weight and liver weight. However, there is no ability
26 to determine liver/body weight ratios and adjust for difference in initial body weight from
27 changes due to TCE exposure. For the groups in which there was no change in body weight after
28 TCE treatment and in which there was no difference in initial body weight between controls and
29 TCE-exposed groups, the reporting of liver weight changes due to TCE exposure is a clearer
30 reflection of TCE-induced effects and the magnitude of such effects. Nevertheless the small
31 number of animals examined in each group is still a limitation on the ability to determine the
32 magnitude of such responses and there statistical significance.
33 In wild-type mice there were no reported significant differences in the initial and final
34 body weight of male or female mice before or after 30 days of TCE exposure. For these groups
35 there was 1.76- and 1.80-fold of control values for liver weight in groups 1 and 2 for female
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-69 DRAFT—DO NOT CITE OR QUOTE
-------
1 mice, and for males 1.84- and 1.62-fold of control values for groups 1 and 2, respectively. For
2 DBA mice there were no reported significant differences in the initial and final body weight of
3 male or female mice before or after 30 days of TCE exposure. For DBA mice there was 1.87-
4 and 1.88-fold of control for liver weight in groups 1 and 2 for female mice, and for males 1.45-
5 and 2.00-fold of control for groups 1 and 2, respectively. These groups represent the most
6 accurate data for TCE-induced changes in liver weight not affected by initial differences in body
7 weight or systemic effects of TCE, which resulted in decreased body weight gain. These results
8 suggest that there is more variability in TCE-induced liver weight gain between groups of male
9 than female mice.
10 The C57BL, B6CBA, NZB, and NMRI groups all had at least one group of male mice
11 with changes in body weight due to TCE exposure. The A/sn group not only had both male
12 groups with decreased body weight after TCE exposure (along with differences between exposed
13 and control groups at the initiation of exposure) but also a decrease in body weight in one of the
14 female groups. Thus, the results for TCE-induced liver weight change in these male groups also
15 reflect changes in body weight. These results suggest a strain-related increased sensitivity to
16 TCE toxicity as reflected by decreased body weight. For C57BL mice, there was 1.65- and
17 1.60-fold of control for liver weight after TCE exposure was reported in groups 1 and 2 for
18 female mice, and for males 1.28-fold (the group with decreased body weight) and 1.82-fold of
19 control values for groups 1 and 2, respectively. For B6CBA mice there was 1.70- and 1.69-fold
20 of controls values for liver weight after TCE exposure in groups 1 and 2 for female mice, and for
21 males 1.21-fold (the group with decreased body weight) and 1.47-fold of control values reported
22 for groups 1 and 2, respectively. For the NZB mice there was 2.09-fold (n = 3) and 2.08-fold of
23 control values for liver weight after TCE exposure in groups 1 and 2 for female mice and for
24 males 2.34- and 3.57-fold (the group with increased body weight) of control values reported for
25 groups 1 and 2, respectively. For the NMRI mice, whose results were reported for one group
26 with 10 mice, there was 1.66-fold of control value for liver weight after TCE exposure for female
27 mice and for males 1.68-fold of control value reported (a group with decreased body weight).
28 Finally, for the A/sn strain that had decreased body weight in all groups but one after TCE
29 exposure and significantly smaller body weights in the control groups before TCE exposure in
30 both male groups, the results still show TCE-related liver weight increases. For the As/n mice
31 there was 1.56- and 1.72-fold (a group with decreased body weight) of control value for liver
32 weight in groups 1 and 2 for female mice and for males 1.62-fold (a group with decreased body
33 weight) and 1.58-fold (a group with decreased body weight) of control values reported for
34 groups 1 and 2, respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-70 DRAFT—DO NOT CITE OR QUOTE
-------
1 The consistency between groups of female mice of the same strain for TCE-induced liver
2 weight gain, regardless of strain examined, is striking. The largest difference within female
3 strain groups occurred in the only strain in which there was a decrease in TCE-induced body
4 weight. For males, even in strains that did not show TCE-related changes in body weight, there
5 was greater variation between groups than in females. For strains in which one group had
6 TCE-related changes in body weight and another did not, the group with the body weight
7 decrease always had a lower liver weight as well. Groups that had increased body weight after
8 TCE exposure also had an increased liver weight in comparison to the groups without a body
9 weight change. These results demonstrate the importance of carefully matching control animals
10 to treated animals and the importance of the effect of systemic toxicity, as measured by body
11 weight decreases, on the determination of the magnitude of liver weight gain induced by TCE
12 exposure. These results also show the increased variation in TCE-induced liver weight gain
13 between groups of male mice and an increase incidence of body weight changes due to TCE
14 exposure in comparison to females, regardless of strain.
15 In terms of strain sensitivity, it is important not only to take into account differing effects
16 on body weight changes due to TCE exposure but also to compare animals of the same age or
17 beginning weight as these parameters may also affect liver weight gain or toxicity induced by
18 TCE exposure. The authors do not state the age of the animals at the beginning of exposure and
19 report, as stated above, a range of initial body weights between the groups as much as 44% for
20 males and 3 8% for females. These differences can be due to strain and age. The differences in
21 final body weight between the groups of controls, when all animals would have been 30 days
22 older and more mature, was still as much as 48% for males and 44% for females. The data for
23 female mice, in which body weight was decreased by TCE exposure only in on group in one
24 strain, suggest that the magnitude of TCE-induced liver weight increase was correlated with
25 body weight of the animals at the beginning of the experiment. For the C57BL and As/n strains,
26 female mice starting weights were averaged 17.5 and 15.5 g, respectively, while the average liver
27 weights were 1.63- and 1.64-fold of control after TCE exposure, respectively. For the B6CBA,
28 wild-type, DBA, and NZB female groups the starting body weights averaged 22.5, 21.0, 23.0,
29 and 21.0 g, respectively, while the average liver weight increases were 1.70-, 1.78-, 1.88-, and
30 2.09-fold of control after TCE exposure. Thus, groups of female mice with higher body weights,
31 regardless of strain, generally had higher increases in TCE-induced liver weight increases. The
32 NMRI group of female mice, did not follow this general pattern and had the highest initial body
33 weight for the single group of 10 mice reported (i.e., 27 g) associated with a 1.66-fold of control
34 value for liver weight. It is probable that the data for these mice had been collected from another
35 study. In fact, the starting weights reported for these groups of 10 mice are identical to the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-71 DRAFT—DO NOT CITE OR QUOTE
-------
1 starting weights reported for 26 mice examined in Kjellstrand et al. (1981). However, while this
2 study reports a 1.66-fold of control value for liver weight after 30 days of TCE exposure, the
3 extrapolated percent liver/body weight given in the 1981 study for 30 days of TCE exposure was
4 1.74-fold of control in female NMRI mice. In the Kjellstrand et al. (1983b) study, discussed
5 below, 10 female mice were reported to have a 1.66-fold of control value for liver weight after
6 30 days exposure to 150-ppm TCE with an initial starting weight of 26.7 g. Thus, these data
7 appear to be from that study. Thus, differences in study design, variation between experiments,
8 and strain differences may account for the differences results reported in Kjellstrand et al.
9 (1983a) for NMRI mice and the other strains in regard to the relationship to initial body weight
10 and TCE response of liver weight gain.
11 These data suggest that initial body weight is a factor in the magnitude of TCE-induced
12 liver weight induction rather than just strain. For male mice, there appeared to be a difference
13 between strains in TCE-induced body weight reduction, which in turn affects liver weight. The
14 DBA and wild-type mice appeared to be the most resistant to this effect (with no groups
15 affected), while the C57BL, B6CBA, and NZB strains appearing to have at least one group
16 affected, and the A/sn strain having both groups of males affected. Only one group of NMRI
17 mice were reported in this study and that group had TCE-induced decreases in body weight. As
18 stated above there appeared to be much greater differences between groups of males within the
19 same strain in regard to liver weight increases than for females and that the increases appeared to
20 be affected by concurrent body weight changes. In general the strains and groups within strain,
21 that had TCE-induced body weight decreases, had the smallest increases in liver weight, while
22 those with no TCE-induced changes in body weight in comparison to untreated animals (i.e.,
23 wild-type and DBA) or had an actual increase in body weight (one group of NZB mice) had the
24 greatest TCE-induced increase in liver weight. Therefore, only examining liver weight in males
25 rather than percent liver/body weight ratios would not be an accurate predictor of strain
26 sensitivity at this dose due to differences in initial body weight and TCE-induced body weight
27 changes.
28
29 E.2.2.6. Kjellstrand et al, 1983b
30 This study was conducted in male and female NMRI mice with a similar design as
31 Kjellstrand et al. (1983a). The ages of the mice were not given by the authors. Animals were
32 housed 10 animals per cage and exposed from 30 to 120 days at concentrations ranging from 37
33 to 3,600 ppm TCE. TCE was stabilized with 0.01% thymol and 0.03% diisopropylene. Animals
34 were exposed continuously with exposure chambers being opened twice a week for change of
35 bedding food and water resulting in a drop in TCE concentration of ~1 hour. A group of mice
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-72 DRAFT—DO NOT CITE OR QUOTE
-------
1 was exposed intermittently with TCE at night for 16 hours. This paradigm results not only in
2 inhalation exposure but, also, oral exposure from TCE adsorption to food and grooming
3 behavior. The authors state that "the different methodological aspects linked to statistical
4 treatment of body and organ weights have been discussed earlier (Kjellstrand et al., 1981). The
5 same air-exposed control was used in three cases." The design of the experiment, in terms of
6 measurement of individual organ and body weights and the inability to assign a percent
7 liver/body weight for each animal, and limitations are similar to that of Kjellstrand et al. (1983b).
8 The exposure design was for groups of male and female mice to be exposed to 37-, 75-, 150-,
9 and 300-ppm TCE continuously for 30 days (n = 10 per gender and group except for the 37 ppm
10 exposure groups) and then for liver weight and body weight to be determined. Additional groups
11 of animals were exposed for 150 ppm continuously for 120 days (n = 10). Intermittent exposure
12 of 4 hours/day for 7 days a week were conducted for 120 days at 900 ppm and examined
13 immediately or 30 days after cessation of exposure (n = 10). Intermittent exposures of
14 16 hours/day at 255-ppm group (n = 10), 8 hours/day at 450 ppm, 4 hours/day at 900 ppm,
15 2 hours/day at 1,800 ppm, and 1 hour/day at 3,600 ppm 7 days/week for 30 days were also
16 conducted (n = 10 per group).
17 As in Kjellstrand et al. (1983a), body weights for individual animals were not recorded in
18 a way that the initial and final body weights could be compared. The approach taken by the
19 authors was to match the control group at the initiation of exposure and compare control and
20 treated average values. At the beginning of the experiment only one group began the experiment
21 with a statistically significant change in body weight between treated and control animals
22 (female mice exposed 16 hours a day for 30 days). In regard to final body weight, which would
23 indicate systemic TCE toxicity, 5 groups had significantly decreased body weight (i.e., males
24 exposed to 150 ppm continuously for 30 or 120 days, males and females exposed continuously to
25 300 ppm for 30 days) and 2 groups significantly increased body weight (i.e., males exposed to
26 1,800 ppm for 2 hours/day and 3,600 ppm for 1 hour/day for 30 days) after TCE exposure. Thus,
27 the accuracy of determining the effect of TCE on liver weight changes, reported by the authors in
28 this study for groups in which body weight were also affected by TCE exposure, would be
29 affected by similar issues as for data presented by Kjellstand et al. (1983a). In addition,
30 comparison in results between the 37-ppm exposure groups and those of the other groups would
31 be affected by difference in number of animals examined (10 vs. 20). As with Kjellstrand et al
32 (1983a), the ages of the animals in this study are not given by the author. Difference in initial
33 body weight (which can be affected by age and strain) reported by Kjellstrand et al. (1983a)
34 appeared to be correlated with the degree of TCE-induced change in liver weight. Although each
35 exposed group was matched to a control group with a similar average weight, the average initial
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-73 DRAFT—DO NOT CITE OR QUOTE
-------
1 body weights in this study varied between groups (i.e., as much as 14% in female control, 16%
2 in TCE-exposed female mice, 12% in male control, and 16% in male exposed mice).
3 For female mice exposed from 37 ppm to 300 ppm TCE continuously for 30 days, only
4 the 300 pm group experienced a 16% decrease in body weight between control and exposed
5 animals. Thus, liver weight increased reported by this study after TCE exposure were not
6 affected by changes in body weight for exposures below 300 ppm in female mice. Initial body
7 weights in the TCE-exposed female mice were similar in each of these groups (i.e., range of
8 29.2-31.6 g, or 8%), with the exception of the females exposed to 150 ppm TCE for 30 days
9 (i.e., initial body weight of 27.3 g), reducing the effects of differences in initial body weight on
10 TCE-induced liver weight induction. Exposure to TCE continuously for 30 days resulted in a
11 dose-dependent change in liver weight in female mice with 1.06-, 1.27-, 1.66-, and 2.14-fold of
12 control values reported for liver weight at 37 ppm, 75 ppm, 150 ppm, and 300 ppm TCE,
13 respectively. In females, the increase at 300 ppm was accompanied by statistically significant
14 decreased body weight in the TCE exposed groups compared to control (-16%). Thus, the
15 response in liver weight gain at that exposure is in the presence of toxicity. However, the TCE-
16 induced increases in liver weight consistently increased with dose of TCE in a linear fashion.
17 For male mice exposed to 37 to 300 ppm TCE continuously for 30 days, both the 150-
18 and 300-ppm-exposed groups experienced a 10 and 18% decrease in body weight after TCE
19 exposure, respectively. The 37- and 75-ppm groups did not have decreased body weight due to
20 TCE exposure but varied by 12% in initial body weight. Thus, there are more factors affecting
21 reported liver weight increases from TCE exposure in the male than female mice, most
22 importantly toxicity. Exposure to TCE continuously for 30 days resulted in liver weights of
23 1.15-, 1.50-, 1.69-, and 1.90-fold of control for 37, 75, 150, and 300 ppm, respectively. The
24 flattening of the dose-response curve for liver weight in the male mice is consistent with the
25 effects of toxicity at the two highest doses, and thus, the magnitude of response at these doses
26 should be viewed with caution. Consistent with Kjellstrand et al. (1983a) results, male mice in
27 this study appeared to have a higher incidence of TCE-induced body weight changes than female
28 mice.
29 The effects of extended exposure, lower durations of exposure but at higher
30 concentrations, and of cessation of exposure were examined for 150 ppm and higher doses of
31 TCE. Mice exposed to TCE at 150 ppm continuously for 120 days were reported to have
32 increased liver weight (i.e., 1.57-fold of control for females and 1.49-fold of control for males),
33 but in the case of male mice, also to have a significant decrease in body weight of 17% in
34 comparison to control groups. Increasing the exposure concentration to 900-ppm TCE and
35 reducing exposure time to 4 hours/day for 120 days also resulted in increased liver weight (i.e.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-74 DRAFT—DO NOT CITE OR QUOTE
-------
1 1.35-fold of control for females and 1.49-fold of controls for males) but with a significant
2 decrease in body weight in females of 7% in comparison to control groups. For mice that were
3 exposed to 150-ppm TCE for 30 days and then examined 120 days after the cessation of
4 exposure, liver weights were 1.09-fold of control for female mice and the same as controls for
5 male mice. With the exception of 1,800 ppm and 3,600 ppm TCE groups exposed at 2 and 1
6 hour, respectively, exposure from 225 ppm, 450 and 900 ppm at 16, 8, and 4 hours, respectively
7 for 30 days did not result in decreased body weight in males or female mice. These exposures
8 did result in increased liver weights in relation to control groups and for female mice the
9 magnitude of increase was similar (i.e., 1.50-, 1.54-, and 1.51-fold of control for liver weight
10 after exposure to 225-ppm TCE 16 hours/day, 450-ppm TCE 8 hours/day, and 900-ppm TCE
11 4 hours/day, respectively). For these groups, initial body weights varied by 13% in females and
12 14% in males. Thus, under circumstances without body weight changes due to TCE toxicity,
13 liver weight appeared to have a consistent relationship with the product of duration and
14 concentration of exposure in female mice. For male mice, the increases in TCE-induced liver
15 weight were more variable (i.e., 1.94-, 1.74-, and 1.61-fold of control for liver weight after
16 exposure to 225-ppm TCE 16 hours/day, 450-ppm TCE 8 hours/day, and 900-ppm TCE
17 4 hours/day, respectively) with the product of exposure duration and concentration did not result
18 in a consistent response in males (e.g., a lower dose for a longer duration of exposure resulted in
19 a greater response than a larger dose at a shorter duration of exposure).
20 Kjellstrand et al. (1983b) reported light microscopic findings from this study and report
21 that
22
23 after 150 ppm exposure for 30 days, the normal trabecular arrangement of the
24 liver cells remained. However, the liver cells were generally larger and often
25 displayed a fine vacuolization of the cytoplasm. The nucleoli varied slightly to
26 moderately in size and shape and had a finer, granular chromatin with a varying
27 basophilic staining intensity. The Kupffer cells of the sinusoid were increased in
28 cellular and nuclear size. The intralobular connective tissue was infiltrated by
29 inflammatory cells. There was not sign of bile stasis. Exposure to TCE in higher
30 or lower concentrations during the 30 days produced a similar morphologic
31 picture. After intermittent exposure for 30 days to a time weighted average
32 concentration of 150 ppm or continuous exposure for 120 days, the trabecular
33 cellular arrangement was less well preserved. The cells had increased in size and
34 the variations in size and shape of the cells were much greater. The nuclei also
35 displayed a greater variation in basophilic staining intensity, and often had one or
36 two enlarged nucleoli. Mitosis was also more frequent in the groups exposed for
37 longer intervals. The vacuolization of the cytoplasm was also much more
38 pronounced. Inflammatory cell infiltration in the interlobular connective tissue
39 was more prominent. After exposure to 150 ppm for 30 days, followed by 120
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-75 DRAFT—DO NOT CITE OR QUOTE
-------
1 days of rehabilitation, the morphological picture was similar to that of the air-
2 exposure controls except for changes in cellular and nuclear sizes.
3
4 Although not reporting comparisons between changes in male and female mice in the results
5 section of the paper, the authors state in the discussion section that "However, liver mass
6 increase and the changes in liver cell morphology were similar in TCE-exposed male and female
7 mice."
8 The authors do not present any quantitative data on the lesions they describe, especially
9 in terms of dose-response. Most of the qualitative description is for the 150-ppm exposure level,
10 in which there are consistent reports of TCE induced body weight decreases in male mice. The
11 authors suggest that lower concentrations of TCE give a similar pathology as those at the
12 150-ppm level, but do not present data to support that conclusion. Although stating that Kupffer
13 cells were increased in cellular and nuclear size, no differential staining was applied light
14 microscopy sections distinguish Kupffer from endothelial cells lining the hepatic sinusoid in this
15 study. Without differential staining such a determination is difficult at the light microscopic
16 level. Indeed, Goel et al. (1992) describe proliferation of sinusoidal endothelial cells after
17 1,000 mg/kg/d and 2,000 mg/kg/d TCE exposure for 28 days in male Swiss mice. However, the
18 described inflammatory cell infiltrates in the Kjellstrand et al. (1983b) study are consistent with
19 invasion of macrophages and well as polymorphonuclear cells into the liver, which could
20 activate resident Kupffer cells. Although not specifically describing the changes as consistent
21 with increased polyploidization of hepatocytes, the changes in cell size and especially the
22 continued change in cell size and nuclear staining characteristics after 120 days of cessation of
23 exposure are consistent with changes in polyploidization induced by TCE. Of note is that in the
24 histological description provided by the authors, although vacuolization is reported and
25 consistent with hepatotoxicity or lipid accumulation, which is lost during routine histological
26 slide preparation, there is no mention of focal necrosis or apoptosis resulting from these
27 exposures to TCE.
28
29 E.2.2.7. Buben and O'Flaherty, 1985
30 This study was conducted with older mice than those generally used in chronic exposure
31 assays (Male Swiss-Cox outbred mice between 3 and 5 months of age) with a weight range
32 reported between 34 to 45 g. The mice were administered distilled TCE in corn oil by gavage
33 5 times a week for 6 weeks at exposure concentrations of either 0, 100, 200, 400, 800, 1,600,
34 2,400, or 3,200 mg TCE/kg/day. While 12-15 mice were used in most exposure groups, the
35 100- and 3,200-mg/kg groups contained 4-6 mice and the two control groups consisted of 24
36 and 26 mice. Liver toxicity was determined by "liver weight increases, decreases in liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-76 DRAFT—DO NOT CITE OR QUOTE
-------
1 glucose-6-phosphate (G6P) activity, increases in liver triglycerides, and increases in serum
2 glutamate-pyruvate transaminase (SGPT) activity." Livers were perfused with cold saline prior
3 to testing for weight and enzyme activity and hepatic DNA was measured.
4 The authors reported the mice to tolerate the 6-week exposed with TCE with few deaths
5 occurring except at the highest dose and that such deaths were related to central nervous system
6 depression. Mice in all dose groups were reported to continue to gain weight throughout the
7 6-week dosing period. However, TCE exposure caused "dose-related increases in liver weight to
8 body weight ratio and since body weight of mice were generally unaffected by treatment, the
9 increases represent true liver weight increases." Exposure concentrations, as low as
10 100 mg/kg/d, were reported to be "sufficient to cause statistically significant increase in the liver
11 weight/body weight ratio," and the increases in liver size to be "attributable to hypertrophy of the
12 liver cells, as revealed by histological examination and by a decrease in the DNA concentration
13 in the livers." Mice in the highest dose group were reported to display liver weight/body weight
14 ratios that were about -75% greater than those of controls and even at the lowest dose there was
15 a statistically significant increase (i.e., control liver/body weight percent was reported to be
16 5.22% ± 0.09% vs. 5.85% ± 0.20% in 100 mg/kg/d exposed mice). The percent liver/body ratios
17 were 5.22% ± 0.09%, 5.84% ± 0.20%, 5.99% ± 0.13%, 6.51% ± 0.12%, 7.12% ± 0.12%,
18 8.51% ± 0.20%, 8.82% ± 0.15%, and 9.12% ± 0.15% for control (n = 24), 100 (n = 5),
19 200 (n = 12), 400 (n = 12), 800 (n = 12), 1,600 (n = 12), 2,400 (n = 12), and 3,200 (n = 4)
20 mg/kg/d TCE. This represents 1.12-, 1.15-, 1.25-, 1.36-, 1.63-, 1.69-, and 1.75-fold of control
21 for these doses. All dose groups of TCE induced a statistically significant increase in liver/body
22 weight ratios. For the 200 through 1,600 mg/kg exposure levels, the magnitudes of the increases
23 in TCE exposure concentrations were similar to the magnitudes of TCE-induced increases in
24 percent liver/body weight ratios (i.e., a ~2-fold increase in TCE dose resulted in ~1.7-fold
25 increase change in percent liver/body weight).
26 TCE exposure was reported to induce a dose-related trend towards increased triglycerides
27 (i.e., control values of 3.08 ± 0.29 vs. 6.89 ± 1.40 at 2,400 mg/kg TCE) with variation of
28 response increased with TCE exposure. For liver triglycerides the reported values in mg/g liver
29 were 3.08 ± 0.29 (n = 24), 3.12 ± 0.49 (n = 5), 4.41 ± 0.76 (n = 12), 4.53 ± 1.05 (n = 12),
30 5.76 ± 0.85 (n = 12), 5.82 ± 0.93 (n = 12), 6.89 ± 1.40 (n = 12), and 7.02 ± 0.69 (n = 4) for
31 control, 100, 200, 400, 800, 1,600, 2,400, and 3,200 mg/kg/d dose groups, respectively.
32 For G6P the values in ug phosphate/mg protein/20 minutes were 125.5 ± 3.2 (n = 12),
33 117.8±6.0(w = 5), 116.4 ± 2.8 (n = 9), 117.3 ± 4.6 (n = 9), 111.7 ± 3.3 (n = 9), 89.9 ± 1.7
34 (n = 9), 83.8 ± 2.1 (n = 8), and 83.0 ± 7.0 (n = 3) for the same dose groups. Only the
35 2,400 mg/kg/d dosing group was reported to be statistically significantly increased for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-77 DRAFT—DO NOT CITE OR QUOTE
-------
1 triglycerides after TCE exposure although there appeared to be a dose-response. For decreases
2 in G6P the 800 mg/kg/d and above doses were statistically significant. The numbers of animals
3 varied between groups in this study but in particular only a subset of the animals were tested for
4 G6P with the authors providing no rationale for the selection of animals for this assay. The
5 differences in the number of animals per group and small number of animals per group affected
6 the ability to determine a statistically significant change in these parameters but the changes in
7 liver weights were robust enough and variation small enough between groups that all TCE-
8 induced changes were described as statistically significant. The livers of TCE treated mice,
9 although enlarged, were reported to appear normal. A dose-related decrease in
10 glucose-6-phophatase activity was reported with similar small decreases (-10%) observed in the
11 TCE exposed groups that did not reach statistical significance until the dose reached 800 mg/kg
12 TCE exposure. SGPT activity was not observed to be increased in TCE-treated mice except at
13 the two highest doses and even at the 2,400 mg/kg dose half of the mice had normal values. The
14 large variability in SGPT activity was indicative of heterogeneity of this response between mice
15 at the higher exposure levels for this indicator of liver toxicity. However, the results of this
16 study also demonstrate that hepatomegaly was a robust response that was observed at the lowest
17 dose tested, was dose-related, and was not accompanied by toxicity.
18 Liver histopathology and DNA content were determined only in control, 400, and
19 1,600 mg/kg TCE exposure groups. DNA content was reported to be significantly decreased
20 from 2.83 ± 0.17 mg/g liver in controls to 2.57 ± 0.14 in 400 mg/kg TCE treated group, and to
21 2.15 ± 0.08 mg/kg liver in the 1,600 mg/kg exposed group. This result was consistent with a
22 decreased number of nuclei per gram of liver and hepatocellular hypertrophy. Liver
23 degeneration was reported as swollen hepatocytes and to be common with treatment. "Cells had
24 indistinct borders; their cytoplasm was clumped and a vesicular pattern was apparent. The
25 swelling was not simply due to edema, as wet weight/dry weight ratios did not increase."
26 Karyorhexis (the disintegration of the nucleus) was reported to be present in nearly all specimens
27 and suggestive of impending cell death. A qualitative scale of negative, 1,2, 3, or 4 was given
28 by the authors to rate their findings without further definition or criterion given for the ratings.
29 "No Karyorhexis, necrosis, or polyploidy was reported in controls, but a score of 1 for
30 Karyorhexis was given for 400 mg/kg TCE and 2 for 1600 mg/kg TCE." Central lobular
31 necrosis reported to be present only at the 1,600 mg/kg TCE exposure level and as a score of 1.
32 "Polyploidy was also characteristic in the central lobular region" with a score of 1 for both 400
33 and 1,600 mg/kg TCE. The authors reported that "hepatic cells had two or more nuclei or had
34 enlarged nuclei containing increased amounts of chromatin, suggesting that a regenerative
35 process was ongoing" and that there were no fine lipid droplets in TCE exposed animals. The
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-78 DRAFT—DO NOT CITE OR QUOTE
-------
1 finding of "no polyploidy" in control mouse liver is unexpected given that binucleate and
2 polyploid hepatocytes are a common finding in the mature mouse liver. It is possible that the
3 authors were referring to unusually high instances of "polyploidy" in comparison to what would
4 be expected for the mature mouse. The score given by the authors for polyploidy did not
5 indicate a difference between the two TCE exposure treatments and that it was of the lowest
6 level of severity or occurrence. No score was given for centrolobular hypertrophy although the
7 DNA content and liver weight changes suggested a dose response. The "Karyorhexis" described
8 in this study could have been a sign of cell death associated with increased liver cell number or
9 dying of maturing hepatocytes associated with the increased ploidy, and suggests that TCE
10 treatment was inducing polyploidization. Consistent with enzyme analyses, centrilobular
11 necrosis was only seen at the highest dose and with the lowest qualitative score, indicating that
12 even at the highest dose there was little toxicity.
13 Thus, the results of this study of TCE exposure for 6 weeks, is consistent with acute
14 studies and show that the region of the liver affected by TCE is the centralobular region, that
15 hepatocellular hypertrophy is observed in that region, and that increased liver weight is induced
16 at the lowest exposure level tested and much lower than those inducing overt toxicity. These
17 authors suggest polyploidization is occurring as a result of TCE exposure although a quantitative
18 dose response cannot be determined from these data.
19
20 E.2.2.8. Channel et al, 1998
21 This study was performed in male hybrid B6C3Fl/CrlBR mice (13 weeks-old,
22 25-30 grams) and focused on indicators of oxidative stress. TCE was administered by oral
23 gavage 5 days a week in corn oil for up to 55 days for some groups. Although the study design
24 indicated that water controls, corn oil controls, and exposure levels of 400, 800, and 1,200 mg/kg
25 day TCE in corn oil, results were not presented for water controls for some parameters measured.
26 Initial body weights and those recorded during the course of the study were not reported for
27 individual treatment groups. Liver samples were collected on study days 2, 3, 6, 10, 14, 21, 28,
28 35, 42, 49, and 56. Histopathology was studied from a single section taken from the median
29 lobe. Thiorbarbiturate acid-reactive substances (TEARS) were determined from whole liver
30 homogenates. Nuclei were isolated from whole liver homogenates and DNA assayed for
31 8-hydroxy-2' deoxyguanosine (8-OHdG). There was no indication that parenchymal cell and
32 nonparenchymal cells were distinguished in the assay. Free radical electron paramagnetic
33 resonance (EPR) for total radicals was analyzed in whole liver homogenates. For peroxisome
34 detection and analysis, livers from 3 mice from the 1,200 mg/kg TCE and control (oil and water)
35 groups were analyzed via electron microscopy. Only centrilobular regions, the area stated by the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-79 DRAFT—DO NOT CITE OR QUOTE
-------
1 authors to be the primary site of peroxisome proliferation, were examined. For each animal, 7
2 micrographs of randomly chosen hepatocytes immediately adjacent to the central vein were
3 examined with peroxisomal area to cytoplasmic area, the number of peroxisomes per unit area of
4 cytoplasm, and average peroxisomal size quantified. Proliferation cell nuclear antigen (PCNA),
5 described as a marker of cell cycle except GO, was examined in histological sections for a
6 minimum of 18 fields per liver section. The authors did not indicate what areas of the liver
7 lobule were examined for PCNA. Apoptosis was detected on liver sections using a apoptosis kit
8 using a single liver section from the median lobe and based on the number of positively labeled
9 cells per 10 mm2 in combination with the morphological criteria for apoptosis of
10 Columbano et al. (1985). However, the authors did not indicate what areas of the liver lobule
11 were specifically examined.
12 The authors reported that body weight gain was not adversely affected by TCE dosing of
13 the time course of the study but did not show the data. No gross lesions were reported to be
14 observed in any group. For TEARS no water control data was reported by the authors. Data
15 were presented for 6 animals per group for the corn oil control group and the 1,200 mg/kg group
16 (error bars representing the SE). No data were presented without corn oil so that the effects of
17 corn oil on the first day of the study (Day 2 of dosing) could not be determined. After 2 and
18 3 days of dosing the corn oil and 1,200-mg/kg TCE groups appeared to have similar levels of
19 TEAR detected in whole liver as nmol TBARS/mg protein. However, by Day 6 the corn oil
20 treated control had a decrease in TEAR that continued until Day 15 where the level was -50% of
21 that reported on Days 2 and 3. The variation between animals as measured by SE was reported
22 to be large on Day 10. By Day 20 there was a slight increase in variation that declined by
23 Day 35 and stayed the same through Day 55. For the TCE exposed group the TBARs remained
24 relatively consistent and began to decline by about Day 20 to a level that similar to the corn oil
25 declines by Day 35. Therefore, corn oil alone had a significant effect on TEAR detection
26 inducing a decline by 6 days of administration that persisted thought 55 days. TCE
27 administration at the 1,200 mg/kg dose in corn oil appeared to have a delayed decline in TBARS.
28 The authors interpreted this pattern to show that lipid peroxidation was elevated in the
29 1,200 mg/kg TCE group at Day 6 over corn oil. However, corn oil alone induced a decrease in
30 TBARs. At no time was TBARS in TCE treatment groups reported to be greater than the initial
31 levels at days 2 and 3, a time in which TCE and corn oil treatment groups had similar levels.
32 Rather than inducing increasing TBARS over the time course of the study TCE, at the
33 1,200 mg/kg dose, appeared to delay the corn oil induced suppression of TBARs detection.
34 Because the authors did not present data for aqueous control animals, the time course of TBARS
35 detection in the absence of corn oil, cannot be established.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-80 DRAFT—DO NOT CITE OR QUOTE
-------
1 For the 800 and 400 mg/kg TCE data the authors presented a figure, without standard
2 error information, for up to 35 days that shows little difference between 400 mg/kg TCE
3 treatment and corn oil suppression of TEAR induction. There was little difference between the
4 patterns of TEAR detection for 800 and 400 mg/kg TCE, indicating that both delayed TEAR
5 suppression by corn oil to a similar extent and did not induce greater TEAR than corn oil alone.
6 For 8-OHdG levels, the authors report that elevations were modest with the greatest
7 increase noted in the 1,200 mg/kg day TCE treatment group of 196% of oil controls on Day 56.
8 Levels fluctuated throughout the study with most of the time points that were elevated showing
9 129% of control for the 1,200 mg/kg/d group. Statistically significant elevations were noted on
10 days 2, 10, 28, 49 and 56 with depression on Day 3. On all other days (i.e., Days 6, 14, 21, 35,
11 and 42) the 8-OHdG values were similar to those of corn oil controls. No statistically significant
12 effects were reported to be observed at lower doses. The figure presented by the authors shows
13 the percent of controls by TCE treatment at 1,200 mg/kg/d but not the control values themselves.
14 The pattern by corn oil is not shown and neither is the standard error of the data. As a percent of
15 control values the variations were very large for many of the data points and largest for the data
16 given at Day 55 in which the authors report the largest difference between control and TCE
17 treatment. There was no apparent pattern of elevation in 8-OHdG when the data were presented
18 in this manner. Because the data for the corn oil control was not given, as well as no data given
19 for aqueous controls, the effects of corn oil alone cannot be discerned.
20 Given that for TEARS corn oil had a significant effect and showed a pattern of decline
21 after 6 days, with TCE showing a delayed decline, it is especially important to discern the effects
22 of corn oil and to see the pattern of the data. At time points when TEARS levels were reported
23 to be the same between corn oil and TCE (Days 42, 49 and 56) the pattern of 8-OHdG was quite
24 different with a lower level at Day 42 a slightly increased level at Day 49 and the highest
25 difference reported at Day 56 between corn oil control and TCE treated animals. The authors
26 report that the pattern of "lipid peroxidation" to be similar between the 1,200 and 800 mg/kg
27 doses of TCE but for there to be no significant difference between 800 mg/kg TCE and corn oil
28 controls. Thus, the pattern of TEARS as a measure of lipid peroxidation and 8-OHdG level in
29 nuclear DNA did not match.
30 In regard to total free radical levels as measured by EPR, results were reported for the
31 1,200 mg/kg TCE as a signal that was subtracted from control values with the authors stating that
32 only this dose level induced an elevation significantly different from controls. Again, aqueous
33 control values were not presented to discern the effects of corn oil or the pattern that may have
34 arisen with time of corn oil administration. The pattern of total free radical level appeared to
35 differ from that of lipid peroxidation and for that of 8-OHdG DNA levels with no changes at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-81 DRAFT—DO NOT CITE OR QUOTE
-------
1 days 2, 3, a peak level at Day 6, a rapid drop at Day 10, mild elevation at Day 20, and a
2 significant decrease at Day 49. The percentage differences between control and treated values
3 reported at Day 6 and 20 by the authors was not proportional to the fold-difference in signal
4 indicating that there was not a consistent level for control values over the time course of the
5 experiment. While differences in lipid peroxidation detection between 1,200 mg/kg TCE and
6 corn oil control were greatest at Day 14, total free radicals showed their biggest change between
7 corn oil controls and TCE exposure on Day 6, time points in which 8-OHdG levels were similar
8 between TCE treatment and corn oil controls. Again, there was no reported difference between
9 corn oil control and the 800 mg/kg TCE exposed group in total free radical formation but for
10 lipid peroxidation the 800 mg/kg TCE exposed group had a similar pattern as that of
11 1,200 mg/kg TCE.
12 Only the 1,200 mg/kg group was evaluated for peroxisomal proliferation at days 6, 10,
13 and 14. Thus, correlations with peroxisome proliferation and other parameters in the report at
14 differing times and TCE exposure concentrations could not be made. The authors report that
15 there was a treatment and time effect for percent peroxisomal area, a "treatment only" effect for
16 number of peroxisome and no effect for peroxisomal size. They also report that hepatocytes
17 examined from corn oil control rats were no different that those from water control rats for all
18 peroxisomal parameter, thus, discounting a vehicle effect. However, there was an effect on
19 peroxisomal size between corn oil control and water with corn oil decreasing the peroxisomal
20 size in comparison to water on all days tested. The highest TCE-induced percent peroxisomal
21 area and number occurred on Day 10 of the 3 time points measured for this dose and the fold
22 increase was -4.5- and ~3.1-fold increase, respectively. The day-10 peak in peroxisomal area
23 and number does not correlate with the reported pattern of free radical or 8-OHdG generation.
24 For cell proliferation and apoptosis, data were given for days 2, 6, 10, 14, and 21 in a
25 figure. PCNA cells, a measure of cells that have undergone DNA synthesis, was elevated only
26 on Day 10 and only in the 1,200 mg/kg/d TCE exposed group with a mean of-60 positive nuclei
27 per 1,000 nuclei for 6 mice (-6%). Given that there was little difference in PCNA positive cells
28 at the other TCE doses or time points studied, the small number of affected cells in the liver
29 could not account for the increase in liver size reported in other experimental paradigms at these
30 doses. The PCNA positive cells as well as "mitotic figures" were reported to be present in
31 centrilobular, midzonal, and periportal regions with no observed predilection for a particular
32 lobular distribution. No data were shown regarding any quantitative estimates of mitotic figures
33 and whether they correlated with PCNA results. Thus, whether the DNA synthesis phases of the
34 cell cycle indicated by PCNA staining were indentifying polyploidization or increased cell
35 number cannot be determined. The authors reported that there was no cytotoxicity manifested as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-82 DRAFT—DO NOT CITE OR QUOTE
-------
1 hepatocellular necrosis in any dose group and that there was no significant difference in
2 apoptosis between treatment and control groups with data not shown. The extent of apoptosis in
3 any of the treatment groups, or which groups and timepoints were studied for this effect cannot
4 be determined. No liver weight or body weight data were provided in this study.
5 These results confirm that as a vehicle corn oil is not neutral in its affects in the liver.
6 The TEARS results indicate a reduction in detection of TEARS in the liver with increasing time
7 of exposure to corn oil alone. Although control animals "treated with water" gavage were
8 studied, only the results for peroxisome proliferation were presented by the study so that the
9 effects of corn oil gavage were not easy to discern. In addition, the data were presented in such a
10 way for 8-OHdG and total free radical changes that the pattern of corn oil administration was
11 obscured. It is not apparent from this study that TCE exposure induces oxidative damage.
12
13 E.2.2.9. Dorfmueller et al, 1979
14 The focus of this study was the evaluation of "teratogenicity and behavioral toxicity with
15 inhalation exposure of maternal rats" to TCE. Female Long-Evans hooded rats (n = 12) of
16 -210 g weight were treated with 1,800 ± 200-ppm TCE for 6 hours/day, 5 days/week, for
17 22 ± 6 days (until pregnancy confirmation) continuing through Day 20 of gestation. Control
18 animals were exposed 22 ± 3 days before pregnancy confirmation. The TCE used in this study
19 contained 0.2% epichlorhydrin. Body weights were monitored as well as maternal liver weight
20 at the end of exposure. Other than organ weight, no other observations regarding the liver were
21 reported in this study. The initial weights of the dams were 212 ± 39 g (mean ± SD) and
22 204 ± 35 g for treated and control groups, respectively. The final weights were 362 ± 32 g and
23 337 ± 48 g for treated and control groups, respectively. There was no indication of maternal
24 toxicity by body weight determinations as a result of TCE exposure in this experiment and there
25 was also no significant difference in absolute or relative percent liver/body weight between
26 control and treated female rats in this study.
27
28 E.2.2.10. Kumar et al., 2001
29 In this study, adult male Wistar rats (130 ± 10 g body weight) were exposed to
30 376 ± 1.76 ppm TCE ("AnalaR grade") for 8, 12, and 24 weeks for 4 hours/day 5 days/week.
31 The ages of the rats were not given by the authors. Each group contained 6 rats. The animals
32 were exposed in whole body chambers and thus, additional oral exposure was probable. Along
33 with histopathology of light microscopic sections, enzymatic activities of alkaline phosphatase
34 and acid phosphatase, glutamic oxoacetate transaminase, glutamic pyruvate transaminase,
35 reduced glutathione and "total sulphydryl" were assayed in whole liver homogenates as well as
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-83 DRAFT—DO NOT CITE OR QUOTE
-------
1 total protein. The authors state that "the size and weight of the liver were significantly increased
2 after 8, 12, and 24 weeks of TCE exposure." However, the authors do not report the final body
3 weight of the rats after treatment nor do they give quantitative data of liver weight changes. In
4 regard to histopathology, the authors state
5
6 After 8 weeks of exposure enlarged hepatocytes, with uniform presence of fat
7 vacuoles were found in all of the hepatocytes affecting the periportal, midzonal,
8 and centrilobular areas, and fat vacuoles pushing the pyknosed nuclei to one side
9 of hepatocytes. Moreover congestion was not significant. After exposure of 12
10 and 24 weeks, the fatty changes became more progressive with marked necrosis,
11 uniformly distributed in the entire organ.
12
13 No other description of pathology was provided in this report. In regard to the description of
14 fatty change, the authors only do conventional H&E staining of sections with no precautions to
15 preserve or stain lipids in their sections. The authors provide a table with histological scoring of
16 simply + or - for minimal, mild or moderate effects and do not define the criteria for that
17 scoring. There is also no quantitative information given as to the extent, nature, or location of
18 hepatocellular necrosis. The authors report "no change was observed in GOT and GPT levels of
19 liver in all the three groups. The GSH level was significantly decreased while TSH level was
20 significantly increased during 8, 12, and 24 weeks of TCE exposure. The acid and alkaline
21 phosphatases were significantly increased during 8, 12, and 24 weeks of TCE exposure." The
22 authors present a series of figures that are poor in quality to demonstrate histopathological
23 TCE-induced changes. No mortality was observed from TCE exposure in any group despite the
24 presence of liver necrosis.
25
26 E.2.2.11. Kawamoto et al, 1988
27 The focus of this study was the long-term effects of TCE treatment on induction of
28 metabolic enzymes in male adult Wistar rats. The authors reported that 8 rats weighing 200 g
29 were treated with 2.0 g/kg TCE in olive oil administered subcutaneously twice a week for
30 15 weeks with 7 rats serving as olive oil controls. In a separate experiment, 5 rats were injected
31 with 1.0 g/kg TCE in olive oil i.p. once a day for 5 continuous days. For comparative purposes
32 groups of 5 rats each were administered 3-methylcholanthrene (20 mg/kg in olive oil i.p.),
33 Phenobarbital (80 mg/kg in saline i.p.) for 4 days as well as ethanol administered in drinking
34 water containing 10% ethanol for 14 days. Microsomes were prepared one week after the last
35 exposure from rats administered TCE for 15 weeks and 24 hours after the last exposure for the
36 other treatments.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-84 DRAFT—DO NOT CITE OR QUOTE
-------
1 Body weights were reported to be slightly less for the TCE treated group than for controls
2 with the initial weights, shown in a figure, to be similar for the first weeks of exposure. At
3 15 weeks there appeared to be -7.5% difference in mean body weights between control and TCE
4 treated rats which the authors reported to not be significantly different. Organ weights at the
5 termination of the experiment were reported to only be different for the liver with a 1.21-fold of
6 control value reported as a percentage of body weight with TCE treatment. The authors report
7 their increase in liver weights in male rats from subcutaneous exposure to TCE in olive oil
8 (2.0 g/kg) to be consistent with the range of liver weight gain in rats reported by Kjellstrand et al.
9 (1981) for 150-ppm TCE inhalation exposure (see comments on that study above). The 5-day
10 i.p. treatment with TCE was also reported to only produce increased liver weight but the data
11 were not shown and the magnitude of the percentage increase was not given by the authors. No
12 liver pathology results were studied or reported as well.
13 Along with an increase in liver weight, 15-week treatment with TCE was reported to
14 cause a significant increase of microsomal protein/g liver of-20% (10.64 ± 0.88 vs.
15 12.58 ± 0.71 mg/g liver for olive oil controls and TCE treatment, respectively). Microsomal
16 cytochrome P450 content was reported to show a mild increase that was not statistically
17 significant of 1.08-fold (1.342 ± 0.205 vs. 1.456 ± 0.159 nmol/mg protein for olive oil controls
18 and TCE treatment, respectively) of control. However, cytochrome P450 content showed
19 1.28-fold of control value (14.28 ± 2.41 vs. 18.34 ± 2.31 nmol/g liver for olive oil controls and
20 TCE treatment, respectively) in terms of g/liver. Chronic treatment of TCE was also reported to
21 cause a significant increase in cytochrome b-5 level (-1.35-fold of control) and NADPH-
22 cytochrome c reductase activity (-1.50-fold of control) in g/liver.
23 The 5-day TCE treatment via the i.p. route of administration was reported to cause a
24 significant increase in microsomal protein (-20%), induce cytochrome P450 (-50% increase
25 g/liver and 22% increase in microsomal protein), but to also increase cytochrome b-5 and
26 NADPH-cytochrome c reductase activity by 50 and 70% in g/liver, respectively. Although
27 weaker, 5-day i.p. treatment with TCE induced an enzyme pattern more similar to that of
28 Phenobarbital and ethanol rather methylcholanthrene (i.e., increased cytochrome P450 but not
29 microsomal protein and NADPH-cytochrome c reductase). Direct quantitative comparisons of
30 vehicle effects and potential impact on response to TCE treatments for 15 weeks subcutaneous
31 exposure and 5-day i.p. exposure could not be made as baseline levels of all enzyme and protein
32 levels changed as a function of age. Of note is that, in the discussion section of the paper, the
33 authors disclose that injection of TCE 2.0 or 3.0 g/kg i.p. for 5 days resulted in paralytic ileus
34 from TCE exposure as unpublished observations. They note that the rationale for injecting TCE
35 subcutaneously was not only that it did not require an inhalation chamber but also guarded
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-85 DRAFT—DO NOT CITE OR QUOTE
-------
1 against peritonitis that sometimes occurs following repeated i.p. injection. In terms of
2 comparison with inhalation or oral results, the authors note that the subcutaneous treatment
3 paradigm will result in TCE not immediately being metabolized but retained in the fatty tissue
4 and that after cessation of exposure TCE metabolites continued to be excreted into the urine for
5 more than 2 weeks.
6
7 E.2.2.12. National Toxicology Program (NTP), 1990
8 E.2.2.12.1. 13-week studies. The NTP conducted a 13 weeks study of 7 week old F344/N rats
9 (10 rats per group) that received doses of 125 to 2,000 mg/kg (males [0, 125, 250, 500, 1,000, or
10 2,000 mg/kg]) and 62.5 to 1,000 mg/kg (females [0, 62.5, 125, 250, 500, or 1,000 mg/kg] TCE
11 via corn oil gavage 5 days per week (see Table E-l). For 7-week old B6C3Flmice (n = 10 per
12 group), the dose levels were reported to be 375 to 6,000 mg/kg TCE (0, 375, 750, 1,500, 3,000,
13 or 6,000 mg/kg). Animals were exposed via corn oil gavage to TCE that was epichlorhydrin
14 free. All rats were reported to survive the 13-week study, but males receiving 2,000 mg/kg
15 exhibited a 24% difference in final body weight. However, there was great variation in initial
16 weights between the dose groups with mean initial weights at the beginning of the study reported
17 to 87, 88, 92, 95, 101, and 83 grams for the control, 125, 250, 500, 1,000, and 2,000 mg/kg dose
18 groups in male rats, respectively. This represents a 22% difference between the highest and
19 lowest initial weights between groups. Thus, changes in final body weight after TCE treatment
20 also reflect differences in starting weights between the groups which in the case of the 500, and
21 1,000 mg/kg groups would results in an lower than expected change in weight due to TCE
22 exposure. For female rats, the mean initial starting weights were reported to be 81, 72, 74, 75,
23 73, and 76 g, respectively for the control, 62.5, 125, 250, 500, and 1,000 mg/kg dose groups.
24 This represents a -13% difference between initial weights. In the case of female rats the larger
25 mean initial weight in the control group would tend to exaggerate the effects of TCE exposure on
26 final body weight. The authors did not report the variation in initial or final body weights within
27 the dose groups. At the lowest doses for male and female rats body mean weights were reported
28 to be decreased by 6 and 7% in male and female rats, respectively. Organ weight changes were
29 not reported for rats.
30 For male mice, mean initial body weights ranged from 19 to 22 g (-16% difference) and
31 for female mice ranged between 18 and 15 g (20% difference), and thus, similar to rats, the final
32 body weights in the groups dose with TCE reflect not only the effects of the compound but also
33 differences in initial weights. For male mice, the mean final body weights were reported to be 3
34 to 17% less than controls for the 375 to 3,000 mg/kg dose. For female mice the percent
35 difference in final body weight was reported to be the same except for the 6,000 mg/kg dose
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-86 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
group but this lack of difference between controls and treated female mice reflected no change in
mice that started at differing weights. Male mice started to exhibit mortality at 1,500 mg/kg with
8/10 surviving the 1,500 mg/kg dose, 3/10 surviving the 3,000 mg/kg dose, and none surviving
the 6,000 mg/kg dose of TCE until the end of the study. For females, 1 animal out of 10 died in
the 750, 1,500, and 3,000 mg/kg dose groups and one surviving the 6,000 mg/kg group. In
general, the magnitude of increase in TCE exposure concentration was similar to the magnitude
of increase in percent liver/body weight for the 750 and 1,500 mg/kg TCE exposure groups in
male B6C3F1 mice and for the 750 to 3,000 mg/kg TCE exposure groups in female mice (i.e., a
2-fold increase in TCE exposure resulted in ~2-fold increase in percent liver/body weight).
Table E-l. Mice data for 13 weeks: mean body and liver weights
Dose (mg/kg
TCE)
Survival
Body weight
(mean in g)
Initial
Final
Liver weight
(mean final in g)
% liver weight/BW
(fold change vs.
control)
Male
0
375
750
1,500
3,000
6,000
10/10
10/10
10/10
8/10
3/10
0/10
21
20
21
19
20
22
36
35
32
29
30
-
2.1
1.74
2.14
2.27
2.78
-
5.8
5.0(0.86)
6.8(1.17)
7.6(1.31)
8.5(1.46)
-
Female
0
375
750
1,500
3,000
6,000
10/10
10/10
9/10
9/10
9/10
1/10
18
17
17
17
15
15
26
26
26
26
26
27
1.4
1.31
1.55
1.8
2.06
2.67
5.5
5.0(0.91)
5.8(1.05)
6.5(1.18)
7.8(1.42)
9.5(1.73)
13
14
15
16
17
18
19
The descriptions of pathology in rats and mice given by this study were not very detailed.
For rats only control and high dose rats were examined histologically. For mice only controls
and the two highest dose groups were examined histologically. Only mean liver weights were
reported with no statistical analyses provided to ascertain quantitative differences between study
groups.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-87 DRAFT—DO NOT CITE OR QUOTE
-------
1 Pathological results were reported to reveal that 6/10 males and 6/10 female rats had
2 pulmonary vasculitis at the highest concentration of TCE. This change was also reported to have
3 occurred in 1/10 control male and female rats. Most of those animals were also reported to have
4 had mild interstitial pneumonitis. The authors report that viral liters were positive during this
5 study for Sendai virus.
6 In mice, liver weights (both absolute and as a percent of body weight) were reported to
7 increase with TCE-exposure level. Liver weights were reported to have increased by more than
8 10% relative to controls for males receiving 750 mg/kg or more and for females receiving
9 1,500 mg/kg or more. The most prominent hepatic lesions detected in the mice were reported to
10 be centrilobular necrosis, observed in 6/10 males and 1/10 females administered 6,000 mg/kg.
11
12 Although centrilobular necrosis was not seen in either males or females
13 administered 3000 mg/kg, 2/10 males had multifocal areas of calcifications
14 scattered throughout their livers. These areas of calcification were considered to
15 be evidence of earlier hepatocellular necrosis. Multifocal calcification was also
16 seen in the liver of a single female mouse that survived the 6000 mg/kg dosage
17 regime. One female mouse administered 3000 mg/kg also had a hepatocellular
18 adenoma, an extremely rare lesion in female mice of this age (20 weeks).
19
20 There appeared to be consistent decrease in liver weight at the lowest dose in both female and
21 male mice after 13 weeks of TCE exposure. Liver weight was increased at exposure
22 concentrations in which there was not increased mortality due to TCE exposure at 13 weeks of
23 TCE exposure.
24
25 E.2.2.12.2. 2-year studies. In the 2-year phase of the NTP study, TCE was administered by
26 corn oil gavage to groups of 50 male and 50 female F344/N rats, and B6C3F1 mice. Dosage
27 levels were 500 and 1,000 mg/kg for rats and 1,000 mg/kg for mice. TCE was administered
28 5 times a week for 103 weeks and surviving animals were killed between weeks 103 and 107.
29 The same number of animals receiving corn oil gavage served as controls. The animals were
30 8 weeks old at the beginning of exposure. The focus of this study was to determine if there was
31 a carcinogenic response due to TCE exposure so there was little reporting of non-neoplastic
32 pathology or toxicity. There was no report of liver weight at termination of the study, only body
33 weight.
34 The authors reported that there was no increase in necrosis in the liver from TCE
35 exposure in comparison to control mice. In control male mice, the incidence of hepatocellular
36 carcinoma (tumors with markedly abnormal cytology and architecture) was reported to be 8/48
37 in controls, and 31/50 in TCE-exposed male mice. For females control mice hepatocellular
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-88 DRAFT—DO NOT CITE OR QUOTE
-------
1 carcinomas were reported in 2/48 of controls and 13/49 of TCE-exposed female mice.
2 Specifically, the authors described liver pathology in mice as follows:
3
4 Microscopically the hepatocellular adenomas were circumscribed areas of
5 distinctive hepatic parenchymal cells with a perimeter of normal appearing
6 parenchyma in which there were areas that appeared to be undergoing
7 compression from expansion of the tumor. Mitotic figures were sparse or absent
8 but the tumors lacked typical lobular organization. The hepatocellular
9 carcinomas had markedly abnormal cytology and architecture. Abnormalities in
10 cytology included increased cell size, decreased cell size, cytoplasmic
11 eosinophilia, cytoplasmic basophilia, cytoplasmic vacuolization, cytoplasmic
12 hyaline bodies, and variations in nuclear appearance. In many instance, several
13 or all of the abnormalities were present in different areas of the tumor. There
14 were also variations in architecture with some of the hepatocellular carcinomas
15 having areas of trabecular organization. Mitosis was variable in amount and
16 location.
17
18 The authors report that the non-neoplastic lesion in male mice differing from controls was focal
19 necrosis in 4 versus 1 animal in the dosed group (8 vs. 2%). There was no fatty metamorphosis
20 in treated male mice versus 2 animals in control. In female mice there was focal inflammation in
21 29 versus 19% of animals (dosed vs. control) and no other changes. Therefore, the reported
22 pathological results of this study did not show that the liver was showing signs of toxicity after
23 two years of TCE exposure except for neoplasia.
24 For hepatocellular adenomas the incidence was reported to be "7/48 control vs. 14/50
25 dosed in males and 4/48 in control vs. 16/49 dosed female mice." The administration of TCE to
26 mice was reported to cause increased incidences of hepatocellular carcinomas in males (control,
27 8/48; dosed, 31/50: p = 0.001) and in females (control 2/48; dosed 13/49; p < 0.005).
28 Hepatocellular carcinomas were reported to metastasize to the lungs in five dosed male mice and
29 one control male mouse, while none were observed in females. The incidences of hepatocellular
30 adenomas were reported to be increased in male mice (control 7/48; dosed 14/50) and in female
31 mice (control 4/48; dosed 16/49; p < 0.05). The survival of both low and high dose male rats and
32 dosed male mice was reported to be less than that of vehicle controls with body weight decreases
33 dose dependent. Female mice body weights were comparable to controls. The authors report
34 adjusted rates of 20.6% for control versus 53.1% for dosed males for adenoma, 22.1% control,
35 and 92.9% for carcinoma in males, and liver carcinoma or adenoma adjusted rates of 100%. For
36 female mice the adjusted rates were reported to be 12.5% adenoma for control versus 55.6% for
37 dosed, and 6.2% control carcinoma versus 43.9% dosed, with liver carcinoma or adenoma
38 adjusted rates of 18.7% for control versus 69.7% for dosed. All of the liver results for male and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-89 DRAFT—DO NOT CITE OR QUOTE
-------
1 female mice were reported to be statistically significant. The administration of TCE was
2 reported to cause earlier expression of tumors as the first animals with carcinomas were
3 57 weeks for TCE-exposed animals and 75 weeks for control male mice.
4 In male rats there was no reported treatment related non-neoplastic liver lesions. In
5 female rats a decrease in basophilic cytological change was reported to be of note in TCE treated
6 rats (-50% in controls but -5% in TCE treatment groups). However, the authors reported that
7 "the results in male F344/N rats were considered equivocal for detecting a carcinogenic response
8 because both groups receiving TCE showed significantly reduced survival compared to vehicle
9 controls (35/70, 70%; 20/50, 40%; 16/50, 32%) and because 20% of the animals in the high-dose
10 group were killed accidently by gavage error." Specifically 1 male control, 3 low-dose males,
11 10 high-dose males, 2 female controls, 5 low-dose females and 5 high-dose female rats were
12 killed by gavage error.
13
14 E.2.2.13. National Toxicology Program (NTP), 1988
15 The studies described in the NTP (1988) TCE report were conducted "to compare the
16 sensitivities of four strains of rats to diisopropylamine-stabilized TCE." However, the authors
17 conclude
18
19 that because of chemically induced toxicity, reduced survival, and incomplete
20 documentation of experimental data, the studies are considered inadequate for
21 either comparing or assessing TCE-induced carcinogenesis in these strains of rats.
22 TCE (more than 99% pure, stabilized with 8ppm diisopropylamine) was
23 administered via corn oil gavage at exposure concentrations of 0, 500 or 1000
24 mg/kg per day, 5 days per week, for 103 weeks to 50 male and female rats of each
25 strain. The survival of "high-dose male Marshal rats was reduced by a large
26 number of accidental deaths (25 animals were accidentally killed).
27
28 However, the report notes survival was decreased at both exposure levels of TCE because of
29 mortality that occurred during the administration of the chemical. The number of animals
30 accidently killed were reported to be: 11 male ACI rats at 500 mg/kg, 18 male ACI rats at
31 1,000 mg/kg, 2 vehicle control female ACI rats, 14 female ACI rats at 500 mg/kg, 12 male ACI
32 rats at 1,000 mg/kg, 6 vehicle control male August rats, 12 male August rats at 500 mg/kg,
33 11 male August rats at 1,000 mg/kg, 1 vehicle control female August rats, 6 female August rats
34 at 500 mg/kg, 13 male August rats at 1,000 mg/kg, 2 vehicle control male Marshal rats, 12 male
35 Marshal rats at 500 mg/kg, 25 male Marshal rats at 1,000 mg/kg, 3 vehicle control female
36 Marshal rats, 14 female Marshal rats at 500 mg/kg, 18 female Marshal rats at 1,000 mg/kg,
37 1 vehicle control male Osborne-Mendel rat, 6 male Osborne-Mendel rats at 500 mg/kg, 7 male
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-90 DRAFT—DO NOT CITE OR QUOTE
-------
1 Osborne-Mendel rats at 1,000 mg/kg, 8 vehicle control female Osborne-Mendel rats, 6 female
2 Osborne-Mendel rats at 500 mg/kg, and 6 female Osborne-Mendel rats at 1,000 mg/kg. The age
3 of the rats "when placed on the study" were reported to differ and were for ACT rats (6.5 weeks),
4 August rats (8 weeks), Marshal rats (7 weeks), and Osborne-Mendel rats (8 weeks). The ages of
5 sacrifice also varied and were 17-18 weeks for the ACT and August rats, and 110-111 weeks for
6 the Marshal rats.
7 Results from a 13-week study were briefly mentioned in the report. For the 13-week
8 duration of exposure, groups of 10 male ACT and August rats were administered 0,125, 250, 500,
9 1,000, or 2,000 mg/kg TCE in corn oil gavage. Groups of 10 female ACT and August rats were
10 administered 0, 62.5, 125, 250, 500, or 1,000 mg/kg TCE. Groups of 10 male Marshal rats
11 received 0, 268, 308, 495, 932, or 1,834 mg/kg and groups of female Marshal rats were given 0,
12 134, 153, 248, 466, or 918 mg/kg TCE. With the exception of 3 male August rats receiving
13 2,000 mg/kg TCE, all animals survived to the end of the 13-week experimental period. "The
14 administration of the chemical for 13 weeks was not associated with histopathological changes."
15 In the 2-year study the report noted that there
16
17 was no evidence of liver toxicity described as non-neoplastic changes in male
18 ACT rats due to TCE exposure with 4% or less incidence of any lesion in control
19 or treated animals. For female ACT rats, the incidence of fatty metamorphosis
20 was 6% in control vehicle, 9% in low dose TCE, and 13% in high dose TCE
21 groups. There was also a 2%, 11%, and 8% incidence of clear cell change,
22 respectively. A 6% incidence of hepatocytomegaly was reported in vehicle
23 control and 15% incidence in the high dose group.
24
25 All other descriptors had reported incidences of less than 4%. For August rats there was also
26 little evidence of liver toxicity. In male August rats there was a reported incidence of 8, 4, and
27 10% focal necrosis in vehicle control, low dose, and high dose, respectively. Fatty
28 metamorphosis was reported to be 8% in control, and 2 and 4% in low and high dose. All other
29 descriptors were reported to be less than 4%. In female August rats, all descriptors of pathology
30 were reported to have a 4% or less incidence except for hepatomegaly, which was 10% for
31 vehicle control, 6% for the low dose and 2% for high dose TCE. For male Marshal rats there
32 was a reported 63% incidence of inflammation, NOS in vehicle control, 12% in low dose and
33 values not recorded at the high dose. There was a reported 6 and 14% incidence of fatty
34 metamorphosis in control and low dose male rats. Clear cell change was 8% in vehicle with all
35 other values 4% or less. For female Marshal rats, all values were 4% or less except for fatty
36 metamorphosis in 6% of vehicle controls. For male Osborne-Mendel rats, there was a reported
37 4, 10, and 4% incidence of focal necrosis in vehicle control, low and high dose respectively. For
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-91 DRAFT—DO NOT CITE OR QUOTE
-------
1 "cytoplasmic change/NOS," there were reported incidences of 26, 32, and 27% in vehicle, low
2 dose, and high dose animals, respectively. All other descriptors were reported to be 4% or less.
3 In female Osborne-Mendel rats there was a reported incidence of 10% of focal necrosis at the
4 low dose with all other descriptors reported at 4% or less.
5 Obviously the negative results in this bioassay are confounded by the killing of a large
6 portion of the animals accidently by experimental error. Still, these large exposure
7 concentrations of TCE did not seem to be causing overt liver toxicity in the rat. Organ weights
8 were not reported in this study, which would have been hard to interpret if they had been
9 reported because of the mortality.
10
11 E.2.2.14. Fukuda et al, 1983
12 In this 104-week bioassay designed primarily to determine a carcinogenic response,
13 female noninbred Crj :CD-1 (ICR) mice and female Crj :CD (S-D) rats 7 weeks of age were
14 exposed to "reagent grade" TCE at 0, 50, 150, and 450 ppm for 7 hours a day, 5 days a week.
15 During the 2-year duration of the experiment inhalation concentrations were reported to be
16 within 2% of target values. The numbers of animals per group were reported to be 49-50 mice
17 and 49-51 rats at the beginning of the experiment. The impurities in the TCE were reported to
18 be 0.128% carbon tetrachloride benzene, 0.019% epichlorohydrin and 0.019%
19 1,1,2-trichloroethane. After 107 weeks from commencement of the exposure, surviving animals
20 were reported to be killed and completely necropsied. "Tumors and abnormal organs as well as
21 other major organs were excised and prepared for examination in H&E sections." No other
22 details of the methodologies used for pathological examination of tissues were given including
23 what areas of the liver and number of sections examined by light microscopy.
24 Body weights were not given but the authors reported that "body weight changes of the
25 mice and rats were normal with a normal range of standard deviation." It was also reported that
26 there were no significant differences in average body weight of animals at specified times during
27 the experiments and no significant difference in mortality between the groups of mice. The
28 report includes a figure showing, that for the first 60 weeks of the experiment, there was a
29 difference in cumulative mortality at the 450 ppm dose in ICR mice and the other groups. The
30 authors reported that significantly increased mortalities in the control group of rats compared to
31 the other dosed groups were observed at 85 weeks and after 100 weeks reflecting many deaths
32 during the 81-85 week and 96-100 week periods for control rats. No significant comparable
33 clinical observations were reported to be noted in each group but that major symptoms such as
34 bloody nasal discharge (in rats), local alopecia (in mice and rats), hunching appearance (in mice)
35 and respiratory disorders (in mice and rats) were observed in some animals mostly after 1 year.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-92 DRAFT—DO NOT CITE OR QUOTE
-------
1 The authors report that "the numbers of different types of tumors were counted and only
2 malignant tumors were counted when both malignant and benign tumors were observed within
3 one organ." They also reported that "all animals were included in the effective numbers except
4 for a few that were killed accidently, severely autolyzed or cannibalized, and died before the first
5 appearance of tumors among the groups." In mice the first tumors were observed at 286 days as
6 thymic lymphoma and most of the malignant tumors appearing later were described as
7 lymphomas or lymphatic leukemias. The incidences of mice with tumors were 37, 36, 54, and
8 52% in the control, 50-, 150- and 450-ppm groups, respectively, by the end of the experiment.
9 "Tumors of the ovary, uterus, subcutaneous tissue, stomach, and liver were observed in the dose
10 groups at low incidences (2-7%) but not in the controls." For the liver, the control, 50- and
11 150-ppm groups were all reported to have no liver tumors with one animal (2%) having an
12 adenoma at the 450 ppm dose. For rats the first tumor was reported to be observed at 410 days
13 and for the incidences of animals with tumors to be 64, 78, 66, and 63% for control, 50-ppm,
14 150-ppm, and 450-ppm TCE, respectively, by the end of the experiment. Most tumors were
15 distributed in the pituitary gland and mammary gland with other tumors reported at a low
16 incidence of 2-4% with none in the controls. For the liver there were no liver tumors in the
17 control or 150-ppm groups but 1 animal (2%) had a cystic cholangioma in 50-ppm group and one
18 animal (2%) had a hepatocellular carcinoma in the 450-ppm group of rats. No details concerning
19 the pathology of the liver or organ weight changes were given by the authors, including any
20 incidences of hepatomegaly or preneoplastic foci. On note is that in these strains, there were no
21 background liver tumors in either strain, indicative of the relative insensitivity of these strains to
22 hepatocarcinogenicity. However, the carcinogenic potential of TCE was reflected by a number
23 of other tumor sites in this paradigm.
24
25 E.2.2.15. Henschler et al, 1980
26 This report focused on the potential carcinogenic response of TCE in mice (NMRI
27 random bred), rats (WIST random bred) and hamsters (Syrian random bred) exposed to 0, 100,
28 and 500-ppm TCE for 6 hours/day 5 days/week for 18 months. The TCE used in the experiment
29 was reported to be pure with the exception of trace amounts of chlorinated hydrocarbons,
30 epoxides and triethanolamines (<0.000025% w/w) and stabilized with 0.0015% triethanolamine.
31 The number of animals in each group was 30 and the ages and initial and final body weights of
32 the animals were not provided in the report. For the period of exposure (8 am-2 pm), animals
33 were deprived of food and water. The exposure period was for 18 months with mice and
34 hamsters sacrificed after 30 months and rats after 36 months. "Deceased animals" were reported
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-93 DRAFT—DO NOT CITE OR QUOTE
-------
1 to be autopsied, spleen, liver, kidneys, lungs and heart weighed, and these organs, as well as
2 stomach, central nervous system, and tumorous tissues, examined in H&E sections.
3 Body weight gain was reported to be normal in all species with no noticeable differences
4 between control and exposed groups but data were not shown. However, a "clearly dose-
5 dependent decrease in the survival rate for both male and female mice" was reported to be
6 statistically significant in both sexes and concentrations of TCE with no other significant
7 differences reported in other species. The increase in mortality was more pronounced in male
8 mice, especially after 50 weeks of exposure. Hence the opportunity for tumor development was
9 diminished due to decreased survival in TCE treated groups. No organ weights were provided
10 for the study due to the design, in which at considerable period of time occurred between the
11 cessation of exposure and the sacrifice of the animals and liver weights changes due to TCE may
12 have been diminished with time. For the 30 autopsied male mice in the control group,
13 1 hepatocellular adenoma and 1 hepatocellular carcinoma was reported. Whether they occurred
14 in the same animal cannot be determined from the data presentation. In the 29 animals in
15 thelOO-ppm TCE exposure group 2 hepatocellular adenomas and 1 mesenchymal liver tumor
16 were reported but no hepatocellular carcinomas also without a determination as whether they
17 occurred in the same animal or not. In the 30 animals autopsied in the 500-ppm-exposure group
18 no liver tumors were reported. In female mice, of the 29 animals autopsied in the control group,
19 30 animals autopsied in the 100 group, and the 28 animals autopsied in the 500-ppm group, there
20 were also no liver tumors reported.
21 In both the 100- and 500-ppm-exposure groups, of male mice especially, low numbers of
22 animals studied, abbreviated TCE exposure duration, and lower numbers of animals surviving to
23 the end of the experiment, limit the power of this study to determine a treatment-related
24 difference in liver carcinogenicity. As discussed in Section E.2.3.2 below, the use of an
25 abbreviated exposure regime or study duration and low numbers of animals examined limits the
26 power of a study to detect a treatment-related response. The lack of any observed background
27 liver tumors in the female mice and a very low background level of 2 tumors in the male mice
28 are indicative of a low sensitivity to detect liver tumors in this paradigm, which may have
29 occurred either through its design, or a low sensitivity of mouse strain used for this endpoint.
30 However, the carcinogenic potential of TCE in mice was reflected by a number of other tumor
31 sites in this paradigm.
32 For rats and hamsters the authors reported "no dose-related accumulation of any kind of
33 tumor in either sex of these species." For male rats there was only 1 hepatocellular
34 adenoma reported at 100 ppm in the 30 animals autopsied and no carcinomas. For female rats
35 there were no liver tumors reported in control animals but, more significantly, at 100 ppm there
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-94 DRAFT—DO NOT CITE OR QUOTE
-------
1 was 1 adenoma and 1 cholangiocarcinoma reported at 100 ppm and at 500 ppm
2 2 cholangioadenomas. Although not statistically significant, the occurrence of this relatively rare
3 biliary tumor was observed in both TCE dose groups in female rats. The difference in survival,
4 as reported in mice, did not affect the power to detect a response in rats, but the low numbers of
5 animals studied, abbreviated exposure duration and apparent low sensitivity to detect a
6 hepatocarcinogenic response suggest a study of low power. Nevertheless, the occurrence of
7 cholangioadenomas and 1 cholangiocarcinoma in female rats after TCE treatments is of concern,
8 especially given the relationship in origin and proximity of the bile and liver cells and the low
9 incidence of this tumor. For hamsters the low background rate of tumors of any kind suggests
10 that in this paradigm, the sensitivity for detection of this tumor is relatively low.
11
12 E.2.2.16. Maltoni et al, 1986
13 The report by Maltoni et al. (1986) included a series of "systematic and integrated
14 experiments (BT 301, 302, 303, 304, 304bis, 305, 306 bis) started in sequence, testing TCE by
15 inhalation and by ingestion." The first experiment (BT 301) was begun in 1976 and the last in
16 1983 with this report representing the complete report of the findings and results of project. The
17 focus of the study was detection of a neoplastic response with only a generalized description of
18 tumor pathology phenotype given and no reporting of liver weight changes induced by TCE
19 exposure.
20 In experiment BT 301, TCE was administered in male and female S-D rats (13 weeks at
21 start of experiment) via olive oil gavage at control, 50 mg/kg or 250 mg/kg exposure levels for
22 52 weeks (4-5 days weekly). The animals (30 male, 30 female for each dose group) were
23 examined during their lifetime. In experiment BT 302, male and female S-D rats (13 weeks old
24 at start of the experiment) were exposed to TCE via inhalation at 0, 100, and 600 ppm, 7 hours a
25 day, 5 days a week, for 8 weeks. The animals (90 animals in each control group, 60 animals in
26 each 100-ppm group, and 72 animals in each 600-ppm group) were examined during their
27 lifetime. In experiment BT 304, male and female Sprague Dawley (S-D) rats (12 weeks old at
28 start of the experiment) were exposed TCE via inhalation at 0, 100, 300, and 600 ppm 7 hours a
29 day, 5 days a week, for 104 weeks. The animals (95 male, 100 female rats control groups, 90
30 animals in each 100-ppm group, 90 animals in each 300-ppm group, and 90 animals in each 600-
31 ppm group) were examined during their lifetime. In experiment BT304bis, male and female S-D
32 rats (12 weeks old at start of the experiment) were exposed to TCE via inhalation at 0, 100, 300,
33 and 600 ppm for 7 hours a day, 5 days a week, for 104 weeks. The animals (40 male, 40 female
34 rats control groups, 40 animals in each 100-ppm group, 40 animals in each 300-ppm group, and
35 40 animals in each 600-ppm group) were examined during their lifetime.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-95 DRAFT—DO NOT CITE OR QUOTE
-------
1 In experiment BT 303, Swiss mice (11 weeks old at the start of the experiment) were
2 exposed to TCE via inhalation in for 8 weeks using the same exposure concentrations as for
3 experiment BT 302. The animals (100 animals in each control group, 60 animals in the
4 100-ppm-exposed group, and 72 animals in each 600-ppm group) were examined during their
5 lifetime. In experiment BT 305, Swiss mice (11 weeks old at the start of the experiment) were
6 exposed to TCE via inhalation in for 78 weeks, 7 hours a day, 5 days a week. The animals
7 (90 animals in each control group, 90 animals in the 100-ppm-exposed group, 90 animals in the
8 300-ppm group, and 90 animals in each 600-ppm group) were examined during their lifetime. In
9 experiment BT 306, B6C3F1 mice (from NCI source) (12 weeks old at the start of the
10 experiment) were exposed to TCE via inhalation in for 78 weeks, 7 hours a day, 5 days a week.
11 The animals (90 animals in each control group, 90 animals in the 100-ppm-exposed group,
12 90 animals in the 300-ppm group, and 90 animals in each 600-ppm group) were examined during
13 their lifetime. In experiment BT 306bis B6C3F1 mice (from Charles River Laboratory as
14 source) (12 weeks old at the start of the experiment) were exposed to TCE via inhalation for
15 78 weeks, 7 hours a day, 5 days a week. The animals (90 animals in each control group,
16 90 animals in the 100-ppm-exposed group, 90 animals in the 300-ppm group, and 90 animals in
17 each 600-ppm group) were examined during their lifetime.
18 In all experiments, TCE was supplied tested and reported by the authors of the study to
19 be was highly purified and epoxide free with butyl-hydroxy-toluene at 20 ppm used as a
20 stabilizer. Extra virgin olive oil was used as the carrier for ingestion experiments and was
21 reported to be free of pesticides. The authors describe the treatment of the animals and running
22 of the facility in detail and report that:
23
24 Animal rooms were cleaned every day and room temperature varied from 19
25 degrees to 22 degrees and was checked 3 times daily. Bedding was changed
26 every two days and cages changes and washed once weekly. The animals were
27 handled very gently and, therefore, were neither aggressive nor nervous.
28 Concentrations of TCE were checked by continuous gas-chromatographic
29 monitoring. Treatment was performed by the same team. In particular, the same
30 person carried out the gavage of the same animals. This is important, since
31 animals become accustomed to the same operators. The inhalation chambers
32 were maintained at 23 ± 2 degrees C and 50 ± 10% relative humidity. Ingestion
33 from Monday to Friday was usually performed early in the morning. The status
34 and behavior of the animals were examined at least three times daily and
35 recorded. Every two weeks the animals were submitted to an examination for the
36 detection of the gross changes, which were registered in the experimental records.
37 The animals which were found moribund at the periodical daily inspection were
38 isolated in order to avoid cannibalism. The animals were weight every two weeks
39 during treatment and then every eight weeks. Animals were kept under
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-96 DRAFT—DO NOT CITE OR QUOTE
-------
1 observation until spontaneous death. A complete necropsy was performed.
2 Histological specimens were fixed in 70% ethyl alcohol. A higher number of
3 samples was taken when particular pathological lesions were seen. All slides
4 were screened by a junior pathologist and then reviewed by a senior pathologist.
5 The senior pathologist was the same throughout the entire project. Analysis of
6 variance was used for statistical evaluation of body weights. Results are
7 expressed as means and standard deviations. Survival time is evaluated using the
8 Kruskal-Wallis test. For different survival rates between groups, the incidence of
9 lesions is evaluated by using the Log rank test. Non-neoplastic, preneoplastic,
10 and neoplastic lesions were evaluated using the Chi-square of Fisher' exact test.
11 The effect of different doses was evaluated using the Cochran-Armitage test for
12 linear trends in proportions and frequencies.
13
14 The authors state that: "Although the BT project on TCE was started in 1976 and most of the
15 experiments were performed from the beginning of 1979, the methodological protocol adopted
16 substantially met the requirements of the Good Laboratory Practices Act." Finally, it was
17 reported that "the experiments ran smoothly with no accidents in relation to the conduct of the
18 experiment and the health of the animals, apart from an excess in mortality in the male B6C3F1
19 mice of the experiment BT 306, due to aggressiveness and fighting among the animals." This is
20 in contrast to the description of the gavage studies conducted by NTP (1990, 1988) in which
21 gavage error resulted in significant loss of experimental animals. Questions have been raised
22 about the findings, experimental conditions, and experimental paradigm of the European
23 Ramazzini Foundation (ERF) from which the Maltoni et al. (1986) experiments were conducted
24 (EFSA, 2006). However, these concerns were addressed by Caldwell et al. (2008a), who
25 concluded that the ERF bioassay program produced credible results that were generally
26 consistent with those of NTP
27 In regards to effects of TCE exposure on survival,
28
29 a nonsignificant excess in mortality correlated to TCE treatment was observed
30 only in female rats (treated by ingestion with the compound) and in male B6C3F1
31 mice. In B6C3F1 mice of the experiment BT 306 bis, the excess in mortality in
32 treated animals was higher (p < 0.05 after 40 weeks) but was not dose correlated.
33 No excess in mortality was observed in the other experiments.
34
35 The authors reported that "no definite effect of TCE on body weight was observed in any of the
36 experiments, apart from experiment BT 306 bis, in which a slight nondose correlated decrease
37 was found in exposed animals."
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-97 DRAFT—DO NOT CITE OR QUOTE
-------
1 In mice, "hepatoma" was the term used by the authors of these studies to describe all
2 malignant tumors of hepatic cells, of different subhistotypes, and of various degrees of
3 malignancy. The authors reported that the hepatomas induced by exposure to TCE
4
5 may be unique or multiple, and have different sizes (usually detected grossly at
6 necropsy). Under microscopic examination these tumors proved to be of the
7 usual type observed in Swiss and B6C3F1 mice, as well as in other mouse strains,
8 either untreated or treated with hepatocarcinogens. They frequently have
9 medullary (solid), trabecular, and pleomorphic (usually anaplastic) patterns. The
10 hepatomas may produce distant metastases, more frequently in the lungs.
11
12 In regard to the induction of "hepatomas" by TCE exposure, the authors report that in
13 Swiss mice exposed to TCE by inhalation for 8 weeks (BT303), the percentage of animals with
14 hepatomas was 1.0% in male mice and 1.0% in female mice in the control group (n = 100 for
15 each gender). For animals exposed to 100 ppm TCE, the percentage in female mice was 1.7%
16 and male mice 5.0% (n = 60 for each gender). For animals exposed to 600 ppm TCE, the
17 percentage in female mice was 0% and in male mice 5.5% (n = 72 for each gender). The
18 relatively larger number of animals used in this bioassay, in comparison to NTP standard assays,
19 allows for a greater power to detect a response. It is also apparent from these results that Swiss
20 mice in this experimental paradigm are a "less sensitive" strain in regard to spontaneous liver
21 cancer induction over the lifetime of the animals. These results suggest that 8 weeks of TCE
22 exposure via inhalation at 100 ppm or 600 ppm may have been associated with a small increase
23 in liver tumors in male mice in comparison to concurrent controls.
24 In Swiss mice exposed to TCE via inhalation for 78 weeks (BT 305), the percentage of
25 animals with hepatomas was reported to be 4.4% in male mice and 0% in female mice in the
26 control group (n = 90 for each gender). For animals exposed to 100 ppm TCE, the percentage in
27 female mice was reported to be 0% and male mice 2.2% (n = 90 for each gender). For animals
28 exposed to 300 ppm TCE, the percentage in female mice was reported to be 0% and in male
29 mice 8.9% (n = 90 for each gender). For animals exposed to 600 ppm TCE, the percentage in
30 female mice was reported to be 1.1% and in male mice 14.4%. As with experiment BT303, there
31 is a consistency in the relatively low background level of hepatomas reported for Swiss mice in
32 this paradigm. After 78 weeks of exposure there appears to be a dose-related increase in
33 hepatomas in male but not female Swiss mice via inhalation exposure.
34 In B6C3F1 mice exposed to TCE by inhalation for 78 weeks (BT306) the percentage of
35 animals with hepatomas was reported to be 1.1% in male mice and 3.3% in female mice in the
36 control group (n = 90 for each gender). For animals exposed to 100 ppm TCE, the percentage in
37 female mice was reported to be 4.4% and in male mice 1.1% (n = 90 for each gender). For
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-98 DRAFT—DO NOT CITE OR QUOTE
-------
1 animals exposed to 300 ppm TCE, the percentage in female mice was reported to be 3.3% and in
2 male mice 4.4% (n = 90 for each gender). For animals exposed to 600 ppm TCE, the percentage
3 in female mice was reported to be 10.0% and in male mice 6.7%. This was the experimental
4 group with excess mortality in the male group due to fighting. The excess mortality could have
5 affected the results. The authors do report that there was a difference in the percentage of males
6 bearing benign and malignant tumors that was due to early mortality among males in experiment
7 BT306. It is unexpected for the liver cancer incidence to be less in male mice than female mice
8 and not consistent with the results reported for the Swiss mice.
9 In B6C3F1 male mice exposed to TCE via inhalation (BT 306 bis) the percentage of
10 animals with hepatomas was reported to be 18.9% in male mice in the control group (n = 90).
11 For animals exposed to 100 ppm TCE, the percentage in male mice was reported to be 21.1%
12 (n = 90). For animals exposed to 300 ppm TCE, the percentage in male mice was reported to be
13 30.0% (n = 90). For animals exposed to 600 ppm TCE, the percentage in male mice was
14 reported to be 23.3%. This experiment did not examine female mice. The authors do report a
15 decrease in survival in mice from this experiment that could have affected results. It is apparent
16 from the BT 306 and BT 306 bis experiments that the background level of liver cancer was
17 significantly different in male mice, although they were supposed to be of the same strain. The
18 finding of differences in response in animals of the same strain but from differing sources has
19 also been reported in other studies for other endpoints (see Section E.3.1.2, below).
20 The authors reported 4 liver angiosarcomas: 1 in an untreated male rat (BT 304); 1 in a
21 male and 1 in a female rat exposed to 600 ppm TCE for 8 weeks (experiment BT302); and 1 in a
22 female rat exposed to 600 ppm TCE for 104 weeks (BT 304). The authors conclude that
23
24 the tumors observed in the treated animals cannot be considered to be correlated
25 to TCE treatment, but are spontaneously arising. These findings are underlined
26 because of the extreme rarity of this tumor in control Sprague Dawley rats,
27 untreated or treated with vehicle materials. The morphology of these tumors is of
28 the liver angiosarcoma type produced by vinyl chloride in this strain of rats.
29
30 In rats treated for 104 weeks, TCE was reported to not affect the percentages of animals
31 bearing benign and malignant tumor and of animals bearing malignant tumors. Moreover, it did
32 not affect the number of total malignant tumors per 100 animals. This study did not report a
33 treatment related increase in liver cancer in rats. The report only explicitly described positive
34 findings so it is assumed that there were no increases in "hepatomas" in rat liver associated with
35 TCE treatment. The authors concluded that "under the tested experimental conditions, the
36 evidence of TCE (without epoxide stabilizer) carcinogen!city, gives the result of TCE treatment-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-99 DRAFT—DO NOT CITE OR QUOTE
-------
1 related hepatomas in male Swiss and B6C3F1 mice. A borderline increased frequency of
2 hepatomas was also seen after 8 weeks of exposure in male Swiss mice." Thus, the increase in
3 liver tumors in both strains of mice exposed to TCE via inhalation reported in this study is
4 consistent with the gavage results from the NTP (1990) study in B6C3F1 mice, where male mice
5 had a higher background level and greater response from TCE exposure than females.
6
7 E.2.2.17. Maltoni et al, 1988
8 This report was an abbreviated description of an earlier study (Maltoni et al., 1986)
9 focusing on the identification of a carcinogenic response in rats and mice by chronic TCE
10 exposure.
11
12 E.2.2.18. Van Duuren et al, 1979
13 This study exposed male and female noninbred HA:ICR Swiss mice at 6-8 weeks of age
14 to distilled TCE with no further descriptions of purity. Gavage feeding of TCE was once weekly
15 in 0.1 mL trioctanoin. Neither initial nor final body weights were reported by the authors. The
16 authors reported that, at the termination of the experiments or at death, animals were completely
17 autopsied with specimens of all abnormal-appearing tissues and organs excised for
18 histopathologic diagnosis. Tissues from the stomachs, livers, and kidneys were reported to be
19 taken routinely for the intragastric feeding experiments. Tissues were reported to be stained for
20 H&E for pathologic examination, but no further description of the lobe(s) of the liver examined
21 or the sections examined was provided by the authors. Results were as only reported the no of
22 mice with forestomach tumors 0.5 mg/mouse of TCE treatment given once a week in 0.1 mL
23 trioctanoin. Mouse body weights were not given so the dose in mg/kg for the mice cannot be
24 ascertained. The protocol used in this experiment kept the mg/mouse constant with a 1 week
25 dosing schedule so that as the mice increased weight with age, the dose as a function of body
26 weight was decreased. The days on test were reported to be 622 for 30 male and female mice.
27 2 male and 1 female mice were reported as having forestomach tumors. For 30 mice treated with
28 trioctanoin alone the number of forestomach tumors was reported to be zero. For mice with no
29 TCE treatment, 5 of 100 male mice were reported to have forestomach tumors and of 8 of
30 60 female mice were reported to have forestomach tumors for 636 and 649 days on test. No
31 results for liver were presented by the authors by the intragastric route of administration
32 including background rates of the incidences of liver tumors or treatment results. The authors
33 note that except for repeated skin applications of certain chemicals, no significant difference
34 between the incidence of distant tumors in treated animals compared with no-treatment and
35 vehicle control groups was noted. Given the uncertainties in regard to dose, the once-a week
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-100 DRAFT—DO NOT CITE OR QUOTE
-------
1 dosing regime, the low number of animals tested with resulting low power, and the lack of
2 reporting of experimental results, the ability to use the results from this experiment in regard to
3 TCE carcinogenicity is very limited.
4
5 E.2.2.19. National Cancer Institute (NCI), 1976
6 This bioassay was "initiated in 1972 according to the methods used and widely accepted
7 at that time" with the design of carcinogenesis bioassays having "evolved since then in some
8 respects and several improvements" having been developed. The most notable changes reported
9 in the foreward of the report are changes "pertaining to preliminary toxicity studies, numbers of
10 controls used, and extent of pathological examination." Industrial grade TCE was tested (99%
11 TCE, 0.19% 1,2,-epoxybutane, 0.04%v ethyl acetate, 0.09% epichlorhydrin, 0.02% TV-methyl
12 pyrrole, and 0.03% diisobutylene) with rats and mice exposed via gavage in corn oil
13 5 times/week for 78 weeks using 50 animals per group at 2 doses with both sexes of Osborne-
14 Mendel rats and B6C3F1 mice. However, for control groups only 20 of each sex and species
15 were used. Rats were killed after 110 weeks and mice after 90 weeks. Rats and mice were
16 initially 48 and 35 days of age, respectively, at the start of the experiment with control and
17 treated animals born within 6 days of each other. Initial weight ranges were reported as ranges
18 for treated and control animals of 168-229 g for male rats, 130-170 g for female rats, 11-22 g
19 for male mice, and 11-18 g for female mice. Animals were reported to be "randomly assigned
20 to treatment groups so that initially the average weight in each group was approximately the
21 same." Mice treated with TCE were reported to be
22
23 maintained in a room housing other mice being treated with one of the following
24 17 compounds: 1,1,2-2-tetrachloroethane, chloroform, 3-chloropropene,
25 chloropicrin, 1,2-dibromochloropropane, 1,2, dibromoethane, ethylene dichloride,
26 1,1-diochloroethane, 3-sulfolene, idoform, methyl chloroform, 1,1,2-
27 trichloroethane, tetrachloroethylene, hexachloroethane, carbon disulfide,
28 trichlorofluoromethane, and carbon tetrachloride. Nine groups of vehicle controls
29 and 9 groups of untreated controls were also housed in this same room.
30
31 The authors note that
32
33 TCE-treated rats and their controls were maintained in a room housing other rats
34 being treated with one of the following compounds: dibromochloropropane,
35 ethylene di chloride, 1,1-dichloroethane, and carbon disulfide. Four groups of
36 vehicle-treated controls were in the same room." Thus, there was the potential of
37 co-exposure to a number of other chemicals, especially for the mice, resulting
38 from exhalation in treated animals housed in the same room, including the control
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-101 DRAFT—DO NOT CITE OR QUOTE
-------
1 groups, as noted by the authors. The authors also noted that "samples of ambient
2 air were not tested for presence of volatile materials" but state that "although the
3 room arrangement is not desirable as is stated in the Guidelines for Carcinogen
4 Bioassay in Small Rodents, there is not evidence the results would have been
5 different with a single compound in a room.
6
7 The initial doses of TCE for rats were reported to be 1,300 and 650 mg/kg. However,
8 these levels were changed based on survival and body weight data "so that the time-weighted
9 average doses were 549 and 1097 mg/kg for both male and female rats." For mice, the initial
10 doses were reported to be 1,000 and 2,000 mg/kg for males and 700 and 1,400 mg/kg for
11 females. The "doses were increased so that the time weighted average doses were 1169 mg/kg
12 and 2339 mg/kg for male mice and 869 and 1739 mg/kg for female mice." The authors reported
13 that signs of toxicity, including reduction in weight, were evident in treated rats, which, along
14 with increased mortality, "necessitated a reduction in doses during the test." In contrast "very
15 little evidence of toxicity was seen in mice, so doses were increased slightly during the study."
16 Doses were "changed for the rats after 7 and 16 weeks of treatment, and for the mice after
17 12 weeks." At 7 weeks of age, male and female rats were dosed with 650mg/kg TCE, at
18 14 weeks they were dosed with 750 mg/kg TCE, and at 23 weeks of age 500 mg/kg TCE. For
19 the high exposure level, the exposure concentrations were 1,300, 1,500, and 1,000 mg/kg TCE,
20 respectively, for the same changes in dosing concentration. For rats the percentage of TCE in
21 corn oil remained constant at 60%. For female mice, the TCE exposure at the beginning of
22 dosing was 700 mg/kg TCE (10% in corn oil) at 5 weeks of age for the "lower dose" level. The
23 dose was increased to 900 mg/kg day (18% in corn oil) at 17 weeks of age and maintained until
24 83 weeks of age. For male mice, the TCE exposure at the beginning of dosing was 1,000 mg/kg
25 TCE (15% in corn oil) at 5 weeks of age for the "lower dose" level. At 11 weeks, the level of
26 TCE remained the same but the percentage of TCE in corn oil was reduced to 10%. The dose
27 was increased to 1,200 mg/kg day at 17 weeks of age (24% in corn oil) and maintained until
28 83 weeks of age. For the "higher dose," the TCE exposure at the beginning of dosing was
29 1,400 mg/kg TCE (10% in corn oil) at 5 weeks of age in female mice. At 11 weeks of age the
30 exposure level of TCE was kept the same but the percentage of TCE in corn oil increased to
31 20%. By 17 weeks of age the exposure concentration of TCE in corn oil was increased to
32 1,800 mg/kg (18% in corn oil) in female mice. For the "higher dose" in male mice, the TCE
33 exposure at the beginning of dosing was 2,000 mg/kg (15% in corn oil) which was maintained at
34 11 weeks in regard to TCE administered but the percent of TCE corn oil was increased to 20%.
35 For male mice the exposure concentration was increased to 2,400 mg/kg (24% in corn oil). For
36 all of the mice treatment continued on a 5 days/week schedule of oral gavage dosing throughout
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-102 DRAFT—DO NOT CITE OR QUOTE
-------
1 the timecourse of treatment (78 weeks of treatment). Thus, not only did the total dose
2 administered to the animals change, but the volumes of vehicle in which TCE was administered
3 changed throughout the experiment.
4 The authors stated that at 37 weeks of age, "To help assure survival until planned
5 termination the dosing schedule was changed for rats to a cycle of 1 week of no treatment
6 followed by 4 weeks of treatment." for male and female rats. Thus, the duration of exposure in
7 rats was also changed. All lobes of the liver were reported to be taken including the free margin
8 of each lobe with any nodule or mass represented in a block 10x5x3 mm cut from the liver
9 and fixed in a marked capsule.
10 Body weights (mean ± SD) were reported to be 193 ± 15.0 g (n = 20), 193 ± 15.8 g
11 (n = 50), and 195 ± 16.7 g(n = 50) for control, low, and high dose male rats at initiation of the
12 experiment. By 1 year of exposure (50 weeks), 20/20 control male rats were still alive to be
13 weighed, 42/50 of the low dose rats were alive and 34/50 of high dose rats were still alive. The
14 body weights of those remaining were decreased by 6.2 and 17% in the low and high dose
15 animals in comparison with the controls. For female rats, the mean body weights were reported
16 to be 146 ± 11.4 g (n = 20), 144 ± 11.0 g (n = 50), and 144 ± 9.5 g (n = 50) for control, low, and
17 high dose female rats at initiation of the experiment. By 1 year of exposure (50 weeks),
18 17/20 control female rats were still alive, 28/50 low dose and 39/50 of the high dose rats were
19 alive. The body weights of those remaining were decreased by 25 and 30% in the low and high
20 dose animals in comparison with the controls. For male mice the initial body weights were
21 17 ± 0.5 g (n = 20), 17 ± 2.0 g (n = 50), and 17 ± 1.1 g (n = 50) for control, low and high doses.
22 By 1 year of exposure (50 weeks), 18/20 control male mice were still alive, 47/50 or the low
23 dose, and 34/50 of the high-dose groups were still alive. The body weights of those remaining
24 were unchanged in comparison to controls. For female mice the initial body weights were
25 14 ± 0.0 g (n = 20), 14 ± 0.6 g (n = 50), and 14 ± 0.7 g (n = 50) for control, low and high doses.
26 By 1 year of exposure (50 weeks), 18/20 control male mice were still alive, 45/50 or the low
27 dose, and 41/50 of the high-dose groups were still alive. The body weights of those remaining
28 were unchanged in comparison to controls.
29 A high proportion of rats were reported to die during the experiment with 17/20 control,
30 42/50 low dose, and 47/50 high dose animals dying prior to scheduled termination. For female
31 rats, 12/20 control, 35/48 low dose, and 37/50 high dose animals were reported to die before
32 scheduled termination with two low dose females reported to be missing and not counted in the
33 denominator for that group. The authors reported that earlier death was associated with higher
34 TCE dose. A decrease in the percentage of tumor-bearing animals was reported to be lower in
35 treated animals and attributed by the authors to be likely related to the decrease in their survival.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-103 DRAFT—DO NOT CITE OR QUOTE
-------
1 A high percentage of respiratory disease was reported to be observed among the rats without any
2 apparent difference in the type, severity, or morbidity as to sex or group. The authors reported
3 that "no significant toxic hepatic changes were observed" but no other details regarding results in
4 the liver of rats. Carbon tetrachloride was administered to rats as a positive control. Alow
5 incidence of both hepatocellular carcinoma and neoplastic nodule was reported to be found in
6 both colony controls (1/99 hepatocellular carcinoma and 0/99 neoplastic nodule in male rats and
7 0/98 hepatocellular carcinoma and 2/98 neoplastic nodules in female rats) and carbon-
8 tetrachloride-treated rats. Hepatic adenomas were included in the description of neoplastic
9 nodules in this study with the diagnosis of hepatocellular carcinoma to be "based on the presence
10 of less organized architecture and more variability in the cells comprising the neoplasms."
11 The authors reported that "increased mortality in treated male mice appears to be related
12 to the presence of liver tumors." For mice both male and female mice the incidences of
13 hepatocellular carcinoma were reported to be high from TCE treatment with 1/20 in age matched
14 controls, 26/50 in low dose and 31/48 in high dose males. Colony controls for male mice were
15 reported to be 5/77 for vehicle and 5/70 for untreated mice. For females mice hepatocellular
16 carcinomas were reported to be observed in 0/20 age matched controls, 4/50 low dose, and
17 11/47 high-dose female mice. Colony controls for female mice were reported to be 1/80 for
18 vehicle and 2/75 for untreated mice. In male mice, hepatocellular carcinomas were reported to
19 be observed early in the study with the first seen at 27 weeks. Hepatocellular carcinomas were
20 not observed so early in low dose male or female mice.
21 The diagnosis of hepatocellular carcinoma was reported to be based on histologic
22 appearance and the presence of metastasis especially to the lung with not other lesions
23 significantly elevated in treated mice. The tumors were reported to be
24
25 varied from those composed of well differentiated hepatocytes in a relatively
26 uniform trabecular arrangement to rather anaplastic lesions in which mitotic
27 figures occurred in cells which varied greatly in size and tinctorial characteristics.
28 Many of the tumors were characterized by the formation of relatively discrete
29 areas of highly anaplastic cells within the tumor proper which were, in turn,
30 surrounded by relatively well differentiated neoplastic cells. In general, various
31 arrangements of the hepatocellular carcinoma occurred, as described in the
32 literature, including those with an orderly cord-like arrangement of neoplastic
33 cells, those with a pseudoglandular pattern resembling adenocarcinoma, and those
34 composed of sheets of highly anaplastic cells with minimal cord or gland-like
35 arrangement. Multiple metaplastic lesions were observed in the lung, including
36 several neoplasms which were differentiated and relative benign in appearance."
37 The authors noted that almost all mice treated with carbon tetrachloride exhibited
38 liver tumors and that the "neoplasms occurring in treated [sic carbon tetrachloride
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-104 DRAFT—DO NOT CITE OR QUOTE
-------
1 treated] mice were similar in appearance to those noted in the trichloroethylene-
2 treated mice.
3
4 Thus, phenotypically this study reported that the liver tumors induced in mice by TCE were
5 heterogeneous and typical of those arising after carbon tetrachloride administration. The
6 descriptions of liver tumors in this study and the tendency of metastasis to the lung are similar to
7 the descriptions provided by Maltoni et al. (1986) for TCE-induced liver tumors in mice via
8 inhalation.
9 In terms of noncancer pathology of the liver, 1 control male rat was reported to display
10 fatty metamorphosis of the liver at 102 weeks. However, for the low dose, 3 male rats were
11 reported to display fatty metamorphosis (90, 110, and 110 weeks), 2 rats to display cystic
12 inflammation (76, 110 weeks), and one rat to display general inflammation (110 weeks). At the
13 high dose, 6 rats were reported to display fatty metamorphosis (12, 35, 49, 52, 52, and
14 58 weeks), 1 rat was reported to display cytomegaly (42 weeks), 2 rats were reported to display
15 centrilobular degeneration (53 and 58 weeks), 1 rats to display diffuse inflammation (62 weeks),
16 1 rat to display congestion (Week 12), and 5 rats to display angiectasis or abnormally enlarged
17 blood vessels which can be manifested by hyperproliferation of endothelial cells and dilatation of
18 sinusoidal spaces (35, 42, 52, 54, and 65 weeks). One control female rat was reported o display
19 fatty metamorphosis of the liver at 110 weeks, and one control female rats to display
20 "inflammation" of the liver at 110 weeks. Of the TCE dosed female rats, only 1 high dose
21 female rat displayed fatty metaphorphosis at Week 96. Thus, for male rats, there was liver
22 pathology present in some rats due to TCE exposure examined from 12 weeks to a year at their
23 time of their premature death. For mice the liver pathology was dominated by the presence of
24 hepatocellular carcinoma with additional hyperplasia noted in 2 mice of the high dose male and
25 female groups and 1 or less mouse exhibiting hyperplasia in the control or low-dose groups.
26 The authors note that "while the absence of a similar effect in rats appears most likely
27 attributable to a difference in sensitivity between the Osborne-Mendel rat and B6C3F1 mouse,
28 the early mortality of rats due to toxicity must also be considered." The conclude that "the test in
29 rats is inconclusive: large numbers of rats died prior to planned termination; in addition, the
30 response of this rat strain to the hepatocarcinogenicity of the positive control compound, carbon
31 tetrachloride, appeared relatively low." Finally, the authors note that "while the results obtained
32 in the present bioassay could possibly have been influenced by an impurity in the TCE used, the
33 extremely low amounts of impurities found make this improbable."
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-105 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.2.20. Herren-Freund et al, 1987
2 This study was gave results primarily in initiated male B6C3 Fl mice that were also
3 exposed to TCE metabolites in drinking water for 61 weeks. However, in Table 1 of the report,
4 results were given for mice that received no initiator but were given 40 mg/L TCE or 2 g/L NaCl
5 as control. The mice were reported to be 28 days of age when placed on drinking water
6 containing TCE. The authors reported that concentrations of TCE fell by about l/2 at the 40 mg/L
7 dose of TCE during the twice a week change in drinking water solution. For control animals
8 (n = 22) body weight at termination was reported to be 32.93 ± 0.54 g, and liver weight was
9 1.80 ± 0.05 g, percent liver/body weight was 5.47% ± 0.16%. For TCE treated animals (n = 32),
10 body weight at termination was reported to be 35.23 ± 0.66 g, and liver weight was
11 1.97 ± 0.10 g, percent liver/body weight was 5.57% ± 0.24%. Thus, hepatomegaly was not
12 reported for this paradigm at this time of exposure. The study reported that for 22 control
13 animals, the prevalence of adenomas was 2/22 animals (or 9%) with the mean number of
14 adenomas per animal to be 0.09 ± 0.06 (SEM). The prevalence of carcinomas in the control
15 group was reported to be 0/22. For 32 animals exposed to 40 mg/L TCE, the prevalence of
16 adenomas was 3/32 animals (or 9%) with the mean number of adenomas per animal to be
17 0.19 ± 0.12 (SEM). The prevalence of animals with hepatocellular carcinomas was 3/32 animals
18 (or 9%) with the mean number of hepatocellular carcinomas to be 0.10 ± 0.05 (SEM). Thus,
19 similar to the acute study of Tucker et al. (1982), significant loss of TCE is a limitation for trying
20 to evaluate TCE hazard in drinking water. However, despite difficulties in establishing
21 accurately the dose received, an increase in adenomas per animal and an increase in the number
22 of animals with hepatocellular carcinomas were reported to be associated with TCE exposure
23 after 61 weeks of exposure. Also of note is that the increase in tumors was reported without
24 significant increases in hepatomegaly at the end of exposure. The authors did not report these
25 increases in tumors as being significant but did not do a statistical test between TCE exposed
26 animals without initiation and control animals without initiation. The low numbers of animal
27 tested limits the statistical power to make such a determination. However, for carcinomas, there
28 was none reported in controls but 9% of TCE-treated mice had hepatocellular carcinomas.
29
30 E.2.2.21. Anna et al, 1994
31 The report focused on presenting incidence of cancer induction after exposure to TCE or
32 its metabolites and included a description of results for male B6C3F1 mice (8 weeks old at the
33 beginning of treatment) receiving 800 mg/kg/d TCE via gavage in corn oil, 5 days/week for
34 76 weeks. There was very limited reporting of results other than tumor incidence. There was no
35 reporting of liver weights at termination of the experiment. Although the methods section of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-106 DRAFT—DO NOT CITE OR QUOTE
-------
1 report gives 800 mg/kg/d as the exposure level, Table 1 in the results section reports that TCE
2 was administered at 1,700 mg/kg/d. This could be a typographical error in the table as a
3 transposition with the dose of "perc" administered to other animals in the same study. The
4 methods section of the report states that the authors based their dose in mice that used in the
5 1990 NTP study. The NTP study only used al,000 mg/kg/d in mice suggesting that the table is
6 mislabeled and suggests that the actual dose is 800 mg/kg/d in the Anna et al. (1994) study. All
7 treated mice were reported to be alive after 76 weeks of treatment. For control animals,
8 10 animals exposed to corn oil, and 10 untreated controls were killed in a 9-day period. The
9 remaining controls were killed at 96, 103, 134 weeks of treatment. Therefore, the control group
10 (all) contains a mixed group of animals that were sacrificed from 76-134 weeks and were not
11 comparable to the animals sacrificed at 76 weeks. At 76 weeks 3 of 10 the untreated and two of
12 the 10 corn oil treated controls were reported to have one small hepatocellular adenoma. None
13 of the controls examined at 76 weeks were reported to have any observed hepatocellular
14 carcinomas. The authors reported no cytotoxicity for TCE, corn oil, and untreated control group.
15 At 76 weeks, 75 mice treated with 800 mg/kg/d TCE were reported to have a prevalence of
16 50/75 animals having adenomas with the mean number of adenomas per animal to be 1.27 ± 0.14
17 (SEM). The prevalence of carcinomas in these same animals was reported to be 30/70 with the
18 mean number of hepatocellular carcinomas per animal to be 0.57 ± 0.10 (SEM). Although not
19 comparable in terms of time till tumor observation, Corn oil control animals examined at much
20 later time points did not have as great a tumor response as did those exposed to TCE. At
21 76-134 weeks 32 mice treated with corn oil were reported to have a prevalence of 4/32 animals
22 having adenomas with the mean number of adenomas per animal to be 0.13 ± 0.06 (SEM). The
23 prevalence of carcinomas in these same animals was reported to be 4/32 with the mean number
24 of hepatocellular carcinomas per animal to be 0.12 ± 0.06 (SEM). Despite only examining one
25 exposure level of TCE and the limited reporting of findings other than incidence data, this study
26 also reported that TCE exposure in male B6C3F1 mice to be associated with increased induction
27 of adenomas and hepatocellular carcinoma, without concurrent cytotoxicity.
28 In terms of liver tumor phenotype, Anna et al. reported the percent of H-ras codon 61
29 mutations in tumors from concurrent control animals (water and corn oil treatment groups
30 combined) examined in their study, historical controls in B6C3 Flmice, and in tumors from TCE
31 or DCA (0.5% in drinking water) treated animals. From their concurrent controls they reported
32 that H-ras codon 61 mutations in 17% (n = 6) of adenomas and 100% (n = 5) of carcinomas. For
33 historical controls (published and unpublished) they reported mutations in 73% (n = 33) of
34 adenomas and mutations in 70% (n = 30) of carcinomas. For tumors from TCE treated animals
35 they reported mutations in 35% (n = 40) of adenomas and 69% (n = 36) of carcinomas, while for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-107 DRAFT—DO NOT CITE OR QUOTE
-------
1 DCA treated animals they reported mutations in 54% (n = 24) of adenomas and in 68% (n = 40)
2 of carcinomas. The authors reported that "in this study, the H-ras codon 61 mutation frequency
3 was not statistically different in liver tumors from dichloroacetic acid and trichloroethylene-
4 treated mice and combined controls (62%, 51% and 69%, respectively)." In regard to mutation
5 spectra in H-ras oncogenes detected B6C3F1 mouse liver "tumors," the authors reported
6 combined results for concurrent and historical controls of 58% AAA, 27% CGA, and 14% CTA
7 substitutions for CAA at Codon 61 out of 58 mutations. For TCE "tumors" the substitution
8 pattern was reported to be 29% AAA, 24% CGA, and 40% CTA substitutions for CAA at Codon
9 61 out of 39 mutations and for DCA 28% AAA, 35% CGA, and 38% CTA substitutions for
10 CAA at Codon 61 out of 40 mutations.
11
12 E.2.2.22. Bull et al, 2002
13 This study primarily presented results from exposures to TCE, DCA, TCA and
14 combinations of DCA and TCA after 52 weeks of exposure with some animals examined at
15 87 weeks. It only examined and described results for liver. In a third experiment, 1,000 mg/kg
16 TCE was administered once daily 7 days a week for 79 weeks in 5% alkamuls in distilled water
17 to 40 B6C3F1 male mice (6 weeks old at the beginning of the experiment). At the time of
18 euthanasia, the livers were removed, tumors identified, and the tissues section of for examination
19 by a pathologist and immunostaining. Liver weights were not reported. For the TCE gavage
20 experiment there were 6 gavage-associated deaths during the course of this experiment among a
21 total of 10 animals that died with TCE treatment. No animals were lost in the control group.
22 The limitations of this experiment were discussed in Caldwell et al. (2008b). Specifically, for
23 the DCA and TCA exposed animals, the experiment was limited by low statistical power, a
24 relatively short duration of exposure, and uncertainty in reports of lesion prevalence and
25 multiplicity due to inappropriate lesions grouping (i.e., grouping of hyperplastic nodules,
26 adenomas, and carcinomas together as "tumors"), and incomplete histopatholology
27 determinations (i.e., random selection of gross lesions for histopathology examination). For the
28 reported TCE results, Bull et al. (2002) reported a high prevalence (23/36 B6C3F1 male mice) of
29 adenomas and hepatocellular carcinoma (7/36) and gave results of an examination of
30 approximately half of the lesions induced by TCE exposure. Tumor incidence data were
31 provided for only 15 control mice and reported as 2/15 (13%) having adenomas and 1/15 (7%)
32 carcinomas. Thus, this study presents results that are consistent with other studies of chronic
33 exposure that show TCE induction of hepatocellular carcinoma in male B6C3F1 mice.
34 For determinations of immunoreactivity to c-Jun as a marker of differences in "tumor"
35 phenotype, Bull et al. (2002) did include all lesions in most of their treatment groups, decreasing
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-108 DRAFT—DO NOT CITE OR QUOTE
-------
1 the uncertainty of his findings. The exceptions were the absence of control lesions and inclusion
2 of only 16/27 and 38/72 lesions for 0.5 g/L DC A + 0.05 g/L TCA and 1 g/kg/day TCE exposure
3 groups, respectively. Immunoreactivity results were reported for the group of hyperplastic
4 nodules, adenomas, and carcinomas. Thus, changes in c-Jun expression between the differing
5 types of lesions were not determined. Bull et al. (2002) reported lesion reactivity to c-Jun
6 antibody to be dependent on the proportion of the DCA and TCA administered after 52 weeks of
7 exposure. Given alone, DCA produced lesions in mouse liver for which approximately half
8 displayed a diffuse immunoreactivity to a c-Jun antibody, half did not, and none exhibited a
9 mixture of the two. After TCA exposure alone, no lesions were reported to be stained with this
10 antibody. When given in various combinations, DCA and TCA coexposure induced a few
11 lesions that were only c-Jun+, many that were only c-Jun-, and a number with a mixed
12 phenotype whose frequency increased with the dose of DCA. For TCE exposure of 79 weeks,
13 TCE-induced lesions also had a mixture of phenotypes (42% c-Jun+, 34% c-Jun-, and 24%
14 mixed) and were most consistent with those resulting from DCA and TCA coexposure but not
15 either metabolite alone.
16 Mutation frequency spectra for the H-ras codon 61 in mouse liver "tumors" induced by
17 TCE (n = 37 tumors examined) were reported to be significantly different than that for TCA
18 (n = 41 tumors examined), with DCA-treated mice tumors giving an intermediate result
19 (n = 64 tumors examined). In this experiment, TCA-induced "tumors" were reported to have
20 more mutations in codon 61(44%) than those from TCE (21%) and DCA (33%). This frequency
21 of mutation in the H-ras codon 61 for TCA is the opposite pattern as that observed for a number
22 of peroxisome proliferators in which the mutation spectra in tumors has been reported to be
23 much lower than spontaneously arising tumors (see Section E.3.4.1.5). Bull et al. (2002) noted
24 that the mutation frequency for all TCE,TCA or DCA was lower in this experiment than for
25 spontaneous tumors reported in other studies (they had too few spontaneous tumors to analyze in
26 this study), but that this study utilized lower doses and was of shorter duration than that of
27 Ferreira-Gonzalez et al. (1995). These are additional concerns along with the effects of lesion
28 grouping in which a lower stage of progression is group with more advanced stages. In a limited
29 subset of tumor that were both sequenced and characterized histologically, only 8 of 34 (24%)
30 TCE-induced adenomas but 9/15 (60%) of TCE-induced carcinomas had mutated H-ras at codon
31 61, which the authors suggest is evidence that this mutation is a late event.
32 The issues involving identification of MO A through tumor phenotype analysis are
33 discussed in detail below for the more general case of liver cancer as well as for specific
34 hypothesized MO As (see Sections E.3.1.4, E.3.1.8, E.3.2.1, and E.3.4.1.5). In an earlier paper,
35 Bull (2000) suggested that "the report by Anna et al (1994) indicated that TCE-induced tumors
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-109 DRAFT—DO NOT CITE OR QUOTE
-------
1 possessed a different mutation spectra in codon 61 of the H-ras oncogene than those observed in
2 spontaneous tumors of control mice." Bull (2000) stated that "results of this type have been
3 interpreted as suggesting that a chemical is acting by a mutagenic mechanism" but went on to
4 suggest that it is not possible to a priori rule out a role for selection in this process and that
5 differences in mutation frequency and spectra in this gene provide some insight into the relative
6 contribution of different metabolites to TCE-induced liver tumors. Bull (2000) noted that data
7 from Anna et al. (1994), Ferreira-Gonzalez et al. (1995), and Maronpot et al. (1995) indicated
8 that mutation frequency in DCA-induced tumors did not differ significantly from that observed
9 in spontaneous tumors, that the mutation spectra found in DCA-induced tumors has a striking
10 similarity to that observed in TCE-induced tumors, and DCA-induced tumors were significantly
11 different than that of TCA-induced liver tumors. What is clear from these observations is the
12 phenotype of TCE-induced tumors appears to be more like DCA-induced tumors (which are
13 consistent with spontaneous tumors), or those resulting from a coexposure to both DCA and
14 TCA, than from those induced by TCA. More importantly, these data suggest that using
15 measures other than dysplasticity and tincture indicate that mouse liver tumors induced by TCE
16 are heterogeneous in phenotype. The descriptions of tumors in mice reported by the NTP and
17 Maltoni et al studies are also consistent with phenotypic heterogeneity as well as consistency
18 with spontaneous tumor morphology.
19
20 E.2.3. Mode of Action: Relative Contribution of Trichloroethylene (TCE) Metabolites
21 Several metabolites of TCE have also been shown to induce liver cancer in rodents with
22 DCA and TCA having been the focus of study as potential active agent(s) of TCE liver toxicity
23 and/or carcinogenesis and both able to induce peroxisome proliferation (Caldwell and Keshava,
24 2006). A variety of DCA effects from exposure have been noted that are consistent with
25 conditions that increase risk of liver cancer (e.g., effects on the cytosolic enzyme glutathione
26 [GST]-S-transferase-zeta, diabetes, and glycogen storage disease), with the pathological changes
27 induced by DCA on whole liver consistent with changes observed in preneoplastic foci from a
28 variety of agents (Caldwell and Keshava, 2006). Chloral hydrate (CH) is one of the first
29 metabolites from oxidative metabolism of TCE with a large fraction of TCE metabolism
30 appearing to go through CH and then subsequent metabolism to TCA and trichloroethanol (Chiu
31 et al., 2006b). Similarities in toxicity may indicate that common downstream metabolites may
32 be lexicologically important, and differences may indicate the importance of other metabolic
33 pathways.
34 Although both induce liver tumors, DCA and TCA have distinctly different actions
35 (Keshava and Caldwell, 2006) and apparently differ in tumor phenotype (see discussion above in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-110 DRAFT—DO NOT CITE OR QUOTE
-------
1 Section E.2.2.8) and many studies have been conducted to try to elucidate the nature of those
2 differences (Caldwell et al., 2008b). Limitations of all of the available chronic studies of TCA
3 and most of the studies of DC A include less than lifetime exposures, varying and small numbers
4 of animals examined, and few exposure concentrations that were relatively high.
5
6 E.2.3.1. Acute studies of Dichloroacetic Acid (DCA)/Trichloroacetic Acid (TCA)
1 The studies in this section focus on studies of DCA and TCA that examine, to the extent
8 possible, similar endpoints using similar experimental designs as those of TCE examined above
9 and that give insight into proposed MO As for all three. Of note for any experiment involving
10 TCA, is whether exposure solutions were neutralized. Unbuffered TCA is commonly used as a
11 reagent to precipitate proteins so that any result from studies using unbuffered TCA could
12 potentially be confounded by the effects on pH.
13
14 E.2.3.1.1. Sanchez and Bull, 1990. In this report TCA and DCA were administered to male
15 B6C3F1 mice (9 weeks of age) and male and female Swiss-Webster mice (9 weeks of age) for
16 up to 14 days. At 2, 4, or 14 days, mice were injected with tritiated thymidine. Experiments
17 were replicated at least once but results were pooled so that variation between experiments could
18 not be determined. B6C3F1 male mice were given DCA or TCA at 0, 0.3 g/L, 1.0 g/L, or
19 2.0 g/L in drinking water (n = 4 for each group for 2 and 5 days, but n = 15 for control and
20 n = 12 for treatment groups at Day 14). Swiss-Webster mice (n = 4) at were exposed to DCA
21 only on Day 14 at 0, 1.0 or 2.0 g/L. Mice were injected with tritiated thymidine 2 hours prior to
22 sacrifice. The pH of the drinking water was adjusted to 6.8-7.2 with sodium hydroxide.
23 Concentrations of TCA and DCA were reported to be stable for a minimum of 3 weeks.
24 Hepatocyte diameters were reported to be determined by randomly selecting 5 different high
25 power fields (400x) in five different sections per animals (total of 25 fields/animal with "cells in
26 and around areas of necrosis, close to the edges of the section, or displaying mitotic figures were
27 not included in the cell diameter measurements." PAS staining was reported to be done for
28 glycogen and lipofuscin determined by autofluorescence. Tritiated thymidine was reported to be
29 given to the animals 2 hours prior to sacrifice. In 2 of 3 replications of the 14-day experiment, a
30 portion of the liver was reported to be set aside for DNA extraction with the remaining group
31 examined autoradiographically for tritiated thymidine incorporation into individual hepatocytes.
32 Autoradiographs were also reported to be examined in the highest dose of either DCA or TCA
33 for the 2- and 5-day treatment groups. Autoradiographs were reported to be analyzed in
34 randomly selected fields (5 sections per animal in 10 different fields) for a total of
35 50 fields/animal and reported as percentage of cells in the fields that were labeled. There was no
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-111 DRAFT—DO NOT CITE OR QUOTE
-------
1 indication by the authors that they characterized differing zones of the liver for preferential
2 labeling. DNA thymidine incorporation results were not examined in the same animals as those
3 for individual hepatoctye incorporation and also not examined at 2- or 5-day time periods. The
4 only analyses reported for the Swiss-Webster mice were of hepatic weight change and
5 histopathology. Variations in results were reported as standard error of the mean.
6 Liver weights were reported but not body weights so the relationship of liver/body weight
7 ratio could not be determined for the B6C3F1 mice. For liver weight, the numbers of animals
8 examined varied greatly between and within treatment groups. The number of control animals
9 examined were reported to be n = 4 on Day 2, n = 8 on day 5 and n = 15 on Day 14. There was
10 also a large variation between control groups in regard to liver weight. Control liver weights for
11 Day 2 were reported to be 1.3 ± 0.1, Day 5 to be 1.5 ±0.05 and for Day 14 to be 1.3 ± 0.04 g.
12 Liver weights in Day 5 control animals were much greater than those for Day 2 and Day 14
13 animals and thus, the means varied by as much as 15%. For DCA, there was no reported change
14 in liver weights compared to controls values at any exposure level of DCA after 2 days of
15 exposure. After 5 days of exposure there was no difference in liver weight between controls and
16 0.3 g/L exposed animals. However, the animals exposed at 1.0 or 2.0 g/L DCA had identical
17 increases in liver weight of 1.7 ± 0.13 and 1.7 ± 0.8 g, respectively. Due to the low power of the
18 experiment, only the 2.0 g/L DCA result was identified by the authors as significantly different
19 from the control value. For TCA there was a slight decrease reported between control values and
20 the 0.3 g/L treatment group (1.2 ± 0.1 g vs. 1.3 ± 0.1 g) but the 1.0 and 2.0 g/L treatment groups
21 had similar slight increases over control (for 1.0 g/L liver weight was 1.5 ± 0.1 and for 2.0 g/L
22 liver weight was 1.4 ± 0.1 g). The same pattern was apparent for the 5-day treatment groups for
23 TCA as for the 2-day treatment groups.
24 For 14 days exposure periods the number of animals studied was increased to!2 for the
25 TCA and DCA treatment groups. After 14 days of DCA treatment, there was a reported dose-
26 related increase in liver weight that was statistically significant at the two highest doses (i.e., at
27 0.3 g/L DCA liver weight was 1.4 ± 0.04, at 1.0 g/L DCA liver weight was 1.7 ± 0.07 g, and at
28 2.0 g/L DCA liver weight was 2.1 ±0.08 g). This was 1.08-, 1.31-, and 1.62-fold of controls,
29 respectively. After 14 days of TCA exposure there was a dose-related increase in liver weight
30 that the authors reported to be statistically significant at all exposure levels (i.e., at 0.3 g/L liver
31 weight was 1.5 ± 0.06, at 1.0 g/L liver weight was 1.6 ± 0.07 g, and at 2.0 g/L liver weight was
32 1.8 ± 0.10 g). This represents 1.15-, 1.23-, and 1.38-fold of control. The authors note that at
33 14 days that DCA-associated increases in hepatic liver weight were greater than that of TCA.
34 What is apparent from these data are that while the magnitude of difference between the
35 exposures was ~6.7-fold between the lowest and highest dose, the differences between TCA
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-l 12 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposure groups for change in liver weight was -2.5. For DC A the slope of the dose-response
2 curve for liver weight increases appeared to be closer to the magnitude of difference in exposure
3 concentrations between the groups (i.e., a difference of 7.7-fold between the highest and lowest
4 dose for liver weight induction). Given that the control animal weights varied as much as 15%,
5 the small number of animals examined, and that body weights were also not reported, there are
6 limitations for making quantitative comparisons between TCA and DCA treatments. However,
7 after 14 days of treatment it is apparent that there was a dose-related increase in liver weight
8 after either DCA or TCA exposure at these exposure levels. For male and female Swiss-Webster
9 mice 1 g/L and 2 g/L DCA treatment (n = 4) was reported to also induce an increase in percent
10 liver/body weight that was similar to the magnitude of exposure difference (see below).
11 Grossly, livers of B6C3F1 mice treated with DCA for 1 or 2 g/L were reported to have
12 "pale streaks running on the surface" and occasionally, discrete, white, round areas were also
13 observed on the surface of these livers. Such areas were not observed in TCA-treated or control
14 B6C3F1 mice. Swiss-Webster mice were reported to have "dose-related increases in hepatic
15 weight and hepatic/body weight ratios were observed. DCA-associated increases in relative
16 hepatic weights in both sexes were comparable to those in B6C3F1 mice. Pale streaks on the
17 surface of the liver were not observed in Swiss-Webster mice. Again there was no significant
18 effect on total body or renal weights (data not shown)." The authors report liver weights for the
19 Swiss-Webster male mice (n = 4 for each group) to be 2.1 ± 0.1 g for controls, 2.1 ± 0.1 g for
20 1.0 g/L DCA and 2.4 ± 0.2 g for 2.0 g/L DCA 14-day treatment groups. The percent liver/body
21 weights for these same groups were reported to be 6.4% ± 0.4%, 6.9% ± 0.2%, and 8.1% ± 0.3%,
22 respectively. For female Swiss-Webster mice (n = 4 for each group) the liver weights were
23 reported to be 1.1 ± 0.1 g for controls, 1.5 ± 0.1 g for 1.0 g/L DCA and 1.7 ± 0.2 g for 2.0 g/L
24 DCA 14-day treatment groups. The percent liver/body weights for these same groups of Swiss
25 mice were reported to be 4.8% ± 0.2%, 6.0% ± 0.2%, and 6.8% ± 0.4%, respectively. Thus,
26 while there was no significant difference in "liver weight" between the control and the 1.0 g/L
27 DCA treatment group for male or female Swiss-Webster mice, there was a statistically
28 significant difference in liver/body weight ratio reported by the authors. These data, illustrate the
29 importance of reporting both measures and the limitations of using small numbers of animals
30 (n = 4 for the Swiss Webster vs. n = 12-14 for B6C3F1 14-days experiments). Relative liver
31 weights were reported by the authors for male B6C3F1 mice only for the 14-day groups, as a
32 function of calculated mean water consumption, as pooled data from the three experiments, and
33 as a figure that was not comparable to the data reported for Swiss-Webster mice. The liver
34 weight data indicate that male mice of the same age appeared to differ in liver weight between
35 the two strains without treatment (i.e., male B6C3F1 mice had control liver weights at 14 days of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-113 DRAFT—DO NOT CITE OR QUOTE
-------
1 1.3 ± 0.04 g for 15 mice, while Swiss-Webster mice had control values of 2.1 ±0.1 for 4 mice).
2 While the authors report that results were "comparable" between the B6C3F1 mice in regard to
3 DCA-induced changes in liver weight, the increase in percent liver/body weight ratios were
4 1.27-fold of control for Swiss-Webster male mice (n = 4) and 1.42-fold of control for female
5 while the increase in liver weight for B6C3F1 male mice (n = 12-14) was 1.62-fold of controls
6 after 14 days of exposure to 2 g/L DC A.
7 The concentration of DNA in the liver was reported as mg hepatic DNA/g of liver. This
8 measurement can be associated with hepatocellular hypertrophy when decreased, or increased
9 cellularity (of any cell type), increased DNA synthesis, and/or increased hepatocellular ploidy in
10 the liver when increased. The number of animals examined for this parameter varied. For
11 control animals there were 4 animals reported to be examined at 2 days, 8 animals examined at
12 5 days, and at 14 days 8 animals were examined. The mean DNA content in control livers were
13 not reported to vary greatly, however, and the variation between animals was relatively low in
14 the 5- and 14-day control groups (i.e., 1.67 ± 0.27 mg DNA/g, 1.70 ± 0.05 mg DNA/g, and
15 1.69 mg DNA/g, for 2-, 5-, or 14-day control animals, respectively). For treatment groups the
16 number of animals reported to be examined appeared to be the same as the control animals. For
17 DC A treatment there did not appear to be a dose-response in hepatic DNA content with the 1 g/L
18 exposure level having the same reported value as control but the 0.3 g/L and 2.0 g/L values
19 reported to be lower (mean values of 1.49 and 1.32 mg DNA/g, respectively). After 5 days of
20 exposure, all treatment groups were reported to have a lower DNA content that the control value
21 (i.e., 1.44 ± 0.06 mg DNA/g, 1.47 ± mg DNA/g, and 1.30 ± 0.14 mg DNA/g, for 0.3, 1.0, and
22 2.0 g/L exposure levels of DCA, respectively). After 14 days of exposure, there was a reported
23 increase in hepatic DNA at the 0.3 g/L exposure level but significant decreases at the 1.0 g/L and
24 2.0 g/L exposure levels (i.e., 1.94 ± 0.20 mg DNA/g, 1.44 ± 0.14 mg DNA/g, and 1.19 ± 0.16 mg
25 DNA/g for the 0.3, 1.0, and 2.0 g/L exposure levels of DCA, respectively). Changes in DNA
26 concentration in the liver were not correlated with the pattern of liver weight increases after
27 DCA treatment. For example, while there was a clear dose-related increase in liver weight after
28 14 days of DCA treatment, the 0.3 g/L DCA exposed group was reported to have a higher rather
29 than lower level of hepatic DNA than controls. After 2 or 5 days of DCA treatment, liver
30 weights were reported to be the same between the 1.0 and 2.0 g/L treatment groups but hepatic
31 DNA was reported to be decreased.
32 For TCA, there appeared to be a dose-related decrease in reported hepatic DNA after
33 2 days of treatment (i.e., 1.63 ± 0.07 mg DNA/g, 1.53 ± 0.08 mg DNA/g, and 1. 43 ± 0.04 mg
34 DNA/g for the 0.3 g/L, 1.0 g/L, and 2.0 g/L exposure levels of TCA, respectively). After 5 days
35 of TCA exposure there was a reported decrease in hepatic DNA for all treatment groups that was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-l 14 DRAFT—DO NOT CITE OR QUOTE
-------
1 similar at the 1.0 g/L and 2.0 g/L exposure groups (i.e., 1.45 ± 0.17 mg DNA/g, 1.29 ± 0.18 mg
2 DNA/g, and 1. 26 ± 0.22 mg DNA/g for the 0.3 g/L, 1.0 g/L, and 2.0 g/L exposure levels of
3 TCA, respectively). After 14 days of TCA treatment, there was a reported decrease in all
4 treatment groups in hepatic DNA content that did not appear to be dose-related (i.e.,
5 1.31±0.17 mg DNA/g, 1.21 ± 0.17 mg DNA/g, and 1. 33 ± 0.18 mg DNA/g for the 0.3 g/L,
6 1.0 g/L, and 2.0 g/L exposure levels of TCA, respectively). Thus, similar to the results reported
7 for DCA, the patterns of liver weight gain did not match those of hepatic DNA decrease for TCA
8 treated animals. For example, although there appeared to be a dose-related increase in liver
9 weight gain after 14 days of TCA exposure, there was a treatment but not dose-related decrease
10 in hepatic DNA content.
11 In regard to the ability to detect changes, the low number of animals examined after
12 2 days of exposure (n = 4) limited the ability to detect a significant change in liver weight and
13 hepatic DNA concentration. For hepatic DNA determinations, the larger number of animals
14 examined at 5 and 14 day time points and the similarity of values with relatively smaller standard
15 error of the mean reported in the control animals made quantitative differences in this parameter
16 easier to determine. However, animals varied in their response to treatment and this variability
17 exceeded that of the control groups. For DCA results reported at 14 days and those for TCA
18 reported at 5 and 14 days, the standard errors for treated animals showed a much greater
19 variability than those of the control animals (range of 0.04-0.05 mg DNA/g for control groups,
20 but ranges of 0.17 to 0.22 mg DNA/g for TCA at 5 days and 0.14 to 0.20 mg DNA/g for DCA or
21 TCA at 14 days). The authors stated that
22
23 the increases in hepatic weights were generally accompanied by decreases in the
24 concentration of DNA. However, the only clear changes were in animals treated
25 with DCA for 5 or 14 days where the ANOVAs were clearly significant (P<0.020
26 and 0.005, respectively). While changes of similar magnitude were observed in
27 other groups, the much greater variation observed in the treated groups resulted in
28 not significant differences by ANOVA ( p = 0.41, 0.66. 0.26, 0.15 for DCA - 2
29 days, and TCA for 2,5, and 14 days, respectively).
30
31 The size of hepatocytes is heterogeneous and correlated with its ploidy, zone, and age of
32 the animal (see Section E. 1.1 above). The authors do not indicate if there was predominance in
33 zone or ploidy for hepatocytes included in their analysis of average hepatocyte diameter in the
34 random selection of 25 fields per animal (n = 3 to 7 animals). There appeared to be a dose-
35 related increase in cell diameter associated with DCA exposure and a treatment but not dose-
36 related increase with TCA treatment after 14 days of treatment. For control B6C3F1 male mice
37 (n = 7) the hepatocyte diameter was reported to be 20.6 ± 0.4 microns. For mice exposed to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-115 DRAFT—DO NOT CITE OR QUOTE
-------
1 DCA hepatocyte diameter was reported to be 22.2 ± 0.2, 25.2 ± 0.6, and 26.0 ± 1.0 microns for
2 0.3 g/L, 1.0 g/L, and 2.0 g/L treated mice (n = 4 for each group), respectively. For mice exposed
3 to TCA hepatocyte diameter was reported to be 22.2 ± 0.2, 22.4 ± 0.6, and 23.2 ± 0.4 microns for
4 0.3 g/L, 1.0 g/L, and 2.0 g/L treated mice (n = 4 for the 0.3 g/L and 1.0 g/L groups and n = 3 for
5 the 2.0 g/L group), respectively. The small number of animals examined limited the power of
6 the experiment to determine statistically significant differences with the authors reporting that
7 only the 1.0 g/L DCA and 2.0 g/L DCA and TCA treated groups statistically significant from
8 control values. The dose-related increases in reported cell diameter were consistent with the
9 dose-related increases in liver weight reported for DCA after 14 days of exposure. However, the
10 pattern for hepatic DNA content did not. For TCA, the dose-related increases in cell diameter
11 were also consistent with the dose-related increases in liver weight after 14 days of exposure.
12 Similar to DCA results, the changes in hepatic DNA content did not correlate with changes in
13 cell size. In regard to the magnitude of increases over control values, the 68 versus 38% increase
14 in liver weight for DCA versus TCA at 2.0 g/L, was less than the 26 and 13% increases in cell
15 diameter for the same groups, respectively. Therefore, for both DCA and TCA exposure there
16 appeared to be dose-related hepatomegaly and increased cell size after 14-days of exposure.
17 The authors reported PAS staining for glycogen content as an attempt to examine the
18 nature of increased cell size by DCA and TCA. However, they did not present any quantitative
19 data and only provided a brief discussion. The authors reported that
20
21 hepatic sections of DCA-treated B6C3F1 mice (1 and 2 g/L) contained very large
22 amounts of perilobular PAS-positive material within hepatocytes. PAS stained
23 hepatic sections from animals receiving the highest concentration of TCA
24 displayed a much less intense staining that was confined to periportal areas.
25 Amylase digesting confirmed the majority of the PAS-positive material to by
26 glycogen. Thus, increased hepatocellular size in groups receiving DCA appears
27 to be related to increased glycogen deposition. Similar increases in glycogen
28 deposition were observed in Swiss-Webster mice.
29
30 There is no way to discern whether DCA-induced glycogen deposition was dose-related and
31 therefore, correlated with increased liver weight and cell diameter. While the authors suggest
32 that Swiss-Webster mice displayed "similar increased in glycogen deposition" the authors did
33 not report a similar increase in liver weight gain after DCA exposure at 14 days (1.27-fold of
34 control percent liver/body weight ratio in Swiss male mice and 1.42-fold in female Swiss-
35 Webster mice vs. 1.62-fold of control in B6C3F1 mice after 14 days of exposure to 2 g/L DCA).
36 Thus, the contribution of glycogen deposition to DCA-induced hepatomegaly and the nature of
37 increased cell size induced by acute TCA exposure cannot be determined by this study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-116 DRAFT—DO NOT CITE OR QUOTE
-------
1 However, this study does show that DC A and TCA differ in respect to their effects on glycogen
2 deposition after short-term exposure and the data suggest that.
3 The authors report that
4
5 localized areas of coagulative necrosis were observed histologically in both
6 B6C3F1 and Swiss-Webster mice treated with DCA at concentrations of 1 and 2
7 g/L for 14 days. The necrotic areas corresponded to the pale streaked areas seen
8 grossly. These areas varied in size, shape and location within sections and
9 occupied up to several mm2. An acute inflammatory response characterized by
10 thin rims of neutrophils was associated with the necrosis, along with multiple
11 mitotic figures. No such areas of necrosis were observed in animals treated at
12 lower concentrations of DCA, or in animals receiving the chemical for 2 or 5
13 days. Mice treated with 2 g/L TCA for 14 days have some necrotic areas, but at
14 such low frequency that it was not possible to determine if it was treatment-
15 related (2 lesions in a total of 20 sections examined). No necrosis was observed
16 in animals treated at the lower concentrations of TCA or at earlier time points.
17
18 Again there were no quantitative estimates given of the size of necrotic areas, variation between
19 animals, variation between strain, or dose-response of necrosis reported for DCA exposure by
20 the authors. The lack of necrosis after 2 and 5 days of exposure at all treatment levels and at the
21 lower exposure level at 14 days of exposure is not correlated with the increases in liver weight
22 reported for these treatment groups.
23 Autoradiographs of randomly chosen high powered fields (400x) (50 fields/animal) were
24 reported as the percentage of cells in the fields that were labeled. There was significant variation
25 in the number of animals examined and in the reported mean percent of labeled cells between
26 control groups. The number of control animals was not given for the 2-day group but for the
27 5-day and 14 day groups were reported to be n = 4 and n = 11, respectively. The mean percent
28 of labeling in control animals was reported at 0.11 ± 0.03, 0.12 ± 0.04, and 0.46 ± 0.07% of
29 hepatocytes for 2-day, 5-day, and 14-day control groups, respectively. Only the 2.0 g/L
30 exposures of DCA and TCA were examined at all 3 times of exposure while all groups were
31 examined at 14 days. However, the number of animals examined in all treatment groups
32 appeared to be only 4 animals in each group. There was not an increase over controls reported in
33 the 2.0 g/L DCA or TCA 2- and 5-day exposure groups in hepatocyte labeling with tritiated
34 thymidine. After 14 days of exposure, there was a statistically significant but very small dose-
35 related increase over the control value after DCA exposure (i.e., 0.46% ± 0.07%,
36 0.64% ± 0.15%, 0.75% ± 0.22%, and 0.94% ± 0.05% labeling of hepatocytes in control, 0.3, 1.0,
37 and 2.0 g/L DCA treatment groups, respectively). For TCA, there was no change in hepatocyte
38 labeling except for a 50% decrease from control values at after 14 days of exposure to 2.0 g/L
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-117 DRAFT—DO NOT CITE OR QUOTE
-------
1 TCA (i.e., 0.46% ± 0.07%, 0.50% ± 0.14%, 0.52% ± 0.26%, and 0.26% ± 0.14% labeling of
2 hepatocytes in control, 0.3, 1.0, and 2.0 g/L TCA treatment groups, respectively). The authors
3 report that
4
5 labeled cells were localized around necrotic areas in these [sic DCA treated]
6 groups. Since counts were made randomly, the local increased in DCA-treated
7 animals at concentrations of 1 and 2 g/L are in fact much higher than indicated by
8 the data. Labeling indices in these areas of proliferation were as high as 30%.
9 Labeled hepatocytes in TCA-treated and the control animals were distributed
10 uniformly throughout the sections. There was an apparent decrease in the
11 percentage of labeled cells in the group of animals treated with the highest dose of
12 TCA. This is because no labeled cells were found in any of the fields examined
13 for one animal.
14
15 The data for control mice in this experiment is consistent with others showing that the liver is
16 quiescent in regard to hepatocellular proliferation with few cells undergoing mitosis (see
17 Section E. 1.1). For up to 14 days of exposure with either DCA or TCA, there is little increase in
18 hepatocellular proliferation except in instances and in close proximity to areas of proliferation.
19 The increases in liver weight reported for this study were not correlated with and cannot be a
20 result of hepatocellular proliferation as only a very small population of hepatocytes is
21 undergoing DNA synthesis. For TCA, there was no increase in DNA synthesis in hepatocytes,
22 even at the highest dose, as shown by autoradiographic data of tritiated thymidine incorporation
23 in random fields.
24 Whole liver sections were examined for tritiated thymidine incorporation from DNA
25 extracts. The number of animals examined varied (i.e., n = 4 for the 2-day exposure groups and
26 n = 8 for 5- and 14-day exposure groups) but the number of control animals examined were the
27 same as the treated groups for this analysis. The levels of tritiated thymidine incorporation in
28 hepatic DNA (dpm/mg DNA expressed as mean x 103 ± SE of n animals) were reported to be
29 similar across control groups (i.e., 56 ± 11, 56 ± 6, and 56 ± 7 dpm/mg DNA, for 2-, 5-, and
30 14-day treatment groups, respectively). After two days of DCA exposure, there appeared to be a
31 slight treatment-related but not dose-related increase in reported tritiated thymidine incorporation
32 into hepatic DNA (i.e., 72 ± 23, 80 ± 6, and 68 ± 7 dpm/mg DNA for 0.3, 1.0, or 2.0 g/L DCA,
33 respectively). After 5 days of DCA exposure, there appeared to be a dose-related increase in
34 reported tritiated thymidine incorporation into hepatic DNA (i.e., 68 ± 18, 110 ± 20, and
35 130 ± 7 dpm/mg DNA for 0.3, 1.0, or 2.0 g/L DCA, respectively). However, after 14 days of
36 DCA exposure, levels of tritiated thymidine incorporation were less than those reported at 5 days
37 and the level for the 0.3 g/L exposure group was less than the control value (i.e., 33 ± 11, 77 ± 9,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-118 DRAFT—DO NOT CITE OR QUOTE
-------
1 and 81 ± 12 dpm/mg DNA for 0.3,1.0, or 2.0 g/L DC A, respectively). After two days of TCA
2 exposure there did not appear to be a treatment-related increase in tritiated thymidine
3 incorporation into hepatic DNA (i.e., 82 ± 16, 52 ± 7, and 54 ± 7 dpm/mg DNA for 0.3, 1.0, or
4 2.0 g/L TCA, respectively). Similar to the reported results for DC A, after 5 days of TCA
5 exposure there appeared to be a dose-related increase in reported tritiated thymidine
6 incorporation into hepatic DNA (i.e., 79 ± 23, 86 ± 17, and 158 ± 33 dpm/mg DNA for 0.3, 1.0,
7 or 2.0 g/L TCA, respectively). After 14 days of TCA exposure there were treatment related
8 increases but not a dose-related increase in reported tritiated thymidine incorporation into hepatic
9 DNA (i.e., 71 ± 10, 73 ± 14, and 103 ± 14 dpm/mg DNA for 0.3, 1.0, or 2.0 g/L TCA,
10 respectively). It would appear that for both TCA and DC A the increase in tritiated thymidine
11 incorporation into hepatic DNA was dose related and peaked after 5 days of exposure. The
12 authors report that the decrease in incorporation into hepatic DNA observed after 14 days of
13 DCA treatment at 0.3 g/L to be statistically significant as well as the increases after 5 and
14 14 days of TCA exposure at the 2.0 g/L level. The small numbers of animals examined, the
15 varying number of animals examined, and the degree of variation in treatment-related effects
16 limits the statistical power of this experiment to detect quantitative changes.
17 Given the limitations of this experiment, determination of an accurate measure of the
18 quantitative differences in tritiated thymidine incorporation into whole liver DNA or that
19 observed in hepatocytes are hard to determine. In general the results for tritiated thymidine
20 incorporation into hepatic DNA were consistent with those for tritiated thymidine incorporation
21 into hepatocytes in that they show that there were at most a small population of hepatocytes
22 undergoing DNA synthesis after up to 14 days of exposure at relative high levels of exposure to
23 DCA and TCA (i.e., the largest percentage of hepatocytes undergoing DNA synthesis for any
24 treatment group was less than 1% of hepatocytes). The highest increases over control levels for
25 hepatic DNA incorporation for the whole liver were reported at the highest exposure level of
26 TCA treatment after 5 days of treatment (3-fold of control) and after 14 days of TCA treatment
27 (2-fold of control). Although the authors report small areas of focal necrosis with concurrent
28 localized increases in hepatocyte proliferation in DCA treated animals exposed tol .0 g/L and
29 2.0 g/L DCA, the levels of whole liver tritiated thymidine incorporation were only slightly
30 elevated over controls at these concentrations, and were decreased at the 0.3 g/L exposure
31 concentration for which no focal necrosis was reported. The whole liver DNA incorporation of
32 tritiated thymidine was not consistent with the pattern of tritiated thymidine incorporation
33 observed in individual hepatocytes. The authors state that "at present, the mechanisms for
34 increased tritiated thymidine uptake in the absence of increased rates of cell replication with
35 increasing doses of TCA cannot be determined." The authors do not discuss the possibility that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-119 DRAFT—DO NOT CITE OR QUOTE
-------
1 the difference in hepatocyte labeling and whole liver DNA tritiated thymidine incorporation
2 could have been due to the labeling representing increased polyploidization rather than cell
3 proliferation, as well as increased numbers of proliferating nonparenchymal and inflammatory
4 cells. The increased cell size due from TCA exposure without concurrent increased glycogen
5 deposition could have been indicative of increased polyploidization. Finally, although both
6 TCA- and DCA-induced increases in liver weight were generally consistent with cell size
7 increases, they were not correlated with patterns of change in hepatic DNA content,
8 incorporation of tritiated thymidine in DNA extracts from whole liver, or incorporation of
9 tritiated thymidine in hepatocytes. In regard to cell size, although increased glycogen deposition
10 with DC A exposure was noted by the authors of this study, lack of quantitative analyses of that
11 accumulation precludes comparison with DCA-induced liver weight gain.
12
13 E.2.3.1.2. Nelson et al, 1989. Nelson and Bull (1988) administered TCE (0, 3.9, 11.4, 22.9,
14 and 30.4 mmol/kg) in Tween 80® via gavage to male Sprague Dawley rats and male B6C3F1
15 mice, sacrificed them fours hours after treatment (n = 4-7), and measured the rate of DNA
16 unwinding under alkaline conditions. They assumed that this assay represented increases in
17 single-strand breaks. For rats there was little change from controls up to 11.4 mmol/kg (1.5 g/kg
18 TCE) but a significantly increased rate of unwinding at 22.9 and 30.4 mmol/kg TCE (~2-fold
19 greater at 30.4 mmol). For mice there was a significantly increased level of DNA unwinding at
20 11.4 and 22.9 mmol. Concentrations above 22.9 mmol/kg were reported to be lethal to the mice.
21 In this same study, TCE metabolites were administered in unbuffered solution using the same
22 assay. DCA was reported to be most potent in this assay with TCA being the lowest, while CH
23 closely approximated the dose-response curve of TCE in the rat. In the mouse the most potent
24 metabolite in the assay was reported to be TCA followed by DCA with CH considerably less
25 potent.
26 The focus of the Nelson et al. (1989) study was to examine whether reported single strand
27 breaks in hepatic DNA induced by DCA and TCA (Nelson and Bull, 1988) were secondary to
28 peroxisome proliferation also reported to be induced by both. Male B6C3F1 mice (25-30 g but
29 no age reported) were given DCA (10 mg/kg or 500 mg/kg) or TCA (500 mg/kg) via gavage in
30 1% aqueous Tween 80® with no pH adjustment. The animals were reported to be sacrificed 1, 2,
31 4, or 8 hours after administration and livers examined for single strand breaks as a whole liver
32 homogenate. In a separate experiment (experiment #2) treatment was parallel to the first
33 (500 mg/kg treatment of DCA or TCA) but levels of PCO activity were measured as an
34 indication of peroxisome proliferation and expressed as umol/min/g liver. In a separate
35 experiment (experiment #3) mice were administered 500 mg/kg DCA or TCA for 10 days with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-120 DRAFT—DO NOT CITE OR QUOTE
-------
1 Clofibrate administered at a dose of 250 mg/kg as a positive control. 24 hours after the last dose,
2 animals were killed and liver examined by light microscopy and PCO activity. Finally, in an
3 experiment parallel in design to experiment #3, single strand breaks were measured in total
4 hepatic DNA after 500 mg/kg exposure to TCA (experiment #4). Electron microscopy was
5 performed on 2 animals/group for vehicle, DCA or TCA treatment, with 6 randomly chosen
6 micrographic fields utilized for peroxisome profiles. These micrographs were analyzed without
7 identification as to what area of the liver lobules they were being taken from. Hence there is a
8 question as to whether the areas which are known to be peroxisome rich were assayed of not.
9 The data from all control groups were reported as pooled data in figures but statistical
10 comparisons were made between concurrent control and treated groups. The results for DNA
11 single strand breaks were reported for "13 control animals" and each experimental time point "as
12 at least 6 animals." DNA strand breaks were reported to be significantly increased over
13 concurrent control by a single exposure to 10 or 500 mg/kg DCA or 500 mg/kg TCA for 1,2, or
14 4 hours after administration but not at 8 or 24 hours. There did not appear to be a difference in
15 the magnitude of response between the 3 treatments (the fraction of unwound DNA was
16 -2.5 times that of control). PCO activity was reported to be not increased over control within
17 24 hours of either DCA or TCA treatment, (n = 6 animals per group). The fraction of alkaline
18 unwinding rates as an indicator of single strand breaks were reported to not be significantly
19 different from controls and TCA-treated animals after 10 days of exposure (n = 5).
20 Relative to controls, body weights were reported to not be affected by exposures to DCA
21 or TCA for 10 days at 500 mg/kg (data were not shown.) (n = 6 per group). However, both DCA
22 and TCA were reported to significantly increase liver weight and liver/body weight ratios (i.e.,
23 liver weights were 1.3 ± 0.05 g, 2.1 ± 0.10 g, and 1.7 ± 0.09 g for control, 500 mg/kg DCA and
24 500 mg/kg TCA treatment groups, respectively while percent liver/body weights were
25 4.9% ± 0.14%, 7.5% ± 0.18%, and 5.7% ± 0.14% for control, 500 mg/kg DCA and 500 mg/kg
26 TCA treatment groups, respectively). PCO activity (umol/min/g liver) was reported to be
27 significantly increased by DCA (500 mg/kg), TCA (500 mg/kg), and Clofibrate (250 mg/kg)
28 treatment (i.e., levels of oxidation were 0.63 ± 0.07, 1.03 ± 0.09, 1.70 ± 0.08, and 3.26 ± 0.05 for
29 control, 500 mg/kg DCA, 500 mg/kg TCA and 250 mg/kg Clofibrate treatment groups,
30 respectively). Thus, the increases were -1.63-, 2.7-, and 5-fold of control for DCA, TCA and
31 Clofibrate treatments. Results from randomly selected electron photomicrographs from 2
32 animals (6 per animal) were reported for DCA and TCA treatment and to show an increase in
33 peroxisomes per unit area that was reported to be statistically significant (i.e., 9.8 ± 1.2, 25.4 ±
34 2.9, and 23.6 ± 1.8 for control, 500 mg/kg DCA and 500 mg/kg TCA, respectively). The 2.5-
35 and 2.4-fold of control values for DCA and TCA gave a different pattern than that of PCO
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-121 DRAFT—DO NOT CITE OR QUOTE
-------
1 activity. The small number of animals examined limited the power of the experiment to
2 quantitatively determine the magnitude of peroxisome proliferation via electron microscopy.
3 The enzyme analyses suggested that both DCA and TCA were weaker inducers of peroxisome
4 proliferation that Clofibrate.
5 The authors report that there was no evidence of gross hepatotoxicity in vehicle or TCA-
6 treated mice. Light microscopic sections from mice exposed to TCA or DCA for 10 days were
7 stained with H&E and PAS for glycogen. For TCA treatment, PAS staining "produced
8 approximately the same intensity of staining and amylase digesting revealed that the vast
9 majority of PAS-positive staining was glycogen." Hepatocytes were reported to be "slightly
10 larger in TCA-treated mice than hepatocytes from control animals throughout the liver section
11 with the architecture and tissue pattern of the liver intact." The histopathology after DCA
12 treatment was reported to be "markedly different than that observed with either vehicle or TCA
13 treatments" with the "most pronounced change in the size of hepatocytes." DCA was reported to
14
15 produce marked cellular hypertrophy uniformly throughout the liver. The
16 hepatocytes were approximately 1.4 times larger in diameter than control liver
17 cells. This hypertrophy was accompanied by an increase in PAS staining;
18 indicating greater glycogen deposition than in TCA-treated and control liver
19 tissue. Multiple white streaks were grossly visible on the surface of the liver of
20 DCA-treated mice. The white areas corresponded with subcapsular foci of
21 coagulative necrosis. These localized necrotic areas were not encapsulated and
22 varied in size. The largest necrotic foci occupied the area of a single lobule.
23 These necrotic areas showed a change in staining characteristics. Often this
24 change consisted of increased eosinophilia. A slight inflammatory response,
25 characterized by neutrophil infiltration, was present. These changed were evident
26 in all DCA-treated mice.
27
28 The results from this experiment cannot inform as to dose-response relationships for the
29 parameters tested with the exception of DNA single strand breaks where 2 concentrations of
30 DCA were examined (10 and 500 mg/kg). For this parameter the 10 mg/kg exposure of DCA
31 was as effective as the 500 mg/kg dose where toxicity was observed. This effect on DNA was
32 also observed before evidence of induction of peroxisome proliferation. The authors did not
33 examine Clofibrate for effects on DNA so whether it too, would have produced this effect is
34 unclear. The results from this study are consistent with those of Sanchez and Bull (1990) for
35 induction of hepatomegaly by DCA and TCA, the lack of hepatotoxicity at this dose by TCA,
36 and the difference in glycogen deposition between DCA and TCA.
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-122 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.3.1.3. Styles et al, 1991. In this report a similar paradigm is used as Nelson et al. (1989)
2 for the determination of repeating that work on single strand breakage and to study DNA
3 synthesis and peroxisome proliferation. In regard to the findings of single strand breaks, Styles
4 et al. (1991) reported for a similar paradigm of 500 mg/kg neutralized TCA administered to male
5 B6C3F1 mice (7-8 weeks of age) examined at 1, 4, 8, and 24 hours after dosing, reported no
6 increased unwinding of DNA 1 or 24 hours after TCA administration. In a separate experiment
7 tritiated thymidine was administered to mice 1 hour before sacrifice at 24, 36, 48, 72, and
8 96 hours after the first dose of 500 mg/kg TCA for 3 days via gavage (n = 5 animals per group).
9 The hepatic DNA uptake of tritiated thymidine was reported to be similar to control
10 levels up to 36 hours after the first dose and then to increase to a level ~6-fold greater than
11 controls by 72 hours after the first dose of TCA. By 96 hours the level of tritiated thymidine
12 incorporation had fallen to ~4-fold greater than controls. The variation, reported by standard
13 deviation (SD) was very large in treated animals (e.g., SD was equal to approximately ±1.3-fold
14 of control for 48 hour time point). Individual hepatocytes were examined with the number of
15 labeled hepatocytes/1,000 cells reported for each animal. The control level was reported to be ~1
16 with a SD of similar magnitude. The number of labeled hepatocytes was reported to decrease
17 between 24 and 36 hours and then to rise slowly back to control levels at 48 hour and then to be
18 significantly increased 72 hours after the first dose of TCA (~9 cells/1,000 with a SD of 3.5) and
19 then to decrease to a level of ~5 cells/1,000. Thus, it appears that increases in hepatic DNA
20 tritiated thymidine uptake preceded those of increased labeled hepatocytes and did not capture
21 the decrease in hepatocyte labeling at 36 hours. By either measure the population of cells
22 undergoing DNA synthesis was small with the peak level being less than 1% of the hepatocyte
23 population. The authors go on to report the zonal distribution of mean number of hepatocytes
24 incorporating tritiated thymidine but no variations between animals were reported. The decrease
25 in hepatocyte labeling at 36 hours was apparent at all zones. By 48 hours there appeared to be
26 slightly more perioportal than midzonal cells undergoing DNA synthesis with centrilobular cells
27 still below control levels. By 72 hours all zones of the liver were reported to have a similar
28 number of labeled cells. By 96 hours the midzonal and centrilobular regions have returned
29 almost to control levels while the periportal areas were still elevated. These results are consistent
30 with all hepatocytes showing a decrease in DNA synthesis by 36 hours and then a wave of DNA
31 synthesis occurring starting at the periportal zone and progressing through to the pericentral zone
32 until 72 hours and then the midzonal and pericentral hepatocytes completing their DNA
33 synthesis activity. Peroxisome proliferation was assessed via electron photomicrographs taken in
34 mice (4 controls and 4 treated animals) given 10 daily doses of 500 mg/kg TCA and killed
35 14 hours after the last dose. No details were given by the authors as to methodology for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-123 DRAFT—DO NOT CITE OR QUOTE
-------
1 peroxisome volume estimate (e.g., how many photos per animals were examined and whether
2 they were randomly chosen). The mean percent cell volume occupied by peroxisome was
3 reported to be 2.1% ± 0.386% and 3.9% ± 0.551% for control and 500 mg/kg TCA, respectively.
4 Given there were no time points examined before 10 days for peroxisome proliferation,
5 correlations with DNA synthesis activity induced by TCA cannot be made from this experiment.
6 However, it is clear from this study that a wave of DNA synthesis occurs throughout the liver
7 after treatment of TCA at this exposure concentration and that it has peaked by 72 hours even
8 with continuous exposure to 96 hours. Whether the DNA synthesis represents polyploidization
9 or cell proliferation cannot be determined from these data as neither can a dose-response.
10
11 E.2.3.1.4. Carter et al, 1995. The aim of this study was to "use correlative biochemical,
12 pathologic and morphometric techniques to characterize and quantify the acute, short-term
13 responses of hepatocytes in the male B6C3F1 mouse to drinking water containing DC A." This
14 report used tritiated thymidine incorporation, DNA concentration, hepatocyte number per field
15 (cellularity), nuclear size and binuclearity (polyploidy) parameters to study 0, 0.5, and 5 g/L
16 neutralized DC A exposures up to 30 days. Male B6C3F1 mice were started on treatment at
17 28 days of age. Tritiated thymidine was administered by miniosmotic pump 5 days prior to
18 sacrifice. The experiment was conduced in two phases which consisted of 5-15 days of
19 treatment (Phase I) and 20-30 days of treatment (Phase II) with 5 animals per group in groups
20 sacrificed at 5-day intervals. Liver sections were stained for H&E, PAS (for glycogen) or methyl
21 green pryonin stain (for RNA). DNA was extracted from liver homogenates and the amount of
22 tritiated thymidine determined as dpm/ug DNA. Autoradiography was performed with the
23 number of hepatocyte nuclei scored in 1,000 hepatocytes selected randomly to provide a labeling
24 index of "number of labeled cells/1000 X 100%." Changes in cellularity, nuclear size and
25 number of multinucleate cells were quantified in H&E sections at 40* power. Hepatocyte
26 cellularity was determined by counting the number of nuclei in 50 microscopic fields with
27 multinucleate cells being counted as one cell and nonparenchymal cells not counted. Nuclear
28 size was also measured in 200 nuclei with the mean area plus 2 SD was considered to be the
29 largest possible single nucleus. Therefore, polyploid diploid cells were identified by the authors
30 but not cells that had undergone polyploidy with increased DNA content in a single nucleus.
31 Mean body weights at the beginning of the experiment varied between 18.7 and 19.6 g in
32 the first 3 exposure groups of Phase I of the study. Through 15 days of exposure there did not
33 appear to be a change in body weight in the 0.5 g/L exposure groups but in the 5 g/L exposure
34 group body weight was reduced at 5, 10 and 15 days with that reduction statistically significant
35 at 5 and. 15 days. Liver weights did not appear to be increased at Day 5 but were increased at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-124 DRAFT—DO NOT CITE OR QUOTE
-------
1 days 10 and 15 in both treatment groups (i.e., means ± S.E.M. for Day 10; 1.36 ± 0.03,
2 1.46 ± 0.03, and 1.59 ± 0.08 g for control, 0.5 and 5 g/L DCA, respectively and for Day 15;
3 1.51 ± 0.06, 1.72 ± 0.05, and 2.08 ± 0.11 g for control, 0.5 and 5 g/L DCA, respectively). The
4 percent liver/body weight followed a similar pattern with the exception that at Day 5 the 5 g/L
5 exposure group had a statistically significant increase over control (i.e., for Day 10;
6 6.00% ± 0.10%, 6.72% ± 0.17%, and 8.21% ± 0.10% for control, 0.5 and 5 g/L DCA,
7 respectively and for Day 15; 6.22 ± 0.08, 6.99 ±0.15, and 10.37 ± 0.27% g for control, 0.5 and
8 5 g/L DCA, respectively).
9 In Phase II of the study, control body weights were smaller than Phase I and varied
10 between 16.6 and 16.9 g in the first 3 exposure groups. Liver weights of controls were also
11 smaller making it difficult to quantitatively compare the two groups in terms of absolute liver
12 weights. However, the pattern of DCA-induced increases in liver weight and percent liver/body
13 weight remained. The patterns of body weight reduction only in the 5 g/L treatment groups and
14 increased liver weight with DCA treatment at both concentrations continued from 20 to 30 days
15 of exposure. For liver weight there was a slight but statistically significant increase in liver
16 weight for the 0.5 g/L treatment groups over controls (i.e., for Day 20; 1.02 ± 0.02, 1.18 ± 0.05,
17 and 1.98 ± 0.05 g for control, 0.5 and 5 g/L DCA, respectively, for Day 25; 1.15 ± 0.03,
18 1.34 ± 0.04, and 2.06 ± 0.12 g for control, 0.5 and 5 g/L DCA, respectively, for Day 30;
19 1.15 ± 0.03, 1.39 ± 0.08, and 1.90 ± 0.12 g for control, 0.5 and 5 g/L DCA, respectively). For
20 percent liver/body weight there was a small increase at 0.5 g/L that was not statistically
21 significant but all other treatments induced increases in percent liver/body weight that were
22 statistically significant (i.e., for Day 20; 4.82% ± 0.07%, 5.05% ± 0.09%, and 9.71% ± 0.11% for
23 control, 0.5 and 5 g/L DCA, respectively, for Day 25; 5.08% ± 0.04%, 5.91% ± 0.09%, and
24 10.38% ± 0.58% for control, 0.5 and 5 g/L DCA, respectively, for Day 30; 5.17% ± 0.09%,
25 6.01% ± 0.08%, and 10.28% ± 0.28% for control, 0.5 and 5 g/L DCA, respectively). Of note is
26 the dramatic decrease in water consumption in the 5 g/L treatment groups that were consistently
27 reduced by 64% in Phase I and 46% in Phase II. The 0.5 g/L treatment groups had no difference
28 from controls in water consumption at any time in the study. The effects of such water
29 consumption decreases would affect body weight as well as dose received. Given the differences
30 in the size of the animals at the beginning of the study and the concurrent differences in liver
31 weights and percent liver/body weight in control animals between the two phases, the changes in
32 these parameters through time from DCA treatments cannot be accurately determined (e.g.,
33 control liver/body weights averaged 6.32% in Phase I but 5.02% in Phase II). However, percent
34 liver/body weight increase were reported to be consistently increased within and between both
35 phases of the study for the 0.5 g/L DCA treatment from 5 days of treatment to 30 days of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-125 DRAFT—DO NOT CITE OR QUOTE
-------
1 treatment (i.e., for Phase I the average increase was 9.5% and for Phase II the average increased
2 was 12.5% for 0.5 g/L DCA treated groups). Although increase at 5 days the nonsignificance of
3 the change may be resultant from the small number of animals examined. The difference in
4 magnitude of dose and percent liver/body weight increase is difficult to determine given that the
5 5 g/L dose of DCA reduced body weight and significantly reduced water consumption by -50%
6 in both phases of the study. Of note is that the differences in DCA-induced percent liver/body
7 weight were ~6-fold for the 15, 25, and 30-day data between the 0.5 and 5 g/L DCA exposures
8 rather than the 10-fold difference in exposure concentration in the drinking water.
9 The incorporation of tritiated thymidine into total hepatic DNA control treatment groups
10 was reported to be 73.34 ± 11.74 dpm/ugDNA at 5 days, 34± 4.12 dpm/ug DNA at 15 days,
11 and 28.48 ± 3.24 dpm/ug DNA at 20 days but was not reported for other treatments. The results
12 for 0.5 g/L treatments were not reported quantitatively but the authors stated that the results
13 "showed similar trends of initial inhibition followed by enhancement of labeling, the changes
14 relative to controls were not statistically significant." For 5 g/L treatment groups the 5-day
15 treated groups DNA tritiated thymidine incorporation was reported to be 42.8% of controls and
16 followed by a transient increase at 15 and 20 days (i.e., 2.65- and 2.45-fold of controls,
17 respectively) but after 25 and 30 days to not be significantly different from controls (data not
18 shown). Labeling indices of hepatocytes were reported as means but variations as either SEM or
19 SD were not reported. Control means were reported as 5.5, 4, 2, 2, 3.2, and 3.5% of randomly
20 selected hepatocytes for 5, 10, 15, 20, 25, and 30 days, respectively, for 4 to 5 animals per group.
21 In contrast to the DNA incorporation results, no increase in labeling of hepatocytes was reported
22 to be observed in comparison to controls for any DCA treatment group from 5 to 30 days of
23 DCA exposure. The 5 g/L treatment group showed an immediate decrease in hepatocyte
24 labeling from Day 5 onwards that gradually increased approximately half of control levels by
25 Day 30 of exposure (i.e., <0.5% labeling index [LI] at Day 5, -1% LI at Day 10, -0.6% LI at
26 Day 20, 1% LI at Day 25 and 2% LI at Day 30). For the 0.5 g/L treatment the labeling index
27 was reported to not differ from controls from days 5 though 15 but to be significantly decreased
28 between days 20 and 30 to levels similar to those observed for the 5 g/L exposures. The
29 relatively higher number of hepatocytes incorporating label reported in this study than others can
30 be reflection of the longer times of exposure to tritiated thymidine. Here, incorporation was
31 shown for 1 weeks worth of exposure and reflects the percent of cell undergoing synthesis during
32 that time period. Also the higher labeling index in control animals at the 5 and 10 day exposure
33 periods is probably a reflection of the age of the animals at the time of study. From the data
34 reported by the authors, there was a correlation between the patterns of total DNA incorporation
35 of label and hepatocyte labeling indices in control groups (i.e., higher level of labeling at 5 days
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-126 DRAFT—DO NOT CITE OR QUOTE
-------
1 than at 15 and 20). However, the patterns of decreased thymidine labeling reported for
2 hepatocytes were not correlated with a transient increase in total DNA thymidine incorporation
3 reported with DCA treatment, especially at the 5 g/L exposure level with a large decrease
4 reported for the number of labeled hepatocytes at the same time an increase in total DNA
5 thymidine incorporation was reported. Although reported to be transiently increased, the total
6 hepatic DNA labeling still represented at most a 2.5-fold increase over control liver, which
7 represents a small population of cells. Given that the study examined hepatocyte labeling in
8 random fields and did not report quantitative zonal differences in proliferation, a more accurate
9 determination of what hepatocytes were undergoing proliferation cannot be made from the LI
10 results. Also although the authors report signs of inflammatory cells for 5-day treatment there is
11 no reference to any inflammatory changes that may have been observed at later time periods
12 when cellular degeneration and loss of nuclei were apparent. Such an increase inflammatory
13 infiltrates can increase the DNA synthesis measurements in the liver. The difference in LI and
14 total DNA synthesis could reflect differences in nonparenchymal cell proliferation or ploidy
15 changes versus mitoses in hepatocytes. Clearly, the increases in liver weight that were reported
16 as early as 5 days of exposure could not have resulted from increased hepatocyte proliferation.
17 The H&E sections were reported to have been fixed in an aqueous solution that reduced
18 glycogen content. However, residual PAS positive material (assumed to be glycogen) was
19 reported to be present indicating that not all of the glycogen had been dissolved. The authors
20 report changes in pathology between 5 and 30 days in control animals that included straightening
21 of hepatocyte cording, decreased mitoses, less clarity and more fine granularity of pericentral
22 hepatocellular cytoplasm, increased numbers of larger nuclei that were not labeled, and reported
23 differences between animals in the amount of glycogen present (i.e., 2 or 3 animals out of the 5
24 had less glycogen than other members of the group with less glycogen in the central and
25 midzonal areas). These changes are consistent with increased polyploidization expected for
26 maturing mice (see Sections E. 1.1 and E. 1.2 above). After 5 days of treatment, 0.5 g/L exposed
27 animals were reported to have livers with fewer mitoses and tritiated thymidine hepatocyte
28 labeling but by 10 days an increase in nuclear size. Labeling was reported to be predominantly
29 in small nuclei. Animals given 0.5 g/L DCA for 15, 20, and 25 days were reported to have
30 "focal cells in the middle zone with less detectable or no cell membranes and loss of the coarse
31 granularity of the cytoplasm" with some cells not having nuclei or cells having a loss of nuclear
32 membrane and apparent karyolysis. "Cells without nuclei because the plane of the section did
33 not pass through the nuclei had the same type of nuclei. Cells without nuclei not related to plane
34 of section had a condensed cytoplasm." Livers from 20-day and later sacrifice groups treated
35 with 0.5 g/L DCA were reported to have normal architecture. After 25 days of treatment
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-127 DRAFT—DO NOT CITE OR QUOTE
-------
1 apoptotic bodies were reported to be observed with fewer nuclei around the central veins nuclei
2 that were larger in central and midzonal areas. In animals treated with 5 g/L DCA the authors
3 report similar features as for 0.5 g/L but in a zonal pattern. Inflammatory cells were reported to
4 not be observed and after 5 and 10 days a marked decrease in labeled nuclei. After 5 days of
5 5 g/L DCA, nuclear depletion in the central and mid-zonal areas was reported. In methyl green
6 pyronin-stained slides a marked loss of cellular membranes was reported at 5 days with a loss of
7 nuclei and formation of "lakes of liver cell debris." After 15 days of treatment there was a
8 reported increase in labeling in comparison to animals sacrificed after 5 or 10 days. The cells
9 nearest to the triads were reported to have clearing of their cytoplasms and an increase in PAS
10 positivity. Hepatocytes of both 0.5 and 5 g/L DCA treatment groups were reported to have
11 "enlarged, presumably polyploidy nuclei." Some of the nuclei were reported to be "labeled,
12 usually in hepatocytes in the mid-zonal area."
13 The morphometric analyses of liver sections were reported to reveal statistically
14 significant changes in cellularity, nuclear size (as measured by either nuclear area or mean
15 diameter of the nuclear area equivalent circle), and multinucleated cells during 30 days exposure
16 to DCA. The authors reported that the concentration of total DNA in the liver, reported as total
17 ug nuclear DNA/g liver, ranged between 278.17 ± 16.88 and 707.00 ± 25.03 in the control
18 groups (i.e., 2-5-fold range). No 0.5 g/L DCA treatment groups differed from their control
19 group in terms of liver DNA concentration. However, for 10 though 30 days of exposure hepatic
20 DNA concentrations were reported to be decreased in the 5 g/L treatment groups (at 5 days there
21 appeared to be -30% increase over control). The number of cells per field was reported to range
22 between 24.28 ± 1.94 and 43.81 ± 1.93 in control livers (i.e., 1.8-fold range). From 5 to 15 days
23 the number of cells/field decreased with 0.5 g/L DCA treatment although only at Day 15 was the
24 change statistically significant. From 20 to 30 days of treatment only the 30 day treatment
25 showed a slight decrease in cells/field and that change was statistically significant. After 5 days
26 of treatment, the number of cells/field was 1.6-fold of control, by 15 days reduced by -20%, and
27 for 20 to 30 days continued to be reduced by as much as 40%. Although the authors reported
28 that the changes in cellularity and DNA concentration to be closely correlated, the patterns in the
29 number of cells/field varied in their consistency with those of DNA concentration (i.e., for days
30 5, 20 and 25 there direction of change with dose was similar between the two parameters but for
31 days 10, 15 and 30 were not). If changes in liver weight were due to hepatocellular hypertrophy,
32 the increased liver size would be matched by a decrease in liver DNA concentration and by the
33 number of cells/field. The large increases in liver/body weight induced by 5 g/L DCA were
34 matched by decreases in liver DNA concentration except for the 5 day exposure group. In
35 general, the small increases in liver/body weight consistently induced by 0.5 g/L treatment from
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-128 DRAFT—DO NOT CITE OR QUOTE
-------
1 Day 5 through 30 were not correlated with DNA concentrations or cells/field. The small number
2 of animal examined for these parameters (i.e., n = 4-5) and the highly variable control values
3 limit the power to accurately detect changes. The apparent dehydration in the animals treated at
4 5 g/L DCA was cited by the authors for the transient increase in cellularity and DNA
5 concentration in the 5-day exposure group. However, drinking water consumption was reported
6 to be similarly reduced at all treatment periods for 5 g/L DCA-treated animals so that all groups
7 would experience the same degree of dehydration.
8 The percentage of mononucleated cells was reported as percent of mononucleated
9 hepatocytes with results given as means but with no reports of variation within groups. The
10 mean control values were reported to range between 60 and 75% for Phase I and between 58 and
11 71% for Phase II of the experiment (n = 4-5 animals per group). The percent of mononucleated
12 hepatocytes was reported to be similar between control and DCA treatment groups at 5- and
13 10-day exposure. At 15 days both DCA treatments were reported to give a similar increase in
14 mononucleated hepatocytes (-80 vs. 60% in control) with only the 5 g/L DCA group statistically
15 significant. The increase in mononucleated cells reported for DCA treatment is similar in size to
16 the variation between control values. For Phase II of the study, DCA treatment was reported to
17 increase the number of mononucleated cells in at all concentrations and exposure time periods in
18 comparison to control values. However, only the increases for the 5 g/L treatments at days 20
19 and 25, and the 0.5 g/L treatment at Day 30 were reported to be statistically significant. Again,
20 small numbers of animals limit the ability to accurately determine a change. However, the
21 consistent reporting of an increasing number of mononucleated cells between 15 and 30 days
22 could be associated with clearance of mature hepatocytes as suggested by the report of DCA-
23 induced loss of cell nuclei.
24 Mean nuclear area was reported to range between 45 and 54 u2 in Phase I and to range
25 between 41 and 48 u2 in Phase II of the experiment with no variation in measurements given by
26 the authors. The only statistically significant differences reported between control and treated
27 groups in Phase I was a decrease from 54 to -42 u2 in the 0.5 g/L DCA 10 day treatment group
28 and a small increase from 50 to -52 u2 15 day treatment group. Clearly the changes reported by
29 the authors as statistically significant did not show a dose-related pattern and were within the
30 range of variation reported between control groups. For Phase II of the experiment both DCA
31 treatment concentrations were reported to induce a statistically significant increase the nuclear
32 area that was dose-related with the exception of Day 30 in which the nuclear area was similar
33 between the 0.5 and 5 g/L treatment groups. The largest increase in nuclear area was reported at
34 20 days for the 5 g/L treatment group (-72 vs. 41 u2 for control). The patterns of increases in
35 nuclear area were correlated with those of increased percentage of mononucleated cells in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-129 DRAFT—DO NOT CITE OR QUOTE
-------
1 Phase II of the study (20-30 days of treatment) as well as the small changes seen in Phase I of
2 the experiment. An increase in nuclear cell area is consistent with increase polyploidization
3 without mitosis as cells are induced towards polyploidization. A decrease in the numbers of
4 binucleate cells in favor of mononucleate cells is consistent with clearance of mature binucleate
5 hepatocyte as well induction of further polyploidization of diploid or tetraploid binucleate cell to
6 tetraploid or octoploid mononucleate cells. The authors suggested that the "large
7 hyperchromatic mononucleated hepatocytes are tetraploid" and suggest that such increases in
8 tetraploid cells have also been observed with nongenotoxic carcinogens and with
9 di(2-ethylhexyl) phthalate (DEHP). In terms of increased cellular granularity observed by the
10 authors with DC A treatment, this result is also consistent with a more differentiated phenotype
11 (Sigal et al., 1999). Thus, these results for DCA are consistent with a DCA induced change in
12 polyploidization of the cells without cell proliferation. The pattern of consistent increase in
13 percent liver/body weight induced by 0.5 g/L DCA treatment from days 5 though 30 was not
14 consistent with the increased numbers of mononucleate cells and increase nuclear area reported
15 from Day 20 onward. The large differences in liver weight induction between the 0.5 g/L
16 treatment group and the 5 g/L treatment groups at all times studied also did not correlate with
17 changes in nuclear size and percent of mononucleate cells. Thus, increased liver weight was not
18 a function of cellular proliferation, but probably included both aspects of hypertrophy associated
19 with polyploidization and increased glycogen deposition induced by DCA. The similar changes
20 reported after short-term exposure for both the 0.5 and 5 g/L exposure concentration were
21 suggested by the authors to indicate that the carcinogenic mechanism at both concentrations
22 would be similar. Furthermore, they suggest that although there is evidence of cytotoxicity (e.g.,
23 loss of cell membranes and apparent apoptosis), the present study does not support that the
24 mechanism of DCA-induced hepatocellular carcinogenesis is one of regenerative hyperplasia
25 following massive cell death nor peroxisome proliferation as the 0.5 g/L exposure concentration
26 has been shown to increase hepatocellular lesions after 100 weeks of treatment without
27 concurrent peroxisome proliferation or cytotoxicity (DeAngelo et al., 1999).
28
29 E.2.3.1.5. DeAngelo et al., 1989. Various strains of rats and mice were exposed to TCA (12
30 and 31 mM) or DCA (16 and 39 mM) for 14 days with S-D rats and B6C3F1 mice exposed to an
31 additional concentration of 6 mM TCA and 8 mM DCA. Although noting that in a previous
32 study that high concentrations of chloracids, the authors did not measure drinking water
33 consumption in this study. This study exposed several strains of male rats and mice to TCA at
34 two concentrations in drinking water (12 mM and 3 ImM neutralized TCA) for 14 days. The
35 conversion of mmols/L or mM TCA is 5 g/L TCA, 2 g/L TCA and 1 g/L for 31 mM, 12 mM,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-130 DRAFT—DO NOT CITE OR QUOTE
-------
1 and 6 mM TCA, respectively. The conversion of mmols/L of mM DC A is 5 g/L DCA, 2 g/L
2 DCA, and 1 g/L DCA for 39 mM, 16 mM and 8 mM DCA, respectively. The strains of mice
3 tested were Swiss-Webster, B6C3F1, C57BL/6, and C3H and for rats were Sprague Dawley,
4 Osborne Mendel, and F344. For the F344 rat and B6C3F1 mice data from two separate
5 experiments were reported for each. The number of animals in each group was reported to be 6
6 for most experiments with the exception of the S-D rats (n = 3 at the highest dose of TCA and
7 n = 4 or 5 for the control and the lower TCA dose), one study in B6C3F1 mice (n = 4 or 5 for all
8 groups), and one study in F344 rats (n = 4 for all groups). The body weight of the controls was
9 reported to range from 269 to 341 g in the differing strains of rats (1.27-fold) and 21 to 28 g in
10 the differing strains of mice (1.33-fold, age not reported). For percent liver/body weight ratios
11 the range was 4.4 to 5.6% in control rats (1.27-fold) and 5.1 to 6.8% in control mice (1.33-fold).
12 As discussed in other studies, the determination of PCO activity appears to be highly
13 variable. This enzyme activity is often used as a proxy for peroxisome proliferation. For PCO
14 activity the range of activity in controls was much greater than for either body weight or percent
15 liver/body weight. For rats there was a 2.8-fold difference in PCO control activity and in mice
16 there was a 4.6-fold difference in PCO activity. Between the two studies performed in the same
17 strain of rat (F344) there was a 2.83-fold difference in PCO activity between controls, and for the
18 two studies in the same strain of mouse (B6C3F1) there was a 3.14-fold difference in PCO
19 activity between controls. Not only were there differences between strains and experiments in
20 the same strain, but also differences in control values between species with a wider range of
21 values in the mice. The lowest level of PCO activity in control rats, expressed as nanomoles
22 NAD reduced/min/mg/protein, was 3.34 and for control mice was 1.40. The highest level
23 reported in control in rats was 9.46 and for control mice was 6.40.
24 These groups of rats and mice were exposed to 2 g/L NaCl, 2 g/L or 5 g/L TCA in
25 drinking water for 14 days and their PCO activity assayed. These doses of TCA did not affect
26 body weight except for the S-D rats, which lost -16% of their body weight. This was also the
27 same group in which only 3 rats survived treatment. The Osborne-Mendel and F344 strains did
28 not exhibit loss of body weight or mortality due to TCA exposure. There was a large variation in
29 response to TCA exposure between the differing strains of rats and mice with a much larger
30 difference between the strains of mice. For the 3 rat strains tested there was a range between 0%
31 change and 2.38-fold of control for PCO activity at the 5 g/L TCA exposure. For the 2 g/L TCA
32 exposure, there was a range of 0% change to 1.54-fold of control for PCO activity. The
33 Osborne-Mendel rats had 1.54-fold of control value for PCO activity at 2 g/L TCA and 2.38-fold
34 of control value for PCO activity reported at 5 g/L, exhibiting the most consistent increase in
35 PCO with increased dose of TCA. Two experiments were reported for F344 rats with one
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-131 DRAFT—DO NOT CITE OR QUOTE
-------
1 reporting a 1.63-fold of control and the other a 1.79-fold of control value for 5 g/L TCA. Only
2 one of the F334 experiments also exposed rats to 2 g/L TCA and reported no change from
3 control values.
4 For the 4 strains of mice tested there was a range of 7.44- to 22.13-fold of control values
5 reported at the 5 g/L TCA exposures and 3.76- to 25.92-fold of control values at the 2 g/L TCA
6 exposures for PCO activity. For the C57BL/6 strain of mice there was little difference between
7 the 5 g/L and 2 g/L TCA exposures and a generally 3-fold higher induction of PCO activity by
8 TCA at the 5 g/L TCA exposure level than for the other mouse strains. Although there was a
9 2.5-fold difference between the 5 g/L and 2 g/L TCA exposure dose, the difference in magnitude
10 of PCO activity between these doses ranged from 0.85- to 2.23-fold for all strains of mice. For
11 the B6C3F1 mice there was a difference between reported increases of PCO activity in the text
12 (i.e., reported as 9.59-fold of control) for one of the experiments and that presented graphically
13 in Figure 2 (i.e., 8.70-fold of control). Nevertheless in the two studies of B6C3 Fl mice, 5 g/L
14 TCA was reported to induce 7.78-fold of control and 8.70-fold of control for PCO activity, and
15 2 g/L TCA was reported to induce 5.56-fold of control and 4.70-fold of control for PCO activity.
16 For the two F344 rat studies in which -200 mg/kg or 5 g/L TCA was administered for 10 or
17 14 days, there was 1.63-fold of control and 1.79-fold of control values reported for PCO activity.
18 Thus, for experiments in which the same strain and dose of TCA were administered, there was
19 not as large a difference in PCO response than between strains and species.
20 Whether increases in percent liver/body weight ratios were similar in magnitude to
21 increased PCO activity can be assessed by examination of the differences in magnitude of
22 increase over control for the 5 g/L and 2 g/L TCA treatments in the varying rat strains and mouse
23 strains. The relationship in exposure concentration was a 2.5:1 ratio for the 5 and 2 g/L doses.
24 For rats treatment of 5 g/L TCA to S-D rats resulted in a significant decrease in body weight and
25 therefore, affected the magnitude of increase in percent liver/body weight ratio for this group.
26 However, for the rest of the rat and mouse data, this dose was not reported to affect body weight
27 so that there is more confidence in the dose-response relationship. For the S-D rat there was no
28 change in the percent liver/body weight ratio at 2 g/L but a 10% decrease at 5 g/L TCA exposure
29 with no change in PCO activity for either. However, for the Osborne-Mendel rats, there was no
30 change in percent liver/body weight ratios for either exposure concentration of TCA, but PCO
31 activity was reported to be 1.54-fold of control at 2 g/L and 2.38-fold of control at 5 g/L TCA.
32 Thus, there was a ratio of 2.5-fold increase in PCO activity between the 5 g/L and 2 g/L
33 treatment groups. For the F344 rats there was a 2-fold difference in liver weight increases (i.e.,
34 12 vs. 6% increase over control) between the two exposure concentrations but 1.6-fold of control
35 value for PCO activity at the 5 g/L TCA exposure concentration and no increase in PCO activity
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-132 DRAFT—DO NOT CITE OR QUOTE
-------
1 at the 2 g/L level. Thus, for the three strains of rats, there did not appear to be a consistent
2 correlation between liver weight induction by TCA and PCO activity.
3 For differing strains of mice, similar concentrations of TCA were reported to vary in the
4 induction of liver weight increases. The range of liver weight induction was 1.26- to 1.66-fold of
5 control values between the 4 strains of mice at 5 g/L TCA and 1.16- to 1.63-fold at 2 g/L TCA.
6 In general, for mice the magnitudes of the difference in the increase in dose between the 5 g/L
7 and 2 g/TCA exposure concentration (2.5-fold) was generally higher than the increase percent
8 liver/body weight ratios at these doses. The differences in liver weight induction between the 2
9 and 5 g/L doses were -40% for the Swiss-Webster, C3H, and for one of the B6C3F1 mouse
10 experiments. For the C57BL/6 mouse there was no difference in liver weight induction between
11 the 2 and 5 g/L TCA exposure groups. For the other B6C3F1 mouse experiments there was a
12 2.5-fold greater induction of liver weight increase for the 5 g/L TCA group than for the 2 g/L
13 exposure group (1.39-fold of control vs. 1.16-fold of control for percent liver/body weight,
14 respectively). For PCO activity the Swiss-Webster, C3H, and one of the B6C3F1 mouse
15 experiments were reported to have ~2-fold difference in the increase in PCO activity between the
16 two doses. For the other B6C3F1 mouse experiment there was only about a 50% increase and
17 for the C57BL/6 mouse data there was 15% less PCO activity induction reported at the 5 g/L
18 TCA dose that at the 2 g/L dose. None of the difference in increases in liver weight or PCO
19 activity in mice from the 2 or 5 g/L TCA exposures were of the same magnitude as the difference
20 in TCA exposure concentration (i.e., 2.5-fold) except for liver weight from the one experiment in
21 B6C3F1 mice. This is also the data used fore comparisons with the Sprague-Dawley rat
22 discussed below.
23 In regard to strain differences for TCA response in mice, there did not appear to be
24 correlations of the magnitude of 5 g/L TCA-induced changes in percent liver/body weight ratio
25 or PCO activity, with the body weights reported for control mice for each strain. The control
26 weights between the 4 strains of mice varied from 21 to 28 grams. The strain with the greatest
27 response (C57B1/6) for TCA-induced changes in percent liver/body weight ratio (i.e., 1.66-fold
28 of control) and PCO activity (22.13-fold of control) had a mean body weight reported to be 26 g
29 for controls. At this dose, the range of percent liver/body weight for the other strains was
30 reported to be 1.26- to 1.39-fold of control and the range of PCO activity reported to be of 7.48-
31 to 8.71-fold of control.
32 Of note is that in the literature, this study has been cited as providing evidence of
33 differences between rats and mice for peroxisomal response to TCA and DCA. Generally the
34 PCO data from the Sprague Dawley rats and B6C3F1 mice at the highest dose of TCA and DCA
35 have been cited. However, the S-D strain was reported to have greater mortality from TCA at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-133 DRAFT—DO NOT CITE OR QUOTE
-------
1 this exposure than the other strains tested (i.e., only 3 rats survived and provided PCO levels)
2 and a lower PCO response (no change in PCO activity over control) that the other two strains
3 tested in this study (i.e., Osborne-Mendel rats was reported to have had 2.38-fold of control and
4 the F344-had a 1.63- to 1.79-fold of control for PCO activity after exposure to 5 g/L TCA with
5 no mortality). The B6C3F1 mouse was reported to have a 7.78- or 8.71-fold of control for PCO
6 activity from 5 g/L TCA exposure. Certainly the male mouse is more responsive to TCA
7 induction of PCO activity. However, as discussed above there are large variations in control
8 levels of PCO activity and in the magnitude and dose-response of TCA-induction of PCO
9 activity between rat and mouse strains and between species. If is not correct to state that the rat
10 is refractory to TCA-induction of peroxisome activity.
11 Unfortunately, the authors chose the S-D rat (i.e., the most unresponsive strain for PCO
12 activity and most sensitive to toxicity) for studies for comparative studies between DC A and
13 TCA effects. The authors also tested for carnitine acetyl CoA transferase (CAT) activity as a
14 marker of peroxisomal enzyme response and took morphometric analysis of peroxisome # and
15 cytoplasmic volume for one liver section for each of two B6C3F1 mice of S-D rats from the
16 5 g/L TCA and 5 g/L DC A treatment groups. Only 6 electron micrograph fields were analyzed
17 from each section (12 fields total) were analyzed without identification as to what area of the
18 liver lobules they were being taken from. Hence there is a question as to whether the areas
19 which are known to be peroxisome rich were assayed of not. Also as noted above, previous
20 studies have indicate that such high concentration of DC A and TCA inhibit drinking water
21 consumption and therefore, raising issues not only about toxicity but also the dose which rats and
22 mice received. The number of peroxisomes per 100 um3 and cytoplasmic volume of
23 peroxisomes was reported to be 6.60 and 1.94%, respectively, for control rats, and 6.89 and
24 0.61% for control mice, respectively. For 5 g/L TCA and 5 g/L DCA the numbers of
25 peroxisomes were reported to be increased to 7.14 and 16.75, respectively in treated Sprague
26 Dawley rats. Thus, there was 2.5- and 1.08-fold of control reported in peroxisome # for 5 g/L
27 DCA and TCA, respectively. The cytoplasmic volume of peroxisomes was reported to be 2.80%
28 and 0.89% for 5 g/L DCA and 5 g/L TCA, respectively (i.e., a 1.44-fold of control and -60%
29 reduction for 5 g/L DCA and 5 g/L TCA, respectively). Thus, 5 g/L TCA was reported to
30 slightly increase the number of peroxisomes and but decrease the percent of the cytoplasmic
31 volume occupied by peroxisome by half. For DCA the reported pattern was for both to increase.
32 PCO activity was reported to increase by a similar magnitude as peroxisome # but not volume in
33 the 5 g/L TCA treated S-D rats. However, although peroxisomal volume was reported to be cut
34 nearly in half and for peroxisome number to be similar, 5 g/L TCA treatment was not reported to
35 change PCO activity in the S-D rat.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-134 DRAFT—DO NOT CITE OR QUOTE
-------
1 For comparisons between DCA and TCA B6C3 Fl mice were examined at 1.0, 2.0, and
2 5.0 g/L concentrations. DCA was reported to induce a higher percent liver/body weight ratio
3 that did TCA at every concentration (i.e., 1.55-, 1.27-, and 1.21-fold of control for DCA and
4 1.39-, 1.16-, and 1.08-fold of control for TCA at 1.0, 2.0, and 5.0 g/L concentrations,
5 respectively). As noted above, for other strains of mice tested and a second experiment with
6 B6C3F1 mice, there was 40% or less difference in percent liver/body weight ratio between the
7 2.0 g/L and 5.0 g/L exposures to TCA but for this experiment there was a 2.5-fold difference.
8 Thus, at 5 g/L there was -40% greater induction of liver weight for DCA than TCA. In the
9 B6C3F1 mice, 5 g/L TCA was reported to increase peroxisome number to 30.75 and cytoplasmic
10 volume to 4.92% (i.e., 4.4- and 8.1-fold of control, respectively). For 5 g/L DCA treatment, the
11 peroxisome number was reported to be 30.77 and 3.75% (i.e., 4.5- and 6.1-fold of control,
12 respectively). While there was no difference in peroxisome number and -40% difference in
13 cytoplasmic volume at the 5.0 g/L exposures of DCA and TCA, there was a greater difference in
14 the magnitude of PCO activity increase. The 5 g/L TCA exposure was reported to induce
15 4.3-fold of control for PCO activity while 5 g/L DCA induced as 9.6-fold of control PCO activity
16 (although a figure in the report shows 8.7-fold of control) which is a -2.5-fold difference
17 between DCA and TCA at this exposure concentration. Thus, for one of the B6C3F1 mouse
18 studies, 5 g/L DCA and TCA treatments were reported to give a similar increase peroxisome
19 number, TCA to induce a 40% greater increase in peroxisomal cytoplasmic volume than DCA
20 and a 2.5-fold greater increase in PCO activity, but DCA to induce -40% greater liver weight
21 induction than TCA.
22 Not only were PCO activity, peroxisome number and cytoplasmic volume occupied by
23 peroxisomes analyzed but also CAT activity as a measure of peroxisome proliferation. For TCA
24 and DCA the results were opposite those reported for PCO activity. In S-D rats control levels of
25 CAT were reported to be 1.81 nmoles of carnitine transferred/min/mg/protein. Exposure to 5 g/L
26 TCA was reported to increase CAT activity by 3.21-fold of control while 5 g/L DCA was
27 reported to induce CAT activity to 10.33-fold of control levels in S-D rats. However, while PCO
28 activity was reported to be the same as controls, and peroxisomal volume decreased, 5 g/L TCA
29 increased CAT activity 3.21-fold of control in these rats. The level of CAT induced by 5 g/L
30 DCA was over 10-fold of control in the rat while peroxisome # was only 2.5-fold of control and
31 cytoplasmic volume 1.4-fold of control. Thus, the fold increases for these three measures were
32 not the same for DCA treatment and for TCA in rats. Nevertheless for CAT, DCA was a
33 stronger inducer in rats than was TCA. In B6C3 Fl mice 5 g/L TCA and 5 g/L DCA induced
34 CAT activity to a similar extent (4.50- and 5.61-fold of control, respectively). The magnitude of
35 CAT induction was similar to that of peroxisome # for both 5 g/L DCA and 5 g/L TCA and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-13 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 lower than PCO activity in DCA-treated mice and cytoplasmic volume in TCA-treated mice by
2 about half. Thus, using CAT as the marker of peroxisome proliferation, the rat was more
3 responsive than the mouse to DC A and nearly as responsive to TCA as the mouse at this high
4 dose in these two specific strains. These data illustrate the difficulty of using only one measure
5 for peroxisome proliferation and shows that the magnitude of increased PCO activity is not
6 necessarily predictive of the peroxisome # or cytoplasmic volume or CAT activity. The
7 difficulty of interpretation of the data from so few animals and sections for the electron
8 microscopy analysis, and the low number of animals for PCO activity and CAT activity (n = 3 to
9 6), the high dose studied (5 g/L), and the selection of a rat strain that appears to be more resistant
10 to this activity but more susceptible to toxicity than the others tested, should be taken into
11 account before conclusions can be made about differences between these chemicals for
12 peroxisome activity between species.
13 Of note is that PCO activity was also shown to be increased by corn oil alone in F344 rats
14 and to potentiate the induction of PCO activity of TCA. After 10 days of exposure to either
15 water, corn oil, 200 mg/kg/d TCA in corn oil or 200 mg/kg TCA in water via gavage dosing,
16 there was 1.40-fold PCO activity from corn oil treatment alone in comparison to water, a
17 1.79-fold PCO activity from TCA in water treatment in comparison to water, and a 3.14-fold
18 PCO activity from TCA in corn oil treatment in comparison to water.
19 The authors provided data for 3 concentrations of DC A and TCA for S-D and for one
20 experiment in the B6C3F1 mouse for examination of changes in body and percent liver/body
21 weight ratios (1, 2, or 5 g/L DCA or TCA) after 14 days of exposure. As noted above, not only
22 did the 5 g/L exposure concentration of DCA result in mortality in the S-D strain of rat, but the
23 5 g/L and 2 g/L concentrations of DCA were reported to decrease body weight (-20 and 25%,
24 respectively). The 5 g/L dose of TCA was also reported to induce a statistically significant
25 decrease in body weight in the S-D rat. There were no differences in final body weight in any of
26 the mice exposed to TCA or DCA. As noted above no TCA or DCA exposure group of S-D rats
27 was reported to have a statistically significant increase in percent liver/body weight ratio over
28 control. For the B6C3F1 male mice, the percent liver/body weight ratio was 1.22-, 1.27-, and
29 1.55-fold of control after exposure to 1, 2, and 5 g/L DCA, respectively, and 1.08-, 1.16-, and
30 1.39-fold of control after exposure to 1, 2, and 5 g/L TCA, respectively. Thus, for DCA there
31 was only a 20% increase in liver weight corresponding to the 2-fold increase between the 1 and
32 2 g/L exposure levels of DCA. Between the 2 and 5 g/L exposure concentrations of DCA there
33 was a 2-fold increase in liver weight corresponding to a 2.5-fold increase in exposure
34 concentration. For TCA, the magnitude of increase in dose was reported to be proportional to
35 the magnitude of increase in percent liver/body weight ratio in the B6C3 Fl male mouse. As
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-136 DRAFT—DO NOT CITE OR QUOTE
-------
1 stated above, the correspondence between magnitude of dose and percent liver weight for TCA
2 exposure in this experiment differed from the other experiment reported for this strain of mouse
3 and also differed from the other 3 strains of mice examined in this study where the magnitude in
4 liver weight gain was much less than exposure concentration.
5
6 E.2.3.2. Subchronic and Chronic Studies of Dichloroacetic Acid (DCA) and Trichloroacetic
1 Acid (TCA)
8 Several experiments have been conducted with exposure to DCA and TCA, generally at
9 very high levels with a limited dose range, for less periods of time than standard carcinogenicity
10 bioassays, and with very limited information on any endpoints other than the liver tumor
11 induction. Caldwell and Keshava (2006) and Caldwell et al. (2008b) have examined these
12 studies for inferences of modes of action for TCE. Key studies are briefly described below for
13 comparative purposes of results reported in TCE studies.
14
15 E.2.3.2.1. Snyder et al., 1995. Studies of TCE have reported either no change or a slight
16 increase in apoptosis only after a relatively high exposure level (Dees and Travis, 1993; Channel
17 et al., 1998). Inhibition of apoptosis, which has been suggested to prevent removal of "initiated"
18 cells from the liver and lead to increased survival of precancerous cells, has been proposed as
19 part of the MOA for peroxisome proliferators (see Section E.3.4). The focus of this study was to
20 examine whether DCA, which has been shown to inhibit DNA synthesis after an initial transient
21 increase (see Section E.2.3.3, below), also alters the frequency of spontaneous apoptosis in mice.
22 This study exposed 28-day old male B6C3F1 male mice (n = 5) to 0, 0.5 or 5.0 g/L buffered
23 DCA in drinking water for up to 30 days (Phase I = 5-15 days exposure and Phase II =
24 20-30 days treatment). Portions of the left lobe of the liver were prepared for histological
25 examination after H&E staining. Hepatocyte number was determined by counting nuclei in
26 50 fields with nonparenchymal cell nuclei excluded on the basis of nuclear size. Multinucleate
27 cells were counted as one cell. Apoptotic cells were visualized by in situ TDT nick end-labeling
28 assay from 2-4 different liver sections from each control or treated animal. The average number
29 of apoptotic cells was then determined for each animal in each group. The authors reported that
30 in none of the tissues examined were necrotic foci observed, there was no any indication of
31 lymphocyte or neutrophil infiltration indicative of an inflammatory response, and suggested that
32 no necrotic cells contributed to the responses in their analysis.
33 Control animals were reported to exhibit apoptotic frequencies ranging from -0.04 to
34 0.085% and that over the 30-day period the frequency rate declined. The authors suggested that
35 this result is consistent with reports of the livers of these young animals undergoing rapid
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-137 DRAFT—DO NOT CITE OR QUOTE
-------
1 changes in cell death and proliferation. They note that animals receiving 0.5 g/L DCA also had a
2 similar trend of decreasing apoptosis with age, supportive of the decrease being a physiological
3 phenomenon. The 0.5 g/L exposure level of DCA was reported to decrease the percentage of
4 apoptotic hepatocytes as the earliest time point studied and to remain statistically significantly
5 decreased from controls from 5 to 30 days of exposure. The rate of apoptosis ranged from
6 -0.025 to 0.060% after 0.5 g/L DCA exposure during the 30-day period (i.e., and -30-40%
7 reduction). Animals receiving the 5.0 g/L DCA dose exhibited a significant reduction at the
8 earliest time point that was sustained at a similar level and statistically significant throughout the
9 time-course of the experiment (percent apoptosis ranged from 0.015-0.030%). The results of
10 this study not only provides a baseline of apoptosis in the mouse liver, which is very low, but
11 also to show the importance of taking into account the effects of age on such determinations.
12 The authors reported that the for rat liver the estimated frequency of spontaneous apoptosis to be
13 -0.1% and therefore, greater than that of the mouse. The significance of the DCA-induced
14 reduction in apoptosis, of a level that is already inherently low in the mouse, for the MOA for
15 induction of cancer is difficult to discern.
16
17 E.2.3.2.2. Mather et al, 1990. This 90-day study in male S-D rats examined the body and
18 organ weight changes, liver enzyme levels, and PCO activity in livers from rats treated with
19 estimated concentrations of 3.9, 35.5, 345 mg/kg day DCA or 4.1, 36.5, or 355 mg/kg/d TCA
20 from drinking water exposures (i.e., 0, 50, 500 and 5,000 ppm or 0.05, 0.5, or 5.0 g/L DCA or
21 TCA in the drinking water). All dose levels of DCA and TCA were reported to result in a dose-
22 dependent decrease in fluid intake at 2 months of exposure. The rats were 9 (DCA) or 10 (TCA)
23 weeks old at the beginning of the study (n = 10/group). Animals with body weights that varied
24 more than 20% of mean weights were discarded from the study. The DCA and TCA solutions
25 were neutralized. The mean values for initial weights of the animals in each test group varied
26 less than 3%. DCA treatment induced a dose-related decrease in body weight that was
27 statistically significant at the two highest levels (i.e., a 6, 9.5, and 17% decrease from control).
28 TCA treatment also resulted in lower body weights that were not statistically significant (i.e.,
29 2.1, 4.4, and 5.9%). DCA treatments were reported to result in a dose-related increase in
30 absolute liver weights (1.01-, 1.13-, and 1.36-fold of control that were significantly different at
31 the highest level) and percent liver/body weight ratios (1.07-, 1.24-, and 1.69-fold of control that
32 were significant at the two highest dose levels). TCA treatments were reported to not result in
33 changes in either absolute liver weights or percent liver/body weight ratios with the exception of
34 statistically significant increase in percent liver/body weight ratios at the highest level of
35 treatment (1.02-fold of control). Total serum protein levels were reported to be significantly
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-13 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 depressed in all animals treated with DCA with animals in the two highest dose groups also
2 exhibiting elevations of alkaline phosphatase. Alanine-amino transferase levels were reported to
3 be elevated only in the highest treatment group. No consistent treatment-related effect on serum
4 chemistry was reported to be observed for the TCA-treated animals with data not shown. In
5 terms of PCO activity, there was only a mild increase at the highest dose of 15% for TCA and a
6 2.5-fold level of control for DCA treatment that were statistically significant. The difference in
7 PCO activity between control groups for the DCA and TCA experiments was reported to be
8 33%. No treatment affect was reported to be apparent for hepatic microsomal enzymes, or
9 measures of immunotoxicity for either DCA or TCA but data were not shown. Focal areas of
10 hepatocellular enlargement in both DCA- and TCA-treated rats were reported to be present with
11 intracellular swelling more severe with the highest dose of DCA treatment. Livers from DCA
12 treated rats were reported to stain positively for PAS, indicating significant amounts of glycogen
13 with TCA treated rats reported to display "less evidence of glycogen accumulation." Of note is
14 that, in this study of rats, DCA was reported to induce a greater level of PCO activity than did
15 TCA.
16
17 E.2.3.2.3. Parrishetal.,1996. Parrish etal. (1996) exposed male B6C3F1 mice (8 weeks old
18 and 20-22 g upon purchase) to TCA or DCA (0, 0.01, 0.5, and 2.0 g/L) for 3 or 10 weeks
19 (n = 6). Livers were excised and nuclei isolated for examination of 8-OHdG and homogenates
20 examined for cyanide insensitive acyl-CoA oxidase (AGO) and laurate hydroxylase activity.
21 The authors noted that control values between experiments varied as much as a factor of 2-fold
22 for PCO activity and that data were presented as percent of concurrent controls. Initial body
23 weights for treatment groups were not presented and thus, differences in mean values between
24 the groups cannot be ascertained.
25 Final body weights were reported to not be statistically significantly changed by DCA or
26 TCA treatments at 21 days or 71 days of treatment (all were within -8% of controls). The mean
27 percent liver/body ratios were reported to be 5.4, 5.3, 6.1, and 7.2% for control, 0.1, 0.5, and
28 2.0 g/L TCA, respectively and 5.4, 5.5, 6.7, and 7.9% for control, 0.1, 0.5, and 2.0 g/L DCA,
29 respectively after 21 days of exposure. This represents 0.98-, 1.13-, and 1.33-fold of control
30 levels with these exposure levels of TCA and 1.02-, 1.24-, and 1.46-fold of control levels with
31 DCA after 21 days of exposure. For 71 days of exposure the mean percent liver/body ratios were
32 reported to be 5.1, 4.6, 5.8, and 6.9% for control, 0.1, 0.5, and 2.0 g/L TCA, respectively and 5.1,
33 5.1, 5.9, and 8.5% for control, 0.1, 0.5, and 2.0 g/L DCA, respectively. This represents 0.90-,
34 1.14-, and 1.35-fold of control with TCA exposure and 1.0-, 1.15-, and 1.67-fold of control with
35 DCA exposure after 71 days of exposure. The magnitude of difference between the 0.1 and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-139 DRAFT—DO NOT CITE OR QUOTE
-------
1 0.5 g/L TCA doses is 5 and 0.5 and 2.0 g/L doses is 4-fold. For the 21-day and 71-day exposures
2 the magnitudes of the increases in percent liver/body weight over control values were greater for
3 DCA than TCA exposure at same concentration with the exception of 0.5 g/L doses at 71 days in
4 which both TCA and DCA induced similar increases. For TCA, the 0.01 g/L dose produces a
5 similar 10% decrease in percent liver/body weight. Although there was a 4-fold increase in
6 magnitude between the 0.5 and 2.0 g/L TCA exposure concentrations, the magnitude of increase
7 for percent liver/body weight increase was 2.5-fold between them at both 21 and 71 days of
8 exposure. For DCA, the 0.1 g/L dose was reported to have a similar value as control for percent
9 liver/body weight ratio. Although there was a 4-fold difference in dose between the 0.5 and
10 2.0 g/L DCA exposure concentrations, there was a ~2-fold increase in percent liver/body weight
11 increase at 21 days and -4.5-fold increase at 71 days.
12 As a percentage of control values, TCA was reported to induce a dose-related increase in
13 PCO activity at 21 days (-1.5-, 2.2-, and ~4.1-fold of control, for 0.1, 0.5, and 2 g/L TCA
14 exposures). Only the 2.0 g/L dose of DCA was reported to induce a statistically significant
15 increase at 21-days of exposure of PCO activity over control (~1.8-fold of control) with the 0.1
16 and 0.5 g/L exposure PCO activity to be slightly less than control values (-20% less). Thus,
17 although there was no increase in percent liver/body weight at 0.1 g/L TCA, the PCO activity
18 was reported to be increased by -50% after 21 days. A 13% increase in liver weight at 0.5 g/L
19 TCA was reported to be associated with 2.2-fold of control level of PCO activity and a 33%
20 increase in liver weight after 2.0 g/L TCA to be associated with 4.1-fold of control level of PCO
21 activity. Thus, increases in PCO activity were not necessarily correlated with concurrent TCA-
22 induced increases in liver weight and the magnitudes of increase in liver weight between 0.5 and
23 2.0 g/L TCA (2.5-fold) was greater than the corresponding increase in PCO activity (1.8-fold of
24 control). Although there was a 20-fold difference in TCA dose, the magnitude of increase in
25 PCO activity between 0.1 and 2.0 g/L TCA was -2.7-fold. As stated above, the 4-fold difference
26 in TCA dose at the two highest levels resulted in a 2.5-fold increase in liver weight. For DCA,
27 the increases in liver weight at 0.1 and 0.5 g/L DCA exposures were not associated with
28 increased PCO activity after 21 days of exposure. The 2.0 g/L DCA exposure concentration was
29 reported to induce 1.8-fold of control PCO activity. After 71 days of treatment, TCA induced a
30 dose-related increase in PCO activity that was approximately twice the magnitude as that
31 reported at 21 days (i.e., -9-fold greater at 2.0 g/L level). After 71 days, for DCA the 0.1 and
32 0.5 g/L doses produced a statistically significant increase in PCO activity (-1.5- and 2.5-fold of
33 control, respectively). The administration of 1.25 g/L clofibric acid in drinking water was used
34 as a positive control and reported to induce ~6-7-fold of control PCO activity at 21 and 71 days
35 of exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-140 DRAFT—DO NOT CITE OR QUOTE
-------
1 Laurate hydroxylase activity was reported to be elevated significantly only by TCA at
2 21 days (2.0 g/L TCA dose only) and to increased to approximately the same extent (-1.4 to
3 1.6-fold of control values) at all doses tested. For 0.1 g/L DCA the laurate hydroxylase activity
4 was reported to be similar to that of 0.1 g/L TCA (~1.4-fold of control) but to be ~1.2-fold of
5 control at both the 0.5 and 2.0 g/L DCA exposures. At 71 days, both the 0.5 and 2.0 g/L TCA
6 exposures induced a statistically significant increase in laurate hydroxylase activity (i.e., 1.6- and
7 2.5-fold of control, respectively) with no change after DCA exposure. The actual data rather
8 than percent of control values were reported for laurate hydroxylase activity. The control values
9 for laurate hydroxylase activity varied 1.7-fold between 21 and 71 days experiments. The results
10 for 8-OHdG levels are discussed in Section E.3.4.2.3, below. Of note is that the increases in
11 PCO activity noted for DCA and TCA were not associated with 8-OHdG levels (which were
12 unchanged, see Section E.3.4.2.3, below) and also not with changes laurate hydrolase activity or
13 percent liver/body weight ratio increases observed after either DCA or TCA exposure. A
14 strength of this study is that is examined exposure concentrations that were lower than those
15 examined in many other short-term studies of DCA and TCA.
16
17 E.2.3.2.4. Bull et al, 1990. The focus of this study was the determination of "dose-response
18 relationships in the tumorigenic response to these chemicals [sic DCA and TCA] in B6C3F1
19 mice, determine the nature of the nontumor pathology that results from the administration of
20 these compounds in drinking water, and test the reversibility of the response." Male and female
21 B6C3F1 mice (age 37 days) were treated from 15 to 52 weeks with neutralized TCA and TCA.
22 A highly variable number and generally low number of animals were reported to be examined in
23 the study with n = 5 for all time periods except for 52 weeks where in males the n = 35 for
24 controls, n =11 for 1 g/L DCA, n = 24 for 2 g/L DCA, n = 11 for 1 g/L TCA, and n = 24 for
25 2 g/L TCA exposed mice. Female mice were only examined after 52 weeks of exposure and the
26 number of animals examined was n = 10 for control, 2 g/L DCA, and 2 g/L TCA exposed mice.
27 "Lesions to be examined histologically for pathological examination were selected by a random
28 process" with lesions reported to be selected from 31 of 65 animals with lesions at necropsy. 73
29 of 165 lesions identified in 41 animals were reported to be examined histologically. All
30 hyperplastic nodules, adenomas and carcinomas were lumped together and characterized as
31 hepatoproliferative lesions. Accordingly there were only exposure concentrations available for
32 dose-response analyses in males and only "multiplicity of hepatoproliferative lesions" were
33 reported from random samples. Thus, these data cannot be compared to other studies and are
34 unsuitable for dose-response with inadequate analysis performed on random samples for
35 pathological examination. The authors state that some of the lesions taken at necropsy and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-141 DRAFT—DO NOT CITE OR QUOTE
-------
1 assumed to be proliferative were actually histologically normal, necrotic, or an abscess as well.
2 It is also limited by a relatively small number of animals examined in regard to adequate
3 statistical power to determine quantitative differences. Similar concerns were raised by
4 Caldwell et al. (2008b) with a subsequent study (e.g., Bull et al., 2002). For example, the
5 authors report that 5/11 animals had "lesions" at 1 g/L TCA at 52 weeks and 19/24 animals had
6 lesions at 2 g/L TCA at 52 weeks. However, while 7 lesions were examined in 5 mice bearing
7 lesions at 1 g/L TCA, only 16 of 30 lesions from 11 of the 19 animals bearing lesions examined
8 in the 2 g/L TCA group. Therefore, almost half of the mice with lesions were not examined
9 histologically in that group along with only half of the "lesions."
10 The authors reported the effects of DC A and TCA exposure on liver weight and percent
11 liver/body changes (m ± SEM) and these results gave a pattern of hepatomegaly generally
12 consistent with short-term exposure studies. The authors report "no treatment produced
13 significant changes in the body weight or kidney weight of the animals (data not shown)" In
14 male mice (n = 5) at 37 weeks of exposure, liver weights were reported to be 1.6 ± 0.1, 2.5 ± 0.1,
15 and 1.9 ± 0.1 g for control, 2 g/L DCA, and 2 g/L TCA exposed mice, respectively. The percent
16 liver/body weights were reported to be 4.1% ± 0.3%, 7.3% ± 0.2%, and 5.1% ± 0.1% for control,
17 2 g/L DCA, and 2 g/L TCA exposed mice, respectively. In male mice at 52 weeks of exposure,
18 liver weights were reported to be 1.7 ±0.1, 2.5 ±0.1, 5.1 ±0.1, 2.2 ±0.1, and 2.7 ±0.1 gfor
19 control (n = 35), 1 g/L DCA (n = 11), 2 g/L DCA (n = 24), 1 g/L TCA (n = 11), and 2 g/L TCA
20 (n = 24) exposed mice, respectively. In male mice at 52 weeks of exposure, percent liver/body
21 weights were reported to be 4.6% ± 0.1%, 6.5% ± 0.2%, 10.5% ± 0.4%, 6.0% ± 0.3%, and
22 7.5% ± 0.5% for control, 1 g/L DCA, 2 g/L DCA, 1 g/L TCA, and 2 g/L TCA exposed mice,
23 respectively. For female mice (n = 10) at 52 weeks of exposure, liver weights were reported to
24 be 1.3 ± 0.1, 2.6 ± 0.1, and 1.7 ± 0.1 g for control, 2 g/L DCA, and 2 g/L TCA exposed mice,
25 respectively. The percent liver/body weights were reported to be 4.8% ± 0.3%, 9.0% ± 0.2%,
26 and 6.0% ± 0.3% for control, 2 g/L DCA, and 2 g/L TCA exposed mice, respectively. Although
27 the number of animals examined varied 3-fold between treatment groups in male mice, the
28 authors reported that all DCA and TCA treatments were statistically increased over control
29 values for liver weight and percent body/liver weight in both genders of mice. In terms of
30 percent liver/body weight ratio, female mice appeared to be as responsive as males at the
31 exposure concentration tested. Thus, hepatomegaly reported at these exposure levels after short-
32 term exposures appeared to be further increased by chronic exposure with equivalent levels of
33 DCA inducing greater hepatomegaly than TCA.
34 Interestingly, after 37 weeks of treatment and then a cessation of exposure for 15 weeks
35 liver weights were assessed in control male mice, 2 g/L DCA treated mice, and 2 g/L TCA
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-142 DRAFT—DO NOT CITE OR QUOTE
-------
1 treated mice (n = 11 for each group but results for controls were pooled and therefore, n = 35).
2 Liver weights were reported to be 1.7 ± 0.1, 2.2 ± 0.1, and 1.9 ± 0.1 g for control, 2 g/L DCA,
3 and 2 g/L TCA exposed mice, respectively. The percent liver/body weights were reported to be
4 4.6% ± 0.1%, 5.7% ± 0.3%, and 5.4% ± 0.2% for control, 2 g/L DCA, and 2 g/L TCA exposed
5 mice, respectively. After 15 weeks of cessation of exposure, liver weight and percent liver/body
6 weight were reported to still be statistically significantly elevated after DCA or TCA treatment.
7 The authors partially attribute the remaining increases in liver weight to the continued presence
8 of hyperplastic nodules in the liver. The authors state that because of the low incidence of
9 lesions in the control group and the two groups that had treatments suspended, all the lesions
10 from these groups were included for histological sectioning. However, the authors present a
11 table indicating that, of the 23 lesions detected in 7 mice exposed to DCA for 37 weeks, 19 were
12 examined histologically. Therefore, groups that were exposed for 52 weeks had a different
13 procedure for tissue examination as those at 37 weeks. In terms of liver tumor induction, the
14 authors stated that "statistical analysis of tumor incidence employed a general linear model
15 ANOVA with contrasts for linearity and deviations from linearity to determine if results from
16 groups in which treatments were discontinued after 37 weeks were lower than would have been
17 predicted by the total dose consumed." The multiplicity of tumors observed in male mice
18 exposed to DCA or TCA at 37 weeks and then sacrificed at 52 weeks were reported by the
19 authors to have a response in animals that received DCA very close to that which would be
20 predicted from the total dose consumed by these animals. The response to TCA was reported by
21 the authors to deviate significantly (p = 0.022) from the linear model predicted by the total dose
22 consumed. Multiplicity of lesions per mouse and not incidence was used as the measure. Most
23 importantly the data used to predict the dose response for "lesions" used a different methodology
24 at 52 weeks than those at 37 weeks. Not only were not all animal's lesions examined but foci,
25 adenomas, and carcinomas were combined into one measure. Therefore, foci, of which a certain
26 percentage have been commonly shown to spontaneously regress with time, were included in the
27 calculation of total "lesions." Pereira and Phelps (1996) note that in initiated mice treated with
28 DCA, the yield of altered hepatocytes decreases as the tumor yields increase between 31 and
29 51 weeks of exposure suggesting progression of foci to adenomas. Initiated and noninitiated
30 control mice also had fewer foci/mouse with time. Because of differences in methodology and
31 the lack of discernment between foci, adenomas, and carcinomas for many of the mice exposed
32 for 52 weeks, it is difficult to compare differences in composition of the "lesions" after cessation
33 of exposure. For TCA treatment the number of animals examined for determination of which
34 "lesions" were foci, adenomas, and carcinomas was 11 out of the 19 mice with "lesions" at
35 52 weeks while all 4 mice with lesions after 37 weeks of exposure and 15 weeks of cessation
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-143 DRAFT—DO NOT CITE OR QUOTE
-------
1 were examined. For DCA treatment the number of animals examined was only 10 out of
2 23 mice with "lesions" at 52 weeks while all 7 mice with lesions after 37 weeks of exposure and
3 15 weeks of cessation were examined. Most importantly, when lesions were examined
4 microscopically then did not all turn out to be preneoplastic or neoplastic. Two lesions appeared
5 "to be histologically normal" and one necrotic. Not only were a smaller number of animals
6 examined for the cessation exposure than continuous exposure but only the 2 g/L exposure levels
7 of DCA and TCA were studied for cessation. The number of animals bearing "lesions" at 37 and
8 then 15 week cessation weeks was 7/11 (64%) while the number of animals bearing lesions at
9 5 weeks was 23/24 (96%) after 2 g/L DCA exposure. For TCA the number of animals bearing
10 lesions at 37 weeks and then 15 weeks cessation was 4/11 (35%) while the number of animals
11 bearing lesions at 52 weeks was 19/24 (80%). While suggesting that cessation of exposure
12 diminished the number of "lesions," conclusions regarding the identity and progression of those
13 lesion with continuous versus noncontinuous DCA and TCA treatment are tenuous.
14 Macroscopically, the "livers of many mice receiving DCA in their drinking water
15 displayed light colored streaks on the surface" at every sacrifice period and "corresponded with
16 multi-focal areas of necrosis with frequent infiltration of lymphocytes." At the light microscopic
17 level, the lesions were described to also be present in the interior of the liver as well. For
18 TCA-treated mice, "similar necrotic lesions were also observed... but at a much lower
19 frequency, making it difficult to determine if they were treatment-related." Control animals were
20 reported not to show degenerative changes. "Marked cytomegaly" was reported for mice treated
21 with either 1 or 2 g/L DCA "throughout the liver" In regard to cell size the authors did not give
22 any description in the methods section of the paper as to how sections were selected for
23 morphometric analysis or what areas of the liver acinus were examined but reported after
24 52 weeks of treatment the long axis of hepatocytes measured (mean ± S.E.) 24.9 ± 0.3,
25 38.5 ± 1.0, and 29.3 ± 1.4 um in control, DCA- and TCA-treated mice, respectively.
26 Mice treated with TCA (2 g/L) for 52 weeks were reported to have livers with
27 "considerable dose-related accumulations of lipofuscin." However, no quantitative analyses
28 were presented. A series of figures representative of treatment showed photographs (1,000*) of
29 lipofuscin fluorescence indicating greater fluorescence in TCA treated liver than control or DCA
30 treated liver.
31 A series of photographs of H&E sections in the report (see Figures 2a, b and c) are shown
32 as representative histology of control mice, mice treated with 2 g/L DCA and 2 g/L TCA. The
33 area of the liver from which the photographs were taken did not include either portal tract or
34 central veins and the authors did not give the zone of the livers from which they were taken. The
35 figure representing TCA treatment shows only a mild increase in cell volume in comparison to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-144 DRAFT—DO NOT CITE OR QUOTE
-------
1 controls, while for DCA treatment the hepatocyte diameter was greatly enlarged, pale stained so
2 that cytoplasmic contents appear absent, nuclei often pushed to the cell perimeter, and the
3 sinusoids appearing to be obscured by the swollen hepatocytes. The apparent reduction of
4 sinusoidal volume by the enlarged hepatocytes raises the possibility of decreased blood flow
5 through the liver, which may have been linked to focal areas of necrosis reported for this high
6 exposure level. In a second set of figures, glycogen accumulation was shown with PAS staining
7 at the same level of power (400x) for the same animals. In control animals PAS positive
8 material was not uniformly distributed between or within hepatocytes but send to show a zonal
9 pattern of moderate intensity. PAS positive staining (which the authors reported to be glycogen)
10 appeared to be slightly less than controls but with a similar pattern in the photograph
11 representing TCA exposure. However, for DCA the photograph showed a uniform and heavy
12 stain within each hepatocyte and across all hepatocytes. The authors stated in the results section
13 of the paper that "the livers of TCA-treated animals displayed less evidence of glycogen
14 accumulation and it was more prominent in periportal than centrilobular portions of the liver
15 acinus." In their abstract they state "TCA produced small increases in cell size and a much more
16 modest accumulation of glycogen." Thus, the statement in the text, which is suggestive that
17 TCA induced an increase in glycogen over controls that was not as much as that induced by
18 DCA, and the statement in the abstract which concludes TCA exposure increased glycogen is not
19 consistent with the photographs. In the photograph shown for TCA there is less not more PAS
20 positive staining associated with TCA treatment in comparison to controls. In Sanchez and Bull
21 (1990) the authors report that "TCA exposure induced a much less intense level of PAS staining
22 that was confined to periportal areas" but do not compare PAS staining to controls but only to
23 DCA treatment. In the discussion section of the paper the authors state "Except for a small
24 increase in liver weight and cell size, the effects produced by DCA were not observed with
25 TCA." Thus, there seems to be a discrepancy with regard to what the effects of TCA are in
26 relation to control animals from this report that has caused confusion in the literature.
27 Kato-Weinstein et al. (2001) reported that in male mice exposed to DCA and TCA the DCA
28 increased glycogen and TCA decreased glycogen content of the liver using chemical
29 measurement of glycogen in liver homogenates and using ethanol-fixed sections stained with
30 PAS, a procedure designed to minimize glycogen loss.
31
32 E.2.3.2.5. Nelson et al, 1990. Nelson et al. (1990) reported that they used the same exposure
33 paradigm as Herren-Freund et al. (1987), with little description of methods used in treatment of
34 the animals. Male B6C3F1 mice were reported to be exposed to DCA (1 or 2 g/L) or TCA (1 or
35 2 g/L) for 52 weeks. The number of animals examined for nontumor tissue was 12 for controls.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-145 DRAFT—DO NOT CITE OR QUOTE
-------
1 The number of animals varied from 2 to 8 for examination of nontumor tissue, hyperplastic
2 nodules, and carcinoma tissues for c-Myc expression. There was no description for how
3 hyperplastic nodules were defined and whether they included adenomas and foci. For the
4 52-week experiments, the results were pooled for lesions that had been obtained by exposure to
5 the higher or lower concentrations of DCA or TCA (i.e., the TCA results are for lesions induced
6 by either 1.0 g/L or 2.0 g/L TCA). A second group of mice were reported to be given either
7 DCA or TCA for 37 weeks and then normal drinking water for the remaining time till 52 weeks
8 with no concentrations given for the exposures to these animals. Therefore, it is impossible to
9 discern what dose was used for tumors analyzed for c-Myc expression in the 37-week treatment
10 groups and if the same dose was used for 37 and 52 week results. Autoradiography was
11 described for 3 different sections per animal in 5 different randomly chosen high power fields
12 per section. The number of hyperplastic nodules or the number of carcinomas per animal
13 induced by these treatments was not reported nor the criteria for selection of lesions for c-myc
14 expression. Apparently a second experiment was performed to determine the expression of
15 c-H-ras. Whereas in the first experiment there were no hyperplastic nodules, in the second
16 1-control animal was reported to have a hyperplastic nodule. The number of control animals
17 reported to be examined for nontumor tissue in the second group was 12. The numbers of
18 animals in the second group was reported to vary from 1 to 7 for examination of nontumor tissue,
19 hyperplastic nodules, and carcinoma tissues for c-H-ras expression. The number of animals per
20 group for the investigation of H-ras did not match the numbers reported for that of c-Myc. The
21 number of animals treated to obtain the "lesion" results was not presented (i.e., how many
22 animals were tested to get a specific number of animals with tumors that were then examined).
23 The number of lesions assessed per animal was not reported.
24 At 52 weeks of exposure, hyperplastic nodules (n = 8 animals) and carcinomas
25 (n = 6 animals) were reported to have ~2-fold expression of c-Myc relative to nontumor tissue
26 (n = 6 animals) after DCA treatment. After 37 weeks of DCA treatment and cessation of
27 exposure, there was a -30% increase in c-Myc in hyperplastic nodules (n = 4 animals) that was
28 not statistically significant. There were no carcinomas reported at this time. After 52 weeks of
29 TCA exposure, there was ~2-fold of nontumor tissue reported for c-Myc in hyperplastic nodules
30 (n = 6 animals) and ~3-fold reported for carcinomas (n = 6 animals). After 37 weeks of TCA
31 exposure there was ~2-fold c-Myc in hyperplastic nodules (n = 2 animals) that was not
32 statistically significant and ~2.6-fold increase in carcinomas (n = 3 animals) that was reported to
33 be statistically significant over nontumor tissue. There was no difference in c-Myc expression
34 between untreated animals and nontumor tissue in the treated animals.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-146 DRAFT—DO NOT CITE OR QUOTE
-------
1 The authors reported that c-Myc expression in TCA-induced carcinomas was "almost 6
2 times that in control tissue (corrected by subtracting nonspecific binding)," and concluded that
3 c-Myc in TCA-induced carcinomas was significantly greater than in hyperplastic nodules or
4 carcinomas and hyperplastic nodules induced by DCA. However, the c-myc expression reported
5 as the number of grains per cells was -2.6-fold in TCA-induced carcinomas and ~2-fold in
6 DCA-induced carcinomas than control or nontumor tissue at 52 weeks. The hyperplastic nodules
7 from DCA- and TCA-treatments at 52 weeks gave identical ratios of ~2-fold. In 3 animals per
8 treatment, c-Myc expression was reported to be similar in "selected areas of high expression" for
9 either DCA or TCA treatments of 52 weeks.
10 There did not appear to be a difference in c-H-ras expression between control and
11 nontumor tissue from DCA- or TCA-treated mice. The levels of c-H-ras transcripts were
12 reported to be "slightly elevated" in hyperplastic nodules induced by DCA (-67%) or TCA
13 (-43%) but these elevations were not statistically significant in comparison to controls.
14 However, carcinomas "derived from either DCA- or TCA-treated animals were reported to have
15 significantly increased c-H-ras levels relative to controls." The fold increase of nontumor tissue
16 at 52 weeks for DCA-induced carcinomas was -2.5-fold and for TCA induced carcinomas
17 -2.0-fold. Again the authors state that "if corrected for nonspecific hybridization, carcinomas
18 expressed approximately 4 times as much c-H-ras than observed in surrounding tissues" Given
19 that control and nontumor tissue results were given as the controls for the expression increases
20 observed in "lesions," it is unclear what this the usefulness of this "correction" is. The authors
21 reported that "focal areas of increased expression of c-H-ras were not observed within
22 carcinomas."
23 The limitations of this experiment include uncertainty as to what doses were used and
24 how many animals were exposed to produce animals with tumors. In addition results of differing
25 doses were pooled and the term hyperplastic nodule, undefined. The authors state that c-Myc
26 expression in itself is not sufficient for transformation and that its over expression commonly
27 occurs in malignancy. They also state that "Unfortunately, the limited amount of tissue available
28 prevented a more serious pursuit of this question in the present study." In regard to the effects of
29 cessation of exposure, the authors do not present data on how many animals were tested with the
30 cessation protocol, what doses were used, and how many lesions comprised their results and
31 thus, comparisons between these results and those from 52 weeks of continuous exposure are
32 hard to make. Quantitatively, the small number of animals, whose lesions were tested, was
33 n = 2-4 for the cessation groups. Bull et al. (1990) is given as the source of data for the
34 cessation experiment (see Section E.2.3.2.1, above).
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-147 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.3.2.6. DeAngelo et al, 1999. The focus of this study was to "determine a dose response
2 for the hepatocarcinogenicity of DC A in male mice over a lifetime exposure and to examined
3 several modes of action that might underlie the carcinogenic process." As DeAngelo et al
4 pointed out, many studies of DC A had been conducted at high concentrations and for less than
5 lifetime studies, and therefore, of suspect relevance to environmental concentrations. This study
6 is one of the few that examined DCA at a range of exposure concentrations to determine a dose-
7 response in mice. The authors concluded that DCA-induced carcinogenesis was not dependent
8 on peroxisome proliferation or chemically sustained proliferation. The number of hepatocellular
9 carcinomas/animals was reported to be significantly increased over controls at all DCA
10 treatments including 0.05 g/L and a no-observed-effect level (NOEL) not observed. Peroxisome
11 proliferation was reported to be significantly increased at 3.5 g/L DCA only at 26 weeks and did
12 not correlate with tumor response. No significant treatment effects on labeling of hepatocytes
13 (as a measure of proliferation) outside proliferative lesions were also reported and thus, that
14 DCA-induced liver cancer was not dependent on peroxisome proliferation or chemically
15 sustained cell proliferation.
16 MaleB6C3Fl mice were 28-30 days of age at the start of study and weighed 18-21 g (or
17 -14% range). They were exposed to 0, 0.05, 0.5, 1.0, 2.0, and 3.5 g/L DCA via drinking water
18 as a neutralized solution. The time-weighted mean daily water consumption calculated over the
19 100-week treatment period was reported to be 147, 153, 158, 151, 147, and 124 (84% of
20 controls) mL/kg/day for 0, 0.05, 0.5, 1, 2, and 3.5 g/L DCA, respectively. The number of
21 animals reported as used for interim sacrifices were 35, 30, 30, 30 and 30 for controls, 0.5, 1.0,
22 2.0, and 3.5 g/L DCA treated groups respectively (i.e., 10 mice per treatment group at interim
23 sacrifices of 26, 52 and 78 weeks). The number of animals at final sacrifice were reported to be
24 50, 33, 24, 32, 14 and 8 for controls, 0.05, 0.5, 1.0, 2.0, and 3.5 g/L DCA treated groups
25 respectively. The number of animals with unscheduled deaths before final sacrifice were
26 reported to be 3, 2, 1, 9, 11 and 8 for controls, 0.05, 0.5, 1.0, 2.0, and 3.5 g/L DCA treated
27 groups respectively. The Authors reported that early mortality tended to occur from liver cancer.
28 The number of animals examined for pathology were reported to be 85, 33, 55, 65, 51, and 41 for
29 controls, 0.05, 0.5, 1.0, 2.0, and 3.5 g/L DCA treated groups respectively. The experiment was
30 conducted in two parts with control, 0.5, 1.0 L, 2.0, and 3.5 g/L groups treated and then 1 months
31 later a second group consisting of 30 control group mice and 35 mice in a 0.05 g/L DCA
32 exposure group studied. The authors reported not difference in prevalence and multiplicity of
33 hepatocellular neoplasms in the two groups so that data were summed and reported together.
34 The number of animals reported as examined for tumors were n = 10 animals, with controls
35 reported to be 35 animals split among 3 interim sacrifice times—exact number per sacrifice time
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-148 DRAFT—DO NOT CITE OR QUOTE
-------
1 is unknown. The number of animals reported "with pathology" and assumed to be included in
2 the tumor analyses from Table 1, and the sum of the number of animals "scheduled for sacrifice
3 that survived till 100 weeks" and "interim sacrifices" do not equal each other. For the 1 g/L
4 DCA exposure group, 30 animals were sacrificed at interim periods, 32 animals were sacrificed
5 at 100 weeks, 9 animals were reported to have unscheduled deaths, but of those 71 animals only
6 65 animals were reported to have pathology for the group. Therefore, some portion of animals
7 with unscheduled deaths must have been included in the tumor analyses. The exact number of
8 animals that may have died prematurely but included in analyses of pathology for the 100 week
9 group is unknown. In Figure 3 of the study, the authors reported prevalence and multiplicity of
10 hepatocellular carcinomas following 79 to 100 weeks of DCA exposure in their drinking water.
11 The number of animals in each dose group used in the tumor analysis for 100 weeks was not
12 given by the authors. Given that the authors included animals that survived past the 78 interim
13 sacrifice period but died unscheduled deaths in their 100 week results, the number must have
14 been greater than those reported as present at final sacrifice. A comparison of the data for the
15 100-week data presented in Table 3a and Figure 3 shows that the data reported for 100 weeks is
16 actually for animals that survived from 79 to 100 weeks. The authors report a dose-response that
17 is statistically significant from 0.5 to 3.5 g/L DCA for hepatocellular carcinoma incidence and a
18 dose-response in hepatocellular carcinoma multiplicity that is significantly increased over
19 controls from 0.05 to 0.5 g/L DCA that survived 79 to 100 weeks of exposure (i.e., 0, 8-, 84-,
20 168-, 315-, and 429 mg/kg/d dose groups with prevalences of 26, 33, 48, 71, 95, and 100%,
21 respectively, and multiplicities of 0.28, 0.58, 0.68, 1.29, 2.47, and 2.90, respectively).
22 Hepatocellular adenoma incidence or multiplicity was not reported for the 0.05 g/L DCA
23 exposure group.
24 In Table 3 of the report, the time course of hepatocellular carcinomas and adenoma
25 development are given and summarized in Table E-2, below.
26 The authors reported hepatocellular carcinomas and number of lesions/animal in mice
27 that survived 79-100 weeks of exposure (they combined exposure groups to be animals after the
28 Week 78 sacrifice time that did and did not make it to 100 weeks). This is the same data
29 reported above for the 100 week exposure with the inclusion of the 0.05 g/L DCA data. The
30 difference between number of animals at interim and final sacrifices and those "with pathology"
31 and used in the tumor analysis but most likely coming from unscheduled deaths is reported in
32 Table E-3 as "extra" and varied across treatment groups.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-149 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table E-2. Prevalence and Multiplicity data from DeAngelo et al. (1999)
Prevalence
52 weeks control = 0% carcinomas, 0% adenoma
0.5 g/L DC A = 0/10 carcinoma, 1/10 adenomas
1.0 g/L DC A = 0/10 carcinomas, 1/10 adenomas
2.0 g/L DC A = 2/10 carcinomas, 0/10 adenomas
3.5 g/L DC A =5/10 carcinomas, 5/10 adenomas
78 weeks control = 10% carcinomas, 10% adenomas
0.5 g/L DC A = 0/10carcinoma, 1/10 adenomas
1.0 g/L DC A = 2/10 carcinomas, 2/10 adenomas
2.0 g/L DC A =5/10 carcinomas, 5/10 adenomas
3.5 g/L DC A = 7/10 carcinomas, 5/10 adenomas
100 weeks control = 26% carcinoma, 10% adenoma
0.5 g/L DC A = 48% carcinoma, 20% adenomas
1.0 g/L DC A = 71% carcinomas, 51.4% adenomas
2.0 g/L DCA = 95% carcinomas, 42.9% adenomas
3.5 g/L DCA = 100% carcinomas, 45% adenomas
Multiplicity
(lesions/animal m ± SEM)
Carcinomas
0
0
0
0.20 ±0.13
0.70 ±0.25
0.10±0.10
0
0.20 ±0.13
1.0 ±0.47
1.20 ±0.37
0.28 ±0.07
0.68 ±0.17
1.29 ±0.17
2.47 ± 0.29
2.90 ±0.40
Adenomas
0
0.10 ±0.09
0.10 ±0.09
0
0.80 ±0.31
0.10 ±0.09
0.10 ±0.09
0.20 ±0.13
1.00 ±-0.42
1.00 ±0.42
0.12 ±0.05
0.32±0.14
0.80±0.17
0.57±0.16
0.64 ±0.23
3
4
5
Table E-3. Difference in pathology by inclusion of unscheduled deaths
from DeAngelo et al. (1999).
Dose = Prevalence of HC
Control = 26%
0.05 g/L = 33%
0.5 g/L = 48%
1 g/L = 71%
2 g/L = 95%
3. 5 g/L =100%
#HC/animal
0.28
0.58
0.68
1.29
2.47
2.9
n = at 100 wk
50
33
24
32
14
8
Extra added in
0
0
1
3
7
3
9
10
11
12
These data show a dose-related increase in tumor formation and decrease in time-to-
tumor associated with DCA exposure at the lowest levels examined. These findings are limited
by the small number of animals examined at 100 weeks but especially those examined at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-150 DRAFT—DO NOT CITE OR QUOTE
-------
1 "interim sacrifice" periods (n = 10). The data illustrate the importance of examining multiple
2 exposure levels at lower concentrations at longer durations of exposure and with an adequate
3 number of animals to determine the nature of a carcinogenic response.
4 Preneoplastic and non-neoplastic hepatic changes were reported to have been described
5 previously and summarized as large preneoplastic foci observed at 52 weeks with multiplicities
6 of 0.1, 0.1, 0.2 and 0.16 for 0.5, 1, 2, and 3.5 g/L DCA exposure respectively. At 100 weeks all
7 values were reported to be significant (0.03, 0.06, 0.14, 0.27 for 0.5, 1, 2, and 3.5 g/L DCA
8 exposure respectively). Control values were not reported by the authors. The authors reported
9 that the prevalence and severity of hepatocellular cytomegaly and of cytoplasmic vacuolization
10 with glycogen deposition to be dose-related and considered significant in all dose groups
11 examined when compared to control liver. However, no quantitative data were shown. The
12 authors reported a severity index of 0 = none, 1 = <25%, 2 = 50-75% and 4 = 75% of liver
13 section for hepatocellular necrosis and report at 26 weeks scores (n = 10 animals) of 0.10 ± 0.10,
14 0.20 ±0.13, 1.20 ±0.38, 1.20 ± 0.39 and 1.10 ± 0.28 for control, 0.5, 1, 2, and 3.5 g/L DCA
15 treatment groups, respectively. Thus, there appeared to be a treatment but not dose-related
16 increase in hepatocellular necrosis that is does not involve most of the liver from 1 to 3.5 g/L
17 DCA at this time point. At 52 weeks of exposure the score for hepatocellular necrosis was
18 reported to be 0, 0, 0.20 ± 0.13, 0.40 ± 0.22 and 1.10 ± 0.43 for control, 0.5, 1, 2, and 3.5 g/L
19 DCA treatment groups, respectively. At 78 weeks of exposure the score for hepatocellular
20 necrosis was reported to be 0, 0, 0, 0.30 ± 0.21 and 0.20 ± 0.13 for control, 0.5, 1, 2, and 3.5 g/L
21 DCA treatment groups, respectively. Finally, the final sacrifice time when more animals were
22 examined the extent of hepatocellular necrosis was reported to be 0.20 ± 0.16, 0.20 ± 0.08,
23 0.42 ±0.15, 0.38 ±0.20 and 1.38 ± 0.42 for control, 0.5, 1,2, and 3.5 g/L DCA treatment
24 groups, respectively. Thus, there was not reported increase in hepatocellular necrosis at any
25 exposure period for 0.5 g/L DCA treatment and the mild hepatocellular necrosis seen at the three
26 highest exposure concentrations at 26 weeks had diminished with further treatment except for the
27 highest dose at up tolOO weeks of treatment. Clearly the pattern of hepatocellular necrosis did
28 not correlate with the dose-related increases in hepatocellular carcinomas reported by the authors
29 and was not increased over control at the 0.5 g/L DCA level where there was a DCA-related
30 tumor increase.
31 The authors cite previously published data and state that CN-insensitive palmitoyl CoA
32 oxidase activity (a marker of peroxisome proliferation) data for the 26 week time point plotted
33 against 100 weeks hepatocellular carcinoma prevalence of animals bearing tumors was
34 significantly enhanced at concentrations of DCA that failed to induce "hepatic PCO" activity.
35 The authors report that neither 0.05 nor 0.5 g/L DCA had any marked effect on PCO activity and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-151 DRAFT—DO NOT CITE OR QUOTE
-------
1 that it was "only significantly increased after 26 weeks of exposure to 3.5 g/L DC A and returned
2 to control level at 52 weeks (data not shown)." In regards to hepatocyte labeling index after
3 treatment for 5 days with tritiated thymidine, the authors report that animals examined in the
4 dose-response segment of the experiment at 26 and 52 weeks were examined but no details of the
5 analysis were reported. The authors comment on the results from this study and a previous one
6 that included earlier time points of study and stated that there were "no significant alterations in
7 the labeling indexes for hepatocytes outside of proliferative lesions at any of the DC A
8 concentrations when compared to the control values with the exception of 0.05 g/L DCA at
9 4 weeks (4.8 ± 0.6 vs. 2.7 ± 0.4 control value; data not shown)."
10 The effects of DCA on body weight, absolute liver weight and percent liver/body weight
11 were given in Table 2 of the paper for 26, 52, 78 and 100 weeks exposure. For 52 and 78 week
12 studies 10 animals per treatment group were examined. Liver weights were not determined for
13 the lowest exposure concentration (0.05 g/L DCA) except for the 100 week exposure period. At
14 26 weeks of exposure there was not a statistically significant change in body weight among the
15 exposure groups (i.e., 35.4 ± 0.7, 37.0 ± 0.8, 36.8 ± 0.8, 37.9 ± 0.6, and 34.6 ± 0.8 g for control,
16 0.5, 1, 2, and 3.5 g/L DCA, respectively). Absolute liver weight was reported to have a dose-
17 related significant increase in comparison to controls at all exposure concentrations examined
18 with liver weight reaching a plateau at the 2 g/L concentration (i.e., 1.86 ± 0.07, 2.27 ±0.10,
19 2.74 ± 0.08, 3.53 ± 0.07, and 3.55 ± 0.1 g for control, 0.5, 1, 2, and 3.5 g/L DCA, respectively).
20 The percent liver/body weight ratio increases due to DCA exposure were reported to have a
21 similar pattern of increase (i.e., 5.25%±0.11%, 6.12%±0.16%, 7.44%±0.12%,
22 9.29% ± 0.08%, and 10.24% ± 0.12% for control, 0.5, 1, 2, and 3.5 g/L DCA, respectively).
23 This represented a 1.17-, 1.41-, 1.77-, and 1.95-fold of control percent liver/body weight at these
24 exposures at 26 weeks.
25 At 52 weeks of exposure there was not a statistically significant change in body weight
26 among the exposure groups except for the 3.5 g/L exposed group in which there was a significant
27 decrease in body weight (i.e., 39.9 ±0.8, 41.7 ±0.8, 41.7 ±0.9, 40.8 ± 1.0, and 35.0 ± 1.1 g for
28 control, 0.5, 1, 2, and 3.5 g/L DCA, respectively). Absolute liver weight was reported to have a
29 dose-related significant increase in comparison to controls at all exposure concentrations
30 examined with liver weight reaching a plateau at the 2 g/L concentration (i.e., 1.87 ± 0.13,
31 2.39 ± 0.04, 2.92 ± 0.12, 3.47 ± 0.13, and 3.25 ± 0.24 g for control, 0.5, 1, 2, and 3.5 g/L DCA,
32 respectively). The percent liver/body weight ratio increases due to DCA exposure were reported
33 to have a similar pattern of increase (i.e., 4.68% ± 0.30%, 5.76% ± 0.12%, 7.00% ± 0.15%,
34 8.50% ± 0.26%, and 9.28% ± 0.64% for control, 0.5, 1, 2, and 3.5 g/L DCA, respectively). For
35 liver weight and percent liver/body weight there was much larger variability between animals
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-152 DRAFT—DO NOT CITE OR QUOTE
-------
1 within the treatment groups compared to controls and other treatment groups. There were no
2 differences reported for patterns of change in body weight, absolute liver weight, and percent
3 liver/body weight between animals examined at 26 weeks and those examined at 52 weeks. At
4 78 weeks of exposure there was not a statistically significant change in body weight among the
5 exposure groups except for the 3.5 g/L exposed group in which there was a significant decrease
6 in body weight (i.e., 46.7 ± 1.2, 43.8 ± 1.5, 43.4 ± 0.9, 42.3 ±0.8, and 40.2 ± 2.2 g for control,
7 0.5, 1, 2, and 3.5 g/L DC A, respectively). Absolute liver weight was reported to have a dose-
8 related increase in comparison to controls at all exposure concentrations examined but none were
9 reported to be statistically significant (i.e., 2.55 ± 0.14, 2.16 ± 0.09, 2.54 ± 0.36, 3.31 ±0.63, and
10 3.93 ± 0.59 g for control, 0.5, 1, 2, and 3.5 g/L DCA, respectively). The percent liver/body
11 weight ratio increases due to DCA exposure were reported to have a similar pattern of increase
12 over control values but only the 3.5 g/L exposure level was reported to be statistically significant
13 (i.e., 5.50% ± 0.35%, 4.93% ± 0.09%, 5.93% ± 0.97%, 7.90% ± 1.55%, and 10.14% ± 1.73% for
14 control, 0.5, 1, 2, and 3.5 g/L DCA, respectively). Finally, for the animals reported to be
15 sacrificed between 90 and 100 weeks there was not a statistically significant change in body
16 weight among the exposure groups except for the 2.0 and 3.5 g/L exposed groups in which there
17 was a significant decrease in body weight (i.e., 43.9 ± 0.8, 43.3 ± 0.9, 42.1 ± 0.9, 43.6 ± 0.7,
18 36.1 ± 1.2, and 36.0 ± 1.3 g for control, 0.05, 0.5, 1, 2, and 3.5 g/L DCA, respectively).
19 Absolute liver weight did not show a dose-response pattern at the two lowest exposure levels but
20 was elevated with the 3 highest doses with the two highest being statistically significant (i.e.,
21 2.59 ± 0.26, 2.74 ± 0.20, 2.51 ± 0.24, 3.29 ± 0.21, 4.75 ± 0.59, and 5.52 ± 0.68 g for control,
22 0.05, 0.5, 1, 2, and 3.5 g/L DCA, respectively). The percent liver/body weight ratio increases
23 due to DCA exposure were reported to have a similar pattern of increase over control values but
24 only the 2.0 and 3.5 g/L exposure levels were reported to be statistically significant (i.e.,
25 6.03% ± 0.73%, 6.52% ± 0.55%, 6.07% ± 0.66%, 7.65% ± 0.55%, 13.30% ± 1.62%, and
26 15.70% ± 2.16% for control, 0.05, 0.5, 1, 2, and 3.5 g/L DCA, respectively).
27 It must be recognized that liver weight increases, especially in older mice, will reflect
28 increased weight due to tumor burden and thus, DCA-induced hepatomegaly will be somewhat
29 obscured at the longer treatment durations. However, by 100 weeks of exposure there did not
30 appear to be an increase in liver weight at the 0.05 and 0.5 g/L exposures while there was an
31 increase in tumor burden reported. Examination of the 0.5 g/L exposure group from 26 to
32 100 weeks shows that slight hepatomegaly, reported as either absolute liver weight increase over
33 control or change in percent liver/body ratio, was present by 26 weeks (i.e., 22% increase in liver
34 weight and 17% increase in percent liver/body weight), decreased with time, and while similar at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-153 DRAFT—DO NOT CITE OR QUOTE
-------
1 52 weeks, was not significantly different from control values at 78 or 100 weeks durations of
2 exposure. However, tumor burden was increased at this low concentration of DC A.
3 The authors present a figure comparing the number of hepatocellular carcinomas per
4 animal at 100 weeks compared with the percent liver/body weight at 26 weeks and show a linear
5 correlation (r = 0.9977). Peroxisome proliferation and DNA synthesis, as measured by tritiated
6 thymidine, were reported to not correlate with tumor induction profiles and were also not
7 correlated with early liver weight changes induced by DCA exposure. Most importantly, in a
8 paradigm that examined tumor formation after up to 100 weeks of exposure, DCA-induced
9 tumor formation was reported to occur at concentrations that did not also cause cytotoxicity and
10 at levels 20 to 40 times lower than those used in "less than lifetime" studies reporting concurrent
11 cytotoxicity.
12
13 E.2.3.2.7. Carter et al, 2003. The focus of this study was to present histopathological
14 analyses that included classification, quantification and statistical analyses of hepatic lesions in
15 male B6C3F1 mice receiving DCA at doses as low as 0.05 g/L for 100 weeks and at 0.5, 1.0, 2.0,
16 and 3.5 g/L for between 26 and 100 weeks. This analysis used tissues from the DeAngelo et al.
17 (1999) (two blocks from each lobe and all lesions found at autopsy). This study used the
18 following diagnostic criteria for hepatocellular changes. Altered hepatic Foci (AHF) were
19 defined as histologically identifiable clones that were groups of cells smaller than a liver lobule
20 that did not compress the adjacent liver. Large foci of cellular alteration (LFCA) were defined as
21 lesions larger than the liver lobule that did not compress the adjacent architecture (previously
22 referred to as hyperplastic nodules by Bull et al., 1990) but had different staining. These are not
23 non-neoplastic proliferative lesions termed "hepatocellular hyperplasia" that occur secondary to
24 hepatic degeneration or necrosis. Adenomas (ADs) showed growth by expansion resulting in
25 displacement of portal triad and had alterations in both liver architecture and staining
26 characteristics. Carcinomas (CAs) were composed of cells with a high nuclear-to-cytoplasmic
27 ration and with nuclear pleomorphism and atypia that showed evidence of invasion into the
28 adjacent tissue. They frequently showed a trabecular pattern characteristic of mouse
29 hepatocellular CAs.
30 The report grouped lesions as eosinophilic, basophilic and/or clear cell, and dysplastic.
31 "Eosinophilic lesions included lesions that were eosinophilic but could also have clear cell,
32 spindle cell or hyaline cells. Basophilic lesions were grouped with clear cell and mixed cell (i.e.,
33 mixed basophilic, eosinophilic, hyaline, and/or clear cell) lesions." The authors reported that
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-154 DRAFT—DO NOT CITE OR QUOTE
-------
1 this grouping was necessary because many lesions had both a basophilic and clear
2 cell component and a few <10 % had an eosinophilic or hyaline
3 component.. .Lesions with foci of cells displaying nuclear pleomorphism,
4 hyperchromasia, prominent nucleoli, irregular nuclear borders and/or altered
5 nuclear to cytoplasmic ratios were considered dysplastic irrespective of their
6 tinctorial characteristics.
7
8 Therefore, Carter et al. (2003) lumped mixed phenotype lesions into the basophilic grouping so
9 that comparisons with the results of Bull et al. (2002) or Pereira (1996), which segregate mixed
10 phenotype from those without mixed phenotype, cannot be done.
11 This report examined type and phenotype of preneoplastic and neoplastic lesions pooled
12 across all time points. Therefore, conclusions regarding what lesions were evolving into other
13 lesions have left out the factor of time. Bannasch (1996) reported that examining the evolution
14 of foci through time is critical for discerning neoplastic progression and described foci evolution
15 from eosinophilic or basophilic lesions to more basophilic lesions. Carter et al. (2003) suggest
16 that size and evolution into a more malignant state are associated with increasing basophilia, a
17 conclusion consistent with those of Bannasch (1996). The analysis presented by Carter et al.
18 (2003) also suggested that there was more involvement of lesions in the portal triad, which may
19 give an indication where the lesions arose. Consistent with the results of DeAngelo et al. (1999),
20 Carter et al. (2003) reported that "DCA (0.05 - 3.5 g/L) increased the number of lesions per
21 animal relative to animals receiving distilled water and shortened the time to development of all
22 classes of hepatic lesions." They also concluded that
23
24 although this analysis could not distinguish between spontaneously arising lesions
25 and additional lesions of the same type induced by DCA, only lesions of the kind
26 that were found spontaneously in control liver were found in increased numbers in
27 animals receiving DCA.. .Development of eosinophilic, basophilic and/or clear
28 cell and dysplastic AHF was significantly related to DCA dose at 100 weeks and
29 overall adjusted for time.
30
31 The authors concluded that the presence of isolated, highly dysplastic hepatocytes in male
32 B6C3F1 mice chronically exposed to DCA suggested another direct neoplastic conversion
33 pathway other than through eosinophilic or basophilic foci.
34 It appears that the lesions being characterized as carcinomas and adenomas in
35 DeAngelo et al. (1999) were not the same as those by Carter et al. (2003) at 100 weeks even
36 though they were from the same tissues (see Table E-4). Carter et al. identified all carcinomas as
37 dysplastic despite tincture of lesion and subdivided adenomas by tincture. If the differing
38 adenoma multiplicities are summed for Carter et al. they do not add up to the same total
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-155 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
multiplicity of adenoma given by DeAngelo et al. It is unclear how many animals were included
in the differing groups in both studies for pathology. The control and high-dose groups differ in
respect to "animals with pathology" between DeAngelo et al. and the "number of animals in
groups" examined for lesions in Carter et al. Neither report gave how many animals with
unscheduled deaths were treated in regards to how the pathology data were included in
presentation of results. Given that DeAngelo et al. represents animals at 100 weeks as also
animals from 79-100 weeks exposure, it is probable that the animals that died after 79 weeks
were included in the group of animals sacrificed at 100 weeks. However, the number of animals
affecting that result (which would be a mix of exposure times) for either DeAngelo et al., or
Carter et al., is unknown from published reports. In general, it appears that Carter et al. (2003)
reported more adenomas/animal for their 100 week animals than DeAngelo et al. (1999) did,
while DeAngelo et al. reported more carcinomas/animal. Carter et al. reported more
adenomas/animal than controls while DeAngelo et al. reported more carcinomas/animal than
controls at 100 weeks of exposure.
Table E-4. Comparison of data from Carter et al. (2003) and DeAngelo et
al. (1999)
Exposure
level of
DCAat
79-100
wk
(g/L)
0
0.05
0.5
1.0
2.0
3.5
Total
adenoma
multiplicity
(Carter)
0.22
0.48
0.44
0.52
0.60
1.48
Total
adenoma
multiplicity
(DeAngelo)
0.12
-
0.32
0.80
0.57
0.64
Total
carcinoma
multiplicity
(Carter)
0.05
<0.025
0.20
0.30
1.55
1.30
Total
carcinoma
multiplicity
(De Angelo)
0.28
0.58
0.68
1.29
2.47
2.90
Sum of
adenomas
and
carcinoma
multiplicity
(Carter)
0.27
-0.50
0.64
0.82
2.15
2.78
Sum of
adenomas
and
carcinoma
multiplicity
(DeAngelo)
0.40
-
1.0
2.09
3.27
3.54
19
20
21
22
23
24
25
26
In order to compare these data with others (e.g., Pereira, 1996) for estimates of
multiplicity by phenotype or tincture it would be necessary to add foci and LFCA together as
foci, and adenomas and carcinomas together as tumors. It would also be necessary to lump
mixed foci together as "basophilic" from other data sets as was done for Carter et al. in
describing "basophilic lesions." If multiplicity of carcinomas and adenomas are summed from
each study to control for differences in identification between adenoma and carcinoma, there are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-156 DRAFT—DO NOT CITE OR QUOTE
-------
1 still differences in the two studies in multiplicity of combined lesions/animal with DeAngelo
2 giving consistently higher estimates. However, both studies show a dose response of tumor
3 multiplicity with DCA and a difference between control values and the 0.05 DCA exposure
4 level. Error is introduced by having to transform the data presented as a graph in Carter et al.
5 (2003). Also no SEM is given for the Carter data.
6 In regard to other histopathological changes, the authors report that
7
8 necrosis was found in 11.3% of animals in the study and the least prevalent toxic
9 or adaptive response. No focal necrosis was found at 0.5 g/L. The incidence of
10 focal necrosis did not differ from controls at 52 or 78 weeks and only was greater
11 than controls at the highest dose of 3.5 g/L at 100 weeks. Overall necrosis was
12 negatively related to the length of exposure and positively related to the DCA
13 dose. Necrosis was an early and transitory response. There was no difference in
14 necrosis 0 and 0.05 g/L or 0.5 g/L. There was an increase in glycogen at 0.5 g/L
15 at the perioportal area. There was no increase in steatosis but a dose-related
16 decrease in steatosis. Dysplastic LFCA were not related to necrosis indicating
17 that these lesions do not represent, regenerative or reparative hyperplasia.
18 Nuclear atypia and glycogen accumulation were associated with dysplastic
19 adenomas. Necrosis was not related to occurrence of dysplastic adenomas.
20 Necrosis was of borderline significance in relation to presence of hepatocellular
21 carcinomas. Necrosis was not associated with dysplastic LFCAs or Adenomas.
22
23 They concluded that "the degree to which hepatocellular necrosis underlies the carcinogenic
24 response is not fully understood but could be significant at higher DCA concentrations (^lg/L)."
25
26 E.2.3.2.8. Stauber and Bull, 1997. This study was designed to examine the differences in
27 phenotype between altered hepatic foci and tumors induced by DCA and TCA. Male B6C3F1
28 mice (7 weeks old at the start of treatment) were treated with 2.0 g/L neutralized DCA or TCA in
29 drinking water for 38 or 50 weeks, respectively. They were then treated with additional
30 exposures (n = 12) of 0, 0.02, 0.1, 0.5, 1.0, 2.0 g/L DCA or TCA for an additional 2 weeks.
31 Three days prior to sacrifice in DCA-treated mice or 5 days for TCA-treated mice, animals had
32 miniosmotic pumps implanted and administered BrdU. Immunohistochemical staining of
33 hepatocytes from randomly selected fields (minimum of 2,000 nuclei counter per animal) from
34 5 animals per group were reported for 14- and 28-day treatments. It was unclear how many
35 animals were examined for 280- and 350-day treatments from the reports. The percentage of
36 labeled cells in control livers was reported to vary between 0.1 and 0.4% (i.e., 4-fold). There
37 was a reported ~3.5-fold of control level for TCA labeling at 14 day time period and a ~5.5-fold
38 for DCA. At 28 days there was ~2.5-fold of control for TCA but a ~2.3-fold decrease of control
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-157 DRAFT—DO NOT CITE OR QUOTE
-------
1 for DCA. At 280 days there was no data reported for TCA but for DCA there was a ~2-fold
2 decrease in labeling over control. At 350 days there was no data for DCA but a reported -2.3-
3 fold decrease in labeling of control with TCA. The authors reported that the increases at Day 14
4 for TCA and DCA exposure and the decrease at Day 28 for DCA exposure were statistically
5 significant although a small number of animals were examined. Thus, although there may be
6 some uncertainty in the exact magnitude of change, there was at most ~5-fold of control labeling
7 for DCA within after 14 days of exposure that was followed by a decrease in DNA synthesis by
8 Day 28 of treatment. These data show that hepatocytes undergoing DNA synthesis represented a
9 small population of hepatocytes with the highest level with either treatment less than 1% of
10 hepatocytes. Rates of cell division were reported to be less than control for both DCA and TCA
11 by 40 and 52 weeks of treatment.
12 In this study the authors reported that there was no necrosis with the 2.0 g/L DCA dose
13 for 52 weeks and conclude that necrosis is a recurring but inconsistent result with chronic DCA
14 treatment. Histological examination of the livers involved in the present study found little or no
15 evidence of such damage or overt cytotoxicity. It was assumed that this effect has little bearing
16 on data on replication rates. Foci and tumors were combined in reported results and therefore,
17 cannot be compared the results Bull et al. (2002) or to DeAngelo et al. (1999). Prevalence rates
18 were not reported. Data were reported in terms of "lesions" with DCA-induced "lesions"
19 containing a number of smaller lesions that were heterogeneous and more eosinophilic with
20 larger "lesions" tending to less numerous and more basophilic. For TCA results using this
21 paradigm, the "lesions" were reported to be less numerous, more basophilic, and larger than
22 those induced by DCA. The DCA-induced larger "lesions" were reported to be more "uniformly
23 reactive to c-Jun and c-Fos but many nuclei within the lesions displaying little reactivity to c-
24 Jun." The authors stated that while most DCA-induced "lesions" were homogeneously
25 immunoreactive to c-Jen and C-Fos (28/41 lesions), the rest were stained heterogeneously. For
26 TCA-induced lesions, the authors reported not difference in staining between "lesions" and
27 normal hepatocytes in TCA-treated animals. Again, of note is that not only were "lesions"
28 comprised of foci and tumors at different stages of progression reported in these results, but that
29 also DCA and TCA results were reported for different durations of exposure.
30
31 E.2.3.2.9. Pereira, 1996. The focus of this study was to report the dose-response relationship
32 for the carcinogenic activity of DCA and TCA in female B6C3F1 mice and the characteristics of
33 the lesions. Female B6C3F1 mice (7-8 weeks of age) were given drinking water with either
34 DCA or TCA at 2.0, 6.67, or 20 mmol/L and neutralized with sodium hydroxide to a pH or
35 6.5-7.5. The control received 20 mmol/L sodium chloride. Conversion of mmol/L to g/L was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-158 DRAFT—DO NOT CITE OR QUOTE
-------
1 as follows: 20.0 mmol/L DCA = 2.58 g/L, 6.67 mmol/L DCA = 0.86 g/L, 2.0
2 mmol/L = 0.26 g/L, 20.0 mmol/L TCA = 3.27 g/L, 6.67 mmol/L TCA = 1.10 g/L, 2.0 mmol/L
3 TCA = 0.33 g/L. The concentrations were reported to be chosen so that the high concentration
4 was comparable to those previously used by us to demonstrate carcinogenic activity. The mice
5 were exposed till sacrifice at 360 (51 weeks), or 576 days (82 weeks) of exposure. Whole liver
6 was reported to be cut into ~3 mm blocks and along with representative section of the visible
7 lesions fixed and embedded in paraffin and stained with H&E for histopathological evaluation of
8 foci of altered hepatocytes, hepatocellular adenomas, and hepatocellular carcinomas. The slides
9 were reported to be evaluated blind. Foci of altered hepatocytes in this study were defined as
10 containing 6 or more cells and hepatocellular adenomas were distinguished from foci by the
11 occurrence of compression at greater than 80% of the border of the lesion.
12 Body weights were reported to be decreased only the highest dose of DCA from
13 40 weeks of treatment onward. For TCA there were only 2 examination periods (Weeks 51 and
14 82) that had significantly different body weights from control and only at the highest dose.
15 Liver/body weight percentage was reported in comparison to concentration graphically and
16 shows a dose-response for DCA with steeper slope than that of TCA at 360 and 576 days of
17 exposure. The authors report that all three concentrations of DCA resulted in increased
18 vacuolation of hepatocytes.(probably due to glycogen removal from tissue processing). Using a
19 score of 1-3, (with 0 indicating the absence of vacuolization, +1 indicating vacuolated
20 hepatocytes in the periportal zone, + 2 indicating distribution of vacuolated hepatocytes in the
21 midzone, and +3 indicating maximum vacuolization of hepatocytes throughout the liver), the
22 authors also reported "the extent of vacuolization of the hepatocytes in the mice administered 0,
23 2.0, 6.67 or 20.0 mmol/1 DCA was scored as 0.0, 0.80 ± 0.08, 2.32 ± 0.11, or 2.95 ± 0.05,
24 respectively."
25 Cell proliferation was reported to be determined in treatment groups containing 10 mice
26 each and exposed to either DCA or TCA for 5, 12, or 33 days with animals implanted with
27 miniosmotic pumps 5 days prior to sacrifice and administered BrdU. Tissues were
28 immunohistochemically stained for BrdU incorporation. At least 2,000 hepatocytes/mouse were
29 reported to be evaluated for BrdU-labeled and unlabeled nuclei and the BrDU-labeling index was
30 calculated as the percentage of hepatocytes with labeled nuclei. Pereira (1996) reported a dose-
31 related increase in BrDU labeling in 2,000 hepatocytes that was statistically significant at 6.67
32 and 20.mmol/L DCA at 5 days of treatment but that labeling at all exposure concentrations
33 decreased to control levels by Day 12 and 33 of treatment. The largest increase in BrdU labeling
34 was reported to be a 2-fold of controls at the highest concentration of DCA after 5 days of
35 exposure. For TCA all doses (2.0, 6.67 and 20 mmol/L) gave a similar and statistically
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-159 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
significant increase in BrDU labeling by 5 days of treatment (~3-fold of controls) but by days 12
and 33 there were no increases above control values at any exposure level. Given the low level
of hepatocyte DNA synthesis in quiescent control liver, these results indicate a small number of
hepatocytes underwent increased DNA synthesis after DCA or TCA treatment and that by
12 days of treatment these levels were similar to control levels in female B6C3F1 mice.
Incidence of foci and tumors in mice administered DCA or TCA (prevalence or number
of animals with tumors of those examined at sacrifice) in this report are given below in
Tables E-5 and E-6.
Table E-5. Prevalence of foci and tumors in mice administered NaCl, DCA,
or TCA from Pereira (1996)
Treatment
N
Foci
Number
%
Adenomas
Number
%
Carcinomas
Number
%
82wks
20.0 mmol NaCl
20.0 mmol DCA
6.67 mmol DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
90
19
28
50
18
27
53
10
17
11
7
11
9
10
11.1
89.5*
39.3*
14.0
61.1*
33.3*
18.9
2
16
7
O
7
3
4
2.2
84.2*
25.0*
6.0
38.9*
11.1
7.6
2
5
1
0
5
5
0
2.2
26.3*
3.6
0
27.8%*
18.5*
0
51 wks
20.0 mmol NaCl
20.0 mmol DCA
6.67 mmol DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
40
20
20
40
20
19
40
0
8
1
0
0
0
3
0
40.0*
5
0
0
0
7.5
1
7
3
0
2
3
3
2.5
35*
15
0
15.8
7.5
2.5
0
1
0
0
5
0
0
0
5
0
0
25*
0
0
13
14
15
16
*p < 0.05.
NaCl = sodium chloride control.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-160 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table E-6. Multiplicity of foci and tumors in mice administered NaCl,
DCA, or TCA from Pereira (1996)
Treatment
N
Foci/mouse
Adenomas/mouse
Carcinomas/mouse
82wks
20.0 mmol NACL
20.0 mmol DCA
6.67 mmol DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
90
19
28
50
18
27
53
0.11 ±0.03
7.95±2.00a
0.39±0.11b
0.14 ±0.05
1.33±0.31a
0.41±0.13b
0.26 ±0.08
0.02 ±0.02
5.58± 1.14a
0.32±0.13b
0.06 ±0.03
0.61±0.22b
0.11 ±0.06
0.08 ±0.04
0.02 ±0.02
0.37±0.17b
0.04 ±0.04
0
0.39±0.16b
0.22±0.10b
0
51 wks
20.0 mmol NACL
20.0 mmol DCA
6.67 mmol DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
40
20
20
40
20
19
40
0
0.60±0.22a
0.05 ±0.05
0
0
0
0.08 ±0.04
0.03 ± 0.03
0.45±0.17a
0.20 ±0.12
0
0.15±0.11
0.21 ±0.12
0.08 ±0.04
0
0.10±0.10
0
0
0.50±0.18b
0
0
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
bp < 0.05.
NaCl = sodium chloride control.
These data show the decreased power of using fewer than 50 mice, especially at shorter
durations of exposure. By 82 weeks of exposure increased adenoma and carcinomas induced by
TCA or DCA treatment are readily apparent.
The foci of altered hepatocytes and the tumors obtained from this study were reported to
be basophilic, eosinophilic, or mixed containing both characteristics and are shown in Tables E-7
and E-8. DCA was reported to induce a predominance of eosinophilic foci and tumors, with over
80% of the foci and 90% of the tumors in the 6.67 and 20.0 mmol/L concentration groups being
eosinophilic. Only approximately half of the lesions were characterized as eosinophilic with the
rest being basophilic in the group administered 2.0 mmol/L DCA. The eosinophilic foci and
tumors were reported to consistently stained immunohistochemically for the presence of GST-Ti,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-161 DRAFT—DO NOT CITE OR QUOTE
-------
1 while basophilic lesions did not stain for GST-Ti, except for a few scattered cells or small areas
2 comprising less than 10% of foci. The foci of altered hepatocytes in the TCA treatment groups
3 were approximately equally distributed between basophilic and eosinophilic in tincture.
4 However, the tumors were predominantly basophilic lacking GST-pi (21 of 28 or 75%) including
5 all 11 hepatocellular carcinomas. The limited numbers of lesions, i.e., 14, in the sodium chloride
6 (vehicle control) group were characterized as 64.3, 28.6, and 7.1% basophilic, eosinophilic, and
7 mixed, respectively.
9
10
11
12
13
14
15
16
17
18
19
20
Table E-7. Phenotype of foci reported in mice exposed to NaCl, DCA, or
TCA by Pereira (1996)
Treatment
at 51 and 82 wk
20.0 mmol NaCl
20.0 mmol DCA
6.67 DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
N
10
150
11
7
22
11
13
% Foci
Basophilic
70
3.3
18.2
42.8
36.4
45.5
38.5
Eosinophilic
30
96.7
81.8
57.2
54.6
54.5
61.5
Mixed
0
0
0
0
9.1
0
0
NaCl = sodium chloride control.
Table E-8. Phenotype of tumors reported in mice exposed NaCl, DCA, or
TCA by Pereira (1996)
Treatment
at 51 and 82 wk
20.0 mmol NaCl
20.0 mmol DCA
6.67 DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
N
4
105
10
3
18
6
4
Tumors
Basophilic
50
2.9
10
0
61.1
100
100
Eosinophilic
25
96.1
90
100
22.2
0
0
Mixed
25.5
1
0
0
16.7
0
0
NaCl = sodium chloride control.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-162 DRAFT—DO NOT CITE OR QUOTE
-------
1 These data for female B6C3F1 mice shows that DC A and TCA treatment induced a
2 mixture of basophilic or eosinophilic foci. The pooling of the data between time and adenoma
3 versus carcinoma decreases the ability to ascertain the phenotype of tumor due to treatment or
4 the progression of phenotype with time as well as the small number of tumor examined at lower
5 exposure concentrations. Foci that occurred at 51 and 82 weeks were presented as one result.
6 Adenomas and carcinoma data were pooled as one endpoint (n = number of total foci or tumors
7 examined). Therefore, evolution of phenotype between less to more malignant stages of tumor
8 were lost.
9
10 E.2.3.2.10. Pereira andPhelps, 1996. The focus of this study was to determine tumor response
11 and phenotype in methyl nitrosourea (MNU)-treated mice after DCA or TCA exposure. The
12 concentrations of DCA or TCA were the same as Pereira (1996). For Pereira (1996) the animals
13 were reported to be 7-8 weeks of age when started on treatment and sacrificed after 360 or 576
14 days of exposure (51 or 82 weeks). For this study and Tao et al. (2004), animals were reported o
15 be 6 weeks of age when exposed to DCA or TCA via drinking water and to be 31 or 52 weeks of
16 age at sacrifice. Thus, exposure time would be-24 or 45 weeks. A control group of non-MNU
17 treated animals was presented for female B6C3F1 mice treated for 31 or 52 weeks and are
18 discussed in Table E-9, below. Although this paradigm appears to be the same paradigm as
19 those reported in Pereira (1996), fewer animals were studied. The number of animals in each
20 group varied between 8 controls and 14 animals in the 2.0 mmol/L treatment groups. In mice
21 that were not treated with MNU, but were treated with either DCA or TCA at 31 weeks, there
22 were no reported statistically significant treatment-related effect upon the yield of foci or altered
23 hepatocytes and liver tumors but the number of animals examined was small and therefore, of
24 limited power to detect a response. The results below indicate a DCA-related increase in foci
25 and percentage of mice with foci.
26 See Section E.4.2.3 for further discussion of the results of coexposures to MNU and DCA
27 or TCA from this study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-163 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table E-9. Multiplicity and incidence data (31 week treatment) from
Pereira and Phelps (1996)
Treatment
20.0 mmol NaCl
20.0 mmol DCA
6.67 DCA
2.0 mmol DCA
20.0 mmol TCA
6.67 mmol TCA
2.0 mmol TCA
No
15
10
10
15
10
10
15
Foci/mouse
0.13 ±0.13
0.40 ±0.16
0.10±0.10
0
0
0
0
incidence %
6.7
40
10
0
0
0
0
Adenomas/mouse
0.13±0.13
0
0
0
0
0
0
incidence %
not reported
0
0
0
0
0
0
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
NaCl = sodium chloride control.
E.2.3.2.11. Ferreira-Gonzalez et al, 1995. The focus of this study was the investigation of
differences in H-ras mutation spectra in hepatocellular carcinomas induced by TCA or DCA in
male B6C3F1 mice. 28-day old mice were exposed for 104 weeks to 0. 1.0 g or 3.5 g/L DCA or
4.5 g/L TCA that was pH adjusted. Tumors observed from this treatment were diagnosed as
either hepatocellular adenomas or carcinomas. DNA was extracted from either spontaneous,
DCA- or TCA-induced hepatocellular carcinomas. Samples for analysis were chosen randomly
in the treatment groups of which 19% of untreated mice had spontaneous liver hepatocellular
carcinomas (0.26 carcinomas/animal), DCA treatment induced 100% prevalence at 3.5 g/L (5.06
carcinomas/animal) and 70.6% carcinomas at 1.0 g/L (1.29 carcinomas/animal). TCA treatment
was reported to induce 73.3% prevalence at 4.5 g/L (1.5 carcinomas/animal). The number of
samples analyzed was 32 for spontaneous carcinomas, 33 for mice treated with 3.5 g/L DCA, 13
from mice treated with 1.0 g/DCA, and 11 from mice treated with 4.5 g/L TCA. This study has
the advantage of comparison of tumor phenotype at the same stage of progression (hepatocellular
carcinoma), for allowance of the full expression of a tumor response (i.e., 104 weeks), and an
adequate number of spontaneous control lesions for comparison with DCA or TCA treatments.
However, tumor phenotype at an endstage of tumor progression reflects of tumor progression
and not earlier stages of the disease process.
There were no ras mutations detected except at H-61 in DNA from spontaneously arising
tumors of control mice. Only 4/57 samples from carcinogen-treated mice were reported to
demonstrate mutation other than in the second exon of H-ras. In spontaneous liver carcinomas,
58% were reported to show mutations in H-61 as compared with 50% of tumor from 3.5 g/L
DCA-treated mice and 45% of tumors from 4.5.g/L TCA-treated mice. Thus, there was a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-164 DRAFT—DO NOT CITE OR QUOTE
-------
1 heterogeneous response for this phenotypic marker for the spontaneous, DCA-, and TCA-
2 treatment induced hepatocellular carcinomas.
3 All samples positive for mutation in the exon 2 of H-ras were sequenced for the
4 identification of the base change responsible for the mutation. The authors noted that H-ras
5 mutations occurring in spontaneously developing hepatocellular carcinomas from B6C3F1 male
6 mice are largely confined to codon 61 and involve a change from CAA to either AAA or CGA or
7 CTA in a ratio of 4:2:1. They noted that in this study, all of the H-ras second codon mutations
8 involved a single base substitution in H-61 changing the wild-type sequence from CAA to AAA
9 (80%), CGA (20%) or CTA for the 18 hepatocellular carcinomas examined. In the 16
10 hepatocellular carcinomas from 3.5 g/L DC A treatment with mutations, 21% were AAA
11 transversions, 50% were CGA transversions, and 29% were CTA transversions. For the
12 6 hepatocellular carcinomas from 1.0 g/L DC A with mutations, 16% were an AAA transversion,
13 50% were a CGA transversion, and 34% were a CTA transversion. For the 5 hepatocellular
14 carcinomas from 4.5 g/L TCA with mutations, 80% were AAA transversions, 20% CGA
15 tranversions, and 0% were CTA transversions. The authors note that the differences in
16 frequency between DC A and TCA base substitutions did not achieve statistical significance due
17 to the relatively small number of tumors from TCA-treated mice. They note that the finding of
18 essentially equal incidence of H-ras mutations in spontaneous tumors and in tumors of
19 carcinogen-treated mice did not help in determining whether DC A and TCA acted as
20 "genotoxic" or "nongenotoxic" compounds.
21
22 E.2.3.2.12. Pereira et al, 2004. Pereira et al. (2004) exposed 7-8 week old female B6C3F1
23 mice treated with "AIN-76A diet" to neutralized 0, or 3.2 g/L DC A in the drinking water and 4.0
24 or 8.0 g/kg L-methionine added to their diet. The final concentration of methionine in the diet
25 was estimated to be 11.3 and 15.3 g/kg. Mice were sacrifice 8 and 44 weeks after exposure to
26 DCA with body and liver weights evaluated for foci, adenomas, and hepatocellular carcinomas.
27 No histological descriptions were given by the authors other than tinctoral phenotype of foci and
28 adenomas for a subset of the data. The number of mice examined was 36 for the DCA + 8.0 g/kg
29 methionine or 4.0 g/kg methionine group sacrificed at 44 weeks. However, for the DCA-only
30 treatment group the number of animals examined was 32 at 44 weeks and for those groups that
31 did not receive DCA but either methionine at 8.0 or 4.0 g/kg, there were only 16 animals
32 examined. All groups examined at 8 weeks had 8 animals per group. Liver glycogen was
33 reported to be isolated from 30-50 mg of whole liver. Peroxisomal acyl-CoA oxidase activity
34 was reported to be determined using lauroyl-CoA as the substrate and was considered a marker
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-165 DRAFT—DO NOT CITE OR QUOTE
-------
1 of peroxisomal proliferation. Whole liver DNA methylation status was analyzed using a 5-MeC
2 antibody.
3 Methionine (8.0 g/kg) and DCA coexposure was reported to result in the death of 3 mice
4 while treatment with methionine (4.0 g/kg) and DCA or methionine (8.0 g/kg) alone was
5 reported to kill one mouse in each group. The authors reported that "There was an increased in
6 body weight during weeks 12 to 36 in the mice that received 8.0 g/kg methionine without DCA.
7 There was no other treatment-related alteration in body weight." However, the authors do not
8 present the data and initial or final body weights were not presented for the differing treatment
9 groups. DCA treatment was reported to increase percent liver/body weight ratios at 8 and
10 44 weeks to about the same extent (i.e., -2.4-fold of control at 8 weeks and 2.2-fold of control at
11 44 weeks). Methionine coexposure was reported to not affect that increase (-2.4-, 2.2-, and
12 2.1-fold of control after DCA treatment alone, DCA/4 g/kg methionine, and DCA/8 mg/kg
13 methionine treatment for 8 weeks, respectively). There was a slight increase in percent
14 liver/body weight ratio associated with 8.0 g/kg methionine treatment alone in comparison to
15 controls (-7%) at 8 weeks with no difference between the two groups at 44 weeks.
16 After 8 weeks of only DCA exposure, the amount of glycogen in the liver was reported to
17 be -2.09-fold of the value for untreated mice (115 vs. 52.5 mg/g glycogen in treated vs. control,
18 respectively at 8 weeks). Both 4 g/kg and 8 g/kg methionine coexposure reduced the amount of
19 DCA-induced glycogen increase in the liver (~1.64-fold of control for DCA/4.0 g/kg methionine
20 and ~1.54-fold of control for DCA/8.0 mg/kg methionine). Thus, for treatment with DCA alone
21 or with the two coexposure levels of methionine, the magnitude of the increase in liver weight
22 was greater than that of the increase in liver glycogen (i.e., 2.42- vs. 2.09-fold of control percent
23 liver/body weight vs. glycogen content for DCA alone, 2.20- vs. 1.64-fold of control percent
24 liver/body weight vs. glycogen content for DC A/4.0 g/kg methionine, 2.10- vs. 1.54-fold of
25 control percent liver/body weight vs. glycogen content for DCA/8.0 g/kg methionine). Thus, the
26 magnitudes of treatment-related increases were higher for percent liver/body weight than for
27 glycogen content in these groups. In regard to percentage of liver mass that glycogen
28 represented, the control value for this study is similar to that presented by Kato-Weinstein et al.
29 (2001) in male mice (-60 mg glycogen per gram liver) and represents -6% of liver mass.
30 Therefore, a doubling of the amount of glycogen is much less than the 2-fold increases in liver
31 weight observed for DCA exposure in this paradigm. These data suggest that DCA-related
32 increases in liver weight gain are not only the result of increased glycogen accumulation, and
33 that methionine coexposure is affecting glycogen accumulation to a much greater extent than the
34 other underlying processes that are contributing to DCA-induced hepatomegaly after 8 weeks of
35 exposure. The authors reported that 8-weeks of DCA exposure alone did not result in a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-166 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant increase in cell proliferation as measured by PCN index (neither data nor methods
2 were shown). This is consistent with other data showing that DC A effects on DNA synthesis
3 were transient and had subsided by 8 weeks of exposure.
4 The levels of lauroyl-CoA oxidase activity were reported to be increased (~1.33-fold of
5 control) by DCA treatment alone at 8 weeks and to be slightly reduced by 8 g/kg methionine
6 treatment alone (~0.83-fold of control). Methionine coexposure was reported to have little effect
7 on DCA-induced increases in lauroyl-CoA oxidase activity. The levels of DNA methylation
8 were reported to be increased by 8.0 g/kg methionine only treatment at 8 weeks ~1.32-fold of
9 control, and reduced by DCA only treatment to ~0.44-fold of control. DCA and 4.0 g/kg
10 methionine coexposure gave similar results as controls (within 2%). Coexposures of DCA and
11 8.0 g/kg methionine treatments were reported to increase DNA methylation 1.22-fold of controls
12 after 8 weeks of coexposure.
13 In the 44-week study, the authors report that foci and hepatocellular adenomas were
14 found. However, the authors do not report the incidences of these lesions in their study groups
15 (how many of the treated animals developed lesions). As noted above, the numbers of animals in
16 these groups varied widely between treatments (e.g., n = 36 for DCA and coexposure to 8.0 g/kg
17 methionine but only n = 16 for 8 g/kg methionine treatment alone). Although reporting
18 unscheduled deaths in the 8.0 g/kg methionine and DCA coexposure groups, the authors did not
19 indicate whether these mortalities occurred in the 44-week or 8-week study groups.
20 Multiplicities of foci and adenoma data were presented. DCA was reported to induce
21 2.42 ± 0.38 foci/mouse and 1.28 ± 0.31 adenomas/mouse (m ± SE) after 44 weeks of treatment.
22 The DCA-induced foci and adenomas were reported to stain as eosinophilic with "relatively
23 large hepatocytes and nuclei." The authors did not present data on the percent of foci and
24 adenomas that were eosinophilic using this paradigm. The addition of 4.0 or 8.0 g/kg methionine
25 to the AIN-76A diet was reported to reduce the number of DCA-induced adenomas/mouse to
26 0.167 ± 0.093 and 0.028 ± 0.028, respectively. However, the addition of 4.0 g/kg methionine to
27 the DCA treatment was reported to increase the number of foci/mouse (3.4 ± 0.46 foci/mouse).
28 The addition of 8.0 g/kg methionine to the DCA treatment was reported to yield
29 0.94 ± 0.24 foci/mouse. There were no foci or tumors in the 16 mice that received either the
30 control diet or the 8.0 g/kg methionine treatment without DCA. The authors did not report
31 whether methionine treatment had an effect on the tincture of the foci or adenomas induced by
32 DCA.
33 Therefore, a very high level of methionine supplementation to an AIN-760A diet, was
34 shown to affect the number of foci and adenomas, i.e., decrease them, after 44 weeks of
35 coexposure to very high exposure concentration of DCA. However, a lower level of methionine
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-167 DRAFT—DO NOT CITE OR QUOTE
-------
1 coexposure increased the incidence of foci at the same concentration of DCA. Methionine
2 treatment alone at the 8 g/kg level was reported to increase liver weight, decrease lauroyl-CoA
3 activity and to increase DNA methylation. No histopathology was given by the authors to
4 describe the effects of methionine alone. Coexposure of methionine with 3.2 g/L DCA was
5 reported to decrease by -25% DCA-induced glycogen accumulation and increase mortality, but
6 not to have much of an effect on peroxisome enzyme activity (which was not elevated by more
7 than 33% over control for DCA exposure alone). The authors suggested that their data indicate
8 that methionine treatment slowed the progression of foci to tumors. Whether, these results
9 would be similar for lower concentrations of DCA and lower concentrations of methionine that
10 were administered to mice for longer durations of exposure, cannot be ascertained from these
11 data. It is possible that in a longer-term study, the number of tumors would be similar. Whether,
12 methionine treatment coexposure had an effect on the phenotype of foci and tumors was not
13 presented by the authors in this study. Such data would have been valuable to discern if
14 methionine coexposure at the 4.0 mg/kg level that resulted in an increase in DCA-induce foci,
15 resulted in foci of a differing phenotype or a more heterogeneous composition than DCA
16 treatment alone.
17
18 E.2.3.2.13. DeAngelo et al, 2008. In this study, neutralized TCA was administered in drinking
19 water to male B6C3 Fl mice (28-30 days old) in three studies. In the first study control animals
20 received 2 g/L sodium chloride while those in the second study were given 1.5 g/L neutralized
21 acetic acid (HAC) to account for any taste aversion to TCA dosing solutions. In a third study
22 deionized water served as the control. No differences in water uptake were reported. Mean
23 initial weights were reported to not differ between the treatment groups
24 (19.5 ± 2.5 g - 21.4 ± 1.6 g or -10% difference). The first study was reported to be conducted at
25 the U.S. EPA laboratory in Cincinnati, OH in which mice were exposed to 2 g/L sodium
26 chloride, or 0.05, 0.5, or 5 g/L TCA in drinking water for 60 weeks. There were 5 animals at
27 each concentration that were sacrificed at 4, 15, 31, and 45 weeks with 30 animals sacrificed at
28 60 weeks of exposure. There were 3 unscheduled deaths in the 0.05 g/L TCA group leaving
29 27 mice at final necropsy. For the other exposure groups there were 29 or 30 animals at final
30 necropsy. In the second study, also conducted in the same laboratory, mice were reported to be
31 exposed to 1.5 g/L neutralized acetic acid or 4.5 g/L TCA for 104 weeks. Serial necropsies were
32 conducted (5 animals per group) at 15, 30, and 45 weeks of exposure and on, 10 animals in the
33 control group at 60 weeks. For this study, a total of 25 animals were sacrificed in interim
34 necropsies in the 1.5 g/L HAC group and 15 in the 4.5 g/L TCA group. There were 7
35 unscheduled deaths in the HAC group and 12 in the 4.5 g/L TCA group leaving 25 animals at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-168 DRAFT—DO NOT CITE OR QUOTE
-------
1 final necropsy and 30 animals in the final necropsy groups, respectively. Study 3 was conducted
2 at the U.S. EPA laboratory in RTF NC. Mice were exposed to deionized water or 0.05 or 0.5 g/L
3 TCA in the drinking water for 104 weeks with serial necropsies (n = 8 per group) conducted at
4 26, 52, and 78 weeks. There were 19-21 animals reported at interim sacrifices and
5 17 unscheduled deaths in the deionized water group, 24 unscheduled deaths in the 0.05 g/L TCA
6 group, and 24 unscheduled deaths in the 0.5 g/L TCA group. This left 34 mice at final necropsy
7 in the control group, 29 mice in the 0.05 g/L TCA group, and 27 mice in the 0.5 g/L group.
8 At necropsy, liver, kidneys, spleen and testes weights were reported to be taken and
9 organs examined for gross lesions. Tissues were prepared for light microscopy and stained with
10 H& E. At termination of the exposure periods, a complete rodent necropsy was reported to be
11 performed. Representative blocks of tissue were examined only in 5 mice from the high dose
12 and control group with the exception of gross lesions, liver, kidney, spleen and testis at interim
13 and terminal sacrifices. If the number of any histopathologic lesions in a tissue was
14 "significantly increased above that in control animals" then that tissue was reported to be
15 examined in all TCA dose groups. For Study #3 a second contract pathologist reviewed 10% of
16 the described hepatic lesions. No "major differences" were reported between the two pathologic
17 diagnoses. The prevalence and multiplicity of hepatic tumors were reported to be derived by
18 performing a histopathologic examination of surface lesions and four sections cut from each of
19 four tissue blocks excised from each liver lobe. Tumor prevalence was reported to be calculated
20 as the percentage of the animals with a neoplastic lesion compared to the number of animals
21 examined. Tumor multiplicity was reported to be calculated by dividing the number of each
22 lesion or combined adenomas and carcinomas by the number of animals examined.
23 Preneoplastic large foci of cellular alteration were also observed over the course of the study.
24 The prevalence and severity of hepatocellular cytoplasmic alterations, inflammation, and
25 necrosis were reported to be determined using a scale based on the amount of liver involved of
26 1 = minimal (occupying 25%), 2 = mild (occupying 25-50%), 3 = moderate (occupying
27 50-75%) and 4 = marked (occupying >75%). The only "significant change outside of the liver"
28 was reported to be testicular degeneration. LDH was determined in arterial blood collected at 30
29 and 60 weeks (Study 1) and 4, 30, and 104 weeks (Study 2). Cyanide insensitive PCO was also
30 reported to be measured. Five days prior to sacrifice, tritiated thymidine (Studies 1 and 2) or
31 BrdU (Study 3) was administered via miniosmotic pumps and the number of hepatocyte nuclei
32 with grain counts >6 were scored in 1,000 cells or chromogen pigment over nuclei (BrdU). The
33 labeling index was calculated by dividing the number of labeled hepatocyte nuclei by total
34 number of hepatocytes scored. Total neoplastic and preneoplastic lesions (multiplicity) were
35 counted individually or combined (adenomas and carcinomas) for each animal. The analysis of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-169 DRAFT—DO NOT CITE OR QUOTE
-------
1 tumor prevalence data were reported to include only those animals examined at the scheduled
2 necropsies or animals surviving to Week 60 (Study 1) or longer than 78 weeks (Studies 2 and 3).
3 The data from all the scheduled necropsies was combined for an overall test of treatment-related
4 effect.
5 For Study #1 (60-week exposure) all TCA treated groups experienced a decrease in
6 drinking water consumption with the decreases in drinking water for the 0.5 and 5 g/L TCA
7 exposure groups reported as statistically significant by the authors. The water consumption in
8 mL/kg-day was reported to be reduced by 11, 17, and 30% in the 0.05, 0.5, and 5 g/L TCA
9 treated groups compared to 2 g/L NaCl control animals as measured by time-weighted mean
10 daily water consumption measured over the study. The control value was reported to be
11 171 mL/kg/day. Although the 0.05 g/L exposure concentrations were not measured, the 0.5 and
12 5 g/L solutions were within 4% of target concentrations. The authors estimated that the mean
13 daily doses were 0, 8 mg/kg, 68 mg/kg and 602 mg/kg per day. For the 102 week studies the
14 mean water consumption with deionized water was reported to be 112 mL/kg/day and
15 132 mL/kg/day for control animals given 1.5 g/L HAC. Therefore, there appeared to be a 35%
16 decrease in water consumption between the controls in Study #1 given 2 g/L NaCl and controls
17 in a Study #3 given deionized water but conducted at a different laboratory. There appeared to
18 be a 23% reduction in water consumption between animals given 2 g/L NaCl and those given
19 1.5 g/L HAC at the same laboratory (Study #2). As the concentrations of TCA were increased,
20 there would be a corresponding increase in the amount of sodium hydroxide needed to neutralize
21 the solutions and a corresponding increase in salts in the solution as well as TCA. The authors
22 did not address nor discuss the differences in drinking water consumption between the differing
23 control solutions between the studies. DeAngelo et al. (1999) reported mean drinking water
24 consumption of 147 mL/kg/day in control mice of over 100 weeks and that the highest dose of
25 DC A (3.5 g/L) reduced drinking water consumption by 26%. Carter et al. (1995) reported that
26 DCA at 5 g/L to decrease drinking water consumption by 64 and 46% but 0.5 g/L DCA to not
27 affect drinking water consumption. While reporting that Study #1 showed that increasing TCA
28 concentration decreased drinking water consumption, the drinking water consumption in Studies
29 #2 and #3 were similar between controls and TCA exposure groups with both being less than the
30 control and low TCA concentration values reported in Study #1 (i.e., in Study #2 the 1.5 g/L
31 HAC and 4.5 g/L TCA drinking water consumption was -130 mL/kg/day and in Study #3 the
32 drinking water consumption was ~112 mL/kg/day for the deionized water control and 0.05 g/L
33 and 0.5 g/L TCA exposure groups). Thus, the drinking water concentrations for Study #3 was
34 -35% less than for the control values for Study #1 and was also -25% less than for DeAngelo et
35 al. (1999). The reasons for the apparently lower drinking water averages for Study #3 and the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-170 DRAFT—DO NOT CITE OR QUOTE
-------
1 lack of effect of the addition of 0.5 g/L TCA that was reported in Study #1 and in other studies,
2 was not discussed by the authors.
3 In Study #1, there was little difference between exposure groups (n = 5) noted for the
4 final body weights (mean range of 27.6-28.1 g) in mice sacrificed after 4 weeks of exposure.
5 However, absolute liver weight and percent liver/body weight ratios increased with TCA dose.
6 The percent liver/body weight ratios were 5.7% ± 0.4%, 6.2% ± 0.3%, 6.6% ± 0.4%, and
7 7.7% ± 0.6% for the 2 g/L NaCl control, 0.05, 0.5, and 5 g/L TCA exposure groups, respectively.
8 These represent 1.09-, 1.16-, and 1.35-fold of control levels that were statistically significant. At
9 15 weeks of exposure the fold increases in percent liver/body weight ratios were 1.14-, 1.16-,
10 and 1.47-fold of controls for 0.05, 0.5, and 5 g/L TCA. At 31 weeks of exposure the fold
11 increases in percent liver/body weight ratios were 0.98-, 1.09-, and 1.59-fold of controls for 0.05,
12 0.5, and 5 g/L TCA. At 45 weeks of exposure the fold increases in percent liver/body weight
13 ratios were 1.13-, 1.45-, and 1.98-fold of controls for 0.05, 0.5, and 5 g/L TCA. At 60 weeks of
14 exposure the percent liver/body weight ratios were 0.94-, 1.25-, 1.60-fold of controls for 0.05,
15 0.5, and 5 g/L TCA. Thus, the range of increase at the lowest level of TCA exposure (i.e.,
16 0.05 g/L) was 0.94- to 1.14-fold of controls. These data consistently show TCA-induced
17 increases in liver weight from 4 to 60 weeks of the study that were dose-related. For the 0.5 g/L
18 exposure group, the magnitude of the increase compared to control was reported to be about the
19 same between weeks 4 and 30 with the highest increase reported to be at Week 45 (1.45-fold of
20 control). In regard to the correspondence with magnitude of difference in dose of TCA and liver
21 weight increase, there was ~2-fold increase in liver weight gain corresponding to 10-fold
22 increases in TCA concentration at 4 weeks of exposure. For the 4 and 15-week exposures there
23 was -3.3- and 3.9-fold difference in liver weight that corresponded to a 100-fold difference in
24 exposure concentration of TCA (i.e., 0.05 vs. 5.0 g/L TCA).
25 The small number of animals examined, n = 5, limit the power of the study to determine
26 the change in percent liver/body weight up to 45 weeks, especially at the lowest dose. However,
27 the 0.05 g/L TCA exposure groups at 4 week and 15 weeks were reported to significantly
28 increase percent liver/body weight ratios. The percent liver/body weight ratios for all of the
29 treatment groups and the ability to detect significant changes were affected by changes in final
30 body weight and changing numbers of animals. After 4 to 30 weeks of exposure, the final body
31 weights of mice increased in control animals but were within 11% of each other between weeks
32 31 and 60. The percent liver/body weight ratios in controls decreased from 4 to 31 weeks and
33 were slightly elevated by 60 weeks compared to the 31-week level. Although control values
34 were changing, there appeared to be no difference between control values and treated values in
35 final body weight for any duration of exposure with the exception of the 5 g/L TCA exposure
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-171 DRAFT—DO NOT CITE OR QUOTE
-------
1 group after 60 weeks of exposure, which was decreased by -15%. At the 31-week and 60-week
2 exposure durations, the 0.05 g/L TCA groups did not have increased percent liver/body weight
3 ratios over controls.
4 In Study #2, conducted in the same laboratory but with a 1.5 g/L HAC solution used for
5 control groups, there was less than 5% difference in final body weights between control mice
6 give HAC and those treated with 4.5 g/L TCA up to 45 weeks. However, final body weight was
7 reduced by TCA treatment by 104 weeks by -15%. Between the interim sacrifices of 15, 30, and
8 45 weeks, the percent liver/body weight ratios in control mice were similar at 15 and 45 weeks
9 (-4.8%) but greater in the 30-week control group (5.3% or -10% greater than other interim
10 control groups). The TCA-induced increases in body weight were 1.60-, 1.40-, and 1.79-fold of
11 control for the 15, 30, and 45 week groups exposed to 4.5 g/L TCA in Study #2. The smaller
12 magnitude of TCA-induced liver weight increase at 30-weeks that that for 15 and 45 weeks, was
13 a reflection of the increased percent liver/body weight ratio reported for the HAC control at that
14 time point.
15 Comparisons can be made between Study #1 and Study #2 for 4.5 g/L or 5.0 g/L TCA
16 exposure levels and controls for 15, 30/31 and 45 weeks of exposure to ascertain the consistency
17 of response from the same laboratory. Although the two studies had differing control solutions
18 and reported different drinking water consumption overall, they were exposing the TCA groups
19 to almost the same concentration of TCA in the same buffered solutions for the same periods of
20 time with the same number of mice per group. Between Study #1 and Study #2, there were
21 consistent percent liver/body weight ratios induced by either 5.0 g/L TCA and 4.5 g/L TCA at
22 weeks 15 and 30/31 (i.e., within 3% of each other). The percent liver/body ratios for these
23 exposure groups ranged from 7.3-7.7% between weeks 15 and 30/31 for the -5.0 g/L TCA
24 exposure in both studies. Final body weights were within 10%. While the percent liver/body
25 weight ratios induced by -5.0 g/L TCA were similar, the magnitude of increase in comparison to
26 the controls was 1.47- and 1.59-fold of control for Study #1, and 1.60- and 1.40-fold of control
27 for Study #2 after 15 and 30/31 weeks of exposure, respectively. At 45 weeks, the percent
28 liver/body weight ratios were within 11% of each other (9.4 vs. 8.4%) and final body weights
29 were within 2% of each for this exposure concentration between the two studies giving a 1.98-
30 and 1.79-fold of control percent liver/body weight, respectively. Thus, the apparent magnitude
31 of TCA-induced increase in percent liver/body weight was affected by control values used as the
32 basis for comparison. The percent liver/body weights reported for either 4.5 g/L TCA or 5.0 g/L
33 TCA exposure groups for weeks 15 and 30/31 was similar between the two studies conducted in
34 the same laboratory.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-172 DRAFT—DO NOT CITE OR QUOTE
-------
1 Study #3 was conducted in a separate laboratory, interim sacrifice times were not the
2 same as for Study #1, the number of animals examined differed (n = 5 for Study #1 and n = 8 for
3 Study #3), and control animals studied for comparative purposes were given different drinking
4 water solutions (deionized water vs. 2 g/L NaCl). Most importantly the body weights reported at
5 52 weeks was much grated than that reported at 45 weeks for Studies #1 and #2. However, a
6 comparison of TC A-induced liver weight gain and the effects of final body weight can be made
7 between the 0.05 and 0.5 g/L TCA exposure groups at 30 weeks (Study #1) and 26 weeks (Study
8 #3), at 45 weeks and 60 weeks (Study #1), and 52 weeks (Study #3). At 31 weeks there was
9 <2% difference in mean final body weights between control and the two TCA-treatment groups
10 in Study #1. There was also little difference between the TCA-treated groups at week in Study
11 #3 at Week 26 and the TCA treatment groups in at Week 31 in Study #1 (i.e., range of
12 42.6-43.5 g for 0.05 and 0.5 g/L TCA treatments in Studies #1 and #3). However, in Study #3,
13 the control value was 12% lower than that of Study #1 for mean final body weight. Based on
14 final body weights, there would be an expectation of similar results between the two studies at
15 the 26 and 30 week time points. At the 45 week (Study #1), and 52-week (Study #3), and
16 60-week (Study #1) durations of exposure, the mean final body weights varied little between
17 their corresponding control groups at each sacrifice time (less than 4% variation between control
18 and TCA-treated groups). However, there was variation in mean final body weights between the
19 differing sacrifice times. Control and TCA-treated groups were reported to have lower mean
20 final body weights at 45 weeks of exposure in Study #1 than at either 30 weeks or at 60 weeks.
21 The 45-week mean final body weights in Study #1 were also reported to be lower than those at
22 52 weeks in Study #3. Control mean body weight values were 28% higher at 52 weeks in Study
23 #3 than 45 weeks in Study #1 and 15% higher for 60 weeks in Study #1. In essence, for
24 Study #1 mean final body weights went down between 31 and 45 weeks of exposure and then
25 went back up at 60 weeks of exposure for control mice (-43, -40, and -44 g for 31, 45, and
26 60 weeks, respectively) as well as for both TCA concentrations. However, for Study #3 final
27 mean body weights went up between 26 and 52 weeks of exposure for control mice (-39 vs.
28 -51 g) and for both TCA concentrations. While for Study #1 the percent liver/body weight
29 ratios were 0.98- and 1.09-fold of control at 31 weeks of exposure, at Week 45 the ratios were
30 1.13- and 1.45-fold of control, and at Week 60 they were 0.94- and 1.25-fold of controls for the
31 0.05 and 0.5 g/L TCA exposure levels, respectively. For Study #3, the pattern differed than that
32 of Study #1. There was a 1.07- and 1.18-fold of control percent liver/body weight for 26 weeks
33 but a 0.92- and 1.04-fold of control percent liver/body weight change at 52 weeks of exposure at
34 0.05 and 0.5 g/L TCA exposure, respectively.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-173 DRAFT—DO NOT CITE OR QUOTE
-------
1 Thus, there appeared to be differences in control and the treatment groups at the 26 week
2 sacrifice groups in Study #3 that was not apparent at the 52-week sacrifice time. Overall, the
3 final body weights appeared to be similar between controls and TCA treatment groups at the
4 52-week sacrifice time in Study #3 and at the 31-, 45-, and 60-week sacrifice times in Study #1.
5 However, although consistent within sacrifice times, the final body weights differed between the
6 various sacrifice times in Studies #1 and #3. The patterns of percent liver/body weight at
7 differing and similar sacrifice times appeared to differ between the Study #1 and Study #3 at the
8 same concentrations of TCA. The largest difference appeared to be between Week 45 group in
9 Study #1 and Week 52 group in Study #3 where both concentrations of TCA were reported to
10 induce increases in percent liver/body weight in one study but to have little difference in the
11 other. The differences in mean final body weights between these two sacrifice times were also
12 the largest although control and TCA-treatment groups had little difference on this parameter.
13 Similar to the work of Kjellstrand et al with TCE (Kjellstrand et al., 1983a), the groups with the
14 lower body weight appeared to have the greatest response in liver weight increase.
15 These data illustrate the variability in findings of percent liver weight induction between
16 laboratories, studies, choice of controls solutions, and the affects of final body weights on this
17 parameter. They also illustrate the limitations for determining either the magnitude or pattern of
18 liver weight increases using a small number of test animals. As animals age the size of their
19 liver changes but also during the latter parts of the lifespan, foci and spontaneously occurring
20 liver tumors can affect liver weight. The results of Study #1 show a consistent dose-response in
21 TCA liver weight increases at 4 and 15 week time periods over a range of concentration from
22 0.05 g/L to 5 g/L TCA.
23 In regard to non-neoplastic pathological changes the authors reported that
24
25 Increased incidences and severity of centrilobular cytoplasmic alterations,
26 inflammation, and necrosis were the only nonproliferative changes seen in livers
27 of animals exposed to TCA for 60 weeks (Tables 7-9; Study 1. Incidences were
28 between 21 and 93%; severity ranged from minimal to mild; and some lesions
29 were transient. Centrilobular cytoplasmic alterations (Table 7) were the most
30 prominent nonproliferative lesion. The incidence and severity were dose related
31 and significantly increased at all TCA concentrations. Centrilobular alterations
32 are a low-grade degeneration of the hepatocytes characterized by an intense
33 eosinophilic cytoplasm with deep basophilic granularity (microsomes) and slight
34 hepatomegaly. The distribution ranged from centrilobular to diffuse. The
35 incidence of inflammation was increased significantly in the 5 g/L TCA treatment
36 group (Table 8), but was significantly lower in the 0.05- and 0.5 g/L groups
37 between 31 and 45 weeks, but abated by 60 weeks. There was a significant dose-
38 related trend, but a significant increase in severity was only found at 5 g/L. No
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-174 DRAFT—DO NOT CITE OR QUOTE
-------
1 alteration in the severity of this lesion was observed. The occurrence and severity
2 of nonproliferative lesions in animals exposed to 0.5 and 4.5 g/L TCA for 104
3 weeks were similar to those observed at 60 weeks (data not shown). No
4 pathology outside the liver was observed except for a significant dose-related
5 trend and incidence of testicular tubular degeneration at 0.5 and 5 g/L TCA.
6
7 The results shown in Table 7 by the authors for the 60-week TCA-exposed mice did not
8 show a dose-response for either incidence or severity of centrilobular cytoplasmic alterations.
9 They reported a 7, 48, 21, and 93% incidence and a 0.10 ± 0.40, 0.70 ± 0.82, 0.34 ± 0.72 and
10 1.60 ± 0.62 mean severity score for control, 0.05, 0.5, and 5.0 g/L TCA exposure groups,
11 respectively. Thus, for control, 0.05 and 0.5 g/L TCA exposure there was less than minimal (i.e.,
12 score of 1 or occupying less than 25% of the microscopic field) severity of this finding for the 27
13 to 30 mice examined in each group. Only slight hepatomegaly is noted by the authors to be
14 included in their description of the centrilobular cytoplasmic alteration. Interestingly, the
15 elevation of this parameter for both incidence and severity in the 0.05 g/L TCA exposed group
16 compared to 0.5 g/L exposure group did not correspond to an increase in percent liver/body
17 weight for this same exposure group. While the percent liver/body weight ratio was 32% higher,
18 the incidence and severity of this lesion were reported to be half that in the 0.5 versus 0.05 g/L
19 exposure groups after 60 days of TCA exposure. Thus, TCA-induced hepatomegaly did not
20 appear to be associated with this centrilobular cytoplasmic change. Similarly the incidence of
21 hepatic inflammation was reported to be 10, 0, 7, and 24% and severity, 0.11 ± 0.40, 0.09 ± 0.30,
22 0.12 ± 0.33, and 0.29 ± 0.48 for control, 0.05, 0.5, and 5.0 g/L TCA exposure groups,
23 respectively. Thus, at no TCA exposure concentration was the incidence more than 24% and the
24 severity was considerably less than minimal. The reported results for hepatic necrosis were
25 pooled from data from the 5 mice exposed for either 30 or 45 weeks (n = 10 total). No
26 incidences of necrosis were reported for either control or 0.05 g/L TCA exposed mice. At
27 0.5 g/L TCA 3/10 mice were reported to have necrosis but at a severity level of 0.50 ± 0.97. At
28 5.0 g/L TCA 5/10 mice were reported to have necrosis but at a severity level of 1.30 ± 1.49. The
29 limitations of the small number of animals pooled in these data are obvious. However, there
30 does not appear to be much more than minimal necrosis at the highest dose of TCA between 30
31 and 45 weeks and this response is reported by the authors to be transient.
32 Serum LDH activity was reported by the authors for 31 and 60 week TCA exposures in
33 Study #1. They state that
34
35 There was a dose-related trend at 31 weeks; serum LDH was significantly
36 increased at 0.5 and 5 g/L TCA (161 ± 39 and 190 ± 44, respectively vs. 100 ± 28
37 IU for the control). LDH activity returned to control levels at 60 weeks.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-175 DRAFT—DO NOT CITE OR QUOTE
-------
1 Similarly, elevated LDH levels were observed at early time periods for 0.5 and
2 4.5 g/L TCA during the 104 week exposure (data not shown: Studies 2 and 3).
3
4 The data presented by the author for Study #1 are from 5 animals/group for the 30-week results
5 and 30 animals/group for the 60-week results. Of interest is for the 60-week data, there appears
6 to be 50% decreased in LDH activity at 0.05 and -25% decrease in LDH activity at 0.5 g/L TCA
7 treatment with the LDH level reported to be the same as control for the 5 g/L TCA exposure
8 group. For the 31-week data, in which only 5 animals were tested in each treatment group, there
9 appeared to be a slight increase at the 0.5 g/L (60% increase over control) and 5 g/L (90%
10 increase over control) treatment groups. The data for necrosis detected by light microscopy and
11 by LDH level is consistent with no changes from control detected at the 0.05 g/L TCA treatment
12 group and less than minimal necrosis of on a 60% increase in LDH level over control reported
13 for 0.5 g/L TCA treatment. Even at the highest dose of 5.0 g/L TCA there is still little necrosis
14 or LDH release reported over control.
15 Data for testicular tubular degeneration was reported for Study #1 after 60-weeks of TCA
16 exposure. The incidence of testicular tubular degeneration was reported to be 7, 0, 14, and 21%
17 for mice exposed to 2.0 g/L NaCl, 0.05, 0.5, and 5.0 g/L TCA. The severity of the lesions was
18 reported to be 0.10 ± 0.40, 0, 0.17 ± 0.47, and 0.21 ± 0.41 with a significant trend with dose
19 reported by the authors for severity and for the 0.5 and 5 g/L treatment groups to be significantly
20 increased over control incidence levels. Of note, similar to the percent liver/body weight ratios
21 and hepatic inflammation values for this data set, the values for testicular tubular degeneration
22 were slightly higher in control mice than 0.05 g/L TCA exposed mice. In regard to mean
23 severity levels for testicular degeneration, although still minimal, there was little difference
24 between the results for reported for the 0.5 g/L TCA and 5.0 g/L TCA exposed mice.
25 In regard to peroxisome proliferation, liver PCO activity was presented for up to
26 60 weeks (Study #1) and 104 weeks (Study #2). Similar to the data for LDH activity, -30
27 animals were examined at the 60-week time point but only 5 animals per exposure group were
28 examined for 4-, 15-, 31-, and 45-week results. The data are presented in a figure and in some
29 instances hard to determine the magnitude of change. Similar to other reports, the baseline level
30 of PCO activity was variable between control groups and ranged 2.7-fold (-1.49 to 4.06 nmol
31 NAD reduced/min/mg protein given by the authors). There appeared to be little change in PCO
32 activity between the 0.05 g/L TCA exposure and control levels for up to 45 weeks of exposure
33 (i.e., the groups with n = 5) in Study #1. For the 60-week group the 0.05 g/L TCA group PCO
34 activity was -1.7-fold of control but was not statistically significant. For the 0.5 g/L TCA
35 treatment groups, the increase ranged from -1.3- to 2.7-fold of control after 4-, 15-, 31-, and 45-
36 weeks of exposure with the largest differences reported at 4 and 60 weeks (i.e., 2.2- and 2.7-fold
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-176 DRAFT—DO NOT CITE OR QUOTE
-------
1 of control, respectively). For the 5.0 g/L TCA exposure groups, the increase ranged from -3.2-
2 to ~5.7-fold of control after 4, 15, 31, and 45 weeks of exposure. While the data at 60-weeks had
3 the most animals examined (-30 vs. 5) with -1.7-, 2.7-, and 4.5-fold of control PCO activity, at
4 this time period the authors report the occurrence of tumors had already occurred. At the earlier
5 time points of 4 and 15 weeks, there was a difference in the magnitude TCA-induced increase in
6 PCO activity. As displayed graphically, at 4 weeks the PCO increase was -1.3-, 2.4-, and
7 5.3-fold of control for 0.05, 0.5, and 5.0 g/L TCA, respectively, while at 15 weeks, the PCO
8 levels were decreased by 5%, increased to 1.3-fold, and increased to 3.2-fold of control with only
9 the 5.0 g/L treatment group difference to be statistically significant.
10 For Study #2 the authors present a figure (Figure #4) that states that PCO values were
11 given for mice given HAC or 4.5 g/L TCA for 4-60 weeks. However, the data presented in #4
12 appears to be for 15-, 30-, 45- and 104-week exposures. The number of mice is not given in the
13 figure but the methods section states that serial section were conducted on 5 mice/group for these
14 interim sacrifice periods. The number of mice examined for PCO activity at 104 weeks was not
15 given by the authors but the number of mice at final sacrifice was given as 25. The levels of
16 PCO in the control tissues varied by -33% for weeks 15 to 45 but there was a -5-fold difference
17 between the level reported at 104 weeks and that for the earlier time periods in control mice
18 shown in the figures (-2.23 vs. 0.41 nmol NAD reduced/min/mg protein as given by the
19 authors). The increase over control induced by 4.5 g/L TCA in Study #2 was shown to be -6.9-,
20 4.8-, 3.6-, and 19-fold of controls for 15, 30, 45 and 104 weeks, respectively.
21 Therefore, at a comparable level of TCA exposure (-5.0 g/L), number of mice examined
22 (n = 5), and durations of exposure (15, 30, and 45 weeks), the increase in PCO activity induced
23 by -5.0 g/L TCA varied between 3.2- to 5.7-fold of control in Study #1 and between 3.6- to
24 6.9-fold of control in Study #2. There was not a consistent pattern between the two studies in
25 regard to level of PCO induction from -5 g/L TCA and duration of exposure. The lowest TCA-
26 induced PCO activity increase was recorded at 15 weeks in Study #1 (i.e., 3.2-fold of control)
27 and highest PCO activity increase was recorded at 15 weeks in Study #2 (i.e., 6.9-fold of
28 control). No PCO data were reported for data in Study #3 with the exception of the authors
29 stating that "PCO activity was significantly elevated for the 0.5 g/L TCA exposure over the 104
30 weeks (study 3). The extent of the increases was similar to those measured for 0.5 g/L TCA
31 (200-375%: data not shown) in Study 1." No other details are given for PCO activity in
32 Study #3.
33 Hepatocyte proliferation was reported by the authors to be assessed by either
34 incorporation of tritiated thymidine (Studies #1 and #2) or BrdU (Study #3) into hepatocyte
35 nuclei. As noted previously, these techniques measure DNA synthesis and not necessarily
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-177 DRAFT—DO NOT CITE OR QUOTE
-------
1 hepatocyte proliferation. The authors did not report if specific areas of the liver were analyzed
2 by autoradiographs or how many autoradiographs were examined in the analyses they conducted.
3 For later time points of examination (60-104 weeks) the authors did not indicate whether
4 hepatocytes in foci or adenomas were excluded from DNA synthesis reports. The authors
5 present data for what are clearly, 31, 45, and 60 week exposure for Study #1 as the percent
6 tritiated thymidine labeled nuclei. An early time point that appears to be 8 weeks is also given.
7 However, for Study #1 only 4 week and 15 week durations were tested so it cannot be
8 established what time period the earlier time point represents. What is very apparent from the
9 data presented for Study #1 is that the baseline level of tritiated thymidine incorporation was
10 relatively high and highly variable for the 5 animals examined (-8% of hepatocytes were
11 labeled). There did not appear to be an apparent pattern of TCA treatment groups at this
12 timepoint with the 0.05 and 5.0 g/L TCA groups having a similar percentage of labeled
13 hepatocytes and for 0.5 g/L TCA reported to have a 60% reduction in labeled hepatocytes. After
14 31 weeks of exposure the control values were reported to be 2% of hepatocytes labeled. The
15 authors report that only the 5.0 g/L TCA group had a statistically significant increase of control
16 and was elevated to -6% of hepatocytes. The two lower doses of TCA had similar reported
17 incidences of labeled hepatocytes of 4.5% that were not reported to be statistically significant.
18 For the 45-week exposure period in Study #1, the control value was reported to be 1.2% with
19 only the 5.0 g/L TCA value reported to be statistically significantly increased at 3.2% and the
20 other two TCA groups to be similar to control. Finally, for the 60 week group from Study #1,
21 the control value was reported to be 0.6% of hepatocytes labeled and the only the 0.5 g/L TCA
22 dose reported to be statistically significantly increased over control at 3.2%. What is clear from
23 this study is that the control value for the unidentified early time point is much higher than the
24 other values. There should not be such a large difference in mature mice nor such a high level.
25 The difference in control values between the earlier time point and the 31 -week time point was
26 4-fold. The difference between the earlier time point and the 45-week time point was ~7-fold.
27 There did not appear to be an increase in hepatocyte tritiated thymidine labeling due to any
28 concentration of TCA at the early unidentified time point (-Week 10 from the figure) from
29 Study #1. There was no dose-response apparent for the other study periods and the percent of
30 hepatocytes labeled were 3% or less. These results indicated DNA synthesis was not increased
31 by 10-60 week exposures to TCA exposure that induced increased liver tumor response.
32 For Study #2 results were reported for tritiated thymidine incorporation into hepatocytes
33 in a figure that was labeled as 4.5 g/L TCA and control tissue for 104 weeks but showed data for
34 15, 30, and 45 weeks of exposure. Of note is that the control values for this study were much
35 lower than that reported for Study #1. The percent of hepatocytes labeled with tritiated
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-178 DRAFT—DO NOT CITE OR QUOTE
-------
1 thymidine was reported to be -2% for the 15 week exposure period and less than 1% for the 30-
2 and 45-week exposure periods. For the 4.5 g/L TCA exposures the percent hepatocytes labeled
3 with tritiated thymidine were -2-4% at all time points with only the 45 week period identified
4 by the authors as statistically significant.
5 For Study #3, rather than tritiated thymidine, BrdU was used as a measure of DNA
6 synthesis. The results are presented in Figure #8 of the report in which the 0.5 g/L TCA
7 concentration is mislabeled as 0 g/L and the figure is mislabeled as having a duration of
8 104 weeks but the data are presented for 26, 52, and 78 weeks of exposure. The percent of
9 hepatocytes at 26 weeks was reported to be -1-2% for the control, 0.05 and 0.5 g/L TCA
10 groups. At 52 weeks the control value was -1% the 0.05 g/L TCA value was less than 0.1% and
11 the 0.5 g/L TCA value was -3.5% but not statistically significant. At 78 weeks of exposure the
12 control value was reported to be -0.2% with only the 0.05 g/L TCA group having a statistically
13 significant increase over control.
14 From these data, the estimated control values for DNA synthesis at similar time points of
15 exposure ranged from 0.4 to 2% at 26-31 weeks and -0.1 to 1.2% at 45-52 weeks. The results
16 for Study #1 and #2 were inconsistent in regard to the magnitude of tritiated thymidine
17 incorporation but consistent in that there was a lot of variability in these measurements, not a
18 consistent pattern with time that was TCA-dose related, and, even at the highest dose of TCA,
19 did not indicate much of an increase in cell proliferation 15-45 weeks of exposure. Similarly the
20 results for Studies #1 and #3 indicate that the two lower doses of TCA there were not generally
21 statistically significant increases in DNA synthesis from 15-45 weeks of exposure although there
22 was an increase in liver tumor response at later time points.
23 The authors reported that "all gross and microscopic histopathological alterations were
24 consistent across the three studies." However, the histological descriptions that follow were
25 focused on the liver for both neoplastic and non-neoplastic parameters. As stated above, only a
26 few animals (n = 5) from the control and high TCA dose level were examined for lesions other
27 than liver, kidneys, spleen and testes. Thus, whether other neoplastic lesions were induced by
28 TCA exposure cannot be determined from this set of studies.
29 Study #1 was conducted for 60 weeks. Although of short duration and using 30 or less
30 animals, the authors reported in the text that
31
32 a significant trend with dose was found for liver cancer. The prevalence and
33 multiplicity of adenomas (38%; 0.55 ± 0.15) or carcinoma (38%; 0.42 ±0.11)
34 were statistically significant at 602 mg/kg/day TCA compared to control (7%;
35 0.07 ± 0.05) [sic for both adenoma and carcinoma the same value was given,
36 mean ± SD]. When either an adenoma or a carcinoma was present, statistical
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-179 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant was seen at both 5 g/L (55%; 1.00 ± 0.19) and 0.5 g/L (38%: 0.52±
2 0.14 TCA exposure groups compared to control (13%; 0.13 ± 0.06). No
3 significant change in liver neoplasia were reported to be observed by the authors
4 at 0.05 g/L TCA. Preneoplastic large foci of cellular alteration (24%) were seen
5 in the 5 g/L TCA control compared to control.
6
7 Although not statically significant, there was an incidence of 15% adenoma in the
8 0.05 g/L TCA treatment group (n = 27) and a multiplicity of 0.15 ± 0.07 adenomas/mouse
9 reported with both values being twice that of the values given for the controls (n = 30). The
10 incidence and multiplicity for carcinomas was approximately the same for the 0.05 g/L TCA
11 treatment group and the control group. Given the small number of animals examined, the study
12 was limited in its ability to determine statistical significance for the lower TCA exposure level.
13 The fold increases of incidence and multiplicity of adenomas at 60 weeks was 2.1-, 3.0-, and
14 5.4-fold of control incidence and 2.1-, 3.4-, and 7.9-fold of control multiplicity for 0.05, 0.5, and
15 5 g/L exposure to TCA. For multiplicity of adenomas and carcinomas combined there was a
16 1.46-, 4.0-, and 7.68-fold of control values. Analysis of tumor prevalence data for this study
17 included only animals examined at scheduled necropsy. Since most animals survived until
18 60 weeks, most were included and a consistent time point for tumor incidence was reported.
19 There are significant discrepancies for reporting of data for tumor incidences in this
20 report for the 104 week data. While the methods section and table describing the dose
21 calculation and animal survival indicate that Study #3 control animals were administered
22 deionized water and those from Study#2 were given HAC, Table 6 of the report gives 2 g/L
23 NaCl as the control solution given for Study #2 and 1.5 g/L HAC for Study #3. A comparison of
24 the descriptions of animal survival and tumor incidence and multiplicity between the results
25 given in DeAngelo et al. (2008) and George et al. (2000) (see Table E-10) shows not only that
26 the control data presented in DeAngelo et al. (2008) for Study #3 to be the same data as that
27 presented by George et al. (2000) previously, but also indicates that rather than 1.5 g/L HAC, the
28 tumor data presented in DeAngelo et al. (2008) is for mice exposed to deionized water.
29 DeAngelo et al. (2008) did not report that these data were from a previous publication.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-180 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table E-10. Comparison of descriptions of control data between George et
al. (2000) and DeAngelo et al. (2008)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Descriptor
Species
Strain
Gender
Age
Source
Mean initial body wt
Water consumption
Laboratory
# Animals at start
# Animals at interim sac.
# Unscheduled deaths
# Animals at final sacrifice
# Animals for pathology
Adenoma incidence
Adenoma multiplicity
Carcinoma incidence
Carcinoma multiplicity
George et al., 2000
Mouse
B6C3F1
Male
28-30 days
Charles River, Portage
19.5 ± 2.5 g
111.7mL/kg/day
RTPNC
72
22
16
34
65
21.40%
0.21 ±0.06
54.80%
0.74 ±0.12
DeAngelo et al., 2008
Mouse
B6C3F1
Male
28-30 days
Charles River, Portage
19.5 ± 2.5 g
112mL/kg/day
RTPNC
72
21
17
34
63
21%
0.21 ±0.06
55%
0.74 ±0.12
RTF NC = Research Triangle Park, North Carolina.
For Studies #2 and #3 tumor prevalence data were reported in the methods section of the
report to include necropsies of animals that survived greater than 78 weeks and thus, included
animals that were scheduled for necropsy but also those which were moribund and sacrificed at
differing times. Thus, for the longer times of study, there was a mixture of exposure durations
that included animals that were ill and sacrificed early and those that survived to the end of the
study. Animals that were allowed to live for longer periods or who did not die before scheduled
sacrifice times had a greater opportunity to develop tumors. However, animals that died early
may have died from tumor-related causes. The mislabeling of the tumor data in DeAngelo et al.
(2008) has effects on the interpretation of results for if the tumor results table was not mislabeled
it would indicated 17 animals were included in the liver tumor analysis that were not included in
the final necropsy and that the 7 unscheduled deaths could not account for the total number of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-181 DRAFT—DO NOT CITE OR QUOTE
-------
1 "extra" mice included in the tumor analysis so some of the animals had to have come from
2 interim sacrifice times (78 weeks or less) and that for Study #3 the data from 9 animals at
3 terminal sacrifice were not used in the tumor analysis. Not only was the control data mislabeled
4 for Study #3, but the control data were also apparently mislabeled for Study #2 as being 2.0 g/L
5 NaCl rather than 1.5 g/L HAC. Of the 42 animals used for the tumor analysis in Study #3, only
6 34 were reported to have survived to interim sacrifice so that 8 animals were included from
7 unscheduled deaths. However, the authors report that there were 17 unscheduled deaths in the
8 study not all were included in the tumor analysis. The basis for the selection of the 8 animals for
9 tumor analysis was not give by the authors.
10 Not only are the numbers of control animals used in the tumor analysis different between
11 two studies (25 mice in Study #2 and 42 mice in Study #3), but the liver tumor results reported
12 for Study #2 and Study #3 were very different. Of the 42 "control" mice examined from Study
13 #3, the incidence and multiplicity of adenomas was reported to be 21% and 0.21 ± 0.06,
14 respectively. For carcinomas, the incidence and multiplicity was reported to be 55% and
15 0.74 ± 0.12, respectively, and for the incidence and multiplicity of adenomas and carcinomas
16 combined reported to be 64% and 0.93 ± 0.12, respectively. For the 25 mice reported by the
17 authors for Study #2 to have been treated with "2.0g/L NaCl" but were probably exposed to
18 1.5 g/L HAC, the incidence and multiplicity of adenomas was 0%. For carcinomas, the
19 incidence and multiplicity was reported to be 12% and 0.20 ± 0.12, respectively and for the
20 incidence and multiplicity of adenomas and carcinomas combined to be 12% and 0.20 ± 0.12,
21 respectively. Therefore, while -64% the 42 control mice in Study #3 were reported to have
22 adenomas and carcinomas, only 12% of the 25 mice were reported to have adenomas and
23 carcinomas in Study #2 for 104-weeks.
24 While the effect of using fewer mice in one study versus the other will be to reduce the
25 power of the study to detect a response, there are additional factors that raise questions regarding
26 the tumor results. Not only were the tumor incidences were reported to be higher in control mice
27 from Study #3 than Study #2, but the number of unscheduled deaths was reported to also be
28 2-fold higher. The age, gender, and strain of mouse were reported to be the same between
29 Study #2 and #3 with only the vehicles differing and weight of the mice to be reported to be
30 different. Although the study by George et al. (2000) describes the same control data set as for
31 Study #3 as being for animals given deionized water, there is uncertainty as to the identity of the
32 vehicle used for the tumor results reported for Study #3 and there are some discrepancies in
33 reporting between the two studies. As discussed below in Section E.2.5, the differences in the
34 weight of the mice between Studies #1, #2, and #3 is critical to the issue of differences in
35 background tumor rate and hence interpretability of the study.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-182 DRAFT—DO NOT CITE OR QUOTE
-------
1 As noted by Leakey et al. (2003b), the greatest correlation with liver tumor incidence and
2 body weight appears between the ages of 20 and 60 weeks in male mice. As reported in
3 Section E.2.5, the mean 45-week body weight reported for control male B6C3F1 mice in the
4 George et al. (2000) study, which is the same control data as DeAngelo et al. (2008) was -50 g.
5 This is a much greater body weight than reported for Study #1 at 45 weeks (i.e., 39.6 g) and for
6 Study #2 at 45 weeks (i.e., 39.4 g). Using probability curves presented by Leakey et al. (2003b),
7 the large background rate of 64% of combined adenomas and carcinomas for Study #3 is in the
8 range predicted for such a large body weight (i.e., -65%). Such a high background incidence
9 compromises a 2-year bioassay as it prevents demonstration of a positive dose-response
10 relationship. Thus, Study #3 of DeAngelo et al. (2008) is not comparable to the results in
11 Study #1 and #2 for the determination of the dose-response for TCA.
12 The accurate determination of the background liver tumor rate is very important in
13 determining a treatment related effect. The very large background level of tumor incidence
14 reported for Study #3 makes the detection of a TCA-related change in tumor incidence at low
15 exposure levels very difficult to determine. Issues also arise as to what the source of the tumor
16 data were in the TCA-treatment and control groups in Study #3. While 29 mice exposed to
17 0.05 g/L TCA were reported to have been examined at terminal sacrifice, 35 mice were used for
18 liver tumor analysis. Similarly, while 27 mice exposed to 0.5 g/L TCA were reported to have
19 been examined at terminal sacrifice, 37 mice were used for tumor analysis. Finally, for the
20 42 control animals examined for tumor pathology in the control group, 34 were examined at
21 terminal sacrifice. Clearly more animals were included in the analyses of tumor incidence and
22 multiplicity than were sacrificed at the end of the experiment. What effect differential addition
23 of the results from mice not sacrificed at 104 weeks and the selection bias that may have resulted
24 from their inclusion on these results cannot be determined. Not only were the background levels
25 of tumors reported to be increased in the control animals in Study #3 compared to Study #2 at
26 104 weeks, but the rate of unscheduled deaths was doubled. This is also an expected
27 consequence of using much larger mice (Leakey et al., 2003b).
28 For the 35 mice examined after 0.05 g/L TCA in Study #3, the incidence and multiplicity
29 of adenomas was reported to be 23% and 0.34 ±0.12, respectively. For carcinomas, the
30 incidence and multiplicity was reported to be 40% and 0.71 ±0.19, respectively, and for the
31 incidence and multiplicity of adenomas and carcinomas combined reported to be 57% and
32 1.11 ± 0.21, respectively. For the 37 mice examined after 0.5 g/L TCA in Study #3, the
33 incidence and multiplicity of adenomas was reported to be 51% and 0.78 ±0.15, respectively.
34 For carcinomas, the incidence and multiplicity was reported to be 78% and 1.46 ±0.21,
35 respectively, and for the incidence and multiplicity of adenomas and carcinomas combined
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-183 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported to be 87% and 2.14 ± 0.26, respectively. Thus, at 0.5 g/L TCA the results presented for
2 this study for the "104 week" liver tumor data were significantly increased over the reported
3 control values. However, these results are identical to those reported in Study #3 for a 10-fold
4 higher concentration of TCA (4.5 g/L TCA) for the same 104 weeks of exposure but in the much
5 larger mice. Of the 36 animals exposed to 4.5 g/L TCA in Study #2 and included in the tumor
6 analysis, 30 animals were reported to be examined at 104 weeks. The incidence and multiplicity
7 of adenomas was reported to be 59% and 0.61 ± 0.16, respectively. For carcinomas, the
8 incidence and multiplicity was reported to be 78% and 1.50 ± 0.22, respectively, and for the
9 incidence and multiplicity of adenomas and carcinomas combined reported to be 89% and
10 2.11 ± 0.25, respectively.
11 The importance of selection and determination of the control values for comparative
12 purposes of tumor induction are obvious from these data. The very large difference in control
13 values between Study #2 and Study #3 is the determinant of the magnitude of the dose response
14 for TCA after 104 weeks of exposure. The tumor response for 0.5 and 4.5 g/L TCA exposure
15 between the two experiments was identical. Therefore, only the background tumor rate
16 determined the magnitude of the response to treatment. If a similar control values (i.e., a
17 historical control value) were used in these experiments, there would appear to be no difference
18 in TCA-tumor response between 0.5 and 4.5 g/L TCA at 104 weeks of exposure. DeAngelo et
19 al. (1999) report for male B6C3F1 mice exposed only water for 79 to 100 weeks the incidence of
20 carcinomas to be 26% and multiplicity to be 0.28 lesions/mouse. For 100-week data, the
21 incidence and prevalence of adenomas was reported to be 10% and 0.12 ± 0.05 and for
22 carcinomas to be 26% and 0.28 ± 0.07. Issues with reporting for that study have already been
23 discussed in Section E.2.3.2.5. However, the data for DeAngelo et al. (1999) are more consistent
24 with the control data for "1.5 g/L HAC" for Study #2 in which there were 0% adenomas and
25 12% carcinomas with a multiplicity of 0.20 ±0.12, than for the control data for Study #3 in
26 which 64% of the control mice were reported to have adenomas and carcinomas and the
27 multiplicity was 0.93 ± 0.12. If either the control data from DeAngelo et al. (1999) or Study #2
28 were used for comparative purposes for the TCA-treatment results of Study #2 or #3, there
29 would be a dose-response between 0.05 and 0.5 g/L TCA but no difference between 0.5 and
30 4.5 g/L TCA after 100 weeks of exposure. The tumor incidence would have peaked at -90% in
31 the 0.5 and 4.5 g/L TCA exposure groups. These results would be more consistent with the
32 60-week results in Study #1 in which 0.5 and 5 g/L TCA exposure groups already had incidences
33 of 38 and 55% of adenomas and carcinomas combined, respectively, compared to the 13%
34 control level. With increased time of exposure the differences between the two highest TCA
35 exposure concentrations may diminish as tumor progression is allowed to proceed further.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-184 DRAFT—DO NOT CITE OR QUOTE
-------
1 However, the use of the larger and more tumor prone mice in Study #3 also increases the tumor
2 incidence at the longer period of study.
3 The authors also presented data for multiplicity of combined adenomas or carcinomas for
4 mice sacrificed at weeks 26, 52, and 78 for Study #3 (n = 8 per group). No indication of
5 variability of response, incidence data, statistical significance, or data for adenomas versus
6 carcinomas, or the incidence of adenomas was reported. The authors reported that "neoplastic
7 lesions were first found in the control and 0.05 g/L TCA groups at 52 weeks. At 78 weeks,
8 adenomas or carcinomas were found in all groups (0.29, 0.20, and 0.57 tumors/animals for
9 control, 0.05 g/L TCA, and 0.5 g/L TCA, respectively)." Because no other data were presented
10 at the 52 and 78 week time points in this study, these results cannot be compared to those
11 presented for Study #1, which was conducted for 60 weeks. Of note, the results presented from
12 Study #1 for 60 weeks of exposure to control, 0.05 g/L or 0.5 g/L TCA exposure in 27-30 mice
13 show a 13, 15, and 38% incidence of hepatocellular adenomas and carcinomas and a multiplicity
14 of 0.13 ± 0.06, 0.19 ± 0.09, and 0.52 ± 0.14, respectively. Both the incidence and multiplicity of
15 adenomas were 2-fold higher in the 0.05 g/L TCA treatment group than for the control.
16 However, the interim data presented by the authors from Study #3 for 52 weeks of exposure in
17 only 8 mice per group gives a higher multiplicity of adenomas and carcinomas for control
18 animals (-0.25) than for either 0.05 or 0.5 g/L TCA treatments. Again, comparisons between
19 Study #2 and #3 are difficult due to difference in mouse weight.
20 Of note, there are no descriptions given in this report in regard to the phenotype of the
21 tumors induced by TCA or for the liver tumors reported to occur spontaneously in control mice.
22 Such information would have been of value as this study reports results for a range of TCA
23 concentration and for 60 and 100 weeks of exposure. Insight could have been gained as to the
24 effects of differing concentrations of TCA exposure, whether TCA-induced liver tumors had a
25 similar phenotype as those occurring spontaneously, as well as information in regard to effects
26 on tumor progression and heterogeneity.
27 Although only examining tissues from 5 mice from the control and high-dose groups only
28 at 104 weeks at organ sites other than the liver, the authors report that
29
30 neoplastic lesions at 104 weeks (Studies #2 and #3) at organ sites other than the
31 liver were found in the lung, spleen, lymph nodes, duodenum (lymphosarcoma),
32 seminal vesicles, skin, and thoracic cavity of control and treated animals. All
33 were considered spontaneous for the male B6C3F1 mouse and did not exceed the
34 tumor incidences when compared to a historical control database (Haseman 1984;
35 NIEHS, 1998).
36
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-185 DRAFT—DO NOT CITE OR QUOTE
-------
1 No data were shown. The limitations involved in examining only 5 animals in the control and
2 high-dose groups, and the need to examine the concurrent control data in each experiment,
3 especially given the large variation in liver tumor response between long-term studies carried out
4 in the two different laboratories used for Study #2 and Study #3 using the same strain and gender
5 of mouse, make assertions regarding extrahepatic carcinogenicity of TCA from this study
6 impossible to support.
7 A key issue raised from this study is whether changes in any of the parameters measured
8 in interim sacrifice periods before the appearance of liver tumors (i.e., 4-15 weeks)
9 corresponded to the induction of liver tumors. The first obstacle for determining such a
10 relationship is the experimental design of these studies in which only a full range of TCA
11 concentrations is treated for 60 weeks of exposure with a small number of animals available for
12 determination of a carcinogenic response (i.e., 30 animals or less in Study #1) and a very small
13 number of animals (n = 5 group) examined for other parameters. Also as stated above, PCO
14 activity was highly variable between controls and between treatment groups (e.g., the PCO
15 activity for Study #1 and #2 at ~5 g/L exposure for 15 weeks). On the other hand, most of the
16 animals that were examined at terminal sacrifice were also utilized for the tumor results without
17 the differential deletion or addition of "extra" animals for the tumor analysis. For the 60-week
18 data in Study #1 there appeared to be a consistent dose-related increase in the incidence and
19 multiplicity of tumors after TCA exposure (Table E-l 1). The TCA-induced increases in liver
20 tumor responses can be compared with both increased liver weight and PCO activity that were
21 also reported to be increased with TCA dose as earlier events. Although the limitations of
22 determining the exact magnitude of responses has already been discussed, as shown below, the
23 incidence and multiplicity of adenomas show a dose-related increase at 60 weeks. However, the
24 magnitude of differences in TCA concentrations was not similar to the magnitude of increased
25 liver tumor induction by TCA after 60 weeks of exposure.
26
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-186 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
Table E-ll. TCA-induced increases in liver tumor occurrence and other parameter over control after 60
weeks (Study #1)
Dose TCA g/L
NaCl
0.05
0.5
5.0
Adenomas
Incidence 7%
15% (2.1 -fold)
21% (3. 0-fold)
3 8% (5 .4-fold)
Multiplicity 0.07
0.15 (2. 1-fold)
0.24 (3. 4-fold)
0.55 (7.9-fold)
Adenomas or carcinomas
Incidence 13%
15% (1.2-fold)
38% (2.9-fold)
55% (4.2-fold)
Multiplicity 0.13
0.19(1. 5-fold)
0.52 (4.0-fold)
1.00 (7. 7-fold)
% liver/body weight
4-week
1.09-fold
1.16-fold
1.35-fold
15 -week
1.14-fold
1.16-fold
1.47-fold
PCO activity
4-week
1.3-fold
2.4-fold
5.3-fold
15 -week
1.0 -fold
1.3-fold
3.2-fold
TO'
w g
"
I
H I
O
H
W
-------
1 First of all, the greater occurrence of TCA-induced increases in adenomas than
2 carcinomas reported after 60 weeks of exposure would be expected for this abbreviated duration
3 of exposure as they would be expected to occur earlier than carcinomas. For adenoma induction,
4 there was a ~2-fold increase between the 0.05 g/L dose of TCA and the control group for
5 incidence (7 vs. 15%) and multiplicity (0.07 vs. 0.15 tumors/animals). However, an additional
6 10-fold increase in TCA dose (0.5 g/L) only resulted in a reported 1.8-fold greater incidence
7 (15 vs. 21%) and 2.2-fold increase in multiplicity (0.15 vs. 0.24 tumors/animal) of control
8 adenoma levels. An additional 10-fold increase in dose (5.0 vs. 0.5 g/L TCA) resulted in a
9 2.2-fold increase in incidence (21 vs. 38%) and 2.9-fold increase in multiplicity (0.24 vs.
10 0.55 tumors/animal) of control adenoma levels. Thus, a 100-fold difference in TCA exposure
11 concentration resulted in differences of 4-fold of control incidence and 6-fold of control
12 multiplicity for adenomas. For adenomas or carcinomas combined (a parameter that included
13 carcinomas for which only the two highest exposure levels of TCA were reported to increase
14 incidence and multiplicity) the incidences were reported to be 13, 15, 38, and 55%, and the
15 multiplicity reported to be 0.13, 0.19, 0.52, and 1.00 for control, 0.05, 0.5, and 5.0 g/L TCA at
16 60 weeks. For multiplicity of adenomas or carcinomas, the 0.05 g/L TCA exposure induced a
17 1.5-fold increase over control. An additional 10-fold increase in TCA (0.5 g/L) induced a 6-fold
18 increase in tumors/animal. An additional 10-fold increase in TCA (5.0 vs. 0.5 g/L) induced an
19 additional 2.2-fold increase in tumors/animal. Therefore, using combinations of adenomas or
20 carcinomas, there was a 13-fold increase in multiplicity that corresponded with a 100-fold
21 increase in dose.
22 The results for adenoma induction at 60 weeks of TCA exposure (i.e., ~2-fold increased
23 incidences and 2- to 3-fold increases in multiplicity with 10-fold increases in TCA dose) are
24 similar to the ~2-fold increase in liver weight gain resulting from 10-fold differences in dose
25 reported at 4-weeks of exposure. For PCO activity there was a -30% increase in PCO activity
26 from control at 0.05 g/L TCA. A 10-fold increase in TCA exposure concentration (0.5 g/L)
27 resulted in an additional ~5-fold increase in PCO activity. However, another 10-fold increase in
28 TCA concentration (0.5 vs. 5 g/L) resulted in a 3-fold increase in PCO activity. The 100-fold
29 increase in TCA dose (0.05 vs. 5 g/L TCA) was correlated with a 14-fold increase in PCO
30 activity. For 15 weeks of TCA exposure there was no difference in 0.05 and control PCO
31 activity and only a 30% difference between the 0.05 and 0.5 g/L TCA exposures. There was a
32 7-fold difference in PCO activity between the 0.5 and 5.0 g/L TCA exposure concentrations.
33 The increases in PCO activity and liver weight data at 15-weeks did not fit the magnitude of
34 increases in tumor multiplicity or incidence data at 60 weeks as well as did the 4-week data.
35 However, the TCA-induced increase in tumors at 60 weeks (especially adenomas) seemed to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-188 DRAFT—DO NOT CITE OR QUOTE
-------
1 correlate more closely with the magnitude of liver weight increase than for PCO activity at both
2 4 and 15 weeks.
3 In regard to Studies #1 and #2 there are consistent periods of study for percent liver/body
4 weight with the consistency of the control values being a large factor in the magnitude of TCA-
5 induced liver weight increases. As discussed above, there were differences in the magnitude of
6 percent liver/body weight increase at the same concentration between the two studies (e.g., a
7 1.47-fold of control percent liver/body weight in the 5 g/L TCA exposed group in Study #1 and
8 1.60-fold of control in Study #2 at 15 weeks). For the two studies that had extended durations of
9 exposure (Studies #2 and #3) the earliest time period for comparison of percent liver/body
10 weight is 26 weeks (Study #3) and 30 weeks (Study #2). If those data sets (26 weeks for
11 Study #3 and 30 weeks for Study #2) are combined, 0.05, 05, and 4.5 g/L TCA gives a percent
12 liver body/weight increase of 1.07-, 1.18-, and 1.40-fold over concurrent control levels. Using
13 this parameter, there appears to be a generally consistent pattern as that reported for Study #1 at
14 weeks 4 and 15. Generally, a 10-fold increase in TCA exposure concentration resulted in
15 ~2.5-fold increased in additional liver weight observed at -30 weeks of exposure which
16 correlated more closely with adenoma induction at 60 weeks than did changes in PCO activity.
17 A similar comparison between Studies of longer duration (Studies #2 and #3) could not be made
18 for PCO activity as data were not reported for Study #3.
19 For 104-week studies of TCA-tumor induction (Studies #2 and #3) the lower TCA
20 exposure levels (0.05 and 0.5 g/L TCA) were assayed in a separate experiment and by a separate
21 laboratory than the high dose (5.0 g/L TCA) and most importantly in larger more tumor prone
22 mice. The total lack of similarity in background levels of tumors in Study #2 and #3, the
23 differences in the number of animals included in the tumor analyses, and the low number of
24 animals examined in the tumor analysis at 104 weeks (less than 30 for the TCA treatment
25 groups) makes the determination of a dose-response TCA-induced liver tumor formation after
26 104-weeks of exposure problematic. The correlation of percent liver/body weight increases with
27 incidence and multiplicity of liver tumors in Study #1 and the similarity of dose-response for
28 early induction of percent liver/body weight gain between Study #1 suggest that there should be
29 a similarity in tumor response. However, as noted above, the 104-week studies had very
30 difference background rates of spontaneous tumors reported in the control mice between
31 Study #2 and #3.
32 Table E-12, below, shows the incidence and multiplicity data for Studies #2 and #3 along
33 with the control data for DeAngelo et al. (1999) for the same paradigm. It also provides an
34 estimate of the magnitude of increase in liver tumor induction by TCA treatments if the control
35 values from the DeAngelo et al. (1999) data set were used as the background tumor rate. As
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-189 DRAFT—DO NOT CITE OR QUOTE
-------
1 shown below, the background rates for Study #2 are more consistent with those of DeAngelo et
2 al. (1999). Whereas there was a 2:1 ratio of multiplicity for adenomas and adenomas and
3 carcinomas between 0.5 and 5.0 g/L TCA after 60 weeks of exposure, there was no difference in
4 any of the data (i.e., adenoma, carcinoma, and combinations of adenoma and carcinoma
5 incidence and multiplicity) for these exposure levels in Study #2 and #3 for 104 weeks. The
6 difference in the incidences and multiplicities for all tumors was 2-fold between the 0.05 and
7 0.5 g/L TCA exposure groups in Study #2. These results are consistent with the two highest
8 exposure levels reaching a plateau of response with a long enough duration of exposure (-90%
9 of animals having liver tumors) and with the 2-fold difference in liver tumor induction between
10 concentrations of TCA that differed by 10-fold, reported in Study #1.
11 If either the control values for Study #2 or the control values from DeAngelo et al. (1999)
12 were used for as the background rate of spontaneous liver tumor formation, the magnitude of
13 liver tumor induction by the 0.05 g/L TCA over control levels differs dramatically from that
14 reported as control tumor rates in Study #3. To put the 64% incidence data for carcinomas and
15 adenomas reported in DeAngelo et al. (2008) for the control group of Study #3 in context, other
16 studies cited in this review for B6C3F1 mice show a much lower incidence in liver tumors in
17 that: (1) the National Cancer Institute (NCI, 1976) study of TCE reports a colony control level of
18 6.5% for vehicle and 7.1% incidence of hepatocellular carcinomas for untreated male B6C3F1
19 mice (n = 70-77) at 78 weeks, (2) Herren-Freund et al. (1987) report a 9% incidence of
20 adenomas in control male B6C3F1 mice with a multiplicity of 0.09 ± 0.06 and no carcinomas
21 (n = 22) at 61 weeks, (3) NTP (1990) report an incidence of 14.6% adenomas and 16.6%
22 carcinomas in male B6C3F1 mice after 103 weeks (n = 48), and (4) Maltoni et al. (1986) report
23 that B6C3F1 male mice from the "NCI source" had a 1.1% incidence of "hepatoma" (carcinomas
24 and adenomas) and those from "Charles River Co." had a 18.9% incidence of "hepatoma" during
25 the entire lifetime of the mice (n = 90 per group). The importance of examining an adequate
26 number of control or treated animals before confidence can be placed in those results in
27 illustrated by Anna et al. (1994) in which at 76 weeks 3/10 control male B6C3F1 mice that were
28 untreated and 2/10 control animals given corn oil were reported to have adenomas but from 76 to
29 134 weeks, 4/32 mice were reported to have adenomas (multiplicity of 0.13 ± 0.06) and
30 4/32 mice were reported to have carcinomas (multiplicity of 0.12 ± 0.06).
31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-190 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-12. TCA-induced increases in liver tumor occurrence after 104 wks (Studies #2 and #3)
to
O k^j
o ^
"I
I
§
***.
S'
1
TO'
w g
I
Dose TCA
Adenomas
Incidence
Multiplicity
Carcinomas
Incidence
Multiplicity
Adenomas or carcinomas
Incidence
Multiplicity
Study #3
1.5g/LHAC(H20?)
0.05 g/L TCA
0.5 g/L TCA
21%
23%
(1.1 -fold)
51%
(2.4-fold)
0.21
0.34
(1.6-fold)
0.78
(3.7-fold)
55%
40%
(0.7-fold)
78%
(1.4-fold)
0.74
0.71
(1.0-fold)
1.46
(2.0-fold)
64%
57%
(0.9-fold)
87%
(1.4-fold)
0.93
1.11
(1.2-fold)
2.14
(2.3-fold)
Study #2
2.0g/LNaCl(HAC?)
4.5 g/L TCA
0%
59%
(?)
0
0.61
(?)
12%
78%
(6.5-fold)
0.20
1.50
(7.5-fold)
12%
89%
(7.4-fold)
0.20
2.14
(11 -fold)
DeAngelo et al., 1999
H2O
0.05 g/TCA (S #3)
0. 5 g/L TCA (S #3)
5.0g/LTCA(S#2)
10%
(2.3-fold)
(5.1 -fold)
(5.9-fold)
0.12
(2.8-fold)
(6.5-fold)
(6.5-fold)
26%
(1.5-fold)
(3.0-fold)
(3.0-fold)
0.28
(2.5-fold)
(5.2-fold)
(5.4-fold)
H I
O >
HH Oq
H TO
H2O = water.
O
H
W
-------
1 Using concurrent control values reported in Study #3, there is no increase in incidence of
2 multiplicity of adenomas and carcinomas for the 0.05 g/L exposure group. However, compared
3 to either the control data from DeAngelo et al. (1999) or the control data from Study #3, there is
4 a -2-3- or ~5-fold increased in incidence or multiplicity of liver tumors, respectively. Thus,
5 trying to determine a correspondence with either liver weight increases or increases in PCO
6 activity at earlier time points will be depend on the confidence placed in the concurrent control
7 data reported in Study #3 in the 104 week studies. As noted previously, the use of larger tumor
8 prone mice in Study #3 limits its usefulness to determine the dose-response for TCA.
9 The authors provide a regression analysis for "tumors/animal" or multiplicity as a percent
10 of control values and PCO activity for the 60-week and 104-week data. Whether adenomas and
11 carcinomas combined or individual tumor type were used was not stated. Also comparing PCO
12 activity at the end of the experiments, when there was already a significant tumor response rather
13 than at earlier time points, may not be useful as an indicator of PCO activity as a key event in
14 tumorigenesis. A regression analysis of these data are difficult to interpret because of the dose
15 spacing of these experiments as the control and 5 g/L exposure levels will basically determine
16 the shape of the dose-response curve. The 0.05 and 0.5 g/L exposure groups in the regression
17 were so close to the control value in comparison to the 5 g/L exposure, that the dose response
18 will appear linear between control and the 5.0 g/L value with the two lowest doses not affecting
19 the slope of the line (i.e., "leveraging" the regression). The value of this analysis is limited by
20 (1) the use of tumor prone larger mice in Study #3 that had large background rates of tumors
21 which make inappropriate the apparent combination of results from Studies #2 and #3 for the
22 multiplicity as percentages of control values (2) the low and varying number of animals analyzed
23 for PCO values and the variability in PCO control values (3) the appropriateness of using PCO
24 values from later time points, and (4) the dose-spacing of the experiment.
25 Similarly, the authors report a regression analysis that compares "percent of
26 hepatocellular neoplasia" which again is indicated by tumor multiplicity with TCA dose as
27 represented by mg/kg/d. This regression analysis also is of limited value for the same reasons as
28 that for PCO with added uncertainty as the exposure concentrations in drinking water have been
29 converted to an internal dose and each study gave different levels of drinking water with one
30 study showing a reduction of drinking water at the 5 g/L level. The authors attempt to identify a
31 NOEL for tumorigenicity using tumor multiplicity and TCA dose. However, it is not an
32 appropriate descriptor for these data, especially given that "statistical significance" of the tumor
33 response is the determinant of the conclusions regarding a dose in which there is no TCA-
34 induced effect. Only the 60-week experiment (i.e., Study #1) is useful for the determination of
35 tumor dose-response due to the issues related to appropriateness of control in Study #3. A power
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-192 DRAFT—DO NOT CITE OR QUOTE
-------
1 calculation of the 60-week study shows that the type II error, which should be >50% and thus,
2 greater than the chances of "flipping a coin," was 41 and 71% for incidence and 7 and 15% for
3 multiplicity of adenomas for the 0.05 and 0.5 g/L TCA exposure groups. For the combination of
4 adenomas and carcinomas, the power was 8 and 92% for incidence and 6 and 56% for
5 multiplicity at 0.05 and 0.5 g/L TCA exposure. Therefore, the designed experiment could accept
6 a false null hypothesis, especially in terms of tumor multiplicity, at the lower exposure doses and
7 erroneously conclude that there is no response due to TCA treatment.
8
9 E.2.3.2.14. DeAngelo et al, 1997. The design of this study appears to be similar to that of
10 DeAngelo et al. (2008) but to have been conducted in F344 rats. 28-30 day old rats that were
11 reported to be of similar weights were exposed to 2.0 g/L NaCl, 0.05, 0.5, or 5.0 g/L TCA in
12 drinking water for 104 weeks. There were groups of animals sacrificed at 15, 30, 45 and
13 60 weeks (n = 6) for PCO analysis. There were 23, 24, 19, and 22, animals reported to be
14 examined at terminal sacrifice at 104 weeks and 23, 24, 20, and 22 animals reported to be used in
15 the liver tumor analysis reported by the authors for the control, 0.05, 0.5, and 5.0 g/L treatment
16 groups, respectively. Complete pathological exams were reported to be performed for all tissues
17 from animals in the high dose TCA group at 104 weeks. No indication is given as to whether a
18 complete necropsy and pathological exam was performed for controls at terminal sacrifice.
19 Tritiated thymidine was reported to be administered at interim sacrifices five days prior to
20 sacrifice and to be examined with autoradiography. The 5 g/L TCA treatment group was reported
21 to have a reduction in growth to 89.3% of controls.
22 For water consumption TCA versus reported to slightly decrease water consumption at all
23 doses with a 7, 8, and 4% decrease in water consumption reported for 0.05, 0.5 and 5.0 g/L TCA,
24 respectively. Body weight was decreased by 5.0 g/L TCA dose only through 78 weeks of
25 exposure to 89.3% of the control value. All of the percent liver/body weight ratios were reported
26 to be slightly decreased (1-4%) by all of the exposure concentrations of TCA but the data shown
27 does not indicate if the liver weight data were taken at interim sacrifice times and appears to be
28 only for animals at terminal sacrifice of 104 weeks.
29 No data were shown for hepatocyte proliferation but the authors reported no TCA
30 treatment effects. For PCO there was a 2.3-fold difference between control values between the
31 15-week and 104-week data. For the 0.05 and 0.5 g/L TCA treatment groups there was not a
32 statistically significant difference reported between control and treated group PCO levels. At
33 15 weeks the PCO activity was reduced by 55%, increased to 1.02-fold, and increased 2.12-fold
34 of control for 0.05, 0.5 and 5.0 g/L TCA exposures, respectively. For the 30 week exposure
35 groups, the 0.05 and 0.5 g/L TCA groups were reported to have PCO levels within 5% of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-193 DRAFT—DO NOT CITE OR QUOTE
-------
1 control level. However, for the 5.0 g/L TCA treatment groups there was ~2-fold of control PCO
2 activity at the 15, 30, 45 and 60 weeks and at 104 weeks there was a 4-fold of control PCO
3 activity. Of note is that the control PCO value was lowest at 104 weeks while the TCA treatment
4 group was similar to interim values.
5 For analysis of liver tumors, there were 20-24 animals examined in each group. Unlike
6 the study of DeAngelo et al. (2008), it appeared that most of the animals that were sacrificed at
7 104 weeks were used in the tumor analysis without addition of "extra" animals or deletion of
8 animal data. The incidence of adenomas was reported to be 4.4, 4.2, 15, and 4.6% and the
9 incidence of hepatocellular carcinomas was reported to be 0, 0, 0, and 4.6% for the control, 0.05,
10 0.5, and 5.0 g/L TCA exposure groups. The multiplicity or tumors/animal was reported to be
11 0.04, 0.08, 0.15, and 0.05 for adenomas and 0, 0, 0, and 0.05 for carcinomas for the control, 0.05,
12 0.5, and 5.0 g/L TCA exposure groups. Although there was an increase in the incidence of
13 adenomas at 0.5 g/L and an increase in carcinomas at 5.0 g/L TCA, they were not reported to be
14 statistically significant by the authors. Neither were the increase in adenoma multiplicity at the
15 0.05 and 0.5 g/L exposures. However, using such a low number of animals per treatment group
16 (n = 20-24) limits the ability of this study to determine a statistically significant increase in tumor
17 response and to be able to determine that there was no treatment-related effect. A power
18 calculation of the study shows that the type II error, which should be >50% and thus, greater than
19 the chances of "flipping a coin," was less than 6% for incidence and multiplicity of tumors at all
20 exposure DCA concentrations with the exception of the incidence of adenomas for 0.5 g/L
21 treatment group (58.7%). Therefore, the designed experiment could accept a false null
22 hypothesis, especially in terms of tumor multiplicity, at the lower exposure doses and erroneously
23 conclude that there is no response due to TCA treatment. Thus, while suggesting a lower
24 response than for mice for TCA-induced liver tumors, the study is inconclusive for determination
25 of whether TCA induces a carcinogenic response in the liver of rats. The experimental design is
26 such that extrahepatic carcinogenicity of TCA in the male rat cannot be determined.
27
28 E.2.3.2.15. DeAngelo et al, 1996. In this study, 28-day-old male F344 rats were given
29 drinking water containing DCA at concentrations of 0, 0.05, 0.5, or 5.0 g/L with another group
30 was provided water containing 2.0 g/L NaCl for 100 weeks. This experiment modified its
31 exposure protocol due to toxicity (peripheral neuropathy) such that the 5.0 g/L group was lowered
32 to 2.5 g/L at 9 weeks and then 2.0 g/L at 23 weeks and finally to 1.0 g/L at 52 weeks. When the
33 neuropathy did not reverse or diminish, the animals were sacrificed at 60 weeks and excluded
34 from the results. Based on measured water intake in the 0, 0.05, and 0.5 g/L groups, the time-
35 weighted average doses were reported to be 0, 3.6, and 40.2 mg/kg/d respectively. This
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-194 DRAFT—DO NOT CITE OR QUOTE
-------
1 experiment was conducted at a U.S. EPA laboratory in Cincinnati and the controls for this group
2 were given 2.0 g/L NaCl (Study #1). In a second study rats were given either deionized water or
3 2.5 g/L DCA, which was also lowered to 1.5 g/L at 8 weeks and to 1.0 g/L at 26 weeks of
4 exposure (Study #2).
5 Although 23 animals were reported to be sacrificed at terminal sacrifice that had been
6 given 2 g/L NaCl, the number of animals reported to be examined in this group for hepatocellular
7 lesions was 3. The incidence data for this group for adenomas was 4.4% so this is obviously a
8 typographical error. The number of rats included in the water controls for tumor analysis was
9 reported to be 33 which was the same number as those at final sacrifice. The number of animals
10 at final sacrifice was reported to be 23 for 2 g/L NaCl, 21 for 0.05 g/L DCA, 23 for 0.5 g/L DCA
11 in experiment #1 and 33 for deionized water and 28 for the initial dose of 2.5 g/L DCA in
12 experiment #2. Although these were of the same strain, the initial body weight was 59.1 g versus
13 76 g for the 2.0 g/L control group versus deionized water group. The treatment groups in both
14 studies were similar to the deionized water group. The percent liver/body weights were greater
15 (4.4 vs. 3.7% in the NaCl vs. deionized water control groups (-20%). The number of
16 unscheduled deaths was greater in Study #2 (22%) than in Study #1 (12%). Interim sacrifice
17 periods were conducted.
18 As with the DeAngelo et al. (2008) study in mice, the number of animals reported at final
19 sacrifice was not the same as the number examined for liver tumors in Study #1 (5 more animals
20 examined than sacrificed at the 0.05 g/L DCA and 6 more animals examined than sacrificed at the
21 0.5 g/L DCA exposure groups) with n = 23,n = 26, and n = 29 for the 2 g/L NaCl, 0.05 g/L DCA
22 and 0.5 g/L DCA groups utilized in the tumor analysis. For Study #2 the same number of rats
23 was reported to be sacrificed as examined. The source of the extra animals for tumor analysis in
24 Study #1, whether from interim sacrifice or unscheduled deaths, was not given by the authors and
25 is unknown. Carcinomas prevalence data were not reported for the control group or 0.05 g/L
26 DCA group in Study #1 and multiplicity data were not reported to the control group, or 0.05 g/L
27 DCA group. Multiplicity was not reported for adenomas in the 0.05 g/L DCA group in Study #1.
28 There was a lack of hepatocyte DNA synthesis and necrosis reported at any dose group
29 carried out to final sacrifice at 100 weeks. The authors reported that the incidence of adenomas to
30 be 4.4% in 2 g/L NaCl control, 0 in 0.05 g/L DCA, and 17.2% in the 0.5 g/L DCA exposure
31 groups. For carcinomas no data were reported for the control or 0.05 g/L DCA group but an
32 incidence of 10.3% was reported for the 0.5 g/L DCA group. The authors reported increased
33 hepatocellular adenomas and carcinomas in male F344 rats although not data were reported for
34 carcinomas in the control and 0.05 g/L exposure groups. They reported that for 0.5 g/L DCA,
35 24.1 versus 4.4% adenomas and carcinomas combined (Study #1) and 28.6 versus 3.0%
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-195 DRAFT—DO NOT CITE OR QUOTE
-------
1 (Study #2) at what was initially 2.5 g/L DCA but continuously reduced). Tumor multiplicity was
2 significantly was reported to be increased in the 0.5 g/L DCA group (0.04 adenomas and
3 carcinomas/animal in control vs. 0.31 in 0.5 g/L DCA in Study #1 and 0.03 in control vs. 0.36 in
4 what was initially 2.5 g/L DCA in Study #2). The issues of use of a small number of animals,
5 additional animals for tumor analysis in Study #1, and most of all the lack of a consistent dose for
6 the 2.5 g/L animals in Study #2, are obvious limitations for establishment of a dose-response for
7 DCA in rats.
8
9 E.2.3.2.16. Richmond et al, 1995. This study was conducted by the same authors as DeAngelo
10 et al. (1996) and appears to report results for the same data set for the 2 g/L NaCl control,
11 0.05 g/L DCA and 0.5 g/L DCA exposed groups. Of note is that while DeAngelo et al. (1996)
12 refer to the 28-day old rats as "weanlings" the same aged rats are referred to as "adults" in this
13 study. Male Fischer 344 rats were administered time-weighted average concentrations of 0, 0.05,
14 0.5, or 2.4 g/L DCA in drinking water. Concentrations were kept constant but due to hind-limb
15 paralysis all 2.4 g/L DCA exposed rats had been sacrificed by 60 weeks of exposure. In the
16 104-week sacrifice time, there were 23 rats reported to be analyzed for incidence of hepatocellular
17 adenomas and carcinomas in the control group, 26 rats in the 0.05 g/L DCA group and 29 rats in
18 the 0.5 g/L DCA exposed group. This is the same number of animals included in the tumor
19 analysis reported in DeAngelo et al. (1996). Tumor multiplicity was not given. Richmond et al.
20 (1995) reported that there was a 4% incidence of adenomas reported in the 2.0 g/L NaCl control
21 animals, 0% at 0.05 g/L DCA, and 21% in the 0.5 DCA group at 104 weeks. These figures are
22 similar to those reported by DeAngelo et al. (1996) for the same data set with the exception of a
23 17.2% incidence of adenomas reported for the 0.5 g/L DCA group. There were no hepatocellular
24 carcinomas reported in the control or 0.05 g/L exposure groups but a 10% incidence reported in
25 the 0.5 g/L DCA exposure group at 104 weeks of exposure. While carcinomas were not reported
26 by DeAngelo et al. (1996) for the control and 0.05 g/L groups they are assumed to be zero in the
27 summary data for carcinomas and adenomas combined. The 10% incidence at 0.5 g/L DCA is
28 similar to the 10.4% incidence reported for this group by DeAngelo et al. (1996). At 60 weeks at
29 2.4 g/L DCA, the incidence of hepatocellular adenoma was reported to be 26% and hepatocellular
30 carcinoma to be 4%. This is not similar to the values reported by DeAngelo for 2.5 g/L DCA that
31 was continuously decreased so that the estimated final concentration was 1.6 g/L DCA for
32 100 weeks for those animals, the incidence of adenomas was reported by DeAngelo et al. (1996)
33 to be 10.7% and carcinomas 21.4%, probably more a reflect of longer exposure time allowing for
34 adenoma to carcinoma progression. The authors did not report any of the results of DCA-induced
35 increases of adenomas and carcinomas to be statistically significant. As it appears the same data
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-196 DRAFT—DO NOT CITE OR QUOTE
-------
1 set was used for the 2.g/L NaCl control, 0.05 g/L DCA and 0.5 g/L DCA exposure groups as was
2 reported in DeAngelo et al. (1996), the same issues arise as regarding the differences in numbers
3 of animals were included in tumor analysis than were reported to have been present at final
4 sacrifice. As stated previously for the DeAngelo et al. (1997) study of TCA in rats, the use of
5 small numbers of rats limits the detection of and ability to determine whether there was no
6 treatment-related effects, especially at the low concentrations of DCA exposure.
7
8 E.2.4. Summaries and Comparisons Between Trichloroethylene (TCE), Dichloroacetic
9 Acid (DCA), and Trichloroacetic Acid (TCA) Studies
10 There are a number of studies to TCE that have reported effects on the liver. However,
11 the study of this compound is difficult as its concentration does not remain stable in drinking
12 water, some studies have been carried out using TCE with small quantities of a carcinogenic
13 stabilizing agent, some studies have been carried out in whole body inhalation chambers that
14 resulted in additional oral administration and for which individual animal data were not recorded
15 throughout the experiment, and the results of gavage studies have been limited by gavage related
16 deaths and vehicle effects. In addition some studies have been conducted using the i.p. route of
17 administration, which results in route-related toxicity and inflammation. For many studies, liver
18 effects consisted of measured increases in liver weight with little or no description of attendant
19 histological changes induced by TCE treatment. A number of studies were conducted at a few
20 relatively high doses with attendant effects on body weight, indicative of systemic toxicity and
21 affecting TCE-induced liver weight gain. Although, many studies have been performed in male
22 mice, the inhalation studies of Kjellstrand et al. indicate that male mice, regardless of strain
23 appear to have a greater variability in response, as measured by TCE-induced liver weight gain,
24 and susceptibility to TCE-induced decreases in body weight than female mice. However, the
25 body of the TCE literature is consistent in identifying the liver as a target of TCE-induced affects
26 and with the most commonly reported change to be a dose-related TCE-induced increase in liver
27 weight in multiple species, strains, and genders from both inhalation and oral routes of exposure.
28 The following sections will not only summarize results for studies of TCE reported in
29 Sections E.2. l-E.2.2, but provide comparison of studies of either TCA or DCA that have used
30 similar paradigms or investigated similar parameters described in Sections E.2.3.1 and E.2.3.2. A
31 synopsis of the results from studies of CH and in comparison with TCE results is presented in
32 Section E.2.5. While the study of Bull et al. (2002), described in Section E.2.2.21, presents data
33 for combinations of DCA or TCA exposure for comparisons of tumor phenotype with those
34 induced by TCE, the examination of coexposure studies of TCE metabolites in rodents that are
35 also exposed to a number of other carcinogens, and descriptions of the toxicity data for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-197 DRAFT—DO NOT CITE OR QUOTE
-------
1 brominated haloacetates that also occur with TCE in the environment, are presented in Section
2 E.4.3.3.
3
4 E.2.4.1. Summary of Results For Short-term Effects of Trichloroethylene (TCE)
5 In regard to early changes in DNA synthesis, the data for TCE is very limited. The study
6 by Mirsalis et al. (1989) used an in vivo-in vitro hepatocyte DNA repair and S-phase DNA
7 synthesis in primary hepatocytes from male Fischer-344 rats (180-300 g) and male and female
8 B6C3F1 mice (20-29 g for male mice and 18-25 g female mice) administered TCE by gavage in
9 corn oil. They reported negative results 2-12 hours after treatment from 50-1,000 mg/kg TCE in
10 rats and mice (male and female) for unscheduled DNA synthesis and repair using 3 animals per
11 group. After 24 and 48 hours of 200 or 1,000 mg/kg TCE in male mice (n = 3) and after 48 hours
12 of 200 (n = 3) or 1,000 (n = 4) mg/kg TCE in female mice, similar values of 0.30 to 0.69% of
13 hepatocytes were reported as undergoing DNA synthesis in those hepatocytes in primary culture
14 with only the 1,000 mg/kg TCE dose in male mice at 48 hours giving a result considered to be
15 positive (-2.2%). No statistical analyses were performed on these measurements, which were
16 obviously limited by both the number of animals examined and the relevance of the paradigm.
17 TCE-induced increases in liver weight have been reported to occur quickly. The
18 inhalation study of Okino et al. (1991) in male rats demonstrates that liver weight and metabolism
19 were increased with as little as 8 hours of TCE exposure (500 and 2,000 ppm) and as early as
20 22 hours after cessation of such exposures with little concurrent hepatic necrosis. Laughter
21 reported increase liver weight in SV129 mice in their 3-days study (see below). Tao et al. (2000)
22 reported a 1.26-fold of control percent liver/body weight in female B6C3Flmice fed 1,000 mg/kg
23 TCE in corn oil for 5 days. Elcombe et al. (1985) and Dees and Travis (1993) reported gavage
24 results in mice and rats after 10 days exposure to TCE which showed TCE-induced increases in
25 liver weight (see below for more detail on dose-response). Tucker et al. (1982) reported that
26 14 days of exposure to 24 mg/kg and 240 mg/kg TCE via gavage to induce a dose-related increase
27 in liver weight in male CD-I mice but did not show the data.
28 TCE-induced increases in percent liver/body weight ratios have been studied most
29 extensively in B6C3F1 and Swiss mice. Both strains have been shown to have a TCE-induced
30 increase in liver tumors from long-term exposure as well (see Section E.2.4.2, below). A number
31 of studies have provided dose-response information for TCE-induced increases in liver weight
32 from 10 days to 13 weeks of exposure in mice. Most studies have reported that the magnitude of
33 increase in TCE exposure concentration is similar to the magnitude increase of percent liver/body
34 weight increase. For example a 2-fold increase in TCE exposure has often resulted in a 2-fold
35 increase in the percent change in liver/body weight over control (i.e., 500 mg/kg TCE induces a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-198 DRAFT—DO NOT CITE OR QUOTE
-------
1 20% increase in liver weight and 1,000 mg/kg TCE induces a 50% increase in liver weight as
2 reported by Elcombe et al., 1985). The range in which this relationship is valid has been reported
3 to vary from 100 mg/kg TCE at 10 days (Dees and Travis, 1993) to 1,600 mg/kg (Buben and
4 O'Flaherty, 1985) at 6 weeks and up to 1,500 mg/kg TCE for 13 weeks (NTP, 1990). The
5 consistency in the relationship between magnitude of liver weight increase and TCE exposure
6 concentration has been reported for both genders of mice, across oral and inhalation routes of
7 exposure, and across differing strains of mice tested. For rats, there are fewer studies with fewer
8 exposure levels tested, but both Berman et al. (1995) and Melnick et al. (1987) report that short-
9 term TCE exposures from 150 mg/kg to -2,000 mg/kg induced percent liver/body weight that
10 increased proportionally with the magnitude of TCE exposure concentration.
11 Dependence of PPARa activation for TCE-liver weight gain has been investigated in
12 PPARa null mice by both Nakajima et al. (2000) and Laughter et al. (2004). After 2 weeks of
13 750 mg/kg TCE exposure to carefully matched SV129 wild-type or PPARa-null male and female
14 mice (n = 6 group), there was a reported 1.50-fold of control in wild-type and 1.26-fold of control
15 percent liver/body weight in PPARa-null male mice by Nakajima et al. (2000). For female mice,
16 there was ~1.25-fold of control percent liver/body weight ratios for both wild-type and PPARa-
17 null mice. Thus, TCE-induced liver weight gain was not dependent on a functional PPARa
18 receptor in female mice and some portion of it may have been in male mice. Both wild-type male
19 and female mice were reported to have similar increases in the number of peroxisome in the
20 pericentral area of the liver and TCE exposure and, although increased 2-fold, were still only -4%
21 of cytoplasmic volume. Female wild-type mice were reported to have less TCE-induced
22 elevation of very long chain acyl-CoA synthetase, D-type peroxisomal bifunctional protein,
23 mitochondrial trifunctional protein a subunits a and P, and cytochrome P450 4A1 than males
24 mice, even though peroxisomal volume was similarly elevated in male and female mice. The
25 induction of PPARa protein by TCE treatment was also reported to be slightly less in female than
26 male wild-type mice (2.17- vs. 1.44-fold of control, respectively).
27 Laughter et al. (2004) also studied SV129 wild-type and PPARa-null male mice treated
28 with 3 daily doses of TCE in 0.1% methyl cellulose for either 3 days (1,500 mg/kg TCE) or
29 3 weeks (0, 10, 50, 125, 500, 1,000, or 1,500 mg/kg TCE 5 days a week). However, not only is
30 the paradigm not comparable to other gavage paradigms, but no initial or final body weights of
31 the mice were reported and thus, the influence of differences in initial body weight on percent
32 liver/body weight determinations could not be ascertained. In the 3-day study, while control
33 wild-type and PPARa-null mice were reported to have similar percent liver/body weight ratios
34 (-4.5%), at the end of the 3-week experiment the percent liver/body weight ratios were reported
35 to be increased in the PPARa-null male mice (5.1%). TCE treatment for 3 days was reported to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-199 DRAFT—DO NOT CITE OR QUOTE
-------
1 increase the percent liver/body weight ratio 1.4-fold of control in the wild-type mice and
2 1.07-fold of control in the null mice. In the 3-week study, wild-type mice exposed to various
3 concentrations of TCE had percent liver/body weights that were reported to be within -2% of
4 control values except for the 1,000 mg/kg and 1,500 mg/kg groups (-1.18- and 1.30-fold of
5 control levels, respectively). For the PPARa-null mice the variability in percent liver/body
6 weight was reported to be greater than that of the wild-type mice in most of the groups and the
7 baseline level of percent liver/body weight ratio also 1.16-fold greater. TCE exposure was
8 apparently more toxic in the null mice with death at the 1,500 mg/kg TCE exposure level
9 resulting in the prevention of recording of percent liver/body weights. At 1,000 mg/kg TCE
10 exposure level there was a reported 1.10-fold of control percent liver/body weight in the PPARa-
11 null mice. None of the increases in percent liver/body weight in the null mice were reported to be
12 statistically significant by Laughter et al. (2004). However, the statistical power of the study was
13 limited due to low numbers of animals and increased variability in the null mice groups. The
14 percent liver/body weight after TCE treatment that was reported in this study was actually greater
15 in the null mice than the wild-type male mice at the 1,000 mg/kg TCE exposure level
16 (5.6% ± 0.4% vs. 5.2% ± 0.5%, for null and wild-type mice, respectively). At 1-weeks and at
17 3-weeks, TCE appeared to induce increases in liver weight in PPARa-null mice, although not
18 reaching statistical significance in this study. At a 1,000 mg/kg TCE exposure for 3 weeks
19 percent liver/body weights were reported to be 1.18-fold of control in wild-type and 1.10-fold of
20 control in null mice. Although the experiments in Laughter et al. for DC A and TCA were not
21 conducted using the same paradigm, the TCE-induced increase in percent liver/body weight more
22 closely resembled the dose-response pattern for DCA than for DCA wild-type SV129 and
23 PPARa-null mice.
24 Many studies have used cyanide-insensitive PCO as a surrogate for peroxisome
25 proliferation. Of note is that several studies have shown that this activity is not correlated with
26 the volume or number of peroxisomes that are increased as a result of exposure to TCE or it
27 metabolites (Nakajima et al.,2000; Elcombe et al., 1985: Nelson et al., 1989). This activity
28 appears to be highly variable both as a baseline measure and in response to chemical exposures.
29 Laughter et al. (2004) presented data showing that WY-14,643 induced increases in PCO activity
30 varied up to 6-fold between experiments in wild-type mice. They also showed that PCO activity,
31 in some instances, was up to 6-fold of wild-type mice values in untreated PPARa-null mice.
32 Parrish et al. (1996) noted that control values between experiments varied as much as a factor of
33 2-fold for PCO activity and thus, their data were presented as percent of concurrent controls.
34 Goldsworthy and Popp (1987) reported that 1,000 mg/kg TCE induced a 6.25-fold of control PCO
35 activity in B6C3F1 mice in two 10-day experiments. However, for F344 rats, the increases over
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-200 DRAFT—DO NOT CITE OR QUOTE
-------
1 control between two experiments conducted at the same dose were reported to vary by >30%.
2 Finally, Melnick et al. (1987) have reported that corn oil administration alone can elevate PCO
3 activity as well as catalase activity.
4 For TCE there are two key 10-days studies (Elcombe et al., 1985; Dees and Travis, 1993)
5 that examine the effects of short-term exposure in mice and rats via gavage exposure and attempt
6 to determine the nature of the dose response in a range of exposure concentrations that include
7 levels below which there is concurrent decreased body weights. Although they have limitations,
8 they reported generally consistent results. In regard to liver weight in mice, gavage exposure to
9 TCE at concentrations ranging from 100 to 1,500 mg/kg TCE produced increases in liver/body
10 weight that was dose-related (Elcombe et al., 1985; Dees and Travis, 1993).
11 Elcombe et al. (1985) reported a small decrease in DNA content with TCE treatment
12 (consistent with hepatocellular hypertrophy) that was not dose-related, increased tritiated
13 thymidine incorporation in whole mouse liver DNA that was that was treatment but not dose-
14 related (i.e., a 2-, 2-, and 5-fold of control values in mice treated with 500, 1,000, and
15 1,500 mg/kg TCE), and slightly increased numbers of mitotic figures that were treatment but not
16 dose-related and not correlated with DNA synthesis as measured by thymidine incorporation.
17 Elcombe et al. (1985) reported an increase in peroxisome volume after TCE exposure that was
18 correlated with the magnitude of increase in peroxisomal-associated enzyme activity at the only
19 dose in which both were tested. Peroxisome increases after TCE treatment in mice livers were
20 identified as being pericentral in location. After TCE treatment, increased peroxisomal volumes
21 in B6C3F1 mice were reported to be not dose-related (i.e., there was little difference between 500
22 to 1,500 mg/kg TCE exposures). The TCE-induced increases in peroxisomal volumes were also
23 not correlated with the reported increases in thymidine incorporation or mitotic activity in mice.
24 Neither TCE-induction of peroxisomes or hepatocellular proliferation, as measured by either
25 mitotic index or thymidine incorporation, was correlated with TCE-induced liver weight
26 increases. Elcombe et al. (1985) only measured PCO activity in a subset of B6C3F1 mice at the
27 1,000 mg/kg TCE exposure level for 10 days of exposure and reported an 8-fold of control PCO
28 activity and a 1.5-fold of control catalase activity. This result was similar to that of Goldsworthy
29 and Popp (1987) who reported 6.25-fold of control PCO activity in male B6C3F1 mice exposed
30 to 1,000 mg/kg/d TCE for 10 days in two separate experiments.
31 Similar to Elcombe et al., who reported no difference in response between 500 and
32 1,000 mg/kg TCE treatments, Dees and Travis (1993) reported that incorporation of tritiated
33 thymidine in DNA from mouse liver was elevated after TCE treatment and the mean peak level of
34 tritiated thymidine incorporation occurred at 250 mg/kg TCE treatment level remaining constant
35 for the 500 and 1,000 mg/kg treated groups. Dees and Travis (1993) specifically report that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-201 DRAFT—DO NOT CITE OR QUOTE
-------
1 mitotic figures, although very rare, were more frequently observed after TCE treatment, found
2 most often in the intermediate zone, and found in cells resembling mature hepatocytes. They
3 reported that there was little tritiated thymidine incorporation in areas near the bile duct epithelia
4 or close to the portal triad in liver sections from both male and female mice. They also reported
5 no evidence of increased lipofuscin and that increased apoptoses from TCE exposure "did not
6 appear to be in proportion to the applied TCE dose given to male or female mice" (i.e., the mean
7 number of apopotosis 0, 0, 0, 1 and 8 for control, 100, 250, 500, and 1,000 mg/kg TCE treated
8 groups, respectively). Both Elcombe et al. (1985) and Dees and Travis (1993) reported no
9 changes in apoptosis other than increased apoptosis only at a treatment level of 1,000 mg/kg TCE.
10 Elcombe et al. (1985) reported increased in percent liver/body weight after TCE treatment
11 in both the Osborne-Mendel and Alderly Park rat strain, although to a smaller extent than in mice.
12 For both strains, Elcombe et al. (1985) reported no TCE-induced changes in body weight at doses
13 ranging from 500 to 1,500 mg/kg. For male Osborne-Mendel rats administration of TCE in corn
14 oil gavage resulted in a 1.18-, 1.26-, and 1.30-fold of control percent liver/body weight at
15 500 mg/kd/day, 1,000 mg/kg/d, and 1,500 mg/kg/d exposures, respectively. For Alderly Park rats
16 those increases were 1.14-, 1.17-, and 1.17-fold of control at the same respective exposure levels
17 for 10 days of exposure. In regard to liver weight increases, Melnick et al. (1987) reported a
18 1.13- and 1.23-fold of control percent liver/body weight in male Fischer 344 rats fed 600 mg/kg/d
19 and 1,300 mg/kg/d TCE in capsules, respectively. There was no difference in the extent of TCE-
20 induced liver increase between the two lowest dosed group administered TCE in corn oil gavage
21 (-20% increase in percent liver/body weight at 600 mg/kd and 1,300 mg/kg TCE) for 14 days.
22 However, the magnitude of increases in percent liver/body weight in these groups was affected by
23 difference between control groups in liver weight although initial and final body weights appeared
24 to be similar. By either type of vehicle, Melnick et al. (1987) reported decreases in body weights
25 in rats treated with concentrations of TCE 2,200 mg/kg/d or greater for 14 days. Similarly, Nunes
26 et al. (2001) reported decreased body weight in S-D rats administered 2,000 mg/kg/d for 7 days in
27 corn oil. Melnick et al. (1987) reported that both exposures to either 600 or 1,300 mg/kg/d TCE
28 in capsules did not result in decreased body weight and caused less than minimal focal necrosis
29 randomly distributed in the liver. At 2,200 and 4,800 mg/kg TCE fed via capsule, Melnick et al.
30 (1987) reported that although there was decreased body weight in rats treated at these exposures,
31 there was little TCE-induced necrosis, and no evidence of inflammation, cellular hypertrophy or
32 edema with TCE exposure. Similarly, Berman et al. (1995) reported increases in liver weight
33 gain at doses as low as 50 mg/kg TCE, no necrosis up to doses of 1,500 mg/kg, and hepatocellular
34 hyper trophy only at the 1,500 mg/kg level in female Fischer 344 rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-202 DRAFT—DO NOT CITE OR QUOTE
-------
1 For rats, Elcombe et al. (1985) reported an increase over untreated rats of 1.13-fold of
2 control PCO activity in Alderly Park rats after 1,000 mg/kg/d TCE exposure for 10 days, while
3 Goldsworthy and Popp (1987) reported a 1.8- and 2.39-fold of control in male Fischer 344 rats at
4 the same exposure in two separate experiments. Melnick et al. (1987) reported PCO activity of
5 1.23- and 1.75-fold of control in male Fischer 344 rats fed 600 mg/kg/d and 1,300 mg/kg/d TCE
6 for 14 days in capsules. For rats treated by gavage with 600 mg/kg/d or 1,200 mg/kg d TCE corn
7 oil, they reported 1.16- and 1.29-fold of control values. However, control levels of PCO were
8 16% higher in corn oil controls than in untreated controls. In addition Melnick et al. (1987)
9 reported little catalase increases in rats fed TCE via capsules in food (less than 6% increase) but a
10 1.18- and 1.49-fold of control catalase activity in rats fed 600 mg/kg/d or 1,200 mg/kg/TCE via
11 corn oil gavage, indicative of a vehicle effect.
12 The data from Elcombe et al. (1985) included reports of TCE-induced pericentral
13 hypertrophy and eosinophilia for both rats and mice but with "fewer animals affected at lower
14 doses." In terms of glycogen deposition, Elcombe report "somewhat" less glycogen pericentrally
15 in the livers of rats treated with TCE at 1,500 mg/kg than controls with less marked changes at
16 lower doses restricted to fewer animals. They do not comment on changes in glycogen in mice.
17 Dees and Travis (1993) reported TCE-induced changes to "include an increase in eosinophilic
18 cytoplasmic staining of hepatocytes located near central veins, accompanied by loss of
19 cytoplasmic vacuolization." Since glycogen is removed using conventional tissue processing and
20 staining techniques, an increase in glycogen deposition would be expected to increase
21 vacuolization and thus, the report from Dees and Travis is consistent with less not more glycogen
22 deposition. Neither study produced a quantitative analysis of glycogen deposition changes from
23 TCE exposure. Although not explicitly discussing liver glycogen content or examining it
24 quantitatively in mice, these studies suggest that TCE-induced liver weight increases did not
25 appear to be due to glycogen deposition after 10 days of exposure and any decreases in glycogen
26 were not necessarily correlated with the magnitude of liver weight gain either.
27 For both rats and mice the data from Elcombe et al. (1985) showed that tritiated thymidine
28 incorporation in total liver DNA observed after TCE exposure did not correlate with mitotic index
29 activity in hepatocytes with both Elcombe et al. (1985) and Dees and Travis (1993) reporting a
30 small mitotic indexes and evidence of periportal hepatocellular hypertrophy from TCE exposure.
31 Neither mitotic index or tritiated thymidine incorporation data support a correlation with TCE-
32 induced liver weight increase in the mouse. If higher levels of hepatocyte replication had
33 occurred earlier, such levels were not sustained by 10 days of TCE exposure. Both Elcombe et al.
34 (1985) and Dees and Travis (1993) present data that represent "a snapshot in time" which does
35 not show whether increased cell proliferation may have happened at an earlier time point and then
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-203 DRAFT—DO NOT CITE OR QUOTE
-------
1 subsided by 10 days. These data suggest that increased tritiated thymidine levels were targeted to
2 mature hepatocytes and in areas of the liver where greater levels of polyploidization occur. Both
3 Elcombe et al. (1985) and Dees and Travis (1993) show that tritiated thymidine incorporation in
4 the liver was ~2-fold of controls between 250-1,000 mg/kg TCE, a result consistent with a
5 doubling of DNA. Thus, given the normally quiescent state of the liver, the magnitude of this
6 increase over control levels, even if a result of proliferation rather than polyploidization, would be
7 confined to a very small population of cells in the liver after 10 days of TCE exposure. Laughter
8 et al. (2004) reported that there was an increase in DNA synthesis after aqueous gavage exposure
9 to 500 and 1,000 mg/kg TCE given as 3 boluses a day for 3 weeks with BrdU given for the last
10 week of treatment. An examination of DNA synthesis in individual hepatocytes was reported to
11 show that 1 and 4.5% of hepatocytes had undergone DNA synthesis in the last week of treatment
12 for the 500 and 1,000 mg/kg doses, respectively. Both Elcombe et al. (1985) and Dees and Travis
13 (1993) show TCE-induced changes for several parameters at the lowest level tested without
14 toxicity and without evidence of regenerative hyperplasia or sustained hepatocellular
15 proliferation. In regards to susceptibility to liver cancer induction, the more susceptible
16 (B6C3F1) versus less susceptible (Alderly Park/Swiss) strains of mice to TCE-induced liver
17 tumors (Maltoni et al., 1988), the "less susceptible" strain was reported by Elcombe et al. (1985)
18 to have, a greater baseline level of liver weight/body weight ratio, a greater baseline level of
19 thymidine incorporation as well as greater responses for those endpoints due to TCE exposure.
20 However, both strains showed a hepatocarcinogenic response after TCE exposure, although there
21 are limitations regarding determination of the exact magnitude of response for these experiments
22 as previously discussed.
23
24 E.2.4.2. Summary of Results For Short-Term Effects of Dichloroacetic Acid (DCA) and
25 TrichloroaceticAcid(TCA): Comparisons With Trichloroethylene (TCE)
26 Short-term exposures from DCA and TCA have been studied either through gavage or in
27 drinking water. Palatability became an issue at the highest level of DCA tested in drinking water
28 experiments (5 g/L) which caused a significant reduction of drinking water intake in mice of 46 to
29 64% (Carter et al., 1995). Decreases in drinking water consumption have also been reported for a
30 range of concentrations of DCA and TCA from 0.05 g/L to 5.0 g/L, in both mice and rats, and
31 with generally the higher concentrations producing the highest decrease in drinking water (Carter
32 et al., 1995; Mather et al., 1990; DeAngelo et al., 1997, 1999, 2008). However, results within
33 studies (e.g., DeAngelo et al., 2008) and between studies have been reported to vary as to the
34 extent of the reduction in drinking water from the presence of TCA or DCA. Some drinking
35 water studies of DCA or TCA have not reported drinking water consumption as well. Therefore,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-204 DRAFT—DO NOT CITE OR QUOTE
-------
1 although in general DCA and TCA studies have do not include vehicle effects, such as corn oil,
2 they have been affected by differences in drinking water consumption not only changing the dose
3 received by the rodents and therefore, potentially the shape of the dose-response curve, but also
4 the effects of dehydration are potentially added to any chemically-related reported effects.
5 Studies have attempted to determine short-term effects on DNA by TCE and its
6 metabolites. Nelson and Bull (1988) administered TCE male Sprague Dawley rats and male
7 B6C3F1 mice measured the rate of DNA unwinding under alkaline conditions 4 hours later. For
8 rats there was a significantly increased rate of unwinding at the two highest dose and for mice
9 there was a significantly increased level of DNA unwinding at a lower dose. In this same study,
10 DCA was reported to be most potent in this assay with TCA being the lowest, while CH closely
11 approximated the dose-response curve of TCE in the rat. In the mouse the most potent metabolite
12 in the assay was reported to be TCA followed by DCA with CH considerably less potent. Nelson
13 and Bull (1988) and Nelson et al. (1989) have reported increases in single strand breaks after
14 DCA and TCA exposure. However, Styles et al. (1991) (for mice) and Chang et al. (1992) (for
15 mice and rats) did not. Austin et al. (1996) note that the alkaline unwinding assay, a variant of the
16 alkaline elution procedure, is noted for its variability and inconsistency depending on the
17 techniques used while performing the procedure. In regard to oxidative damage as measured by
18 TEARS for lipid peroxidation and 8-OHdG levels in DNA, increases appear to be small (less than
19 50% greater than control levels) and transient after DCA and TCA treatment in mice (see Section
20 E.3.4.2.3) with TCE results confounded by vehicle or route of administration effects.
21 Although there is no comparative data for TCE, the study of Styles et al. (1991) is
22 particularly useful for determining effects of TCA from 1 to 4 days of exposure in mice. Styles et
23 al. (1991) reported no change in "hepatic" DNA uptake of tritiated thymidine up to 36 hours, a
24 peak at 72 hours (~6-fold of control), and falling levels by 96 hours (~4-fold of controls) after
25 500 mg/kg TCA gavage exposure. Incorporation of tritiated thymidine observed for individual
26 hepatocytes decreased between 24 and 36 hours, rose slowly back to control levels at 48 hours,
27 significantly increased by 72 hours, and then decreased by 96 hours. Thus, increases in "hepatic"
28 DNA tritiated thymidine uptake did not capture the decrease observed in individual hepatocytes at
29 36 hours. By either measure the population of cells undergoing DNA synthesis was small with
30 the peak level being less than 1% of the hepatocyte population. Zonal distribution of labeled
31 hepatocytes were decreased at 36 hours in all zones, appeared to be slightly greater in perioportal
32 than midzonal cells with centrilobular cells still below control levels by 48 hours, similarly
33 elevated over controls in all zones by 72 hours, and to have returned to near control levels in the
34 midzonal and centrilobular regions but with periportal areas still elevated by 96 hours. These
35 results are consistent with all hepatocytes showing a decrease in DNA synthesis by 36 hours and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-205 DRAFT—DO NOT CITE OR QUOTE
-------
1 then a wave of DNA synthesis to occur, starting at the periportal zone and progressing through the
2 liver acinus that is decreased by 4 days after exposure.
3 Along with changes in liver weight, DNA synthesis, and glycogen accumulation, several
4 studies of DCA and TCA have focused on the extent of peroxisome proliferation as measured by
5 changes in peroxisome number, cytoplasmic volume and enzyme activity induction as potential
6 "key events" occurring from shorter-term exposures that may be linked to chronic effects such as
7 liver tumorigenicity. As noted above in Section E.2.4.1, TCE-induced liver weight gain has been
8 reported to not be dependent on a functional PPARa receptor in female mice while some portion
9 of increased liver weight may have been in male mice. Also as noted cyanide-insensitive PCO
10 has also been reported to not be correlated with the volume or number of peroxisomes that are
11 increased as a result of exposure to TCE or it metabolites (Nakajima et al., 2000; Elcombe et al.,
12 1985: Nelson et al., 1989) and to be highly variable both as a baseline measure and in response to
13 chemical exposures (e.g., variation of up to 6-fold between after WY-14,643 exposure in mice).
14 Also as noted, above the vehicle used in many TCE gavage experiments, corn oil, has been
15 reported to elevate PCO activity as well as catalase activity.
16 A number of short-term studies have examined the effects of TCA and DCA on liver
17 weight increases and evidence of peroxisome proliferation and changes in DNA synthesis. In
18 particular two studies of DCA and TCA used a similar paradigm presented by Elcombe et al.
19 (1985) and Dees and Travis (1993) for TCE effects in mice. Nelson et al. (1989) report findings
20 from gavage doses of unbuffered TCA (500 mg/kg) and DCA (500 mg/kg) in male B6C3F1 mice
21 and Styles et al. (1991) also providing data on peroxisome proliferation using the same paradigm.
22 Nelson et al. (1989) reported levels of PCO activity in mice administered 500 mg/kg DCA or
23 TCA for 10 days with 250 mg/kg Clofibrate administration serving as a positive control. DCA
24 and TCA exposure were reported to not affect body weight, but both to significantly increase liver
25 weight (1.63-fold of control for DCA and 1.30-fold of control for TCA treatments), and percent
26 liver/body weight ratios (1.53-fold of control for DCA and 1.16-fold of control for DCA
27 treatments). PCO activity was reported to be significantly increased by -1.63-, 2.7-, and 5-fold of
28 control for DCA, TCA and Clofibrate treatments, respectively and indicated that both DCA and
29 TCA were weaker inducers of this activity than Clofibrate. Results from randomly selected
30 electron photomicrographs showed an increase in peroxisomes per unit area but gave a different
31 pattern than PCO enzyme activity (i.e., 2.5- and 2.4-fold of control peroxisome volume for DCA
32 and TCA, respectively). Evidence of gross hepatotoxicity was reported to not occur in vehicle or
33 TCA-treated mice. Light microscopic sections were reported to show TCA and control
34 hepatocytes to have the same intensity of PAS staining, but with slightly larger hepatocytes
35 occurring in TCA-treated mice throughout the liver section with architecture and tissue pattern of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-206 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
the liver intact. For DCA, the histopathology was reported to be markedly different than control
mice or TCA treated mice. DCA was reported to induce a marked increase in the size of
hepatocytes throughout the liver with an approximately 1.4-fold of control diameter that was
accompanied by increased PAS staining (indicative of glycogen deposition). All DCA-treated
mice were reported to have multiple white streaks grossly visible on the surface of the liver
corresponding with subcapsular foci of coagulative necrosis that were not encapsulated, varied in
size, and accompanied by a slight inflammatory response characterized by neutrophil infiltration.
A quantitative comparison of effects from equivalent exposures of TCE, TCA, and DCA
(500 mg/kg for 10 days in mice via corn oil gavage for TCE) shown in Table E-13 can be drawn
between the Elcombe et al. (1985), Dees and Travis (1993), Styles et al. (1991), and Nelson et al.
(1989) data for relationship to control values for percent liver/body weight, PCO, and
qualitatively for glycogen deposition.
Table E-13. Comparison of liver effects from TCE, TCA, and DCA (10-day
exposures in mice)
Model
Expo-
sure
% Liver/body
wt.
Peroxisome
volume
Peroxisome
enzyme
activity
Glycogen
deposition
Nelson et al., 1989a
B6C3F1 male
TCA
DCA
1.16-fold
1.53-fold
2.4-fold
2.5-fold
2.7-fold
1.63 -fold
No change
Increased
Styles et al., 1991
B6C3F1 male
TCA
NR
1.9-fold
NR
NR
Elcombe et al., 1985
B6C3F1 male
Alderly Park male (Swiss)
TCE
TCE
1.20-fold
1.43-fold
8-fold
4-fold
NR
NR
NR
NR
Dees and Travis, 1993
B6C3F1 male
B6C3F1 female
TCE
TCE
1. 05-fold"
1.18-fold
NR
NR
NR
NR
NR
NR
17
18
19
20
21
22
23
24
""Unbuffered. NR = not reported as no analysis was performed for this dose or the authors did not report this finding
(i.e., did not note a change in glycogen in description of exposure-related changes).
bStatistically significant although small increase.
Although using a similar species, route of exposure, and dose, the comparison of
responses for TCE and its metabolites shown above are in male mice and also are reflective of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-207 DRAFT—DO NOT CITE OR QUOTE
-------
1 variability in strain, and variability and uncertainty of initial body weights. As described in more
2 detail in Section E.2.2, initial age and body weight have an impact on TCE-related increases in
3 liver weight. Male mice have been reported to have greater variability in response than female
4 mice within and between studies and most of the comparative data for the 10-day 500 mg/kg
5 doses of TCE or its metabolites were from studies in male mice. Corn oil, used as the vehicle for
6 TCE gavage studies but not those of its metabolites, has been noted to specifically affect
7 peroxisomal enzyme induction, body weight gain, and hepatic necrosis, specifically, in male mice
8 (Merrick et al., 1989). Corn oil alone has also been reported to increase PCO activity in F344 rats
9 and to potentiate the induction of PCO activity of TCA (DeAngelo et al., 1989). Thus,
10 quantitative inferences regarding the magnitude of response in these studies are limited by a
11 numb er of factors.
12 The variability in the magnitude of TCE-induced increases in percent liver/body weight
13 across studies in readily apparent but for TCE, TCA and DCA there is an increase in liver weight
14 in mice at this dose after 10 days of exposure. The volume of the peroxisomal compartment in
15 hepatocytes was reported to be more greatly increased from TCE-treatment by Elcombe et al.
16 (1985) than for either TCA or DCA by Nelson et al. (1989) or Styles et al. (1991). However, the
17 control values for the B6C3F1 mice were half that of the other strain reported by Elcombe et al.
18 (1985) and this parameter in general did not match the pattern of PCO activity values reported for
19 TCA and DCA (Nelson et al., 1989). There is no PCO activity data at this dose for TCE but
20 Elcombe et al. (1985) reported that the magnitude of TCE-induced increase in peroxisome
21 volume was similar to that of PCO activity at the only dose where both were tested (1,000 mg/kg
22 TCE). However, Elcombe et al. (1985) reported increased peroxisomal volumes in B6C3F1 mice
23 after 10 days of TCE treatment were not dose-related (i.e., there was little difference between 500,
24 1,000, and 1,500 mg/kg TCE exposures in the magnitude of TCE-induced increases in
25 peroxisomal volume). The lack of dose-response for TCE-induced peroxisomal volume increases
26 was not consistent with increases in percent liver/body weight that increased with increasing TCE
27 exposure concentration. Also as noted above, PCO activity appears to be highly variable in
28 untreated and treated rodents and to vary between experiments and between studies.
29 From the above comparison it is clears that TCE, DCA and TCA exposures were
30 associated with increased liver weight in mice but a question arises as to what changes account
31 for the liver weight increases. For TCE and TCA 500 mg/kg treatments, changes in glycogen
32 were not reported in the general descriptions of histopathological changes (Elcombe et al., 1985;
33 Styles et al., 1991; Dees and Travis, 1993) or were specifically described by the authors as being
34 similar to controls (Nelson et al., 1989). However, for DCA, glycogen deposition was
35 specifically noted to be increased with treatment, although no quantitative analyses was presented
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-208 DRAFT—DO NOT CITE OR QUOTE
-------
1 that could give information as to the nature of the dose-response (Nelson et al., 1989). Issues in
2 regard to not only whether TCE and its metabolites each gives a similar response for a number of
3 parameters, but what potential changes may be associated with carcinogenicity from long-term
4 exposures can be examined by a comparison of the dose-response curves for these parameters
5 from a range of exposure concentrations and durations of exposure. In addition, if glycogen
6 accumulation results from DC A exposure, what proportion of DCA-induced liver weight
7 increases result from such accumulation or other events that may be similar to those occurring
8 with TCE exposure (see Section E.4.2.4, below)?
9 As noted above in Section E.2.4.1., TCE-induced changes in liver weight appear to be
10 proportional to the exposure concentration across route of administration, gender and rodent
11 species. As an indication of the potential contribution of TCE metabolites to this effect, a
12 comparison of the shape of the dose-response curves for liver weight induction for TCE and its
13 metabolites is informative. A number of studies of TCA and DCA in drinking water, conducted
14 from 10-days to 4 weeks, have attempted to measure changes in liver weight induction,
15 peroxisomal enzyme activity, and changes in DNA synthesis predominantly in mice to provide
16 insight into the MOA(s) for liver cancer induction (Parrish et al., 1996; Sanchez and Bull, 1990;
17 Carter et al., 1995; DeAngelo et al., 1989, 2008).
18 Direct comparisons are harder to make between the drinking water studies of DCA and
19 TCA and the gavage studies of TCE (Tables E-14, E-15, and E-16). Similar to 10-day gavage
20 exposures to TCE, 14-day exposures to TCA or DCA via drinking water were reported to induce
21 dose-related increases in liver weight in male B6C3F1 mice (0.3, 1.0, and 2.0 g/L TCA or DCA)
22 with a greater increase in liver weight from DCA than TCA at 2 g/L and a difference in the shape
23 of the dose-response curve (Sanchez and Bull, 1990). They reported a 1.08-, 1.31-, and 1.62-fold
24 of control liver weight for DCA and a 1.15-, 1.22-, and 1.38-fold of control values for TCA at 0.3
25 g/L, 1.0 g/L and 2.0 g/L concentrations, respectively (n = 12-14 mice). While the magnitude of
26 difference between the exposures was ~6.7-fold between the lowest and highest dose, the
27 differences between TCA exposure groups for change in percent of liver weight was -2.5, but for
28 DCA the slope of the dose-response curve for liver weight increases appeared to be closer to the
29 magnitude of difference in exposure concentrations between the groups (i.e., a difference of
30 7.7-fold between the highest and lowest dose for liver weight induction).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-209 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
w g
I
Table E-14. Liver weight induction as percent liver/body weight fold-of-control in male B6C3F1 mice from
DCA or TCA drinking water studies
Concentration
(g/L)
Duration of exposure
14 or 15 days
20 or 21 days
25 days
28 or 30 days
Mean for
average of days
14-30
DCA
0.1
0.3
0.5
1.0
2.0
5.0
1.08-fold
1.12-fold
1.31 -fold
1.62-fold
1.67-fold
1.02-fold
1.24-fold, 1.05-fold
1.46-fold, 2.01 -fold
1.16-fold
2.04-fold
1.16-fold
1.99-fold, 1.42-fold
1.02-fold
1.08-fold
1.15-fold
1.31 -fold
1.83 -fold
1.67-fold
TCA
0.05
0.1
0.3
0.5
1.0
2.0
3.0
5.0
1.15-fold
1.23 -fold, 1.08-fold
1.38-fold, 1.16-fold, 1.26-fold
1.39-fold, 1.35-fold
0.98-fold
1.1 3 -fold
1.3 3 -fold
1.09-fold
1.16-fold
1.33-fold
1.09-fold
0.98-fold
1.15-fold
1.15-fold
1.16-fold
1.30-fold
1.3 3 -fold
1.37-fold
H I
O
H
W
-------
to
O k^j
o ^
"I
I
§
^.
S'
1
TO'
Table E-15. Liver weight induction as percent liver/body weight fold-of-control in male B6C3F1 or Swiss mice
from TCE gavage studies
Concentration
(mg/kg/d)
10 days
28 days
42 days
Mean for average of
days 10-42
B6C3F1
100
250
500
600
1,000
1,200
1,500
2,400
1.00-fold
1.00-fold
1.20-fold, 1.06-fold
1.50-fold, 1.17-fold, 1.50-fold
1.47-fold
1.36-fold
1.64-fold
1.81 -fold
1.00-fold
1.00-fold
1.1 3 -fold
1.36-fold
1.39-fold
1.64-fold
1.47-fold
1.81 -fold
Swiss
100
200
400
500
800
1,000
1,500
1,600
2,000
2,400
1.43 -fold
1.56-fold
1.75-fold
1.32-fold
1.41 -fold
1.38-fold
1.69-fold
1.12-fold
1.15-fold
1.25-fold
1.36-fold
1.63 -fold
1.12-fold
1.15-fold
1.25-fold
1.38-fold
1.36-fold
1.49-fold
1.75-fold
1.63-fold
1.38-fold
1.69-fold
^
to
I
o a.
H I
O >
HH Oq
H ^
O
H
W
-------
Table E-16. B6C3F1 and Swiss (data sets combined)
2
3
4
5
9
10
11
12
13
14
15
16
17
18
19
20
Concentration (mg/kg/d)
100
200
250
400
500
600
800
1,000
1,200
1,500
1,600
2,000
2,400
Mean for average of days 10-42
1.06-fold
1.15-fold
1.00-fold
1.25-fold
1.26-fold
1.36-fold
1.36-fold
1.49-fold
1.64-fold
1.61 -fold
1.63-fold
1.38-fold
1.75-fold
DeAngelo et al. (1989) reported that after 14 days of exposure to 5 g/L or 2 g/L TCA in
male mice, the magnitudes of the difference in the increase in dose (2.5-fold) was generally
higher than the increase percent liver/body weight ratios at these doses (i.e., -40% for the Swiss-
Webster, C3H, and for one of the B6C3F1 mouse experiments, and for the C57BL/6 mouse there
was no difference in liver weight induction between the 2 and 5 g/L TCA exposure groups).
There was a range in the magnitude of percent liver/body weight ratio increases between the
strains of mice with liver weight induction reported to range between 1.26- to 1.66-fold of control
values for the 4 strains of mice at 5 g/L TCA and to range between 1.16- to 1.63 -fold of control
values at 2 g/L TCA. One strain, B6C3F1, was chosen to compare responses between DC A and
TCA. At 1 g/L, 2 g/L and 5 g/L TCA or DCA, DCA was reported to induce a greater increase in
liver weight that TCA (i.e., 1.55- vs. 1.39-fold of control percent liver/body weight ratio for
5.0 g/L DCA vs. TCA, respectively). At the 5 g/L exposures DCA induced -40% greater percent
liver/body weight than TCA. Although as noted above, the majority of the data from this study in
mice did not indicate that the magnitude of difference in exposure concentration was the same as
that of liver weight induction for TCA, in the particular experiment that examined both DCA and
TCA, the increase in percent liver/body weight ratios were similar to the magnitude of difference
in dose between the 2 g/L and 5 g/L exposure concentrations for both DCA and TCA (i.e., 2- to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-212 DRAFT—DO NOT CITE OR QUOTE
-------
1 2.5-fold increase in liver weight change corresponding to a 2.5-fold difference in exposure
2 concentration).
3 Carter et al. (1995) examined 0.5 and 5.0 g/L exposures to DCA in B6C3F1 male mice
4 and reported that percent liver/body weights were increased consistently from 0.5 g/L DCA
5 treatment from 5 days to 30 days of treatment (i.e., a range of 1.05- to 1.16-fold of control). For
6 5.0 g/L DCA exposure the range of increase in percent liver/body weight was reported to be 1.37-
7 to 2.04-fold of control for the same time period. At the 15 days of exposures the percent
8 liver/body weight ratios were 1.67- and 1.12-fold of control for 5.0 and 0.5 g/L DCA and at
9 30 days were 1.99- and 1.16-fold, respectively. The difference in magnitude of dose and percent
10 liver/body weight increase is difficult to determine given that the 5 g/L dose of DCA reduced
11 body weight and significantly reduced water consumption by -50%. The differences in DCA-
12 induced percent liver/body weights were ~6-fold for the 15, 25, and 30-day data between the 0.5
13 and 5 g/L DCA exposures rather than the 10-fold difference in exposure concentration in the
14 drinking water.
15 Parrish et al. (1996) reported that for male B6C3F1 mice exposed to TCA or DCA (0,
16 0.01, 0.5, and 2.0 g/L) for 3 or 10 weeks, the 4- to 5-fold magnitude of difference in doses
17 resulted in increases in percent liver/body weight for the 21-day and 71-day exposures that were
18 greater for DCA than TCA. The percent liver/body weight ratio were 0.98-, 1.13-, and 1.33-fold
19 of control levels at 0.1, 0.5, and 2.0 g/L TCA and for DCA were 1.02-, 1.24-, and 1.46-fold of
20 control levels, respectively, after 21 days of exposure. Both TCA and DCA exposures at 0.1 g/L
21 resulted in difference in percent liver/body weight change of 2% or less. For TCA, although there
22 was a 4-fold increase in magnitude between the 0.5 and 2.0 g/L TCA exposure concentrations, the
23 magnitude of increase for percent liver/body weight increase was 2.5-fold between them at both
24 21 and 71 days of exposure. For DCA, the 4-fold difference in dose between the 0.5 and 2.0 g/L
25 DCA exposure concentrations were reported to result in a ~2-fold increase in percent liver/body
26 weight increase at 21 days and ~4.5-fold increase at 71 days.
27 DeAngelo et al. (2008) studied 3 exposure concentrations of TCA in male B6C3F1 mice,
28 which were an order of magnitude apart, for 4 weeks of exposure. The percent liver/body weight
29 ratios were 1.09-, 1.16-, and 1.35-fold of control levels, for 0.05, 0.5, and 5.0 g/L TCA exposures,
30 respectively. The 10-fold differences in exposure concentration of TCA resulted in ~2-fold
31 differences in percent liver/body weight increases. No dose-response inferences can be drawn
32 from the 4-week study of DCA and TCA in B6C3F1 male mice by Kato-Weinstein et al. (2001)
33 but 2 g/L DCA and 3 g/L TCA in drinking water were reported to induce percent liver/body
34 weights of 1.42- and 1.33-fold of control, respectively (n = 5).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-213 DRAFT—DO NOT CITE OR QUOTE
-------
1 The majority of short-term studies of DC A and TCA in mice have been conducted in the
2 B6C3F1 strain and in males. Studies conducted from 14 to 30 days show a consistent increase in
3 percent liver/body weight induction by TCA or DCA. Accordingly an examination of all of the
4 data from Parrish et al. (1996), Sanchez and Bull (1990), Carter et al. (1995), Kato-Weinstein et
5 al. (2001), and DeAngelo et al. (1989, 2008) from 14 to 30 days of exposure in male B6C3F1
6 mice can give an approximation of the dose-response differences between DCA and TCA for liver
7 weight induction as shown in Table E-14 and Figure E-l, below. Although the data for B6C3F1
8 mice from Sanchez and Bull (1990) is reported as the fold of liver weight rather that percent
9 liver/body weight increase, it is included in the comparison as both reflect increase in liver
10 weight. Similar data can be assessed for TCE for comparative purposes. Short duration studies
11 (10-42 days) were selected because (1) in chronic studies, liver weight increases are confounded
12 by tumor burden, (2) multiple studies are available, and (3) in this duration range, Kjellstrand et
13 al. (1981) reported that TCE-induced increases in liver weight plateau, and (4) TCA studies do
14 not show significant duration-dependent differences in this duration range. These comparisons
15 are presented in Table E-14.
16 DeAngelo et al. (1989) and Carter et al. (1995) used up to 5 g/L DCA and TCA in their
17 experiments with Carter et al. (1995) noting a dramatic decrease in water consumption in the
18 5 g/L DCA treatment groups (46-64% reduction) which can affect body weight as well as dose
19 received. DeAngelo et al. (1989) did not report drinking water consumption. The drinking water
20 consumption was reported by DeAngelo et al. (2008) to be reduced by 11, 17, and 30% in the
21 0.05, 0.5, and 5 g/L TCA treated groups compared to 2 g/L NaCl control animals over 60 weeks.
22 DeAngelo et al. (1999) reported mean drinking water consumption to be reduced by 26% in mice
23 exposed to 3.5 g/L DCA over 100 weeks. Carter et al. (1995) reported that DCA at 5 g/L to
24 decrease drinking water consumption by 64 and 46% but 0.5 g/L DCA to not affect drinking
25 water consumption. Thus, it appears that the 5 g/L concentrations of either DCA or TCA can
26 significantly affect drinking water consumption as well as inducing reductions in body weight.
27 Accordingly, an estimation of the shape of the dose-response curve for comparative purposes
28 between DCA or TCA drinking water studies is best examined at concentrations at 2 g/L or less,
29 especially for DCA.
30
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-214 DRAFT—DO NOT CITE OR QUOTE
-------
Male B6C3F1 mice liver weight for TCA and DCAin drinking water- days 14-30
1.0
Concentration of DCA or TCA (g/l)
1 Figure E-l. Comparison of average fold-changes in relative liver weight to
2 control and exposure concentrations of 2 g/L or less in drinking water for
3 TCA and DCA in male B6C3F1 mice for 14-30 days (Parrish et al.,1996;
4 Sanchez and Bull, 1990; Carter et al., 1995; Kato-Weinstein et al., 2001;
5 DeAngelo et al., 1989, 2008). (Reproduced from Section 4.5.)
6
7
8 The dose-response curves for similar concentrations of DCA and TCA are presented in
9 Figure E-l for durations of exposure from 14-28 days in the male B6C3F1 mouse, which was the
10 most common sex and strain used. For this comparative analysis an average is provided between
11 two values for a given concentration and duration of exposure for comparison with other doses
12 and time points. As noted in the discussion of individual experiments, there appears to be a linear
13 correlation between dose in drinking water and liver weight induction up to 2 g/L of DCA.
14 However, the shape of the dose-response curve for TCA appears to be quite different (i.e., lower
15 concentrations of TCA inducing larger increase that does DCA but then the response reaching an
16 apparent plateau for TCA at higher doses while that of DCA continues to increase). As shown by
17 DeAngelo et al. (2008), 10-fold differences in the magnitude of exposure concentration to TCA
18 corresponded to ~2-fold differences in liver weight induction increases. In addition, TCA studies
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-215 DRAFT—DO NOT CITE OR QUOTE
-------
1 did not show significant duration-dependent difference in liver weight induction in this duration
2 range as shown in Table E-14.
3 Of interest is the issue of how the dose-response curves for TCA and DCA compare to
4 that of TCE in a similar model and dose range. Since TCA and DCA have strikingly different
5 dose-response curves, which one if either best fits that of TCE and thus, can give insight as to
6 which is causative agent for TCE's effects in the liver? In the case of the TCE database in the
7 mouse two strains have been predominantly studied, Swiss and B6C3F1, and both have been
8 reported to get liver tumors in response to chronic TCE exposure. Rather than administered in
9 drinking water, oral TCE studies have been conducted via oral gavage and generally in corn oil
10 for 5 days of exposure per week. The study by Goel et al. (1992) was conducted in ground-nut
11 oil. Vehicle effects, the difference between daily and weekly exposures, the dependence of TCE
12 effects in the liver on its metabolism to a variety of agents capable inducing effects in the liver,
13 differences in response between strains, and the inherent increased variability in use of the male
14 mouse model all add to increased difficulty in establishing the dose-response relationship for TCE
15 across studies and for comparisons to the DCA and TCA database. Despite difference in
16 exposure route, etc., a consistent pattern of dose-response emerges from combining the available
17 TCE data. The effects of oral exposure to TCE from 10-42 days on liver weight induction is
18 shown in Figure E-2 using the data of Elcombe et al. (1985), Dees and Travis (1993), Goel et al.
19 (1992), Merrick et al. (1989), Goldsworthy and Popp (1987), and Buben and O'Flaherty (1985).
20 More detailed discussion of the 4- to 6-week studies is presented in Section E.2.4.3, below (e.g.,
21 for Merrick et al., 1989; Goel et al., 1992; Buben and O'Flaherty, 1985). For this comparative
22 analysis an average is provided between two values per concentration and duration of exposure
23 for comparison with other doses and time points. As shown by the 10-day data in B6C3 Fl mice,
24 there are significant differences in response between studies of male B6C3F1 mice at the same
25 dose of TCE. This variability is similar to findings from inhalation studies of TCE in male mice
26 (Kj ell strand etal., 1983a).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-216 DRAFT—DO NOT CITE OR QUOTE
-------
Male mice liver weight forTCE oral gavage - days 10-42
2.0
•2> 1.6-1
1.4 -
1.2 -
1.0 I • •
0 500 1000 1500 2000 2500 3000
Concentration of TCE (mg/kg/day)
Male mice liver weight for TCE oral gavage - days 10-42
2.0
•2> 1.6H
1.4 -
1.2 -
1.0
B6C3F1 and Swiss
Plot 2 Rear
500 1000 1500 2000 2500
Concentration of TCE (mg/kg/day)
3000
1 Figure E-2. Comparisons of fold-changes in average relative liver weight
2 and gavage dose of (top panel) male B6C3F1 mice for 10-28 days of
3 exposure (Merrick et al., 1989; Elcombe et al., 1985; Goldsworthy and
4 Popp, 1987, Dees and Travis, 1993) and (bottom panel) in male B6C3F1
5 and Swiss mice. (Reproduced from Section 4.5.)
772/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-217 DRAFT—DO NOT CITE OR QUOTE
-------
1 As shown in Figure E-2, oral TCE administration in male B6C3F1 and Swiss mice
2 appeared to induce a dose-related increase in percent liver/body weight that was generally
3 proportional to the increase in magnitude of dose, though as expected, with more variability than
4 observed for a similar exercise for DCA or TCA in drinking water. Common exposure
5 concentrations between B6C3F1 and Swiss mice were 100, 500, 1,000, 1,500 and 2,400 mg/kg/d
6 TCE which corresponded to a 5-, 2-, 1.5-, and 1.6-fold difference in the magnitude of dose. For
7 the data from studies in B6C3 Fl mice, there was no increase reported at 100 mg/kg/d TCE but
8 between 500 and 1,000, 1,000 and 1,500, and 1,500 and 2,400 mg/kg/d TCE the magnitude of
9 difference in doses matched that of the magnitude of increase in percent liver/body weight (i.e., a
10 2.6-, 1.4-, and 1.7-fold increase in liver weight was matched by a 2-, 1.5-, and 1.6-fold increase in
11 TCE exposure concentration at these exposure intervals). However, only 10-day was available
12 for doses between 100 and 500 mg/kg in B6C3F1 mice and at the lower doses, a 10-day interval
13 may have been too short for the increase in liver weight to have been fully expressed. The
14 database for the Swiss mice, which has more data from 28 and 42 days of exposure, support this
15 conclusion. At 28-42 days of exposure there was a much greater increase in liver weight from
16 TCE exposure in Swiss mice than the 10-day data in B6C3F1 mice. In Figure E-2, the 10-day
17 data are included for comparative purpose for the B6C3F1 data set and the Swiss and B6C3F1
18 data sets combined. Both the combined TCE data and that for only B6C3F1 mice shows a
19 correlation with the magnitude of dose and magnitude of percent liver/body weight increase. The
20 slope of the dose-response curves are both closer to that of DCA than TCA. The correlation
21 coefficients for the linear regressions presented for the B6C3F1 data are R2 = 0.861 and for the
22 combined data sets is R2 = 0.712. Comparisons of the slopes of the dose-response curves indicate
23 that TCA is not responsible for TCE-induced liver effects. In this regression all data points were
24 treated equally although some came from several sets of data and others did not. Of note is that
25 the 2,000 mg/kg TCE data point in the combined data set, which is much lower in liver weight
26 response than the other data, is from one experiment (Goel et al., 1992), from 6 mice, at one time
27 point (28 days), and one strain (Swiss). Deletion of these data point from the rest of the 23 used
28 in the study results in a better fit to the data of the regression analysis.
29 A more direct comparison would be on the basis of dose rather than drinking water
30 concentration. The estimations of internal dose of DCA or TCA from drinking water studies have
31 been reported to vary with DeAngelo et al. reporting DCA drinking water concentrations of 1.0,
32 2.0, and 5.0 g/L to result in 90, 166, and 346 mg/kg/d, respectively. For TCA, 0.05, 0.5, 1.0, 2.0,
33 and 5 g/L drinking water exposures were reported to result in 5.8 (range 3.6-8.0), 50 (range of
34 32.5 to 68), 131, 261, and 469 (range 364 to 602) mg/kg/d doses. The estimations of internal dose
35 of DCA or TCA from drinking water studies, while varying considerably (DeAngelo et al., 1989,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-218 DRAFT—DO NOT CITE OR QUOTE
-------
1 2008), nonetheless suggest that the doses of TCE used in the gavage experiments were much
2 higher than those of DC A or TCA. However, only a fraction of ingested TCE is metabolized to
3 DCA or TCA, as, in addition to oxidative metabolism, TCE is also cleared by glutathione (GSH)
4 conjugation and by exhalation.
5 While DCA dosimetry is highly uncertain (see Sections E.3.3 and E.3.5), the mouse
6 physiologically based pharmacokinetic (PBPK) model, described in Section E.3.5 was calibrated
7 using extensive in vivo data on TCA blood, plasma, liver, and urinary excretion data from
8 inhalation and gavage TCE exposures, and makes robust predictions of the rate of TCA
9 production. If TCA were predominantly responsible for TCE-induced liver weight increases, then
10 replacing administered TCE dose (e.g., mg TCE/kg/day) by the rate of TCA produced from TCE
11 (mg TCA/kg/day) should lead to dose-response curves for increased liver weight consistent with
12 those from directly administered TCA. Figure E-3 shows this comparison using the PBPK
13 model-based estimates of TCA production for 4 TCE studies from 28-42 days in the male NMRI,
14 Swiss, and B6C3F1 mice (Kjellstrand et al., 1983b; Buben and O'Flaherty, 1985; Merrick et al.,
15 1989; Goel et al., 1992) and 4 oral TCA studies in B6C3F1 male mice at 2 g/L or lower drinking
16 water exposure (DeAngelo et al., 1989, 2008; Parrish et al., 1996; Kato-Weinstein et al., 2001)
17 from 14-28 days of exposure. The selection of the 28-42 day data for TCE was intended to
18 address the decreased opportunity for full expression of response at 10 days. PBPK modeling
19 predictions of daily internal doses of TCA in terms of mg/kg/d via produced via TCE metabolism
20 would be are indeed lower than the TCE concentrations in terms of mg/kg/d given orally by
21 gavage. The predicted internal dose of TCA from TCE exposure studies are of a comparable
22 range to those predicted from TCA drinking water studies at exposure concentrations in which
23 palability has not been an issue for estimation of internal dose. Thus, although the TCE data are
24 for higher exposure concentrations, they are predicted to produce comparable levels of TCA
25 internal dose estimated from direct TCA administration in drinking water.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-219 DRAFT—DO NOT CITE OR QUOTE
-------
2.5
05
-------
1 dose to 1.75-fold at the highest dose administered orally in Buben and O'Flaherty (1985) and over
2 2-fold in the inhalation study of Kjellstrand et al. (1983b). For this analysis is unlikely that strain
3 differences can account for this inconsistency in the dose-response curves. TCE-induced
4 increases in liver weight appear to be generally similar between B6C3F1 and Swiss male mice
5 (see Table E-14) via oral exposure and between NMRI male and female mice after inhalation,
6 although the NMRI strain appeared to be more prone to TCE-induced toxicity in male mice and
7 for females to have a smaller TCE-induced liver weight increase than other strains (Kjellstrand et
8 al., 1983b). As noted previously, the difference in response between strains and between studies
9 in the same strain for TCE liver weight increases can be highly variable. Little data exist to
10 examine this issue for TCA studies although DeAngelo et al. (1989) report a range of 1.16- to
11 1.63-fold of control percent liver/body weight increase after 14 days exposure at 2 g/L TCA in the
12 Swiss-Webster, C3H, C57BL/6, and B6C3F1 strains, with differences also noted between
13 2 studies of the B6C3F1 mouse.
14 Furthermore, while as noted previously, oral studies appear to report a linear relationship
15 between TCE exposure concentration and liver weight induction, the inclusion of inhalation
16 studies on the basis of internal dose led to a highly consistent dose-response curve for among
17 TCE study. Therefore, it is unlikely that differing routes of exposure can explain the
18 inconsistencies in dose-response. The PBPK model predicted that matching average TCA
19 production by TCE with the equivalent average dose from drinking water-administered TCA also
20 led to an equivalent area-under-the-curve (AUC) of TCA in the liver. Moreover, Dees and Travis
21 (1993) administered 100 to 1,000 mg/kg/d TCA by gavage to male and female B6C3F1 mice for
22 11 days, and did not observe increases in liver/body weight ratios more than 1.28-fold, no higher
23 than those observed with drinking water exposures. Finally, the dose-response consistency
24 between TCE inhalation and gavage studies argues against route of exposure significantly
25 impacting liver weight increases. Thus, no level of TCA administration appears able account for
26 the continuing increase in liver weights observed with TCE, quantitatively inconsistent with TCA
27 being the predominant metabolite responsible for TCE-induced liver weight changes. Thus,
28 involvement of other metabolites, besides TCA, is implicated as the causes of TCE-induced liver
29 effects.
30 Additional analyses do, however, support a role for oxidative metabolism in TCE-induced
31 liver weight increases, and that the parent compound TCE is not the likely active moeity
32 (suggested previously by Buben and O'Flaherty [1985]). In particular, the same studies are
33 shown in Figure E-4 using PBPK-model based predictions of the AUC of TCE in blood and total
34 oxidative metabolism, which produces chloral, trichloroethanol, DCA, and other metabolites in
35 addition to TCA. The dose-response relationship between TCE blood levels and liver weight
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-221 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
increase, while still having a significant trend, shows substantial scatter and a low R2 of 0.43. On
the other hand, using total oxidative metabolism as the dose metric leads to substantially more
consistency dose-response across studies, and a much tighter linear trend with an R2 of 0.90 (see
Figure E-4). A similar consistency is observed using liver-only oxidative metabolism as the dose
metric, with R2 of 0.86 (not shown). Thus, while the slope is similar between liver weight
increase and TCE concentration in the blood and liver weight increase and rate of total oxidative
metabolism, the data are a much better fit for total oxidative metabolism.
CD
w
to
iq
c\i
p
c\i
R=0.426'
co
CD
iq
c\i
p
c\i
oo
R2=0.8955
100 200 300 400 500
Daily AUC TCE in Blood (mg-hr/l)
500 1000 1500
Daily TCE Oxidized (mg/kg-d)
9 Figure E-4. Fold-changes in relative liver weight for data sets in male
10 B6C3F1, Swiss, and NRMI mice reported by TCE studies of duration
11 28-42 days (Kjellstrand et al., 1983b; Buben and O'Flaherty, 1985;
12 Merrick et al., 1989; Goel et al., 1992) using internal dose metrics predicted
13 by the PBPK model described in Section E.3.5: (A) dose metric is the
14 median estimate of the daily AUC of TCE in blood, (B) dose metric is the
15 median estimate of the total daily rate of TCE oxidation. Lines show linear
16 regression. Use of liver oxidative metabolism as a dose metric gives results
17 qualitatively similar to (B), with R2 = 0.86. (Reproduced from Section 4.5.)
18
19
20 As stated in many of the discussions of individual studies, there is a limited ability to
21 detect a statistically significant change in liver weight change in experiments that use a relatively
22 small number of animals. Many experiments have been conducted with 4-6 mice per dose group.
23 The experiments of Buben and O'Flaherty used 12-14 mice per group giving it a greater ability to
24 detect a TCE-induced dose response. In some experiments greater care was taken to document
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-222 DRAFT—DO NOT CITE OR QUOTE
-------
1 and age and weight match the control and treatment groups before the start of treatment. The
2 approach taken above for the analyses of TCE, TCA and DCA uses data across several data sets
3 and gives a more robust description of these dose-response curves, especially at lower exposure
4 levels. For example, the data from DeAngelo et al. (2008) for TCA-induced percent liver/body
5 weight ratio increases in male B6C3F1 mice were only derived from 5 animals per treatment
6 group after 4 weeks of exposure. The 0.05 and 0.5 g/L exposure concentrations were reported to
7 give a 1.09- and 1.16-fold of control percent liver/body weight ratios, which were consistent with
8 the increases noted in the cross-study database above. However, a power calculation shows that
9 the type II error, which should be >50% and thus, greater than the chances of "flipping a coin,"
10 was only a 6 and 7% and therefore, the designed experiment could accept a false null hypothesis.
11 Although the qualitative similarity to the linear dose-response relationship between DCA
12 and liver weight increases is suggestive of DCA being the predominant metabolite responsible for
13 TCE liver weight increases, due to the highly uncertain dosimetry of DCA derived from TCE, this
14 hypothesis cannot be tested on the basis of internal dose. Similarly, another TCE metabolite, CH,
15 has also been reported to induce liver tumors in mice, however, there are no adequate comparative
16 data to assess the nature of liver weight increases induced by this TCE metabolite (see Section
17 E.2.5, below). Whether its formation in the liver after TCE exposure correlates with TCE-
18 induced liver weight changes cannot be determined. Of note is the high variability in total
19 oxidative metabolism reported in mice and humans of Section 3.3, which suggests that the
20 correlation of total TCE oxidative metabolism with TCE-induced liver effects should lead not
21 only to a high degree of variability in response in rodent bioassays which is the case (see Section
22 E.2.4.4, below) but also make detection of liver effects more difficult in human epidemiological
23 studies (see Section 4.3.2). What mechanisms or events are leading to liver weight increases for
24 DCA, TCA and TCE can be examined by correlations between changes in glycogen content,
25 hepatocyte volume, and evidence of polyploidization noted in short-term assays.
26 Data have been reported regarding the nature of changes the TCE and its metabolites
27 induce in the liver and are responsible for the reported increases in liver weight. Increased liver
28 weight may result from increased size or hypertrophy of hepatocytes through changes in glycogen
29 deposition, but also through increased polyploidization. Increased cell number may also
30 contribute to increased liver weight. As noted above in Section E.2.4.1, hepatocellular
31 hypertrophy appeared to be related to TCE-induced liver weight changes after short-term
32 exposures. However, neither glycogen deposition, DNA synthesis, or increases in mitosis appear
33 to be correlated with liver weight increases. In particular DNA synthesis increases were similar
34 from 250-1,000 mg/kg and peroxisomal volume was similar between 500 and 1,500 mg/kg TCE
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-223 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposures after 10 days. Autoradiographs identified hepatocytes undergoing DNA synthesis in
2 "mature" hepatocytes that were in areas where polyploidization typically takes place in the liver.
3 By 14 days of exposure, Sanchez and Bull (1990) reported that both dose-related TCA-
4 and DCA-induced increases in liver weight were generally consistent with changing cell size
5 increases, but were not correlated with patterns of change in hepatic DNA content, incorporation
6 of tritiated thymidine in DNA extracts from whole liver, or incorporation of tritiated thymidine in
7 hepatocytes. There are conflicting reports of DNA synthesis induction in individual hepatocytes
8 for up to 14 days of DCA or TCA exposure and a lack of correlation with patterns observed for
9 this endpoint and those of whole liver thymidine incorporation. The inconsistency of whole liver
10 DNA tritiated thymidine incorporation with that reported for hepatocytes was noted by the
11 Sanchez and Bull (1990) to be unexplained. Carter et al. (1995) also report a lack of correlation
12 between hepatic DNA tritiated thymidine incorporation and labeling in individual hepatocytes in
13 male mice. Carter et al. (1995) reported no increase in labeling of hepatocytes in comparison to
14 controls for any DCA treatment group from 5 to 30 days of DCA exposure. Rather than increase
15 hepatocyte labeling, DCA induced a decrease with no change reported from days 5 though 15 but
16 significantly decreased levels between days 20 and 30 for 0.5 g/L that were similar to those
17 observed for the 5 g/L exposures.
18 The most comparable time period between TCE, TCA and DCA results for whole liver
19 thymidine incorporation is the 10- and 14-day durations of exposure when peak tritiated
20 thymidine incorporation into individual hepatocytes and whole liver for TCA and DCA have been
21 reported to have already passed (Styles et al., 1991; Sanchez and Bull, 1990; Pereira, 1996; Carter
22 et al., 1995). Whole liver DNA synthesis was elevated over control levels by ~2-fold after from
23 250 to 1,000 mg/kg TCE exposure after 10 days of exposure but did not correlate with mitosis
24 (Elcombe et al., 1985; Dees and Travis, 1993). After 3 weeks of exposure to TCE, Laughter et al.
25 (2004) reported in individual hepatocytes that 1 and 4.5% of hepatocytes had undergone DNA
26 synthesis in the last week of treatment for the 500 and 1,000 mg/kg TCE levels, respectively.
27 More importantly, these data show that hepatocyte proliferation in TCE-exposed mice at 10 days
28 of exposure or for DCA- or TCA-exposed mice for up to 14 days of exposure is confined to a
29 very small population of cells in the liver.
30 In regard to cell size, although increased glycogen deposition with DCA exposure was
31 noted by Sanchez and Bull (1990), lack of quantitative analyses of that accumulation in this study
32 precludes comparison with DCA-induced liver weight gain. Although not presenting a
33 quantitative analysis, Sanchez and Bull (1990) reported DCA-treated B6C3F1 mice to have large
34 amounts of PAS staining material and Swiss-Webster mice to have similar increase despite
35 reporting differences of DCA-induced liver weight gain between the two strains. The lack of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-224 DRAFT—DO NOT CITE OR QUOTE
-------
1 concordance of the DC A-induced magnitude of increase in liver weight with that of glycogen
2 deposition is consistent with the findings for longer-term exposures to DCA reported by
3 Kato-Weinstein et al. (2001) and Pereira et al. (2004) in mice (see Section E.2.4.4, below).
4 Carter et al. (1995) reported that in control mice there was a large variation in apparent glycogen
5 content and also did not perform a quantitative analysis of glycogen deposition. The variability
6 of this parameter in untreated animals and the extraction of glycogen during normal tissue
7 processing for light microscopy makes quantitative analyses for dose-response difficult unless
8 specific methodologies are employed to quantitatively assess liver glycogen levels as was done
9 by Kato-Weinstein et al. (2001) and Pereira et al. (2004).
10 Although suggested by their data, polyploidization was not examined for DCA or TCA
11 exposure in the study of Sanchez and Bull (1990). Carter et al. (1995) reported that hepatocytes
12 from both 0.5 and 5 g/L DCA treatment groups were reported to have enlarged, presumably
13 polyploidy nuclei with some hepatocyte nuclei labeled in the mid-zonal area. There were
14 statistically significant changes in cellularity, nuclear size, and multinucleated cells during
15 30 days exposure to DCA. The percentage of mononucleated cells hepatocytes was reported to
16 be similar between control and DCA treatment groups at 5- and 10-day exposure. However, at
17 15 days and beyond, DCA treatments were reported to induce increases in mononucleated
18 hepatocytes. At later time periods there were also reports of DCA-induced increases nuclear
19 area, consistent with increased polyploidization without mitosis. The consistent reporting of an
20 increasing number of mononucleated cells between 15 and 30 days could be associated with
21 clearance of mature hepatocytes as suggested by the report of DCA-induced loss of cell nuclei.
22 The reported decrease in the numbers of binucleate cells in favor of mononucleate cells is not
23 typical of any stage of normal liver growth (Brodsky and Uryvaeva, 1977). The linear dose-
24 response in DCA-induced liver weight increase was not consistent with the increased numbers of
25 mononucleate cells and increase nuclear area reported from Day 20 onward by Carter et al.
26 (1995). Specifically, the large differences in liver weight induction between the 0.5 g/L
27 treatment group and the 5 g/L treatment groups at all times studied also did not correlate with
28 changes in nuclear size and percent of mononucleate cells. Thus, DCA-induced increases in liver
29 weight were not a function of cellular proliferation, but probably included hypertrophy associated
30 with polyploidization, increased glycogen deposition and other factors.
31 In regard to necrosis, Elcombe et al. (1985) reported only small incidence of focal
32 necrosis in 1,500 mg/kg TCE-exposed mice and no necrosis at exposures up to 1,000 mg/kg for
33 10 days as did Dees and Travis (1993). Sanchez and Bull (1990) report DCA-induced localized
34 areas of coagulative necrosis both for B6C3F1 and Swiss-Webster mice at higher exposure
35 levels (1 or 2 g/L) by 14 days but not at the 0.3 g/L level or earlier time points. For TCA
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-225 DRAFT—DO NOT CITE OR QUOTE
-------
1 treatment, necrosis was reported to not be associated with TCA treatment for up to 2 g/L and up
2 to 14 days of exposure. Carter et al. (1995) reported that mice given 0.5 g/L DCA for 15, 20,
3 and 25 days had midzonal focal cells with less detectable or no cell membranes, loss of the
4 coarse granularity of the cytoplasm, with some cells having apparent karyolysis, but for liver
5 architecture to be normal.
6 As for apoptosis, Both Elcombe et al. (1985) and Dees and Travis (1993) reported no
7 changes in apoptosis other than increased apoptosis only at a treatment level of 1,000 mg/kg
8 TCE. Rather than increases in apoptosis, peroxisome proliferators have been suggested to
9 inhibit apoptosis as part of their carcinogenic MOA (see Section E.3.4.1). However, the age and
10 species studied appear to greatly affect background rates of apoptosis. Snyder et al. (1995)
11 report that control mice were reported to exhibit apoptotic frequencies ranging from -0.04 to
12 0.085%, that over the 30-day period of their study the frequency rate of apoptosis declined, and
13 suggest that this pattern is consistent with reports of the livers of young animals undergoing
14 rapid changes in cell death and proliferation. They reported rat liver to have a greater the
15 estimated frequency of spontaneous apoptosis (~0.1%) and therefore, greater than that of the
16 mouse. Carter et al. (1995) reported that after 25 days of 0.5 g/L DCA treatment apoptotic
17 bodies were reported as well as fewer nuclei in the pericentral zone and larger nuclei in central
18 and midzonal areas. This would indicate an increase in the apoptosis associated potential
19 increases in polyploidization and cell maturation. However, Snyder et al. (1995) report that
20 mice treated with 0.5 g/L DCA over a 30-day period had a similar trend as control mice of
21 decreasing apoptosis with age. The percentage of apoptotic hepatocytes decreased in DCA-
22 treated mice at the earliest time point studied and remained statistically significantly decreased
23 from controls from 5 to 30 days of exposure. Although the rate of apoptosis was very low in
24 controls, treatment with 0.5 g/L DCA reduced it further (-30-40% reduction) during the 30-day
25 study period. The results of this study not only provide a baseline of apoptosis in the mouse
26 liver, which is very low, but also to show the importance of taking into account the effects of
27 age on such determinations. The significance of the DCA-induced reduction in apoptosis
28 reported in this study, from a level that is already inherently low in the mouse, to account for the
29 MOA for induction of DCA-induced liver cancer is difficult to discern.
30
31 E.2.4.3. Summary Trichloroethylene (TCE) Subchronic and Chronic Studies
32 The results of longer-term (Channel et al., 1998; Toraason et al., 1999; Parrish et al.,
33 1996) studies of "oxidative stress" for TCE and its metabolites are discussed in
34 Section E.3.4.2.3. Of note are the findings that the extent of increased enzyme activities
35 associated with peroxisome proliferation do not appear to correlate with measures of oxidative
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-226 DRAFT—DO NOT CITE OR QUOTE
-------
1 stress after longer term exposures (Parrish et al., 1996) and single strand breaks (Chang et al.,
2 1992).
3 Similar to the reports of Melnick et al. (1987) in rats, Merrick et al. (1989) report that
4 vehicle (aqueous or gavage) affects TCE-induced toxicity in mice. Vehicle type made a large
5 difference in mortality, extent of liver necrosis, and liver weight gain in male and female
6 B6C3F1 mice after 4 weeks of exposure. The lowest dose used in this experiment was
7 600 mg/kg/d in males and 450 mg/kg/d in females. Administration of TCE via gavage using
8 Emulphor resulted in mortality of all of the male mice and most of the female mice at a dose in
9 corn oil that resulted in few deaths. However, use of Emulphor vehicle induced little if any
10 focal necrosis in males at concentrations of TCE in corn oil gavage that caused significant focal
11 necrosis, indicating vehicle effects.
12 As discussed above in Section E.2.4.2, the extent of TCE-induced liver weight increases
13 was consistent between 4 and 6 weeks of exposure and between 10-day and 4 week exposure at
14 higher dose levels. In general, the reported elevations of enzymatic markers of liver toxicity and
15 results for focal hepatocellular necrosis were not consistent and did not reflect TCE dose-
16 responses observed for induction of liver weight increases (Merrick et al., 1989). Female mice
17 given corn oil and male and female mice given TCE in Emulphor were reported to have "no to
18 negligible necrosis" although they had increased liver weight from TCE exposure. Using a
19 different type of oil vehicle, Goel et al. (1992) exposed male Swiss mice to TCE in groundnut
20 oil at concentrations ranging from 500 to 2,000 mg/kg for 4 weeks and reported no changes in
21 body weight up to 2,000 mg/kg, although there was a 15% decrease at the highest dose, but
22 increases TCE-induced increase in percent liver/body weight ratio. At a dose of 1,000 and
23 2,000 mg/kg, liver swelling, vacuolization, and widespread degenerative necrosis of hepatocytes
24 was reported along with marked proliferation of "endothelial cells" but no quantitation
25 regarding the extent or location of hepatocellular necrosis was reported, nor whether there was a
26 dose-response relationship in these events. They reported a TCE-related dose-response in
27 catalase, liver protein but decreased induction at the 2,000 mg/kg level where body weight had
28 decreased.
29 Three studies were published by Kjellstrand et al. that examined effects of TCE
30 inhalation primarily in mice using whole body inhalation chambers (Kjellstrand et al., 1981,
31 1983a, b). Liver weight changes were used as the indication of TCE-induced effects. The
32 quantitative results from these experiments had many limitations due to their experimental
33 design including failure to determine body weight changes for individual animals and inability
34 to determine the exact magnitude of TCE due to concurrent oral TCE ingestion from food and
35 grooming behavior. An advantage of this route of exposure is that there were not confounding
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-227 DRAFT—DO NOT CITE OR QUOTE
-------
1 vehicle effects. The results from Kjellstrand et al. (1981) are particularly limited by
2 experimental design errors but reported similar increases in liver weight gain in gerbils and rats
3 exposed at 150 ppm TCE. For rats, Kjellstrand et al. (1981) do report increases in liver/body
4 weight ratios of 1.26- and 1.21-fold of control in male and female rat 30 days of continuous
5 TCE inhalation exposure. The unpublished report of Woolhiser et al. (2006) reports 1.05-,
6 1.07-, and 1.13-fold of control percent liver/body weight changes in 100-, 300- and
7 1,000-ppm-exposure groups that are exposed for 6 hours/day, 5 days/week for 4 weeks in
8 groups of 8 female S-D rats. At the two highest exposure levels, body weight was reduced by
9 TCE exposure. If the 150 ppm continuous exposure concentrations of Kjellstrand are analogous
10 to 750-ppm-exposures using the paradigm of Woolhiser et al. (2006). Therefore, the very
11 limited inhalation database for rats does indicate TCE-related increases in liver weight.
12 The study of Kjellstrand et al. (1983 a) employed a more successful experimental design
13 that recorded liver weight changes in carefully matched control and treatment groups to
14 determine TCE-treatment related effects on liver weight in 7 strains of mice after 30 days of
15 continuous inhalation exposure at 150 ppm TCE. Individual animal body weight changes were
16 not recorded so that such an approach cannot take into account the effects of body weight
17 changes and determine a relative percent liver/body weight ratio. The data presented in this
18 report was for absolute liver weight changes between treated and nontreated groups with
19 carefully matched average body weights at the initiation of exposure. A strength of the
20 experimental design is its presentation of results between duplicate experiments and thus, to
21 show the differences in results between similar exposed groups that were conducted at different
22 times. This information gives a measure of variability in response with time. Mouse strain
23 groups, that did not experience TCE-induced decreased body weight gain in comparison to
24 untreated groups (i.e., DBA and wild-type mice), represented the most accurate determination of
25 TCE-induced liver weight changes given that systemic toxicity that affects body weight can also
26 affect liver weight. The C57BL, B6CBA, and NZB groups all had at least one group out of two
27 of male mice with changes in final body weight due to TCE exposure. Only one group of NMRI
28 mice were reported in this study and that group had TCE-induced decreases in final body
29 weight. The A/sn group not only had both male groups with decreased final body weight after
30 TCE exposure (along with differences between exposed and control groups at the initiation of
31 exposure) but also a decrease in body weight in one of the female groups and thus, appears to be
32 the strain with the greatest susceptibility to TCE-induced systemic toxicity. In strains of male
33 mice in which there was no TCE-induced affects on final body weight (wild-type and DBA), the
34 influence of gender on liver weight induction and variability of the response could be more
35 readily assessed. In wild-type mice there was a 1.76- and 1.80-fold of control liver weight in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-228 DRAFT—DO NOT CITE OR QUOTE
-------
1 groups 1 and 2 for female mice, and for males a 1.84- and 1.62-fold of control liver weight for
2 groups 1 and 2, respectively. For DBA mice there was a 1.87- and 1.88-fold of control liver
3 weight in groups 1 and 2 for female mice, and for males a 1.45- and 2.00-fold of control liver
4 weight for groups 1 and 2, respectively. Of note, as described previously, the size of the liver is
5 under strict control in relation to body size. An essential doubling of the size of the liver is a
6 profound effect with the magnitude of liver weight size increase physiologically limited.
7 Overall, the consistency between groups of female mice of the same strain for TCE-
8 induced liver weight gain, regardless of strain examined, was striking as was the lack of body
9 weight changes at TCE exposure levels that induced body weight changes in male mice. In the
10 absence of body weight changes, the difference in TCE-response in female mice appeared to be
11 reflective of strain and initial weight differences. Groups of female mice with higher body
12 weights, regardless of strain, generally had higher increases in TCE-induced liver weight
13 increases. For the C57BL and As/n strains, female mice starting weights were averaged 17.5
14 and 15.5 g, while the average liver weights were 1.63- and 1.64-fold of control after TCE
15 exposure, respectively. For the B6CBA, wild-type, DBA, and NZB female groups the starting
16 body weights averaged 22.5, 21.0, 23.0, and 21.0 g, while the average liver weights were 1.70-,
17 1.78-, 1.88-, and 2.09-fold of control after TCE exposure, respectively. The NMRI group of
18 female mice, did not follow this general pattern and had the highest initial body weight for the
19 single group of 10 mice reported (i.e., 27 g) associated with 1.66-fold of control liver weight.
20 The results of Kjellstrand et al. (1983a) suggested that there was more variability
21 between male mice than female mice in relation to TCE-induced liver weight gain. More strains
22 exhibited TCE-induced body weight changes in male mice than female mice suggesting
23 increased susceptibility of male mice to TCE toxicity as well as more variability in response.
24 Initial body weight also appeared to be a factor in the magnitude of TCE-induced liver weight
25 induction rather than just strain. In general, the strains and groups within strain that had TCE-
26 induced body weight decreases had smaller TCE-induced increase in liver weight. Therefore,
27 only examining liver weight in males as an indication of TCE treatment effects would not be an
28 accurate predictor of strain sensitivity nor the magnitude or response at doses that also affect
29 body weight. The results from this study show that comparison of the magnitude of TCE
30 response, as measured by liver weight increases, should take into account, strain, gender, initial
31 body weight and systemic toxicity. It shows a consistent pattern of increased liver weight in
32 both male and female mice after TCE exposure of 150 ppm for 30 days.
33 Kjellstrand et al. (1983b) presented data in the NMRI strain of mice (a strain that
34 appeared to be more prone to TCE-induced toxicity in male mice and a smaller TCE-induced
35 increase in liver weight in female mice) after inhalation exposure of 37 to 300 ppm TCE. They
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-229 DRAFT—DO NOT CITE OR QUOTE
-------
1 used the same experimental paradigm as that reported in Kjellstrand et al. (1983a) except for
2 exposure concentration. For female mice exposed to concentrations of TCE ranging from 37 to
3 300 ppm TCE continuously for 30 days, only the 300 pm group experienced a 16% decrease in
4 body weight between control and exposed animals and therefore, changes in TCE-induced liver
5 weight increases were affected by changes in body weight only for that group. Initial body
6 weights in the TCE-exposed female mice were similar in each of these groups (i.e., range of
7 29.2-31.6 g, or 8%), with the exception of the females exposed to 150 ppm TCE for 30 days
8 (i.e., initial body weight of 27.3 g), reducing the effects of differences in initial body weight on
9 TCE-induced liver weight induction. Exposure to TCE continuously for 30 days was reported to
10 result in a linear dose-dependent increase in liver weight in female mice with 1.06-, 1.27-, 1.66-,
11 and 2.14-fold of control liver weights reported at 37 ppm, 75 ppm, 150 ppm, and 300 ppm TCE,
12 respectively. In male mice there were more factors affecting reported liver weight increases
13 from TCE exposure. For male mice both the 150- and 300-ppm-exposed groups experienced a
14 10 and 18% decrease in final body weight after TCE exposure, respectively. The 37- and 75-
15 ppm groups did not have decreased final body weight due to TCE exposure but varied by 12%
16 in initial body weight. TCE-induced increases in liver weight were reported to be 1.15-, 1.50-,
17 1.69-, and 1.90-fold of control for 37, 75, 150, and 300 ppm TCE exposure in male mice,
18 respectively. The flattening of the dose-response curve at the two highest doses is consistent
19 with the effects of toxicity on final body weight.
20 Kjellstrand et al. (1983b) noted that liver mass increase and the changes in liver cell
21 morphology were similar in TCE-exposed male and female mice and report that after 150 ppm
22 exposure for 30 days, liver cells were generally larger and often displayed a fine vacuolization
23 of the cytoplasm, changes in nucleoli appearance, Kupffer cells of the sinusoid to be increased
24 in cellular and nuclear size, the intralobular connective tissue was infiltrated by inflammatory
25 cells and for exposure to TCE in higher or lower concentrations during the 30 days to produce a
26 similar morphologic picture. For mice that were exposed to 150 ppm TCE for 30 days and then
27 examined 120 days after the cessation of exposure, liver weights were 1.09-fold of control for
28 TCE-exposed female mice and the same as controls for TCE-exposed male mice. However, the
29 livers were not the same as untreated liver in terms of histopathology. The authors reported that
30 "after exposure to 150 ppm for 30 days, followed by 120 days of rehabilitation, the
31 morphological picture was similar to that of the air-exposure controls except for changes in
32 cellular and nuclear sizes." The authors did not present any quantitative data on the lesions they
33 describe, especially in terms of dose-response, and most of the qualitative description is for the
34 150-ppm-exposure level in which there are consistent reports of TCE induced body weight
35 decreases in male mice. Although stating that Kupffer cells were increased in cellular and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-230 DRAFT—DO NOT CITE OR QUOTE
-------
1 nuclear size, no differential staining was applied to light microscopy sections and used to
2 distinguish Kupffer from endothelial cells lining the hepatic sinusoid in this study. Without
3 differential staining such a determination is difficult at the light microscopic level and a question
4 remains as to whether theses are the same cells as described by Goel et al. (1992) as a
5 proliferation of sinusoidal endothelial cells after exposures of 1,000 and 2,000 mg/kg/d TCE
6 exposure for 28 days in male Swiss mice. As noted in Section E.2.4.2, the discrepancy in DNA
7 synthesis measures between hepatocyte examinations of individual hepatocytes and whole liver
8 measures in several reports of TCE metabolite exposure, is suggestive of increased DNA
9 synthesis in the nonparenchymal cell compartment of the liver. Thus, nonparenchymal cell
10 proliferation is suggested as an effect of subchronic TCE exposures in mice without concurrent
11 focal necrosis via inhalation studies (Kjellstrand et al., 1983b) and with focal necrosis in the
12 presence of TCE in a groundnut oil vehicle (Goel et al., 1992).
13 Although Kjellstrand et al. (1983b) did not discuss polyploidization, the changes in cell
14 size and especially the continued change in cell size and nuclear staining characteristics after
15 120 days of cessation of exposure are consistent with changes in polyploidization induced by
16 TCE that were suggested in studies from shorter durations of exposure (Elcombe et al., 1985;
17 Dees and Travis, 1993) and of longer durations (e.g., Buben and O'Flaherty, 1985). Of note is
18 that in the histological description provided by Kjellstrand et al. (1983b), there is no mention of
19 focal necrosis or apoptosis resulting from these exposures to TCE to mice. Vacuolization is
20 reported and consistent with hepatotoxicity or lipid accumulation, which is lost during routine
21 histological slide preparation. The lack of reported focal necrosis in mice exposed through
22 inhalation is consistent with reports of gavage experiments of TCE in mice that do not use corn
23 oil as the vehicle (Merrick et al., 1989).
24 Buben and O'Flaherty (1985) reported the effects of TCE via corn oil gavage after six
25 weeks of exposure at concentrations ranging from 100 to 3,200 mg/kg d. This study was
26 conducted with older mice than those generally used in chronic exposure assays (Male Swiss-
27 Cox outbred mice between 3 and 5 months of age). Liver weight increases, decreases in liver
28 G6P activity, increases in liver triglycerides, and increases in SGPT activity were examined as
29 parameters of liver toxicity. Few deaths were reported during the 6-week exposure period
30 except at the highest dose and related to central nervous system depression. TCE exposure
31 caused dose-related increases in percent liver/body weight with a dose as low as 100 mg/kg/d
32 were reported to cause a statistically significant increase (i.e., 112% of control). The increases
33 in liver size were attributed to hepatocyte hypertrophy, as revealed by histological examination
34 and by a decrease in the liver DNA concentration, and although enlarged, were reported to
35 appear normal. A dose-related trend toward triglyceride concentration was also noted. A dose-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-231 DRAFT—DO NOT CITE OR QUOTE
-------
1 related decrease in glucose-6-phophatase activity was reported with similar small decreases
2 (-10%) observed in the TCE exposed groups that did not reach statistical significance until the
3 dose reached 800 mg/kg TCE exposure. SGPT activity was not observed to be increased in
4 TCE-treated mice except at the two highest doses and even at the 2,400 mg/kg dose half of the
5 mice had normal values. The large variability in SGPT activity was indicative of heterogeneity
6 of this response between mice at the higher exposure levels for this indicator of liver toxicity.
7 Such variability of response in male mice is consistent with the work of Kjellstrand et al. Thus,
8 the results from Buben and O'Flaherty (1985) suggest that hepatomegaly is a robust response
9 that was reported to be observed at the lowest dose tested, dose-related, and not accompanied by
10 overt toxicity.
11 In terms of histopathology, Buben and O'Flaherty (1985) reported swollen hepatocytes
12 with indistinct borders; their cytoplasm was clumped and a vesicular pattern was apparent and
13 not simply due to edema in TCE-treated male mice. Karyorhexis (the disintegration of the
14 nucleus) was reported to be present in nearly all specimens from TCE-treated animals and
15 suggestive of impending cell death, not present in controls, and to appear at a low level at
16 400 mg/kg TCE exposure level and slightly higher at 1,600 mg/kg TCE exposure level. Central
17 lobular necrosis was present only at the 1,600 mg/kg TCE exposure level and at a very low
18 level. Buben and O'Flaherty report increased polyploidy in the central lobular region for both
19 400 mg/kg and 1,600 mg/kg TCE and described as hepatic cells having two or more nuclei or
20 enlarged nuclei containing increased amounts of chromatin, but at the lowest level of severity or
21 occurrence. Thus, the results of this study are consistent with those of shorter-term studies via
22 gavage, which report hepatocellular hypertrophy in the centralobular region, increased liver
23 weight induced at the lowest exposure level tested and at a level much lower than those inducing
24 overt toxicity, and that TCE exposure is associated with changes in ploidy.
25 The National Toxicology Program 13-week study of TCE gavage exposure in 10 F344/N
26 rats (125 to 2,000 mg/kg [males] and 62.5 to 1,000 mg/kg [females]) and in B6C3Flmice (375
27 to 6,000 mg/kg) reported all rats survived the 13-week study, but males receiving 2,000 mg/kg
28 exhibited a 24% difference in final body weight. The study descriptions of pathology in rats and
29 mice were not very detailed and included only mean liver weights. The rats had increased
30 pulmonary vasculitis at the highest concentration of TCE and that viral liters were positive for
31 Sendai virus and no liver effects were noted for them in the study. For mice, liver weights (both
32 absolute and percent liver/body weight) were reported to increase in a dose-related fashion with
33 TCE -exposure and to be increased by more than 10% in 750 mg/kg TCE-exposed males and
34 1,500 mg/kg or more TCE-exposed females. Hepatotoxicity was reported as centrilobular
35 necrosis in 6/10 males and 1/10 females exposed to 6,000 mg/kg TCE and multifocal areas of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-232 DRAFT—DO NOT CITE OR QUOTE
-------
1 calcifications scattered throughout 3,000 mg/kg TCE exposed male mice and only a single
2 female 6,000 mg/kg dose, considered to be evidence of earlier hepatocellular necrosis. One
3 female mouse exposed to 3,000 mg/kg TCE also had a hepatocellular adenoma, an extremely
4 rare lesion in female mice of this age (20 weeks). However, at the lowest dose of exposure, was
5 a consistent decrease in liver weigh in female and male mice after 13 weeks of TCE exposure.
6 Kawamoto et al. (1988) exposed rats to 2 g/kg TCE subcutaneously for 15 weeks and
7 reported TCE-induced increases in liver weight. They also reported increase in cytochrome
8 P450, cytochrome b-5, and NADPH cytochrome c reductase. The difficulties in relating this
9 route of exposure to more environmentally relevant ones is discussed in Section E.2.2.11.
10 For 2-year or lifetime studies of TCE exposure a consistent hepatocarcinogenic response
11 has been observed in mice of differing strains and genders and from differing routes of
12 exposure. However, for rat studies some studies have been confounded by mortality from
13 gavage error or the toxicity of the dose of TCE administered. In some studies, a relative
14 insensitive strain of rat has been used. However, in general it appears that the mouse is more
15 sensitive than the rat to TCE-induced liver cancer. Three studies give results the authors
16 consider to be negative for TCE-induced liver cancer in mice, but have either design and/or
17 reporting limitations, or are in strains and paradigms with apparent low ability for liver cancer
18 induction or detection.
19 Fukuda et al. (1983) reported a 104-week inhalation bioassay in female Crj:CD-l (ICR)
20 mice and female Crj :CD (S-D) rats exposed to 0, 50, 150 and 450 ppm TCE (n = 50). There
21 were no reported incidences of mice or rats with liver tumors for controls indicative of relatively
22 insensitive strains used in the study for liver effects. While TCE was reported to induce a
23 number of other tumors in mice and rats in this study, the incidence of liver tumors was less than
24 2% after TCE exposure. Of note is the report of cystic cholangioma reported in 1 group of rats.
25 Henschler et al. (1980) exposed NMRI mice and WIST random bred rats to 0, 100, and
26 500 ppm TCE for 18 months (n = 30). This study is limited by short duration of exposure, low
27 number of animals, and low survival in rats. Control male mice were reported to have one
28 hepatocellular carcinoma and 1 hepatocellular adenoma with the incidence rate unknown. In the
29 100 ppm TCE exposed group, 2 hepatocellular adenomas and 1 mesenchymal liver tumor were
30 reported. No liver tumors were reported at any dose of TCE in female mice or controls. For
31 male rats, only 1 hepatocellular adenomas at 100 ppm was reported. For female rats no liver
32 tumors were reported in controls, but 1 adenoma and 1 cholangiocarcinoma was reported at
33 100 ppm TCE and at 500 ppm TCE, 2 cholangioadenomas, a relatively rare biliary tumor, was
34 reported. The difference in survival in mice, did not affect the power to detect a response, as
35 was the case for rats. However, the low number of animals studied, abbreviated exposure
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-233 DRAFT—DO NOT CITE OR QUOTE
-------
1 duration, and apparently low sensitivity of this paradigm (i.e., no background response in
2 controls) suggests a study of limited ability to detect a TCE carcinogenic liver response. Of note
3 is that both Fukuda et al. (1983) and Henschler et al. (1980) report rare biliary cell derived
4 tumors in rats in relatively insensitive assays.
5 Van Duuren et al. (1979), exposed mice to 0.5 mg/mouse to TCE via gavage once a
6 week in 0.1 mL trioctanion (n = 30). Inadequate design and reporting of this study limit that
7 ability to use the results as an indicator of TCE carcinogenicity.
8 The NCI (1976) study of TCE was initiated in 1972 and involved the exposure of
9 Osborn-Mendel rats and B6C3F1 mice to varying concentrations of TCE. The animals were
10 coexposed to a number of other carcinogens as exhalation as multiples studies and control
11 animals all shared the same laboratory space. Treatment duration was 78 weeks and animals
12 received TCE via gavage in corn oil at 2 doses (n = 20 for controls, but n = 50 for treatment
13 groups). For rats, the high dose was reported to result in significant mortality (i.e., 47/50 high-
14 dose rats died before scheduled termination of the study). A low incidence of liver tumors was
15 reported for controls and carbon tetrachloride positive controls in rats from this study. In
16 B6C3F1 mice, TCE was reported to increase incidence of hepatocellular carcinomas in both
17 doses and both genders of mice (-1,170 and 2,340 mg/kg for males and 870 and 1,740 mg/kg
18 for female mice). Hepatocellular carcinoma diagnosis was based on histologic appearance and
19 metastasis to the lung. The tumors were described in detail and to be heterogeneous "as
20 described in the literature" and similar in appearance to tumors generated by carbon
21 tetrachloride. The description of liver tumors in this study and tendency to metastasize to the
22 lung are similar to descriptions provided by Maltoni et al. (1986) for TCE-induced liver tumors
23 in mice via inhalation exposure.
24 For male rats, noncancer pathology in the NCI (1976) study was reported to include
25 increased fatty metamorphosis after TCE exposure and angiectasis or abnormally enlarged blood
26 vessels. Angiectasis can be manifested by hyperproliferation of endothelial cells and dilatation
27 of sinusoidal spaces. The authors conclude that due to mortality, "the test is inconclusive in
28 rats." They note the insensitivity of the rat strain used to the positive control of carbon
29 tetrachloride exposure.
30 The NTP (1990) study of TCE exposure in male and female F344/N rats, and B6C3F1
31 mice (500 and 1,000 mg/kg for rats and 1,000 mg/kg for mice) is limited in the ability to
32 demonstrate a dose-response for hepatocarcinogenicity. There was also little reporting of
33 non-neoplastic pathology or toxicity and no report of liver weight at termination of the study.
34 However, by the end of a 2-year cancer bioassay, liver tumor induction can be a significant
35 factor in any changes in liver weight. No treatment-related increase in necrosis in the liver was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-234 DRAFT—DO NOT CITE OR QUOTE
-------
1 observed in mice. A slight increase in the incidence of focal necrosis was noted for TCE-
2 exposed male mice (8 vs. 2% in control) with a slight reduction in fatty metamorphosis in
3 treated male mice (0 treated vs. 2 control animals) and in female mice a slight increase in focal
4 inflammation (29 vs. 19% of animals) and no other changes. Therefore, this study did not show
5 concurrent evidence of liver toxicity but did show TCE-induced neoplasia after 2 years of TCE
6 exposure in mice. The administration of TCE was reported to cause earlier expression of tumors
7 as the first animals with carcinomas were 57 weeks for TCE-exposed animals and 75 weeks for
8 control male mice.
9 The NTP (1990) study reported that TCE exposure was associated with increased
10 incidence of hepatocellular carcinoma (tumors with markedly abnormal cytology and
11 architecture) in male and female mice. Hepatocellular adenomas were described as
12 circumscribed areas of distinctive hepatic parenchymal cells with a perimeter of normal
13 appearing parenchyma in which there were areas that appeared to be undergoing compression
14 from expansion of the tumor. Mitotic figures were sparse or absent but the tumors lacked
15 typical lobular organization. Hepatocellular carcinomas had markedly abnormal cytology and
16 architecture with abnormalities in cytology cited as including increased cell size, decreased cell
17 size, cytoplasmic eosinophilia, cytoplasmic basophilia, cytoplasmic vacuolization, cytoplasmic
18 hyaline bodies and variations in nuclear appearance. Furthermore, in many instances several or
19 all of the abnormalities were present in different areas of the tumor and variations in architecture
20 with some of the hepatocellular carcinomas having areas of trabecular organization. Mitosis
21 was variable in amount and location. Therefore, the phenotype of tumors reported from TCE
22 exposure was heterogeneous in appearance between and within tumors.
23 For rats, the NTP (1990) study reported no treatment-related non-neoplastic liver lesions
24 in males and a decrease in basophilic cytological change reported from TCE-exposure in female
25 rats. The results for detecting a carcinogenic response in rats were considered to be equivocal
26 because both groups receiving TCE showed significantly reduced survival compared to vehicle
27 controls and because of a high rate (e.g., 20% of the animals in the high-dose group) of death by
28 gavage error.
29 The NTP (1988) study of TCE exposure in four strains of rats to "diisopropylamine-
30 stabilized TCE" was also considered inadequate for either comparing or assessing TCE-induced
31 carcinogenesis in these strains of rats because of chemically induced toxicity, reduced survival,
32 and incomplete documentation of experimental data. TCE gavage exposures of 0, 500 or
33 1,000 mg/kg per day (5 days per week, for 103 weeks) male and female rats was also marked by
34 a large number of accidental deaths (e.g., for high-dose male Marshal rats 25 animals were
35 accidentally killed). Results from a 13-week study were briefly mentioned in the report and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-23 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 indicated exposure levels of 62.5-2,000 mg/kg TCE were not associated with decreased survival
2 (with the exception of 3 male August rats receiving 2,000 mg/kg TCE) and that the
3 administration of the chemical for 13 weeks was not associated with histopathological changes.
4 In regard to evidence of liver toxicity, the 2-year study of TCE exposure reported no evidence of
5 TCE-induced liver toxicity described as non-neoplastic changes ACT, August, Marshal, and
6 Osborne-Mendel rats. Interestingly, for the control animals of these four strains there was, in
7 general, a low background level of focal necrosis in the liver of both genders. In summary, the
8 negative results in this bioassay are confounded by the killing of a large portion of the animals
9 accidently by experimental error but TCE-induced overt liver toxicity was not reported.
10 Maltoni et al. (1986) reported the results of several studies of TCE via inhalation and
11 gavage in mice and rats. A large number of animals were used in the treatment groups but the
12 focus of the study was detection of a neoplastic response with only a generalized description of
13 tumor pathology phenotype given and limited reporting of non-neoplastic changes in the liver.
14 Accidental death by gavage error was reported not to occur in this study. In regards to effects of
15 TCE exposure on survival, "a nonsignificant excess in mortality" correlated to TCE treatment
16 was observed only in female rats (treated by ingestion with the compound) and in male B6C3F1
17 mice. TCE-induced effects on body weight were reported to be absent in mice except for one
18 experiment (BT 306 bis) in which a slight nondose correlated decrease was found in exposed
19 animals. "Hepatoma" was the term used to describe all malignant tumors of hepatic cells, of
20 different subhistotypes, and of various degrees of malignancy and were reported to be unique or
21 multiple, and have different sizes (usually detected grossly at necropsy) from TCE exposure. In
22 regard to phenotype tumors were described as usual type observed in Swiss and B6C3F1 mice,
23 as well as in other mouse strains, either untreated or treated with hepatocarcinogens and to
24 frequently have medullary (solid), trabecular, and pleomorphic (usually anaplastic) patterns.
25 Swiss mice from this laboratory were reported to have a low incidence of hepatomas without
26 treatment (1%). The relatively larger number of animals used in this bioassay (n = 90 to 100), in
27 comparison to NTP standard assays, allows for a greater power to detect a response.
28 TCE exposure for 8 weeks via inhalation at 100 ppm or 600 ppm may have been
29 associated with a small increase in liver tumors in male mice in comparison to concurrent
30 controls during the life span of the animals. In Swiss mice exposed to TCE via inhalation for
31 78 weeks there a reported increase in hepatomas associated with TCE treatment that was dose-
32 related in male but not female Swiss mice. In B6C3F1 mice exposed via inhalation to TCE for
33 78 weeks, the results from one experiment indicated a greater increase in liver cancer in females
34 than male mice but in a second experiment in males there was a TCE-exposure associated
35 increase in hepatomas. Although the mice were supposed to be of the same strain, the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-236 DRAFT—DO NOT CITE OR QUOTE
-------
1 background level of liver cancer was significantly different in male mice. The finding of
2 differences in response in animals of the same strain but from differing sources has also been
3 reported in other studies for other endpoints (see Section E.3.1.2). However, for both groups of
4 male B6C3F1 mice the background rate of liver tumors over the lifetime of the mice was less
5 than 20%.
6 For rats, there were 4 liver angiosarcomas reported (1 in a control male rat, 1 both in a
7 TCE-exposed male and female at 600 ppm TCE for 8 weeks, and 1 in a female rat exposed to
8 600 ppm TCE for 104 weeks) but the specific results for incidences of hepatocellular
9 "hepatomas" in treated and control rats were not given. Although the Maltoni et al. (1986)
10 concluded that the small number was not treatment-related, the findings were brought forward
11 because of the extreme rarity of this tumor in control S-D rats, untreated or treated with vehicle
12 materials. In rats treated for 104 weeks, there was no report of a TCE treatment-related increase
13 in liver cancer in rats. This study only presented data for positive findings so it did not give the
14 background or treatment-related findings in rats for liver tumors in this study. Thus, the extent
15 of background tumors and sensitivity for this endpoint cannot be determined. Of note is that the
16 S-D strain used in this study was also noted in the Fukuda et al. (1983) study to be relatively
17 insensitive for spontaneous liver cancer and to also be negative for TCE-induced hepatocellular
18 liver cancer induction in rats. However, like Fukuda et al. (1983) and Henschler et al. (1980),
19 that reported rare biliary tumors in insensitive strains of rat for hepatocellular tumors, Maltoni et
20 al. (1986) reported a relatively rare tumor type, angiosarcoma, after TCE exposure in a relatively
21 insensitive strain for "hepatomas." As noted above, many of the rat studies were limited by
22 premature mortality due to gavage error or premature mortality (Henschler et al., 1980; NCI,
23 1976; NTP, 1990, 1988), which was reported not occur in Maltoni et al. (1986).
24 There were other reports of TCE carcinogenicity in mice from chronic exposures that
25 were focused primarily on detection of liver tumors with limited reporting of tumor phenotype
26 or non-neoplastic pathology. Herren-Freund et al. (1987) reported that male B6C3 Fl mice
27 given 40 mg/L TCE in drinking water had increased tumor response after 61 weeks of exposure.
28 However, concentrations of TCE fell by about !/2 at this dose of TCE during the twice a week
29 change in drinking water solution so the actual dose of TCE the animals received was less than
30 40 mg/L. The percent liver/body weight was reported to be similar for control and TCE-
31 exposed mice at the end of treatment. However, despite difficulties in establishing accurately
32 the dose received, an increase in adenomas per animal and an increase in the number of animals
33 with hepatocellular carcinomas were reported to be associated with TCE exposure after 61
34 weeks of exposure and without apparent hepatomegaly. Anna et al. (1994) reported tumor
35 incidences for male B6C3F1 mice receiving 800 mg/kg/d TCE via gavage (5 days/week for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-237 DRAFT—DO NOT CITE OR QUOTE
-------
1 76 weeks). All TCE-treated mice were reported to be alive after 76 weeks of treatment.
2 Although the control group contained a mixture of exposure durations (76-134 weeks) and
3 concurrent controls had a very small number of animals, TCE-treatment appeared to increase the
4 number of animals with adenomas, the mean number of adenomas and carcinomas, but with no
5 concurrent TCE-induced cytotoxicity.
6
7 E.2.4.4. Summary of Results For Subchronic and Chronic Effects of Dichloroacetic Acid
8 (DCA) and Trichloroacetic Acid (TCA): Comparisons With Trichloroethylene
9 (TCE)
10 There are no similar studies for TCA and DCA conduced at 6 weeks and with the range
11 of concentrations examined in Buben and O'Flaherty (1985) for TCE. In general, many studies
12 of DCA and TCA have been conducted at few and high concentrations, with shortened durations
13 of exposure, and varying and low numbers of animals to examine primarily a liver tumor
14 response in mice. However, the analyses presented in Section E.2.4.2 gives comparisons of
15 administered TCA and DCA dose-responses for liver weight increases for a number of studies in
16 combination as well as comparing such dose-responses to that of TCE and its oxidative
17 metabolism. As stated above, many subchronic studies of DCA and TCA have focused on
18 elucidating a relationship between dose and hypothesized events that may be indicators of
19 carcinogenic potential that have been described in chronic studies with a focus on indicators of
20 peroxisome proliferation and DNA synthesis. Many chronic studies have focused on the nature
21 of the DCA and TCA carcinogenic response in mouse liver through examination of the tumors
22 induced.
23 Most all of the chronic studies for DCA and TCA have been carried out in mice. As the
24 database for examination of the ability of TCE to induce liver tumors in rats includes several
25 studies that have been limited in ability determine a carcinogenic response in the liver, the
26 database for DCA and TCA in rats is even more limited. For TCA, the only available study in
27 rats (DeAngelo et al., 1997) has been frequently cited in the literature to indicate a lack of
28 response in this species for TCA-induced liver tumors. Although reporting an apparent dose-
29 related increase in multiplicity of adenomas and an increase in carcinomas over control at the
30 highest dose, DeAngelo et al. (1997) use such a low number of animals per treatment group
31 (n = 20-24) that the ability of this study to determine a statistically significant increase in tumor
32 response and to be able to determine that there was no treatment-related effect are limited. A
33 power calculation of the study shows that the type II error, which should be >50%, was less than
34 8% probability for incidence and multiplicity of all tumors at all exposure DCA concentrations
35 with the exception of the incidence of adenomas and adenomas and carcinomas for 0.5 g/L
36 treatment group (58%) in which there was an increased in adenomas reported over control
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-23 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 (15 vs. 4%) that was the same for adenomas and carcinomas combined. Therefore, the designed
2 experiment could accept a false null hypothesis and erroneously conclude that there is no
3 response due to TCA treatment. Thus, while suggesting a lower response than for mice for liver
4 tumor induction, it is inconclusive for determination of whether TCA induces a carcinogenic
5 response in the liver of rats.
6 For DCA, there are two reported long-term studies in rats (DeAngelo et al., 1996;
7 Richmond et al., 1995) that appear to have reported the majority of their results from the same
8 data set and which consequently were subject to similar design limitations and DCA-induced
9 neurotoxicity in this species. DeAngelo et al. (1996) reported increased hepatocellular
10 adenomas and carcinomas in male F344 rats exposed for 2 years. However, the data from
11 exposure concentrations at a 5 g/L dose had to be discarded and the 2.5 g/L DCA dose had to be
12 continuously lowered during the study due to neurotoxicity. There was a DCA-induced
13 increased in adenomas and carcinomas combined reported for the 0.5 g/L DCA (24.1 vs. 4.4%
14 adenomas and carcinomas combined in treated vs. controls) and an increase at a variable dose
15 started at 2.5 g/L DCA and continuously lowered (28.6 vs. 3.0% adenomas and carcinomas
16 combined in treated vs. controls). Only combined incidences of adenomas and carcinomas for
17 the 0.5 g/L DCA exposure group was reported to be statistically significant by the authors
18 although the incidence of adenomas was 17.2 versus 4% in treated versus control rats.
19 Hepatocellular tumor multiplicity was reported to be increased in the 0.5 g/L DCA group
20 (0.31 adenomas and carcinomas/animal in treated vs. 0.04 in control rats) but was reported by
21 the authors to not be statistically significant. At the starting dose of 2.5 g/L that was
22 continuously lowered due to neurotoxicity, the increased multiplicity of hepatocellular
23 carcinomas was reported by the authors to be to be statistically significant
24 (0.25 carcinomas/animals vs. 0.03 in control) as well as the multiplicity of combined adenomas
25 and carcinomas (0.36 adenomas and carcinomas/animals vs. 0.03 in control rats). Issues that
26 affect the ability to determine the nature of the dose-response for this study include (1) the use
27 of a small number of animals (n = 23, n =21 and n = 23 at final sacrifice for the 2.0 g/L NaCl
28 control, 0.05 and 0.5 g/L treatment groups) that limit the power of the study to both determine
29 statistically significant responses and to determine that there are not treatment-related effects
30 (i.e., power) (2) apparent addition of animals for tumor analysis not present at final sacrifice
31 (i.e., 0.05 and 0.5 g/L treatment groups), and (3) most of all, the lack of a consistent dose for the
32 2.5 g/L DCA exposed animals. Similar issues are present for the study of Richmond et al.
33 (1995) which was conducted by the same authors as DeAngelo et al. (1996) and appeared to be
34 the same data set. The Richmond et al. (1995) data for the 2 g/L NaCl. 0.05 g/L DCA and
35 0.5 g/L DCA exposure groups were the same data set reported by DeAngelo et al. (1996) for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-239 DRAFT—DO NOT CITE OR QUOTE
-------
1 these groups. Additional data was reported for F344 rats administered and 2.5 g/L DC A that,
2 due to hind-limb paralysis, were sacrificed 60 weeks (DeAngelo et al., 1996). Tumor
3 multiplicity was not reported by the authors. There was a small difference in reports of the
4 results between the two studies for the same data for the 0.5 g/L DC A group in which Richmond
5 et al. (1995) reported a 21% incidence of adenomas and DeAngelo et al. (1996) reported a
6 17.2% incidence. The authors did not report any of the results of DCA-induced increases of
7 adenomas and carcinomas to be statistically significant. The same issues discussed above for
8 DeAngelo et al. (1996) apply to this study. Similar to the DeAngelo study of TCA in rats
9 (DeAngelo et al., 1997) the study of DCA exposure in rats reported by DeAngelo et al. (1996)
10 and Richmond et al. (1995), the use of small numbers of rats limits the detection of treatment-
11 related effects and the ability to determine whether there was no treatment related effects
12 (Type II error), especially at the low concentrations of DCA exposure.
13 For mice the data for both DCA and TCA is much more extensive and has shown that
14 both DCA and TCA induced liver tumors in mice. Many of the studies are for relatively high
15 concentrations of DCA or TCA, have been conducted for a year or less, and have focused on the
16 nature of tumors induced to ascertain potential MO As and to make inferences as to whether
17 TCE-induced tumors in mice are similar. As shown previously in Section E.2.4.2, the dose-
18 response curves for increased liver weight for TCE administration in male mice are more similar
19 to those for DCA administration and TCE oxidative metabolism than for direct TCA
20 administration. There are two studies in male B6C3F1 mice that attempt to examine multiple
21 concentrations of DCA and TCA for 2-year studies (DeAngelo et al., 1999, 2008) at doses that
22 do not induce cytotoxicity and attempt to relate them to subchronic changes and peroxisomal
23 enzyme induction. However, the DeAngelo et al. (2008) study was carried out in B6C3F1 mice
24 that were of large size and prone to liver cancer and premature mortality limiting its use for the
25 determination of TCA-dose response in a 2-year bioassay. One study in female B6C3F1 mice
26 describes the dose-response for liver tumor induction at a range of DCA and TCA
27 concentrations after 51 or 82 weeks (Pereira, 1996) with a focus on the type of tumor each
28 compound produced.
29 DeAngelo et al. (1999) conducted a study of DCA exposure to determine a dose
30 response for the hepatocarcinogenicity of DCA in male B6C3F1 mice over a lifetime exposure
31 and especially at concentrations that did not illicit cytotoxicity or were for abbreviated exposure
32 durations. DeAngelo et al. (1999) used 0.05, 0.5, 1.0, 2.0, and 3.5 g/L exposure concentrations
33 of DCA in their 100-week drinking water study. The number of animals at final sacrifice was
34 generally low in the DCA treatment groups and variable (i.e., n = 50, n = 33, n = 24, n = 32,
35 n= 14, and n = 8 for control, 0.05, 0.5, 1, 2.0, and 3.5 g/L DCA exposure groups). It was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-240 DRAFT—DO NOT CITE OR QUOTE
-------
1 apparent that animals that died unscheduled deaths between weeks 79 and 100 were included in
2 data reported for 100 weeks. Although the authors did not report how many animals were
3 included in the 100-week results, it appeared that the number was no greater than 1 for the
4 control, 0.05, and 0.5 exposure groups and varied between 3 and 7 for the higher DCA exposure
5 groups. The multiplicity or number of hepatocellular carcinomas/animals was reported to be
6 significantly increased over controls in a dose-related manner at all DCA treatments including
7 0.05 g/L DCA, and a NOEL reported not to be observed by the authors (i.e., 0.28, 0.58, 0.68,
8 1.29, 2.47, and 2.90 hepatocellular carcinomas/animal for control, 0.05, 0.5, 1.0, 2.0, and 3.5 g/L
9 DCA). Between the 0.5 and 3.5 g/L exposure concentrations of DCA the magnitude of increase
10 in multiplicity was similar to the increases in magnitude in dose. The incidence of
11 hepatocellular carcinomas were reported to be increased at all doses as well but not reported to
12 be statistically significant at the 0.05 g/L exposure concentration. However, given that the
13 number of mice examined for this response (n = 33), the power of the experiment at this dose
14 was only 16.9% to be able to determine that there was not a treatment related effect. The
15 authors did not report the incidence or multiplicity of adenomas for the 0.05 g/L exposure group
16 in the study and neither did they report the incidence or multiplicity of adenomas and
17 carcinomas in combination. For the animals surviving from 79 to 100 weeks of exposure, the
18 incidence and multiplicity of adenomas peaked at 1 g/L while hepatocellular carcinomas
19 continued to increase at the higher doses. This would be expected where some portion of the
20 adenomas would either regress or progress to carcinomas at the higher doses.
21 DeAngelo et al. (1999) reported that peroxisome proliferation was significantly
22 increased at 3.5 g/L DCA only at 26 weeks, not correlated with tumor response, and to not be
23 increased at either 0.05 or 0.5 g/L treatments. The authors concluded that DCA-induced
24 carcinogenesis was not dependent on peroxisome proliferation or chemically sustained
25 proliferation, as measured by DNA synthesis. DeAngelo et al. (1999) reported not only a dose-
26 related increase in DCA-induced liver tumors but also a decrease in time-to-tumor associated
27 with DCA exposure at the lowest levels examined. In regards to cytotoxicity there appeared to
28 be a treatment but not dose-related increase in hepatocellular necrosis that did not involve most
29 of the liver from 1 to 3.5 g/L DCA exposures for 26 weeks of exposure that decreased by
30 52 weeks with no necrosis observed at the 0.5 g/L DCA treatment for any exposure period.
31 Hepatomegaly was reported to be absent by 100 weeks of exposure at the 0.05 and
32 0.5 g/L exposures while there was an increase in tumor burden reported. However, slight
33 hepatomegaly was present by 26 weeks in the 0.5 g/L group and decreased with time. Not only
34 did the increase in multiplicity of hepatocellular carcinomas increase proportionally with DCA
35 exposure concentration after 79-100 weeks of exposure, but so did the increases in percent
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-241 DRAFT—DO NOT CITE OR QUOTE
-------
1 liver/body weight. DeAngelo et al. (1999) presented a figure comparing the number of
2 hepatocellular carcinomas/animal at 100 weeks compared with the percent liver/body weight at
3 26 weeks that showed a linear correlation (r = 0.9977) while peroxisome proliferation and
4 DNA synthesis did not correlate with tumor induction profiles. The proportional increase in
5 liver weight with DCA exposure was also reported for shorter durations of exposure as noted in
6 Section E.2.4.2. The findings of the study illustrates the importance of examining multiple
7 exposure levels at lower concentrations, at longer durations of exposure and with an adequate
8 number of animals to determine the nature of a carcinogenic response. Although Carter et al.
9 (1995) suggested that there is evidence of DCA-induced cytotoxicity (e.g., loss of cell
10 membranes and apparent apoptosis) at higher levels, the 0.5 g/L exposure concentration has
11 been shown by DeAngelo et al. (1999) to increase hepatocellular tumors after 100 weeks of
12 treatment without concurrent peroxisome proliferation or cytotoxicity in mice.
13 As noted in detail in E. 2.3.2.13, DeAngelo et al. (2008) exposed male B6C3F1 mice to
14 neutralized TCA in drinking water to male B6C3 Fl mice in three studies. Rather than using
15 5 exposure levels that were generally 2-fold apart, as was done in DeAngelo et al. (1999) for
16 DCA, DeAngelo et al. (2008) studied only 3 doses of TCA that were an order of magnitude
17 apart which limits the elucidation of the shape of the dose-response curve. In addition
18 DeAngelo et al. (2008) contained 2 studies, each conducted in a separate laboratories, for the
19 104-week data so that the two lower doses were studied in one study and the highest dose in
20 another. The first study was conducted using 2 g/L NaCl, or 0.05, 0.5, or 5 g/L TCA in drinking
21 water for 60 weeks (Study #1) while the other two were conducted for a period of 104 weeks
22 (Study #2 with 2.5 g/L neutralized acetic acid or 4.5 g/L TCA exposure groups and Study #3
23 with deionized water, 0.05 and 0.5 g/L TCA exposure groups). In the studies reported in
24 DeAngelo et al. (2008) a small number of animals has been used for the determination of a
25 tumor response (~n = 30 at final necropsy), but for the data for liver weight or PCO activity at
26 interim sacrifices the number was even smaller (n = 5). The percent liver/body weight changes
27 at 4 weeks in Study #1 have been included in the analysis for all TCA data in Section E.2.4.2,
28 and are consistent with that data. Although there was a 10-fold difference in TCA exposure
29 concentration, there was a 9, 16, and 35% increase in liver weight over control for the 0.05, 0.5,
30 and 5 g/L TCA exposures. PCO activity varied 2.7-fold as baseline controls but the increase in
31 PCO activity at 4 weeks was 1.3-, 2.4-, and 5.3-fold of control for the 0.05, 0.5, and 5 g/L TCA
32 exposure groups in Study #1. The incidence data for adenomas observed at 60 weeks was 2.1-,
33 3.0-, and 5.4-fold of control values and the fold increases in multiplicity were similar after 0.05,
34 0.5, and 5.0 g/L TCA. Thus, in general the dose-response for TCA-induced liver weight
35 increases at 4 weeks was similar to the magnitude of induction of adenomas at 60 weeks. Such
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-242 DRAFT—DO NOT CITE OR QUOTE
-------
1 a result is more consistent with the ability of TCA to induce tumors and increases in liver weight
2 at low doses with little change with increasing dose as shown by this study and the combined
3 data for TCA liver weight induction by administered TCA presented in Section E.2.4.2.
4 While the 104-week data from Study's #2 and #3 could have been more valuable for
5 determination of the dose-response as it would have allowed enough time for full tumor
6 expression, serious issues are apparent for Study #3, which was reported to have a 64%
7 incidence rate of adenomas and carcinomas for controls while that of Study #2 was 12%. As
8 stated in Section E.2.3.2.13, the mice in Study #3 were of larger size than those of either Study
9 #1 or #2 and the large background rate of tumors reported is consistent with mice of these size
10 (Leakey et al., 2003b). However, the large background rate and increased mortality for these
11 mice limit their use for determining the nature of the dose-response for TCA liver
12 carcinogenicity. Examination of the data for treatment groups shows that there was no
13 difference in any of the results between the 0.5 g/L (Study #3) and 5 g/L (Study #2) TCA
14 exposure groups (i.e., adenoma, carcinoma, and combinations of adenoma and carcinoma
15 incidence and multiplicity) for 104 weeks of exposure. For these same exposure groups, but at
16 60 weeks of exposure (Study #1), there was a 2-fold increase in multiplicity for adenomas, and
17 for adenomas and carcinomas combined between the 0.5 and 5.0 g/L TCA exposure groups. At
18 the two lowest doses of 0.05 and 0.5 g/L TCA from Study #3 in the large tumor prone mice, the
19 differences in the incidences and multiplicities for all tumors were 2-fold at 104 weeks. These
20 results are consistent with (1) the two highest exposure levels reaching a plateau of response
21 after a long enough duration of exposure for full expression of the tumors (i.e., -90% of animals
22 having liver tumors at the 0.5 and 5 g/L exposures) with the additional tumors observed in a
23 tumor-prone paradigm. Thus, without use of the 0.05 and 0.5 g/L TCA data from Study #3,
24 only the 4.5 g/L TCA data from Study #2 can be used for determination of the TCA cancer
25 response in a 2-year bioassay.
26 To put the 64% incidence data for carcinomas and adenomas reported in DeAngelo et al.
27 (2008) for the control group of Study #3 in context, other studies cited in this review for male
28 B6C3F1 mice show a much lower incidence in liver tumors with: (1) NCI (1976) study of TCE
29 reporting a colony control level of 6.5% for vehicle and 7.1% incidence of hepatocellular
30 carcinomas for untreated male B6C3F1 mice (n = 70-77) at 78 weeks, (2) Herren-Freund et al.
31 (1987) reporting a 9% incidence of adenomas in control male B6C3F1 mice with a multiplicity
32 of 0.09 ± 0.06 and no carcinomas (n = 22) at 61 weeks, (3) NTP (1990) reporting an incidence
33 of 14.6% adenomas and 16.6% carcinomas in male B6C3F1 mice after 103 weeks (n = 48), and
34 (4) Maltoni et al. (1986) reporting that B6C3F1 male mice from the "NCI source" had a 1.1%
35 incidence of "hepatoma" (carcinomas and adenomas) and those from "Charles River Co." had a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-243 DRAFT—DO NOT CITE OR QUOTE
-------
1 18.9% incidence of "hepatoma" during the entire lifetime of the mice (n = 90 per group). The
2 importance of examining an adequate number of control or treated animals before confidence
3 can be placed in those results in illustrated by Anna et al. (1994) in which at 76 weeks
4 3/10 control male B6C3F1 mice that were untreated and 2/10 control animals given corn oil
5 were reported to have adenomas but from 76 to 134 weeks, 4/32 mice were reported to have
6 adenomas (multiplicity of 0.13 ± 0.06) and 4/32 mice were reported to have carcinomas
7 (multiplicity of 0.12 ± 0.06). Thus, the reported combined incidence of carcinomas and
8 adenomas of 64% reported by DeAngelo et al. (2008) for the control mice of Study #3, not only
9 is inconsistent and much higher than those reported in Studies #1 and #2, but also much higher
10 than reported in a number of other studies of TCE.
11 Trying to determine a correspondence with either liver weight increases or increases in
12 PCO activity after shorter periods of exposure will be depend whether data reported in Study #3
13 in the 104 week studies can be used. DeAngelo et al. (2008) report a regression analyses that
14 compare "percent of hepatocellular neoplasia," indicated by tumor multiplicity, with TCA dose,
15 represented by estimations of the TCA dose in mg/kg/d, and with PCO activity for the 60-week
16 and 104-week data. Whether adenomas and carcinomas combined or individual tumor type
17 were used in these analysis was not reported by the authors. Concerns arise also from
18 comparing PCO activity at the end of the experiments, when there was already a significant
19 tumor response, rather than at earlier time points. Such PCO data may not be useful as an
20 indicator key event in tumorigenesis when tumors are already present. In addition regression
21 analyses of these data are difficult to interpret because of the dose spacing of these experiments
22 as the control and 5 g/L exposure levels will basically determine the shape of the dose-response
23 curve. The 0.05 and 0.5 g/L exposure levels are close to the control value in comparison to the
24 5 g/L exposure level, the dose response appears to be linear between control and the 5.0 g/L
25 value with the two lowest doses not affectly changing the slope of the line (i.e., "leveraging" the
26 regression). Thus, the value of these analyses is limited by (1) use of data from Study #3 in a
27 tumor prone mouse that is not comparable to those used in Studies #1 and #2, (2) the
28 appropriateness of using PCO values from later time points and the variability in PCO control
29 values (3) the uncertainty of the effects of palatability on the 5 g/L TCA results which were
30 reported in one study to reduce drinking water consumption, and (4) the dose-spacing of the
31 experiment.
32 DeAngelo et al. (2008) attempt to identify a NOEL for tumorigenicity using tumor
33 multiplicity data and estimated TCA dose. However, it is not an appropriate descriptor for these
34 data, especially given that "statistical significance" of the tumor response is the determinant
35 used by the authors to support the conclusions regarding a dose in which there is no TCA-
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-244 DRAFT—DO NOT CITE OR QUOTE
-------
1 induced effect. Due to issues related to the appropriateness of use of the concurrent control in
2 Study #3, only the 60-week experiment (i.e., Study #1) is useful for the determination of tumor
3 dose-response. Not only is there not allowance for full expression of a tumor response at the
4 60-week time point but a power calculation of the 60-week study shows that the type II error,
5 which should be >50% and thus, greater than the chances of "flipping a coin," was 41 and 71%
6 for incidence and 7 and 15% for multiplicity of adenomas for the 0.05 and 0.5 g/L TCA
7 exposure groups. For the combination of adenomas and carcinomas, the power calculation was
8 8 and 92% for incidence and 6 and 56% for multiplicity at 0.05 and 0.5 g/L TCA exposure.
9 Therefore, the designed experiment could accept a false null hypothesis, especially in terms of
10 tumor multiplicity, at the lower exposure doses and erroneously conclude that there is no
11 response due to TCA treatment.
12 Pereira (1996) examined the tumor induction in female B6C3 Fl mice and demonstrate
13 that foci, adenoma, and carcinoma development in mice are dependent on duration of exposure,
14 or period of observation in the case of controls, for full expression of a carcinogenic response.
15 In control female mice a 360- versus 576-day observation period showed that at 360 days no
16 foci or carcinomas and only 2.5% of animals had adenomas whereas by 576 days of observation,
17 11% had foci, 2% adenomas, and 2% had carcinomas. For DCA and TCA treatments, foci,
18 adenomas, and carcinoma incidence and multiplicity did not reach full expression until
19 82 weeks at the 3 doses employed (2.58 g/L DCA, 0.86 g/L DCA, 0.26 g/L DCA, 3.27 g/L
20 TCA, 1.1.0 g/L TCA, and 0.33 g/L TCA). Although the numbers of animals were relatively low
21 and variable at the two highest doses (18-28 mice) there were 50-53 mice studied at the lowest
22 dose level and 90 animals studied in the control group. The results of Pereira (1996) show that
23 not only were the incidence of mice with foci, adenoma, and carcinomas greatly increased with
24 duration of exposure, but that concentration also affected the nature and magnitude of the
25 response in female mice. At 2.86 g/L, 0.86 g/L, 0.26 g/L DCA exposures and controls, after 82
26 weeks the incidence of adenomas in female B6C3 Fl mice was reported to be 84.2, 25.0, 6.0,
27 and 2.2%, respectively, and carcinomas to be 26.3, 3.6, 0, and 2.2%, respectively. For the
28 multiplicity or number of tumors/animal at these same exposure levels of DCA, the multiplicity
29 was reported to be 5.58, 0.32, 0.06, and 0.02 adenomas/animal, and 0.37, 0.04, 0, and
30 0.02 carcinomas/animal. Thus, for DCA exposure in female mice, for ~3-fold increases in DCA
31 exposure concentration, after 82 weeks of exposure there was a similar magnitude of increase in
32 adenomas incidence with much greater increases in multiplicity. For hepatocellular carcinoma
33 induction, there was no increase in the incidence or multiplicity or carcinomas between the
34 control and 0.33 g/L DCA dose. At 3.27, 1.10, and 0.33 g/L TCA and controls, after 82 weeks
35 the incidence of adenomas in female B6C3F1 mice was reported to be 38.9, 11.1, 7.6, and 2.2%,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-245 DRAFT—DO NOT CITE OR QUOTE
-------
1 respectively, and carcinomas to be 27.8, 18.5, 0, and 2.2%, respectively. At these same
2 exposure levels of TCA, the multiplicity was reported to be 0.61, 0.11, 0.08, and
3 0.02 adenomas/animal, and 0.39, 0.22, 0, and 0.02 carcinomas/animal, respectively. Thus, for
4 TCA, the incidences of adenomas were lower at the two highest doses than DCA and the
5 ~3-fold differences in dose between the two lowest doses only resulted in -50% increase in
6 incidences of adenomas. For incidence of carcinomas the ~3-fold difference in dose between
7 the two highest doses only resulted in -50% increase in carcinoma incidence. A similar pattern
8 was reported for multiplicity after TCA exposure. Foci were also examined and, in general.,
9 were similar to adenomas regarding incidence and multiplicity. Thus, the dose-response curve
10 for tumor induction in female mice differed between DCA and TCA after 82 weeks of exposure
11 with TCA having a much less steep dose-response curve than DCA. This is consistent with the
12 pattern of liver weight increases reported for male B6C3F1 mice in Section E.2.4.2.
13 DeAngelo et al. (1999) report a linear increase in incidence and multiplicity of
14 hepatocellular carcinomas that is proportional to dose and as well as proportional to the
15 magnitude of liver weight increase from subchronic exposure to DCA. However, the studies of
16 DeAngelo et al. (2008) and Pereira (1996) are suggestive that TCA induced increase in tumor
17 incidence are less proportional to increases in dose as are liver weight increases from subchronic
18 exposure. Given that TCE subchronic exposure also induced an increase in liver weight that
19 was proportional to dose (i.e., similar to DCA but not TCA), it is of interest as to whether the
20 dose-response for TCE induced liver cancer in mice was similar. The database for TCE, while
21 consistently showing a induction of liver tumors in mice, is very limited for making inferences
22 regarding the shape of the dose-response curve. For many of these experiments multiplicity was
23 not given only liver tumor incidence. NTP (1990), Bull et al. (2002), Anna et al. (1994)
24 conducted gavage experiments in which they only tested one dose of-1,000 mg/kg/d TCE. NCI
25 (1976) tested 2 doses that were adjusted during exposure to an average of 1,169 mg/kg/d and
26 2,339 mg/kg/d in male mice with only 2-fold dose spacing in only 2 doses tested. Maltoni et al.
27 (1988) conducted inhalation experiments in 2 sets of B6C3F1 mice and one set of Swiss mice at
28 3 exposure concentrations that were 3-fold apart in magnitude between the low and mid-dose
29 and 2-fold apart in magnitude between the mid- and high-dose. However, for one experiment in
30 male B6C3F1 mice, the mice fought and suffered premature mortality and for two the
31 experiments in B6C3F1 mice, although using the same strain, the mice were obtained from
32 differing sources with very different background liver tumor levels. For the Maltoni et al.
33 (1988) study a general descriptor of "hepatoma" was used for liver neoplasia rather than
34 describing hepatocellular adenomas and carcinomas so that comparison of that data with those
35 from other experiments is difficult. More importantly, while the number of adenomas and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-246 DRAFT—DO NOT CITE OR QUOTE
-------
1 carcinomas may be the same between treatments or durations of exposure, the number of
2 adenomas may decrease as the number of carcinomas increase during the course of tumor
3 progression. Such information is lost by using only a hepatoma descriptor. Maltoni et al.
4 (1988) did not report an increase over control for 100 ppm TCE for the Swiss group and one of
5 the B6C3F1 groups and only a slight increase (1.12-fold) in the second B6C3F1 group. At
6 300 ppm TCE exposure, the incidences of hepatoma were 2-fold of control values for the Swiss,
7 4-fold of control for group of B6C3F1 mice, and 1.6-fold of control for the other group of
8 B6C3F1 mice. At 600 ppm TCE the incidences of hepatoma were 3.3-fold of control for the
9 Swiss group, 6.1-fold of control for one group of B6C3F1 mice, and 1.2-fold for the other group
10 of B6C3F1 mice. Thus, for each group of TCE exposed mice in the Maltoni et al. (1988)
11 inhalation study, the background levels of hepatomas and the shape of the dose-response curve
12 for TCE-hepatoma induction were variable. However, an average of the increases, in terms of
13 fold of control, between the 3 experiments gives a ~2.9-fold increase between the low- and mid-
14 dose (100 ppm and 300 ppm) and ~1.4-fold increase between the mid- and high-dose (300 ppm
15 and 600 pm) groups. Although such a comparison obviously has a high degree of uncertainty
16 associated with it, it suggests that the magnitude of TCE-induced hepatoma increases over
17 control is similar to the 3- and 2-fold difference in the magnitude of exposure concentrations
18 between these doses. Therefore, the increase in TCE-induced liver tumors would roughly
19 proportional to the magnitude of exposure dose. This result would be similar to the result for
20 the concordance of the increases in liver weight and exposure concentration observed 28-42 day
21 exposures to TCE (see Section E.2.4.2) using oral data from B6C3F1 and Swiss mice, and
22 inhalation data from NMRI mice. The available inhalation data for TCE induced liver weight
23 dose-response is from one study in a strain derived from Swiss mice (Kjellstrand et al., 1983b)
24 and was conducted in male and female mice with comparable doses of 75 ppm and 300 ppm
25 TCE. However, male mice of this strain exhibited decreased body weight at the 300 ppm level,
26 which can affect percent liver/body weight increases. The magnitude of TCE-induced increases
27 in liver weight between the 75 ppm and 300 ppm exposures were ~1.80-fold for males (1.50 vs.
28 1.90-fold of control liver weights) and 4.2-fold for females (1.27- vs. 2.14-fold of control liver
29 weight) in this strain. Female mice were examined in one study each of Swiss and B6C3F1
30 mice by Maltoni et al. (1988). Both the Swiss and B6C3F1 studies reported increases in
31 incidences of hepatomas over controls only at the 600 ppm TCE level in female mice indicating
32 less of a response than males. Similarly, the Kjellstrand et al. (1983b) data also showed less of a
33 response in females compared to males in terms TCE induction of liver weight at the 37 to
34 150 ppm range of exposure in NMRI strain. While the data for TCE dose-response of liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-247 DRAFT—DO NOT CITE OR QUOTE
-------
1 tumor induction is very limited, it is suggestive of a correlation of TCE-induced increases in
2 liver weight correlating liver tumor induction with a pattern that is dissimilar to that of TCA.
3 Of those experiments conducted at -1,000 mg/kg/d gavage dose of TCE in male
4 B6C3F1 mice for at least 79 weeks (Bull et al., 2002; NCI, 1976; Anna et al., 1994; NTP, 1990)
5 the control values were conducted in varying numbers of animals (some as low as n = 15, i.e.,
6 Bull et al., 2002) and with varying results. The incidence of hepatocellular carcinomas ranged
7 from 1.2 to 16.7% (NCI, 1976; Anna et al., 1994, NTP, 1990) and the incidence of adenomas
8 ranged from 1.2 to 14.6% (Anna et al., 1994; NTP, 1990) in control B6C3F1 mice. After
9 -1,000 mg/kg/d TCE treatment, the incidence of carcinomas ranged from 19.4 to 62%
10 (Bull et al., 2002; NCI, 1976; Anna et al., 1994; NTP, 1990) with 3 of the studies (NCI, 1976;
11 Anna et al., 1994; NTP, 1990) reporting a range of incidences between 42.8 to 62.0%). The
12 incidence of adenomas ranged from 28 to 66.7% (Bull et al., 2002; Anna et al., 1994; NTP,
13 1990). These data are illustrative of the variability between experiments to determine the
14 magnitude and nature of the TCE response in the same gender (male), strain (B6C3F1), time of
15 exposure (3/4 studies were for 76-79 weeks and 1 for 2 years duration), and roughly the same
16 dose (800-1,163 mg/kg/d TCE). Given, that the TCE-induced liver response, as measured by
17 liver weight increase, is highly correlated with total oxidative metabolism to a number of agents
18 that are hepatoactive agents and hepatocarcinogens, the variability in response from TCE
19 exposure would be expected to be greater than studies of exposure to a single metabolite such as
20 TCA or DC A.
21 Caldwell et al. (2008b) have commented on the limitations of experimental paradigms
22 used to study liver tumor induction by TCE metabolites and show that 51-week exposure
23 duration has consistently produced a tumor response for these chemicals, but with greater lesion
24 incidence and multiplicity at 82 weeks. As reported by DeAngelo et al. (1999) and Pereira
25 (1996), full expression of tumor induction in the mouse does not occur until 78 to 100 weeks of
26 DCA or TCA exposure, especially at lower concentrations. Thus, use of abbreviated exposure
27 durations and concurrently high exposure concentrations limits the ability of such experiments
28 to detect a treatment-related effect with the occurrence of additional toxicity not necessarily
29 associated with tumor-induction. Caldwell et al. (2008b) present a table that shows that the
30 differences in the ability of the studies to detect treatment-related effects could also be attributed
31 to a varying and low number of animals in some exposure groups and that because of the low
32 numbers of animals tested at higher exposures, the power to detect a statistically significant
33 change is very low and in fact for many of the endpoints is considerably less than "50%
34 chance." Table E-17 from Caldwell et al. (2008b) illustrates the importance of experimental
35 design and the limitations in many of the studies in the TCE metabolite database.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-248 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
Table E-17. Power calculations" for experimental design described in text,
using Pereira et al. as an example
Exposure concentration11 in female
B6C3F1 mice (Pereira, 1996;Pereira
and Phelps, 1996)
20.0 mmol/L NaCl (control) (82 wks)
2.58g/LDCA(82wks)
0.86g/LDCA(82wks)
0.26 g/L DCA (82 wks)
3.27g/LTCA(82wks)
1.10g/LTCA(82wks)
0.33 g/L TCA (82 wks)
Number
of
animals
90
19
28
50
18
27
53
Power
calculation
for foci
Null
hypothesis
0.03
0.74
0.99
0.15
0.60
0.93
Power
calculation for
adenomas
Null
hypothesis
0.03
0.20
0.98
0.09
0.64
0.91
Power
calculation for
carcinomas
Null hypothesis
0.13
0.91
-
0.14
0.3
-
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
aThe power calculations represent the probability of rejecting the null hypothesis when in fact the alternate
hypothesis is true for tumor multiplicity (i.e., the total number of lesions/number of animals). The higher the
power number calculated, the more confidence we have in the null hypothesis. Assumptions made included:
normal distribution for the fraction of tumors reported, null hypothesis represents what we expected the control
tumor fraction to be, the probability of a Type I error was set to 0.05, and the alternate hypothesis was set to four
times the null hypothesis value.
bConversion of mmol/L to g/L from the original reports of Pereira (1996) and Pereira and Phelps
(1996) is as follows: 20.0 mmol/L DCA = 2.58 g/L, 6.67 mmol/L DCA = 0.86 g/L, 2.0 mmol/L
= 0.26 g/L, 20.0 mmol/L TCA = 3.27 g/L, 6.67 mmol/L TCA =1.10 g/L, 2.0 mmol/L TCA =
0.33 g/L.
Bull et al. (1990) examined male and female B6C3F1 mice (age 37 days) exposed from
15 to 52 weeks to neutralized DCA and TCA (1 or 2 g/L) but tumor data were not suitable for
dose response. They reported effects of DCA and TCA exposure on liver weight and percent
liver/body changes that gave a pattern of hepatomegaly generally consistent with short-term
exposure studies. Only 10 female mice were examined at 52 weeks but the female mice were
reported to be as responsive as males at the exposure concentration tested. After 37 weeks of
treatment and then a cessation of exposure for 15 weeks, liver weights percent liver/body weight
were reported to be elevated over controls which Bull et al. (1990) partially attribute the
remaining increases in liver weight to the continued presence of hyperplastic nodules in the liver.
Macroscopically, livers treated with DCA were reported to have multifocal areas of necrosis and
frequent infiltration of lymphocytes on the surface and an interior of the liver. For TCA-treated
mice, similar necrotic lesions were reported but at such a low frequency that they were similar to
controls. Marked cytomegaly was reported from exposure to either 1 or 2 g/L DCA throughout
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-249 DRAFT—DO NOT CITE OR QUOTE
-------
1 the liver. Cell size was reported to be increased from TCA and DCA treatment with DCA
2 producing the greatest change. The 2 g/L TCA exposures were observed to have increased
3 accumulations of lipofuscin but no quantitative analysis was done. Photographs of light
4 microscopic sections, that were supposed to be representative of DCA and TCA treated livers at
5 2 g/L, showed such great hepatocellular hypertrophy from DCA treatment that sinusoids were
6 obscured. Such a degree of cytomegaly could have resulted in reduction of blood flow and
7 contributed to focal necrosis observed at this level of exposure.
8 As discussed in Sections E.3.2 and E.3.4.2.1, glycogen accumulation has been described
9 to be present in foci in both humans and animals as a result from exposure to a wide variety of
10 carcinogenic agents and predisposing conditions in animals and humans. Bull et al (1990)
11 reported that glycogen deposition was uniformly increased from 2 g/L DCA exposure with
12 photographs of TCA exposure showing slightly less glycogen staining than controls. However,
13 the abstract and statements in the paper suggest that there was increased PAS positive material
14 from TCA treatment that has caused confusion in the literature in this regard. Kato-Weinstein et
15 al. (2001) reported that in male B6C3F1 mice exposed to DCA and TCA, the DCA treatment
16 increased glycogen and TCA decreased glycogen content of the liver by using both chemical
17 measurement of glycogen in liver homogenates and by using ethanol-fixed sections stained with
18 PAS, a procedure designed to minimize glycogen loss. Kato-Weinstein et al. (2001) reported
19 that glycogen rich and poor cells were scattered without zonal distribution in male B6C3F1 mice
20 exposed to 2 g/L DCA for 8 weeks. For TCA treatments they reported centrilobular decreases in
21 glycogen and -25% decreases in whole liver by 3 g/L TCA. Kato-Weinstein et al. (2001)
22 reported whole liver glycogen to be increased ~1.50-fold of control (90 vs. 60 mg glycogen/g
23 liver) by 2 g/L DCA after 8 weeks exposure male B6C3F1 mice with a maximal level of
24 glycogen accumulation occurring after 4 weeks of DCA exposure. Pereira et al. (2004) reported
25 that after 8 weeks of exposure to 3.2 g/L DCA liver glycogen content was 2.20-fold of control
26 levels (155.7 vs. 52.4 mg glycogen/g liver) in female B6C3F1 mice. Thus, the baseline level of
27 glycogen content reported by (-60 mg/g) and the increase in glycogen after DCA exposure was
28 consistent between Kato-Weinstein et al. (2001) and Pereira et al. (2004). However, the increase
29 in liver weight reported by Kato-Weinstein et al. (2001) of 1.60-fold of control percent
30 liver/body weight cannot be accounted for by the 1.50-fold of control glycogen content.
31 Glycogen content only accounts for 5% of liver mass so that 50% increase in glycogen cannot
32 account for the 60% increase liver mass induced by 2 g/L DCA exposure for 8 weeks reported by
33 Kato-Weinstein (2001). Thus, DCA-induced increases in liver weight are occurring from other
34 processes as well. Carter et al. (2003) and DeAngelo et al. (1999) reported increased glycogen
35 after DCA treatment at much lower doses after longer periods of exposure (100 weeks). Carter
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-250 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported increased glycogen at 0.5 g/L DCA and DeAngelo et al. (1999) at 0.03 g/L DCA in
2 mice. However, there is no quantitation of that increase.
3 The issues involving identification of MO A through tumor phenotype analysis are
4 discussed in detail below for the more general case of liver cancer as well as for specific
5 hypothesized MO As (see Sections E.3.1.4, E.3.1.8, E.3.2.1, andE.3.4.1.5). For TCE and its
6 metabolites, c-Jun staining, H-rats mutation, tincture, heterogeneity in dysplacity have been used
7 to describe and differentiate liver tumors in the mouse.
8 Bull et al. (2002) reported 1,000 mg/kg TCE administered via gavage daily for 79 weeks
9 in male B6C3F1 mice to produce liver tumors and also reported deaths by gavage error (6 out of
10 40 animals). The limitations of the experiment are discussed in Caldwell et al. (2008b).
11 Specifically, for the DCA and TCA exposed animals, the experiment was limited by low
12 statistical power, a relatively short duration of exposure, and uncertainty in reports of lesion
13 prevalence and multiplicity due to inappropriate lesions grouping (i.e., grouping of hyperplastic
14 nodules, adenomas, and carcinomas together as "tumors"), and incomplete histopatholology
15 determinations (i.e., random selection of gross lesions for histopathology examination). For the
16 TCE results, a high prevalence (23/36 B6C3F1 male mice) of adenomas and hepatocellular
17 carcinoma (7/36) was reported. For determinations of immunoreactivity to c-Jun, as a marker of
18 differences in "tumor" phenotype, Bull et al. (2002) included all lesions in most of their
19 treatment groups, decreasing the uncertainty of his findings. However, for immunoreactivity
20 results hyperplastic nodules, adenomas, and carcinomas were grouped and thus, changes in c-Jun
21 expression between the differing types of lesions were not determined. Bull et al. (2002)
22 reported lesion reactivity to c-Jun antibody to be dependent on the proportion of the DCA and
23 TCA administered after 52 weeks of exposure. Given alone, DCA was reported to produce
24 lesions in mouse liver for which approximately half displayed a diffuse immunoreactivity to a c-
25 Jun antibody, half did not, and none exhibited a mixture of the two. After TCA exposure alone,
26 no lesions were reported to be stained with this antibody. When given in various combinations,
27 DCA and TCA coexposure induced a few lesions that were only c-Jun+, many that were only
28 c-Jun-, and a number with a mixed phenotype whose frequency increased with the dose of DCA.
29 For TCE exposure of 79 weeks, TCE-induced lesions were reported to also have a mixture of
30 phenotypes (42% c-Jun+, 34% c-Jun-, and 24% mixed) and to be most consistent with those
31 resulting from DCA and TCA coexposure but not either metabolite alone.
32 Stauber and Bull (1997) exposed male B6C3F1 mice (7 weeks old at the start of
33 treatment) to 2.0 g/L neutralized DCA or TCA in drinking water for 38 or 50 weeks, respectively
34 and then exposed (n = 12) to 0, 0.02, 0.1, 0.5, 1.0, 2.0 g/L DCA or TCA for an additional 2
35 weeks. Foci and tumors were combined in reported results as "lesions" and prevalence rates
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-251 DRAFT—DO NOT CITE OR QUOTE
-------
1 were not reported. The DCA-induced larger "lesions" were reported to be more "uniformly
2 reactive to c-Jun and c-Fos" but many nuclei within the lesions displaying little reactivity to c-
3 Jun. Stauber and Bull (1997) stated that while most DCA-induced "lesions" were
4 homogeneously immunoreactive to c-Jun and C-Fos (28/41 lesions), the rest were stained
5 heterogeneously. For TCA-induced lesions, the authors reported no difference in staining
6 between "lesions" and normal hepatocytes in TCA-treated animals. These results are slightly
7 different that those reported by Bull et al. (2002) for DC A, who report c-Jun positive and
8 negative foci in DCA-induced liver tumors but no mixed lesions. Because "lesions" comprised
9 of foci and tumors, different stages of progression reported in these results. The duration of
10 exposures also differed between DC A and TCA treatment groups that can affect phenotype. The
11 shorter duration of exposure can also prevent full expression of the tumor response.
12 Stauber et al. (1998) presented a comparison of in vitro results with "tumors" from
13 Stauber and Bull (1997) and note that 97.5% of DCA-induced "tumors" were c-Jun + while none
14 of the TCA-induced "tumors" were c-Jun +. However, the concentrations used to give tumors in
15 vivo for comparison with in vitro results were not reported. This appears to differ from the
16 heterogeneity of result for c-Jun staining reported by Bull et al. (2002) and Stauber and Bull
17 (1997). There was no comparison of c-Jun phenotype for spontaneous tumors with the authors
18 stating that because of such short time, no control tumors results were given. However, the
19 results of Bull et al. (2002) and Stauber and Bull (1997), do show TCA-induced lesions to be
20 uniformly c-Jun negative and thus, the phenotypic marker was able to show that TCE-induced
21 tumors were more like those induced by DC A than TCA.
22 The premise that DCA induced c-Jun positive lesions and TCA-induced c-Jun negative
23 lesions in mouse liver was used as the rationale to study induction of "transformed" hepatocytes
24 by DCA and TCE treatment in vitro. Stauber et al. (1998) isolated primary hepatocytes from
25 5-8 week old male B6C3F1 mice (n = 3) and subsequently cultured them in the presence of
26 DCA or TCA. In a separate experiment 0.5 g/L DCA was given to mice as pretreatment for
27 2 weeks prior to isolation. The authors assumed that the anchorage-independent growth of these
28 hepatocytes was an indication of an "initiated cell." DCA and TCA solutions were neutralized
29 before use. After 10 days in culture with DCA or TCA (0, 0.2, 0.5 and 2.0 mM), concentrations
30 of 0.5 mM or more DCA and TCA both induced an increase in the number of colonies that was
31 statistically significant, increased with dose with DCA, and slightly greater for DCA. In a time
32 course experiment the number of colonies from DCA treatment in vitro peaked by 10 days and
33 did not change through days 15-25 at the highest dose and, at lower concentrations of DCA,
34 increased time in culture induced similar peak levels of colony formation by days 20-25 as that
35 reached by 10 days at the higher dose. Therefore, the number of colonies formed was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-252 DRAFT—DO NOT CITE OR QUOTE
-------
1 independent of dose if the cells were treated long enough in vitro. However, not only did
2 treatment with DCA or TCA induce anchorage independent growth but untreated hepatocytes
3 also formed larger numbers of colonies with time, although at a lower rate than those treated
4 with DCA. The level reached by untreated cells in tissue culture at 20 days was similar to the
5 level induced by 10 days of exposure to 0.5 mM DCA. The time course of TCA exposure was
6 not tested to see if it had a similar effect with time as did DCA. The colonies observed at
7 10 days were tested for c-Jun expression with the authors noting that "colonies promoted by
8 DCA were primarily c-Jun positive in contrast to TCA promoted colonies that were
9 predominantly c-Jun negative." Of the colonies that arose spontaneously from tissue culture
10 conditions, 10/13 (76.9%) were reported to be c-Jun +, those treated with DCA 28/34 (82.3%)
11 were c-Jun +, and those treated with TCA 5/22 (22.7%) were c-Jun +. Thus, these data show
12 heterogeneity in cell in colonies but with more were c-Jun + colonies occurring by tissue culture
13 conditions alone and in the presence of DCA, rather than in the presence of TCA. The authors
14 reported that with time (24, 48, 72, and 96 hours) of culture conditioning the number of c-Jun+
15 colonies was increased in untreated controls. The authors reported that DCA treatment delayed
16 the increase in c-Jun+ expression induced by tissue culture conditions alone in untreated controls
17 while TCA treatment was reported to not affect the increasing c-Jun+ expression that increased
18 with time in tissue culture. This results seems paradoxical given that DCA induced a higher
19 number of colonies at 10 days of tissue culture than TCA and that most of the colonies were
20 c-Jun positive. The number of colonies was greater for pretreatment with DCA, but the
21 magnitude of difference over the control level was the same after DCA treatment in vitro without
22 and without pretreatment. As to the relationship of c-Jun staining and peroxisome proliferators
23 as a class, as pointed out by Caldwell and Keshava (2006), although Bull et al. (2004) have
24 suggested that the negative expression of c-jun in TCA-induced tumors may be consistent with a
25 characteristic phenotype shown in general by peroxisome proliferators as a class, there is no
26 supporting evidence of this.
27 An approach to determine the potential MO As of DCA and TCA through examination of
28 the types of tumors each "induced" or "selected" was to examine H-ras activation
29 (Ferreira-Gonzalez et al., 1995; Anna et al., 1994; Bull et al., 2002; Nelson et al., 1990). This
30 approach has also been used to try to establish an H-ras activation pattern for "genotoxic" and
31 "nongenotoxic" liver carcinogens compounds and to make inferences concerning peroxisome
32 proliferator-induced liver tumors. However, as noted by Stanley et al. (1994), the genetic
33 background of the mice used and the dose of carcinogen may affect the number of activated
34 H-ras containing tumors that develop. In addition, the stage of progression of "lesions" (i.e., foci
35 vs. adenomas vs. carcinomas) also has been linked the observance of H-ras mutations. Fox et al.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-253 DRAFT—DO NOT CITE OR QUOTE
-------
1 (1990) note that tumors induced by phenobarbital (0.05% drinking water (H2O), 1 year),
2 chloroform (200 mg/kg corn oil gavage, 2 times weekly for 1 year) or Ciprofibrate (0.0125%
3 diet, 2 years) had a much lower frequency of H-ras gene activation than those that arose
4 spontaneously (2-year bioassays of control animals) or induced with the "genotoxic" carcinogen
5 benzidine-2 hydrochloric acid (HC1; 120 ppm, drinking H^O, 1 year) in mice. In that study, the
6 term "tumor" was not specifically defined but a correlation between the incidence of H-ras gene
7 activation and development of either a hepatocellular adenoma or hepatocellular carcinoma was
8 reported to be made with no statistically significant difference between the frequency of H-ras
9 gene activation in the hepatocellular adenomas and carcinomas. Histopathological examination
10 of the spontaneous tumors, tumors induced with benzidine-2HCL, Phenobarbital, and chloroform
11 was not reported to reveal any significant changes in morphology or staining characteristics.
12 Spontaneous tumors were reported to have 64% point mutation in codon 61 (n = 50 tumors
13 examined) with a similar response for Benzidine of 59% (n = 22 tumors examined), whereas for
14 Phenobarbital the mutation rate was 7% (n = 15 tumors examined), chloroform 21%
15 (n = 24 tumors examined) and Ciprofibrate 21% (n = 39 tumors examined). The Ciprofibrate-
16 induced tumors were reported to be more eosinophilic as were the surrounding normal
17 hepatocytes. Hegi et al. (1993) tested Ciprofibrate-induced tumors in the NIH3T3
18 cotransfection-nude mouse tumorigenicity assay, which the authors state is capable of detecting a
19 variety of activated proto-oncogenes. The tumors examined (Ciprofibrate-induced or
20 spontaneously arising) were taken from the Fox et al. study (1990), screened previously, and
21 found to be negative for H-ras activation. With the limited number of samples examined,
22 Hegi et al. concluded that ras proto-oncogene activation or activation of other proto-oncogenes
23 using the nude mouse assay were not frequent events in Ciprofibrate-induced tumors and that
24 spontaneous tumors were not promoted with it. Using the more sensitive methods, the H-ras
25 activation rate was reported to be raised from 21 to 31% for Ciprofibrate-induced tumors and
26 from 64 to 66% for spontaneous tumors. Stanley et al. (1994) studied the effect of
27 methylclofenapate (MCP) (25 mg/kg for up to 2 years), a peroxisome proliferator, in B6C3F1
28 (relatively sensitive) and C57BL/10J (relatively resistant) mice for H-ras codon 61 point
29 mutations in MCP-induced liver tumors (hepatocellular adenomas and carcinomas). In the
30 B6C3F1 mice the number of tumors with codon 61 mutations was 11/46 and for C57BL/10J
31 mice 4/31. Unlike the findings of Fox et al. (1990), Stanley et al. (1994) reported an increase in
32 the frequency of mutation in carcinomas, which was reported to be twice that of adenomas in
33 both strains of mice, indicating that stage of progression was related to the number of mutations
34 in those tumors, although most tumors induced by MCP did not have this mutation.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-254 DRAFT—DO NOT CITE OR QUOTE
-------
1 In terms of liver tumor phenotype, Anna et al. (1994) reported that the H-ras codon 61
2 mutation frequency was not statistically different in liver tumors from DCA and TCE-treated
3 mice from a highly variable number of tumors examined. In regard to mutation spectra in H-ras
4 oncogenes in control or spontaneous tumors, the patterns were slightly different but mostly
5 similar to that of DCA-induced tumors (0.5% in drinking water). From their concurrent controls
6 they reported that H-ras codon 61 mutations in 17% (n = 6) of adenomas and 100% (n = 5) of
7 carcinomas. For historical controls (published and unpublished) they reported mutations in 73%
8 (n = 33) of adenomas and mutations in 70% (n = 30) of carcinomas. For tumors from TCE
9 treated animals they reported mutations in 35% (n = 40) of adenomas and 69% (n = 36) of
10 carcinomas, while for DCA treated animals they reported mutations in 54% (n = 24) of
11 adenomas and in 68% (n = 40) of carcinomas. Anna et al. (1994) reported more mutations in
12 TCE-induced carcinomas than adenomas.
13 The study of Ferreira-Gonzalez et al. (1995) in male B6C3 Fl mice has the advantage of
14 comparison of tumor phenotype at the same stage of progression (hepatocellular carcinoma), for
15 allowance of the full expression of a tumor response (i.e., 104 weeks), and an adequate number
16 of spontaneous control lesions for comparison with DCA or TCA treatments. However, tumor
17 phenotype at an endstage of tumor progression reflects of tumor progression and not earlier
18 stages of the disease process. In spontaneous liver carcinomas, 58% were reported to show
19 mutations in H-61 as compared with 50% of tumor from 3.5 g/L DCA-treated mice and 45% of
20 tumors from 4.5.g/L TCA-treated mice. Thus, there was a heterogeneous response for this
21 phenotypic marker for the spontaneous, DCA-, and TCA-treatment induced hepatocellular
22 carcinomas and not a pattern of reduced H-ras mutation reported for a number of peroxisome
23 proliferators. A number of peroxisome proliferators have been reported to have a much smaller
24 mutation frequency that spontaneous tumors (e.g., 13-24% H-ras codon 61 mutations after
25 Methylclofenopate depending on mouser strain, Stanley et al. [1994]: 21 to 31% for
26 Ciprofibrate-induced tumors and from 64 to 66% for spontaneous tumors, Fox et al. [1990] and
27 Hegietal. [1993]).
28 Bull (2000) suggested that "the report by Anna et al (1994) indicated that TCE-induced
29 tumors possessed a different mutation spectra in codon 61 of the H-ras oncogene than those
30 observed in spontaneous tumors of control mice." Bull (2000) stated that "results of this type
31 have been interpreted as suggesting that a chemical is acting by a mutagenic mechanism" but
32 went on to suggest that it is not possible to a priori rule out a role for selection in this process
33 and that differences in mutation frequency and spectra in this gene provide some insight into the
34 relative contribution of different metabolites to TCE-induced liver tumors. Bull (2000) noted
35 that data from Anna et al. (1994), Ferreira-Gonzalez et al. (1995), and Maronpot et al. (1995)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-255 DRAFT—DO NOT CITE OR QUOTE
-------
1 indicated that mutation frequency in DCA-induced tumors did not differ significantly from that
2 observed in spontaneous tumors. Bull (2000) also noted that the mutation spectra found in DCA-
3 induced tumors has a striking similarity to that observed in TCE-induced tumors, and DCA-
4 induced tumors were significantly different than that of TCA-induced liver tumors.
5 Bull et al. (2002) reported that mutation frequency spectra for the H-ras codon 61 in
6 mouse liver "tumors" induced by TCE (« = 37 tumors examined) were reported to be
7 significantly different than that for TCA (n = 41 tumors examined), with DCA-treated mice
8 tumors giving an intermediate result (n = 64 tumors examined). In this experiment, TCA-
9 induced "tumors" were reported to have more mutations in codon 61 (44%) than those from TCE
10 (21%) and DCA (33%). This frequency of mutation in the H-ras codon 61 for TCA is the
11 opposite pattern as that observed for a number of peroxisome proliferators in which the number
12 of mutations at H-ras 61 in tumors has been reported to be much lower than spontaneously
13 arising tumors (see Section E.3.4.1.5). Bull et al. (2002) noted that the mutation frequency for
14 all TCE,TCA or DCA tumors was lower in this experiment than for spontaneous tumors reported
15 in other studies (they had too few spontaneous tumors to analyze in this study), but that this
16 study utilized lower doses and was of shorter duration than that of Ferreira-Gonzalez et al.
17 (1995). These are additional concerns in addition to the effects of lesion grouping in which a
18 lower stage of progression is group with more advanced stages. In a limited subset of tumors
19 that were both sequenced and characterized histologically, only 8 of 34 (24%) TCE-induced
20 adenomas but 9/15 (60%) of TCE-induced carcinomas were reported to have mutated H-ras at
21 codon 61, which the authors suggest is evidence that this mutation is a late event.
22 Thus, in terms of H-ras mutation, the phenotype of TCE-induced tumors appears to be
23 more like DCA-induced tumors (which are consistent with spontaneous tumors), or those
24 resulting from a coexposure to both DCA and TCA (Bull et al., 2002), than from those induced
25 by TCA. As noted above, Bull et al. (2002) reported the mutation frequency spectra for the H-
26 ras codon 61 in mouse liver tumors induced by TCE to be significantly different than that for
27 TCA, with DCA-treated mice tumors giving an intermediate result and for TCA-induced tumors
28 to have a H-ras profile that is the opposite than those of a number of other peroxisome
29 proliferators. More importantly, these data suggest that using measures, other than dysplasticity
30 and tincture, mouse liver tumors induced by TCE are heterogeneous in phenotype.
31 With regard to tincture, Stauber and Bull (1997) reported the for male B6C3F1 mice,
32 DCA-induced "lesions" contained a number of smaller lesions that were heterogeneous and more
33 eosinophilic with larger "lesions" tending to less numerous and more basophilic. For TCA
34 results using this paradigm, the "lesions" were reported to be less numerous, more basophilic,
35 and larger than those induced by DCA. Carter et al. (2003) used tissues from the DeAngelo et al.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-256 DRAFT—DO NOT CITE OR QUOTE
-------
1 (1999) and examined the heterogeneity of the DCA-induced lesions and the type and phenotype
2 of preneoplastic and neoplastic lesions pooled across all time points. Carter et al. (2003)
3 examined the phenotype of liver tumors induced by DCA in male B6C3 Fl mice and the shape
4 of the dose-response curve for insight into its MO A. They reported a dose-response of
5 histopathologic changes (all classes of premalignant lesions and carcinomas) occurring in the
6 livers of mice from 0.05-3.5 g/L DCA for 26-100 weeks and suggest foci and adenomas
7 demonstrated neoplastic progression with time at lower doses than observed DCA genotoxicity.
8 Preneoplastic lesions were identified as eosinophilic, basophilic and/or clear cell (grouped with
9 clear cell and mixed cell) and dysplastic. Altered foci were 50% eosinophilic with about 30%
10 basophilic. As foci became larger and evolved into carcinomas they became increasingly
11 basophilic. The pattern held true through out the exposure range. There was also a dose and
12 length of exposure related increase in atypical nuclei in "noninvolved" liver. Glycogen
13 deposition was also reported to be dose-dependent with periportal accumulation at the 0.5 g/L
14 exposure level. Carter et al. (2003) suggested that size and evolution into a more malignant state
15 are associated with increasing basophilia, a conclusion consistent with those of Bannasch (1996)
16 and that there a greater periportal location of lesions suggestive as the location from which they
17 arose. Consistent with the results of DeAngelo et al. (1999), Carter et al. (2003) reported that
18 DCA (0.05-3.5 g/L) increased the number of lesions per animal relative to animals receiving
19 distilled water, shortened the time to development of all classes of hepatic lesions, and that the
20 phenotype of the lesions were similar to those spontaneously arising in controls. Along with
21 basophilic and eosinophilic lesions or foci, Carter et al. (2003) concluded that DCA-induced
22 tumors also arose from isolated, highly dysplastic hepatocytes in male B6C3F1 mice chronically
23 exposed to DCA suggesting another direct neoplastic conversion pathway other than through
24 eosinophilic or basophilic foci.
25 Rather than male B6C3F1 mice, Pereira (1996) studied the dose-response relationship for
26 the carcinogenic activity of DCA and TCA and characterized their lesions (foci, adenomas and
27 carcinomas) by tincture in females (the generally less sensitive gender). Like the studies of TCE
28 by Maltoni et al. (1986), female mice were also reported to have increased liver tumors after
29 TCA and DCA exposures. Pereira (1996) pool lesions were pooled for phenotype analysis so the
30 affect of duration of exposure could not be determined nor adenomas separated from carcinomas
31 for "tumors." However, as the concentration of DCA was decreased the number of foci was
32 reported by Pereira (1996) to be decreased but the phenotype of the foci to go from primarily
33 eosinophilic foci (i.e., -95% eosinophilic at 2.58 g/L DCA) to basophilic foci
34 (-57% eosinophilic at 0.26 g/L). For TCA the number of foci was reported to -40 basophilic
35 and -60 eosinophilic regardless of dose. Spontaneously occurring foci were more basophilic by
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-257 DRAFT—DO NOT CITE OR QUOTE
-------
1 a ratio of 7/3. Pereira (1996) described the foci of altered hepatocytes and tumors induced by
2 DCA in female B6C3F1 mice to be eosinophilic at higher exposure levels but at lower or
3 intermittent exposures to be half eosinophilic and half basophilic. Regardless of exposure level,
4 half of the TCA-induced foci were reported to be half eosinophilic and half basophilic with
5 tumors 75% basophilic. In control female mice, the limited numbers of lesions were mostly
6 basophilic, with most of the rest being eosinophilic with the exception of a few mixed tumors.
7 The limitations of descriptions tincture and especially for inferences regarding peroxisome
8 proliferator from the description of "basophilia" is discussed in Section E.3.4.1.5.
9 The results appear to differ between male and female B6C3F1 mice in regard to tincture
10 for DCA and TCA at differing doses. What is apparent is that the tincture of the lesions is
11 dependent on the stage of tumor progression, agent (DCA or TCA), gender, and dose. Also what
12 is apparent from these studies is the both DCA and TCA are heterogeneous in their tinctoral
13 characteristics as well as phenotypic markers such as mutation spectra or expression of c-Jun.
14 The descriptions of tumors in mice reported by the NCI, NTP, and Maltoni et al. studies
15 are also consistent with phenotypic heterogeneity as well as consistency with spontaneous tumor
16 morphology (see Section E.3.4.1.5). As noted in Section E.3.1, hepatocellular carcinomas
17 observed in humans are also heterogeneous. For mice, Maltoni et al. (1986) described malignant
18 tumors of hepatic cells to be of different subhistotypes, and of various degrees of malignancy and
19 were reported to be unique or multiple, and have different sizes (usually detected grossly at
20 necropsy) from TCE exposure. In regard to phenotype tumors were described as usual type
21 observed in Swiss and B6C3F1 mice, as well as in other mouse strains, either untreated or treated
22 with hepatocarcinogens and to frequently have medullary (solid), trabecular, and pleomorphic
23 (usually anaplastic) patterns. For the NCI (1976) study, the mouse liver tumors were described
24 in detail and to be heterogeneous "as described in the literature" and similar in appearance to
25 tumors generated by carbon tetrachloride. The description of liver tumors in this study and
26 tendency to metastasize to the lung are similar to descriptions provided by Maltoni et al. (1986)
27 for TCE-induced liver tumors in mice via inhalation exposure. The NTP (1990) study reported
28 TCE exposure to be associated with increased incidence of hepatocellular carcinoma (tumors
29 with markedly abnormal cytology and architecture) in male and female mice. Hepatocellular
30 adenomas were described as circumscribed areas of distinctive hepatic parenchymal cells with a
31 perimeter of normal appearing parenchyma in which there were areas that appeared to be
32 undergoing compression from expansion of the tumor. Mitotic figures were sparse or absent but
33 the tumors lacked typical lobular organization. Hepatocellular carcinomas were reported to have
34 markedly abnormal cytology and architecture with abnormalities in cytology cited as including
35 increased cell size, decreased cell size, cytoplasmic eosinophilia, cytoplasmic basophilia,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-258 DRAFT—DO NOT CITE OR QUOTE
-------
1 cytoplasmic vacuolization, cytoplasmic hyaline bodies and variations in nuclear appearance.
2 Furthermore, in many instance several or all of the abnormalities were reported to be present in
3 different areas of the tumor and variations in architecture with some of the hepatocellular
4 carcinomas having areas of trabecular organization. Mitosis was variable in amount and
5 location. Therefore, the phenotype of tumors reported from TCE exposure was heterogeneous in
6 appearance between and within tumors from all 3 of these studies.
7 Caldwell and Keshava (2006) report
8
9 that Bannasch (2001) and Bannasch et al. (2001) describe the early phenotypes of
10 preneoplastic foci induced by many oncogenic agents (DNA-reactive chemicals,
11 radiation, viruses, transgenic oncogenes and local hyperinsulinism) as
12 insulinomimetic. These foci and tumors have been described by tincture as
13 eosinophilic and basophilic and to be heterogeneous. The tumors derived from
14 them after TCE exposure are consistent with the description for the main tumor
15 lines of development described by Bannasch et al (2001) (see Section 3.4.1.5).
16 Thus, the response of liver to DCA (glycogenosis with emergence of glycogen
17 poor tumors) is similar to the progression of preneoplastic foci to tumors induced
18 from a variety of agents and conditions associated with increased cancer risk.
19
20 Furthermore Caldwell and Keshava (2006) note that Bull et al. (2002) report expression of
21 insulin receptor (IR) to be elevated in tumors of control mice or mice treated with TCE, TCA and
22 DCA but not in nontumor areas suggesting that this effect is not specific to DCA.
23 There is a body of literature that has focused on the effects of TCE and its metabolites
24 after rats or mice have been exposed to "mutagenic" agents to "initiate" hepatocarcinogenesis
25 and this is discussed in Section E.4.2, below. TCE and its metabolites were reported to affect
26 tumor incidence, multiplicity, and phenotype when given to mice as a coexposure with a variety
27 of "initiating" agents and with other carcinogens. Pereira and Phelps (1996) reported that MNU
28 alone induced basophilic foci and adenomas. MNU and low concentrations of DCA or TCA in
29 female mice were reported to induce heterogeneous for foci and tumor with a higher
30 concentration of DCA inducing more eosinophilic and a higher concentration of TCA inducing
31 more tumors that were basophilic. Pereira et al. (2001) reported that not only dose, but gender
32 also affected phenotype in mice that had already been exposed to MNU and were then exposed
33 to DCA. As for other phenotypic markers, Lantendresse and Pereira (1997) reported that
34 exposure to MNU and TCA or DCA induced tumors that had some commonalities, were
35 heterogeneous, but for female mice were overall different between DCA and TCA as
36 coexposures with MNU.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-259 DRAFT—DO NOT CITE OR QUOTE
-------
1 Stop experiments which attempt to ascertain the whether progression differences exist
2 between TCA and DC A have used higher concentrations at much lower durations of exposure.
3 A question arises as to whether the differences in results between those animals in which
4 treatment was suspended in comparison to those in which had not had been conducted so that full
5 expression of response had not been allowed rather than "progression" as well as the effects of
6 using large doses. After 37 weeks of treatment and then a cessation of exposure for 15 weeks
7 Bull et al. (1990) reported that after 15 weeks of cessation of exposure, liver weight and percent
8 liver/body weight were reported to still be statistically significantly elevated after DCA or TCA
9 treatment. The authors partially attribute the remaining increases in liver weight to the continued
10 presence of hyperplastic nodules in the liver. In terms of liver tumor induction, the authors
11 stated that "statistical analysis of tumor incidence employed a general linear model ANOVA
12 with contrasts for linearity and deviations from linearity to determine if results from groups in
13 which treatments were discontinued after 37 weeks were lower than would have been predicted
14 by the total dose consumed." The multiplicity of tumors observed in male mice exposed to DCA
15 or TCA at 37 weeks and then sacrificed at 52 weeks were reported by the authors to have a
16 response in animals that received DCA very close to that which would be predicted from the
17 total dose consumed by these animals. The response to TCA was reported by the authors to
18 deviate significantly (p = 0.022) from the linear model predicted by the total dose consumed.
19 Multiplicity of lesions per mouse and not incidence was used as the measure. Most importantly
20 the data used to predict the dose response for "lesions" used a different methodology at 52 weeks
21 than those at 37 weeks. Not only were not all animal's lesions examined, but foci, adenomas,
22 and carcinomas were combined into one measure. Therefore, foci, of which a certain percentage
23 have been commonly shown to spontaneously regress with time, were included in the calculation
24 of total "lesions." Pereira and Phelps (1996) note that in MNU-treated mice that were then
25 treated with DCA, the yield of altered hepatocytes decreases as the tumor yields increase
26 between 31 and 51 weeks of exposure suggesting progression of foci to adenomas. Initiated and
27 noninitiated control mice were reported to also have fewer foci/mouse with time. Because of
28 differences in methodology and the lack of discernment between foci, adenomas, and carcinomas
29 for many of the mice exposed for 52 weeks, it is difficult to compare differences in composition
30 of the "lesions" after cessation of exposure in the Bull et al. (1990) study. For TCA treatment
31 the number of animals examined for determination of which "lesions" were foci, adenomas, and
32 carcinomas was 11 out of the 19 mice with "lesions" at 52 weeks while all 4 mice with lesions
33 after 37 weeks of exposure and 15 weeks of cessation were examined. For DCA treatment the
34 number of animals examined was only 10 out of 23 mice with "lesions" at 52 weeks while all
35 7 mice with lesions after 37 weeks of exposure and 15 weeks of cessation were examined. Most
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-260 DRAFT—DO NOT CITE OR QUOTE
-------
1 importantly, when lesions were examined microscopically then did not all turn out to be
2 preneoplastic or neoplastic. Two lesions appeared "to be histologically normal" and one
3 necrotic. Not only were a smaller number of animals examined for the cessation exposure than
4 continuous exposure but only the 2 g/L exposure levels of DCA and TCA were studied for
5 cessation. The number of animals bearing "lesions" at 37 and then 15 week cessation weeks was
6 7/11 (64%) while the number of animals bearing lesions at 52 weeks was 23/24 (96%) after
7 2 g/L DCA exposure. For TCA the number of animals bearing lesions at 37 weeks and then
8 15 weeks cessation was 4/11 (35%) while the number of animals bearing lesions at 52 weeks was
9 19/24 (80%). While suggesting that cessation of exposure diminished the number of "lesions,"
10 conclusions regarding the identity and progression of those lesion with continuous versus
11 noncontinuous DCA and TCA treatment are tenuous.
12
13 E.2.5. Studies of Chloral Hydrate (CH)
14 Given that total oxidative metabolism appears to be highly correlated with TCE-induced
15 increases in liver weight in the mouse rather than merely the presence of TCA, other metabolites
16 are of interest as potential agents mediating the effects observed for TCE. Recently Caldwell
17 and Keshava provided a synopsis of the results of more recent studies involving CH (Caldwell
18 and Keshava, 2006). A large fraction of TCE oxidative metabolism appears to go through CH,
19 with subsequent metabolism to TCA and trichloroethanol (Chiu et al., 2006b). Merdink et al.
20 (2008) demonstrated that CH administered to humans can be extremely variable and complex in
21 its pharmacokinetic behavior with a peak plasma concentration of CH in plasma 40-50 times
22 higher than observed at the same time interval for other subjects. Studies of CH toxicity in
23 rodents are consistent with the general presumption that oxidative metabolites are important for
24 TCE-induced liver tumors, but whether CH and its metabolites are sufficient to explain all of
25 TCE liver tumorigenesis remains unclear, particularly because of uncertainties regarding how
26 DCA may be formed (Chiu et al., 2006b). Studies of CH may enable a comparison between
27 toxicity of TCE and CH and may help elucidate its role in TCE effects. As with other TCE
28 metabolites, the majority of the studies have focused on the mouse liver tumor response. For
29 rats, while the limited data suggests that there is less of a response than mice to CH, those studies
30 are limited in power or reporting.
31 Daniel et al. (1992) exposed adult male B6C3F1 (C57B1/6JC male mice bred to
32 C3Heb/Fej female mice) 28-day old mice to CH, 2-chloroacetaldehyde, or DCA in 2 different
33 phases (I and II) with initial weights ranging from 9.4 to 13.6 g. The test compounds were
34 buffered and administered in drinking water for 30 and 60 weeks (n = 5 for interim sacrifice),
35 and for 104 weeks (n = 40). The concentration of CH was 1 g/L and for DCA 0.5 g/L and the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-261 DRAFT—DO NOT CITE OR QUOTE
-------
1 estimated doses of DCA were 85, 93, and 166 mg/kg/d for the DCA group I, DCA group II, and
2 CH exposed group, respectively. Microscopic examination of tissues was conducted for all
3 tissues for five animals of the CH groups with liver, kidneys, testes, and spleen, in addition to all
4 gross lesions, reported to be examined microscopically in all of the 104-week survivors. The
5 initial body weight for drinking water controls was reported to be 12.99 ± 3.04 g for group I
6 (n = 23) and 10.48 ± 1.70 for group II (n = 10). For DCA treated animals, initial body weights
7 were 13.44 ± 2.57 g for group I (n = 23) and 9.65 ± 2.72 g for group II (n = 10). For the CH
8 treated group the initial body weights were reported to be 10.42 ± 2.49 g(n = 40). It is not clear
9 from the report what control group best matched, if any, the CH group. Thus, the mean initial
10 body weights of the groups as well as the number of animals varied considerably in each group
11 (i.e., -40% difference in mean body weights at the beginning of the study). The number of
12 animals surviving till the termination of the experiment was 10, 10, 16, 8, and 24 for the control
13 group I, control group II, DCA group I, DCA group II, and CH groups, respectively. An
14 increase in absolute and relative liver weight versus reported to be observed at 30 weeks for
15 DCA and CH groups and at 60 weeks for CH but data were not shown in the study. At 104
16 weeks, the data for the surviving control groups were combined as was that for the 2 DCA
17 treatment groups. Of note was that for CH treated survivors (n = 24) water consumption was
18 significantly reduced in comparison to controls. Absolute liver weight was reported to be
19 2.09±0.6g, 3.17± 1.3 gand2.87± 1.1 g for control, DCA and CH treatment groups,
20 respectively. The % liver to body weight was reported to be similarly elevated (1.57-fold of
21 control for DCA and 1.41-fold of control for CH) at 104 weeks. At 104 weeks the treatment-
22 related liver lesions in histological sections were reported to be most prominently
23 hepatocytomegaly and vacuolization in DCA-treated animals. Cytomegaly was also reported to
24 be in 5, 92, and 79% of control, DCA and CH treatment groups, respectively. Cytomegaly in CH
25 treated mice was described as minimal and associated with an increased number of basophilic
26 granules (rough endoplasmic reticulum). Hepatocellular necrosis and chronic active
27 inflammation were reported to be mildly increased in both prevalence and severity in all treated
28 groups. The histological findings, from interim sacrifices (n = 5), were considered by the
29 authors to be unremarkable and were not reported. Liver tumors were increased by DCA and
30 CH treatment. The percent incidence of liver carcinomas and adenomas combined in the
31 surviving animals was 15, 75, and 71% in control, DCA and CH treated mice, respectively. In
32 the CH treated group, the incidence of hepatocellular carcinoma was 46%. The number of
33 tumors/animals was also significantly increased with CH treatment. Most importantly,
34 morphologically the authors noted that there did not appear to be any discernable differences in
35 the visual appearance of the DCA- and CH-induced tumors.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-262 DRAFT—DO NOT CITE OR QUOTE
-------
1 George et al. (2000) exposed male B6C3F1 mice and male F344/N rats to CH in drinking
2 water for 2 years (up to 162.6 mg/kg/d). Target drinking water concentrations were 0, 0.05, 0.5,
3 and 2 g/L CH in rats and 0, 0.05, 0.5 and 1.0 g/L CH in mice. Groups of animals (n = 6/group)
4 were sacrificed at 13 (rats only), 26, 52 and 78 weeks following the initiation of dosing with
5 terminal sacrifices at Week 104. A complete pathological examination was performed on 5 rats
6 and mice from the high-dose group, with examination primarily of gross lesions except for liver,
7 kidney, spleen and testes. BrdU incorporation was measured in the interim sacrifice groups in
8 rats and mice with PCO examined at 26 weeks in mice. In rats, the number of animals surviving
9 >78 weeks and examined for hepatocellular proliferative lesions was 42, 44, 44, and 42 for the
10 control, 7.4, 37.4 and 163.6 mg/kg/d CH treatment groups, respectively. Only 32, 36, 35, and
11 32 animals were examined at the final sacrifice time. Only the lowest treatment group had
12 increased liver tumors, which were marginally significantly increased by treatment. The percent
13 of animals with hepatocellular adenomas and carcinomas was reported to be 2.4, 14.3, 2.3 and
14 6.8% in male rats. In mice, preneoplastic foci and adenomas were reported to be increased in the
15 livers of all CH treatment groups (13.5-146.6 mg/kg/d) at 104 weeks. The incidences of
16 adenomas were reported to be statistically increased at all dose levels, the incidences of
17 carcinomas significantly increased at the highest dose, and time-to-tumor decreased in all CH-
18 treatment groups. The percent incidence of hepatocellular adenomas was reported to be 21.4,
19 43.5, 51.3, and 50% in control, 13.5, 65.0, and 146.6 mg/kg day treatment groups, respectively.
20 The percent incidence of hepatocellular carcinomas was reported to be 54.8, 54.3, 59.0, and
21 84.4% in these same groups. The resulting percent incidence of hepatocellular adenomas and
22 carcinomas was reported to be 64.3, 78.3, 79.5, and 90.6%. The number of mice surviving
23 >78 weeks was reported to be 42, 46, 39, and 32 and the number surviving to final sacrifice to be
24 34, 42, 31, and 25 for control, 13.5, 65.0 and 146.56 mg/kg/d, respectively. CH exposure was
25 reported to not alter serum chemistry, hepatocyte proliferation (i.e., DNA synthesis), or hepatic
26 PCO activity (an enzyme associated with PPARa agonism) in rats and mice at any of the time
27 periods monitored (all interim sacrifice periods for BrdU incorporation, 52 or 78 weeks for
28 serum enzymes, and 26 weeks for PCO) with the exception of 0.58 g/L CH at 26 weeks slightly
29 increasing hepatocyte labeling (-2-3-fold increase over controls) in rats and mice but the percent
30 labeling still represented 3% or less of hepatocytes. With regard to other carcinogenic endpoints
31 only five animals were examined at the high dose, thereby limiting the study's power to
32 determine an effect. Control mice were reported to have a high spontaneous carcinoma rate
33 (54%), thereby limiting the ability to detect a treatment-related response. No descriptions of the
34 foci or tumor phenotype were given. However, of note is the lack of induction of PCO response
35 with CH at 26 weeks of administration in either rats or mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-263 DRAFT—DO NOT CITE OR QUOTE
-------
1 Leakey et al. (2003a) studied the effects of CH exposure (0, 25, 50, and 100 mg/kg,
2 5 days/week, 104-105 weeks via gavage) in male B6C3F1 mice with dietary control used to
3 manipulate body growth (n = 48 for 2 year study and n = 12 for the 15-month interim study).
4 Dietary control was reported to decrease background liver tumor rates (incidence of 15-20%)
5 and was reported to be associated with decreased variation in liver-to-body weight ratios, thereby
6 potentially increasing assay sensitivity. In dietary-controlled groups and groups fed ad libitum,
1 liver adenomas and carcinomas (combined) were reported to be increased with CH treatment.
8 With dietary restriction there was a more discernable CH tumor-response with overall tumor
9 incidence reduced, and time-to-tumor increased by dietary control in comparison to ad libitum
10 fed mice. Incidences of hepatocellular adenoma and carcinoma overall rates were reported to be
11 33, 52, 49, and 46% for control, 25, 50, and 100 mg/kg ad libitum-fed mice, respectively. For
12 dietary controlled mice the incidence rates were reported to be 22.9, 22.9, 29.2, and 37.5% for
13 controls, 25, 50, and 100 mg/kg CH, respectively. Body weights were matched and carefully
14 controlled in this study.
15 After 2 years of CH treatment the heart weights of ad libitum-fed male mice administered
16 100 mg/kg CH were reported to be significantly less and kidney weights of the 50 and 100
17 nig/kg less than vehicle controls. No other significant organ weight changes due to CH treatment
18 were reported to be observed in either diet group except for liver. The liver weights of CH
19 treated groups for by dietary groups were reported to be increased at 2 years and the absolute
20 liver weights of dosed groups to be generally increased at 15 months with percent liver/body
21 weight ratios increased in CH treated dietary-controlled mice at 15 months. There was 1.0-,
22 0.87-, and 1.08-fold of control percent liver/body weight for ad libitum fed mice exposed to 25,
23 50, and 100 mg/kg CH, respectively. For dietary controlled mice, there was 1.05-, 1.08-, and
24 1.11 -fold of control percent liver/body weight for the same dose groups at 15 months. Thus,
25 there was no corresponding dose-response for percent liver/body weight in the ad libitum-fed
26 mice, which were reported to show a much larger variation in liver-to-body-weight ratios (i.e.,
27 the standard deviation and standard errors were 2- to 17-fold lower in dietary controlled groups
28 than for ad libitum-fed groups). Liver weight increases at 15-months did not correlate with
29 2-year tumor incidences with this group. However, for dietary controlled groups the increase in
30 percent liver/body weights at 15 months were generally correlated with increases in liver tumors
31 at 2 years. The incidences of peripheral or focal fatty change were reported to be increased in all
32 CH-treated groups of ad libitum-fed mice at 15 months (approximately half the animals showed
33 these changes for all dose groups, with no apparent dose-response). Of the enzymes associated
34 with PPARa agonism (total CYP, CYP2B isoform, CYP4A, or lauric acid p-hydroxylase
35 activity), only CYP4A and lauric acid P-hydroxylase activity were significantly increased at
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-264 DRAFT—DO NOT CITE OR QUOTE
-------
1 15 months of exposure in the dietary-restricted group administered 100 mg/kg CH with no other
2 groups reported showing a statistically significant increased response (n = 12/group). Although
3 not statistically significant, the 100 mg/kg CH exposure group of ad libitum-fed mice also had an
4 increase in CYP4A and lauric acid p-hydroxylase activity. The authors reported that the increase
5 in magnitude of CYP4A and lauric acid P-hydroxylase activity at 100 mg/kg CH at 15 months in
6 dietary controlled mice correlated with the increase incidence of mice with tumors. However,
7 there was no correlation of tumor incidence and the increased enzyme activity associated with
8 peroxisome proliferation in the ad libitum-fed mice. No descriptions of liver pathology were
9 given other than incidence of mice with fatty liver changes. Hepatic malondialdehyde
10 concentration in ad libitum fed and dietary controlled mice did not change with CH exposure at
11 15 months but the dietary controlled groups were all approximately half that of the ad libitum-
12 fed mice. Thus, while overall increased tumors observed in the ad libitum diet correlated with
13 increased malondialdehyde concentration, there was no association between CH dose and
14 malondialdehyde induction for either diet.
15 Induction of peroxisome-associated enzyme activities was also reported for shorter times
16 of CH exposure. Seng et al. (2003) described CH toxicokinetics in mice at doses up to
17 1,000 mg/kg/d for 2 weeks with dietary control and caloric restriction slightly reducing acute
18 toxicity. Lauric acid P-hydroxylase and PCO activities were reported to be induced only at doses
19 >100 mg/kg in all groups, with dietary-restricted mice showing the greatest induction.
20 Differences in serum levels of TCA, the major metabolite remaining 24 hr after dosing, were
21 reported not to correlate with hepatic lauric acid P-hydroxylase activities across groups.
22 Leuschner and Beuscher (1998) examined the carcinogenic effects of CH in male and
23 female S-D rats (69-79 g, 25-29 days old at initiation of the experiment) administered 0, 15, 45,
24 and 135 mg/kg CH in unbuffered drinking water 7 days/week (n = 50/group) for 124 weeks in
25 males and 128 weeks in females. Two control groups were noted in the methods section without
26 explanation as to why they were conducted as two groups. The mean survival for males was
27 similar in treated and control groups with 20, 24, 20, 24, and 20% of Ccontrol I, Control II, 15,
28 45, and 135 mg/kg CH-treated groups, respectively, surviving till the end of the study. For
29 female rats, the percent survival was 12, 30, 24, 28, and 16% for of Control I, Control II, 15, 45,
30 and 135 mg/kg CH-treated groups, respectively. The authors report no substance-related
31 influence on organ weights and no macroscopic evidence of tumors or lesions in male or female
32 rats treated with CH for 124 or 128 weeks. However, no data are presented on the incidence of
33 tumors using this paradigm, especially background rates. The authors report a statistically
34 significant increase in the incidence of hepatocellular hypertrophy in male rats at the 135 mg/kg
35 dose (14/50 animals vs. 4/50 and 7/50 in controls I and II). For female rats, the incidence of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-265 DRAFT—DO NOT CITE OR QUOTE
-------
1 hepatocellular hypertrophy was reported to be 10/50 rats (Control I) and 16/50 (Control II) rats
2 with 18/50, 13/50 and 12/50 female rats having hepatocellular hypertrophy after 15, 45, and
3 135 mg/kg CH, respectively. The lack or reporting in regard to final body weights, histology,
4 and especially background and treatment group data for tumor incidences, limit the interpretation
5 of this study. Whether this paradigm was sensitive for induction of liver cancer cannot be
6 determined.
7 From the CH studies in mice, there is an apparent increase in liver adenomas and
8 carcinomas induced by CH treatment by either drinking water or gavage with all available
9 studies performed in male B6C3F1 mice. However, the background levels of hepatocellular
10 adenomas and carcinomas in these mice in George et al. (2000) and body weight data from this
11 study show it is from a tumor prone mouse. Comparisons with concurrent studies of mice
12 exposed to DC A revealed that while both CH and DC A induced hepatomegaly and cytomegaly,
13 DCA-induced cytomegaly was accompanied by vacuolization while that of CH to be associated
14 with increased number of basophilic granules (rough endoplasmic reticulum) which would
15 suggest separate effects. However, the morphology of the CH-induced tumors was reported to
16 be similar between DC A and CH-induced tumors (Daniel et al., 1992). Using a similar paradigm
17 (2-year study of B6C3F1 male mice), De Angelo et al. (1999) and Carter et al. (2003) described
18 DCA-induced tumors to be heterogeneous. This is the same description given for TCE-induced
19 tumors in the studies by NTP, NCI, and Maltoni et al. and to be a common description for tumors
20 caused by a variety of carcinogenic agents. Similar to the studies cited above for CH, DeAngelo
21 et al. (1999) reported that PCO levels were only elevated at 26 weeks at 3.5 g/L DCA and had
22 returned to control levels by 52 weeks. Similar to CH, no increased tritiated thymidine was
23 reported for DCA at 26 and 52 weeks with only 2-fold of control values reported at 0.05 g/L at
24 4 weeks. Leakey et al. (2003a) reported that ad libitum fed male mice exhibited a similar degree
25 of increased incidence of peripheral or focal fatty change at 15 months for all CH doses but not
26 enzymes associated with peroxisome proliferation. While dietary restriction seemed to have
27 decreased background levels of tumors and increased time-to-tumor, CH-gave a clear dose-
28 response in dietary restricted animals. However, while the overall level of tumor induction was
29 reduced there was a greater induction of PPARa enzymes by CH. Induction of liver tumors by
30 CH observed in ad libitum fed mice were not correlated with PPARa induction, with dietary
31 restriction alone appearing to have greater levels of lauric acid co-hydrolase activity in control
32 mice at 15 months. Seng et al. (2003) report that lauric acid p-hydroxylase and PCO were
33 induced only at exposure levels >100 mg/kg CH, again with dietary restricted groups showing
34 the greatest induction. Such data argues against the role of peroxisome proliferation in CH-liver
35 tumor induction in mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-266 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.2.6. Serum Bile Acid Assays
2 Serum bile acids (SB A) have been suggested as a sensitive indicator of hepatotoxicity to
3 a variety of halogenated solvents with an advantage of increased sensitivity and specificity over
4 conventional liver enzyme tests that primarily reflect the acute perturbation of hepatocyte
5 membrane integrity and "cell leakage" rather than liver functional capacity (i.e., uptake,
6 metabolism, storage, and excretion functions of the liver) (Bai et al., 1992b; Neghab et al., 1997).
7 While some studies have reported negative results, a number of studies have reported elevated
8 SBA in organic solvent-exposed workers in the absence of any alterations in normal liver
9 function tests. These variations in results have been suggested to arise from failure of some
10 methods to detect some of the more significantly elevated SBA and the short-lived and reversible
11 nature of the effect (Neghab et al., 1997). Neghab et al. (1997) have reported that occupational
12 exposure to l,l,2-trichloro-l,2,2-trifluoroethane and trichloroethylene has resulted in elevated
13 SBA and that several studies have reported elevated SBA in experimental animals to chlorinated
14 solvents such as carbon tetrachloride, chloroform, hexachlorobutadiene, tetrachloroethylene,
15 1,1,1-trichloroethane, and trichloroethylene at levels that do not induce hepatotoxicity (Bai et al.,
16 1992a, b; Hamdan and Stacey, 1993; Wang and Stacey, 1990). Toluene, a nonhalogenated
17 solvent, has also been reported to increase SBA in the absence of changes in other hepatobiliary
18 functions (Neghab and Stacey, 1997). Thus, disturbance in SAB appears to be a generalized
19 effect of exposure to chlorinated solvents and nonchlorinated solvents and not specific to TCE
20 exposure.
21 Neghab et al. (1997) reported that 8 hour time-weighted averages exposure to TCE of
22 8.9 ppm, measured in the breathing zone using a charcoal tube personal sampler for the whole
23 mean duration of exposure of 3.4 years, to have not significant changes in albumin, bilirubin,
24 alkaline phosphatase, alanine aminotransferase, 5'-nucleosidase, y-glutamyltransferase, but to
25 have significantly increased total serum bile acids. Not only were total bile acids significantly
26 increased in these TCE-exposed workers compared to controls (~2-fold of control), but,
27 specifically, deoxycholic acid and subtotal of free bile acids were increased. Neghab et al.
28 (1997) do not show the data, but also report that "despite the apparent overall low level of
29 exposure, there was a very good correlations (r = 0.94) between the degree of increase in serum
30 concentration of total bile acids and level of TCE." Neghab et al. (1997) note that while a
31 sensitive indicator or exposure to such solvents in asymptomatic workers, there is no indication
32 that actual liver injury occurs in conjunction with SAB increases.
33 Wang and Stacey (1990) administered TCE in corn oil via i.p. injection to male S-D rats
34 (300-500 g) at concentrations of 0.01, 0.1, 1, 5, and 10 mmol/kg on 3 consecutive days (n = 4, 5,
35 or 6) with liver enzymes and SBA examined 4 hours after the last TCE treatment. At these dose,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-267 DRAFT—DO NOT CITE OR QUOTE
-------
1 there were not differences between treated and control animals in regard to alkaline phosphatase
2 and sorbitol dehydrogenase concentrations and an elevation of alanine aminotransferase only at
3 the highest dose. However, there was generally a reported dose-related increase in cholic acid,
4 chenodeoxycholic acid, deoxycholic acid, taurocholic acid, tauroursodeoxycholic acid with
5 cholic acid and taurochlolic acid increased at the lowest dose. The authors report that
6 "examination of liver sections under light microscopy yielded no consistent effects that could be
7 ascribed to trichloroethylene." In the same study a rats were also exposed to TCE via inhalation
8 (n = 4) at 200 ppm for 28 days, and 1,000 ppm for 6 hours/day. Using this paradigm, cholic acid
9 and taurocholic acid were significantly elevated at the 200 ppm level, (-10- and ~5-fold of
10 control, respectively) with very large standard errors of the mean. At the 1,000 ppm level
11 (6 hours, day) cholic acid and taurocholic acid were elevated to ~2-fold of control but neither
12 was statistically significant. The large variability in responses between rats and the low number
13 of rats tested in this paradigm limit its ability to determine quantitative differences between
14 groups. Nevertheless, without the complications associated with i.p. exposure (see
15 Section E.2.2.1, above), both inhalation exposure of TCE at a relative low exposure level was
16 also associated with increased SB A levels. The authors stated that "no increases in alanine
17 amino transferase levels were observed in the rats exposed to trichloroethylene via inhalation."
18 No histopathology results were reported for rats exposed via inhalation. As stated by Wang and
19 Stacey (1990), "intraperitoneal injection is not particularly relevant to humans" which was the
20 rationale given for the inhalation exposure experiments in the study. They point out that
21 intestinal interactions require consideration because a major determinant of SB A is their
22 absorption from the gut and intestinal flora may play a role in bile acid metabolism. They also
23 note that grooming done by the experimental rats would probably give small exposure via
24 ingestion of TCE as well. However, Wang and Stacey (1990) reported consistent results in terms
25 of TCE-induced changes in SBA at relatively low concentrations by either inhalation or i.p.
26 routes of exposure that were not associated with other measures of toxicity.
27 Hamdan and Stacey (1993) administered TCE in corn oil (1 mmol/kg) in male Sprague
28 Dawley rats (300-400 g) and followed the time-course of SBA elevation, TCE concentration and
29 trichloroethanol in the blood at 2, 4, 8, and 16 hours after dosing (n = 4,5, or 6 per group). Liver
30 and blood concentration of TCE were reported to peak at 4 hours while those of trichloroethanol
31 peaked at 8 hours after dosing. TCE levels were not detectable by 16 hours in either blood or
32 liver while those of trichloroethanol were still elevated. Elevations of SBA were reported to
33 parallel those of TCE with cholic acid and taurochloate acid reported to show the highest levels
34 of bile acids. The dose given was based on that reported by Wang and Stacey (1990) to give no
35 hepatotoxicity but an increase in SBA. The authors state that liver injury parameters were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-268 DRAFT—DO NOT CITE OR QUOTE
-------
1 checked and found unaffected by TCE exposure but do not show the data. Thus, it was TCE
2 concentration and not that of its metabolite that was most closely related to changes in SB A and
3 after a single exposure, the effect was reversible. In an in vitro study by Bai and Stacey (1993),
4 TCE was studied in isolated rat hepatocytes with TCE reported to cause a dose-related
5 suppression of initial rates of cholic acid and taurocholic acid but with no significant effects on
6 enzyme leakage and intracellular calcium contents, further supporting a role for the parent
7 compound in this effect. The authors noted that the changes in SBA result from interference
8 with a physiological process rather "than an event associated with significant pathological
9 consequences."
10
11 E.3. STATE OF SCIENCE OF LIVER CANCER MODES OF ACTION (MOAs)
12 The experimental evidence in mice shows that TCE and its metabolites induce foci,
13 hepatocellular adenomas, and carcinomas that are heterogeneous in nature as indicated by
14 phenotypic differences in tincture, mutational markers, or gene expression markers. The tumors
15 induced by TCE are reflective of phenotypes that are either similar to those induced by mixtures
16 of DC A and TCA exposure, or more like those induced by DC A. These tumors have been
17 described to be similar also to those arising spontaneously in mice or from chemically induced
18 hepatocarcinogenesis and to arise from preneoplastic foci, and in the case of DC A, single
19 dysplastic hepatocytes as well as foci. HCC observed in humans also has been described to be
20 heterogeneous and to be associated with formation of preneoplastic nodules. Although several
21 conditions have been associated with increased risk of liver cancer in humans, the mechanism of
22 HCC is unknown at this time. A great deal of attention has been focused on predicting which
23 cellular targets (e.g., "stem-cell" or mature hepatocyte) are associated with HCC as well as on
24 phenotypic markers in HCC that can provide insight not only into MO A and origin of tumor, but
25 also for prediction of clinical course. Examination of pathways and epigenetic changes
26 associated with cancer, and the relationship of these changes to liver cancer are also discussed
27 below. The field of cancer research has been transformed by the recent discoveries of epigenetic
28 changes and their role in cancer and chronic disease states. The following discussion describes
29 these advances but also the issues involved with the technologies that have emerged to describe
30 them (see Section E.3.1.2, below). Exposure to TCE and its metabolites, like many others,
31 induces a heterogeneous response, even in a relatively homogeneous genetic paradigm as the
32 experimental laboratory rodent model. The importance of phenotypic anchoring is a major issue
33 in the study of any MO As using these new technologies of gene expression pattern. Although a
34 large amount of information is now available using microarray technologies and transgenic
35 mouse models, specifically for TCE and in study of suggested MO As for TCE and its
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-269 DRAFT—DO NOT CITE OR QUOTE
-------
1 metabolites, use of these approaches has limitations that need to be considered in the
2 interpretation of data and conclusions derived from such data, especially quantitative
3 conclusions.
4 For TCE and its metabolites, the extent of acute to subchronic induction of hepatomegaly
5 correlated with hepatocellular carcinogenicity, although each had differing factors contributing
6 to that hepatomegaly from periportal glycogen deposition to hepatocellular hypertrophy and
7 increased polyploidy. The extent of transient DNA synthesis, peroxisome proliferation, or
8 cytotoxicity was not correlated with carcinogenicity. Hepatomegaly is also a predictor of
9 carcinogenicity for a number of other compounds in mice and rats. Allen et al. (2004) examined
10 the NTP database (87 compounds for rat and 83 for mice) and tried to correlate specific
11 hepatocellular pathology in prechronic studies with carcinogenic endpoints in the chronic 2-year
12 assays. The best single predictor of liver cancer in mice was hepatocellular hypertrophy.
13 Hepatocellular cytomegaly and hepatocyte necrosis also contributed, although the numbers of
14 positive findings were less than hypertrophy. With regard to genotoxicity studies, there was no
15 evidence of a correlation between mouse liver tumor chemicals and Salmonella or micronucleus
16 assay outcome. None of the prechronic liver lesions examined were correlated with either
17 Salmonella or Micronucleus assays. In rats no single prechronic liver lesions (when considered
18 individually) was a strong predictor of liver cancer in rats. The most predictive lesions was
19 hepatocellular hypertrophy. There was not significant correlation between liver tumors/toxicity
20 and the 2 mutagenicity measures. Although the lack of correlation with the mutagenicity assays
21 could be interpreted as rodent assays predominantly identifying nongenotoxic liver carcinogens,
22 this conclusion could be questioned because it is solely dependent on Salmonella mutagenicity
23 and additional genotoxic endpoints could conceivably shift the association between liver cancer
24 and genotoxicity towards a more positive correlation. As to questions of the usefulness of the
25 mouse bioassay, the two mutagenicity assays did not correlate with rat results either and an
26 important indicator for carcinogenicity would be lost.
27 Examination of tumor phenotype from TCE, DC A and TCA exposures in mice shows a
28 large heterogeneity, which is also consistent with the heterogeneity observed in human HCC (see
29 Section E.3.1.8, below). The heterogeneity of tumor phenotype has been correlated with survival
30 outcome and tumor aggressiveness in humans and in transgenic mouse models that share some of
31 the same perturbations in gene pathway expression (see Sections E.3.1.8 and E.3.2.1, below).
32 An examination of common pathway disturbances that may be common to all cancers and those
33 of liver tumors shows that there are pathways in common, but that there is greater heterogeneity
34 in disturbance of hepatic pathways in cancer that may make is useful as a marker of disturbances
35 indicative of different targets of carcinogenicity depending on the cellular context and target.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-270 DRAFT—DO NOT CITE OR QUOTE
-------
1 Thus, although primate and human liver may not be as susceptible to HCC as the rodent liver,
2 the pathways leading to HCC in rodents and humans appear to be similar and heterogeneous,
3 with some indicative of other susceptible cellular targets for neoplasia in a differing context.
4
5 E.3.1. State of Science for Cancer and Specifically Human Liver Cancer
6 E.3.1.1. Epigenetics and Disease States (Transgenerational Effects, Effects of Aging and
1 Background Changes)
8 Recently, Wood et al. (2007) published their work on "genomic landscapes" of human
9 breast and colorectal cancers that significantly forwards the understanding of "key events"
10 involved with induction of cancer. They state that there are -80 DNA mutations that alter amino
11 acid in a typical cancer but that examination of the overall distribution these mutations in
12 different cancers of the same type leads to a new view of cancer genome landscapes: they are
13 composed of a handful of commonly mutated genes "mountains" but are dominated by a much
14 larger number of infrequently mutated gene "hills."
15
16 Statistical analyses suggested that most of the ~ 80 mutation in an individual
17 tumor were harmless and that <15 were likely to be responsible for driving the
18 initiation, progression, or maintenance of the tumor.. .Historically the focus of
19 cancer research has been on the gene mountains, in part because they were the
20 only alterations that could be identified with available technologies. However,
21 our data show that vast majority of mutations in cancers do not occur in such
22 mountains. This new view of cancer is consistent with the idea that a large
23 number of mutations, each associated with a small fitness advantage, drive tumor
24 progression. It is the "hills" and not the "mountains" that dominate the cancer
25 genomic landscape.
26
27 The large number of "hills" actually reflects alterations in a much smaller number of cell
28 signaling pathways. Indeed, pathways rather than individual genes appear to govern the course
29 of tumorigenesis.
30
31 It is becoming increasingly clear that pathways rather than individual genes
32 govern the course of tumorigenesis. Mutations in any of several genes of a single
33 pathway can thereby cause equivalent increases in net cell proliferation.... This
34 new view of cancer is consistent with the idea that a large number of mutations,
35 each associated with a small fitness advantage, drive tumor progression.
36
37 Thus, when pathways are altered the same phenotype can arise from alterations in any of several
38 genes.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-271 DRAFT—DO NOT CITE OR QUOTE
-------
1 Consistent with the arguments put forth by Wood et al. (2007) for mutations in cancer is
2 the additional insight into pathway alterations by epigenomic mechanisms, which can act
3 similarly as mutation. Weidman et al. (2007) report that
4
5 cell phenotype is not only dependent on its genotype but also on its unique
6 epigenotype, which is shaped by developmental history and environmental
7 exposures. The human and mouse genome projects identified approximately
8 15,500 and 29,000 CpG islands, respectively. Hypermethylation of CpG-rich
9 regions of gene promoters inhibit expression by blocking the initiation of
10 transcription. DNA methylation is also involved in the allelic inactivation of
11 imprinted genes, the silencing of genes on the inactive X chromosome, and the
12 reduction of expression of transposable elements. Because epigenomic
13 modifications are copied after DNA synthesis by DNMT1, they are inherited
14 during somatic cell replication.. .Inherited and spontaneous or environmentally
15 induced epigenetic alterations are increasingly being recognized as early
16 molecular events in cancer formation. Furthermore, such epigenetic alterations
17 are potentially more adverse than nucleotide mutations because their effects on
18 regional chromatin structure can spread, thereby affecting multiple genetic loci.
19 Although tumor suppressor gene silencing by DNA methylation occurs frequently
20 in cancer, genome-wide hypomethylation is one of the earliest events to occur in
21 the genesis of cancer. Demethylation of the genome can lead to the reactivation
22 of transposable elements, thereby altering the transcription of adjacent genes, the
23 activation of oncogenes such as H-Ras, and biallelic expression of imprinted loci
24 (e.g., loss of IGF2 imprinting).
25
26 Thus, epigenetic modification may be worse than mutation in terms of cancer induction.
27 Dolinoy et al. (2007) report on the role of environmental exposures on the epigenome,
28 especially during critical periods of development and their role in adult disease susceptibility.
29 They report that
30
31 aberrant epigenetic gene regulation has been proposed as a mechanism of action
32 for nongenotoxic carcinogenesis, imprinting disorders, and complex disorders
33 including Alzheimer's disease, schizophrenia, asthma, and autism. Epigenetic
34 modifications are inherited not only during mitosis but also can be transmitted
35 transgenerationally (Rakyan et al., 2002; Rakyan et al., 2003; Anway et al., 2005).
36 The influence on environmental factors on epigenetic gene regulation may also
37 persist transgenerationally despite lack of continued exposure in second, third,
38 and fourth generations (Anway et al., 2005). Therefore if the genome is
39 compared to the hardware in a computer, the epigenome is the software that
40 directs the computer's operation... The epigenome is particularly susceptible to
41 deregulation during gestation, neonatal development, puberty and old age.
42 Nevertheless, it is most vulnerable to environmental factors during embryogenesis
43 because DNA synthetic rate is high, and the elaborate DNA methylation pattern
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-272 DRAFT—DO NOT CITE OR QUOTE
-------
1 and chromatin structure required for normal tissue development is established
2 during early development... 83 imprinted genes have been identified in mice and
3 humans with 29 or about one third being imprinted in both species. Since
4 imprinted genes are functionally haploid, they are denied the protection from
5 recessive mutations that diploidy would normally afford. Imprinted genes that
6 have been linked to carcinogenesis include IGF2 (bladder, lung, ovarian and
7 others), IGF2R (breast, colon, lung, and others), and Neuronatin (pediatric
8 leukemia).
9
10 Bjornsson et al. (2008) recently reported that not only were there time-dependent changes
11 in global DNA methylation within the same individuals in 2 separate populations in widely
12 separated geographic locations, these changes showed familial clustering in both increased and
13 decreased methylation. These results were not only suggested to support the relationship of age-
14 related loss of normal epigenetic patterns as a mechanism for late onset of common human
15 diseases but also that losses and gains of DNA methylation observed over time in different
16 individuals could contribute to disease with the example provided of cancer which is associated
17 with both hypomethylation and hypermethylation through activation of oncogenes and silencing
18 of tumor suppressor genes. The study also showed considerable interindividual age variation,
19 with differences accruing over time within individuals that would be missed by studies that
20 employed group averaging.
21 The review by Reamone-Buettner and Borlak (2007) provide insight into the role of
22 noncoding RNAs in diseases such as cancer. They report that
23
24 a large number of noncoding RNAs (ncRNAs) play important role in regulating
25 gene expressions, and advances in the identification and function of eukaryotic
26 ncRNAs, e.g., microRNAs and their function in chromatin organization, gene
27 expression, disease etiology have been recently reviewed. The regulatory
28 pathways mediated by small RNAs are usually collectively referred to as RNA
29 interference (RNAi) or RNA-mediated silencing. RNAi can be triggered by small
30 double-stranded RNA (dsRNA) either introduced exogenously into cells as small
31 interfering siRNAs or that have been produced endogenously from small non-
32 coding RNAs known as microRNAs (miRNAs). The dsRNAs are
33 characteristically cleaved by the ribonuclease Ill-enzyme Dicer into 21- to 23 nt
34 duplexes and the resulting fragments base-pair with complementary mRNA to
35 target cleavage or to repress translation.. .Two mechanisms exist of miRNA-
36 mediated gene regulation, degradation of the target mRNA, and translational
37 repression. Whether one or the other of these mechanisms is used depends on the
38 degree of the complementary between the miRNA and target mRNA. For a near
39 perfect match, the Argonaute protein in the RNA-induced silencing complex
40 (RISC) cleaves the mRNA target, which is destined for subsequent degradation by
41 ribonucleases. In the situation of a less degree of complimentarity, commonly
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-273 DRAFT—DO NOT CITE OR QUOTE
-------
1 occurring in humans, the translational repression mechanism is used to control
2 gene expression. However, the exact mechanism for translational inhibition is
3 unclear.
4
5 The varying degrees in complimentarity would help explain the large number of genes that could
6 be affected by miRNA and pleiotropic response.
7 The review by Feinberg et al. (2006) specifically addresses the epigenetic progenitor
8 origin of human cancer. They conclude that epigenetic alterations are ubiquitous and serve as
9 surrogate alterations for genetic change (oncogene activation, tumor-suppressor-gene silencing),
10 by mimicking the effect of genetic change. They report that:
11
12 Advances in characterizing epigenetic alterations in cancer include global
13 alterations, such as hypomethylation of DNA and hypoacetylation of chromatin,
14 as well as gene-specific hypomethylation and hypermethylation. Global DNA
15 hypomethylation leads to chromosomal instability and increased tumour
16 frequency, which has been shown in vitro and in vivo in mouse models, as well as
17 gene-specific oncogene activation, such as R-ras in gastric cancer, and cyclin D2
18 and maspin in pancreatic cancer. In addition, the silencing of tumour-suppressor
19 genes is associated with promoter DNA hypermethylation and chromatin
20 hypoacetylation, which affect divergent genes such as retinoblastoma 1 (RBI),
21 p!6 (also known as cyclin-dependent kinase inhibitor 2A (CDKN2A), von
22 Hippel-Lindau tumor suppressor (VHL), and MutL protein homologue (MLH1).
23
24 Genetic mechanisms are not the only path to gene disruption in cancer.
25 Pathological epigenetic changes - non-sequence-based alteration that are inherited
26 through cell division - are increasingly being considered as alternatives to
27 mutations and chromosomal alterations in disrupting gene function. These
28 include global DNA hypomethylation, hypermethylation and hypomethylation of
29 specific genes, chromatin alterations and loss of imprinting. All of these can lead
30 to aberrant activation of growth-promoting genes and aberrant silencing of
31 tumour-suppressor genes.
32
33 Most CG dinucleotides are methylated on cytosine residues in vertebrate
34 genomes. CG methylation is heritable, because after DNA replication the DNA
35 methyltransferase 1, DNMT1, methylates unmethylated CG on the base-paired
36 strand. CG dinucleotides within promoters within promoters tend to be protected
37 from methylation. Although individual genes vary in hypomethylation, all
38 tumours have shown global reduction of DNA methylation. This is a striking
39 feature of neoplasia.
40
41 In addition to global hypomethylation, promoters of individual genes show
42 increased DNA methylation levels. Hypermethylation of tumour-suppressor
43 genes can be tumour-type specific. An increasing number of genes are found to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-274 DRAFT—DO NOT CITE OR QUOTE
-------
1 be normally methylated at promoters but hypomethylated and activated in the
2 corresponding tumours. These include R-RAs in gastric cancer, melanoma
3 antigen family A, 1(MAGE1) in melanoma, maspin in gastric cancer, S100A4 in
4 colon cancer, and various genes in pancreatic cancer.
5
6 Our genetic material is complexed with proteins in the form of histones in a one-
7 to-one weight ratio. Core histones H2A, H2B, H3 and H4 form nucleosome
8 particles that package 147 bp of DNA, and the linker histone HI packages more
9 DNA between core particles, forming chromatin. It is chromatin and not just
10 DNA, that is the substrate for all processes that affect genes and chromosomes. In
11 recent years, it has become increasingly evident that chromatin, like DNA
12 methylation, can impart memory to genetic activity. There are dozens of post-
13 translational histone modifications. Studies in many model systems have shown
14 that particular histone modifications are enriched at sites of active chromatin
15 (histone H3 and H4 hyperacetylation, lysing at 4 and H3 (H3-K4) dimethylation
16 and trimethylation, and H3-K79 methylation) and others are enriched at sites of
17 silent chromatin (H3-K9 and H3-K27 methylation). These and other histone
18 modifications survive mitosis and have been implicated in chromatin memory.
19
20 Overproduction of key histone methyltransferases that catalyze the methylation of
21 either H3-K4 or H3-K27 residues are frequent events in neoplasia. Global
22 reductions in monoacetylated H4-K16 and trimethylated H4-K20 are general
23 features of cancer cells.
24
25 Genomic imprinting is parent-of-origin-specific gene silencing. It results from a
26 germ-line mark that causes reduced or absent expression of a specific allele of a
27 gene in somatic cells of the offspring. Imprinting is a feature of all mammals
28 affecting genes that regulate cell growth, behaviour, signaling, cell cycle and
29 transport; moreover, imprinting is necessary for normal development. Imprinting
30 is important in neoplasia because both gynogenotes (embryos derived only from
31 the maternal genetic complement) and androgenotes (embryos derived only from
32 the paternal genetic complement) form tumours - ovarian teratomas, and
33 hydtidiform moles/ choriocarcinomas, respectively. Loss of imprinting (LOI)
34 refers to activation of the normally silenced allele, or silencing of the normally
35 active allele, of an imprinted gene. LOI of the insulin-like growth factor 2 gene
36 (IGF2) accounts for half of Wilms tumours in children. LOI of IGF2 is also a
37 common epigenetic variant in adults and is associated with a fivefold increased
38 frequency of colorectal neoplasia. LOI of IGF2 might cause cancer by increasing
39 the progenitor cell population in the kidney in Wilm's tumor and in the
40 gastrointestinal tract in colorectal cancer.
41
42 Feinberg et al. (2006) propose that epigenetic changes can provide mechanistic unity to
43 understanding cancer, they can occur earlier and set the stage for genetic alterations, and have
44 been linked to the pluripotent precursor cells from which cancers arise. "To integrate the idea of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-275 DRAFT—DO NOT CITE OR QUOTE
-------
1 these early epigenetic events, we propose that cancer arises in three steps; an epigenetic
2 disruption of progenitor cells, an initiating mutation and genetic and epigenetic plasticity."
3
4 The first step involves an epigenetic disruption of progenitor cells in a given
5 organ or system, which leads to a polyclonal precursor population of neoplasia-
6 ready cells. These cells represent a main target of environmental, genetic and
7 age-dependent exposure that largely accounts for the long latency period of
8 cancer. Epigenetic disruption might perturb the normal balance between
9 undifferentiated progenitor cells and differentiated committed cells within a given
10 anatomical compartment, either in number or in their capacity for aberrant
11 differentiation, which provides a common mechanism of neoplasia.
12
13 All tumours show global changes in DNA methylation, and DNA methylation is
14 clonally inherited through cell division. Because the conventional genetic
15 changes in cancer are also clonal, global hypomethylation would have to occur
16 universally, at the same moment as the mutational changes, which seems unlikely.
17 This suggests that global DNA hypomethylation (and global reductions of specific
18 histone modifications) precedes genetic change in cancer. Similarly,
19 hypermethylation of tumour-suppressor genes has been observed in the normal
20 tissue of patients in which the same gene is hypermethylated in the tumour tissue.
21 Recent data demonstrate LOT of IGF2 throughout the normal colonic epithelium
22 of patients who have LOI-associated colorectal cancer. LOT is associated with
23 increased risk of intestinal cancers in both humans and mice. A specific change
24 in the epithelium is seen in mice that are engineered to have biallelic expression
25 of IGF2 - a shift in the proportion of progenitor to differentiated cells throughout
26 the epithelium; a similar abnormality was observed in humans with LOT of IGF2.
27
28 The proposed existence of the epigenetically disrupted progenitors of cancer
29 implies that the earliest stages in neoplastic progression occur even before what a
30 pathologist would recognize as a benign pre-neoplastic lesion. Such alterations
31 are inherently polyclonal. This is in contrast with the widely accepted model of
32 cancer as a monoclonal disorder that arises from an initiating mutation- a model
33 that was proposed and accepted when little was known about epigenetic
34 phenomena in cancer.
35
36 Thus, Feinberg et al. (2006) provide a hypothesis for the latency period of cancer and
37 suggest that epigenetic changes predate mutational ones in cancer. Tissues that look
38 phenotypically "normal" may harbor epigenetic changes and predispositions toward neoplasia.
39 In regard to what cells may be targets or epigenetic changes that can be "progenitor cells" in the
40 case of cancer, Feinberg et al. (2006) define such cell having "capacity for self-renewal and
41 pluripotency - over their tendency toward limited replicative potential and differentiation."
42 Within the liver, there are multiple cell types that would fit such a definition including those who
43 are considered "mature" (see Section E.3.1.4, below). Feinberg et al. (2006) also note that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-276 DRAFT—DO NOT CITE OR QUOTE
-------
1 epigenetic states can be continuously modified to become heterogeneous at all states of the
2 neoplastic process.
3
4 Telomere erosion results in chromosome shortening and uncapped ends that begin
5 to fuse and the resulting dicentric chromosomes break at anaphase. DNA
6 palindromes have recently been found to form at high levels in cancer cells. Like
7 telomere erosion, DNA palindrome formation can lead to genetic instability by
8 initiating bridge-breakage-fusion cycles. However, it is not known how or
9 exactly when palindromes form, although they appear early in cancer progression.
10 Epigenetic instability can also promote cancer through pleiotropic alterations in
11 the expression of genes that modify chromatin.
12
13 Epigenetic changes are reversible but the changes can initiate irreversible genetic
14 changes. Permanent epigenetic changes can have an epigenetic basis. On a
15 background of cancer-associated epigenetic instability, the effects of mutations in
16 oncogenes and tumour -suppressor genes might be exacerbated. Therefore the
17 risk of developing malignancy would be much higher for a given mutations event
18 if it occurred on the background of epigenetic disruption.
19
20 The environmental dependence of cancer fits an epigenetic model generally for
21 human disease - the environment might influence disease onset not simply
22 through mutational mechanisms but in epigenetically modifying genes that are
23 targets for either germline or acquired mutation; that is, by allowing genetic
24 variates to be expressed. Little is known about epigenetic predispositions to
25 cancer, but a recent twin study indicates that, similar to cancer risk, global
26 epigenetic changes show striking increase with age.
27
28 Environmental insults might affect the expression of tumour-progenitor genes,
29 leading to both genetic and epigenetic alterations. Liver regeneration after tissue
30 injury leads to widespread hypomethylation and hypermethylation of individual
31 genes; both of these epigenetic changes occur in cancer.
32
33 In regard to the implications of epigenomic changes and human susceptibility to toxic
34 insult, the review by Szyf (2007) provides additional insights.
35
36 The basic supposition in the field has been that the interindividual variations in
37 response to xenobiotic are defined by genetic differences and that the main hazard
38 anticipated at the genomic level from xenobiotic is mutagenesis or physical
39 damage to DNA. In accordance with this basic hypothesis, the main focus of
40 attention in pharmacogenetics has been on identifying polymorphisms in genes
41 encoding drug metabolizing enzymes and receptors. New xenobiotics were
42 traditionally tested for their genotoxic effects. However, it is becoming clear that
43 epigenetic programming plays an equally important role in generating
44 interindividual phenotypic differences, which could affect drug response.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-277 DRAFT—DO NOT CITE OR QUOTE
-------
1 Moreover, the emerging notion of the dynamic nature of the epigenome and its
2 responsibility to multiple cellular signaling pathways suggest that it is potentially
3 vulnerable to the effects of xenobiotics not only during critical period in
4 development but also later in life as well. Thus, non-genotoxic agents might
5 affect gene function through epigenetic mechanisms in a stable and long-term
6 fashion with consequences, which might be indistinguishable form the effects of
7 physical damage to the DNA. Epigenetic programming has the potential to
8 persist and even being transgenerationally transmitted (Anway et al., 2005) and
9 this possibility creates a special challenge for toxicological assessment of safety
10 of xenobiotics. Any analysis of interindividual phenotype diversity should
11 therefore take into account epigenetic variations in addition to genetic sequence
12 polymorphisms. Whereas, a germ-line polymorphism is a static property of an
13 individual and might be mapped in any tissue at any point in life, epigenetic
14 differences must be examined at different time points and at diverse cell types.
15
16 Karpinets and Foy (2005) propose that epigenetic alterations precede mutations and that
17 succeeding mutations are not random but in response to specific types of epigenetic changes the
18 environment has encouraged. This mechanism was also suggested as to both explain the delayed
19 effects of toxicant exposure and the bystander effect of radiation on tumor development, which
20 are inconsistent with the accepted mechanism of direct DNA damage.
21
22 In a study of ionizing radiation, non-irradiated cells acquired mutagenesis through
23 direct contact with cells whose nuclei had previously been irradiated with alpha-
24 particles (Zhou et al., 2003). Molecular mechanisms underlying these
25 experimental findings are not known but it is believed that it may be a
26 consequence of bystander interactions involving intercellular signaling and
27 production of cytokines (Lorimore et al., 2003).
28
29 Caldwell and Keshava (2006) report that
30
31 aberrant DNA methylation has emerged in recent years as a common hallmark of
32 all types of cancers with hypermethylation of the promoter region of specific
33 tumor suppressor genes and DNA repair genes leading to their silencing (an effect
34 similar to their mutation), and genomic hypomethylation (Ballestar and Esteller,
35 2002; Berger and Daxenbichler, 2002; Herman et al.,1998; Pereira et al. 2004;
36 Rhee et al., 2002). Whether DNA methylation is a consequence or cause of cancer
37 is a long-standing issue (Ballestar and Esteller, 2002). Fraga et al. (2004, 2005)
38 report global loss of monoacetylation and trimethylation of histone H4 as
39 common a hallmark of human tumor cells but suggest genomone-wide loss of 5-
40 methylcytosine (associated with the acquisition of a transformed phenotype) does
41 not exist as a static predefined value throughout the process of carcinogenesis but
42 as a dynamic parameter (i.e., decreases are seen early and become more marked in
43 later stages).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-278 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.3.1.2. Emerging Technologies, DNA andsiRNA, miRNA Microarrays—Promise and
2 Limitations for Modes of Action (MOAs)
3 Currently new approaches are emerging for the study of changes in gene expression and
4 protein production induced by chemical exposure that could be related to their toxicity and serve
5 as an anchor for determining similar patterns between rodent models and human diseases or risks
6 of chemically-induced health impacts. Such approaches have the promise to extend the
7 definitions of "genotoxic" and "nongenotoxic" effects which with the advent of epigenomic
8 study have become obsolete as they assume only alteration of the DNA sequence is important in
9 cancer induction and progression. However, not only is phenotypic anchoring an issue in regard
10 to the differing cell types, regions, and lobes of the liver (see Section E.I. 2, above), it is also an
11 issue for overall variability of response between animals and is critical for interpretation of
12 microarray and other genomic database approaches. As shown in the discussions of TCE effects
13 in animal models, TCE treatment resulted in a large variability in response between what are
14 supposed to be relatively homogeneous genetically similar animals and there was an apparent
15 difference in response between studies using the same paradigm. It is important that as varying
16 microarray approaches and analyses of TCE toxicity or of potential MO As are published, the
17 issue of phenotypic anchoring at the cellular to animal level is addressed. Several studies of
18 TCE microarray results and those of PPARa agonists have been reported in the literature in an
19 attempt to discern MO As. Issues related to conduct of these experiments and interpretation of
20 their results are listed below.
21 Perhaps one of the most important studies of this issue has been reported by Baker et al.
22 (2004). The ILSIHESI formed a hepatotoxicity working group to evaluate and compare
23 biological and gene expression responses in rats exposed to well-studied hepatotoxins (Clofibrate
24 and methapyrilene), using standard experimental protocol and to address the following issues: (a)
25 how comparable are the biological and gene expression data from different laboratories running
26 identical in vivo studies (b) how reproducible are the data generated across laboratories using the
27 same microarray platform (c) how do data compare using different microarray platforms; (d)
28 how do data compare using RNA from pooled and individual animals; (e) do the gene expression
29 changes demonstrate time- and dose-dependent responses that correlate with known biological
30 markers of toxicity? (Baker et al., 2004). The rat model studied was the male S-D rat (57 or
31 60-66 days of age) exposed to 250 or 25 mg/kg/d Clofibrate for 1, 3 or 7 days. Two separate in
32 vivo studies were conducted: one at Abbott Laboratories and on at GlaxoSmithKline (GSK, in
33 United Kingdom [UK]). There was a difference in biological response between the two
34 laboratories. The high dose (250 mg/kg/d) group at Day 3 had a 15% increase in liver weight
35 relative to body weight in the GSK study, compared with a 3% liver weight increase in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-279 DRAFT—DO NOT CITE OR QUOTE
-------
1 Abbott study. At 7-days, there was a 31% liver weight increase in the GSK study and 15% in the
2 Abbott study. Observed changes in clinical chemistry parameters also indicated difference in the
3 biological response of the in vivo study concordant with difference in liver weight. A significant
4 reduction in total cholesterol levels was seen in the GSK study at the high dose for all time
5 points. However, the Abbott study demonstrated a significant reduction only at one dose and
6 time point. The incidence of mitotic figures also differed between the labs. In both studies there
7 was a 2-3 times greater Acyl-CoA enzyme (ACOX) activity at the high dose but no difference
8 from control in the low dose. Again the GSK lab gave greater response. For microarrays, GSK
9 and ULR pooled samples from each treatment group of four animals. U.S. EPA did some of the
10 microarray analyses as well as GSK and ULR (GSK in UK). It is apparent that although the
11 changes in genes were demonstrated by both laboratories, there were quantitative differences in
12 the fold change values observed between the two sites.
13 The U.S. EPA analyzed gene expression in individual RNA samples obtained from Day 7
14 high and low-dose animals that had been treated at Abbot. GSK (U.S.) and ULR analyzed gene
15 expression in pooled RNA from Day 7 high and low dose animals treated at GSK (UK). Gene
16 expression data from individual animal samples indicated that 7 genes were significantly
17 upregulated (maximum of 7.2-fold) and 12 were down regulated (maximum of 4.3-fold decrease)
18 in the high-dose group. The low-dose group generated only one statistically significant gene
19 expression change, namely heat shock protein 70 (HSP70). In comparison, expression changes
20 in the 7-day pooled high-dose samples analyzed by GSK (U.S.) ranged from 43.3-fold to a
21 3.5-fold decrease. Changes in these same samples analyzed by ULR ranged from a 4.9-fold
22 increase to a 4.3-fold decrease. As an example, the microarray fold change at 7-day 250 mg/kg/d
23 Clofibrate showed a 3.8-fold increase for U.S. EPA individual animals sampled, and 2.2-fold
24 increase for pooled samples by ULR, and a 20.3-fold increase in pooled samples by GSK (U.S.)
25 for CYP4A1 (Baker et al., 2004). Thus, these results show a very large difference not only
26 between treatment groups but between pooled an nonpooled data and between labs analyzing the
27 same RNA.
28 Not only was there a difference in DNA microarray results but a comparison of gene
29 expression data from Day 7 high-dose samples obtained using quantitative realtime PCR versus
30 data generated using cDNA microarrays has shown a quantitative difference but qualitative
31 similar patterns. Although both methods of quantitative real time PCR on the pooled sample
32 showed the PPARa gene to be down regulated, the GSK (U.S.) pooled sample microarray
33 analysis indicated upregulation; the URL pooled and U.S. EPA individual microarray analyses
34 showed no change. The microarray for PPARa at 7-day 250 mg/kg/d Clofibrate showed no
35 change for individual animals (U.S. EPA), no change for pooled samples (ULR) and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-280 DRAFT—DO NOT CITE OR QUOTE
-------
1 upregulation of 1.8-fold value for pooled samples for GSK(U.S.). The quantitative real time
2 PCR on the pooled sample using Taqman gave a 4.5-fold down regulation and using SYBR
3 Green gave a 1.2-fold down regulation of PPARa.
4 Baker et al. (2004) reported that the pooling of samples for microarray analysis has been
5 used in the past to defray the cost of microarray experiments, reduce the effect of biological
6 variation, and in some cases overcome availability of limiting amounts of tissues. Unfortunately
7 this approach essentially produced a sample size (ri) of one animal. Repeated microarray
8 experiments with such pooled RNA produces technical replicates as opposed to true biological
9 replicates and thus, does not allow calculation of biologically significant changes in gene
10 expression between different dose groups or time points. Another possible consequence of
11 pooling is to mask individual gene changes and leave open the possibility of introducing error
12 due to individual outlier responses.
13 Woods et al. (2007a) note that
14
15 because toxicogenomics is a relatively novel technology, there are a number of
16 limitations that must be resolved before array data is widely accepted. Microarray
17 studies have been touted as being highly sensitive for detecting toxic responses at
18 much earlier time points and/or lower doses than histopathology, clinical
19 chemistry or other traditional toxicological assays can detect. However, based on
20 the nature of the assay, measurements of extreme levels of gene expression - low
21 or high -are thought to be unreliable. Also the reproducibility of microarray
22 experiments has raised concerns. "Batch effects" based on the day, user, and
23 laboratory environment have been observed in array datasets. To address these
24 concerns, confirmation of microarray-derived gene expression profiles is typically
25 performed using quantitative real time polymerase chain reaction (RT-PCR) or
26 Northern blot analysis.
27
28 In addition to the issues raised above, Waxman and Wurmbach (2007) raise issues
29 regarding how quantitative realtime PCR experiments are conducted. They state that cancer
30 development affects almost all pathways and genes including the "housekeeping" genes, which
31 are involved in the cell's common basic functions (e.g., glyceraldehyde-3-phosphate
32 dehydrogenase [GADPH], beta actin [ACTB], TATA-binding protein, ribosomal proteins, and
33 many more). However, "many of these genes are often used to normalize quantitative real-time
34 RT-PCR (qPCR) data to account for experimental differences, such as differences in RNA
35 quantity and quality, the overall transcriptional activity and differences in cDNA synthesis.
36 GADPH and ACTB are most commonly used for normalization, including studies of cancer."
37 Waxman and Wurmbach (2007) suggest that despite the fact that it has been shown that these
38 genes are differentially expressed in cancers, including colorectal-, prostate-, and bladder-cancer,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-281 DRAFT—DO NOT CITE OR QUOTE
-------
1 some qPCR studies on hepatocellular carcinoma used GAPDH or ACTB for normalization.
2 Since many investigations on cancer include multiple comparisons, and analyze different stages
3 of the disease, such as normal tissue, preneoplasm, and consecutive stages of cancer, "it crucial
4 to find an appropriate gene for normalization" whose expression is constant throughout all
5 disease stage and not response to treatment. For liver cancers associated with exposure to
6 hepatitis C virus (HCV), Waxman and Wurmbach (2007) reported that differing states, including
7 preneoplastic lesions (cirrhosis and dysplasia) and consecutive stages of hepatocellular
8 carcinoma, had differential expression of "housekeeping" genes and that using them for
9 normalization had an effect on the fold change of qPCR data and on the general direction (up or
10 down) of differentially expressed genes. For example, GAPDH was strongly upregulated in
11 advanced and very advanced stages of hepatocellular carcinoma (in some samples up to 7-fold)
12 and ACTB was up-regulated 2- to 3-fold in many advanced and very advanced tumor samples.
13 Waxman and Wurmbach (2007) conclude that
14
15 microarray data are known to be highly variable. Due to its higher dynamic range
16 qPCR is thought to be more accurate and therefore is often used to corroborate
17 microarray results. Mostly, general direction (up and down-regulation) and rank
18 order of the fold-changes are similar, but the levels of the fold changes of
19 microarray experiments differ compared to qPCR data and show a marked
20 tendency of being smaller. This effect is more pronounced as the fold change is
21 very high.
22
23 In relation to use of gene expression and indicators of cancer causation, Volgelstein and
24 Kinzler (2004) make important points regarding their use:
25
26 Levels of gene expression are unreliable indicators of causation because
27 disturbance of any network invariably leads to a multitude of such changes only
28 peripherally related to the phenotype. Without better ways to determine whether
29 an unmutated but interesting candidate gene has a causal role in neoplasia, cancer
30 researchers will likely be spending precious time working on genes only
31 peripherally related to the disease they wish to study.
32
33 This is important caveat for gene expression studies for MOA that are "snapshots in time"
34 without phenotypic anchoring and even more applicable to experimental paradigms where there
35 is ongoing necrosis or toxicity in addition to gene changes that may or may not be associated
36 with neoplasia.
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-282 DRAFT—DO NOT CITE OR QUOTE
-------
1 For an endpoint that is not as complex as neoplasia, there are issues regarding uses of
2 microarray data. In regard to the determination of acute liver toxicity caused by one of the most
3 studied hepatotoxins, acetaminophen, and its correlation with microarray data, Beyer et al.
4 (2007) also have reported the results of a landmark study examining issues regarding use of this
5 approach.
6
7 The biology of liver and other tissues in normal and disease states increasingly is
8 being probed using global approaches such as microarray transcriptional profiling.
9 Acceptance of this technology is based principally on a satisfactory level of
10 reproducibility of data among laboratories and across platforms. The issue of
11 reproducibility and reliability of genomics data obtained from similar
12 (standardized) biological experiments performed in different laboratories is
13 crucial to the generation and utility of large databases of microarray results.
14 While several recent studies uncovered important limitation of expression
15 profiling of chemical injury to cells and tissues (Baker et al 2004; Beekman et al
16 2006; Ulrich et al 2004), determining the effects of intralaboratory variables on
17 the reproducibility, validity, and general applicability of the results that are
18 generated by different laboratories and deposited into publicly available databases
19 remains a gap... The National Institutes of Environmental Health Sciences
20 (NIEHS) established the Toxicogenomics Research Consortium to apply the
21 collective and specialized expertise from academic institutions to address issues in
22 integrating gene expression profiling, bioinformatics, and general toxicology.
23 Key elements include developing standardized practices for gene expression
24 studies and conducting systematic assessments of the reproducibility of traditional
25 toxicity endpoints and microarray data within and among laboratories. To this
26 end the consortium selected the classical hepatotoxicant acetaminophen (APAP)
27 for its proof of concept experiments. Despite more than 30 years of research on
28 APAP, we are far from a complete understanding of the mechanisms of liver
29 injury, risk factors, and molecular markers that predict clinical outcome after
30 poisoning. APAP-induced hepatotoxicity was performed at seven geographically
31 dispersed Centers. Parallel studies with N-acetyl-m-aminophenol (AMAP), the
32 non-hepatotoxic isomer of APAP, provided a method to isolate transcripts
33 associated with hepatotoxicity (Beyer et al., 2007).
34
35 Beyer et al identified potential sources of interlaboratory variability when microarray
36 analyses were conducted by one laboratory on RNA samples generated in different laboratories
37 but using the same experimental paradigm and source of animals. Toxic injury by APAP
38 showed variability across Centers and between animals (e.g., percent liver affected by necrosis
39 [<20 to 80% at one time period and 0 to 60% at another], control animal serum ALT [3-fold
40 difference], and in glutathione depletion [<5 to >60%] between centers). There was concordance
41 between APAP toxicity as measured in individual animals (rather than expressed as just a mean
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-283 DRAFT—DO NOT CITE OR QUOTE
-------
1 with SE) and transcript!onal response. Of course the variability between gene platforms and
2 processing of the microarrays had been reduced by using the same facility to do all of the
3 microarray analyses. However, the results show that phenotypic anchoring of gene expression
4 data are required for biologically meaningful meta-analysis of genomic experiments.
5 Woods et al. (2007a) note that
6
7 improvements should continue to be made on statistical analysis and presentation
8 of microarray data such that it is easy to interpret. Prior to the current advances in
9 bioinformatics, the most common way of reporting results of microarray studies
10 involved listing differentially expressed genes, with little information about the
11 statistical significance or biological pathways with which the genes are
12 associated.
13
14 However, there are issues with the use of "Classifiers" or predictive genomic computer programs
15 based on genes showing altered expression in association with the observed toxicities.
16
17 Although these metrics built on different machine learning algorithms could be
18 useful in estimating the severity of potential toxi cities induced by compounds, the
19 applications of these classifiers in understanding the mechanisms of drug-induced
20 toxicity are not straightforward. In particular this approach is unlikely to
21 distinguish the upstream causal genes from the downstream responsive genes
22 among all the genes associated with an induced toxicity. Without knowledge of
23 the causal sufficiency order, designing experiments to test predicted toxicity in
24 animal models remains difficult" (Dai et al., 2007).
25
26 Ulrich (2003) states limitation of microarray analysis to study nuclear receptors (e.g., PPARa).
27
28 Nuclear receptors comprise a large group of ligand-activated transcription factors
29 that control much of cellular metabolism. Toxicogenomics is the study of the
30 structure and output of the entire genome as it related and responds to adverse
31 xenobiotic exposure. Traditionally, the genes regulated by nuclear receptors in
32 cells exposed to toxins have been explored at the mRNA and protein levels using
33 northern and western blotting techniques. Though effective when studying the
34 expression of individual genes, these approaches do not enable the understanding
35 of the myriad of genes regulated by individual receptors or of the crosstalk
36 between receptors.. .Discovery of the multiple genes regulated by each receptor
37 type has thus been driven by technological advances in gene expressional
38 analysis, most commonly including differential display, RT-PCR and DNA
39 microarrays., and in the development or receptor transgenic and knockout animal
40 models. There is much cross talk between receptors and many agonists interact
41 with multiple receptors. Off target effects cannot be predicted by target
42 specificity. Though RCR can affect transcription directly, much of its effects are
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-284 DRAFT—DO NOT CITE OR QUOTE
-------
1 exerted through heterodimeric hinging with other nuclear receptors (PXR, CAR,
2 PPARa, PPARy, FXR, LXR, TR) (Ulrich, 2003).
3
4 Another tool recent developed is gene silencing by introduction of siRNA. Dai et al.
5 (2007) note issues involved in the siRNA to change gene expression for exploration of MO A etc.
6 to include the potential of off-target effects, incomplete knockdown, and nontargeting of splice
7 variants by the selected siRNA sequence. Using knockdown of PPARa in mice, Dai et al. (2007)
8 report "PPARa knockdown was variable between mice ranging from ~ 80 % knockdown to little
9 or no knockdown and that differing siRNAs gave different patterns of gene expression with some
10 grouped with PPARa -/- null mice but others grouped with expression patterns of mice injected
11 with control siRNA or Ringers buffer alone and showing no PPARa knockdown." Dai et al
12 concluded that it is possible that it is the change in PPARa levels that is important for perturbing
13 expression of genes modulated by PPARa rather than the absolute levels of PPARa. Not only is
14 the finding of variability in knockdowns by siRNA technologies important but The finding that
15 level of PPAR is not necessarily correlated with function and that it could be the change and not
16 absolute level that matters in modulation in gene expression by PPARa is of importance as well.
17 How an animal responds to decreased PPARa function may also depend on its gender. Dai et al.
18 (2007) observed more dramatic phenotypes in female vs. male mice treated with siRNA and
19 noted that in aged PPARa -/- mice, Costet et al. (1998) have reported sexually dimorphic
20 phenotypes including obesity and increased serum triglyceride levels in females, and steatosis
21 and increased hepatic triglyceride levels in males.
22 In regard to the emerging science and preliminary reports of the effects of microRNA as
23 oncogenes and tumor suppressors and of possible importance to hypothesized MO As for liver
24 cancer, the same caveats as described for DNA microarray analyses all apply along with
25 additional uncertainties. miRNAs repress their targeted mRNAs by complementary base pairing
26 and induction of the RNA interference pathway. Zhang et al. (2007) report Northern blot
27 detection of gene expression at the mRNA level and its correlation with miRNA expression in
28 cancer cells as well as realtime PCR. These PCR-based analyses quantify miRNA precursors
29 and not the active mature miRNAs. However, they report that the relationship between
30 pri-miRNA and mature miRNA expression has not been thoroughly addressed and is critical in
31 order to use real time PCR analysis to study the function of miRNAs in cancers. They go on to
32 state that
33
34 although Northern Blotting is a widely used method for miRNA analysis, it has
35 some limitations, such as unequal hybridization efficiency of individual probes
36 and difficulty in detecting multiple miRNAs simultaneously. For cancer studies,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-285 DRAFT—DO NOT CITE OR QUOTE
-------
1 it is important to be able to compare the expression pattern of all known miRNAs
2 between cancer cells and normal cells. Thus, it is better to have methods which
3 detect all miRNA expression at a single time.. .Although Northern blot analysis,
4 real-time PCR, and miRNA microarray can detect the expression of certain
5 miRNAs and determine which miRNAs may be associated with cancer formation,
6 it is difficult to determine whether or not miRNAs play a unique role in cancers.
7 Also these techniques cannot directly determine the correlation between mRNA
8 expression levels and whether the up-regulation or down-regulation of certain
9 miRNAs is the cause of cancer or a downstream effect of the disease.. .Many
10 miRNA genes have been found that are significantly overexpressed in different
11 cancers. All of them appear to function as oncogenes; however, only a few of
12 them have been well characterized.
13
14 Zhang et al. (2007) suggest that bioinformatic studies indicate that numerous genes are the
15 targets of miR-17-92: more than 600 for miR-19a and miR-20, two members of the miR-17-92
16 cluster.
17 Cho (2007) state that
18
19 though more than 530 miRNAs have been identified in human, much remains to
20 be understood about their precise cellular function and role in the development of
21 diseases.. .Although each miRNA can control hundreds of target genes, it remains
22 a great challenge to identify the accurate miRNA targets for cancer research.
23
24 Thus, miRNAs have multiple targets so, like other transcription factors, may have pleotropic
25 effects that are cell, timing, and context specific.
26 Vogelstein and Kinzler (2004) state "in the last decade many important gene responsible
27 for the genesis of various cancers have been discovered." Most importantly they and others
28 suggest that pathways rather than individual gene expression should be the focus of study. As a
29 specific example, Volgelstein and Kinzler note
30
31 another example of the reason for focusing on pathways rather than individual
32 genes has been provided by studies of TP53 tumor-suppressor gene. The p53
33 protein is a transcription factor that normally inhibits cell growth and stimulates
34 cell death when induced by cellular stress. The most common way to disrupt the
35 p53 pathway is through a point mutation that inactivates its capacity to bind
36 specifically to its cognate recognition sequence. However, there are several other
37 ways to achieve the same effects, including amplification of the MDM2 gene and
38 infection with DNA tumor viruses whose products bind to p53 and functionally
39 inactivate it.
40
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-286 DRAFT—DO NOT CITE OR QUOTE
-------
1 In regard to cellular anchoring for gene expression or pathway alterations associated with
2 cancer and the importance of "context" of gene expression changes, Vogelstein and Kinzler
3 (2004) give several examples.
4
5 In solid tumors the important of the interactions between stroma and epithelium is
6 becoming increasingly recognized (e.g., the importance of the endothelial
7 cell)... One might expect that a specific mutation of a widely expressed gene
8 would have identical or at least similar effects in different mammalian cell types.
9 But this is not in general what is observed. Different effects of the same mutation
10 are not only found in distinct cell types; difference can even be observed in the
11 same cell types, depending on when the mutation occurred during the tumorigenic
12 process. The RAS gene mutations provide informative examples of these
13 complexities. KRAS2 gene mutation in normal pancreatic duct cells seem to
14 initiate the neoplastic process, eventually leading to the development of
15 pancreatic cancer. The same mutations occurring in normal colonic or ovarian
16 epithelial cells lead to self-limiting hyperplastic or borderline lesions that do not
17 progress to malignancy. In many human and experimental cancers, RAS genes
18 seem to function as oncogenes. But RAS genes can function as suppressor genes
19 under other circumstances, inhibiting tumorigenesis after administration of
20 carcinogens to mice. These and similar observation on other cancer genes are
21 consistent with the emerging notion that signaling molecules play multiple roles
22 at multiple time, even in the same cell type. However, the biochemical bases for
23 such variations among cancer cells are almost unknown.
24
25 In regard to the major pathways and mediators involved in cancer several investigators
26 have reported a coherent set that are involved in many types of cancers. Vogelstein and Kinzler
27 (2004) note that major pathways and mediators include p53, RB, WNT, E-cadherin, GL1, APC,
28 ERK, RAS:GTP, P13K,SMAD, RTK BAD, BAX, and H1F1. In regard to coherence and site
29 concordance between animal and human data, the disturbance of a pathway in one species may
30 result in the different expression of tumor pattern in another but both linked to a common
31 endpoint of cancer. Thus, pathways rather than a single mutation should be the focus of MO A
32 and cancer as several actions can be manifested by one pathway or change at one time that lead
33 to cancer.
34 Vogelstein and Kinzler (2004) also note that pathways that are common to "cancer" are
35 also operative in liver cancer where, as a heterogeneous disease, multiple pathways have been
36 implicated in differing manifestations of this disease. Thus, liver cancer may be an example in
37 its multiple forms that are analogous to differing sites being affected by common pathways
38 leading to "cancer." Pathway concordance may not always show up as site concordance as
39 expression of cancer between species. Liver cancer may be the example where many pathways
40 can lead a cancer that is characterized by its heterogeneity.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-287 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.3.1.3. Etiology, Incidence and Risk Factors for Hepatocellular Carcinoma (HCC)
2 The review article of Farazi and DePinho (2006) provides and excellent summary of the
3 current state of human liver cancer in terms of etiology and incidence. The 5-year survival rate
4 of individuals with liver cancer in the United States is only 8.9% despite aggressive conventional
5 therapy with lethality of liver cancer due in part from its resistance to existing anticancer agents,
6 a lack of biomarkers that can detect surgically respectable incipient disease, and underlying liver
7 disease that limits the use of chemotherapeutic drugs. Chen et al. (2002) report that surgical
8 resection is considered the only "curative treatment" but >80 of patients have widespread HCC at
9 the time of diagnosis and are not candidates for surgical treatment. Among patients with
10 localized HCC who undergo surgery, 50% suffer a recurrence. Primary liver cancer is the fifth
11 most common cancer worldwide and the third most common cause of cancer mortality. HCC
12 accounts for between 85 and 90% of primary liver cancers (El-Serag and Rudolph, 2007). Seitz
13 and Stickel (2006) report that epidemiological data from the year 2000 indicate that more than
14 560,000 new cases of HCC occurred worldwide, accounting for 5.6% of all human cancers and
15 that HCC is the fifth most common malignancy in men and the eighth in women. Overall,
16 incidence rates of HCC are higher in males compared to females. In almost all populations,
17 males have higher liver cancer rates than females, with male:female ratios usually averaging
18 between 2:1 and 4:1 and the largest discrepancies in rates (>4:1) found in medium-risk European
19 populations (El-Serag and Rudolph, 2007). Experiments show a 2- to 8-fold of control HCC
20 development in male mice as well supporting the hypothesis that androgens influence HCC
21 progression rather than sex-specific exposure to risk factors (El-Serag and Rudolph, 2007).
22 El-Serag and Rudolph (2007) also report that
23
24 in almost all areas, female rates peak in the age group 5 years older than the peak
25 age group for males. In low risk population (e.g., U.S.) the highest age-specific
26 rates occur among persons aged 75 and older. A similar pattern is seen among
27 most high-risk Asian populations. In contrast male rats in high-risk African
28 populations (e.g., Gambia) ten to peak between ages 60 and 65 before declining,
29 whereas female rates peak between 65 and 70 before declining.
30
31 Age adjusted incidence rates for HCC are extremely high in East and Southeast Asia and
32 in Africa but in Europe, there is a gradually decreasing prevalence from South to North. HCC
33 incidence rates also vary greatly among different populations living in the same region and vary
34 by race (e.g., for all ages and sexes in the United States, HCC rates are 2 times higher in Asian
35 than in African Americans, whose rates are 2 times higher than those in whites) ethnic variability
36 likely to include differences in the prevalence and acquisition time of major risk factors for liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-288 DRAFT—DO NOT CITE OR QUOTE
-------
1 disease and HCC (El-Serag and Rudolph, 2007). Worldwide HCC incidence rate doubled during
2 the last two decades and younger age groups are increasingly affected (El-Serag, 2004). The
3 high prevalence of HCC in Asia and Africa may be associated with widespread infection with
4 hepatitis B virus (HBV) and HCV but other risk factors include chronic alcohol misuse, non
5 alcoholic fatty liver disease (NAFLD), tobacco, oral contraceptives, and food contamination with
6 aflatoxins (Seitz and Stickel, 2006). El-Serag and Rudolph (2007) report HCC to be the fastest
7 growing cause of cancer-related death in men in the United States with age-adjusted HCC
8 incidence rates increasing more than 2-fold between 1985 and 2002 and that, overall, 15-50% of
9 HCC patients in the United States have no established risk factors.
10 Although liver cirrhosis is present in a large portion of patients with HCC, it is not always
11 present. Fattovich etal. (2004) report that
12
13 differences of geographic area, method of recruitment of the HCC cases (medical
14 or surgical) and the type of material studied (liver biopsy specimens, autopsy, or
15 partial hepatectomies) may account for the variable prevalence of HCC without
16 underlying cirrhosis (7% to 54%) quoted in a series of studies. Percutaneous liver
17 biopsy specimens are subject to sampling error. However, only a small
18 proportion of patients with HCC without cirrhosis have absolutely normal liver
19 histology, the majority of them showing a range of fibrosis intensity from no
20 fibrosis are all to septal and bridging fibrosis, necroinflammation, steatosis, and
21 liver cell dysplasia.
22
23 Farazi and DePinho (2006) note that for diabetes, a higher indices of HCC has been
24 described in diabetic patients with no previous history of liver disease associated with other
25 factors. El-Serag and Rudolph (2007) report that in their study of VA patients (173,643 patients
26 with and 650,620 patients without diabetes), that HCC incidence doubled among patients with
27 diabetes and was higher among those with a longer follow-up of evaluation. "Although most
28 studies have been conducted in low HCC rate areas, diabetes also has been found to be a
29 significant risk factor in areas of high HCC incidence such as Japan. Taken together, available
30 data suggest that diabetes is a moderately strong risk factor for HCC."
31 NAFLD and nonalcoholic steatohepatitis contribute to the development of fibrosis and
32 cirrhosis and therefore, might also contribute to HCC development. The pathogenesis of
33 NAFLD includes the accumulation of fat in the liver which can lead to reactive oxygen species
34 in the liver with necrosis factor a (TNFa) elevated in NAFDL and alcoholic liver disease (Seitz
35 and Stickel, 2006). Abnormal liver enzymes not due to alcohol, viral hepatitis, or iron overload
36 are present in 2.8 to 5.5% of the United States general population and may be due to NAFLD in
37 66 to 90% of cases (Adams and Lindor, 2007). Primary NAFLD occurs most commonly and is
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-289 DRAFT—DO NOT CITE OR QUOTE
-------
1 associated with insulin-resistant states, such as diabetes and obesity with other conditions
2 associated with insulin resistance, such as polycystic ovarian syndrome and hypopituitarism also
3 associated with NAFLD (Adams and Lindor, 2007). The steatotic liver appears to be susceptible
4 to further hepatotoxic insults, which may lead to hepatocyte injury, inflammation, and fibrosis,
5 but the mechanisms promoting progressive liver injury are not well defined (Adams and Lindor,
6 2007). Substrates derived from adipose tissue such as FFA, TNF-a, leptin, and adiponectin have
7 been implicated with oxidative stress appearing to be important leading to subsequent lipid
8 peroxidation, cytokine induction, and mitochondrial dysfunction. Liver disease was the third
9 leading cause of death among NAFLD patients compared to the 13th leading cause among the
10 general population, suggesting that liver-related mortality is responsible for a proportion of
11 increased mortality risk among NAFLD patients (Adams and Lindor, 2007).
12 The relative risk for HCC in type 2 diabetics has been reported to be approximately 4 and
13 increases to almost 10 for consumption of more than 80 g of alcohol per day (Hassan et al.,
14 2002). El-Serag and Rudolph (2007) report that
15
16 it has been suggested that many cryptogenic cirrhosis and HCC cases represent
17 more severe forms of nonalcoholic fatty liver disease (NAFLD), namely
18 nonalcoholic steato hepatitis (NASH). Studies in the United States evaluating risk
19 factors for chronic liver disease or HCC have failed to identify HCV, HBV, or
20 heavy alcohol intake in a large proportion of patients (30-40%). Once cirrhosis
21 and HCC are established, it is difficult to identify pathologic features of NASH.
22 Several clinic-based controlled studies have indicated that HCC patients with
23 cryptogenic cirrhosis tend to have clinical and demographic features suggestive of
24 NASH (predominance of women, diabetes, and obesity) as compared with age-
25 and sex-matched HCC patients of well defined vial or alcoholic etiology. The
26 most compelling evidence for an association between NASH and HCC is indirect
27 and come from studies examining HCC risk with 2 conditions strongly associated
28 with NASH: obesity and diabetes. In a large prospective cohort in the US,
29 followed up for 16 years, liver cancer mortality rates were 5 times greater among
30 men with the greatest baseline body mass index (range 35-40) compared with
31 those with a normal body mass index. In the same study, the risk of liver cancer
32 was not as increase in women, with a relative risk of 1.68. Two other population-
33 based cohort studies from Sweden and Denmark found excess HCC risk
34 (increased 2- to 3-fold) in obese men and women compared with those with a
35 normal body mass index.. .Finally, liver disease occurs more frequently in those
36 with more severe metabolic disturbances, with insulin resistance itself shown to
37 increase as the disease progresses. Several developed countries most notably the
38 United States, are in the midst of a burgeoning obesity epidemic. Although the
39 evidence linking obesity to HCC is relatively scant, even small increase in risk
40 related to obesity could translate into a large number of HCC cases.
41
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-290 DRAFT—DO NOT CITE OR QUOTE
-------
1 Thus, even a small increase in risk related to obesity could result in a large number of HCC cases
2 and the latency of HCC may make detection of increased HCC risk not detectable for several
3 years.
4 Other factors are involved as not every cirrhotic liver progresses to HCC. Seitz and
5 Stickel (2006) suggest that 90 to 100% of those who drink heavily suffer from alcoholic fatty
6 liver, 10-35% of those evolve to alcoholic steatohepatitis, 8-20% of those evolve to alcoholic
7 cirrhosis, and 1-2% of those develop HCC. HCV infects approximately 170 million individuals
8 worldwide with approximately 20% of chronic HCV cases developing liver cirrhosis and 2.5%
9 developing HCC. Infection with HBV, a noncytopathic, partially double stranded hepatotropic
10 DNA virus classified as a member of the hepadnaviridae family, is also associated with liver
11 cancer risk with several lines of evidence supporting the direct involvement of HBV in the
12 transformation process (Farazi and DePinho, 2006). El-Serag and Rudolph (2007) suggest that
13
14 Epidemiologic research has shown that the great majority of adult-onset HCC
15 cases are sporadic and that many have at lease 1 established non-genetic risk
16 factor such as alcohol abuse or chronic HCV or HBV infection. However, most
17 people with these known environmental risk factors never develop cirrhosis or
18 HCC, whereas a sizable minority of HCC case develop among individuals without
19 any known risk factors... Genetic epidemiology studies in HCC, similar to several
20 other conditions, have fallen short of early expectations that they rapidly and
21 unequivocally would result in identification of genetic variants conveying
22 substantial excess risk of disease and thereby establish the groundwork for
23 effective genetic screening for primary prevention.
24
25 E.3.1.4. Issues Associated with Target Cell Identification
26 Another outstanding and important question in HCC pathogenesis involves the cellular
27 origin of this cancer. The liver is made up of a number of cell types showing different
28 phenotypes and levels of differentiation. Which cell types are targets of hepatocarcinogens and
29 are those responsible for human HCC is a matter of intense debate. Studies over the last decade
30 provide evidence of several types of cells in the liver that can repopulate the hepatocyte
31 compartment after a toxic insult. "Indeed, although the existence of a liver stem cell is often
32 debated, most experts agree that progenitor liver cells are activated, in response to significant
33 exposure to hepatotoxins. Also, progenitor cells derived from nonhepatic sources, such as bone
34 marrow and pancreas, have been demonstrated recently to be capable of differentiating into
35 mature hepatocytes under correct microenvironmental conditions" (Gandillet et al., 2003). At
36 present, analyses of human HCCs for oval cell markers, comparison of their gene-expression
37 patterns with rat fetal hepatoblasts and the cellular characteristics of HCC from various animal
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-291 DRAFT—DO NOT CITE OR QUOTE
-------
1 models have provided contrasting results about the cellular origin of HCC and imply dual origins
2 from either oval cells or mature hepatocytes. The failure to identify a clear cell of origin for
3 HCC might stem from the fact that there are multiple cells of origin, perhaps reflecting the
4 developmental plasticity of the hepatocyte lineage. The resolution of the HCC cell of origin
5 issue could affect the development of useful preventative strategies to target nascent neoplasms,
6 foster an understanding of how HCC-relevant genetic lesions function in that specific cell-
7 development context and increase our ability to develop more accurate mouse models in which
8 key genetic events are targeted to the appropriate cellular compartment (Farazi and DePinho,
9 2006). Two reviews by Librecht (2006) and Wu and Chen (2006) provide excellent summaries
10 of the issues involved in identifying the target cell for HCC and the review by Roskams et al.
11 (2004) provides a current view of the "oval cell" its location and human equivalent. Recent
12 reports by Best and Coleman (2007) suggest another type of liver cell is also capable of
13 proliferation and differentiating into small hepatocytes (i.e., small hepatocyte-like progenitor
14 cell).
15 The review by Librecht (2006) provides an excellent description of the controversy and
16 data supporting different views of the cells of origin for HCC.
17
18 In recent years, the results of several studies suggest that human liver tumors can
19 be derived from hepatic progenitor cells rather than from mature cell types. The
20 available data indeed strongly suggest that most combined hepatocellular-
21 cholangiocarcinomas arise from hepatic progenitor cells (HPCs) that retained
22 their potential to differentiate into the hepatocyte and biliary lineages. Hepatic
23 progenitor cells could also be the basis for some hepatocellular carcinomas and
24 hepatocellular adenomas, although it is very difficult to determine the origin of an
25 individual hepatocellular carcinoma. There is currently not enough data to make
26 statements regarding a hepatic progenitor cell origin of cholangiocarcinoma. The
27 presence of hepatic progenitor cell markers and the presence and extent of the
28 cholangiocellular component are factors that are related the prognosis of
29 hepatocellular carcinomas and combined hepatocellular-cholangiocarcinomas,
30 respectively... The traditional view that adult human liver tumors arise from
31 mature cell types has been challenged in recent decades.. .HPCs are small
32 epithelial cells with an oval nucleus, scant cytoplasm and location in the bile
33 ductules and canals of Hering. HPCs can differentiate towards the biliary and
34 hepatocytic lineages. Differentiation towards the biliary lineage occurs via
35 formation of reactive bile ductules, which are anastamosing ductules lined by
36 immature biliary cells with a relatively large and oval nucleus surrounded by a
37 small rim of cytoplasm. Hepatocyte differentiation leads to the formation of
38 intermediate hepatocyte-like cells, which are defined as polygonal cells with a
39 size intermediate between than of HPCs and hepatocytes. In most liver diseases,
40 hepatic progenitor cells are "activated" which means that they proliferate and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-292 DRAFT—DO NOT CITE OR QUOTE
-------
1 differentiate towards the hepatocytic and/or biliary lineages. The extent of
2 activation is correlated with disease severity.. .HPCs and their immediate biliary
3 and hepatocytic progeny not only have a distinct morphology, but they also
4 express several markers, with many also present in bile duct epithelial cells.
5 Immunohistochemistry using antibodies against these markers facilitates the
6 detection of HPCs. The most commonly used markers are cytokeratin (CK) 19
7 and CK7...The proposal that a human hepatocellular carcinoma does not
8 necessarily arise from mature hepatocyte, but could have HPC origin, has
9 classically been based on three different observations. Each of them, however,
10 gives only indirect evidence that can be disputed.. .Firstly, it has been shown that
11 HPCs are the cells of origin of HCC in some animal models of
12 hepatocarcinogenesis, which has led to the suggestion that this might also be the
13 case in humans. However, in other animal models, the HCCs arise from mature
14 hepatocytes and not from HPCs or reactive bile ductular cells (Bralet et al 2002;
15 Lin et al 1995- DEN treated rats). Since it is currently insufficiently clear which
16 of these animal models accurately mimics human hepatocarcinogenesis, one
17 should be careful about extrapolating data regarding HPC origin of HCC in
18 animal models to the human situation... Secondly, liver diseases that are
19 characterized by the presence of carcinogens and development of dysplastic
20 lesions also show HPC activation. Therefore, the suggestion has been made that
21 HPCs form a "target population" for carcinogens, but this is only a theoretical
22 possibility not supported by experimental data.. .Thirdly, several studies have
23 shown that a considerable proportion of HCCs express one or more HPC markers
24 that are not present in normal mature hepatocytes. Due to the fact that most HPC
25 markers are also expressed in the biliary lineage, the term "biliary marker" has
26 been used in some of these studies. The "maturation arrest" hypothesis states that
27 genetic alterations occurring in a HPC, or its immediate progeny, cause aberrant
28 proliferation and prevent its normal differentiation. Further accumulation of
29 genetic alterations eventually leads to malignant transformation of these
30 incompletely differentiated cells. The resulting HCC expresses HPC markers as
31 evidence of its origin. However, expression of HPC markers can also be
32 interpreted in the setting of the "dedifferentiation" hypothesis, which suggests that
33 the expression of HPC markers is acquired during tumor progression as a
34 consequence of accumulating mutations. For example, experiments in which
35 human HCC cells lines were transplanted into nude mice have nicely shown that
36 the expression of HPC marker, CK19, steadily increased when the tumors became
37 increasingly aggressive and metastasized to the lung, Thus, the expression of
38 CK19 in a HCC does not necessarily mean that the tumor has a HPC origin, but it
39 can also be mutation-induced, acquired expression associated with tumor
40 progression. Both possibilities are not mutually exclusive. For an individual
41 HCC that expresses a HPC marker, it remains impossible to determine whether
42 this marker reflects the cellular origin and/or is caused by tumor progression.
43 This can only be elucidated by determining whether HCC contains cells that are
44 ultrastructurally identical to HPCs in nontumor liver.
45
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-293 DRAFT—DO NOT CITE OR QUOTE
-------
1 Similarly, the review by Wu and Chen (2006) also presents a valuable analysis of these
2 issues and state:
3
4 The question of whether hepatocellular carcinomas arises from the differentiation
5 block of stem cells or dedifferentiation of mature cells remains controversial.
6 Cellular events during hepatocarcinogenesis illustrate that HCC may arise for
7 cells at various stages of differentiation in the hepatic stem cell lineage... The role
8 of cancer stem cells has been demonstrated for some cancers, such as cancer of
9 the hematopoietic system, breast and brain. The clear similarities between normal
10 stem cell and cancer stem cell genetic programs are the basis of the a proposal
11 that some cancer stem cells could derived form human adult stem cells. Adult
12 mesenchymal stem cells (MSC) may be targets for malignant transformation and
13 undergo spontaneous transformation following long-term in vitro culture,
14 supporting the hypothesis of cancer stem cell origin. Stem cells are not only units
15 of biological organization, responsible for the development and the regeneration
16 of tissue and organ systems, but are also targets of carcinogenesis. However, the
17 origin of the cancer stem cell remains elusive... Three levels of cells that can
18 respond to liver tissue renewal or damage have been proved (1) mature liver cells,
19 as "unipotential stem cells," which proliferate under normal liver tissue renewal
20 and respond rapidly to liver injury, (2) oval cells, as bipotential stem cells, which
21 are activated to proliferate when the liver damage is extensive and chronic or if
22 proliferation of hepatocytes is inhibited; and (3) bone marrow stem cells, as
23 multipotent liver stem cells, which have a very long proliferation potential. There
24 are two major nonexclusive hypotheses of the cellular origin of cancer; from stem
25 cells due to maturation arrest or from dedifferentiation of mature cells. Research
26 on hepatic stem cells in hepatocarcinogenesis has entered a new era of
27 controversy, excitement and great expectations...The two major hypotheses about
28 the cellular origination of HCC have been discussed for almost 20 years. Debate
29 has centered on whether or not HCC originates from the differentiation block of
30 stem cells or dedifferentiation of mature cells. Recent research suggests that HCC
31 may originate from the transdifferentiation of bone marrow cells. In fact, there
32 might be more than one type of carcinogen target cell. The argument about the
33 origination of HCC becomes much clearer when viewed from this viewpoint:
34 poorly differentiated HCC originate from bone marrow stem cells and oval cells,
35 while well-differentiated HCC originates form mature hepatocytes... The cellular
36 events during hepatocarcinogenesis illustrate that HCC may arise from cells at
37 various stages of differentiation in the hepatocyte lineage. There are four levels
38 of cells in the hepatic stem cell lineage: bone marrow cell, hepato-pancreas stem
39 cell, oval cell and hepatocyte. HSC and the liver are known to have a close
40 relationship in early development. Bone marrow stem cells could differentiate
41 into oval cells, which could differentiate into heptatocytes and duct cells. The
42 development of pancreatic and liver buds in embryogenesis suggests the existence
43 of a common progenitor cells to both the pancreas and liver. All of the four levels
44 of cells in the stem cell lineage may be targets of hepatocarcinogenesis.
45
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-294 DRAFT—DO NOT CITE OR QUOTE
-------
1 Along with the cell types described as possible targets and participants in HCC, Best and
2 Coleman (2007) describe yet another type of cell in the liver that can respond to hepatocellular
3 injury, which they term small hepatocyte-like progenitor cells and conclude that they are not the
4 progeny of oval cells, but represent a distinct liver progenitor cell population. Another potential
5 regenerative cell is the small hepatocyte-like progenitor cell (SHPC). SHPCs share some
6 phenotypes with hepatocytes, fetal hepatoblasts, and oval cells, but are phenotypically distinct.
7 They express markers such as albumin, transferring, and alpha-fetoprotein (AFP) and possess
8 bile canaliculi and store glycogen.
9 A recent review by Roskams et al. (2004) provides a current view of the "oval cell" its
10 location and human equivalent. They conclude that
11
12 while similarities exist between the progenitor cell compartment of human and
13 rodent livers, the different rodent models are not entirely comparable with the
14 human situation, and use of the same term has created confusion as to what
15 characteristics may be expected in the human ductular reaction. For example, a
16 defining feature of oval cells in many rodent models of injury is production of
17 alpha-fetoprotein, whereas ductular reactions in humans rarely display such
18 expression. Therefore we suggest that the "oval cell" and "oval -like cell" no
19 longer be used in description of human liver.
20
21 In the chronic hepatitis and cancer model of Vig et al. (2006) it is not the oval cells or
22 SHPCs that are proliferating but the mature hepatocytes, thus, supporting theories that it is not
23 only oval cells that are causing proliferations leading to cancer. Vig et al. (2006) also report that
24 studies in mice an humans indicate that oval cells also may give rise to liver tumors and that oval
25 cells commonly surround and penetrate human liver tumors, including those caused by hepatitis
26 B. Tarsetti et al. (1993) suggest that although some studies have suggested that oval cells are
27 directly involved in the formation of HCC others assert that HCC originates from preneoplastic
28 foci and nodules derived from hepatocytes and report that HCC evolved in their model of liver
29 damage from hepatocytes, presumably hepatocellular nodules, and not from oval cells. They
30 also suggest that proliferation alone may not lead to cancer. Recent studies that follow the
31 progression of hepatocellular nodules to HCC in humans (see Section E.3.2.4, below) suggest an
32 evolution from nodule to tumor.
33
34 E.3.1.5. Status of Mechanism of Action for Human Hepatocellular Carcinoma (HCC)
35 The underlying molecular mechanisms leading to hepatocarcinogenesis remain largely
36 unclear (Yeh et al., 2007). Although HCC is multistep, and its appearance in children suggest a
37 genetic predisposition exists, the inability to identify most of the predisposing genes and how
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-295 DRAFT—DO NOT CITE OR QUOTE
-------
1 their altered expression relates to histological lesions that are the direct precursors to HCC, has
2 made it difficult to identify the rate limiting steps in hepatocarcinogenesis (Feitelson et al.,
3 2002). Calvisi et al. (2007) report that although the major etiological agents have been
4 identified, the molecular pathogenesis of HCC remains unclear and that while deregulation of a
5 number of oncogenes (e.g., c-Myc, cyclin Dl and fi-catenin and tumor suppressor genes
6 including PJ6INK4A, P53, E-cadherin, DLC-1, andpRb) have been observed at different
7 frequencies in HCC, the specific genes and the molecular pathways that play pivotal roles in
8 liver tumor development have not been identified. Indeed rather than simple patterns of
9 mutations, pathways that are common to cancer have been identified through study of tumors
10 and through transgenic mouse models. Branda and Wands (2006) state that the molecular factors
11 and interactions involved in hepatocarcinogenesis are still poorly understood but are particularly
12 true with respect to genomic mutations, "as it has been difficult to identify common genetic
13 changes in more than 20% to 30% of tumors." As well as phenotypically heterogeneous, "it is
14 becoming clear that HCCs are genetically heterogeneous tumors." The descriptions of
15 heterogeneity of tumors and of pathway disruptions common to cancer are also shown for liver
16 tumors (see Sections E.3.1.6 and E.3.1.8, below). However, many of these studies focus on the
17 end process and of examination of the genomic phenotype of the tumor for inferences regarding
18 clinical course, aggressiveness of tumor, and consistency with other forms of cancer. As stated
19 above, the events that produce these tumors from patients with conditions that put them at risk,
20 are not known.
21 El-Serag and Rudolph (2007) suggest that risk of HCC increases at the cirrhosis stage
22 when liver cell proliferation is decreased and that acceleration of carcinogenesis at this stage may
23 result from telomere shortening (resulting in limitations of regenerative reserve and induction of
24 chromosomal instability), impaired hepatocyte proliferation (resulting in cancer induction by loss
25 of replicative competition), and altered milieu conditions that promote tumor cell proliferation.
26
27 When telomeres reach a critically short length, chromosome uncapping induces
28 DNA damage signals, cell-cycle arrest, senescence, or apoptosis. Telomeres are
29 critically short in human HCC and on the single cell level telomere shortening
30 correlated with increasing aneuploidy in human HCC...Chemicals inhibiting
31 hepatocyte proliferation accelerate carcinogen-induced liver tumor formation in
32 rats as well as the expansion and transformation of transplanted hepatocytes. It is
33 conceivable that abnormally proliferating hepatocytes would not expand in
34 healthy regenerating liver but would expand quickly and eventually transform in
35 the growth restrained cirrhotic liver... .Liver mass is controlled by growth factors
36 - mass loss through could provide a growth stimulatory macroenvironment. For
37 the microenvironment, cirrhosis activates stellate cells resulting in increased
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-296 DRAFT—DO NOT CITE OR QUOTE
-------
1 production of extracellular matrix proteins, cytokines, growth factors, and
2 products of oxidative stress.
3
4 Like other cancers, genomic instability is a common feature of human HCC with various
5 mechanisms thought to contribute, including telomere erosion, chromosome segregation defects,
6 and alteration in DNA damage-response pathways. In addition to genetic events associated with
7 the development of HCC (p53 inactivation, mutation in p-catenin, overexpression of ErbB
8 receptor family members, and overexpression of the MET receptor whose ligand is HGF) various
9 cancer-relevant genes seem to be targeted on the epigenetic level (methylation) in human HCC
10 (Farazi and DePinho, 2006). Changes in methylation have been detected in the earliest stages of
11 hepatocarcinogenesis and to a greater extent in tumor progression (Lee et al., 2003). Seitz and
12 Stickel (2006) report that aberrant DNA hypermethylation (a silencing effect on genes) may be
13 associated with genetic instability as determined by the loss of heterozygosity and microsatellite
14 instability in human HCC due to chronic viral hepatitis and that modifications of the degree of
15 hepatic DNA methylation have also been observed in experimental models of chronic
16 alcoholism. Farazi and DePinho (2006) report that two of the key molecules that involved in
17 DNA damage response, p53 and BRCA2, seem to have roles in destabilizing the HCC genome
18 (Collin, 2005). The inactivation of p53 through mutation or viral oncoprotein sequestration is a
19 common event in HCC and p53 knock in mouse models containing dominant point mutations
20 have been shown to cause genomic instability. However, Farazi and DePinho (2006) note that
21 despite documentation of deletions or mutations in these and other DNA damage network genes,
22 their direct roles in the genomic instability of HCC have yet to be established in many genetic
23 model systems.
24 Telomere shortening has been described as a key feature of chronic hyperproliferative
25 liver disease (Urabe et al., 1996; Miura et al., 1997; Rudolf and DePinho, 2001: Kitada et al.,
26 1995), specifically occurring in the hepatocyte compartment. These observations have fueled
27 speculation that telomere shortening associated with chronic liver disease and hepatocyte
28 turnover contribute to the induction of genomic instability that drives human HCC (Farazi and
29 DePinho, 2006). Defects in chromosome segregation during mitosis result in aneuploidy, a
30 common cytogenetic feature of cancer cell including HCC (Farazi and DePinho, 2006).
31 Several studies have attempted to categorize genomic changes in relation to tumor state.
32 In general, high levels of chromosomal instability seem to correlate with the de-differentiation
33 and progression of HCC (Wilkens et al., 2004). Several studies have suggested certain
34 chromosomal changes to be specific to dysplastic lesions, early -stage and late-stage HCCs, and
35 metastases. It is important to note that the studies that have attempted to compare genomic
36 profiles and tumor state are few in number, often did not classify HCCs on the basis of etiology,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-297 DRAFT—DO NOT CITE OR QUOTE
-------
1 and used relatively low-resolution genome-scanning platforms (Farazi and DePinho, 2006).
2 Farazi and DePinho (2006) note that it should be emphasized that although genome-etiology
3 correlates reported in some studies, are intriguing, several studies have failed to uncover
4 significant differences in genomic changes between different etiological groups, although the
5 outcome might related to small sample sizes and the low-resolution genome-scanning platform
6 used.
7
8 E.3.1.6. Pathway and Genetic Disruption Associated with Hepatocellular Carcinoma (HCC)
9 and Relationship to Other Forms ofNeoplasia
10 In their landmark paper, Hanahan and Weinberg (2000) suggested that the vast catalog of
11 cancer cell genotypes were a manifestation of six essential alterations in cell physiology that
12 collectively dictate malignant growth; self-sufficiency in growth signals, insensitivity to growth
13 -inhibitory (antigrowth signals), elevation of programmed cell death (apoptosis), limitless
14 replication potential, sustained angiogenesis, and tissue invasion and metastasis. They proposed
15 that these six capabilities are shared in common by most and perhaps all types of human tumors
16 and, while virtually all cancers must acquire the same six hallmark capabilities, their means of
17 doing so would vary significantly, both mechanistically and chronologically. It was predicted
18 that in some tumors, a particular genetic lesions may confer several capabilities simultaneously,
19 decreasing the number of distinct mutational steps required to complete tumorigenesis. Loss of
20 the p53 tumor suppressor was cited as an example that could facilitate both angiogenesis and
21 resistance to apoptosis and to enable the characteristic of genomic instability. The paths that
22 cells could take on their way to becoming malignant were predicted to be highly variable, and
23 within a give cancer type, mutation of a particular target genes such as ras orp53 could be found
24 only in a subset of otherwise histologically identical tumors. Furthermore, mutations in certain
25 oncogenes and tumor suppressor genes could occur early in some tumor progression pathways
26 and late in others. Genes known to be functionally altered in "cancer" were identified as
27 including Fas,Bcl2, Decoy R, Bax, Smads, TFGpR, pi5, pi6, Cycl D, Rb, human papilloma
28 virus E7, ARF, PTEN, Myc, Fos, Jun, Ras, Abl, NF1, RTK, transforming growth factor alpha
29 (TGF-a), Integrins, E-cadherin, Src, p-catenin, APC, and WNT.
30 Branda and Wands (2006) report that two signal transduction cascades that appear to be
31 very important are insulin/IFG-1/IRS-l/MAPK and Wnt/Frizzled/p-catenin pathways which are
32 activated in over 90% of HCC tumors (Branda and Wands, 2006). Feitelson et al. (2002)
33 reported that
34
35 In addition to NF-KB, up-regulated expression of rhoB has been reported in some
36 HCCs. RhoB is in the ras gene family, is associated with cell transformation, and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-298 DRAFT—DO NOT CITE OR QUOTE
-------
1 may be a common denominator to both viral and non-viral hepatocarcinogenesis.
2 Activation of ras and NF-KB, combined with down regulation of multiple negative
3 growth regulatory pathways, then, may contribute importantly to early steps in
4 hepatocarcinogenesis. Thus viral proteins may alter the patterns of hepatocellular
5 gene expression by transcript!onal trans-regulation... Another early event appears
6 to involve the mutation of p-catenin, which is a component of the Wnt signal
7 transduction pathway whose target genes include c-myc, c-jun, cyclin Dl,
8 fibronectin, the connective tissue growth factor WISP, and matrix
9 metaolloproteinases.
10 Boyault et al. (2007) report that
11
12 altogether, the principle carcinogenic pathways known to be deregulated in HCC
13 are inactivation of TP53, Wnt/wingless activation mainly through CTNNB1
14 mutations activating P-catenin- and AXIN1-inactivating mutations,
15 retinoblastoma inactivation through RB 1 and CDKN2A promoter methylation and
16 rare gene mutations, insulin growth factor activation through IGF2
17 overexpression, and IGF2R-inactiving mutations.
18
19 El-Serag and Rudolph suggest that "in general, the activation of oncogenic pathways in
20 human HCC appears to be more heterogeneous compared with other cancer types." El-Serag
21 and Rudolph (2007) report that the p53 pathway is a major tumor-suppressor pathway that
22 (1) limits cell survival and proliferation (replicative senescence) in response to telomere
23 shortening (2) induces cell-cycle arrest in response to oncogene activation (oncogene-induced
24 senescence), (3) protects genome integrity, and (4) is affected at multiple levels in human HCC.
25 "p53 mutations occur in aflatoxin induced HCC (>50%) and with lower frequency (20-40%) in
26 HCC not associated with aflatoxin." In addition,
27
28 the vast majority of human HCC overexpresses gankyrin, which inhibits both Rb
29 checkpoint and p53 checkpoint function...The p!6/Rb checkpoint is another
30 major pathway limiting cell proliferation in response to telomere shortening,
31 DNA damage, and oncogene activation. In human HCC the Rb pathway is
32 disrupted in more than 80% of cases, with repression of p 16 by promoter
33 methylation being the most frequent alteration. Moreover, expression of gankyrin
34 (an inhibitor of p53 and Rb checkpoint function) is increased in the vast majority
35 of human HCCs, indicating that the Rb checkpoint is dysfunctional in the vast
36 majority of human HCCs.. .The frequent inactivation of p53 in human HCC
37 indicates that abrogation of p53-dependent apoptosis could promote
38 hepatocarcinogenesis. The role of impairment of p53-independent apoptosis for
39 hepatocarcinogenesis remains to be defined.. .Activation of the P-catenin pathway
40 frequently occurs in mouse and human HCC involving somatic mutations, as well
41 as transcriptional repression of negative regulators. An activation of the Akt
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-299 DRAFT—DO NOT CITE OR QUOTE
-------
1 signaling and impaired expression of phosphatase and tensin homolog (PTEN) (a
2 negative regulator of Akt) have been reported in 40-60% of Human HCC.
3
4 They suggest that although Myc is a potent oncongene inducing hepatocarcinogenesis in mouse
5 models the data on human HCC are heterogeneous and further studies are required.
6
7 E.3.1.7. Epigenetic Alterations in Hepatocellular Carcinoma (HCC)
8 The molecular pathogenesis of HCC remains largely unknown but it is presumed that the
9 development and progression of HCC are the consequence of cumulative genetic and epigenetic
10 events similar to those described in other solid tumors (Calvisi et al., 2006). Calvisi et al. (2007)
11 provide a good summary of DNA methylation status and cancer as well as its status in regard to
12 HCC:
13
14 Aberrant DNA methylation occurs commonly in human cancers in the forms of
15 genome-wide hypomethylation and regional hypermethylation. Global DNA
16 hypomethylation (also known as demethylation) is associated with activation of
17 protooncogenes, such as c-Jun, c-Myc, and c-HA-Ras, and generation of genomic
18 instability. Hypermethylation on CpG islands located in the promoter regions of
19 tumor suppressor genes results in transcriptional silencing and genomic
20 instability. CpG hypermethylation (also known as de novo methylation) acts as
21 an alternative and/or complementary mechanisms to gene mutations causing gene
22 inactivation, and it is now recognized as an important mechanism in
23 carcinogenesis. Although the mechanism(s) responsible for de novo methylation
24 in cancer are poorly understood, it has been hypothesized that epigenetic silencing
25 depends on activation of a number of proteins known as DNA methyltransferases
26 (DNMTs) that posses de novo methylation activity. The importance of DNMTs
27 in CpG methylation was substantiated by the observation that genetic disruption
28 of both DNMT1 and DNMT3b genes in HCT116 cell lines nearly eliminated
29 methyltransferase activity. However, more recent findings indicate that the
30 HCT116 cells retain a truncated, biologically active form of DNMT1 and
31 maintain 80% of their genomic methylation. Further reduction of DNMT1 levels
32 by a siRNA approach resulted in decreased cell viability, increased apoptosis,
33 enhanced genomic instability, checkpoint defects, and abrogation of replicative
34 capacity. These data show that DNTM1 is required for cell survival and suggest
35 that DNTM1 has additional functions that are independent of its methyltransferase
36 activity. Concomitant overexpression of DNMT1, -3A, and -3b has been found in
37 various tumors including HCC. However, no changes in the expression of
38 DNMTs were found in other neoplasms, such as colorectal cancer, suggesting the
39 existence of alternative mechanisms. In HCC, a novel DNMT3b splice variant,
40 known as DNMT3b4 is overexpressed. DNMT3b4 lacks DNMT activity and
41 competes with DNMT2b3 for targeting of pericentromeric satellite regions in
42 HCC, resulting in DNA hypomethylation of these regions and induction of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-300 DRAFT—DO NOT CITE OR QUOTE
-------
1 chromosomal instability, further linking aberrant methylation and generation of
2 genomic alterations.
3
4 It is now well accepted that methylation changes occur early and ubiquitously in
5 cancer development. The case has been made that tumor cell heterogeneity is
6 due, in part, to epigenetic variation in progenitor cells and that epigenetic
7 plasticity together with genetic lesions drive tumor progression (Feinberg et al.,
8 2006).
9
10 A growing number of genes undergoing aberrant CpG island hypermethylation in
11 HCC have been discovered, suggesting that de novo methylation is an important
12 mechanism underlying malignant transformation in the liver. However, most of
13 the previous studies have focused on a single or a limited number of genes, and
14 few have attempted to analyze the methylation status of multiple genes in HCC
15 and associated chronic liver diseases. In addition, the functional consequence(s)
16 of global DNA hypomethylation and CpG island hypermethylation in human liver
17 cancer has not been investigated to date. Furthermore, to our knowledge no
18 comprehensive analysis of CpG island hypermethylation involving activation of
19 signaling pathways has been performed.
20
21 Calvisi et al. (2007) report that global gene expression profiles show human HCC to
22 harbor common molecular features that differ greatly from those of nontumorous surrounding
23 tissues, and that human HCC can be subdivided into 2 broad but distinct subclasses that are
24 associated with length of patient survival. They further suggest that aberrant methylation is a
25 major event in both early and late stages of liver malignant transformation and might constitute a
26 critical target for cancer risk assessment, treatment, and chemoprevention of HCC. Calvisi et al.
27 (2007) conducted analysis of methylation status of genes selected based on their capacity to
28 modulate signaling pathways (Ras, Jak/Stat, Wingless/Wnt, and RELN) and/or biologic features
29 of the tumors (proliferation, apoptosis, angiogenesis, invasion, DNA repair, immune response,
30 and detoxification). Normal livers were reported to show the absence of promoter methylation
31 for all genes examined. At least 1 of the genes involved in inhibition of Ras (ARH1, CLU,
32 DAB2, HDAB21P, HIN-1, HRASL, LOX, NORE1A, PAR4, RASSF1A, RASSF2, RASSF3,
33 RASSF4, RIG, RRP22, and SPRY2 and -¥), Jak/Stat (ARH1,CIS, SHP1, PIAS-1, PIAS-y, SOCS1,
34 -2, and -3, SYK, and GRIM-19}, and Wnt/p-catenin (APC, E-cadherin, y-catenin, SFRP1, -2, -4,
35 and -5, DKK-1 and -3, WIF-1 and HDPR1} pathways was affected by de novo methylation in all
36 HCC. A number of these genes were also reported to be highly methylated in the surrounding
37 nontumorous liver. In contrast, inactivation of at least 1 of these genes implicated in the RELN
38 pathway (DAB1, reelin) was detected differentially in HCC of subclasses of tumor that had
39 difference in tumor aggressiveness and progression. Epigenetic silencing of multiple tumor
40 suppressor genes maintains activation of the Ras pathway with a major finding in the Calvisi et
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-301 DRAFT—DO NOT CITE OR QUOTE
-------
1 al. (2007) study to be the concurrent hypermethylation of multiple inhibitors of the Ras pathway
2 with Ras was significantly more active in HCC than in surrounding or normal livers. Also
3 important, was the finding that no significant associations between methylation patterns and
4 specific etiologic agents (i.e., HVB, HVC, ethanol, etc.) were detected further substantiating the
5 conclusion that aberrant methylation is a ubiquitous phenomenon in hepatocarcinogenesis.
6
7 Current evidence suggests that hypomethylation might promote malignant
8 transformation via multiple mechanisms, including chromosome instability,
9 activation of protooncogenes, reactivation of transposable elements, and loss of
10 imprinting... The degree of DNA hypomethylation progressively increased from
11 nonneoplastic livers to fully malignant HCC, indicating that genomic
12 hypomethylation is an important prognostic factor in HCC, as reported for brain,
13 breast, and ovarian cancer.
14
15 Calvisi et al. (2007) also report that regional CpG hypermethylation was also enhanced during
16 the course of HCC disease and that the study of tumor suppressor gene promoters showed that
17 CpG methylation was frequently detected both in surrounding nontumorous livers and HCC.
18
19 E.3.1.8. Heterogeneity of Preneoplastic and Hepatocellular Carcinoma (HCC) Phenotypes
20 A very important issue for the treatment of HCC in humans is early detection. Research
21 has focused on identification of lesions that will progress to HCC and to also determine from the
22 phenotype of the nodule and genetic expression its cell source, likely survival, and associations
23 with etiologies and MO As. As with rodent models where preneoplastic foci have been observed
24 to be associated with progression to adenoma and carcinoma, nodules observed in humans with
25 high risk for HCC have been observed to progress to HCC. In humans, histomorphology of
26 HCC is notoriously heterogeneous (Yeh et al., 2007). Although much progress has been made,
27 there is currently not universally accepted staging system for HCC partly because of the natural
28 course of early HCC is unknown and the natural progression of intermediated and advanced
29 HCC are quite heterogeneous (Thorgeirsson, 2006). Nodules are heterogeneous as well with
30 differences in potential to progress to HCC. Chen et al. (2002) report that standard clinical
31 pathological classification of HCC has limited valued in predicting the outcome of treatment as
32 the phenotypic diversity of cancer is accompanied by a corresponding diversity in gene
33 expression patterns. There is also histopathological variability in the presentation of HCC in
34 geographically diverse regions of the world with some slow growing, differentiated HCC
35 nodules surrounded by a fibrous capsule are common among Japanese but, in contrast, a
36 "febrile" form of HCC, characterized by leukocytosis, fever, and necrosis within a poorly
37 differentiated tumor to be common in South African blacks (Feitelson et al., 2002).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-302 DRAFT—DO NOT CITE OR QUOTE
-------
1 A multistep process is suggested histologically, where HCC appears within the context of
2 chronic hepatitis and/or cirrhosis within regions of the liver cell dysplasia or adenomatous
3 hyperplasia (Feitelson et al., 2002). Kobayashi et al. (2006) report that the higher the grade of
4 the nodule the higher the percentage that will progress to HCC with 18.8% of all nodules and
5 regenerative lesions going on to become HCC, 53.3% remaining unchanged, and 27.9%
6 disappearing in the observation period of 0.1 to 8.9 years. Borzio et al. (2003) report that the rate
7 of liver malignant transformation was 40% in larger regenerative nodules, low-grade dysplastic,
8 and high-grade dysplastic nodules with higher grade of dysplasia extranodular detection of large
9 cell change and hyperchronic pattern associated with progression to HCC. Yeh et al. (2007)
10 report that nuclear staining for Ki-67 and Topo II-a (a nuclear protein targeted by several
11 chemotherapeutic agents) significantly increased in the progression from cirrhosis, through high
12 grade dysplastic nodules to HCC whereas the scores for TGF-a in these lesions showed an
13 inverse relationship. "In comparison with 18 HCC arising in noncirrhotic livers, the expression
14 of TGF-a is significantly stronger in cirrhotic liver than in noncirrhotic parenchyma and its
15 expression is also stronger in HCC arising in cirrhosis than in HCC arising in noncirrhotic
16 patients." They concluded that initiation in cirrhotic and noncirrhotic liver may have different
17 pathways with Transforming growth factor-a (a mitogen activated the EFGR) playing a relative
18 more important role in HCC from cirrhotic liver. Over expression of TGF-a in the liver of
19 transgenic mice induced increased proliferation, dysplasia, adenoma and carcinoma. Yeh et al.
20 (2007) concluded that such high-grade dysplastic nodules are precursor lesions in
21 hepatocarcinogenesis and that TGF-a may play an important role in the early events of liver
22 carcinogenesis.
23 Moinzadeh et al. (2005) reported in a meta-analysis of all available (n = 785) HCCs that
24 gains and losses of chromosomal material were most prevalent in a number of chromosomes and
25 that amplifications and deletions occurred on chromosomal arms in which oncogenes (e.g., MYC
26 and 8q24) and tumor suppressor genes (e.g., RBI on 13ql4) are located as well a modulators of
27 the WNT-signaling pathway. However, in multifocal HCC, nodules arising de novo within a
28 single liver have a different spectrum of genetic lesions. "Hence, there are likely to be many
29 paths to hepatocellular carcinoma, and this is why it has been difficult to assign specific
30 molecular alterations to changes in hepatocellular phenotype, clinical, or histopathological
31 changes that accompany tumor development" (Feitelson et al., 2002).
32 Serum AFP is commonly used as tumor marker for HCC. Several reports have linked
33 HCC to cytokines in an attempt to find more specific markers of HCC. Jia et al. (2007) report
34 that AFP marker allows for identification of a small set of HCC patients with smaller tumors,
35 and these patients have a relatively long-term survival rate following curative treatment.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-303 DRAFT—DO NOT CITE OR QUOTE
-------
1 Presently the only approach to screen for the presence of HCC in high-risk
2 populations is the combination of serum AFP and ultrasonagraphy. However,
3 elevated AFP is only observed in about 60 to 70% of HCC patients and to a lesser
4 extent (33-65%) in patients with smaller HCCs. Moreover, nonspecific elevation
5 of serum AFP has been found in 15% to 58% of patients with chronic hepatitis
6 and 11% to 47% of patients with liver cirrhosis.
7
8 Soresi et al. (2006) report that serum interleukin (IL)-6 levels are low in physiological
9 conditions, but increase considerably pathological conditions such as trauma, inflammation and
10 neoplasia. In tumors IL-6 may be involved in promoting the differentiation and growth of target
11 cells. "Many works have reported high serum IL-6 levels in various lifer diseases such as acute
12 hepatitis, primary biliary cirrhosis, chronic hepatitis (hepatitis C) and HCV-correlated liver
13 cirrhosis and in hepatocellular carcinoma." Soresi et al. (2006) report that patients with HCC
14 group had higher IL-6 values than those with cirrhosis and that "higher-staged" patients had the
15 highest IL-6 levels. Hsia et al (2007) also examined IL-6, IL-10 and hepatocyte growth factor
16 (HGF) as potential markers for HCC.
17
18 The expression of IL-6 or IL-10 or higher level of HGF or AFP was observed only
19 0-3% of normal subjects. Patients with HCC more frequently had higher IL-6 and
20 IL-10 levels, where as HGF levels in HCC patients were not significantly elevated
21 compared to patients with chronic hepatitis or non-HCC tumors (but greater than
22 controls). Among patients with low AFP level, IL-6 or IL-10 expression was
23 significantly associated with the existence of HCC. Patients with large HCC (>5
24 cm) more often had increased IL-6, IL-10 or AFP levels. Serum levels of IL-6
25 and IL-10 are frequently elevated in patients with HCC but not in benign liver
26 disease or non-HCC tumors.
27
28 Nuclear DNA content and ploidy have also been the subjects of several studies through
29 the years for identification of pathways for prediction of survival or origin of tumors. Nakajima
30 et al. (2004) report that p53 loss can contribute to the propagation of damaged DNA in daughter
31 cells through the inability to prevent the transmission of inaccurate genetic material, considered
32 to be one of the major mechanisms for the emergence of aneuploidy in tumors with inactivated
33 p53 protein and the increasing ploidy in HCC was associated with disturbance in p53. McEntee
34 et al. (1991) reported that specimens from 74 patients who underwent curative resection for
35 primary HCC and analyzed for DNA content, (i.e., tumors were classified as DNA aneuploid if a
36 separate peak was present from its standard large diploid peak [2C] and tetraploid peak [4C])
37 33% were DNA diploid, 30% were DNA tetraploid/polyploidy, and 37% were aneuploid of the
38 primary tumors examined. Nontumor controls were diploid and survival was not different
39 between patients with diploid versus nondiploid tumors. Zeppa et al. (1998) reported ploidy in
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-304 DRAFT—DO NOT CITE OR QUOTE
-------
1 84 hepatocellular carcinomas diagnosed by fine-needle aspiration biopsy to have 68 cases that
2 were aneuploid and 16 euploid (9 diploid and 7 polyploid), with median survival of 38 months
3 for patients with diploid HCC and 13 months for aneuploid HCC. Lin et al. (2003) report in their
4 study of fine needle aspiration of HCC that
5
6 the ratio of S and G2/M periods of DNA, which reflect cell hyperproliferation, in
7 the group with HCC tumors> 3cm in diameter were markedly higher than those of
8 the group with nodules< 3 cm in diameter and the group with hyperplastic
9 nodules... DNA analysis of aspiration biopsy tissues acquired from intrahepatic
10 benign hyperplastic nodules showed steady diploid (2c) peak that stayed in Gl
11 period. DNA analysis of aspiration biopsy tissues acquired from HCC nodules
12 showed S period of hyperproliferation and G2/M period. The DNA analysis of
13 HCC nodules showed aneuploid peak.
14
15 They concluded that in regard to the biological behavior of the cell itself, that the normal tissue,
16 reactive tissue and benign tumor all have normal diploid DNA but, like most other malignant
17 tumors, "HCC appears to have polyploid DNA, especially aneuploid DNA." Attallah et al.
18 (1999) report small needle liver biopsy data to show HCC to be 21.4% diploid, 50% aneuploid
19 and 28.6% tetraploid and that higher ploidies (aneuploid and tetraploid) were observed in human
20 liver cancer than residual tissues, although in some cases there was increased aneuploidy
21 (cirrhosis, 37%, hepatitis -50%). Of note for the study is the lack of appropriate control tissue
22 and uncertainty as to how some of their diploid cells could have been binucleate tetraploid cells.
23 Anti et al. (1994) reported reduction in binuclearity in the chronic hepatitis and cirrhosis groups
24 that was significantly correlated with a rise in the diploid/polyploidy ratio and that precancerous
25 and cancerous nodules within cirrhotic liver show an increased tendency toward diploidy or the
26 emergence of aneuploid populations. They note that a number of investigators have noted
27 significantly increased hepatocyte diploidization during the early stages of chemically induced
28 carcinogenesis in rat liver, but other experimental findings indicate that malignant transformation
29 can occur after any type of alteration in ploidy distribution. On the other hand, Melchiorri et al.
30 (1994) note that several studies using flow cytometric or image cytometric methods reported
31 high DNA ploidy values in 50-77% of the examined HCCs and that the presence of aneuploidy
32 was significantly related to a poor patient prognosis. They report that the DNA content of
33 mononucleated and binucleated hepatocytes, obtained by ultrasound-guided biopsies of
34 10 macroregenerative nodules without histologic signs of atypia from the lesions with the greater
35 fraction of mononucleated hepatocytes were diagnosed as HCCs during the clinical follow-up
36 with results also suggesting that diploid and tetraploid stem cell lines are the main lines of the
37 HCCs as well as a reduction in the percentage of binucleated hepatocytes in HCC. Gramantieri
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-305 DRAFT—DO NOT CITE OR QUOTE
-------
1 et al. (1996) report that the percentage of binucleated cells was reduced in most of HCC they
2 studied (i.e., the mean percentage of binucleated cells 9% in comparison to 24% found in normal
3 liver) and that most HCC, as many other solid neoplasms, showed altered nuclear parameters.
4 Along with reporting pathways that are perturbed in HCC, emerging evidence also shows
5 that signatures of pathway are predictive of clinical characteristics of HCC. A number of studies
6 have examined gene expression in tumors to try to determine which pathways may have been
7 disturbed in an attempt to predict survival and treatment options for the patients and to
8 investigate possible MO As for the tumor induction and progression. Chen et al. (2002)
9 described a systematic characterization of gene expression patterns in human liver cancers using
10 cDNA microarrays to study tumor and nontumor liver tissues in HCC patients, and of note did
11 quality assurance on their microarray chips (many studies do not report that they have done so),
12 and examined the effects of hepatitis virus on its subject and identified people with it. Most
13 importantly, Chen et al. (2002) provided phenotypic anchoring of each tumor with its genetic
14 profile rather than pooling data. The hierarchical analysis demonstrated that clinical samples
15 could be divided into two major clusters, one representing HCC samples and the other with a few
16 exceptions, representing nontumor liver tissues. Most importantly, expression patterns varied
17 significantly among the HCC and nontumor liver samples and that samples from HBV-infected,
18 hepatitis C virus infected, and noninfected individuals were interspersed in the HCC branch.
19 Thus, tumors from people infected with HVB, HVC and noninfected people with HCC were
20 interspersed in the HCC pattern and could be discerned based on etiology. One cluster of genes
21 was highly expressed in HCC samples compared with nontumor liver tissues included a
22 "proliferation cluster" comprised of genes whose functions are required for cell-cycle
23 progression and whose expression levels correlate with cellular proliferation rates with most of
24 the genes in this cluster are specifically expressed in the G2/M phase. Gene profiles for HCC
25 were consistent with fewer molecular features of differentiated normal hepatocytes. Chen et al.
26 (2002) noted that both normal and liver tumors are complex tissue compose of diverse cells and
27 that distinct patterns of gene expression seemed to provide molecular signatures of several
28 specific cell types including expression of two clusters of genes associated with T and B
29 lymphocytes, presumably reflecting lymphocytic infiltration into liver tissues, and genes
30 associated with stellate cell activation. This important finding acknowledges that HCC are not
31 only heterogeneous in hepatocyte phenotype but are made up of many other nonparenchymal cell
32 types and that gene expression patterns reflect that heterogeneity. A gene cluster was also
33 identified at a higher level in HCC that included several genes typically expressed in endothelial
34 cells, including CD34, which is expressed in endothelial cells in veins and arteries but not in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-306 DRAFT—DO NOT CITE OR QUOTE
-------
1 endothelial cells of the sinusoids in nontumor liver and which may reflect disruption of the
2 molecular program that normally regulate blood vessel morphogenesis in the liver.
3 Of great importance was the investigation by Chen et al. (2002) of whether samples from
4 multiple sites in a single HCC tumor, or multiple separate tumor nodules in one patient, would
5 share a recognizable gene expression signature. With a few instructive exceptions, all the tumor
6 samples from each patient clustered were reported to cluster together. To further examine the
7 relationship among multiple tumor samples from individual patients, they calculated the pairwise
8 comparison for all pairs of samples and samples some primary tumors multiple times. Tumor
9 patterns of gene expression were more highly correlated those seen in samples from the same
10 patient than other patients but every tumor had a distinctive and characteristic gene expression
11 pattern, recognizable in all samples taken from different areas of the same tumor. For multiple
12 discrete tumor masses obtained from six patients, three of these patients had multiple tumors
13 with a shared distinctive gene expression pattern but in three other patients, expression patterns
14 varied between tumor nodules and the difference provided new insights into the sources of
15 variation in molecular and biological characteristics of cancers. Thus, in some patients multiple
16 tumors were from the same clone, as demonstrated by a similar gene expression profile, but for
17 some patients multiple tumors were arising from differing clones within the same liver. In
18 regard to whether the distinctive expression patterns characteristic of each tumor reflect the
19 individuality of the tumor or are determined by the patient in whom the tumor arose, analysis of
20 the expression patterns observed in the two tumor nodules from one patient showed that the two
21 tumors were not more similar than those of an arbitrary pair of tumors from different patients.
22 These results show the heterogeneity of HCC and that "one gene pattern" will not be
23 characteristic of the disease.
24 However, HCC did have a pattern that differed from other cancers. Chen et al. (2002)
25 analyzed the expression patterns of 10 randomly selected HCC samples and 10 liver metastases
26 of other cancers and reported that the HCC samples and the metastatic cancers clustered into two
27 distinct groups, based on difference in their patterns of gene expression. Although some of the
28 HCC samples were poorly differentiated and expressed the genes of the liver-specific cluster at
29 very low levels compared to with either normal liver or well-differentiated HCC, the genes of the
30 liver-specific cluster were reported to be consistently expressed at higher levels in HCC than in
31 tumors of nonliver origin. Metastatic cancers originating from the same tissue typically clustered
32 together, expressing genes characteristic of the cell types of origin. Thus, liver cancer was
33 distinguishable from other cancer even though very variable in expression and differentiation
34 state.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-307 DRAFT—DO NOT CITE OR QUOTE
-------
1 In an attempt to create molecular prognostic indices that can be used for identification of
2 distinct subclasses of HCC that could predict outcome, Lee et al. (2004a) report two subclasses
3 of HCC patients characterized by significant differences in the length of survival. They also
4 identified expression profiles of a limited number of genes that accurately predicted the length of
5 survival. Total RNAs from the 19 normal livers, including "normal liver in HCC patients," were
6 pooled and used as a reference for all microarray experiments and thus variations between
7 patients, and especially differences due to conditions predisposing HCC, were not determined.
8 DNA microarray data using hierarchical clustering was reported to yield two major clusters, one
9 representing HCC tumors, and the other representing nontumor tissues with a few exceptions that
10 were not characterized by the authors. Lee et al. (2004a) report that, along with 2 distinctive
11 subtypes of gene expression patterns in HCC, there was heterogeneity among HCC gene
12 expression profiles and that one group had an overall survival time of 30.8 months and the other
13 83.7 months. Only about half the patients in each group were reported to have cirrhosis.
14 Expression of typical cell proliferation markers such as PCNA and cell cycle regulators such as
15 CDK4, CCNB1, CCNA2, and CKS2 was greater in one class than the other of HCC.
16 The report by Boyault et al. (2007) attempted to compare etiology and genetic
17 characterization of the tumors they produce and confirms the heterogeneity of HCC, some
18 without attendant genomic instability. Boyault et al. (2007) reported that genetic alterations are
19 indeed closely associated with clinical characteristics of HCC that define 2 mechanisms of
20 hepatocarcinogenesis.
21
22 The first type of HCC was associated with not only a high level of chromosome
23 instability and frequent TP53 and AXIN1 mutations but also was closely linked to
24 HBV infections and a poor prognosis. Conversely, the second subgroup of HCC
25 tumors was chromosome-stable, having a high incidence of activating p-catenin
26 alteration and was not associated with viral infection.
27
28 Boyault et al. (2007) reported that in a series of 123 tumors, mutations in the CTNNB1
29 (encoding P-catenin), TP53, ACINI, TCF1, PIK3CA and KRAS genes in 34, 31, 13, 5, 2, and
30 1 tumors were identified, respectively. No mutations were found in NRAS, HRAS, and EGFR.
31 Hypermethylation of the CDKN2A and CDH1 promoter was identified in 35 and 16% of the
32 tumors, respectively. Boyault et al. (2007) grouped tumors by genomic expression as well as
33 other factors. HCC groups associated with high rate of chromosomal instability were reported to
34 be enriched with over expression of cell-cycle/proliferation/DNA metabolism genes. They
35 concluded that "the primary clinical determinant of class membership is HBV infection and the
36 other main determinants are genetic and epigenetic alterations, including chromosome instability,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-308 DRAFT—DO NOT CITE OR QUOTE
-------
1 CTNNB1 and TP53 mutations, and parental imprinting. Tumors related to HCV and alcohol
2 abuse were interspersed across subgroups G3-G6." Boyault et al. (2007) suggested that there
3 results indicate that HBV infection early in life leads to a specific type of HCC that has immature
4 features with abnormal parental gene imprinting selections, possibly through the persistence of
5 fetal hepatocytes or alternatively through partial dedifferentiation of adult hepatocytes. "These
6 Gl tumors are related to high-risk populations found in epidemiological studies."
7
8 E.3.2. Animal Models of Liver Cancer
9 There are obvious differences between rodents and primate and human liver, and there is
10 a difference in background rates of susceptibility to hepatocarcinogenesis. With strains of mice
11 there are large differences in responses to hepatotoxins (e.g., acetaminophen) and to
12 hepatocarcinogens as well as background rates of hepatocarcinogenecity. Maronpot (2007)
13 reports that modulators of murine hepatocarcinogenesis, such as diet, hormones, oncogenes,
14 methylation, imprinting, and cell proliferation/apoptosis are among multiple mechanistically
15 associated factors that impact this target organ response in control as well as in treated mice, and
16 suggests that there is no one simple paradigm to explain the differential strain sensitivity to
17 hepatocarcinogenesis. Because of the variety of studies with differing protocols used to generate
18 susceptibility data, direct comparisons among strains and stocks is problematic but in regard to
19 susceptibility to carcinogenicity the C3H/HeJ and C57BL/6J mouse have been reported to have
20 up to a 40-fold difference in liver tumor multiplicity (Maronpot, 2007). However, as noted
21 above, TCE causes liver tumors in C6C3F1 and Swiss mice with studies of trichloroethylene
22 metabolites dichloroacetic acid, trichloroacetic acid, and CH suggesting that both dichloroacetic
23 acid and trichloroacetic acid are involved in trichloroethylene-induced liver tumorigenesis.
24 Many effects reported in mice after dichloroacetic acid exposure are consistent with conditions
25 that increase the risk of liver cancer in humans and can involve GST Xi, histone methylation, and
26 overexpression of insulin-like growth factor-II (IGF-II; Caldwell and Keshava, 2006). The
27 heterogeneity of liver phenotype observed in mouse models is also consistent with human HCC.
28 These data lend support to the qualitative relevance of the mouse model for TCE-induced cancer
29 risk.
30 Bannasch et al. (2003) made important observations that have implications regarding the
31 differences in susceptibility between rodent and human liver cancer. They stated that
32
33 Although the classification of such nodular liver lesions in rodents as hyperplastic
34 or neoplastic has remained controversial, persistent nodules of this type are
35 considered neoplasms, designated as adenomas. In human pathology, the
36 situation appears to be paradoxical because adenomas are only diagnosed in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-309 DRAFT—DO NOT CITE OR QUOTE
-------
1 noncirrhotic liver, yet a confusing variety terms avoiding the clearcut
2 classification as an adenoma has been created for nodular lesions in liver
3 cirrhoses, not withstanding that the vast majority hepatocellular carcinomas
4 develop in cirrhotic livers. Even if a portion of these nodular lesions would be
5 regarded as adenomas, being integrated into an adenoma-carcinoma sequence as
6 observed in many animal experiments, clinical and epidemiological records of
7 liver neoplasms, including both benign and malignant forms, would increase
8 considerably. This would not only bring hepatic neoplasia further into focus of
9 human neoplasia in general, but also shed new light on the classification of some
10 chemicals producing high incidence of liver neoplasms in rodents, but appearing
11 harmless to humans according to epidemiological evaluations solely based on the
12 incidence of hepatocellular carcinoma in exposed populations.
13
14 Thus, that in humans only HCCs are recorded but in animals adenomas are counted as
15 neoplasms, may indicate that the scope of the problem of liver cancer in humans may be
16 underestimated.
17 Tumor phenotype differences have been reported for several decades through the work of
18 Bannasch et al. The predominant cell line of foci of altered hepatocytes (FAH) have excess
19 glycogen storage early in development that appears to be similar to that shown by DC A
20 treatment. Bannasch et al. (2003) report that "the predominant glycogenotic-basophilic cell line
21 FAH reveals that there is an overexpression of the insulin receptor, the IGF-1 receptor, the
22 insulin receptor substrates-1/2 and other components of the insulin-stimulated signal transduction
23 pathway." Bannasch states that foci of this type have increased expression of GST-u and insulin
24 has also been shown to induce the expression of GST-pi but that hyperinsulin-induced foci do
25 not show increased GST-u. Cellular dedifferentiation during progression from glycogenotic to
26 basophilic cell populations is associated with downregulation in insulin signaling. The
27 amphophilic-basophilic cell lineage of peroxisome proliferators and hepadnaviridae were
28 reported to have foci that mimic effects of thyroid hormone with mitochondrial proliferation and
29 activation of mitochondrial enzymes. Bannasch et al. (2003) state that
30
31 the unequivocal separation of 2 types of compounds, usually classified as
32 initiators and promoters, remains a problem at the level of the foci because at least
33 the majority of chemical hepatocarcinogens seem to have both initiating and
34 promoting activity, which may differ in quantitative rather than qualitative terms
35 from one compound to another.. .Whereas genetic mutations have been
36 predominantly postulated to initiate hepatocarcinogenesis for many years, more
37 recently epigenetic changes have been increasingly discussed as a plausible cause
38 of the evolution of preneoplastic foci characterized by metabolic changes
39 including the expression of GSTpi.
40
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-310 DRAFT—DO NOT CITE OR QUOTE
-------
1 Su and Bannasch (2003) report that glycogen-storing foci represents early lesion with the
2 potential to progress to more advance glycogen-poor basophilic lesions through mixed cell foci
3 and resulting hyperproliferative lesions and are associated with HCC in man. Small-cell change
4 (SCC) of liver parenchyma (originally called liver cell dysplasia of small cell size) is reported to
5 share cytological and histological similarities to early well defined HCC. Close association
6 between SCC and more advanced (basophilic) foci indicates that foci often progress to HCC
7 through SCC in humans. SCC were reported to be present in all basophilic foci. Previous
8 studies were cited that showed that the biochemical phenotype of human FAH, mainly including
9 glycogen storing clear cell foci and clear cell-predominated mixed cell foci, were observed in
10 more than 50% of cirrhotic livers with or without HCC. FAH of clear and mixed cell types were
11 observed in almost all livers bearing HCC, and in chronic liver diseases without HCC but at a
12 lower frequency. Su and Bannasch (2003) report that
13
14 the finding of mixed cell foci (MCF) mainly in livers with high-risk or
15 cryptogenenic cirrhosis indicates that these are more advanced precursor lesions
16 in man, in line with earlier observations in experimental animals. Considering
17 their preferential emergence in cirrhotic livers of the high-risk group, their
18 unequivocally elevated proliferative activity, and the resulting large size with
19 frequent nodular transformation, we suggest that mixed cell populations are
20 endowed with a high potential to progress to HCC in humans, as previously
21 shown in rats.
22
23 In human HCC, irregular areas of liver parenchyma with marked cytoplasmic amphophilia,
24 phenotypically similar to the amphophilic preneoplastic foci in rodent liver exposed to different
25 hepatocarcinogenic chemicals (e.g., DHEA a peroxisome proliferator) or the hepadnaviruses
26 were reported to present in 45% of the specimens from cirrhotic livers examined. "However,
27 more data are needed to elucidate the nature of the oncocytic and amphophilic lesions regarding
28 their role in HCC development."
29 With respect to the ability respond to a mitogenic stimulus, differences between primate
30 and rodent liver response to a powerful stimulus, such as partial hepatectomy, have been noted
31 that indicate that primate and human liver respond differently (and much more slowly) to such a
32 stimulus. Gaglio et al. (2002) report after 60% partial hepatectomy in Rhesus macaques
33 (Macaca mulatto), the surface area of the liver remnant was restored to its original preoperative
34 value over a 30 day period. The maximal liver regeneration occurred between days 14 and 21,
35 with thickening of liver cell plates, binucleation of hepatocytes, Ki-67 and PCNA expression
36 (occurring in hepatocytes throughout the lobule at a maximum labeling index of 30%), and
37 mitoses parallel increased most prominently between posthepatectomy days 14 and 30.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-311 DRAFT—DO NOT CITE OR QUOTE
-------
1 However, cytokines associated with inducing proliferation were elevated much earlier. TGF-a,
2 IL-6, HGF, IL-6 and TNF-a mRNA persisted until Day 14, with peak elevations of IL-6, TNF-a,
3 occurring 24 hours later surgery, and IL-6 reduced to control levels by Day 14. Gaglio et al.
4 (2002) suggest that their results clearly indicate that the pattern and timing of liver regeneration
5 observed in this nonhuman primate model are significantly different when comparing different
6 species (e.g., peak expression of Ki-67 in a 60% partial hepatectomy model in rats occurs within
7 hours following partial hepatectomy) and that the difference in timing and pattern of maximal
8 hepatocellular regeneration cannot be explained simply by differences in size of animals (e.g.,
9 60% partial hepatectomy in dogs produced liver regeneration peaks at 72 hours with weights
10 approximating the weights of the Rhesus macaques). They note that previous studies in humans,
11 who underwent 40-80% partial hepatectomy, reveal a similar delay in peak liver regeneration
12 based on changes in serum levels of ornithine decarboxylase and thymidine kinase, further
13 highlighting significant interspecies differences in liver regeneration. For C57BL/6 X 129 mice
14 Fujita et al. (2001) report that after partial hepatectomy, the liver had recovered more than 90%
15 of its weight within 1 week. This difference in response to a mitogenic stimulus has impacts on
16 the interpretations of comparisons between rodent and primate liver responses to chemical
17 exposures which give a transient increases in DNA synthesis or cell proliferation such as PPARa
18 agonists. Also, as stated above, the primate and human liver, while having a significant
19 polyploidy compartment, do not have the extent of polyploidization and the early onset of that
20 has been observed in the rodent. However, as noted by Lapis et al. (1995), exposure to DEN has
21 proven to be a highly potent hepatocarcinogen in nonhuman primates, inducing malignant
22 tumors in 100% of animals with an average latent period of 16 months when administered at
23 40 mg/kg intraperitoneally every 2 weeks.
24 In regard to species extrapolation of epigenomic changes between humans and rodents,
25 Weidman et al. (2007) caution that
26
27 Although we do predict some overlap between mouse and human candidate
28 imprinted genes identified through our machine-learning approach, it is likely that
29 the most significant criterion in species-specific identification will differ. This
30 difference underscored the importance for increased caution when assessing
31 human risk from environmental agents that alter the epigenome using rodent
32 models; the molecular pathways targeted may be independent.
33
34 Despite species differences, the genome of the mouse has been sequenced and many
35 transgenic mouse models are being used to study the consequences of gene expression
36 modulation and pathway perturbation to study human diseases and treatments. However, the use
37 of transgenic models must be used with caution in trying to determine to determine MOAs and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-312 DRAFT—DO NOT CITE OR QUOTE
-------
1 the background effects of the transgene (including background levels of toxicity) and specificity
2 of effects must be taken into account for interpretation of MO A data, especially in cases where
3 the knockout in the mouse causes significant liver necrosis or steatosis (Keshava and Caldwell,
4 2006; Keshava and Caldwell, 2006; Caldwell and Keshava, 2006; Caldwell et al., 2008b). For
5 the determination of effects of pathway perturbation and similarity to human HCC phenotype,
6 mouse transgenic models have been particularly useful with tumors produced in such models
7 shown to correlate with tumor aggressiveness and survival to human counterparts.
8
9 E.3.2.1. Similarities with Human and Animal Transgenic Models
10 Mice transgenic for transforming growth factor -a (a member of the EGF family and a
11 ligand for the ErfB receptors) develop HCCs (Farazi and DePinho, 2006). Compound TGFa and
12 MYC transgenic mice show increase hepatocarcinogenesis that is associated with the disruption
13 of TGF-pl signaling and chromosomal losses, some of which are syntenic to those in human
14 HCCs that include the retinoblastoma (KB) tumor suppressor locus (Sargent, 1999). Lee et al.
15 (2004b) investigated whether comparison of global expression patterns of orthologous genes in
16 human and mouse HCCs would identify similar and dissimilar tumor phenotypes, and thus,
17 allow the identification of the best-fit mouse models for human HCC. The molecular
18 classification of HCC on the basis of prognosis in Lee et al. (2004a) was further compared with
19 gene-expression profiles of HCCs from seven different mouse models (Lee et al., 2004b).
20 Lee et al. (2004b) characterized the gene expression patters of 68 HCC from seven different
21 mouse models; two chemically induced (Ciprofibrate and diethylnitrosamine), four transgenic
22 (targeted overexpression of Myc, E2F1, Myc andE2Fl, andMyc and Tgfa in the liver). HCCs
23 from some of these mice (MYC, E2F1 and MYC-E2F1 transgenics) showed similar gene-
24 expression patterns to the ones of HCCs from patients with better survival. Murine HCCs
25 derived for MYC-TGF-a transgenic model or diethylnitrosamine-treated mice showed similar
26 gene-expression patterns to HCCs from patients with poor survival. The authors report that Myc
27 Tgfa transgenic mice typically have a poor prognosis, including earlier and higher incident rates
28 of HCC development, higher mortality, higher genomic instability and higher expression of poor
29 prognostic markers (e.g., AFP) and that Myc and Myc/E2fl transgenic mice have relatively
30 higher frequency of mutation in p-catenin (CatnV) and nuclear accumulation of p-catenin that are
31 indicative of lower genomic instability and better prognosis in human HCC.
32 Lee et al. (2004b) indentified three distinctive HCC clusters, indicating that gene
33 expression pattern of mouse HCC are clearly heterogeneous and reported that Ciprofibrate-
34 induced HCCs and HCCs from Acox -/- mice were closely clustered and well separated from
35 other mouse models. However, are several issues regarding this study that give limitations to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-313 DRAFT—DO NOT CITE OR QUOTE
-------
1 some of its conclusions regarding the Acox -/- mouse and Ciprofibrate treatment. The Acox -/-
2 mouse is characterized by profound hepatonecrosis, which confounds conclusions regarding
3 gene expression related to PPARa agonism made by the authors. There was very limited
4 reporting of the animal models (DEN and Clofibrate) protocols used. Only three tumors were
5 examined for Clofibrate treatment and it is unknown if the tumors were from the same animals.
6 Similarly only three tumors were examined from DEN treatment, which has been shown to
7 produce heterogeneous tumors and to produce necrosis in some paradigms of exposure.
8 Myc/E2Fl and E2F1 mice were split in both clusters that were compared with human HCCs.
9 The authors used previously published data from Meyer et al. (2003) for tumors from Acoxl"1"
10 null mice, DENA-treated mice and Ciprofibrate-treated mice.
11 Meyer et al. (2003) examined three tumors from 2 C57BL/6J mice fed Ciprofibrate for
12 19 months and three tumors from 2 C57BL/6J mice injected with DEN at 2-3 months but the age
13 at which tumors appear was not given by the authors. Pooled mRNA from animals of varying
14 age (5-15 months old) was used for controls. mRNAs that differed by 2-fold in tumors were
15 reported to be: 60 genes up-regulated and 105 genes down-regulated in Acoxl"1" null mice
16 tumors; 136 genes up-regulated and 156 genes down-regulated in Ciprofibrate-induced tumors;
17 and 61 genes up-regulated and 105 genes down-regulated in DEN-induced tumors. The authors
18 state that "Each tumor class revealed a somewhat different unique expression pattern." There
19 were "genes that were general liver tumor markers in all three types of tumors" with 38 genes
20 commonly deregulated in all three tumor types. On note, the cell cycle genes (CDK4,
21 CDC25Am CDC7 and MAPK3) cited by Lee et al. (2004b) as being more highly expressed in
22 DEN-induced tumors were not reported to be changed in DEN tumors in Meyer et al. (2003) or
23 to be altered in the Acoxl"1" null mice or mice treated with Ciprofibrate. Finally, the distinction
24 between groups may be dominated by gene expression changes in a large number of genes that
25 are related to PPAR activation but not related to hepatocarcinogenesis.
26 Calvisi et al. (2004a) used transgenic mice to study pathway alterations and tumor
27 phenotype and to further examine the premise that genomic alterations (genetic and epigenetic)
28 characteristic of HCC can describe tumors into 2 broad categories, the first category
29 characterized by activation of the Wnt/Wingless pathway via disruption of p-catenin function
30 and chromosomal stability and the second by chromosomal instability. Increased coexpression
31 of c-myc with TGF-a or E2F-1 transgenic mice was reported to result in a dramatic synergistic
32 effect on liver tumor development when compared with respective monotransgenic lines,
33 including shorter latency period, and more aggressive phenotype whereas P-catenin activation is
34 relatively common in HCCs developed in c-myc and c-myc/TGF-pl transgenic mice, rare in the
35 c-myc/TGF-a transgenic line which also has genomic instability. Calvisi et al. (2004a) also
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-314 DRAFT—DO NOT CITE OR QUOTE
-------
1 report that p-catenin staining correlated with histopathologic type of liver tumors. Eosinophilic
2 tumors with abnormal nuclear staining of P-catenin were predominant in neoplastic lesions
3 characteristic of c-myc and c-myc/E2Fl lesions. Poorly differentiated HCCs with basophilic or
4 clear-cell phenotypes developed more frequently in c-myc/TGF-a and TGF-a mice and often
5 showed a reduction or loss of P-catenin immunoreactivity. P-catenin mutation was associated
6 with a more benign phenotype. Calvisi et al. (2004a) note that the relationship between
7 P-catenin activation, tumor grade, and clinical outcome in human HCC remains controversial.
8
9 There are studies that show a significant correlation between P-catenin nuclear
10 accumulation, a high grade of HCC tumor differentiation, and a better prognosis,
11 whereas others find that nuclear accumulation of P-catenin may be associated
12 with poor survival or that it does not affect clinical outcome.
13
14 Calvisi et al. (2004b) report for E-cadherin a variety of morphologenetic events, including
15 cell migration, separation, and formation of boundaries between cell layers and differentiation of
16 each cell layer into functionally distinct structures. Loss of expression of E-cadherin was
17 reported to result in dedifferentiation, invasiveness, lymph node or distant metastasis in a variety
18 of human neoplasms including HCC and that the role of E-cadherin might be more complex that
19 previously believed.
20
21 In order to elucidate the role of E-cadherin in the sequential steps of liver
22 carcinogenesis, we have analyzed the expression patterns of E-cadherin in a
23 collection of preneoplastic and neoplastic liver lesions from c-Myc, E2F1,
24 c-Myc/TGF-a and c-Myc/E2Fl transgenic mice. In particular, we have
25 investigated the relevance of genetic, epigenetic, and transcript onal mechanisms
26 on E-cadherin protein expression levels. Our data indicate that loss of E-cadherin
27 contributes to HCC progression in c-Myc transgenic mice by promoting cell
28 proliferation and angiogenesis, presumably through the upregulation of HIF-la
29 and VEGF proteins.
30
31 The c-Myc line, was most like wild-type and lost E-cadherin in the tumors. c-Myc/TGF-a
32 dysplastic lesion were reported to show overexpression of E-cadherin mainly in pericentral areas
33 with E2F1 clear cell carcinoma showed intense staining of E-cadherin. Reduction or loss of E-
34 cadherin expression is primarily determined by loss of heterozygosity at the E-cadherin locus or
35 by its promoter hypermethylation in human HCC Calvisi et al. (2004b) determined the status of
36 the E-cadherin locus and promoter methylation in wild-type livers and tumors from transgenic
37 mice by microsatellite analysis and methylation specific PCR, respectively.
38
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-315 DRAFT—DO NOT CITE OR QUOTE
-------
1 Wild-type livers and HCCs, regardless of their origins, showed the absence of
2 LOH at the E-cadherin locus. E-cadherin promoter was not hypermethylated in
3 wild-type, c-Myc/TGF-a and E2F1 livers. No E-cadherin promoter
4 hypermethylation was detected in c-Myc and c-Myc/E2Fl HCCs with normal
5 levels of E-cadherin protein. In striking contrast, seven of 20 (35%) of c-Myc and
6 two of four (50%) c-Myc/E2Fl HCCs with downregulation of E-cadherin
7 displayed E-cadherin promoter hypermethylation. These results suggest that
8 promoter hypermethylation might be responsible for E-cadherin downregulation
9 in a subset of c-Myc and c-Myc/E2Fl HCCs... The molecular mechanisms
10 underlying down-regulation of E-cadherin in c-Myc tumors remain poorly
11 understood at present. No LOH at the E-cadherin locus was detected in the c-
12 Myc HCCs whereas only a subset of c-Myc tumors displayed hypermethylation of
13 the E-cadherin promoter. Furthermore, no association was detected between
14 E-cadherin downregulation and protein levels of transcriptional repressers, Snail,
15 Slug or the tumor suppressor WT1, in disagreement with the finding that
16 overexpression of Snail suppresses E-cadherin in human HCC... E-cadherin might
17 play different and apparently opposite roles, which depend on specific tumor
18 requirements in both human and murine liver carcinogenesis.
19
20 Importantly, the results of Calvisi et al. (2004b) show that hypermethylation of promoters can be
21 associated with down regulation of a gene in mouse liver tumors similar to human HCC and that
22 tumors can have the same behavior with methylation change as with loss of hetererozygosity.
23 This report also gives evidence of the usefulness of the mouse model to study human liver
24 cancer as it shows the similarity of dysfunctional regulation in mouse and human cancer and the
25 heterogeneity within and between mouse lines tumors with differing dysfunctions in gene
26 expression. This parallels human cancer where there is heterogeneity in tumors from one person
27 and every tumor has its own signature. Finally, this report correlates differing pathway
28 perturbations with mouse liver phenotypes similar to those reported in experimental
29 carcinogenesis models and for TCE and its metabolites.
30 Farazi and DePinho (2006) suggest that
31
32 as comparative array CGH analysis of various murine cancers has shown that such
33 aberrations often target syntenic loci in the analogous human cancer type, we
34 further suggest that comparative genomic analysis of available mouse model of
35 mouse HCC might be particularly helpful in filtering through the complex human
36 cancer genome. Ultimately, mouse models that share features with human HCCs
37 could serve as valuable tools for gene identification and drug development.
38 However, one needs to keep in mind key differences between mice and humans.
39 For example, as noted in certain human HCC cases, telomere shortening might
40 drive the genomic instability that enables the accumulation of cancer-relevant
41 changes for hepatocarcinogenesis. As mice have long telomeres, this aspect of
42 hepatocarcinogenesis might be fundamentally different between the species and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-316 DRAFT—DO NOT CITE OR QUOTE
-------
1 provide additional opportunities for model refinement and testing of this
2 mechanism through use of a telomere deficient mouse model. These and other
3 cross-species difference, and limitations in the use of human cell-culture systems,
4 must be considered in any interpretation of data from various model systems
5 (Farazi and DePinho, 2006).
6
7 Thus, these mouse models of liver cancer inductions are qualitatively able to mimic human liver
8 cancer and support the usefulness of mouse models of cancer.
9
10 E.3.3. Hypothesized Key Events in HCC Using Animal Models
11 E.3.3.1. Changes in Ploidy
12 As stated above in Section E. 1.1, increased polyploidization has been associated with
13 numerous types of liver injury and appears to result from exposure to TCE and its metabolites as
14 well as changes in the number of binucleate cells. Hortelano et al. (1995) reported that cytokines
15 and NO can affect ploidy and further suggests a role of these changes for carcinogenesis in
16 general. Vickers and Lucier (1996) noted that while both DEN and 17 a-ethinylestradiol have
17 been reported to enhance the proportion of diploid hepatocytes, initiators like N-
18 nitrosomorpholine are reported to increase the proportion of hypertrophied and polyploidy
19 hepatocytes. The relationship of such changes to cancer induction has been studied in transgenic
20 mouse models and in models involved with mitogens of differing natures.
21 Melchiorri et al. (1993) report the response pattern of the liver to acute treatment with
22 primary mitogens in regard to ploidy changes occurring in rat liver following two different types
23 of cell proliferation: compensatory regeneration induced by surgical partial hepatectomy (PH)
24 and direct hyperplasia induced by the mitogens lead nitrate and Nafenopin (a PPARa agonist) in
25 8 week old male Wistar rats. Feulgen stain was used and DNA content quantified by image
26 cytometry in mononucleate and binucleate cells. Mitotic index was determined in the same
27 samples. The term "diploid" was used to identify cells with a single, diploid nucleus and
28 tetraploid for cells containing 2 diploid nuclei or one tetraploid nucleus referred (bi- and
29 mononucleate, respectively). Octoploid cells were identified as either binucleate or
30 mononucleate.
31
32 During liver regeneration following surgical PH an increase in the mitotic index
33 with a peak at 24 hours was observed. The most striking effect associated with
34 the regenerative response was the almost complete disappearance of binucleate
35 cells, tetraploid (2 X 2c) as well as octoploid (4 X 2c) with only < 10% of the
36 control values being present 3 days after PH... Concomitantly, an increase in
37 mononucleate tetraploid (4c) as well as mononucleate octoploid (8c) cells was
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-317 DRAFT—DO NOT CITE OR QUOTE
-------
1 observed, resulting at 3 days after PH in a population made up of almost entirely
2 (98%) by mononucleated cells.
3
4 However, lead nitrate treatment was reported to induce rapid increase in the formation of
5 binucleate cells occurring 3 days after treatment, their number accounting for 40% of the total
6 cell population versus 22% binucleate cells in control rats and 2% in PH animals killed at the
7 same time point. The increased binuclearity was reported to be observed only in the 4 X 2c cells
8 (25 vs. 6% of the controls) and in 8 X 2c cells (3.7 vs. 0.1% of controls). The increase in 4 X 2c
9 and 8 X 2c cells was reported to be accompanied by a concomitant reduction in 2 X 2c cells with
10 the change induced in cellular ploidy by lead nitrate resulting in 37% of cells being either 8c or
11 16c. However, at the same time point, cells having a ploidy higher than 4c were reported to
12 account for only 11% in PH rats and 9% in control animals. Changes in the ploidy pattern were
13 reported to be preceded by an increased mitotic activity, which was maximal 48 hours after
14 treatment with lead nitrate. The increase in mitotic index in lead nitrate-treated rats was
15 associated with a striking increase in the labeling index of hepatocytes (60.1 vs. 3% of control
16 rats) and to an almost doubling of hepatic DNA content in 3 days after lead nitrate. Melchiorri et
17 al. (1993) concluded that the entire cell cycle appeared to be induced by lead nitrate but that the
18 finding of a high increase of binucleate cells suggested that lead nitrate-induced liver growth,
19 unlike liver regeneration induced by partial hepatectomy, was characterized by an uncoupling
20 between cell cycle and cytokinesis. This raised questions whether lead nitrate-induced liver
21 growth resulted in a true increase in cell number or is only the expression of an increased
22 hepatocyte ploidy. They reported that part of the increase in DNA content observed 3 days after
23 lead nitrate was indeed expression of polyploidizing process due to acytokinetic mitoses but that
24 a consistent increase in cells number (+26%) was also induced by lead nitrate treatment.
25 After Nafenopin treatment, Melchiorri et al. (1993) reported that the increase in DNA
26 content was increased 22% over controls and was much lower than induced by lead nitrate and
27 that Nafenopin did not induce significant changes in binucleate cell number. However, a shift
28 towards a higher ploidy class (8c) was reported to be observed following Nafenopin and the 21%
29 increase in DNA content seen after Nafenopin treatment was almost entirely due to increase in
30 the ploidy state with only 7% increase in cell number.
31 Melchiorri et al. (1993) examined whether hepatocytes characterized by high ploidy
32 content (highly differentiated cells) would be preferentially eliminated by apoptosis. An increase
33 in apoptotic bodies was reported to be associated with the regression phase after lead nitrate
34 treatment (when liver mass is reduced) but despite the elimination of excess DNA, the changes in
35 ploidy distribution induced by lead nitrate were found to persist suggested that polyploidy cells
36 were not preferentially eliminated by apoptosis during the regression phase of the liver.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-318 DRAFT—DO NOT CITE OR QUOTE
-------
1 Melchiorri et al. (1993) note that other studies in rat exposed to the mitogens cyproterone acetate
2 (CPA) and the peroxisome proliferator MCP also reported a very strong decline in binucleate
3 cells with a concomitant increase in mononucleate tetraploid cells in the liver similar to the
4 pattern described after partial hepatectomy.
5 Lalwani et al. (1997) reported the results of 1,000 ppm WY-14,643 exposure in male
6 Wistar rats after 1, 2, and 4 weeks and suggested that an early wave of nuclear division occurred
7 at the early stages of exposure without cumulative effects on cell proliferation. Consistent with
8 hepatomegaly, WY-14,643-treated were reported to exhibit multifocal hepatocellular
9 hypertrophy and karyomegaly by routine microscopic analysis. For binucleate hepatocytes, there
10 were no reported differences between WY-14,643 and controls for days 4 and 11 but an increase
11 in the number at Day 25 in WY-14,643-treated animals compared to controls. Increases in the
12 diameter of nuclei were shown by WY-14,643 treatment from Day 11 and 25 with increasing
13 numbers of cells displaying larger nuclear diameters. The mitotic index was reported not to be
14 significantly changed in WY-14,643 treated rats compared to controls. Mitotic figures did not
15 appear to survive the treatment necessary for flow cytometric analyses. PCNA was increased on
16 Day 4 in WY-14,643- treated animals compared to controls whereas no differences were found
17 on days 11 and 25. However, immunohistochemistry was reported to show remarkable increases
18 in BrdU-labeled nuclei in liver sections after 4 days of labeling with the populations of BrdU-
19 labeled cell declining over the course of treatment. The labeling index was high and
20 approximately 80% of the BrdU-labeled cells were in periportal areas. PCNA-expressing cells
21 were increased in the periportal area of the liver. Intense nuclear staining of PCNA was evident
22 as an indicator of DNA replication in S phase. Microscopic examination showed BrdU labeling
23 only in periportal hepatocytes, whereas no significant labeling was observed in nonparenchymal
24 cells, indicating that the replicative activity was confined to the liver cells. Lalwani et al. (1997)
25 suggested that their results showed that events related to cell proliferation occur in the initial
26 phase of WY-14,643 treatment in rats but not followed by changes in the rate of DNA synthesis
27 as the treatment progressed. They note that Marsman et al. (1988) observed constant increases in
28 DNA synthesis by [3H]-thymidine authoradiography with up to 1 year of continuous
29 administration of WY-14,643, whereas the rate of DNA synthesis or the BrdU labeling index in
30 their study declined after the first 4 weeks of treatment. They suggest that the increased
31 percentage of cells appearing in G2-M phase and the analysis of liver nuclear profiles suggest
32 that the progression of these additional cells (i.e., cells that are stimulated to enter the cell cycle
33 by the test agent) through the cell cycle is arrested in the late stages of the cell cycle. They state
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-319 DRAFT—DO NOT CITE OR QUOTE
-------
1 Unlike BrdU labeling, which demonstrated DNA synthesis activity over the 4-day
2 labeling period, the PCNA labeling index represents levels of the protein product
3 at an interval post treatment. PCNA expression in cells exposed to chemicals or
4 to WY may not provide true representation of S phase or proliferative activity
5 because PCNA-expressing nuclei were also found in GO=G1 and G2-M phases.
6
7 Lalwani et al. (1997) concluded that cell proliferation alone does not appear to constitute a
8 determining process leading to tumors in most tissues and sustained cell replication may not be a
9 primary feature of peroxisome proliferator-induced hepatocarcinogenesis. Miller et al. (1996)
10 note that studies with MCP in Alpk:AP rats indicate that DNA synthesis occurs primarily in one
11 hepatocyte subpopulation as defined by ploidy status, the binucleated tetraploid (2 X 2N)
12 hepatocytes and that this preferential hepatocyte DNA synthesis is manifested by dramatic
13 alterations in hepatocyte ploidy subclasses, i.e., significant increases in mononucleate tetraploid
14 (4N) hepatocytes concomitant with decreases in 2 X 2N hepatocytes. They reported results in
15 male Fischer 344 rats were 13 weeks old (an agent in which polyploidization had reached a
16 plateau) exposed to 1,000 ppm WY-14,643 and MCP (gavage via corn oil at 8 mg/mL or
17 25 mg/kg MCP once daily) for 2, 5, and 10 days (n = 4). WY-14,643 and MCP were reported to
18 induce significant increases in the octoploid hepatocyte class that coincided with decreases in the
19 tetraploid hepatocyte class. However, MCP did not induce this shift until Day 5 of exposure.
20 These results show an approximate doubling of mononuclear octoploid (8N) hepatoctyes but still
21 a very low number of the total hepatocyte population that does not reach greater than 7% and is
22 still only approximately twice that of control values and thus, does not present itself with a very
23 large target population. There was no real effect on 4N hepatocytes due to these treatments and
24 the percent of hepatocytes that were 4N stayed -70% and were thus, the majority cell type in the
25 liver. Miller et al. (1996) note the importance of maturation and/or strain for these analyses there
26 are maturation-dependent differences in the distribution and mitogenic sensitivity of hepatoctyes
27 in the various subclasses.
28 Hasmall and Roberts (2000) note that despite their differing abilities to induced liver
29 cancer, both DCB (a nonhepatocarcinogen in Fischer 344 rats) and DEHP, at the doses and
30 routes used in the NTP bioassays, induced similar profiles of S-phase LI. A large and rapid peak
31 during the first 7 days (1,115 and 1,151 % of control for DEHP and DCB, respectively) was
32 followed by a return to control levels. They suggest that the size of the S-phase response does
33 not necessarily determine hepatocarcinogenic risk and that the subpopulation in which S-phase is
34 induced may be a better correlate with subsequent hepatocarcinogenecity. They compared the
35 effects on polyploidy/nuclearity and on the distribution of S-phase labeled cells with ETU, the
36 peroxisome proliferator MCP, and phenobarbitone. Male F334 rats 7-9 weeks old were exposed
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-320 DRAFT—DO NOT CITE OR QUOTE
-------
1 to MCP (0.1% in diet), ETU 83 ppm diet, phenobarbitone (500 mg/mL drinking water) for 7
2 days. The number of rats for 7 day study was not given by the authors. Hasmall and Roberts
3 (2000) reported that treatment of rats with MCP, ETU or phenobarbitone for 7 days had no
4 significant effect on the ploidy profile as compared with corn oil controls (data not shown) but
5 that MCP and phenobarbitone did induce significant changes in nuclearity. MCP reduced the
6 2 X 2N population and increased the 8N population. Phenobarbitone similarly increased the
7 proportion of cells in the 4N population. ETU had no effect on the nuclearity profile as
8 compared with control. However, what the authors describe for their results in polidy and
9 nuclearity are different than those presented in their figures. There were significant differences
10 between controls that the authors did not characterize and there appeared to be a greater
11 difference between controls than some of the treatments.
12 Gupta (2000) report that in transgenic mice with overexpression of TGF-a, liver-cell
13 turnover increases, along with the onset of hepatic polyploidy, whereas hepatocellular carcinoma
14 originating in these animals contain more diploid cells. They note that coexpression of c-Myc
15 and TGF-a transgenes in mouse hepatocytes was associated with greater degrees of polyploidy
16 as well as increased development of hepatocellular carcinoma. Gupta (2000) notes that in the
17 presence of ongoing liver injury and continuous depletion of parenchymal cells, hepatic
18 progenitor cells (including oval cells) are eventually activated but what roles polyploid cells play
19 in this process requires further study. In the working model by Gupta (2000), sustained disease
20 by chronic hepatitis, metabolic disease, toxins, etc., may lead to hepatocyte polyploidy and loss,
21 and the emergence of rapidly cycling progenitor or escape cell clones with the onset of liver
22 cancer.
23 Conner et al. (2003) describe the development of transgenic mouse models in which
24 E2F1 and/or c-Myc was overexpressed in mouse liver. The E2F1 and c-Myc transcription
25 factors are both involved in regulating key cellular activities including growth and death and,
26 when overexpressed, are capable of driving quiescent cells into S-phase in the absence of other
27 mitogenic stimuli and are potent inducers of apoptosis operating at least through one common
28 pathway involving p53. Deregulation of their expression is also frequently found in cancer cells
29 (Conner et al., 2003). Conner et al. (2003) reported that although both c-Myc and E2F1 mono-
30 transgenic mice were prone to liver cancer, E2F1 mice developed HCC more rapidly and with a
31 higher frequency and that the combined expression of these two transcription factors
32 dramatically accelerated HCC growth compared to either E2F1 or c-Myc mono-transgenic mice.
33 All three transgenic lines were reported to show a low but persistent elevation of hepatocyte
34 proliferation before an onset of tumor growth. Ploidy was shown to be affected differently by
35 c-Myc and E2F1, and suggested distinct differences by which these two transcription factors
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-321 DRAFT—DO NOT CITE OR QUOTE
-------
1 control liver proliferation/maturation. Both transgenic alterations induced liver cancer but had
2 differing effects on polyploidization suggestive that liver cancer can arise from either type of
3 mature hepatocyte.
4 c-Myc single-transgenic mouse showed a continuous high cell proliferation that preceded
5 the appearance of preneoplastic lesions, which was also true, although to a lesser extent, in the
6 E2F1 mouse. At 15 weeks of age, all of the transgenic mouse lines were reported to have a high
7 incidence (>60%) of hepatic dysplasia with mitotic indices equivalent in c-Myc/E2Fl, and c-
8 Myc livers, but 2-fold higher than the mitotic index in E2F1 and very low in wild-type mice.
9 Thus, the combination of the two transgenes did not have an additive effect on proliferation. An
10 analysis of the DNA content in hepatocyte nuclei isolated from 4- to 15-week old mice was
11 reported to show that in young wild-type livers, the majority of nuclei had a diploid DNA
12 content with a smaller proportion of tetraploid nuclei. As the mice aged, the number of
13 tetraploid and octoploid nuclei increased consistent with the previous findings of others.
14 However, c-Myc mice were reported to demonstrate a premature polyploidization with the
15 number of 2N nuclei in c-Myc livers almost 2-fold less, while the proportion of 4N nuclei
16 increased more than 2.5-fold at 4 weeks of age. The most prominent ploidy alteration was an
17 increase in the fraction of hepatocytes with octaploid nuclei (~200-fold higher). The percentage
18 of polyploidy cells was reported to continue to rise in 15 week old c-Myc livers. The majority of
19 hepatocytes had nuclei with 4N and 8N DNA content, with an attendant increase in binucleated
20 hepatocytes and increase in average cell size. In striking contrast, E2F1 hepatocytes were
21 reported not to undergo normal polyploidization with aging. The majority of E2F1 nuclei were
22 reported to remain in the diploid state and to be almost identical in E2F1 mice at 4 and 15 weeks
23 of age. The percentage of binucleated hepatocytes was also reduced. In c-Myc/E2Fl mice, the
24 age-related changes in ploidy distribution were reported to resemble those found in both c-Myc
25 and in E2F1 single transgenic mice. At a young age, c-Myc/E2Fl mice, similar to E2F1 mice,
26 were reported to retain significantly more diploid nuclei than c-Myc mice. However, as mice
27 aged, the majority of c-Myc/E2Fl hepatocytes, similar to c-Myc cells but in contrast to findings
28 in E2F1 cells, became polyploid. Consistent with a more progressive polyploidization, the DNA
29 content was significantly higher in both c-Myc/E2Fl and c-Myc livers. Conner et al. (2003)
30 report that other known modulators of ploidy in the liver are the tumor suppressor p53, pRb, and
31 the cell cycle inhibitor p21 as well as, genes involved in the control of the cell cycle progression
32 such as cyclin A, cyclin B, cyclin D3, and cyclin E.
33 Along with increased liver cancer, Conner et al. (2003) note that the C-Myc mice also
34 experienced a persistent liver injury as evidenced by significant elevation of circulating levels of
35 aspartate aminotransferase, alanine aminotransferase, and alkaline phosphatase along with the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-322 DRAFT—DO NOT CITE OR QUOTE
-------
1 appearance of a frequent oval/ductular proliferation. However, oval cell proliferation may be a
2 marker of hepatocyte damage but not be the cells responsible for tumor induction (Tarsetti et al.,
3 1993). Conner et al. (2000) report that if E2F1 is overexpressed in the liver, there is both
4 oncogenic and tumor-suppressive properties. In regard to liver morphological changes, E2F1
5 transgenic mice were reported to uniformly develop pericentral dysplasia and foci adjacent to
6 portal tracts followed by the abrupt appearance of adenomas and subsequent malignant
7 conversion with all of the animals having foci by 2-4 months and by 8-10 months most having
8 adenomas with dysplastic changes remaining confined to the pericentral regions of the liver
9 lobule. In regard to phenotype, the majority of the foci were composed of small round cells, with
10 clear-cell phenotype but eosinophilic, mixed, and basophilic foci were also seen. In adenomas
11 with malignant transformation to HCC, there appeared to be high mitotic indices, blood vessel
12 invasion, and central collection of deeply basophilic cells with large nuclei giving a "nodule- in-
13 nodule" appearance. Macrovesicular hepatic steatosis was first noted in some E2F1 transgenic
14 livers at 6-8 months and by 10-12 months 60% of animals had developed prominent fatty
15 change. Hepatic steatosis has been noted in several transgenic mouse models of liver
16 carcinogenesis (Conner et al., 2000). These results raise interesting points of regional difference
17 in tumor formation which can be lost in analyses using whole liver and that the phenotype of foci
18 and tumors are similar to those seen from chemical carcinogenesis. The occurrence of
19 hepatotoxicity in these transgenic mice is also of note.
20
21 E.3.3.2. Hepatocellular Proliferation and Increased DNA Synthesis
22 Caldwell et al. (2008b) have presented a discussion of the role of proliferation in cancer
23 induction. They state that
24
25 in the case of CC14 exposure, hepatocyte proliferation may be related to its ability
26 to induce liver cancer at necrogenic exposure levels, but the nature of this
27 proliferation is fundamentally different from peroxisome proliferators or other
28 primary mitogens that cause hepatocyte proliferation without causing cell death
29 (Coni et al., 1993; Ledda-Columbano et al., 1993, 1998, 2003; Menegazzi et al.,
30 1997; Columbano and Ledda-Columbano, 2003). After initiation with a
31 mutagenic agent, the transient proliferation induced by primary mitogens has not
32 been shown to lead to cancer-induction, while partial hepatectomy or necrogenic
33 treatments of CC14 result in the development of tumors [Ledda-Columbano et al.,
34 1993; Gelderblom et al., 2001].
35
36 Roskams et al. (2003) notes that partial hepatectomy does not cause hepatocellular carcinoma in
37 normal mice without initiation. Melchiorri et al. (1993) report that a series of studies has shown
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-323 DRAFT—DO NOT CITE OR QUOTE
-------
1 that acute proliferative stimuli provided by primary mitogens, unlike those of the regenerative
2 type such as those elicited by surgical or chemical partial hepatectomy, do not support the
3 initiation phase and do not effectively promote the growth of initiated cells (Columbano et al.,
4 1990; Columbano et al., 1987; Ledda-Columbano et al., 1989). They note that, the finding that
5 most of these chemicals, with the exception of WY, induce only a very transient increase in cell
6 proliferation raises the question whether such a transient induction of liver cell proliferation
7 might be related to liver cancer appearing 1-2 years later. They note that mitogen-induced liver
8 growth differs from compensatory regeneration in several aspects (1) it does not require an
9 increased expression of hepatocyte growth factor mRNA in the liver (2) it is not necessarily
10 associated with an immediate early genes such as c-fos and c-jun; (3) it results in an excess of
11 tissue and hepatic DNA content that is rapidly eliminated by apoptotic cell death following
12 withdrawals of the stimulus.
13 Other studies have questioned the importance of a brief wave of DNA synthesis in
14 induction of liver cancer. Chen et al. (1995) note that Jirtle et al. (1991) and Schulte-Hermann et
15 al. (1986) reported that during a 2-week period of treatment with lead, DNA synthesis was
16 increased most in centralobular hepatocytes and that the predominantly centrilobular distribution
17 of the labeled nuclei may have been due largely to the brief wave of mitogenic response, because
18 from the fifth day onward DNA synthesis activity returned to control level even though lead
19 nitrate treatment continued. They concluded that sustained cell proliferation may be more
20 important than a brief wave of increased DNA synthesis. Chen et al. (1995) also noted that a
21 number of different agents acting via differing MO As will induce periportal proliferation.
22 Vickers and Lucier (1996) reported that mitogenic response induced by acute 17
23 a-ethinylestradiol administration is randomly distributed throughout the hepatic lobule, while
24 continuous administration increases the proportion of diploid cells. Richardson et al. (1986)
25 reported that the lobular distribution of the correlation of hepatocyte initiation and akylation
26 reported in their model of carcinogenicity did "not support that early proliferation is associated
27 with cancer as at 7 days there is a transient increase in the lobes least likely to get a tumor and no
28 difference between the lobes at 14 and 28 days DEN although there is a difference in tumor
29 formation between the lobes." Cells undergoing DNA synthesis may not be in the same zone of
30 the liver where other hypothesized "key events" take place.
31 Tanaka et al. (1992) note that the distribution of hepatocyte proliferation in the periportal
32 area was in contrast to the distribution of peroxisome proliferation in the centrilobular area of
33 Clofibrate treated rats. Melnick et al. (1996) note that replicative DNA synthesis commonly has
34 been evaluated by measurement of the fraction of cells incorporating BrdU or tritiated thymidine
35 into DNA during S-phase of the cell cycle (S-phase labeling index), but that the S-phase labeling
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-324 DRAFT—DO NOT CITE OR QUOTE
-------
1 index would not be identical to the cell division rate when replication of DNA does not progress
2 to formation of two viable daughter cells. "The general view at an international symposium on
3 cell proliferations and chemical carcinogenesis was that although cell replication is involved
4 inextricably in the development of cancers, chemically enhanced cell division does not reliably
5 predict carcinogenicity (Melnick et al ,1993)." They note that the finding that enzyme-altered
6 hepatic foci were not induced in rats fed WY-14,643 for 3 weeks followed by partial
7 hepatectomy indicates that early high levels of replicative DNA synthesis and peroxisome
8 proliferation are not sufficient activities for initiation of hepatocarcinogenesis. Baker et al.
9 (2004) reported that, similar to the pattern of transient increases in DNA synthesis reported for
10 TCE metabolites, Clofibrate exposure induced the upregulation of a variety of cell proliferation-
11 associated genes (e.g., G2/M specific cyclin Bl, cyclin-dependent kinase 1, DNA topoisomerase
12 II alpha, c-myc protooncogene, pololike serien-threonine protein kinase, and cell divisions
13 control protein 20) began on or before Day 1 and peaked at some point between days 3 and 7.
14 By Day 7, cell proliferation genes were down regulated. The chronology of this gene expression
15 agrees with the histologic diagnosis of mitotic figures in the tissue, where an increase in mitotic
16 figures was detected in the Day 1 and most notably Day 3 high and low-dose groups. However,
17 by Day 7, the incidence of mitotic figures had decreased. The clustering of genes associated
18 with the G2/M transition point suggests that in the rats, the polyploid cells arrested at G2/M are
19 those that are proceeding through the cell cycle.
20 A dose-response for increased DNA-synthesis also seems to be lacking for the model
21 PPARa agonist, WY-14,643 suggesting that the transient increases in DNA synthesis reported by
22 Eacho et al. (1991) for this compound at lower levels that then increase later at necrogenic
23 exposure levels, are not related to its carcinogenic potential. Wada et al. (1992) reported that in
24 male Fischer 344 rats exposed to a range of WY-14,643 concentrations (5-1,000 ppm) that liver
25 weight gain occurred at the lowest dose that gave a sustained response for many weeks but gave
26 increased cell labeling only in the first week. Peroxisomes proliferation, as measure by electron
27 microscopy, increases started at 50 ppm exposures. By enzymatic means, peroxisomal activities
28 were elevated at the 5 ppm dose. Of note is the reported difference in distribution in
29 hepatocellular proliferation, which was not where the hypertrophy or where the lipofuscin
30 increases were observed. The authors note that these data suggest that 50 and 1,000 ppm WY-
31 14,643 should give the same carcinogenicity if peroxisome proliferation or sustained
32 proliferation are the "key events." The study of Marsman et al. (1992) is very important in that it
33 not only shows that clofibric acid (another PPAR a agonist) does not have sustained
34 proliferation, but it also shows that it and WY-14,643 at 50 ppm did not induce apoptosis in rats.
35 It is probable that use of WY-14,643 at high concentrations may induce apoptosis in a manner
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-325 DRAFT—DO NOT CITE OR QUOTE
-------
1 not applicable to other peroxisome proliferators or to treatment with WY-14,643 at 50 ppm.
2 This study also confirmed that exposure to WY-14,643 at 50 ppm and WY-14,643 at 1,000 ppm
3 induces similar effects in regards to hepatocyte proliferation and peroxisomal proliferation.
4 The study by Eacho et al. (1991) also gives a reference point for the degree of
5 hepatocytes undergoing transient DNA synthesis from WY-14,643 and Clofibrate and how much
6 smaller it is for TCE and its metabolites, which generally involve less than 1% of hepatocytes.
7
8 The labeling index of BrdU was 7.2% on day 3 and 15.5% on day 6 after clofibric
9 acid but by day 10 and 30 labeling index was the same as controls at -1-2%... .For
10 WY the labeling index was 34.1% at day 3 and 18.6% at day 6. At day 10 the
11 labeling index was 3.3% and at day 30 was 6%, representing 6.6- and 15-fold of
12 respective controls. Control levels were -0.5 to 1%.... The labeling index was
13 increased to 32% by 0.3% LY171883 and to 52% by 0.05% Nafenopin. The
14 0.005% and 0.1% dietary doses of WY increased the 7 day labeling index to a
15 comparable level (55% - 58%).
16
17 Yeldani et al. (1989) report results showing that until foci appear, cell proliferation has
18 ceased to increase over controls after the first week for ciprofibrate-induced
19 hepatocarcinogenesis. The results also show the importance of using age matched controls and
20 not pooled controls for comparative purposes of proliferation as well as how low proliferative
21 rates are in control animals. The results of Barass et al. (1993) are important in suggesting that
22 age of animals is important when doing quantitation of labeling indexes. Studies such as that
23 conducted by Pogribny et al. (2007) that only give the replication rate as a ratio to control will
24 make the proliferation levels look progressive when in fact they are more stable with time as it is
25 just the controls that change with age as a comparison point.
26
27 E.3.3.3. Nonparenchymal Cell Involvement in Disease States Including Cancer
28 The recognition that not only parenchymal cells but also nonparenchymal cells play a
29 role in HCC has resulted in studies of their role in initiation as well as progression of neoplasia.
30 The role of the endothelial cell in controlling angiogenesis, a prerequisite for neoplastic
31 progression, and the role of the Kupffer cell and its regulation of the cytokine milieu that
32 controls many hepatocyte functions and responses have been reported. However, as pointed out
33 by Pikarsky et al. (2004) and by the review by Nickoloff et al. (2005) the roles of inflammatory
34 cytokines in cancer are context and timing specific and not simple. For TCE, nonparenchymal
35 cell proliferation has been observed after inhalation (Kjellstrand et al., 1983b) and gavage
36 (Goel et al., 1992) exposures of-4 weeks duration.
37
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-326 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.3.3.3.1. Epithelial cell control of liver size and cancer—angiogenesis.
2 The epithelium is key in controlling restoration after partial hepatectomy and not
3 surprisingly HCC growth. Greene et al. (2003) hypothesized that the control of physiologic
4 organ mass was similar to the control of tumor mass in the liver and that specifically, the
5 proliferation of hepatocytes after partial hepatectomy, like the proliferations of neoplastic cells in
6 tumors, requires the synthesis of new blood vessels to support the rapidly increasing mass. They
7 report that a peak in hepatocyte production of vascular endothelial growth factor (VEGF), an
8 endothelial mitogen, corresponds to an increase of VEGF receptor expression on endothelial
9 cells after partial hepatectomy and the rate of endothelial proliferation.. Fibroblast growth factor
10 and transforming growth factor-alpha (TGfox), which stimulate endothelial cells, are secreted by
11 hepatoctyes 24 hours after partial hepatectomy. However, endothelial cells were reported to
12 secrete hepatocyte growth factor, a potent hepatocyte mitogen, that is also proangiogenic. The
13 secretion of transforming growth factor -beta by (TGfox) endothelial cells 72 hours after partial
14 hepatectomy was reported to inhibit hepatocyte proliferation. Thus, Greene et al. (2003)
15 suggested that endothelial cells and hepatocytes of the regenerating liver influence each other,
16 and both populations are required for the regulation of the regenerative process.
17
18 E.3.3.3.2. Kupffer cell control of proliferation and cell signals, role in early and late effects
19 Vickers and Lucier (1996) have reported that Kupffer cells are increased in number in
20 prenoplastic foci but are decreased in hepatocellular carcinoma, and that other studies have
21 demonstrated that both sinusoidal endothelial cells and Kupffer cells within hepatocellular
22 carcinoma cells in humans stain positive for mitotic activity although the number of
23 nonparenchymal cells compared to parenchymal cells may be reduced. Lapis et al. (1995)
24 reported that Kupffer cells contain lysozyme in their cytoplasmic granules, vacuoles and
25 phagosomes, some cells show a positive reaction in the rough endoplasmic reticulum,
26 perinuclear cisternae and the Golgi zone, and that in human monocytes the lysozyme is
27 colocalized with the CD68 antigen and myeloperoxidase. They also report that, in rodent
28 hepatocarcinogenesis, increased numbers of Kupffer cells were observed in preneoplastic foci,
29 whereas abnormally low numbers were present following progression to hepatocellular
30 carcinoma. They also note that "the Kupffer cell count in human HCC has also been shown to
31 be very low and varies with different histological form." They reported that for monkey HCCs,
32 that the proportion of endothelial elements remained constant (the parenchymal/endothelial cell
33 ratio), however, there was a striking reduction in the areas occupied by Kupffer cells. While
34 healthy control livers contained the highest number of Kupffer cells, in the tumor-bearing cases
35 the nonneoplastic, noncirrhotic liver adjacent to the HCC nodules had a significantly lower
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-327 DRAFT—DO NOT CITE OR QUOTE
-------
1 number of Kupffer cells and the number decreased further in the nonneoplastic portions of
2 cirrhotic livers. Within HCC nodules the Kupffer cell count was greatly reduced with no
3 significant changes were observed between the cirrhotic areas and the carcinomas, however, the
4 tumors contained fewer lysozyme and CD68 positive cells. Lapis et al. (1995) note that
5
6 since other cell types within the liver sinusoids (monocytes and polypmorphs) and
7 portal macrophage were also positive, it was important to identify the star-like
8 morphology of the Kupffer cells. The results of the two independent observers
9 assessment of the morphology and enumeration of Kupffer cells were quite
10 consistent and differed by only 3%." "The loss of Kupffer cells in the HCC may
11 possibly result from capillarization of the sinusoids, which has been observed
12 during the process of liver cirrhosis and carcinogenesis. Capillarization entails the
13 sinusoidal lining endothelial cells losing their fenestrations.
14
15 E.3.3.3.3. Nf-kB and TNF-a - context, timing and source of cell signaling molecules
16 A large body of literature has been devoted to the study of nuclear factor K B for its role
17 not only in inflammation and a large number of other processes, but also for its role in
18 carcinogenesis. However, the effects of these cytokines are very much dependent on their
19 cellular context and the timing of their modulation. As described by Adli and Baldwin (2006),
20
21 The classic form of NF-kB is composed of a heterodimer of the p50 and p65
22 subunits, which is preferentially localized in the cytoplasm as an inactive complex
23 with inhibitor proteins of the IkB family. Following exposure to a variety of
24 stimuli, including inflammatory cytokines and LPS, IkBs are phosphorylated by
25 the IKKa/P complexes then accumulate in the nucleus, where they
26 transcriptionally regulate the expression of genes involved in immune and
27 inflammatory responses.
28
29 The five members of the mammalian NF-kB family, p65 (RelA), RelB, c-Rel, P50/pl05
30 (NF-KB 1) and p52/plOO (NF-kB2), exist in unstimulated cells as homo- or heterodimers bound
31 to IkB family proteins. Transcriptional specificity is partially regulated by the ability of specific
32 NF-kB dimmers to preferentially associate with certain members of the IkB family. Individual
33 NF-kB responses can be characterized as consisting of waves of activation and inactivation of
34 the various NF-kB members (Hayden and Ghosh, 2004). While the function of NF-kB in many
35 contexts have been established, it is also clear that there is great diversity in the effects and
36 consequences of NF-kB activation with NF-kB subunits not necessarily regulating the same
37 genes in an identical manner and in all of the different circumstances in which they are induced.
38 The context within which NF-kB is activated, be it the cell type or the other stimuli to which the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-328 DRAFT—DO NOT CITE OR QUOTE
-------
1 cell is exposed, is therefore, a critical determinant of the NF-kB behavior (Perkins and Gilmore,
2 2006).
3 Balkwill et al. (2005) report that
4
5 the NF-KB pathway has dual actions in tumor promotion: first by preventing cell
6 death of cells with malignant potential, and second by stimulating production of
7 proinflammatory cytokines in cells of infiltrating myeloid and lymphoid cells.
8 The proinflammatory cytokines signal to initiated and/or otherwise damaged
9 epithelial cells to promote neoplastic cell proliferation and enhance cell survival.
10 However, the tumor promoting role of NF-KB may not always predominate. In
11 some cases, especially early cancers, activation of this pathway may be tumor
12 suppressive (Perkins, 2004). Inhibiting NF-KB in keratinocytes promotes
13 squamous cell carcinogenesis by reducing growth arrest and terminal
14 differentiation of initiated keratinocytes (Seitz et al., 1998).
15
16 Other inflammatory mediators have also been associated with oncogenesis. Balkwill et al.
17 (2005) reported that TNFa is frequently detected in human cancers (produced by epithelial tumor
18 cells, as in for instance, ovarian and renal cancer) or stromal cells (as in breast cancer). They
19 also report that the loss of hormonal regulation of IL-6 is implicated in the pathogenesis of
20 several chronic diseases, including B cell malignancies, renal cell carcinoma, and prostate,
21 breast, lung, colon, and ovarian cancers. Over 100 agents, such as antioxidants, proteosome
22 inhibitors, NSAIDs, and immunosuppressive agents are NF-KB inhibitors with none being
23 entirely specific (Balkwill et al., 2005). Thus, alterations in these cytokines, and the cells that
24 produce them, are implicated as features of "cancer" rather than specific to HCC.
25 Balkwill et al. (2005) report that
26
27 Two mouse models of inflammation-associated cancer now implicate the gene
28 transcription factor NF-KB and the inflammatory mediator known as tumor-
29 necrosis factor a (TNF- a) in cancer progression. Using a mouse model of
30 inflammatory hepatitis that predisposes mice to liver cancers, Pikarsky et al.
31 present evidence that the survival of hepatocytes - liver cells - and their
32 progression to malignancy are regulated by NF-KB. NF-KB is an important
33 transcription factor that controls cell survival by regulating programmed cell
34 death, proliferation, and growth arrest. Pikarsky et al. find that the activation state
35 of NF-KB, and its localization in the cell, can be controlled by TNF-a produced by
36 neighboring inflammatory cells (collectively known as stromal cells).
37
38 Pikarsky et al. (2004) reported that that the inflammatory process triggers hepatocyte NF-KB
39 through upregulation of TNF-a in adjacent endothelial and inflammatory cells. Switching off
40 NF-KB in mice from birth to seven months of age, using hepatocyte-specific inducible IicB-super
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-329 DRAFT—DO NOT CITE OR QUOTE
-------
1 represser transgene, had no effect on the course of hepatitis, nor did it affect early phases of
2 hepatocyte transformation. By contrast, suppressing NF-KB inhibition through anti-TNF-a
3 treatment or induction of the IicB-super represser in later stages of tumor development resulted in
4 apoptosis of transformed hepatocytes and failure to progress to hepatocellular carcinoma. The
5 Mdr2 knockout hepatocytes in Pikarsky's model of hepatocarcinogenicity were distinguishable
6 from wild-type cells by several abnormal features; high proliferation rate, accelerated
7 hyperploidy and dysplasia. Pikarsky et al. (2004) reported that NF-KB knockout and double
8 mutant mice displayed comparable degrees of proliferation, hyperploidy and dysplasia implying
9 that NF-KB is not required for early neoplastic events. Thus, activation of NF-KB was not
10 important in the early stages of tumor development, but was crucial for malignant conversion.
11
12 Greten et al reporting in Cell, come to a similar conclusion by studying a mouse
13 colitis-associated cancer model. Their work does not directly implicate TNF-a,
14 but instead found enhanced production of several pro-inflammatory mediators
15 (cytokines) including TNF-a,, in the tumor microenvironment during the
16 development of cancer. An important feature of both studies is that NF-KB
17 activation was selectively ablated in different cell compartments in developing
18 tumor masses, and at different stages of cancer development.
19
20 Balkwill et al. (2005) also note that TNF-a and NF-KB have many different effects, depending on
21 the context in which they are called into play and the cell type and environment.
22 In contrast, El-Serag and Rudolph (2007) note that "the influence of inflammatory
23 signaling on hepatocarcinogenesis can be context dependent; deletion of Nf-KB-dependent
24 inflammatory responses enhanced HCC formation in carcinogen treated mice (Sakurai et al.,
25 2006)." Similarly, deletion of Nf-KB essential modulator/I kappa P kinase (NEMO/IKK), an
26 activator of Nf-KB, induced steatohepatitis and HCC in mice (Luedde et al., 2007). Maeda et al.
27 (2005) reported that hepatocyte specific deletion of IKKP (which prevents NF-kB activation)
28 increased DEN-induced hepatocarcinogenesis and that a deletion of IKKP in both hepatocytes
29 and hematopoietic-derived cells, however, had the opposite effect, decreasing compensatory
30 proliferation and carcinogenesis. They suggest that these results, differ from previous suggestion
31 that the tumor-promoting function of NF-kB is excreted in hepatocytes (Pikarsky et al., 2004),
32 and suggest that chemicals or viruses that interfere with NF-kB activation in hepatocytes may
33 promote HCC development.
34 Alterations in NF-kB levels have been suggested as a key event for the
35 hepatocarcinogenicity by PPARa agonists. The event associated with PPAR effects has been
36 the extent of NF-kB activation as determined through DNA binding. As reported by Tharappel
37 et al. (2001), NF-kB activity is assayed with electrophoretic modibility shift assay with nuclear
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-330 DRAFT—DO NOT CITE OR QUOTE
-------
1 extracts prepared from frozen liver tissue as a measure of DNA binding of NF-kB. Increase
2 transcription of downstream targets of NF-kB activity have also been measured. It has been
3 suggested that PPARa may act as a protective mechanism against liver toxicity. Ito et al. (2007)
4 cite repression of NF-kB by PPARa to be the rationale for their hypothesis that PPARa-null
5 mice may be more vulnerable to tumorigenesis induced by exposure to environmental
6 carcinogens. However, as shown in Section E.3.4.1.2, although DEHP was reported to also
7 induce glomerularnephritis more often in PPARa-null mice, as suggested Kamijo et al. (2007) to
8 be due of the absence of PPARa- dependent anti-inflammatory effect of antagonizing the
9 oxidative stress and NF-KB pathway, there was no greater or lesser susceptibility to DEHP-
10 induced liver carcinogenicity in the PPARa null mice.
11 Because PPARa is known to exert anti-inflammatory effects by inducing expression of
12 iKBa, which antagonizes NFicB signaling, the expression of IicBa has been measured in some
13 studies (Kamijo et al., 2007) as well as expression of TNR1 mRNA to evaluate the sensitivity to
14 the inflammatory response. Ito et al. (2007) report that in wild-type mice there did not appear to
15 be a difference between controls and DEHP treatment for p65 immunoblot results. DEHP
16 treatment was also reported to not induce p65 or p52 mRNA either or influence the expression
17 levels of TNFa, IkBa, IkBp and IL-6 mRNA in wild-type mice. Tharappel et al. (2001) treated
18 rats with WY-14,643, gemfibrozil or Dibutyl phthalate and reported elevated NF-kB DNA
19 binding in rats with WY-14,642 to have sustained response but not others. WY-14,643 increased
20 DNA binding activity of NF-kB at 6, 34 or 90 days. Gemfibrozil and DEHP increased NF-kB
21 activity to a lesser extent and not at all times in rats. For gemfibrozil, there was only a 2-fold
22 increase in binding at 6 days with no increase at 34 days and increase only in low dose at 90
23 days. In rats treated with Dibutyl phthalate, there no change at 6 days, at 34 days there was an
24 increase at high and low dose, at 90 days only low dose animals showed a change. In pooled
25 tissue from WY-14,643- treated animals, the complex that bound the radiolabeled NF-kB
26 fragment did contain both p50 and p65. Both WY-14,643 and gemfibrozil were reported to
27 produce tumors in rats with Dibutyl pthalate untested in rats for carcinogenicity. Thus, early
28 changes in NF-kB were not supported as a key event and WY-14,643 to have a pattern that
29 differed from the other PPARa agonists examined.
30 In regard to the links between inflammation and cancer, Nickoloff et al. (2005) in their
31 review of the issue, caution that such a link is not simple. They note that
32
33 dissecting the mediators of inflammation in cutaneous carcinogenic pathways has
34 revealed key roles for prostaglandins, cyclooxygenase-2, tumor necrosis factor-a,
35 AP-1, NF-KB, signal transducer and activator of transcription (STAT)3, and
36 others. Several clinical conditions associated with inflammation appear to
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-331 DRAFT—DO NOT CITE OR QUOTE
-------
1 predispose patients to increased susceptibility for skin cancer including discoid
2 lupus erythematosus, dystrophic epidermolysis bullosa, and chronic wound sites.
3 Despite this vast collection of data and clinical observations, however, there are
4 several dermatological setting associated with inflammation that do not
5 predispose to conversion to lesions into malaignancies such as psoriasis, atopic
6 dermatitis, and Darier's disease.
7
8 Nickoloff et al. (2005) suggest that such a
9
10 link may not be as simple as currently portrayed because certain types of
11 inflammatory processes in skin (and possibly other tissues as well) may also serve
12 a tumor suppressor function. Over the past few months, several publications in
13 leading biomedical journals grappled with an important issue in oncology, namely
14 defining potential links between chronic tissue damage, inflammation, and the
15 development of cancer. Balkwill and Coussens (2004) reviewed the role of the
16 NF-KB signal transduction pathway that can regulate inflammation and also
17 promote malignancy. Their review summarized the latest findings revealed in a
18 letter to Nature by Pikarsky et al. (2004). Using Mdr2 knockout mice in which
19 hepatitis is followed by hepatocellular carcinoma, Pikarsky et al. implicated
20 TNFa upregulation in tumor promotion of HCC, and suggest that TNFa and NF-
21 KB are potential targets for cancer prevention in the context of chronic
22 inflammation. A similar conclusion was reached with respect to NF-KB by an
23 independent group of investigators using a model of experimental dextran sulfate-
24 induced colitis, in which inactivation of the 1KB kinase resulted in reduced
25 colorectal tumors (Greten et al ., 2004). Although there are many other clinical
26 condition supporting the concept of inflammation is a critical component of tumor
27 progression (e.g., reflux esophagitis/esophageal cancer; inflammatory bowel
28 disease/colorectal cancer), there is at least one notable example that does not fit
29 this paradigm. As described below, psoriasis is a chronic cutaneous inflammatory
30 disease, which is seldom if ever accompanied by cancer suggesting the
31 relationship between tissue repair, inflammation, and development may not be as
32 simple as portrayed by the aforementioned reviews and experimental results.
33 Besides psoriasis, other noteworthy observations pointing to more complexity
34 include the observation that in the Mdr2 knockout mice, we rarely detect bile duct
35 tumors despite extensive inflammation, NF-KB activation, and abundant
36 proliferation of bile ducts in portal spaces (Pikarsky et al., 2004). Moreover, in a
37 skin-cancer mouse model, NF-KB was shown to inhibit tumor formation (Dajee et
38 al., 2003). Thus, the composition of inflammatory mediators, or the properties of
39 the responding epithelial cells (e.g., signaling machinery, metabolic status), may
40 dictate either tumor promotion or tumor suppression. Chronic inflammation and
41 tissue repair can trigger pro-oncogenic events, but also that tumor suppressor
42 pathways may be upregulated at various sites of injury and chronic cytokine
43 networking.
44 One cannot easily dismiss the many dilemmas raised by the psoriatic
45 plaque that confound a simple link between the tissue repair, inflammation, and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-332 DRAFT—DO NOT CITE OR QUOTE
-------
1 carcinogenesis. Since it is easily visible to the naked eye, and patients may suffer
2 from such lesions for decades, it is difficult to argue that various skin cancers
3 such as squamous cell carcinoma, basal cell carcinoma, or melanoma actually do
4 develop within plaques by are being overlooked by patients and dermatologists.
5 Remarkably, psoriatic plaques are intentionally exposed to mutagenic agents
6 including excessive sunlight, topical administration of crude coal tar, or parenteral
7 DNA cross-linking agent -psoralen followed by ultraviolet light. Moreover these
8 treatments are known to induce skin cancer in nonlesional skin. Thus since
9 psoriatic skin is characterized by altered differentiation, angiogenesis, increased
10 telomerase activity, proliferative changes, and apoptosis resistance, one would
11 expect that each and every psoriatic plaque would be converted to cancer, or at
12 least serve as fertile soil for the presence of non-epithelial skin cancers over
13 time... .In conclusion, it would seem prudent to remember the paradigm proposed
14 by Weiss (1971) in which he suggested that premalignant cells do not comprise an
15 isolated island, but are a focus of intense tissue interactions. The myriad
16 inflammatory effects of the tumor microenvironment are important for
17 understanding tumor development, as well as tumor suppression and senescence,
18 and for the design for efficacious prevention strategies against inflammation-
19 associate cancer (Nickoloff et al., 2005).
20
21 E.3.3.4. Gender Influences on Susceptibility
22 As discussed previously, male humans and rodents are generally more likely to get HCC.
23 The increased risk of liver tumors from estrogen supplements in women has been documented.
24 In mice male TCE exposure has been shown to have greater variability in response and greater
25 effects on body weight in males (Kjellstrand et al., 1983a, b) but to also induce dose-related
26 increases in liver weight and carcinogenic response in female mice as well as males (see
27 Section E.2.3.3.2). Recent studies have attempted to link differences in inflammatory cytokines
28 and gender differences in susceptibility.
29 Lawrence et al. (2007) suggest that
30
31 studies of Naugler et al. (2007) and Rakoff-Nahoum and Medzhitov (2007),
32 advance our understanding of the mechanisms of cancer-related inflammation.
33 They describe an important role for an intracellular signaling protein called
34 MyD88 in the development of experimental liver and colon cancers in mice.
35 MyD88 function has been well characterized in the innate immune response
36 (Akira and Takeda, 2004), relaying signals elicited by pathogen-associated
37 molecules and by the inflammatory cytokine interleukin-1 (IL-1).... The
38 conclusion from Naugler et al. (2007) and Rakoff-Nahoun and Medzhitov is that
39 MyD88 may function upstream of NF-KB in cells involved in inflammation-
40 associated cancer. Immune cells infiltrate the microenvironment of a tumor.
41 Naugler et al. (2007) and Rakoff-Nahoun and Medzhitov (2007) suggest that the
42 development of liver and intestinal cancers in mice may depend on a signaling
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-333 DRAFT—DO NOT CITE OR QUOTE
-------
1 pathway in infiltrating immune cells that involved the protein MyD88, the
2 transcription factor NF-KB, and the pro-inflammatory cytokine 11-6. TLR binds a
3 ligand which acts on MyD88 which acts on NF-KB which leads to secretion of
4 inflammatory cytokine IL-6 which leads to promotion of tumor cell survival and
5 proliferation.
6
7 Naugler et al. (2007) suggested gender disparity in MyD88-dependent IL-6 production
8 was linked to differences in cancer susceptibility using the DEN model (a mutagen with
9 concurrent regenerative proliferation at a single high dose) with a single injection of DEN.
10 Partial hepatectomy was reported to induce no gender-related difference in IL-6 increase. After
11 DEN treatment the male mouse had 275 ng/mL as the peak IL-6 levels 12 hours after DEN and
12 for female mice the peak was reported to be 100 ng/mL 12 hours after DEN administration. This
13 is only about a 2.5-fold difference between genders. 11-6 mRNA induction was reported for mice
14 4 hours after DEN while at 4 hours, at a time when there was no difference in serum IL-6
15 between male and female mice. It was not established that the 4-hour results in mRNA
16 translated to the differences in serum at 12 hour between the sexes. The magnitude of mRNA
17 differences does not necessarily hold the same relationship as the magnitude in serum protein. In
18 fact, there was not a linear correlation between mRNA induction and IL-6 serum levels.
19 A number of issues complicate the interpretation of the results of the study. The study
20 examined an acute response for the chronic endpoint of cancer and may not explain the
21 differences in gender susceptibility for agents that do not cause necrosis. The DEN was
22 administered in 15-day old mice (which had not reached sexual maturity) for tumor information
23 at a much lower dose than used in short-term studies of inflammation and liver injury in which
24 mature mice were used. If large elevations of IL-6 are the reason for liver cancer, why does not
25 a partial hepatectomy induce liver cancer in itself? The percentage of proliferation at 36 and 48
26 hours after partial hepatectomy was the same between the sexes. If a 2.5-fold difference in IL-6
27 confers gender susceptibility, it should do so after partial hepatectomy and lead to cancer. For
28 female mice, partial hepatectomy showed alterations in a number of parameters. However,
29 partial hepatectomy does not cause cancer alone. The 5-fold increase 4 hours after DEN
30 induction of IL-6 mRNA in male mice is in sharp contrast to the 27-fold induction of IL-6 1 hour
31 after partial hepatectomy (in which at 4 hours the IL-6 had diminished to 6-fold). There
32 appeared to be variability between experiments. For example, the difference in males between
33 experiments appears to be the same magnitude as the difference between male and female in one
34 experiment and the baseline of IL-6 mRNA induction appeared to be highly variable between
35 experiments as well as absolute units of ALT in serum 24 and 48 hours after DEN treatment that
36 tended to be greater that the effects of treatments. The experiments used very few animals
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-334 DRAFT—DO NOT CITE OR QUOTE
-------
1 (n = 3) for most treatment groups. Of note is that the MyD88 -/- male mice still had a
2 background level of necrosis similar to that of WT mice at 48 hours after DEN treatment, a time,
3 long after the peak of IL-6 mRNA induction and IL-6 serum levels were reported to have
4 peaked. One of the key issues regarding this study is whether difference in IL-6 reported here
5 lead to an increase proliferation and does that difference within 48 hours of a necrotizing dose of
6 a carcinogen change the susceptibility to cancer? This report shows that male and female mice
7 have a difference in necrosis after CCL4 and a difference in proliferation. Are early differences
8 in IL-6 at 4 hours related to the same kind of stimulus that leads to necrosis and concurrent
9 proliferation? The amount of proliferation (as measured by DNA synthesis) between male and
10 female mice 48 hours after DEN was very small and the study was conducted in a very few mice
11 (n = 3). At 36 hours the degree of proliferation was almost the same between the genders and
12 about 0.6% of cells. The baseline of proliferation also differed between genders but the variation
13 and small number of animals made it insignificant statistically. At 48 hours the differences in
14 proliferation between male and female mouse were more pronounced but still quite low (2% for
15 males and -1% for females). Is the change in proliferation just a change in damage by the agent?
16 Given the large variation in serum ALT and by inference necrosis, is there an equal amount of
17 variability in proliferation? This study gives only limited information for DEN treatment.
18 The difference in incidence of HCC was reported to be greater than that of "proliferation"
19 between genders and of other parameters although differences in tumor multiplicity or size
20 between the genders are never given in the paper. Most importantly, comparisons between the
21 short-term changes in cytokines and indices of acute damage are for adult animals that are
22 sexually mature and at doses that are 4 times (100 vs. 25 mg/kg) that of the sexually immature
23 animals who are going through a period of rapid hepatocyte proliferation (15 day old animals).
24 It is therefore, difficult to extrapolate between the two paradigms to distinguish the effects of
25 hormones and gender on the response. Finally, the work of Rakoff-Nahoum and Medzhitov
26 (2007) showed that it is the effect of tumor progression and not initiation that is affected by
27 MyD88 (a signaling adaptor to Toll-like receptors). Thus, examination of parameters at the
28 initiation phase at necrotic doses for liver tumors may not be relevant.
29
30 E.3.3.5. Epigenomic Modification
31 There are several examples of chemical exposure to differing carcinogens that have lead
32 to progressive loss of DNA methylation (i.e., DNA hypomethylation) including TCE and its
33 metabolites. The evidence for TCE and its metabolites is specifically discussed in
34 Section E.3.4.2.2, below. Other examples of carcinogens exposures or conditions that have been
35 noted to change DNA methylation are early stages of tumor development include ethionine
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-335 DRAFT—DO NOT CITE OR QUOTE
-------
1 feeding, phenobarbitol, arsenic, dibromoacetic acid, and stress. However, it has not yet been
2 established whether epigenetic changes induced by carcinogens and found in tumors play a
3 causative role in carcinogenesis or are merely a consequence of the transformed state (Tryndyak
4 et al., 2006).
5 Pogribny et al. (2007) report the effects of WY-14,643 on global mouse DNA
6 hypomethylation exposed at 1,000 ppm for 1 week, 5 weeks, or 5 months. What is of particular
7 note in this study is that at this exposure level, one commonly used for MOA studies using
8 WY-14,643 to characterize the effects of PPARa agonists as a class, there was significant
9 hepatonecrosis and mortality reported by Woods et al. (2007b). Both wild-type and PPARa -/-
10 null mice were examined. In wild-type mice DNA syntheses was elevated 3-, 13-, and 22-fold of
11 time-matched controls after 1 week, 5 weeks, and 5 months of WY 14,543 treatment. Changes
12 in ploidy were not examined. After 5 weeks of exposure, the ratio of unmethylated CpG cites in
13 whole liver DNA was the same for WY-14,643 treatment and control but by 5 months there was
14 an increase in hypomethylation in WY-14,643 treated wild-type mice. The authors did not report
15 whether foci were present or not which could have affected this result. The similarity in
16 hypomethylation at 5 days and 5 weeks, a time point that also had a small probability of foci
17 development, is suggestive of foci affecting the result at 5 months. For PPAR -/- mice there was
18 increased hypomethylation reported at 1 week and 5 weeks after WY-14,643 treatment that was
19 not statistically significant with so few animals studied. At 5 months the null mice had
20 decreased hypomethylation compared to 1 and 5 weeks. The authors note that, methylation of c-
21 Myc genes was reported to not be affected by long-term dietary treatment with WY-14,643 even
22 though WY-14,643-related hypomethylation of c-Myc gene early after a single dose of WY-
23 14,643 has been observed (Ge et al., 2001a). The authors concluded "thus, alterations in the
24 genome methylation patterns with continuous exposure to nongenotoxic liver carcinogens, such
25 as WY, may not be confined to specific cell proliferation-related genes."
26 Pogribny et al. (2007) reported Histone H3 and H4 trimethylation status in wild-type and
27 PPAR null mice to show a rapid and sustained loss of histone H3K9 and histone H4K20
28 trimethylation in wild-type mice fed WY-14,643 from 1 week to 5 months. There was no
29 progressive loss in histone hypomethylation, with the same amount of demethylation occurring
30 at 5 days, 5 weeks, and 5 months in wild-type mice fed WY-14,643. The change from control
31 was -60% reduction. The control values with time were not reported and all controls were
32 pooled to give one value (n = 15). For PPAR -/-1 mice there was a slight decrease with WY-
33 14,643 treatment (-15%) reported. In wild-type mice, WY-14,643 treatment was reported to
34 have no effect on the major histone methyltransferase, Suv39hl, while expression of another
35 (PRDM/Rizl) increased significantly as early as on week of treatment and remained elevated for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-336 DRAFT—DO NOT CITE OR QUOTE
-------
1 up to five months. The effect on expression of Suv420h2 (responsible for histone H4K20
2 trimethylation) was more gradual and the amounts of this protein in livers of mice fed Wy-
3 14m643 were reported to be lower than in control. The authors did not examine these
4 parameters in the null mice so the relationship of these effects to receptor activation cannot be
5 determined. Pogribny et al. (2007) report hypomethylation of retroelements (LTRLAP, LINE1
6 and LINE2 retrotransposons) following long-term exposure to WY-14,643, which the authors
7 concluded, can have effects on the stability of the genome. Again, these results are for whole
8 liver that may contain foci. Nevertheless, these findings raise questions about other target organs
9 and a more general mechanism for WY-14,643 effects than a receptor mediated one. The lack of
10 effects on c-Myc and the irrelevance of the transient proliferation through it reported here gives
11 more evidence of the irrelevance of a MO A dependent on transient proliferation. The authors
12 noted that studies show that a sustained loss of DNA methylation in liver is an early and
13 indispensable event in hepatocarcinogenesis induced by long-term exposure of both genotoxic
14 and nongenotoxic carcinogens in rodents. Thus, this statement argues against making such a
15 distinction in MO A for "genotoxic" and "nongenotoxic" carcinogens. Finally, the use of a dose
16 which Woods et al. (2007b) demonstrate to have significant hepatonecrosis and mortality, limits
17 the interpretation of these results and their relevance to models of carcinogenesis without
18 concurrent necrosis.
19 Strain sensitivity to hepatocarcinogenicity has been investigated in terms of short-term
20 changes in methylation. Bombail et al. (2004) reported that a tumor-inducing dose of
21 phenobarbital reduced the overall level of liver DNA methylation in a tumor-sensitive (B6C3F1)
22 mouse strain but that the same dose of phenobarbital did not alter global methylation level in a
23 more tumor-resistant strain (C57BL/6), although the compound increased hepatocyte
24 proliferation as measured by increased DNA synthesis in both strains (Counts et al., 1996).
25 Bombail et al. reported that "In a similar study, Watson and Goodman (2002) used a PCR-based
26 technique to measure DNA methylation changes specifically in GC-rich regions of the mouse
27 genome." Watson and Goodman (2002) found that, that in these areas of the genome, exposure
28 to phenobarbital caused an increase in methylation in dosed animals compared with control
29 animals. Again, the change was more pronounced in tumor-prone C3FI/He and B6C3F1 strains
30 than in the less sensitive C57BL/6 strain. They also reported increased DNA synthesis in
31 C57BL/6 mice but decreased global methylation in the B6C3F1 strain after PB administration
32 1-2 weeks. The lifetime spontaneous tumor rates were reported to be less than 5% in C57BL/6
33 mice but up to 80% in C3H/He mice. Counts et al. (1996) reported cell proliferation and global
34 hepatic methylation status in relatively liver tumor susceptible B6C3F1 with relatively resistant
35 C57BL6 mice following exposure to PB and/or chlorine/methionine deficient (CMD) diet. Cell
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-337 DRAFT—DO NOT CITE OR QUOTE
-------
1 proliferation (i.e, DNA synthesis) was reported to be higher in C57BL/6 mice while transient
2 hypomethylation occurred to a greater extent in B6C3F1 mice after phenobarbital treatment.
3 Dual administration of CMD and PB led to enhanced cell proliferation and greater global
4 hypomethylation with similar trends in terms of strain sensitivities in comparison to with either
5 treatment alone (i.e., greater increase in cell proliferation in C57BL/6 and greater levels of
6 hypomethylation in B6C3F1). Thus, the authors concluded that B6C3F1 mice have relatively
7 low capacity to maintain the nascent methylation status of their hepatic DNA. However, on the
8 whole, the control values for methylation for the C57BL/6 mice appear to be slightly higher than
9 the B6C3F1 mice. Thus, claims that the liver tumor sensitive B6C3F1 had more global
10 hypomethylation after a promoting stimulus, which could be related to tumor sensitivity, is
11 tempered by the fact that resistant strain had a higher control baseline of methylation. The
12 baseline level of LI or hepatocyte proliferation also appears to be slightly higher in the C57BL/6
13 mouse. In addition, the largest strain difference in hypomethylation after a CMD diet was at
14 Week 12 (135% of control for the B6C3F1 strain and 151% of control for the C57BL/6 strain)
15 and this pattern was opposite that for the 1 week time point. Thus, the suggestion by Counts et
16 al. (1996), that the inability to maintain methylation status by the B6C3F1 strain, is also not
17 supported by the longer duration data for CMD diet.
18
19 E.3.4. Specific Hypothesis for Mode of Action (MOA) of Trichloroethylene (TCE)
20 Hepatocarcinogenicity in Rodents
21 E.3.4.1. PPARa Agonism as the Mode of Action (MOA) for Liver Tumor Induction—The
22 State of the Hypothesis
23 PPARa receptor activation has been suggested to be the MOA for TCA liver tumor
24 induction and for TCE liver tumor induction to occur primarily as a result of the presence of its
25 metabolite TCA (NAS, 2006). However, as discussed previously (see Section E.2.1.10), TCE-
26 induced increases in liver weight have been reported in male and female mice that do not have a
27 functional PPARa receptor (Nakajima et al., 2000). The dose-response for TCE-induced liver
28 weight increases differs from that of TCA (see Section E.2.4.2). The phenotype of the tumors
29 induced by TCE have been described to differ from those by TCA and to be more like those
30 occurring spontaneously in mice, those induced by DC A, or those resulting from a combination
31 of exposures to both DCA and TCA (see Section E.2.4.4). As to whether TCA-induced tumors
32 are induced through activation of the PPARa receptor, the tumor phenotype of TCA-induced
33 mouse liver tumors has been reported to have a pattern of H-ras mutation frequency that is
34 opposite that reported for other peroxisome proliferators (see Section E.2.4.4.; Bull et al., 2002;
35 Stanley et al., 1994; Fox et al., 1990; Hegi et al., 1993). While TCE, DCA, and TCA are weak
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-338 DRAFT—DO NOT CITE OR QUOTE
-------
1 peroxisome proliferators, liver weight induction from exposure to these agents has not correlated
2 with increases in peroxisomal enzyme activity (e.g., PCO activity) or changes in peroxisomal
3 number or volume. However, liver weight induction from subchronic exposures appears to be a
4 more accurate predictor of carcinogenic response for DCA, TCA, and TCE in mice (see
5 Section E.2.4.4). The database for cancer induction in rats is much more limited than that of
6 mice for determination of a carcinogenic response to these chemicals in the liver and the nature
7 of such a response.
8 The MOA for peroxisome proliferators has been the subject of research and debate for
9 several decades. It has evolved from an "oxidative damage" due to increased peroxisomal
10 activity to a MOA framework example developed by Klaunig et al. (2003) that described causal
11 inferences for hepatocarcinogenesis after a chemical exposure was shown to activate of the
12 PPAR-a receptor with concurrent perturbation of cell proliferation and apoptosis, and selective
13 clonal expansion. Of note although inhibition of apoptosis was proposed as part of the sequellae
14 of PPARa activation, as noted in Section E.2.4.1, no changes in apoptosis in mice exposed to
15 TCE have been reported with the exception of mild enhanced apoptosis at 1,000 mg/kg/d dose
16 but more importantly that for mice the rate of apoptosis decreases as mice age and appear to be
17 lower than that of rats. While DCA exposure has been noted to reduce apoptosis, the
18 significance of DCA-induced reduction in apoptosis from a level that is already inherently low in
19 the mouse, is difficult to apply as the MOA for DCA-induced liver cancer.
20 Klaunig et al. based causal inferences on the attenuation of these events in PPAR-a-null
21 mice in response to the prototypical agonist WY-14,643 with a number of intermediary events
22 considered to be associative (e.g., expression of peroxisomal and nonperoxisome genes,
23 peroxisome proliferation, inhibition of gap junction intracellular communication, hepatocyte
24 oxidative stress as well as Kupffer cell-mediated events). The data set for DEHP was
25 prominently featured as an example of "PPAR-a induced hepatocarcinogenesis." For DEHP
26 PPAR-a activation was described as the initial key event with evidence lacking for a direct effect
27 but supported primarily supported by evidence from PPAR-a-knockout mice treated with
28 WY-14,643. Klaunig et al. concluded that".. .all the effects observed are due only to the
29 activation of this receptor and the downstream events resulting from this activation and that no
30 other modes of action are operant"
31 Although that PPARa receptor activation is the sole MOA for DEHP has been cited by
32 several reports (including IARC, 2000), several articles have questioned the adequacy of this
33 proposed MOA (Melnick, 2001, 2002, 2003; Melnick et al., 2007; FIFRA SAP, 2004; Caldwell
34 and Keshava, 2006; Caldwell et al., 2008b; Keshava and Caldwell, 2006; and Keshava et al.,
35 2007; Guyton et al. 2009). New information is now available that also questions several of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-339 DRAFT—DO NOT CITE OR QUOTE
-------
1 assumptions inherent in the proposed MOA by Klaunig et al. and the dismissal of PPARa
2 agonists as posing a health risk to humans. Specific questions have been raised about the use of
3 WY-14,643 as a prototype for PPARa (especially at necrogenic doses) and use of the PPARa -/-
4 null mouse in abbreviated bioassays to determine carcinogenic hazard.
5
6 E.3.4.1.1. Heterogeneity of PPARa agonist effects and inadequacy of WY-14,643 paradigm
1 as prototype for class. Inferences regarding the carcinogenic risk posed to humans by PPARa
8 agonists have been based on limited epidemiology studies in humans that were not designed to
9 detect such effects. However, as noted by Nissen et al. (2007) the PPARa receptor is pleiotropic,
10 highly conserved, has "cross talk" with a number of other nuclear receptors, and plays a role in
11 several disease states. "The fibrate class of drugs, which are PPARa agonists intended to treat
12 dyslipidemia and hypercholesterolemia, have recently been associated with a number of serious
13 side effects." While these reports of clinical side effects are for acute or subchronic conditions
14 and do not (and would not be expected to) be able to detect liver cancer from fibrate treatment,
15 they clearly demonstrate that compounds activating the PPAR receptors may produce a spectrum
16 of effects in humans and the difficulty in studying and predicting the effects from PPAR
17 agonism. Graham et al. (2004) recently reported significantly increased incidence of
18 hospitalized rhabdomyolysis in patients treated with fibrates both alone and in combination with
19 statins. Even though pharmaceutical companies have spent a great deal of effort to develop
20 agonists which are selective for desired effects, the pleiotropic nature of the receptor continues to
21 be an obstacle.
22 Also, fibrates, WY-14,643 and other PPARa agonists are pan agonists for other PPARs.
23 Shearer and Hoekstra (2003) note that fibrates, including Fenofibrate, Clofibrate, Bezafibrate,
24 Ciprofibrate, Gemfibrozil, and Beclofibrate are all drugs that were discovered prior to the
25 cloning of PPARa and without knowledge of their mechanism of action but with optimization of
26 lipid lowering activity carried out by administration of candidates to rodents. They report that
27 many PPARa ligands, including most of the common fibrate ligands, show only modest
28 selectivity over the other subtypes with, for example, fenofibric acid and WY-14,643 showing
29 < 10-fold selectivity for activation of human PPARa compared to PPARy and/or PPAR5. In
30 human receptor transactivation assays they report:
31 Human receptor transactivation assays of median effective concentration (ECso):
32
33 WY-14,643 = 5.0 urn for PPARa, 60 urn for PPAR y, 35 urn for PPAR6.
34 Clofibrate = 55 urn for PPARa, -500 urn for PPAR y, inactive at 100 urn for PPAR6
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-340 DRAFT—DO NOT CITE OR QUOTE
-------
1 Fenofibrate = 30 urn for PPARa, 300 urn for PPAR y, inactive at 100 urn for PPAR6
2 Bezafibrate = 50 urn for PPARa, 60 urn for PPAR y, 20 urn for PPAR6.
3
4 Murine receptor transactivation assay of ECso:
5
6 WY = 0.63 urn for PPARa, 32 urn for PPAR y, inactive at 100 urn for PPAR6
7 Clofibrate = 50 urn for PPARa, -500 urn for PPAR y, inactive at 100 urn for PPAR6
8 Fenofibrate = 18 urn for PPARa, 250 urn for PPAR y, inactive at 100 urn for PPAR6
9 Bezafibrate = 90 urn for PPARa, 55 urn for PPAR y, 110 urn for PPAR6.
10
11 Thus, these data show the relative effective concentrations and "potency for PPAR
12 activity" of various agonists in humans and rodents, rodent and human responses may vary
13 depending on agonist, agonists vary in what they activate between the differing receptors, and
14 that there is a great deal of transactivation of these drugs.
15 For fibrates specifically, a study by Nissen et al. (2007) reports that in current practice,
16 2 fibrates, Gemfibrozil and Fenobibrate, are still widely used to treat a constellation of lipid
17 abnormalities known as atherogenic dyslipidemia and note that currently available fibrates are
18 weak ligands for the PPARa receptor and may interact with other PPAR systems. They note that
19 the pharmaceutical industry has sought to develop new, more potent and selective agents within
20 this class but, most importantly, that none of the novel PPARa agonists has achieved regulatory
21 approval and that according to a former safety officer in the U.S. Food and Drug Administration
22 (El-Hage, 2007) that more than 50 PPAR modulating agents have been discontinued due to
23 various types of toxicity (e.g., elevations in serum creatinine, rhabdomylosis, "multi-species,
24 multi-site increases in tumor with no safety margin for clinical exposures," and adverse
25 cardiovascular outcomes) but without scientific publications describing the reasons for
26 termination of the development programs. Nissen et al. report differences in effect between a
27 more highly selective and potent PPARa agonist and the less potent and specific one in humans.
28 They note
29
30 a recent large study of Fenofibrate in patients with diabetes showed no significant
31 reduction in morbidity but a trend toward increased all-cause mortality (Keech et
32 al. 2005, 2006). Whether this potential increase in mortality is derived from
33 compound specific toxicity of Fenofibrate or is an adverse effect of PPARa
34 activation remains uncertain."
35
36 In addition to the lack of publication of effects from PPAR agonists in human
37 trials in which toxicity can be examined as noted by Nissen et al., the Keech study
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-341 DRAFT—DO NOT CITE OR QUOTE
-------
1 is illustrative of the problem in trying to ascertain liver effects from fibrate
2 treatment in humans as the focus of the outcomes was coronary events in a study
3 of 5 years duration in a older diabetic population. As stated above, the challenges
4 the pharmaceutical industry and the risk assessor face in determining the effects
5 of PPAR agonists is "that these compounds and drugs modulate the activity of a
6 large number of genes, some of which produce unknown effects."
7
8 Nissen et al. further note that
9
10 Accordingly, the beneficial effects of PPAR activation appear to be associated
11 with a variety of untoward effects which may include, oncogenesis, renal
12 dysfunction, rhabdomylosis, and cardiovascular toxicity. Recently, the FDA
13 began requiring 2-year preclinical oncogenicity studies for all PPAR-modulating
14 agents prior to exposure of patients for durations of longer than 6 months
15 (El-Hage, 2007).
16
17 Guyton et al. (2009) further explore the status of the PPARa epidemiological database and
18 describe its inability to discern a cancer hazard from the available data. Thus, while existing
19 evidence for liver cancer in humans is null rather than negative, there remains a concern for
20 oncogenicity and many obstacles for determining such effects through human study. The
21 heterogeneity in response to PPARa agonists and the heterogeneity of effects they cause
22 (Keshava and Caldwell, 2006) are evident from these reports.
23 Many studies have used the effects of WY-14,643 at a very high dose and extrapolated
24 those findings to PPARa agonists as a class. However, this diverse group of chemicals have
25 varying potencies and effects for the "key events" described by Klaunig et al. (2003) (Keshava
26 and Caldwell, 2006). The standard paradigm used with WY-14,643 to induced liver tumors in
27 all mice exposed to 1 year (an abbreviated bioassay), uses a large dose that has also has been
28 reported to produced liver necrosis, which can have an effect of cell proliferation and gene
29 expression patterns, and to also induce premature mortality (Woods et al., 2007b). As stated
30 above, WY-14,643 also has a short peak of DNA synthesis that peaks after a few days of
31 exposure, recedes, and then unlike most PPARa agonists studied (e.g., Clofibrate, clofibric acid,
32 Nafenopin, Ciprofibrate, DEHP, DCA, TCA and LY-171883) has a sustained proliferation at the
33 doses studied (Tanaka et al., 1992; Barrass et al., 1993; Marsman et al., 1992; Eacho et al., 1991;
34 Lake et al., 1993; Yeldani et al., 1989; David et al., 1999; Marsman et al., 1988; Carter et al.,
35 1995; Sanchez and Bull, 1990). Clofibrate has been shown to have a decrease in proliferation
36 gene expression shortly after its peak (see Section E.3.2.2). As shown in above for WY-14,643,
37 hepatocellular increases in DNA synthesis did not appear to have a dose-response (see
38 Section E.3.4.2), only WY-14,643 had a sustained elevation of Nf-KB (gem and dibutyl phthalate
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-342 DRAFT—DO NOT CITE OR QUOTE
-------
1 did not) (see Section E.3.4.3.3), and the effects on DNA methylation occurred at 5 months and
2 not earlier time points (when Foci were probably present) and effects of histone trimethylation
3 were observed to be the same from 1 weeks to 5 months (see Section E.3.4.5). Such effects on
4 the epigenome suggest other effects of WY-14,643, other than receptor activation, are not
5 specific to just WY-14,643 and are found in a number of conditions leading to cancer and in
6 tumor progression (see Sections E.3.2.1 and E.3.2.7.).
7 In their study of PPARa-independent short-term production of reactive oxygen species
8 from induced by large concentrations of WY-14,643 and DEHP in the diet, Woods et al. (2007c)
9 examined short-term exposures to (0.6% w/w DEHP or 0.05% or 500 pm WY-14,643 for 3 days,
10 1 weeks or 3 weeks) and reported that WY-14,643 induced a dramatic increase in bile flow that
11 was not observed from DEHP exposure. By 1 week of exposure there was a 5% increase in bile
12 flow for DEHP treatment but a 240% increase in bile flow for WY-14,643 treatment. By
13 3 weeks the difference in bile volume between treated and control was 12% for DEHP and
14 1,100% for WY-14,643 treated animals. In this study oxygen radical formation, as measured by
15 spin trapping in the bile, was reported to be decreased after 3 days of treatment after DEHP and
16 WY-14,643 treatment. However, the large changes in bile flow by WY-14,643 treatment limit
17 the interpretation of these data along with a small number of animals examined in this study
18 (e.g., 6 control and DEHP animals and 3 animals exposed to WY-14,643 at 3 days), a 30%
19 variation in percent liver/body weight ratios between control groups, and the insensitivity of the
20 technique. In an earlier study oxidative stress appears to be correlated with neither cell
21 proliferation nor carcinogenic potency (Woods et al., 2006). Woods et al. (2006) reported
22 WY-14,643Y or DEHP to induce an increase in free radicals at 2 hrs, a decrease at 3 days then
23 an increase at 3 weeks for both. However, radical formation did not correlate with the
24 proliferative response, as DEHP fails to produce a sustained induction of proliferative response
25 in rodent liver but WY-14,643 does, and both WY-14,643 and DEHP gave a similar pattern of
26 radical formation that did not vary much from controls which is in contrast to their carcinogenic
27 potency.
28 Although assumed to be a reflection of cell proliferation in many studies of WY-14,643
29 and by Klaunig et al. (2003), DNA synthesis recorded using the standard exposure paradigm for
30 WY-14,643, can also be a reflection of hepatocyte, nonparenchymal cell or inflammatory cell
31 mitogenesis (in the case of necrosis induced inflammation), from changes in hepatocyte ploidy,
32 or a combination of all. Other peroxisome proliferators have been shown to have a decrease in
33 proliferation gene expression shortly after their peaks (e.g., Clofibrate, see Section E.3.2.2) and
34 both Methylclofenapate and Nafenopin have been shown to increase cell ploidy with Nafenopin
35 having the majority of its DNA synthesis a reflection of increased ploidy with only a small
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-343 DRAFT—DO NOT CITE OR QUOTE
-------
1 percentage as increases in cell number (see Section E.3.4.1). Several authors have also noted
2 increases in ploidy for WY-14,643 (see Section E.3.4.1).
3 The Tg.AC genetically modified mouse was used to study 14 chemicals administered by
4 the topical and oral (gavage and/or diet) routes by Eastin et al. (2001). Clofibrate was considered
5 clearly positive in the topical studies but not WY-14,643 regardless of route of administration.
6 Based on the observed responses, it was concluded by the workgroup (Assay Working Groups)
7 that the Tg.AC model was not overly sensitive and possesses utility as an adjunct to the battery
8 of toxicity studies used to establish human carcinogenic risk. The difference in result between
9 Clofibrate and WY-14,643 is indicative of a different MOA for the two compounds.
10 Similarly, at large exposure concentrations Boerrigter (2004) investigated the response of
11 male and female lacZ-plasmid transgenic mice treated at 4 months of age with 6 doses of
12 2,333 mg/kg DEHP, 200 mg/kg WY-14,643 or 90 mg/kg Clofibrate over a two week period.
13 Mutation frequencies were assayed at 21 days following the last exposure. DEHP and WY-
14 14,643 were shown to significantly elevate the mutant frequency in both male and female liver
15 DNA while Clofibrate, at the dose level studied, was apparently nonmutagenic in male and
16 female liver (i.e., six-dose exposure to DEHP or WY-14,643 over a two week period
17 significantly increased the mutant frequency in liver of both female and male mice by
18 approximately 40%). The author noted that
19
20 the laxZ plasmid-based transgenic mouse mutation assay is somewhat unique
21 among other commercially available models (e.g. mutamouse and big blue), by
22 virtue of its ability to accurately quantify both point mutations and large deletions
23 including those which originate in the lacZ plasmid catamer and extend into the 3'
24 flanking genomic region. It should be noted that to date there is no single, agreed
25 upon protocol for conducting mutagenicity assays with transgenic rodents
26 although several aspects have been upon by the Transgenic Mutation Assays
27 workgroup of the International Workshop on Genotoxicity Procedures.
28
29 For several chemicals both rats and mice demonstrate evidence of receptor activation
30 through peroxisome proliferation and peroxisome-related gene expression but only one develops
31 cancer. The herbicide, 2,4-dichlorophenoxyacetic acid (2,4-D), is a striking example of the
32 problems that would be associated with only using evidence of PPARa receptor activation to
33 make conclusions about MOA of liver tumors. 2,4-D is structurally similar to the PPARa
34 agonist Clofibrate and has been shown at similar concentrations to increase peroxisome number
35 and size, increase hepatic carnitine acetyltransferase activity and catalase, and decrease serum
36 triglycerides and cholesterol in rats (Vainio et al., 1983). Peroxisome number was also increased
37 in Chinese hamsters to a similar level as with Clofibrate at the same exposure concentration after
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-344 DRAFT—DO NOT CITE OR QUOTE
-------
1 9 days of exposure to 2,4-D (Vainio et al., 1982). In mice, Lundgren et al. (1987) report that
2 2,4-D exposure statistically increased the liver-somatic index over controls after a few days
3 exposure and increased mitochondrial protein, microsomal protein, carnitine acetyltransferase,
4 PCO activity, cytochrome oxidase, cytosolic epoxide hydrolase, microsomal epoxide hydrolase,
5 microsomal P450 content, and hepatic cytosolic epoxide hydrolase in mouse liver. Thus, 2,4-D
6 activates the PPARa receptor, with associated changes in peroxisome-related gene expression, in
7 multiple species and at similar doses to Clofibrate. However, Charles et al. (1996) and Charles
8 and Leeming (1998) report that in several 2-year studies that there were no 2,4-D-induced
9 increases in liver tumors in F344 rats, CD-I rats, B6C3F1 mice and CD-I mice. Another
10 example, is provided by Gemfibrozil, known as (5-2[2,5-dimethylphenoxy]
11 2-2-dimethylpentanoic acid) and [2,2-dimethyl-5-(2,5-xylyoxy) valeric acid], a therapeutic agent
12 that activates the PPARa receptor and is a peroxisome proliferator, but is carcinogenic only in
13 male rats but not female rats, nor in either gender of mouse (Contrera et al., 1997). Gemfibrozil
14 causes tumors in pancreas, liver, adrenal, and testes of male rats and causes increases in absolute
15 and relative liver weights in both rats and mice (Fitzgerald et al., 1981). Gemfibrozil, is a highly
16 effective lipid and cholesterol lowering drugs in humans and in mice (Olivier et al., 1988).
17 However, although Gemfibrozil activates the PPARa receptor and induces peroxisome
18 proliferation in mice, it does not induce liver tumors in that species. In the long-term study of
19 Bezafibrate, Hays et al. (2005) note that the role of this receptor in hepatocarcinogenesis has
20 only been examined using one relatively specific PPARa agonist (WY-14,643) and report that
21 Bezafibrate can induce the expression of a number of PPARa target genes (acyl CoA oxidase
22 and CYP4a) and increased liver weight in PPARa knockout mice that is not dependent on
23 activation of PPARP or PPARy. As noted by Boerrigter (2004),
24
25 In contrast to DEHP and WY-14,643, Clofibrate produced hepatocellular
26 carcinomas in rats only while no increase in the incidence of tumors was reported
27 in mice (Gold and Zeiger 1997). However, Clofibrate induces peroxisome
28 proliferation in both rats and mice (Lundgren and DePierre 1989) but only
29 produced hepatocellular carcinomas in rats (Gold and Zeiger, 1997).
30
31 Melnick et al. (1996) noted that similar levels of peroxisomal induction were observed in rats
32 exposed to DEHP and di(2-ethylhexyl) adipate (DEHA) at doses comparable to those used in the
33 bioassays of these chemicals. However, DEHP but not DEHA gave a positive liver tumor
34 response in 2-year studies in rats. In an evaluation of the carcinogenicity of tetrachloroethylene,
35 an expert panel of the International Agency for Research on Cancer concluded that the weak
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-345 DRAFT—DO NOT CITE OR QUOTE
-------
1 induction of peroxisome proliferation by this chemical in mice was not sufficient to explain the
2 high incidence of liver tumors observed in an inhalation bioassay.
3 In adult animals, apoptosis acts as a safeguard to prevent cells with damaged DNA from
4 progressing to tumor, but like cell proliferation, alterations in apoptosis are common to many
5 MO As. In addition, only short-term data are available on changes in apoptosis due to PPARa
6 agonists, and long-term changes have not been investigated (Rusyn et al., 2006). For example,
7 although a decrease in apoptosis has also suggested to be an important additional molecular
8 event that may affect the number of cells in rodent liver following exposure to the peroxisome
9 proliferator DEHP, apoptosis rates have not investigated past 4 days of exposure and thus, the
10 time-course of this event is uncertain. The antiapoptotic effects of PPAR agonists appear to be
11 also dependent on nonparenchymal cells (i.e., Kupffer cells) which do not express PPARa and
12 could be a transient event (Rusyn et al., 2006). Morimura et al. (2006) report evidence for
13 exposure to WY-14,643 that does not support a role for PPARa-mediated apoptosis in tumor
14 formation (see Section E.3.5.1.3, below) as well as appearing to be specific to WY-14,643 (see
15 Section E.3.4.3.3).
16 The lack of a causal relationship of transient DNA synthesis increases and
17 hepatocarcinogenesis has been raised by many (Caldwell et al., 2008b) and is discussed in
18 Section E.3.4.2 as well as the changes in ploidy (see Section E.3.4.1). In regard to gene
19 expression profiles, many studies have focused on gene profiles during the early transient
20 proliferative phase or have identified genes primarily associated with peroxisome proliferation as
21 "characteristic" or relevant to those associated with tumor induction. Several have focused on
22 the number of genes whose expression "goes up" or "goes down" from a small number of
23 animals. Caldwell and Keshava (2006) presented information on WY-14,643, dibutyl phthalate,
24 Gemfibrozil and DEHP, and noted inconsistent results between PPARa agonists, paradoxes
25 between mRNA and protein expression, strain, gender, and species differences in response to the
26 same chemical, and time-dependent differences in response for several enzymes and glutathione.
27
28 E.3.4.1.2. New information on causality and sufficiency for PPARa receptor activation. In
29 its review of the U.S. EPA's draft risk assessment of perfluorooctanoic acid (PFOA), the Science
30 Advisory Panel (FIFRA SAP, 2004) expressed concerns about whether PPARa agonism
31 constitutes the sole MOA for PFOA effects in the liver and the relevance to exposed fetuses,
32 infants, and children. In part based on uncertainties regarding the Klaunig et al. (2003) proposed
33 MOA, they concluded that the tumors induced by PFOA were relevant to human risk assessment.
34 The hypothesis that activation of the PPARa receptor is the sole mode of action
35 hepatocarcinogenesis induced by DEHP and many other chemicals is further called into question
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-346 DRAFT—DO NOT CITE OR QUOTE
-------
1 by recent studies. In the case of DEHP, Klaunig et al. (2003) assumed that WY-14,643 and
2 DEHP would operate through the same key events and that long-term bioassays of DEHP in
3 PPARa -/- knockout mice would be negative and hence demonstrate the need for receptor
4 activation for hepatocarcinogenesis from DEHP.
5 The fallacy of these assumptions is illustrated by the recent report of the first 2-year
6 bioassay of DEHP in PPARa -/- knockout mice (Sv/129 background strain) that reported DEHP-
7 induced hepatocarcinogenesis (Ito et al., 2007). Further discussion is provided by Guyton et al.
8 (2009). Similar to other studies, the PPAR -/- mice had slightly increased liver weights in
9 comparison to controls and treated wild-type mice (-12% increase over controls). In fact
10 statistical analysis of the incidence data show that adenomas were significantly increased in
11 PPARa -/- mice compared with wild-type mice exposed to 500 ppm DEHP and that a significant
12 dose-response trend for adenomas and adenomas plus carcinomas was observed in PPARa -/-
13 mice (Figure E-5). Overall, the cancer incidences were consistent with a previous study of
14 DEHP (David et al., 1999) in B6C3F1 mice at the same doses for nearly the same exposure
15 duration. A strength of this study is that it was conducted at much lower more environmentally
16 relevant doses that did not significantly increase liver enzymes as indications of toxicity. As
17 noted by Kamija et al. (2007), DEHP was reported also to induce glomerularnephritis more often
18 in PPARa-null mice because of the absence of PPARa-dependent anti-inflammatory effect of
19 antagonizing the oxidative stress and NF-KB pathway (Kamijo et al., 2007). Thus, these data
20 support that hypothesis that there is no difference in liver tumor incidences between PPARa -/-
21 mice and wild-type mice in a standard nonabbreviated exposure bioassay that does not exceed
22 the maximal tolerated doses and that DEHP can induce hepatotoxicity as well as other effects
23 independent of action of the PPARa receptor.
24 The study of Yang et al. (2007a) informs as to the sufficiency of PPARa receptor
25 activation and subsequent molecular event for hepatocarcinogenesis in mice. The study used a
26 VP16PPARa transgene under control of the liver-enriched activator protein (LAP) promoter to
27 activate constitutively the PPARa receptor in mouse hepatocytes. LAP-VP16PPARa transgenic
28 mice showed a number of effects associated with PPARa receptor activation including decreased
29 serum triglycerides and free fatty acids, peroxisome proliferation, enhanced hepatocyte DNA
30 synthesis and induction of cell-cycle genes and those described as "PPARa targets" to
31 comparable levels reported for WY-14,643 exposure. Hepatocyte proliferation, as determined by
32 the labeling index of hepatocyte nuclei, was increased after 2 weeks of WY-14,643 treatment
33 over controls (20.5 vs. 1.6% in control livers) with the LAP-VP16PPARa mice giving a similar
34 results (20.8 vs. 1.0% in control livers). The authors noted that transgenic mice did not appear to
35 have positive labeling of nonparenchymal cell nuclei that were present in the WY-14,643 treated
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-347 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
animals. The transferase-mediated dUTP nick end-labeling assay results were reported to show
that there was no difference in apoptosis in wild-type mice treated with WY-14,643, the
transgenic mice, or controls. In a small number of animals, microsomal genes (CYP4A),
peroxisomal (Acox, BIEN—the bifunctional enzyme) and mitochondrial fatty oxidation genes
(LCAD—long chain acyl CoA dehydrogenase and VLCAD—very long chain acyl CoA
dehydrogenase) were expressed in the transgenic mice with WY-14,643 also increasing
expression of these genes in wild-type mice but with less lipoprotein lipase (LPL) than the
transgenic mice. Hepatic CoA oxidation, were increased to a similar level in wild-type mice
treated with WY-14,643 and the transgenic mice (n = 3-4) and were statistically different than
controls. LAP- VP16PPARa transgenic mice (8 weeks of age) exhibited hepatomegaly (-50
increase percent body/liver weight over controls), and an accumulation of lipid due to
triglycerides but not cholesterol. However, compared to wild-type mice exposed to WY-14,643
for two weeks, the extent of hepatomegaly was reduced (i.e., percent liver/body weight increase
of ~2.5-fold with WY-14,643 treatment), no hepatocellular hypertrophy or eosinophilic
cytoplasms and no evidence of nonparenchymal cell proliferation were observed in the
LAP-VP16PPARa transgenic mice.
18
19
20
21
22
23
24
25
Adenomas
100
|>|>m DEHP
500
in Ito-Wild • no-knockout n David-79wk n David-total
+ - p<=0.05 by 1 -tail Fisher exact test as compared to
control; * - p <=0.05 by 1 - and 2-tail Fisher exact test as
compared to control in same study
Carcinomas
100
|>|>m DEHP
500
n Ito-Wild • Ito-knockout
nDavid-79wk n David-total
No statistically-significant differences across all
studies and doses.
Figure E-5. Comparison of Ito et al. and David et al. data for DEHP tumor
induction from Guyton et al. (2009).
At ~1 year of age, Yang et al. (2007a) reported there to be no evidence of preneoplastic
lesions or hepatocellular neoplasia in LAP- VP16PPARa transgenic mice, in contrast to results
after 11 months of exposure to WY-14,643 in wild-type mice. Microscopic examination of liver
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-348 DRAFT—DO NOT CITE OR QUOTE
-------
1 sections were consistent with the gross findings, as hepatocellular carcinomas and hepatic lesions
2 were observed in the long-term WY-14,643 treated wild-type mice, but not in >20
3 LAP-VP16PPARa mice at the age of over 1 year in the absence of dox. There was no
4 quantitative information on tumors given nor of foci development in the WY-14,643 mice. As
5 noted by Yang et al. (2007a), PPARa activation only in mouse hepatocytes is sufficient to induce
6 peroxisome proliferation and increased DNA synthesis but not to induce liver tumors. Thus,
7 "hepatocyte proliferation" indentified by Klaunig et al. (2003) as a "causal event" in their
8 PPARa MOA is not sufficient to induce hepatocarcinogenesis. These data not only call into
9 question the adequacy of the MOA hypothesis proposed by Klaunig et al. (2003) but suggest
10 multiple mechanisms and also multiple cell types may be involved in hepatocarcinogenicity
11 caused by chemicals that are also PPARa agonists.
12
13 E.3.4.1.3. Use of the PPAR-/- knockout and humanized mouse. Great importance has been
14 attached to the results reported for PPARa -/- mice and their humanized counterpart with respect
15 to inferences regarding the MOA or peroxisome proliferators and whether short-term chemical
16 exposures or abbreviated bioassays conducted with these mice can show that a PPARa MOA is
17 involved. Consequently, the use of these models warrants scrutiny. Compared to untreated
18 wild-type mice, liver weights in knockout mice or humanized mice have been reported to be
19 elevated (Voss et al., 2006; Laughter et al., 2004; Morimura et al., 2006) and within 10% of each
20 other (Peters et al., 1997). In order to be able to assign affects to a test chemical tested in
21 knockout mice, a better characterization is needed of the baseline differences between PPARa -/-
22 knockout and wild-type mice. This is particularly important for examining weak agonists
23 because the changes they induce may be small and need to be confidently distinguished from
24 differences due to the loss of the receptor alone. As shown by the Ito et al. (2007) study and as
25 noted by Maronpot et al. (2004), there is a need for lifetime studies to characterize background or
26 spontaneous tumor patterns and life spans (including those of the background strain). While the
27 original work by Lee et al. (1995) describes "the mice homozygous for the mutation were viable,
28 healthy, and fertile and appeared normal," the authors did not describe the survival curves for
29 this model nor their background tumor rate. In fact, further work has shown that they carry a
30 background of chronic conditions, including: (1) chronic diseases such as obesity and steatosis
31 (Akiyama et al., 2001; Costet et al., 1998); (2) altered hepatic of hepatocellular structure and
32 function, such as vacuolated hepatocytes (Voss et al., 2006; Anderson et al., 2004), also seen in
33 "humanized" mice (Cheung et al., 2004); and (3) altered lipid metabolism, including reduced
34 glycogen stores, blunted hepatic and cardiac fatty acid oxidation enzyme system response to
35 fasting, elevated plasma free fatty acids, fatty liver (steatosis), impaired gluconeogenesis, and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-349 DRAFT—DO NOT CITE OR QUOTE
-------
1 significant hepatic insulin resistance (Lewitt el al., 2001). Howroyd et al. (2004) reported
2 decreased longevity and enhancement of age-dependent lesions in PPARa -/- mice.
3 These baseline differences from wild-type mice may render them more susceptible to
4 toxic responses or shorten their lifespans with chemical exposure. For example, after
5 administration of 250 microliters CCVkg, all male and 40% of female PPARa knockout mice
6 were dead or moribund after 2 days of treatment, whereas 25% of male wild-type mice and none
7 of the female wild-type mice exhibited outward signs of toxicity (Anderson et al., 2004). Hays
8 et al. (2005) reported that 100% of PPARa knockout have cholestasis after 1 year of Bezafibrate
9 treatment with higher bile acid concentration than wild-type mice. Lewitt et al. (2001) noted that
10 male knockout mice have more marked accumulation of hepatic fat, hypercholesterolemia and to
11 be particularly sensitive to fasting with some dying if fasted for more than 24 hours. Sexual
12 dimorphism but especially increased susceptibility of the male mouse has been reported for
13 knockout mice with pure Sv/129 backgrounds (Lewitt et al., 2001; Anderson et al., 2004) as well
14 as those with a suggested C57BL/6N background (Djouadi et al., 1998, Costet et al., 1998).
15 Akiyama et al. (2001) showed an apparent greater sexual dimorphism in mice with a pure Sv/129
16 background than C57BL/6N in regard to weight gain from 2 to 9 months but not in changes in
17 body weight or liver weight between wild-type and knockout animals. Adipose tissue, serum
18 triglycerides and cholesterol were altered in the knockout animals. Given that the experiment
19 was only carried out for 9 months, changes in body fat, liver weight and lipid levels may be
20 greater as the animals get older and steatosis is more prevalent. The dramatic effect on survival
21 as well as gender difference by the increased expression of lipoprotein lipase in the PPARa
22 knockout mouse with further genetic modification is demonstrated by Nohammer et al. (2003)
23 who reported 50% mortality in 6 months and 100% mortality within 11 months of age while
24 females survived. These differences could affect the results of tumor induction for PPARa
25 agonists with less potency than WY-14,643 that do not produce tumors so rapidly. In addition,
26 these studies suggest the need for careful consideration of the effects of use of different
27 background strains for the knockout and the need for careful characterization of the background
28 responses of the mouse model and the effects of the use of different background strains for the
29 knockout. Morimura et al. (2006) reported that, using the B6 background strain, there were only
30 foci at time periods but knockouts with the SV129 background had multiple tumors after WY-
31 14,643 treatment.
32 PPARa knockout mice have also been used to examine the dependence of PPARa on
33 changes in cell signaling, protein production, or liver weight. However, to be useful, the changes
34 incurred just by loss of the PPARa should also be well described. Reported differenced between
35 PPARa-knockout and wild-type mice can impact the sensitivity and specificity of these markers
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-350 DRAFT—DO NOT CITE OR QUOTE
-------
1 of for the hypothesized MOA. In regards to altered cell signaling, Wheeler et al. (2003) note that
2 in normal cells p21waf and p27klpl inhibit the Cdk/cyclin complexes responsible for cell cycle
3 progression through Gl/S transition. While these cellular signaling molecules are down-
4 regulated in response to partial hepatectomy in normal mice, they remain elevated in PPARa
5 knockout mice along with decreased DNA synthesis. Fumonisins are hepatocarcinogens that
6 have been associated changes in apoptosis and tissue generation, and increased acyl-CoA
7 oxidase and CYP4A (markers of PPARa activation) (Martinez-Larranaga et al., 1996). Voss et
8 al. (2006) report that the average number of hepatic apoptotic foci per mouse induced by
9 Fumonisins were 3-fold higher and liver mitotic figures counts were 2-fold lower in PPARa
10 knockout in comparison to wild-type mice, thus, illustrating a difference in proliferative response
11 in the mice. PPARa-null mice have been reported to have increased apoptosis and decreased
12 mitosis with fumonisin treatment. Voss et al. (2006) also report several differences in gene
13 expression in wild-type and PPARa knockout mice that ranged from 0.3 to 483% of the activity
14 of wild-type mice. The complex expression patterns of gene expression and determination of
15 their mechanistic implications in regard to hepatotoxicity and carcinogenicity are difficult.
16 Certainly the large number of genes whose expression is affected by WY-14,643 (1,012 genes as
17 cited by Voss et al., 2006) illustrates such complexity. Voss et al. (2006) conclude that studies
18 should consider dose- and time course-related effect as well as species and strain-related
19 differences in the expression of gene products.
20 The "humanized" PPARa mouse has a human copy of PPARa inserted into a PPARa
21 knockout mouse. It is inserted in a tetracycline response system so that in the absence of DOX
22 only human PPARa is transcribed in humanized mouse liver and not in other tissues. A rigorous
23 examination of newly emerging studies regarding the "humanized" mouse is warranted. There
24 are two papers that have been published using the humanized PPARa mouse (Cheung et al.,
25 2004; Morimura et al., 2006). Many of the issues described above for PPARa -/- mice are of
26 concern for the humanized knockout mouse. In addition, the placement of the humanized PPAR
27 gene is a potential confounding factor, as discussed by Morimura et al. (2006):
28
29 It also cannot be ruled out that the hPPARa mice are resistant to the hepatotoxic
30 effects of peroxisome proliferators due to the site of expression of the human
31 receptor. The cDNA was placed under control of the tetracycline regulatory
32 system and the liver-specific Cebp/B promoter that is preferentially expressed in
33 hepatocytes.
34
35 In the Cheung et al. (2004) report, the humanized mouse was fed WY-14,643 for 2 or
36 8 weeks (age not given for the mice). WY-14,643 and Fenobrate were reported to decrease
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 51 DRAFT—DO NOT CITE OR QUOTE
-------
1 serum total triglyceride levels in wild and humanized mice to about the level seen in PPARa -/-
2 mice (which were already suppressed without treatment). Hepatomegaly and increase in
3 hepatocyte size were observed in the PPARa -humanized mice fed WY-14,643 for 2 weeks but
4 less than that of wild mice. By contrast, Morimura et al., (2006) state that the humanized mice
5 did not exhibit hepatomegaly after treatment with WY-14,643. Cheung et al (2004) present
6 figures that show increased vacuolization of hepatocytes in a control humanized mouse in
7 comparison to wild-type mice. Vacuolization increased with WY-14,643 treatment in the
8 humanized mouse. Therefore, there was a background level of liver dysfunction in these mice
9 even with humanized PPARa. Vacuolization is consistent with fatty liver observed in the
10 nonhumanized PPARa -/- mouse. The authors reported that the humanized mouse did not have
11 increased #s of peroxisomes after WY treatment. However, they present a figure for genes
12 encoding peroxisomal, mitochondrial, and microsomal fatty acid oxidation enzymes that shows
13 they were still markedly increased in PPARa -humanized mice following 8 weeks of exposure to
14 WY-14,643. Therefore, there is a paradox in these reported results.
15 Morimura et al. (2006) provided a useful example to illustrate the many issues associated
16 with interpreting studies with genetically-altered animals. While this study is suggestive of a
17 difference in susceptibility to tumor induction between wild-type and PPARa humanized mice, a
18 conclusion that human PPARa is refractory to liver tumor induction is not sufficiently supported
19 by this study. This study had uneven durations of exposure and follow-up and reported
20 substantial toxicity or mortality that limit the interpretation of the observed tumor rates. For
21 example, the 6 week-old male "humanized" mice had a 44-week experimental period but for
22 wild-type mice that period was 38 weeks. In addition, for humanized mice, 10 mice were treated
23 with 0.1% WY-14,643 with 20 controls, but for wild-type mice, 9 mice were given 0.1% WY
24 with 10 controls. Furthermore, wild-type, WY-14,643-treated animals had suppressed growth
25 and only a 50% survival to 38 weeks, so an effective LD50 has been used for this length of
26 exposure. Specifically, of the 10 wild-type WY-14,643 treated mice, 3 died of toxicity and 2
27 were killed due to morbidity and their tissues examined. Humanized mice had similar growth for
28 animals treated with WY-14,643 or controls with only one mouse killed because of morbidity.
29 Therefore, the reported results, including tumor numbers, are for a mixture of different exposure
30 durations and ages of animals. In addition the results of the study were reported for only on
31 exposure level.
32 Furthermore, it is interesting that while control humanized mice had no adenomas,
33 WY-14,643 treated humanized mice had one. Morimura et al. (2006) noted that this adenoma
34 had a morphology "similar to spontaneous mouse liver tumor with basophilic and clear
35 hepatocytes," whereas the tumors in wild-type mice treated with WY-14,643 were more
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 52 DRAFT—DO NOT CITE OR QUOTE
-------
1 diffusely basophilic. If the humanized animals were allowed to live their natural lifespan, this
2 raises the possibility that WY-14,643 may induce tumors that are similar to other carcinogens
3 rather than those that have been described as "characteristic" of peroxisome proliferators (see
4 Section E.3.5.1.5) when human PPARa is present. Therefore, the humanized PPARa rather than
5 mouse PPARa may have an association with a tumor phenotype characteristic of other MO As
6 but this study need to be carried out for a longer period of exposure and with more animals to
7 make that determination. The baseline tumor response of PPARa humanized mice needs to be
8 characterized as well as tumors exposure to WY-14,643 or other carcinogens acting through
9 differing MO As. The numbers of foci were not reported, but "altered foci" were detected in one
10 humanized mouse with WY-14,643 treatment and one without treatment. The phenotypes of the
11 foci were not given by the authors.
12 As discussed above, changes in liver weights have been associated with susceptibility to
13 liver tumor induction and the issues regarding baseline differences in PPARa -/- mice are equally
14 relevant for PPARa humanized mice. Morimura et al. (2006) reported that absolute liver weight
15 for control humanized mice at 44 weeks was 1.57 g (n = 10). The absolute liver weight for wild
16 control mice was 1.1 g (n = 9) at 38 weeks. The final body weights differed by 14% but liver
17 weights differed by 30%. Therefore, even though comparing different aged mice, the control
18 humanized mice had greater liver size than the wild-type control mice on an absolute and relative
19 basis. This is consistent with humanized knockout mice having greater sized livers and a
20 baseline of hepatomegaly. With treatment, Morimura et al. (2006) report that PPARa humanized
21 mice treated with WY-14,643 had greater absolute and relative liver weights than controls but
22 less elevations than wild-type treated animals. However, because half of the wild-type animals
23 died, it is difficult to discern if liver weights were reported for moribund animals sacrificed as
24 well as animals that survived to 38 weeks for wild-type mice treated with WY-14,643. However,
25 it appears that moribund animals were included that were sacrificed early for treated groups and
26 that values from the animal killed at 27 weeks were added in with those surviving till 45 weeks
27 in the PPARa humanized mice treated with WY-14,643 group.
28 With respect to the gene expression results reported by Morimura et al. (2006), it is
29 important to note that they are for liver homogenates with a significant portion of the nuclei from
30 nonparenchymal cell of the liver (e.g., Kupffer and stellate cells). Thus, the results represent
31 changes resulting from a mixture of cell types and from differing zones of the liver lobule, with
32 potentially different gene changes merged together. Livers without macroscopic nodules were
33 used for western blot and but could have contained small foci in the homogenate as well. The
34 gene expression results were also reported for an exposure level of WY-14,643 that is an LD50 in
35 wild-type mice and could reflect toxicity responses rather than carcinogenic ones. The samples
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-353 DRAFT—DO NOT CITE OR QUOTE
-------
1 were also obtained at the end of the experiment (with a mix of durations of exposure) and may
2 not reflect key events in the causation of the cancer but events that are downstream.
3 These limitations notwithstanding, it is interesting that expression of p53 gene was
4 reported by Morimura et al. (2006) to be increased in PPARa humanized mice treated with
5 WY-14,643 compared to all other groups. Furthermore, of the cell cycle genes that were tested,
6 (i.e., CD-I, Cyclin-dependent Kinases 1 and 4, and c-myc) there was a slightly greater level of
7 c-myc and CD-I in control PPARa humanized mice than control wild-type mice as a baseline.
8 This could indicate that there was already increased cell cycling going on in the control PPARa
9 humanized mouse and could be related to the increased liver size. Treatment with WY-14,643
10 induced an increase in cycling genes in wild-type mice in relation to its control, but whether that
11 induction was greater than control levels for PPARa humanized mice for c-myc and CDk4 was
12 not reported by the authors. Apoptosis genes were reported to have little difference between
13 control PPARa humanized and wild-type mice but to have a greater response induced by
14 WY-14,643 in humanized mice forp53 andp2L There was no consistent or large change in
15 apoptosis genes in response to exposure to WY-14,643 in wild-type mice. The increased
16 response of apoptosis genes in PPARa humanized mice without corresponding tumor formation
17 does not support that response as a key event in the MO A (neither does the lack of response from
18 WY-14,643 in wild-type mice). For genes associated with PPARa peroxisomal (Acox),
19 microsomal (CYP4a) mitochondrial fatty oxidation (Mead) and especially malic enzyme, there
20 was a greater response in wild-type than PPARa humanized mouse after treatment with
21 WY-14,643. However, this is somewhat in contrast to Cheung et al. (2004), who reported
22 increased in some genes encoding peroxisomal, mitochondrial, and microsomal fatty oxidation
23 enzymes in the PPARa humanized mouse after treatment with WY-14,643.
24 The results reported by Yang et al. (2007b) use another type of "humanized" mouse to
25 study PPARa effects. Yang et al. (2007b) used a PPARa humanized transgenic mouse on a
26 PPAR -/- background that has the complete human PPARa (hPPARa) gene on a PAC genomic
27 clone, introduced onto the mouse PPARa-null background and express hPPARa not only in the
28 liver but also in other tissues. Mice were administered WY-14,643 or Fenofibrate [0.1% or 0.2%
29 (w/w)]. The authors show a figure representing expression of the hPPARa for two mice with the
30 tissue used for the genotyping exhibiting great variation in expression between the two cloned
31 mice as indicated by intensity of staining. The authors state that in agreement with mRNA
32 expression, hPPARa protein was highly expressed in the liver of hPPARaPAC mice to an extent
33 similar to the mPPARa in wild-type mice. They report that following two weeks of Fenofibrate
34 treatment, a robust induction of mRNA expression of genes encoding enzymes responsible for
35 peroxisomal (Acox), mitochondrial (MCAD and LCAD), microsomal (CYP4A) and cytosolic
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 54 DRAFT—DO NOT CITE OR QUOTE
-------
1 (ACOT) fatty acid metabolism were found in liver, kidney and heart of both wild-type and
2 hPPARaPAC mice indicating that hPPARa functions in the same manner as mPPARa to regulate
3 fatty acid metabolism and associated genes. However, the authors did no measures in
4 Fenofibrate treated animals, only WY-14,643, raising the issue of whether there was a difference
5 in the relative mRNA expression of genes for ACOX etc. and lipids between the two
6 peroxisomal proliferator treatments. The expression of enzymes associated with PPARa
7 induction was presented only for mice treated with Fenofibrate. However, the lipids results were
8 presented only for mice treated with WY-14,643. Therefore, it cannot be established that these
9 two agonists give the same response for both parameters. Also for the enzymes, the relative
10 expressions compared to wild-type controls, the absolute expression, and variation between
11 animals is not reported. It appears that the peroxisomal enzyme induction by Fenofibrate is the
12 same in the wild-type and transgenic mice. However, in Figure 4 of the paper the mice treated
13 with WY-14,643 instead of Fenofibrate were presented for the peroxisomal membrane protein 70
14 (PMP70) in total liver protein gel. There appears to be more PMP70 in the transgenic mice than
15 wild-type mice as a baseline. The PMP70 appeared to be similar after WY-14,643 treatment.
16 However, only one gel was given and no other quantitation was given by the authors.
17 The authors state that "in addition WY-14,643 and Fenofibrate treatment produced
18 similar effect to the liver specific humanized PPARa mouse line (Cheung et al 2004)."
19 However, the results were not the same between Fenofibrate and WY-14,643 and the mouse line
20 used by Cheung et al. had background differences in response and pathology. In one figure in
21 the paper there appears to be a difference in background level of serum total triglyceride between
22 the wild-type and hPPARaPAC mice that the authors do not note. The power of using such few
23 mice does not help discern any significant differences in background level of triglycerides. The
24 authors note that WY-14,643 treatment also resulted in decreased serum triglycerides levels in
25 hPPARaPAC mice consistent with the induction of expression of genes encoding fatty acid
26 metabolism and that the hypolipidemic effects of fibrates are generally explained by increased
27 expression of LPL and decreased expression of apolipoprotein C- III (Apo C-III) (Auwerx et al.,
28 1996). However, the alteration of these genes by WY-14,643 treatment was only observed in
29 wild-type mice and not in hPPARaPAC mice suggesting that the hypolipidemic effect observed in
30 hPPARaPAC mice are not through LPL and APO C-III. The authors do not note that there could
31 be a difference in the regulation of these pathways by the transgene rather than how the normal
32 gene is regulated and the pathways it affects. The rationale for examining this question with
33 WY-14,643 treatment rather than with Fenofibrate treatment is not addressed by the authors,
34 especially since the other "markers" of peroxisomal gene induction appear to be affected by
35 Fenofibrate in the wild-type and hPPARaPAC mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-355 DRAFT—DO NOT CITE OR QUOTE
-------
1 Hepatomegaly was reported to be observed in the hPPARaPAC mice following two weeks
2 of WY-14,643 treatment as revealed by the increase liver to body weight ratio compared to
3 untreated hPPARaPAC mice but to be markedly lower when compared to wild-type mice under
4 the same treatment. Histologically, the livers of the wild-type mice treated with WY-14,643
5 were hypertrophic with clear eosinophilic regions. These phenotypic effects were observed in
6 both wild-type and hPPARaPAC mice. The percent liver/body weight was reported to increase
7 from -4% in wild-type mice to -9% after WY-14,643 treatment and from -4% in hPPARaPAC to
8 little less that 6% after treatment with WY-14,643. In wild-type mice treated with WY-14,643
9 the labeling index was 21.8% compared with 1.1% in untreated wild-type controls. In
10 hPPARaPAC mice, WY-14,643 treatment was reported to give an average labeling index of 1.0%
11 compared with 0.8% in the untreated control hPPARaPAC mice. Treatment with WY-14,643
12 treatment was reported to result in a marked induction in the expression of CDK4 and cyclin Dl
13 in the livers of wild-type mice but to be unaffected hPPARaPAC mice treated with WY-14,643.
14 These data were reported to be in agreement with the liver-specific PPARa-humanized mice that
15 showed not increase in incorporation of BrdU into hepatocytes upon treatment with WY-14,643
16 (Cheung et al., 2004) and further confirmed that activation of hPPARa dose not induce
17 hepatocyte proliferation. However, the authors present a figure as an example with one liver
18 each with no quantitation given by the authors for BrdU incorporation. It is not clear whether the
19 pictures were taken from the same area of the liver or how representative they are. The numbers
20 of mice were never reported for the labeling index. The data presented do suggest that there was
21 hypertrophy and hepatomegaly in the humanized mice and but not proliferation in this particular
22 WY,-14,643 model. Of interest would be investigation of proliferation by other peroxisome
23 proliferators besides WY-14,643 at this necrogenic dose as it is WY-14,643 that is the anomaly
24 to continue to induce proliferation or DNA synthesis at 2 weeks. The photomicrographs
25 presented by the authors are so small and at such low magnification that little detail can be
26 discerned from them. There are no portal triads or central veins to orient the reader as to what
27 region of the liver has been affected and where if any there would be hepatocellular
28 vacuolization.
29 To determine whether peroxisome proliferation occurred in the hPPARaPAC mice upon
30 administration of PPs, Yang et al. (2007b) examined by Western Blot analysis the protein levels
31 of the major PMP70 a marker of peroxisome proliferation). After two weeks treatment of
32 1,000 ppm WY-14,643, induction of PMP70 was reported to be observed in the wild-type mice
33 as well as in hPPARaPAC mice. The authors suggested that this result indicates that peroxisomal
34 proliferator treatment induced peroxisomal proliferation in hPPARaPAC mice. The results of this
35 study indicate that hepatomegaly and peroxisome proliferation occur in this humanized mouse
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-356 DRAFT—DO NOT CITE OR QUOTE
-------
1 model when treated with large concentrations of WY-14,643. Thus, these results are inconsistent
2 with claims that peroxisome proliferators cannot cause hepatomegaly or peroxisome proliferation
3 in humans or that humans are refractory to these effects. Like the lipid effects, they suggest a
4 broader spectrum of effects may occur in humans and decreases the specificity of these effects as
5 species specific. However, due to the model compound being WY-14,643 at a necrogenic dose
6 of 1,000 ppm, the effect may not be seen in humans using the lower potency peroxisome
7 proliferators. It would have been useful for this study to include an examination of these effects
8 with Fenofibrate rather than WY-14,643 and then attempting to extrapolate such effects to other
9 peroxisome proliferators. The authors often attribute the effects of peroxisome proliferators to
10 those reactions induced by WY-14,643 and do not acknowledge that the changes induced by
11 WY-14,643 may be different. This is especially true in regards to hepatocellular DNA synthesis
12 in which other peroxisome proliferators can cause liver tumors without the sustained
13 proliferation that WY-14,643 induces, especially at a necrogenic dose.
14 Yang et al. (2007b) report the results of induction of various genes by WY-14,643 in
15 wild-type and hPPARaPAC mice by microarray analysis followed by confirmation and
16 quantitation by qPCR and report that more genes were induced by WY-14,643 in wild-type mice
17 than in hPPARaPAC mice. They report that
18
19 importantly, the oncogene c-myc was not induced in hPPARaPAC mice.
20 Moreover, genes encoding cell surface proteins such as Anxa2, CD39, CD63,
21 Ly6D, and CD24a, and several other genes such as Cidea, Cidec, DhrsS and
22 Hsdllb were also not induced in hPPARaPAC mice. Interestingly, Sult2al was
23 only induced in hPPARaPAC mice and not in WT mice; this gene is also induced
24 in human hepatocytes by PP (Fang et al., 2005). The regulation of several of
25 these genes has previously been demonstrated through a PPARa-dependent
26 mechanism. Additional studies will be necessary to fully explore the molecular
27 regulatory mechanism and the functional implication associated with these
28 differently regulated genes.
29
30 The authors do not indicate the context of how the mice were treated, whether these are pooled
31 results, and when the samples were taken. It is assumed to be whole liver. As stated in Section
32 E.3.2.2 above, there are several limitations for interpretations of the results such as those
33 presented by Yang et al. (2007b) which include the lack of phenotypic anchoring for the results.
34 The authors have shown changes from whole liver and have listed changes in genes between
35 wild-type and humanized mice on a PPAR -/- background that in itself with bring about changes
36 in gene expression. The authors acknowledge difficulties in determining what their reported
37 gene changes mean.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-357 DRAFT—DO NOT CITE OR QUOTE
-------
1 Yang et al. (2007b) report that "activation of PPARa alters hepatic miRNA expression
2 (Shah et al., 2007)." They report that let-7C, a miRNA critical in cell growth and shown to
3 target c-myc, was inhibited by WY-14,643 treatment in wild-type mice and that the expression
4 levels of both pri-let-7C and mature let-7C were significantly higher in hPPARaPAC mice
5 compared to wild-type mice. Treatment with WY-14,643 was reported to decrease the
6 expression of Pri-let-7C and mature let-7C in wild-type mice but in hPPARaPAC mice. The
7 authors note that
8
9 in addition, the induction of c-myc by WY-14,643 treatment in wild type mice did
10 not occur in WY-14,643 treated hPPARaPAC mice. This is in agreement with the
11 previous observation in liver-specific humanized PPARa (Shah et al 2007) and
12 further indicates the activation of human PPARa does not cause a change in
13 hepatic miRNA and c-myc gene expression.
14
15 A qPCR analysis of pri-let-7C following 2 weeks WY-14,632 treatment was reported for wild-
16 type and hPPARaPAC mice (n = 3-4). There appeared to be -20 times more let-7C expression in
17 hPPARaPAC mice than control wild mice as a baseline. The gel given by the authors showed a
18 very small difference in wild-type mice in let-7C northern blot analysis between a control wild-
19 type and WY-14,643-treated wild-type mouse. There appeared to be no difference in the
20 hPPARaPAC mice between control and WY-14,643 treatment and a larger stained area than the
21 control wild-type mice. The relative c-Muc expression between the hPPARaPAC mice and wild-
22 type control mice did not correlate with changes in let-7C expression. Thus, the amount of
23 decrease by treatment with WY-14,632 in wild-type mice appeared to be extremely small
24 compared to the much greater baseline expression in the hPPARaPAC mice. The change brought
25 by WY-14,632 treatment in wild-type mice was a small change compared to the 20-fold greater
26 baseline expression in the hPPARaPAC mice. The authors stated that the expression of the c-Myc
27 regulator was higher in the hPPARaPAC mice indicating over regulation of cell division and an
28 inability for hepatocytes to proliferate. However, their results showed that there was a greater
29 difference in regulatory baseline function of the PPAR using this paradigm and this construct.
30 Are these differences due to human PPAR or to the way PPAR was put back into PPAR -/-
31 mouse and expected to function? If the experiment included mouse PPAR put back in this way
32 on a null background, what would such an experiment show? Are these results representative of
33 the PPAR or how it is now controlled and expressed? In addition, what would the study of other
34 peroxisome proliferators besides WY-14,643 show in regard to changes in miRNA. Are these
35 results reflective of a just the transient effect that is prolonged in a special case? As discussed in
36 Section E.3.2.2 there are issues with microarray data in addition to the newly emerging field of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-358 DRAFT—DO NOT CITE OR QUOTE
-------
1 miRNA arrays, which include phenotypic anchoring and whether they are from whole liver or
2 pooled samples. The results given in this report are for relative Let-7C expression given and not
3 absolute values. The changes in baseline Let-7C expression between the wild-type and the
4 hPPARaPAC mice did not correlate with the magnitude of difference in northern blot analysis and
5 did not correlate at all with c-myc expression reported in this study. Thus, a direct correlation
6 between the effect of Let-7C expression and function and effects from WY-14,643 was not
7 supported. The relative expression was reported but the variation of baseline expression of the
8 "PPAR controlled genes" was not. Given that one of the first figures reported a large difference
9 between animals in expression of the human PPAR gene in the transgenic animals, how did this
10 difference affect the results given here as relative changes downstream?
11 Yang et al. (2007b) conclude that the hPPARaPAC mice represent the most relevant model
12 for humans since, the tissue distribution of PPARa is similar to that observed in wild-type mice
13 and the hPPARa in hPPARaPAC mice is under regulation of its native promoter. Indeed up-
14 regulation of hepatic mPPARa in wild-type mice by fasting was mirrored by the hPPARa in
15 hPPARaPAC mice. However, there was no demonstration that the artificial chromosome that is
16 replicating along with other DNA is controlled sterically by the same control since it is not on
17 the mouse genome in the same place as the native PPAR. There is also not a demonstration of
18 how stable the baseline of PPAR DNA expression is in this mouse model—does it vary as much
19 or more than native PPAR between mice? The authors state that
20
21 induction of PPARa target genes for fatty acid metabolism and a decrease in
22 serum triglycerides by PP in hPPARaPAC mice indicates that hPPARa is
23 functional in the mouse environment with respects to regulation of fatty acid
24 metabolism. This is in agreement with the liver-specific PPARa humanized mice
25 that also exhibit these responses (Cheung et al., 2004). Indeed the DNA binding
26 domain of hPPARa is 100% homologous with that of the mouse suggesting that
27 both bind to the same PPRE binding site in the promoter region of target genes.
28 Transfection of hPPAR into murine hepatocytes increased PPs induced
29 peroxisome proliferation related effects (Macdonald et al., 1999). These results
30 suggest that hPPARa and mPPARa do not differ in induction of target genes with
31 known PPRE.
32
33 However, replacement with human PPAR in the Cheung et al. model is not sufficient to prevent
34 the same types of toxicity as seen with PPAR knockouts on the hepatocytes such as steatosis.
3 5 Yang et al. (2007b) note that
36
37 the increased LPL and decreased expression of apo C-III are proposed to explain
38 the hypolipidemic effects of PPS (Auwerx et al., 1996). However, hPPARaPAC
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-359 DRAFT—DO NOT CITE OR QUOTE
-------
1 mice treated with PP exhibit lowered serum triglycerides without alteration of the
2 expression of LPL and apo C-III. This indicates the hypolipidemic effects in
3 rodents are mediated via other molecular regulatory mechanisms. It is also
4 suggested that the activation of PPARa by PPs stimulates hepatic fatty acid
5 oxidation and thereby diminishing their incorporation into triglycerides and
6 secretion of VLDL (Froyland et al., 1997). Consistent with this idea, a robust
7 induction of the genes encoding enzymes for fatty acid oxidation by PP in
8 hPPARaPAC mice were observed. Thus, the exact mechanism by which PPs exert
9 their hypolipidemic effects needs reexamination.
10
11 However, the use of two different peroxisome proliferators (i.e., WY-14,643 and Fenofibrate) for
12 two types of effects (peroxisomal and lipid) may be the cause of some paradoxes here in terms of
13 MO A for lipid effects. The baseline differences in the hPPARaPAC mice for serum total
14 triglycerides was not explored by these authors and the small number of animals used make
15 conclusions difficult about the magnitude of difference. The differences in baseline expression
16 for LPL are not discernable in the graphic representation of the results.
17 Yang et al. (2007b) note that
18
19 on the other hand, the difference in the affinity of ligands for the human and
20 mouse PPARa receptor was proposed to account for the species difference. The
21 ligand binding domain of hPPARa is 94% homologous with that of the mouse. In
22 vitro transactivation assays have previously shown that WY has a higher affinity
23 for rodent PPARa than human PPARa, while Fenofibrate has similar affinity for
24 rodent and human PPARa (Shearer and Hoekstra, 2003; Sher et al., 1993). In the
25 present study WY and Fenofibrate exhibit the same capacity to induce known
26 PPARa target genes in the liver, kidney and heart in both wild-type and
27 hPPARaPAC mice.
28
29 The statement by the authors that Fenofibrate and WY-14,643 had the same affinity "as shown
30 by this study" is not correct. The two treatments were not studied for the same enzymes or genes
31 in the data reported in the study. Both WY-14,643 and Fenofibrate can induce PPARa targets
32 but it was not shown to the same extent. Yang et al. (2007b) state that
33
34 This is in agreement with the liver-specific PPARa humanized mice that also
35 exhibit a similar capacity to induce PPARa target genes in liver by WY and
36 Fenofibrate (Cheung et al., 2004). Thus, the ligand affinity difference between
37 mouse and human PPARa may not be critical under the conditions of these
38 studies.
39
40 Alternatively, these results could reflect that these studies were conducted with two different
41 agonists with different affinities and responses due to receptor activation.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-360 DRAFT—DO NOT CITE OR QUOTE
-------
1 Finally, a useful comparison to make are the differences between wild-type mice,
2 PPARa -/- mice that serve as the background for the transgenic human mouse models, and both
3 transgenic models. The small and variable number of animals examined in these studies is
4 readily apparent. The results of the Cheung et al. (2004) humanized mouse model and those
5 reported for Yang et al. (2007b) show differences in the study designs including PPARa agonists
6 studied for particular effects and results reported for similar treatments (see Table E-18).
7 As shown above, the effect on the PPARa -/- by the knockout included decreased
8 triglyceride levels and slightly increased liver weight. Although treatment with WY-14,643 and
9 Fenofibrate were reported to decrease triglyceride levels in wild-type mice, paradoxically so did
10 knocking out the receptor. Exposures to WY-14,643 appeared to induce a slight increase and
11 Fenofibrate a slight decrease in triglyceride levels in PPARa -/- mice but the variability of
12 response and small number of animals in the experiments limited the ability to discern a
13 quantitative difference in the treatments. In the study by Cheung et al. (2004) it appears that the
14 insertion of humanized PPARa restored the baseline and treatment responses for triglyceride
15 levels. Overall, the results reported by Yang et al. (2007b) appeared to show a lower level of
16 triglycerides in control wild-type mice that was similar in magnitude to the treatment effect
17 reported by Fenofibrate by Cheung et al. (2004). However, there also appeared to be restoration
18 of this effect in the humanized mouse model of Yang et al. (2007b). In regard to DNA
19 synthesis, both Cheung et al. (2004) and Yang et al. (2007b) only gave results for WY-14,643
20 and for different durations of exposure so they were not comparable. It appeared that -60% of
21 hepatocytes were labeled by 8 weeks of WY-14,643 treatment (Cheung et al., 2004) compared
22 to -20% after 2 weeks of exposure. Again this highlights the difference between using
23 WY-14,643 as a model for the PPARa as a class at times when almost all other PPARa agonists
24 have ceased to increase DNA synthesis or have reductions in this parameter. The background
25 changes due to the PPARa -/- knockout were not reported so that the effects of the knockout
26 could not be ascertained. It appeared that insertion of humanized PPARa did not result in
27 restoration of WY-14,643 -induced DNA synthesis. The correlation with this parameter and
28 any focal areas of necrosis were not discussed by the authors of the study. In regard to
29 hepatomegaly, Fenofibrate and WY-14,643 appeared to both give an increase in liver weight in
30 the humanized mouse model of Cheung et al. (2004) with little effect in the knockout mouse.
31 For Fenofibrate there was little difference in liver weight gain in the wild-type mouse and that of
32 the humanized mouse model of Cheung et al. (2004). However, Fenofibrate was not tested in
33 the humanized mouse model of Yang et al. (2007b). In that model only WY-14,643 was used
34 but there was still an increase in liver weight. Thus, in terms of effects on liver weight gain and
35 triglyceride levels both models gave comparable results and appeared to indicate that insertion
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-361 DRAFT—DO NOT CITE OR QUOTE
-------
1 humanized PPARa would restore some of the effects of the knockout. However, the results
2 from both experiments highlight the need for adequate numbers of animals and other PPARa
3 agonists to be tested besides WY-14,463 at such a high dose and certainly for longer periods of
4 time to ascertain whether such manipulations will affects carcinogenicity.
5
6 E.3.4.1.4. NF-icB activation. NF-KB activation has also been proposed as a key event in the
7 induction of liver cancer through PPARa activation. As discussed in Sections E.3.2.6 and
8 E.3.4.3.3, activation of the NF-KB pathway is implicated in carcinogenesis, nonspecific for a
9 particular MOA for liver cancer, and is context dependent on its effects. Its specific actions
10 depend on the cell type and type of agent or signal that activates translocation of the complex.
11 NF-KB is not only involved in biological processes other than tumor induction, but also exhibits
12 some apparently contradictory behaviors (Perkins and Gilmore, 2006). Although many studies
13 point to a tumor-promoting function of NF-KB subunits, evidence also exists for tumor
14 suppressor functions. NF-KB actions are associated with TNF and INK among many other cell
15 signaling systems and molecules and it has functions that alter proliferation and apoptosis. NF-
16 KB activation reported in some studies may be associated with early Kupffer cell responses and
17 be associative but not key events in the carcinogenic process. However, most assays look at total
18 NF-KB expression in the whole liver and at the early periods of proliferation and apoptosis. The
19 origin of the NF-KB is crucial as to its effect in the liver. For instance, hepatocyte specific
20 deletion of IKKp increased DEN-induced hepatocarcinogenesis but a deletion of IKKp in both
21 hepatocytes and Kupffer cells however, were reported to have the opposite effect (Maeda et al.,
22 2005).
23
24 E.3.4.1.5. Phenotype as an indicator of a PPARa mode of action (MOA). As discussed
25 previously (see Sections E.3.1.5, and E.3.1.8) FAH precede both hepatocellular adenomas and
26 carcinomas in rodents and, in humans with chronic liver diseases that predispose them to
27 hepatocellular carcinomas. Striking similarities in specific changes of the cellular phenotype of
28 preneoplastic FAH are emerging in experimental and human hepatocarcinogenesis, irrespective
29 of whether this was elicited by chemicals, hormones, radiation, viruses, or, in animal models, by
30 transgenic oncogenes or Helicobacter hepaticus. Several authors have noted that the detection
31 of phenotypically similar FAH in various animal models and in humans prone to developing or
32 bearing hepatocellular carcinomas favors the extrapolation from data obtained in animals to
33 humans (Bannasch et al., 2003; Su and Bannasch, 2003; Bannasch et al., 2001). In regard to
34 phenotype by tincture Caldwell and Keshava (2006) state:
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-362 DRAFT—DO NOT CITE OR QUOTE
-------
Table E-18. Comparison between results for Yang et al. (2007b) and Cheung et al. (2004)a
Effect
Triglycerides
BrdU
incorporation
Wild type mice
Cheung
(n = 6-9)
Control 145 mg/mL
0. 1% WY-14,643 60 mg/mL
(2 wks)
0.2% Fenofibrate 85 mg/mL
(2 wks)
Yang
(« = 4-6)
Control 95 mg/mL
0.1 % WY-14,643 55 mg/mL
(2wks)
Cheung
(« = 5)
Control 1.6%
0.1% WY-14,643 57.9%
(8 wks)
Yang
(« = 4-6)
Control 1.1%
0.1% WY-14,643 21.8%
(2 wks)
PPAR -/- knockout mice
Cheung
(n = 6-9)
Control 100 mg/mL
0. 1% WY-14,643 1 15 mg/mL
(2 wks)
0.2% Fenofibrate 85 mg/mL
(2 wks)
Not done
Humanized mice (liver only)
Cheung
(n = 6-9)
Control 175 mg/mL
0.1%WY-14,643 60 mg/mL
(2 wks)
0.2% Fenofibrate 85 mg/mL
(2 wks)
Cheung
(« = 5)
Control 1.6%
0.1% WY-14,643 2.8%
(8 wks)
Humanized PAC mice
Yang
(n = 4-6)
Control 120 mg/mL
0.1%WY-14,643 75 mg/mL
(2 wks)
Yang
(n = 4-6)
Control 0.8%
0.1% WY-14,643 1.0%
(2 wks)
-------
Table E 18. Comparison between results for Yang et al. (2007b) and Cheung et al. (2004) (continued)
Effect
Hepatomegaly1"
(% liver body
weight ratio)
Wild type mice
Cheung
(n = 5-9)
Control 4%
0.1%WY-14,643 11%
(2 wks)
0.2% Fenofibrate 8.5%
(2 wks)
Yang
(» = 4-6)
Control 4%
0.1%WY-14,643 9%
(2 wks)
PPAR -/- knockout mice
Cheung
(n = 5-9)
Control 5%
0.1%WY-14,643 5%
(2 wks)
0.2% Fenofibrate 5.5%
(2 wks)
Humanized mice (liver only)
Cheung
(n = 5-9)
Control 4.5%
0.1%WY-14,643 7%
(2 wks)
0.2% Fenofibrate 7%
(2 wks)
Humanized PAC mice
Yang
(n = 4-6)
Control 4%
0.1% WY 6%
(2 wks)
aThe ages of the humanized knockout mice are not given for Cheung et al. (2004) but are 8-10 weeks for Yang et al. (2007b).
Percentages are approximate values extrapolated from figures for hepatomegaly.
-------
1 In addition, the term "basophilic" in describing preneoplastic foci or tumors can
2 be misleading. The different types of FAH have been related to three main
3 preneoplastic hepatocellular lineages: 1) the glycogenotic-basophilic cell lineage,
4 2) its xenomorphic-tigroid cell variant, and 3) the amphophilic-basophilic cell
5 lineage. Specific changes of the cellular phenotype of the first two lineages of
6 FAHs are similar in experimental and human hepatocarcinogenesis, irrespective
7 of whether they were elicited by DNA-reactive chemicals, hormones, radiation,
8 viruses, transgenic oncogenes and local hyperinsulinism as described by the first
9 two FAHs and this similarity favors extrapolation from data obtained in animals
10 to humans (Bannasch et al., 2003; Su and Bannasch, 2003; Bannasch et al.,
11 2001). In contrast, the amphophilic cell lineage of hepatocarcinogenesis has
12 been observed mainly after exposure of rodents to peroxisome proliferators or to
13 hepadnaviridae (Bannasch et al., 2001).
14
15 Bannasch (1996) describes "amphophilic" FAH and tumors induced by
16 peroxisome proliferators to maintain the phenotype as the foci progress to
17 tumors. They are glycogen poor from the start with increased numbers of
18 mitochondria, peroxisomes and ribosomes. The author further states that the
19 "homogenous basophilic" descriptions by others of foci induced by WY are
20 really amphophilic. Agents other than peroxisome proliferators can induce
21 "acidophilic" or "eosinophilic" (due to increased smooth endoplasmic reticulum)
22 or glycognotic foci which tend to progress to basophilic stages (due to increased
23 ribosomes).
24
25 Tumors and foci induced by peroxisome proliferators have been suggested to
26 have a phenotype of increased mitochondrial proliferation and mitochondrial
27 enzymes (thyromimetic rather than insulinomimetic) (Keshava and Caldwell,
28 2006).
29
30 Tumors from peroxisome proliferators in Kraupp-Grasl et al. (1990) and
31 Grasl-Kraupp et al. (1993) for rat liver tumors were characterized as weakly basophilic with
32 some eosinophilia and as similar to the description given by Bannasch et al as amphophilic.
33 However, a number of recent studies indicate that other "classic" peroxisome proliferators may
34 have a different phenotype than has been attributed to the class through studies of WY-14,643.
35 A recent study of DEHP, another peroxisome proliferator assumed to induce liver tumors
36 through activation of the PPARa receptor, reported the majority of liver FAH to be of the first
37 two types after a lifetime of exposure to DEHP with a dose-related tendency for increased
38 numbers of amphophilic FAHs in rats (Voss et al., 2005). As stated previously, the MO A of
39 DEHP-induced liver tumors in mice also appears not to be dependent on PPARa activation.
40 Michel et al. (2007) report the phenotype of tumors and foci in rats treated with clofibric
41 acid at a very large dose (5,000 ppm for 20 months) and note that in controls the first type of
42 foci to appear was tigroid on Day 264 and their incidence increased with time representing the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-365 DRAFT—DO NOT CITE OR QUOTE
-------
1 most abundant type in this group. They report no adenomas or carcinomas after up to 607 days
2 after giving saline injection in the control animals. DEN treatment was examined up to 377
3 days only with tigroid, eosinophilic and clear cell foci observed at that time. Clofibric acid was
4 examined up to 607 days with tigroid and clear cell foci reported to be the first to appear on Day
5 264 no other foci class. By Day 377, there were tigroid, eosinophilic and clear cell foci but no
6 basophilic foci reported with clofibric acid treatment and, although only a few animals were
7 examined, 2/5 had adenomas but not carcinomas. By Day 524 all types of foci were seen
8 (including basophilic for the first time) and there were adenomas and carcinomas in 2/5 animals.
9 By 607 days a similar pattern was observed without adenomas but 3/6 animals showing
10 carcinomas. Although the number of animals examined is very small, these results indicate that
11 clofibric acid was not inducing primarily "basophilic foci" as reported for peroxisome
12 proliferators but the first foci are tigroid and clear cell foci. Basophilic foci did not appear until
13 Day 524 similar to control values for foci development and distribution. However, unlike
14 controls, clofibric acid induced eosinophilic and clear cell foci earlier. This is inconsistent with
15 the phenotype ascribed to peroxisome proliferators as exemplified by WY-14,643.
16 In regard to GST-u and y-transpeptidase (GGT), Rao et al. (1986) fed 2 male F344 rats a
17 diet of 0.1% WY-14,643 for 19 months or 3 F344 rats 0.025% Ciprofibrate for 15-19 months
18 and reported "altered areas,"(AA) "neoplastic nodules" (NN), and hepatocellular carcinomas
19 (HCC). For WY-14,643 treatment 107 AA, 75 NN, and 5 HCC, and for Ciprofibrate treatment
20 107 AA, 27 NN, and 16 HCC were identified. In the WY-14,643-treated rats, HCC, and NN
21 were both GGT and GST-7i negative (96-100%) with 87% of AA was negative for both. In
22 Ciprofibrate-treated rats NN and HCC were negative for both markers (95%) but only 46% of
23 AA were negative for both markers. Thus, a different pattern for tumor phenotype was reported
24 for WY-14,643 and another peroxisome proliferator, Ciprofibrate, in this study as well.
25 In addition, GGT phenotype is reported not to be specific to weakly basophilic foci.
26 GGT staining was reported to be negative in eosinophilic tumors after initiation and promotion.
27 Kraupp-Grasl et al. (1990) note differences among PPARa agonists in their ability to promote
28 tumors and suggest they not necessarily be considered a uniform group. Caldwell and Keshava
29 (2006) suggest that the reports of a simple designation of "basophilic" is not enough to associate
30 a foci as caused by peroxisome proliferators (Bannasch, 1996; Grasl-Kraupp et al.,1993;
31 Kraupp-Grasl et al., 1990). Increased basophilia of tumors and increased numbers of
32 carcinomas is consistent with the progressive basophilia described by Bannasch (1996), as many
33 adenomas progress to carcinomas.
34 It should be noted that the amphophilic foci and tumors described by Bannasch et al.
35 were primarily studied in rats. Morimura et al. (2006) noted that WY-14,643 induced diffusely
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-366 DRAFT—DO NOT CITE OR QUOTE
-------
1 basophilic tumors in mice and therefore, identified the WY-14,643 tumors in a way consistent
2 with the descriptions of amphophilic tumors by Bannasch et al. The tumor induced by
3 WY-14,643 in their humanized mouse was reported to be similar to those arising spontaneously
4 in the mouse. However, the mouse response could differ from the rat, especially for PPARa
5 agonists other than WY-14,643.
6 H-ras activation and mutation studies have attempted to assign a pattern to peroxisome
7 proliferator-induced tumors as noted in Section E.2.3.3.2, above. However, also as noted in
8 Section E.2.3.3.2, the genetic background of the mice used, the dose of carcinogen and the stage
9 of progression of "lesions" (i.e., foci vs. adenomas vs. carcinomas) may affect the number of
10 activated H-ras containing tumors that develop. Fox et al. (1990) note that tumors induced by
11 Ciprofibrate (0.0125% diet, 2 years) had a much lower frequency of H-ras gene activation than
12 those that arose spontaneously (2-year bioassays of control animals) or induced with the
13 "genotoxic" carcinogen benzidine-2 HC1 (120 ppm, drinking H2O, 1 year) and that the
14 Ciprofibrate-induced tumors were reported to be more eosinophilic as were the surrounding
15 normal hepatocytes than spontaneously occurring tumors. Anna et al. (1994) also stated that
16 mice treated with Ciprofibrate had a markedly lower frequency of tumors with activated H-ras
17 but that the spectrum of mutations in tumors was similar those in "spontaneous tumors."
18 Hegi et al. (1993) tested Ciprofibrate-induced tumors from Fox et al. (1990) in the NIH3T3
19 cotransfection-nude mouse tumorigenicity assay and concluded that ras protooncogene
20 activation, were not frequent events in Ciprofibrate-induced tumors and that spontaneous tumors
21 were not promoted with it. Stanley et al. (1994) studied the effect of MCP, a peroxisome
22 proliferator, in B6C3F1 (relatively sensitive) and C57BL/10J (relatively resistant) mice for
23 H-ras codon 61-point mutations in MCP-induced liver tumors (hepatocellular adenomas and
24 carcinomas). In the B6C3F1 mice, -24% of MCP-induced tumors had codon 61 mutations and
25 for C57BL/10J mice -13%. The findings of an increased frequency of H-ras mutation in
26 carcinomas compared to adenomas in both strains of mice is suggestive that these mutations
27 were related to stage of progression. Thus, in mice, the phenotype of tumors did not appear to
28 be readily distinguishable from spontaneous tumors based on tincture for peroxisome
29 proliferators other than WY-14,643, but did have more of a signature in terms of H-ras mutation
30 and activation.
31 The expression of c-Jun has been used to discern TCE tumors from those of its
32 metabolites. However, as pointed out by Caldwell and Keshava (2006), although Bull et al.
33 (2004) have suggested that the negative expression of c-jun in TCA-induced tumors may be
34 consistent with a characteristic phenotype shown in general by peroxisome proliferators as a
35 class, there is no supporting evidence of this. While increased mitochondrial proliferation and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-367 DRAFT—DO NOT CITE OR QUOTE
-------
1 mitochondrial enzymes (thyromimetic rather than insulinomimetic) properties have been
2 ascribed to peroxisome proliferator-induced tumors, the studies cited in Bull et al. (2004) have
3 not examined TCA-induced tumors for these properties.
4
5 E.3.4.1.6. Human relevance. In its framework for making conclusions about human
6 relevance, the U.S. EPA Cancer Guidelines (U.S. EPA, 2005) asks that critical similarities and
7 differences between test animals and humans be identified. Humans possess PPARa at sufficient
8 levels to mediate the human hypolipidemic response to peroxisome-proliferating fibrate drugs.
9 Fenofibrate and Ciprofibrate induce treatment related increases in liver weight, hypertrophy,
10 numbers of peroxi somes, numbers of mitochondria, and smooth endoplasmic reticulum in
11 cynomologous monkeys at 15 days of exposure (Hoivik et al., 2004). Given the species
12 difference in the ability to respond to a mitogenic stimulus such as partial hepatectomy (see
13 Section E.3.3) lack of hepatocellular DNA synthesis at this time point is not unexpected and, as
14 Rusyn et al. (2006) note, examination at differing time point may produce differing results. It is
15 therefore, generally acknowledged that "a point in the rat and mouse key events cascade where
16 the pathway is biologically precluded in humans in principle cannot be identified."(Klaunig et
17 al., 2003; NAS, 2006). Thus, from a qualitative standpoint, the effects described above are
18 plausible in humans.
19 As for quantitative differences, there are two key issues. First, as stated in the Cancer
20 Guidelines, when considering human relevance, "Any information suggesting quantitative
21 differences between animals and humans is flagged for consideration in the dose-response
22 assessment." Therefore, while Klaunig et al. (2003) and NAS (2006) go on to suggest that
23 "this mode of action is not likely to occur in humans based on differences in several key steps
24 when taking into consideration kinetic and dynamic factors," under the Cancer Guidelines,
25 such "kinetic and dynamic factors" need to be made explicit in the dose-response assessment,
26 and should not be part of the qualitative characterization of hazard. Second, the discussion
27 above points to the lack of evidence supporting associations between the postulated events and
28 carcinogenic potency. Thus, because interspecies differences in carcinogenicity do not appear
29 to be associated with interspecies differences in postulated events, they do not provide reliable
30 metrics with which to make inferences about relative human sensitivity.
31
32 E.3.4.2. Other Trichloroethylene (TCE) Metabolite Effects That May Contribute to its
33 Hepatocarcinogenicity
34 While the focus of most studies of TCA has been its effects on peroxisomal proliferation,
35 DCA has been investigated for a variety of effects that are also observed either in early stages of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-368 DRAFT—DO NOT CITE OR QUOTE
-------
1 oncogenesis (glycogen deposition) or conditions that predispose patients to liver cancer. Some
2 studies have examined microarray profiles in attempt to study the MOA or TCE (see
3 Section E.3.2.2 for caveats regarding such approaches). Caldwell and Keshava have provided a
4 review of these studies, which is provided below.
5
6 E.3.4.2.1. DCA effects and glycogen accumulation correlations with cancer. As noted
7 previously, DCA administration has been reported to increase the observable amount of
8 glycogen in mouse liver via light microscopy and, although to not be primarily responsible
9 for DCA-induced liver mass increases, to be increase whole liver glycogen as much by 50%
10 (Kato-Weinstein et al., 2001). Given that TCE and DCA tumor phenotypes indicate a role for
11 DCA in TCE hepatocarcinogenicity (see Section E.2.3.3.2, above), Caldwell and Keshava (2006)
12 described the correlations with effects induced by DCA that have been associated with
13 hepatocarcinogenicity.
14
15 A number of studies suggest DCA-induced liver cancer may be linked to its
16 effects on the cytosolic enzyme glutathione (GST)-S-transferase-zeta. GST-zeta
17 is also known as maleylacetoacetate isomerase and is part of the tyrosine
18 catabolism pathway whose disruption in type 1 hereditary tyrosinemia has been
19 linked to increased liver cancer risk in humans. GST-zeta metabolizes
20 maleylacetoacetate (MAA) to fumarylacetoacetate (FAA) which displays
21 apoptogenic, mutagenic, aneugenic, and mitogenic activities (Bergeron et al.,
22 2003; Jorquera and Tanguay, 2001; Kim et al., 2000). Increased cancer risk has
23 been suggested to result from FAA and MAA accumulation (Tanquary et al.
24 1996). Cornett et al. (1999) reported DCA exposure in rats increased
25 accumulation of maleylacetone (a spontaneous decarboxylation product of
26 MAA), suggesting MAA accumulation. Ammini et al. (2003) report depletion of
27 the GST-zeta to be exclusively a post-transcriptional event with genetic ablation
28 of GST-zeta causing FAA and MAA accumulation in mice. Schultz et al. (2002)
29 report that elimination of DCA is controlled by liver metabolism via GST-zeta in
30 mice, and that DCA also inhibits the enzyme (and thus its own elimination) with
31 young mice being the most sensitive to this inhibition. On the other hand, older
32 mice (60 weeks) had a decreased capacity to excrete and metabolize DCA in
33 comparison with younger ones. The authors suggest that exogenous factors that
34 deplete or reduce GST-zeta will decrease DCA elimination and may increase its
35 carcinogenic potency. They also suggest that, due to suicide inactivation of
36 GST-zeta, an assumption of linear kinetics can lead to an underestimation of the
37 internal dose of DCA at high exposure rates. In humans, GST-zeta has been
38 reported to be inhibited by DCA and to be polymorphic (Tzeng et al 2000;
39 Blackburn et al., 2001, 2000). Board et al. (2001) report one variant to have
40 significantly higher activity with DCA as a substrate than other GST zeta
41 isoforms, which could affect DCA susceptibility.
42
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-369 DRAFT—DO NOT CITE OR QUOTE
-------
1 Individuals with glycogen storage disease or with poorly controlled diabetes have
2 excessive storage of glycogen in their livers (glycogenosis) and increased risk of
3 liver cancer (LaVecchia et, 1994; Adami et al., 1996; Wideroffet al., 1997;
4 Rake et al., 2002). In an animal model where hepatocytes are exposed to a local
5 hyperinsulinemia from transplanted islets of Langerhans and the remaining tissue
6 is hypoinsulinemic, insulin induces alterations that resemble preneoplastic foci of
7 altered hepatocytes (FAH) and develop into hepatocellular tumors in later stages
8 of carcinogenesis (Evert et al., 2003). A number of studies have reported
9 suppression of apoptosis, decreases in insulin, and glycogenosis in mice liver by
10 DCA at levels that also induce liver tumors (Bull, 2004; Bull et al., 2004;
11 Lingohr et al., 2001). In isolated murine hepatocytes, Lingohr et al. (2002)
12 reported DCA-induced glycogenosis was dose related, occurred at very low
13 doses (10 uM), occurred without the presence of insulin, was not affected by
14 insulin addition, was dependent on phosphatidylinositol 3-kinase (P13K)
15 activity, and was not a result of decreased glycogen breakdown. The authors
16 noted that PI3K is also known to regulate cell proliferation and apoptosis in
17 hepatocytes, and that understanding these mechanisms may be important to
18 understanding DCA-induced carcinogenesis. They also report insulin receptor
19 (IR) protein levels decreased to 30% of controls in mice liver after up to 52
20 weeks of DCA treatment. Activation of the IR is also the principal pathway by
21 which insulin stimulates glycogen synthetase (the rate limiting enzyme of
22 glycogen biosynthesis). However, in DCA-induced liver tumors IR protein was
23 elevated as well as mitogen-activated protein kinase (a downstream target protein
24 of the IR) phosphorylation. DCA-induced tumors were glycogen poor (Lingohr
25 et al., 2001). The authors suggest that normal hepatocytes down-regulate
26 insulin-signaling proteins in response to the accumulation of liver glycogen
27 caused by DCA and that the initiated cell population, which does not accumulate
28 glycogen and is promoted by DCA treatment, responds differently from normal
29 hepatocytes to the insulin-like effects of DCA.
30
31 Gene expression studies of DCA show a number of genes identified with cell
32 growth, tissue remodeling, apoptosis, cancer progression, and xenobiotic
33 metabolism to be altered in mice liver at high doses (2 g/L DCA) in drinking
34 water (Thai et al., 2001, 2003). After 4 weeks, RNA expression was altered in 4
35 known genes (alpha-1 protease inhibitor, cytochrome B5, stearoyl-CoA
36 desaturase and caboxylesterase) in two mice (Thai et al., 2001). Except for Co-A
37 desaturase, a similar pattern of gene change was reported in DCA-induced
38 tumors (10 tumors from 10 different mice) after 93 weeks. Using cDNA
39 microarray in the same mice, Thai et al. (2003) identified 24 genes with altered
40 expression, of which 15 were confirmed by Northern blot analysis after 4 weeks
41 of exposure. Of the 15 genes, 14 revealed expression suppressed two- to fivefold
42 and included: MHR 23 A, cytochrome P450 (CYP), 2C29, CYP 3A11, serum
43 paraoxonase/arylesterase 1, liver carboxylesterase, alpha-1 antitrypsin, ER p72,
44 GST-pi 1, angiogenin, vitronectin precursor, cathepsin D, plasminogen precursor
45 (contains angiostatin), prothrombin precursor and integrin alpha 3 precursor. An
46 additional gene, CYP 2A4/5, had a twofold elevation in expression. After 93
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-370 DRAFT—DO NOT CITE OR QUOTE
-------
1 weeks of treatment with 3.5 g/L DC A, Northern blot analyses of total RNA
2 isolated from DCA-induced hepatocellular carcinomas showed similar alteration
3 of expression (11 of 15). It was noted that peroxisome proliferator-activated
4 receptor (PPAR)a and IR gene expression were not changed by DCA treatment.
5 Genes involved in glycogen or lipid metabolism were not tested.
6
7 Although it has not been possible to determine directly whether DCA is produced
8 from TCE at carcinogenic levels, there is indirect evidence that DCA is formed
9 from TCE in vivo and contributes to liver tumor development. Pretreatment with
10 either DCA or TCE inhibits GST-zeta while TCA pretreatment does not (Schultz
11 et al., 2002; Bull et al., 2004). TCE treatment decreased Vmax for DCA
12 metabolism to 49% of control levels with a 1 g/kg TCE dose resembling effects
13 those of 0.05 g/L DCA (Schultz et al., 2002).
14
15 E.3.4.2.2. Genetic profiling data for Trichloroethylene (TCE): gene expression and
16 methylation status studies. Caldwell and Keshava (2006) and Keshava and Caldwell (2006)
17 report on both genetic expression studies and studies of changes in methylation status induced by
18 TCE and its metabolites (see Sections E.2.3.2 and E.2.3.3, above) as well as differences and
19 difficulties in the patterns of gene expression between differing PPARa agonists. In
20 Section E.4.2.2 (below), the effects of coexposures of DCA, TCA and Chloroform on
21 methylation status are discussed. In particular are concerns for the interpretation of studies that
22 employ pooling of data as well as interpretation of "snapshots in time of multiple gene
23 changes." For the Laughter et al. (2004) study in particular, it is not clear whether transcription
24 arrays were performed on pooled data (no data on variability between individual animals was
25 provided and the methodology section of the report is not transparently written in this regard).
26 The issue of phenotypic anchoring also arises as data on percent liver/body weight indicates
27 significant variability within TCE treatment groups, especially in PPARa-null mice. For studies
28 of gene expression using microarrays Bartosiewicz et al. (2001) used a screening analysis of
29 148 genes for xenobiotic-metabolizing enzymes, DNA repair enzymes, heat shock proteins,
30 cytokines, and housekeeping gene expression patterns in the liver in response TCE. The TCE-
31 induced gene induction was reported to be highly selective; only Hsp 25 and 86 and Cyp2a were
32 up-regulated at the highest dose tested. Collier et al. (2003) reported differentially expressed
33 mRNA transcripts in embryonic hearts from S-D rats exposed to TCE with sequences down-
34 regulated with TCE exposure appearing to be those associated with cellular housekeeping, cell
35 adhesion, and developmental processes. TCE was reported to induce up-regulated expression of
36 numerous stress-response and homeostatic genes.
37 For the Laughter et al. (2004) study, transcription profiles using macroarrays containing
38 approximately 1,200 genes were reported in response to TCE exposure. Forty-three genes were
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-371 DRAFT—DO NOT CITE OR QUOTE
-------
1 reported to be significantly altered in the TCE-treated wild-type mice and 67 genes significantly
2 altered in the TCE-treated PPARa knockout mice. Out of the 43 genes expressed in wild-type
3 mice upon TCE exposure, 40 genes were reported by the authors to be dependent on PPARa and
4 included genes for CYP4al2, epidermal growth factor receptor, and additional genes involved in
5 cell growth. However, the interpretation of this information is difficult because in general,
6 PPARa knockout mice have been reported to be more sensitive to a number of hepatotoxins
7 partly because of defects in the ability to effectively repair tissue damage in the liver
8 (Shankar et al., 2003; Mehendale, 2000) and because a comparison of gene expression profiles
9 between controls (wild-type and PPARa knockout) were not reported.
10 As stated previously, knockout mice in this study also responded to TCE exposure with
11 increased liver weight, had increased background liver weights, and also had higher baseline
12 levels of hepatocyte proliferation than wild-type mice. Nakajima et al. (2000) reported that the
13 number of peroxisomes in hepatocytes increased by 2-fold in wild-type mice but not in PPARa
14 knockout mice. However, TCE induced increased liver weight in both male and female wild-
15 type and knockout mice, suggesting hepatic effects independent of PPARa activation. In
16 regards to toxicity, after three weeks of TCE treatment (0 tol,500 mg/kg via gavage), Laughter
17 et al. (2004) reported toxicity at thel,500 mg/kg level in the knockout mice that was not
18 observed in the wild-type mice — all knockout mice were moribund and had to be removed
19 from the study. Differences in experimental protocol made comparisons between TCE effects
20 and those of its metabolites difficult in this study (see Section E.2.1.13, above).
21 As reported by Voss et al. (2006), dose-, time course-, species-, and strain-related
22 differences should be considered in interpreting gene array data. The comparison of differing
23 PPARa agonists presented in Keshava and Caldwell (2006) illustrate the pleiotropic and varying
24 liver responses of the PPARa receptor to various agonists, but did imply that these responses
25 were responsible for carcinogenesis.
26 As discussed above in Section E.3.3.5 and in Caldwell and Keshava (2006),
27
28 Aberrant DNA methylation has emerged in recent years as a common hallmark of
29 all types of cancers, with hypermethylation of the promoter region of specific
30 tumor suppressor genes and DNA repair genes leading to their silencing (an effect
31 similar to their mutation) and genomic hypomethylation (Ballestar and Esteller,
32 2002; Berger and Daxenbichler, 2002; Herman et al., 1998; Pereira et al., 2004;
33 Rhee et al., 2002). Whether DNA methylation is a consequence or cause of cancer
34 is a long-standing issue (Ballestar and Esteller, 2002). Fraga et al. (2004, 2005)
35 reported global loss of monoacetylation and trimethylation of histone H4 as a
36 common hallmark of human tumor cells; they suggested, however, that
37 genomewide loss of 5-methylcytosine (associated with the acquisition of a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-372 DRAFT—DO NOT CITE OR QUOTE
-------
1 transformed phenotype) exists not as a static predefined value throughout the
2 process of carcinogenesis but rather as a dynamic parameter (i.e., decreases are
3 seen early and become more marked in later stages).
4
5 Although little is known about how it occurs, a hypothesis has also been proposed that
6 that the toxicity of TCE and its metabolites may arise from its effects on DNA methylation status.
7 In regard to methylation studies, many are coexposure studies as they have been conducted in
8 initiated animals, and as stated above, some are very limited in regard to the reporting and
9 conduct of the study. Cal dwell and Keshava (2006) reviewed the body of work regarding TCE,
10 DCA, and TCA for this issue. Methionine status has been noted to affect the emergence of liver
11 tumors. As noted by Counts et al. (1996) a choline/methionine deficient diet for 12 months did
12 not increase liver tumor formation in C3H/HeN mice but is tumorigenic to B6C3F1 mice. Tao et
13 al. (2000) and Pereira et al. (2004) have studied the effects of excess methionine in the diet to see
14 if it has the opposite effects as a deficiency (i.e., and reduction in a carcinogenic response rather
15 than enhancement). As noted above for Tao et al. (2000), the administration of excess
16 methionine in the diet is not without effect. The data of Tao et al. (2000) suggest that percent
17 liver/body weight ratios are affected by short-term methionine exposure (300 mg/kg) in female
18 B6C3F1 mice. Pereira et al. (2004) reported that very high level of methionine supplementation
19 to an AIN-760A diet, affected the number of foci and adenomas after 44 weeks of coexposure to
20 3.2.g/L DCA. While the highest concentration of methionine (8.0 g/kg) was reported to decrease
21 both the number of DCA-induce foci and adenomas, the lower level of methionine coexposure
22 (4.0 g/kg) increased the incidence of foci. Coexposure of methionine (4.0 or 8.0 g/kg) with 3.2
23 g/L DCA was reported to decrease by -25% DCA-induced glycogen accumulation, increase
24 mortality, but not to have much of an effect on peroxisome enzyme activity (which was not
25 elevated by more than 33% over control for DCA exposure alone). Methionine treatment alone
26 at the 8 g/kg level was reported to increase liver weight, decrease lauroyl-CoA activity and to
27 increase DNA methylation. The authors suggested that their data indicate that methionine
28 treatment slowed the progression of foci to tumors. Given that increasing hypomethylation is
29 associated with tumor progression, decreased hypomethylation from large doses of methionine
30 are consistent with a slowing of progression. Whether, these results would be similar for lower
31 concentrations of DCA and lower concentrations of methionine that were administered to mice
32 for longer durations of exposure, cannot be ascertained from these data. It is possible that in a
33 longer-term study, the number of tumors would be similar. Whether, methionine treatment
34 coexposure had an effect on the phenotype of foci and tumors was not presented by the authors in
35 this study. Such data would have been valuable to discern if methionine coexposure at the 4.0
36 mg/kg level that resulted in an increase in DCA-induce foci, resulted in foci of a differing
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-373 DRAFT—DO NOT CITE OR QUOTE
-------
1 phenotype or a more heterogeneous composition than DCA treatment alone. Finally, a decrease
2 in tumor progression by methionine supplementation is not shown to be a specific event for the
3 MOA for DCA-induced liver carcinogenicity.
4 Tao et al. (2000) reported that 7 days of gavage dosing of TCE (1,000 mg/kg in corn oil),
5 TCA (500 mg/kg, neutralized aqueous solution), and DCA (500 mg/kg, neutralized aqueous
6 solution) in 8-week old female B6C3F1 mice resulted in not only increased liver weight but also
7 increased hypomethylation of the promoter regions of c-Jun and c-Myc genes in whole liver
8 DNA (data shown for 1-2 mice per treatment). Treatment with methionine was reported to
9 abrogate this response only at a 300 mg/kg i.p. dose with 0-100 mg/kg doses of methionine
10 having no effect. Ge et al. (2001b) reported DCA- and TCA-induced DNA hypomethylation and
11 cell proliferation in the liver of female mice at 500 mg/kg and decreased methylation of the
12 c-Myc promoter region in liver, kidney and urinary bladder. However, increased "cell
13 proliferation" preceded hypomethylation. Ge et al. (2002) also reported hypomethylation of the
14 c-myc gene in the liver after exposure to the peroxisome proliferators 2,4-dichlorophenoxyacetic
15 acid (2,4-D)(l,680 ppm), dibutyl phthalate (20,000 ppm), Gemfibrozil (8,000 ppm), and
16 WY-14,643 (50-500 ppm, with no effect at 5 or 10 ppm) after six days in the diet. Caldwell and
17 Keshava (2006) concluded that hypomethylation did not appear to be a chemical-specific effect
18 at these concentrations. As noted above in Section E.3.3.5, chemical exposure to a number of
19 differing carcinogens have been reported to lead to progressive loss of DNA methylation..
20 Caldwell and Keshava (2006) also note similar changes in methylation after initiation and
21 treatment with DCA or TCA.
22
23 After initiation by N-methyl-N-nitrosourea (25 mg/kg) and exposure to 20 mmL/L
24 DCA or TCA (46 weeks), Tao et al. (2004) report similar hypomethylation of
25 total mouse liver DNA by DCA and TCA with tumor DNA showing greater
26 hypomethylation. A similar effect was noted for region-2 (DMR-2) of the
27 insulin-like growth factor-II (IGF-II) gene. The authors suggest that
28 hypomethylation of total liver DNA and the IGF-II gene found in non-turnorous
29 liver tissue would appear to be the result of a more prolonged activity and not cell
30 proliferation, while hypomethylation of tumors could be an intrinsic property of
31 the tumors. Over expression of IGF-II gene in liver tumors and preneoplastic foci
32 has been shown in both animal models of hepatocarcinogenesis and humans, and
33 may enhance tumor growth, acting via the over-expressed IGF-I receptor (Scharf
34 et al., 2001; Werner and Le Roith, 2000). IGF-I is the major mediator of the
35 effects of the growth hormone; it thus has a strong influence on cell proliferation
36 and differentiation and is a potent inhibitor of apoptosis (Furstenberger et al.,
37 2002). Normally, expression of IGF-II in liver is greater during the fetal period
38 than the adult, but is over-expressed in human hepatocarcinomas due to activation
39 of fetal promoters (Scharf et al., 2001) and loss of imprinting (Khandawala et al.,
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-374 DRAFT—DO NOT CITE OR QUOTE
-------
1 2000). Takeda et al. (1996) report IGF-II expression in the liver is monoallelic
2 (maternally imprinted) in the fetal period is relaxed during the postnatal period,
3 (resulting in biallelic expression), and is imbalanced in human hepatocarcinomas
4 (leading to restoration of monoallelic IG-II expression).
5
6 However, Bull (2004) and Bull et al. (2004) have recently suggested that hypomethylation
7 and peroxisome proliferation occur at higher exposure levels than those that induce liver tumors
8 for TCE and its metabolites. They report that a direct comparison in the no-effect level or low-
9 effect level for induction of liver tumors in the mouse and several other endpoints shows that, for
10 TCA, liver tumors occur at lower concentrations than peroxisome proliferation in vivo but that
11 PPARa activation occurs at a lower dose than either tumor formation or peroxisome
12 proliferation. A similar comparison for DCA shows that liver tumor formation occurs at a much
13 lower exposure level than peroxisome proliferation, PPARa activation, or hypomethylation. In
14 addition, they report that these chemicals are effective as carcinogens at doses that do not
15 produce cytotoxicity.
16
17 E.3.4.2.3. Oxidative Stress. Several studies have attempted to study the possible effects of
18 "oxidative stress" and DNA damage resulting from TCE exposures. The effects of induction of
19 metabolism by TCE, as well as through coexposure to ethanol, have been hypothesized in itself
20 to increase levels of "oxidative stress" as a common effect for both exposures (see
21 Section E.4.2.4, below). Oxidative stress has been hypothesized to be the MOA for peroxisome
22 proliferators as well, but has been found to neither be correlated with cell proliferation nor
23 carcinogenic potency of peroxisome proliferators (see Section E.3.4.1.1). As a MOA, it is not
24 defined or specific as the term "oxidative stress" is implicated as part of the pathophysiologic
25 events in a multitude of disease processes and is part of the normal physiologic function of the
26 cell and cell signaling.
27 In regard to measures of oxidative stress, Rusyn et al. (2006) noted that although an
28 overwhelming number of studies draw a conclusion between chemical exposure, DNA damage,
29 and cancer based on detection of 8-OHdG, a highly mutagenic lesion, in DNA isolated from
30 organs of in vivo treated animals, a concern exists as to whether increases in 8-OHdG represent
31 damage to genomic DNA, a confounding contamination with mitochondrial DNA, or an
32 experimental artifact. As described in Section E.2.2.8, the study by Channel et al. (1998)
33 demonstrated that corn oil as vehicle had significant effects on measures of "oxidative stress"
34 such as TEARS. Also as noted previously (see Sections E.2.1.1 and E.2.2.11), studies of TCE
35 which employ the i.p. route of administration can be affected by inflammatory reactions resulting
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-375 DRAFT—DO NOT CITE OR QUOTE
-------
1 from that routes of administration and subsequent toxicity that can involve oxygen radical
2 formation from inflammatory cells.
3 The issues with interpretation of the Channel et al. (1998) study of TCE administered via
4 corn oil gavage to mice have already been discussed in Section E.2.1.7, above. The TEARS
5 results indicated suppression of TEARS with increasing time of exposure to corn oil alone with
6 data presented in such a way for 8-OHdG and total free radical changes that the pattern of corn
7 oil administration was obscured. It was not apparent from that study that TCE exposure induced
8 oxidative damage in the liver.
9 Toraason et al. (1999) measured 8-OHdG and a "free radical-catalyzed isomer of
10 arachidonic acid and marker of oxidative damage to cell membranes, 8-Epi-prostaglandin F2a
11 (SepiPGF)," excretion in the urine and TEARS (as an assessment of malondialdehyde and marker
12 of lipid peroxidation) in the liver and kidney of male Fischer rats (150-200 g) exposed to single
13 0, 100, 500, or 1,000 mg/kg TCE i.p. injections in Alkamuls vehicle (n = 6/group). Two
14 sequential urine samples were collected 12 hours after injection and animals were sacrificed at
15 24 hours with DNA collected from liver tissues and TEARS measured in liver homogenates. The
16 mean body weights of the rats were reported to vary by 13% but the liver weights varied by 44%
17 after the single treatments of TCE. In contrast to the large volume of the literature that reports
18 TCE-induced increases in liver weight, the 500 and 1,000 mg/kg exposed rats were reported to
19 have reduced liver weight by 44% in comparison to the control values. Using this paradigm, 500
20 mg/kg TCE was reported to induce stage II anesthesia and a 1,000 mg/kg TCE to induce Level III
21 or IV (absence of reflex response) anesthesia and burgundy colored urine with 2/6 rats at 24
22 hours comatose and hypothermic. The animals were sacrificed before they could die and the
23 authors suggested that they would not have survived another 24 hours. Thus, using this paradigm
24 there was significant toxicity and additional issues related to route of exposure. Urine volume
25 declined significantly during the first 12 hours of treatment and while water consumption was not
26 measured, it was suggested by the authors to be decreased due to the moribundity of the rats.
27 Given that this study examined urinary markers of "oxidative stress" the effects on urine volume
28 and water consumption, as well as the profound toxicity induced by this exposure paradigm, limit
29 the interpretation of the study. The authors noted that because both using volume and creatinine
30 excretion were affected by experimental treatment, urinary excretion of 8-OHdG changed
31 significantly based on the mode of data expression. Excretion of SepiPGF was reported to be no
32 different from controls 12-24 hours and decreased 24 hours after TCE exposure at the two
33 highest levels. Excretion of 8-OHdG was reported to not be affected by any exposure level of
34 TCE and, if expressed on the basis of 24-hours, decreased. TEARS concentration per gram of
35 liver was reported to be increased at the 500 and 1,000 mg/kg TCE exposure levels (~2-3-fold).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-376 DRAFT—DO NOT CITE OR QUOTE
-------
1 The effects of decreased liver size in the treated animals for this measure in comparison to
2 control animals, was not discussed by the authors. For 8-OHdG measures in the liver and
3 lymphocytes, the authors reported that "cost prohibited analysis of all of the tissues samples" so
4 that a subset of animals was examined exhibiting the highest TEARS levels. The number of
5 animals used for this determination was not given nor the data except for 500 mg/kg TCE
6 exposure level. TCE was reported to increase 8-OHdG/dG in liver DNA relative to controls to
7 about the same extent in lymphocytes from blood and liver (~2-fold) with the results for liver
8 reported to be significant. The issues of bias in selection of the data for this analysis, as well as
9 the issues already stated for this paradigm limit interpretation of these data while the authors
10 suggest that evidence of oxidative damage was equivocal.
11 DCA and TCA have also been investigated using similar measures. Larson and Bull
12 (1992) exposed male B6C3F1 mice [26 ± 3 g (SD)] to a single dose of 0, 100, 300, 1,000, or
13 2,000 mg/kg/d TCA or 0, 100, 300, or 1,000 mg/kg/d DCA in distilled water by oral gavage
14 (n = 4). Fischer 344 rats (237 ± 4 g) received a single oral dose of 0, 100, or 1,000 mg/kg DCA
15 or TCA (n = 4 or 5) TEARS was measured from liver homogenates and assumed to be
16 malondialdehyde. The authors stated that a preliminary experiment had shown that maximal
17 TEARS was increased 6 hours after a dose of DCA and 9 hours after a dose of TCA in mice (data
18 shown) and that by 24 hours TEARS concentrations had declined to control values (data not
19 shown). However, time-course information in rats was not presented and the same times used for
20 both species, (i.e., 6- and 9-hours time periods after administration of DCA and TCA) for
21 examination of TEARS activity. A dose of 100 mg/kg DCA (rats or mice) or TCA (mice) did
22 not elevate TEARS concentrations over that of control liver with this concentration of TCA not
23 examined in rats. For TCA, there was a slight dose-related increase in TEARS over control
24 values starting at 300 mg/kg in mice (i.e., 1.68-, 2.02-, and 2.70-fold of control for 300, 1,000,
25 and 2,000 mg/kg TCA). For DCA there were similar increases over control for both the 300 and
26 1,000 mg/kg dose levels in mice (i.e., 3.22- and 3.45-fold of control, respectively). For rats the
27 1,000 and 2,000 mg/kg levels of TCA were reported to show a statistically significant increase in
28 TEARS over control (i.e., 1.67- and 2.50-fold, respectively) with the 300 and 1,000 mg/kg level
29 of DCA showing similar increases but with only the 300 mg/kg-induced change statistically
30 significant different than control values (i.e., 3.0- and 2.0-fold of control, respectively). Of note,
31 is the report that the induction of TEARS in mice is transient and had subsided within 24 hours of
32 a single dose of DCA or TCA, that the response in mice appeared to be slightly greater with DCA
33 than TCA at similar doses, and that for DCA, there was similar TEARS induction between rats
34 and mice at similar dose levels.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-377 DRAFT—DO NOT CITE OR QUOTE
-------
1 A study by Austin et al. (1996) appears to a follow-up publication of the preliminary
2 experiment cited in Larson and Bull (1992). Male B6C3F1 mice (8 weeks old) were treated with
3 single doses of DC A or TCA in buffered solution (300 mg/kg) with liver examined for 8-OHdG.
4 The authors stated that in order to conserve animals, controls were not employed at each time
5 point. For DCA the time course of 8-OHdG was studied at 0, 4, 6, and 8 hours after
6 administration and for TCA at 0, 6, 8, and 10 hours after of a 300 mg/kg dose (n = 6). There was
7 a statistically significant increase over controls in 8-OHdG for the 4- and 6-hour time points for
8 DCA (~1.4- and 1.5-fold of control, respectively) but not at 8 hours in mice. For TCA, there was
9 a statistically significant increase in 8-OHdG at 8 and 10 hours for TCA (-1.4- and 1.3-fold of
10 control, respectively).
11 The results for PCO and liver weight for Parrish et al. (1996) are discussed in
12 Section E.2.3.2.2 above for male B6C3F1 mice exposed to TCA or DCA (0, 0.01, 0.5, and
13 2.0 g/L) for 3 or 10 weeks (n = 6). The study focused on an examination of the relationship with
14 measures of peroxisome proliferation and oxidative stress. The dose-related increase in PCO
15 activity at 21 days (-1.5-, 2.2-, and ~4.1-fold of control, for 0.1, 0.5, and 2.g/L TCA) was
16 reported not to be increased similarly for DCA. Only the 2.0 g/L dose of DCA was reported to
17 induce a statistically significant increase at 21-days of exposure of PCO activity over control
18 (~1.8-fold of control). After 71 days of treatment, TCA induced dose-related increases in PCO
19 activities that were approximately twice the magnitude as that reported at 21 days (i.e., ~9-fold
20 greater at 2.0 g/L level). Treatments with DCA at the 0.1 and 0.5 g/L exposure levels produced
21 statistically significant increase in PCO activity of ~1.5- and 2.5-fold of control, respectively.
22 The administration of 1.25 g/L clofibric acid in drinking water, used as a positive control, gave
23 ~6-7-fold of control PCO activity at 21 and 71 days exposure.
24 Parrish et al. (1996) reported that laurate hydroxylase activity was reported to be elevated
25 significantly only by TCA at 21 days and to approximately the same extent (-1.4 to 1.6-fold of
26 control) increased at all doses tested. At 71 days both the 0.5 and 2.0 g/L TCA exposures
27 induced a statistically significant increase in laurate hydroxylase activity (i.e., 1.6- and 2.5-fold of
28 control, respectively) with no change reported after DCA exposure. The actual data rather than
29 percent of control values were reported for laurate hydroxylase activity with the control values
30 varying 1.7-fold between 21 and 71 days experiments. Levels of 8-OHdG in isolated liver nuclei
31 were reported to not be altered from 0.1, 0.5, or 2.0 g/L TCA or DCA after 21 days of exposure
32 and this negative result was reported to remain even when treatments were extended to 71 days of
33 treatment. The authors noted that the level of 8-OHdG increased in control mice with age (i.e.,
34 ~2-fold increase between 71-day and 21-day control mice). Clofibric acid was also reported not
35 to induce a statistically significant increase of 8-OHdG at 21 days, but to produce an increase
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-378 DRAFT—DO NOT CITE OR QUOTE
-------
1 (~1.4-fold of control) at 71 days. Thus, the increases in PCO activity noted for DCA and TCA
2 were not associated with 8-OHdG levels (which were unchanged) and, also, not with changes
3 laurate hydrolase activity observed after either DCA or TCA exposure. Of note is the variability
4 in both baseline levels of PCO and laurate hydrolase activity. Also of note, is that the authors
5 report taking steps to minimize artifactual responses for their 8-OHdG determinations. The
6 authors concluded that their data does not support an increase in steady state oxidative damage to
7 be associated with TCA initiation of cancer and that extension of treatment to time periods
8 sufficient to insure peroxisome proliferation failed to elevate 8-OHdG in hepatic DNA. The
9 increased 8-OHdG at 10 weeks after Clofibrate administration but lack of 8-OHdG elevation at
10 similar levels of PCO induction by were also noted by the authors to suggest that peroxisome
11 proliferative properties of TCA were not linked to oxidative stress or carcinogenic response.
12 As noted above for the study of Leakey et al. (2003a) (see Section E.2.3.4), hepatic
13 malondialdehyde concentration in ad libitum fed and dietary controlled mice did not change
14 with CH exposure at 15 months but the dietary controlled groups were all approximately half
15 that of the ad libitum-fed mice. Thus, while overall increased tumors observed in the ad libitum
16 diet correlated with increased malondialdehyde concentration, there was no association between
17 CH dose and malondialdehyde induction for either diet.
18
19 E.4. EFFECTS OF COEXPOSURES ON MODE OF ACTION (MOA)—INTERNAL
20 AND EXTERNAL EXPOSURES TO MIXTURES INCLUDING ALCOHOL
21 Caldwell et al. (2008b) recently published a review of the issues and studies involved
22 with the effects of coexposures to TCE metabolites that could be considered internal (i.e., an
23 internal coexposure for the liver) and coexposures to metabolites and other commonly occurring
24 chemicals that are present in the environment. As they stated:
25
26 Human exposure to a pollutant rarely occurs in isolation. EPA's Cumulative
27 Exposure project and subsequent National Air Toxics Assessment have
28 demonstrated that environmental exposure to a number of pollutants, classified
29 as potential human carcinogens, is widespread [U.S. EPA, 2006;Woodruff et al.,
30 1998]. Interactions between carcinogens in chemical mixtures found in the
31 environment have been a concern for several decades. Furthermore, how these
32 interactions affect the mode of action (MOA) by which these chemicals operate
33 and how such effects may modulate carcinogenic risk is of concern as well.
34 Thus, an understanding of the MOA(s) of a pollutant can help elucidate its
35 potential carcinogenic risk to humans, and can also help identify susceptible
36 subpopulations through their intrinsic factors (e.g., age, gender, and genetic
37 polymorphisms of key metabolic and clearance pathways) and extrinsic factors
38 (e.g. co-exposures to environmental contaminants, ethanol consumption, and
39 pharmaceutical use). Trichloroethylene (TCE) can be a useful example for
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-379 DRAFT—DO NOT CITE OR QUOTE
-------
1 detailing the difficulties and opportunities for investigating such issues because,
2 for TCE, there is both internal exposure to a "chemical mixture" of multiple
3 carcinogenic metabolites [Chiu et al., 2006a, b] and co-exposures from
4 environmental contamination of TCE metabolites, and from pollutants that share
5 common metabolites, metabolic pathways, MO As, and targets of toxicity with
6 TCE.
7
8 Typically, ground water or contaminated waste sites can have a large number of
9 pollutants that vary in regard to information available to support the
10 characterization of their potential hazard, and that have differing MO As and
11 targets. For example, Veeramachaneni et al. (2001) reported reproductive effects
12 in male rabbits, resulting from exposure to drinking water containing
13 concentrations of chemicals typical of ground water near hazardous waste sites.
14 The drinking water exposure mixture contained arsenic, chromium, lead,
15 benzene, chloroform, phenol, and TCE. Even at 45 weeks after the last
16 exposure, mating desire/ability, sperm quality, and Ley dig cell function were
17 subnormal. However, while the exposure levels are relevant to human
18 environmental exposures, design of this study precludes a conclusion as to which
19 individual toxicant, or combination of the seven toxicants, caused the effects.
20 Thus, this study exemplifies he problems associated with studying a multi-
21 mixture milieu. Studies of the interactions of TCE metabolites or common co-
22 exposures that report the interactions of 2 or 3 chemicals at one time are easier to
23 interpret.
24
25 Since EPA published its 2001 draft assessment, several approaches have been
26 reported that include examination of tumor phenotype, gene expression, and
27 development of physiologically-based pharmacokinetic (PBPK) models to assess
28 possible effects of co-exposure. They attempt to predict whether such co-
29 exposures would produce additivity of response or if co-exposure would change
30 the nature of responses induced by TCE or its metabolites. In addition, new
31 studies on co-exposure to DBA may help identify a co-exposure of concern.
32 These studies may give potential insights into possible MO As and modulators of
33 TCE toxicity. More recent information on the toxicity of individual metabolites
34 of TCE [Caldwell and Keshava, 2006] may be helpful in trying to identify which
35 are responsible for TCE toxicity, but may also identify the effects of
36 environmental co-exposures.
37
38 Recently, EPA sought advice from the National Academy of Sciences (NAS)
39 [Chiu et al., 2006a] with the NAS charge questions including the following. (1)
40 What TCE metabolites, or combinations of metabolites, may be plausibly
41 involved in the toxicity of TCE? (2) What chemical co-exposures may plausibly
42 modulate TCE toxicity? (3) What can be concluded about the potential for
43 common drinking water contaminants such as other solvents and/or haloacetates
44 to modulate TCE toxicity? (4) What can be concluded about the potential for
45 ethanol consumption to modulate TCE toxicity? Thus, the understanding of the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 80 DRAFT—DO NOT CITE OR QUOTE
-------
1 effects of co-exposure, in the context of MO A, is an important element in
2 understanding the risk of a potential human carcinogen.
3
4 U.S. EPA's draft TCE risk assessment [U.S. EPA, 2001] identified several
5 factors involving co-exposure to TCE metabolites, environmental contaminants,
6 and ethanol that could lead to differential sensitivity to TCE toxicity. Research
7 needs identified there, as well as in previous reviews [Bull, 2000; Pastino et al.,
8 2000], included further elucidation of the interaction of TCA and DCA in TCE-
9 induced liver tumors and a better understanding of the functional relationships
10 among risk factors. The complexity of TCE's potential interactions with
11 chemical co-exposures from either common environmental co-contaminants or
12 common behaviors such as alcohol consumption mirrors the complexity of the
13 metabolism and the actions of TCE metabolites. Thus, TCE presents a good case
14 study for further exploration of the effects of co-exposure on MO A.
15
16 The following sections first reiterates the findings of Bull et al. (2002) in regard to
17 simple coexposures of DCA and TCA which can be experienced as an internal coexposure after
18 TCE exposure. A number of studies have examined the effects of TCE or its metabolites after
19 previous exposure to presumably genotoxic carcinogen to not only determine the effect of the
20 coexposure on liver carcinogenicity but also to use such paradigms to distinguish between the
21 effects of TCA and DCA. Finally, not only is TCE a common coexposure with its own
22 metabolites, but is also a common coexposure with other solvents, and the brominated analogues
23 of TCA and DCA. The available literature is examined for potential similarities in target and
24 effects that may cause additional concern. The effects of ethanol on TCE toxicity is examined
25 as well as the potential pharmacokinetic modulation of risk using recently published reports of
26 PBPK models that may be useful in predicting coexposure effects.
27
28 E.4.1. Internal Coexposures to Trichloroethylene (TCE) Metabolites: Modulation of
29 Toxicity and Implications for TCE Mode of Action (MOA)
30 Exposure to TCE will produce oxidative metabolites in the liver as an internal
31 coexposure. As stated above, the phenotypic analysis of TCE-induced tumors have similarities
32 to combinations of DCA and TCA and in some reports to resemble more closely DCA-induced
33 tumors in the mouse. Results from Bull et al. (2002) are presented in Section E.2.2.22 for the
34 treatment of mice to differing concentrations of DCA and TCA in combination and the
35 resemblance of tumor phenotype to that of TCE. In regard to cancer dose-response, the most
36 consistent treatment-related increase in response occurred with combinations of exposure to
37 DCA and TCA that appeared to increase lesion multiplicity when compared to effects from
38 individual chemicals separately. Bull et al. (2002) presented results for "selected" lesions
39 examined for pathology characterization that suggest coexposure of 0.5 g/L DCA with either 0.5
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 81 DRAFT—DO NOT CITE OR QUOTE
-------
1 or 2 g/L TCA had a greater than additive effect on the total number of hyperplastic nodules. In
2 addition coexposure to 0.1 g/L DCA and 2 g/L TCA was reported to have a greater than additive
3 effect on the total number of adenomas, but not carcinomas, induced. The random selection of
4 lesions for the determination of potential treatment-related effects on incidence and multiplicity,
5 rather than characterization of all lesions, increases the uncertainty in this finding.
6
7 E.4.2. Initiation Studies as Coexposures
8 There is a body of literature that has focused on the effects of TCE and its metabolites
9 after rats or mice have been exposed to "mutagenic" agents to "initiate" hepatocarcinogenesis.
10 Given that most of these "initiating agents" have many effects that are not only mutagenic but
11 also epigenetic, that the dose and exposure paradigm modify these effects, that "initiators" can
12 increased tumor responses alone, and the tumors that arise from these protocols are reflective of
13 simultaneous actions of both "initiator" and "promoter," paradigms that first expose rats or mice
14 to a "mutagen" and then to other carcinogenic agents can be described as a coexposure
15 protocols. As stated previously, DEN and 7V-nitrosomorpholine have been reported to increase
16 differing populations of mature hepatocytes with DEN not only being a mutagen but also able to
17 induce concurrent hepatocyte regeneration at a high dose. Thus, the effects of the TCE or its
18 metabolites are hard to discern from the effects of the "initiating" agent in terms of MO A. As
19 demonstrated in the studies of Pereira et al. (1997) below, the gender also determines the nature
20 of the tumor response using these protocols. In addition, when the endpoint for examination is
21 tumor phenotype the consequences of tumor progression are hard to discern from the MOA of
22 the agents using paradigms of differing concentrations, different durations of exposure, lesions
23 counted as "tumors" to include different stages of tumor progression (foci to carcinoma), and
24 highly variable and low numbers of animals examined. However, differences in phenotype of
25 tumors resulting from such coexposures, like the coexposure studies cited above for just TCE
26 metabolites, can help determine that exposure to TCE metabolites results in differing actions as
27 demonstrated by differing effects in the presence of cocarcinogens. As stated above, Kraupp-
28 Grasl et al. (1990) use the same approach and note differences among PPARa agonists in their
29 ability to promote tumors suggest they should not necessarily be considered a uniform group.
30
31 E.4.2.1. Herren-Freund et al, 1987
32 The results of TCE exposure alone were reported previously (E.2.2.17) for this study.
33 This study's focus was on the effect of TCE, TCA, DCA and Phenobarbital on
34 hepatocarcinogenicity in male B6C3F1 mice after "initiation" at 15 days with 2.5 or 10 ug/g
35 body weight of ethylnitrosourea (ENU) and then subsequent exposure to TCE and other
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 82 DRAFT—DO NOT CITE OR QUOTE
-------
1 chemicals in drinking water begging at 4 weeks of age (an age when the liver is already
2 undergoing rapid growth). DCA and TCA were given in buffered solutions and sodium chloride
3 given in the water of control animals. The experiment was reported to be terminated at 61
4 weeks because the "mice started to exhibit evidence of tumors." Concentrations of TCE were 0,
5 3 and 40 mg/L, of DCA and TCA 0, 2 and 5 g/L, and of Phenobarbital 0 and 500 mg/L. The
6 number of animals examined in each group ranged from 16 to 32. ENU alone in this paradigm
7 was reported to induce statistically significant increases in adenomas and hepatocellular
8 carcinomas (39% incidence of adenomas and 39% incidence of carcinomas vs. 9 and 0% for
9 controls) at the 10 ug/g dose (n = 23), but not at 2.5 ug/g dose (n = 22). The effects of high
10 doses of DCA and TCA alone have already been discussed for other studies, as well as the lack
11 of statistical power using a paradigm with so few and variable numbers of animals, the
12 limitations of an abbreviated duration of exposure which does not allow for full expression of a
13 carcinogenic response, and problems of volatilization of TCE in drinking water. DCA and TCA
14 treatments at these levels (5 g/L) were reported to increase adenomas and carcinomas
15 irrespective of ENU pretreatment and to approximately the same extent with and without ENU.
16 TCE at the highest dose was reported to increase the number of animals with adenomas (37 vs.
17 9% in control) and carcinomas (37 vs. 0% in controls) but only the # of adenomas/animal was
18 statistically significant as the number of animals examined was only 19 in the TCE group.
19 Phenobarbital was reported to have no effect on ENU tumor induction using this paradigm.
20
21 E.4.2.2. Parnettetal, 1986
22 This study used a rat liver foci bioassay (y-glutamyltranspeptidase, i.e., GGT) for hepatic
23 foci after at 3 and 6 month using protocols that included partial hepatectomy, DEN (10 mg/kg)
24 or TCA (1,500 ppm in drinking water) treatment, and then promotion with 5,000 ppm TCA (i.e.,
25 5 g/L) for 10, 20, or 30 days and phenobarbital (500 ppm) in male S-D rats (5-6 weeks old at
26 partial hepatectomy). The number of animals per group ranged from 4-6. PCO activities were
27 given for various protocols involving partial hepatectomy, DEN, TCA and Phenobarbital
28 treatments but there was no controls values given that did not have a least one of these
29 treatments. Overall, it appeared there was a slight decrease of PCO activity in rats treated with
30 partial hepatectomy/DEN/Phenobarbital treatments and a slight increase over other treatments
31 for rats treated with partial hepatectomy/DEN/5,000 ppm TCA or just TCA from 2 weeks to
32 6 months of sampling. In regard to GGT-positive foci, the partial
33 hepatectomy/DEN/Phenobarbital group (n = 6) was reported to have more positive foci at 3 or
34 6 months than rats "initiated" with TCA and PB after partial hepatectomy or partial
35 hepatectomy/Phenobarbital treatment alone (2.05 foci/cm2 vs. -.05-0.10 foci/cm2 for all other
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 83 DRAFT—DO NOT CITE OR QUOTE
-------
1 groups). The number of GOT positive foci in rats without any treatment were not studied or
2 presented by the authors. For "promotion" protocols the number of GGT positive foci induced
3 by the partial hepatectomy/DEN/Phenobarbital protocol at 3 and 6 months, appeared to be
4 reduced when Phenobarbital exposure was replaced by TCA coexposure but there was no dose-
5 response between the 50, 500 and 5,000 ppm. However, TCA treatment along with partial
6 hepatectomy and DEN treatment did increase the levels of foci (means of 0.71-0.39 foci/cm2 at
7 3 months and 1.83-2.45 foci/cm2 at 6 months) over treatment of just partial hepatectomy and
8 DEN (0.05 ± 0.20 foci/cm2 at 3 months and 0.30 ± 0.39 foci/cm2 at 6 months). For the TCA
9 animals treated only with 5,000 ppm TCA, the number of GGT positive foci at 3 months was
10 0.23 ±0.16 foci/cm2 and at 6 months 0.03 ± 0.32 foci/cm2 with no values for untreated animals
11 presented. For the positive control (partial hepatectomy/DEN/Phenobarbital) the number of
12 GGT positive foci increased from 3 to 6 months (1.65 ± 0.23 foci/cm2 and at 6 months
13 7.61±0.72 foci/cm2). The authors concluded that
14
15 although TCA is reported to cause hepatic peroxisomal stimulation in rats and
16 mice, the results of this study indicate that it is unlikely TCA's effects are related
17 to the promoting ability seen here. The minimal stimulation of, 10 to 20% over
18 controls of peroxisomal associated, PCO activity in TCA exposed rats was seen
19 only at the 5000 ppm level and only within the promotion protocol. This finding
20 is in contrast to the promoting activity seen at all three concentrations of TCA.
21
22 E.4.2.3. Pereira and Phelps, 1996
23 The results for mice that were not "initiated" by exposure to MNU, but exposed to DCA
24 or TCA, are discussed in Section E.2.3.2.6. However, differences in responses after initiation
25 are useful for showing differences between single and coexposures as well as differences
26 between DCA and TCA effects. On Day 15 of age, female B6C3F1 mice received an i.p.
27 injection of MNU (25 mg/kg) and at 7 weeks of age received DCA (2.0, 6.67, or 20 mmol/L),
28 TCA (2.0, 6.67 mmol, or 20 mmol/L), or NaCl continuously for 31 or 51 weeks of exposure.
29 The number of animals studied ranged from 6 to 10 in 31-week groups and 6 to 39 in the
30 52-week groups. There was a "recovery group" in which mice received either 20 mmol/L
31 DCA (2.58 g/L DCA) (n = 12) or TCA (3.27 g/L TCA) (n = 11) for 31 weeks and then
32 switched to saline for 21 weeks until sacrifice at 52 weeks. Strengths of the study included the
33 reporting of hepatocellular lesions as either foci, adenomas, or carcinomas and the presentation
34 of incidence and multiplicity of each separately reported for the treatment paradigms.
35 Limitations included the low and variable number of animals in the treatment groups.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 84 DRAFT—DO NOT CITE OR QUOTE
-------
1 MNU was reported to not "significantly" induce foci or altered hepatocytes, adenomas,
2 or carcinomas at 31 (n= 10) or 51 weeks (n = 39). However, MNU did increase the incidence
3 and number/mouse of foci, adenomas and carcinomas at the 52 week sacrifice time in
4 comparison to saline controls, albeit at lower levels than observed in DCA or TCA
5 cotreatments groups (e.g., 10 vs. 0% foci, 17.5 vs. 2.5% adenomas, and 10 vs. 0% incidence of
6 carcinomas at 52 weeks for MNU-treated mice vs. saline control). Coexposure of DCA
7 (20.0 mmol/L) for 52 weeks in MNU-treated mice increased the number of foci and
8 hepatocellular adenomas with the authors reporting "the yield of total lesions/mouse increased
9 as a second order function of the concentration of DCA (correlation coefficients > 0.998)."
10 TCA coexposure in MNU-treated mice was reported not to result in a significant difference in
11 yield of foci or altered hepatocytes with either continuous 52 week or 31-week exposure, but
12 exposures to 20.0 or 6.67 mmol/L TCA did result in increased yield of liver tumors with both
13 exposure protocols (see below).
14 For TCA treatment in MNU treated mice, the incidences of foci were similar (12.5 vs.
15 18.2%) but the number of foci/mouse was ~3-fold greater in the cessation protocol than with
16 continuous exposure. The incidence of adenomas was reported to be the same (-66%) as well
17 as the number of adenomas/animal between continuous and cessation exposures. For
18 carcinomas, there was a greater incidence for mice with continuous TCA exposure (83 vs.
19 36%) as well as a greater number of carcinomas/mouse (~4-fold) than for those initiated mice
20 with cessation of TCA exposure. As noted above, the number of animals treated with TCA
21 was low and variable (e.g., 23 mice studied at 52 weeks 20.0 mmol/L TCA, and 6 mice at
22 52 weeks 6.67 mmol/L TCA), limiting the ability to discern a statistically significant effect in
23 regard to dose-response. The concentration-response relationship for tumors/mouse after 31
24 and 51 weeks was reported to be best represented by linear progression.
25 A comparison of results for animals treated with MNU and 20.0 mmol/L DCA or TCA
26 for 31 weeks and sacrificed at 31 weeks and those which were treated with MNU and DCA or
27 TCA for 31 weeks and then sacrificed at 52 weeks is limited by the number of animals exposed
28 (n = 10 for 31 week sacrifice DCA or TCA, n = 12 for DCA recovery group, and n = 11 for
29 TCA recovery group). No carcinoma data were reported for animals exposed at 31 weeks and
30 sacrificed at 31 weeks making comparisons with recovery groups impossible for this parameter
31 and thus, determinations about progression from adenomas to carcinomas. For the MNU and
32 DCA-treated animals, the incidence or number of animals reported to have foci at 31 weeks
33 was reported to be 80% but 38.5% for in the recovery group. For adenomas, the incidence was
34 reported to be 50% for DCA-treated animals at 31 weeks and 46.2% for the recovery group.
35 For MNU and TCA-treated animals, the incidence of foci at 31 weeks was reported to 20 and
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-385 DRAFT—DO NOT CITE OR QUOTE
-------
1 18.2% for the recovery group. For adenomas, the incidence was reported to be 60% for the
2 TCA-treated animals at 31 weeks and 63.6% for the recovery group. Thus, this limited data set
3 shows a decrease in incidence of foci for the MNU and DC A-treated recovery group but no
4 change in incidence of foci for TCA or for adenomas for DC A- or TCA-treatment between
5 those sacrificed at 31 weeks and those sacrificed 21 weeks later. In regard to multiplicity, the
6 number of foci/mouse was reported to be 2.80 ± 0.20 for the 31-week DC A group and
7 0.46 ± 0.18 for the recovery group (mean ± SEM). The number of adenomas/mouse was
8 reported to be 1.80 ± 0.83 for the 31-week group and 0.69 ± 0.26 for the recovery group. Thus,
9 both the number of foci and adenomas per mouse was reported to be decreased after the
10 recovery period for MNU and DCA treated mice. Given that the number of animals with foci
11 was decreased by half, the concurrent decrease in foci/mouse is not surprising. For TCA
12 treatments, the numbers of foci/mouse were reported to be 0.20 ± 0.13 for the 31-week group
13 and 0.45 ± 0.31 for the recovery group. The number of adenomas/mouse for TCA-treatment
14 groups was reported to be 1.30 ± 0.45 for the 31-week group and 0.91 ± 0.28 for the recovery
15 group. For the MNU and TCA-treated mice, the numbers of foci/mouse were reported to be
16 increased and the number of adenomas/mouse reported to be slightly lower. Because
17 carcinoma data are not presented for the 31 week group, it is impossible to determine whether
18 the TCA adenomas regressed to foci or the TCA adenomas progressed to carcinomas and more
19 foci apparent with increased time.
20 For the comparison of the numbers of foci, adenomas, or carcinomas per mouse that
21 were reported for the mice exposed at 31 weeks and sacrificed and those exposed for 52 weeks,
22 issues arise as to the impact of such few animals studied at 31 weeks, and the differing
23 incidences of lesions reported for these mice on tumor multiplicity estimates. The number of
24 animals studied who treated with MNU and 20.0 mmol/L DCA or TCA for 31 weeks and then
25 sacrificed was n = 10, while the number of animals exposed to 20.0 mmol/L DCA or TCA for
26 52 weeks was 24 for the DCA group and 23 for the TCA group. The number of animals treated
27 at lower concentrations of DCA or TCA were even lower at the 31-week sacrifice (e.g., n = 6
28 for MNU and 6.67 mmol/L DCA at 31 weeks) and also for the 52-week durations of exposure
29 (e.g., n = 6 for MNU and 6.6.7 mmol/L TCA).
30 At 31 weeks, 80% of the animals were reported to have foci and 50% to have foci after
31 52 weeks of exposure to 20.0 mmol/L DCA and MNU treatment. Thus, similar to the
32 "recovery" experiment, the number of animals with foci decreased even with continuous
33 exposure between 31 and 52 weeks. For adenomas, 20.0 mmol DCA exposure for 31 weeks
34 was reported to induce adenomas in 50% of mice and after 52 weeks of exposure to induce
35 adenomas in 73% of mice. For TCA, the number of animals with foci was reported to be 20%
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 86 DRAFT—DO NOT CITE OR QUOTE
-------
1 at 31 weeks and 12% at 52 weeks after exposure to 20.0 mmol/L TCA after MNU treatment
2 and similar to the incidence of foci reported for the TCA-recovery group. For 20.0 mmol TCA,
3 adenomas reported in 60% of mice after 31 weeks and in 67% of mice after 52 weeks of
4 exposure and also similar to the incidence of adenomas reported for the TCA-recovery group.
5 In regard to multiplicity, the number of foci/mouse was decreased from 2.80 ± 0.20 to
6 1.46 ± 0.48 between 31 weeks and 52 weeks of 20.0 mmol DCA in MNU exposed mice. The
7 number of adenomas/mouse was reported to be increased from 1.80 ± 0.83 to 3.62 ± 0.70
8 between 31 weeks and 52 weeks of 20.0 mmol DCA and MNU exposed mice. For
9 20.0 mmol/L TCA, the number of foci/mouse was 0.20 ± 0.13 and 0.13 ± 0.7 for 31- and
10 52-week exposures. The number of adenomas/mouse was reported to be 1.30 ± 0.45 and
11 1.29 ± 0.24 for 31- and 52-week exposures. Thus, by only looking at foci and adenoma
12 multiplicity data, there would not appear to be a change between 31 and 52-weeks. However,
13 during progression a shift may occur such that foci become adenomas with time and adenomas
14 become carcinomas with time. For carcinomas there was no data reported for 31-week
15 exposure in MNU and DCA- or TCA-treated mice. However, at 52 weeks 20.0 mmol DCA
16 was reported to induce carcinomas in 19.2% of mice and 20.0 mmol TCA to induce carcinomas
17 in 83% of mice. The corresponding numbers of carcinomas/mouse was 0.23 ±0.10 for
18 20.0 mmol/L DCA treatment and 2.79 ± 0.48 for 20.0 mmol/L TCA treatment at 52 weeks in
19 MNU treated mice. Thus, although fewer than 20% of MNU-treated mice were reported to
20 have foci at 20.0 mmol TCA, by 52 weeks almost all had carcinomas with -67% also having
21 adenomas. For DCA, many more mice had foci at 31 weeks (80%) than for TCA and by
22 52 weeks -70% had adenoma with only -20% reported to have carcinomas. The incidence
23 data are suggestive that as these high doses of DCA and TCA, TCA was more efficient
24 inducing progression of a carcinogenic response than DCA in MNU-treated mice.
25 The authors interpret the decrease in foci and adenomas between animals treated with
26 MNU and 20.0 mmol/L DCA for 31 weeks and sacrificed and those sacrificed 21 weeks later
27 to indicate that these lesions were dependent on continued exposure. However, the total
28 number of lesions cannot be ascertained because carcinoma data were not reported for 31-week
29 exposures. Carcinomas were reported in the recovery group at 52 weeks
30 (0.15 ± 0.10 carcinomas/mouse in 15.4% of animals). Of note is that not only did the number
31 of foci/mouse and incidence decrease between the 31-week group and the recovery group, but
32 also between 31- and 52-weeks of continuous exposure for the MNU and 20.0 mmol/L DCA
33 treated groups. Although derived from very few animals, the 6.67 mmol/L DCA group
34 reported no change for foci/mouse but a decrease in the incidence of foci between 31- and
35 52-weeks of exposure in MNU treated mice (i.e., 0.67 ± 0.18 foci/mouse in 50% of the animals
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 87 DRAFT—DO NOT CITE OR QUOTE
-------
1 at 31 weeks and 0.50 ± 0.34 foci/mouse in 20% of mice treated for 52 weeks). The numbers of
2 foci/mouse for both MNU-treated and untreated control mice were reported to be decreased
3 between 31 and 51 weeks as well.
4 As noted in Section E.3.1.8. the number of "nodules" in humans, which may be
5 analogous to foci and adenomas, can spontaneously regress with time rather than becoming
6 hepatocellular carcinomas. Also as tumors get larger with progression, the number of
7 tumors/mouse can decrease due to coalescence of tumors and difficulty distinguishing between
8 them. While data are suggestive of a decrease in the number of adenomas/mouse after
9 cessation of DC A exposure, the incidence data are similar between the 31-week exposure and
10 recovery groups. Of note is that the number of carcinomas/mouse and the incidence of
11 carcinomas was reported to be similar between the MNU-treated mice exposed continuously to
12 20.0 mmol/L DCA for 52 weeks and those which were treated for 31 weeks and then sacrificed
13 at 52 weeks. Also of note is that, although incidences and multiplicities of foci and adenomas
14 was reported to be relatively low in the 2.0 mmol/L DCA exposure groups, at 52-weeks 40% of
15 the mice tested had carcinomas with 0.70 ± 0.40 carcinomas/mouse. This was a greater
16 percentage of animals with carcinomas and multiplicity than that reported for the highest dose
17 of DCA. This result suggests that the effects in regard to tumor progression, and specifically
18 for carcinoma induction, differ between the lowest and highest doses used in this experiment.
19 However, the low numbers of animals examined for the lower doses, 31-weeks exposures, and
20 in the recovery group decrease the confidence in the results of this study in regard to the effects
21 of cessation of exposure on tumor progression.
22 In regard to tumor phenotype, in MNU-treated female mice that were not also exposed
23 to either DCA or TCA, all four foci and 86.7% of 15 adenomas were reported to be basophilic
24 and 13.3% eosinophilic at the end of the 52 week-study. However, when MNU-treated female
25 mice were also exposed to DCA the number eosinophilic foci and tumors increased with
26 increasing dose after 52 weeks of continuous exposure. At the 20.0 mmol/L level all 38 foci
27 examined were eosinophilic and 99% of the tumors (almost all adenomas) were eosinophilic.
28 At the 2.0 mmol/L DCA exposure there were no foci examined but about 5 of 9 tumors
29 examined (~2:1 carcinoma:adenoma ratio) were basophilic and the other 4 were eosinophilic.
30 For TCA coexposure in MNU-treated mice, the 20 mmol/L TCA treatment was reported to
31 give results of 1 of the 3 foci examined to be basophilic and 2 that were eosinophilic. For the
32 98 tumors examined (~2:1 carcinoma/adenoma ratio) 71.4% were reported to be basophilic and
33 28.6% were eosinophilic. At the 2.0 mmol/L TCA exposure level, the 2 foci examined were
34 reported to be basophilic while the 6 tumors (all adenomas) were reported to be 50%
35 eosinophilic and 50% basophilic. Thus, after 52 weeks female mice treated with MNU and a
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-388 DRAFT—DO NOT CITE OR QUOTE
-------
1 high dose of DC A had eosinophilic foci and adenomas and those treated with the high dose of
2 TCA had a mixture of basophilic and eosinophilic foci and tumors with a 3:1 ratio of tumors
3 (mostly carcinomas) being basophilic. At the lower doses of either DCA or TCA the tumors
4 tended to be mostly carcinomas for DCA and adenomas for TCA but both were -50%
5 basophilic and 50% eosinophilic. The tumors observed from MNU treatment alone were all
6 adenomas and mostly 87% basophilic. Thus, not only did treatment concentrations of DCA
7 and TCA give a different result for tumor multiplicity and incidence, but also for tumor
8 phenotype in MNU treated female mice. Eosinophilic foci and tumors were reported to be
9 consistently GST-Ti positive while basophilic lesions "did not contain GST-Ti, except for a few
10 scattered cells or very small area comprising less than 5% of the tumor."
11 Thus, exposure to either DCA or TCA increased incidence and number of animals with
12 lesions (foci, adenomas, or carcinomas) in MNU- versus nontreated mice (see
13 Section E.2.3.2.6, above). These results suggest that the pattern of foci, adenoma and
14 carcinoma incidence, multiplicity, and progression appeared to differ between TCA and DCA
15 in MNU-treated female mice. However, the low and variable number of animals used in this
16 study, make quantitative inferences between DCA and TCA exposures in "initiated" animals,
17 problematic.
18
19 E.4.2.4. Taoetal.,2000
20 The source of liver tumors for this analysis was reported to be the study of Pereira and
21 Phelps (1996). Samples of liver "tumors" and "noninvolved" liver was homogenized for
22 protein expression for c-Jun and c-Myc and therefore, contained homogeneous cell types for
23 study. The term "liver tumors" was not defined so it cannot be ascertained as to whether the
24 lesions studied were altered foci, hepatocellular adenomas, or carcinomas. Liver tissues were
25 reported to be frozen prior to study which raises issues of m-RNA quality. Although this study
26 reports that there were no MNU-induced "tumors" the original paper of Pereira and Phelps
27 (1996) reports that there were four foci and 15 adenomas in MNU-only treated mice. The
28 authors reported no difference in c-Jun and c-Myc m-RNA from DCA or TCA-induced tumors
29 from mice "initiated" with MNU. DNA methyltransferase was reported to be decreased in
30 noninvolved liver in MNU-only treated mice in comparison to that from TCA- and DCA-
31 treated mice. For a comparison between noninvolved liver and tumors, tumors were reported
32 to have a greater level than did noninvolved liver.
33
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-3 89 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.4.2.5. Lantendresse andPereira, 1997
2 This study used the tumors from Pereira and Phelps (1996), except for the MNU-treated
3 only groups and those groups treated with either DCA or TCA but not MNU initiation, to further
4 study various biomarkers. The omissions were cited as to be due to insufficient tissue. For
5 immunohistochemical evaluation of the molecular biomarkers other than GST-u, liver
6 specimens from 7 MNU/20.0 mmol DCA- (i.e., 2.58 g/L DCA) treated and 6 MNU/20.0 mmol
7 TCA - (i.e., 3.27 g/L TCA) treated female mice randomly selected. For GST-u, the number of
8 animals from which lesion specimens were derived, was 24 MNU/DCA-treated and
9 23 MNU/TCA-treated mice. The DCA treated mice were reported to have 1-9 lesions/mouse
10 and TCA treated mice 1-3 lesions/mouse. The number of lesions examined for each biomarker
11 varied greatly. For TCA-induced foci, no foci were examined for any biomarker except
12 3 lesions for GST-u, while for DCA 12-15 foci were examined for each biomarker and
13 38 lesions examined for GST-u. Similarly for TCA-induced adenomas, there were 8-10 lesions
14 examined for all biomarkers with 32 lesions examined GST-u, while for DCA 12 lesions for all
15 biomarkers with 94 lesions examined for GST-u. Finally, for TCA-induced carcinomas there
16 were 3-4 lesions examined per group with 64 lesions examined for GST-u, while for DCA-
17 induced carcinomas there were no lesions examined for any biomarker except 3 examined for
18 GST-u. The biomarkers used were: GST-u, TGF-a, TGF-P, c-Jun, c-Fos, c-Myc, cytochrome
19 oxidase CYP2E1, and cytochrome oxidase CYP4A1.
20 MNU/DCA treatment was reported to produce "predominantly eosinophilic lesions" with
21
22 in general, the hepatocytes of DCA-promoted foci and tumors were less
23 pleomorphic and uniformly larger and had more distinctive cell borders than the
24 hepatocytes in lesions caused by TCA. Parenchymal hepatocytes of DCA-
25 promoted mice were uniformly hypertrophied, with prominent cell borders, and
26 the cytoplasm was markedly vacuolated, which was morphologically consistent
27 with the previous description of glycogen deposition in these lesions. In contrast,
28 TCA-promoted proliferative lesions tended to be basophilic, as previously
29 reported, and were composed of hepatocytes with less distinct cell borders, slight
30 cytoplasmic vacuolization, and greater variability in nuclear size and cellular size.
31
32 The hepatocytes of altered foci and hepatocellular adenomas from MNU-treated female
33 mice also treated with DCA were reported to stain positively for TGF-a, c-Jun, c-Myc,
34 CYP2E1, CYP4A1, and GST-u. The authors do not present the data for foci and adenomas
35 separately but as an aggregate and as the number of lesions with <50% cells stained or the
36 number of lesions with >50% cells stained either "minimally to mildly" or "moderately to
37 densely" stained. Because no carcinomas for DCA were examined and especially because no
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-390 DRAFT—DO NOT CITE OR QUOTE
-------
1 foci for TCA analyses were included in the aggregates, it is difficult to compare the profile
2 between TCA and DCA exposure in initiated animals and to separate these results from the
3 effects of differences in tumor progression. Thus, any differences seen in these biomarkers due
4 to progression from foci to adenoma in DC A-induced lesions or from progression of adenoma to
5 carcinoma in TCA-induce lesions, was lost. If the results for adenomas had been reported
6 separately, there would have been a common stage of progression from which to compare the
7 DCA and TCA effects on initiated female mice liver tumors. For DCA-induced "lesions"
8 (-50% foci and -50% adenomas), most lesions had >50% cells staining with moderate to dense
9 levels for TGF-a, and CYP2E1, CYP4A1, and GST-7i and most lesions had <50% cells staining
10 for even minimally to mild staining for TGF-P and c-Fos. For c-Jun and c-Myc the aggregate
11 DCA-induced "lesions" were heterogeneous in the amount of cells and the intensity of cell
12 staining for these biomarkers in MNU-treated female mice.
13 For the TCA "lesions" (-60% adenomas and -30% carcinomas) the authors note that
14
15 in general, the hepatocytes of tumors promoted by TCA demonstrated variable
16 immunostaining. With the exception of c-Jun, greater than 50% of the
17 hepatocytes in TCA lesions were essentially negative or stained only minimally to
18 mildly for the protein biomarkers studies. In some instances, particularly in TCA-
19 promoted tumors, there was regional staining variability within the lesions,
20 including immunoreactivity for c-Jun and c-Myc proteins, consistent with clonal
21 expansion or tumor progression.
22
23 As stated above, the term "lesion" refers to foci and adenomas for DCA but for adenomas and
24 carcinomas for TCA making inferences as to differences in the actions of the two compounds
25 through the comparisons of biomarkers confounded by the effects of tumor progression. The
26 largest differences in patterns between TCA induced "lesions" and those by DCA appeared to be
27 TGF-a (with no lesions having >50% cells stained mildly or moderately/densely for TCA-
28 induced lesions), CYP2E1 (with few lesions having >50% stained moderately/densely for TCA-
29 induced lesions), CYP4A1 (with no lesions having >50% stained mildly or moderately/densely
30 for TCA-induced lesions), and GST-u (with all lesions having <50% cells stained even mildly
31 for TCA-induced lesions). However, as shown by these data, while the "lesions" induced by
32 TCA and DCA had some commonalities within each treatment, there was heterogeneity of
33 lesions produced by both treatments in female mice already exposed to MNU. Overall, the
34 tumor biomarker pattern suggests differences in the effects of DCA and TCA through
35 differences in tumor phenotype they induce as coexposures with MNU treated female mice.
36 The authors note that nonlesion parenchymal hepatocytes in DCA-treated initiated mice
37 stained mostly negative for CYP2E1 and CYP4A1, while in TCA-treated mice staining patterns
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-391 DRAFT—DO NOT CITE OR QUOTE
-------
1 in parenchymal nonlesions hepatocytes were centrilobular for CYP2E1 and panlobular for
2 CYP4A1 (a pattern for CYP4A1 that is opposite of that found in the TCA-induced lesions).
3
4 E.4.2.6. Perdraetal, 1997
5 This study used a similar paradigm as that of Pereira and Phelps (1996) to study
6 coexposures of TCA and DCA to female B6C3F1 mice already exposed to MNU. At 15 days
7 the mice received 25 mg/kg MNU and starting at 6 weeks of age neutralized solutions of either
8 0, 7.8, 15.6, 25.0 mmol/L DCA (n = 30 for control and 25 mmol/L DCA and n = 20 for 7.8 and
9 15.6 mmol/L DCA), 6.0 or 25.0 mmol/L TCA (n = 30 for 25.0 mmol/L TCA and n = 20 for
10 6.0 TCA), or combinations of DCA and TCA that included 25.0 mmol/L TCA + 15.6 mmol/L
11 DCA (n = 20), 7.8 mmol/L DCA + 6.0 mmol/L TCA (n = 25), 15.6 mmol/L DCA + 6.0 mmol/L
12 TCA (45), 25.0 mmol/L DCA + 6.mmol/L TCA (n = 25). The corresponding concentrations of
13 DCA and TCA in g/L is 25 mmol = 3.23 g/L, 15.6 mmol = 2.01 g/L and 7.8 mmol = 1.01 g/L
14 DCA and 25 mmol = 4.09 g/L and 6.0 mmol = 0.98 g/L TCA. Accordingly, the number of
15 animals at the beginning of the study varied between 20 and 45. At terminal sacrifice (after
16 44 weeks of exposure) the numbers of animals examined were less with the lowest number
17 examined to be 17 mice in the 7.8 mmol/L DCA group and the largest to be 42 in the
18 15.6 mmol/L DCA + 6.0 mmol/L TCA exposed group.
19 The authors reported that only a total of eight hepatocellular carcinomas were found in
20 the study (i.e., 25.0 mmol/L DCA induced 3 carcinomas, 7.8 mmol DCA + 6.0 mmol TCA
21 induced one carcinoma, and 25.0 mmol/L TCA induced 4 carcinomas). Thus, they presented
22 data for foci/mouse, and adenomas/mouse and their sum of both as "total lesions." The
23 incidences of lesions (i.e., how many mice in the groups had lesions) were not reported. The
24 shortened duration of exposure (i.e., 44 weeks), the omission of carcinomas from total "lesion"
25 counts (precluding consideration of progression of adenomas to carcinomas), the lack of
26 reporting of tumor incidences between groups, and the variable and low numbers of animals
27 examined in each group make quantitative inferences regarding additivity of these treatments
28 difficult. MNU treated mice did have a neoplastic response, albeit low using this paradigm. For
29 mice that were only exposed to MNU (n = 30 at terminal sacrifice) the mean number of foci,
30 adenomas and "lesions" per mouse were 0.21, 0.07 and 0.28, respectively. No data were given
31 for mice without MNU treatment but few lesions would be expected in controls. Pereira and
32 Phelps (1996) reported that saline-only treatment in 40 female mice for 51 weeks resulted in 0%
33 foci, 0.03 adenomas/mouse in 2.5% of mice, and 0% carcinomas. In general, it appeared that
34 the numbers of foci, adenomas and the combination of both reported as "lesions" per mouse that
35 would have been predicted by the addition of multiplicities given for DCA, TCA, and MNU
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-392 DRAFT—DO NOT CITE OR QUOTE
-------
1 treatments alone, were similar to those observed as coexposure treatments. The largest numbers
2 of foci and adenomas/mouse were reported for the 25.0 mmol/L DCA and 6.0 mmol/L TCA
3 treatments in MNU treated mice (mean of 6.57 "lesions"/mouse) with the lowest number
4 reported for 7.8 mmol/L DCA and 6 mmol/L TCA (mean of 1.16 "lesions'Vmouse).
5 The authors reported that the foci of altered hepatocytes were predominantly eosinophilic
6 in DCA-treated female mice initiated with MNU, while those observed after MNU and TCA
7 treatment were basophilic. MNU treatment alone induced 4 basophilic and 2 eosinophilic foci,
8 and 2 basophilic adenomas. MNU and DCA treatment was reported to produce only
9 eosinophilic foci and adenomas at the 25.0 mmol/L DCA exposure level. At the 7.8 mmol/L
10 DCA level of treatment in MNU-treated mice, 2 foci were basophilic, 4 were eosinophilic and
11 the 1 adenoma observed was reported to be eosinophilic. Thus, the concentration of exposure
12 appeared to alter the tincture of the foci observed after MNU and DCA exposure using this
13 paradigm. In this study, MNU and TCA treatment was reported to induce foci and adenomas
14 that were all basophilic at both 25.0 mmol/L TCA and 6.0 mmol/L TCA exposures. After
15 7.8 mmol/L DCA + 6.0 mmol/L TCA exposure, 2/23 foci were basophilic and 21/23 foci were
16 reported to be eosinophilic while all 4 adenomas reported for this group were eosinophilic.
17 Irrespective of treatment, eosinophilic foci for were reported to be GST-u positive and
18 basophilic foci to be GST-u negative. An exception was the 4 carcinomas in the group treated
19 with 25 mmol/L TCA which were reported to be predominantly basophilic but contained small
20 areas of GST-u positive hepatocytes.
21 It should be noted that the increased dose (up to 3.23 g/L DCA and 4/09 g/L TCA) raises
22 issues of toxicity and effects on water consumption as other studies have noted toxicity at highly
23 doses of DCA and TCA. The use of an abbreviated duration of exposure in the study raises
24 issues of sensitivity of the bioassay at the lower doses used in the experiment. In particular, was
25 enough time provided to observe the full development of a tumor response? Finally, a question
26 arises as what can be concluded from the low numbers of foci examined in the study and the
27 affect of such low numbers on the ability to discern differences in these foci by treatment. As
28 with Pereira and Phelps, there appeared to be a difference the nature of the response induced by
29 coexposure of MNU to relatively high versus low DCA concentrations. Of note is that while
30 this experiment reported no hepatocellular carcinomas at the lowest dose of DCA at 44 weeks
31 (7.8 mmol DCA), Pereira and Phelps (1996) reported that in 9 mice treated with MNU and
32 2.0 mmol DCA for 52 weeks, there were no foci but 20% of mice had adenomas
33 (0.20 adenomas/mouse) and 40% of mice had carcinomas (0.70 carcinomas/mouse).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-393 DRAFT—DO NOT CITE OR QUOTE
-------
1 These results suggest that DCA coexposure affects TCA-induced lesions. The authors
2 concluded that mixtures of DCA and TCA appear to be at least additive and likely synergistic
3 and similar to the pathogenesis of DCA.
4
5 E.4.2.7. Tao et al, 1998
6 The focus of this study was an examination of tumors resulting from MNU and DCA or
7 TCA exposure in mice with the source of tumors was reported to be the study of Pereira et al.
8 (1997). Thus, similar concerns discussed above for that study paradigm are applicable to the
9 results of this study. The authors stated that there were also two recovery groups in which
10 exposure was terminated 1 week prior to euthanization at Week 44. The Pereira et al. (1997)
11 study does not report a cessation group in the study. In this study the number of animals treated
12 in the cessation group, the incidences of tumors in the mice, and the number of tumors examined
13 were not reported. Another group of female B6C3F1 mice (7-8 weeks old) were reported to not
14 be administered MNU but given 25 mmol/L DCA (3.23 g/L DCA), 25 mmol TCA (4.09 g/L
15 TCA), or control drinking water for 11 days (n = 7).
16 Hepatocellular adenomas in DCA-treated mice, adenomas and carcinomas in TCA-
17 treated mice were reported to be analyzed for percent-5-methylcytosine in the DNA of tumor
18 tissues. The levels of 5-methylcytosine in liver DNA of mice administered DCA or TCA for
19 11 days were reported to be reduced in comparison to control tissues (reduced to -36% of
20 control for DCA and -41% of control for TCA with the control value reported to be -3.5% of
21 DNA methylated). The number of animals examined was reported to be 7-10 animals per
22 group.
23 For control liver from mice that had received MNU but not DCA or TCA, and
24 noninvolved liver after 44 weeks of exposure to either, the levels of 5-methylcytosine were
25 similar and not different from the -3.5% of DNA methylated in untreated mice in the 11-days
26 experiment. Thus, initial decreases in methylated DNA shown by exposure to DCA or TCA
27 alone for 11 days, were not observed in "noninvolved" liver of animals exposed to either DCA
28 or TCA and MNU.
29 In regard to tumor tissues, the level of 5-methylcytosine in DNA of hepatocellular
30 adenomas receiving DCA and MNU was reported to be decreased by 36% in comparison to
31 noninvolved liver from the same animals. When exposure to DCA was terminated for 1 week
32 prior to sacrifice the level of 5-methylcytosine in the adenomas was reported to be higher and no
33 longer differed statistically from the noninvolved liver from the same animal or liver from
34 control animals only administered MNU. The number of samples was reported to be
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-394 DRAFT—DO NOT CITE OR QUOTE
-------
1 9-16 samples without identification as to how many samples were used for each tumor analysis
2 or how many animals provided the samples (i.e., were most of the adenomas from on animal?)
3 For TCA the 5-methylcytosine level was reported to be reduced by 40% in hepatocellular
4 adenomas and 51% reduction in hepatocellular carcinomas in comparison to noninvolved liver
5 from the same animals. These levels were also reported to be less than that the control animals
6 administered only MNU. Termination of exposure to TCA 1 week prior to sacrifice was
7 reported to not produce a statistically significant change in the level of 5-methylcytosine in
8 either adenomas or carcinomas. The levels of 5-methylcytosine were reported to be lower in
9 carcinomas than adenomas (-20% reduction) and to be lower in the "recovery" carcinomas than
10 continuous carcinomas (-25%) but were not reported as statistically significant. The results are
11 reported to have been derived from 8-16 "samples each." Again information on the number of
12 animals with tumors, whether the tumors were from primarily from one animal, and which DNA
13 results are from 8 versus 16 samples, was not provided by the authors. Given that Pereira et al.
14 (1997), the source for material of this study, reported that treatment of MNU and 25.0 mmol/L
15 TCA treatment for 44 weeks induced only 4 carcinomas, a question arises as to how many
16 carcinomas were used for the 44-week 5-methylcytosine results in this study for carcinomas
17 (i.e., how can 8-16 samples arise from 4 carcinomas?). In addition, a question arises as to
18 whether there was a difference in tumor-response in those animals with and without one week of
19 cessation of exposure which cannot be discerned from this report. The use of highly variable
20 number of samples between analysis groups and lack of information as to how many tumors
21 were analyzed adds uncertainty to the validity of these findings. There did not appear to be a
22 difference in methylation activity from short-term exposure to either DCA or TCA alone in
23 whole liver DNA extracts. However, the authors conclude that the difference in methylation
24 status between tumors resulting from MNU and DCA or TCA exposures supports differences in
25 the action between DCA and TCA.
26
27 E.4.2.8. Stauber et al, 1998
28 In this study, 5-8 week old male B6C3F1 mice were used for isolation of primary
29 hepatocytes which were subsequently isolated and cultured in DCA or TCA. In a separate
30 experiment 0.5 g/L DCA was given to mice as pretreatment for 2 weeks prior to isolation. The
31 authors note that and indication of an "initiated cell" is anchorage-independent growth. DCA
32 and TCA solutions were neutralized before use. The primary hepatocytes from 3 mice per
33 concentration were cultured for 10 days with DCA or TCA colonies (8 cells or more)
34 determined in quadruplicate. The levels of DCA used were 0, 0.2, 0.5 and 2.0 mM DCA or
35 TCA. At concentrations of 0.5 mM or more DCA and TCA both induced an increase in the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-395 DRAFT—DO NOT CITE OR QUOTE
-------
1 number of colonies that was statistically significant and increased with dose with DCA giving a
2 slightly greater effect. The authors noted that concentrations greater than 2.0 mM were
3 cytotoxic but did not show data on toxicity for this study.
4 Of great interest is the time-course experiment from this study in which the number of
5 colonies from DCA treatment in vitro peaked by 10 days and did not change through days
6 15-25 at the highest dose. For the lower concentrations of DCA, increased time in culture
7 induced similar peak levels of colony formation by days 20-25 as that reached by 10 days at the
8 higher dose. Therefore, the number of colonies formed was independent of dose if the cells
9 were treated long enough in vitro. The number of colonies that formed in control hepatocyte
10 cultures also increased with time but at a lower rate than those treated with DCA (2.0 mM DCA
11 gave ~2-fold of control by 25 days of exposure to hepatocytes in culture). However, the level
12 reached by cells untreated in tissue culture alone by 20 days was similar to the level induced by
13 0.5 mM DCA by 10 days of exposure. This finding raises the issue of what these "colonies"
14 represent as tissue culture conditions alone transform these cells to what the authors suggest is
15 an "initiated" state. TC A exposure was not tested with time to see if it had a similar effect with
16 time as did DCA.
17 At 10 days, colonies were tested for c-Jun expression with the authors noting that
18 "colonies promoted by DCA were primarily c-Jun positive in contrast to TCA promoted
19 colonies that were predominantly c-Jun negative." For colonies that arose spontaneously from
20 tissue culture conditions, 10/13 (76.9%) were reported to be c-Jun +, those treated with DCA
21 28/34 (82.3%) were c-Jun +, and those treated with TCA 5/22 (22.7%) were c-Jun +. These
22 data show heterogeneity in cell in colonies although more were c-Jun + with DCA than TCA.
23 The number of colonies reported in the c-Jun labeling results represent sums between
24 experiments and thus, present total numbers of the control and the of colonies derived from
25 doses of DCA and TCA at 0.2 to 2.0 mM at 10 days. Thus, changes in colony c-Jun+ labeling
26 due to increasing dose cannot be determined. The authors reported that with time (24, 48, 72,
27 and 96 hours) of culture conditioning the number of c-Jun+ colonies was increased in untreated
28 controls. DCA treatment was reported to delay the increase in c-Jun+ expression induced by
29 tissue culture conditions alone in untreated controls. TCA treatment was reported to not affect
30 the increasing c-Jun+ expression that increased with time in tissue culture. In this instance,
31 tissue culture environment alone was shown to transform cells and can be viewed as a
32 "coexposure." DCA pretreatment in vivo was reported to increase the number of colonies after
33 plating which reached a plateau at 0.10 mM and gave changes as at low a concentration of
34 0.02mM DCA administered in vitro. The background level of colony formation varied between
35 controls (i.e., 2-fold different in pretreatment experiments and nonpretreatment experiments).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-396 DRAFT—DO NOT CITE OR QUOTE
-------
1 Therefore, although the number of colonies was greater for pretreatment with DCA, the
2 magnitude of difference over the control level was the same after DCA treatment in vitro with
3 and without pretreatment.
4 The authors presented a comparison of "tumors" from Stauber and Bull (1997) and state
5 that DCA tumors were analyzed after 38 weeks of treatment but that TCA tumors were analyzed
6 after 52 weeks. They note that 97.5% of DCA-induced "tumors" were c-Jun + while none of the
7 TCA-induced "tumors" were c-Jun +. The concentrations used to give tumors in vivo for
8 comparison with in vitro results were not reported. What was considered to be "tumors" from
9 the earlier report for this analysis was also not noted. Stauber and Bull (1997) reported results
10 for combination of foci and tumors raising issues as to what was examined in this report. The
11 authors stated that because of such short time, no control tumors results were given. The short
12 and variable time of duration of exposure increases the possibility of differences between the in
13 vivo data resulting from differences in tumor progression as well as a decreased ability by the
14 shortened time of observation for full expression of the tumor response.
15
16 E.4.3. Coexposures of Haloacetates and Other Solvents
17 As noted by Caldwell et al. (2008b), drinking water exposure data suggest coexposure of
18 TCE and its haloacetic acid metabolites, TCA and DCA, is not an uncommon event as DCA and
19 TCA are the two most abundant haloacetates in most water supplies (Weisel et al., 1999;
20 Boorman et al., 1999). Dibromoacetic acid (DBA) concentrations have also been reported to
21 range up to approximately 20 ug/L in finished water and distribution systems (Weinberg et al.,
22 2002). Caldwell et al. (2008b) have also noted that coexposure in different media also occurs
23 with solvents like perchloroethylene (PERC) that may share some MO As, targets of toxicity,
24 and common metabolites that can therefore, potentially affect TCE health risk (Wu and Schaum,
25 2000). Some of the information contain in the following sections have been excerpted from the
26 discussions by Caldwell et al. (2008b) regarding the implications for the risk of TCE exposure
27 as modulated by coexposures to haloacetates and other solvents that have been studied and
28 reported in the literature.
29
30 E.4.3.1. Carbon tetrachloride, Dichloroacetic Acid (DCA), Trichloroacetic Acid (TCA):
31 Implications for Mode of Action (MOA)from Coexposures
32 Studies of specific combinations of TCE and chemicals colocated in contaminated areas
33 have been reported by Caldwell et al. (2008b). For carbon tetrachloride
34
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-397 DRAFT—DO NOT CITE OR QUOTE
-------
1 Pretreatment with TCE in drinking water at levels as low as 15 mM for three days
2 has been reported to increase susceptibility to liver damage to subsequent
3 exposure to a single IP injection of 1 mM/kg carbon tetrachloride (CCU) in
4 Fischer 344 rats [Steup et al., 1991]. Potential mechanistic explanations for this
5 observation included altered metabolism, decreased hepatic repair capability,
6 decreased detoxification ability, or combination of one or more of the above
7 activities. Simultaneous administration of an oral dose of TCE (0.5ml/kg) has
8 also been reported to increase the liver injury induced by an oral dose of 0.05
9 ml/kg CCU [Steup et al., 1993]. The authors suggested that TCE appeared to
10 impair the regenerative activity in the liver, thus leading to increased damage
11 when CCU is given in combination with TCE.
12
13 As discussed above in Section E.4.2, initiation studies are in themselves a coexposure.
14 The study of Bull et al. (2004) is included here as it not only used a coexposure of vinyl
15 carbamate with TCE metabolites, but also used carbon tetrachloride as a coexposure as well.
16 The rationale for this approach was that coexposure of TCE (and therefore, to its metabolites)
17 and CCU are likely to occur as they are commonly found together at contaminated sites. Bull et
18 al. (2004) hypothesized that modification of tumor growth rates is an indication of promotion
19 rather than effects on tumor number, and that by studying tumor growth rates they could classify
20 carcinogens by their MO As. B6C3F1 male mice were initiated with vinyl carbamate (3 mg/kg)
21 at 2 weeks of age and then treated with DCA, TCA, CC14, (0.1, 0.5, or 2.0 g/L for DCA and
22 TCA; 50, 100 or 500 mg/kg CCL4 in 5% Alkamuls via gavage) in pair-wise combinations of the
23 three for 18 to 36 weeks. The exposure level of CCL4 to 5, 20 and 50 mg/kg was reported to be
24 reduced at Week 24 due to toxicity for CC14. The number of mice in each group was reported to
25 be 10 with the study divided into 5 segments. There were evidently differences between
26 treatment segments as the authors state that "because of some significant quantitative
27 differences in results that were obtained with replicate experiments treated in different time
28 frames, the simultaneous controls have been used in the analysis and presentation of these data."
29 As with Bull et al. (2002), the interpretation of the results of the study is limited by a low
30 number of animals per group, short duration time of exposure and limited examination and
31 reporting of results. For example, a sample of 100 out of the 8,000 lesions identified in the
32 study was examined to verify that the general descriptor of neoplastic and nonneoplastic lesion
33 was correctly labeled with "tumors" describing a combination of hyperplastic nodules,
34 adenomas, and carcinomas. No incidence data were reported by the authors, but general lesion
35 growth information included mean lesion volume and multiplicity of lesions (numbers of
36 lesions/mouse). Using these reported indices, there appeared to be differences in treatment-
37 related effects.
38
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-398 DRAFT—DO NOT CITE OR QUOTE
-------
1 As discussed in Caldwell et al. (2008b):
2
3 Each treatment was examined alone and then in differing combinations with each
4 other. Mice initiated with vinyl-carbamate, but without further exposure to the
5 other toxicants, were reported to have a few lesions that were of small size during
6 the examination period (20-36 weeks). At 30 weeks of CC14 exposure, there was
7 a dose-related response reported for multiplicity but mean lesion size was smaller
8 at the highest dose in initiated animals. At 36 weeks, DCA exposure was reported
9 to increase multiplicity at the two highest exposure levels and increased lesion
10 size at all levels compared to initiated-only animals. However, at a similar level
11 of induction, multiplicity and mean size of those lesions resulting from DCA
12 treatment were reported to be much smaller in comparison with CC14 treatment
13 (i.e., a 20-fold difference for lesion volume). At 36 weeks, treatments with the
14 same concentration of TCA or DCA induced similar multiplicity, but the mean
15 lesion volume was reported to be approximately 4-fold greater in tumors induced
16 by DCA as compared to TCA, and in animals treated with DCA multiplicity had
17 reached a plateau by 24 weeks rather than 36 for those treated with TCA.
18
19 Thus, using multiplicity of lesions and lesion volume as indicators of differences in
20 MO A, exposure to CCU, DCA, and TCA appeared to produce distinct differences in results in
21 animals previously treated with vinyl carbamate.
22 As discussed in Caldwell et al. (2008b):
23
24 Simultaneous coexposure of differing combinations of CCU, DCA, and TCA were
25 reported to give more complex results between 24 and 36 weeks of observation
26 but to show that coexposure had effects on lesion multiplicity and volume in
27 initiated animals. At 36 weeks, TCA coexposure appeared to reduce the lesion
28 volume of either DCA- or CCU-induced lesions after vinyl carbamate treatment.
29 Similarly, DCA coexposure was reported to reduce the lesion volume of either
30 TCA- or CCU-induced lesions when each was given alone after vinyl carbamate
31 treatment. With regard to multiplicity, TCA coexposure was reported to reduce
32 DCA-induced multiplicity only at the lowest dose of TCA while coexposure with
33 DCA increased multiplicity of CCl4-induced lesions at all exposure levels. At 24
34 weeks, there appeared to be little effect on mean lesion volume by any of the
35 coexposures but DCA coexposure decreased multiplicity of TCA-induced lesions
36 (up to 3-fold) while TCA treatment slightly increased the number of CCU-induced
37 multiplicity (1.6-fold). This study confirms that short duration of exposure to all
38 three of these chemicals can cause lesions in already exposed to vinyl carbamate,
39 and suggests that combinations of these agents differentially influence lesion
40 number and growth rates. The authors have interpreted their results to indicate
41 differences in MOA between such treatments. However, the limitations of the
42 study limit conclusions regarding how such coexposure may be able to affect
43 toxicity and tumor induction and what the MOA is for each of these agents. This
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-399 DRAFT—DO NOT CITE OR QUOTE
-------
1 is especially true at lower and more environmentally relevant concentrations
2 given for longer durations to uninitiated animals.
3
4 E.4.3.2. Chloroform, Dichloroacetic Acid (DCA), and Trichloroacetic Acid (TCA)
5 Coexposures: Changes in Methylation Status
6 In Section E.3.4.2.2, information on the effects of TCE and its metabolites was presented
7 in regard to effects on methylation status. After 7 days of gavage dosing, TCE, TCA and DCA
8 were reported to increased hypomethylation of the promoter regions of c-Jun and c-Myc genes
9 in mouse whole liver DNA, however, Caldwell and Keshava (2006) concluded that
10 hypomethylation did not appear to be a chemical-specific effect at the concentration used. Bull
11 et al. (2004) suggested that hypomethylation occurs at higher exposure levels than those that
12 induce liver tumors for TCE and its metabolites. Along with studies of methylation changes
13 induced by a exposure to a single agent a Pereira et al. (2001) have attempted to examine the
14 effects on methylation changes from coexposures. This study was also reviewed by Caldwell et
15 al. (2008b).
16 Pereira et al. (2001) hypothesized that changes in the methylation status of DNA can be a
17 key event for MO A for DCA- and TCA-induced liver carcinogenicity through changes in gene
18 regulation, and that chloroform (CHCb) coexposure may result in modification of DNA
19 methylation. As discussed in Caldwell et al. (2008b),
20
21 After 17 days of exposure of exposure to CHC13 (0, 400, 800, 1,600 mg/L, n = 6
22 mice per treatment group) female B6C3F1 mice were coexposed to DCA or TCA
23 (500 mg/kg) during the last 5 days of exposure to chloroform. As noted by
24 Caldwell et al. (2007b), Pereira et al. (2001) reported the effects of
25 hypomethylation of the promoter region of c-Myc gene and on expression of its
26 mRNA in the whole livers of female B6C3F1 mice and thus, these results
27 represent composite changes in DNA methylation status from a variety of cell
28 types and for hepatocytes lumped from differing parts of the liver lobule. When
29 given alone, DCA, TCA, and to a lesser extent, the highest concentration of
30 CHCls (1,600 mg/L), was reported to decrease methylation of the c-myc promoter
31 region. Coadministration of CHC13 (at 800 and 1,600 mg/L) was reported to
32 decrease DCA-induced hypomethylation while exposure to CHCb had no effect
33 on TCA-induced hypomethylation. Treatment with DCA, TCA, and, to a lesser
34 extent CHCb, was reported to increase levels of c-myc mRNA. While expression
35 of c-myc mRNA was increased by DCA or TCA treatment, increasing
36 coexposures to CHCb were reported to attenuate the actions of DCA but not
37 TCA. Thus, differences in methylation status and expression of the c-myc gene
38 induced by DCA or TCA exposure was reported to be differentially modulated by
39 coexposure to CHC13. The authors suggest these differences support differing
40 actions by DCA and TCA. However, whether these changes represent key events
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-400 DRAFT—DO NOT CITE OR QUOTE
-------
1 in the induction of liver cancer is a matter of debate, especially as a "snapshot in
2 time" approach for such a nonspecific endpoint.
3
4 In a coexposure study in which an "initiating agent" was used as a coexposure along with
5 other coexposure, Pereira et al. (2001) treated male and female 15-day old B6C3F1 mice with
6 MNU (a cause of liver and kidney tumors) and then, starting at 5 weeks of age, treated them
7 with DCA (3.2 g/L) or TCA (4.0 g/L) along with coexposure to CHC13 (0, 800, or 1,600 mg/L)
8 for 36 weeks. Mice were reported to be examined for evidence of promotion of liver and kidney
9 tumors. The numbers of animals in the exposure groups were highly variable, ranging from 25
10 (female initiated mice exposed to DCA) to 6 (female initiated mice exposed to DCA and
11 1,600 mg/L CHCb), thus, limiting the power of the study to ascertain treatment-related changes.
12 However, unlike Bull et al. (2004), all liver tissues were examined with incidences of foci,
13 adenomas, carcinomas, and both adenoma and carcinoma reported separately for treatment
14 groups. Multiplicity for a combination of adenomas and carcinomas were reported as well as
15 the tincture of foci and tumors.
16 Although as noted by Caldwell et al. (2008b):
17
18 [T]he statistical power of the study to detect change was very low, an examination
19 of the pattern of tumors induced by coexposure to MNU and TCE metabolites in
20 female mice suggested that: (1) DCA exposure increased the incidence of
21 adenomas but not carcinomas; (2) TCA increased incidence of carcinomas with
22 little change in adenoma incidence; (3) coexposure to 800 and 1,600 mg/L of
23 CHCls decreased adenoma incidence by DCA treatment but not TCA; and (4)
24 CHC13 coexposure decreased multiplicity of TCA-induced tumors and foci, but
25 not for DCA. Caldwell et al. (2008) also note that this study suggests a gender-
26 related effect on tumor induction from this study with; (1) adenoma incidences
27 similar in male and female mice treated with DCA, but carcinoma incidence
28 greater in males; (2) adenoma and carcinoma incidences greater in males than
29 females treated with TCA; (3) tumor multiplicity similar in both genders for DCA
30 treatments, but lower in females mice for TCA; and (4) less of an inhibitory effect
31 by CHCb on adenoma incidence from DCA exposure in male mice.
32
33 Pereira et al. (2001) also described the tinctural characteristics of the specific lesions
34 induced by their coexposure treatments. Both foci and tumors induced by DCA exposure in
35 "initiated" mice were reported to be over 95% eosinophilic in females, while in males, 89% of
36 the foci were eosinophilic and 91% of tumors were basophilic. Thus, not only was there a
37 gender-related difference in the incidences of tumors and foci but also foci and tumor
38 phenotype. CHCb coexposure was reported to change the DC A-induced foci from primarily
39 eosinophilic to basophilic (i.e., 11 vs. 75% basophilic) in male mice coexposed to MNU. In
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-401 DRAFT—DO NOT CITE OR QUOTE
-------
1 male and female mice, TCA-induced tumors and foci were basophilic with no effect
2 on phenotype in MNU treated mice.
3
4 E.4.3.3. Coexposures to BrominatedHaloacetates: Implications for Common Modes of
5 Action (MOAs) and Background Additivity to Toxicity
6 As noted by Caldwell et al. (2008b), along with chlorinated haloacetates and other
7 solvents, "coexposures with TCE and brominated haloacetates may occur through drinking
8 water. These compounds may affect TCE toxicity in a similar fashion to their chlorinated
9 counterparts. As bromide concentrations increase, brominated haloacetates increase in the water
10 supply."
11 Kato-Weinstein et al. (2001) administered dibromoacetate (DBA), bromochloroacetate
12 (BCA), bromodichloroacetate (BDCA), TCA, and DC A in drinking water at concentrations of
13 0.2-3 g/L for 12 weeks to B6C3F1 male mice. The focus of the study was to determine the
14 similarity in action between the brominated and chlorinated haloacetates. Each of the
15 haloacetates, given individually, were reported to increase liver/body weight ratios in a dose-
16 dependent manner. The dihaloactates, DCA, BCA and DBA, caused liver glycogen
17 accumulation both by chemical measurements in liver homogenates and in ethanol-fixed liver
18 sections (to preserved glycogen) stained with PAS. For DCA, a maximal level of glycogen
19 increase was observed at 4 weeks of exposure at a 2 g/L exposure concentration. They report a
20 1.60-fold of control percent liver/body weight and 1.50-fold of control glycogen content after
21 8 weeks of exposure to 2 g/L DCA in male B6C3F1 mice. The baseline level of glycogen
22 content (-60 mg/g) and the increase in glycogen after DCA exposure was consistent with the
23 results reported by Pereira et al. (2004). The percent liver/body weight data for control mice
24 was for animals sacrifice at 20 weeks of age. The 4-12 week exposure to DCA were staggered
25 so that all animals would be 20 weeks of age at sacrifice. Thus, the animals were at differing
26 ages at the beginning of DCA treatments between the groups. However, as with Pereira et al.
27 (2004) the -10% increase in liver mass that the glycogen increases represent are lower than the
28 total increase in liver mass reported for DCA exposure. The authors noted possible
29 contamination of BCA with small percentages of DCA and DBA in their studies (i.e., 84%
30 BCA, 6% DCA and 8% DBA). The trihaloacetates (TCA and low concentrations of BDCA)
31 were reported to produce slight decreases in liver glycogen content, especially in the central
32 lobular region in cells that tended to accumulate glycogen in control animals. These effects on
33 liver glycogen were reported at the lowest dose examined (i.e., 0.3 g/L). At the highest
34 concentration, BDCA was reported to induce a pattern of glycogen distribution similar to that of
35 DCA in mice.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-402 DRAFT—DO NOT CITE OR QUOTE
-------
1 All dihaloacetates were reported to reduce serum insulin levels at high concentrations.
2 Conversely, trihaloacetates were reported to have no significant effects on serum insulin levels.
3 For the study of peroxisome proliferation and DNA synthesis, mice were treated to BCA, DBA,
4 and BDCA for 2, 4, or 26 weeks. The effects on DNA synthesis were small for all brominated
5 haloacetates with only DBA reported to show a significant increase in DNA synthesis at 2 and 4
6 weeks but not at 26 weeks (increase in DNA synthesis was 3-fold of the highest control level).
7 Of note is the highly variable level of DNA synthesis reported for control values that varied to a
8 much higher degree (~3-6-fold variation within control groups at the same time points) than did
9 treatment-related changes. DBA was the only brominated haloacetate that was reported to
10 consistently increased PCO activity as a percentage of control values (actual values and
11 variability between controls were not reported) with a 2-3-fold increase in PCO activity at 0.3
12 to 3.0 g/L DBA. DBA-induced PCO activity increases were reported to be limited to 2-4 weeks
13 of treatment in contrast to TCA, which the authors reported to increase PCO activity
14 consistently over time.
15 Tao et al. (2004) reported DNA methylation, glycogen accumulation and peroxisome
16 proliferation after exposure of female B6C3F1 mice and male Fischer 344 rats exposed to 1 or
17 2 g/L DBA in drinking water for 2 to 28 days. DBA was reported to induce dose-dependent
18 DNA hypomethylation in whole mouse and rat liver after 7 days of exposure with suppression
19 sustained for the 28-day exposure period. The expression of mRNA for c-Myc in mice and rats
20 and mRNA expression of the IGF-II gene in female mice were reported to be increased during
21 the same period. Both rats and mice were reported to exhibit increased glycogen with mice
22 having increased levels at 2 day and rats at 4 days. DBA was reported to cause an increase in
23 lauroyl-CoA oxidase activity (a marker of peroxisome proliferation) in both mice (after 7 days)
24 and rats (after 4 days) that was sustained for 28 days. Methylation changes reported here for
25 DBA exposure in rats and mice are consistent with those reported for TCA and DCA by Pereira
26 et al. (2001) in mice. The pattern of glycogen accumulation was also similar to that reported for
27 DCA by Kato-Weinstein et al. (2001) and suggests that the brominated analogues of TCE
28 metabolites exhibited similar actions as their chlorinated counterparts. In regard to peroxisomal
29 enzyme activities Kato-Weinstein et al. (2001) reported PCO activity to be limited to 2-4 weeks
30 with Tao et al. (2004) reporting lauroyl-CoA oxidase activity to be sustained for the lengths of
31 the study (28-days) for DBA.
32 As noted by Caldwell et al. (2008b), "given the similarity of DCA and DBA effects, it is
33 plausible that DBA exposure also induces liver cancer. Melnick et al. (2007) reported the
34 results of DBA exposure to F344/N rats and B6C3F1 mice exposed to DBA for 3 months or
35 2 years in drinking water (0, 0.05, 0.5, or 1.0 g/L DBA for 2 years). Neoplasms at multiple sites
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-403 DRAFT—DO NOT CITE OR QUOTE
-------
1 were reported in both species exposed to DBA for 2 years with no effects on survival and little
2 effect on mean body weight in either species. Similar to TCE, DCA and TCA, the liver was
3 reported to be a target of DBA exposure. After 3-months of exposure, there were dose-related
4 increases in hepatocellular vacuolization and liver weight reported in rats and mice described as
5 'glycogen-like.'" The authors report that the major neoplastic effect of DBA in rats was
6 induction of malignant mesotheliomas in males and increased incidence of mononuclear cell
7 leukemia in males and females. For mice, the major neoplastic effect of DBA exposure was
8 reported to be the increased incidence of hepatocellular adenomas and carcinomas at all
9 exposure levels. In addition to these liver tumors, hepatoblastomas were also reported to be
10 increased in all exposure groups of male mice and exceeded historical control rates. The
11 incidence of alveolar/bronchiolar adenoma and carcinoma was reported to be increased in the
12 0.5 g/L group of male mice along with marginal increases in alveolar hyperplasia in
13 DBA-treated groups. The authors reported that the increases in hepatocellular neoplasms were
14 not associated with hepatocellular necrosis or regenerative hyperplasia and concluded that an
15 early increase in hepatocyte proliferation were not likely involved in the MOA for DBA because
16 no increases in hepatocyte DNA labeling index were observed in mice exposed for 26 days and
17 the slight increase that occurred in male F344 rats was not accompanied by an increase in liver
18 tumor response.
19 As noted by Caldwell et al. (2008b),
20
21 [T]he results of Kato-Weinstein et al. (2001), Tao et al. (2004), and Melnick et al.
22 (2007) are generally consistent for DBA and show a number of activities that may
23 be common to TCE metabolites (i.e., brominated and chlorinated haloacetate
24 analogues generally have similar effects on liver glycogen accumulation, serum
25 insulin levels, peroxisome proliferation, hepatocyte DNA synthesis, DNA
26 methylation status, and hepatocarcinogenicity). It is therefore, plausible that these
27 effects may be additive in situations of coexposure. However, as noted by
28 Melnick et al. (2007), methylation status, events associated with PPARa agonism,
29 hepatocellular necrosis, and regenerative hyperplasia are not established as key
30 events in the MOA of these agents, and the MO As for DCA- and DBA-induced
31 liver tumors are unknown.
32
33 E.4.3.4. Coexposures to Ethanol: Common Targets and Modes of Action (MOAs)
34 As noted in the U.S. EPA's draft TCE assessment (U.S. EPA, 2001), alcohol
35 consumption is a common coexposure that has been noted to affect TCE toxicity with TCE
36 exposure cited as potentially increasing the toxicity of methanol and ethanol, not only by
37 altering their metabolism to aldehydes, but also by altering their detoxification (e.g., similar to
38 the "alcohol flush" reported for those who have an inactive aldehyde dehydrogenase allele). As
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-404 DRAFT—DO NOT CITE OR QUOTE
-------
1 noted by Caldwell et al. (2008b) "chemical co-exposures from both the environment and
2 behaviors such as alcohol consumption may have effects that overlap with TCE in terms of
3 active agents, pharmacokinetics, pharmacodynamics, and/or target tissue toxicity."
4 Caldwell et al. (2008b) also note:
5
6 In their review of solvent risk (including TCE), Brautbar and Williams (2002)
7 suggest that laboratory testing that is commonly used by clinicians to detect liver
8 toxicity may not be sensitive enough to detect early liver hepatotoxicity from
9 industrial solvents and that the final clinical assessment of hepatotoxicity and
10 industrial solvents must take into account synergism with medications, drugs of
11 use and abuse, alcohol, age-dependent toxicity, and nutrition. Although many of
12 these factors may be important, the focus in this section is on the effects of
13 ethanol. Contemporary literature reports effects similar to those of TCE's and
14 previous reports indicate ethanol consumption impacts TCE toxicity in humans,
15 affects the pharmacokinetics and toxicity of TCE in rats, and is also a risk factor
16 for cancer.
17
18 The association between malignant tumors of the upper gastrointestinal tract and
19 liver and ethanol consumption is based largely on epidemiological evidence, and
20 thought to be causally related [Bradford et al., 2005; Badger et al., 2003].
21 Studies of the mechanisms of ethanol carcinogenicity have suggested the
22 importance of its metabolism, including induction of CYP2E1 associated
23 increases in production of reactive oxygen species and enhanced activation of a
24 variety of pro-carcinogens, alteration of retinol and retinoic acid metabolism, and
25 the actions of the metabolite acetaldehyde [Badger et al., 2003]. While ethanol is
26 primarily metabolized by alcohol dehydrogenase, it undergoes simultaneous
27 oxidation to acetate by hepatic P450s, primarily CYP2E1. Both chronic ethanol
28 consumption as well as TCE treatment induces CYP2E1 [U.S. EPA, 2001].
29 Oneta et al. (2002) report that even at moderate chronic ethanol consumption,
30 hepatic CYP2E1 is induced in humans, which they suggest, may be of
31 importance in the pathogenesis of alcoholic liver disease; of ethanol, drug, and
32 vitamin A interactions; and in alcohol-associated carcinogenesis. Induction of
33 CYP2E1 can cause oxidative stress to the liver from nicotinamide dinucleotide
34 phosphate (NADPH)-dependent reduction of dioxygen to reactive products even
35 in the absence of substrate, and subsequent apoptosis [Badger et al., 2003].
36 Bradford et al. (2005) suggest that CYP2E1, and not NADPH oxidase, is
37 required for ethanol-induced oxidative DNA damage to rodent liver but that
38 NADPH oxidase-derived oxidants are critical for the development of ethanol -
39 induced liver injury.
40
41 There is increasing evidence that acetaldehyde, which is toxic, mutagenic, and
42 carcinogenic, rather than alcohol is responsible for its carcinogenicity [Badger et
43 al., 2003]. Mitochondrial aldehyde dehydrogenase (ALDH2) disposes of
44 acetaldehyde generated by the oxidation of ethanol, and ALDH2 inactivity
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-405 DRAFT—DO NOT CITE OR QUOTE
-------
1 through mutation or polymorphism has been linked to esophageal cancer in
2 humans (everyday drinkers and alcoholics) [Badger et al., 2003]. For instance,
3 increased esophageal cancer risk was reported for patients with the ALDH3 * 1
4 polymorphism as well as increased acetaldehyde in their saliva. TCE exposure
5 has also been reported to induce a similar alcohol flush in humans which may be
6 linked to its ability to decrease ALDH activities at relatively low concentrations
7 and thus conferring a similar status to individuals with inactive ALDH2 allele
8 [Wang et al., 1999]. Whether the MO A for the buildup of acetaldehyde after
9 ethanol and TCE co-exposure is: (1) the induction of CYP2E1 by TCE resulting
10 in increased metabolism to acetaldehyde; (2) inhibition of ALDH and thus
11 reduced clearance of acetaldehyde, or (3) a number of other actions are
12 unknown. Crabb et al. (2001) reported 20-30% reductions in ALDH2 protein
13 level by PPARa agonists (Clofibrate treatment in rats and WY treatment in both
14 wild and PPARa-null mice). This could be another pathway for TCE-induced
15 effects on ethanol metabolism. It is an intriguing possibility that the reported
16 association between the increased risk of human esophageal cancer and TCE
17 exposure [Scott and Chiu, 2006] could be related to TCE effects on
18 mitochondrial ALDH, given a similar association of this endpoint with ethanol
19 consumption or ALDH polymorphism.
20
21 Finally, ethanol ingestion may have significant effects on TCE
22 pharmacokinetics. Baraona et al. (2002 a,b) reported that chronic, but not acute,
23 ethanol administration increased the hepatotoxicity of peroxynitrite, a potent
24 oxidant and nitrating agent, by enhancing concomitant production of nitric oxide
25 and superoxide. They also reported that nitric oxide mediated the stimulatory
26 effects of ethanol on blood flow. Ethanol markedly enhanced portal blood flow
27 (2-fold increase), with no changes in the hepatic, splenic, or pancreatic arterial
28 blood flows in rats.
29
30 E.4.3.5. Coexposure Effects on Pharmacokinetics: Predictions Using Physiologically Based
31 Pharmacokinetic (PBPK) Models
32 Along with experimental evidence that has focused on chronic and acute experiments
33 using rodents, the potential pharmacokinetic modulation of risk has also been recently published
34 reports using PBPK models that may be useful in predicting coexposure effects. Caldwell et al.
35 (2008b) also examined and discussed these approaches and note:
36
37 An important issue for prediction of the effects and relationship on MO As by
38 co-exposure is the degree to which modulation of TCE toxicity by other agents
39 can be quantified. Pharmacokinetics or the absorption, distribution, metabolism,
40 and elimination of an agent, can be affected by internal and external co-exposure.
41 Such information can help to identify the chemical species that may be causally
42 associated with observed toxic responses, the MO A, and ultimately identify
43 potentially sensitive subpopulations for an effect such as carcinogenicity.
44
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-406 DRAFT—DO NOT CITE OR QUOTE
-------
1 Physiologically based pharmacokinetic (PBPK) models are often used to
2 estimate and predict the lexicologically relevant dose of foreign compounds in
3 the body and have been developed to predict effects on pharmacokinetics that are
4 additive or less or greater than additive. One of the first such models was
5 developed for TCE [Andersen et al., 1987]. Given that TCE, PERC, and methyl
6 chloroform (MC) are often found together in contaminated groundwater, Dobrev
7 et al. (2001) attempted to investigate the pharmacokinetic interactions among the
8 three solvents to calculate defined "interaction thresholds" for effects on
9 metabolism and expected toxicity. Their null hypothesis was defined as
10 competitive metabolic inhibition being the predominant result for TCE given in
11 combination with other solvents. Gas uptake inhalation studies were used to test
12 different inhibition mechanisms. A PBPK model was developed using the gas
13 uptake data to test multiple mechanisms of inhibitory interactions (i.e.,
14 competitive, noncompetitive, or uncompetitive) with the authors reporting
15 competitive inhibition of TCE metabolism by MC and PERC in simulations of
16 pharmacokinetics in the rat. Occupational exposures to chemical mixtures of the
17 three solvents within their Threshold Limit Value (TLV)/TWA limits were
18 predicted to result in a significant increase (22%) in TCE blood levels compared
19 with single exposures.
20
21 Dobrev et al. (2002) extended this work to humans by developing an interactive
22 human PBPK model to explore the general pharmacokinetic profile of two
23 common biomarkers of exposure, peak TCE blood levels, and total amount of
24 TCE metabolites generated in rats and humans. Increases in the TCE blood
25 levels were predicted to lead to higher availability of the parent compound for
26 GSH conjugation, a metabolic pathway that may be associated with kidney
27 toxicity/carcinogenicity. A fractional change in TCE blood concentration of
28 15% for combined TLV exposures of the three chemicals (25/50/350 ppm of
29 PERC/TCE/MC) resulted in a predicted 27% increase of the S-(l, 2-
30 dichlorovinyl)-L-cysteine (DCVC) metabolites, indicating a nonlinear risk
31 increase due to combined exposures to TCE. Binary combinations of the
32 solvents produced GST-mediated metabolite levels almost twice as high as the
33 expected rates of increase in peak blood levels of the parent compound. The
34 authors suggested that using parent compound peak blood levels (a less sensitive
35 biomarker) would result in two to three times higher (i.e., less conservative)
36 estimates of potentially safe exposure levels. In regard to the detection of
37 metabolic inhibition by PERC and MC, the simulations showed TCE blood
38 concentrations to be the more sensitive dose metric in rats, but the total of TCE
39 metabolites to be the more sensitive dose measure in humans. Finally,
40 interaction thresholds were predicted to be occurring at lower levels in humans
41 than rats.
42
43 Thrall and Poet (2000) investigated the pharmacokinetic impact of low-dose
44 co-exposures to toluene and TCE in male F344 rats in vivo using a real-time
45 breath analysis system coupled with PBPK modeling. The authors report that,
46 using the binary mixture to compare the measured exhaled breath levels from
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-407 DRAFT—DO NOT CITE OR QUOTE
-------
1 high- and low-dose exposures with the predicted levels under various metabolic
2 interaction simulations (competitive, noncompetitive, or uncompetitive
3 inhibition), the optimized competitive metabolic interaction description yielded
4 an interaction parameter Ki value closest to the Michaelis-Menten affinity
5 parameter (KM) of the inhibitor solvent. This result suggested that competitive
6 inhibition is the most plausible type of metabolic interaction between these two
7 solvents.
8
9 Isaacs et al. (2004) have reported gas uptake co-exposure data for CHCb and
10 TCE. The question as to whether it is possible to use inhalation data in
11 combination with PBPK modeling to distinguish between different metabolic
12 interactions was addressed using sensitivity analysis theory. Recommendations
13 were made for design of optimal experiments aimed at determining the type of
14 inhibition mechanisms resulting from a binary co-exposure protocol. This paper
15 also examined the dual nature of inhibition of each chemical in the pair to each
16 other, which is that TCE and CHC13 were predicted to interact in a competitive
17 manner. Even though as stated by Dobrev et al. (2001), other solvents inhibit
18 TCE metabolism, it is also possible to quantify the synergistic interaction that
19 TCE has on other solvents, using techniques such as gas uptake inhalation
20 exposures.
21
22 Haddad et al. (2000) has developed a theoretical approach to predict the
23 maximum impact that a mixture consisting of co-exposure to dichloromethane,
24 benzene, TCE, toluene, PERC, ethylbenzene, m-, p-, and o-xylene, and styrene
25 would have on venous blood concentration due to metabolic interactions in
26 Sprague-Dawley rats. Two sets of experimental co-exposures were conducted.
27 The first study evaluated the change in venous blood concentration after a 4 hour
28 constant inhalation exposure to the 10 chemical mixtures. This experiment was
29 designed to examine metabolic inhibition for this complex mixture. The second
30 study was designed to study the impact of possible enzyme induction by using
31 the same inhalation co-exposure after a 3 day pretreatment with the same 10
32 chemical mixture. The resulting venous concentration measurements for TCE
33 from the first study were consistent with metabolic inhibition theory. The 10-
34 chemical mixture was the most complex co-exposure used in this study. The
35 authors stated that as mixture complexity increased, the resulting parent
36 compound concentration time courses changed less, an observation which is
37 consistent with metabolic inhibition. For the pretreatment study, the authors
38 found a systematic decrease in venous concentration (due to higher metabolic
39 clearance) for all chemicals except PERC. Overall, these studies suggest a
40 complex metabolic interaction between TCE and other solvents.
41
42 A PBPK model for TCE including all its metabolites and their interactions can
43 be considered a mixtures model where all metabolites have a common starting
44 point in the liver. An integrated approach taking into account TCE metabolites
45 and their metabolic inhibition and interactions among each other is suggested in
46 Chiu et al. (2006b).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-408 DRAFT—DO NOT CITE OR QUOTE
-------
1 E.5. POTENTIALLY SUSCEPTIBLE LIFE STAGES AND CONDITIONS THAT
2 MAY ALTER RISK OF LIVER TOXICITY AND CANCER
3 As described in Sections E.I.2, E.3.2.2, E.3.2.6, E.4.2.1, E.4.2.2, E.4.2.3, and E.4.2.4,
4 there are a number of conditions that are associated with increased risk of liver cancer and
5 toxicity that include age, use of a number of prescription medications including fibrates and
6 statins, disease state (e.g., diabetes, NALD, viral infections) and exposure to external
7 environmental contaminants that have an affect on TCE toxicity and targets. Obviously
8 epigenetic and genetic factors play a role in determining the risk to the individual. In terms of
9 liver cancer, there is general consensus that despite the associations that have been made with
10 etiological factors and the risk of liver cancer, the mechanism is still unknown. The MOA of
11 TCE toxicity is also unknown but exposure to TCE and its metabolites have shown in rodent
12 models to induce liver cancer and in a fashion that is not consistent with only a hypothesized
13 MOA of PPARa receptor activation that is in need of revision. However, multiple TCE
14 metabolites have been shown to also induce liver cancer with varying effects on the liver that
15 have also been associated with early stages of neoplasia (glycogen storage) or other actions
16 associated with risk of hepatocarcinogenicity. The growing epidemic of obesity has been
17 suggested to increase the risk of liver cancer and may reasonably increase the target population
18 for TCE effects on the liver.
19 Lifestyle factors such as ethanol ingestion have not only been shown to increase liver
20 cancer risk in those who already have fatty liver, but also to increase the toxicity of TCE.
21 However, as noted by Caldwell et al. (2008b), while there is evidence to suggest that TCE
22 exposure may increase the risk of liver toxicity and cancer, there are not data to support a
23 quantitative estimate of how coexposures may modulate that risk.
24
25 These findings can also serve to alert the risk manager to the possibility that
26 multiple internal and external exposures to TCE that may act via differing MO As
27 for the production of liver effects. This information suggests a possible lack of
28 "zero" background exposures and can help identify potential susceptible
29 populations.
30
31 Background levels of haloacetates in drinking water may add to the cumulative
32 exposure an individual receives via the metabolism of TCE. The brominated
33 haloacetates apparently share some common effects and pathways with their
34 chlorinated counterparts. Thus, concurrent exposure of TCE, its metabolites, and
35 other haloacetates may pose an additive response as well as an additive dose.
36 However, personal exposures are difficult to ascertain and the effects of such co-
37 exposures on toxicity are hard to quantify. EPA's guidance on cumulative risk
38 assessments directs "each office to take into account cumulative risk issues in
39 scoping and planning major risk assessments and to consider a broader scope that
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-409 DRAFT—DO NOT CITE OR QUOTE
-------
1 integrates multiple sources, effects, pathways, stressors, and populations for
2 cumulative risk analyses in all cases for which relevant data are available" [U.S.
3 EPA, 1997]. Widespread exposure to possible background levels of TCE
4 metabolites or co-contaminants and other extrinsic factors have the potential to
5 affect TCE toxicity. However, the available data for co-exposures on TCE
6 toxicity appears inadequate for quantifying these effects, particularly at
7 environmental levels of contamination and exposure. Thus, the risk manager and
8 assessor are going to be limited by not having information regarding either (1)
9 the type of exposure data necessary to assess the magnitude of co-exposures that
10 may affect toxicity, or (2) the potential quantitative impacts of these co-
11 exposures that would enable specific adjustments to risk. Nonetheless, the risk
12 manager should be aware that qualitatively a case can be made that extrinsic
13 factors may affect TCE toxicity.
14
15 E.6. UNCERTAINTY AND VARIABILITY
16 Along with general conclusions about the coherence of data that enable conclusions
17 about effects on the liver shown through experimental studies of TCE, there have also been
18 extensive discussions throughout this report regarding the specific limitations of experimental
19 studies whose design was limited by small and varying groups of animals and variability in
20 control responses as well as reporting deficiencies. Section E.3.2.5 has brought forward the
21 uncertainty in the MO A for liver cancer in general. The consistency of different animal models
22 with human HCC is described in Section E.3.3, with Section E.3.2.2 providing a discussion of
23 the promise and limitations of emerging technologies to study the MO As of liver can in general
24 and for TCE specifically. Issues regarding the target cell for HCC and the complexities of
25 studying the MOA for a heterogeneous disease are described in Sections E.3.2.4 and E.3.2.8,
26 respectively. Finally, the uncertainty regarding key events in how activation of the PPARa
27 receptor my lead to hepatocarcinogenesis and the problems with extrapolation of results using
28 the common paradigm to study them (exposure to high levels of WY-14,643 in abbreviated
29 bioassays in knockout mice) are outlined in Section E.3.5.1. As such uncertainties are identified
30 future research can focus on resolving them.
31
32 E.7. REFERENCES
33 Adami HO, Chow WH, Nyrene O, Berne C, Linet MS, Ekbom A, et al. 1996. Excess risk of primary liver cancer in
34 patients with diabetes mellitus. J Natl Cancer Inst 88:1472-1477.
35 Adams LA, LindorKD. 2007. Nonalcoholic fatty liver disease. AnnEpidemiol 17: 863-869.
36 Adli M, Baldwin AS. 2006. IKK-i/IKKe control constitutive, cancer cell-associated NF-KB activity via regulation
37 of Ser-536 p65/RelA phosphorylation. J of Biological Chemistry 281(37): 26076-26984.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-410 DRAFT—DO NOT CITE OR QUOTE
-------
1 Aikata H, Takaishi H, Kawakami Y, Takahashi S, Kitamoto M, Nakanish T, Nakamura Y, Shjimanmoto F,
2 Kajiyama G, Ide T. 2000. Telomere reduction in human liver tissues with age and chronic inflammation. Exp Cell
3 Res 256:578-582.
4 Akira S, Takeda K, 2004. Toll-like receptor signaling. Nat Rev Immunol. 4(7): 449-511
5 Akiyama TE, Nicol CJ, Fievet C, Staels B, Ward JM, Auwerx J, Lee SST, Gonzalez FJ, Peters JM. 2001.
6 Peroxisome proliferator-activated receptor-a regulates lipid homeostasis, but is not associated with obesity. J Biol
7 Chem. 276(42):39088-39093.
8 Allen DG, Pearse G, Haseman JK, Maronpot RR. 2004. Prediction of rodent carcinogenesis: an evaluation of
9 prechronic liver lesions as forecasters of liver tumors in NTP carcinogenicity studies. Toxicol Pathol 32: 393-401.
10 Ammini CV, Fernandez-Canon J, Shroads AL, Cornett R, Cheung J, James MO, et al. 2003. Pharmacologic or
11 genetic ablation of maleylacetoacetate isomerase increases levels of toxic tyrosine catabolites in rodents. Biochem
12 Pharmacol 66:2029-2038.
13 Andersen ME, Gargas ML, Clewell HJ, SeverynKM. 1987. Quantitative evaluation of the metabolic interactions
14 between trichloroethylene and 1,1-dichloroethylene in vivo using gas uptake methods
15 Anderson SP, Howroyd P, Liu J, Qian X, Bahnemann R, Swanson C, Kwak MK, Kensler TW, Corton JC. 2004.
16 The transcriptional response to a peroxisome proliferator-activated receptor alpha agonist includes increased
17 expression of proteome maintenance genes. J Biol Chem. 279:52390-8.
18 Anna CH, Maronpot RR, Pereira MA, Foley JF, Malarkey DE, Anderson MW. 1994. ras proto-oncogene activation
19 in dichloroacetic-, trichloroethylene- and tetrachloroethylene-induced liver tumors in B6C3F1 mice. Carcinogenesis
20 15:2255-2261.
21 Anti M, Marra G, Ludovico Rapaccini G, Rumi C, Bussa S, Fadda G, Vecchio FM, Valenti A, Percesepe A, Pompili
22 M, Armelad F, Gentiloni N. 1994. DNA ploidy pattern in human chronic liver diseases and hepatic nodular lesions.
23 Cancer 73(2): 281-288.
24 Anway MD, Cupp AS, Uzumcu M, Skinner MK. 2005. Epigenetic transgenerational actions of endocrine
25 disrupters and male fertility. Science. 308: 1466-1469.
26 Attallah AB, Tabll AA, Salem SF, El-Sadany M, Ibrahim TA, Osman S, El-Dosoky IM. 1999. DNA ploidy of liver
27 biopsies from patients with liver cirrhosis and hepatocellular carcinoma: a flow cytometric analysis. Cancer letters
28 142:65-69.
29 Austin EW, Parrish JM, Kinder DH, Bull RJ. 1996. Lipid peroxidation and formation of 8-hydroxydeoxygaunosine
30 from acute doses of halogenated acetic acids. Fundamental and Applied Toxicology 31: 77-82.
31 Auwerx J, Schoonjans K, Fruchart JC, Staels B. 1996. Transcriptional control of triglyceride metabolism: fibrates
32 and fatty acids change the expression of the LPL and apo C-III genes by activating the nuclear receptor PPAR.
33 Atherosclerosis 124 (suppl). S29-S37.
34 Badger TM, Ronis MJ, Seitz HK, Albano E, Ingelman-Sundberg M, Lieber CS. 2003. Alcohol metabolism: Role in
35 toxicity and carcinogenesis. Alcohol Clin Exp Res 27:336-347
36 Bai CL, Canfield PJ, Stacey NH. 1992a. Effects of hexachloro-1,3- butadiene and 1,1,2,2-tetrachloroethylene on
3 7 individual serum bile acids. Toxicol Ind Health 8:191-203.
38 Bai CL, Canfield PJ, Stacey NH. 1992b. Individual serum bile acids as early indicators of carbon tetrachloride- and
39 chloroform-induced liver injury. Toxicology 75:221-234.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-411 DRAFT—DO NOT CITE OR QUOTE
-------
1 Bai CL, Stacey NH. 1993. Mechanism of trichloroethylene-induced elevation of individual bile acids. II. In vitro
2 and in vivo interference by trichloroethylene with bile acid transport in isolated rat hepatocytes. Toxicol and Appl
3 Pharmacol 121:296-302.
4 Balkwill F, Coussens LM. 2004. An inflammatory Link. The NF-KB protein is a key player in inflammation, It
5 now seems that it might also activate signaling pathways, in both cancer cells and tumour-associated inflammatory
6 cells, that promote malignancy. Nature 431:405-406.
7 Balkwill F, Charles KA, Mantovani A. 2005. Smoldering and polarized inflammation in the initiation and
8 promotion of malignant disease. CancerCell 7:211-217.
9 Baker VA, Harries HM, Waring JF, Duggan CM, Ni HA, Jolly RA, Yoon LW, De Souza AT, Schmid JE, Brown
10 RH, Ulrich RG, Rockett JC. 2004. Clofibrate-induced gene expression changes in rat liver: a cross-laboratory
11 analysis using membrane cDNA arrays. Environmental Health Perspectives 112(4): 428-438.
12 Ballestar E, Esteller M. 2002. The impact of chromatin in human cancer: linking DNA methylation to gene
13 silencing. Carcinogenesis 23:1103-1109
14 BannaschP. 1996. Pathogenesis of hepatocellular carcinoma: sequential cellular, molecular, and metabolic changes.
15 Prog Liver Dis. 16:161-97.
16 Bannasch P. 2001. Hormonal and hormone-like effects eliciting hepatocarcinogenesis. Folia Histochem Cytobio
17 139(Suppl2):28-29.
18 Bannasch P, Nehrbass D, Kopp-Schneider A. 2001 Significance of hepatic preneoplasia for cancer
19 chemoprevention. IARC SciPubl 154:223-240.
20 Bannasch P, Haertel T, Su Q. 2003. Significance of hepatic preneoplasia in risk identification and early detection of
21 neoplasia. Toxicol Pathol. 31(1): 134-139.
22 Baraona E, Shoichet L, Navder K, Lieber CS. 2002a. Mediation by nitric oxide of the stimulatory effects of ethanol
23 on blood flow. Life Sci 70:2987-2995.
24 Baraona E, Zeballos GA, Shoichet L, Mak KM, Lieber CS. 2002b. Ethanol consumption increases nitric oxide
25 production in rats, and its peroxynitrite-mediated toxicity is attenuated by polyenylphosphatidylcholine. Alcohol
26 Clin Exp Res 26:883-889.
27 Barass NC, Price RJ, Lake BG, Orton TC. 1993. Comparison of the acute and chronic mitogenic effects of the
28 peroxisome proliferators methylclofenapate and clofibric acid in rat liver. Carcinogenesis 14(7): 1451-1456.
29 Bartosiewicz M, Perm S, Buckpitt A. 2001. Applications of gene arrays in environmental toxicology: fingerprints of
3 0 gene regulation associated with cadmium chloride, benzo(a)pyrene,and trichloroethylene. Environ Health Perspect
31 109:71-74.
32 Beekman JM, Boess F, Hildebrand H, Kalkuhl A Suter L. 2006. Gene expression analysis of the hepatoxicant
33 methylpyrilene in primary rat hepatocytes. An interlaboratory study. Environ Health Perspect 114: 92-99.
34 Berger J, Daxenbichler G. 2002. DNA methylation of nuclear receptor genes—possible role in malignancy. J Steroid
35 BiochemMol Bio 180:1-11.
36 Bergeron A, Jorquera R, Tanguay RM. 2003. La tyrosinemie hereditare: unemaladie du stress du reticulum
37 endoplasmique?][inFrench] Med Sci (Paris) 19(10):976-980.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-412 DRAFT—DO NOT CITE OR QUOTE
-------
1 Herman E, Schlicht M, Moser VC, MadPhail RC. 1995. A multidisciplinary approach to toxicological screening: I,
2 Sytem toxicity. JToxicol Environ Health 45(2): 127-43.
3 BestDH, Coleman WB. 2007. Treatment with 2-AAF blocks the small hepatocyte-like progenitor cell response in
4 retrorsine-exposed rats. JHepatol 46: 1055-1063.
5 Beyer RP, Fry RC, Lasarev MR, McConnachi LA, Meira LB, Palmer VS, Powel CL, Ross PK, Bammler TK,
6 Bradford BU, Cranson AB, Cunningham ML, Fannin RD, Higgins GM, Hurban P, Kayton RJ, Kerr KF, Kosyk O,
7 Lobenhofer ED, Sieber SO, Vliet PA. Weis BK, Wolfinger R, Woods CG, Freedman JH, Linney E, Kaufman WF,
8 Kavanagh TJ, Paules RA, Rusyn I, Samson LD, Spencer PS, Suk W, Tennant RJ, Zarbl H, and members of the
9 Toxicogenomics Research Consortium. 2007. Multicenter study of acetaminophen hepatoxocity reveals the
10 importance of biological endpoint in genomic analysis. Toxicological Sciences 99(1): 326-337.
11 Bjornsson HT, Sigurdsson MI, Fallin MD, Irizarry RA, Aspelund T, Cui H, Yu W, Rongione MA, Ekstrom TJ,
12 Harris TB, Launer LJ, Eiriksdottir G, Leppert MF, Sapienza C, Gudnason V, Feinberg AP. 2008. Intraindividual
13 change over time in DNA methylation with familial clustering. JAMA 299(24): 2877-2883
14 Blackburn AC, Tzeng HF, Anders MW, Board PG. 2000. Discovery of a functional polymorphism in human
15 glutathione transferase zetaby expressed sequence tag database analysis. Pharmacogenetics 10:49-57.
16 Blackburn AC, Coggan M, Tzeng HF, Lantum H, Polekhina G, Parker MW, et al. 2001. GSTZ Id: a new allele of
17 glutathione transferase zeta and maleylacetoacetate isomerase. Pharmacogeneticsll:671-678.
18 Board PG, Chelvanayagam G, Jermiin LS, Tetlow N, Tzeng HF, Anders MW, et al. 2001. Identification of novel
19 glutathione transferases and polymorphic variants by expressed sequence tag database analysis. Drug Metab Dispos
20 29(4Pt2):544-547.
21 Boerrigter ME. 2004. Mutagenecity of the peroxisome proliferates clofibrate, Wyeth 14,643 and di-2-ethylhexl
22 phthalate in the lacZ plasmid-based transgenic mouse mutation assay. Journal of Carcinogenesis 3(1): 1-7
23 Bombail V, Moggs JG, Orphanides G. 2004. Perturbation of epigenetic status by toxicants. Toxicology letters.
24 149:51-58.
25 Boorman GA, Dellarco V, Dunnick JK, Chapin RE, Hunter S, Hauchman F, Gardner H, Cox M, Sills RC. 1999.
26 Drinking water disinfection byproducts: Review and approach to toxicity evaluation. Environ Health Perspect
27 107:207-217.
28 Borzio M, Fargion S, Borzio F, Fancanzani AL, Croce AM, Stroffolini T, Oldani S, Cotichini R, Roncalli M. 2003.
29 Impact of large regenerative, low grade and high grade dysplastic nodules in hepatocellular carcinoma development.
30 Journal of Hepatology 39: 208-214.
31 Boyault S, Rickman DS, de Reynies A, Balabaud C, Rebouissou S, Jeannot E, Herault A, Saric J, Belghiti J, Franco
32 D, Bioulac-Sage P, Laurent-Puig P, Zucman-Rossi J. 2007. Trancriptome classification of HCC is related to gene
33 alterations and to new therapeutic targets. Hepatology 45(1): 42-52.
34 Bradford BU, Kono H, Isayama F, Kosyk O, Wheeler MD, Akiyama TE, Bleye L, Krausz KW, Gonzalez FJ, Koop
35 DR, Rusyn I. 2005. Cytochrome P450 CYP2E1, but not nicotinamide adenine dinucleotide phosphate oxidase, is
36 required for ethanol-induced oxidative DNA damage in rodent liver. Hepatology 41:336-344.
3 7 Bralet MP, Pichard V, Ferry N. 2002. Demonstration of direct lineage between hepatocytes and hepatocellular
38 carcinoma in diethylnitrosamine-treated rats. Hepatology 36: 623-630.
39 Branda M, Wands JR. 2006. Signal transduction cascades and hepatitis B and C related hepatocellular carcinoma.
40 Hepatology 43(5): 892-902.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-413 DRAFT—DO NOT CITE OR QUOTE
-------
1 Brautbar N, Williams JII. 2002. Industrial solvents and liver toxicity: Risk assessment, risk factors and mechanisms.
2 Int J Hyg Environ Health 205:479-491.
3 Brodsky WY, Uryvaeva IV. 1977. Cell polyploidy: its relation to tissue growth and function. Int Rev Cytol 50:
4 275-332.
5 Buben JA, O'Flaherty EJ. 1985. Delineation of the role of metabolism in the hepatotoxicity of trichloroethylene and
6 perchloroethylene: a dose-effect study. Toxicology and Applied Pharmacology. 78: 105-122.
7 Bull RJ. 2000. Mode of action of liver tumor induction by trichloroethylene and its metabolites, trichloroacetate and
8 dichloroacetate. Environ Health Perspect 108 (S2): 241-259.
9 Bull, RJ. 2004. Trichloroethylene and liver tumors in mice. Presentation at the U.S. EPA Symposium on New
10 Scientific Research Related to the Health Effects of Trichlorothylene. 27
11 Bull RJ, Sanchez IM, Nelson MA, Larson JL, Lansing AJ. 1990. Liver tumor induction in B6C3F1 mice by
12 dichloroacetate and trichloroacetate. Toxicology 63:341-359.
13 Bull RJ, Orner GA, Cheng RS, Stillwell L, Stauber AJ, Sasser LB, Lingohr MK, Thrall BD. 2002. Contribution of
14 dichloroacetate and trichloroacetate to liver tumor induction in mice by trichloroethylene. Toxicol Appl Pharmacol
15 182:55-65.
16 Bull RJ, Sasser LB, Lei XC. 2004. Interactions in the tumor-promoting activity of carbon tetrachloride,
17 trichloroacetate, and dichloroacetate in the liver of male B6C3F1 mice. Toxicology 199:169-183.
18 Caldwell JC, Keshava N. 2006. Key issues in the modes of action and effects of trichloroethylene metabolites for
19 liver and kidney tumorigenesis. Environ Health Perspect 114:1457-1463.
20 Caldwell JC, Jinot J, DeVoney D, Gift J. 2008(a). Evaluation of evidence for infection as the mode of action for
21 induction of rat lymphoma. Environ MolMutagen 49(2): 155-64.
22 Caldwell JC, Keshava N, Evans MV. 2008(b). Difficulty of mode of action determination for trichloroethylene: an
23 example of complex interactions of metabolites and other chemical exposures. Environ Mol Mutagen 49(2):
24 142-54
25 Calvisi DF, Factor VM, Ladu S, Conner EA, Thorgeirsson SS. 2004a. Disruption of (3-catenin pathway or genomic
26 instability defines two distinct categories of liver cancer in transgenic mice. Gastroenterology 126: 1374-1386.
27 Calvisi DF, Ladu S, Conner EA, Factor VM, Thorgeirson SS. 2004b. Disregulation of E-cadherin in transgenic
28 models of liver cancer. Laboratory Investigation 84: 1137-1147.
29 Calvisi DF, Ladu S, Gorden A, Farina M, Conner EA, Lee JS, Factor VM, Thorgeirsson SS. 2006. Ubiquitous
30 activation of Ras and Jak/S tat pathway sin human HCC. Gastroenterology 130: 1117-1128.
31 Calvisi DF, Ladu S, Gorden A, Farina M, Lee JS, Conner EA, Schroeder I, Factor VM, Thorgeirsson SS. 2007.
32 Mechanistic and prognostic significance of aberrant methylation in the molecular pathogenesis of human
33 hepatocellular carcinoma. J Clin Invest 117(9): 2713-2711.
34 Carter JH, Carter HW, DeAngelo AB. 1995. Biochemical pathologic and morphometric alterations induced in male
35 B6C3F1 mouse liver by short-term exposure to dichloroacetic acid. Toxicology Letters 81: 55-71.
36 Carter JH, Carter HW, Deddens JA, Hurst BM, George MH, DeAngelo AB. 2003. A 2-year dose-response study of
37 lesion sequences during hepatocellular carcinogenesis in the male B6C3F1 mouse given the drinking water chemical
3 8 dichloroacetic acid. Environ Health Perspect lll(l):53-64.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-414 DRAFT—DO NOT CITE OR QUOTE
-------
1 Charbonneau Mm Couture J, Plaa GL. 1991. Inhalation versus oral administration of acetone: effect of the vehicle
2 on the potentiation of CCL4- induced liver injury. Toxicology Letters 57: 47-54.
3 Chang LW, Daniel FB, DeAngelo AB. 1992. Analysis of DNA strand breaks induced in rodent liver in vivo,
4 hepatocytes in primary culture, and a human cell line by chlorinated acetic acids and chlorinated acetaldehydes.
5 Environ Mol Mutagen 20L 277-288.
6 Channel SR, Lantendresse JR, Kidney JK, Grabau JH, Lane JW, Steel-Goodwin L, Gothaus MC. 1998. A
7 subchronic exposure to trichloroethylene causes lipid peroxidation and hepatocellular proliferation in male B6C3F1
8 mouse liver. Tox Sci 43: 145-154.
9 Charles JM and Leeming MM. 1998. Chronic dietary toxicity/oncogenicity studies on 2,4-dichlorophenoxybutyric
10 acid in rodents. Toxicological Sciences 46:21-30.
11 Charles JM, Bond DM, Jeffries TK, Yano BL, Stott WT, Johnson KA, Cunny HC, Wilson RD, Bust JS. 1996.
12 Chronic dietary toxicity/oncogenicity studies on 2,4-dichlorophenoxyacetic acid in rodents. Fundamental and
13 applied toxicology 11:166-172.
14 Chen ZY, White CC, He CY, Liu YF, Eaton DL. 1995. Zonal differences in DNA synthesis activity and
15 cytochrome p450 gene expression in livers of Male F344 rats treated with five nongenotoxic carcinogens. J Environ
16 Pathol, Toxicol and Oncol 14(2) 83-99.
17 Chen C, Cheung ST, So S, Fan ST, Barry S, Higgins J, Lai KM, Ji J, Dudoit S, Ng IOI, van de Rijn M, Botstein D,
18 Brown PO. 2002. Gene expression patters in human liver cancers. Molecular Biology of the Cell. 13:1929-1939.
19 Cheung C, Akiyama TE, Ward JM, Nicol CJ, Feigenbaum L, Vinson C, Gonzalez FJ. 2004. Diminished
20 hepatocellular proliferation in mice humanized for the nuclear receptor peroxisome proliferator-activated receptor a.
21 Cancer Res 64: 3849-3854.
22 Chiu WA, Caldwell JC, Keshava N, Scott CS. 2006a. Key scientific issues in the health risk assessment of
23 trichloroethylene. Environ Health Perspect 114:1445-1449.
24 Chiu WA, Okino M, Lipscomb JC, Evans MV. 2006b. Issues in the pharmacokinetics of trichloroethylene and its
25 metabolites. Environ Health Perspect 114:1450-1456.
26 Cho WCS. 2007. OncomiRs: the discovery and progress of microRNAs in cancers. Molecular Cancer 6:50
27 doi: 10.1186/1476-4598-6-60.
28 Cogger VC, Warren A, Fraser R, NgU M, McLean AJ, Le Couteur DG. 2003. Hepatic sinusoidal
29 pseudocapillarization with aging in the nonhuman primate. Exp Gerontol. 38: 1101-1107.
30 Collier JM, Selmin O, Johnson PD, Runyan RB. 2003. Trichloroethylene effects on gene expression during cardiac
31 development. Birth Defects Res A Clin Mol Teratol 67:488-495.
32 Collin SM. 2005. Mechanisms leading to chromosomal instability. Semin. Cancer Biol. 15: 33-42.
33 Columbano A and Ledda-Columbano GM. 2003. Mitogenesis by ligands of nuclear receptors: an attractive model
34 for the study of the molecular mechanisms implicated in liver growth. Cell Death and Differentiation. 10 :S19-S21
35 Columbano A, Ledda-Columbano GM, Lee G, Rajalakshmi S, Sarma DSR. 1987. Instability of mitogen-induced
36 liver hyperplasia to support the induction of enzyme-altered islands induced by liver carcinogens. Cancer Res. 47:
37 5557-5559.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-415 DRAFT—DO NOT CITE OR QUOTE
-------
1 Columbano A, Ledda-Columbano GM, Curto M, Ennas GM, Coni P, Sarma DSR, Pani P. 1990. Cell proliferation
2 and promotion of rat liver carcinogenesis: different effect of hepatic regeneration and mitogen-induced hyperplasia
3 on the development of enzyme altered foci. Carcinogenesis 11: 771-776.
4 Columbano A, Ledda-Columbano GM, Coni, Faa G, Liguori C, Santacruz G, Pani P. 1985. Occurrence of cell
5 death (apoptosis) during the involution of liver hyperplasia. Lab Inv. 52. 670-675.
6 Coni p, Simbula G, Carcereri De Prati A, Menegazzi M, Suzyki H, Sarma D, Ledda-Columbano GM, Columbano A.
7 1993. Differences in the steady-state levels of c-fos, c-jun, and c-myc messenger RNA during mitogen-induced liver
8 growth and compensatory regeneration. Hepatology 17:1109-1117.
9 Conner EA, Lemmer ER, Omori M, Wirth PJ, Factor VM, Thorgeirsson SS. 2000. Dual functions of E2F-1 in a
10 transgenic mouse model of liver carcinogenesis. Oncogene 19:5054-5062.
11 Conner EA, Lemmer ER, Sanchez A, Factor VM, Thorgeirsson SS. 2003. E2F1 blocks and c-Myc accelerates
12 hepatic ploidy in transgenic mouse models. Bioch Biophys Res Comm302: 114-120.
13 Contrera JF, Jacobs AC, DeGeorge JJ. 1997. Carcinogenicity testing and the evaluation of regulatory requirements
14 for Pharmaceuticals. Reg Toxicol and Pharmacol. 25:130-145.
15 Cornett R, James MO, Henderson GN, Cheung J, Shroads AL, Stacpoole PW. 1999. Inhibition of glutathione S-
16 transferase zeta and tyrosine metabolism by dichloroacetate: a potential unifying mechanism for its altered
17 biotransformation and toxicity. Biochem Biophys Res Commun 262(3):752-756.
18 Costet P, Legendre C, More J, Edgar A, Galtier P, Pineau T. 1998. Peroxisome proliferator-activated receptor
19 alphaisoform deficiency leads to progressive dyslipidemia with sexually dimorphic obesity and steatosis. J Biol
20 Chem 273:29577-29585.
21 Counts JL, Sarmiento JI, Harbions ML, Downing JC, McClain RM, Goodman JI. 1996. Carcinogenesis. 17(6):
22 1251-1257.
23 Crabb DW, Pinaire J, Chou WY, Sissom S, Peters JM, Harris RA, Stewart M. 2001. Peroxisome proliferator-
24 activated receptors (PPAR) and the mitochondria! aldehyde dehydrogenase (ALDH2) promoter in vitro and in vivo.
25 Alcohol ClinExp Res 25:945-952.
26 Dai X, De Souza AT, Dai H, Lewis DL, Lee CK, Spencer AG, Herweijer H, Hagstrom JE, Linsley PS, Bassett DE,
27 Ulrich RG, He YD. 2007. PPARa siRNA-treated expression profiles uncover the causal sufficiency network for
28 compound-induced liver hypertrophy. PLoS Computational Biology 3(3): 389-401.
29 Dajee M, Lazarov M, Zhang JY, Cai T, Green CL, Russell AJ, Marinkovich MP, Tao S, Lin Q, Kubo Y, Khavari
30 PA. 2003. NF-kappaB blockade and oncogenic Ras trigger invasive human epidermal neoplasia. Nature. 421:
31 639-643.
32 Daniel FB, DeAngelo AB, Stober JA, Olson GR, Page NP. 1992. Hepatocarcinogenicity of chloral hydrate, 2-
33 chloroacetaldehyde, and dichloroacetic acid in the male B6C3F1 mouse. Fundamental and Applied Toxicology 19:
34 159-168.
3 5 David RM, Moore MR, Cifone MA, Finney DC, Guest D. 1999. Chronic peroxisome proliferation and
3 6 hepatomegaly associated with the hepatocellular tumorigenesis of di(2-ethylhexyl)phthalate and the effects of
37 recovery. Toxicol Sci 50(2): 195-205.
3 8 DeAngelo AB, Daniel FB, McMillan I, P Wernsing, Savage RE. 1989. Species and strain sensitivity to the
3 9 induction of peroxisome proliferation by chloroacetic acids. Toxicology and Applied Pharmacology 101: 285-298.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-416 DRAFT—DO NOT CITE OR QUOTE
-------
1 DeAngelo, AB; Daniel, FB; Most, BM; et al. (1996) The carcinogenicity of dichloroacetic acid in the male Fischer
2 344 rat. Toxicology 114:207-221.
3 DeAngelo AB, Daniel FB, Most BM, Olson GR. 1997. Failure of monochloroacetic acid and trichloroacetic acid
4 administered in the drinking water to produce liver cancer in male F344/N rats. Journal of Toxicology and
5 Environmental Health 52: 425-445.
6 DeAngelo AB, George MH, House DE. 1999. Hepatocarcinogenicity in the male B6C3F1 mouse following a
7 lifetime exposure to dichloroacetic acid in the drinking water: dose-response determination and modes of action. J
8 Toxicol Environ Health 58:485-507.
9 DeAngelo AB, Daniel FB, Wong DM, George MH. 2008. The induction of hepatocellular neoplasia by
10 trichloroacetic acid administered in the drinking water of the male B6C3F1 mouse. J Toxicol and Environmental
11 Health Part A 71: 1056-1068.
12 Dees C, Travis C. 1993. The mitogenic potential of trichloroethylene inB6C3Fl mice. Toxicology Letters
13 69:129-137.
14 Djouadi F, Weinheimer CJ, Saffitz JE, Pitchford C, Bastin J, Gonzalex F, Kelly DP. 1998. A gender-related defect
15 in lipid metabolism and glucose homeostasis in peroxisome proliferator-activator Receptor a-deficient mice. J of
16 Clin Invest. 102 (6):1083-1091.
17 Dobrev ID, Andersen ME, Yang RS. 2001. Assessing interaction thresholds for trichloroethylene in combination
18 with tetrachloroethylene and 1,1,1 -trichloroethane using gas uptake studies and PBPK modeling. Arch Toxicol
19 75:134-144.
20 Dobrev ID, Andersen ME, Yang RS. 2002. In silico toxicology: Simulating interaction thresholds for human
21 exposure to mixtures of trichloroethylene, tetrachloroethylene, and 1,1,1-trichloroethane. Environ Health Perspect
22 110:1031-1039.
23 Dolinoy DC, Weidman JR, Jirtle RL. 2007. Epigenetic gene regulation: linking early developmental environment
24 to adult disease. Reproductive Toxicology 23: 297-307.
25 Dorfmueller MA, Henne SP, York RG, Bornschein RL, Manson JM. 1979. Evaluation of teratology and behavioral
26 toxicity with inhalation exposure of maternal rats to trichloroethylene. Toxicology 14:153-166.
27 Eacho PI, Larder TL, Brodhecker CA. 1991. Hepatocellular DNA synthesis in rats given peroxisome proliferating
28 agents: comparison of Wy-14,643 to clofibric acid, nafenopinandLY171883. Carcinogenesis 12(9): 1557-1561.
29 Eastin WC, Mennear JH, Tennant RW, Stall RE, Branstetter DG, Bucher JR, McCullough B, Binder RL, Spalding
30 JW, Mahler JF. 2001. Tg.AC genetically altered mouse: assay working group overview of available data. Toxicol
31 Pathol 29(suppl):60-80.
32 Elcombe CR, Rose MS, Pratt IS. 1985. Biochemical, histological and ultrastructural changes in rat and mouse liver
33 following the administration of trichloroethylene: possible relevance to species differences in hepatocarcinogenicity.
34 Toxicology and Applied Pharmacology, 79: 365-376.
35 El-Hage J (2007. Center for Drug Evaluation and Research. U.S. Food and Drug Administration. Peroxisome
36 proliferator-activated receptor (PPAR) antagonists: preclinical and clinical cardiac safety considerations.
37 http://www.fda.gov/cder/present/DIA2006/El-Hage_CardiacSafety.ppt Accessibility verified February 27, 2007.
38 El-SeragHB. 2004. Hepatocellular carcinoma: recent trends in the United States. Gastroenterology 127 (5 Supp 1):
39 S27-S34.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-417 DRAFT—DO NOT CITE OR QUOTE
-------
1 El-Serag HB, Rudolph KL. 2007. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis.
2 Gastroenterology 132: 2557-2576.
3 European Food Safety Authority (EFSA). 2006. Opinion of the scientific panel AFC related to a new long-term
4 carcinogenicity study on aspartame. Available online May 11, 2006:
5 http://www.efsa. eu.int/science/afc/afc_opinions/147 l_en.html.
6 Evert M, Schneider-Stock R, Dombrowski F. 2003. Apolipoprotein A-IV mRNA overexpression in early
7 preneoplastic hepatic foci induced by low-number pancreatic islet transplants in streptozotocin-diabetic rats. Pathol
8 Res Pract 199(6):373-379.
9 Fang HL, Strom SC, Cai H, Falany CN, Kocarek TA, Runge-Morris M. 2005. Regulation of human hepatic
10 hydroxysteroid sulgotranferase gene expression by the peroxisome proliferator-activated receptor alpha transcription
11 factor. MolPharmacol 67: 1257-1267.
12 Farazi PA, DePinho RA. 2006. Hepatocellular carcinoma pathogenesis: from genes to environment. Nature
13 Reviews Cancer 6: 674-687.
14 Fattovich G, Stroffolini T, Zagni I, Donate F. 2004. Hepatocellular carcinoma in cirrhosis: incidence and risk
15 factors. Gastroenterology, 127: S35-S50.
16 Feinberg AP, Ohlsson R, Henikoff S. 2006. The epigenetic progenitor origin of human cancer. Nature Reviews
17 Genetics 7: 21-33.
18 Feitelson MA, Sun B, Satiroglu- Tufan ML, Liu J, Pan J, Lian Z. 2002. Genetic mechanisms of
19 hepatocarcinogenesis. Oncogene 21: 2593-2604.
20 Ferreira-Gonzalez DeAngelo AB, Nasim S, Garret CT. 1995. Ras oncogene activation during hepatocarcinogenesis
21 in B6 C3F1 male mice by dichloroacetic and trichloroacetic acids. Carcinogenesis 16(3): 495-500.
22 FIFRA SAP (Federal Insecticide, Fungicide, and Rodenticide Act Science Advisory Panel). 2004. A Set of
23 Scientific Issues Being Considered by the Environmental Protection Agency Regarding Proposed Science Policy:
24 PPAR-a Agonist-Mediated Hepatocarcinogenesis in Rodents and Relevance to Human Health Risk Assessment.
25 Minutes No. 2003-05 (5 March 2004), Federal Insecticide, Fungicide, and Rodenticide Act Science Advisory Panel.
26 Available: http://www.epa.gov/oscpmont/sap/2003/december9/meetingminutes.pdf/ [accessed 20 September 2005].
27 Fitzgerald JE, Sanyer JL, Schardein JL, Lake RS, McGuire EJ, de la Inglesia FA. 1981. Carcinogen bioassay and
28 mutagenicity studies with the hypolipidemic agent gemfibrozil. J Natl Cancer Inst. 67(5): 1105-16.
29 Fox TR, Schumann AM, Watanabe PG, Yano BL, Maher BM. 1990. Mutational analysis of H-ras oncogene in
30 spontaneous C57B176 X C3H/He mouse liver tumors and tumors induced with genotoxic and nongenotoxic
31 hepatocarcinogens. Cancer Research 50: 4014-4019.
32 Fraga MF, Herranz M, Espada J, Ballestar E, Paz MF, Ropero S, et al. 2004. A mouse skin multistage
33 carcinogenesis model reflects the aberrantDNA methylation patterns of human tumors. Cancer Res 64:5527-5534.
34 Fraga MF, Ballestar E, Villar-Garea A, Boix-Chornet M, Espada J, Schotta G, et al. 2005. Loss of acetylation at
35 Ly s 16 and trimethylation at Ly s20 of histone H4 is a common hallmark of human cancer. Nat Genet 3 7:3 91 -400
36 Froyland L, Madsen L, Vaagenes H, Totland GK, Auwerx J, Kryvi H, Staels B, Berge RK. 1997. Mitochondrion is
37 the principal target for nutritional and pharmacological control of triglyceride metabolism. J Lipid Res. 38:
38 1851-1858.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-418 DRAFT—DO NOT CITE OR QUOTE
-------
1 Fujita J, Marino MW, Wada H, Jungbluth AA, Mackrell PJ, Rivadeneira DE, Stapleton PP, Daly JM. 2001. Effect
2 of TNF gene depletion on liver regeneration after partial hepatectomy in mice. Surgery 129(1): 48-54.
3 Fukuda K, Takemoto K, Tsuruta H. 1983. Inhalation carcinogenicity of trichloroethylene in mice and rats.
4 Industrial Health 21: 243-254.
5 Furstenberger G, Senn HJ. 2002 Insulin-like growth factors and cancer. Lances Oncol 3:298-302
6 Gaglio PJ, Liu H, Dash S, Cheng S, Dunne B, Ratterree M, Baskin G, Blanchard J, Bohm R, Theise ND, LaBrecque
7 D. 2002. Liver regeneration investigated in nonhuman primate model (Macaca mulatto). JHepatol 37: 625-632.
8 Gandillet A, Alexandre E, Royer C, Cinqualbre J, Jaeck D, Richert L. 2003. Hepatocyte ploidy in regenerating
9 livers after partial hepatectomy, drug-induced necrosis, and cirrhosis. Eur Surg Res 35:148-160.
10 Ge R, Wang W, Kramer PM, Yan S, tao L, Pereira MA. 2001a. Wy-14,643-induced hypomethylation of the c-myc
11 gene in mouse liver. Toxicol Sci 62: 28-35.
12 Ge R, Yang S, Kramer PM, Tao L, Pereira MA. 200 Ib. The effect of dichloroacetic acid and trichloroacetic acid on
13 DNA methylation and cell proliferation in B6C3F1 mice. JBiochemMol Toxicol 15:100-106.
14 Ge R, Tao L, Kramer PM, Cunningham ML, Pereira MA. 2002. Effect of peroxisome proliferators on the
15 methylation and protein level of c-myc protooncogene inB6C3Fl mice liver. J Biochem Mol Toxicol 16(1): 41-47.
16 Gelderblom WCA, Galendo D, Abel S, Swanevelder S, Marasas WFO, Wild CP. 2001. Cancer initiation by
17 fumonisinBl in rat liver-Role of cell proliferation. Cancer Lett 169:127-137.
18 George MH, Moore T, Kilburn S, Olson GR, DeAngelo AB. 2000. Carcinogenicity of Chloral hydrate
19 administered in drinking water to the Male F344/N rat and Male B6C3F1 mouse. Toxicologic Pathology 28(4):
20 610-618.
21 Goel SK, Rao GS, Pandya KP, Shanker R. 1992. Trichloroethylene toxicity in mice: a biochemical, hematological
22 and pathological assessment. Indian Journal of Experimental Biology 30: 402-406.
23 Gold LS, Zeiger E. 1997. Handbook of carcinogenic potency and genotoxicity databases. Boca Raton: CRC Press.
24 Goldsworthy TL, Popp JA. 1987. Chlorinated hydrocarbon-induced peroxisomal enzyme activity in relation to
25 species and organ carcinogenicity. Toxicol and Appl Pharmacol 88:225-233.
26 Gorla GR, Malhi H, Gupta S. 2001. Polyploidy associated with oxidative injury attenuates proliferation potential of
27 cells. J Cell Science. 114(16): 2943-2951.
28 Graham D J, Staffa JA, Shatin D, Andrade SE, Schech SD, La GL, et al. 2004. Incidence of hospitalized
29 rhabdomyolysis in patients treated with lipid-lowering drugs. JAMA 292:2585-2590.
30 Gramantieri L, Melchiorri C, Chieco P, Gaiani S, Stecca B, Casali A, Bolondi L. 1996. Alteration of DNA ploidy
31 and cell nuclearity in human hepatocellular carcinoma associated with HBV infection. J Hepatol 25: 848-853.
32 Grasl-Kraupp B, Waldho"r T, Huber W, Schulte-Hermann R. 1993. Glutathione S-transferase isoenzyme patters in
3 3 different subtypes of enzyme-altered rat liver foci treated with the peroxisome proliferator nafenopin or with
34 Phenobarbital. Carcinogenesis 14:2407-2412.
3 5 Greene AK, Wiener S, Puder M, Yoshida A, Shi B, Perez-Atayde AR, Efstathiou JA, Holmgren L, Adamis AP,
36 Rupnick M, Folkman J, O;Reily MS. 2003. Endothelial-directed hepatic regeneration after partial hepatectomy.
37 Ann Surgery 237(4): 530-535.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-419 DRAFT—DO NOT CITE OR QUOTE
-------
1 Greten FR, Eckmann L, Greten TF, Park JM, Li ZW, Egan LJ, Kagnoff MF, Karin M. 2004. IKKbeta links
2 inflammation and tumorigenesis in a mouse model of colitis-associated cancer. Cell 118(3):285-96
3 Gupta S. 2000. Hepatic polyploidy and liver growth control. Seminars in Cancer Biology. 10: 161-171.
4 Haddad S, Charest-Tardif G, Krishnan K. 2000. Physiologically based modeling of the maximal effect of metabolic
5 interactions on the kinetics of components of complex chemical mixtures. J Toxicol Environ Health 61:209-223
6 Hamdan H, Stacey NH. 1993. Mechanisms of trichloroethylene-induced elevation of individual serum bile acids: I.
7 Correlation of trichloroethylene concentrations to bile acids in rat serum. Toxicol and Appl Pharmacol 121:
8 291-295.
9 Hanahan D, Weinberg RA. 2000. The hallmarks of cancer. Cell 100: 57-70.
10 Hasmall SC, Roberts RA. 2000. The nongenotoxic hepatocarcinogen diethylhexylphthalate and methylclofenebrate
11 induce DNA synthesis preferentially in octoploid rat hepatocytes. Toxicologic pathology 28(4): 503-509.
12 Haseman et. al. 1984. Statistical issues in the design, analysis and interpretation of animal carcinogenicity studies.
13 Environ Health Persp 58: 385-392.
14 Hassan MM, Frame A, Part YZ, El-SeragHB. 2002. Rising prevalence of hepatitis C virus infection among
15 patients recently diagnosed with hepatocellular carcinoma in the United States. J Clin Gastroenterol 35: 266-269
16 Hayden MS, Ghosh S. 2004. Signaling to NF-kappaB. Genes Dev 18: 2195-2224.
17 Hays T, Rusyn I, Burns AM, Kennett MJ, Ward JM, Gonzalez FJ, et al. 2005. Role of peroxisome proliferator-
18 activated receptor a (PPAR a) in benzafibrate-induced hepatocarcinogenesis and cholestasis. Carcinogenesis
19 26(l):219-227.
20 Hegi ME, Fox TR, Belinsky SA, Devereux TR, Anderson MW. 1993. Analysis of activated protooncogenes in
21 B6C3F1 mouse liver tumors induced by ciprofibrate, a potent peroxisome proliferator. Carcinogenesis 14(1):
22 145-149.
23 Henschler D, Romen W, Elsasser HM, Reichaert D, Eder E, Radwan Z. 1980. Carcinogenicity study of
24 trichloroethylene by longterm inhalation in three animal species. Arch Toxicol 43: 237-248 (1980).
25 Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, et al. 1998. Incidence and functional consequences of
26 hMLHl promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci USA 95(12):6870-6785.
27 Herren-Freund SL, Pereiira MA, Khoury MD, Olson G. 1987. The carcinogenicity of trichloroethylene and its
28 metabolites trichloroacetic acid and dichloroacetic acid, in mouse liver. Toxicol Appl Pharmacol. 90: 183-189.
29 Herrera E, Samper E, Martin-Caballero M, Flores JM, Lee HW, Blasco MA. 1999. Disease states associated with
30 telomerase deficiency appear earlier in mice with short-telomeres. The EMBO Journal 18(11): 2050-2960.
31 Hoivik DJ, Quails CW, Mirabile RC, Cariello NF, Kimbrough CL, Colton HM, Anderson SP, Santostefano MJ,
32 Morgan PJO, Dahl RR, Brown AR, Zhao Z, Mudd PN, Oliver WB, Brown HR, and Miller RT. 2004. Fibrates
3 3 induce hepatic peroxisome and mitochondria! proliferation without evidence of cellular proliferation and oxidative
34 stress in cynomolgus monkeys. Carcinogenesis 25(9): 1757-1769.
35 Hortelano S, Dewez B, Genaro AM, Diaz-Guerra MJ, Bosca' L. 1995. Nitric oxide is released in regenerating liver
36 after parial hepatectomy. Hepatology 21(3): 776-786.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-420 DRAFT—DO NOT CITE OR QUOTE
-------
1 Hsia CY, Huo TI, Chiang SY, Lu MF, Sun CL, Wu JC, Lee PC, Chi CW, Lui WY, Lee SD. 2007. Evaluation of
2 interleukin-6 , interleukin-10 and human hepatocyte growth factor as tumor markers for hepatocellular carcinoma.
3 Eur J Surg Oncol 33(2): 208-12.
4 Howroyd P, Swanson C, Dunn C, Cattley RC, Gorton JC. 2004. Decreased longevity and enhancement of age-
5 dependent lesions in mice lacking the nuclear receptor peroxisome proliferator-activated receptor a (PPARa).
6 Toxicologic Pathology 32:591-599.
7 IARC 2000. Di(2-ethylhexyl) phthalate. In: IARC monographs on the evaluation of carcinogenic risk of chemicals
8 to humans Vol 77 Some industrial chemicals. Lyon, France: International Agency for Research on Cancer, 148.
9 Available online at http://monographs.iarc.fr/ENG/Monographs/vol 177.pdf-141.
10 Isaacs KK, Evans MV, Harris TR. 2004. Visualization-based analysis for a mixed-inhibition binary PBPK model:
11 Determination of inhibition mechanism. J Pharmacokinet Pharmacodyn 31:215-242.
12 Ito Y, Yamanoshita O, Asaeda N, Tagawa Y, Lee CH, Aoyama T, Ishihara G, Furuhashi K, Kamijima M, Gonzalez
13 FJ, Nakajima T. 2007. Di(2-ethylhexyl)phthalate induced hepatic tumorigenesis through a peroxisome proliferator-
14 activated receptor a-independent pathway. J Occup Health 49:172-182.
15 Jia HL, Ye QH, Qin LX, Budhu A, Forgues M, Chen Y, Liu YK, Sun HC, Wang L, Lu HZ, Shen F, Tang ZH, Wang
16 XW. 2007. Gene expression profiling reveals potential biomarkers of human hepatocellular carcinoma. Clin
17 Cancer Res 13(4): 1133-1139.
18 Jirtle RL, Meyer SA, brokenbrough JS. 1991. Liver tumor promoter Phenobarbital: a biphasic modulator of
19 hepatocyte proliferation. Prog Clin Biolo Res 369: 209-216.
20 Jorquera R, Tanguay RM. 2001. Fumarylacetoacetate, the metabolite accumulating in hereditary tyrosinemia,
21 activates the ERK pathway and induces mitotic abnormalities and genomic instability. Hum Mol Genet
22 10:1741-1752.
23 Kamijo Y, Hora K, Nakajima T, Kono K, Takahashi K, Ito Y, Higuchi M, Kiyosawa K, Shigematsu H, Gonzalez Fj,
24 Aoyama T. 2007. Peroxisome proliferator-activated receptor a by long-term exposure to the plasticizer Di-(2-
25 ethylhexyl)phthalate. Journal of American Society of Nephrology. 18(1): 176-88.
26 Karpinets TV, Foy BD. 2005. Tumorigenesis: the adaptation of mammalian cells to sustained stress environment by
27 epigenetic alterations and succeeding matched mutations. Carcinogenesis 26(8): 1323-1334.
28 Kato-Weinstein J, Stauber AJ, Orner GA., Thrall BD, Bull PJ. 2001. Differential effects of dihalogenated and
29 trihalogenated acetates in the liver of B6C3 Fl mice. Journal of Applied Toxicology 21: 81-89.
30 Kawamoto T, Hobara T, Nakamura K, Imamura A, Ogino K, Kobayashi H, Iwamoto S, Sakai T. 1988. Induction
31 of cytochrome P450, cytochrome b-5, NADPH cytochrome c reductase and change in cytochrome P-450 isozymes
32 with long-term trichloroethylene treatment. Toxicology 53: 239-249.
33 Keech A. Simes RJ, Barter P, et al 2005. FIELD study investigators. Effects of long-term fenofibrate therapy on
34 cardiovascular events in 9795 people with type 2 diabetes mellitus (the FIELD study): randomized controlled trial
35 [published correction appears in Lancet, 2006:368:1420] Lancet 366:1849-1861.
36 Keech A, Simes J, Barter P, Best J, Scott R, Taskinen MR. 2006. Correction to the FIELD study report. Lancet
37 368(9545): 1415.
3 8 Keshava N and Caldwell JC. 2006. Key Issues in the Role of Peroxisome Proliferator-Activated Receptor Agonism
39 and Cell Signaling in Trichloroethylene Toxicity. Environmental Health Perspectives. 114 (9): 1464-1470.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-421 DRAFT—DO NOT CITE OR QUOTE
-------
1 Keshava N, Chiu W, Caldwell JC. 2007. Keshava and Caldwell respond. Environmental Health Perspectives.
2 115(1):A4-A5.
3 Khandawala HM, McCutcheon, IE, Flyvbjerg A, Friend KE. 2000. The effects of insulin-like growth factors on
4 tumorigenesis and neoplastic growth. Endocr. Rev. 21(3), 215-244.
5 Kim HJ, Odend'hal S, Bruckner JV. 1990. Effect of oral dosing vehicles on the acute hepatoxicity of carbon
6 tetrachloride in rats. Toxicol Appl Pharmacol 102: 34-49.
7 Kim SZ, Kupke KG, lerardi-Curto L, Holme E, Greter J, Tanguay RM, et al. 2000. Hepatocellular carcinoma
8 despite long-term survival in chronic tyrosinaemia I. J Inherit Metab Dis 23:791-804.
9 Kitada T, Seki S, Kawakita N, Kuroki T, Monna T. 1995. Telomere shortening in chronic liver diseases. Biochem
10 Biophys Res Commun 211: 33-39.
11 Kjellstrand P, Lanke J, Bjerkemo M, Zetterqvist L, Mansoon L. 1980. Irreversible effects of trichloroethylene
12 exposure on the central nervous system. Scand J Work Environ Health. 6: 40-47.
13 Kjellstrand P, Kanje M, Mansson L, Bjerkemo M. Mortensen I, Lanke J, HolmquistB. 1981. Toxicology 21:
14 105-115.
15 Kjellstrand P, Holmquist B, Mandal N, Bjerkemo M. 1983 (a). Effects of continuous trichloroethylene inhalation
16 on different strains of mice. Acta Pharmacol et Toxicol 53: 369-374.
17 Kjellstrand P, Holmquist B, Aim P, Kanje M, Romare S, Jonsson I, Mannson L, Bjerkemo M. 1983 (b).
18 Trichloroethylene: further studies of the effects on body and organ weights and plasma butyryIcholinesterase
19 activity in mice. Acta Pharmacol et Toxicol 53: 375-384.
20 Klaunig JE, Babich MA, Baetc.ke KP, Cook JC, Corton JC, David RM, et al. 2003. PPARa agonist-induced rodent
21 tumors: modes of action and human relevance. Crit Rev Toxicol 33:655-780.
22 Kobayashi M, Ikeda K, Hosaka T, Sezaki H, Someya T, Akuta N, Suzuki F, Suzuki Y, Saitoh S, Arase Y, Kumada
23 H. 2006. Dysplastic nodules frequently develop into hepatocellular carcinoma in patients with chronic viral
24 hepatitis and Cirrhosis. Cancer 106:636-647.
25 Kraupp-Grasl B, Huber W, Putz Bm Gerbracht U, Schulte-Herman R. 1990. Tumor promotion by peroxisome
26 proliferator nafenopin involving specific subtype of altered foci in rat liver. Cancer Research 50:3701-3706.
27 Kudryavtsev BN, Kudryavtseva MV, Sakuta GA, Stein GI. 1993. Human hepatocyte polyploidization kinetics in
28 the course of life cycle. Virchows Arch B Cell Pathol Incl Mol Pathol 64 (6):387-393.
29 Kumar P, PrasadAK, MajiBK, ManiU, DuttaKK. 2001. Hepatotoxic alterations induced by inhalation of
30 trichloroethylene (TCE) in rats. Biomedical and Environmental Sciences 14 : 325-332.
31 Lake BG, Evans JG, Cunninghame ME, Price RJ. 1993. Comparison of the hepatic effects of nafenopin and WY-
32 14,643 on peroxisome proliferation and cell replication in the rat and Syrian hamster. Environ Health Perspect
33 101:241-248.
34 Lalwani ND, Dethloff LA, Hasking JR, Robertson DG, DE LA Iglesia FA. 1997. Increased nuclear ploidy, not cell
3 5 proliferation, is sustained in the peroxisome proliferated-treated rat liver. Toxicologic Pathology 25(2) 165-176.
36 Lantendresse JR, Pereira MA. 1997. Dissimilar characteristics of N-methyl-N-nitrosurea-initiated foci and tumors
37 promoted by dichloroacetic acid or trichloroacetic acid in the liver of female B6C3F1 mice. Toxicologic Pathology
38 25(5): 433-440.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-422 DRAFT—DO NOT CITE OR QUOTE
-------
1 Lapis K, Zalatnai A, Timar F, Thorgeirsson UP. 1995. Quantitative evaluation of lysozome- and CD68 -positive
2 Kupffer cells in diethylnitrosoamine-induced hepatocellular carcinomas in monkeys. Carcinogenesis 16(12):
3 3083-3085.
4 Larson JL, Bull RJ. 1992. Metabolism of lipoperoxidative activity of trichloroacetate and dichloroacetate in rats
5 and mice. Toxicology and Applied Pharmacology 115L 268-277.
6 Laughter AR, Dunn CS, Swanson CL, Howroyd P, Cattley RC, Gorton JC. 2004. Role of the peroxisome
7 proliferator-activated receptor a (PPARa) in responses to trichloroethylene and metabolites, trichloroacetate and
8 dichloroacetate in mouse liver. Toxicology 203:83-98.
9 LaVecchia C, Negri E, Franceschi S, D'Avanzo B, Boyle P. 1994. A case-control study of diabetes mellitus and
10 cancer risk. Br J Cancer 70:950-953.
11 Lawrence T, Hagemann T, Balkwill F. 2007. Sex, cytokines, and cancer. Science 317: 51-52.
12 Leakey JEA, Seng JE, Lantendresse JR, Hussain N, Allen LJ, Allaben WT. 2003a. Dietary controlled
13 carcinogenicity study of chloral hydrate in male B6C3F1 mice. Toxicology and Applied Pharmacology 193:
14 266-280.
15 Leakey JEA, Seng JE, Allaben WT. 2003b. Body weight considerations in the B6C3F1 mouse and the use of
16 dietary control to standardize background tumor incidence in chronic bioassays. Toxicology and Applied
17 Pharmacology 193: 237-265.
18 Ledda-Columbano GM, Columbano A, Curto M, Ennas MG, Coni P, Sarma DSR, Pani P. 1989. Further evidence
19 that mitogen-induced cell proliferation does not support the formation of enzyme-altered islands in rat liver by
20 carcinogens. Carcinogenesis 10: 847-850.
21 Ledda-Columbano GM, Coni P, Simbula G, Zeddam and Columbano A 1993. Compensatory regeneration, mitogen
22 induced liver growth, and multistage chemical Carcinogenesis.
23 Ledda-Columbano GM, Curto M, Piga R, Zedda AI, Menegazzi M, Sartori C, Shinozuka H, Bluethmann H, Poli V,
24 Ciliberto G, Columbano A. 1998. In vivo hepatocyte proliferation is inducible through a TNF, IL-6-independent
25 pathway. Oncogene 17:1039-1044.
26 Ledda-Columbano GM, Perra A, Concas D, Cossu C, Molotzu F, Sartori C, Shinozuka H, Columbano A. 2003.
27 Different effects of the liver mitogens triiodo-thyronine and ciprofibrate on the development of rat hepatocellular
28 carcinoma. Toxicol Pathol 31:113-120.
29 Lee SST, Pineau T, Drago J, Lee EJ, Owens JW, Kroetz DL, Fernandez-Salguero PM, Westphal H, Gonzalez FJ.
30 1995. Targeted disruption of the a isoform of the peroxisome proliferator-activated receptor gene in mice results in
31 abolishment of the pleiotropic effects of peroxisome proliferators. Molecular and Cellular Biology.
32 15(6):3012-3022.
33 Lee S, Lee HJ, Kim JH, Lee HS, Jang JJ, Kang GH. 2003. Aberrant CpG island hypermethylation along multistep
34 hepatocarcinogenesis. Am J Pathol 163: 1371-13787
35 Lee JS, Chu IS, Heo J, Calvisi DF, Sun Z, Roskams T, Durnez A, Demetrius AJ, Thorgeirsson SS. 2004(a).
3 6 Classification of prediction of survival in hepatocellular carcinoma by gene expression profiling. Hepatology 40(3):
37 667-676.
3 8 Lee JS, Chu IS, Mikaelyan A, Calvisi DF, Heo J, Reddy JK, Thorgeirsson SS. 2004(b). Application of comparative
39 functional genomics to identify best-fit mouse models to study human cancer. Nature Genetics.36(12): 1306-1311.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-423 DRAFT—DO NOT CITE OR QUOTE
-------
1 Librecht L. 2006. Hepatic progenitor cells in human liver tumor development. World J Gastroenterol 12 (39:
2 6261-6265.
3 Leuschner J, Beuscher N. 1998. Studies on the mutagenic and carcinogenic potential of chloral hydrate. Arzneim-
4 Forsch./Drug Res. 48(11): 961-968.
5 Lewitt MS, Brismar K, Wang J, Wivall-Helleryd IL, Sindelar P,Gonzalez FJ, et al. 2001. Responses of insulin-like
6 growth factor (IGF)-I and IGF-binding proteins to nutritional status in peroxisome proliferator-activated receptor-a
7 knockout mice. GrowthHorm IGF Res 11:303-313.
8 Lin YZ, Brunt EM, Bowling W, Hafenrichter DG, Kennedy SC, Flye MW, Ponder KP. 1995. Ras-transduced
9 diethylnitrosamine-treated hepatocytes develop into cancers of mixed phenotype in vivo. Cancer Res 55:
10 5242-5250.
11 Lin LW, Yin XY, He YM, Gao HD, Lin XD. 2003. Biological characteristics of HCC by ultrasound-guided
12 aspiration biopsy and it clinical application. World J Gastroenterol 9(5): 941-945.
13 Lingohr MK, Bull RJ, Kato-Weinstein J, Thrall BD. 2002. Dichloroacetate stimulates glycogen accumulation in
14 primary hepatocytes through an insulin-independent mechanism. Toxicol Sci 68(2):508-515.
15 Lingohr MK, Thrall BD, Bull RJ. 2001. Effects of dichloroacetate (DCA) on serum insulin levels and insulin-
16 controlled signaling proteins in livers of male B6C3F1 mice. Toxicol Sci 59(1): 178-184.
17 Lorimore SA, Coates PJ, Wright EG. 2003. Radiation-induced genomic instability and bystander effects: interrelated
18 nontargeted effects of exposure to ionizing radiation. Oncogene 22: 7058-7069.
19 Luedde T, Beraza N, Kotsikoris V, van Loo G, Nenci A, De Vos R, Roskams T, Trautwein C, Pasparakis M. 2007.
20 Deletion of NEMO/IKKgamma in liver parenchymal cells causes steatohepatitis and hepatocellular carcinoma.
21 Cancer Cell 11: 119-132.
22 Lundgren B, DePeirre JW. 1989. Proliferation of peroxisomes and induction of cytosolic and microsomal epoxide
23 hydrolase in different strains of mice and rats after dietary treatment with clofibrate. Xenobiotica 19: 876-881.
24 Lundgren B, Meijer J, DePeirre JW. 1987. Induction of cytosolic and microsomal epoxide hydrolases and
25 proliferation of peroxisomes and mitochondria in mouse liver after dietary exposure to p-chlorophenoxyacetic acid,
26 2,4-Dichlorophenoxyacetic acid and 2,4,5-trichlorophenoxy acetic acid. Biochemical Pharmacology
27 36(6):815-821.
28 Macdonald N, Holden RR, Roberts RA. 1999. Additional peroxisome proliferator-activated receptor alpha to guinea
29 pig hepatocytes confers increased responsiveness to peroxisome proliferators. Cancer Res 59: 4776-4780.
30 Maeda S, Kamata H, Luo JL, Leffert H, Karin M. 2005. IKK(3 couples hepatocyte death to cytokine-driven
31 compensatory proliferation that promotes chemical hepatocarcinogenesis. Cell. 121:977-990.
32 Malarkey DE, Johnson K, Ryan L, Boorman G, Maronpot RR. 2005. New insights into functional aspects of liver
33 morphology. Toxicol Pathol 33: 27-34.
34 Maltoni C, Lefemine G, Cotti G. 1986. Experimental research on trichloroethylene carcinogenesis. Archive of
35 Research on Industrial Carcinogenesis Vol 5 Princeton Scientific Publishing. Maltoni and Mehlman editors. Pages
36 45-153.
37 Maltoni C, Lefemine G, Cotti G, Perino G. 1988. Long-term carcinogenicity bioassays on trichloroethylene
3 8 administered by inhalation to Sprague-Dawley rats and Swiss and B6C3F1 mice. Annals of the New York Academy
39 of Sciences 534: 316-343.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-424 DRAFT—DO NOT CITE OR QUOTE
-------
1 Maronpot RR. 2007. Biological basis of differential susceptibility to hepatocarcinogenesis among mouse strains.
2 Background paper prepared for the National Academy of Sciences for use by the Standing Committee on Risk
3 Analysis Issues & Reviews. Workshop #3: Mouse Liver Tumors, Benefits and Constraints on Use in Human Health
4 Risk Assessment, Qualitative and Quantitative Aspects. Oct 14, 2007.
5 Maronpot RR, Fox T,, Malarkey DE, Goldsworthy TL. 1995. Mutations in the ras proto-oncogene clues to etiology
6 and molecular pathogenesis of mouse liver tumors. Toxicology 101: 125-156.
7 Maronpot RR, Flake G, Huff J. 2004. Relevance of animals carcinogenesis findings to human cancer predictions
8 and prevention. Toxicol Pathol 32(Suppl. 1): 40-48.
9 Marsman DS, Cattley RC, Conway JG, Popp JA. 1988. Relationship of hepatic peroxisome proliferation and
10 replicative DNA synthesis to the hepatocarcinogenicity of the peroxisome proliferators di(2-ethylhexyl)phthalate
11 and [4-chloro-6-(2,3-xylidino)-2-pyrimidinylthio]acetic acid (Wy-14,643) in rats. Cancer Res 48:6739-6744.
12 Marsman DS, Goldsworthy TL, Popp JA. 1992. Contrasting hepatocyte peroxisome proliferation, lipofuscin
13 accumulation and cell turnover for the hepatocarcinogen WY-14,643 and clofibric acid.. Carcinogenesis 13(6):
14 1011-1017.
15 Martinez-Larranaga MR, Anadon A, Diaz MJ, Fernandez B, Sevil B, Fernandez-Cruz ML, Fernandez MC, Martinez
16 MA, AntonR. 1996. Induction of cytochromeP4501Al and P4504A1 activities and peroxisomal proliferation by
17 fumonisinBl. Toxicol and Appl Pharmacol 141: 185-194.
18 Mather GG, ExonJH, KollerLD. 1990. Subchronic 90 day toxicity of dichloroacetic and trichloroacetic acid in
19 rats. Toxicology 64: 71-80.
20 McEntee GP, Batts KA, Katzmann JA, Ilstrup DM Nargorney DM. 1991. Relationship of nuclear DNA content to
21 clinical and pathologic findings in patients with primary hepatic malignancy. Surgery. 111(4): 376-379.
22 McLean AJ, Cogger VC, Chong GC, Warrant A, Markus AM, Dahlstrom JE, Le Couter DG. 2003. Age-related
23 pseudocapillarization of the human liver. J Pathol 200: 112-117.
24 Mehendale HM. 2000. PPAR-alpha: a key to the mechanism of hepatoprotection by clofibrate. Toxicol Sci.
25 57(2): 187-90.
26 Melchiorri C, Chieco P, Zedda AI, Coni P, Ledda-Columbano GM, Columbano A. 1993. Ploidy and nuclearity of
27 rat hepatocytes after compensatory regeneration or mitogen-induced liver growth. Carcinogenesis 14(9):
28 1825-1830.
29 Melchiorri C, Bolondi L, Chieco P, Pagnoni M, Gramantieri L, Barbara L. 1994. Diagnostic and prognostic value of
30 DNA ploidy and cell nuclearity in ultrasound-guided liver biopsies. Cancer 74(6): 1713-1719.
31 Melnick RL, Jameson CW, Goehl TJ, Maronpot RR, Collins B J, Greenwell A, Harrington FW, Wilson RE,
32 Tomaszewski KE, Agarwal DK. 1987. Toxicity of microencapsulated trichloroethylene in Fischer 344 rats.
33 Fundamental and Applied Toxicology 8: 432-442.
34 Melnick RL, Huff J, Barrett JC, Maronpot RR, Lucier G, Portier CJ. 1993. Cell proliferation and chemical
35 carcinogenesis: Symposium Overview. Environ Health Persp 101(Suppl 5): 3-8.
36 Melnick RL, Kohn MC, Portier CJ. 1996. Implications for risk assessment of suggested nongenotoxic mechanisms
37 of chemical carcinogenesis. Environmental Health Perspectives 104(1): 123-134.
3 8 Melnick RL. 2001. Is peroxisome proliferation an obligatory precursor step in the carcinogenicity of di(2-
39 ethylhexyl)phthalate (DEHP)? Environ Health Perspect 109(5): 437-442.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-425 DRAFT—DO NOT CITE OR QUOTE
-------
1 Melnick RL. 2002. The IARC evaluation of di(2-ethylhexyl)phthalate (DEHP): a flawed decision based on an
2 untested hypothesis. Int J Occup Environ Health 8(3): 284-286.
3 Melnick RL. 2003. Suppression of crucial information in the IARC evaluation of DEHP. Int J Occup Environ Health
4 9(1): 84-85.
5 Melnick RL, Thayer KA, Bucher JR. 2007. Conflicting views on chemical carcinogenesis arising from the design
6 and evaluation of rodent carcinogenicity studies. Environ Health Perspectives. Doi: 10.1289/ehp.9989 Online
7 November 7, 2007.
8 Menegazzi M, Carcereri-De Prati A, Suzuki H, Shinozuka H, Pibiri Mo, Piga R, Columbano A, and Ledda-
9 Columbano GM. 1997. Liver cell proliferation induced by nafenopin and cyproterone acetate is not associated with
10 increases in activation of transcription factors NF-KB and AP-1 or with expression of tumor necrosis factor a.
11 Hepatology. 25(3):585-591.
12 Merdink JL, Robison LM, Stevens DK, Hu M, Parker JC, Bull RJ. 2008. Kinetics of chloral hydrate and its
13 metabolites in male human volunteers. Toxicology 245: 130-140.
14 Merrick BA, Robinson M, Condi LW. 1989. Differing hepatotoxicity and lethality after subacute trichloroethylene
15 exposure in aqueous or corn oil gavage vehicles in B6C3F1 mice. Journal of Applied Toxicology. 9(1):15-21.
16 Meyer K, Lee JS, Dyck PA, Cao WQ, Rao MS, Thorgeirsson SS, Reddy JK. 2003. Molecular profiling of
17 hepatocellular carcinomas developing spontaneously in acyl-CoA deficient mice: comparison with liver tumors
18 induced by wild-type mice by a peroxisome proliferator and a genotoxic carcinogen. Carcinogenesis 24(5):
19 975-84.
20 Michel C, Desdouets C, Slaoui M, Isaacs KR, Roberts RA, Boitier E. 2007. Diethylnitrosamine initiation does not
21 alter clofibric acid-induced hepatocarcinogenesis in the rat. Toxicological Sciences. 99(1): 58-69.
22 Miller RT, Shah RS, Cattley RC, Popp JA. 1996. The peroxisome proliferator WY-14,643 and methylclofenapate
23 induce hepatocellular ploidy alterations and ploidy-specific DNA synthesis in F344 rats. Toxicology and Applied
24 Pharmacology 138:317-323.
25 Minamishima YA, NakayamaK, NakayamaKI. 2002. Recovery of liver mass without proliferation of hepatocytes
26 after partial hepatectomy in Skpl-deficient mice. Cancer Research 62; 995-999.
27 Mirsalis JC, Tyson CK, Steinmetz KL, Loh EK, Hamilton CM, Bakke JP, Spalding JW. 1989. Measurement of
28 unscheduled DNA synthesis and S-phase synthesis in rodent hepatocytes following in vivo treatment: testing of 24
29 compounds. Environmental and Molecular Mutagenesis 14: 155-164.
30 Miura N Miura N, Horikawa I, Nishimoto A, Ohmura H, Ito H, Hirohashi S, Shay JW, Oshimura M. 1997.
31 Progressive telomere shortening and telomerase reactivation during hepatocellular carcinogenesis. Cancer Genet
32 Cytogenet 93(1): 56-62.
33 Moinzadeh P, Breuhahn K, Stutzer H, Schirmacher P. 2005. Chromosome alterations in human hepatocellular
34 carcinomas correlated with aetiology and histological grade-results of an exploratory CGH meta-analysis. British
35 Journal of Cancer 92:935-941.
36 Morimura K, Cheung C, Ward JM, Reddy JK, Gonzalez FJ. 2006. Differential susceptibility of mice humanized for
37 peroxisome proliferator-activated receptor {alpha} to Wy-14,643 induced liver tumorigenesis. Carcinogenesis.
38 27:1074-1080.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-426 DRAFT—DO NOT CITE OR QUOTE
-------
1 Nakajima T, Kamino Y, Usuda N, Liang Y, Fukushima Y, Kametani K, Gonzalez FJ, Aoyama T. 2000. Sex-
2 dependent regulation of hepatic peroxisome proliferation in mice by trichloroethylene via peroxisome proliferator-
3 activated receptor a (PPARa). Carcinogenesis 21(4): 677-682.
4 Nakajima T, Moriguchi M, Mitsumoto Y, Sekoguchi S, Nishikawa T, Takshima H, Watanabe T, Katagishi T,
5 Kimura H, Okanoue T, Kagawa K. 2004. Centrosome aberration accompanied with p53 mutation can induce
6 genetic instability in hepatocellular carcinoma. Modern Pathology 17: 722-727.
7 NAS (National Academy of Sciences). 2006. Assessing the Human Health Risks of Trichloroethylene: Key
8 Scientific Issues. BEST-K-03-06-A. Available: http://www8.nationalacademies.org/cp/projectview.aspx?key=127
9 [accessed 14 March 2006].
10 National Cancer Institute (NCI) Technical Report No. 2. Carcinogenesis bioassay of trichloroethylene. (Cas no. 79-
11 01-6). NCI-CG-TR-2. February 1976.
12 National Toxicology Program (NTP) Technical Report No. 273. Toxicology and Carcinogenesis studies of
13 trichloroethylene (Cas No. 79-01-6) in four strains of rats (ACI, August, Mashal, Osborne-Mendel) (gavage studies).
14 April 1988.
15 National Toxicology Program (NTP) Technical Report No. 243. Carcinogenesis studies of trichloroethylene
16 (without epichlorhydrin) (Cas No. 79-01-6) inF344/NRatsandB6C3Fl mice (gavage studies). May 1990.
17 Naugler WE, Sakurai T, Kim S, Maeda S, Kim K, Elsharkawy AM, Karin M. 2007. Gender disparity in liver
18 cancer due to sex differences in MyD88-dependent IL-6 production. Science 317: 121-124.
19 Neghab M, Stacey NH. 1997. Toluene-induced elevation of serum bile acids: relationship to bile acid transport. J
20 Toxicol and Environ Health. 52: 249-268.
21 Neghab M, Qu Sm, Bai CL, Caples J, Stacey NH. 1997. Raised concentration of serum bile acids following
22 occupational exposure to halogenated solvents, l,l,2-trichloro-l,2,2-trifluoroethane and trichloroethylene. Int Arch
23 Occup Environ Health 70: 187-194.
24 Nelson MA, Bull RJ. 1988. Induction of strand breaks in DNA by trichloroethylene and metabolites in rat and
25 mouse liver in vivo. Toxicology and Applied Pharmacol 94: 45-54.
26 Nelson MA, Lansing AJ, Sanches AM, Bull RJ. 1989. Dichloroacetic acid and trichloroacetic acid-induced DNA
27 strand breaks are independent of peroxisome proliferation. Toxicology 58:239-248.
28 Nelson MA, Sanchez, Bull RJ, Sylvester SR. 1990. Increased expression of c-myc and c-H-ras in dichloroacetate
29 and trichloroacetate-induced liver tumors in B6C3F1 mice. Toxicology 64: 47-57.
30 Nickoloff BJ, Ben-Neriah, Pikarsky E. 2005. Inflammation and cancer: is the link as simple as we think? J Invest
31 Dermatol 124(6): x-xiv.
32 National Institute of Environmental Health Science. 1998. Tumor incidence in control animals by route and vehicle
33 of administration, B6C3F1 mice, Research Triangle Park NC, NIEHS Toxicology Data Management Systems.
34 Nissen SE, Nicholls SJ, Wolski K, Howey DC, McErlean E, Want MD, Gomez EV, Russo JM. 2007. Effects of a
3 5 potent and selective PPAR-a agonist in patients with atherogenic dyslipidemia or hypercholesterolemia: two
36 randomized controlled trials. JAMA 297 (12): 1362-1373.
37 Nohammer C, Brunner F, Wolkart G, Staber PB, Steyrer E, Gonzalez FJ, Zechner R, Hoefler G. 2003. Myocardial
3 8 dysfunction and male mortality in peroxisome proliferator-activated receptor alpha knockout mice overexpressing
39 lipoprotein lipase in muscle. Lab Invest. 83:259-69.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-427 DRAFT—DO NOT CITE OR QUOTE
-------
1 Nunes J, Ehrich M, Robertson J. 2001. Toxicosis associated with dual oral exposure of rats to lead and
2 trichloroethylene. Toxicologic Pathology. 29(4): 451-457.
3 Okino T, Nakajima T, Nakano M. 1991. Morphological and biochemical analyses of trichloroethylene hepatoxicity:
4 difference in ethanol- and Phenobarbital-pretreated rats. Toxicology and Applied Pharmacology 108: 379-389.
5 Olivier P, Plancke MO, Marzin D, Clavey V, Sauzieres J, Fruchart JC. 1988. Effects of fenofibrate, gemfibrozil,
6 and nicotinic acid on plasma lipoprotein levels in normal and hyperlipidemic mice: A proposed model for drug
7 screening. Atherosclerosis. 70(1-2): 107-114.
8 Oneta CM, Lieber CS, Li J, Ruttimann S, Schmid B, Lattmann J, Rosman AS, Seitz HK. 2002. Dynamics of
9 cytochrome P4502E1 activity in man: Induction by ethanol and disappearance during withdrawal phase. J Hepatol
10 36:47-52.
11 Parrish JM, Austin EW, Stevens DK, Kinder DH, Bull RJ. 1996. Haloacetate-induced oxidative damage to DNA in
12 the liver of male B6C3F1 mice. Toxicology 110:103-111.
13 Parnell MJ, Koller LD, Exon JH, Arnzen JM. 1986. Trichloroacetic acid effects on rat liver peroxisome and
14 enzyme-altered foci. Environmental Health Perspectives 69:73-79.
15 Pastino GM, Yap WY, Carroquino M. 2000. Human variability and susceptibility to trichloroethylene. Environ
16 Health Perspect. 108 Suppl 2:201-14.
17 Pereira MA. 1996. Carcinogenic activity of dichloroacetic acid and trichloroacetic acid in the liver of female
18 B6C3F1 mice. Fundamental and Applied Toxicology 31: 192-199.
19 Pereira MA, PhelpsBJ. 1996. Promotion by dichloroacetic acid and trichloroacetic acid of N-methyl-N-nitrosurea-
20 initiated cancer in the liver of female B6C3F1 mice. Cancer Letters 102: 133-141.
21 Pereira MA, Li K, Kramer PM. 1997. Promotion by mixtures of dichloroacetic acid and trichloroacetic acid of N-
22 methyl-N-nitrosourea-initiated cancer in the liver of female B6C3F1 mice. Cancer Letters 115:15-23.
23 Pereira MA, Kramer PM, ConranPB, Tao L. 2001. Effect of chloroform on dichloroacetic acid and trichloroacetic
24 acid-induced hypomethylation and expression of the c-myc gene and on their promotion of liver and kidney tumors
25 inmice. Carcinogenesis 22(9): 1511-1519.
26 Pereira MA, Wang W, Kramer PM, Tao L. 2004. Prevention by methionine of dichloroacetic acid-induced liver
27 cancer and DNA hypomethylation in mice. Toxicological Sciences 77: 243-248.
28 Perkins ND. 2004. NF-KB:tumor promoter or suppressor? Trends Cell Biol 14: 64-69.
29 Perkins ND, Gilmore TD. 2006. Good cop, bad cop: the different faces of NF-icB. Cell death and Differentiation.
30 13:759-772.
31 Peters JM, Cattley RC, Gonzalez FJ. 1997. Role of PPAR alpha in the mechanism of action of the nongenotoxic
32 carcinogen and peroxisome proliferator Wy-14,643. Carcinogenesis. 18:2029-33.
33 Pikarsky E, Porat RM, Stein I, Abramovitch R, Amit S, Kasem S. Gutkovich-Pyest E, Urieli-Shoval S, Galun E,
34 Ven-Neriah Y. 2004. NF-icB functions as a tumour promoter in inflammation-associated cancer. Nature 431:
35 461-466.
36 Pogribny IP, Tryndyak VP, Woods CG, Witt SE, Rusyn I. 2007. Epigenetic effects of the continuous exposure to
37 peroxisome proliferator WY-14,643 in mouse liver are dependent on peroxisome proliferator activated receptor a.
3 8 Mutation Research 625:62-71.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-428 DRAFT—DO NOT CITE OR QUOTE
-------
1 Popper H. 1986. Aging and the liver. In: Popper H, Schaffner F, Editors. Progress in liver disease. Vol. III. Orlando
2 (FL): Grune and Stratton: 659-683.
3 Rake JP, Visser G, Labrune P, Leonard JV, Ullrich K, Smit GP. 2002. Glycogen storage disease type I: diagnosis,
4 management, clinical course and outcome. Results of the European Study on Glycogen Storage Disease Type I
5 (ESGSD I). Eur J Pediatr 161(Suppl 1):S20-S34.
6 Rakoff-Nahoum S, Medzhitov R. 2007. Regulation of spontaneous intestinal tumorigenesis through the adaptor
7 protein MyD88. Science 317: 124-127.
8 Rakyan VK, BlewittME, DrukerR, Preis JI, WhitelawE. 2002. Metastable epialleles in mammals. Trends Genet
9 18:348-351.
10 Rakyan VK, Chong S, Champ ME, Cuthbert PC, Morgan HD, Luu KV, Whitelaw E. 2003. Transgenerational
11 inheritance of epigenetic states at the muring AxinFu allele occurs after maternal and paternal transmission. Proc
12 Natl Acad Sci USA 100: 2538-2543.
13 Ramdhan DH, Kamijima M, Yamada N, Ito Y, Yanagiba Y, Nakamura D, Okamura A, Ichihara G, Aoyama T,
14 Gonzalez FJ, Nakajima T. 2008. Molecular mechanisms of trichloroethylene-induced hepatoxicity mediated by
15 CYP2E1. Toxicol and Appl Pharmacol 231: 300-307.
16 Rao MS, Tatematsu M, Subbarao V, Ito N, Reddy JK. 1986. Analysis of peroxisome proliferator-induced
17 preneoplastic and neoplastic lesions of rat liver fro placenta! form of glutathione S-transferase and y-
18 glutamyltranspeptidase. Cancer Research 46: 5287-5290.
19 Reamone-Buettner SM, Borlak J. 2007. A new paradigm in toxicology and teratology; altering gene activity in the
20 absence of DNA sequence variation. Reproductive Toxicology 24: 20-30.
21 Rhee I, Bachman KE, Park BH, Jair KW, Yen RW, et al. 2002. DNMT1 and DNMT3b cooperate to silence genes in
22 human cancer cells. Nature 416(6880):552-556.
23 Richardson FC, Boucheron JA, Dyroff MC, Popp JA, Swenberg JA. 1986. Biochemical and morphological studies
24 of heterogeneous lobe responses in hepatocarcinogenesis. Carcinogenesis 7(2): 247-251.
25 Richmond, RE; Carter, JH; Carter, HW; et al. (1995) Immunohistochemical analysis of dichloroacetic acid (DCA)-
26 induced hepatocarcinogenesis in male Fischer (F344) rats. Cancer Lett 92:67-76.
27 Roskams T, Yang SQ, Koteish A, Durnez A, DeVos R, Huang X, Achten R, Verslype C, Diehl AM. 2003.
28 Oxidative stress and oval cell accumulation in mice and humans with alcoholic and nonalcoholic fatty liver disease.
29 Am JPathol 164(4): 1301-1311.
30 Roskams TA, Theise ND, Balabaud C, Bhagat G, Bhagat PS, Bioulac-Sage P, Brunt EM, Crawford JM, Crosby HA,
31 Desmert V, Finegold MJ, Geller SA, Gouw ASH, Hytiroglou P, Knisely AS, Kojiro M, Lefkowitch JH, Nakanuma
32 Y, Olynyk JK, Park YN, Portman B, Saxena R, Scheur PJ, Stain AJ, Thung SN, Wanless IR, West AB. 2004.
33 Nomenclature of the finer branches of the biliary tree: canals, ductules, and ductular regions in human livers. Special
34 Article. Hepatology 39(6): 1739-1745.
3 5 Rudolf KL, DePinho RA. 2001. In the liver biology and pathology (eds Arias I, Fausto N, Boyer J, Chisart F and
36 Shafritz D) 1001-1011 (Lippincott Williams and Wilkins, Philadelphia)
37 Rusyn I, Bradham CA, Cohn L, Schoonhoven R, Swenberg JA, Brenner DA, Thurman RG. 1999. Corn oil rapidly
3 8 activates nuclear factor - KB in hepatic Kupffer cells by oxidant-dependent mechanisms. Carcinogenesis 20 (11):
39 2095-2100.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-429 DRAFT—DO NOT CITE OR QUOTE
-------
1 Rusyn I, Peters JM, Cunningham ML. 2006. Modes of action and species-specific effects of di-(2-
2 ethylhexl)phthalate in the liver. Critical Reviews in Toxicology. 36:459-479.
3 Sakurai T, Maeda S, Chang L, Karin M. 2006. Loss of hepatic NR-kappa B activity enhances chemical
4 hepatocarcinogenesis through sustained c-Jun N-terminal kinase 1 activation. Proc Natl Acad Sci USA 103 :
5 10544-10551.
6 Sanchez IM, Bull RJ. 1990. Early induction of reparative hyperplasia in the liver of B6C3F1 mice treated with
7 dichloroacetate and trichloroacetate. Toxicology 64: 33-46.
8 Sanz N, Diez-Fernandez C, Cascales M. 1996. Variations of hepatocyte antioxidant and DNA ploidy in rats aged 2
9 to 8 months. Biochimica et Biophysica Acta 1315: 123-130.
10 Sargent LM, 1999. Nonrandom cytogenetic alterations in hepatocellular carcinoma from transgenic mice
11 overexpressing c-Myc and transforming growth factor-a in the liver. Am J Pathol 154: 1047-1055.
12 Scharf JG, Dombrowski F, Ramadori G. 2001. The IGF axis and hepatocarcinogenesis. Mol Pathol 54:138-144.
13 SchmuckerDL. 2001. Liver Function and Phase I Drug Metabolism in the Elderly, a paradox. Drugs Aging
14 18(11): 837-851.
15 Schmucker DL. 2005. Age-related changes in liver structure and function: implications for disease?. Experimental
16 Gerontology 49: 650-659.
17 SchmuckerDL, Mooney JS, Jones AL. 1977. Age-related changes in the hepatic endoplasmic reticulum: a
18 quantitative analysis. Science 197:1005-1008.
19 Schmucker DL, Mooney JS, Jones AL. 1978. Stereological analysis of hepatic fine structure in the Fischer 344 rat.
20 Influence of sublobular location and animals age. J Cell Biol 78: 319-337.
21 Schwartz-Arad D, Zajicek G, Barrfeld E. 1989. The streaming liver IV: DNA content of the hepatocyte increases
22 with its age. Liver 9: 93-99.
23 Scott CS, Chiu WA. 2006. Trichloroethylene cancer epidemiology: A consideration of select issues. Environ Health
24 Perspect 114:1471-1478.
25 Schulte-Hermann R, Timmermann-Trosiener I, SchupplerJ. 1986. Facilitated expression of adaptive responses to
26 Phenobarbital in putative pre-stages of liver cancer. Carcinogenesis 7(10): 1651-1655.
27 Schultz IR, Merdink JL, Gonzalez-Leon A, Bull RJ. 2002. Dichloroacetate toxicokinetics and disruption of tyrosine
28 catabolism in B6C3F1 mice: dose-response relationships and age as a modifying factor. Toxicology 173:229-247
29 Seigel JH, Kasmin FE. 1997. Biliary tract diseases in the elderly: management and outcomes. Gut 41: 433-435.
30 Seitz CS, Lin Q, Deng H, Khavari PA. 1998. Alterations in NF-KB function in transgenic epithelial tissue
31 demonstrate a growth inhibitory role for NF-KB. Proc Natl Acad Sci USA 95: 2307-2312.
32 Seitz HK, Stickel F. 2006. Risk factors and mechanisms of hepatocarcinogenesis with special emphasis on alcohol
33 and oxidative stress. Biol Chem 387(4): 349-60.
34 Seng JE, Agrawal N, Horsley ETM, Leakey TI, Scherer EM, Xia S, Allaben WT, Leakey JEA. 2003. Toxicokinetics
35 of chloral hydrate in ad libitum-fed, dietary controlled and calorically restricted male B6C3F1 mice following short-
36 term exposure. Toxicol and Appl Pharmacol 193(2): 281-292.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-430 DRAFT—DO NOT CITE OR QUOTE
-------
1 Shah YM, Morimura K, Yang Q, Tanabe T, Takagi M, Gonzalez FJ. 2007. Peroxisome proliferator-activated
2 receptor a regulates a microRNA-mediated signaling cascade responsible for Hepatocellular proliferation.
3 Molecular and cellular biology 27(12):4238-4247.
4 Shankar K, Vaidya VS, Gorton JJ, Bucci TJ, Liu J, Waalkes MP, Mehendale HM. 2003. Activation of PPAR-a in
5 streptozotocin-induced diabetes is essential for resistance against acetaminophen toxicity. FASEB J.
6 17(12): 1748-50
7 Shearer BG, Hoekstra. 2003. Recent advances in peroxisome proliferator-activated receptor science. Current
8 Medicinal Chemistry 10:267-280.
9 Sher T, Yi HF, McBride OW, Gonzalez FJ. 1993. cDNA cloning chromosomal mapping, and functional
10 characterization of the human peroxisome proliferator activated receptor. Biochemistry 32: 5598-5604.
11 Sigal SH, Rajvanshi P, Gorla GR, Sokhi RP, Saxena R, Gebhard DR, Reid LM, Gupta S. 1999. Partial
12 hepatectomy-induced polyploidy attenuates hepatocyte replication and activates cell aging events. Am J Physiol 176
13 (Gastrointest. Liver Physiol. 39): G1260-G1272.
14 SnyderRD, Pullman J, Carter JH, Carter HW, DeAngelo AB. 1995. In vivo administration of dichloroacetic acid
15 suppresses spontaneous apoptosis in muring hepatocytes. Cancer Research 55: 3702-3705.
16 Soni MG, Mangipudy RS, Mumtax MM, Mehendale HM. 1998. Tissue repair response as a function of dose
17 during trichloroethylene hepatoxicity. lexicological Sciences 42: 158-165.
18 Soni MB, Ramaiah SK, Mumtaz MM, Clewell H, Mehendale HM. 1999. Toxicant-inflicted injury and stimulated
19 tissue repair are opposing toxicodynamic forces in predictive toxicology. Regulatory Toxicology and
20 Pharmacology. 29: 165-174.
21 Soresi M, Giannitrapani L, D'Antona F, Florena AM, La Spada E, Terranova A, Cervello M, D'Alessandro N,
22 Montalto G. 2006. Interleukin-6 and its soluble receptor in patients with liver cirrhosis and hepatocellular
23 carcinoma. World J Gastroenterol 12(16): 2563-2568.
24 Stanley LA, Blackburn DR, Devereaux S, Foley J, Lord PG, Maronpot RR, Orton TC, Anderson MW. 1994. Ras
25 mutations in methylclofenapate-induced B6C3F1 and C57BL/10J mouse liver tumors. Carcinogenesis 15(6):
26 1125-1131.
27 Stauber AJ, Bull RJ. 1997. Differences in phenotype and cell replicative behavior of hepatic tumors induced by
28 dichloroacetate (DCA) and trichloroacetate (TCA). Toxicology and Applied pharmacology 144: 235-246.
29 Stauber AJ, Bull RJ, Thrall BD. 1998. Dichloroacetate and trichloroacetate promote clonal expansion of
30 anchorage-independent hepatocytes in vivo and in vitro. Toxicology and Applied Pharmacology 150: 287-294.
31 Steup DR, Wiersma D, McMillan DA, Sipes IG. 1991. Pretreatment with drinking water solutions containing
32 trichloroethylene or chloroform enhances the hepatotoxicity of carbon tetrachloride in Fischer 344 rats. Fundam
33 Appl Toxicol. 16(4):798-809.
34 Steup DR, Hall P, McMillan DA, Sipes IG. 1993. Time course of hepatic injury and recovery following
3 5 coadministration of carbon tetrachloride and trichloroethylene in Fischer-344 rats. Toxicol Pathol 21:327-334.
36 Styles JA, Wyatt I, Coutts C. 1991. Trichloroacetic acid: studies on uptake and effects on hepatic DNA and liver
37 growth in mouse. Carcinogenesis. 12 (9): 1715-1719.
38 Su Q, Bannasch P. 2003. Relevance of hepatic preneoplasia for human hepatocarcinogenesis. Toxicol Pathol. 31:
39 126-133.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-431 DRAFT—DO NOT CITE OR QUOTE
-------
1 Szyf M. 2007. The dynamic epigenome and its implications in toxicology. Toxicological Sciences. 100(1): 7-23.
2 Takubo K, Kaminishi M. 2001. Diseases of the digestive tract and telomere lengths: significance and problems of
3 telomere measurements. Nippon Shokakibyo GakkaiZasshi 98: 144-150.
4 Takubo K, Nakamura K, Izumiyama N, Furugori E, Sawabe M, Arai T, Esaki Y, Mafane K, Kammori M, Fujiwara
5 M, Kato M, Oshimura M, Sasajima K. 2000. Telomere shortening with aging in human liver. J Gerontol A: Biol
6 MedSci 55: 533-536.
7 Takeda S, Kondo M, Kumada T, Koshikawa T, Ueda R, Nishio M, et al. 1996. Allelic-expression imbalance of the
8 insulin-like growth factor 2 gene in hepatocellular carcinoma and underlying disease. Oncogene 2(7): 1589-1592.
9 Tanaka K, Smithy PF, Stromberg PC, Eydelloth RS, Herold EG, Grossman S J, Frank JD, Hertzog PR, Soper KA,
10 Keenan KP. 1992. Studies of early hepatocellular proliferation and peroxisomal proliferation in Sprague-Dawley
11 rats treated with tumorigenic doses of clofibrate. Toxicology and Applied Pharmacology 116:71-77.
12 Tanguay RM, Jorquera R, Poudrier J, St-Louis M. 1996. Tyrosine and its catabolites: from disease to cancer. Acta
13 Biochem Pol 43:209-216.
14 Tao L, Kramer PM, Ge R, Pereira MA. 1998. Effect of dichloroacetic acid and trichloroacetic acid on DNA
15 memylation in liver and tumors of female B6C3F1 mice. Toxicological Sciences 43: 139-144.
16 Tao L, Yan S, Xie M, Kramer PM, Pereira MA. 2000. Effect of trichloroethylene and its metabolites,
17 dichloroacetic acid and trichloroacetic acid, on the methylation and expression of c-Jun and c-Myc protooncogenes
18 in mouse liver: prevention by methionine. Toxicological Sciences 54: 399-407.
19 Tao L, Li Y, Kramer PM, Wang W, Pereira MA. 2004. Hypomethy lation of DNA and the insulin-like growth
20 factor-II gene in dichloroacetic and trichloroacetic acid-promoted mouse liver tumors. Toxicology 196: 127-136.
21 Tarsetti F, Lenzi R, Salvi R, Schuler E, Rijhsinghani K, Lenzen R, Tavoloni N. 1993. Liver carcinogenesis
22 associated with feeding of ethionine in a choline-free diet: evidence against the role of oval cells in the emergence of
23 hepatocellular carcinoma. Hepatology 18(3): 596-603.
24 Thai SF, Allen JW, DeAngelo AB, George MH, Fuscoe JC. 2001. Detection of early gene expression changes by
25 differential display in the livers of mice exposed to dichloroacetic acid. Carcinogenesis. 22:1317-1322.
26 Thai SF, Allen JW, DeAngelo AB, George MH, Fuscoe JC. 2003. Altered gene expression in mouse livers after
27 dichloroacetic acid exposure. Mutat Res 543:167-180.
28 Tharappel JC, Cunningham ML Spear BT, Glauert HP. 2001. Differential activation of hepatic NF-icB in rats and
29 hamsters by the peroxisome proliferators Wy-14,643, Gemfibrozil, and dibutyl phthalate. Toxicol Sci 62: 20-27.
30 Thrall KD, Poet TS. 2000. Determination of biokinetic interactions in chemical mixtures using real-time breath
31 analysis and physiologically based pharmacokinetic modeling. J Toxicol Environ Health 59:653-70.
32 Thorgeirsson SS. 2006. See "a tumor progression model for hepatocellular carcinoma; bioinformatic analysis of
33 genomic data" by Poon TCW, Wong N, Lai PBS, Rattray M, Johnson PJ, Sung JJY on pager 1262. Editorial
34 comment. Gastroenterology 131(4): 1344-1346.
35 ToraasonM, Clark J, DankovicD, MathiasP, Skaggs S, Walker C, WerrenD. 1999. Oxidative stress and DNA
36 damage in Fischer rats following acute exposure to trichloroethylene or perchloroethylene. Toxicology 138: 43-53.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-432 DRAFT—DO NOT CITE OR QUOTE
-------
1 Tryndyak VP, Muskhelishvili L, Kovalchuk O, Rodriguez-Juarez R, Montgomery B, Churchwell MI, Ross SA,
2 Beland FA, Pogribny IP. 2006. Effect of long-term tamoxifen exposure on genotoxic and epigenetic changes in rat
3 liver; implications for tamoxifen-induced hepatocarcinogenesis. Carcinogenesis 27(8) 1713-1720.
4 Tucker AN, Sanders VM, Barnes DW, Bradshaw TJ, White KL, Sain LE, Borzelleca JF, Munson AE. 1982.
5 Toxicology of trichloroethylene in the mouse. Toxicology and Applied Pharmacology 62: 351-357.
6 Tzeng HF, Blackburn AC, Board PG, Anders MW. 2000. Polymorphism- and species-dependent inactivation of
7 glutathione transferase zeta by dichloroacetate. Chem Res Toxicol 13(4):231-236.
8 UlrichRG. 2003. The toxicogenomics of nuclear receptor agonists. Current Opinion in Chemical Biology
9 7:505-510.
10 Ulrich RG, Rockett JC, Gibson GG, Pettit SD. 2004. Overview of an interlaboratory collaboration on evaluating the
11 effects of model hepatotoxicants on hepatic gene expression. Environ Health Perspect 112: 423-427.
12 Urabe Y Nouso K, Higashi T, Nakatsukasa H, Hino N, Ashida K, Kinugasa N, Yoshida K, Uematsu S, Tsuji T.
13 1996. Telomere length in human liver diseases. Liver 16: 293-297.
14 U.S. Environmental Protection Agency. 1997. Cumulative risk assessment guidance: Phase I Planning and Scoping.
15 Memorandum from Carol M. Browner, Administrator, and Fred Hansen, Deputy Administrator, U.S. Environmental
16 Protection Agency, Washington, DC.
17 U.S. EPA (Environmental Protection Agency). 2001. Trichloroethylene Health Risk Assessment: Synthesis and
18 Characterization. U.S. Environmental Protection Agency. Office of Research and Development. External Review
19 Draft. Available: http://oaspub.epa.gov/eims/eimscomm.getfile?p_dowload_id=4580 [accessed 3/3/06].
20 U.S. EPA (Environmental Protection Agency). (2005) Guidelines for Carcinogen Risk Assessment. EPA/630/P-
21 03/001F. Washington DC: Risk Assessment Forum, U.S. Environmental Protection Agency. Available:
22 http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=439797 [Accessed 8/05].
23 U.S. Environmental Protection Agency. 2006. National Air Toxics Assessment. Available at:
24 http://www.epa.gov/ttn/atw/natal999/.
25 Vainio H, Nickels J, Linnainmaa K. 1982. Phenoxy acid herbicides cause peroxisome proliferation in Chinese
26 hamsters. ScandJ work environ health 8:70-73.
27 Vainio H, Linnainmaa K, Kahonen M, Nickels J, Hietanen E, Marniemi J, Peltonen P. 1983. Hypolipidemia and
28 peroxisome proliferation induced by phenoxyacetic acid and herbicides in rats. Biochemical Pharmacology 32(18):
29 2775-2779.
30 Van Duuren BL, Goldschmidt BM, Loewengart G, Smith AC, Meichionne S, Seidman I, Roth D. 1979.
31 Carcinogenicity of halogenated olefinic and aliphatic hydrocarbons in mice. JNCI63(6): 1433-1439.
32 Veeramachaneni DN, Palmer JS, AmannRP. 2001. Long-term effects on male reproduction of early exposure to
33 common chemical contaminants in drinking water. HumReprod. 16(5):979-87.
34 Vickers AEM, Lucier GW. 1996. Estrogen receptor levels and occupancy in hepatic sinusoidal endothelial and
3 5 Kupffer cells are enhanced by initiation with diethylnitroamine and promotion with 17 a-ethinylestradiol in rats.
36 Carcinogenesis 17(6): 1235-1242.
37 Vielhauer V, Sarafoff M, Gais P, Rabes HM. 2001. Cell type-specific induction of O6-alkylguanine-DNA
3 8 alkyltransferase mRNA expression in rat liver during regeneration, inflammation and preneoplasia. J Cancer Res
39 ClinOncol 127: 591-602.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-433 DRAFT—DO NOT CITE OR QUOTE
-------
1 Vig P, Russo FP, Edwards RJ, Tadrous PJ, Wright NA, Thomas HC, Alison MR, Forbes SJ. 2006. The source of
2 parenchymal regeneration after chronic hepatocellular liver injury in mice. Hepatology 43(2): 316-323.
3 Vogelstein G, Kinzler KW. 2004. Cancer genes and the pathways they control. Nature Medicine 10(8): 789-799.
4 Vollmar B, Pradarutti S, Richter S, Menger MD. 2002. In vivo quantification of ageing changes in rat liver from
5 early juvenile to senescent life. Liver 22: 330-341.
6 Voss C, Zerban H, Bannasch P, Berger MR. 2005. Lifelong exposure to di-(2-ethylhexyl)-phthalated induced
7 tumors in liver and testes of Sprague-Dawley rats. Toxicology 2006: 359-371.
8 Voss KA, Riley R, Dunn C, Gorton JC. 2006. The role of tumor necrosis factor alpha and the peroxisome
9 proliferator-activated receptor alpha in modulating the effects of fumonisin in mouse liver. Toxicology. 222:165-74.
10 WadaN, MarsmanDS, Popp JA. 1992. Dose-related effects of the hepatocarcinogen, WY-14,643, on peroxisomes
11 and cell replication. Fundamental and Applied Toxicology 18: 149-154.
12 Wang G, Stacey NH. 1990. Elevation of individual serum bile acids on exposure to trichloroethylene or a-
13 naphthylisothioncyanate. Toxicol and Appl Pharmacol 105: 209-215.
14 Wang RS, Nakajima T, Honma T. 1999. Trichloroethylene inhibits aldehyde dehydrogenase only for aliphatic
15 aldehydes of short chains in rats. Toxicology 132:9-18.
16 Wanson JC, Bernaert D, Penasse W, Mosselmans R, Bannasch P. 1980. Separation in distinct subpopulations by
17 Elutriation of liver cells following exposure of rats to N-Nitrosomorpholine. Cancer Research 40: 459-471.
18 Wantanabe T, Shimada H, Tanaka Y. 1978. Human hepatocytes and aging: a cytophotometrical analysis in 35
19 sudden-death cases. Vichows Arch B Cell Pathol. 27(4): 307-316.
20 Watson RE, Goodman JI. 2002. Effects of Phenobarbital on DNA methylation in GC-rich regions of hepatic DNA
21 from mice that exhibit different levels of susceptibility to liver tumorigenesis. Toxicol Sci 68:51-58.
22 Waxman S, Wurmbach E. 2007. De-regulation of common housekeeping genes in hepatocellular carcinoma. BMC
23 genomics 8:243 doi:10.1186/1471-2164-8-243.
24 Weibel ER, Staubli W, Gnage HR, Hess F. 1964. Correlated morphometric and biochemical studies on the liver
25 cell. I. Morphometric model, stereological methods, and normal morphometric data for rat liver. J Cell Biol. 21:
26 68-91.
27 Weidman JR, Dolinoy DC, Murphy SK, Jirtle RL. 2007. Cancer susceptibility: epigenetic manifestation of
28 environmental exposures. The Cancer Journal 13(1): 9-16.
29 Weinberg HS, Krasner SW, Richardson SD, Thruston AD. 2002. The occurrence of disinfection by-products (DBFs)
30 of health concern in drinking water. Results of a nationwide DBF occurrence study. U.S. Environmental Protection
31 Agency. EPA/600/R-02/068.
32 Weisel CP, Kim H, Haltmeier P, Klotz JB. 1999. Exposure estimates to disinfection by-products of chlorinated
33 drinking water. Environ Health Perspect 107:103-110.
34 Weiss PA. 1971. A cell is not an island entire of itself. Perspect Biol Med 14: 182-205.
3 5 Werner H, Le Roith D. 2000. New concepts in regulation and function of the insulin-like growth factors:
36 implications for understanding normal growth and neoplasia. Cell Mol Life Sci 57:932-942.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-434 DRAFT—DO NOT CITE OR QUOTE
-------
1 Wheeler MD, Smutney OM, Check JF, Rusyn I, Schulte-Hermann R, Thurman RG. 2003. Impaired Ras membrane
2 association and activation in PPARa knockout mice after partial hepatectomy. Am J Physiol Gastointest Liver
3 Physiol. 284: G302-G312.
4 Wideroff L, Gridley G, Mellemkjaer L, Chow WH, Linet M, Keehn S, et al. 1997. Cancer incidence in a population-
5 based cohort of patients hospitalized with diabetes mellitus in Denmark. JNatl Cancer Inst 89:1360-1365.
6 Wilkens L, Flemming P, Gebel M, Bleck J, Terkamp C, Wingen L, Kreipe H, Schlegelberger B, 2004. Induction of
7 aneuploidy by increasing chromosomal instability during dedifferentiation of hepatocellular carcinoma. Proc Natl
8 AcadSciUSA 101: 1309-1314.
9 Wood LD, Parsons DW, Jones S, et al. 2007. The genomic landscapes of human breast and colorectal cancers.
10 Science 318: 1108-1113.
11 Woodruff TJ, Axelrad DA, Caldwell J, Morello-Frosch R, Rosenbauam A. 1998. Public health implications of 1990
12 air toxics concentrations across the United States. Environ Health Perspect 106:245-251.
13 Woods CG, Burns AM, Maki A, Bradford BU, Cunningham ML, Connor HD, Kadiiska MB, Mason RP, Peters JM,
14 Rusyn I. 2006. Sustained formation of alpha-(4-pyridyl-l-oxide)-N-tert-butylnitrone radical adducts in mouse liver
15 by peroxisome proliferators is dependent upon peroxisome proliferator activated receptor-alpha, but not NADPH
16 oxidase. Free Radic Biol Med. 42(3):335-42.
17 Woods CG, Vanden Heuvel JP, Rusyn I. 2007a. Genomic profiling in nuclear receptor-mediated toxicity.
18 Toxicologic Pathol 35:474-494.
19 Woods CG, Burns AM, Bradford BU, Ross PK, Kosyk O, Swenburg JA, Cunningham ML, Rusyn I. 2007b. Wy-
20 14,643-induced cell proliferation and oxidative stress in mouse liver are independent of NADPH oxidase.
21 Toxicological Sciences 98(2): 366-374.
22 Woods CG, Burns AM, Maki A, Bradford BU, Cunningham ML, Connor HD, Kadiiska MB, Mason RP, Peters JM,
23 Rusyn I. 2007c. Sustained formation of a-(4-pyridy 1-1-oxide) -N-tert-butylnitrone radical adducts in mouse liver
24 by peroxisome proliferators is dependent upon perosixome proliferator-activated receptor a, but not NADPH
25 oxidase. Free Radical Biology & Medicine 42L 335-342.
26 Woolhiser MR, Krieger SM, Thomas J, Hotchkiss JA. 2006. Report from the Dow Chemical Company to ATSDR
27 and EPA, Study ID # 031020.Trichloroethylene (TCE): immunotoxicity potential in CD rats following a 4-week
28 vapor inhalation exposure.
29 WuXZ, ChenD. 2006. Originof hepatocellular carcinoma: role of stem cells. J GastroenterolHepatol21(7):
30 1093-8.
31 Wu C, Schaum J. 2000. Exposure assessment of trichloroethylene. Environ Health Perspect 108(Suppl 2):359-363.
32 Yang Q, Ito S, Gonzalez FJ. 2007a. Hepatocyte-restricted constitutive activation of PPARi induced
33 hepatoproliferationbut not hepatocarcinogenesis. Carcinogenesis. 28(6): 1171-1177.
34 Yang Q, Nagano T, Shah Y, Cheung C, Ito S, Gonzalez FJ. 2007b. The PPARa-humanized mouse: a model to
3 5 investigate the species differences in liver toxicity mediated by PPARa. Toxicol Sciences. Published on line
36 August 9, 2007.
37 Yeh MM, Larson AM, Campbell JS, Fausto N, Rulyak SJ, Swanson PE. 2007. The expression of transforming
3 8 growth factor -a in cirrhosis, dysplastic nodules, and hepatocellular carcinoma; an immunohistochemical study in
39 70 cases. Am JSurg Pathol 31(5): 681-689.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-43 5 DRAFT—DO NOT CITE OR QUOTE
-------
1 Yeldani AV, Milano M, Subbarao V, Reddy JK, Rao MS. 1989. Evaluation of liver cell proliferation during
2 ciprofibrate-induced hepatocarcinogenesis. Cancer Letters 47: 21-27.
3 Zajicek G, Schwartz-Arad D, Barrfeld E. 1989. The streaming liver. V: Time and age-dependent changes of
4 hepatocy te DNA content, following partial hepatectomy .Liver9:164-171.
5 Zajicek G, Schwartz-Arad D. 1990. Streaming liver VII: DNA turnover in acinus zone-3. Liver 10:137-140.
6 Zajicek G, Arber N, Schwartz-Arad. 1991. Streaming liver VIII: Cell production rates following partial
7 hepatectomy. Liver 11: 347-351.
8 Zeppa P, Benincasa G, Troncone G, Lucariello A, Zabatta A, Cochand-Priollet B, Fulciniti F, Vetrani A, De Rosa G,
9 Palombini L. 1998. Retrospective evaluation of DNA ploidy of hepatocarcinoma on cytological samples.
10 Diagnostic Cytopathology 19(5): 323-329.
11 Zhang B, Pan Z, Cobb GP, Anderson TA. 2007. microRNAs as oncogenes and tumor suppressors. Developmental
12 Biology 302: 1-12.
13 Zhou H, Randers-Pehrson G, Geard CR, Brenner D J, Hall EJ, Hei TK. 2003. Interaction between radiation-induced
14 adaptive response and bystander mutagenesis in mammalian cells. RadiatRes 160: 512-516.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 E-436 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX F
TCE Noncancer Dose-Response Analyses
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix F: TCE Noncancer Dose-Response Analyses
LIST OF TABLES F-iv
LIST OF FIGURES F-v
APPENDIX F: TCE NONCANCER DOSE-RESPONSE ANALYSES F-l
F.I. DATA SOURCES F-l
F.2. DOSIMETRY F-l
F.2.1. Estimates of Trichlorethylene (TCE) in Air From Urinary Metabolite
Data Using Ikedaetal. (1972) F-l
F.2.1.1. Results for Chiaetal. (1996) F-l
F.2.1.2. Results for Mhirietal. (2004) F-4
F.2.2. Dose Adjustments to Applied Doses for Intermittent Exposure F-4
F.2.3. Physiologically Based Pharmacokinetic (PBPK) Model-Based Internal
Dose Metrics F-5
F.3. DOSE-RESPONSE MODELING PROCEDURES F-5
F.3.1. Models for Dichotomous Response Data F-5
F.3.1.1. Quantal Models F-5
F.3.1.2. Nested Dichotomous Models F-6
F.3.2. Models for Continuous Response Data F-6
F.3.3. Model Selection F-6
F.3.4. Additional Adjustments for Selected Data Sets F-7
F.4. DOSE-RESPONSE MODELING RESULTS F-8
F.4.1. Quantal Dichotomous and Continuous Modeling Results F-8
F.4.2. Nested Dichotomous Modeling Results F-8
F.4.2.1. Johnson etal. (2003) Fetal Cardiac Defects F-8
F.4.2.2. Narotsky etal. (1995) F-12
F.4.3. Model Selections and Results F-20
F.5. DERIVATION OF POINTS OF DEPARTURE F-27
F.5.1. Applied Dose Points of Departure F-27
F.5.2. Physiologically Based Pharmacokinetic (PBPK) Model-Based Human
Points of Departure F-27
F.6. SUMMARY OF POINTS OF DEPARTURE (PODs) FOR CRITICAL
STUDIES AND EFFECTS SUPPORTING THE INHALATION
REFERENCE CONCENTRATION (RfC) AND ORAL REFERENCE
DOSE(RfD) F-28
F.6.1. National Toxicology Program (NTP, 1988)—Benchmark Dose (BMD)
Modeling of Toxic Nephropathy in Rats F-28
F.6.1.1. Dosimetry and Benchmark Dose (BMD) Modeling F-29
F.6.1.2. Derivation of HEC99 and HED99 F-29
F.6.2. National Cancer Institute (NCI, 1976)—Lowest-Observed-Adverse-
Effect Level (LOAEL) for Toxic Nephrosis in Mice F-29
F.6.2.1. Dosimetry F-31
F.6.2.2. Derivation of HEC99 and HED99 F-31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-ii DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS (continued)
F.6.3. Woolhiser et al. (2006)—Benchmark Dose (HMD) Modeling of
Increased Kidney Weight in Rats F-32
F.6.3.1. Dosimetry and Benchmark Dose (BMD) Modeling F-32
F.6.3.2. Derivation of HEC99 and HED99 F-34
F.6.4. Keil et al. (2009)—Lowest-Observed-Adverse-Effect Level (LOAEL)
for Decreased Thymus Weight and Increased Anti-dsDNA and
Anti-ssDNA Antibodies in Mice F-34
F.6.5. Keil et al. (2009)—Lowest-Observed-Adverse-Effect Level (LOAEL)
for Decreased Thymus Weight and Increased Anti-dsDNA and
Anti-ssDNA Antibodies in Mice F-34
F.6.5.1. Dosimetry F-34
F.6.5.2. Derivation of HEC99 and HED99 F-35
F.6.6. Johnson et al. (2003)—Benchmark Dose (BMD) Modeling of Fetal
Heart Malformations in Rats F-35
F.6.6.1. Dosimetry and Benchmark Dose (BMD) Modeling F-36
F.6.6.2. Derivation of HEC99 and HED99 F-36
F.6.7. Peden-Adams et al. (2006)—Lowest-Observed-Adverse-Effect Level
(LOAEL) for Decreased PFC Response and Increased Delayed-Type
Hypersensitivity in Mice F-37
F.7. REFERENCES F-38
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-iii DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF TABLES
F-l. Dose-response data from Chia et al. (1996) F-l
F-2. Data on TCE in air (ppm) and urinary metabolite concentrations in workers
reported by Ikedaetal. (1972) F-2
F-3. Estimated urinary metabolite and TCE air concentrations in dose groups from
Chia etal. (1996) F-4
F-4. Data on fetuses and litters with abnormal hearts from Johnson et al. (2003) F-9
F-5. Comparison of observed and predicted numbers of fetuses with abnormal hearts
from Johnson et al. (2003), with and without the high-dose group, using a nested
model F-9
F-6. Results of nested log-logistic model for fetal cardiac anomalies from Johnson
et al. (2003) without the high-dose group, on the basis of applied dose
(mg/kg/d in drinking water) F-10
F-7. Results of nested log-logistic model for fetal cardiac anomalies from Johnson
et al. (2003) without the high-dose group, using the TotOxMetabBW34 dose
metric F-l 3
F-8. Results of nested log-logistic model for fetal cardiac anomalies from Johnson
et al. (2003) without the high-dose group, using the AUCCBld dose metric F-14
F-9. Analysis of LSCs with respect to dose fromNarotsky et al. (1995) F-15
F-10. Results of nested log-logistic and Rai-VanRyzin model for fetal eye defects
from Narotsky et al. (1995), on the basis of applied dose (mg/kg/d in drinking
water) F-l 6
F-l 1. Comparison of results of nested log-logistic (without LSC or 1C) and quantal
log-logistic model for fetal eye defects fromNarotsky et al. (1995) F-18
F-12. Results of nested log-logistic and Rai-VanRyzin model for prenatal loss from
Narotsky et al. (1995), on the basis of applied dose (mg/kg/d in drinking water) F-20
F-l3. Model selections and results for noncancer dose-response analyses F-23
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-iv DRAFT—DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
F-l . Regression of TCE in air (ppm) and TCA in urine (mg/g creatinine) based on
data from Ikeda et al. (1972) [[[ F-3
F-2. BMD modeling of Johnson et al. (2003) using nested log-logistic model, with
applied dose, without LSC, with 1C, and without the high-dose group, using a
BMR of 0.05 extra risk (top panel) or 0.01 extra risk (bottom panel) ........................... F-ll
F-3. BMD modeling of Johnson et al. (2003) using nested log-logistic model, with
TotOxMetabBW34 dose metric, without LSC, with 1C, and without the high-dose
group, using a BMR of 0.01 extra risk [[[ F-l 3
F-4. BMD modeling of Johnson et al. (2003) using nested log-logistic model, with
AUCCBld dose metric, without LSC, with 1C, and without the high-dose group,
using a BMR of 0.01 extra risk [[[ F-14
F-5. BMD modeling of fetal eye defects from Narotsky et al. (1995) using nested
log-logistic model, with applied dose, with both LSC and 1C, using a BMR of
0.05 extra risk [[[ F-17
F-6. BMD modeling of fetal eye defects from Narotsky et al. (1995) using nested
log-logistic model, with applied dose, without either LSC or 1C, using a BMR of
0.05 extra risk [[[ F-18
F-7. BMD modeling of fetal eye defects from Narotsky et al. (1995) using nested Rai-
VanRyzin model, with applied dose, without either LSC or 1C, using a BMR of
0.05 extra risk [[[ F-19
F-8. BMD modeling of prenatal loss reported in Narotsky et al. (1995) using nested
log-logistic model, with applied dose, without LSC, with 1C, using a BMR of
0.05 extra risk (top panel) or 0.01 extra risk (bottom panel) ......................................... F-21
F-9. BMD modeling of prenatal loss reported in Narotsky et al. (1995) using nested
Rai-VanRyzin model, with applied dose, without LSC, with 1C, using a BMR
-------
LIST OF FIGURES (continued)
F-12. Derivation of HECgg and HED99 corresponding to the rodent idPOD from NTP
(1988) toxic nephrosis in mice [[[ F-32
F-13. BMD modeling of Woolhiser et al. (2006) for increased kidney weight in female
S-Drats [[[ F-33
F-14. Derivation of FtECgg and F£ED99 corresponding to the rodent idPOD from
Woolhiser et al. (2006) for increased kidney weight in rats .......................................... F-35
F-15. Derivation of FtECgg and FtEDgg corresponding to the rodent idPOD from Keil
et al. (2009) for decreased thymus weight and increased anti-dsDNA and
anti-ssDNA antibodies in mice [[[ F-36
F-16. Derivation of FtECgg and FtEDgg corresponding to the rodent idPOD from Johnson
et al. (2003) for increased fetal cardiac malformations in female S-D rats using the
-------
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
APPENDIX F: TCE NONCANCER DOSE-RESPONSE ANALYSES
F.I. DATA SOURCES
Data sources are cited in the body of this report in the section describing dose-response
analyses (see Chapter 5).
F.2. DOSIMETRY
This section describes some of the more detailed dosimetry calculations and adjustments
used in Section 5.1.
F.2.1. Estimates of Trichlorethylene (TCE) in Air From Urinary Metabolite Data Using
Ikeda et al. (1972)
F.2.1.1. Results for Chia et al (1996)
Chia et al. (1996) demonstrated a dose-related effect on hyperzoospermia in male
workers exposed to trichloroethylene (TCE), lumping subjects into four groups based on range of
trichloroacetic acid (TCA) in urine (see Table F-l).
Table F-l. Dose-response data from Chia et al. (1996)
TCA, mg per g creatinine
0.8to<25
50 to <75
75to<100
>100to 136.4
No. of subjects
37
18
8
5
No. with hyperzoospermia
6
8
4
3
Minimum and maximum TCA levels are reported in the text of Chia et al. (1996), the other data, in their
Table 5.
Data from Ikeda et al. (1972) were used to estimate the TCE exposure concentrations
corresponding to the urinary TCA levels reported by Chia et al. (1996). Ikeda et al. (1972)
studied 10 workshops, in each of which TCE vapor concentration was "relatively constant."
They measured atmospheric concentrations of TCE and concentrations in workers' urine of total
trichloro compounds (TTC), TCA, and creatinine, and demonstrated a linear relation between
TTC/creatinine (mg/g) in urine and TCE in the work atmosphere. Their data are tabulated as
geometric means (the last column was calculated by us, as described in Table F-2).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-1 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table F-2. Data on TCE in air (ppm) and urinary metabolite concentrations
in workers reported by Ikeda et al. (1972)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
n
9
5
6
4
4
5
5
5
4
4
TCE
(ppm)
3
5
10
25
40
45
50
60
120
175
TTC
(mg/L)
39.4
45.6
60.5
164.3
324.9
399
418.9
468
915.3
1210.9
TCA
(mg/L)
12.7
20.2
17.6
77.2
90.6
138.4
146.6
155.4
230.1
235.8
TTC (mg/g
creatinine)
40.8
42.4
47.3
122.9
221.2
337.7
275.8
359
518.9
1040.1
TCA (mg/g
creatinine)
13.15127
18.78246
13.76
57.74729
61.68273
117.137
96.52012
119.2064
130.4478
202.5399
These data were used to construct the last column "TCA.cr.mg.g" (mg TCA/g creatinine),
as follows: TCA (mg/g creatinine) = TCA (mg/L) x TTC (mg/g creatinine)/TTC (mg/L). The
regression relation between TCE (ppm) and TCA (mg/g creatinine) was evaluated using these
data. Ikeda et al. (1972) reported that the measured values are lognormally distributed and
exhibit heterogeneity of variance, and that the reported data (above) are geometric means. Thus,
the regression relation between loglO(TCA [mg/g creatinine]) and loglO(TCE [ppm]) was used,
assuming constant variances and using number of subjects "n" as weights. Figure F-l shows the
results.
Next, a Berkson setting for linear calibration was assumed, in which one wants to predict
X(TCE, ppm) from means for 7 (TCA, mg/g creatinine), with substantial error in 7(Snedecor
and Cochran, 1980). Thus, the inverse prediction for the data of Chia et al. (1996) was used to
infer their mean TCE exposures. The relation based on data from Ikeda et al. (1972) is
loglO(TCA, mg/g creatinine) = 0.7098 + 0.7218*loglO(TCE, ppm)
and the inverse prediction is
(Eq. F-l)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-2 DRAFT—DO NOT CITE OR QUOTE
-------
4
5
6
7
8
9
10
11
12
13
14
log10(TCA, mg/g.creatinine in urine) = 0.7098 + 0.7218 * log10(TCE, ppm)
D)
5
o
o -
8-
10
50
100
TCE, ppm
Coefficients:
Value Std. Error t value Pr(>|t|)
(Intercept) 0.7098 0.1132 6.2688 0.0002
loglO(TCE.ppm) 0.7218 0.0771 9.3578 0.0000
Residual standard error: 0.3206 on 8 degrees of freedom
Multiple R-Squared: 0.9163
F-statistic: 87.57 on 1 and 8 degrees of freedom, the p-value
is 0.0000139
Figure F-l. Regression of TCE in air (ppm) and TCA in urine (nig/g
creatinine) based on data from Ikeda et al. (1972).
loglO(TCE) = [loglO(TCA) - 0.7098]/0.7218
TCE, ppm = 10A( [loglO(TCA) - 0.7098]/0.7218)
(Eq. F-2)
Because of the lognormality of data reported by Ikeda et al. (1972), the means of the
logarithms of the ranges for TCA (mg/g creatinine) in Chia et al. (1996), which are estimates of
the median for the group, were used. The results are shown in Table F-3.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-3 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table F-3. Estimated urinary metabolite and TCE air concentrations in dose
groups from Chia et al. (1996)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
TCA, mg per g
Creatinine
0.8to<25
50 to <75
75to<100
>100to 136.4
Estim. TCA
median3
4.47
61.2
86.6
117
LoglO(TCA
median)
0.650515
1.787016
1.937531
2.067407
Estim. ppm
TCEb
0.827685
31.074370
50.226119
76.008668
a 10A(mean[loglO(TCA limits in first column)]).
b 10A([loglO(TCA median)] -0.7098)70.7218.
Dose-response relations for the data of Chia et al. (1996) were modeled using both the
estimated medians for TCA (mg/g creatinine) in urine and estimated TCE (ppm in air) as doses.
The TCE-TCA-TTC relations are linear up to about 75 ppm TCE (Figure 1 of Ikeda et al. 1972),
and certainly in the range of the benchmark dose (BMD). As noted below (see Section F.2.2),
the occupational exposure levels are further adjusted to equivalent continuous exposure for
deriving the point of departure (POD).
F.2.1.2. Results for Mhiri et al (2004)
The lowest-observed-adverse-effect level (LOAEL) group for abnormal trigeminal nerve
somatosensory evoked potential reported in Mhiri et al. (2004) had a urinary TCA concentration
of 32.6 mg TCA/mg creatinine. Using Eq. F-2, above gives an occupational exposure level =
10A([loglO(32.6) - 0.7098]/0.7218) = 12.97404 ppm. As noted below (see Section F.2.2), the
occupational exposure levels are further adjusted to equivalent continuous exposure for deriving
the POD.
F.2.2. Dose Adjustments to Applied Doses for Intermittent Exposure
The nominal applied dose was adjusted for exposure discontinuity (e.g., exposure for
5 days per week and 6 hours per day reduced the dose by the factor [5/7]*[6/24]). The
physiologically based pharmacokinetic (PBPK) dose metrics took into account the daily and
weekly discontinuity to produce an equivalent average dose for continuous exposure. No dose
adjustments were made for duration of exposure or a less-than-lifetime study, as is typically done
for cancer risk estimates, though in deriving the candidate reference values, an uncertainty factor
for subchronic-to-chronic exposure was applied where appropriate.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-4 DRAFT—DO NOT CITE OR QUOTE
-------
1 For human occupational studies, inhalation exposures (air concentrations) were adjusted
2 by the number of work (vs. nonwork) days and the amount of air intake during working hours as
3 a fraction of the entire day (10 m3 during work/20 m3 for entire day). For the TCE ppm in air
4 converted from urinary metabolite data using Ikeda et al. (1972), the work week was 6 days, so
5 the adjustment for number of work days is 6/7.
6
7 F.2.3. Physiologically Based Pharmacokinetic (PBPK) Model-Based Internal Dose Metrics
8 PBPK modeling was used to estimate levels of dose metrics corresponding to different
9 exposure scenarios in rodents and humans (see Section 3.5). The selection of dose metrics for
10 specific organs and endpoints is discussed under Section 5.1.
11 The PBPK model requires an average body weight. For most of the studies, averages
12 specific to each species, strain, and sex were used. Where these were not reported in the text of
13 an article, data were obtained by digitizing the body weight graphics (Maltoni et al., 1986) or by
14 finding the median of weekly averages from graphs (National Cancer Institute [NCI], 1976;
15 National Toxicology Program [NTP], 1990, 1988). Where necessary, default adult body weights
16 specific to the strain were used (U.S. EPA, 1994).
17
18 F.3. DOSE-RESPONSE MODELING PROCEDURES
19 Where adequate dose-response data were available, models were fitted with the
20 BenchMark Dose Software (BMDS) (http://www.epa.gov/ncea/bmds) using the applicable
21 applied doses or PBPK model-based dose metrics for each combination of study, species, strain,
22 sex, endpoints, and benchmark response (BMR) under consideration.
23
24 F.3.1. Models for Dichotomous Response Data
25 F.3.1.1. Quantal Models
26 For dichotomous responses, the log-logistic, multistage, and Weibull models were fitted.
27 These models adequately describe the dose-response relationship for the great majority of data
28 sets, specifically in past TCE studies (Filipsson and Victorin, 2003). If the slope parameter of
29 the log-logistic model was less than 1, indicating a supralinear dose-response shape, the model
30 with the slope constrained to 1 was also fitted for comparison. For the multistage model, an
31 order one less than the number of dose groups was used, in addition to the 2nd-order multistage
32 model if it differed from the preceding model, and the first-order ('linear') multistage model
33 (which is identical to a Weibull model with power parameter equal to 1). The Weibull model
34 with the power parameter unconstrained was also fitted t.
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-5 DRAFT—DO NOT CITE OR QUOTE
-------
1 F.3.1.2. Nested Dichotomous Models
2 In addition, nested dichotomous models were used for developmental effects in rodent
3 studies to account for possible litter effects, such maternal covariates or intralitter correlation.
4 The available nested models in BMDS are the nested log-logistic model, the Rai-VanRyzin
5 models, and the NCTR model. Candidates for litter-specific covariates (LSC) were identified
6 from the studies and considered legitimate for analysis if they were not significantly dose-related
7 (determined via regression, analysis of variance). The need for a LSC was indicated by a
8 difference of at least 3 in the Akaike Information Criteria (AIC) for models with and without a
9 LSC. The need to estimate intralitter correlations (1C) was determined by presence of a high
10 correlation coefficient for at least one dose group and by AIC. The fits for nested models were
11 also compared with the results from quantal models.
12
13 F.3.2. Models for Continuous Response Data
14 For continuous responses, the distinct models available in BMDS were fitted: power
15 model (power parameter unconstrained and constrained to >1), polynomial model, and Hill
16 model. Both constant variance and modeled variance models were fit; but constant variance
17 models were used for model parsimony unless the/>-value for the test of homogenous variance
18 was <0.10, in which case the modeled variance models were considered. For the polynomial
19 model, model order was selected as follows. A model of order 1 was fitted first. The next higher
20 order model (up to order «-l) was accepted if AIC decreased more than 3 units and the/>-value
21 for the mean did not decrease.
22
23 F.3.3. Model Selection
24 After fitting these models to the data sets, the recommendations for model selection set
25 out in U.S. Environmental Protection Agency (U.S. EPA)'s Benchmark Dose Technical
26 Guidance Document (Inter-Agency Review Draft, U.S. EPA, 2008b) were applied. First, models
27 were generally rejected if the/>-value for goodness of fit was <0.10. In a few cases in which
28 none of the models fit the data with/? > 0.10, linear models were selected on the basis of an
29 adequate visual fit overall. Second, models were rejected if they did not appear to adequately fit
30 the low-dose region of the dose-response relationship, based on an examination of graphical
31 displays of the data and scaled residuals. If the benchmark dose lower bound (BMDL) estimates
32 from the remaining models were "sufficiently close" (a criterion of within 2-fold for "sufficiently
33 close" was used), then the model with the lowest AIC was selected. The AIC is a measure of
34 information loss from a dose-response model that can be used to compare a set of models.
35 Among a specified set of models, the model with the lowest AIC is considered the "best." If two
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-6 DRAFT—DO NOT CITE OR QUOTE
-------
1 or more models share the lowest AIC, the BMD Technical Guidance Document (U.S. EPA,
2 2008b) suggests that an average of the BMDLs could be used, but averaging was not used in this
3 assessment (for the one occasion in which models shared the lowest AIC, a selection was made
4 based on visual fit). If the BMDL estimates from the remaining models are not sufficiently
5 close, some model dependence is assumed. With no clear biological or statistical basis to choose
6 among them, the lowest BMDL was chosen as a reasonable conservative estimate, as suggested
7 in the Benchmark Dose Technical Guidance Document., unless the lowest BMDL appeared to be
8 an outlier, in which case further judgments were made.
9
10 F.3.4. Additional Adjustments for Selected Data Sets
11 In a few cases, the dose-response data necessitated further adjustments in order to
12 improve model fits.
13 The behavioral/neurological endpoint "number of rears" from Moser et al. (1995)
14 consisted of counts, measured at five doses and four measurement times (with eight observations
15 each). The high dose for this endpoint was dropped because the mean was zero, and no
16 monotone model could fit that well. Analysis of means and standard deviations for these counts
17 suggested a Box-Cox power transform (Box et al., 1978) of !/2 (i.e., square root) to stabilize
18 variances (i.e., the slope of the regression of log [standard deviation (SD)] on log[mean] was
19 0.46, and the relation was linear and highly significant). This information was helpful in
20 selecting a suitable variance model with high confidence (i.e., variance constant, for square-root
21 transformed data). Thus, the square root was taken of the original individual count data, and the
22 mean and variance of the transformed count data were used in the BMD modeling.
23 The high-dose group was dropped due to supra-linear dose-response shapes in two cases:
24 fetal cardiac malformations from Johnson et al. (2003) and decreased PFC response from
25 Woolhiser et al. (2006). Johnson et al. (2003) is discussed in more detail below (see
26 Section F.4.2.1). For Woolhiser et al. (2006), model fit near the BMD and the lower doses as
27 well as the model fit to the variance were improved by dropping the highest dose (a procedure
28 suggested in U.S. EPA (2008b).
29 In some cases, the supralinear dose-response shape could not be accommodated by these
30 measures, and a LOAEL or no-observed-adverse-effect level (NOAEL) was used instead. These
31 include NCI (1976) (toxic nephrosis, >90% response at lowest dose), Keil et al. (2009)
32 (autoimmune markers and decreased thymus weight, only two dose groups in addition to
33 controls), and Peden-Adams et al. (2006) (developmental immunotoxicity, only two dose groups
34 in addition to controls).
35
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-7 DRAFT—DO NOT CITE OR QUOTE
-------
1 F.4. DOSE-RESPONSE MODELING RESULTS
2 F.4.1. Quantal Dichotomous and Continuous Modeling Results
3 The documents Appendix.linked.files\AppF.Non-cancer.Plots.TCE.contin.DRAFT.pdf
4 and Appendix.linked.files\AppF.Non-cancer.Plots.TCE.dichot.DRAFT.pdf show the fitted
5 model curves. The graphics include observations (group means or proportions), the estimated
6 model curve (solid red line) and estimated BMD, with a BMDL. Vertical bars show 95%
7 confidence intervals for the observed means. Printed above each plot are some key statistics
8 (necessarily rounded) for model goodness of fit and estimated parameters. Printed in the plots in
9 the upper left are the BMD and BMDL for the rodent data, in the same units as the rodent dose.
10 More detailed results, including alternative BMRs, alternative dose metrics, quantal
11 analyses for endpoints for which nested analyses were performed, etc. are documented in the
12 several spreadsheets. Input data for the analyses are in the following documents:
13 Appendix.linked.files\AppF.Non-cancer.Input.Data.TCE.contin.DRAFT.pdf and
14 Appendix.linked.files\AppF.Non-cancer.Input.Data.TCE.dichot.DRAFT.pdf The documents
15 Appendix.linked.files\AppF.Non-cancer.Results.TCE.contin.DRAFT.pdf and
16 Appendix.linked.files\AppF.Non-cancer.Results.TCE.dichot.DRAFT.pdf present the data and
17 model summary statistics, including goodness-of-fit measures (Chi-square goodness-of-fit
18 /7-value, AIC), parameter estimates, BMD, and BMDL. The group numbers "GRP" are arbitrary
19 and are the same as GRP in the plots. Finally, note that not all plots are shown in the documents
20 above, since these spreadsheets include many "alternative" analyses.
21
22 F.4.2. Nested Dichotomous Modeling Results
23 F.4.2.1. Johnson et al (2003) Fetal Cardiac Defects
24 F.4.2.1.1. Results using applied dose. The biological endpoint was frequency of rat fetuses
25 having cardiac defects, as shown in Table F-4. Individual animal data were kindly provided by
26 Dr. Johnson (personal communication from Paula Johnson, University of Arizona, to Susan
27 Makris, U.S. EPA, 26 August 2009). Cochran-Armitage trend tests using number of fetuses and
28 number of litters indicated significant increases in response with dose (with or without including
29 the highest dose).
30 One suitable candidate for a LSC was available: female weight gain during pregnancy.
31 Based on goodness of fit, this covariate did not contribute to better fit and was not used. Some
32 ICs were significant and these parameters were included in the model.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-8 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
Table F-4. Data on fetuses and litters with abnormal hearts from Johnson et
al. (2003)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Dose group
(mg/kg/d):
0
0.00045
0.048
0.218
129
Fetuses
Number of pups:
Abnormal heart:
606
13
144
0
110
5
181
9
105
11
Litters
Number of litters:
Abnormal heart:
55
9
12
0
9
4
13
5
9
6
With the high dose included, the chi-square goodness of fit was acceptable, but some
residuals were large (1.5 to 2) for the control and two lower doses. Therefore, models were also
fitted after dropping the highest dose. For these, goodness of fit was adequate, and scaled
residuals were smaller for the low doses and control. Predicted expected response values were
closer to observed when the high dose was dropped, as shown in Table F-5:
Table F-5. Comparison of observed and predicted numbers of fetuses with
abnormal hearts from Johnson et al. (2003), with and without the high-dose
group, using a nested model
Dose group (mg/kg/d):
Observed:
Abnormal hearts (pups)
0
13
0.00045
0
0.048
5
0.218
9
129
11
Predicted expected:
With high dose
Without high dose
19.3
13.9
4.5
o o
J.J
3.5
3.4
5.7
10
11
—
Accuracy in the low-dose range is especially important because the BMD is based upon
the predicted responses at the control and the lower doses. Based on the foregoing measures of
goodness of fit, the model based on dropping the high dose was used.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-9 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
The nested log-logistic and Rai-VanRyzin models were fitted; these gave essentially the
same predicted responses and POD. The former model was used as the basis for a POD; results
are in Table F-6 and Figure F-2.
Table F-6. Results of nested log-logistic model for fetal cardiac anomalies
from Johnson et al. (2003) without the high-dose group, on the basis of
applied dose (mg/kg/d in drinking water)
Model
NLOG
NLOG
NLOG
NLOG
NLOG
NLOG*
LSC?
Y
Y
N
N
N
N
1C?
Y
N
N
Y
Y
Y
AIC
246.877
251.203
248.853
243.815
243.815
243.815
Pval
NA(df=0)
0.0112
0.0098
0.0128
0.0128
0.0128
BMR
0.01
0.01
0.01
0.1
0.05
0.01
BMD
0.252433
0.238776
0.057807
0.71114
0.336856
0.064649
BMDL
0.03776
0.039285
0.028977
0.227675
0.107846
0.020698
9
10
11
12
13
* Indicates model selected (Rai-VanRyzin model fits are essentially the same).
NLOG = "nested log-logistic" model.
LSC analyzed was female weight gain during pregnancy.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-10 DRAFT—DO NOT CITE OR QUOTE
-------
Nested Logistic Model with 0.95 Confidence Level
0.12
0.1
| 0.08
<0.06
o
0.04
0.02
0
Nested Logistic
BMDL
BMD
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
dose
13:3608/272008
1
2
3
4
5
6
1
Nested Logistic Model with 0.95 Confidence Level
0.12
0.1
ID 0.08
o
£
< 0.06
c
% 0.04
03
0.02
0
Nested Logistic
BMDL
BMP
0 0.05
13:3708/272008
0.1
dose
0.15
0.2
Figure F-2. BMD modeling of Johnson et al. (2003) using nested log-logistic
model, with applied dose, without LSC, with 1C, and without the high-dose
group, using a BMR of 0.05 extra risk (top panel) or 0.01 extra risk (bottom
panel).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-l 1 DRAFT—DO NOT CITE OR QUOTE
-------
1 F.4.2.1.2. Chi-square Goodness of Fit Test for nested log-logistic. The BMDS choice of
2 subgroups did not seem appropriate given the data. The high-dose group of 13 litters was
3 subdivided into three subgroups having sums of expected counts 3, 3, and 2. However, the
4 control group of 55 litters could have been subdivided because expected response rates for
5 controls were relatively high. There was also concern that the goodness of fit might change with
6 alternative choices of sub groupings.
7 An R program was written to read the BMDS output, reading parameters and the table of
8 litter-specific results (dose, covariate, estimated probability of response, litter size, expected
9 response count, observed response count, scaled chi-square residual). The control group of
10 55 litters was subdivided into three subgroups of 18, 18, and 19 litters. Control litters were
11 sampled randomly without replacement 100 times, each time creating 3 subgroups—i.e.,
12 100 random assignments of the 55 control litters to three subgroups were made. For each of
13 these, the goodness-of-fit calculation was made and the/>-value saved. Within these
14 100/>-values, >75% were >0.05, and >50% had ^-values >0.11, this indicated that the model is
15 acceptable based on goodness-of-fit criteria.
16
17 F.4.2.1.3. Results using physiologically basedpharmacokinetic (PBPK) model-based dose
18 metrics. The nested log-logistic model was also run using the dose metrics in the dams of total
19 oxidative metabolism scaled by body weight to the 3/4-power (TotOxMetabBW34) and the area-
20 under-the-curve of TCE in blood (AUCCBld). As with the applied dose modeling, LSC
21 (maternal weight gain) was not included, but 1C was included, based on the criteria outlined
22 previously (see Section F.3.1.2). The results are summarized in Table F-7 and Figure F-3 for
23 TotOxMetabBW34 and Table F-8 and Figure F-4 for AUCCBld.
24
25 F.4.2.2. Narotsky et al (1995)
26 Data were combined for the high doses in the single-agent experiment and the lower
27 doses in the 'five-cube' experiment. Individual animal data were kindly provided by Dr.
28 Narotsky (personal communications from Michael Narotsky, U.S. EPA, to John Fox, U.S. EPA,
29 19 June 2008, and to Jennifer Jinot, U.S. EPA, 10 June 2008). Two endpoints were examined:
30 frequency of eye defects in rat pups and prenatal loss (number of implantation sites minus
31 number of live pups on postnatal day 1).
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-12 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table F-7. Results of nested log-logistic model for fetal cardiac anomalies
from Johnson et al. (2003) without the high-dose group, using the
TotOxMetabBW34 dose metric
Model
NLOG
NLOG
NLOG
NLOG*
NLOG
LSC?
Y
Y
N
N
N
1C?
Y
N
Y
Y
N
AIC
246.877
251.203
243.815
243.815
248.853
Pval
NA(df=0)
0.0112
0.0128
0.0128
0.0098
BMR
0.01
0.01
0.1
0.01
0.01
BMD
0.174253
0.164902
0.489442
0.0444948
0.0397876
BMDL
0.0259884
0.0270378
0.156698
0.0142453
0.0199438
5
6
7
8
9
10
11
12
13
14
15
* Indicates model selected. BMDS failed with the Rai-VanRyzin and NCTR models.
NLOG = "nested log-logistic" model.
LSC analyzed was female weight gain during pregnancy.
Nested Logistic Model with 0.95 Confidence Level
0.1
§5.08
•§).04
2
45.02
0
ted Logistic
BMDL
BMP
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
dose
12:4402/062009
Figure F-3. BMD modeling of Johnson et al. (2003) using nested log-logistic
model, with TotOxMetabBW34 dose metric, without LSC, with 1C, and
without the high-dose group, using a BMR of 0.01 extra risk.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-13 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table F-8. Results of nested log-logistic model for fetal cardiac anomalies
from Johnson et al. (2003) without the high-dose group, using the AUCCBld
dose metric
Model
NLOG
NLOG
NLOG*
NLOG*
NLOG
LSC?
Y
Y
N
N
N
1C?
Y
N
Y
Y
N
AIC
246.877
251.203
243.816
243.816
248.853
Pval
NA(df=0)
0.0112
0.0128
0.0128
0.0098
BMR
0.01
0.01
0.1
0.01
0.01
BMD
0.00793783
0.00750874
0.0222789
0.00202535
0.00181058
BMDL
0.00118286
0.00123047
0.00712997
0.000648179
0.000907513
5
6
7
8
9
10
11
12
13
14
15
* Indicates model selected. BMDS failed with the Rai-VanRyzin and NCTR models.
NLOG = "nested log-logistic" model.
LSC analyzed was female weight gain during pregnancy.
Nested Logistic Model with 0.95 Confidence Level
Nisted Logistic
0.1
|J0.08
I
<0.06
c
'•§0.04
-0.02
0
BMDL
BMP
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
dose
12:4202/062009
Figure F-4. BMD modeling of Johnson et al. (2003) using nested log-logistic
model, with AUCCBld dose metric, without LSC, with 1C, and without the
high-dose group, using a BMR of 0.01 extra risk.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-14 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
10
11
12
13
14
15
16
17
Two LSCs were considered, with analyses summarized in Table F-9. The number of implants is
unrelated to dose, as inferred from regression and analysis of variance, and was considered as a
LSC for eye defects. As number of implants is part of the definition for the endpoint of prenatal
loss, it is not considered as a LSC for prenatal loss. A second LSC, the dam body weight on
gestation day (GD) 6 (damBW6) was significantly related to dose and is unsuitable as a litter-
specific covariate.
Table F-9. Analysis of LSCs with respect to dose from Narotsky et al. (1995)
Relation of litter-specific covariates to dose
Implants:
damBW6:
none
significant
TCE
0
10.1
32
101
320
475
633
844
1,125
Mean
Implants
9.5
10.1
9.1
7.8
10.4
9.7
9.6
8.9
9.6
Mean
damBW6
176.0
180.9
174.9
170.1
174.5
182.4
185.3
182.9
184.2
Using expt as covariate, e.g., damBW6 ~ TCE.mg.kgd + expt
Linear regression
AoV (ordered factor)
p = 0.7486
p = 0.1782
p = 0.0069
p = 0.0927
Two LSCs were considered, with analyses summarized in Table F-9. The number of
implants is unrelated to dose, as inferred from regression and analysis of variance, and was
considered as a LSC for eye defects. As number of implants is part of the definition for the
endpoint of prenatal loss, it is not considered as a LSC for prenatal loss. A second LSC, the dam
body weight on GD 6 (damBW6) was significantly related to dose and is unsuitable as a litter-
specific covariate.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-15 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
F.4.2.2.1. Feta/ eye defects. The nested log-logistic and Rai-VanRyzin models were fitted to
the number of pups with eye defects reported by Narotsky et al. (1995), with the results
summarized in Table F-10.
Table F-10. Results of nested log-logistic and Rai-VanRyzin model for fetal
eye defects from Narotsky et al. (1995), on the basis of applied dose (mg/kg/d
in drinking water)
Model
NLOG
NLOG
NLOG
NLOG
NLOG
NLOG
RAI
RAI
RAI
RAI
RAI
RAI
LSC?
Y
Y
N
N
N
N
Y
Y
N
N
N
N
1C?
Y
N
Y
N
N
N
Y
N
Y
N
N
N
AIC
255.771
259.024
270.407
262.784
262.784
262.784
274.339
264.899
270.339
262.481
262.481
262.481
Pval
0.3489
0.0445
0.2281
0.0529
0.0529
0.0529
0.1047
0.0577
0.2309
0.0619
0.0619
0.0619
BMR
0.05
0.05
0.05
0.10
0.05
0.01
0.05
0.05
0.05
0.10
0.05
0.01
BMD
875.347
830.511
622.342
691.93
427.389
147.41
619.849
404.788
619.882
693.04
429.686
145.563
BMDL
737.328a
661.629
206.460
542.101
264.386
38.7117b
309.925
354.961
309.941
346.52
214.843
130.938b
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
a Graphical fit at the origin exceeds observed control and low dose responses and slope is quite flat (see Figure F-5),
fitted curve does not represent the data well.
b Indicates model selected.
NLOG = "nested log-logistic" model; RAI = Rai-VanRyzin model.
LSC analyzed was implants.
Results for the nested log-logistic model suggested a better model fit with the inclusion of
the LSC and 1C, based on AIC. However, the graphical fit (see Figure F-5) is strongly sublinear
and high at the origin where the fitted response exceeds the observed low-dose responses for the
control group and two low-dose groups. An alternative nested log-logistic model without either
LSC or 1C (see Figure F-6), which fits the low-dose responses better, was selected. Given that
this model had no LSC and no 1C, the nested log-logistic model reduces to a quantal log-logistic
model. Parameter estimates and the ^-values were essentially the same for the two models (see
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-16 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
5
Table F-l 1). A similar model selection can be justified for the Rai-Van Ryzin model (see
Figure F-7). Because no LSC and no 1C were needed, this endpoint was modeled with quantal
models, using totals of implants and losses for each dose group, which allowed choice from a
wider range of models (those results appear with quantal model results in this appendix).
Nested Logistic Model with 0.95 Confidence Level
6
7
0.5
0.4
Affected
o
w
|0.2
LL
0.1
0
INCSICU i_uyii>iio
L
^
r
£<5
i <
0 200
<
400
> <
>
BMDL
600
dose
— -~~^
•"^
800
/
- :
1
:
:
BMD
1000
17:2708/042008
Figure F-5. BMD modeling of fetal eye defects from Narotsky et al. (1995)
using nested log-logistic model, with applied dose, with both LSC and 1C,
using a BMR of 0.05 extra risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-17 DRAFT—DO NOT CITE OR QUOTE
-------
Nested Logistic Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
0.5
0.4
T3
-------
RaiVR Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0.5
RaiVR
T3
1
0.4
0.3
•-
t3
ro
0.1
0
BMDL
BMD
0 200
17:2508/042008
400
600
dose
800
1000
Figure F-7. BMD modeling of fetal eye defects from Narotsky et al. (1995)
using nested Rai-VanRyzin model, with applied dose, without either LSC or
1C, using a BMR of 0.05 extra risk.
F.4.2.2.2. Narotsky et al. (1995) prenatal loss. The nested log-logistic and Rai-VanRyzin
models were fitted to prenatal loss reported by Narotsky et al. (1995), with the results
summarized in Table F-12.
The BMDS nested models require a LSC, so dam body weight on GD6 ("damBW6") was
used as the LSC. However, damBW6 is significantly related to dose and, so, is not a reliable
LSC. Number of implants could not be used as a LSC because it was identified as number at risk
in the BMDS models. These issues were obviated because the model selected did not employ
the LSC.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-19 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
3
4
Table F-12. Results of nested log-logistic and Rai-VanRyzin model for
prenatal loss from Narotsky et al. (1995), on the basis of applied dose
(mg/kg/d in drinking water)
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Model
NLOG
NLOG
NLOG
NLOG
NLOG
NLOG
RAI
RAI
RAI
RAI
RAI
RAI
LSC?
Y
Y
N
N
N
N
Y
Y
N
N
N
N
1C?
Y
N
N
Y
Y
Y
Y
N
N
Y
Y
Y
AIC
494.489
627.341
628.158
490.766
490.766
490.766
491.859
626.776
626.456
488.856
488.856
488.856
Pval
0.2314
0.0000
0.0000
0.2509
0.2509
0.2509
0.3044
0.0000
0.0000
0.2983
0.2983
0.2983
BMR
0.10
0.10
0.10
0.10
0.05
0.01
0.10
0.10
0.10
0.10
0.05
0.01
BMD
799.723
790.96
812.92
814.781
738.749
594.995
802.871
819.972
814.98
814.048
726.882
562.455
BMDL
539.094
694.673
725.928
572.057
447.077
252.437 *
669.059
683.31
424.469
678.373
605.735
468.713 *
* Indicates model selected.
NLOG = "nested log-logistic" model; RAI = Rai-VanRyzin model.
LSC analyzed was dam body weight on GD6.
For the nested log-logistic models, the AIC is much larger when the 1C is dropped, so the
1C is needed in the model. The LSC can be dropped (and is also suspect because it is correlated
with dose). The model with 1C and without LSC was selected on the basis of AIC (shown in
Figure F-8). For the Rai-VanRyzin models, the model selection was similar to that for the nested
log-logistic, leading to a model with 1C and without LSC, which had the lowest AIC (shown in
Figure F-9).
F.4.3. Model Selections and Results
The final model selections and results for noncancer dose-response modeling are
presented in Table F-13.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-20 DRAFT—DO NOT CITE OR QUOTE
-------
1
2
4
5
6
Nested Logistic Model with 0.95 Confidence Level
0.8
0.7
Tg 0.6
"G
£ 0.5
<
c 0.4
o
£ °-3
2
u- 0.2
0 1
w . i
0
INCOICU i_vjyiono :
-
-
-
-
-
-
-
- -T-
1
= T
i
-
BMDL
>
__-— —~"
>
^/
/
/ <
/
/
/
/
/
/
H
\
H
> ;
'-.
'-_
H
Y '-
BMD i
0 200
16:4408/202008
400
600
dose
800
1000
Nested Logistic Model with 0.95 Confidence Level
0.8
0.7
S 0.6
o
£0.5
c 0.4
o
'•6 0.3
ui 0.2
0.1
0
NCOICU i_uyioiio
n^-
1
: j
BMDL
I "
R
V1D
/
/
^
^
~
|
:
0 200
16:4508/202008
400
600
dose
800
1000
Figure F-8. BMD modeling of prenatal loss reported in Narotsky et al.
(1995) using nested log-logistic model, with applied dose, without LSC, with
1C, using a BMR of 0.05 extra risk (top panel) or 0.01 extra risk (bottom
panel).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-21 DRAFT—DO NOT CITE OR QUOTE
-------
RaiVR Model with 0.95 Confidence Level
0.8
0.7
"CD 0.6
£ 0.5
c 0.4
o
t5 0.3
CO
LJ- 0.2
0.1
0
^divrx
r
r
r
;
r
r "T
| 1
0
- T
_L ^
- I J
200 400
>
BMDL
600
BMD
/
/
/
\
'-._
'-._
> ;
:
j
800 1000
dose
16:4608/202008
RaiVR Model with 0.95 Confidence Level
4
5
6
0.8
0.7
^ 0.6
sg 0-5
c 0.4
0
ro °'3
^ 0.2
0.1
0
r\ai vr\
h
h
§-
r _
; 1
: $
- i -
BMDL
BMC
)
\
\
\
/ \
/
-
j
0 200
16:4608/202008
400
600
dose
800
1000
Figure F-9. BMD modeling of prenatal loss reported in Narotsky et al.
(1995) using nested Rai-VanRyzin model, with applied dose, without LSC,
with 1C, using a BMR of 0.05 extra risk (top panel) or 0.01 extra risk (bottom
panel).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-22 DRAFT—DO NOT CITE OR QUOTE
-------
Table F-13. Model selections and results for noncancer dose-response analyses
to
- a
o g"
vo 5'
I
Co
1
TO'
GRP
Study/run
abbrev. !
Species
Sex
Strain
Exp.
route
Endpoint
Dose metric
BMR
type
BMR
BMD/
BMDL
BMDL
Model
Rep.
BMD
Notes
Dichotomous models
3
7
13
13
13
14
36
38
38
38
38
39
49
49
49
49
Chia et al.,
1996
Narotsky
etal., 1995
Narotsky
etal.,
1995.sa
Narotsky
etal.,
1995.sa
Narotsky
etal.,
1995.sa
Johnson
etal.,
2003.drophi
Griffin etal.,
2000
Maltoni
etal., 1986
Maltoni
etal., 1986
Maltoni
etal., 1986
Maltoni
etal., 1986
Maltoni
etal., 1986
NTP, 1988
NTP, 1988
NTP, 1988
NTP, 1988
human
rat
rat
rat
rat
rat
mice
rat
rat
rat
rat
rat
rat
rat
rat
rat
M
F
F
F
F
F
F
M
M
M
M
M
F
F
F
F
workers.elec. factory
F344
F344
F344
F344
Sprague.Dawley
MRL++
Sprague.Dawley
Sprague.Dawley
Sprague.Dawley
Sprague.Dawley
Sprague.Dawley
Marshall
Marshall
Marshall
Marshall
inhal
oral.gav
oral.gav
oral.gav
oral.gav
oral.dw
oral.dw
inhal
inhal
inhal
inhal
oral.gav
oral.gav
oral.gav
oral.gav
oral.gav
N.hyperzoospermia
N. pups. eye. defects
N. dams, w.resorbed. litters
N. dams, w.resorbed. litters
N. dams, w.resorbed. litters
N. litters. abnormal. hearts
portal. infiltration
megalonucleocytosis
megalonucleocytosis
megalonucleocytosis
megalonucleocytosis
megalonucleocytosis
toxic nephropathy
toxic nephropathy
toxic nephropathy
toxic nephropathy
appl.dose
appl.dose
appl.dose
AUCCBId
TotMetabBW34
appl.dose
appl.dose
appl.dose
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
appl.dose
appl.dose
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
extra
0.1
0.01
0.01
0.01
0.01
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.05
0.05
0.05
0.05
2.14
1.46
5.47
5.77
1.77
2.78
2.67
1.22
1.18
1.19
1.13
1.53
1.45
1.45
1.46
1.45
1.43
60.1
32.2
17.5
77.5
0.0146
13.4
40.2
0.0888
0.086
53.8
33.8
9.45
0.0132
0.0129
2.13
loglogistic.1
multistage
multistage. 2
multistage. 2
weibull
loglogistic.1
loglogistic.1
multistage
loglogistic
loglogistic
weibull
multistage. 2
loglogistic.1
loglogistic.1
loglogistic.1
loglogistic.1
3.06
806
570
327
156
0.0406
35.8
49.2
0.105
0.102
61
51.8
28.9
0.0404
0.0397
6.5
a
b
c
d
e
to
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
- a
o g"
vo 5'
I
Co
1
TO'
Table F-13. Model selections and results for noncancer dose-response analyses (continued)
GRP
Study/run
abbrev. 5
Species
Sex
Strain
Exp.
route
Endpoint
Dose metric
BMR
type
BMR
BMD/
BMDL
BMDL
Model
Rep.
BMD
Notes
Mested dichotomous models
NA
NA
NA
NA
Johnson
etal.,
2003.drophi
Johnson
etal.,
2003.droph
Johnson
etal.,
2003.drophi
Narotsky
etal., 1995
rat
rat
rat
rat
F
F
F
F
Sprague.Dawley
Sprague.Dawley
Sprague.Dawley
F344
oral.dw
oral.dw
oral.dw
oral.gav
N. pups. abnormal. hearts
N. pups. abnormal. hearts
N. pups. abnormal. hearts
N. prenatal. loss
appl.dose
TotOxMetabBW34
AUCCBId
appl.dose
extra
extra
extra
extra
0.01
0.01
0.01
0.01
3.12
3.12
3.12
1.2
0.0207
0.0142
0.000648
469
loglogistic.lC
loglogistic.lC
loglogistic.lC
RAI.IC
0.711
814
b
b
b
Continuous models
2
5
8
19
21
23
26
34sq
49
51
Land etal.,
1981
Carney
et al., 2006
Narotsky
etal., 1995
Crofton and
Zhao. 1997
George
etal., 1986
George
etal., 1986
George
etal., 1986
Moser et al.,
1995+persc
om
George
etal., 1986
Buben and
O'Flaherty,
1985
mouse
rat
rat
rat
rat
rat
rat
rat
rat
mouse
M
F
F
M
F
F
F
F
F
M
(C57B1xC3H)F1
Sprague-Dawley
(Crl:CD)
F344
Long-Evans
F344
F344
F344
F344
F344
SwissCox
inhal
inhal
oral.gav
inhal
oral. food
oral. food
oral. food
oral.gav
oral. food
oral.gav
pet. abnormal. sperm
gm.wgt.gain.GD6.9
gm.wgt.gain.GD6.20
dB.auditory.threshold(16kHz)
litters
live. pups
Foffspring.BWgm.day21
no. rears
traverse.time.21do
Liverwt.pctBW
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
appl.dose
standard
relative
relative
absolute
standard
standard
relative
standard
relative
relative
0.5
0.1
0.1
10
0.5
0.5
0.05
1
1
0.1
1.33
2.5
1.11
1.11
1.69
1.55
1.41
1.64
1.98
1.26
46.9
10.5
108
274
179
152
79.7
248
72.6
81.5
polynomial. constvar
hill
polynomial. constvar
polynomial. constvar
polynomial. constvar
polynomial. constvar
polynomial. constvar
polynomial. constvar
power
hill. constvar
125
62.3
312
330
604
470
225
406
84.9
92.8
b,f
to
i
o §
H I
O >
HH Oq
H ^
O
H
W
-------
to
- a
o g"
vo 5'
I
Co
1
TO'
Table F-13. Model selections and results for noncancer dose-response analyses (continued)
GRP
51
51
58
58
58
SO.Rp
SO.Rp
SO.Rp
S3
62
52
55
55
55
55
57
57
57
Study/run
abbrev. 5
Buben and
O'Flaherty,
1985
Buben and
O'Flaherty,
1985
Kjellstrand
etal, 1983b
Kjellstrand
etal, 1983b
Kjellstrand
etal, 1983b
Kjellstrand
etal, 1983b
Kjellstrand
etal, 1983b
Kjellstrand
etal, 1983b
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Woolhiser
et al, 2006
Species
mouse
mouse
mouse
mouse
mouse
mouse
mouse
mouse
rat
rat
rat
rat
rat
rat
rat
rat
rat
rat
Sex
M
M
M
M
M
M
M
M
F
F
F
F
F
F
F
F
F
F
Strain
SwissCox
SwissCox
NMRI
NMRI
NMRI
NMRI
NMRI
NMRI
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
CD (Sprague-
Dawley)
Exp.
route
oral.gav
oral.gav
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
inhal
Endpoint
Jverwt.pctBW
Liverwt.pctBW
Liverwt.pctBW
Liverwt.pctBW
Liverwt.pctBW
Kidneywt.pctBW
Kidneywt.pctBW
Kidneywt.pctBW
Antibody. Forming Cells
Antibody. Forming Cells
Antibody. Forming Cells
kidney. wt. perl OOgm
kidney. wt. perl OOgm
kidney. wt. perl OOgm
kidney .wt. perl OOgm
liver. wt. perl OOgm
iver.wt. perl OOgm
iver.wt. perl OOgm
Dose metric
AMetLiv! BW34
TotOxMetabBW34
appl.dose
AMetLiv1BW34
TotOxMetabBW34
appl.dose
AMetGSHBW34
TotMetabBW34
appl.dose
AUCCBId
TotMetabBW34
appl.dose
ABioactDCVCBW34
AMetGSHBW34
TotMetabBW34
appl.dose
AMetLiv! BW34
TotOxMetabBW34
BMR
type
relative
relative
relative
relative
relative
relative
relative
relative
standard
standard
standard
relative
relative
relative
relative
relative
relative
relative
BMR
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
1
1
1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
BMD/
BMDL
1.08
1.08
1.36
1.4
1.3
1.17
1.18
1.17
1.94
1.44
1.5
4.29
4.27
4.28
1.47
4.13
1.53
1.53
BMDL
28.6
37
21.6
22.7
73.4
34.7
0.17
71
31.2
149
40.8
15.7
0.0309
0.032
40.8
25.2
46
48.9
Model
polynomial.constvar
polynomial. constvar
hill
hill
hill
polynomial
polynomial
polynomial
power.constvar
polynomial
polynomial
hill. constvar
hill. constvar
hill. constvar
polynomial.constvar
hill. constvar
polynomial.constvar
polynomial.constvar
Rep.
BMD
28.4
36.7
30.4
32.9
97.7
47.1
0.236
95.2
60.6
214
61.3
54.3
0.103
0.107
52.3
70.3
56.1
59.8
Notes
b
to
o §
H I
O >
HH Oq
H ^
O
H
W
-------
o Table F-13. Model selections and results for noncancer dose-response analyses (continued)
to
\ ^ "Eight-stage multistage model.
§ *;• 'Dropped highest dose.
HH Oq
H TO
O
H
W
-------
1 F.5. DERIVATION OF POINTS OF DEPARTURE
2 F.5.1. Applied Dose Points of Departure
3 For oral studies in rodents, the POD on the basis of applied dose in mg/kg/d was taken to
4 be the BMDL, NOAEL, or LOAEL. NOAELs and LOAELs were adjusted for intermittent
5 exposure to their equivalent continuous average daily exposure (for BMDLs, the adjustments
6 were already performed prior to BMD modeling).
7 For inhalation studies in rodents, the POD on the basis of applied dose in ppm was taken
8 to be the BMDL, NOAEL, or LOAEL. NOAELs and LOAELs were adjusted for intermittent
9 exposure to their equivalent continuous average daily exposure (for BMDLs, the adjustments
10 were already performed prior to BMD modeling). These adjusted concentrations are considered
11 human equivalent concentrations, in accordance with U.S. EPA (1994), as TCE is considered a
12 Category 3 gas (systemically acting) and has a blood-air partition coefficient in rodents greater
13 than that in humans (see Section 3.1).
14
15 F.5.2. Physiologically Based Pharmacokinetic (PBPK) Model-Based Human Points of
16 Departure
17 As discussed in Section 5.1.3, the PBPK model was used for simultaneous interspecies
18 (for endpoints in rodent studies), intraspecies, and route-to-route extrapolation based on the
19 estimates from the PBPK model of the internal dose points of departure (idPOD) for each
20 candidate critical study/endpoints. The following documents contain figures showing the
21 derivation of the human equivalent doses and concentrations (human equivalent doses [HEDs]
22 and human equivalent concentrations [HECs]) for the median (50th percentile) and sensitive (99th
23 percentile) individual from the (rodent or human) study idPOD. In each case, for a specific
24 study/endpoint(s)/sex/species (in the figure main title), and for a particular dose metric (Y-axis
25 label), the horizontal line shows the original study idPOD (a BMDL, NOAEL, or LOAEL as
26 noted) and where it intersects with the human 99th percentile (open square) or median (closed
27 square) exposure-internal-dose relationship:
28 Appendix.linked.files\AppF.Non-cancer.FIECs.Plots.human.inhalation.studies.TCE.DRAFT.pdf
29 Appendix.linked.files\AppF.Non-cancer.FIECs.Plots.rodent.inhalation.studies.TCE.DRAFT.pdf
30 Appendix.linked.files\AppF.Non-cancer.FIECs.Plots.rodent.oral.studies.TCE.DRAFT.pdf
31 Appendix.linked.files\AppF.Non-cancer.FIEDs.Plots.human.inhalation.studies.TCE.DRAFT.pdf
32 Appendix.linked.files\AppF.Non-cancer.FIEDs.Plots.rodent.inhalation.studies.TCE.DRAFT.pdf
33 Appendix.linked.files\AppF.Non-cancer.FIEDs.Plots.rodent.oral.studies.TCE.DRAFT.pdf
34 The original study internal doses are based on the median estimates from about 2,000
35 "study groups" (for rodent studies) or "individuals" (for human studies), and corresponding
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-27 DRAFT—DO NOT CITE OR QUOTE
-------
1 exposures for the human median and 99th percentiles were derived from a distribution of 2,000
2 "individuals." In both cases, the distributions reflect combined uncertainty (in the population
3 means and variances) and population variability.
4 In addition, as part of the uncertainty/variability analysis described in Section 5.1.4.2, the
5 POD for studies/endpoints for which BMD modeling was done was replaced by the LOAEL or
6 NOAEL. This was done to because there was no available tested software for performing BMD
7 modeling in such a context and because of limitations in time and resources to develop such
8 software. However, the relative degree of uncertainty/variability should be adequately captured
9 in the use of the LOAEL or NOAEL. The graphical depiction of the HECgg or HEDgg using
10 these alternative PODs is shown in the following files:
11 Appendix, linked. files\AppF .Non-
12 cancer.HECs.AltPOD.Plots.rodent.inhalation.studies.TCE.DRAFT.pdf
13 Appendix, linked. files\AppF .Non-
14 cancer.HECs.AltPOD.Plots.rodent.oral.studies.TCE.DRAFT.pdf
15 Appendix, linked. files\AppF .Non-
16 cancer.HEDs.AltPOD.Plots.rodent.inhalation.studies.TCE.DRAFT.pdf
17 Appendix, linked. files\AppF .Non-
18 cancer.HEDs.AltPOD.Plots.rodent.oral.studies.TCE.DRAFT.pdf
19
20 F.6. SUMMARY OF POINTS OF DEPARTURE (PODs) FOR CRITICAL STUDIES
21 AND EFFECTS SUPPORTING THE INHALATION REFERENCE CONCENTRATION
22 (RfC) AND ORAL REFERENCE DOSE (RfD)
23 This section summarizes the selection and/or derivation of PODs from the critical studies
24 and effects that support the inhalation reference concentration (RfC) and oral reference dose
25 (RfD). In particular, for each endpoint, the following are described the dosimetry (adjustments
26 of continuous exposure, PBPK dose metrics), selection of BMR and BMD model (if BMD
27 modeling was performed), and derivation of the human equivalent concentration or dose for a
28 sensitive individual (if PBPK modeling was used). Section 5.1.3.1 discusses the dose metric
29 selection for different endpoints.
30
31 F.6.1. National Toxicology Program (NTP, 1988)—Benchmark Dose (BMD) Modeling of
32 Toxic Nephropathy in Rats
33 The critical endpoint here is toxic nephropathy in female Marshall rats (NTP, 1988),
34 which was the most sensitive sex/strain in this study, although the differences among different
35 sex/strain combinations was not large (BMDLs differed by <3-fold).
36
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-28 DRAFT—DO NOT CITE OR QUOTE
-------
1 F.6.1.1. Dosimetry and Benchmark Dose (BMD) Modeling
2 Rats were exposed to 500 or 1,000 day, 5 days/week, for 104 weeks. The primary dose
3 metric was selected to be average amount of dichlorovinyl cysteine (DCVC)
4 bioactivated/kgyVday, with median estimates from the PBPK model for the female Marshall rats
5 in this study of 0.47 and 1.1.
6 Figure F-10 shows BMD modeling for the dichotomous models used (see Section F.5.1,
7 above). The log-logistic model with slope constrained to >1 was selected because (1) the log-
8 logistic model with unconstrained slope yielded a slope estimate <1 and (2) it had the lowest
9 AIC.
10 The idPOD of 0.0132 mg DC VC bioactivated/kgyVday was a BMDL for a BMR of 5%
11 extra risk. This BMR was selected because toxic nephropathy is a clear toxic effect. This BMR
12 required substantial extrapolation below the observed responses (about 60%); however, the
13 response level seemed warranted for this type of effect and the ratio of the BMD to the BMDL
14 was not large (1.56 for the selected model).
15
16 F.6.1.2. Derivation of HEC99 and HED99
17 The F£EC99 and F£ED99 are the lower 99th percentiles for the continuous human exposure
18 concentration and continuous human ingestion dose that lead to a human internal dose equal to
19 the rodent idPOD. The derivation of the HEC99 of 0.0056 ppm and HED99 of 0.00338 mg/kg/d
20 for the 99th percentile for uncertainty and variability are shown in Figure F-l 1. These values are
21 used as this critical effect's POD to which additional uncertainty factors (UFs) are applied.
22
23 F.6.2. National Cancer Institute (NCI, 1976)—Lowest-Observed-Adverse-Effect Level
24 (LOAEL) for Toxic Nephrosis in Mice
25 The critical endpoint here is toxic nephrosis in female B6C3F1 mice (NCI, 1976), which
26 was the most sensitive sex in this study, although the LOAEL for males differed by less than
27 50%.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-29 DRAFT—DO NOT CITE OR QUOTE
-------
NTP.1988 kidney toxic nephropathy rat Marshall F oral.gav(GRP 49)
BMR: 0.05 extra
loglogistic, Pval = 1, AIC=123
background 0, intercept 0.74, slope 0.31
0.0
\
0.2
0.4
i i r
0.6 0.8 1.0
ABioactDCVCBW34
loglogistic, Pval = 0.44, AIC = 122
background 0, intercept 1, slope 1
0.0 0.2
ABioactDCVCBW34
1.0
<
£=
O
CD
0.8 -
0.6 -
0.4 -
0.2 -
0.0 -
multistage-1, Pval = 0.05, AIC = 126
Background 0, Beta(1) 1.4, Beta(2) 0
BMDandBMDL, O.OB58, 0.0288
0.0
I
0.2
\
\
\
ABioactDCVCBW34
\
0.4 0.6 0.8 1.0
multistage-2, Pval = 0.05, AIC = 126
Background 0, Beta(1) 1.4, Beta(2) 0
0.0 0.2
0.6 0.8 1.0
ABioactDCVCBW34
1
2
3
4
multistage-1, Pval = 0.05, AIC = 126
Background 0, Beta(1) 1.4
0.0 0.2 0.4 0.6 0.8
ABioactDCVCBW34
1.0
weibull, Pval = 1, AIC = 123
Background 0, Slope 1.1, Power 0.19
0.8 -
Ju 0.6 -
%
£=
O
CD
0.4 -
0.2 -
0.0 -
BMDandBMDL, 9.1Be-08, NA
0.0 0.2 0.4 0.6 0.8
ABioactDCVCBW34
1.0
Figure F-10. BMD modeling of NTP (1988) toxic nephropathy in female
Marshall rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-30 DRAFT—DO NOT CITE OR QUOTE
-------
MTP.1988
BMDL for systemic kidney toxic.nephropathy in
F Marshall rat
MTP.1988
BMDL for systemic kidney toxic.nephropathy in
F Marshall rat
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
O -,
m
o
o
o -
o -
o -=
m
o
o
ro
o
in
o -
1 101 102 103 104
101 102 103 104
TCE inhalation (ppm)
TCE oral (mg/kg-d)
Figure F-ll. Derivation of HECgg and HEDgg corresponding to the rodent
idPOD from NTP (1988) toxic nephropathy in rats.
F.6.2.1. Dosimetry
Mice were exposed to a time-weighted average of 869 and 1,739 mg/kg/d, 5 days/week,
for 78 weeks. BMD modeling was not performed because the response at the LOAEL was
>90%. The primary dose metric was selected to be average amount of TCE conjugated with
glutathione (GSH)/kg Yd. In this study, the lower dose group was exposed to two different dose
levels (700 mg/kg/d for 12 weeks and 900 mg/kg/d for 66 weeks). The median estimates from
the PBPK model for the two dose levels were 0.583 and 0.762 mg TCE conjugation with
GSH/kgyVd. Applying the same time-weighted averaging gives an idPOD LOAEL of 0.735 mg
TCE conjugation with GSH/kgyVd.
F.6.2.2. Derivation ofHEC99 andHED99
The HECgg and HED99 are the lower 99th percentiles for the continuous human exposure
concentration and continuous human ingestion dose that lead to a human internal dose equal to
the rodent idPOD. The derivation of the HEC99 of 0.50 ppm and HED99 of 0.30 mg/kg/d for the
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-31 DRAFT—DO NOT CITE OR QUOTE
-------
1 99th percentile for uncertainty and variability are shown in Figure F-12. These values are used as
2 this critical effect's POD to which additional UFs are applied.
NCI. 1976
TWA-LOAEL for systemic kidney toxic.nephropathy in
F B6C3F1 mouse
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
•5J-
Q
m
1 10 10 10 10
NCI. 1976
TWA-LOAEL for systemic kidney toxic.nephropathy in
FB6C3F1 mouse
•5J-
m
10 10 10
TCE inhalation (ppm)
1 10 10 10 10
TCE oral (mg/kg-d)
Figure F-12. Derivation of HECgg and HEDgg corresponding to the rodent
idPOD from NTP (1988) toxic nephrosis in mice.
F.6.3. Woolhiser et al. (2006)—Benchmark Dose (BMD) Modeling of Increased Kidney
Weight in Rats
The critical endpoint here is increased kidney weights in female Sprague-Dawley (S-D)
rats (Woolhiser et al., 2006).
F.6.3.1. Dosimetry and Benchmark Dose (BMD) Modeling
Rats were exposed to 100, 300, and 1000, 6 hours/day, 5 days/week, for 4 weeks. The
primary dose metric was selected to be average amount of DCVC bioactivated/kg Vday, with
median estimates from the PBPK model for this study of 0.038, 0.10, and 0.51.
Figure F-13 shows BMD modeling for the continuous models used (see Section F.5.2,
above). The Hill model with constant variance was selected because it had the lowest AIC and
because other models with the same AIC either were a power model with power parameter <1 or
had poor fits to the control data set.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-32 DRAFT—DO NOT CITE OR QUOTE
-------
Woolhiser.etal.2006 Kidney kidney.wt.perlOOgm rat CD (Sprague-Dawley) F inhal (GRP 65)
BMR: 0.1 relative
power, P(V) = 0.81, P(M) = 0.92, AIC = -128
lalpha -5, rho 2, control 0.81, slope 0.19, power 0.44
0.0
ABioactDCVCBW34
0.5
power, P(V) = 0.89, P(M) = 0.87, AC = -130
alpha 0.0049, rho NA control 0.81, slope 0.19, power 0.44
1.00 -
0.0 0.1 0.2 0.3 0.4 0.5
ABioactDCVCBW34
power, P(V) = 0.81, P(M) = 0.38, ftC = -128
lalpha -5.1, rho 1.4, control 0.83, slope 0.23, power 1
1.00 -
c 0.95 -
g 0.90 -
E 0.85 -
0.80 -
0.75 -
BMDandBMDL 0.356, 0.234
1
0.0
\
0.1
I
0.2
0.3
ABioactDCVCBW34
\
0.4
0.5
power, P(V) = 0.89, P(M) = 0.4, AC =-130
alpha 0.0052, rho NA control 0.83, slope 0.23, power 1
1.00 -
0.1 0.2 0.3 0.4
ABioactDCVCBW34
i
0.5
1.00 -
c 0.95 -
80.90
E 0.85 -
0.80 -
0.75 -I
polyn, P(V) = 0.81, P(M) = 0.38, AC =-128
lalpha-5.1, rho 1.4, betaOO.83, betal 0.23
BMD and BMDL 0.356, 0.234
1
0.0
I
0.1
I
0.2
I
0.3
I
0.4
ABioactDCVCBW34
0.5
polyn, P(V) = 0.89, P(M) = 0.4, AIC = -130
alpha 0.0052, rho NA, betaO 0.83, betal 0.23
1.00 -
c 0.95 -
g 0.90 -
E 0.85 -
0.80 -
0.75 -I
BMDandBMDL, 0.36, 0.243
i
0.0
0.1
i
0.2
i
0.3
i
0.4
ABioactDCVCBW34
i
0.5
hill, P(V) = 0.81, P(M) = 0.65, AIC = -128
lalpha -5, rho 2.2, Intercept 0.81, v 0.18, n 1, k 0.15
1.00 -
0.5
hill, P(V) = 0.89, P(M) = 0.6, AIC = -130
alpha 0.005, rho NA intercept 0.81, v 0.18, n 1, k 0.15
1.00 -
ABioactDCVCBW34
0.1 0.2 0.3 0.4
ABioactDCVCBW34
i
0.5
2 Figure F-13. BMD modeling of Woolhiser et al. (2006) for increased kidney
3 weight in female S-D rats.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-33 DRAFT—DO NOT CITE OR QUOTE
-------
1 The idPOD of 0.0309 mg DCVC bioactivated/kgyVday was a BMDL for a BMR of 10%
2 weight change, which is the BMR typically used by U.S. EPA for body weight and organ weight
3 changes. The response used in each case was the organ weight as a percentage of body weight,
4 to account for any commensurate decreases in body weight, although the results did not differ
5 much when absolute weights were used instead.
6
7 F.6.3.2. Derivation of HEC99 and HED99
8 The HECgg and F£ED99 are the lower 99th percentiles for the continuous human exposure
9 concentration and continuous human ingestion dose that lead to a human internal dose equal to
10 the rodent idPOD. The derivation of the HEC99 of 0.0131 ppm and HED99 of 0.00791 mg/kg/d
11 for the 99th percentile for uncertainty and variability are shown in Figure F-14. These values are
12 used as this critical effect's POD to which additional UFs are applied.
13
14 F.6.4. Keil et al. (2009)—Lowest-Observed-Adverse-Effect Level (LOAEL) for Decreased
15 Thymus Weight and Increased Anti-dsDNA and Anti-ssDNA Antibodies in Mice
16 The critical endpoints here are decreased thymus weight and increased anti-dsDNA and
17 anti-ssDNA antibodies in female B6C3F1 mice (Keil et al., 2009).
18
19 F.6.5. Keil et al. (2009)—Lowest-Observed-Adverse-Effect Level (LOAEL) for Decreased
20 Thymus Weight and Increased Anti-dsDNA and Anti-ssDNA Antibodies in Mice
21 The critical endpoints here are decreased thymus weight and increased anti-dsDNA and
22 anti-ssDNA antibodies in female B6C3F1 mice (Keil et al., 2009).
23
24 F.6.5.1. Dosimetry
25 Mice were exposed to 1400 and 14000 ppb of TCE in drinking water, with an average
26 dose estimated by the authors to be 0.35 and 3.5 mg/kg/d, for 30 weeks. The dose-response
27 relationships were sufficiently supralinear that BMD modeling failed to produce an adequate fit.
28 The primary dose metric was selected to be the average amount of TCE metabolized/kg3/4/day.
29 The lower dose group was the LOAEL for both effects, and the median estimate from the PBPK
30 model at that exposure level was 0.139 mg TCE metabolized/kg3/4/day, which is used as the
31 rodent idPOD.
32
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-34 DRAFT—DO NOT CITE OR QUOTE
-------
Woolhiser.etal.2006
BMDL for systemic kidney weight.increased in
F Spraaue-Daw lev rat
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
m
a
ro
o
in
o -=
Woolhiser.etal.2006
BMDL for systemic kidney weight.increased in
F Spraaue-Dawlev rat
O -=
m
a
ro
o
in
10 4 10 3 10 2 10 1 1 101 102 103 104
10 4 10 3 10 2 10
1 101 102 103 104
TCE inhalation (ppm)
TCE oral (mg/kg-d)
Figure F-14. Derivation of HECgg and HEDgg corresponding to the rodent
idPOD from Woolhiser et al. (2006) for increased kidney weight in rats.
F.6.5.2. Derivation ofHEC99 andHED99
The F£EC99 and F£ED99 are the lower 99th percentiles for the continuous human exposure
concentration and continuous human ingestion dose that lead to a human internal dose equal to
the rodent idPOD. The derivation of the HEC99 of 0.0332 ppm and HED99 of 0.0482 mg/kg/d for
the 99th percentile for uncertainty and variability are shown in Figure F-15. These values are
used as this critical effect's POD to which additional UFs are applied.
F.6.6. Johnson et al. (2003)—Benchmark Dose (BMD) Modeling of Fetal Heart
Malformations in Rats
The critical endpoint here is increased fetal heart malformations in female S-D rats
(Johnson et al., 2003).
772/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-3 5 DRAFT—DO NOT CITE OR QUOTE
-------
Keil.etal.2009
mouse B
-------
1 for the 99th percentile for uncertainty and variability are shown in Figure F-16. These values are
2 used as this critical effect's POD to which additional UFs are applied.
Johnson.etal.2003
BMDL for developmental heart malformations in
F Soraaue-Daw lev rat
Johnson.etal.2003
BMDL for developmental heart malformations in
FSoraaue-Dawlev rat
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
m
.Q
O -=
m
.Q
O -=
O -
10 4 10 3 10 2 10 1 1 101 102 103 104
TCE inhalation (ppm)
10 510 410 310 210
102 103 104
TCE oral (mg/kg-d)
Figure F-16. Derivation of HECgg and HED99 corresponding to the rodent
idPOD from Johnson et al. (2003) for increased fetal cardiac malformations
in female S-D rats using the total oxidative metabolism dose metric.
F.6.7. Peden-Adams et al. (2006)—Lowest-Observed-Adverse-Effect Level (LOAEL) for
Decreased PFC Response and Increased Delayed-Type Hypersensitivity in Mice
The critical endpoints here are decreased PFC response and increased delayed-type
hypersensitivity in mice exposed pre- and postnatally (Peden-Adams et al., 2006).
Mice were exposed to 1400 and 14,000 ppb in drinking water, with an average dose in
the dams estimated by the authors to be 0.37 and 3.7 mg/kg/d, from GDO to postnatal ages of 3
or 8 weeks. The dose-response relationships were sufficiently supralinear that BMD modeling
failed to produce an adequate fit. In addition, because of the lack of an appropriate PBPK model
and parameters to estimate internal doses given the complex exposure pattern (placental and
lactational transfer, and pup ingestion postweaning), no internal dose estimates were made.
Therefore, the LOAEL of 0.37 mg/kg/d on the basis of applied dose was used as the critical
effect's POD to which additional UFs are applied.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-37 DRAFT—DO NOT CITE OR QUOTE
-------
1 F.7. REFERENCES
2 Box GEP, Hunter WG, Hunter JS. (1978). Statistics for Experimenters, New York: John Wiley & Sons.
3 Buben, JA; O'Flaherty, EJ. (1985) Delineation of the role of metabolism in the hepatotoxicity of trichloroethylene
4 and perchloroethylene: a dose-effect study. Toxicol Appl Pharmacol 78:105-122.
5 Carney, EW; Thorsrud, BA; Dugard, PH; Zablotny, CL. (2006) Developmental toxicity studies in Crl:Cd (SD) rats
6 following inhalation exposure to trichloroethylene and perchloroethylene. Birth Defects Research (Part B)
7 77:405-412.
8 Chia SE, Ong CN, Tsakok MF, Ho A. (1996) Semen parameters in workers exposed to trichloroethylene. Reprod
9 Toxicol 10(4):295-299.
10 Crofton, KM; Zhao, X. (1997) The ototoxicity of trichloroethylene: extrapolation and relevance of high-
11 concentration, short-duration animal exposure data. Fundam Appl Toxicol 38(1):101-106.
12 Filipsson, A.F., K. Victoria (2003). Comparison of available benchmark dose softwares and models using
13 trichloroethylene as a model substance. Regulatory Toxicology and Pharmacology 37:343-355.
14 George, JD; Reel, JR; Myers, CB; Lawton, AD; Lamb, JC. (1986) Trichloroethylene: reproduction and fertility
15 assessment in F344 rats when administered in the feed. RTI Project No. 310-2344, NTP-86-085. National Institute
16 of Environmental Health Sciences, National Toxicology Program, RTF, NC.
17 Griffin JM, Gilbert KM, Lamps LW, Pumford MR. 2000. CD4(+) T-cell activation and induction of autoimmune
18 hepatitis following trichloroethylene treatment in MRL+/+ mice. Toxicol Sci 57:345-352.
19 Ikeda M, Otsuji H, Imamura T, Komoike Y. (1972) Urinary excretion of total trichloro-compounds,
20 trichloroethanol, and trichloroacetic acid as a measure of exposure to trichloroethylene and tetrachloroethylene. Brit
21 JIndMed29(3):328-33.
22 Johnson, PD; Goldberg, SJ; Mays, MZ; Dawson, BV. (2003) Threshold of trichloroethylene contamination in
23 maternal drinking waters affecting fetal heart development in the rat. Environ Health Perspect 111(3):289-292.
24 Keil, DE; Peden-Adams, MM; Wallace, S; Ruiz, P; Gilkeson, GS. (2009) Assessment of trichloroethylene (TCE)
25 exposure in murine strains genetically-prone and non-prone to develop autoimmune disease. Journal of
26 Environmental Science and Health, Part A 44: 443-453.
27 Kjellstrand P, Holmquist B, Aim P, Kanje M, Romare S, Jonsson I, Mansson L, Bjerkemo M. (1983a)
2 8 Trichloroethylene: further studies of the effects on body and organ weights and plasma butyryIcholinesterase
29 activity in mice. Acta Pharmacol Toxicol (Copenh) 53(5):375-84.
30 Kjellstrand P, Holmquist B, Mandahl N, Bjerkemo M. (1983b) Effects of continuous trichloroethylene inhalation on
31 different strains of mice. Acta Pharmacol Toxicol (Copenh) 53(5):369-74.
32 Land, PC; Owen, EL; Linde, HW. (1981) Morphologic changes in mouse spermatozoa after exposure to inhalational
33 anesthetics during early spermatogenesis. Anesthesiology 54:53-56.
34 Maltoni, C; Lefemine, G; Cotti, G. (1986) Experimental research on trichloroethylene carcinogenesis. In: Maltoni,
35 C; Mehlman MA., eds. Vol. 5. Archives of research on industrial carcinogenesis. Princeton, NJ: Princeton Scientific
36 Publishing;
37 Mhiri, C; Choyakh, F; Ben, HM; et al. (2004) Trigeminal somatosensory evoked potentials in trichloroethylene-
38 exposed workers. Neurosciences 9(2): 102-107.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-3 8 DRAFT—DO NOT CITE OR QUOTE
-------
1 Moser, Cheek & MacPhail. A multidisciplinary approach to toxicological screening III. Neurobehavioral toxicity. J
2 Toxicol. Environ. Hlth., 1995, 45, 173-210.
3 Narotsky, MG; Weller, EA; Chinchilli, VM; Kavlock, RJ. (1995) Nonadditive developmental toxicity in mixtures of
4 trichloroethylene, di(2-ethylhexyl) phthalate, and heptachlor in a 5 x 5 x 5 design. Fundam Appl Toxicol
5 27:203-216.
6 NCI (National Cancer Institute). (1976) Carcinogenesis bioassay of trichloroethylene. Division of Cancer Cause and
7 Prevention, National Cancer Institute, U.S. Department of Health, Education, and Welfare, DHEW Publication No.
8 (NIH) 76-802, Technical Report Series No. 2, 218 pages; NCI-CG-TR-2; NTIS PB254122.
9 http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr002.pdf.
10 NTP (National Toxicology Program). (1988) Toxicology and carcinogenesis studies of trichloroethylene (CAS no.
11 79-01-6) in four strains of rats (ACI, August, Marshall, Osborne-Mendel) (gavage studies). Public Health Service,
12 U.S. Department of Health and Human Services; NTP TR-273; NIH Publication No. 88-2529. Available from the
13 National Institute of Environmental Health Sciences, Research Triangle Park, NC, and the National Technical
14 Information Service, Springfield, VA; PB88-218896. http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr273.pdf.
15 NTP (National Toxicology Program). (1990) Carcinogenesis Studies of Trichloroethylene (Without Epichlorhydrin)
16 (CAS No. 79-01-6) in F344/N Rats and B6C3F1 Mice (Gavage Study). NTP TR 243. Research Triangle Park, NC:
17 U.S. Department of Health and Human Services.
18 Peden-Adams MM, Eudaly JG, Heesemann LM, Smythe J, Miller J, Gilkeson GS, et al. (2006). Developmental
19 immunotoxicity of trichloroethylene (TCE): studies in B6C3F1 mice. J Environ Sci Health A Tox Hazard Subst
20 Environ Eng 41:249-271.
21 Snedecor GW, Cochran WG. (1980). Statistical Methods (7th ed.), Ch. 9.12 and Ch 9.14 (pp. 169-172)
22 U.S. EPA (Environmental Protection Agency) (2008b). Benchmark Dose Technical Guidance (Inter-Agency
23 Review Draft).
24 U.S. EPA (Environmental Protection Agency). (1994) Methods for derivation of inhalation reference concentrations
25 and application of inhalation dosimetry. Environmental Criteria and Assessment Office, Office of Health and
26 Environmental Assessment, Washington, Washington, DC; EPA/600/8-90/066F. Available from: National
27 Technical Information Service, Springfield, VA; PB2000-500023.
28 Woolhiser, MR; Krieger, SM; Thomas, J; Hotchkiss, JA. (2006) Trichloroethylene (TCE): Immunotoxicity potential
29 in CD rats following a 4-week vapor inhalation exposure. Dow Chemical Company, Toxicology & Environmental
30 Research and Consulting, Midland, MI, Study ID 031020, July 5, 2006, unpublished.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 F-39 DRAFT—DO NOT CITE OR QUOTE
-------
APPENDIX G
TCE Cancer Dose-Response Analyses with
Rodent Cancer Bioassay Data
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-i DRAFT-DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix G: TCE Cancer Dose-Response Analyses With Rodent Cancer
Bioassay Data
LIST OF TABLES G-iii
LIST OF FIGURES G-v
APPENDIX G: TCE CANCER DOSE-RESPONSE ANALYSES WITH RODENT
CANCER BIO AS SAY DAT A G-6
G.I. DATA SOURCES G-6
G.I.I. Numbers at Risk G-6
G.I.2. Cumulative Incidence G-6
G.2. INTERNAL DOSE METRICS AND DOSE ADJUSTMENTS G-7
G.3. DOSE ADJUSTMENTS FOR INTERMITTENT EXPOSURE G-8
G.4. RODENT TO HUMAN DOSE EXTRAPOLATION G-9
G.5. COMBINING DATA FROM RELATED EXPERIMENTS IN MALTONI
ETAL. (1986) G-10
G.6. DOSE-RESPONSE MODELING RESULTS G-ll
G.7. MODELING TO ACCOUNT FOR DOSE GROUPS DIFFERING IN
SURVIVAL TIMES G-16
G.7.1. Time-to-Tumor Modeling G-17
G.7.2. Poly-3 Calculation of Adjusted Number at Risk G-18
G.8. COMBINED RISK FROM MULTIPLE TUMOR SITES G-19
G.8.1. Methods G-19
G.8.1.1. Single Tumor Sites G-19
G.8.1.2. Combined Risk From Multiple Tumor Sites G-20
G.8.2. Results G-21
G.9. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK)-MODEL
UNCERTAINTY ANALYSIS OF UNIT RISK ESTIMATES G-40
G. 10. REFERENCES G-42
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-ii DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF TABLES
G-1. Internal dose metrics used in dose-response analyses, identified by "X" G-7
G-2. Experiments BT304 and BT304bis, female Sprague-Dawley rats, Maltoni et al.
(1986) G-12
G-3. Experiments BT304 and BT304bis, male Sprague-Dawley rats, Maltoni et al.
(1986): leukemias G-13
G-4. Experiments BT304 and BT304bis, male Sprague-Dawley rats, Maltoni et al.
(1986): kidney adenomas + carcinomas G-14
G-5. Experiments BT304 and BT304bis, male Sprague-Dawley rats, Maltoni et al.
(1986): testis, Ley dig cell tumors G-15
G-6. Rodent to human conversions for internal dose metric TotOxMetabBW34 G-20
G-7. Rodent to human conversions for internal dose metric TotMetabBW34 G-21
G-8. Female B6C3F1 mice—applied doses: data G-22
G-9. Female B6C3F1 mice—applied doses: model selection comparison of model fit
statistics for multistage models of increasing order G-22
G-10. Female B6C3F1 mice—applied doses: BMD and risk estimates (inferences for
BMR of 0.05 extra risk at 95% confidence level) G-23
G-ll. B6C3F1 female mice inhalation exposure—applied doses G-25
G-12. B6C3F1 female mice—applied doses: model selection comparison of model fit
statistics for multistage models of increasing order G-25
G-13. B6C3F1 female mice inhalation exposure—applied doses (inferences for 0.05
extra risk at 95% confidence level) G-26
G-14. Maltoni Sprague-Dawley male rats—applied doses G-28
G-15. Maltoni Sprague-Dawley male rats—applied doses: model selection comparison
of model fit statistics for multistage models of increasing order G-28
G-16. Maltoni Sprague-Dawley male rats—applied doses G-29
G-17. Female B6C3F1 mice—internal dose metric (total oxidative metabolism): data G-31
G-18. Female B6C3F1 mice—internal dose: model selection comparison of model fit
statistics for multistage models of increasing order G-31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-iii DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF TABLES (continued)
G-19. Female B6C3F1 mice—internal dose metric (total oxidative metabolism): BMD
and risk estimates G-32
G-20. B6C3F1 female mice inhalation exposure—internal dose metric (total oxidative
metabolism) G-34
G-21. B6C3F1 female mice—internal dose: model selection comparison of model fit
statistics for multistage models of increasing order G-34
G-22. B6C3F1 female mice inhalation exposure—internal dose metric (total oxidative
metabolism) (inferences for 0.05 extra risk at 95% confidence level) G-35
G-23. Maltoni Sprague-Dawley male rats—internal dose metric (total metabolism) G-37
G-24. Maltoni Sprague-Dawley male rats—internal dose model selection comparison
of model fit statistics for multistage models of increasing order G-37
G-25. Maltoni Sprague-Dawley male rats—internal dose metric (total metabolism)
(inferences for 0.01 extra risk at 95% confidence level) G-38
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-iv DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
G-l. Female B6C3F1 mice—applied doses: combined and individual tumor extra-risk
functions G-24
G-2. Female B6C3F1 mice—applied doses: posterior distribution of BMDc for
combined risk G-24
G-3. B6C3F1 female mice inhalation exposure—applied doses: combined and
individual tumor extra-risk functions G-27
G-4. B6C3F1 female mice inhalation exposure—applied doses: posterior
distribution of BMDc for combined risk G-27
G-5. Maltoni Sprague-Dawley male rats—applied doses: combined and individual
tumor extra-risk functions G-30
G-6. Maltoni Sprague-Dawley male rats—applied doses: posterior distribution of
BMDc for combined risk G-30
G-7. Female B6C3F1 mice—internal dose metric (total oxidative metabolism):
combined and individual tumor extra-risk functions G-33
G-8. Female B6C3F1 mice—internal dose metric (total oxidative metabolism):
posterior distribution of BMDc for combined risk G-33
G-9. B6C3F1 female mice inhalation exposure—internal dose metric: combined
and individual tumor extra-risk functions G-36
G-10. B6C3F1 female mice inhalation exposure—internal dose metric: posterior
distribution of BMDc for combined risk G-36
G-l 1. Maltoni Sprague-Dawley male rats—internal dose metric: combined and
individual tumor extra-risk functions G-39
G-12. Maltoni Sprague-Dawley male rats—internal dose metric: posterior distribution
of BMDc for combined risk G-39
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-v DRAFT-DO NOT CITE OR QUOTE
-------
1 APPENDIX G: TCE CANCER DOSE-RESPONSE ANALYSES WITH RODENT
2 CANCER BIOASSAY DATA
O
4
5 G.I. DATA SOURCES
6 Trichloroethylene (TCE) cancer endpoints were identified in Maltoni et al. (1986),
7 National Cancer Institute (NCI, 1976), National Toxicology Program (NTP, 1988, 1990), Fukuda
8 et al. (1983), and Henschler et al. (1980). These data were reviewed and tabulated in
9 spreadsheets, and the numbers were verified. All endpoint data identified by authors as having a
10 statistically significant response to dose were tabulated, and data that had marginally significant
11 trends with dose were also reviewed. For all endpoints for which dose-response model estimates
12 were presented, trends were verified using the Cochran-Armitage or the Poly-3 test.
13
14 G.I.I. Numbers at Risk
15 The numbers of animals at risk are not necessarily those used by the authors; instead, as
16 the number at risk, the number alive at 52 weeks was used (if the first cancer of the type of
17 interest was observed at later than 52 weeks) or the number alive at the week when the first
18 cancer of the type of interest was observed. In general, the data of Maltoni et al. (1986) were
19 presented in this way, in their tables titled "Incidence of the different types of tumors referred to
20 specific corrected numbers." In a few cases in Maltoni et al. (1986), the time of first occurrence
21 was later than 52 weeks, so an alternative number at risk was used from another column (for
22 another cancer) in the same table having a first occurrence close to 52 weeks. For NTP (1988,
23 1990) and for NCI (1976), the week of the first observation and the numbers alive at that week
24 were determined from the appendix tables. For Fukuda et al. (1983), the reported "effective
25 number of mice" in their Table 2 was used, which is consistent with numbers alive at
26 40-42 weeks (when the first tumor, a thymic lymphoma, was observed) in their mortality curve.
27 For Henschler et al. (1980), the number of mice alive at Week 36 (from their Figure 1), which is
28 when the first tumor was observed (according to their Figure 2), was used.
29
30 G.1.2. Cumulative Incidence
31 Maltoni et al. (1986) conducted a lifetime study, in which rodents were exposed for
32 104 weeks (rats) or 78 weeks (mice), and allowed to live until they died "naturally." Maltoni
33 et al. (1986) reported cumulative incidence on this basis, and it was not possible for us to
34 determine incidence at any fixed time such as 104 weeks on study. For Henschler et al. (1980),
35 the number of mice with tumors observed by Week 104 (their Figure 2) was used. The
36 cumulative incidence reported by Fukuda et al. (1983) at 107 weeks (after 104 weeks of
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
exposure) was used. For the NCI (1976) and NTP (1988, 1990) studies, the reported cumulative
incidence at 103 to 107 weeks (study time varied by study and species) was used.
G.2. INTERNAL DOSE METRICS AND DOSE ADJUSTMENTS
Physiologically based pharmacokinetic (PBPK) modeling was used to estimate levels of
dose metrics corresponding to different exposure scenarios in rodents and humans (see
Section 3.5). The selection of dose metrics for specific organs and endpoints is discussed under
Section 5.2. Internal dose metrics were selected based on applicability to each major affected
organ. The dose metrics used with our cancer dose-response analyses are shown in Table G-l.
Table G-l. Internal dose metrics used in dose-response analyses, identified
by "X"
Dose metric units
ABioactDCVCBW34 (mg/wk-kg3/4)
AMetGSHBW34 (mg/wk-kg3/4)
AMetLivlBW34 (mg/wk-kg3/4)
AMetLngBW34 (mg/wk-kg3/4)
AUCCBld (mg-hr/L-wk)
TotMetabBW34 (mg/wk-kg3/4)
TotOxMetabBW34 (mg/wk-kg3/4)
Liver
0
0
X
0
0
0
X
Lung
0
0
0
X
X
0
X
Kidney
X
X
0
0
0
X
0
Other
0
0
0
0
X
X
0
The PBPK model requires the rodent body weight as an input. For most of the studies,
central estimates specific to each species, strain, and sex (and substudy) were used. These were
estimated by medians of body weights digitized from graphics in Maltoni et al. (1986), by
medians of weekly averages in NTP (1990, 1988), and by averages over the study duration of
weekly mean body weights tabulated in NCI (1976).
For the studies by Fukuda et al. (1983) and Henschler et al. (1980), mouse body weights
were not available. After reviewing body weights reported for similar strains by two
laboratories1 and in the other studies reported for TCE, it was concluded that a plausible range
for lifetime average body weight is 20-35 g, with a median near 28 g. For these two studies,
1http://phenome.jax.org/pub-
cgi/phenome/mpdcgi?rtn=meas%2Fdatalister&req=Cbody+weight&pan=2&noomit=&datamode=measavg,
http://www.hilltoplabs.com/public/growth.html.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-7 DRAFT-DO NOT CITE OR QUOTE
-------
1 internal dose metrics for these three average body weights (20, 28, and 35 g) were computed.
2 The percentage differences between the internal dose metrics for the intermediate body weight
3 (BW) of 28 g and the low and high average BW of 20 gm and 35 g were then evaluated. Internal
4 dose metrics were little affected by choice of body weight. For all dose metrics, the differences
5 were less than ±13%. A body weight of 28 g was used for these two studies.
6 The medians (from the Markov chain Monte Carlo posterior distribution) for each of the
7 dose metrics for the rodent were used in quantal dose-response analyses. The median is probably
8 the most appropriate posterior parameter to use as a dose metric, as it identifies a "central"
9 measure and it is also a quantile, making it more useful in nonlinear modeling. The "multistage"
10 dose-response functions are nonlinear. One is interested in estimating the expected response.
11 The expected value of a nonlinear function of dose is under- or overestimated when the mean
12 (expected value) of the dose is used, depending on whether the function is concave or convex.
13 (This is Jensen's Inequality: for a real convex function f(X), f[E(X)] < E[f(X)].) For the
14 dose-response function, one is interested in E[f(X)], so using E(X) (estimated by the posterior
15 mean) as the dose metric will not necessarily predict the mean response. Using the posterior
16 median rather than the mean as the dose metric should lead to a response function that is closer
17 to the median response. However, if the estimated dose-response function is close to linear, this
18 source of distortion may be small, and the mean response might be predicted reasonably well by
19 using the posterior mean as the dose metric. The mean and median are expected to be rather
20 different because the posterior distributions are skewed and approximately lognormal.
21 Therefore, results based on the posterior median and the posterior mean dose metric were
22 compared before deciding to use the median.
23
24 G.3. DOSE ADJUSTMENTS FOR INTERMITTENT EXPOSURE
25 The nominal applied dose was adjusted for exposure discontinuity (e.g., exposure for
26 5 days per week and 6 hours per day reduced the dose by the factor [(5/7) * (6/24)]), and for
27 exposure durations less than full study time (up to 2 years) (e.g., the dose might be reduced by a
28 factor [78 wk/104 wk]). The PBPK dose metrics took into account the daily and weekly
29 discontinuity to produce an equivalent dose for continuous exposure. The NCI (1976) gavage
30 study applied one dose for weeks 1-12 and another, slightly different dose for weeks 13-78;
31 PBPK dose metrics were produced for both dose regimes and then time-averaged (e.g., average
32 dose = (12/78) x Dl + (66/78) x D2). For Henschler et al. (1980), Maltoni et al. (1986), and NCI
33 (1976), a further adjustment of (exposure duration/study duration) was made to account for the
34 fact that exposures ended prior to terminal sacrifice, so that the dose metrics reflect average
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-8 DRAFT-DO NOT CITE OR QUOTE
-------
1 weekly values over the exposure period. Finally, for NCI (1976), the dose metrics were then
2 adjusted for early sacrifice2 (at 91 weeks rather than 104 weeks) by a factor of (91 wk/104 wk)3.3
O
4 G.4. RODENT TO HUMAN DOSE EXTRAPOLATION
5 Adjustments for rodent-to-human extrapolation were applied to the final results—the
6 benchmark dose (BMD), benchmark dose lower bound (BMDL), and cancer slope factor
7 (potency), which is calculated as benchmark response (BMR)/BMDL, e.g., 0.10/BMDLi0.
8 For the PBPK dose metrics, a ratio between human and laboratory animal internal dose
9 was determined by methods described in Section 3.5. The cancer slope factor is relevant only for
10 very low extra risk (typically on the order of 10~4 to 10"6), thus very low dose, and it was
11 determined that the relation between human and animal internal dose was linear in the low-dose
12 range for each of the dose metrics used, hence this ratio was multiplied by the animal dose (or
13 divided into the cancer slope factor) to extrapolate animal to human dose or concentration.
14 For the experimentally applied dose, default interspecies extrapolation approaches were
15 used. These are provided for comparison to results based on PBPK metrics. To extrapolate
16 animal inhalation exposure to human inhalation exposure, the "equivalent" human exposure
17 concentration (i.e., the exposure concentration in humans that is expected to give the same level
18 of response that was observed in the test species) was assumed to be identical to the animal
19 inhalation exposure concentration, i.e., "ppm equivalence." This assumption is consistent with
20 U.S. Environmental Protection Agency recommendations (U.S. EPA, 1994) for deriving a
21 human equivalent concentration for a Category 3 gas for which the blood:air partition coefficient
22 in laboratory animals is greater than that in humans (see Section 3.1 for discussion of the TCE
23 blood:air partition coefficient). To extrapolate animal oral exposure to equivalent human oral
24 exposure, animal dose was scaled up by body weight to the 3/4-power using the factor
25 (BWRuman/BWAnimai)0'75- To extrapolate animal inhalation exposure to human oral exposure, the
26 following equation (Eq. G-l) was used;4
27
2For studies of less than 2 years (i.e., with terminal kills before 2 years), the doses are generally adjusted by the
study length ratio to a power of three (i.e., a factor [length of study in wk/104 wk]3) to reflect the fact that the
animals were not observed for the full standard lifetime (U.S. EPA, 1980).
3For studies of less than 2 years (i.e., with terminal kills before 2 years), the doses are generally adjusted by the
study length ratio to a power of three (i.e., a factor [length of study in wk/104 wk]3) to reflect the fact that the
animals were not observed for the full standard lifetime (U.S. EPA, 1980).
4ToxRisk version 5.3, © 2000-2001 by the KS Crump Group, Inc.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-9 DRAFT-DO NOT CITE OR QUOTE
-------
1 Animal, equivalent oral intake, mg/kg/d =
2 ppm * [A4WTCE/24A5]5 * MK * (60 min/hr) * (103 mg/g) * [24 hr/BWkg] (Eq. G-l)
O
4 with units
5
6 ppm * [g/mol + L/mol ] * L/min * (min/hr) * (mg/g) * [hr/day + kg] (Eq. G-2)
7
8 which reduces to
9
10 ppm * [7.738307 * MV/BWkg] (Eq. G-3)
11
12 where
13 ppm = animal inhalation concentration, 1/106, unitless
14 MV = minute volume (breathing rate) at rest, L/minute.
15
16 Minute volume (MV) was estimated using equations from U.S. EPA (1994, p. 4-27),
17
18 Mouse ln(MV) = 0.326 + 1.05 * ln(BWkg) (Eq. G-4)
19 Rat ln(MV) = -0.578 + 0.821 * ln(BWkg). (Eq. G-5)
20
21 Animal equivalent oral intake was converted to human equivalent oral intake by
22 multiplying by the rodent to human ratio of body weights to the power +0.25.6
23 To extrapolate animal oral exposure to equivalent human inhalation exposure, the
24 calculation above was reversed to extrapolate the animal inhalation exposure.
25
26 G.5. COMBINING DATA FROM RELATED EXPERIMENTS IN MALTONI ET AL.
27 (1986)
28 Data from Maltoni et al. (1986) required decisions by us regarding whether to combine
29 related experiments for certain species and cancers.
30 In experiment BT306, which used B6C3F1 mice, males experienced unusually low
31 survival, reportedly because of the age of the mice at the outset and resulting aggression. The
5Molecular weight of TCE is 131.39; there are 24.45 L of perfect gas per g-mol at standard temperature and
pressure, U.S. EPA (1994).
6Find whole animal intake from mg/kg/d * BWAmmai- Scale this allometrically by (BWHuman/BWAmmai)°'75 to
extrapolate whole human intake. Divide by human body weight to find mg/kg/d for the human. The net effect is
Animal mg/kg/d * (BWAmmai/BWHuman)A0.25 = Human mg/kg/d.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-10 DRAFT-DO NOT CITE OR QUOTE
-------
1 protocol was repeated (for males only), with an earlier starting age, as experiment BT306bis, and
2 male survival was higher (and typical for such studies). The rapid male mortality in experiment
3 BT306 apparently censored later-developing cancers, as suggested by the low frequency of liver
4 cancers for males in BT306 as compared to BT306bis. Data for the two experiments clearly
5 cannot legitimately be combined. Therefore only experiment BT306bis males were used in the
6 analyses.
7 Experiments BT304 and BT304bis, on rats, provide evidence in male rats of leukemia,
8 carcinomas of the kidney, and testicular (Leydig cell) tumors, and provide evidence in female
9 rats for leukemia. Maltoni et al. (1986, p. 46) stated "Since experiments BT 304 and BT 304bis
10 on Sprague-Dawley rats were performed at the same time, exactly in the same way, on animals
11 of the same breed, divided by litter distribution within the two experiments, they have been
12 evaluated separately and comprehensively." The data were also analyzed separately and in
13 combination.
14 The data and modeling results for these tumors in the BT304 and BT304bis experiments
15 are tabulated in Tables G-2 through G-5, below. It was decided that it was best to combine the
16 data for the two experiments. There were no consistent differences between experiments, and no
17 firm basis for selecting one of them. Our final analyses are, therefore, based on the combined
18 numbers and tumor responses for these two experiments.
19
20 G.6. DOSE-RESPONSE MODELING RESULTS
21 Using BenchMark Dose Software (BMDS), the multistage quantal model was fitted using
22 the applicable dose metrics for each combination of study, species, strain, sex, organ, and BMR
23 (extra risk) value under consideration. A multistage model of order one less than the number of
24 dose groups (g) was fitted. This means that in some cases the fitted model could be strictly
25 nonlinear at low dose (estimated coefficient "bl" was zero), and in other cases, higher-order
26 coefficients might be estimated as zero so the resulting model would not necessarily have order
27 (#groups-l). Because more parsimonious, lst-order models often fit such data well, based on our
28 extensive experience and that of others (Nitcheva et al., 2007), if the resulting model was not a
29 lst-order multistage, then lower-order models were also fitted, down to a lst-order multistage
30 model. This permitted us to screen results efficiently.
31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-l 1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
Table G-2. Experiments BT304 and BT304bis, female Sprague-Dawley rats,
Maltoni et al. (1986). Number alive is reported for week of first tumor
observation in either males or females.a These data were not used for
dose-response modeling because there is no consistent trend (for the combined
data, there is no significant trend by the Cochran-Armitage test, and no significant
differences between control and dose groups by Fisher's exact test).
Exposure
Concen.
(ppm)
0
100
300
600
0
100
300
600
0
100
300
600
No.
alive
No. rats
with this
cancer
Proportion
with
cancer
Multistage model fit statistics11
Model
order
/7-Value
AIC
BMDio
BMDLio
Experiment BT304, female rats, leukemias, TV alive at 7 weeks
105
90
90
90
7
6
0
7
0.067
0.067
0.000
0.078
No adequately fitting model
Experiment BT304bis, female rats, leukemias, TV alive at 7 weeks
40
40
40
40
0
3
2
4
0.000
0.075
0.050
0.100
1
0.202
70.4
127
58.7
Experiments BT304 and BT304bis, female rats, leukemias, combined data
145
130
130
130
7
9
2
11
0.048
0.069
0.015
0.085
3
0.081
227
180
134
1
9
10
11
12
13
14
15
a First tumor occurrences were not reported separately by sex.
b Models of orders 3 were fitted; the highest-order nonzero coefficient is reported in column "Model order."
BMDL was estimated for extra risk of 0.10 and confidence level 0.95. Exposure concentrations were multiplied
by (7/24) * (5/7) = 0.20833 before fitting the models, to adjust for exposure periodicity (i.e., the time-averaged
concentrations were about 20% of the nominal concentrations).
AIC - Akaike Information Criteria.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
Table G-3. Experiments BT304 and BT304bis, male Sprague-Dawley rats,
Maltoni et al. (1986): leukemias. Number alive is reported for week of first
tumor observation in either males or females.a
Exposure
concen.
(ppm)
0
100
300
600
0
100
300
600
0
100
300
600
No.
alive
No. rats
with this
cancer
Proportion
with
cancer
Multistage model fit statistics11
Model
order
/7-Value
AIC
BMDio
BMDLio
Experiment BT304, male rats, leukemias, TV alive at 7 weeks
95
90
90
89
6
10
11
9
0.063
0.111
0.122
0.101
1
0.429
238
NA
NA
Experiment BT304bis, male rats, leukemias, TV alive at 7 weeks
39
40
40
40
3
3
O
6
0.077
0.075
0.075
0.150
3
0.979
102
143
71.9
Combined data for BT304 and BT304bis, male rats, leukemias
134
130
130
129
9
13
14
15
0.067
0.100
0.108
0.116
1
0.715
337
269
111
5
6
7
8
9
10
11
12
13 AIC—Akaike Information Criteria.
Tirst tumor occurrences were not reported separately by sex.
bModels of orders 3 were fitted; the highest-order nonzero coefficient is reported in column "Model order." BMDL
was estimated for extra risk of 0.10 and confidence level 0.95. Exposure concentrations were multiplied by
(7/24)*(5/7) = 0.20833 before fitting the models, to adjust for exposure periodicity (i.e., the time-averaged
concentrations were about 20% of the nominal concentrations). "NA" indicates the BMD or BMDL could not be
solved because it exceeded the highest dose.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-13 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
Table G-4. Experiments BT304 and BT304bis, male Sprague-Dawley rats,
Maltoni et al. (1986): kidney adenomas + carcinomas. Number alive is
reported for week of first tumor observation in either males or females.a
Exposure
concen.
(ppm)
0
100
300
600
0
100
300
600
0
100
300
600
No.
alive
No. rats
with this
cancer
Proportion
with cancer
Multistage model fit statistics11
Model
order
/7-Value
AIC
BMDio
BMDLio
Experiment BT304 male rats, kidney adenomas + carcinomas, TV alive at 47 weeks
87
86
80
85
0
1
0
4
0.000
0.012
0.000
0.047
3
0.318
50.1
173
134
Experiment BT304bis, male rats, kidney adenomas + carcinomas, TV alive at 53
weeks
34
32
36
38
0
0
0
1
0.000
0.000
0.000
0.027
3
0.988
13.0
266
173
Combined data for BT304 and BT304bis, male rats, kidney adenomas + carcinomas
121
118
116
123
0
1
0
5
0.000
0.008
0.000
0.041
3
0.292
60.5
181
144
6 a First tumor occurrences were not reported separately by sex.
7 b Models of orders three were fitted; the highest-order nonzero coefficient is reported in column "Model order."
8 BMDL was estimated for extra risk of 0.10 and confidence level 0.95. Exposure concentrations were multiplied by
9 (7/24)*(5/7) = 0.20833 before fitting the models, to adjust for exposure periodicity (i.e., the time-averaged
10 concentrations were about 20% of the nominal concentrations). "NA" indicates the BMD or BMDL could not be
11 solved because it exceeded the highest dose.
12
13 AIC - Akaike Information Criteria.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
Table G-5. Experiments BT304 and BT304bis, male Sprague-Dawley rats,
Maltoni et al. (1986): testis, Leydig cell tumors. Number alive is reported for
week of first tumor observation.21
Exposure
concen.
(ppm)
0
100
300
600
0
100
300
600
0
100
300
600
No.
alive
No. rats
with this
cancer
Proportion
with
cancer
Multistage model fit statistics11
Model
order
/7-Value
AIC
BMDio
BMDLio
Experiment BT304, male rats, Leydig cell tumors, TV alive at 47 weeks
87
86
80
85
5
11
24
22
0.057
0.128
0.300
0.259
1
0.0494
309
41.5
29.2
Experiment BT304bis, male rats, Leydig cell tumors, TV alive at 53 weeks
34
32
36
38
1
5
6
9
0.029
0.156
0.167
0.237
1
0.369
117
54.5
30.9
Combined data for BT304 and BT304bis, male rats, Leydig cell tumors
121
116
116
122
6
16
30
31
0.050
0.138
0.259
0.254
1
0.0566
421
44.7
32.7
5
6
7
8
9
10
11
12
13
14
a Numbers alive reported for weeks as close as possible to Week 52 (first tumors observed at weeks 81, 62,
respectively, for the two experiments).
b Models of orders three were fitted; the highest-order nonzero coefficient is reported in column "Model order."
BMDL was estimated for extra risk of 0.10 and confidence level 0.95. Exposure concentrations were multiplied
by (7/24)*(5/7) = 0.20833 before fitting the models, to adjust for exposure periodicity (i.e., the time-averaged
concentrations were about 20% of the nominal concentrations). "NA" indicates the BMD or BMDL could not be
solved because it exceeded the highest dose.
AIC - Akaike Information Criteria.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-15 DRAFT-DO NOT CITE OR QUOTE
-------
1 The document Appendix.linked.files\AppG.Cancer.Rodents.Plots.TCE.DRAFT.pdf
2 shows the fitted model curves. The graphics include observations (as proportions, i.e.,
3 cumulative incidence divided by number at risk), the estimated multistage curve (solid red line)
4 and estimated BMD, with a BMDL. Vertical bars show 95% confidence intervals for the
5 observed proportions. Printed above each plot are some key statistics (necessarily rounded) for
6 model goodness of fit and estimated parameters. Printed in the plots at upper left are the BMD
7 and BMDL for the rodent data, in the same units as the rodent dose. Within the plot at lower
8 right are human exposure values (BMDL and cancer slope factor for continuous inhalation and
9 oral exposures) corresponding to the rodent BMDL. For applied doses, the human equivalent
10 values were calculated by "default" methods,7 as discussed above, and then only for the same
11 route of exposure as the rodent, and they are in units of rodent dose. For internal dose metrics,
12 the human values are based upon the PBPK rodent-to-human extrapolation, as discussed in
13 Section 5.2.1.2.
14 The document Appendix.linked.files\AppG.Cancer.Rodents.Results.TCE.DRAFT.pdf
15 presents the data and model summary statistics, including goodness-of-fit measures (Chi-square
16 goodness-of-fit/7-value, Akaike Information Criteria [AIC]), parameter estimates, BMD, BMDL,
17 and "cancer slope factor" ("CSF"), which is the extra risk divided by the BMDL. Much more
18 descriptive information appears also, including the adjustment terms for intermittent exposure,
19 and the doses before applying those adjustments. The group "GRP" numbers are arbitrary, and
20 are the same as GRP numbers in the plots. There is one line in this table for each dose-response
21 graph in the preceding document. Input data for the analyses are in the file
22 Appendix.linked.files\AppG.Cancer.Rodents.Input.Data.TCE.DRAFT.pdf Finally, the values
23 and model selections for the results used in Section 5.2 are summarized in the file
24 Appendix.linked.files\AppG.Cancer.Rodents.model.selections.TCE.DRAFT.pdf (primary dose
25 metrics in bold).
26
27 G.7. MODELING TO ACCOUNT FOR DOSE GROUPS DIFFERING IN SURVIVAL
28 TIMES
29 Differential mortality among dose groups can potentially interfere with (i.e., censor) the
30 occurrence of late-appearing cancers. Usually the situation is one of greater mortality rates at
31 higher doses, caused by toxic effects, or, sometimes, by cancers other than the cancer of interest.
32 Statistical methods of estimation (for the cancer of interest) in the presence of competing risks
33 assume uninformative censoring.
7For oral intake, dose (BMDL) is multiplied by the ratio of animal to human body weight (60 kg female, 70 kg male)
taken to the 1A power. For inhalation exposures, ppm equivalence is assumed.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-16 DRAFT-DO NOT CITE OR QUOTE
-------
1 For bioassays with differential early mortality occurring primarily before the time of the
2 1st tumor or 52 weeks (whichever came first), the effects of early mortality were largely
3 accounted for by adjusting the tumor incidence for animals at risk, as described above, and the
4 dose-response data were modeled using the multistage model.
5 If, however, there was substantial overlap between the appearances of cancers and
6 progressively differential mortality among dose groups, it was necessary to apply methods that
7 take into account individual animal survival times. Two such methods were used here:
8 time-to-tumor modeling and the poly-3 method of adjusting numbers at risk. Three such studies
9 were identified, all with male rats (see Table 5-27). Using both survival-adjustment approaches,
10 BMDs and BMDLs were obtained and unit risks derived. Section 5.2.1.3 presents a comparison
11 of the results for the three data sets and for various dose metrics.
12
13 G.7.1. Time-to-Tumor Modeling
14 The first approach used to take into account individual survival times was application of
15 the multistage Weibull (MSW) time-to-tumor model. This model has the general form
16
17 P(d,0=1- exp[-(?o + qid + q2d2 + ...+ q^) * (t- r0)z], (Eq. G-6)
18
19 where P(d,t) represents the probability of a tumor by age t for dose d, and parameters z > 1,
20 t0 > 0, and qt > 0 for / = 0,1,.. .,&, where k = the number of dose groups; the parameter t0
21 represents the time between when a potentially fatal tumor becomes observable and when it
22 causes death. The MSW model likelihood accounts for the left-censoring inherent in
23 "Incidental" observations of nonfatal tumors discovered upon necropsy and the right-censoring
24 inherent in deaths not caused by fatal tumors. All of our analyses used the model for incidental
25 tumors, which has no t0 term, and which assumes that the tumors are nonfatal (or effectively so,
26 to a reasonable approximation). This seems reasonable because the tumors of concern appeared
27 relatively late in life and there were multiple competing probable causes of death (especially
28 toxic effects) operating in these studies (also note that cause of death was not reported by the
29 studies used). It is difficult to formally evaluate model fit with this model because there is no
30 applicable goodness-of-fit statistic with a well-defined asymptotic distribution. However, plots
31 of fitted vs. observed responses were examined.
32 A computer program ("MSW") to implement the multistage Weibull time-to-tumor
33 model was designed, developed and tested for U.S. EPA by Battelle Columbus (Ohio). The
34 MSW program obtains maximum likelihood estimates for model parameters and solves for the
35 BMDL (lower confidence limit for BMD) using the profile-likelihood method. The model, with
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-17 DRAFT-DO NOT CITE OR QUOTE
-------
1 documentation for methodology (statistical theory and estimation, and numerical algorithms) and
2 testing, was externally reviewed by experts in June 2007. Reviews were generally positive and
3 confirmed that the functioning of the computer code has been rigorously tested. (U.S. EPA and
4 Battelle confirmed that MSW gave results essentially identical to those of "ToxRisk," a program
5 no longer commercially issued or supported.) U.S. EPA's BMDS Web site provided reviewers'
6 comments and U.S. EPA's responses.8 The MSW program and reports on statistical and
7 computational methodology and model testing will be made available in 2009 (after
8 implementing some changes to reporting features and error-handling).
9 Results of this modeling are shown in the file
10 Appendix.linked.files\AppG.Cancer.Rodents.TimetoTumor.Results.TCE.DRAFT.pdf.
11
12 G.I.2. Poly-3 Calculation of Adjusted Number at Risk
13 To obtain an independent estimate of a point of departure using different assumptions, it
14 was thought desirable to compare time-to-tumor modeling to an alternative survival-adjustment
15 technique, "poly-3 adjustment" (Portier and Bailer, 1989), applied to the same data. This
16 technique was used to adjust the tumor incidence denominators based on the individual animal
17 survival times. The adjusted incidence data then served as inputs for U.S. EPA's BMDS
18 multistage model, and multistage model selection was conducted as described in Section 5.2.
19 A detailed exposition is given by Piegorsch and Bailer (1997), Section 6.3.2. Each
20 tumor-less animal is weighted by its fractional survival time (survival time divided by the
21 duration of the bioassay) raised to the power of 3 to reflect the fact that animals are at greater
22 risk of cancer at older ages. Animals with tumors are given a weight of 1. The sum of the
23 weights of all the animals in an exposure group yields the effective survival-adjusted
24 denominator. The "default" power of 3 (thus, "poly-3") was assumed, which was found to be
25 representative for a large number of cancer types (Portier et al., 1986). Algebraically,
26
27 Nadj = ^,w, (Eq.G-7)
28
8At http://www.epa.gov/ncea/bmds/response.html under title "2007 External Review of New Quanta! Models;" use
links to comments and responses.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-18 DRAFT-DO NOT CITE OR QUOTE
-------
1 where
2 wt = 1 if tumor is present
3 Wj = (tj/Tf if tumor is absent at time of death (£)
4 T = duration of study. TV was rounded to the nearest integer.9
5
6 Calculations are reproduced in the spreadsheets linked above.
7
8 G.8. COMBINED RISK FROM MULTIPLE TUMOR SITES
9 For bioassays that exhibited more than one type of tumor response in the same sex and
10 species (these studies have a row for "combined risk" in the "Endpoint" column of Table 5-27,
11 Section 5.2), the cancer potency for the different tumor types combined was estimated. The
12 combined tumor risk estimate describes the risk of developing tumors for any (not all together)
13 of the tumor types that exhibited a TCE-associated tumor response; this estimate then represents
14 the total excess cancer risk. The model for the combined tumor risk is also multistage, with the
15 sum of the stage-specific multistage coefficients from the individual tumor models serving as the
16 stage-specific coefficients for the combined risk model (i.e., for each
17 q, a =q + a + ... + a , where the q s are the coefficients for the powers of dose and k is
1f i [combined] 1i\ 2i2 *ik 1i r
18 the number of tumor types being combined) (Bogen, 1990; NRC, 1994). This model assumes
19 that the occurrences of two or more tumor types are independent. The resulting model equation
20 can be readily solved for a given BMR to obtain a maximum likelihood estimate (BMD) for the
21 combined risk. However, the confidence bounds for the combined risk estimate are not
22 calculated by available modeling software. Therefore, a Bayesian approach was used to estimate
23 confidence bounds on the combined BMD. This approach was implemented using the freely
24 available WinBUGS software (Spiegelhalter et al., 2003), which applies Markov chain Monte
25 Carlo computations. Use of WinBUGS has been demonstrated for derivation of a distribution of
26 BMDs for a single multistage model (Kopylev et al., 2007) and can be straightforwardly
27 generalized to derive the distribution of BMDs for the combined tumor load.
28
29 G.8.1. Methods
30 G.8.1.1. Single Tumor Sites
31 Cancer dose-response models were fitted to data using BMDS. These were multistage
32 models with coefficients constrained to be non-negative. The order of model fitted was (g- 1),
9Notice that the assumptions required for significance testing and estimating variances of parameters are changed by
this procedure. The Williams-Bieler variance estimator is described by Piegorsch and Bailer (1997). Our multistage
modeling did not take this into account, so the resulting BMDL may be somewhat lower than could be obtained by
more laborious calculations.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-19 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
where g is the number of dose groups. For internal dose metrics, the values shown in tables
above were used.
The multistage model was modified for U.S. EPA NCEA by Battelle (under contract
EPC04027) to provide model-based estimates of extra risk at a user-specified dose and
profile-likelihood confidence intervals for that risk. Thus, confidence intervals for extra risk in
addition to BMDs could be reported.
G.8.1.2. Combined Risk From Multiple Tumor Sites
The multistage model identified by BMDS10 was used in a WinBUGS script to generate
posterior distributions for model parameters, the BMD and extra risk at the same dose specified
for the BMDS estimates. The burn-in was of length 10,000, then 100,000 updates were made
and thinned to every 10th update for sample monitoring. From a WinBUGS run, the sample
histories, posterior distribution plots, summary statistics, and codas were archived.
Codas were then imported to R and processed using R programs to compute BMD and
the extra risk at a specific dose for each tumor type. BMD and extra risk for the combined risk
function (assuming independence) were also computed following Bogen.n Results were
summarized as percentiles, means, and modes (modes were based upon the smoothed posterior
distributions). The extra risks across tumor types at a specific dose (10 or 100 was used) were
also summed.
BMDLs for rodent internal doses, reported below, were converted to human external
doses using the conversion factors in Tables G-6 and G-7 (based on PBPK model described in
Section 3.5).
Table G-6. Rodent to human conversions for internal dose metric
TotOxMetabBW34
Route
Inhalation, ppm
Oral, mg/kg/d
Sex
F
M
F
M
Human (mean)
9.843477
9.702822
15.72291
16.4192
10The highest-order model was used, e.g., if BMDS estimates were gamma = 0, beta.l > 0, beta.2 = 0, beta.3 > 0, the
model in WinBUGS allowed beta.2 to be estimated (rather than being fixed at zero).
nBogen, K.T. 1990. Uncertainty in Environmental Health Risk Assessment. London: Taylor & Francis
[Chapter IV]. NRC (National Research Council). 1994. Science and Judgement in Risk Assessment. Washington,
DC: National Academy Press [Chapter 11, Appendix 1-1, Appendix 1-2].
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-20 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-7. Rodent to human conversions for internal dose metric
TotMetabBW34
4
5
6
7
Route
Inhalation, ppm
Oral, mg/kg/d
Sex
F
M
F
M
Human (mean)
11.84204
11.69996
18.76327
19.6
The application of rodent to human conversion factors is as follows:
Given rodent internal dose/) in some units of TotOxMetabBW34, divide by tabled value Y
above to find human exposure in ppm or mg/kg/d.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Example: ppm (human) = Z)(rodent)/7
ppm (human female mean) = 500 (internal units)/9.843477
= 50.80 ppm
(Eq. G-8)
G.8.2. Results
The results follow in this order:
Applied doses
NCI, 1976, Female B6C3F1 mice, oral gavage, liver and lung tumors and lymphomas
(see Tables G-8 through G-10 and Figures G-l and G-2)
Maltoni, 1986, Female B6C3F1 mice, inhalation (expt. BT306), liver and lung tumors
(see Tables G-l 1 through G-l3 and Figures G-3 and G-4)
Maltoni, 1986, Male Sprague-Dawley rats, inhalation (expt. BT304), kidney tumors,
testis Ley dig Cell tumors, and lymphomas (see Tables G-l 4 through G-l 6 and
Figures G-5 and G-6)
Internal Doses
NCI, 1976, Female B6C3F1 mice, oral gavage, liver and lung tumors and lymphomas
(see Tables G-l7 through G-l9 and Figures G-7 and G-8)
Maltoni, 1986, Female B6C3F1 mice, inhalation (expt. BT306), liver and lung tumors
(see Tables G-20 through G-22 and Figures G-9 and G-10)
Maltoni, 1986, Male Sprague-Dawley rats, inhalation (expt. BT304), kidney tumors,
Testis Leydig Cell tumors, and lymphomas (see Tables G-23 through G-25 and
Figures G-11 and G-l2)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-21 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-8. Female B6C3F1 mice—applied doses: data
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Dosea
0
356.4
713.3
Nb
18
45
41
Liver
hepatocellular
carcinomas
0
4
11
Lung
adenomas +
carcinomas
1
4
7
Hematopoietic
lymphomas +
sarcomas
1
5
6
a Doses were adjusted by a factor 0.41015625, accounting for exposure 5/7 days/week, exposure duration 78/91
weeks, and duration of study (91/104)A3. Averaged applied gavage exposures were low-dose 869 mg/kg/d,
high dose 1,739 mg/kg/d.
b Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study.
Source: NCI (1976).
Table G-9. Female B6C3F1 mice—applied doses: model selection
comparison of model fit statistics for multistage models of increasing order
* Largest in absolute value.
Source: NCI (1976).
Tumor site
Liver
Lung
Lymphomas + sarcomas
Model
order,
*selected
2
1*
2
1*
2
1*
Coeff.
estimates
equal
zero
y
y
NA
NA
P2
NA
AIC
78.68
77.52
78.20
76.74
77.28
77.28
Largest*
scaled
residual
0
-0.711
0
-0.551
0.113
0.113
Goodness
of fit
/j-value
1
0.6698
1
0.4649
0.8812
0.8812
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-10. Female B6C3F1 mice—applied doses: BMD and risk estimates
(inferences for BMR of 0.05 extra risk at 95% confidence level)
Parameters used in model
p- Value for HMDS model
BMD05 (from BMDS)
BMD05 (median, mode— WinBUGS)
BMDL (BMDS)*
BMDL (5th percentile, WinBUGS)
BMDos for combined risk (median,
mode, from WinBUGS)
BMDL for combined risk (5th
percentile, WinBUGS)
Liver
hepatocellular
carcinomas
qO,ql
0.6698
138.4
155.5, 135.4
92.95
97.48
Lung
adenomas +
carcinomas
qO,ql
0.6611
295.2
314.5,212.7
144.3
150.7
Hematopoietic
lymphomas +
sarcomas
qO,ql
0.8812
358.8
352.3,231.7
151.4
157.7
84.99, 78.95
53.61
BMDS maximum likelihood risk estimates
Risk at dose 100
Upper 95% CL
Sum of risks at dose 100
0.03640
0.05749
0.01722
0.03849
0.01419
0.03699
0.06781
WinBUGS Bayes risk estimates
Risk at dose 100: mean, median
Upper 95% CL
Comb, risk at dose 100 mean, median
Comb, risk at dose 100, upper 95% CL
0.0327, 0.0324
0.0513
0.0168,0.0161
0.0334
0.0152,0.0143
0.0319
0.06337, 0.0629
0.09124
4
5
6
7
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
Source: NCI (1976).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-23 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
_
en
ro
-i— *
x
LU
CM
O
CO
O
p
CD
O
O
- vertical solid, BMDc
i-- vertical dash, BMDLc
0
50
100
Dose
150
200
Figure G-l. Female B6C3F1 mice—applied doses: combined and individual
tumor extra-risk functions.
CO
c
0
Q
o
CM
O
LO
p
o
LO
O
O
d
o
o
p
o __
T
I I
200 400
600 800
1000
N = 300000 Bandwidth = 1.602
5
6
7
Figure G-2. Female B6C3F1 mice—applied doses: posterior distribution of
BMDc for combined risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-24 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-ll. B6C3F1 female mice inhalation exposure—applied doses
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Dosea
0
15.6
46.9
93.8
Liver hepatomas/Arb
3/88
4/89
4/88
9/85
Lung adenomas +
carcinomas/A^
2/90
6/90
7/89
14/87
'Doses adjusted by a factor 0.133928571, accounting for exposure 7/24 hours/day x 5/7 days/week, and
exposure duration 78/104 weeks. Applied doses were 100, 300, and 600 ppm.
' Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study.
Source: Maltoni (1986).
Table G-12. B6C3F1 female mice—applied doses: model selection
comparison of model fit statistics for multistage models of increasing order
Tumor Site
Liver
Lung
Model
order,
*selected
3
2
1*
3
2
1*
Coeff.
estimates
equal zero
P2
PI
NA
P2
P2
NA
AIC
154.91
153.02
153.47
195.91
193.91
193.91
Largest*
scaled
residual
0.289
0.330
-0.678
0.741
0.714
0.714
Goodness
of fit
/j-value
0.7129
0.8868
0.7223
0.3509
0.6471
0.6471
*Largest in absolute value.
Source: Maltoni (1986).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-25 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-13. B6C3F1 female mice inhalation exposure—applied doses
(inferences for 0.05 extra risk at 95% confidence level)
Parameters used in model
p- Value for BMDS model
BMD05 (from BMDS)
BMD05 (median, mode— WinBUGS)
BMDL (BMDS)*
ms combo.exe BMDosC, BMDLc
BMD05 (5th percentile, WinBUGS)
BMDos for combined risk (median,
mode, from WinBUGS)
BMDL for combined risk (5th percentile,
WinBUGS)
Liver hepatomas
qO,ql
0.7223
72.73
71.55,56.79
37.13
Lung adenomas +
carcinomas
qO,ql
0.06471
33.81
34.49,31.65
21.73
32.12, 16.22
37.03
22.07
23.07,20.39
15.67
BMDS maximum likelihood risk estimates
Risk at dose 10
Upper 95% CL
Sum of risks at dose 10
0.0070281
0.0151186
0.0150572
0.0250168
0.0220853
WinBUGS Bayes risk estimates: means (medians)
Risk at dose 10: mean, median
Upper 95% CL
Comb, risk at dose 10: mean, median
Comb, risk at dose 10: upper 95% CL
0.007377, 0.007138
0.01374
0.01489, 0.01476
0.02
0.02216, 0.02198
0.03220
4
5
6
7
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
Source: Maltoni (1986).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-26 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
o
CO
o .
— vertical solid, BMDc
- - vertical dash, BMDLc
.£2 CN
01
0
50
100
150
200
Dose
Figure G-3. B6C3F1 female mice inhalation exposure—applied doses:
combined and individual tumor extra-risk functions.
CO
c
CD
Q
CD
0
O
p
o
CM
O
O
O
I
T
T
I
0 100 200 300
N = 300000 Bandwidth = 0.4731
Figure G-4. B6C3F1 female mice inhalation exposure—applied doses:
posterior distribution of BMDc for combined risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-27 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-14. Maltoni Sprague-Dawley male rats—applied doses
Dosea
0
20.8
62.5
125
Kidney adenomas
+ carcinomas/A^
0/121
1/118
0/116
5/123
Leukemias/Arb
9/134
13/130
14/130
15/129
Testis, Leydig
cell tumors/A^
6/121
16/116
30/116
31/122
4
5
6
7
a Doses adjusted by a factor 0.208333333, accounting for exposure 7 hours/day x 5/7 days/week. Applied doses
were 100, 300, and 600 ppm.
b Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study.
9
10
11
12
13
14
15
Table G-15. Maltoni Sprague-Dawley male rats—applied doses: model
selection comparison of model fit statistics for multistage models of
increasing order
Tumor site
Kidney
Leukemia
Dropping high dose
Testis
Dropping high dose
Model
order*
3
2
1*
3
2
1
2
1*
3
2
1
2
1*
Coeff.
estimates
equal
zero
P1,P2
y
y
P2,P3
P2
NA
P2
NA
P2, P3
P2
NA
P2
NA
AIC
60.55
61.16
59.55
336.8
336.8
336.8
243.7
243.7
421.4
421.4
421.4
277.6
277.6
Largest+
scaled
residual
1.115
-1.207
-1.331
0.537
0.537
0.537
0.512
0.512
-1.293
-1.293
-1.293
0.291
0.291
Goodness
of fit
/j-value
0.292
0.253
0.4669
0.715
0.715
0.715
0.529
0.529
0.057
0.057
0.057
0.728
0.728
* Model order selected + largest in absolute value
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-16. Maltoni Sprague-Dawley male rats—applied doses
Parameters used in models
p- Value for BMDS model
BMDoi (from BMDS)
BMDoi (median, mode— WinBUGS)
BMDL (BMDS)*
BMDL (5th percentile, WinBUGS)
BMDoi for combined risk (median,
mode, from WinBUGS)
BMDL for combined risk (5th
percentile, WinBUGS)
Kidney
adenomas +
carcinomas
qO,ql
0.4669
41.47
46.00,35.71
22.66
23.23
Leukemia
(high dose
dropped)
qO,ql
0.5290
14.5854
12.32, 8.021
5.52597
5.362
Testis, Leydig
cell tumors (high
dose dropped)
qO,ql
0.7277
2.46989
2.497, 2.309
1.77697
1.789
1.960, 1.826
1.437
BMDS maximum likelihood risk estimates
Risk at dose 10
Upper 95% CL
Sum of risks at dose 10
Risk at dose 1
Upper 95% CL
Sum of risks at dose 1
0.0024208
0.0048995
0.0068670
0.0202747
0.0398747
0.0641010
0.0002423
0.0004911
0.0006888
0.0020462
0.0040609
0.0066029
WinBUGS Bayes risk estimates: means (medians)
Risk at dose 10: mean, median
Upper 95% CL
Comb, risk at dose 10, mean, median
Comb, risk at dose 10, upper 95% CL
Risk at dose 1: mean, median
Upper 95% CL
Comb, risk at dose 1, mean, median
Comb, risk at dose 1, upper 95% CL
0.002302,
0.002182
0.004316
0.008752,
0.008120
0.01860
0.03961, 0.03945
0.05462
0.05020, 0.04998
0.06757
2.305e-04,
2.184e-04
4.325e-04
8.800e-04,
8.150e04
1.876e-03
0.004037,
0.004017
0.005601
0.005143,0.005114
0.006971
3
4
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-29 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
vertical solid, BMDc
vertical dash, BMDLc
Figure G-5. Maltoni Sprague-Dawley male rats—applied doses: combined
and individual tumor extra-risk functions.
5
6
7
c
CD
Q
cq
o
CD
CD
CNI
o
p
o
I
I
T
I
T
T
2 4 6 8 10 12 14
N = 300000 Bandwidth = 0.03059
Figure G-6. Maltoni Sprague-Dawley male rats—applied doses: posterior
distribution of BMDc for combined risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-30 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
4
5
6
7
8
9
10
11
12
13
14
15
Table G-17. Female B6C3F1 mice—internal dose metric (total oxidative
metabolism): data
Internal dose"
0
549.8
813.4
JV"
18
45
41
Liver
hepatocellular
carcinomas
0
4
11
Lung
adenomas +
carcinomas
1
4
7
Hematopoietic
lymphomas +
sarcomas
1
5
6
"Internal dose, Total Oxidative Metabolism, adjusted for body weight, units [mg/(wk-kg3/4)]. Internal doses were
adjusted by a factor 0.574219, accounting for exposure duration 78/91 weeks, and duration of study
(91/104)A3. Before adjustment, the median internal doses were 957.48 and 1416.55 (mg/wk-kg374).
^Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study.
Source: NCI (1976).
Table G-18. Female B6C3F1 mice—internal dose: model selection
comparison of model fit statistics for multistage models of increasing order
Tumor site
Liver
Lung
Lymphomas + sarcomas
BMD,
BMDL
505, 284
367, 245
742, 396
780, 380
870, 389
839, 390
Model
order*
2*
1
2*
1
2
1*
Coeff.
estimates
equal zero
Y,PI
y
PI
NA
NA
NA
AIC
77.25
78.86
76.33
76.74
79.26
77.27
Largest+
scaled
residual
-0.594
-1.083
-0.274
-0.551
0
-0.081
Goodness
of fit
/j-value
0.7618
0.3542
0.7197
0.4649
1
0.9140
16
17
18
19
* Model order selected + largest in absolute value.
Source: NCI (1976).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-31 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
Table G-19. Female B6C3F1 mice—internal dose metric (total oxidative
metabolism): BMD and risk estimates (values rounded to 4 significant
figures) (inferences for BMR of 0.05 extra risk at 95% confidence level)
Parameters used in models
p- Value for HMDS model
BMD05 (from BMDS)
BMD05 (median, mode from WinBUGS)
BMDL (BMDS)*
BMDL (5th percentile, WinBUGS)
BMD05 for Combined Risk (median,
mode, from WinBUGS)
BMDL for Combined Risk (5th percentile,
WinBUGS)
Liver
hepatocellular
carcinomas
qO, ql, q2
0.7618
352.4
284.8, 292.5
138.1
162.6
Lung
adenomas +
carcinomas
qO, ql, q2
0.7197
517.8
414.3,299.9
193.0
195.4
Hematopoietic
lymphomas +
sarcomas
qO,ql
0.9140
423.8
409.8, 382.6
189.5
226.2
136.1, 121.1
85.65
BMDS maximum likelihood risk estimates
Risk at dose 100
Upper 95% CL
Sum of risks at dose 100
0.004123
0.04039
0.001912
0.02919
0.0120315
0.0295375
WinBUGS Bayes risk estimates
Risk at dose 100: mean, median
Upper 95% CL
Comb, risk at dose 100 mean, median
Comb, risk at dose 100, upper 95% CL
0.01468,
0.01311
0.03032
0.01284,
0.01226
0.02590
0.009552,
0.008286
0.021410
0.03663, 0.03572
0.05847
5
6
7
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
Source: NCI (1976).
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-32 DRAFT-DO NOT CITE OR QUOTE
-------
.
10
o:
I
LU
00
O
C\j
O
p
O
0
200
400
600
800
Dose
2
3
4
Figure G-7. Female B6C3F1 mice—internal dose metric (total oxidative
metabolism): combined and individual tumor extra-risk functions.
c
CD
O
CO
O
CD
O
O
O
O -
O
p
O __
I I I I I I
100 200 300 400 500 600
5
6
7
N = 300000 Bandwidth = 3.023
Figure G-8. Female B6C3F1 mice—internal dose metric (total oxidative
metabolism): posterior distribution of BMDc for combined risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-33 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Table G-20. B6C3F1 female mice inhalation exposure—internal dose metric
(total oxidative metabolism)
Internal dosea
0
280.946
622.530
939.105
Liver
hepatomas/A7b
3/88
4/89
4/88
9/85
Lung adenomas +
carcinomas/A^
2/90
6/90
7/89
14/87
a Internal dose, Total Oxidative Metabolism, adjusted for body weight, units (mg/lwk-kg374]).
Internal doses were adjusted by a factor 0.75, accounting for exposure duration 78/104 weeks.
Before adjustment, median internal doses were 374.5945, 830.0405, 1252.14 (mg/[wk-kg3/4]).
b Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study
Source: Maltoni (1986).
Table G-21. B6C3F1 female mice—internal dose: model selection
comparison of model fit statistics for multistage models of increasing order
Tumor site
Liver
Lung
Model
order,
*selected
3*
2
1
3
2
1*
Coeff.
estimates
equal
zero
P1,P2
PI
NA
P2
NA
NA
AIC
153.1
153.4
154
195.8
195.9
194
Largest+
scaled
residual
-0.410
-0.625
-0.816
-0.571
-0.671
-0.776
Goodness
of fit
/j-value
0.8511
0.7541
0.5571
0.3995
0.3666
0.6325
* Model order selected + largest in absolute value.
Source: Maltoni (1986).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-34 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
Table G-22. B6C3F1 female mice inhalation exposure—internal dose metric
(total oxidative metabolism) (inferences for 0.05 extra risk at 95% confidence
level)
Parameters used in models
p- Value for HMDS model
BMD05 (from BMDS)
BMD05 (median, mode— WinBUGS)
BMDL (BMDS)*
ms_combo BMDosC, BMDLc
BMDL (5th percentile, WinBUGS)
BMDos for combined risk (median, mode, from
WinBUGS)
BMDL for combined risk (5th percentile,
WinBUGS)
Liver
hepatomas
qO, ql, q2, q3
0.5571
813.7
672.9, 648.0
419.7
Lung adenomas +
carcinomas
qO,ql
0.6325
366.7
382.8,372.1
244.6
412.76, 189.23
482.7
251.1
286.7,263.1
199.5
BMDS maximum likelihood risk estimates
Risk at dose 100
Upper 95% CL
Sum of risks at dose 100
0.006284
0.01335
0.01389
0.02215
0.02017
WinBUGS Bayes risk estimates: means (medians)
Risk at dose 100: mean, median
Upper 95% CL,
Comb, risk at dose 100 mean, median
Comb, risk at dose 100, upper 95% CL
0.003482,
0.002906
0.008279
0.01337,
0.01331
0.02022
0.01637,0.01621
0.02455
5
6
7
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
Source: Maltoni (1986).
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-3 5 DRAFT-DO NOT CITE OR QUOTE
-------
cc
x
LJJ
1
2
3
4
5
Figure G-9. B6C3F1 female mice inhalation exposure
combined and individual tumor extra-risk functions.
800
-internal dose metric:
CD
O
O
O _
o
p
•(•« o
0
Q
CN ~
O
O
d
o -
O
p
o
J
6
7
I I I I I I
200 400 600 800 1000 1400
N = 300000 Bandwidth = 5.053
Figure G-10. B6C3F1 female mice inhalation exposure—internal dose
metric: posterior distribution of BMDc for combined risk.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-36 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
4
5
6
7
Table G-23. Maltoni Sprague-Dawley male rats—internal dose metric (total
metabolism)
Internal dosea
0
214.6540
507.0845
764.4790
Kidney adenomas +
carcinomas/A^
0/121
1/118
0/116
5/123
Leukemias/A7*1
9/134
13/130
14/130
15/129
Testis, Leydig
cell tumors/A^
6/121
16/116
30/116
31/122
a Internal dose, Total Oxidative Metabolism, adjusted for body weight, units [mg/(wk-kg3/4)].
b Numbers at risk are the smaller of (a) time of first tumor observation or (b) 52 weeks on study.
9
10
11
12
Table G-24. Maltoni Sprague-Dawley male rats—internal dose model
selection comparison of model fit statistics for multistage models of
increasing order
Tumor site
Kidney
Leukemias
Testis, Leydig cell tumors
Model
order,
*selected
3
2
1*
3
2
1*
3
2
1*
Coeff.
estimates
equal zero
y,p2
y
y
P2,P3
P2
NA
P2,P3
P2
NA
AIC
61.35
61.75
60.32
336.5
336.5
336.5
417.7
417.7
417.7
Largest*
scaled
residual
-1.264
-1.343
-1.422
0.479
0.479
0.479
1.008
1.008
1.008
Goodness
of fit
/7-value
0.262
0.246
0.370
0.828
0.828
0.828
0.363
0.363
0.363
13
14
: Largest in absolute value.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-37 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Table G-25. Maltoni Sprague-Dawley male rats—internal dose metric (total
metabolism) (inferences for 0.01 extra risk at 95% confidence level)
Parameters used in models
p- Value for BMDS model
BMDoi (from BMDS)
BMDoi (median, mode — WinBUGS)
BMDL (BMDS)*
BMDL (5th percentile, WinBUGS)
BMDoi for combined risk (median,
mode, from WinBUGS)
BMDL for combined risk (5th
percentile, WinBUGS)
Kidney adenomas
+ carcinomas
qO,ql
0.3703
295.1
161.3
Leukemias
qO,ql
0.8285
145.8
65.29
Testis, Leydig
cell tumors
qO,ql
0.3626
26.65
20.32
20.97, 19.73
16.14
BMDS maximum likelihood risk estimates
Risk at dose 100
Upper 95% CL
Sum of risks at dose 100
Risk at dose 10
Upper 95% CL
Sum of risks at dose 10
0.003400
0.0068784
0.0068694
0.0169134
0.0370162
0.0504547
0.04729
0.0003406
0.0006900
0.0006891
0.0017044
0.0037648
0.0051638
0.004795
WinBUGS Bayes risk estimates: means (medians)
Risk at dose 100: mean, median
Upper 95% CL
Comb, risk at dose 100 — mean, median
Comb, risk at dose 100, upper 95% CL
Risk at dose 100 — mean, median
Upper 95% CL
Comb, risk at dose 10 — mean, median
Comb, risk at dose 10, upper 95% CL
0.003191,
0.003028
0.006044
7.691e-03,
7.351e-03
1.539e-02
0.03641,
0.03641
0.04769
0.04688, 0.04680
0.060380
3.196e-04,
3.032e04
6.060000e-04
7.726e-04,
7.376e04
1.550000e-03
0.003705,
0.003703
0.004874000
0.004793, 0.0047820
0.006208
4
5
* All confidence intervals are at 5% (lower) or 95% (upper) level, one-sided.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-3 8 DRAFT-DO NOT CITE OR QUOTE
-------
ir:
LJJ
O
CNJ
O
LO
CD
O
O
p
d
O
O
vertical solid, BMDc
- vertical dash, BMDLc
I I I I I I
0 100 200 300 400 500
1
2
3
4
Dose
Figure G-ll. Maltoni Sprague-Dawley male rats—internal dose metric:
combined and individual tumor extra-risk functions.
Distribution of BMDc for combined risk
£•
'w
CO
O
oo
p
o
O
O _
O
O
O
20 40 60 80 100
N = 300000 Bandwidth = 0.2732
5
6
1
Figure G-12. Maltoni Sprague-Dawley male rats—internal dose metric:
posterior distribution of BMDc for combined risk.
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-39 DRAFT-DO NOT CITE OR QUOTE
-------
1 G.9. PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK)-MODEL
2 UNCERTAINTY ANALYSIS OF UNIT RISK ESTIMATES
3 As discussed in Section 5.2, an uncertainty analysis was performed on the unit risk
4 estimates derived from rodent bioassays to characterize the impact of pharmacokinetic
5 uncertainty. In particular, two sources of uncertainty are incorporated: (a) uncertainty in the
6 rodent internal doses for each dose group in each chronic bioassay and (b) uncertainty in the
7 relationship between exposure and the human population mean internal dose at low exposure
8 levels.
9 A Bayesian approach provided the statistical framework for this uncertainty analysis.
10 Rodent bioassay internal dose-response relationships were modeled with the multistage model,
1 1 with general form
12
1 3 P(id) = 1 - exp[-(<7o + qiid + q2id2 + ... + ^/ 0, for /' = 0, 1, ...,&. Since the BMD (in internal dose units) for a given BMR can
17 be derived from the multistage model parameters qt, it is sufficient to estimate the posterior
18 distribution of qt given the combined bioassay data (for each dose group/ the number
19 responding^, the number at risk «,, and the administered dose 4) and the rodent
20 pharmacokinetic data, for which the posterior distribution can be derived using the Bayesian
21 analysis of the PBPK model described in Section 3.5. In particular, the posterior distribution of
22 qt can be expressed as
23
24 P(q[l} \Dbloassay Dpk) oc P(q[l]) P(ym \ qm nv]) P(idv] \dm, Dpk) (Eq. G- 1 0)
25
26 Here, the first term after the proportionality P(q\t]) is the prior distribution of the multistage
27 model parameters (assumed to be noninformative), the second term P(y\j^q\t\ %]) is the likelihood
28 of observing the bioassay response given a particular set of multistage parameters and the
29 number at risk (the product of binomial distributions for each dose group), and P(id^\d^ Dpk) is
30 the posterior distribution of the rodent internal doses id\j], given the bioassay doses and the
3 1 pharmacokinetic data used to estimate the PBPK model parameters.
32 The distribution of unit risk (URici = BMRIBMD) estimates in units of "per internal dose"
33 is then derived deterministically from the distribution of multistage model parameters:
34
35 P(URld\Dbloassay Dpk.rodent) = \P(q^\Dbloassay Dplc.rodent) 5[UR - BMR/BMD(q[l])] dqv] (Eq. G-l 1)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-40 DRAFT-DO NOT CITE OR QUOTE
-------
1 Here 6 is the Dirac delta-function. Then, the distribution of unit risk estimates in units of "per
2 human exposure" (per mg/kg/d ingested or per continuous ppm exposure) is derived by
3 converting the unit risk estimate in internal dose units:
4
J f \UJ\human\J-Jbioassay J-Jpk-rodent) 1-t^ \UJ\id\J-J bioassay J-Jpk-rodent) f (J^conversion\-L'pk-human)
6 §(URhuman — URjd x idconversion) dldCOnversion (Eq. G-12)
7
8 Here, idconversion is the population mean of the ratio between internal dose and administered
9 exposure at low dose (0.001 ppm or 0.001 mg/kg/d), and P(idconversion\Dpk-human) is its posterior
10 distribution from the Bayesian analysis of the human PBPK model.
11 This statistical model was implemented via Monte Carlo as follows. For each bioassay,
12 for a particular iteration r(r= 1... nr~),
13
14 (1) A sample of rodent PBPK model population parameters (\a^L)rodent,r was drawn from the
15 posterior distribution. Using these population parameters, a single set of group rodent
16 PBPK model parameters Qmdent,r was drawn from the population distribution. As
17 discussed in Section 3.5, for rodents, the population model describes the variability
18 among groups of rodents, and the group-level parameters represent the "average"
19 toxicokinetics for that group.
20 (2) Using Qmdent,r, the rodent PBPK model was run to generate a set of internal doses /'<%r for
21 the bioassay.
22 (3) Using this set of internal doses /<%r, a sample g^ was selected from the distribution
23 (conditional on /£%r) of multistage model parameters, generated using the WinBUGS,
24 following the methodology of Kopylev et al. (2007).
25 (4) The unit risk in internal dose units URtd,r = BMR/BMD(q^r) was calculated based on the
26 multistage model parameters.
27 (5) A sample of human PBPK model population parameters (u,2)toma«,r was drawn from the
28 posterior distribution. Using these population parameters, multiple sets of individual
29 human PBPK model parameters Qhuman,r,[s] (s= l...ns~) were generated. A continuous
30 exposure scenario at low exposure was run for each individual, and the population mean
31 internal dose conversion was derived by taking the arithmetic mean of the internal dose
32 conversion for each individual: idconversior,:r = Sum(idconversion>r:S)/ns.
33 (6) The sample for the unit risk in units per human exposure was calculated by multiplying
34 the sample for the unit risk in internal dose units by the sample for the population internal
35 dose conversion: URhUman,r ~ URld,r x idconversiorirr.
36
37 In practice, samples for each of the above distributions were "precalculated," and
38 inferences were performed by re-sampling (with replacement) according to the scheme above.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-41 DRAFT-DO NOT CITE OR QUOTE
-------
1 For the results described in Section 5.2, a total of nr = 15,000 samples was used for deriving
2 summary statistics. For calculating the unit risks in units of internal dose, the BMDs were
3 derived by re-sampling from a total of 4.5x 106 multistage model parameter values (1,500 rodent
4 PBPK model parameters from the Bayesian analysis described in Section 3.5, for each of which
5 there were conditional distributions of multistage model parameters of length 3,000 derived
6 using WinBUGS). The conversion to unit risks in units of human exposure was re-sampled from
7 500 population mean values, each of which was estimated from 500 sampled individuals.
8 The file
9 Appendix.linked.files\AppG.Cancer.Rodents.Uncertainty.Analysis.TCE.DRAFT.pdf contains
10 summary statistics (mean, and selected quantiles from 0.01 to 0.99) from these analyses, and is
11 the source for the results presented in Chapter 5 (see Tables 5-34 and 5-35). Histograms of the
12 distribution of unit risks in per unit human exposure are in the file
13 Appendix.linked.filesVAppG.Cancer.Rodents.uncertainty.CSF-
14 inhal.histograms.inhalation.bioassays.TCE.DRAFT.pdf for the rodent inhalation bioassays and
15 Appendix.linked.filesVAppG.Cancer.Rodents.uncertainty.CSF-
16 oral.histograms.oral.bioassays.TCE.DRAFT.pdf for the rodent oral bioassays. Route-to-route
17 extrapolated unit risks are in the files
18 Appendix.linked.filesVAppG.Cancer.Rodents.uncertainty.CSF-
19 inhal.histograms.oral.bioassays.TCE.DRAFT.pdf (inhalation unit risks extrapolated from oral
20 bioassays) and Appendix.linked.files\AppG.Cancer.Rodents.uncertainty.CSF-
21 oral.histograms.inhalation.bioassays.TCE.DRAFT.pdf (oral unit risks extrapolated from
22 inhalation bioassays). Each figure shows the uncertainty distribution for the male and female
23 combined population risk per unit exposure (transformed to base-10 logarithm), with the
24 exception of testicular tumors, for which only the population risk per unit exposure for males is
25 shown.
26
27 G.10. REFERENCES
28 Bogen, K.T. 1990. Uncertainty in Environmental Health Risk Assessment. London: Taylor & Francis.
29 Fukuda, K; Takemoto, K; Tsuruta, H. (1983) Inhalation carcinogenicity of trichloroethylene in mice and rats. Ind
30 Health 21:243-254.
31 Henschler D, Romen W, Elsasser HM, Reichaert D, Eder E, Radwan Z. 1980. Carcinogenicity study of
32 trichloroethylene by longterm inhalation in three animal species. Arch Toxicol 43: 237-248 (1980).
33 Kopylev, L; Chen, C; White, P. (2007) Towards quantitative uncertainty assessment for cancer risks: central
34 estimates and probability distributions of risk in dose-response modeling. Regul Toxicol Pharmacol 49(3):203-207.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-42 DRAFT-DO NOT CITE OR QUOTE
-------
1 Maltoni, C; Lefemine, G; Cotti, G. (1986) Experimental research on trichloroethylene carcinogenesis. In: Maltoni,
2 C; Mehlman MA., eds. Vol. 5. Archives of research on industrial carcinogenesis. Princeton, NJ: Princeton Scientific
3 Publishing;
4 NCI (National Cancer Institute). (1976) Carcinogenesis bioassay of trichloroethylene. Division of Cancer Cause and
5 Prevention, National Cancer Institute, U.S. Department of Health, Education, and Welfare, DHEW Publication No.
6 (NIH) 76-802, Technical Report Series No. 2, 218 pages; NCI-CG-TR-2; NTIS PB254122.
7 http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr002.pdf.
8 Nitcheva DK, Piegorsch WW, West RW. (2007). On use of the multistage dose-response model for assessing
9 laboratory animal carcinogenicity, Regulatory Toxicology and Pharmacology 48:135-147.
10 NRC (National Research Council). 1994. Science and Judgment in Risk Assessment. Washington, DC: National
11 Academy Press
12 NTP (National Toxicology Program). (1988) Toxicology and carcinogenesis studies of trichloroethylene (CAS no.
13 79-01-6) in four strains of rats (ACI, August, Marshall, Osborne-Mendel) (gavage studies). Public Health Service,
14 U.S. Department of Health and Human Services; NTP TR-273; NIH Publication No. 88-2529. Available from the
15 National Institute of Environmental Health Sciences, Research Triangle Park, NC, and the National Technical
16 Information Service, Springfield, VA; PB88-218896. http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr273.pdf.
17 NTP (National Toxicology Program). (1990) Carcinogenesis Studies of Trichloroethylene (Without Epichlorhydrin)
18 (CAS No. 79-01-6) in F344/N Rats and B6C3F1 Mice (Gavage Study). NTP TR 243. Research Triangle Park, NC:
19 U. S Department of Health and Human Services.
20 Piegorsch WW, Bailer AJ, 1997, Statistics for Environmental Biology and Toxicology (Chapman & Hall, London).
21 See Ch. 6.3.2
22 Portier CJ, Bailer AJ. 1989. Testing for increased carcinogenicity using a survival-adjusted quanta! response test.
23 Fund Appl Toxicol 12:731-737.
24 Portier CJ, Hedges JC, Hoel DG. 1986. Age-specific models of mortality and tumor, onset for historical control
25 animals in the National Toxicology Program's carcinogenicity experiments. Cancer Research 46:4372-4378.
26 Spiegelhalter, D; Thomas, A; Best, N; et al. (2003) WinBUGS user manual. Version 1.4. Available online at
27 www.mrc-bsu.cam.ac.uk/bugsAVinBUGS/manuall4.pdf.
28 U.S. EPA (Environmental Protection Agency). (1994) Methods for derivation of inhalation reference concentrations
29 and application of inhalation dosimetry. Environmental Criteria and Assessment Office, Office of Health and
30 Environmental Assessment, Washington, Washington, DC; EPA/600/8-90/066F. Available from: National
31 Technical Information Service, Springfield, VA; PB2000-500023.U.S. EPA. 1980. Water Quality Criteria
32 Documents; Availability. Fed Reg 45(231), page 79352.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 G-43 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX H
Lifetable Analysis and Weighted Linear
Regression based on Results from
Charbotel et al. (2006)
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 H-i DRAFT—DO NOT CITE OR QUOTE
-------
CONTENTS—Appendix H: Lifetable Analysis and Weighted Linear Regression based on
Results from Charbotel et al. (2006)
APPENDIX H: LIFETABLE ANALYSIS AND WEIGHTED LINEAR REGRESSION
BASED ON RESULTS FROM CHARBOTEL ET AL. (2006) H-l
H.I. LIFETABLE ANALYSIS H-l
H.2. EQUATIONS USED FOR WEIGHTED LINEAR REGRESSION OF
RESULTS FROM CHARBOTEL ET AL. (2006) H-l
H.3. REFERENCES H-4
777/5 document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 H-ii DRAFT—DO NOT CITE OR QUOTE
-------
1 APPENDIX H: LIFETABLE ANALYSIS AND WEIGHTED LINEAR REGRESSION
2 BASED ON RESULTS FROM CHARBOTEL ET AL. (2006)
3
4
5 H.l. LIFETABLE ANALYSIS
6 A spreadsheet illustrating the extra-risk calculation for the derivation of the lower 95%
7 bound on the effective concentration associated with a 1% extra risk (LECoi) for renal cell
8 carcinoma (RCC) incidence is presented in Table H-l.
9
10 H.2. EQUATIONS USED FOR WEIGHTED LINEAR REGRESSION OF RESULTS
11 FROM CHARBOTEL ET AL. (2006) (source: Rothman [1986], p. 343-344)
12 Linear model: RR = 1 + bX
13
14 where RR = risk ratio, X = exposure, and b = slope
15
16 b can be estimated from the following equation:
17
18 b = ^ ^ (Eq. H-l)
.7=2
19
20 where y' specifies the exposure category level and the reference category (j = 1) is ignored.
21 The standard error of the slope can be estimated as follows:
22
23 SE(b)* . (Eq.H-2)
24
25 The weights, Wj, are estimated from the confidence intervals (as the inverse of the variance):
26
(Eq. H-3)
2 xl.96
28
29 where RRj is the 95% upper bound on the RRj estimate (for they'th exposure category) and RR± is
30 the 95% lower bound on the RRj estimate.
31
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 H-1 DRAFT—DO NOT CITE OR QUOTE
-------
to
O k^j
o ^
"I
I
§
***.
£3'
1
TO'
i
Table H-l. Extra-risk calculation3 for environmental exposure to 1.82 ppm TCE (the LEC0i for RCC
incidence)11 using a linear exposure-response model based on the categorical cumulative exposure results of
Charbotel et al. (2006), as described in Section 5.2.2.1.2.
A
Interval
number
(0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
B
Age
interval
<1
1-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-59
80-84
C
All cause
mortality
(xl05/yr)
685.2
29.9
14.7
18.7
66.1
94
96
107.9
151.7
231.7
352.3
511.7
734.8
1140.1
1727.4
2676.4
4193.2
6717.2
Extra risk = (Rx - Ro)/(l -
D
RCC
incidence
(xl05/yr)
0
0
0
0.1
0.1
0.2
0.7
1.6
3.2
6.3
11
17.3
26.2
36.2
44.6
49
51.6
44.4
E
All cause
hazard
rate
(h")
0.0069
0.0012
0.0007
0.0009
0.0033
0.0047
0.0048
0.0054
0.0076
0.0116
0.0176
0.0256
0.0367
0.0570
0.0864
0.1338
0.2097
0.3359
F
Prob. of
surviving
interval
0.9932
0.9988
0.9993
0.9991
0.9967
0.9953
0.9952
0.9946
0.9924
0.9885
0.9825
0.9747
0.9639
0.9446
0.9173
0.8747
0.8109
0.7147
G
Prob. of
surviving
up to
interval
1.0000
0.9932
0.9920
0.9913
0.9903
0.9871
0.9824
0.9777
0.9725
0.9651
0.9540
0.9373
0.9137
0.8807
0.8319
0.7631
0.6675
0.5412
H
RCC
cancer
hazard
rate
(h)
0.000000
0.000000
0.000000
0.000005
0.000005
0.000010
0.000035
0.000080
0.000160
0.000315
0.000550
0.000865
0.001310
0.001810
0.002230
0.002450
0.002580
0.002220
Ro =
I
Cond.
prob. of
RCC
incidence
in interval
(Ro)
0.000000
0.000000
0.000000
0.000005
0.000005
0.000010
0.000034
0.000078
0.000155
0.000302
0.000520
0.000801
0.001175
0.001549
0.001777
0.001750
0.001554
0.001021
0.010736
J
Exp.
duration
mid
interval
(xtime)
0.5
3
7.5
12.5
17.5
22.5
27.5
32.5
37.5
42.5
47.5
52.5
57.5
62.5
67.5
72.5
77.5
82.5
K
Cum.
exp. mid
interval
(jedose)
2.77
16.61
41.52
69.20
96.88
124.56
152.24
179.91
207.59
235.27
262.95
290.63
318.31
345.99
373.67
401.35
429.03
456.71
L
Exposed
RCC
hazard
rate
(hx)
0.000000
0.000000
0.000000
0.000006
0.000006
0.000013
0.000049
0.000117
0.000245
0.000504
0.000919
0.001507
0.002375
0.003409
0.004358
0.004961
0.005407
0.004809
M
Exposed
all cause
hazard
rate
(h*x)
0.0069
0.0012
0.0007
0.0009
0.0033
0.0047
0.0048
0.0054
0.0077
0.0118
0.0180
0.0262
0.0378
0.0586
0.0885
0.1363
0.2125
0.3384
N
Exposed
prob. of
surviving
interval
(qx)
0.9932
0.9988
0.9993
0.9991
0.9967
0.9953
0.9952
0.9946
0.9924
0.9883
0.9822
0.9741
0.9629
0.9431
0.9153
0.8726
0.8086
0.7129
0
Exposed
prob. of
surviving
up to
interval
(Sx)
1.0000
0.9932
0.9920
0.9913
0.9903
0.9871
0.9824
0.9777
0.9724
0.9650
0.9537
0.9367
0.9124
0.8786
0.8286
0.7584
0.6617
0.5351
Rx =
P
Exposed
cond.
prob. of
RCC in
interval
(Rx)
0.000000
0.000000
0.000000
0.000006
0.000006
0.000013
0.000048
0.000114
0.000237
0.000484
0.000869
0.001393
0.002127
0.002909
0.003456
0.003518
0.003223
0.002183
0.020586
Ro) = 0.00996
^§
H I
O >
HH Oq
H TO
O
H
W
-------
to
\D
rj
S
a
<§»
TO
TO'
Si
i.
§
^ Si
•^ st
O s
^H c^
hH Oq
H TO
W S
Column A: interval index number (/').
Column B: 5-year age interval (except <1 and 1-4) up to age 85.
Column C: all-cause mortality rate for interval / (x 105/year) (2004 data from NCHS [2007]).
Column D: RCC incidence rate for interval / (x 105/year) (2001-2005 SEER data [http://seer.cancer.gov]).
Column E: all-cause hazard rate for interval /' (h*,) [= all-cause mortality rate x number of years in age interval].0
Column F: probability of surviving interval / without being diagnosed with RCC (q,) [= exp(-/z*,)].
Column G: probability of surviving up to interval /' without having been diagnosed with RCC (S,) [Sj = 1; St = St-i x qt_^ for i > 1].
Column H: RCC incidence hazard rate for interval /' (h,) [= RCC incidence rate x number of years in interval].
Column I: conditional probability of being diagnosed with RCC in interval /' [= (hj/h*,) x Sf x (l-q,)], i.e., conditional upon surviving up to interval /' without
having been diagnosed with RCC [Ro, the background lifetime probability of being diagnosed with RCC = the sum of the conditional probabilities
across the intervals].
Column J: exposure duration (in years) at mid-interval (xtime).
Column K: cumulative exposure mid-interval (xdose) [= exposure level (i.e., 1.82 ppm) x 365/240 x 20/10 x xtime] (365/240 x 20/10 converts continuous
environmental exposures to corresponding occupational exposures).
Column L: RCC incidence hazard rate in exposed people for interval / (hx,) [= h,• x (l + p x xdose), where (3 = 0.001205 + (1.645 x 0.0008195) = 0.002554]
[0.001205 per ppm x year is the regression coefficient obtained from the weighted linear regression of the categorical results (see Section 5.2.2.1.2).
To estimate the LEC0i, i.e., the 95% lower bound on the continuous exposure giving an extra risk of 1%, the 95% upper bound on the regression
coefficient is used, i.e., MLE + 1.645 x SE].
Column M: all-cause hazard rate in exposed people for interval /' (h*x,) [= h*t + (hxt - h,)].
Column N: probability of surviving interval /' without being diagnosed with RCC for exposed people (qx,) [= exp(-h*x,)].
Column O: probability of surviving up to interval /' without having been diagnosed with RCC for exposed people (Sx,) [Sx, = 1; Sxt = Sxt-i x qx^, for /> 1].
Column P: conditional probability of being diagnosed with RCC in interval / for exposed people [= (hxj/h*x,) x Sxt x (l-qx,)] (Rx, the lifetime probability of
being diagnosed with RCC for exposed people = the sum of the conditional probabilities across the intervals).
a Using the methodology of BEIRIV (1988).
b The estimated 95% lower bound on the continuous exposure level of TCE that gives a 1% extra lifetime risk of RCC.
0 For the cancer incidence calculation, the all-cause hazard rate for interval /' should technically be the rate of either dying of any cause or being diagnosed with
the specific cancer during the interval, i.e., (the all-cause mortality rate for the interval + the cancer-specific incidence rate for the interval—the cancer-specific
mortality rate for the interval [so that a cancer case isn't counted twice, i.e., upon diagnosis and upon death]) x number of years in interval. This adjustment
was ignored here because the RCC incidence rates are small compared with the all-cause mortality rates.
MLE = maximum likelihood estimate, SE = standard error.
O
H
W
-------
1 H.3. REFERENCES
2 BEIR ([Committee on the] Biological Effects of Ionizing Radiation). (1988) Health risks of radon and other
3 internally deposited alpha emitters. BEIR IV. Washington, DC: National Academy Press.
4 Charbotel B, Fevotte J, Hours M, Martin J-L, Bergeret A. (2006) Case-control study on renal cell cancer and
5 occupational exposure to trichloroethylene. Part II: epidemiological aspects. Ann Occup Hyg 50: 777-787.
6 NCHS (National Center for Health Statistics). (2007) National Vital Statistics Reports, vol. 55, no. 19; August 21,
7 2007, Table 3. National Center for Health Statistics, Hyattsville, MD.
8 RothmanKJ. (1986) Modern Epidemiolgy. Little, Brown and Company, Boston.
This document is a draft for review purposes only and does not constitute Agency policy.
10/20/09 H-4 DRAFT—DO NOT CITE OR QUOTE
------- |