United States
Environmental Protection
Agency
Office of Research
and Development
Washington, DC 20460
NCEA-1-0503
January 16, 2002
External Review Draft
Perchlorate Environmental
Contamination:
Toxicological Review and
Risk Characterization
External
Review
Draft
(Do Not Cite
or Quote)
Notice
This document is an external review draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
090ECB99 cov-0
-------
NCEA-1-0503
January 16, 2002
Internal Review Draft
Perchlorate Environmental
Contamination: Toxicological
Review and Risk Characterization
Notice
This document is an external review draft. It has not been
formally released by EPA and should not at this stage be
construed to represent Agency policy. It is being circulated
for comment on its technical accuracy and policy implications.
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Disclaimer
This document is an external draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
11
-------
Table of Contents
Page
List of Tables xi
List of Figures xvi
Contributors and Reviewers xxvi
Preface xxxi
Acknowledgments xxxiii
EXECUTIVE SUMMARY E-l
1. INTRODUCTION 1-1
1.1 PRODUCTION USES AND SOURCES OF PERCHLORATE
CONTAMINATION 1-1
1.2 EVOLUTION OF ANALYTICAL DETECTION METHODS AND
EMERGING OCCURRENCE DATA 1-4
1.3 HEALTH AND ECOTOXICOLOGY RISK ASSESSMENTS-
HISTORICAL OVERVIEW 1-16
1.3.1 Overview of Perchlorate Health Risk Assessment 1-16
1.3.2 Overview of Ecotoxicology Screening Level Assessment 1-19
1.4 RISK CHARACTERIZATION AND REGULATORY AGENDA 1-20
1.4.1 U.S. Environmental Protection Agency Regulatory Plans 1-20
1.4.2 State Regulatory Plans 1-21
1.5 SUMMARY 1-23
2. PHYSICOCHEMICAL CHARACTERISTICS 2-1
3. TOXICOKINETICS/TOXICODYNAMICS AND MODE-OF-ACTION TESTING
STRATEGY 3-1
3.1 ABSORPTION, DISTRIBUTION, METABOLISM, AND ELIMINATION
OF PERCHLORATE 3-1
3.1.1 Human Studies 3-2
3.1.2 Laboratory Animal Studies 3-3
3.2 IODINE METABOLISM AND THYROID PHYSIOLOGY 3-8
3.3 TOXICOKINETICS OF PERCHLORATE 3-13
3.4 TOXICODYNAMICS OF THYROID HORMONE PERTURBATIONS 3-15
3.4.1 Carcinogenic Effects 3-15
3.4.2 Neurodevelopmental Deficits and Other Potential Adverse Effects
Resulting from Thyroid Hormone Disruption 3-21
3.5 DEVELOPMENT OF A TOXICITY TESTING STRATEGY BASED ON
MODE OF ACTION 3-29
4. HUMAN HEALTH EFFECTS DATA 4-1
4.1 EPIDEMIOLOGICAL DATA 4-2
4.1.1 Ecological Studies 4-3
iii
-------
Table of Contents
(cont'd)
Page
4.1.2 Occupational Studies 4-13
4.2 CLINICAL STUDIES 4-19
4.2.1 Studies in Healthy Human Subjects 4-19
4.2.2 Studies in Patients with Graves' Disease 4-25
4.2.2.1 Hematological Effects 4-27
4.3 SUMMARY OF CONCLUSIONS REGARDING HUMAN HEALTH
EFFECTS STUDIES 4-28
5. TOXICOLOGICAL EFFECTS IN LABORATORY ANIMAL STUDIES 5-1
5.1 CHRONIC STUDIES AND GENOTOXICITY ASSAYS 5-7
5.1.1 Cancer Studies 5-11
5.1.2 Genotoxicity Assays 5-12
5.1.2.1 In Vitro Assays 5-13
5.1.2.2 In Vivo Assays 5-15
5.1.2.3 Summary of Genotoxicity Battery Results 5-16
5.2 GENERAL TOXICITY: SHORT-TERM AND SUBCHRONIC TESTING .... 5-17
5.2.1 Historical Data 5-17
5.2.2 Caldwell etal. (1995) 14-Day Study 5-19
5.2.2.1 Thyroid Histology Data 5-20
5.2.2.2 Thyroid and Pituitary Hormone Analyses 5-21
5.2.3 The 90-Day Testing Strategy Bioassay in Rats 5-23
5.2.3.1 General Toxicity, Thyroid Histopathology Results, and
Satellite Reproductive Assay 5-25
5.2.3.2 Thyroid and Pituitary Hormone Analyses 5-27
5.3 DEVELOPMENTAL NEUROTOXICITY STUDIES 5-33
5.3.1 The 1998 Developmental Neurotoxicity Study 5-33
5.3.1.1 Results of General Toxicity Measures, Neurohistology,
and Morphology 5-35
5.3.1.2 Evaluation of Thyroid Histopathology 5-38
5.3.1.3 Thyroid and Pituitary Hormone Analyses 5-41
5.3.1.4 Behavioral Evaluations 5-43
5.3.2 Motor Activity Study (Bekkedal et al., 2000) 5-47
5.3.2.1 EPA and NIEHS Statistical Analyses of Motor Activity
Effects 5-48
5.3.3 The 2001 "Effects Study" 5-52
5.3.3.1 Results of General Toxicity Measures 5-53
5.3.3.2 Evaluation of Thyroid Histopathology 5-53
5.3.3.3 Thyroid and Pituitary Hormone Analyses 5-57
5.3.3.3.1 Maternal Hormone Analyses 5-59
5.3.3.3.2 Fetal and Neonatal Hormone Analyses 5-59
5.3.3.4 Brain Morphometry Effects 5-60
iv
-------
Table of Contents
(cont'd)
Page
5.4 DEVELOPMENTAL STUDIES 5-74
5.4.1 Historical Studies 5-74
5.4.2 Segment II Developmental Toxicity Study in Rabbits 5-76
5.4.2.1 Results of Maternal Examinations and Thyroid
Histopathology 5-78
5.4.2.2 Developmental Endpoints 5-78
5.4.2.3 Maternal Thyroid and Pituitary Hormone Analyses 5-79
5.4.3 Segment II Developmental Study in Rats 5-81
5.4.3.1 Results of Maternal Examinations 5-82
5.4.3.2 Developmental Endpoints 5-82
5.4.3.3 Conclusions Regarding Developmental Toxicity in Rats 5-83
5.5 TWO-GENERATION REPRODUCTIVE TOXICITY STUDY 5-83
5.5.1 General Toxicity Results and Evaluation of Reproductive Parameters . .. 5-85
5.5.2 Evaluation of Thyroid Histology 5-86
5.5.2.1 Colloid Depletion, Hypertrophy, and Hyperplasia 5-87
5.5.2.2 Bayesian Analysis of Tumor Incidence 5-88
5.5.3 Thyroid and Pituitary Hormone Analyses 5-91
5.6 IMMUNOTOXICITY STUDIES 5-92
5.6.1 Results for General Toxicity, Organ Weight, and Cellularity
Measures 5-97
5.6.2 Evaluation of Thyroid Histology 5-97
5.6.3 Thyroid and Pituitary Hormone Analyses 5-98
5.6.4 Results of Immune Function Assays 5-99
5.6.5 Results for Evaluations of Hematological Parameters 5-102
5.6.6 Results Summary 5-103
6. CONSTRUCTION OF PBPK MODELS TO ADDRESS PERCHLORATE'S
MODE-OF -ACTION 6-1
6.1 MODE-OF-ACTION FRAMEWORK AND UNDERLYING MODELING
APPROACH 6-3
6.1.1 Parallelogram Approach to Interspecies Extrapolation 6-4
6.1.2 Extending the Parallelogram Approach to Various Experimental
Life Stages 6-8
6.2 ADULT RAT AND HUMAN MODEL STRUCTURES 6-9
6.2.1 Data and Methods 6-15
6.2.1.1 Studies in Laboratory Rats 6-15
6.2.1.1.1 Acute iv Experiments in Rats 6-18
6.2.1.1.2 Drinking Water Studies in Rats 6-19
6.2.1.2 Human Studies 6-19
6.2.1.2.1 Human Iodide Kinetic Data 6-19
v
-------
Table of Contents
(cont'd)
Page
6.2.1.2.2 Perchlorate Kinetics and Inhibition of Thyroid
Iodide Uptake 6-20
6.2.1.2.3 Supporting Kinetic Studies 6-21
6.2.2 Adult Male Rat Model Development 6-22
6.2.2.1 Physiologic Parameters and Tissue Partition Coefficients 6-22
6.2.2.2 Chemical-Specific Parameters 6-23
6.2.2.2.1 Affinity Constants and Maximum Velocities
for Active Transport Processes 6-23
6.2.2.2.1 Effective Partitions, Permeability Area Cross
Products and Clearance Values 6-25
6.2.2.3 Adult Male Rat Model Simulation Results and Validation 6-25
6.2.3 Human Model Development 6-34
6.2.3.1 Physiologic Parameters and Tissue Partition Coefficients 6-34
6.2.3.2 Chemical-Specific Parameters 6-35
6.2.3.2.1 Affinity Constants and Maximum Velocities
for Active Transport Processes 6-35
6.2.3.2.2 Effective Partitions, Permeability Area Cross
Products, and Clearance Values 6-35
6.2.3.3 Adult Human Model Parameterization and Validation 6-35
6.2.4 Summary 6-48
PREGNANT RAT AND FETAL MODEL STRUCTURE 6-51
6.3.1 Data and Methods 6-56
6.3.1.1 AFRL/HEST Experiments in Laboratory Rats 6-57
6.3.1.1.1 Drinking Water Study 6-57
6.3.1.1.2 Preliminary Iodide Kinetics Study 6-57
6.3.1.1.3 Iodide Inhibition Kinetics Study 6-58
6.3.1.2 Data Published in the Literature 6-58
6.3.1.2.1 Versloot et al., 1997 6-58
6.3.1.2.2 Sztanyik and Turai, 1988 6-58
6.3.1.2.3 Feldman et al., 1961 6-59
6.3.2 Pregnant Rat and Fetus Model Development 6-59
6.3.2.1 Physiological Parameters and Tissue Partition Coefficients .... 6-59
6.3.2.1.1 Maternal Tissues 6-60
6.3.2.1.2 Maternal Blood Flow 6-61
6.3.2.1.3 Fetal Tissues 6-61
6.3.2.1.4 Fetal Blood Flow 6-63
6.2.2.2 Chemical-Specific Parameters 6-63
6.3.2.2.1 Affinity Constants and Maximum Velocities
for Active Uptake Processes 6-63
6.3.2.2.2 Effective Partitioning Permeability Area Cross
Products and Clearance Values 6-64
vi
-------
Table of Contents
(cont'd)
Page
6.3.2.3 Pregnant Rat and Fetus Model Parameterization
and Validation 6-65
6.3.2.3.1 Perchlorate Model Parameterization 6-65
6.3.2.3.2 Iodide Model Parameterization 6-68
6.3.3 Model Validation 6-71
6.3.4 Summary 6-77
6.4 LACTATING AND NEONATAL RAT MODEL STRUCTURE 6-79
6.4.1 Data and Methods 6-85
6.4.1.1 AFRL/HEST Experiments in Laboratory Rats 6-86
6.4.1.1.1 Drinking Water Study 6-86
6.4.1.1.2 Cross-fostering Study 6-86
6.4.1.1.3 Perchlorate Kinetics Study 6-87
6.4.1.1.4 Iodide Inhibition Kinetics Study 6-87
6.4.1.2 Data Published in the Literature 6-88
6.4.1.2.1 Sztanyik and Turai, 1988 6-88
6.4.1.2.2 Potter et al., 1959 6-88
6.4.2 Lactating and Neonatal Rat Model Development 6-88
6.4.2.1 Physiological Parameters and Partition Coefficients 6-88
6.4.2.1.1 Maternal Tissues 6-88
6.4.2.1.2 Neonatal Tissues 6-90
6.4.2.1.3 Blood Flows 6-91
6.4.2.2 Chemical-Specific Parameters 6-91
6.4.2.2.1 Affinity Constants and Maximum Velocities
for Active Uptake Processes 6-91
6.4.2.2.2 Effective Partitions, Permeability Area Cross
Products and Clearance Values 6-92
6.4.2.3 Lactating Rat and Neonate Model Parameterization and
Validation 6-92
6.4.2.3.1 Perchlorate Model Parameterization 6-92
6.4.2.3.2 Iodide Model Parameterization 6-95
6.4.3 Model Validation 6-97
6.4.4 Summary 6-101
6.5 APPLICATION OF PBPK MODEL STRUCTURES TO INTERSPECIES
EXTRAPOLATION 6-105
6.5.1 Sensitivity Analysis of Proposed Adult Male Rat Model 6-106
6.5.2 Derivation of Human Equivalent Exposure Estimates 6-113
6.5.3 Summary 6-121
7. DOSE-RESPONSE ASSESSMENTS FOR HUMAN HEALTH 7-1
7.1.1 Key Event and Weight of the Evidence 7-4
7.1.2 Dosimetric Adjustment of Exposures Associated with Effect Levels .... 7-10
vii
-------
Table of Contents
(cont'd)
Page
7.1.2.1 Choice of Dose Metric 7-12
7.1.3 Point of Departure Analysis 7-16
7.1.4 Application of Uncertainty Factors 7-20
7.1.5 Operational Derivation of the Reference Dose 7-23
7.1.5.1 Comparison with Derivation Considering Human Data 7-25
7.1.5.2 Comparison with Derivation Based on Tumor Data 7-27
7.1.5.2.1 Choice of Dose-Response Procedure 7-27
7.1.5.3 Possible Susceptibility 7-30
7.1.6 Designation of Confidence Levels 7-30
7.2 INHALATION REFERENCE CONCENTRATION 7-31
7.3 SUMMARY 7-31
APPENDIX 7A: Correlation Analyses 7A-1
APPENDIX 7B: Benchmark Dose Statistics for Hormone Analyses 7B-1
7B.1 Benchmark Dose Estimates Submitted to U.S. Environmental
Protection Agency 7B-2
7B.2 U.S. Environmental Protection Agency Benchmark Dose
Estimates for Thyroid and Pituitary Hormones 7B-4
7B.3 Summary of U.S. Environmental Protection Agency Benchmark
Dose Analyses 7B-6
8. SCREENING ECOLOGICAL RISK ASSESSMENT FOR PERCHLORATE 8-1
8.1 INTRODUCTION 8-1
8.1.1 Management Goals and Decisions 8-1
8.1.2 Scope, Complexity, and Focus 8-2
8.2 PROBLEM FORMULATION 8-5
8.2.1 Assessment Endpoints 8-5
8.2.1.1 Fish Community Richness and Productivity 8-5
8.2.1.2 Aquatic Invertebrate Community Richness and Productivity .... 8-5
8.2.1.3 Aquatic Plant Richness and Productivity 8-6
8.2.1.4 Soil Invertebrate Community Richness and Productivity 8-6
8.2.1.5 Terrestrial Plant Richness and Productivity 8-6
8.2.1.6 Population Productivity of Herbivorous Wildlife 8-6
8.2.2 Conceptual Models 8-6
8.2.3 Analysis Plan 8-9
8.3 ANALYSIS 8-9
8.3.1 Characterization of Exposure 8-9
8.3.1.1 Water Concentrations 8-9
8.3.1.2 Aquatic Bioaccumulation 8-11
8.3.1.3 Soil Levels 8-12
8.3.1.4 Uptake by Vegetation 8-13
Vlll
-------
Table of Contents
(cont'd)
Page
8.3.1.5 Herbivore Exposure 8-15
8.3.2 Characterization of Effects 8-17
8.3.2.1 Aquatic Organisms 8-17
8.3.2.2 Terrestrial Organisms 8-22
9. EVALUATION OF EVIDENCE FOR INDIRECT EXPOSURE 9-1
9.1 FERTILIZERS AS SOURCES OF PERCHLORATE SALTS 9-2
9.1.1 The Potential Role of Fertilizers 9-2
9.1.2 Raw Material Use 9-3
9.1.3 Fertilizer Analysis Studies 9-4
9.1.4 Complicating Factors 9-8
9.2 MONITORING FATE AND TRANSPORT IN LIVING PLANTS 9-9
9.2.1 Difficulties in Analyzing Plant Tissues and Other Environmental
Samples for Perchlorate 9-9
9.2.2 Ecological Transport 9-10
9.2.3 Extrapolating to Food Plants 9-14
9.3 SUMMARY 9-16
10. MAJOR RISK CHARACTERIZATION CONCLUSIONS 10-1
10.1 HUMAN HEALTH 10-1
10.1.1 Hazard Potential 10-1
10.1.2 Dose Response 10-2
10.1.3 Risk Characterization 10-4
10.1.3.1 Direct Exposures 10-4
10.1.3.2 Indirect Exposures 10-5
10.1.4 Major Uncertainties and Research Needs 10-6
10.2 ECOTOXICOLOGY 10-6
10.2.1 Aquatic Life 10-6
10.2.2 Risks to Consumers of Aquatic Life 10-8
10.2.3 Terrestrial Life 10-8
10.2.3.1 Plants 10-8
10.2.3.2 Soil Invertebrates 10-9
10.2.3.3 Herbivores 10-10
10.2.3.4 Carnivores 10-11
10.2.4 Uncertainties 10-11
10.2.4.1 Uncertainties Concerning Aquatic Risks 10-11
10.2.4.2 Uncertainties Concerning Terrestrial Risks 10-12
10.2.5 Research Needs 10-13
10.2.5.1 Exposure 10-14
10.2.5.2 Effects 10-15
10.2.5.3 Site-Specific Investigations 10-15
ix
-------
Table of Contents
(cont'd)
Page
10.3 CHARACTERIZATION PROGRESS SUMMARY 10-16
11. REFERENCES 11-1
APPENDIX A: Schematics of Study Designs for Neurodevelopmental,
Two-Generation Reproductive and Development Studies A-l
APPENDIX B: List of Abbreviations and Acronyms B-l
x
-------
List of Tables
Number Page
1-1 Occurrence and Potential Sources of Perchlorate Releases to the Environment
as of Noveber, 2001 1-11
2-1 Gibbs Free Energies of Formation for Selected Anions in Aqueous Solution 2-2
2-2 Physicochemical Properties of Ammonium and Alkali Metal Perchlorates
at 25 °C 2-3
3-1 Percent Inhibition of Iodide Uptake in the Thyroid Gland of Spraque Dawley
Rats Dosed with Perchlorate 3-8
3-2 Interspecies and Intraspecies Differences in Thyroid Structure and T3, T4, and
Thyroid Stimulating Hormones 3-14
3-3 Mechanisms of Antithyroid-Mediated Neoplasia in Rodents 3-17
3-4 Main Symptoms and Effects of Hypothyroidism 3-23
3-5 Dietary Reference Intakes for Iodide 3-27
3-6 Minimum Database for Derivation of an Oral Reference Dose 3-31
3-7 Factors for Uncertainties in Applied Extrapolations Used to Derive Reference
Doses 3-31
3-8 Perchlorate Peer Review Recommended Studies Summary 3-33
4-1 Thyroid Disorders and Their Approximate Prevalences in the Human Neonatal
Period 4-5
4-2 Urine and Serum Perchlorate Values Before, During, and After the Ingestion
of 10 mg of Serum Perchlorate Daily for 14 Days 4-19
4-3 Urine and Serum Iodine Values Before, During, and After the Ingestion of
10 mg of Serum Perchlorate Daily for 14 Days 4-20
4-4 Thyroid 123I Uptakes Before, During, and After the Ingestion of 10 mg
Serum Perchlorate Daily for 14 Days 4-20
4-5 Summary of Human Population Studies 4-28
xi
-------
List of Tables
(cont'd)
Number Page
5-1 Benchmark Dose and Benchmark Dose Lower Confidence Limit Estimates
Calculated from Wolf (2000, 2001) Thyroid Histopathology Data 5-3
5-2 A Comparison of No-Observed-Adverse-Effect Levels and Lowest-Observed-
Adverse-Effect Levels from the Original 1998 Analyses and the 2001 Reanalyses
for Hormone and Morphometry on Thyroid Follicular Lumen Size 5-8
5-3 Benchmark Dose and Benchmark Dose Lower Confidence Limit Estimates from
Thyroid Histopathology in the "Effects Study" 5-56
5-4 NOAELs and LOAELs for Effects on Thyroid and Pituitary Hormones from
the Argus 2001 "Effects Study" 5-57
5-5 Summary of Immunotoxicity Test Results 5-101
6-1 Physiological Parameters for the Adult Male Rat and Human Physiologically
Based Pharmacokinetic Models 6-13
6-2 Chemical-Specific Parameters for the Adult Male Rat and Human Physiologically
Based Pharmacokinetic Models 6-16
6-3 Physiological Parameters for the Pregnant Rat and Fetus Physiologically Based
Pharmacokinetic Model 6-53
6-4 Perchlorate-Specific Parameters for the Pregnant Rat and Fetus Physiologically
Based Pharmacokinetic Model 6-54
6-5 Iodide-Specific Parameters for the Pregnant Rat and Fetus Physiologically
Based Pharmacokinetic Model 6-55
6-6 Physiological Parameters for Lactating Dam and Neonate Physiologically
Based Pharmacokinetic Model 6-81
6-7 Perchlorate-Specific Parameters for Lactating Dam and Neonate Physiologically
Based Pharmacokinetic Model 6-82
6-8 Iodide-specific Parameters for Lactating Dam and Neonate Physiologically
Based Pharmacokinetic Model 6-84
6-9 Sensitivity Analysis for Physiological Parameters in the Adult Male Rat
Model at 0.1 mg/kg Perchlorate Dose 6-108
xii
-------
List of Tables
(cont'd)
Number Page
6-10 Sensitivity Analysis for Chemical Specific Parameters in the Adult Male Rat
Model at 0.1 mg/kg Perchlorate Dose 6-109
6-11 Sensitivity Analysis for Physiological Parameters in the Adult Male Rat
Model at 1.0 mg/kg Perchlorate Dose 6-111
6-12 Sensitivity Analysis for Chemical-Specific Parameters in the Male Rat
Model at 1.0 mg/kg Perchlorate Dose 6-112
6-13 "Up-regulated" Values of Vmaxc_T/> After Perchlorate Drinking Water
Exposure in the Adult Male Rat Model 6-115
7-1 Physiologically Based Pharmacokinetic-Model Calculated Human Equivalent
Exposures to Various Experimental Doses in the Male Rat for 15 and 70 kg
Human Based on Perchlorate Area Under the Curve in Serum or Thyroid as
the Dose Metric 7-10
7-2 Ratio of Physiologically Based Pharmacokinetic-Derived Perchlorate Area
Under the Curve Serum Concentrations in Drinking Water for Various
Experimental Life Stages 7-11
7-3 Physiologically Based Pharmacokinetic-Model Calculated Human Equivalent
Exposures to Various Experimental Life Stages in the Rat Using Serum
Perchlorate Area Under the Curve as the Dose Metric 7-11
7-4 Physiologically Based Pharmacokinetic-Model Calculated Human Equivalent
Exposures to Various Experimental Doses in the Adult Male Rat for 15 and
70 kg Human Based on % Iodide Uptake Inhibition in the Thyroid 7-12
7-5 Physiologically Based Pharmacokinetic-Model Predicted % Inhibition of
Iodide Uptake in the Thyroid 7-13
7-6 Ratios of Physiologically Based Pharmacokinetic-Derived % Iodide Uptake
Inhibition in Drinking Water for Various Experimental Life Stages 7-13
7-7 Physiologically Based Pharmacokinetic-Model Calculated Human Equivalent
Exposures to Various Experimental Life Stages in the Rat Using % Iodide
Uptake Inhibition in the Thyroid as the Dose Metric 7-14
7-8 Default Dose-Response Procedures for Thyroid Carcinogens 7-24
xiii
-------
List of Tables
(cont'd)
Number Page
7-9 Data Demonstrating Antithyroid Activity 7-24
7A-1 Pearson's Rank Correlations Between Thyroid Hormones and Thyroid Stimulating
Hormone in Rats of the Caldwell et al. (1995) 14-Day Study 7A-1
7A-2 Spearman's Rank Correlations Between The Rank Order of Hormone Levels
and Histological Severity Rating Decrease in Follicular Lumen Size in Rats of
the Caldwell et al. (1995) 14-Day Study 7A-2
7A-3 Pearson's Rank Correlations Between Thyroid Hormones and Thyroid
Stimulating Hormone in Rats for the Combined 14- and 90-Day Data of
the Springborn Laboratories, Inc. (1998) Subchronic Rat Study 7A-2
7A-4 Pearson's Rank Correlations Between Thyroid Hormones and Thyroid
Stimulating Hormone for the 14-Day Data of The Springborn Laboratories,
Inc. (1998) Subchronic Rat Study 7A-2
7A-5 Pearson's Rank Correlations Between Thyroid Hormones and Thyroid
Stimulating Hormone of the 90-Day Data of the Springborn Laboratories,
Inc. (1998) Subchronic Rat Study 7A-3
7A-6 Pearson's Rank Correlations Between Thyroid Hormones and Thyroid
Stimulating Hormone for the F1 Rat Pups on PND5 in the Developmental
Neurotoxicity Study 7A-3
7B-1 Continuous Functions Used in Benchmark Dose Modeling 7B-1
7B-2 Benchmark Dose Estimates for Male Hormone Data of Caldwell et al.
(1995) 14-Day Rat Study, Using Kodell-West Algorithm 7B-3
7B-3 Coefficients and Goodness-of-Fit Statistics of Kodell-West (Quadratic
Polynomial) Model Fits to Male Hormone Data of Caldwell et al. (1995)
14-Day Rat Study 7B-4
7B-4 Benchmark Dose Estimates Using Power Function Fit to Combined
Male and Female Hormone Data of Caldwell et al. (1995) 14-day Rat Study 7B-5
7B-5 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates Using
Power Function Fit to Combined Male and Female Hormone Data of Caldwell
et al. (1995) 14-day Rat Study 7B-6
xiv
-------
List of Tables
(cont'd)
Number Page
7B-6 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates for
Combined Male and Female Hormone Data of 14-day Time Point in the
Springborn Laboratories, Inc. (1998) Subchronic Study 7B-7
7B-7 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates for
Combined Male and Female Hormone Data of 14-day Time Point in the
Springborn Laboratories, Inc. (1998) Subchronic Study 7B-8
7B-8 Benchmark Dose Estimates for Combined Male and Female Hormone Data
of 90-day Time Point in the Springborn Laboratories, Inc. (1998) Subchronic
Study 7B-9
7B-9 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates for
Combined Male and Female Hormone Data of 90-day Time Point in the
Springborn Laboratories, Inc. (1998) Subchronic Study 7B-10
7B-10 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates for
Hormone and Thyroid Morphometry Data of F1-Generation Pups at PND5
in the Developmental Neurotoxicity Study 7B-11
7B-11 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates for
Hormone Data of F1-Generation Pups at PND5 in the Developmental
Neurotoxicity Study 7B-12
7B-12 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates Using
the Linear Model Fit to the Motor Activity Data of F1-Generation Pups at
PND14 in the Developmental Neurotoxicity Study 7B-12
7B-13 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates Using
the Power Model Fit to the Hormone Data of Female Rabbits on Gestation
Day 29 in the Developmental Study 7B-13
7B-14 Benchmark Dose and Benchmark Dose 95% Lower Limit Estimates Using
the Power Model Fit to the Hormone Data of Female Rabbits on Gestation
Day 29 in the Developmental Study 7B-13
8-1 Results of Perchlorate Toxicity Tests in Aquatic and Terrestrial Species 8-18
8-2 Procedure for Deriving Tier II Water Quality Values for Sodium Perchlorate 8-20
xv
-------
List of Figures
Number Page
1-1 Sources and pathways of groundwater contamination for perchlorate 1-5
1-2 Distribution of perchlorate detected in public water supply sources in California ... 1-8
1-3 Locations of specific perchlorate manufacturers or users identified through
EPA Information Request responses from current manufacturers 1-9
1 -4 Locations of reported environmental releases of perchlorate to groundwater,
surface water, or soil 1-10
1 -5 Considerations for comprehensive characterization of perchlorate
contamination 1 -24
1 -6 Structure and membership of executive committee; subcommittees areas and
co-chairs of the Interagency Perchlorate Steering Committee 1-27
2-1 Chemical structure of perchlorate 2-1
3-1 Schematic representation of thyroid hormone biosynthesis and secretion in a
single thyroid follicular cell 3-10
3-2 Schematic of the hypothalamic-pituitary-thyroid axis and feedback mechanisms. .. 3-12
3-3 Comparison of the molecular dimensions for the perchlorate and iodide anions. ... 3-14
3-4 Schematic of antithyroid effects that influence thyroid carcinogenesis 3-17
3-5 Proliferative changes involved in the multistage characterization of thyroid
follicular cell neoplasia in rodents represent a morphologic continuum 3-19
3-6 Possible molecular events in human thyroid follicular carcinogenesis 3-20
3-7 Schematic representation of the role of the placenta in thyroid hormone
metabolism during human pregnancy 3-22
3-8 Approximate timing of major insults to the brain resulting from
hypothyroxinemia, superimposed on major neurodevelopmental events
in humans 3-25
3-9 Timelines of developmental processes in the nervous system of rats and
humans 3-26
xvi
-------
List of Figures
(cont'd)
Number Page
3-10 Schematic illustrating that a high confidence oral reference dose is based on
data that address all potentially critical stages over a lifetime 3-30
3-11 Schematic characterization of comprehensive exposure-dose-response continuum
and the evolution of protective to predictive dose-response estimates 3-32
3-12 Mode-of-action for perchlorate proposed by the U.S. EPA 3-36
5-1 Benchmark dose and benchmark dose lower limit estimates recalculated for
thyroid histopathology based on 2000 Pathology Working Group review 5-5
5-2 Distribution of benchmark dose and benchmark dose limit estimates shown
by "box and whisker" plots of colloid depletion, hypertrophy, and hyperplasia
from rat studies recalculated for thyroid histopathology based on 2000
Pathology Working Group review 5-6
5-3 Effects in the Caldwell et al. (1995) study of 14-day drinking water administration
of ammonium perchlorate to Spraque-Dawley rats on serum total T3, T4, and
thyroid stimulating hormone concentrations 5-22
5-4 Effects in the Caldwell et al. (1995) study of 14-day drinking water administration
of ammonium perchlorate to Spraque-Dawley rats on serum rT3 and
thyroglobulin concentrations 5-24
5-5 Effects from 90-day drinking water administration of ammonium perchlorate
to Spraque-Dawley rats on serum total T3 concentrations 5-29
5-6 Effects from 90-day drinking water administration of ammonium perchlorate
to Spraque-Dawley rats on serum total T4 concentrations 5-30
5-7 Effects from 90-day drinking water administration to ammonium perchlorate
to Spraque-Dawley rats on serum total thyroid stimulating hormone 5-31
5-8 Effects from maternal drinking water administration of ammonium perchlorate
to Spraque-Dawley rats on thyroid gland follicular lumen size in F1-generation
offspring on PND5 5-40
xvii
-------
List of Figures
(cont'd)
Number Page
5-9 Effects from maternal drinking water administration of ammonium perchlorate
to Spraque-Dawley rat F1-generation pups on serum total T3, T4, and thyroid
stimulating hormone concentrations 5-42
5-10 The effects of developmental exposure to perchlorate on motor activity in male
rats on PND14 5-44
5-11 Bayesian estimates of the posterior densities for the expected increase in the
logarithm of the number of ambulatory movements at the final habituation
time per unit dose increase of ammonium perchlorate 5-50
5-12 Bayesian estimate of the posterior density for the expected increase in the
logarithm of the number of ambulatory movements at the final habituation
time per unit dose increase of ammonium perchlorate for the combined data
from the two studies of motor activity effects 5-53
5-13 Lower confidence limit on the dose of ammonium perchlorate in drinking water
that produced a 10% increase in the incidence of colloid depletion in the thyroid
gland as a function of post-natal age of rat pups 5-56
5-14 Topograph of the approximate anatomical landmarks on the ventral and
dorsal surfaces of the brain used for making the morphometry measurements 5-64
5-15 Profile analysis of brain morphometry measurements for PND21 rat pup
brain regions 5-69
5-16 Effects from ammonium perchlorate in drinking water administration in
pregnant New Zealand rabbits during GD6 to GD28 on T3, T4, and
thyroid stimulating hormone concentrations 5-80
6-1 Mode-of-action for perchlorate proposed by the U.S. EPA 6-3
6-2 Schematic of thyroid and pituitary hormone levels with associated pathology
after acute versus chronic dosing with perchlorate 6-5
6-3 Schematic of parallelogram approach used for interspecies extrapolation 6-6
6-4 Illustration of how human equivalent exposure is calculated using physiologically
based pharmacokinetic models 6-7
XVlll
-------
List of Figures
(cont'd)
Number Page
6-5 Schematic of extended parallelogram approach used for perchlorate due to
effects at different life stages 6-8
6-6 Schematic for the adult male rat and human physiologically based
pharmacokinetic models of perchlorate and iodide distribution 6-11
6-7 Adult male rat physiologically based pharmacokinetic model predictions after
an acute intravenous dosing with radiolabeled perchlorate 6-26
6-8 Simulations illustrating the necessity of including plasma binding in the adult
male rat physiologically based pharmacokinetic model structure 6-27
6-9 Adult male rat physiologically based pharmacokinetic model predictions versus
data time course of perchlorate concentrations in the thyroid and cumulative
excreted perchlorate in the urine 6-29
6-10 Validation for male rat physiologically based pharmacokinetic model of
perchlorate disposition 6-30
6-11 Male rat physiologically based pharmacokinetic model predictions versus
data time course of iodide concentrations at two doses of 125I" with carrier,
0.033 mg/kg or 0.33 mg/kg, in the thyroid or in the serum 6-31
6-12 Male rat physiologically based pharmacokinetic model predictions versus data
time course of thyroid perchlorate concentrations in male rats during ingestion
of 30, 10, 3.0, 1.0, 0.1, or 0.01 mg/kg-day in drinking water for 14 days 6-32
6-13 Male rat physiologically based pharmacokinetic model predictions versus data
time course of iodide uptake inhibition in male rats administered perchlorate
either by a single iv dose or in drinking water for 14 days, followed by an iv
dose of 33 /ug/kg 125I"with carrier 6-33
6-14 Human physiologically based pharmacokinetic model predictions versus mean
131I" concentration time course in serum, thyroid, gastric juice, and urine 6-36
6-15 Simulations illustrating the necessity of including plasma binding in the human
physiologically based pharmacokinetic model structure 6-37
6-16 Human physiologically based pharmacokinetic model predictions versus data
of the observed cumulative urine excretion in male subjects dosed with
perchlorate 0.5, 0.1, or 0.02 mg/kg-day for 14 days 6-38
xix
-------
List of Figures
(cont'd)
Number Page
6-17 Human physiologically based pharmacokinetic model predictions versus
data of one subject's serum perchlorate concentration and corresponding
48-hour cumulative urine perchlorate 6-39
6-18 Human physiologically based pharmacokinetic model predictions versus
data of one subject's serum perchlorate concentration and corresponding
48-hour cumulative urine perchlorate 6-39
6-19 Human physiologically based pharmacokinetic model predictions versus
data of 48-hour cumulative urine perchlorate shown for two different subjects .... 6-40
6-20 Human physiologically based pharmacokinetic model predictions versus
data for thyroid radioactive iodide uptake on day 14 of perchlorate exposure at
0.5 mg/kg-day for a healthy female and male 6-41
6-21 Human physiologically based pharmacokinetic model predictions versus
data for thyroid radioactive iodide uptake on day 14 of perchlorate exposure
at 0.1 mg/kg-day for a healthy female and male 6-42
6-22 Human physiologically based pharmacokinetic model predictions versus
data for thyroid radioactive iodide uptake on day 14 of perchlorate exposure
at 0.02 mg/kg-day for a healthy female and male 6-43
6-23 Human physiologically based pharmacokinetic model predictions versus
data for thyroid radioactive iodide uptake on day 14 of perchlorate exposure
at 0.007 mg/kg-day for a healthy female and male 6-44
6-24 Validation for human physiologically based pharmacokinetic model 6-46
6-25 Validation for human physiologically based pharmacokinetic model 6-47
6-26 Validation for human physiologically based pharmacokinetic model 6-48
6-27 Schematic for the pregnant dam and fetal rat physiologically based
pharmacokinetic model of perchlorate and iodide distribution 6-52
6-28 Simulations illustrating the necessity of including plasma binding in the pregnant
dam and fetal rat physiologically based pharmacokinetic model structure 6-66
xx
-------
List of Figures
(cont'd)
Number Page
6-29 Pregnant dam and fetal rat physiologically based pharmacokinetic model
predictions versus data time course of perchlorate concentrations in maternal
serum and thyroid on GD20 6-67
6-30 Pregnant dam and fetal rat physiologically based pharmacokinetic model
predictions versus data time course of perchlorate concentrations in pooled
fetal serum on GD20 6-68
6-31 Pregnant dam and fetal rat physiologically based pharmacokinetic model
predictions versus data time course of 125I' radiolabeled iodide concentrations
in maternal serum, thyroid, mammary gland, and placenta on GD20 6-69
6-32 Pregnant dam and fetal rat physiologically based pharmacokinetic model
predictions versus data time course of 125I" radiolabeled iodide concentrations
in fetal serum on GD20 after an iv injection to the dam with 2.19 ng/kg 125I" 6-70
6-33 Validation for pregnant dam and fetal rat physiologically based pharmacokinetic
model 6-72
6-34 Validation for pregnant dam and fetal rat physiologically based pharmacokinetic
model 6-73
6-35 Validation for pregnant dam and fetal rat physiologically based pharmacokinetic
model 6-74
6-36 Validation for pregnant dam and fetal rat physiologically based pharmacokinetic
model 6-76
6-37 Schematic for the lactating dam and neonatal rat physiologically based
pharmacokinetic model of perchlorate and iodide distribution 6-80
6-38 Simulations illustrating the necessity of including plasma binding in the
lactating dam and neonatal rat physiologically based pharmacokinetic model
structure 6-93
6-39 Lactating dam and neonatal rat physiologically based pharmacokinetic model
predictions versus data time course of perchlorate concentrations in the
maternal thyroid and milk on PND5 and PND10 at doses in drinking water
to the dam of 10.0, 1.0, 0.1, or 0.01 mg/kg-day perchlorate 6-94
xxi
-------
List of Figures
(cont'd)
Number Page
6-40 Lactating dam and neonatal rat physiologically based pharmacokinetic model
predictions versus data time course of perchlorate concentrations in the serum
of male and female neonates on PND5 and PND10 at doses in drinking water
to the dam of 10.0, 1.0, 0.1, or 0.01 mg/kg-day perchlorate 6-94
6-41 Lactating dam and neonatal rat physiologically based pharmacokinetic model
predictions versus data time course of iodide concentrations in the maternal
serum or thyroid and in male or female neonatal pups on PND10 after an iv
dose to the lactating dams of 2.10 ng/kg 125I' 6-96
6-42 Validation for lactating dam and neonatal rat physiologically based
pharmacokinetic model 6-98
6-43 Validation for lactating dam and neonatal rat physiologically based
pharmacokinetic model 6-100
6-44 Validation for lactating dam and neonatal rat physiologically based
pharmacokinetic model 6-102
6-45 Validation for lactating dam and neonatal rat physiologically based
pharmacokinetic model 6-103
6-46 Upregulation of maximal capacity of active transport into the thyroid follicle
for perchlorate optimized by fitting to drinking water data in the rat 6-115
6-47 Michaelis-Menten fit of the "acute" male rat area under the curve for serum
and thyroidal perchlorate in ng/L hr 6-117
6-48 Michaelis-Menten fit of the "chronic" male rat area under the curve for serum
and thyroidal perchlorate . . 6-118
6-49 Michaelis-Menten fit of the human area under the curve for serum and thyroidal
perchlorate on exposure Day 2 6-120
6-50 Michaelis-Menten fit of the human equivalent exposure of perchlorate in
drinking water derived from the area under the curve for serum or thyroid
versus percent predicted inhibition in the rat after an "acute" iv dose 6-122
xxii
-------
List of Figures
(cont'd)
Number Page
7-1 Physiologically based pharmacokinetic-model calculated human equivalent
exposures to various experimental doses in the male rat for 15 and 70 kg
human based on perchlorate area under the curve in serum or thyroid as the
dose metric 7-10
7-2 Ratio of physiologically based pharmacokinetic-derived perchlorate area
under the curve serum concentrations in drinking water for various experimental
life stages 7-11
7-3 Physiologically based pharmacokinetic-model calculated human equivalent
exposures to various experimental life stages in the rat using serum perchlorate
area under the curve as the dose metric 7-11
7-4 Physiologically based pharmacokinetic-model calculated human equivalent
exposures to various experimental doses in the adult male rat for 15 and 70 kg
human based on % iodide uptake inhibition in the thyroid 7-12
7-5 Physiologically based pharmacokinetic-model predicted % inhibition of iodide
uptake in the thyroid 7-13
7-6 Ratios of physiologically based pharmacokinetic-derived % iodide uptake
inhibition in drinking water for various experimental life stages 7-13
7-7 Physiologically based pharmacokinetic-model calculated human equivalent
exposures to various experimental life stages in the rat using % iodide uptake
inhibition in the thyroid as the dose metric 7-14
7-8 Default dose-response procedures for thyroid carcinogens 7-24
7-9 Data demonstrating antithyroid activity 7-24
7A-1 Pearson's rank correlations between thyroid hormones and thyroid stimulating
hormone in rats of the Caldwell et al. (1995) 14-day study 7A-1
7A-2 Spearman's rank correlations between the rank order of hormone levels and
histological severity rating decrease in follicular lumen size in rats of the
Caldwell et al. (1995) 14-day study 7A-2
7A-3 Pearson's rank correlations between thyroid hormones and thyroid stimulating
hormone in rats for the combined 14- and 90-day data of the Springborn
Laboratories, Inc. (1998) subchronic rat study 7A-2
xxiii
-------
List of Figures
(cont'd)
Number Page
7A-4 Pearson's rank correlations between thyroid hormones and thyroid stimulating
hormone for the 14-day data of the Springbom Laboratories, Inc. (1998)
subchronic rat study 7A-2
7A-5 Pearson's rank correlations between thyroid hormones and thyroid stimulating
hormone of the 90-day data of the Springbom Laboratories, Inc. (1998)
subchronic rat study 7A-3
7A-6 Pearson's rank correlations between thyroid hormones and thyroid stimulating
hormone for the F1 rat pups on PND5 in the developmental neurotoxicity
study 7A-3
7B-1 Continuous functions used in benchmark dose modeling 7B-1
7B-2 Benchmark dose estimates for male hormone data of Caldwell et al.
(1995) 14-day rat study, using Kodell-West algorithm 7B-3
7B-3 Coefficients and goodness-of-fit statistics of Kodell-West (Quadratic
Polynomial) model fits to male hormone data of Caldwell et al. (1995)
14-day rat study 7B-4
7B-4 Benchmark dose estimates using power function fit to combined male and
female hormone data of Caldwell et al. (1995) 14-day rat study 7B-5
7B-5 Benchmark dose and benchmark dose 95% lower limit estimates using power
function fit to combined male and female hormone data of Caldwell et al. (1995)
14-day rat study 7B-6
7B-6 Benchmark dose and benchmark dose 95% lower limit estimates for combined
male and female hormone data of 14-day time point in the Springbom
Laboratories, Inc. (1998) subchronic study 7B-7
7B-7 Benchmark dose and benchmark dose 95% lower limit estimates for combined
male and female hormone data of 14-day time point in the Springbom
Laboratories, Inc. (1998) subchronic study 7B-8
7B-8 Benchmark dose estimates for combined male and female hormone data of
90-day time point in the Springbom Laboratories, Inc. (1998) subchronic
study 7B-9
xxiv
-------
List of Figures
(cont'd)
Number Page
7B-9 Benchmark dose and benchmark dose 95% lower limit estimates for
combined male and female hormone data of 90-day time point in the
Springborn Laboratories, Inc. (1998) subchronic study 7B-10
7B-10 Benchmark dose and benchmark dose 95% lower limit estimates for hormone
and thyroid morphometry data of F1-Generation pups at PND5 in the
developmental neurotoxicity study 7B-11
7B-11 Benchmark dose and benchmark dose 95% lower limit estimates for hormone
data of F1-generation pups at PND5 in the developmental neurotoxicity study . .. 7B-12
7B-12 Benchmark dose and benchmark dose 95% Lower limit estimates using the
linear model fit to the motor activity data of F1-generation pups at PND14
in the developmental neurotoxicity study 7B-12
7B-13 Benchmark dose and benchmark dose 95% lower limit estimates using the
power model fit to the hormone data of female rabbits on gestation day 29
in the developmental study 7B-13
7B-14 Benchmark dose and benchmark dose 95% lower limit estimates using the
power model fit to the hormone data of female rabbits on gestation day 29
in the developmental study 7B-13
8-1 A conceptual model of exposure of ecological endpoint receptors to perchlorate . . . 8-7
xxv
-------
Contributors and Reviewers
Chemical Manager/Assessment Author
Annie M. Jarabek*
National Center for Environmental Assessment
Immediate Office of the Director
U.S. Environmental Protection Agency
Washington, DC
Contributing Assessment Authors
Members of the U.S. Environmental Protection Agency perchlorate risk assessment review
team for the 2002 revised external peer review draft are listed below in alphabetical order. Those
identified by an asterisk (*) also served on the 1998 review team.
Special acknowledgment is paid to Dr. Eric Clegg who served as a contributing author to the
1998 assessment but who has since departed service with the Agency.
The authors also gratefully acknowledge the insights and contributions provided by Drs. Robert
MacPhail (NHEERL) and Bob Sonawane (NCEA-W).
Randy Bruins, Ph.D.*
National Center for Environmental
Assessment
Cincinnati, OH
Harlal Choudhury, Ph.D.*
National Center for Environmental
Assessment
Cincinnati, OH
Tim Collette, Ph.D.
National Exposure Research Laboratory
Athens, GA
Kevin Crofton, Ph.D.*
National Health and Environmental Effects
Laboratory
Research Triangle Park, NC
Vicki Dellarco, Ph.D.
Office of Pollution, Prevention, and Toxic
Substances
Washington, DC
David B. Dunson, Ph.D.
National Institutes of Environmental Health
Sciences
Research Triangle Park, NC
Andrew Geller, Ph.D.*
National Health and Environmental Effects
Laboratory
Research Triangle Park, NC
Michael Griffith, Ph.D.
National Center for Environmental
Assessment
Cincinnati, OH
Jean Harry, Ph.D.
National Institute of Environmental Health
Sciences
Research Triangle Park, NC
Brian H. Hill, Ph.D.
National Center for Environmental
Assessment
Cincinnati, OH
xxvi
-------
Contributing Assessment Authors
(cont'd)
Gary Kimmel, Ph.D.*
National Center for Environmental
Assessment
Washington, DC
Kevin Mayer*
Region 9
San Francisco, CA
Allan Marcus, Ph.D.*
National Center for Environmental
Assessment
Research Triangle Park, NC
Robert Park
National Institute of Occupational Safety
and Health
Cincinnati, OH
John M. Rogers, Ph.D.
National Health and Environmental Effects
Laboratory
Research Triangle Park, NC
Ralph Smialowicz, Ph.D.*
National Health and Environmental Effects
Laboratory
Research Triangle Park, NC
Glenn Suter, Ph.D.*
National Center for Environmental
Assessment
Cincinnati, OH
Douglas C. Wolf, DVM, Ph.D.
National Health and Environmental Effects
Laboratory
Research Triangle Park, NC
Edward Urbansky, Ph.D.*
National Risk Management Research
Laboratory
Cincinnati, OH
Internal EPA Reviewers
The following EPA scientists contributed their insights on internal Agency review of the revised
assessment.
David Bayliss
National Center for Environmental
Assessment
Washington DC
Nancy Beck, Ph.D.
National Center for Environmental
Assessment
Washington DC
Todd Borci
Region 1
Boston, MA
Joyce Donohue, Ph.D.
Office of Water
Washington DC
Lynn Flowers, Ph.D., D.A.B.T.
National Center for Environmental
Assessment
Immediate Office IRIS Staff
Washington DC
Sarah Levinson
Region 1
Boston, MA
XXVI1
-------
Internal EPA Reviewers (2001) (cont'd)
Chris Lau, Ph.D.
National Health and Environmental Effects
Research Laboratory
Research Triangle Park, NC
Susan Makris, Ph.D.
Office of Pollution, Prevention, and Toxic
Substances
Washington DC
Vince Nabholz, Ph.D.
Office of Pollution, Prevention, and Toxic
Substances
Washington DC
Cheryl Overstreet
Region 6
Kansas City, KA
Cornell Rosiu
Region 1
Boston, MA
MaryJane Selgrade, Ph.D.
National Health and Environmental Effects
Research Laboratory
Research Triangle Park, NC
Carolyn Smallwood
National Center for Environmental
Assessment
Cincinnati,OH
Sharon K. Taylor, DVM, Ph.D.
National Center for Environmental
Assessment
Washington DC
The following scientists in the California Environmental Protection Agency, Office of
Environmental Health Hazard Assessment were also asked under contract to provide the National
Center for Environmental Assessment with comments on the November 15, 2001 internal review
draft:
George V. Alexeef, Ph.D., D.A.B.T. Robert A. Howd, Ph.D.
Anna M. Fan, Ph.D. David Ting, Ph.D.
xxviii
-------
1998 External Peer Review Draft
The 1998 external peer review draft was reviewed at a public workshop held in San
Bernadino, CA on February 10-11, 1999. The workshop was sponsored by the EPA Office of
Solid Waste and Emergency Response (OSWER) and the Office of Water (OW), and conducted
by Research Triangle Institute. Members of the 1998 external review panel included:
Dr. Melvin Andersen
Colorado State University
Center for Environmental Toxicology and
Technology
Fort Collins, CO
Dr. David Brusick
Covance Laboratories, Inc.
Vienna, VA
Dr. Rick Cardwell
Parametrix, Inc.
Kirkland, WA
Dr. Charles Emerson
University of Massachusetts Medical Center
Worcester, MA
Dr. Joseph Haseman
National Institute of Environmental Health
Sciences
Biostatistics Branch
Research Triangle Park, NC
Dr. Curtis Klaassen (Chair)
University of Kansas Medical Center
Kansas City, KS
Dr. Susan Porterfield (Unable to Attend)
Medical College of Georgia
Augusta, GA
Dr. Rochelle Tyl
Research Triangle Institute
Center for Life Sciences and Technology
Research Triangle Park, NC
Dr. Kimber White
Medical College of Virginia
Strauss Immunotoxicity Research
Laboratory
Room 2011
Richmond, VA 23298
Dr. R. Thomas Zoeller
University of Massachusetts
Department of Biology
Morrill Science Center
Amherst, MA
xxix
-------
Contributors and Reviewers
(cont'd)
Internal EPA Reviewers (1998 External Review Draft)
The following EPA scientists contributed their insights on internal Agency review of the 1998
assessment:
Dorothy Canter, Ph.D.
Office of Solid Waste and Emergency
Response
Washington, DC
Christopher Cubbison, Ph.D.
National Center for Environmental
Assessment
Cincinnati, OH
Joyce Donohue, Ph.D.
Office of Water
Washington, DC
Richard Hill, Ph.D.
Office of Prevention, Pesticides, and Toxic
Substances
Washington, DC
Carole Kimmel, Ph.D.
National Center for Environmental
Assessment
Washington, DC
Chris Lau, Ph.D.
National Human and Ecological Effects
Research Laboratory
Research Triangle Park, NC
Susan Makris, Ph.D.
Office of Prevention, Pesticides, and Toxic
Substances
Washington, DC
Vince Nabholz, Ph.D.
Office of Prevention, Pesticides, and Toxic
Substances
Washington, DC
William van der Schalie, Ph.D.
National Center for Environmental
Assessment
Washington, DC
Jennifer Seed, Ph.D.
Office of Prevention, Pesticides, and Toxic
Substances
Washington, DC
Sherry Selevan, Ph.D.
National Center for Environmental
Assessment
Washington, DC
MaryJane Selgrade, Ph.D.
National Human and Ecological Effects
Research Laboratory
Research Triangle Park, NC
Carolyn Smallwood
National Center for Environmental
Assessment
Cincinnati, OH
Stan Smucker, Ph.D.
U.S. EPA Region 9, Superfund Division
San Francisco, CA
Dan Stralka, Ph.D.
U.S. EPA Region 9, Superfund Division
San Francisco, CA
xxx
-------
Preface
The purpose of this review is to provide scientific support and rationale for hazard
identification and dose-response assessments based on the emerging data for both human health
and ecological effects caused by exposures to perchlorate. It is not intended to be a
comprehensive treatise on the chemical or the toxicological nature of perchlorate.
In Chapter 10, the U.S. Environmental Protection Agency (EPA) has characterized its
overall confidence in the quantitative and qualitative aspects of hazard and dose-response (U.S.
Environmental Protection Agency, 1995) for both the human health and ecotoxicological effects
of perchlorate. Matters considered in this characterization include knowledge gaps,
uncertainties, quality of data, and scientific controversies. This characterization is presented in
an effort to make apparent the limitations of the individual assessments and to aid and guide the
risk assessor in the ensuing steps of the risk assessment process.
Development of these hazard identifications and dose-response assessments for perchlorate
have followed the general guidelines for risk assessments set forth by the National Research
Council (1983). Other EPA guidelines that were used in the development of this health risk
assessment include the Assessment of Thyroid Follicular Cell Tumors (U.S. Environmental
Protection Agency, 1998a), Guidelines for Neurotoxicity Risk Assessment (U.S. Environmental
Protection Agency, 1998b), 1996 Proposed Guidelines for Carcinogen Risk Assessment (Federal
Register, 1996), Guidelines for Reproductive Toxicity Assessment (U.S. Environmental
Protection Agency, 1996a), Use of the Benchmark Dose Approach in Health Risk Assessment
(Crump et al., 1995), Methods for Derivation of Inhalation Reference Concentrations and
Application of Inhalation Dosimetry (U.S. Environmental Protection Agency, 1994), Proposed
Interim Policy for Particle Size and Limit Concentration Issues in Inhalation Toxicology Studies
(Whalan and Redden, 1994), Guidelines for Developmental Toxicity Risk Assessment (Federal
Register, 1991), Recommendations for and Documentation of Biological Values for Use in Risk
Assessment (U.S. Environmental Protection Agency, 1988), The Risk Assessment Guidelines of
1986 (U.S. Environmental Protection Agency, 1987), and the Guidelines for Ecological Risk
Assessment (U.S. Environmental Protection Agency, 1998c).
The document presents the hazard identification or dose-response assessment for noncancer
toxicity for each route of exposure, either the oral reference dose (RfD) or the inhalation
xxxi
-------
reference concentration (RfC). The RfD and RfC are meant to provide information on long-term
effects other than carcinogenicity, although more recently, the value of mode-of-action
information to inform the potential for a continuum from noncancer toxicity as precursor lesions
to carcinogenicity presented as tumors has been recognized (Federal Register, 1996; Wiltse and
Dellarco, 1996). Consideration of this continuum is especially pertinent to the evaluation of the
potential toxicity of perchlorate. When such a continuum can be characterized, the dichotomous
approaches to "noncancer" versus "cancer" toxicity can be harmonized into one route-specific
estimate. The objective is to select a prominent toxic effect or key event that is pertinent to the
chemical's key mode of action, defined as a chemical's influence on molecular, cellular, and
physiological functions (Wiltse and Dellarco, 1996). A harmonized approach to the
neurodevelopmental and neoplastic effects of perchlorate is proposed herein.
In a default characterization without mode-of-action information, the RfD typically is
based, in part, on the assumption that a threshold exists for certain toxic effects, both for the
individual and the population; whereas, a threshold may not exist for other carcinogenic effects.
Thus, if the critical toxic effect is prevented, then all toxic effects are prevented. In general, the
RfD or RfC is an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily
oral exposure or continuous inhalation exposure to the human population (including sensitive
subpopulations) that is likely to be without deleterious noncancer effects during a lifetime. The
oral RfD is expressed in units of milligrams per kilogram per day. The inhalation RfC considers
toxic effects for both the respiratory tract as the portal of entry, as well as for effects remote to
the respiratory tract (extra-respiratory or systemic effects). The RfC is expressed in units of
milligrams per cubic meter.
The carcinogenicity assessment is meant to provide information on three aspects of the
carcinogenic risk assessment for perchlorate: the EPA classification and quantitative estimates
of risk from both oral and inhalation exposure. The classification reflects a weight-of-evidence
judgment of the likelihood that the agent is a human carcinogen and the conditions under which
the carcinogenic effects may be expressed.
The screening-level ecological risk assessment of environmental contamination by
perchlorate follows the Guidelines for Ecological Risk Assessment (U.S. Environmental
Protection Agency, 1998c).
xxxii
-------
Acknowledgments
The authors are indebted to the following individuals who imparted their insights, data,
experimental analysis, and expertise to improve specific areas of the report or to facilitate the
process of its development and review.
As noted in the introduction (Chapter 1), this assessment could not have been accomplished
without the cooperation of individuals who work for the governmental entities represented in the
Interagency Perchlorate Steering Committee. Each of the subcommittee members contributed to
discussions as the process evolved, via stakeholder forums or meetings, and the integrated
approach to the overall risk characterization framework began to materialize. Special
acknowledgment for oversight of the testing strategy endeavor, notably communication with the
contract labs, expediting data delivery, and writing reports goes to Lt. Col. Dan Rogers (U.S. Air
Force Materiel Command); Dr. Richard Stotts, Dr. Dave Mattie, and Capt. David Tsui (Air Force
Research Laboratory/Human Effectiveness Directorate [AFRL/HEST], Operational Toxicology
Branch); and Cornell Long and Dr. Ron Porter (AFRL/HEST, Human Systems Center).
Special thanks is paid to three scientists who were especially dedicated to development of
the physiologically-based pharmacokinetic (PBPK) models for AFRL/HEST: Elaine Merrill,
Rebecca Clewell, and Dr. Jeffrey Gearhart. Terri Sterner is particularly appreciated for her
attention to reference retrieval and electronic transfer of data.
Other individuals at AFRL/HEST at Wright-Patterson Air Force Base should be noted for
their invaluable technical contributions: Dr. William Baker, for all the histopathology analysis
and reports that he generated in a short period of time in 1998; Latha Narayanan, for her expert
and reliable analyses of thyroid and pituitary hormone data over the years; and Drs. Jeff Fisher
and Kyung Yu for their work on iodide and perchlorate kinetics. Deirdre Mahle is also
acknowledged for her contributions to the experimental work.
The Perchlorate Study Group (PSG), particularly Michael Girard, is recognized for its aid
in sponsoring studies and ensuring timely data delivery in appropriate formats for EPA analyses.
Toxicology Excellence for Risk Assessment (TERA) (notably Michael Dourson, Joan
Dollarhide, and Jacqueline Patterson), also was very responsive in this regard on behalf of the
PSG.
xxxiii
-------
Additional gratitude is expressed to those individuals who submitted unpublished reports
and literature summaries for consideration by the EPA review team. Notable contributions were
provided by Dr. Steven Lamm of Consultants in Epidemiology and Occupational Health, Inc.,
Drs. Gay Goodman and Richard Pleus of Intertox, and Dr. Monte Greer of Oregon Health
Sciences University.
Drs. William Farland and George Alapas are paid special thanks for management and
sheparding of this assessment internally in the agency and for allocating expertise and funds from
across ORD and NCEA to support the analyses. Drs. Amy Mills and Susan Rieth of the IRIS
program are appreciated for their dedication to managing the peer review contract and associated
logistics. Kate Schalk and staff at Environmental Research Group, Inc. are acknowledged for
conducting the peer review.
The authors are indebted to the dedication and expertise of the following individuals of
OAO Corporation for their roles in 1998 document production: John Barton, for project
coordination and technical editing; Carolyn Perry, Bettye Kirkland, and Yvonne Harrison, for
word processing; Dave Leonhard and Veda Williams, for graphic arts; and David Belton, for
reference retrieval and editing.
The following are all gratefully acknowledged for professional production of the 2002
external review draft document: Carol Seagle, for technical editing; Carolyn Perry and Kelly
Quinones for word processing; John Bennett, for reference retrieval and editing; and Dave
Leonhard and Diane Caudhill for graphic arts.
Richard Wilson (EPA, NCEA), as always, is applauded for his tireless and cheerful
dedication to the task of photocopying both drafts of the document and its supporting materials
throughout the years.
xxxiv
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
EXECUTIVE SUMMARY
The purposes of this document is to present an assessment that updates previous
provisional values issued by the U.S. Environmental Protection Agency (EPA) for an oral
reference dose (RfD) for perchlorate and revises the assessment previously released as a draft
external review document (U.S. Environmental Protection Agency, 1998d). The objective of this
assessment is to derive a human health risk estimate, based on an evaluation of its potential to
cause toxicity or cancer, and to provide a screening-level ecological risk assessment for
perchlorate based on all toxicity data that recently have become available to the Agency as of fall
2001. Another important objective was to evaluate the evidence for indirect exposures, i.e.,
those exposures not by direct ingestion of contaminated water. This revised assessment
incorporates data from new studies and analyses in response-level to recommendations made at a
previous peer review of the 1998 draft (Research Triangle Institute, 1999). Most of these data
were obtained as results of a testing strategy that was designed with knowledge of the mode of
action for perchlorate toxicity that identified major data gaps in the data available prior to 1997.
This executive summary concisely presents key findings from the present assessment.
SUMMARY FINDINGS
Sources of Perchlorate Contamination and Occurrence
• Perchlorate is an oxidizing anion that originates as a contaminant in ground and surface waters
from the dissolution of ammonium, potassium, magnesium, or sodium salts. Perchlorate is
exceedingly mobile in aqueous systems and can persist for many decades under typical ground
and surface water conditions.
• Ammonium perchlorate is manufactured for use as the oxidizer component and primary
ingredient in solid propellant for rockets, missiles, and fireworks. Because it is a reducing
agent, it can undergo a variety of intramolecular redox reactions that lead to the release of
gaseous products. Through such reactions, it acts as a thrust booster.
• Perchlorate salts are also used on a large scale as a component of air bag inflators. Perchlorate
salts are also used in nuclear reactors and electronic tubes, as additives in lubricating oils, in
January 16, 2002
E-l
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
tanning and finishing leather, as a mordant for fabrics and dyes, in electroplating, in aluminum
refining, and in rubber manufacture, as a mordant for fabrics and dyes, and in the production of
paints and enamels. Chemical fertilizer had been reported to be a potential source of
perchlorate contamination, but new investigations by the Agency have determined that this is
not an issue for agricultural applications.
• Large-scale production of perchlorate-containing chemicals in the United States began in the
mid-1940s. Because of its shelf life, perchlorate must be washed out of the United States'
missile and rocket inventory to be replaced with a fresh supply. Thus, large volumes have been
disposed of in various states since the 1950s.
• Perchlorate began to be discovered at various manufacturing sites and in well water and
drinking water supplies within the months following the April 1997 development of an ion
chromatography analytical method that achieved a method detection limit (MDL) of
approximately 1 ppb and a minimum reporting limit (MRL) of 4 ppb. There are 20 states with
confirmed releases in ground or surface water. There are 40 states that have confirmed
perchlorate manufacturers or users based on EPA Information Request responses.
In California, most of the locations where perchlorate has been detected are associated with
facilities that have manufactured or tested solid rocket fuels for the Department of Defense or
the National Aeronautics and Space Administration.
• To date, there has not been a systematic national survey of perchlorate occurrence and a
National Primary Drinking Water Regulation for perchlorate does not currently exist.
Perchlorate was placed on the Contaminant Candidate List (CCL) in March 1998. The CCL
lists priority contaminants (defined as either known or anticipated to occur in public water
systems) in need of research, guidance development, regulatory determinations, or monitoring
by the states. Perchlorate was listed as a contaminant that required additional research and
occurrence information before regulatory determinations could be considered.
• Perchlorate was placed on the Unregulated Contaminants Monitoring Rule (UCMR) in March
1999 (Federal Register, 1999) to gather needed exposure information. Under the UCMR, all
large public water systems and a representative sample of small public water systems were
required to monitor for perchlorate beginning in January 2001. This effort does not extend to
investigating potential sources in ground and surface water that have not migrated into public
water supplies. Identification of the magnitude and extent of perchlorate occurrence in the
January 16, 2002
E-2
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
environment is important in assessing the routes of exposure to humans and for determining the
different types of organisms and ecosystems that may be affected.
• In early 2000, an analytical method to detect perchlorate in drinking water (EPA Method 314.0)
using ion chromatography was published as a direct final rule (Federal Register, 2000). The
EPA Method 314.0 was also approved as a monitoring method for the UCMR (Federal
Resister, 2000). The MDL for the method is 0.53 ppb and the MRL is 4 ppb. Improvements
developed commercially in the analytical capabilities may lower the MRL to the sub-part per
billion level in the near future.
• Adequate exposure characteristics of transport and transformation in the environment are also
absent. Preliminary biotransport studies at six contaminated sites indicate a potential for
uptake into plant and animal tissues in ecosystems. Extension of analytical methods to detect
perchlorate in plant and animal tissues awaits validation before a conclusive determination can
be made.
An Integrated Approach to Comprehensive Risk Characterization
• Perchlorate is of concern for several reasons. First, there were uncertainties in the toxicological
database available that could be used to evaluate the potential for perchlorate to produce human
health effects when present at low levels in drinking water. The purpose of the targeted
toxicity testing strategy was to develop a database to address key data gaps. Secondly, the
actual extent of the occurrence of perchlorate in ground and surface waters is not known at this
time. Additionally, the efficacy of different treatment technologies for various water uses (such
as drinking water or agricultural applications) and different scales (i.e., large or small volumes)
is still being determined. Finally, the extent and nature of ecological impact or transport and
transformation phenomena in various environmental media have only, as yet, been studied
superficially.
•To adequately and comprehensively characterize the risks posed by perchlorate contamination
and to develop scientifically-based management strategies that effectively mitigate the potential
risks posed by perchlorate contamination, several advances are essential. The analytical
methods used to characterize various exposures must be accurate and precise. The exposure
estimates cannot be gauged with respect to their risk unless robust health and ecological risk
estimates are available. Treatment technologies should be targeted to levels of concern and
January 16, 2002
E-3
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
tailored to the intended water use. Technology transfer is necessary so that all affected parties
and concerned citizens are apprised of accurate and reliable information that is up to date with
the evolving state of the science.
• The toxicity testing strategy was expedited through a unique partnership between the
Department of Defense and EPA, together with members of an Interagency Perchlorate
Steering Committee (IPSC), which includes other governmental representatives from the
National Institute for Environmental Health Sciences (NIEHS) and affected state, tribal, and
local governments.
• The charge of the IPSC is to facilitate and coordinate accurate accounts of related technological
issues (occurrence surveys, health assessment, ecotoxicology assessment, treatability, waste
stream handling, and analytical detection). This assessment is intended to address the need for
evaluation of perchlorate's potential to cause human health effects or impact on ecological
systems, based on currently available data.
Physicochemical Characteristics
• As an oxidant, perchlorate is kinetically nonlabile. This means the reduction of the central
chlorine atom from an oxidation state of +7 (perchlorate) to -1 (chloride ion) occurs extremely
slowly. Sorption is not expected to attenuate perchlorate because it absorbs weakly to most soil
minerals. Natural chemical reduction in the environment is not expected to be significant.
These two factors account for perchlorate being both very mobile in aqueous systems and
persistent for many decades under typical ground and surface water conditions.
• The activation energy to perchlorate reduction is so high that it cannot be expected to act as an
oxidant under human physiological conditions (i.e., dilute solution, unelevated temperatures,
neutral pH). This is supported by absorption, distribution, metabolism, and elimination studies
that show perchlorate is excreted virtually unchanged in the urine after absorption.
Hazard Identification and Mode of Action Testing Strategy
• The health effects and toxicity database available in the spring of 1997 was determined to be
inadequate for quantitative risk assessment by an independent (non-EPA) peer review. A
testing strategy was developed based on a hazard identification using the available data and the
suspected mode of action for perchlorate to target testing on potential effects of perchlorate.
January 16, 2002
E-4
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Data from this effort was used to support the previous EPA draft assessment and this revised
assessment in 2002.
•To design a testing strategy based on the mode of action for a chemical, it is necessary to
understand its toxicokinetics and toxicodynamics. Perchlorate is readily absorbed from the
intestinal tract, and oral uptake is considered to be the major route of exposure. Because of its
high charge, perchlorate does not pass readily through the skin. Exposure via inhalation is
expected to be negligible because the vapor pressure of perchlorate salts and acids is expected
to be low at room temperatures. Droplet size during showering likely would preclude
inhalation of perchlorate-contaminated water as an aerosol. Perchlorate is known to inhibit the
uptake of iodide in the thyroid at the sodium (Na+)-iodide (I-) symporter, or NIS, thereby
causing a reduction in the hormones thyroxine (T4) and triiodothyronine (T3). When these
hormones enter the blood circulation, they are bound to plasma proteins. There may be other
locations of inhibition of iodide transport in the gland. Perchlorate itself is not metabolized in
the thyroid or peripheral tissues.
• Control of circulating concentrations of these hormones is regulated primarily by a negative
feedback known as the hypothalamic-pituitary-thyroid axis or feedback system involving three
organs: (1) the thyroid, which produces T4 and T3; (2) the pituitary gland which produces
TSH; and (3) the hypothalamus, which also responds to and helps to maintain optimal T4 and
T3 levels. The hypothalamus stimulates the pituitary gland through thyrotrophic-releasing
hormone (TRH) to produce thyroid stimulating hormone (TSH), which then prompts the
thyroid to produce T4 and T3. Cells in the hypothalamus and pituitary gland respond to the
levels of circulating T4 and T3, such that when thyroid production levels are low, there is a
signal to increase the output of TRH and TSH. Circulating hormone levels (T4, T3, and TSH)
can be monitored readily to serve as biomarkers of exposure and effect of agents that disrupt
the status of this negative feedback system.
• The hypothalamic-pituitary-thyroid feedback system for regulation of thyroid hormones is
conserved across species. Differences in plasma protein binding between rats and humans
account for differences in the circulating half-life of the hormones and in thyroid response to
TSH between the species. New studies since 1999 have confirmed that the inhibition of iodide
uptake by perchlorate at the NIS is essentially the same sensitivity across species. This is
January 16, 2002
E-5
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
important when considering decrements in T4 as important to neurodevelopmental effects
versus neoplasia that results in the gland due to stimulation by TSH.
• Given its mode of action as an inhibitor of iodide uptake that results in disturbances of the
hypothalamic-pituitary-thyroid axis, concerns arose about the potential for perchlorate to cause
carcinogenic, neurodevelopmental, developmental, reproductive, and immunotoxic effects.
Further, there is concern for ecotoxicology effects on various aquatic and terrestrial plants and
animals.
• The human health testing strategy for perchlorate developed in 1997 originally included eight
different recommended studies to address data gaps and enhance the mechanistic information
on the mode of action. The goal of these studies was to provide a comprehensive database on
which to arrive at a revised human health risk assessment with greater confidence than previous
recommended provisional values. These studies are described briefly below.
(1) A 90-day oral bioassay to identify other target tissues in young adult rats; to provide data
on the effects of repeated exposures to perchlorate on T3, T4, and TSH levels; to
evaluate recovery of effects after 30 days; and to screen for some reproductive
parameters. A genotoxicity assay also was performed on rats from the terminal sacrifice.
(2) A neurodevelopmental study in rats to evaluate the potential for functional and
morphological effects in offspring from the mother exposed during pregnancy and
lactation.
(3) A Segment n developmental study in rabbits to evaluate the potential for perchlorate to
cause birth defects and to provide data on thyroid hormone effects in a second species
other than the rat.
(4) A two-generation reproductive toxicity study to evaluate the potential for perchlorate to
cause deficits in reproductive performance in adult rats and for toxicity in the young
offspring.
(5) Absorption, distribution, metabolism, and elimination (ADME) studies to characterize
the pharmacokinetics of perchlorate in laboratory animals and humans and to provide
data necessary to allow construction of models for quantitative description of different
internal dose metrics and interspecies extrapolation.
January 16, 2002
E-6
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
(6) Mechanistic studies that characterize the effects of perchlorate on the iodide uptake
mechanism across species as a link with the ADME studies to aid in the quantitative
extrapolation of dose across species.
(7) A battery of genotoxicity assays to evaluate the potential for carcinogenicity by
evaluating the potential for direct effects on deoxyribonucleic acid (DNA).
(8) Immunotoxicity studies to evaluate the potential for perchlorate to disrupt immune
function, including cell-mediated and humoral toxicity.
• After the External Peer Review in 1999, additional studies were performed to replicate the
neurodevelopmental study (i.e., changes in brain morphometry and motor activity); determine
the developmental toxicity potential in rats versus rabbits; investigate additional aspects of
immunotoxicity; and develop a consistent nomenclature and scoring system for the
histopathological lesions in the thyroid gland. Additional pharmacokinetic data was also
developed into physiologically-based pharmacokinetic (PBPK) models of perchlorate and
iodide distribution.
• A battery of ecological screening tests as part of the 1997 testing strategy was conducted as
part of the 1997 testing strategy in laboratory organisms representative of ecological receptors
across soil, sediment, and water to evaluate dose-response relationships. These were
considered to be a tier of tests to give an idea of gross toxicity that would determine the need
and types of tests to be performed in the next tier. The tests did not measure the amount of
perchlorate in the tissues of the species being tested. Based on stakeholder input and the need
for a more focused battery of tests, lettuce was substituted for duckweed because of Tribal
concerns regarding the sizable lettuce crop along the Colorado river. The following species
were selected for the first round of testing:
(1) Daphnia magna (water flea) to represent an aquatic invertebrate
(2) Ceriodaphnia magna (water flea) to represent an aquatic invertebrate
(3) Lactuca sativa (lettuce) to represent a vascular plant
(4) Pimephales promelas (fathead minnow) to represent an aquatic invertebrate
(5) Eisenia foetida (earthworm) to represent a soil invertebrate
(6) Microtus pennsylvanicus (meadow vole) to represent an herbivore
• Other studies in the set of tests included the Frog Embryo Teratogenesis Assay: Xenopus
(FETAX) and a phytoremediation study to examine uptake, distribution, and degradation in
January 16, 2002
E-7
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
experimental systems with rooted cuttings of woody plants, including willow, Eastern
Cottonwood, and eucalyptus.
• Additional studies, some of chronic duration, on effect levels in aquatic animals, an aquatic
plant, a terrestrial plant, and a soil invertebrate have been performed since 1999. A study of
perchlorate occurrence in six selected sites with known or suspected contamination also
examined perchlorate concentrations in site media and in various ecological receptors.
Human Health Assessment
• The testing strategy confirmed that the target tissue for perchlorate toxicity was the thyroid
gland. Anti-thyroid effects included iodide uptake inhibition, perturbations of T3, T4, and TSH
hormones, and thyroid histopathology in adult, fetal, and postnatal rats across studies with a
range of experimental design. Thyroid weight in these studies was also effected. Other than
effects in the thyroid, no effects were observed in rabbits of the developmental study, but the
developmental study in rats identified 30 mg/kg-day as the lowest observed adverse effect level
(LOAEL).
• Competitive inhibition of iodide uptake at the NIS by perchlorate is the key event leading to
both potential neurodevelopmental and neoplastic sequelae. The decrement in iodide uptake
leads to subsequent drops in T4 and T3 that can lead to permanent neurodevelopmental
deficits. Because of strong correlations between changes in iodide uptake inhibition with
decrements in T3 and T4; between T3 and T4 with changes in TSH; and between changes in
T3, T4, or TSH with thyroid histopathology, an assessment model was proposed that used the
changes in T3, T4, and TSH as the precursor lesions to subsequent effects that potentially could
lead to thyroid tumors or to altered neurodevelopment. This assessment approach essentially
harmonizes noncancer and cancer approaches because it is presumed that the no-observed-
adverse-effect-level (NOAEL) for the precursor lesions will preclude any subsequent sequelae
at higher doses.
° Thyroid tumors were observed in previous studies in rats exposed in long-term bioassays at
high doses. Thyroid tumors were more recently also diagnosed in the first-generation (Fl)
adults (second parental generation [P2]) at 19 weeks in a two-generation reproductive study.
Both the latency and incidence of these tumors were significant relative to the entirety of the
National Toxicology Program data base for this type of tumor and in this strain of rat. These
January 16, 2002
E-8
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
effects and the demostration of a progression with duration of effects on hormones and thyroid
histopathology in the 90-day study raised the concern that extended exposures to perchlorate
may change the hypothalamic-pituitary-feedback system or the cellular sensitivity and demand
for thyroid hormones.
• The rat model is considered relevant yet conservative for human health risk assessment of
potential thyroid neoplasia because of the differences in thyroid structure and hormone
half-lives. Perchlorate was demonstrated to be nongenotoxic in the testing battery employed,
suggesting the antithyroid effects are an indirect mode of action for thyroid tumor formation.
• Due to the age- and time-dependent nature of the key event of perchlorate toxicity and its
anti-thyroid effects, the revised RfD was based on weight-of-the-evidence approach to the
entire data base. The RfD is proposed to be protective of both neurodevelopmental and
neoplastic sequelae. An administered dose of 0.01 mg/kg-day was supported as a lowest-
observed-adverse-effect level (LOAEL) based on effects on brain morphometry in pups from a
PND21 sacrifice in a neurodevelopmental study that repeated similar observations made in a
similar 1998 study, hormonal effects indicative of hypothyroidism (decreased T4 and increased
TSH) in the dams of those same pups on GD21, thyroid histopathology and hormone changes
in these same pups at various developmental stages (GD21, PND4, PND9, and PND21),
thyroid histopathology and hormone changes at the 14- and 90-day sacrifice dates in a
subchronic study, and indications of immunotoxicity (dermal contact hypersensitivity).
• A human equivalent exposure (HEE) was calculated using physiologically-based
pharmacokinetic (PBPK) models for interspecies adjustment based on the area under the curve
(AUC) of perchlorate in the serum as the dose metric. The HEE for the maternal dams was
chosen for operational derivation because brain morphometry effects may have been
programmed in utero and because the dams of effected pups were hypothyroid.
• A composite uncertainty factor of 300 was used to address uncertainties in the extrapolations
required for the RfD derivation. A three-fold factor for intraspecies variability was retained
due to the variability observed in the data and PBPK modeling for the adult humans and
because the subjects used to develop the models did not provide kinetic data for the potentially
susceptible population. There was also uncertainty in the parallelogram approach to extending
the adult structures to predict doses for different life stages in the human. A full factor of ten
was applied to extrapolate the LOAEL for the adverse effects (brain morphometry, colloid
January 16, 2002
E-9
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
depletion and hormone changes) observed in various studies at the 0.01 mg/kg-day dosage
level. A three-fold factor for duration was applied due to the concern for the biological
importance of the statistically significant increase in tumors observed in the F1-generation pups
(second parental, P2 generation) at 19 weeks and the evidence for progression of effects with
extended exposure in the 90-day study. The finding of tumors at 19 weeks raised concern for
in utero programming, i.e., that disruption of thyroid hormones in the developing fetus may
predispose the developing neonate and adult to future insults to the thyroid gland. This factor
can also be viewed as part of a data base deficiency since there are no adequate long-term
bioassays of perchlorate. Finally, a three-fold factor was applied for inaccurate characterization
of immunotoxicity since recent studies reinforced concern for this potential endpoint. Because
the test article was ammonium perchlorate, an adjustment factor of 0.85 was made for the
percent of molecular weight of the salt from ammonium (15.35%), so that the RfD is expressed
for perchlorate as the anion alone. This was done to be compatible with the analytical methods
that measure the anion in environmental samples and because most perchlorate salts readily
dissolve in water. The resultant revised RfD value for perchlorate is 0.00003 mg/kg-day.
Confidence in the principal study, the data base and the RfD were all designated as medium.
Screening Ecological Risk Assessment
• A secondary acute value of 5 mg/L (as perchlorate) was derived to be protective of 95% of
aquatic organisms during short-term exposures with 80% confidence. The secondary chronic
value of 0.6 (as perchlorate) likewise was derived to be protective of 95% of aquatic organisms
during short-term exposures with 80% confidence. These values were derived based on
sodium perchlorate and are probably protective even if ammonium perchlorate is the
contaminant released. Calculated ammonia-nitrogen concentrations corresponding to those
values are below the acute and chronic ambient water quality criteria for ammonia, regardless
of pH.
• For terrestrial plants, the quartile inhibitory concentrations for growth in soil and sand were
78 mg/kg (293 mg/L) and 41 mg/kg (160 mg/L), respectively. A factor of 10 was applied to
account for interspecies variance to obtain a screening benchmark of 4 mg/kg.
January 16, 2002
E-10
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
• Because of limited data on effects for soil invertebrates, a conservative estimate of a threshold
for soil community effects was derived at 1 mg/kg. The equivalent aqueous phase benchmark
is 2.8 mg/L.
• A factor of 10 for interspecies variance and LOAEL to NOAEL extrapolation was applied to
the human health risk LOAEL estimate based on rat data (0.01 mg/kg-day) to obtain a
screening benchmark of 0.001 mg/kg-day for the representative herbivore (meadow vole)
because it also is a rodent. The population-level implications of this effect are unknown, but it
seems likely that such effects on the thyroid could diminish survivorship and fecundity, which
would diminish population production.
• Data are available showing that perchlorate accumulates in the tissues of exposed fish,
amphibians, and invertebrates. However, data are insufficient to determine whether perchlorate
is concentrated in those tissues to levels exceeding the levels of exposure. By contrast, several
studies have shown that perchlorate is taken up and concentrated in aerial plant parts, especially
leaves, although studies designed for the purpose of quantifying plant concentration factors
have not yet been conducted.
Uncertainties and Assessment Research Needs
• Accurate exposure information is a requisite for risk characterization for both human and
ecological assessments. These data should include transport and transformation processes,
notably the fate of perchlorate in irrigated soils because of the potential for evaporative
concentration.
• Research concerning the human health risks of perchlorate needs to better characterize the
dose-response for perchlorate inhibition of iodide uptake in adults, fetuses, and neonates. More
definitive studies linking iodide uptake inhibition and the degree of perturbation of the
hypothalamic-pituitary-thyroid axis (i.e., changes in T3, T4, and TSH levels) and association
with neurobehavioral problems, thyroid changes, and neoplastic sequelae may continue to
improve the confidence in the assessment. Understanding the relative sensitivity of laboratory
animal assays of neurodevelopmental effects versus epidemiological studies of
neuropsychological development also needs to be advanced. Research on potential factors
influencing sensitivity is critically requisite. Animal models of thyroid impairment such as
iodide deficiency and "womb to tomb" exposure designs should be explored.
January 16, 2002
E-l 1
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
• Because only a screening tier of tests has been performed, the major uncertainty derives from
data gaps. Data on bioaccumulation in aquatic biota would allow evaluation of exposure of
organisms that feed on fish and other aquatic organisms. Effects of perchlorate on algae and
aquatic macrophytes are required to estimate risks to aquatic primary producers. Data on
bioaccummulation in aquatic plants are necessary to assess direct impact to primary consumers
(i.e., planktonic and benthic invertebrate communities). Data on accumulation in terrestrial
vascular plants also should be investigated further. The factor applied for the use of subchronic
data in fish could be addressed by chronic effect testing. Effects also should be determined in
nondaphnid invertebrates and of dietary exposure in birds and herbivorous or litter-feeding
invertebrates.
Risk Characterization
• As noted above, the lack of exposure information precludes comparison with the human health
and ecological toxicity assessment for accurate characterization of risk. Indirect human
exposure pathways can be addressed best by a new EPA document, Methodology for Assessing
Health Risks Associated with Multiple Pathway of Exposure to Combustor Emissions, which is
scheduled for final release in January 2002.
• Noncancer neurobehavioral effects have been shown at lower doses. The estimate for
perchlorate has been based on precursor effects considered protective for both the thyroid
neoplasia and neurodevelopmental effects. It is appropriate for comparison against direct oral
exposures. The frequency and magnitude of exposure are key attributes for characterization
compared with those assumptions of continuous lifetime exposure assumed in the derivation.
The degree to which the particular suspected population at risk fits with the assumptions used
in the RfD derivation should be kept in mind when performing any risk characterization.
Further, RfD estimates are not intended to serve as a "bright line" because, by definition, there
is an order-of-magnitude uncertainty around the estimate. This typically translates into a range
of threefold below to threefold above the RfD.
• Ecological risk could not be precluded nor accurately characterized because of the significant
data gaps described above.
January 16, 2002
E-12
Draft-Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
1. INTRODUCTION
The purpose of this document is to revise the previous human health and ecological risk
assessment external review draft (ERD) document (U.S. Environmental Protection Agency,
1998d). This revision is based on recommendations made at the 1999 external peer review
(Research Triangle Institute, 1999). The peer review panel recommended some alternative
analyses and several additional studies. This revised assessment addresses these
recommendations and is based on all data made available to the Agency as of Fall 2001;
including new studies from the perchlorate testing strategy. The purpose of this chapter is to
provide background information on the current status of perchlorate (C104 ) contamination in the
United States and an historical perspective on how certain issues of concern have evolved to
prominence. The role of this risk assessment will be placed in context with respect to the overall
integrated approach to addressing the perchlorate contamination and regulatory readiness.
1.1 PRODUCTION USES AND SOURCES OF PERCHLORATE
CONTAMINATION
Perchlorate is an oxidizing anion that originates as a contaminant in ground and surface
waters from the dissolution of perchloric acid and of the salts including ammonium, potassium,
magnesium, or sodium. With the exception of potassium perchlorate, each of these compounds
is extremely soluble. Potassium perchlorate is regarded as sparingly soluble; however, it may
dissolve completely under the appropriate environmental conditions.
Ammonium perchlorate is the oxidizer and primary ingredient (by mass) in solid propellant
for rocket motors. For example, ammonium perchlorate (NH4C104) makes up 69.7% of the
propellant for the space shuttle rocket motors and 65 to75% of the Stage I motors of the
Minuteman in and 68% of the Titan missile motors (Rogers, 1998). Because the ammonium ion
is a reducing agent, ammonium perchlorate can undergo a variety of intramolecular redox
reactions that lead to the release of gaseous products. The explosive decomposition shown in
Equation 1-1 is induced thermally and occurs at temperatures below 300 °C (Schilt, 1979a).
January 16, 2002
1-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
4 NH4C104(s) - 2 Cl2(g) + 3 02(g) + 2 N20(g) + 8 H20(g)
(1-1)
Through such reactions, ammonium perchlorate also acts as a thrust booster. Even after such
decomposition, the dichlorine and dioxygen thus produced remain capable of engaging in
subsequent redox reactions with fuels.
Specific uses of various perchlorate salts include: solid rocket fuel oxidizer, flares, and
pyrotechnics (potassium); solid rocket fuel oxidizer, explosives, chemical processes and
pyrotechnics (ammonium); precursor to potassium and ammonium perchlorate and in explosives
(sodium); and military batteries (magnesium) (Rogers, 1998). Perchlorate salts also are used on a
large scale as a component of air bag inflators. Other industrial or commercial applications of
perchlorate salts include their use in nuclear reactors and electronic tubes; as additives in
lubricating oils; in tanning and finishing leathers; as a mordant for fabrics and dyes;
in electroplating, aluminum refining, and rubber manufacture; and in the production of paints and
enamels (Siddiqui et al., 1998). A 1998 report raised the concern that chemical fertilizer is
a potential source of perchlorate contamination (TRC Environmental Corporation, 1998). More
recent studies limit concern to certain types of fertilizer containing Chilean caliche (Urbansky,
2000; U.S. Environmental Protection Agency, 2001 a,b; Urbansky and Collette, 2001); however,
production practices have been changed to address that issue. Besides their large-scale
commercial uses, perchlorate salts often are employed on a small scale in laboratory chemical
studies as ionic strength adjustors or as noncomplexing counterions. Some still are used in
medical diagnostics in thyroid function tests. Perchloric acid is used for various laboratory
applications requiring a strong acid. Wet ashing organic matter with perchloric acid still is
performed today as a means of preparation for certain samples. Anhydrous magnesium
perchlorate (Mg(C104)2) is a strong desiccant; however, historically, Anhydrone®, a slightly
hydrated form of Mg(C104)2, has been used to collect the water formed in combustion analyses.
The large-scale production of perchlorate-containing chemicals in the United States began
in the mid-1940s. The approximate percentages sold for specific end uses are 92% as an
oxidizer, 7% as an explosive, and 1% other uses (American Pacific Corporation, 1998). The
typical volume of production ranged from 1 to 15 million lb per year (Rogers, 1998) although
production in the 1980's was generally 20 to 30 million pounds per year (Kerr-McGee Chemical
LLC, 1998; American Pacific Corporation, 1998). Solid rocket fuel inventories are growing at a
January 16,2002 1 -2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
significant rate as systems reach the end of their service life and as treaties mandate motor
disposal. The current disposal method for these motors is open burning or open detonation, both
of which are becoming increasingly difficult to perform under intense public and regulatory
pressure based, in part, on concern over incomplete destruction via these methods. Currently, the
large solid rocket motor disposal inventory shows 55 million lb of propellant awaits disposal, and
this number is expected to be over 164 million lb by the year 2005 (Siddiqui et al., 1998).
A significant portion of this inventory contains ammonium perchlorate, which now can be
reclaimed and recycled into new motor propellants. The accepted method for removal and
recovery of solid rocket propellant from rocket motors is high-pressure water washout. This
method generates large amounts of aqueous solution containing low concentrations of
ammonium perchlorate. Although ammonium perchlorate can be recovered from these aqueous
solutions, it is cost-prohibitive to remove it entirely. Most of the locations where perchlorate has
been detected in ground or surface waters are in areas associated with development, testing, or
manufacture of aerospace materials. Explosives and fireworks manufacturing and disposal
operations have also been implicated in a number of environmental releases. Laboratory
activities and fertilizer operations are potential sources of contamination in relatively few known
instances. Perchlorate contamination also may occur where mining activities use explosives
extensively (Siddiqui et al., 1998).
When ammonium perchlorate is released to water, the salt is highly soluble and dissociates
completely releasing ammonium (NH4) and perchlorate (C104):
Its high solubility is not affected by pH or temperature. It is likely that most of the ammonium
has been biodegraded, and the cation in the environment is best viewed as mostly sodium (Na+)
or possibly hydrogen (H+), especially where contamination levels are below 100 ppb;
nevertheless, those regions with high concentrations of perchlorate ion probably retain a small
fraction of ammonium ion (Urbansky, 1998a). At those sites where contamination has occurred
for decades, very little (if any) ammonium ion has been found. To date, there has been no
quantitative determination of the cations responsible for the charge balance.
NH4C104(s) ^>° NH4+(aq) + C104~(aq).
(1-2)
January 16, 2002
1-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
As an oxidant, perchlorate is kinetically nonlabile. This means that reduction of the central
chlorine atom from an oxidation state of +7 (perchlorate) to - 1 (chloride ion) occurs extremely
slowly. This will be elaborated on in Chapter 2 in the discussion of physicochemical
characteristics. Sorption is not expected to attenuate perchlorate concentrations because it
absorbs weakly to most soil minerals. Natural chemical reduction in the environment is also not
expected to be significant. Together, these two factors account for perchlorate's high mobility
and persistence for many decades under typical groundwater and surface water conditions.
Figure 1-1 summarizes the various pathways through which perchlorate can reach ground and
surface water sources.
1.2 EVOLUTION OF ANALYTICAL DETECTION METHODS AND
EMERGING OCCURRENCE DATA
The Region 9 Office of the U.S. Environmental Protection Agency (EPA) first became
aware of the potential contamination issues with perchlorate in 1985 when samples measured
with a colorimetric method reported contamination in 14 wells ranging from 0.11 to 2.6 ppm
(Takata, 1985). The Region 9 office requested assistance from the Centers for Disease Control
and Prevention (CDC) to evaluate the potential health effects of these levels of perchlorate
(Takata, 1985). In response the CDC recommended validation of the colorimetric measures but
could not address the potential for toxicity of the chemical because of toxicity data
insufficiencies (Margolis, 1986). The CDC also recommended additional testing to determine
potential target tissues and the effects from long-term, low-level exposures. The absence of a
valid analytical method to measure low concentrations of perchlorate and the lack of data with
which to characterize the risk of toxicity led Region 9 of EPA to focus on chemicals other than
perchlorate at these sites. By the early 1990s, however, perchlorate at detectable levels
(>1 mg/L) was found in monitoring wells at a California Superfund site, and EPA Region 9
increased its effort to establish a human-health-based reference dose (RfD) in order to help gauge
the risk of the contamination that was beginning to be characterized. In 1997, after perchlorate
was discovered in a number of California water supplies, the California Department of Health
Services (CA DHS) adopted 18 ppb as its provisional action level.
January 16, 2002
1-4
DRAFT-DO NOT QUOTE OR CITE
-------
E3
C
P
CT\
K>
O
o
N»
U>
H
i
o
o
o
H
o
c!
o
Perchlorate for Rocket Building
Perchbrate for Explosive M ining
Perchlorate Manufacturing Sites
H
Perchbrate for Auto Airbags
Perchbrate for Chrome Plating
Perchbrate for Other Uses
Aqueous Waste Tailings
Contaminated Solid Waste
Atmospheric Deposition
Disposal in Ponds/Lagoons
Open-Pit Burning Disposal
Hazardous Waste Landfills
J
Hazardous Waste Landfill y
Disposal in Ponds, Lagoons, etc.
Dissociation Ammonium Perchlorate -^ Ammonium + Perchlorate
Bio-Reduction Perchlorate Chlorate ^ Chlonde
Chemical Conversion Chlorate Perchlorate
Perchlorate
i
Chlorate
Chlorite
=1=4=
y Chlonde
=4=
groundwater flow
perchlorate, chlorate, chlonte, chloride
perchlorate, chlorate, chlonte, chloride ^ groundwater flow
Figure 1-1. Sources and pathways of groundwater contamination for perchlorate. (Modified from Siddiqui et al., 1998.)
O
*
n
3
m
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
In January 1997, the California Department of Health Services' Division of Drinking Water
and Environmental Management requested the Sanitation and Radiation Laboratory Branch
(SRLB) test for perchlorate in drinking water wells potentially affected by groundwater migrating
from the Aerojet facility near Sacramento. Based on its provisional action level, Region 9 of
EPA indicated that a reporting limit of at least 4 ppb would be necessary. However, procedures
to measure perchlorate at such low levels were not available. An ion chromatographic (IC)
method was capable of detecting 400 ppb; and, during the previous year, Aerojet had improved
the method to detect 100 ppb. By March 1997, SRLB and an analytical equipment manufacturer
had developed an IC method that achieved a method detection limit of approximately 1 ppb and a
reporting limit of 4 ppb. Within several months following the March 1997 development of the
low-level (4 ppb) IC detection method, perchlorate was discovered at various manufacturing sites
and in well water and drinking water supplies in California, Nevada, and Utah.
Efforts in several additional laboratories helped improve the IC method (Eldridge et al.,
1999; Urbansky, 2000). Although IC is the dominant analytical method used at this time, a
variety of additional techniques are being refined for perchlorate analysis, including: mass
spectrometry, Raman spectrometry, capillary electrophoresis, and others (Urbansky, 2000).
Recent publications have reported detection of perchlorate in tap water at levels as low as 0.1 ppb
(Handy et al., 2000; Koester et al., 2000).
In March 1999, EPA included perchlorate in the Unregulated Contaminant Monitoring
Rule (UCMR) (Federal Register, 1999). Under the UCMR, all large public water systems and a
representative sample of small public water systems were required to monitor for perchlorate
beginning in January 2001. The EPA Method 314.0 for the analysis of perchlorate in drinking
water using IC methods was published in early 2000 as a direct final rule (Federal Register,
2000). The EPA Method 314.0 was also approved as a monitoring method for the UCMR
(Federal Register, 2000). However, this effort does not extend to investigating potential sources
in groundwater and surface water that have not migrated into public water supplies. Additional
information about the UCMR is available at the web site http://www.epa/gov/safewater/
ucmr.html.
The CA DHS adopted 18 ppb as its provisional action level in 1997 after perchlorate was
discovered in a number of California water supplies. The CA DHS also added perchlorate to the
list of unregulated chemicals for which monitoring is required in 1999 (Title 22, California Code
January 16, 2002
1-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of Regulations §64450). By September 2001, over 2,800 sources of public water supply had
been sampled in California by water supply agencies responding to CA DHS requirements. Most
of these sources represent water supply wells. Of the sources sampled, 206 (over 7 percent) had
perchlorate concentrations greater than 5 ppb in at least two samples (Figure 1-2). Most of these
wells have as their source groundwater plumes that have spread as far as nine miles from the site
of original release.
At this time, there has not been a systematic national survey of perchlorate occurrence.
Several states and EPA regions are taking significant steps to test water supplies for perchlorate.
notably the states of Arizona, Utah, and Texas, EPA Regions 6 and 7, and Suffolk County,
New York. Figure 1-3 indicates states with confirmed perchlorate manufacturers or users, and
Figure 1-4 indicates those states in which facilities have, in response to reported releases, directly
measured perchlorate in groundwater or surface water. Table 1-1 describes these locations. The
data published in Siddiqui et al., 1998 (drinking water systems in New Mexico, Indiana,
Pennsylvania, and Iowa) are displayed in Figure 1-3 and in Table 1-1, but they have not been
independently confirmed.
Information on other potential sites across the country is being gathered from the
Department of Defense (DoD) and National Aeronautics and Space Administration (NASA)
searches and from EPA information requests made to perchlorate manufacturers. The EPA has
notified state, tribal, and local governments when it has evidence of perchlorate manufacture and
use in these governmental jurisdictions. The American Water Works Association Research
Foundation is coordinating a survey to characterize possible perchlorate contamination of
drinking water sources in areas of high risk. The EPA will build on these survey data and other
information to discover potential sources and evaluate threats to water resources.
Region 9 officials have collected information concerning detected perchlorate releases in
20 different states (Table 1-1). For two of these states, Pennsylvania and Indiana, the only
reported releases have not been confirmed by a state or federal agency and should be considered
questionable until the detections can be independently validated. In Washington State, propellant
was observed scattered around open burn/open detonation sites although results of solid rocket
chemical analyses of groundwater samples are not yet available. In California, most areas where
perchlorate has been detected are associated with facilities that have manufactured, tested, or
disposed of solid rocket fuels and propellants for DoD or NASA. Two facilities that
January 16, 2002
1-7
DRAFT-DO NOT QUOTE OR CITE
-------
California Overview Map
0Ru
• River
IP*
Russian Sacramento\
(See Inset) \
\
California Perchlorate
Detections in
Drinking Water Sources
O
Santa
Clara
\N
\
s
o
Sacramento Inset Map
A
O
o©
oo
° O
O
100 Miles
O
6 Miles
Source of well localions and
perchlorate values:
California Department
of Health Services
O 4-9 ug/L
O 9 - 17-9 Hg/L
O 18 - 31-9 Mg/L
O 32 - too ng/L
© > 100 ng/L
O
O
o
Los Angeles Area Inset Map
h' « -"i
/.®p -
San Fernando Valley ^ Pasadena
coco^, XX San Ber
San Gabriel Valley ^ 0 OO
CO " ^
San Bernardino
Los Angeles
o
o o u
,0 O
° <*> %
of* °
CP Orange
\o° County
Riverside
o 5 10
%
o
CO
Ftogion 9 GIS Cantvr
December 5, aooi
Figure 1-2. Distribution of perchlorate detected in public water supply sources in
California. Also noted are several large sites of groundwater contamination
that include perchlorate releases (Mayer, 2001).
January 16, 2002
1-8
DRAFT-DO NOT QUOTE OR CITE
-------
p
3
C
S»
ON
K>
O
o
K>
O
>
|-d
H
a
o
z
o
H
O
c
o
H
tfl
O Figure 1-3. Locations of specific perchlorate manufacturers or users identified through responses to EPA Information
n Requests from current manufacturers (identifying shipments of at least 500 pounds in any year) and
H through investigations by state and local authorities (Mayer, 2001).
m
• Perchlorate Manufacturers and Users
Major Rivers
State does not contain a known manufacturer or user
State contains a known manufacturer or user
-------
p
3
5=
ON
K)
O
o
K)
O
ja
>
TCl
•H
6
o
z
o
H
Figure 1-4. Locations of reported environmental releases of perchlorate to groundwater, surface water, or soil.
O Perchlorate measured in four water supplies in New Mexico, Iowa, Indiana, and Pennsylvania has been
w published in Siddiqui et al., 1998, but has not been confirmed independently by EPA or state authorities.
^ Monitoring for perchlorate releases in most states is very limited or nonexistent (Mayer, 2001).
O
H
m
o Perchlorate Releases (Unconfirmed)
Major Rivers
i State with no reported perchlorate release
j State with a reported perchlorate release
-------
TABLE 1-1. OCCURRENCE AND POTENTIAL SOURCES OF PERCHLORATE
RELEASES TO THE ENVIRONMENT AS OF NOVEMBER, 2001" (Mayer, 2001)
State
Location
Suspected Source
Type of Contamination
Max.
Cone, ppb
AL
Redstone Army Arsenal -
NASA Marshall Space Flight
Center
Huntsville, AL
Propellant Manufacturing,
Testing, Research,
Disposal
Monitoring Well
Springs/Seeps
19,000
37
AZ
Apache Nitrogen Products
Benson, AZ
Explosives Manufacturing
Monitoring Well
670
AZ
Aerodyne
Gila River Ind. Res.,
Chandler, AZ
Propellant Testing
Monitoring Well
18
AZ
Davis Monthan AFB
Tucson, AZ
Explosives/Propellant
Disposal
Soil
Not
confirmed
AZ
Unidynamics Phoenix Inc.
Phoenix Goodyear Airport,
Goodyear, AZ
Explosives/Ordnance
Manufacturing
Monitoring Well
80
AZ
Universal Propulsion
Phoenix, AZ
Rocket Manufacturing
Soil
—
AZ
Unidynamics Phoenix Inc.
Whiter Tanks Disposal Area
Maricopa County, AZ
Explosi ves/Ordnance
Disposal
Public Water Supply Well
(Unconfirmed Report)
Soil
(4)
AR
Atlantic Research
East Camden, AR
Rocket Manufacturing
Disposal - Open
Burn/Open Detonation
Monitoring Well
Surface Water
Soil
1,500
480,000
CA
Aerojet General also affects
Mather AFB
Rancho Cordova, CA
Rocket Manufacturing
Public Water Supply Well
Monitoring Well
260
640,000
CA
Alpha Explosives
Lincoln, CA
Explosives Manufacturing
Monitoring Well
Reported in Surface Water
67,000
CA
Boeing/Rocketdyne, NASA at
Santa Susana Field Lab
U.S. DOE
Santa Susana, CA
Rocket Research, Testing
and Production
Monitoring Well
750
CA
Edwards AFB
Jet Propulsion Lab, North Base
Edwards, CA
Rocket Research
Monitoring Well
300
CA
El Toro Marine Corps Air
Station
Orange Co., CA
Explosives Disposal
Monitoring Well
380
CA
Lawrence Livermore National
Laboratory Site 300
Tracy, CA
U.S. DOE
Explosives Research
Monitoring Well
84
January 16, 2002
1-11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1 (cont'd). OCCURRENCE AND POTENTIAL SOURCES OF
PERCHLORATE RELEASES TO THE ENVIRONMENT AS OF NOVEMBER, 2001"
(Mayer, 2001)
State
Location
Suspected Source
Type of Contamination
Max.
Cone, ppb
CA
Lockheed Propulsion Upper
Santa Ana Valley
Redlands, CA
Rocket Manufacturing
Public Water Supply Well
87
CA
NASA - Jet Propulsion Lab
Raymond Basin
Pasadena, CA
Rocket Research
Public Water Supply Well
54
CA
Rialto, CA
Fireworks Facility (?)
B.F. Goodrich (?)
Rocket Research and
Manufacturing
Public Water Supply Well
(inactive)
811
CA
San Fernando Valley
Glendale, CA
Grand Central Rocket (?)
Rocket Manufacturing
Monitoring Well
84
CA
San Gabriel Valley
Baldwin Park, CA
Aerojet Rocket
Manufacturing
Public Water Supply Well
Monitoring Well
159
2,180
CA
San Nicholas Island
Ventura Co., CA
U.S. Navy Firing Range
Public Water Supply
(Springs)
12
CA
Stringfellow Superfiind Site
Glen Avon, CA
Hazardous Waste
Disposal Facility
Monitoring Well
Private Well
682,000
37
CA
UTC (United Technologies)
San Jose, CA
Rocket Testing
Monitoring Well
180,000
CA
Whittaker-Bermite Ordnance
Santa Clanta, CA
Ordnance Manufacturing
Public Water Supply Well
47
CA
Whittaker Ordnance
Hollister, CA
Ordnance Manufacturing
Private Well
Monitoring Well
810
88
IN
American Waterworks Service
Greenwood, IN
Unknown Source
Public Water Supply Well
(Unconfirmed Report)
(4)
IA
American Water Works Service
Clinton, IA
Unknown Source
Public Water Supply Well
(Unconfirmed Report)
(6)
IA
Ewart, IA
Unknown Source
Livestock Well
29
IA
Hills, IA
Unknown Source
Private Well
30
IA
Napier, IA
Agriculture (?)
Private Well
10
KS
Herington, KS
Ammunition Facility
Monitoring Well
9
MA
Massachusetts Military Res.
Barnstable Co., MA
Disposal - Open Bum/
Open Detonation
Monitoring Well
300
MD
Naval Surface Warfare Center
Indian Head, MD
Propellant Handling
Waste Discharge to
Surface Water
>1,000
January 16, 2002
1-12
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1 (cont'd). OCCURRENCE AND POTENTIAL SOURCES OF
PERCHLORATE RELEASES TO THE ENVIRONMENT AS OF NOVEMBER, 2001"
(Mayer, 2001)
State
Location
Suspected Source
Type of Contamination
Max.
Cone, ppb
MD
White Oak Fed. Research
Center
(Naval Surface Warfare Center)
White Oak, MD
Propellant Handling
Monitoring Well
72
MO
ICI Explosives
Joplin, MO
Explosives Facility
Monitoring Well
107,000
MO
Lake City Army Amm. Plant
Independence, MO
Propellant Handling
Monitoring Well
70
NE
Lewiston, NE
Agricultural Chemical
Facility
Shallow Private Well
5
NE
Mead, NE
Fireworks Facility
Monitoring Well
24
NV
Kerr-McGee/BMI
Henderson, NV
Chemical Manufacturing
Public Water Supply
Monitoring Well
Surface Water
24
3,700,000
120,000
NV
PEPCON
Henderson, NV
Chemical Manufacturing
Monitoring Well
600,000
NM
American Water Works Service
Clovis, NM
Unknown
Public Water Supply Well
(Unconfirmed Report)
(4)
NM
Ft. Wingate Depot Activity
Gallup, NM
Explosives Disposal
Monitoring Well
2,860
NM
Holloman AFB
Alamogordo, NM
Rocket Testing
Monitoring Well
Seasonal Surface Water
Soil
40
16,000
NM
Los Alamos National Lab
Los Alamos, NM
U.S. Dept. of Energy Lab
Chemicals
Public Water Supply Well
Monitoring Well
Deep Borehold Water
3
220
1,662
NM
Melrose Air Force Range
Melrose, NM
Explosives
Public Water Supply Well
25
NM
White Sands Missile Range
White Sands, NM
Rocket Testing
Monitoring Well
Soil
21,000
NY
West Hampton
Suffolk County, NY
Unknown Source(s)
Public Water Supply Well
Monitoring Well
16
3,370
NY
Yaphank
Suffolk County, NY
Fireworks
Private Well
Monitoring Well
26
122
PA
American Water Works Service
Yardley, PA
Unknown
Public Water Supply Well
(Unconfirmed Report)
(5)
January 16, 2002
1-13
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 1-1 (cont'd). OCCURRENCE AND POTENTIAL SOURCES OF
PERCHLORATE RELEASES TO THE ENVIRONMENT AS OF NOVEMBER, 2001"
(Mayer, 2001)
State
Location
Suspected Source
Type of Contamination
Max. Cone,
ppb
TX
Longhorn Army Ammunition
Depot
Karnak, TX
Propellant Handling
Monitoring Well
Reported in Surface Water
Soil
169,000
TX
McGregor Naval Weapons Plant
McGregor, TX
Propellant Handling
Monitoring Well
Reported in Surface Water
Soil
91,000
TX
PANTEX Plant (USDOE)
Amarillo, TX
Explosives
Monitoring Well
5
TX
Red River Army Depot
Texarkana, TX
Propellant Handling
Monitoring Well
80
UT
Alliant Tech Systems
Magna, UT
Rocket Manufacturing
Public Water Supply Well
16
UT
Thiokol
Promontory, UT
Rocket Manufacturing
Well Supply Well
(Inactive)
42
WA
Camp Bonneville
near Vancouver, WA
Explosives/Propellant
Disposal
Soil
—
WV
Allegheny Ballistics Lab
Rocket Center, WV
Rocket Research,
Production, Open
Burn/Open Detonation
Surface Discharge of
Groundwater Extraction
400
"Data reported to EPA Region 9 as of November 2001. All reports have been confirmed by federal, state, or
county agencies except where noted. Soil concentrations are not listed.
1 manufactured ammonium perchlorate in Nevada were found to have released perchlorate to
2 groundwater resulting in low levels (4 to 24 ppb) in Lake Mead and the Colorado River. This
3 water is used for drinking, irrigation, and recreation for millions of people in Nevada, California,
4 Arizona, and by Native American tribes.
5 The concentrations reported in wells and surface water vary widely. At one facility near
6 Henderson, NV, perchlorate in groundwater monitoring wells was measured as high as 0.37%
7 (3.7 million ppb). The highest level of perchlorate reported in any public water supply well was
8 800 ppb in an inactive well in California. Few active public water supply wells have perchlorate
9 greater than 100 ppb, and none are reported at this level outside of California.
January 16, 2002
1-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Perchlorate was found in a number of water supply wells on Long Island, NY, including
several downgradient from a fireworks facility. It has been speculated that the wide distribution
pattern of the New York contamination could be a result of low levels of perchlorate contained in
fertilizer imported from Chile (TRC Environmental Corporation, 1998; Urbansky, 2000; Suffolk
County Department of Health Services, 2001 a,b). Agricultural chemicals also have been
implicated as a potential source of perchlorate contamination in Nebraska at a shallow well near
a speciality fertilizer facility (Williams, 2000). After state and federal officials in Region 7
added perchlorate analyses in their program testing hundreds of rural wells for fertilizers and
agricultural chemicals. Their results showed that fertilizer application to farmlands is an unlikely
source of perchlorate in Midwestern states.
In addition to discoveries at facilities involved with rocket propellants, explosives, and
fireworks, a number of perchlorate detections have been made at current or former military
facilities where propellants and explosives were disposed of by detonation and burning.
Cooperation from Department of Defense (DoD) and Department of Energy (DoE) officials will
continue to be important for examining these types of potential sources.
In the past three years, the increasing interest in investigating the environment has resulted
in increasing detections. It is likely that regional positive efforts at detection may largely explain
the distribution of known areas of release to the environment (Figure 1-4) when compared to the
potential distribution suggested in Figure 1-3. As the efforts for detection become more uniform
nationwide, the occurrence of perchlorate in the environment may more closely resemble the
pattern of perchlorate usage.
It is important to distinguish between minimum detection limit (MDL) and the minimum
reporting limit (MRL), which is also called the practical quantitation limit (PQL). MDLs are
calculated from the precision of replicate low level measurements and are assumed to reflect
99% confidence that a trace concentration above zero can be detected. MRLs are higher values
that reflect actual quantifiable concentrations. The EPA calculated and published an MDL for
Method 314 (Ion Chromatography) at 0.53 ng/L (Federal Register, 2000). This was derived
through the analysis of 7 replicate samples fortified at 2.0 yug/L. Based upon this result, an MRL
for perchlorate was established at 4.0 ///L. Dionex, the manufacturer of the ion chromatography
column, published an MDL of 0.2 //g/L and MRL of 2.0 //g/L (Dionex, 2000).
January 16, 2002
1-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Method 314 does not represent the lowest possible MRL or MDL. Unpublished
improvements in the ion chromatography method may lower the MRL to the sub-part per billion
level (Yates, 2001). Several research and commercial laboratories have been developing mass
spectrometry methods to detect sub-ppb levels of perchlorate (Urbansky et al., 1999; Magnuson
et al., 2000 a,b; Urbansky, 2000; Handy et al., 2000; Koester et al., 2000; Winkler, 2001). It is
reasonable to expect that a reliable sub-ppb MRL for perchlorate will be commercially available
in the very near future. The Agency encourages development of these emerging methods (e.g.,
LC/MS/MS) to eliminate interferences that can be encountered by extending IC methods for
low-level analysis in a variety of matrices (e.g., soil or plants and animal tissues). The market
demand for this capability may determine the commercial availability and expense of this
method. Regulatory pressure to ensure protection or water supplies and to maintain treatment
process control is also a factor driving the development of lower reporting limits for perchlorate.
Thorough method validation and quality assurance information must be complied to establish a
standard analytical method in the sub-ppb range for various media.
1.3 HEALTH AND ECOTOXICOLOGY RISK ASSESSMENTS-
HISTORICAL OVERVIEW
This section briefly summarizes how the assessments for the health and ecotoxicology risks
of perchlorate contamination have evolved. This document represents the revised assessment
that incorporates additional data and analyses recommended at the external peer review convened
by the Agency in February, 1999 (Research Triangle Institute, 1999).
1.3.1 Overview of Perchlorate Health Risk Assessment
The EPA Region 9 office requested evaluation of the toxicology data from the EPA
Superfund Technical Support Center (Stralka, 1992). The EPA Superfund Technical Support
Center issued a provisional RfD in 1992 (Dollarhide, 1992) and a revised provisional RfD in
1995 (Dollarhide, 1995) based on a literature review (Environmental Resources Management,
Inc., 1995) submitted by the Perchlorate Study Group (PSG). Ideally, an RfD is based on a
database that evaluates an array of endpoints that address potential toxicity during various critical
life stages, from developing fetus through adult and reproductive stages. The provisional RfD
January 16, 2002
1-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
values (1992 and 1995) were based on an acute study in which single doses of potassium
perchlorate caused the release of iodide (I") from the thyroids of patients with Graves' disease, an
autoimmune condition that results in hyperthyroidism. It was difficult to establish a
dose-response for the effects on thyroid function from daily or repeated exposures in normal
humans from the data on patients with Graves' disease because of a variety of confounding
factors, including that the disease itself has effects; that often only a single exposure, rather than
repeated exposures was tested; that only one or two doses were employed; and that often the only
effect monitored was iodide release from the thyroid or control of the hyperthyroid state.
Nevertheless, a no-observed-adverse-effect-level (NOAEL) was determined to be
0.14 mg/kg-day based on release of iodide in the thyroid, followed by incomplete inhibition of
iodide uptake. Uncertainty factors that ranged from 300 to 1,000 were applied to account for
data missing on additional endpoints and extrapolations required to calculate a lifetime human
exposure level. The provisional RfD values issued are listed as such by EPA because they did
not undergo the internal EPA and external peer review required of estimates available on the
EPA's Integrated Risk Information System (IRIS). Standard assumptions for ingestion rate and
body weight were applied to the RfD to calculate the reported range in the groundwater cleanup
guidance levels of 4 to 18 ppb.
In recognition of the potential influence of the reduced analytical detection limit, a
reevaluation of the provisional 1992 and 1995 RfDs that serve as the basis of the provisional
action level was warranted. An external non-EPA peer review convened in March 1997 to assess
an analogous RfD derivation by an independent organization (Toxicology Excellence for Risk
Assessment, 1997) determined that the health effects and toxicity data were insufficient for a
credible quantitative risk analysis (Toxicology Excellence for Risk Assessment, 1998a). The
external peer review panel concluded that the limited database was insufficient to rule out effects
of perchlorate on other organs, so it could not be determined unequivocally that the effect on the
thyroid was the critical effect. In particular, the reviewers were concerned that developmental
toxicity, notably neurological development affected by hypothyroidism during pregnancy, could
be another critical effect of perchlorate that had not been examined adequately in studies to date.
In response to the March 1997 external peer review of the provisional RfD value, a subsequent
external peer review of experts was convened in May 1997 to recommend and prioritize a set of
studies to address the key data gaps and to reduce uncertainties in various extrapolations
January 16, 2002
1-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
(Toxicology Excellence for Risk Assessment, 1998b). The objective of the new studies is to
provide a comprehensive database that will support development of a robust RfD estimate that
reduces the uncertainties inherent in the provisional values. The strategical basis of the new
battery of toxicity studies is discussed in Chapter 3. These data featured prominently in the
external peer review draft of the assessment issued by the EPA in December 1998. At the
subsequent external peer review convened by the Agency in February 1999, recommendations
were made for additional studies and analyses, including completion of those on studies that were
only available as preliminary data at that time (Research Triangle Institute, 1999). The EPA
committed to a second external peer review and a revised risk assessment in order to benefit from
the additional insights that these data might bring to bear. The purpose of this current revised
document is to incorporate all of the data from new studies and to respond to recommendations
made at the previous external peer review.
Because the Agency is committed to utilizing the latest available science to support its
human and ecotoxicological risk estimates, the Office of Research and Development (ORD)
issued interim guidance in 1999 to its risk assessors and risk managers to be followed until this
revised assessment became publicly available (Noonan, 1999). The recommendation was to
continue using the standing provisional RfD range of 0.0001 to 0.0005 mg/kg-day for
perchlorate-related risk assessment activities because of the significant concerns and
uncertainties that remained to be addressed. This recommendation was based on the
determination that important new analyses on emerging data would likely have an impact on the
previously proposed health risk benchmark in the 1998 external review draft (U.S.
Environmental Protection Agency, 1998d) and that, while the new estimates would reflect greater
accuracy, the resultant revised risk estimate could be either higher or lower.
This recommendation helped to ensure that the Agency bases its risk management
decisions on the best available peer reviewed science and was in keeping with the full and open
participatory process embodied by the proposed series of external peer review workshops.
It should be noted that, due to the uncertainty of whether the final proposed revised oral human
health risk benchmark would increase or decrease based on the new data and analyses, the
standing provisional RfD range was the more conservative of the estimates available at the time
of the interim guidance and, therefore, more likely to be protective of public health. The
recommendation was also consistent with Agency practice that existing toxicity estimates remain
January 16, 2002
1-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
in effect until the review process to revise them is completed. The steps necessary to complete
this assessment are outlined in Section 1.4. Once completed, this assessment will be included on
the Agency's Integrated Risk Information System (IRIS).
1.3.2 Overview of Ecotoxicology Screening Level Assessment
The mobility and persistence of perchlorate discussed in the beginning of this chapter also
may pose a threat to ecological receptors and whole ecosystems either by direct harm to
organisms or by indirectly affecting their ability to survive and reproduce. There were very
limited data in 1997 with which to evaluate the effects of perchlorate on ecological systems; nor
were there data about the possible uptake of perchlorate into agricultural products irrigated by
contaminated water. Analytical tests had been derived to detect perchlorate in water, but little
work had been done to extend these methods to testing plant and animal tissues or food crops for
perchlorate.
Searches of available databases revealed minimal information on the ecological effects of
ammonium perchlorate or any of perchlorate's other salts. Little data exist to describe
perchlorate's ecological effects on various soil, sediment, or aquatic receptors, including aquatic
vertebrates, aquatic or sediment invertebrates, and bacteria or plants. The data that were
available suggested effects on thyroid-hormone-mediated development in the South African
clawed frog, Xenopus laevis, in the range of 50 to 100 ppm, and 1,000 ppm had been shown to
completely block the metamorphosis of tadpoles. Effects on development and population growth
also had been indicated in the freshwater lamprey at 100 ppm and the freshwater hydra at
350 ppm. Mortality was observed in cold-water trout (6,000 to 7,000 ppm) and Daphnia magna
(670 ppm). Effects on seed germination and growth of agricultural plants were reported at
10 ppm.
Under the auspices of the Ecological/Transport and Transformation Subcommittee of the
Interagency Perchlorate Steering Committee (IPSC, see Section 1.5), the U.S. Air Force (USAF)
Detachment 1, Human Systems Center, Brooks Air Force Base (AFB), in conjunction with EPA,
developed a proposal for a battery of screening-level bioassays in laboratory-reared organisms
representative of soil, sediment, and water column receptors, to evaluate dose-response
relationships. The identified tests focus on identifying gross (direct) toxicity tests whose
endpoints can include mortality, growth, and reproductive success. Bioassays with standard
January 16,2002 1-19 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
protocols and general regulatory acceptance were chosen. Although these were screening-level
tests and provided only an indication of gross toxicity, they provided the dose-response
information required to make decisions about the need for a next tier of tests to be completed
(e.g., bioavailability, bioaccumulation, histopathology).
Additional new studies were recommended at the 1999 external peer review in the
ecotoxicology arena as well, and some additional data has become available that improves the
information base somewhat. Most significantly, additional data are available on effect levels in
aquatic animals, an aquatic plant, a terrestrial plant, and a soil invertebrate; and some of these
data are for chronic durations. In addition, surveys have been conducted at several sites of
known or suspected perchlorate contamination with analysis of environmental and biological
materials for perchlorate levels. While these new data have been incorporated in the current
revision and are described in Chapter 8, the knowledge in this arena requires that the ecological
assessment must still be characterized as a screening level rather than definitive. The number of
species is still quite low and the site surveys aimed only to describe the range of exposures at the
sites. The ecotoxicological review will undergo the same peer review process as the health risk
assessment that is described in Section 1.4.
1.4 RISK CHARACTERIZATION AND REGULATORY AGENDA
This section briefly describes pending regulatory activities that this evaluation and
characterization will likely influence. Particular emphasis is placed on the revised health risk
assessment and ecotoxicology assessments.
1.4.1 U.S. Environmental Protection Agency Regulatory Plans
The Safe Drinking Water Act (SDWA), enacted by Congress in 1974 and amended in 1986
and again in 1996 (U.S. Code, 1996), provides the basis for safeguarding public drinking water
systems from contaminants that pose a threat to public health. The purpose of the SDWA is to
protect public health by ensuring that public drinking water systems provide tap water that is safe
for drinking and bathing. Within EPA, the Office of Ground Water and Drinking Water
January 16, 2002
1-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
develops National Primary Drinking Water Regulations (NPDWR) to control the levels of
contaminants that may occur in public drinking water systems.
The 1996 amendments to the SDWA require EPA to publish a list of contaminants that are
not currently subject to a NPDWR and are known or anticipated to occur in public water systems.
This list, known as the Contaminant Candidate List (CCL), is the source of priority contaminants
for research, guidance development, regulatory determinations, and monitoring by the states.
The SDWA requires EPA to determine whether or not to regulate at least five contaminants from
the CCL by 2001. The CCL also must be reviewed and updated every 5 years; the next review is
scheduled for 2003.
With broad public input and consultation with the scientific community, a draft CCL was
published on October 6, 1997. The draft CCL specifically requested comment on whether to
include perchlorate on the CCL based on the limited information EPA had received on
perchlorate's occurrence in drinking water supplies at the time of publication. As a result of the
public comments and the obtainment of additional occurrence information, EPA determined that
sufficient information exists to raise concern over perchlorate's potential public health impact
and added perchlorate to the final CCL published on March 2, 1998.
The CCL consists of 50 chemical and 10 microbiological contaminants and is divided into
two categories: (1) contaminants for which sufficient information exists to begin to make
regulatory determinations in 2001, and (2) contaminants for which additional research and
occurrence information is necessary before regulatory determinations can be made. Perchlorate
falls into the latter category because of the need for additional research in the areas of health
effects, treatment technologies, analytical methods, and extent of occurrence.
1.4.2 State Regulatory Plans
The CA DHS and the California EPA Office of Environmental Health Hazard Assessment
(CA EPA OEHHA) reviewed the EPA risk assessment reports for perchlorate and established its
action level at 18 ppb based on the provisional RfD values from the EPA Superfund Technical
Support Center. The CA DHS advises water utilities to remove drinking water supplies from
service if they exceed the 18-ppb action level. If the contaminated source is not removed from
service because of system demands, and if drinking water provided by the utility exceeds the
action level, the CA DHS advises the utility to notify its customers. On August 1, 1997, the CA
January 16, 2002 1 -21 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
DHS informed drinking water utilities of its intention to develop a regulation requiring
monitoring of perchlorate as an unregulated chemical. Legislative action to establish a state
drinking water standard for perchlorate by January 2000 (California Senate Bill 1033 [California
State Senate, 1998]) was vetoed by the governor after passage by both houses. The governor
supported prioritizing the regulation of perchlorate in drinking water but objected to the strict
time schedule required.
In July 2001, the CA EPA OEHHA posted a notice on its web site indicating that it was
initiating a risk assessment for perchlorate in connection with the development of a public health
goal (PHG) for a number of chemicals in drinking water (www.oehha.ca.gov/public_info/public/
phgannounc.html). PHGs are concentrations of chemicals in drinking water that are not
anticipated to produce adverse health effects following long-term exposures. These goals are
non-regulatory in nature but are to be used as the health basis with which to update the state
primary drinking water standards established by CA DHS for chemicals in drinking water subject
to regulation. A 45-day public comment period will be provided after posting, followed by a
public workshop. Scientific peer reviews are arranged through the University of California. The
overall process will include time for revisions, further public comment, and responses to
comments. The new PHGs are scheduled for publication in 2003.
New York, Arizona, and Texas also initially adopted the level of 18 ppb as their version of
advisory levels for water supply systems. Texas and Arizona health departments revised their
perchlorate advisory levels based on research presented in EPA's December 1998 External
Review Draft Toxicity Assessment. In July 1999, Texas arrived at a value of 22 ppb in drinking
water by calculating the exposure of a 15 kg child drinking 0.64 liter per day and using the
reference dose proposed in the 1998 EPA ERD document. Texas revised this value to 4 ppb in
October 2001 based in part on the interim ORD guidance (Noonan, 1999). Arizona derived a
14 ppb level in March 2000, based on a 15 kg child drinking 1 liter per day and using the
proposed RfD in the 1998 EPA ERD document. New York State has continued to use 18 ppb as
the advisory level for perchlorate in drinking water.
The Nevada Division of Environmental Protection (NDEP) has authority under Nevada
Water Pollution Control Regulations to address pollutants in soil or groundwater. The state's
Corrective Action Regulations direct NDEP to establish action levels for hazardous substances,
pollutants, or contaminants, using drinking water standards such as a maximum contaminant
January 16, 2002
1-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
level (MCL), health advisories, or background or protective levels (determined by IRIS or the
equivalent). In August 1997, Nevada determined that the action level of 18 ppb, as established
by EPA, would be the recommended action level for cleanup, pending a more current risk
assessment.
1.5 SUMMARY
Perchlorate contamination is a concern for several reasons. First, there are uncertainties in
the toxicological database that is used to address the potential of perchlorate to produce human
health effects when present at low levels in drinking water. Additionally, the actual extent of
perchlorate occurrence in ground and surface waters and other media (soils or plant and animal
tissues) is unknown—a problem compounded by limits to the analytical detection method. The
efficacy of different treatment technologies for various water uses, including drinking and
irrigation, is also not well established. Finally, the nature and extent of ecological effects and
details about transport and transformation phenomenon in various environmental media have
been studied only superficially. EPA aims to more comprehensively characterize the risks to
human and ecological health from perchlorate contamination through the integrative approach
presented in Figure 1-5.
Thus, a number of key pieces of information and scientific advances are essential to
adequately characterize the risks of perchlorate contamination and to develop scientifically-based
management strategies that effectively mitigate the potential risks posed by perchlorate
contamination. Accurate characterization of exposures relies on reliable analytical detection
methods. The exposure estimates cannot be gauged with respect to their risk unless a robust
health risk estimate is available. Treatment technologies should be targeted to levels of concern
and tailored to the intended water use. Technology transfer is necessary so that all affected
parties and concerned citizens are appraised of accurate and reliable information that is
up-to-date with the evolving state-of-the-science. The purpose of the revised risk
characterizations presented in this document is to serve in this integrative approach by providing
more robust risk estimates than those that currently exist provisionally in order to better gauge
the potential human health and ecological impacts.
January 16, 2002
1-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Personal Uses
Ingestion
Dermal
Inhalation
- Fruits
-Animals
- Soil
Route of
Exposure
Figure 1-5. Considerations for comprehensive characterization of perchlorate
contamination. (Modified from Underwood, 1998.)
The National Center for Environmental Assessment (NCEA) in the Office of Research and
Development (ORD) of EPA first evaluated the emerging information and new human
health/toxicity and ecotoxicity data from the testing strategy (see Chapter 3) and issued an
external peer review draft in December 1998. In February 1999, an external peer review
workshop was convened. The peer review panel endorsed the conceptual approach proposed by
NCEA and recommended additional studies and analyses. This revised risk characterization
document represents a response to those recommendations and includes data made available to
the EPA as of Fall 2001.
As with any risk assessment, incorporation of new data is an iterative process. Because of
regulatory schedule constraints, this assessment has gone forward with the recognition that new
data may warrant further revision at a future date. Data in additional analyses that are warranted
and which will be arriving in the period between the issuance of the external peer review draft
and the external peer review workshop are identified herein and may be presented at that
workshop.
January 16, 2002
1-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Independent, external peer review of the study protocols, toxicity studies, and revised risk
assessment for perchlorate will be critical to ensuring that future decisions will be protective of
human health and that the potential for ecotoxicology is characterized appropriately. The IRIS
program will oversee the external peer review and has tasked a qualified contractor to manage
the peer review of technical issues related to the development of the human health and
ecotoxicology assessments, including system design, conduct of toxicity studies, statistical
analysis of data, designation of effect levels, selection of critical effect and uncertainty factors,
and risk characterization. The peer review will be conducted by a panel of technical experts
selected by contractors in ecotoxicology; neurotoxicology; developmental, reproductive, genetic,
and general toxicology; endocrinology; pathology; biostatistics; dose-response modeling; and
risk assessment.
The risk characterization assessment package, supporting studies, and study protocols for
the new data will be distributed to the peer review panel in advance of the peer review meeting.
The peer reviewers are charged with evaluating experimental protocols, performance, and results
for any new studies since 1999 in addition to how the data are used in this risk assessment. Peer
reviewers will independently review the risk assessment package and supporting studies and will
submit their written comments to the IRIS contractor prior to the peer review meeting. The IRIS
contractor will compile the peer reviewers comments and distribute them to all of the reviewers
prior to the meeting which will be held on March 5 and 6, 2002. Sacramento was selected as a
location for its accessibility to stakeholders and peer reviewers. The public will be invited to
attend and observe the peer review meeting and may obtain pre-meeting comments at that time.
Following the peer review meeting, the peer review panel will generate a report detailing their
comments on the reference dose package and supporting studies. NCEA then will generate a
responsiveness summary report that will discuss how comments made by the peer reviewers have
been addressed. The revised risk characterization will be issued subsequently by EPA as a final
IRIS assessment after Agency consensus review across offices and laboratories and a final IRIS
program clearance.
It should be noted that this assessment effort was accomplished in an expedited time frame
through the partnership and cooperation of a number of governmental entities. The IPSC was
formed in January 1998 to bring together government representatives from EPA; DoD; the
National Institute for Environmental Health Sciences (NIEHS); and affected state, tribal, and
January 16, 2002
1-25
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
local governments. Participation in IPSC also has been solicited from other governmental
entities. The purpose of the IPSC is to facilitate and coordinate accurate accounts of related
technological issues (occurrence, health effects, treatability, waste stream handling, analytical
detection, and ecological impacts) and to create information-transfer links for interagency and
intergovernmental activities regarding these areas of concern.
Figure 1 -6 shows the structure of the IPSC, members of its executive committee, and
co-chairs of the subcommittees. Note that a subcommittee exists for each of the outstanding
controversial issues regarding perchlorate contamination. These are identified in the
comprehensive characterization framework presented in Figure 1-5. Research to obtain
additional data and the development of new methods and applications is underway in these
human health and ecotoxicology areas, as well as in most of the others, to ensure that the state-
of-the-science is brought to bear in addressing the unique issues of perchlorate contamination.
The IPSC collaborated in 1998 with EPA ORD NCEA on a draft report to a Congressional
committee that assesses the state-of-the-science on the health effects of perchlorate on humans
and the environment and the extent of perchlorate contamination. The report also contained
recommendations for future research to address emerging issues (U.S. Environmental Protection
Agency, 1998e). This report will be finalized and sent to Congress after the IRIS file is
completed. Updates on activities of IPSC can be found on the EPA Office of Water (OW) web
site at the following address: http://www.epa.gov./ogwdw/ccl/perchlor/perchlo.html. Discussion
papers presented by the IPSC present additional information on the areas (e.g., analytical and
treatment technology) that have not been discussed in detail herein.
January 16, 2002
1-26
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 1-6. Structure and membership of the executive committee, subcommittees areas,
and co-chairs of IPSC. The IPSC is designed to ensure an integrated approach
to addressing the perchlorate contamination challenge and to provide accurate
accounts of technical issues to stakeholders. (OSWER = Office of Solid Waste
and Emergency Response, NCEA = National Center for Environmental
Assessment, DoD = Department of Defense, USAF = U.S. Air Force, OW =
Office of Water, NERL = National Exposure Research Laboratory, OERR =
Office of Emergency Response and Remediation, NRMRL = National Risk
Management Research Laboratory, Cal DHS = California Department of
Health Services, USN = U.S. Navy, UT DEQ = Utah Department of
Environmental Quality).
January 16, 2002
1-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
2. PHYSICOCHEMICAL CHARACTERISTICS
This chapter provides an overview of the physicochemical properties of perchlorate. These
are important to understanding the persistence of perchlorate in the environment and to
understanding how perchlorate is processed in various plants and animals. Additional
toxicokinetic and toxicodynamic information can be found in Chapters 3 and 6; additional data
on transport and transformation, including biotransport, are discussed in Chapters 8 and 9.
In the solid state, the perchlorate anion has been determined by X-ray diffraction to have a
nearly perfect tetrahedral geometry with the four oxygen atoms at the vertices and the chlorine
atom at the center as shown in Figure 2-1. In aqueous solution, the geometry is probably
perfectly tetrahedral. The average chlorine-to-oxygen bond distance is 1.42 pm (Schilt, 1979b),
and the oxygen-to-oxygen distance is 2.43 pm. The partial molar ionic volume is 44.5 cmVmol
at 25 °C, compared with 36.7 for iodide.
Figure 2-1. Chemical structure of perchlorate,
January 16, 2002
2-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
Perchlorate is widely known to be a very poor complexing agent and is used extensively as
a counter anion in studies of metal cation chemistry, especially in nonaquous solution (Urbansky,
1998). In this application, it is comparable with other noncomplexing or weakly ligating anions,
such as trifluoromethanesulfonate (triflate [CF3SOf ]), tetrafluoroborate (BF^-), and, to a lesser
extent, nitrate (NOj). Some exceptions are known, but are rare, such as some copper
compounds (Burke et al., 1982). All of these anions have a highly delocalized (CF3SOj, NOj,
ClO^-) or sterically blocked (BF4) monovalent anionic charge and large volume. The low charge
density reduces their affinity for cations and their extent of aquation (see Table 2-1).
TABLE 2-1. GIBBS FREE ENERGIES OF FORMATION FOR
SELECTED ANIONS IN AQUEOUS SOLUTION (Urbansky, 1998)
Anion
AG°, kJ Mol"1
bf4-
-1,490
po43'
-1,019
S042"
-744
hco3-
-587
OH-
-157
Cl-
-131
no3-
-109
Br-
-104
cio4-
-8.5
cio3-
-8.0
This low association with cations is responsible for the extremely high solubilities of perchlorate
salts in aqueous and nonaqueous media. As noted, the ammonium and the alkali metal salts of
perchlorate generally are readily soluble in water. Salts of the smaller univalent cations (i.e.,
ammonium [NH4+], lithium [Li+], and sodium [Na+]) are very soluble; whereas, those of the
larger univalent cations are less so (i.e., potassium [K+], rubidium [Rb+], and cesium [Cs+]).
Quaternary ammonium salts are less soluble still. The outstanding example is sodium
January 16,2002 2-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 perchlorate, which is extremely soluble (>8 mol dm'3). Table 2-2 lists these solubilities as well
2 as other key physicochemical properties.
3
4
TABLE 2-2. PHYSICOCHEMICAL PROPERTIES OF AMMONIUM AND
ALKALI METAL PERCHLORATES AT 25 °C (Schilt, 1979).
Magnitude of Physicochemical Property of Perchlorate
Physical Property
nh4
Li
Na
K
Rb
Cs
Molecular Weight (g mol'1)
117.49
106.40
122.44
138.55
Density
1.952
2.429
2.499
2.5298
2.9
3.327
Solubility (w/w %)
Water
24.922
59.71
209.6
2.062
1.338
2.000
Methanol
6.862
182.25
51.36
0.105
0.000
0.093
Ethanol
1.907
151.76
14.71
0.012
0.009
0.011
n-Propanol
0.387
105.00
4.888
0.010
0.006
0.006
Acetone
2.260
136.52
51.745
0.155
0.095
0.150
Ethyl Acetate
0.032
95.12
9.649
0.001
0.016
0.000
Ethyl Ether
0.000
113.72
0.000
0.000
0.000
0.000
Thermochemical data
Mi?, kJ mol"1
-290.4
-384.0
-385.7
-435.5
-434.7
-434.7
AG°, kJ mol'1
-88.9"
-254c
-255"
-304
-306
-307
AS°, kJ mol"1
186b
130c
142b
151
161
175
A//°h, kJ mol"1
-26.6
26.1
14.7
50.6
56.8
55.6
Magnetic susceptibility
(xlO6)
46.3
32.8
37.6
47.4
—
69.9
Molar refraction
17.22
—
13.58
15.27
—
—
"Thermochemical data converted from kcal/mol using 1,000 cal = 4.184 J.
'Weast (1989).
cDean (1985).
1 Because of their large solubilities, the health risk assessment for perchlorate anion (C104~)
2 is appropriate for perchlorate salts, including ammonium perchlorate [CASRN 7790-98-97,
3 sodium perchlorate [CASRN 7601-89-07, potassium perchlorate [CASRN 7778-74-77, and
January 16, 2002
2-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
lithium perchlorate [CASRN 7791 -03-9/. The estimate is not appropriate to characterize the risk
of effects of perchloric acid (HC104) [CASRN 7601-90-37 because it is a strong acid, and the
dominant mode of toxicity is the irritating action of the hydrogen ion on skin and mucous
membranes.
Perchlorate can be a strong oxidizing agent under certain conditions as indicated by its high
reduction potential; therefore, the question has arisen as to whether or not it has the potential to
behave as an oxidant in biological systems. The thermodynamics of the halogen oxoanions and
oxoacids to participate in redox reactions are well understood. Under standard conditions in 1 M
acid, where the species is reduced to chloride, the oxidizing strength and standard reduction
potential (E°) increase as follows: Cl2 < HOC1 < HC102 < C103" < C104~. The reduction
potentials for the oxoanions increase with increasing acidity or decreasing pH (i.e., they are
stronger oxidizing agents in acidic solution). Consider, for example, the reduction of
chlorine(VII) to chlorine(V) under both acidic and alkaline conditions. In 1.0 M H+(aq) solution
(pH = 0),
C104" + 2 H+ + 2 e" - C103" + H20, E° = 1.20 V. (2-1)
In 1.0 M OH"(aq) solution (pH = 14),
C104" + H20 + 2 e" - CIO3- + 2 OJT, E° = 0.37 V. (2-2)
The effect of pH can be explained in terms of Le Chatelier's principle. In Reaction 2-1,
hydrogen ion is plentiful and acts a reactant; this drives the reaction forwards. In Reaction 2-2,
hydroxide ion is a product of the reaction and is already present at 1 M. This reduces the driving
force for this reaction to take place. The reaction is still spontaneous, as shown by the positive
value of E°; nonetheless, the driving force is considerably smaller for this case.
Thermodynamically, perchlorate is a stronger oxidant in the chlorine oxoanion series at the
extremes of the pH scale; however, such extremes are difficult to achieve in vivo (Tsui, 1998).
In Chapter 1, perchlorate anion was described as a nonlabile oxidant. Although the driving
force for reduction is very high, the activation energy required to start the process is also very
high. With the chlorine oxoanions, kinetic lability runs counter to the thermodynamic stability.
January 16, 2002
2-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
That is, the most stable species, hypochlorite (C10~), reacts fastest; whereas, the least stable
species, perchlorate (C104~), reacts the slowest. It is important to note that the activation energy
required for the reduction of perchlorate to take place is a function not only of the perchlorate,
but also of the chemical nature of the reductant. With common reducing agents (e.g., thiosulfate,
sulfite, or ferrous ions), the activation energy is too high for any reaction to be observed. In fact,
this property (lack of lability) is exploited routinely in chemical studies where perchlorate salts
are used to control the ionic medium and strength, but do not themselves react.
An alternative way of expressing the thermodynamic driving force for a reaction is the
Gibbs free energy function. Although the driving force for redox reactions is often conveniently
expressed in terms of the potential, there are practical limitations to this approach. For example,
in the decomposition reaction of ammonium perchlorate in Equation 1-1, an electric potential
cannot be measured. The Gibbs free energy of reaction, AG°n, is a measure of the energy
available to do work when a reaction is performed under constant pressure at standard state
conditions.1 When ammonium perchlorate explodes, the gaseous products push against the
surrounding air and thereby perform expansion work on the atmosphere.2 AG^, specifies the
maximal nonexpansion mechanical work that can be obtained from a chemical reaction carried
out at constant temperature and pressure.3 If the nonexpansion work is the electrical work of a
redox process, then an additional relationship applies (Equation 2-3), where n is the number of
electrons transferred; F is the Faraday constant, 96,485 C (mol e)"1; and E° is the electric
potential for the reaction under standard state conditions.
'This is the case with reactions occurring exposed to the open air, rather than inside a sealed container.
In a sealed container, where volume is constant and pressure changes, a different thermodynamic quantity, the
Helmholtz free energy AA^, is used instead. The superscript circle indicates standard state conditions (i.e., solution
concentrations of 1 mol dm"3 and gas pressures of 1 bar). All thermodynamic data are for a temperature of 298 K.
All of the thermodynamic relationships herein apply at other conditions, and reference tables exist only for standard
conditions. For other conditions, appropriate corrections must be made.
2Expansion work (fV^) is significant only when a reaction has a net change in the number of gas molecules
and can be calculated from the equation of state for a perfect gas: W = -PA V = AnRT (where P = pressure (atm),
V= volume (L), n = number of moles, R = ideal gas constant (L atm k"1), and T = temperature (K) and T and P are
constant). For reactions occurring in the condensed phases, Wcxp ~ 0.
3To obtain the maximal nonexpansion work, it is assumed that the process occurs reversibly so the loss of
energy as heat is minimized. Although this is approximately true for an electrochemical cell, most chemical
reactions do not take place under conditions that approach reversibility. For example, explosions are so irreversible
because so much internal energy is lost as heat that the nonexpansion work is much smaller than AG^.
January 16, 2002
2-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
= -^max = ~nFE° (T, P constant)
(2-3)
The negative sign is necessary because the work done on the environment represents a loss of
free energy from the chemical system. Nonexpansion work includes, but is not limited to,
causing an electric current to flow or lifting an object against gravity. Whenever a chemical
reaction has the ability to do work on the surroundings, it will take place spontaneously.4 AG°
is calculated as follows using Hess's law:
The Gibbs free energy of formation, AG°, is calculated for the formation of a compound from its
standard state as an element; consequently, AG° = 0 for Cl2(g) and 02(g). For Reaction 1-1,
AG^ = 2AG° [N20(g)] + 8AG° [H20(g)] - 4AG° [nh4cio4(s)]
This large negative value for AG^, suggests that the decomposition of ammonium perchlorate is
spontaneous and has a large quantity of energy available to do work. When 4 moles (468 g) of
ammonium perchlorate decompose, enough energy is released to lift a 1 kg mass 130 km, heat
and completely boil 0.5 kg of water (starting from 25 °C), or power a 100-W light bulb for 3.5 h.
Each molecule contains a large amount of potential chemical energy; however, a handful of
ammonium perchlorate will not spontaneously explode. The free energy is not released because
the reaction kinetics are too slow at room temperature—only an infinitesimal fraction of the
molecules possesses enough energy to reach the activation energy of the transition state at any
point. The activation energy for the reaction between an ammonium cation and a perchlorate
anion also is too great for a reaction to occur.
headers who have studied thermodynamics will recall that the determining factor for the spontaneity of a
chemical process is a net increase in the entropy of the universe (i.e., Ai^ > 0). It can be shown that AG^ =
-TAS^V; therefore, A^,v > 0 means AG^ < 0, and A^v > 0 means AG^, < 0 (because T> 0). As a consequence of
these relationships, it can be stated definitively that negative free energy available to do positive nonexpansion work
is a measure of the thermodynamic spontaneity of a chemical reaction. This implies that any chemical reaction
capable of performing positive nonexpansion work will occur spontaneously. Conversely, positive free energy
suggests that the reverse reaction is spontaneous.
AG^ = S AG° (all products) - E AG° (all reactants).
(2-4)
= 2(104) + 8(-229) - 4(-89) kJ = -1,268 kJ.
(2-5)
January 16, 2002
2-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
The distinction between thermodynamic spontaneity and kinetic lability must be
emphasized. A reaction with AG^, « 0 and E° » 0 is thermodynamically favored, but may be
so slow as to take virtually an infinite amount of time to occur (as is the case with most
perchlorate reductions). On the other hand, a reaction that occurs very quickly may have a very
small driving force. Reaction rates are fast when the combined internal energies of the reactants
closely approach the activation energy required to form the transition state. In a similar case, the
kinetic barrier (activation energy) explains why an open gas jet does not burst into flame until the
heat of a match is applied.
It is well established that, in aqueous solution, chlorine(I), chlorine(HT), and chlorine(V)
species undergo their most facile reductions via nucleophilic attack at the chlorine atom rather
than at the oxygen atom. When oxoanions are dissolved in water, the rate of net oxygen atom
exchange (Equation 2-6) can be used to understand how reactions proceed:
0C10„" + H20 ^ 0C1O„~ + H20, 0 a labeled oxygen atom; 0 ^ n <, 3. (2-6)
Reaction 2-6 proceeds through an associative mechanism in which the incoming water molecule
attacks the central chlorine atom. Consider the simplest example, hypochlorous acid, for which
the following mechanism is the accepted explanation (where 0 is again a labeled oxygen atom):
H20
HOC1 + H20 ~ [HO CI OH,]* ~ OH" + C10H + FT — H20 + C10H (2-7)
The aquated species [HO - "CI - • ¦0H2]t represents the activated complex and is the transition state
of Reaction 2-7; the proton is not directly transferred from the labeled water molecule to the
hydroxide that is part of the HOC1 molecule. Rather, a proton is lost to the bulk water of the
solution form the activated complex, and another proton is gained. This activated complex may
revert back to reactants or proceed to products.5 As the number of oxygen atoms increases, the
water has greater difficulty accessing the reaction site. The oxidation state of the chlorine
increases by 2 with each additional oxygen atom; accordingly, the chlorine becomes more and
5Note that AG°„ = 0 because the reactants and products are chemically identical. This suggests a process at
equilibrium in which the forward and reverse rates are balanced.
January 16, 2002
2-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
more electron-poor and holds the oxygen atoms closer to share their electrons. (This factor will
be expanded on further when perchlorate is examined specifically.)
In perchlorate, which contains chlorine(VII), the central chlorine atom is sterically blocked
from the attack of an incoming reducing agent by the tetrahedrally oriented oxygen atoms.
As the oxidation state of the central chlorine atom increases, the strength of the chlorine-oxygen
bonds also increases. The electron-deficient chlorine(VII) draws electron density from the
oxygen ligands resulting in increased 0(/?7t:)-C1(J7t) back donation despite the high
electronegativity of the oxygen atoms. Increased O-Cl bond strength thus further complicates
oxoanion reduction by making oxygen-atom abstraction even more difficult.
Perchloric acid normally exhibits oxidizing behavior when heated and concentrated. When
cold and dilute, HC104 acts only as a strong Bronsted-Lowry acid with no more oxidizing
character than other mineral acids, such as sulfuric or hydrochloric acids. In the absence of free
H+, as in vivo, a reducer or a catalyst with a lot of free potential energy would be requisite to
increase the rate (Tsui, 1998).
All observable perchlorate reductions reported in the literature are initiated via oxygen
atom abstraction by air-sensitive transition metal species (Urbansky, 1998). The metal cations
that react with perchlorate are all sensitive to atmospheric oxygen because they are strong
(thermodynamically) and labile (kinetically facile) reductants. None of these metal ions would
survive under human physiologic conditions. Certainly, any reductant capable of reacting with
perchlorate, such as Tim(aq) (Earley et al., 2000), Ch3Re02 (Abu-Omar et al., 1996), or certain
Rev complexes (Abu-Omar et al., 2000) would not survive in a physiologic environment. Thus,
the activation energy to perchlorate reduction is so high that perchlorate cannot be expected to
act as an oxidant under human physiological conditions (i.e., dilute solution, moderate
temperatures, and nearly neutral pH). This is supported by absorption, distribution, metabolism,
and elimination studies that show perchlorate is excreted virtually unchanged after absorption
(see Chapters 3 and 6).
A catalyst increases the rate of chemical reactions by reducing the activation energy,
increasing the number of collisions, or properly orienting chemical reactants. Many catalysts
reduce the activation energy, but some have multiple effects. When a perchlorate ion collides
with a reducing agent, the two entities can recoil unaffected or they can interact. If they interact,
the entity they form is called an activated complex and is a transition state from which they can
January 16, 2002
2-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
separate or react. If they have sufficient internal energy (enough to overcome the activation
energy), the species will react. For perchlorate, this means an oxygen atom is transferred to the
reductant. If a catalyst is involved, it can act as an intermediate, removing oxygen atoms from
the perchlorate and transferring them to the reductant. In the case of the rhenium (V) catalysts,
the coordinated rhenium center accepts oxygen atoms from (and is therefore oxidized by) the
perchlorate. This oxidized species (now containing Rev") then transfers an oxygen atom to (and
is therefore reduced by) any reducing agent; however, the authors used thioethers and mercaptans
for this purpose (Abu-Omar et al., 2000). Of particular interest in this work was that the
conditions were not nearly so forcing as what is normally required for perchlorate reduction. The
reaction took place at roughly neutral pHs and ambient temperatures.
Some bacteria have catalysts (i.e, enzymes known as reductases) that allow the microbes to
use perchlorate as an oxidant in anaerobic metabolic pathways. Although oxygen is a stronger
oxidant than perchlorate, bacteria will utilize perchlorate under low-oxygen conditions. For
example, perchlorate-reducing monera use perchlorate reductases under conditions where
conventional inorganic chemistry suggests that perchlorate reduction should be imperceptibly
slow (Urbansky, 1998; Logan, 1998). Over the past few years, there has been a profusion of
work in this area, mostly slanted towards bioremediation (Coates et al., 1999, 2000; Logan, 2001;
Nzengung and Wang, 2000).
This chapter provides a brief summary of some physiochemical properties of the
perchlorate anion, especially the salient features that might bear on its environmental and
toxicological chemistry. Additional chemical issues are explored in some depth in Chapter 9 as
related to analysis of environmental samples. Additional chemical-specific issues as related to
the pharmacokinetics of perchlorate in organisms are discussed in Chapters 3 and 6.
January 16, 2002
2-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
3. TOXICOKINETICS/TOXICODYNAMICS AND
MODE-OF-ACTION TESTING STRATEGY
This chapter explains the rationale that was the basis of the testing strategy which was
designed to evaluate the potential critical targets for perchlorate and to establish a database
robust enough to support a quantitative risk assessment. Aspects of the toxicokinetics and
toxicodynamics of perchlorate and its interaction with the thyroid are discussed as the basis for
the development of a testing strategy based on the mode of action of perchlorate. Mode of action
is defined as a chemical's influence on molecular, cellular, and physiological functions (Federal
Register, 1996; Wiltse and Dellarco, 1996). Understanding the mode of action helps to interpret
the relevancy of the laboratory animal and human data to inform the most appropriate
dose-response procedure (see Chapter 7).
As discussed in Chapter 2, perchlorate salts dissolve readily in water. The resultant anion
is easily absorbed from the gastrointestinal tract. However, because of its high charge, neither
perchlorate, nor other electrolytes applied from aqueous solution or aqueous media penetrate the
skin readily (Scheuplein and Bronaugh, 1983). Uptake of inorganic ions such as perchlorate
through the skin is typically less than 10% and frequently less than 1%. Exposure via inhalation
of fumes or vapors is considered negligible because the vapor pressure of perchlorate salts and
acids is low at room temperatures. The risk from exposure to particles would depend on the
particle size distribution. Thus, the ingestion route is the major concern for the risk posed by the
perchlorate contamination and is the focus of this characterization.
3.1 ABSORPTION, DISTRIBUTION, METABOLISM, AND
ELIMINATION OF PERCHLORATE
Limited absorption, distribution, metabolism, and elimination (ADME) studies were in
existence prior to the testing strategy discussed in Section 3.5. Although experimental studies in
laboratory species and humans had been performed using radiolabeling techniques, most were at
high concentrations, and the published data were expressed simply as thyroid:blood ratios of
January 16, 2002
3-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
radioactivity counts that provided no information on internal dose to biological tissues. Oral
drinking water administration, the most relevant to the contamination issue, was not the norm.
Time-course studies were very limited and essentially nonexistent for repeated administration.
More importantly, no data existed on the co-administration of iodide (I ) and perchlorate even
though data were necessary to develop a physiologically based pharmacokinetic model (Fisher,
1998a). The following section describes the limited pharmacokinetic information that was
considered when the data gap was highlighted during the development of protocols for the testing
strategy. The development of physiologically-based pharmacokinetic models that describe
ADME for perchlorate with data from the testing strategy will be discussed in Chapter 6.
Perchlorate appears to be eliminated rapidly, primarily in the urine (>90%), and virtually
unchanged from both rats (Eichler and Hackenthal, 1962) and humans (Anbar et al., 1959).
Durand (1938) measured urinary elimination from two human subjects who ingested 794 mg of
sodium perchlorate in 100 g of water. Urinary elimination accounted for 50% of the
administered dose within 5 hr and 95% within 48 hr. Half-lives have been reported for the rat
ranging from <8 hr (95% in 60 hr) to -20 hr (Wolff, 1998). Stanbury and Wyngaarden (1952)
reported that perchlorate appears in the urine within 10 to 15 min of oral dosing and that peak
plasma levels occur within 3 hr. Perchlorate was reported to undergo a two-phased urinary
elimination process in rats and calves. In rats, the first phase accounted for approximately 96%
of the administered dose and had a half-life of 1 to 2 hr. The second phase accounted for 4% and
had a half-life that ranged from 72 to 80 h. In calves, the first-phase half-life was reported to be
2 to 2.5 hr, and the second 23 to 27 hr (Selivanova et al., 1986, as cited in Allred, 1998). The
kinetics of long-term administration of perchlorate have not been characterized. The distribution
and metabolism of perchlorate and its relevance to potential toxicity in the thyroid will be
discussed in greater detail in Section 3.3 following discussions of iodine metabolism and thyroid
physiology in Section 3.2.
3.1.1 Human Studies
The majority of the human data on perchlorate ADME prior to the strategy was comprised
of the therapeutic case and clinical studies of Graves' disease patients described in Section 4.2.2.
These studies established the effect of perchlorate on the sodium (Na+)-iodide (I ) symporter
(NIS) but were of limited use in establishing quantitative dose-response relationships.
January 16,2002 3-2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Anbar et al. (1959) demonstrated that perchlorate was not metabolized in humans. Four
patients were administered 200 mg (approximately 2.9 mg/kg using a default body weight of
70 kg) double-labeled K36C11804, and urine was collected 3 h after dosing. Perchlorate was found
to be excreted at approximately 200 //g/min in the urine. Total urine radioactivity was
distributed between 36C1,36C1I804", 36C104' and 36C1" and indicated that perchlorate was excreted
unchanged in the urine. No human data existed with which to adequately characterize the
pharmacokinetics of perchlorate during steady-state, low-dose, repeated administration.
3.1.2 Laboratory Animal Studies
Although the perchlorate discharge test has been performed in rats (Atterwill et al., 1987),
the procedure is very different than that used in humans and does not readily allow for
comparison or extrapolation. Rats are dosed intraperitoneally (ip) with 100 /^L (1 yuCi) 125I", then
dosed ip with potassium perchlorate at 5, 10, 25, or 50 mg/kg body weight from 1 to 6 h
afterwards. Results are expressed as thyroid:blood ratios, which differ from how most human
data are expressed. Additionally, the time points at which uptake is measured are highly
dissimilar to those used in human studies.
Anbar et al. (1959) also attempted to confirm the lack of perchlorate accumulation and lack
of metabolism in the thyroid in rats. White rats were injected ip with 36KC104, and the specific
activity per gram of tissue was measured at 30 min, 4 hr, and 12 hr. The activity was greatest in
the thyroid and peaked at 4 h. The salivary and adrenal glands also had high activity levels.
Rabbits also were tested; the thyroid activity levels were again the highest of any tissue and
peaked at 2 h. Rabbit testes had the next highest specific activities.
In one of the only co-administration studies, Anbar et al. (1959) simultaneously
administered 131I" and 36C104" in equimolar concentrations. The thyroid:blood specific activity for
iodide was slightly higher than the ratio for perchlorate (1.80 and 1.69, respectively).
Halmi et al. (1956) examined iodide uptake in male Sprague-Dawley rats when active
transport was completely blocked via sodium perchlorate. The rats were first administered 6 mg
of propylthiouracil (PTU) subcutaneously to prevent iodide organification. Iodide uptake was
prevented by administration of 100, 200, or 400 mg sodium perchlorate with half of each dose
administered along with the PTU and the other half administered 45 min later with 5 to 50 nCi
131r. The rats were sacrificed 1.0 to 1.5 h after the iodide administration. Perchlorate reduced the
January 16, 2002
3-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
thyroid:blood ratio from 22.7 to 0.45; radioiodide was found to account for 30% of the thyroid
gland volume when it entered the gland by diffusion alone. Rats sacrificed 4.0 to 4.5 h after
iodide administration produced similar results, indicating that equilibrium is reached prior to
1.0 to 1.5 h. The distribution of radioiodide in other tissues also was examined. Perchlorate did
not affect the organ:serum iodide ratios in the following organs: submaxillary salivary gland,
parotid salivary gland, pituitary gland, adrenal glands, testes, spleen, kidneys, lung, skin, or
diaphragm. However, perchlorate administration did affect the stomach wall:serum and gastric
juice:serum iodide ratios (0.36 and 0.75, respectively) compared with the ratios for controls
administered sodium chloride (1.45 and 15.8, respectively). This suggested a gastric iodide
pump subject to inhibition by perchlorate and, as will be discussed in Chapter 6, the
gastrointestinal tract is another tissue with NIS.
Goldman and Stanbury (1973) administered 0.1 yuCi of the potassium salt of 36Cl-labeled
perchlorate (K36C104) by ip injection to male Sprague-Dawley rats that had been maintained on a
low-iodine diet for 4.5 to 5.0 weeks prior to dosing (approximately 40 fj.g stable perchlorate per
injection). The radionucleide retention in the thyroid, expressed as percent of dose per gram of
tissue, was recorded at 2 h (6 rats), 4 h (5 rats), 8 h (6 rats), 24 h (6 rats), 48 h (6 rats), and 96 h
(5 rats). The peak was reported to appear around 4 h and then to fall to approximately 5% of this
peak value after approximately 96 h. An exponential function was used to estimate a half-life of
20 h. Urinary excretion data indicated that the disappearance rate from the plasma and thyroid
and the appearance rate in the urine corresponded closely although the question was raised as to
whether there is some curvilinearity to the urinary excretion, which may suggest limited
saturation. The retained dose and its standard deviation in tissues at 96 h were reported as
0.142 ± 0.1, 0.125 ± 0.09, 0.098 ± 0.03, 0.048 ± 0.04, and background for the thyroid, kidney,
spleen, liver, and brain, respectively.
Chow et al. (1969) compared the uptake of radiolabeled perchlorate and iodide ions with
stable ions in normal and thyroid-impaired rodents. Intact male Sprague-Dawley rats were
injected ip with 0.1, 0.2, or 5.0 meq/kg stable potassium perchlorate (14, 28, or 690 mg/kg,
respectively) 2 h prior to sacrifice. The specific activity of the chlorine label (36C1) was
25.2 ^Ci/mmol. Thyroid impairment was affected by pretreatment with thyroid-stimulating
hormone (TSH) (1 international unit [IU] TSH in 0.9% saline solution ip 18 h prior to
perchlorate administration), hypophysectomization (removal of the pituitary), TSH and
January 16, 2002
3-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
hypophysectomization, or PTU (0.1% PTU in drinking water for 2 weeks prior to perchlorate
administration). Perchlorate at the 0.1- and 0.2-meq/kg dose levels was found to preferentially
concentrate in the rat thyroid as compared to the plasma, and the concentration was related
inversely to dose. The high dose level did not result in the concentration of radiolabeled
perchlorate in the thyroid. Rats pretreated with TSH or PTU also concentrated perchlorate at the
lower dose levels. At the two lower levels, hypophysectomized rats were not able to concentrate
perchlorate compared with intact rats, but the thyroid perchlorate concentration at the high dose
level did not differ between intact and altered rats. In a second subset of the same study, rats
were exposed to 0.005, 0.01, 0.02, 0.05, or 0.10 meq/kg perchlorate (0.69, 1.4, 2.8, 6.9, or
14 mg/kg, respectively) under the same general conditions. The concentration of radiolabeled
perchlorate in the thyroid again was related inversely to perchlorate dose. Male albino guinea
pigs also were exposed to the same doses. The guinea pigs displayed the same relationships as
the rats, but concentrated more perchlorate in the thyroid compared to plasma levels.
Chow and Woodbury (1970) demonstrated that perchlorate is actively sequestered by the
thyroid gland at low doses but that the capacity of the symporter to actively sequester perchlorate
is exceeded at higher doses. Male Sprague-Dawley rats were functionally nephrectomized by
ligating the renal pedicle of both kidneys 24 h before the rats were sacrificed. Perchlorate was
administered as the radiolabeled potassium salt (K36C104) in solution by ip injection at 0.005,
0.1, or 2.0 mmol/kg stable potassium perchlorate (0.69, 14, and 280 mg/kg body weight,
respectively, assuming 0.266 kg body weight; actual weight 226 ± 4 g) 2 to 240 min before
sacrifice. A group of control rats received [14C]-insulin, 35S04"2 or 36C1" 2 h prior to sacrifice to
determine thyroid follicle volume and intrafollicular membrane potential. Concentrations of
perchlorate in the thyroid and plasma were measured at 0.033, 0.067, 0.13, 0.2, 0.50, 1.0, 2.0,
and 4.0 h after sacrifice. Again, perchlorate was actively sequestered by the thyroid gland at the
low dose, but the capacity of the symporter to actively sequester perchlorate was exceeded at the
higher doses (e.g., the thyroid:plasma [milligrams per gram:milligrams per liter] ratios at 15 min
or 4 h post-dosing were 6.4, 0.69, and 0.36 or 13.8, 0.93, and 0.44 at the 0.5, 14.0, or
280.0 mg/kg doses, respectively). These data suggest that maximal inhibition by perchlorate of
active uptake of iodide probably occurs below 14 mg/kg potassium perchlorate (10.0 mg/kg as
perchlorate). If perchlorate-induced inhibition of active iodide uptake is substantial, iodide still
may enter the thyroid by diffusion, but in a smaller amount. Likewise, if inhibition of iodide
January 16, 2002
3-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
uptake by perchlorate is incomplete, then iodide still may be actively sequestered into the thyroid,
again in a smaller amount. Thus, perchlorate-induced thyroid hormone perturbations may
plateau in adult rats dosed with perchlorate greater than approximately 5 to 10 mg/kg of
perchlorate (Fisher, 1998a).
Wolff and Maurey (1962) demonstrated the competitive nature of the perchlorate inhibition
in sheep thyroid tissue slices incubated at 37 °C for 100 min. This study showed that the
Km constants for anion accumulation and the constants for inhibition of accumulation were
identical within the error of the method.
Eichler and Hackenthal (1962) presented perchlorate elimination data for male and female
Wistar rats dosed subcutaneously with 0.2, 1.0, or 6.0 of the 36C1" sodium perchlorate salt
(Na36C104) per 100 g body weight (2, 10, or 60 mg/kg). The elimination curves showed nearly
linear, rapid excretion of perchlorate until 6 hr, at which time the curve slope started to decrease.
The rate of excretion increased with dose. The elimination rates of the different doses prior to
24 h were significantly different from each other but were similar after 24 h. Over 60 hr, 93.4 to
97.4% of the administered dose was recovered, again suggesting that perchlorate was not
metabolized.
In a recent review (Von Burg, 1995), perchlorate elimination curves in rats and calves were
described as biphasic in both species. For rats, 96% of administered perchlorate is eliminated
with a half-life of 1 to 2 hr. The second portion of the curve accounts for 4% of the dose, with
half-life of 72 to 80 hr. Calves have a faster overall rate of elimination, but the initial elimination
is slower. The first-phase half-life was 2.0 to 2.5 hr, and the second-phase half-life ranged from
23 to 27 hr.
An intravenous (iv) study performed at AFRL/HEST in Sprague-Dawley rats with
perchlorate to characterize its inhibition of iodide uptake supports the conclusion that there is
inhibition at low concentrations and there is a gradual plateau at higher concentrations (Meyer,
1998). Rats were dosed once by iv tail-vein injection with either 0.01, 0.1, 1.0, or 3.0 mg/kg of
cold (i.e., not radiolabeled) ammonium perchlorate in saline. Perchlorate was administered as
ammonium perchlorate, and the data are presented as milligrams perchlorate per kilogram body
weight. Two hours after dosing with perchlorate, the rats were dosed again by iv tail-vein
injection with 33 Mg/kg l25I dissolved in saline. Rats were sacrificed at selected times (n = 6 per
time point) up to 24 h. Total and free 125I were measured in serum, thyroid, and urine.
January 16,2002 3-6 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Perchlorate serum, thyroid, tissue, and urine analyses began in January 1999 and are reported in
Chapter 6. For control comparison, rats were dosed once by iv tail-vein injection with 33 (J.g/kg
nonradiolabeled iodide and 125I mixed in physiologic saline. Rats (n = 6) were sacrificed at the
same selected time points up to 24 hr.
Table 3-1 shows the percent of inhibition of 125I uptake as measured by bound l25I in the
thyroid. Inhibition of 125I uptake into the thyroid by perchlorate was measured by bound or free
l25I in the thyroid at various time points after the single-dose of perchlorate. Because the 125I was
administered 2 hr after dosing with ammonium perchlorate, these time points correspond to 4, 8,
and 11 h after dosing. The most profound inhibitory effects were found at the 1.0- and 3.0-mg
perchlorate/kg dose group; however, the trend for 125I inhibition is evident at the 0.01- and
0.1-mg/kg levels (Meyer, 1998). By 24 h (26 h after dosing with perchlorate), inhibitory effects
on l25I uptake were still observed at the 1.0- and 3.0-mg/kg dose groups.
Recovery of 125I in urine 24 hr after dosing with 125I (26 h after ammonium perchlorate) was
between 79 and 88% for control 125I-dosed rats and perchlorate-dosed rats. The control 125I-dosed
rats excreted 79.5% (SD ± 5.50) of the l25I dose over the 24-hr period; whereas, the perchlorate-
dosed rats excreted 87% (SD ± 7.84), 86% (SD ± 4.47), 87.8 (SD ± 20.20) and 79.3 (SD ± 10.58)
of the 125I dose in urine at the 0.010.1-, 1.0-, and 3.0-mg/kg dose levels, respectively. The
amount of 125I in serum was elevated in the perchlorate-dosed animals compared to the control
l25I-dosed rats for up to 6 hr in all dose groups, suggesting that thyroid function was altered by
perchlorate and that a transient "discharge" of organified 125I occurred as reported in studies
summarized in Chapter 3. Free l25I levels in serum were similar between perchlorate-dosed and
control 125I-dosed rats (Meyer, 1998). These results are consistent with those of Chow et al.
(1969) and Chow and Woodbury (1970). The pattern for the inhibition of iodide uptake, albeit
only after a single dose, is strikingly similar to the patterns shown for the thyroid hormone
decreases. Consequently, data on the species differences (i.e., rat versus human in particular) in
perchlorate inhibition of the symporter will provide a basis for evaluating the degree of
uncertainty that should be applied when utilizing laboratory animal data as the model for humans
(see Chapter 7).
Repeated dose studies in rats (Fisher, 1998a) and in humans (Channel, 1998a) to establish
the kinetics of perchlorate at steady-state performed by AFRL/HEST to further characterize the
inhibition of iodide uptake by perchlorate are discussed in Chapter 6.
January 16, 2002
3-7
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-1. PERCENT INHIBITION OF IODIDE UPTAKE IN THE
THYROID GLAND OF SD RATS DOSED WITH PERCHLORATE (Meyer, 1998)
Dose
[Iodide]
Percentage of
Time Points3
(mg perchlorate/kg)
Og/g)
Inhibition
2 hr
Controlb
24.4
—
0.01
21.3
13
0.1
18.6
24
1
7.4
70
3
2.99
88
6 hr
Control15
46.5
—
0.01
36.7
21
0.1
32.0
31
1
19.2
59
3
9.13
80
9 hr
Control15
55
—
0.01
49.2
11
0.1
39.2
29
1
24.7
55
3
10.0
82
"Time points correspond to dosing with 125I and to 4, 6, and 11 hr after dosing with ammonium perchlorate.
bDosed with only iodide (33 //g/kg).
1 3.2 IODINE METABOLISM AND THYROID PHYSIOLOGY
2 Iodine plays a central role in thyroid physiology as both a constituent of thyroid hormones
3 and a regulator of thyroid gland function. Like perchlorate, iodine is absorbed efficiently from
4 the gastrointestinal tract. Iodine in organic form is converted mostly to iodide before absorption
5 (Cavalieri, 1997). The kidneys account for about two-thirds of the iodide cleared from plasma
6 and more than 90% of the iodide cleared from the body. Sweat and breast milk account for
7 various fractions of iodide loss, and fecal elimination constitutes approximately 1% of total body
8 iodide clearance.
January 16, 2002
3-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The thyroid gland concentrates iodide against an electrochemical gradient by a carrier-
mediated mechanism driven by adenosine triphosphate (ATP). The activation energy required
for perchlorate reduction is so high that it cannot act as an oxidant under physiological conditions
(i.e., dilute solution, moderate temperatures, and neutral pH). Plasma membrane experiments
indicate that the sodium cation (Na+) and iodide cotransport are electrogenic, with a
thermodynamically downhill transport of approximately two Na+ ions driving one iodide ion
against its electrochemical gradient into the cell. The transport is sensitive to ouabain, an
inhibitor of ATPase. The molecule responsible for the transport of iodide has been named the
sodium (Nat)/iodide (I') symporter or NIS. The thyroid thus has a specialized ability to
concentrate iodide selectively from the surroundings where the concentration is very low (10"8 to
10"7 M) and where the concentration of chloride ions is in the order of 0.01 to 0.1 M. The
transport is "active," not only by electrochemical criteria, but also by metabolic ones: it does not
occur in the cold, it requires oxygen, and, as mentioned, it is a function of the ATP level.
In addition to the thyroid, other organs that can concentrate iodide include the salivary glands,
gastric mucosa, choroid plexus, mammary glands, and the placenta. Iodide secreted into the
saliva and gastric juice is reabsorbed in the small intestine (Cavalieri, 1997).
Nevertheless, it is essentially only in the thyroid that the newly concentrated iodide can be
metabolized further to form thyroid hormone; and, only in the thyroid, does TSH regulate the
process. Thyroid hormones play numerous and profound roles in regulating metabolism, growth,
development, and maintenance of homeostasis. It is generally thought that these actions result
from the effects of the thyroid hormones on protein synthesis (Hill et al., 1989).
Figure 3-1 shows a schematic representation of thyroid hormone biosynthesis and secretion
in a single thyroid follicular cell. The thyroid hormones are stored as amino acid residues in
thyroglobulin (Tg), a protein constituting most of the colloid in the thyroid follicles. In situ, the
follicular cell displays functional and structural polarity: the vascular space is at the bottom, and
the lumen of the follicle is at the top. The striated circle straddling the basolateral membrane
represents the iodide transporter. The process of thyroid hormone biosynthesis is first stimulated
by TSH binding to the follicular cell TSH receptor and cyclic adenosine monophosphate (cAMP)
activation (Hard, 1998). The protein portion of Tg is synthesized on rough endoplasmic
reticulum (ER), and carbohydrate moieties are added by the Golgi apparatus (GA).
Thyroglobulin proceeds to the apical surface in secretory vesicles (small open circles) that
January 16, 2002
3-9
DRAFT-DO NOT QUOTE OR CITE
-------
Apical
Surface
Basolateral
Membrane
Extracellular
Space
Follicular Lumen
(Colloid)
Figure 3-1. Schematic representation of thyroid hormone biosynthesis and secretion in a
single thyroid follicular cell. (Modified from Hill et al., 1989; Cavalieri, 1997;
and Fisher, 1996.)
1 fuse with the cell membrane and discharge their contents into the follicular lumen. Iodide enters
2 the cell by active transport, and then, at the apical surface, is oxidized by thyroid peroxidase
3 (TPO). The hydrogen-peroxide-generating system is represented by hydrogen peroxide (H202).
4 Organification occurs at or near this apical cell-colloid interface; the oxidized iodide is
5 incorporated into tyrosine residues in peptide linkage in Tg. Two iodinated tyrosyl groups couple
6 in ether linkage to form tetraiodothyronine (T4), which initially remains trapped in Tg. Hormone
7 secretion first involves pinocytosis of colloid-containing iodinated Tg (large open circle) at the
8 apical border of the follicular lumen and resolution into vesicles that fuse with lysosomes (LY,
9 dark circle). Lysosome proteolysis (striated circle) then converts Tg to amino acids, T4,
January 16, 2002
3-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
triiodothyronine (T3), diiodotyrosine (DIT) and monoiodotyrosine (MIT). Iodotryosine
dehalogenase regenerates iodide from MIT and DIT for reuse within the thyroid or release into
the blood, accounting for the iodide leak in the chronic state of iodine excess in certain thyroid
disorders. Type I iodothyronine deiodinase converts a fraction of the free T4 to T3. Both
hormones (T4 and T3) are released into the blood circulation by a process that is not well
understood. The thyroid also releases Tg, of which some is iodinated and some uniodinated as
newly synthesized protein.
Although T4 is by far the major hormone secreted by the thyroid (typically at 8 to 10 times
the rate of T3), T4 is considered a prohormone because about 33% of the T4 secreted undergoes
5'-deiodination to T3 in the peripheral tissues and T3 is about fourfold more potent than T4.
Another 40% undergoes deiodination of the inner ring to yield the inactive material, reverse
triiodothyronine (rT3), which recently has been postulated to play an inhibitory role on the
conversion of T4 to T3. T3 is regarded as the active hormone because it is the form that appears
to activate a response by nuclear DNA. Upon entering the circulation, both T4 and T3 are bound
and transported in strong, but not covalent, association with plasma proteins.
The major plasma-protein carrier in humans is thyroxine-binding globulin, a glycoprotein
with a very high affinity for T4 and a lower affinity for T3. In rats, the T4 and T3 are bound to
prealbumin (PA) or albumin with a weaker attachment. Control of the circulating concentrations
of these hormones is regulated primarily by a negative feedback involving three organs: (1) the
thyroid, which produces thyroid hormone, and (2) the pituitary gland and (3) hypothalamus,
which respond to and help maintain optimal T3 and T4 levels (Hill et al., 1998). Figure 3-2
shows the schematic for this hypothalamic-pituitary-axis and the feedback mechanisms.
The hypothalamus stimulates the pituitary gland through thyrotropin-releasing hormone
(TRH) to produce TSH, which prompts the thyroid to produce T4 and T3. Once secreted into the
blood, T4 and T3 are bound to plasma proteins (thyroid-binding globulin [TBG] in humans or
prealbumin [PA] and albumin in rats). In addition to the aforementioned conversion of T4 to T3
in peripheral tissues, thyroid hormone also is metabolized irreversibly in the liver by uridine
diphosphyl glucuronosyl transferases (UDPGTs) to either glucuronic (T4) or sulfate (mainly T3)
conjugates that are excreted in bile. A portion of the conjugated material is hydrolyzed in the
intestine, and the free hormones thus released are reabsorbed into the blood via enterohepatic
circulation. The remaining portion of the conjugated material is excreted in the feces.
January 16, 2002
3-11
DRAFT-DO NOT QUOTE OR CITE
-------
Influences from Periphery
:via Nervous System
(")
(")
Hypothalamus
(Paraventricular Nuclei)
TRH
(+)
Pituitary
(Thyrotroph)
[ifa ,+
v
Thyroid
Thyroglobuliri
,a0
^Peroxidase
I2 + Tyrosine 1
Deiodination-
,MIT ,
DIT,
• T4, T3
Plasma/Blood
J)
T4, T3
PA v
PP-TH
x TBG
Bound to
Plasma
_ Proteins
I'Trapping
(Thiocyanate^
Perchlorate )
Proteolysis
Organic Binding-coupling T4j T3 Release
Thiourea, PTU
Sulfonamides
Methimazole
Aminotrazole
(lodideA
V. Excess/
Target Tissue
(Nuclear Receptor)
(+)
Liver
UDIJGTs
T4-GLUC
Bilary
Excretion
Figure 3-2. Schematic of the hypothalamic-pituitary-thyroid axis and feedback
mechanisms (PP-TH = plasma protein-thyroid hormone, PTU =
propylthiouracil, UDPGT = uridine diphosphyl glucuronosyl transferase,
T4 GLUC = T4-glucuronide conjugate). (Modified from U.S. Environmental
Protection Agency, 1998a; Hill et al., 1998; and Capen, 1997).
January 16, 2002
3-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Cells in the hypothalamus and pituitary gland respond to levels of circulating T4 and T3
such that when thyroid production levels are high, there is a signal to reduce the output of (TRH)
and TSH. Similarly, when thyroid hormone levels are low, the pituitary is prompted to deliver
more TSH to the thyroid in order to increase the output of T4 and T3. This negative feedback
loop helps the body respond to varying demands for thyroid hormone and to maintain hormone
homeostasis. Thus circulating T4, T3, and TSH are monitored readily in experimental animals
and humans and so may serve as biomarkers of exposure to and indicators of the effects from
agents that disrupt the status of the hypothalamic-pituitary-thyroid axis (U.S. Environmental
Protection Agency, 1998a).
In the absence of thyroid-binding globulin, as in the rat and mouse, a greater fraction of
thyroid hormone is free of protein binding and subject to metabolism and removal from the body.
As a consequence, the half-life of T4 in the rat is only about 1 to 24 hr, in contrast to the 6 to
7 day half-life in humans. Rats compensate for the increased turnover rate by secreting more
TSH from the pituitary gland. Table 3-2 provides the interspecies and intraspecies differences in
both thyroid hormone and gland structure between rats and humans. The consequences of
disrupting the status of the hypothalamic-pituitary-axis will be discussed in Section 3.4.
3.3 TOXICOKINETICS OF PERCHLORATE
Because of the complex anatomy of the thyroid follicle, all of the locations where
perchlorate inhibition is exerted remain to be established (Wolff, 1998). Perchlorate has been
established as a competitive inhibitor of iodide uptake across the basolateral membrane (i.e., acts
by the inhibition at NIS). Figure 3-3 shows a comparison of the molecular dimensions of
perchlorate and iodide. The following potency series was constructed for monovalent anion-
based inhibition of iodide transport in thyroid slices: Tc04" ^C104" > Re04" > SCN" > BF4" > I" >
N03" > Br" > CI" (Wolff, 1998). However, it is not clear whether this anion sequence, measured
at very high concentrations, has any mechanistic relation to what occurs at low concentrations in
the thyroid. It is important to determine which solution properties of the anions determine this
sequence (e.g., crystal radius, hydrated radius, hydration enthalpy, charge density). Strong base
anion-exchange resins (usually a large cation with a weak field) exhibit a marked preference for
C104" (e.g., compared to CI ); thus, it seems likely that selectivity for iodide or perchlorate in the
January 16, 2002
3-13
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 3-2. INTERSPECIES AND INTRASPECIES DIFFERENCES IN
THYROID STRUCTURE AND T3, T4, AND TSH HORMONES
(U.S. Environmental Protection Agency, 1998a)
Parameter
Human
Rat
Thyroxine-binding globulin
Present
Essentially absent
T4 Half-life
5 to 6 Days
0.5 to 1 Day
T3 Half-life
1 Day
0.25 Day
T4 Production rate/kg body weight
1 x
10 x that in humans
TSH
1 x
6 to 60 x that in humans
Follicular cell morphology
Low cuboidal
Cuboidal
Sex differences
Serum TSH
Ma=F"
M 5 2 x F
Cancer sensitivity
F = 2.5 x M
M > F
°M = male, F = female.
Perchlorate (CIO/):
Max. Dimension:
(xyz) = 5.420
angstroms
Min. Dimension:
(xyz) = 4.724
angstroms
Iodide (I ):
Max. Dimension:
(xyz) = 4.400
angstroms
Min. Dimension:
(xyz) = 4.400
angstroms
Figure 3-3. Comparison of the molecular dimensions for the perchlorate (left) and iodide
(right) anions.
January 16,2002 3-14 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
thyroid may be based on an anion-exchange mechanism using a large cation such as a quaternary
amine (e.g., arginine) (Wolff, 1989).
Perchlorate also has been used to stimulate the efflux of iodide already stored in the
follicular lumen of the gland (Atterwill et al., 1987). The exact nature of the mechanism for this
effect has not been established, however. Transport of iodide out of the cell is downhill
electrically, but this could be accounted for by the high concentration gradient that is established
from follicular lumen (iodide stored in the colloid) to the basolateral and extracellular space.
This may be the rate-limiting aspect for perchlorate efflux effect. Perchlorate added to the apical
side of a polarized thyroid cell monolayer is substantially less effective than when added to the
basolateral side (Wolff, 1998). Moreover, perchlorate rapidly increases the secretory response to
TSH, and TSH increases iodide efflux before it increases iodide influx, suggesting that additional
control points may exist.
Thus, perchlorate appears to have no effect on the iodination process itself but, rather,
displaces iodide by competitive uptake at the NIS. Perchlorate is concentrated by thyroid tissue
in a manner similar to iodide, but it is not significantly metabolized in the gland nor peripherally,
as mentioned previously. It is not unequivocally established whether there are additional effects
of perchlorate on iodide transport within the thyroid. Pharmacokinetic studies with perchlorate,
both acute and particularly once steady state has been achieved, have provided some useful data
with which to gain insight on this issue. The potential impacts as health endpoints of interest for
human health risk assessment of this perturbation in the hypothalamic-pituitary-thyroid axis and
hormone economy will be discussed in Section 3.4.
3.4 TOXICODYNAMICS OF THYROID HORMONE PERTURBATIONS
Given the established mode of action for perchlorate as the inhibition of iodide uptake at
the NIS, it is important to distinguish the temporal aspects with respect to potential adverse tissue
response.
3.4.1 Carcinogenic Effects
In higher organisms, when demands for more thyroid hormone are small, existing thyroid
follicular cells can meet the demand. With increased need, as a result of certain chemical
January 16,2002 3-15 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
exposures or iodine deficiency, the thyroid responds by increasing the size (hypertrophy) and
number (hyperplasia) of thyroid follicular cells to enhance hormone output. With continued TSH
stimulation, there is actual enlargement of the thyroid (goiter) and, at least in rodents, eventual
neoplasia of the thyroid follicular cells. Because TSH-producing pituitary cells also are
stimulated, they too sometimes undergo hyperplasia and neoplasia (U.S. Environmental
Protection Agency, 1998a; Hill et al., 1998). The EPA Assessment of Thyroid Follicular Cell
Tumors (U.S. Environmental Protection Agency, 1998a), as well as reviews recommended
therein, provides details about thyroid follicular cell carcinogenesis. Figure 3-4 shows
schematically the possible antithyroid effects that could influence carcinogenesis. Note that
effects, not only in the thyroid but also in peripheral tissues and the liver, may cause demand on
thyroid hormone production such that the TSH stimulation of the thyroid to produce more
hormone is enlisted. Table 3-3 lists mechanisms of antithyroid-mediated neoplasia in rodents.
The potential for an indirect effect of perchlorate has been established, but genotoxicity
information was required to evaluate its potential for direct effects. As will be discussed in
Section 3.5, a battery of genetoxicity assays was included in the testing strategy.
Long-term perturbations in the hypothalamic-pituitary-thyroid axis by the various
influences listed in Table 3-3 are more likely to predispose the laboratory rat to a higher
incidence of proliferative lesions (Capen, 1997). One factor that may play a role in this
interspecies quantitative difference in sensitivity to thyroid stimulation is the influence of protein
carriers of thyroid hormones in the blood (Table 3-2). Both humans and rodents have
nonspecific, low-affinity protein carriers of thyroid hormones (e.g., albumin). However, in
humans, other primates, and dogs, there is a high-affinity binding protein, thyroxine-binding
globulin, which binds T4 (and T3 to a lesser degree). This protein is missing in rodents and
lower vertebrates. As previously indicated, T4 is bound to proteins with lower affinity in the
rodent and is more susceptible to removal from the blood, by metabolism, and through excretion
than in dogs and primates.
In keeping with this finding, the serum half-life of T4 is much shorter in rats (less than
1 day) than it is in humans (5 to 9 days); this difference in T4 half-life results in a 10-fold greater
requirement for exogenous T4 in the rat with a nonfunctioning thyroid than in the adult human.
Serum T3 levels also show a species difference: the half-life in the rat is about 6 hr; whereas, it is
about 24 hr in humans. High thyroid hormone synthetic activity is demonstrated in thyroid
January 16, 2002
3-16
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 3-4. Schematic of antithyroid effects that influence thyroid carcinogenesis. (U. S.
Environmental Protection Agency, 1998a; and Hill et al., 1998).
TABLE 3-3. MECHANISMS OF ANTITHYROID-MEDIATED
NEOPLASIA IN RODENTS (U.S. Environmental Protection Agency, 1998a).
• DNA Directed
- X rays
-131I
- Genotoxic chemicals
• Indirect
- Partial thyroidectomy
- Transplantation of TSH-secreting pituitary tumors
- Iodide deficiency
- Chemicals inhibiting iodide uptake
- Chemicals inhibiting thyroid peroxidase
- Chemicals inhibiting TH
- Chemicals inhibiting conversion of T3 and T4
- Chemical inhibiting hepatic thyroid hormone metabolism and excretion
January 16, 2002
3-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
follicles in rodents, where the follicles are relatively small and are surrounded by cuboidal
epithelium. Follicles in primates demonstrate less activity and are large with abundant colloid,
and follicular cells are relatively flattened (low cuboidal) (McClain, 1992).
The accelerated production of thyroid hormones in the rat is driven by serum TSH levels
that are probably about 6- to 60-fold higher than in humans. This assumes a basal TSH level in
rats and humans of 200 ng/mL and 5 ^U/mL, respectively, and a potency of human TSH of 1.5 to
15 IU/mg of hormone (U.S. Environmental Protection Agency, 1998a). Thus, it appears that the
rodent thyroid gland is chronically stimulated by TSH levels to compensate for the increased
turnover of thyroid hormones. It follows that increases in TSH levels above basal levels in rats
could more readily move the gland towards increased growth and potential neoplastic change
than in humans. In addition to considerations about the influence of serum thyroid hormone
carrier proteins, there are differences between humans and laboratory animals in size and life
span and in the pharmacokinetics and pharmacodynamics of endogenous and exogenous
chemicals. Any comparison of thyroid carcinogenic responses across species should be
cognizant of all these factors.
A number of goitrogenic compounds, those that either interfere with thyroid hormone
synthesis or secretion, have been demonstrated to result in thyroid follicular cell adenomas in
rats. Excessive secretion of TSH alone has been reported to produce a high incidence of thyroid
follicular cell adenomas. The pathogenic mechanism of thyroid follicular cell tumor
development in rodents involves a sustained excessive stimulation of the thyroid by TSH. In the
multistage model of this pathogenesis, the proliferative lesions often begin as hyperplasia, may
proceed to the development of benign tumor (adenomas), and infrequently develop into
malignant tumors (Figure 3-5).
The precise molecular steps in the carcinogenic process leading to thyroid follicular cell
cancer have not been elucidated totally although significant insights into the problem have been
described (Farid et al., 1994; Said et al., 1994). Normal cell division in the thyroid seems to be
affected by an interplay among several mitogenic factors, namely TSH, insulin-like growth
factor-1 (IGF-1), insulin, epidermal growth factor (EGF), and possibly fibroblast growth factor
(FGF). Additionally, other factors, such as transforming growth factor P, certain interferons, and
interleukin 1, may inhibit growth.
January 16, 2002
3-18
DRAFT-DO NOT QUOTE OR CITE
-------
Morphologic Continuum 11
Normal
Hyperplasia
Adenoma
Carcinoma
Significance in Risk Assessment
Figure 3-5. Proliferative changes involved in the multistage characterization of thyroid
follicular cell neoplasia in rodents represent a morphologic continuum.
Although these lesions typically are classified as discrete entities, the overlap in
morphologic features should be emphasized because only imprecise criteria to
separate borderline proliferative lesions exist. Thyroid neoplasia in rodents is
considered relevant to human risk assessment (U.S. Environmental Protection
Agency, 1998a) but thought to be protective (Capen, 1997).
1 Figure 3-6 shows the possible molecular events in human thyroid follicular carcinogenesis.
2 In spite of the potential qualitative similarities, there is evidence that humans may not be as
3 sensitive quantitatively to thyroid cancer development from thyroid-pituitary disruption as are
4 rodents. Rodents readily respond to reduced iodide intake with the development of cancer;
5 whereas, humans develop profound hyperplasia with "adenomatous" changes with only
6 suggestive evidence of malignancy. Even with congenital goiters from inherited blocks in
7 thyroid hormone production, only a few malignancies have been found in humans. Thus, despite
8 a common physiology in regard to the thyroid-pituitary feedback system, the role of disruption of
9 this axis in human cancer development is much less convincing. EPA has adopted the following
10 science
January 16, 2002
3-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
Figure 3-6. Possible molecular events in human thyroid follicular carcinogenesis (
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
• Adverse rodent noncancer thyroid effects (e.g., thyroid enlargements) following short- and
long-term reductions in thyroid hormone levels are presumed to pose human noncancer health
hazards.
The new data on the antithyroid activity of perchlorate that has resulted from the testing
strategy will be evaluated in Chapter 7 according to criteria provided in the guidance (U.S.
Environmental Protection Agency, 1998a) to determine the likelihood that the chemical would
act indirectly, via disruption of the thyroid-pituitary axis, or directly on DNA.
3.4.2 Neurodevelopmental Deficits and Other Potential Adverse Effects
Resulting from Thyroid Hormone Disruption
As expressed by the external review panel convened by Toxicology Excellence for Risk
Assessment (TERA) in 1997, there was concern about other potential adverse effects of
perchlorate-induced hypothyroidism. Humans respond as do experimental animals in regard to
short- and mid-term disturbances in thyroid functioning from various anti-thyroid stimuli such as
iodide deficiency, partial thyroidectomy (surgically or 13'I" induced), and goitrogenic chemicals
such as thionamides (U.S. Environmental Protection Agency, 1998a). For instance, thyroid
hormone is critical to normal brain and physical development. This dependency begins in the
uterus and extends to 3 years of age in humans. Thus, there was concern that hypothyroidism
during pregnancy could result in neurodevelopmental effects.
The role of the placenta in thyroid hormone metabolism is shown in Figure 3-7. Although
the fetus is initially dependent on maternal thyroid hormone levels, the potential for disruption of
fetal hormone production remains once the fetal thyroid assumes this function because
perchlorate can cross the placenta. Disruption of circulating thyroid hormones can have
drastically different effects on fetuses and infants than on adults, depending on the developmental
stage at exposure (Table 3-4). It is important to emphasize that even transient disruption may
lead to permanent effects in the developing organism.
Chemical-induced alterations in thyroid hormone homeostasis are known to adversely
affect the development of many organ systems, including the nervous and reproductive systems
(Porterfield, 1994; Jannini et al., 1995). Severe developmental hypothyroidism caused by iodine
deficiencies or a congenital condition has devastating effects on fetal and postnatal development,
including mental deficiencies and hearing, speech, and motor deficits (Porterfield, 1994; Sher
January 16, 2002
3-21
DRAFT-DO NOT QUOTE OR CITE
-------
Mother
+ TBG
+ T4
+ T3
+ T4
+ TSH
Iodine
TRH
TSH —
rT3 +
T4 ¦=
T3
PTU
Methimazole
CIO/
Placenta
Estrogens
hCG
TRH
£> rT3"
£> T2
Fetus
-~
-~
TRH
* rT3
-~ T4
-~ T3
-~
-~
-~
Figure 3-7. Schematic representation of the role of the placenta in thyroid hormone
metabolism during human pregnancy. The placenta produces estrogens and
hCG that increase maternal TBG levels and stimulate maternal thyroid
hormone production, respectively. Both activities tend to increase maternal
T4 and T3 concentrations and to inhibit maternal TSH secretion. Iodide and
TRH readily cross the placenta, and the placenta itself synthesizes TRH. The
placenta is impermeable to TSH and only partially permeable to T4 and T3.
Placental Type IU iodothyronine monodeiodinase enzymes degrade T4 to rT3
and T3 to 3,3'-diiodothyronine (T2). Propylthiouracil and methimazole
readily cross the placenta. Given its physicochemical characteristics and
similarity to iodide, perchlorate also is anticipated to cross readily. (Modified
from Fisher, 1996 and Underwood, 1998).
January 16, 2002
3-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
TABLE 3-4. MAIN SYMPTOMS AND EFFECTS OF HYPOTHYROIDISM
Developmental
Adult
(Transient disruption leads to permanent effects.) (Transient disruption leads to transient effects.)
et al., 1998). It is important to emphasize that these effects are caused by a lack of thyroid
hormones alone, rather than by tumor development or thyroid hypertrophy/hyperplasia due to
increases in TSH. Thus, the important species comparison may be perchlorate's action of iodide
uptake inhibition at the NIS. In fact, data discussed in Chapters 5 and 6 show that the sensitivity
of the NIS is quite similar across species.
During development, thyroid hormones regulate cell proliferation, migration, and
differentiation. Intracellularly, THs bind to thyroid hormone receptors that interact with thyroid
response elements to alter expression of messenger ribonucleic acids (mRNAs) and subsequent
protein synthesis. The pituitary-thyroid TSH feedback loop may or may not be activated during
development, depending on the mechanism of action of the chemical. The adversity of
congenital hypothyroidism, usually less severe than endemic cretinism, can be ameliorated via
early postnatal thyroxine therapy. In contrast, the effects of developmental iodine deficiency can
not be corrected with only postnatal therapy, indicating that iodine deficiency during pregnancy
is the causative action (Cao et al., 1994). Clearly, xenobiotics that contribute to fetal or maternal
hypothyroidism or hypothyroxenemia are of concern.
Since the previous external peer review, studies reported in the clinical and epidemiological
literature have reinforced concerns for deficits in neuropsychological development related to
maternal thyroid deficiency. Haddow et al. (1999) showed an effect on IQ scores in children
(ages seven to nine) who had normal thyroid function at birth but were born to women with
abnormal thyrotropin levels versus children born to a matched cohort of women with normal
thyrotropin levels as controls. Haddow et al. (1999) concluded that even mild and probably
• Delayed reflex ontogeny
• Impaired fine motor skills
• Deaf-mutism, spasticity
• Gait disturbances
• Mental retardation
• Speech impairments
• Run down, slow, depressed
• Sluggish, cold, tired
• Dryness and brittleness of hair
• Dry and itchy skin, constipation
• Muscle cramps
• Increased menstrual flow
» Thyroid tumors in rodents
January 16, 2002
3-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
asymptomatic hypothyroidism in pregnant women can adversely affect their children's
subsequent performance on neuropsychological tests.
Pop et al. (1995) noted an average impairment of 10.5 IQ points in the offspring of mothers
with high thyroid peroxidase antibody (TPO-Ab) titers during pregnancy. In a later prospective
study these same researchers evaluated developmental indices at 3 weeks, 10 months, 1 and
2 years of age and demonstrated that a maternal free T4 blood level that was less than the 10th
percentile of first trimester values (10.4 pmol/L in their study series) was associated with
distinctly impaired psychomotor development whether or not TSH and TPO-Abs were elevated
(Pop, et al., 1999). Smit et al. (2000) reported a similar relationship between free T4 and early
neurodevelopment of children born from treated hypothyroid women.
Morreale de Escobar et al. (2000) evaluated epidemiological, clinical, and basic research
data to ascertain if the principal factor leading to neurodevelopmental deficits in children was
related to maternal hypothyroidism, whether clinical or subclinical (as defined by TSH higher
than the 98th percentile of the normal population); or if they were instead related to maternal
hypothyroxinemia per se (decrement in T4 without concomitant increase in TSH). These
researchers concluded that conditions resulting in hypothyroxinemia alone (a low for gestational
age circulating maternal free T4 level whether or not TSH was increased) poses an increased risk
for poor neuropsychological development of the fetus. T4 is the required substrate for the
ontogenically-regulated generation of T3 in the amounts needed for optimal brain development,
both temporally and spatially. Normal maternal T3 concentrations did not seem to prevent the
potential damage of a low T4 supply (Morreale de Escobar et al., 2000). Hypothyroxinemia
seems to be much more frequent in pregnant women than either clinical or subclinical
hypothyroidism and autoimmune thyroid disease (AITD), especially in regions where the iodine
intake of the pregnant woman is inadequate to meet her increased needs for T4 (Morreale de
Escobar et al., 2000).
Figure 3-8 illustrates the windows of susceptibility for insults to the brain resulting from
hypothyroxinemia. A similar map has been developed for rats, and time lines have begun to be
compared and correlated (Rice and Barone, 2000), as shown in Figure 3-9. Morreale de Escobar
et al. (2000) reported findings that altered early migration of cortical cells can be observed in rats
with severe iodine deficiency. Porterfield (2000) has also discussed the potential for
January 16, 2002
3-24
DRAFT-DO NOT QUOTE OR CITE
-------
iodine-deficiency cretinism
mixed neurological and hypothyroid manifestations
neurological myxedematous
maternal thyroidal autoantibodies
congenital hypothyroidism
cochlea
cerebral cortex
striatum j
Subarachnoid pathways
Myelination
ts it
Major events in the CNS
Corpus callosum
cerebellum -
dentate of hippocampus
_L
T
2 3 4 5 6
Gestational age in six months
Birth (term)
Prematurity
Figure 3-8. Approximate timing of major insults to the brain resulting from
hypothyroxinemia, superimposed on major neurodevelopmental events in
humans. Conditions resulting in early maternal hypothyroxinemia, combined
to later impairment of the fetal thyroid, are the most damaging, with central
nervous system (CNS) damage that is irreversible at birth. The most frequent
cause is maternal iodine deficiency (ID) and the presence of maternal
autoimmune thyroid disease (AITD). Unless ID is also present, the CNS
damage in congenital hypothyroidism is preventable by early postnatal
treatment because the normal maternal thyroxinemia has avoided damage to
the brain until birth. If maternal hypothyroxinemia persists, normal maternal
concentrations of T3 do not protect the fetal brain because of its dependence
on intracerebral regulation of local T3 availability by deiodinating pathways
using T4 as a substrate. Interruption of the contribution of maternal T4 in
premature infants with an immature thyroid may also underlie their increased
risk of neurodevelopmental problems, the more severe the earlier their birth.
The question mark indicates that it is unknown whether very early CNS
development, corresponding to a period when the general morphogenesis of
the pros encephalon (neurolation and segmentation) is being determined, is
thyroid hormone sensitive or not (Morreale de Escobar et al., 2000).
January 16, 2002
3-25
DRAFT-DO NOT QUOTE OR CITE
-------
I Embryonic I
Fetal
Birth
Postnatal
Ovulation
— Fertilization
Implantation
K
Functional organization
H
Histogenesis
H
Organogenesis
GD 5-6 GD 15
|GD 8-9 GD 111
Adolescence,
GD 21-22
I-
NeurulaUon
Proliferation and migration
(A)
PND 15
Myeli nation
PND 35-45
Figure 3-9. Timelines of developmental processes in the nervous system of rats (a) and
humans (b). Rat timeline is compared to timing of fertilization, organogenesis,
and histogenesis. Human perinatal period is scaled in months and the
postnatal development is scaled in years (Rice and Barone, 2000).
January 16,2002
3-26
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
neurotoxicity and altered brain development that may result from exposure to environmental
chemicals that disrupt thyroid function even on a transient basis.
These concerns for the potential adverse effects of perchlorate on T4 and T3, especially
during pregnancy, are compounded by the growing appreciation that women of childbearing age
have relatively low iodide intake. A January 2001 report by the National Academy of Sciences
(NAS) concerning the dietary reference intake of trace-mineral nutrients, including iodine,
indicated that less than 25% of the total population was below the estimated average requirement
for iodide and stressed a need to look at levels of adequacy for susceptible age groups and status
during pregnancy and lactation. The higher requirements during this time indicate a potential
susceptibility as shown in Table 3-5. The NAS also cautions against using urinary iodine as a
biomarker for iodine status unless the data are from 24-hour collections or are normalized against
creatinine. Other reports suggest that the level of iodide intake is less than a third of the range
recommended for pregnant women by the World Health Organization (WHO) (Caron et al.,
1997).
TABLE 3-5. DIETARY REFERENCE INTAKES (DRI) FOR IODIDE
(National Academy of Sciences, 2001)
Age or Status
Adequate Intake (AI)
/ig/day
Estimated Average
Requirement (EAR)
Mg/day
Recommended Dietary
Allowance (RDA)
Mg/day
0-6 months
110
6-12 months
130
1 -3 years
65
90
4-8 years
65
90
9-13 years
73
120
14-18 years
95
150
19-15 years
95
150
51 + years
95
150
Pregnancy
160
220
Lactation
209
290
January 16, 2002
3-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The prevalence of abnormal thyroid function continues to be debated and this is
confounded by the variable definitions of the disease state as well as the different measures of
thyroid function (Canaris et al., 2000). Most reports are still defined by TSH levels rather than
for hypothyroxinemia per se, but recent presentations suggest that TSH is a poor test to assess the
severity of tissue hypothyroidism (Meier et al., 2001), and recommendations in the epidemiologic
literature are proposing that screening of pregnant women should include the determination of
free T4 (Morreale de Escobar et al., 2000). Age, sex and dietary iodine levels are confounding
factors, although virtually all studies report higher prevalence rates for hypothyroidism
(as defined by increased TSH) in women with age (Canaris et al., 2000). Rates as high as 24%
among women older than 60 years have been reported. Suppressed TSH levels have been
associated with decreased bone density, increased risk of atrial fibrillation, premature atrial beats,
and effects on serum lipids notably elevated serum cholesterol levels.
Together these findings strongly suggest that a susceptible population of particular concern
for perchlorate exposure is pregnant women with hypothyroxinemia and that the iodine
deficiency represents an additional potential insult that could exacerbate the effects of perchlorate
toxicity. The elderly, especially women, represent another potentially susceptible population, as
well as people with cardiac dysfunction or risk factors such as elevated serum cholesterol.
As mentioned above, reproductive toxicity was also a concern as a potential effect of
perchlorate's mode of action. In females, thyroid hormones appear to have a role in stimulating
the onset of human chorionic gonadotropin (hCG) production by the placenta early in pregnancy.
Human chorionic gonadotropin is essential for the maintenance of pregnancy. Therefore, a
hypothyroid condition has potential to interfere with normal placental function and fetal
survival, as well as the potential to interfere with lactation. Suppression of thyroid hormone
secretion with radioactive iodine or goitrogens reduces milk yield in lactating animals. This
effect may be caused by suppression of placental lactogen production. Thyroid-releasing
hormone is known to play a role in prolactin release during the estrous cycle. Additionally, the
thyroid is necessary for the transition to the anestrus state in seasonally breeding species.
In summary, effects on thyroid hormone levels have roles in estrous cycle regulation, pregnancy
maintenance, fetal growth, and lactation.
In males, the primary effects of hypothyroidism appear to occur during testicular
development. The testis is responsive to thyroid hormones only during a limited time during the
January 16, 2002
3-28
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
perinatal and prepubertal periods. Thyroid hormone is a major regulator of seminiferous
epithelium development by inducing the normal differentiation of Sertoli cells, gonocytes, and
Leydig cells, and by limiting the proliferation of those cell types. In the hypothyroid condition,
those cells proliferate beyond the norm, and the steroidogenic function of the Leydig cells, on a
per-cell basis (but not necessarily in total), is impaired. Secretory activity of the Sertoli cells also
appears to be impaired. In boys, untreated hypothyroidism is associated with marked and
precocious testis enlargement, but low androgen activity. In a small study, hypothyroid men had
complaints of reduced libido that was probably related to a defective leutenizing hormone
response to gonadotropin-releasing hormone.
The inclusion of an immunological evaluation of mice exposed to perchlorate was
warranted because of evidence from earlier clinical studies that indicated a link between the
treatment of Graves' disease with perchlorates and serious hematological effects that may be
linked to immune mechanisms. A small number of patients undergoing perchlorate therapy have
been reported to develop aplastic anemia, agranulocytosis, lymphadenopathy, leukopenia, or skin
rashes. The antithyroid drugs propylthiouracil and methimazoles are reported to exert their
effects on the hematopoietic system through immune mechanisms. Because the use of these
antithyroid drugs by a small number of patients also resulted in sequelae similar to that of some
patients under perchlorate treatment, it has been postulated that perchlorate also may act via the
immune system.
3.5 DEVELOPMENT OF A TOXICITY TESTING STRATEGY BASED
ON MODE OF ACTION
Because the RfD is intended to be a lifetime dose-response estimate, the typical objective
of a database to support such a quantitative assessment is to evaluate a comprehensive array of
testing endpoints that represent various life stages during which potential effects could occur
(e.g., the developing fetus through adult) and for effects on reproductive capability (shown
schematically in Figure 3-10). As discussed in the previous sections, thyroid hormone
deficiencies, such as those induced by perchlorate, can affect normal metabolism, growth, and
development. No robust data existed prior to this time to evaluate other potential target tissues or
effects. There were limited data on effects caused by long-term exposures and no data with
January 16, 2002
3-29
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
) * o
0 i
V s
Lnrr hTI \i 1
• v"
'if?
Tl «
Reproductive
Developmental
General Toxicity
Figure 3-10. Schematic illustrating that a high confidence RfD is based on data that
address all potentially critical stages over a lifetime.
which to evaluate the effects of perchlorate in potentially susceptible populations such as in
developing fetuses, nor were there data on the effects of perchlorate on the reproductive capacity
of male or female laboratory animals. Table 3-6 shows the minimum database for derivation of
an RfD with low confidence (a 90-day bioassay) and the rationale for other tests typically
included to bolster the confidence in the derivation-the same suite of tests that has been
discussed for perchlorate. These data typically also reduce the uncertainty for which uncertainty
factors are applied (see Table 3-7), either because the absence of data on a suspected endpoint
(e.g., developmental toxicity) has been addressed or because mechanistic data provide insight on
the relevance of the laboratory animal model, including the magnitude of interspecies and
intrahuman variability in toxicokinetics and toxicodynamics. Any individual chemical database
may fall in between this range of high and low certainty, depending on the quality of the
individual studies and whether the dose response for suspected endpoints is characterized well.
The objective of the testing strategy was to provide a comprehensive database that
described the mode-of-action-based pathogenesis in quantitative terms so that the resultant
estimate could be more predictive and ultimately support the development of a robust RfD
estimate that reduced the uncertainties inherent in the provisional, presumably protective values
(see Figure 3-11).
January 16, 2002
3-30
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
TABLE 3-6. MINIMUM DATABASE FOR DERIVATION OF AN
ORAL REFERENCE DOSE
Mammalian Database" Confidence Comments
Two chronic oral bioassays in different Highb Minimum database for high confidence
species
One two-generation reproductive study
Two developmental toxicity studies in
different species
One subchronic oral bioassay Low Minimum database for estimation of an RfD
"Rationale is to use different species to evaluate variability in species sensitivity unless a particular laboratory
animal model is more appropriate.
bRationale is to address all potentially critical life stages.
TABLE 3-7. FACTORS FOR UNCERTAINTIES IN APPLIED EXTRAPOLATIONS
USED TO DERIVE REFERENCE DOSES®
10„ -
- Human to sensitive human
10A -
- Experimental animal to human
10s -
- Subchronic to chronic duration
10L ¦
- LOAEL(HEE)" to NOAEL(HEE)"
10D ¦
- Incomplete to complete database
MF -
- Modifying factor. Professional assessment of scientific uncertainties of the study and database not
explicitly addressed above. Default for the MF is 1.0 (e.g., applied for small sample size or poor
exposure characterization).
"HEE = human equivalent exposure.
As illustrated in Figure 3-11, it is ultimately desirable to have a comprehensive
biologically-based dose-response model that incorporates the mechanistic determinants of
chemical disposition, toxicant-target interactions, and tissue responses integrated into an overall
quantitative model of the pathogenesis (Jarabek, 1995a). Because the internal tissue dose of the
chemical or its toxic moiety in a target tissue is not always proportional to the applied dose of a
compound, emphasis has been placed on the need to distinguish clearly between the exposure
concentration and the dose to critical target tissues. Consequently, the term "exposure-dose-
response" has been recommended as more accurate and comprehensive (Andersen et al., 1992).
This expression refers, not only to the determination of the quantitative relationship between
exposure concentrations and target tissue dose, but also to the relationship between tissue dose
and the observed or expected responses in laboratory animals and humans. The process of
January 16, 2002
3-31
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Chemical
Exposure
Concentration
"Dose"
Toxicological
Response
Protective
Predictive
Exposure
Default
Exposure
Mechanisms .
y '
Disposition Models
Tissue
Dose
Exposure
Mechanisms
r
Tissue
Dose
Toxicant
Tissue
Interaction
Disposition Models
. Mechanisms .
Toxicant-Target Models
Exposure
Mechanisms
Tissue
Dose
Response Qualitative
Response
Response
Response
t
Quantitative
Mechanisms
Mechanisms
Disposition Models
Toxicant-Target Models Tissue Response Models
Figure 3-11. Schematic characterization of comprehensive exposure-dose-response
continuum and the evolution of protective to predictive dose-response
estimates (U.S. Environmental Protection Agency, 1994 and Jarabek 1995b).
determining the exposure-dose-response continuum is achieved by linking the mechanisms or
critical biological factors that regulate the occurrence of a particular process and the nature of the
interrelationships among these factors. This can be especially important for interspecies
extrapolation and to understanding intrahuman variability.
Dose-response estimates based on characterization of the exposure-dose-response
continuum at the rudimentary ("black box") level necessarily incorporate large uncertainty
factors to ensure that the estimates are protective in the presence of substantial data gaps. With
each progressive level, incorporation and integration of mechanistic determinants allow
elucidation of the exposure-dose-response continuum and, depending on the knowledge of model
parameters and fidelity to the biological system, a more accurate characterization of the
pathogenesis process (Jarabek, 1995a). Because of the increase in accuracy of the
characterization with each progressive level, dose-response estimates also progress from more
protective to factually-based (predictive).
January 16, 2002
3-32
DRAFT-DO NOT QUOTE OR CITE
-------
Eight new studies were recommended as part of the original testing strategy after the May
1997 external peer review to provide such a comprehensive array of endpoints. These studies are
described below along with the role they were anticipated to play in informing the revised health
risk assessment (see Table 3-8).
TABLE 3-8. PERCHLORATE PEER REVIEW RECOMMENDED
STUDIES SUMMARY
Study
Description
Potential Use in Assessment
90-Day subchronic bioassay
+ THa + reproductivity +
genotoxicity + recovery
Developmental
neurotoxicity + TH
Developmental study + TH
Two-Generation
reproductive toxicity + TH
ADME studies
Mechanistic studies
Genotoxicity assays
Immunotoxicity
Tests for other target tissues;
evaluates effect on TH in young adult
rats; reproductive parameters added;
mouse micronuclei and a recovery
group
Evaluates nervous system in fetal and
postnatal rats; TH in does
(PO-generation) and pups
(F1-generation)
Evaluates birth defects in rabbits;
TH in does at end of gestation
Evaluates fertility of adult rats and
toxicity in offspring over two
generations; TH in parents
(FO-generation) and offspring
(Fl- and F2-generations)
Characterize absorption, distribution,
metabolism, and elimination in rats
and humans; iodine inhibition and
perchlorate kinetices and hormone
homeostasis
Evaluate mechanism of TH response
and sensitivity in rats and humans
Test for toxicity to DNA
Evaluates immune system structure
and function
Minimum database for RfD dose-
response for TH in young adult rats;
additional information on others; may
allow decrease in uncertainty factor
(UF) for database deficiencies
Potentially critical effect; comparison of
developmental versus adult effects on
TH
Potentially critical effect; data in second
species for TH effects; may reduce UF
for database deficiencies
Potentially critical effect; may reduce
UF for database deficiencies
Interspecies extrapolation
Interspecies extrapolation; determine
susceptible subpopulation
Mode-of-action information for thyroid
neoplasia; may reduce UF for database
deficiencies
Potentially critical effect; may reduce
UF for database deficiencies
"Thyroid hormones (T4 and T3); Thyroid stimulating hormone (TSH), a pituitary hormone, was also assayed in
those studies.
January 16, 2002
3-33
DRAFT-DO NOT QUOTE OR CITE
-------
]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
(1) 90-Day Subchronic Oral Bioassay Study. This study was considered the minimum data
requirement for derivation of an oral RfD. The study aimed to identify other target tissues,
to test young adult rats, and to provide data on the effect of repeated exposure to perchlorate
on thyroid hormone levels. The 30-day recovery phase, i.e., evaluation of the thyroid status
30 days after perchlorate was stopped, would provide data necessary to characterize its
anti-thyroid effects with respect to carcinogenicity (U.S. Environmental Protection Agency,
1998a). These data were collected to allow reduction of the uncertainty factor applied for
database deficiencies.
(2) Developmental Neurotoxicity Study. This study was designed to evaluate the potential for
developmental neurotoxicity of perchlorate by assessing functional and morphological
endpoints in offspring from the mother exposed during pregnancy and lactation.
Neurotoxicity endpoints were likely to be a critical effect, and the developing organism a
sensitive subpopulation. It was hoped that these data would allow reduction of the
uncertainty factors applied for intrahuman variability and database deficiencies.
(3) Segment II Developmental Study. This study was conducted to evaluate the potential for
perchlorate to cause birth defects in rabbits and to evaluate a potentially critical effect and
subpopulation. This study also was conducted to provide data on the thyroid hormone
effects in a second species (in addition to rats). These data might allow reduction of the
uncertainty factor applied for database deficiencies.
(4) Two-Generation Reproductive Toxicity Study. This study was designed to evaluate the
potential for perchlorate to cause deficits in reproductive performance in adult rats and for
toxicity in the young offspring. The primary goal of this study was to identify a potentially
critical effect and to allow for reduction of the uncertainty factor applied for database
deficiencies.
(5) Absorption, Distribution, Metabolism, and Elimination Studies. These ADME studies
aimed to understand the pharmacokinetics (i.e., how perchlorate is absorbed, distributed,
metabolized, and excreted) of perchlorate in test animals and humans. These data were to
January 16, 2002
3-34
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
provide information to support construction of quantitative extrapolation of dose across
species (e.g., rat to human).
(6) Perchlorate Mechanism Studies. These studies provided a link to the pharmacokinetic
studies and were conducted via a comparison of existing literature and of new in vitro and
in vivo data that evaluated the effects of perchlorate on the iodide uptake mechanism across
species to aid in the quantitative extrapolation of dose.
(7) Genotoxicity Assays. These studies evaluated the potential for carcinogenicity by
evaluating mutations and toxic effects on DNA. These data were useful to determining
whether the benign thyroid tumors were likely to be a result of the proposed threshold
pathogenesis process.
(8) Immunotoxicity Studies. These studies were planned to evaluate the potential for
perchlorate to disrupt immune function and identify a potentially critical effect. These data
would help to reduce the uncertainty factor applied for database deficiencies. Because
concern was raised for these potential adverse effects based on the previous clinical
experience with treatment of Graves' disease patients, these studies were considered
necessary to a comprehensive database for perchlorate.
In the 1998 external review draft (U.S. Environmental Protection Agency, 1998d), a model
based on mapping the events of the mode of action for perchlorate was proposed as shown in
Figure 3-12. The key event was identified as the inhibition of iodide uptake at the NIS, followed
by decreases in thyroid hormones and increases in TSH. Both the potential neurodevelopmental
and neoplastic sequelae of this perturbation in thyroid hormone economy were proposed as
downstream adverse health outcomes. The conceptual model was endorsed by the external peer
review panel in 1999 (Research Triangle Institute, 1999), and additional studies were
recommended to reevaluate indications of developmental and neurodevelopmental in rats for
effects observed in the 1998 database. Delineating the continuum of histopathological changes
in the thyroid was also recommended. The results of all the studies in the testing strategy (both
January 16, 2002
3-35
DRAFT-DO NOT QUOTE OR CITE
-------
• i-
Exposure
Effect^
Noncancer Cancer;
¦ ¦ a a i a a
¦IBB I I g
¦ i a a iaa
Susceptibility
Figure 3-12. Mode-of-action model for perchlorate toxicity proposed by the U.S. EPA
(U.S. Environmental Protection Agency, 1998d). Schematic shows the
exposure-dose-response continuum considered in the context of biomarkers
(classified as measures of exposure, effect, and susceptibility) and level of
organization at which toxicity is observed (U.S. Environmental Protection
Agency, 1994; Schulte, 1989). The model maps the toxicity of perchlorate on
this basis by establishing casual linkage or prognostic correlations of
precursor lesions.
1 "old " 1998 and "new" 2001), as well as additional studies now available in the literature, will be
2 reported together with EPA's interpretation and evaluation in Chapter 5.
January 16, 2002
3-36
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
4. HUMAN HEALTH EFFECTS DATA
The available data on the human health effects of perchlorate exposures are limited. Until
the emerging concern regarding environmental contamination, the majority of the studies were
clinical reports on patients treated with potassium perchlorate for Graves' disease. The
non-EPA, independent peer review held in March 1997 (Toxicology Excellence for Risk
Assessment, 1998a) concluded that the experimental design limitations of the studies prior to that
time precluded their use in quantitative dose-response assessment. The CA DHS also determined
in 1997 that there were major limitations on the human studies. Nevertheless, the studies were
useful in hazard identification and supported the conceptual model for the mode of action of
perchlorate available at the time as described in Chapter 3.
Since the external peer review of the previous 1998 external review draft held in 1999 by
the U.S. Environmental Protection Agency (Research Triangle Institute, 1999), some ecological
studies have been performed that have addressed the limitations in the human data with some
success. Two occupational populations with inhalation exposure to perchlorate were also
studied, and some additional clinical studies in healthy adults performed. On December 14,
2001, after internal peer review of this document, the Agency articulated its interim policy on the
use of third-party studies submitted by regulated entities (U.S. Environmental Protection Agency,
2001c). For these purposes, EPA is considering "third party studies" as studies that have not
been conducted or funded by a federal agency pursuant to regulations that protect human
subjects. Under the interim policy, the Agency will not consider or rely on any such human
studies (third-party studies involving deliberate exposure of human subjects when used to
identify or quantify toxic endpoints such as those submitted to establish a NOAEL or NOEL for
systemic toxicity of pesticides) in its regulatory decision making, whether previously or newly
submitted. Some of the clinical studies contained in this database fall in this category of studies
not to be considered. However, the scientific and technical strengths and weaknesses of these
studies were described before this Agency policy was articulated. Therefore, because of the
scientific shortcomings of these studies, they will not be used as "principal studies" in the
derivation of a RfD. The ethical issues surrounding the conduct of these studies or their use for
January 16, 2002
4-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
regulatory purposes in light of the Agency's interim policy will not be discussed in this
document. The Agency is requesting that the National Academy of Sciences conduct an
expeditious review of the complex scientific and ethical issues posed by EPA's possible use of
third-party studies which intentionally dose human subjects with toxicants to identify or quantify
their effects.
4.1 EPIDEMIOLOGICAL DATA
To be informative to quantitative dose-response analysis for risk assessment applications,
epidemiological studies must pose research questions that are based on appropriate physiological
issues relevant to the mode of action for the chemical and its toxic effect. In some contexts, a
sufficient specification may take relatively simple form. For example, with occupational cancer,
the generally assumed underlying mechanisms lead to a simple test: does exposure to a substance
or mixture specified as a dependent parameter, X, at time t, increase the incidence of specific
cancers at time t2 > t, + a, where a >0 is some lag time. The relation of risk at t2 to the history of
prior exposure may be a complex one, but almost always, risk is an increasing function of
exposure at various time intervals, X(lt). This test may require controlling for confounding
factors, which is usually not difficult when relevant detailed information is available.
In contrast, determining the effect of an environmental exposure on a regulated system
could be more of a challenge. Thus, cancers whose risk depends on endocrine status introduce
increased complexity. Environmental perturbations of physiological systems that have inherent
variability over time and are imbedded in control networks that function to minimize disruption
make it a challenge to determine which endpoints to measure. Cross-sectional assessments
during chronic exposures may capture variability in some regulated biological parameters while
other parameters will tend to stabilize at "normal" levels despite substantial environmental
impact on production and function. In such instances it can be difficult to distinguish alterations
due to the xenobiotic from the variation that occurs in response to other environmental factors.
Short-term fluctuations in exposure often have no effect independent of cumulative dose for
chronic diseases such as lung cancer or other respiratory diseases but may be important for
endocrine system functions that affect neurodevelopmental, hyperplastic, neoplastic, immune, or
autoimmune events (Park, 2001).
January 16,2002 4-2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The effect of the perchlorate anion on the hypothalamic-pituitary-thyroid feedback system
is an example of a regulated system that is potentially difficult to characterize. Important effects
may be evident as shifts in average levels of measurable factors, but more important effects may
involve alterations in transient responses to demands on the regulated system (Park, 2001).
Multiple covariates that may influence potential perchlorate health effects include iodine
availability, age, gender, ethnicity, health status, diet, and possibly social class. For neonates, the
birth process itself stimulates an endocrine cascade with the amplitudes of endpoint variation
depending on birth weight, gestational age, age at sampling (in hours), and possibly
environmental temperature. Post-partum developmental risk factors for the neonate and growing
child include perchlorate exposure via lactation or consumption of contaminated water.
Individual perchlorate exposures are difficult to measure or even estimate in population-
based studies. This makes their usefulness to quantitative dose-response analysis limited,
particularly if confounding variables are not controlled and small population sizes are evaluated.
The few population-based studies from geographic areas that have experienced perchlorate
contamination offer little help beyond indicating that clinical thyroid disease is not greatly
increased in populations with sustained drinking water contamination as high as 15 /ig/L in the
past. However, most of the studies have principally evaluated thyroid function or hormone status
and have not evaluated neurodevelopmental or other deficits in children or adults resulting from
perturbed thyroid function over sustained periods of exposure.
4.1.1 Ecological Studies
Rockette and Arena (1983) reviewed death certificates for workers known to have been
exposed to perchloric acid, magnesium perchlorate, and other chemicals in a U.S. chemical plant.
Because the workers had received multiple chemical exposures, the authors could not associate
an elevated death rate for a particular time period or work area and a specific chemical.
The Environmental Health Investigations Branch within the CA DHS, under a cooperative
agreement with ATSDR, conducted health assessment activities and consultations on the
Aerojet-General Corporation Superfund site in Sacramento County, CA (California Department
of Health Services, 1997; 1998a,b,c,d,e). A preliminary health review (California Department of
Health Services, 1997) analyzed several statewide databases for possible perchlorate-related
outcomes during the suspected years of contamination within the zip codes most likely exposed.
January 16,2002 4-3 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
In California, thyroid hormone levels in newborns are measured and kept on file by the Genetic
Disease Branch of the Centers for Disease Control and Prevention. Data for the period 1985
through 1996 from relevant zip codes was assessed for a total of 11,814 thyroid hormone screens.
Although an extrapolation of the statewide rate predicted there would be 3.76 cases of
hypothyroidism, four cases were observed. In the non-exposed areas, six cases of
hypothyroidism were found although 6.41 cases were predicted. These data suggested no
association between residence in the potentially-exposed zip codes and neonatal hypothyroidism.
The TSH levels (ascertained only in neonates with initially low T4 levels) in the potentially-
exposed areas were statistically significantly lower than those in the nonexposed areas. The
database also was evaluated for diagnosis of goiter among the first five reported hospitalized
individuals residing in the zip code of most likely contamination from the years 1991 to 1995.
Because there are so many diseases or conditions that can produce a goiter other than perchlorate
ingestion, and because the database can not differentiate this aspect, it was concluded that these
data would not be useful in determining the prevalence of thyroid enlargement due to perchlorate
in the affected water district. The same zip code also was evaluated for agranulocytosis or
aplastic anemia as one of the top five diagnoses for the years 1991 to 1995. There were a total of
76 cases in 5 years, less than the statewide rate of 41.6 per year. The rate for aplastic anemia was
3.8 hospitalizations per 100,000 individuals per year, a rate higher than the statewide rate of 2.2.
However, all but one of the hospitalizations also had an additional diagnosis of cancer with
chemotherapy or radiation treatment; these treatments are likely explanations for this
observation; acquired immunodeficiency syndrome (AIDS) may be another. The registry also
was searched for cases of childhood leukemia (either acute lymphocytic leukemia or acute
myelogenous leukemia). Again, the rate for the potentially exposed zip code was less than the
corresponding rate for California.
The CA DHS concluded that the data on goiter, agranulocytosis, and aplastic anemia did
not indicate an increase in incidence; however, these data do not provide definitive causitive
information because other likely causes for these conditions existed. Increases in the incidence
of decreased neonatal thyroid hormone levels, hypothyroidism, or childhood leukemia rates were
not observed. The CA DHS noted that the major limitation on studies of this nature is that
imposed by the absence of good exposure estimates and the absence of data on transport and
transformation models which would provide dose reconstruction for the affected population. It is
January 16, 2002
4-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
unclear when the contaminated plume entered the drinking water supply; consequently, the time
period analyzed may have been too broad. Improving this exposure information was one of the
recommendations made in the report to Congress regarding perchlorate (U.S. Environmental
Protection Agency, 1998e). Finally, that perchlorate is not specific for producing thyroid
dysfunction or hematological abnormalities makes assessing these outcome surveys difficult.
Table 4-1 shows the approximate prevalence of these disorders in the neonatal period
(1:30,000 to 1:100,000), and suggests that studies with large numbers of subjects may be
necessary to detect subtle effects.
Based on these results, the CA DHS investigated several other water service areas for
exposure (California Department of Health Services, 1998a,b,c,d,e) and ascertained that
complete exposure pathways to perchlorate contaminated water existed in several areas. These
studies reinforced the need for this document which attempts to properly characterize the risk
posed by perchlorate contamination by providing better exposure estimates and a revised health
risk estimate.
Since the 1999 external peer review, eight new population studies have been performed.
One of these studies has examined effects in the general population (Li et al., 2001), another in
school-age children (Crump et al., 2000), and six have been devoted to evaluating neonatal
endocrine status in areas with contaminated drinking water (Crump et al., 2000; Lamm et al.,
1999; Li et al., 2000a,b; Brechner et al., 2000; Schwartz, 2001). In each study, the critical
covariates were captured with varying degrees of success and only one study (Schwartz, 2001)
offers a convincing description of neonatal perchlorate effects (Park, 2001).
In a study of the general population, Li et al. (2001) investigated physician-generated
medical insurance claims for thyroid problems in a Medicaid insured population in Nevada,
comparing all counties that were known not to have perchlorate contaminated drinking water
with the one county that had contamination at approximately 10 /^g/L. This was a study of
period-prevalence, i.e., the proportion of the population that had claims for thyroid-related
disorders anytime during a two-year period. Incident cases could not be identified within this
database. Thyroid patients were defined as having one or more of the following diagnoses of
thyroid disease according to the International Classification of Diseases, 9th Revision (ICD-9):
(1) simple and unspecified goiter (ICD-9 Code 240); (2) non-toxic nodular goiter (ICD-9 Code
241); (3) thyrotoxicosis with or without goiter (ICD-9 Code 242); (4) congenital hypothyroidism
January 16, 2002
4-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
TABLE 4-1. THYROID DISORDERS AND THEIR APPROXIMATE
PREVALENCES IN THE HUMAN NEONATAL PERIOD (Fisher, 1996).
Thyroid Dysgenesis
Agenesis
Hypogenesis
Ectopia
1:4000
Thyroid Dyshormonogenesis
1:30,000
TSH unresponsiveness
Iodide trapping defect
Organification defect
Defect in thyroglobulin
Iodotyrosine deiodinase deficiency
Hypothalamic-Pituitary Hypothyroidism 1:100,000
Hypothalamic-pituitary anomaly
Panhypopituitarism
Isolated TSH deficiency
Thyroid hormone resistance
Transient Hypothyroidism 1:40,000
Drug induced
Maternal antibody induced
Idiopathic
(ICD-9 Code 243); (5) acquired hypothyroidism (ICD-9 Code 244); (6) thyroiditis (ICD-9 Code
245); (7) other disorders of the thyroid (ICD-9 Code 246) and (8) malignant neoplasms of the
thyroid gland (ICD-9 Code 193). Two of these disorders have very low prevalence: congenital
hypothyroidism (0.01%) and thyroid cancer (0.02%).
Comparisons were made between the exposed county, which includes Las Vegas, and
(a) an unexposed county with a similar large city (Reno), and (b) all other counties (unexposed).
There were no statistically significant period-prevalence rate differences between the exposed
county and the two categories of comparison counties; however, the differences between the
comparison county groups themselves were quite large, indicating that either important
confounding risk factors were not controlled or estimates were unstable due to the small numbers
of cases in the comparison counties. For acquired hypothyroidism, prevalences (%) in the two
categories of unexposed counties were significantly different (Reno: 1.17 [95% CI = 1.05 to
1.30, using a normal approximation to the Poisson distribution for number of cases] and other
January 16, 2002
4-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
counties: 1.44 [95% CI = 1.29 to 1.59]). Age, gender, ethnicity, iodine intake, and other
important risk factors were unavailable in this database and there could have been differential
under- or over-diagnosis in this Medicaid population. Interestingly, when comparing the two
counties with large urban centers and restricting focus to the 6 (out of 8) more prevalent
outcomes (total n=3069), all 6 showed elevated (but not individually significant) rate ratios for
the exposed county, ranging from 1.01 to 1.89. While these findings appear to rule out a large
perchlorate-related excess (i.e., greater than two-fold) for some thyroid disorders such as
acquired hypothyroidism (appearing as routine medical insurance claims), the study had a
statistical power of less than 0.5 to detect a 50% excess for several specific thyroid disorders
(i.e., the observed relative rises exceeded 1.50 but were not statistically significant).
Unfortunately, owing to potentially overwhelming confounding (e.g., related to age, gender,
ethnicity, or iodine intake) or because of small numbers of cases in the comparison counties, little
else can be concluded from this study.
The Crump et al. (2000) study of school children (mean age 7.3 years) in three Chilean
cities permitted comparisons on effects of drinking water with widely varying perchlorate
content: 0, 5, and 100 ppb. A total of 162 school-age children were studied, 127 of whom had
lifelong residence in their respective cities. Controlling for age, gender, and urinary iodine,
a highly significant trend of increasing T4 levels—the opposite to the expected direction for
effects on T4 from perchlorate—was observed with increasing perchlorate content in the water.
The city with the highest concentrations (100 ppb) had a significant five-fold excess in family
history of thyroid-related problems. Children in all three cities had elevated goiter prevalence,
but it was highest in the city with intermediate concentrations (5 ppb) which was believed to also
have iodine deficiencies. A variable introduction of iodized salt in earlier years may have
affected these observations. It is not known what role boiling drinking water may have played or
how the microbiological quality of drinking water varied across the cities studied. Ethnic and
socioeconomic attributes were thought to be similar across the three groups of children but were
not controlled for in the analysis. Whether ambient indoor and outdoor temperatures may have
played a role in thyroid functional status was not investigated. It would appear that uncontrolled
confounding effects, particularly from environmental or other factors, make it difficult to
interpret the observed effects of drinking water contaminated with perchlorate at levels as low as
5 ppb on thyroid function in this study. Controlling for urinary iodine in the analyses would
January 16, 2002
4-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
better address whether iodine deficiency differences across the three cities studied may have
distorted the association of T4 changes with perchlorate exposure. The paradoxical trend
observed in this study remains unexplained.
Crump et al. (2000) also studied newborns screened for hypothyroidism by a heel-stick
blood sample between February 1996 and January 1999 in the same three Chilean cities.
A systematic laboratory error gamma counter contamination occurred between December 1, 1997
and June 30, 1998 which caused TSH to be reported very low (0.1 //U/mL) for a high proportion
(29.1%) of the blood samples analyzed. The error was reported to be limited to this 7-month
period and to have affected a similar proportion of samples from each of the three cities. All data
obtained during the 7-month period in question were excluded, leaving 9,784 neonatal records
for analysis. Analysis revealed a statistically significant decline in TSH (log-transformed) with
increasing city-perchlorate levels, a trend opposite to that hypothesized. The analysis was
adjusted for gender and age at screening as categorical variables in days but covariates lacking
included iodine intake (known to be low in one city), ethnicity, and birth weight. The ages at
screening differed across the three cities studied; the median ages were 3,4, and 6 for the
unexposed, low, and high perchlorate studies, respectively. Other important environmental
factors may have played a role such as ambient temperatures, caloric intake, and social class.
This paradoxical finding parallels the similar result in school age children in the same Chilean
population discussed above, and remains unexplained.
Lamm et al. (1999) examined rates of congenital hypothyroidism in 7 counties of California
and Nevada with perchlorate contaminated drinking water. This outcome is defined as a result of
a mandatory screening program at birth that involves a preliminary T4 determination followed by
a TSH assay in a prescribed subset with low T4. Age at screen is not considered in this
procedure for selecting candidates for TSH determination and screening age distributions by
county were not reported. County-specific levels of perchlorate contamination were unavailable.
Rates for the California births were adjusted only for Hispanic ethnicity, observed to be a risk
factor in this and other studies (Brechner et al., 2000; Schwartz, 2001). The county rate ratios for
congenital hypothyroidism ranged from 0.6 to 1.1 relative to the statewide expected rates and
were not statistically significant for all exposed counties combined, the rate ratio was 1.03 (95%
CI = 0.90 to 1.16). Expected rates based on the non-exposed counties of the two states were not
used. Had only non-exposed counties been used for comparison (given that the exposed counties
January 16, 2002
4-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
comprise a substantial fraction but assuming it is less than half of the state's population) the
resulting rate ratios for the exposed counties would have been 1% or higher. Most critically
lacking in the analysis was classification on age at time of blood sample for the screening test.
Birth weight and further detail on ethnicity and other risk factors were also unavailable.
Therefore, it is likely that uncontrolled confounding has played a role in this study, making it
difficult to interpret and allows for some role of perchlorate in the almost two-fold observed
variation in risk of neonatal hypothyroidism across counties.
Li et al. (2000a) compared the mean monthly T4 levels derived from mandatory screening
of all newborns in Las Vegas (exposed) and Reno (unexposed), controlling for birth weight
(within the restricted range 2.5-4.5 kg) and for age at sample (days 1, 2 or 3 versus 4), for the
period April 1998 through June 1999. Statistical differences in the mean birth weight and mean
age at time of sample were noted for the Las Vegas (n = 17,308) and Reno (n = 5,882) newborns.
The exposure variable was based on monthly measurements made on Las Vegas finished water
by the Southern Nevada Water Authority using IC with a detection limit of 4 ppb. The source of
the Las Vegas water supply, Lake Mead, is known to have thermal stratification that causes
seasonal variation in drinking water perchlorate content. The water supply in Reno comes from
the mountains via Lake Tahoe, the Truckee River, and local wells. Tests of these water sources
for Reno were reported to detect no perchlorate (data not shown nor was it specified if these
measurements were made monthly). A highly significant period or seasonal effect was observed
for both cities (perhaps suggesting an ambient temperature effect), but no difference was
observed between cities during the period of exposure (7 out of the 15 months of observation
when perchlorate content was high in Las Vegas drinking water). Highly significant age effects
were observed, but the dependence of these age effects on exposure (i.e., an exposure interaction)
was not examined. For reasons that are obscure, T4 levels reported in this study were
considerably higher than those reported by others (17 versus 7-10 /ig/dL). The restriction on
birth weight would be inappropriate if birth weight were an intervening variable (i.e., itself
affected by thyroid changes resulting from perchlorate exposure). Regressions on first trimester
and 9-month cumulative exposures using monthly perchlorate levels and grouping birth
outcomes by month in Las Vegas and Reno revealed no trends for T4 differences between the
two cities although more powerful analyses could have been performed using individual
observations. This study suggests that clinically significant individual neonatal T4 differences
January 16, 2002
4-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
have not resulted from current perchlorate exposures although the possibility of important
variation with exposure conditional on neonatal age was not examined.
In a parallel study design, Li et al. (2000b) studied TSH levels in Las Vegas and Reno
newborns over an eleven-month period from December 1998 to October 1999. Las Vegas water
had measurable perchlorate levels in 8 of the 11 months. The perchlorate exposure measures and
assumptions were the same as in Li et al. (2000a). TSH levels were determined on screening
samples that were below the 10th percentile on T4 in each daily batch of samples collected
throughout the state, selected without regard to age at screening. TSH levels from the two cities
for birth weights restricted to 2.5-4.5 kg were compared adjusting for gender and age at screen
(days 2-7 versus 8-30). Births whose screening sample was taken on the first day were excluded
because those TSH levels were considered unstable. The study did not report whether the age at
screen distribution differed between the two cities. Ethnicity and other risk factors were not
available. Using a log-transformed TSH level to facilitate statistical testing, they found no
difference in TSH levels between the two populations (a very small negative effect was estimated
for TSH with exposure), however, the log transformation may have suppressed important
differences at the high end of the TSH distribution and the analysis was not restricted to the
8 months when exposure differed between the two cities. Examination of an exposure with age
interaction was not reported. Excluding births screened on the first day may have further
obscured differences arising from perchlorate exposure, differences that pertain to thyroid
responsiveness. This study suggests that TSH levels in newborns after the first day did not differ
substantially between two cities with and without perchlorate contamination of drinking water as
estimated by monthly measurements.
Brechner et al. (2000) studied TSH levels in Arizona newborns assayed over a three-year
period between October 1994 and the end of December 1997 in the Arizona Newborn Screening
Program. In this program, total T4 is assayed in daily batches of specimens received from all
over the state. TSH is measured in selected samples, representing approximately 10% of the
samples with the lowest T4 levels from each batch. TSH levels were compared between two
cites, Flagstaff and Yuma, representing areas of nonexposure and exposure to perchlorate. Zip
codes were used to determine that Yuma was the only area with essentially all of its drinking
water supplied by the Colorado River below Lake Mead. Exposure data were not available for
the period between 1994 and 1997. Measurements made by the U.S. Environmental Protection
January 16, 2002
4-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Agency Region 9 laboratory in August 1999 reported perchlorate levels at 6 ppb in both raw and
finished water for Yuma and not detectable in Flagstaff water. Because the water processing
facilities have not changed in either city and perchlorate is known to persist for long periods,
Brechner and colleagues presumed that comparable differences between the perchlorate levels in
the two cities existed during the period of analysis. Controlling for age at screen (days 0, 1 -4,
5+) and Hispanic ethnicity, these investigators found a statistically significant elevation in TSH
for the exposed population in Yuma (crude TSH: 19.9 versus 13.4 mU/L; adjusted TSH effect
not reported). However, the age-at-screening distributions differed considerably between these
two cities presenting a possibility for some residual confounding on age. In Yuma (exposed)
5.9% of newborns were screened in the first 24 hours when TSH levels peak (mean TSH =
30 mU/L), compared with 2.4% of Flagstaff newborns (mean TSH = 23 mU/L). Thirty-one
percent of Yuma births were screened at day 6 compared with 46% of Flagstaff births.
Additionally, because of this negative association between age and exposure, the analysis of
variance procedure employed had the potential for bias arising from colinearity. The age and
exposure effect estimates would be jointly affected: overestimating exposure and
underestimating age effects, or visa versa. Other factors not controlled included gender and birth
weight. This study offers positive support for an association of increased neonatal TSH with
perchlorate exposures; however, similar to other studies on this question, it has some unresolved
methodological issues, most notably the strong association between age at screen and perchlorate
exposure.
There is a subtle form of bias in the Brechner and other studies where TSH was determined
on a low - T4 percentile subset of all births that mixes on a daily basis ages at screen for samples
from all over the state. Bloods with low T4 are selected, but the T4 distribution depends on age.
Births with screen ages that usually have higher T4 (typically after 24 hr) are less likely to be
selected for TSH determination; conversely, at ages under 24 hr, births are more likely to be
selected. Both summary and age-specific TSH comparisons would be unbiased with respect to
exposure effects only if the same age at screen distributions were obtained in both the exposed
and unexposed populations. The effect of this bias on estimation of overall perchlorate exposure
effects is difficult to predict, depending in part on how perchlorate exposure affects T4 as well as
on its effects on TSH, and on how sampling age varied with exposure status. It is conceivable
that this bias could explain some of the elevated TSH in perchlorate-exposed neonates of the
January 16, 2002
4-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Brechner et al. (2000) study, but the same sampling bias was potentially present in the Li et al.
(2000b) study that found no effect. The latter study, however, excluded neonatal blood samples
taken during the first 24 hours. That is the period when the strongest perchlorate-related
differences were observed in the Brechner et al. (2000) study.
Schwartz (2001) analyzed both T4 and TSH levels for all California newborns screened in
1996, making use of detailed covariate information on age, birth weight, ethnicity and birth
multiplicity. Perchlorate exposure was assigned using the mothers' postal zip codes that were
linked to state water testing data on all drinking water sources. These estimates of perchlorate
levels were ultimately collapsed into four exposure categories: 0, 1-2, 3-12, 13+ ppb. This level
of exposure detail far exceeded that of any other studies, very likely resulting in the least
exposure misclassifications.
An analysis of covariance (ANCOVA) model was used in this analysis. The ANCOVA
model is a multiple linear regression model that can simultaneously estimate effects for levels
categorical variables like gender as well as for continuous variables like age or birth weight.
Controlling for age at screening (6-hour increments up to 48 hours), gender, single versus
multiple birth, birth weight (in 5 levels), and ethnicity (20 categories), a highly statistically
significant declining trend was observed for T4 with the four perchlorate exposure levels (0,
-9.7, -11.2, - 18.2). T4 levels in this model declined with age (relative to its final level after
48 hours) until about 18 hrs (-50 mg/dL below final level) and then increased over the next 30
hours (to 36 mg/dL above final level) before assuming its final level after 48 hours. For TSH
(log-transformed), there was a significant increasing trend with perchlorate exposure (0, 0.029,
0.03, 0.128), and the TSH level followed a more rapid time course increasing immediately after
birth, then declining to a final level by 24 hours. Substantial birth weight, gender, ethnicity and
birth multiplicity effects were observed for T4, and smaller effects were observed for TSH.
The models specified in this study tested for uniform additive exposure effects for T4 and
TSH across all covariate categories, including baseline shifts. Another issue of considerable
physiological interest would have been whether the amplitudes of the T4 and TSH surges
depended on perchlorate exposure with baseline levels relatively unaffected, which could be
tested by evaluating an interaction between age and exposure. An examination of interaction was
not reported. The bias in TSH measurements introduced by the T4-triggered sample selection
described above for other effects studies would also affect the Schwartz study. This bias would
January 16, 2002
4-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
not affect inferences on exposure effects if the age at screen distribution were similar across the
four exposure levels. These distributions were not reported in the Schwartz study.
The Schwartz study also modeled the effect of two screening performance criteria on the
same set of predictors: (a) "presumptive positive criterion" and (b) a positive finding of
congenital hypothyroidism. Not surprisingly, these models did not predict the standard screening
outcomes well because the screening algorithm does not take into account the several very
important predictors identified in this study. Rather, finding a presumptive positive is based
entirely on T4 without regard to age at screen, birth weight, etc. Similarly, identifying a case of
congenital hypothyroidism is based only on T4-triggered sample selection and subsequent TSH
determination (>25 /iU/ml).
The Schwartz study is by far the most convincing of the neonatal studies, being based on
the most elaborate exposure assignment and the most detailed collection of covariate information
pertaining to neonatal thyroid function. It is unlikely that bias arising from the TSH sampling
could produce such a consistent TSH exposure response and would play no role in the stronger
(based on narrower confidence intervals for the parameter estimates) exposure response observed
for T4.
4.1.2 Occupational Studies
There are two publications investigating workers in ammonium perchlorate production
(Gibbs et al., 1998; Lamm et al., 1999). The route of exposure for each was by inhalation to
perchlorate dust, introducing a considerable uncertainty in dose-response analysis especially due
to poor characterization of particle size distribution. Both studies were also cross-sectional in
design and, therefore, subject to survivor bias in that workers experiencing adverse effects could
have left employment. This issue was not addressed in either study. It would have been
particularly noteworthy had any former employee no longer in the cohort experienced thyroid
disorders, aplastic anemia, or related hematological disorders, each of which have been reported
in settings where perchlorate is used for short periods at higher doses in the treatment of disease
(Lawrence et al., 1999). The airborne exposures that were characterized corresponded to daily
doses on the order of 20 to 50 mg and possibly higher as the air-sampling methods excluded
large particulate (> 50 //m) that could add considerable mass to the daily inhaled or ingested
January 16, 2002
4-13
DRAFT-DO NOT QUOTE OR CITE
-------
]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
dose. In the study that investigated this (Lamm et al., 1999), the daily absorbed dose based on
urinary perchlorate actually exceeded the inhaled dose.
There was no clear evidence for any perchlorate effect on thyroid function, as defined by
the investigators, in these two cross-sectional occupational studies. However, historical exposure
classification was limited in one study and absent in the other. Former employees were lost to
follow-up, and neither study controlled for potential confounding arising from body mass,
environmental temperatures, or socioeconomic status. There was no measurement of thyroid
iodine status or of any index thyroid dynamic responsiveness that conceivably could be altered
even though steady-state TSH and T4 levels appear to be in the normal range. Because of the
cross-sectional design and measured endpoints, the studies did not evaluate the dynamics of
hypothalamic-pituitary-thyroid feedback that are likely important in target populations such as
hypothyroxinemic pregnant women and their fetuses.
Gibbs et al. (1998) performed a case control occupational epidemiology study to evaluate
thyroid function and standard clinical blood test parameters of liver, kidney, and bone marrow
function in employees exposed to ammonium perchlorate airborne dust at a production facility
and an associated cross-blending facility. Exposure estimates were based on multiple samples
(average sample number = 17) for eight homogenous exposure groups defined by similar job
activities: control, maintenance/foreman, and six discrete operator job categories. Personal
breathing zone samples (n = 119) were used for the work categories and full-shift area samples
were used for the control group (n = 19). The control exposure was not zero but was several
orders of magnitude below any exposure category. In their 1997 analyses, when ammonium was
quantified using National Institute for Occupational Safety and Health Method 6016 which had a
minimum reporting limit of 0.017 mg/m3, concentrations in a large number of the samples were
reported as undetectable. The 1998 analyses were performed using the modified EPA 300.0
methodology that determines perchlorate using ion chromatography and has a reporting limit of
approximately 0.00004 mg/m3.
Effects were examined in either a single-shift design (pre- and post-shift parameter
measurements) or working lifetime design based on medical surveillance data that included
thyroid examination since 1996 (blood tests, physical exam, and history since 1994). Dose was
reconstructed based on personnel records for job type and area samples.
January 16, 2002
4-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Despite the lack of particle size distribution data, an inhaled "dose" was calculated for a
single shift as (Gibbs et al., 1998):
f
\
respiratory
rate
x
inhalation
concentration
v y
c
\
exposure
duration
v y
x
\
fraction
absorbed
V /
(4-1)
Working lifetime exposure estimates were calculated as:
^ (mean group exposure) x (years in exposure group) x 2,000, (4-2)
where the 2,000 was an average of the number of hours worked yearly based on typical overtime
rates at the facilities.
Daily respiratory rates of 0.0165 m3/kg-hr and 0.0068 m3/kg-hr were estimated for "active"
and "sedentary" workers, respectively, based on Beals et al. (1996). These estimates are slightly
lower than the default EPA respiratory rates and are moderately lower than those recommended
by the International Commission on Radiological Protection in its recent human respiratory tract
model (International Commission on Radiological Protection, 1994). Average body weights of
the workers were larger than the typical default body weights. Because current practice usually
scales ventilation rate based on body weight, higher ventilation rates were expected.
The absence of particle size diameter and distribution data is a significant limitation of the
study because this data is required to assess the potential inhalability of the ammonium
perchlorate aerosol. Data from another production facility indicate the majority of particles are
200 (xm (Hancock, 1998). Particles larger than 30 jum are typically not inhalable by humans
(U.S. Environmental Protection Agency, 1996b). Further, there was no mention of face volume
performance of the personal samplers using 5-//m filters although this is an important
consideration in dusty environments when the particles under investigation have a large diameter.
This consideration is especially important here because the filter cassettes were changed when
respirators were used. Even if a 5-//m particle diameter could be assumed, the inhaled "dose"
calculation should have included an adjustment for inhalability and deposition efficiency to
January 16, 2002
4-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
calculate the deposition fraction, approximately 0.3 at 5 (U.S. Environmental Protection
Agency, 1996b).
The assumption about the solubility of the inhaled particles is also problematic because this
parameter is particle-diameter dependent. The particle diameter dictates the location
(extrathoracic, tracheobronchial, pulmonary) where the particle deposits and the local milieu and
clearance vary with location also influence solubility (U.S. Environmental Protection Agency,
1996b; Snipes et al., 1997). The solubility of cesium chloride (CsCl) in beagles was used to
estimate a fraction absorbed. Although CsCl and NH4C104 may have similar solubilities,
additional uncertainty is introduced because the CsCl particle diameter or inhalability function
for the beagles was not taken into account; and the hydroscopicity, which influences the initial
deposition site, may not be the same. The assumptions about dose could have been validated
with a mass balance approach. For example, perchlorate could have been measured in the blood
when samples were taken for thyroid hormone analyses. Additionally, urine samples could have
been monitored for perchlorate because it is excreted in the urine. These additional
measurements would have afforded some confidence that the inhaled dose estimates were
reasonable.
Standard clinical thyroid profiles included a total serum T4, triiodothyronine resin uptake,
and TSH. Bone marrow function was evaluated during medical surveillance examinations with
standard tests from the complete blood count which included hemoglobin, hematocrit, red blood
cell count, mean corpuscular volume, white blood cell count, and platelet count. Standard serum
chemistries were used to assess kidney (serum creatinine level and blood urea nitrogen) and liver
(serum glutamyl pyruvic transaminase [SGPT], serum glutamyl oxaloacetic transaminase
[SGOT], g-glutamyl transpeptidase [GGTP], and alkaline phosphatase) functions.
Dependent variables for the single-shift study were the cross-shift change in measures of
thyroid function. Explanatory variables included race, gender, age, hours awake prior to the
pre-shift test, number of hours slept during the most recent period prior to the test, time of day,
and shift length. Dependent variables for the working lifetime included measures of thyroid,
bone marrow, liver, and kidney functions. For the thyroid tests, an additional explanatory
variable was used to indicate whether the measurement was from a routine physical in 1996 or
from a pre-shift or a post-shift examination in 1997 or 1998. The dose variables were group
(control, low dose, or high dose) and estimated cumulative exposure. The dose group
January 16, 2002
4-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
designation was an arbitrary stratification of <8 mg/kg-day and >8 mg/kg-day. Multiple
regression was used to analyze the relationship between effect measures and explanatory
variables. A sequential approach was used to determine whether a dependent variable should be
log-transformed and whether any outliers (defined as a value corresponding to a residual larger in
absolute value than three standard deviations) should be eliminated from an analysis.
Estimated doses for the single shift-study ranged from 0.0002 to 0.436 mg/kg-day with a
mean of 0.036 mg/kg-day and median of 0.013 mg/kg-day. The dose estimate was not a
significant predictor of thyroid function parameters measured in 83 control (65 male, 18 female)
or 18 exposed (15 male, 3 female) individuals. Working lifetime exposure estimates ranged from
0.5 to 7.0 (mean 3.5) mg/kg for the low-dose group and from 8.0 to 88.0 (mean 38.0) mg/kg for
the high-dose group. The duration of exposure ranged from 1 to 27 years (mean 8.3).
No significant correlations were detected in any measures of thyroid, bone marrow, liver, or
kidney function; however, significant gender and race differences were apparent in the clinical
tests of bone marrow, liver, and kidney functions. Females were slightly lower in hemoglobin,
hematocrit, SGPT, GGTP, and creatinine than males; black workers were slightly lower than
whites in hemoglobin and hematocrit and slightly higher in creatinine.
The only significant finding (p = 0.01) was that cross-shift TSH changes were greater for
those who worked a 12-h shift than for those who worked 8-h shifts, accounting for a
0.45 urinary IU/mL increase across the shift. This was attributed to the influence of circadian
changes in serum TSH. However, the TSH increase (10%) across a single work shift in an
exposed group (n = 18) compared to an unexposed group (n=83) was observed in groups that
together comprised less than half of employees eligible for study. Comparison of workers in
three groups (unexposed, low and high cumulative exposure) resulted in consistent patterns for
all thyroid parameters in which the unexposed group had values intermediate between those of
the low and the high cumulative dose groups. This suggests that important confounding was
present (i.e., that the comparison group, which apparently included office workers, differed from
the exposed groups on other important risk factors) as well. For thyroid (TSH) and liver
outcomes (SGOT, GGPT, SGPT), there were subtle indications of exposure effects: the standard
deviation increased substantially in the high dose group, as did the average values but not the
percentiles up to the 75th, suggesting that a small subgroup had undergone a considerable upward
January 16, 2002
4-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
excursion. Statistical tests (regression analysis) of these effects were severely limited by the
apparent confounding that affected baseline levels.
In the second study of ammonium perchlorate workers, Lamm et al. (1999) assembled a
comparison group at the same facility from an unrelated process thought to have low exposure to
inhaled perchlorate. Workers were classified using presumptive exposure based on visible dust
generated. Pre- and post-shift urine samples were collected to measure urinary perchlorate,
iodine and creatinine levels. Post-shift blood samples were analyzed for complete blood count
(CBC), hemoglobin, hematocrit and additional red cell parameters (mean corpuscular volume,
mean corpuscular hemoglobin, and mean corpuscular concentration). A clinical chemistry panel
was also run on post-shift serum samples. Thyroid parameters included TSH, free T4, T4, T3,
thyroid hormone binding ratio, thyroid peroxidase antibodies, and clinical examination. Urinary
perchlorate measurements were used to calculate a post-shift level of perchlorate (mg) per g of
creatinine as an excretion dose, D:
D = k[Ei -0.354 E0]/0.646. (4-3)
The right hand term in brackets is the post-shift adjusted level in mg perchlorate per gram of
creatinine. Perchlorate absorption was calculated as a time-weighted average exposure using an
assumption that the percent absorbed which is excreted is 95%. The human adult creatinine
excretion rate was then used to link perchlorate excretion rate in terms of creatinine to rates in
terms of time, so that the exposure dose was then calculated as:
12 hours x 60 minutes / hour x 0.001 g/mg x 1 mg creatinine/min x [post-shift]/0.646. (4-4)
While particle size distribution data were collected, this information was not utilized in the
analyses. Inhalation exposure was instead categorized into either "total" or "respirable". While
these categories correlated with each other to a good degree (r = 0.82), perchlorate absorption
(mg/shift) did not correlate as well to respirable ( r = 0.45) as it did to total particles (r = 0.54).
The comparison group had current absorbed doses equal to 20% of the low perchlorate-exposed
group and 3% of the high exposed group even though the inhaled dose of the comparison group
was 4% of that of the low dose and 0.02% of the high dose group. This suggests that there was
January 16, 2002
4-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
considerable exposure misclassification, arising perhaps from general environmental
contamination at the work site or in clothing. In one subject, urinary perchlorate increased over a
12 hr period during which there was thought to be no exposure. No significant associations were
observed between perchlorate exposure and thyroid parameters; however, measures of
cumulative exposure were not considered. Suggestions of increasing trends for T3, T4, and
maximum-T3 were not statistically significant but were based on small numbers (numbers of
workers in exposure groups: 21 for the unexposed versus 14, 8, and 15 in the low, medium, and
high exposure groups).
4.2 CLINICAL STUDIES
The historical clinical data on perchlorate have been predominantly case reports of patients
whose results would be confounded either with thyroid disease or other pharmaceutical agents.
A few more recent studies have begun to evaluate thyroid function in healthy volunteers. This
section will discuss the available data on thyroid function from several clinical studies. A more
formal development of the pharmacokinetic data in humans is presented in Chapter 6.
4.2.1 Studies in Healthy Human Subjects
Few data are available to demonstrate the effects of perchlorate on healthy individuals and
issues of ethics are likely to preclude clinical evaluation in sensitive populations such as pregnant
women. Exposure duration to perchlorate is typically short, from a few days to 4 weeks.
Burgi et al. (1974) examined the effects of perchlorate on the secretion of endogenous
iodine by the normal human thyroid gland. Five healthy volunteers (3 males, 2 females;
ages 24 to 27 years) received tracers of 125I-iodide and l31I-thyroxine for 17 days, followed by
600 mg/day perchlorate (9.7 mg/kg-day, based on actual reported average body weight of
61.8 kg) for 8 days. Urine and serum were analyzed for 125I and 13II to determine if perchlorate
can cause the discharge of endogenous, as well as exogenous iodide, from the thyroid. Results
show that this dose of perchlorate also was sufficient to completely block iodide uptake by the
thyroid. In addition, perchlorate caused a 65% increase in excretion of nonthyroxine iodide over
background. The authors attributed this increase to additional secretion of endogenous iodide
January 16, 2002
4-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
from the thyroid. Treatment with carbimazole plus perchlorate caused an additional increase in
the secretion of nonthyroxine iodide, suggesting that perchlorate causes only a partial release of
endogenous iodide. This study suggests a Lowest-Observed-Adverse-Effect-Level (LOAEL) of
9.7 mg/kg-day for thyroid effects in healthy patients.
Brabant et al. (1992) administered potassium perchlorate to five healthy male volunteers
(age 25 to 28 years) to study changes in TSH concentration and release in response to a decrease
in iodine supply to the thyroid. During the first 4 weeks of the study, the volunteers were given
200 /Ug/day iodine. After iodine supplementation was discontinued, the volunteers were
administered 900 mg/day of potassium perchlorate orally for 4 weeks to induce a state of iodine
depletion. At the end of the 4-week perchlorate treatment, levels of thyroid hormones were
measured. Although perchlorate treatment had no effect on thyroid volume or levels of
triiodothyronine (T3) and thyroxine (T4), intrathyroidal iodide concentration and serum levels of
TSH were decreased significantly, and serum levels of thyroglobulin were nearly doubled. The
authors speculate that the decrease of TSH, which is the opposite of the expected response, may
be an early adaptive mechanism to the iodine deficiency induced by perchlorate. They suggest
that, early in iodide deficiency, the thyroid becomes more sensitive to TSH creating a feedback
mechanism that decreases TSH levels. Only as iodide deficiency becomes more prolonged do
TSH levels increase. This study defined a LOAEL of 13 mg/kg-day for thyroid effects. In a
follow-up study, Brabant (1994) repeated the earlier studies with perchlorate treatment lasting
longer than 4 weeks. As a result of the longer treatment, thyroid volumes increased in all
subjects although TSH levels did not increase.
Lawrence et al. (2000) performed a 14-day clinical study with nine euthyroid volunteers
(ages 22 to 30 years). Each subject was enrolled after a normal complete physical exam that
included a thyroid exam. Blood was obtained for baseline measurement of thyroid function tests,
TPO antibodies, CBC, and routine blood chemistries. A spot urine was obtained for routine
urinalysis. All baseline tests were normal.
Ten mg of perchlorate as potassium perchlorate was dissolved in 1-L bottles of spring
water. Each subject was instructed to consume the 1-L bottle intermittently during waking hours.
Assuming a body weight of 70 kg, this dosage is equivalent to 0.14 mg/kg-day. Blood specimens
were drawn between 8:00 and 9:00 a.m and 24-hour urine samples were obtained on days 7 and
14 during exposure and then again after another 14 days after perchlorate was discontinued.
January 16, 2002
4-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Thyroid function was assessed by assays for TSH, free thyroxine index (FTI), total T3, (TT3) and
T4. Blood chemistries and CBC were also measured. Baseline thyroid radioactive iodine uptake
(RAIU) was measured using 123I at 4, 8, and 24 hours after the ingestion of 150 /^Ci 1231.
As reported by the authors, statistical analysis for the thyroid RAIU was carried out by
analysis of variance (ANOVA) with post hoc pairwise comparisons using Tukey's method. The
outcome measure variable was log-transformed to achieve greater homoscedasticity and a more
Gaussian distribution. Serial analyses were done: a three-factor ANOVA with factors as patient,
treatment, and time and a set of two-factor ANOVAs, one for each of the three times. The
analogous mixed-model ANOVAs were also run with subject as a random effect to confirm that
repeated measures among the subjects did not affect the results. Statistical analyses of the
thyroid function tests and urine and serum perchlorate and iodine values were carried out by
ANOVA and Student Newman Keuls (SNK).
Urine and serum perchlorate levels at baseline and during and after ingestion of the daily
10 mg perchlorate dose are presented in Table 4-2. Perchlorate levels returned to baseline after
the two week recovery period. There was also no significant changes in urinary iodine excretion
during, or 2 weeks after stopping the perchlorate administration as shown in Table 4-3. The
authors note that the iodide ingestion of the volunteers was not controlled in the diet and were
variable. It may also be worthwhile to note that the urinary iodine values are relatively high (see
Chapter 3), indicating a potential protective status in these subjects for the inhibition of the NIS
by perchlorate.
TABLE 4-2. URINE AND SERUM PERCHLORATE (C104) VALUES BEFORE,
DURING, AND AFTER THE INGESTION OF 10 mg OF C104 DAILY FOR 14 DAYS
(Lawrence et al., 2000)
Time Urine Perchlorate3 (mg/24 hr) Serum Perchlorate3 (//g/mL)
Baseline < 0.5 0
7 Days C104 7.7 ± 0.8a 0.61 ± 0.02
14 Days C104" 7.5 ± 1.0 0.59 ± 0.02
14 Days After C104" < 0.5 0
°Mean ± SE.
January 16, 2002
4-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
TABLE 4-3. URINE AND SERUM IODINE VALUES BEFORE, DURING, AND
AFTER THE INGESTION OF 10 mg OF C104 DAILY FOR 14 DAYS
(Lawrence et al., 2000)
Time
Urine Iodine3 Oug/24 hr)
Serum Iodine" (//g/dL)
Baseline
254 ± 69
6.5 ±0.42a
7 Days C104~
233 ± 49
6.2 ±0.34
14 Days C104'
385 ±123
6.4 ±0.37
14 Days After C104"
208 ± 42
6.3 ±0.57
"Mean ± SE.
A highly significant decrease in the 123I thyroid RAIU with respect to baseline
measurements at all three time points was noted (Table 4-4), 34%, 39%, and 41% at 4, 8, and
24 hours. The decrease averaged over all three time points was 38%. Two weeks after
perchlorate was discontinued, the thyroid RAIU values were significantly higher at all three time
points (average increase of 25%), indicating a rebound that may represent upregulation of the
N1S. The time course of the iodine inhibition could not be calculated since the subjects drank the
dose ad libitum over the day and there was evidence that the full 10 mg/day dose was not
achieved for at least some subjects because the average daily urinary excretion of perchlorate was
7.6 for the 2-week course of perchlorate administrations. There was a corresponding increase in
urinary iodide excretion during dosing followed by a drop below baseline during rebound. T3
levels were observed to rise throughout the 28-day trial (trend not tested).
In a subsequent study reported as a letter to the editor by these same investigators,
Lawrence et al. (2001) used nine healthy male volunteers and a dose of 3 mg/day (.04 mg/kg-day
assuming 70 kg body weight) and again observed decreased RAIU. The mean 8-hour decrease
from baseline was reported to be at 10% and at 24-hours to be 10.3%. Neither were significant
based on Tukey paired t-test (data not shown). The RAIU after stopping the perchlorate
ingestion for 14 days rebounded as in the first study and was reported to be an increase of 22% at
8 hours and 18% at 24 hours (p < 0.02). It is worthwhile to note when evaluating these results
that these data (Lawrence et al., 2000; 2001) were evaluated for use in the physiologially-based
pharmacokinetic (PBPK) models described in Chapter 6, but the data were excluded due to the
January 16, 2002
4-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
TABLE 4-4. THYROID 123I UPTAKES BEFORE, DURING, AND AFTER THE
INGESTION OF 10 mg C104 DAILY FOR 14 DAYS (Lawrence et al., 2000)
Time
Thyroid
Baseline
,23I Uptake3 (% Dose)
14 days on C104"
14 days after C104"
4 Hours
12.5 ± 1.3
8.2 ±0.7"
16.6 ± 2.4°
8 Hours
17.3 ± 1.9
10.6 ± 1.0"
21.9 ± 2.8°
24 Hours
23.6 ±2.6
14.0 ± 1.6b
27.1 ± 3.3d
"mean ± S.E.
bp < 0.01 vs. baseline and after C104".
c/j < 0.01 vs. baseline.
Ap < 0.05 vs. baseline.
lack of availability of all records to the QA/QC process and unresolved issues regarding sample
sequences. Variability of serum and urine perchlorate results, potentially due to the unstructured
drinking water regimen (Merrill, 2001 a,b) was noted. Serum levels from the 0.04 mg/kg-day
dose group ranged from non-detect to 495 mg/L on days when the subjects were supposed to
have consumed perchlorate. Given this variability and the unknown consequence of a 10%
change in thyroid RAIU of a small sample of healthy euthyroid individuals to potentially
hypothyroid or hypothyroxinemic pregnant women, it would be difficult to designate this effect
as a No-Observed-Adverse-Effect-Level (NOAEL) with any confidence.
Greer et al. (2000) described a third study of RAIU in healthy euthyroid subjects in an
abstract. Perchlorate was dissolved in 400 ml of drinking water at one of three doses to twenty-
four euthyroid volunteers (4 males and 4 non-pregnant females per dose; 18 to 57 years old).
The subjects were instructed to drink 100 ml at 4 set times throughout the day for 14 days.
Measurement of 8- and 24-hour RAIU was performed prior to perchlorate ingestion (baseline),
on exposures days 2 and 14, and on post-exposure Day 15. Expressed as a percentage of baseline
(mean ± S.E.), the abstract reports 24-hour RAIU values for the 0.02, 0.1 and 0.5 mg/kg-day dose
groups as: 83 ± 5.6, 59 ± 3.5 and 31 ± 2.6 on exposure day 2; 85 ± 5.6, 57 ± 4.7, and 34 ± 4.5
on exposure day 14; and 111 ±5.1, 96 ± 12, and 108 ± 12 on post-exposure Day 15. These
correspond to RAIU inhibition values expressed as % of baseline (where indicates inhibition
relative to baseline) for the 0.02, 0.1 and 0.5 mg/kg-day dose groups of -17, -41, and -69 on
January 16, 2002
4-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
exposure Day 2; -15, -43, and -66 on exposure Day 14; and +11, -4, and +8 on post-exposure
Day 15. The authors report no difference between males and females and that a linear log-dose
relationship was observed with the regression slopes indistinguishable between the 8- and
24-hour measurements (data not shown).
In other unpublished data provided in Merrill (2001a; Attachment #7) these same
investigators tested seven euthyroid subjects (six non-pregnant females and one male) at a dose
of 0.007 mg/kg-day. Expressed as a percent of baseline, the average 8- and 24-hour RAIU
inhibition values measured on exposure Day 14 were -6.2 and -1.8%. The inhibition values
ranged from -38.6% to +27.9% of baseline at the 8-hour time point and -26.7 to +39% of
baseline at the 24-hour time point. The range for the post-exposure Day 14 RAIU inhibition
values was - 19.3 to +45% of baseline. No measurements were made on Day 2 when the RAIU
inhibition would have been greater. There was no RAIU inhibition measured on post-exposure
Day 15. In the Greer et al. (2000) abstract, the authors estimate the no effect level at
0.007 mg/kg-day.
In order to evaluate whether the 0.007 mg/kg-day dose had a sufficient sample size to
detect a difference of the observed magnitude as in the other doses tested, the EPA calculated the
power of the usual t-test for the 14-day exposure data. A log transform of the ratio of the
individual values at Day 14 to their baseline values was based on the non-central t distribution.
The power at the 0.007 mg/kg-day dose was low (0.1) compared to the other doses: 0.95, 0.998,
and 0.999 at 0.02, 0.1, and 0.5 mg/kg-day.
The EPA has also been made aware of another human clinical study being performed at
Loma Linda and funded by Lockheed Martin (Beck, 2001). The study is not yet completed
because the objective sample size for each dose group has not yet been attained. Human
euthyroid volunteers (male and non-pregnant females) have been dosed with perchlorate in gel
caps at 0.007, 0.014, and 09.04 mg/kg-day. Measurements were made at baseline, 3-months,
6-months, and after recovery from exposure for RAIU, T3, T4, and TSH levels. These dosages
are the same as already tested so the added value to the human database, especially with respect
to the now prominent concern for neurodevelopmental effects secondary to hypothyroxinemia or
even transient decrements in T4, is not readily apparent. The additional data may potentially
reduce the variability and low power due to the small sample sizes of the previous studies if
sufficiently comparable in design.
January 16, 2002
4-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
4.2.2 Studies in Patients with Graves' Disease
Potassium perchlorate had been used to treat Graves' disease in humans; consequently,
most of the prior data on perchlorate effects on humans are in patients with this disease. Graves'
disease is an autoimmune disorder which causes patients to carry immunoglobulins in their blood
that bind to TSH receptors on thyroid cells and act like TSH to stimulate DNA synthesis and cell
divisions, leading to a hyperthyroid state. Symptoms of the disease include increased synthesis
and secretion of iodide-containing hormones into the blood by the thyroid gland, thyroid gland
enlargement, increased basal metabolism, and weight loss. Perchlorate inhibits the excessive
synthesis and secretion of thyroid hormones by inhibiting the uptake of iodide into the thyroid
and causes an efflux (discharge) of accumulated iodide in the gland.
Stanbury and Wyngaarden (1952) evaluated therapeutic perchlorate use in patients (n = 8,
although reporting of exact numbers for various aspects [e.g., different dose levels] of the study
is sketchy) with Graves' disease and found that perchlorate caused the discharge of iodide
accumulated in the thyroid and blocked the uptake of iodide into the thyroid. Within 30 min of
administration, a single dose of 100 mg potassium perchlorate caused the nearly complete release
(-80%) of 13II from the thyroids of Graves' disease patients previously treated with tracer
amounts of 131I and 1 -methyl-2-mercaptoimidazole (MMIA). MMIA was given to cause
accumulation of l31I in the thyroid because MMIA prevents the oxidation of iodide ion to iodine
and its attachment to tyrosyl groups (see Chapter 3). A single dose of 10 mg perchlorate
appeared to cause a -50% release of accumulated iodine. The authors reported that perchlorate
doses as low as 3 mg caused detectable, but incomplete, release of iodide from the thyroid
(although quantitative data for doses less than 10 mg were not presented). In addition, Stanbury
and Wyngaarden (1952) reported that the uptake of tracer levels of 131I into the thyroid glands of
two patients with Graves' disease was markedly inhibited for as long as 6 hr when 100 mg of
potassium perchlorate was given orally 1 h prior to administration of the tracer. Beyond 6 h,
uptake of13'I recommenced. Inhibition of iodide uptake also occurred in three patients without
MMIA treatment. The authors stated that no toxic effects were encountered in any patients who
were given, in more than three doses, a total not exceeding 600 mg potassium perchlorate. This
January 16, 2002
4-25
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
study was used to identify a LOAEL of 1.4 mg/kg-day1 for complete release of iodine from the
thyroid for the RfD reviewed in March 1997 (Toxicology Excellence for Risk Assessment,
1997). Because it was not clear what degree of iodide efflux constitutes an adverse effect, a
NOAEL was not designated for this study. An expert peer review panel later determined this
study was inadequate for RfD derivation (Toxicology Excellence for Risk Assessment, 1998b).
Godley and Standbury (1954) report using potassium perchlorate to treat 24 patients with
Graves' disease. Patients were treated with 600 to 1,200 mg/day (typically 200 mg every 8 h)
for at least 11 weeks with a few patients treated as long as 45 to 52 weeks. A decrease in iodide
uptake was observed. Five patients became euthyroid after continuous administration for
28 weeks. Two patients developed gastrointestinal problems that were assumed to result from
perchlorate treatment. In one of these patients, these effects occurred at 600 mg/day, but the dose
that the other patient received is not specified. Other side effects of antithyroid agents, such
hematological changes, liver damage, and skin rash, were not observed. This study suggested a
LOAEL of 9 mg/kg-day in humans for short-term exposures.
Crooks and Wayne (1960) observed one case of skin rash and three cases of nausea (12%)
among 35 patients treated with 600 mg/day (9 mg/kg-day) and 165 patients given 1,000 mg/day
(14 mg/kg-day). All patients had diffuse goiters and exophthalmos, classic signs of Graves'
disease. In another group of 10 patients given 1,500 mg/day (21 mg/kg-day) and 40 patients
given 2,000 mg/day (29 mg/kg-day), five cases of skin rash, two cases of nausea, and one case of
agranulocytosis occurred (16%). Leukocyte counts returned to normal in the patient with the
agranulocytosis when perchlorate treatment was stopped. The length of treatment was unclear
but generally appears to have been less than 8 weeks although it appears that one patient was
monitored for 22 weeks. The authors report that the "time to cure" Graves' disease using
perchlorate is approximately 9 weeks. The authors also report that 1 of 12 infants born of
mothers given 600 to 1,000 mg/day was born with a very slightly enlarged thyroid that returned
to normal size in 6 weeks; no other abnormalities were noted. This study suggested a LOAEL
between 9 and 14 mg/kg-day.
'Unless otherwise indicated, for human studies in which the actual body weight of the subjects was not
reported, the dose in milligrams per kilogram per day was calculated assuming a body weight of 70 kg. Thus, a dose
of 100 mg/day 70 kg is 1.4 mg/kg-day.
January 16, 2002
4-26
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Morgans and Trotter (1960) reported that 3% of 180 patients treated with 400 to
1,000 mg/day (6 to 14 mg/kg-day) potassium perchlorate and 18% of 67 patients treated with
1,200 to 2,000 mg/day (17 to 29 mg/kg-day) displayed a variety of adverse reactions that
included skin rash, sore throat, gastrointestinal irritation, and lymphadenopathy. Reactions
occurred within 2 to 3 weeks of drug administration. This study suggested a LOAEL between
6 and 14 mg/kg-day.
Connell (1981) reported a case study of a single 72-year-old female Graves' disease patient
who was treated with 200 mg/day (3 mg/kg-day) potassium perchlorate for 22 years without any
indication of adverse side effects. Thyrotoxicosis recurred 4 weeks after stopping potassium
perchlorate administration, suggesting that this dose level provided sufficient clinical control of
the hyperthyroidism. The study also suggested that the adverse reactions seen at higher doses
may not occur at lower doses, even after long-term treatment.
4.2.2.1 Hematological Effects
Between 1961 and 1966, the occurrence of severe hematological side effects in patients
receiving long-term potassium perchlorate treatment for Graves' disease led to a decreased use of
potassium perchlorate as a therapeutic agent. Several authors (Hobson, 1961; Johnson and
Moore, 1961; Fawcett and Clarke, 1961; Rrevans et al., 1962; Gjemdal, 1963) report case studies
in which a single patient suffered fatal aplastic anemia after treatment doses ranging from 6 to
14 mg/kg-day. The duration of treatment ranged from 3 mo (Johnson and Moore, 1961) to 8 mo
(Hobson, 1961). In all cases, patients were started at the high end of the treatment range for a
period of time and then were reduced to the lower end of the treatment range after the appearance
of side effects. In two cases (Hobson, 1961; Gjemdal, 1963), patients had co-exposures to other
drugs. Other case reports are available that report nonfatal agranulocytosis in patients treated
with 14mg/kg-day for 12 days (Southwell and Randall, 1960) or 3 mo (Sunar, 1963). Barzilai
and Sheinfeld (1966) report that 11% of 76 patients developed leukopenia or other unspecified
side effects after treatment with 1,000 mg/day (14 mg/kg-day) for a little as 2 mo. Within this
group, there was one case of fatal aplastic anemia and one case of fatal agranulocytosis.
These studies suggest that doses in the range of 6 to 14 mg/kg-day may represent a frank
effect level in patients with Graves' disease although there were questions as to whether these
effects were caused by the disease itself, whether there was some contamination, or whether the
January 16, 2002
4-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
effects occurred only at high doses. A review by Wenzel and Lente (1984) concluded that the
"severe adverse reactions, such as agranulocytosis, were likely to occur only when large doses of
more than 1,000 mg potassium perchlorate were administered." There is no information to
suggest that humans without Graves' disease would have a similar reaction to perchlorate.
Antithyroid drugs appear to exert their effects on the hematopoietic system through an
immune mechanism. Wing and Fantus (1987) reviewed the adverse effects of two antithyroid
drugs, propylthiouracil and methimazole, and concluded that most reactions were related to the
immunologic effects of these drugs. They noted that skin rash and granulocytopenia were among
the most commonly reported adverse effects of these drugs. Less commonly reported effects
include aplastic anemia, leukopenia, and antibodies to insulin and glucagon. In fact, Wing and
Fantus (1987) recommend that patients be instructed to report skin rash immediately, as this may
be an early sign of adverse immune reaction caused by the antithyroid drugs. Although these
authors did not include perchlorate in their investigation, the similarity of the effects seen after
perchlorate treatment—including rash, leukopenia, agranulocytosis, and aplastic anemia—
suggest that perchlorate also may act in a similar fashion to induce an immune effect.
There is a tight functional connectivity between the immune and endocrine systems which
is mediated, at least in part, by shared receptors and mediators among the systems (Kammuller,
1995). Thus, although the mechanism of perchlorate action on the hematopoietic system is not
known, it is likely to be an immune reaction. Although it is possible that perchlorate may cause
hematological effects in healthy humans, it appears that Graves' disease patients are likely to be
more sensitive to this type of immune-induced adverse effect than are healthy people. The
increased sensitivity to immunologic function in Graves' disease patients arises because of the
underlying abnormal immunologic function in Graves' disease. Immunoreactivity to antithyroid
drugs is another expression of the compromised immune system in these patients (Wall et al.,
1984; Wing and Fantus, 1987).
4.3 SUMMARY OF CONCLUSIONS REGARDING HUMAN HEALTH
EFFECTS STUDIES
The recent human studies support the established effect of perchlorate at the NIS. Using
these data as the basis for quantitative dose-response assessment is more difficult. Of the five
January 16, 2002
4-28
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
population studies investigating the effects of perchlorate exposures on TSH levels in newborns
(Lamm et al., 1999; Li et al. 2000b; Crump et al., 2000; Brechner et al., 2000; Schwartz, 2001),
the Brechner et al. (2000) study had a somewhat better exposure classification owing to a more
narrow, but still ecological, geographical focus (two small cities) and Schwartz had a relatively
detailed exposure classification down to the level of zip codes. Only these two studies had
positive findings in newborns. The restriction of birth weight in Li et al. (2000b) could have
reduced study sensitivity if thyroid endpoints in non-normal birth weights are especially effected
by perchlorate. The strong dependence of thyroid endpoints on birth weight observed in several
studies raises the possibility that birth weight itself could be an intervening variable in
perchlorate effects. That is, perchlorate exposure may affect birth weight. This would be a
testable hypothesis in several of the studies. If birth weight were an intervening variable, birth
weight restriction in the Li et al. (2000a,b) studies or controlling for birth weight as a confounder
in the Li et al. (2000a,b), Brechner et al. (2000) and Schwartz (20001) studies may have resulted
in an underestimation of perchlorate exposure effects.
In the one study that reported age-specific perchlorate exposure effects on TSH (Brechner
et al., 2000), the largest effect was in the first 24 hours after birth. This observed exposure-age
interaction was not statistically evaluated. The study with the strongest findings (Schwartz,
2001) actually focused only on the first 2 days after birth. Therefore, excluding day-one screened
births as in the Li et al. (2000b) study may severely reduce or eliminate the ability to detect a
perchlorate effect.
The well-known TSH surge at birth is thought to represent a response to temperature
change (Schwartz, 2001). This suggests that ambient temperatures - prenatal and perinatal -
might be important determinants of thyroid endpoints. The strong period/seasonal effect
observed in the Li et al. (2000b) study supports this temperature conjecture and the unexpected
trends across Chilean cities in the Crump et al. (2000) and variations across U.S. counties in the
Lamm et al. (1999) and Schwartz (2001) investigations could also be related to temperature.
It should also be noted that all of the studies in this review examined endpoints that may be
insensitive to the consequences of altered thyroid function. No detailed models of thyroid
dynamic response were postulated with subsequent analysis of relevant endpoints that would
reliably detect the specific perchlorate- or environmentally-induced defects. Nonetheless, one
study examining neonatal thyroid status in the first five days found a perchlorate effect that was
January 16, 2002
4-29
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
greatest in the first 24 hours and that rapidly declined over the next two days, suggesting
alteration of thyroid response to the birth event. The issue of iodine depletion in exposed
populations was not directly evaluated although experimental evidence of short-term depletion in
adults at high doses was observed.
All of the observational field studies utilized "ecological" exposure rather than individual-
specific dose measurements; the relative specificity of the dose metric varied widely from
"exposed/not exposed", to an average concentration in drinking water for a given zip code. The
occupational studies used air sampling to estimate homogeneous exposure groups. Nevertheless,
there was evidence of perchlorate effects on neonatal thyroid status, with the studies by Brechner
et al. (2000) and Schwartz (2001) contributing the most compelling observations, and iodine
depletion was observed experimentally. The presence of exposure misclassification and
potentially serious confounding in many of the studies makes interpretation difficult and allows
for the possibility of missed effects even at the level of current thyroid function (e.g., steady state
levels of TSH or T4). The full implications of these findings are unclear; however, they should
be taken seriously, especially in populations already at risk for thyroid deficiency. These
considerations are summarized in Table 4-5.
The present review differs from a recent summary co-authored by two major participants in
industry-funded perchlorate research (Soldin et al., 2001). That review argues that there is now
sufficient evidence to recommend safe levels for regulatory purposes. The authors see no
immediate need for refinement of the physiological issues underlying the existing epidemiologic
study designs or for new initiatives in evaluating such issues in human populations. Potentially
important aspects of the mode-of-action for perchlorate not well addressed in the available
human studies include: (1) short-term effects of variable exposure during pregnancy, for
example, on critical neurodevelopmental effects; (2) the effects of iodine depletion on the T4 or
TSH surge response at birth, i.e., whether the effect of perchlorate on fetal thyroid status depends
additionally on prior cumulative exposure; (3) the equilibration of this regulated system under
chronic exposure and the masking of potential deficiency states such that steady-state T4 or TSH
levels appear normal despite substantial impact on production and function; and (4) the special
situation of populations or individuals with inadequate iodine intake where thyroid
responsiveness may be compromised.
January 16, 2002
4-30
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
The recent clinical data (Lawrence et al., 2000; 2001; Greer et al., 2000) may be more
useful in helping to characterize the potential effects on thyroid function if the mode of action
framework is superimposed on the interpretation of the data (i.e., that prevention of significant
iodide inhibition would preclude adverse neurodevelopmental and neoplastic sequelae).
However, given the current controversy in evaluating thyroid status, particularly in pregnant
women, it is difficult to ascertain the degree of iodide inhibition to designate as adverse. Further,
there is considerable uncertainty associated with using small sample sizes of euthyroid
individuals as the basis of such a determination, so that the use of a factor to account for this in a
risk derivation would be warranted, particularly when the variability as noted is considered and
the range of inhibition of iodide uptake at levels suggested to be "No-Observed-Adverse-Effect-
Levels" include values as great as 38.6% below baseline. A discussion considering these human
clinical data in comparison to the laboratory animal toxicological study results can be found in
Section 7.1.5.1.
January 16, 2002
4-31
DRAFT-DO NOT QUOTE OR CITE
-------
03
3
C
P
On
K)
O
O
K)
TABLE 4-5. SUMMARY OF HUMAN POPULATION STUDIES (Park, 2001)
Publication
Study Population
CIO/ Source and
Levels
Duration
Outcomes
studied
Findings
Problems/Comment
u>
o
o
z
o
H
o
c
o
H
w
o
o
HH
H
m
Gibbs JP, Ahmad R,
Crump KS, et al
JOEM 1998, 40:1072-
1082 Evaluation of a
population with
occupational exposure
to airborne ammonium
percltlorate for
possible acute or
chronic effects on
thyroid function.
Lamm SH, Braverman
LE, Li FX, et al JOEM
1999;41:248-260.
Thyroid health status of
ammonium perchlorate
workers • a Cross-
sectional occupational
health study
Lawrence JE, Lamm
SI I, Braverman LE
J Endocrinol Invest
1999, 22 405-407. The
use ofperchlorate for
the prevention of
thyrotoxicosis in
patients given iodine
rich contrast agents
Kerr-McGee workers
in voluntary medical
surveillance 1994-98;
170 out of 254 did
survey; 130 did single
shifl evaluation
Radiocontrast patient
series
Airborne exposure to
AP in 8 homogenous
exposure groups
0.04-627 fim/m'
using closed face
cassettes
American Pacific
workers 37 AP and
21 azide workers: full
feasible participation,
all from same site with
same other work
attributes
I day
1-27 yr.
Airborne exposure in
3 AP groups based
on visible dust level;
total and rcspirable
AP by individual
closed-face samplers
10-11 hrs on subset
from each exp group,
levels: total dust
(mg/day)- .01, .34,
6.57, 59 4; resp
fraction (nig/day)1
.02, .09, .60, 8.6
Therapeutic high oral
doses (1000 mg) in
day prior to contrast
agent
1 day
n=58,
6 days
n=2
1 day
T3U, T4, FTI,
TSII, liver,
kidney and
hematol fen
T4: 7.5 A(g/dL
TSH: 2.0 ^lU/ml
Urine AP, T3, T4,
FTI, TSH, THBR,
and hematologic
fen
T4- 7.0 pg/dL
TSII: 2.6
JLrU/lnl
Misc thyroid
parameters
Indication of increase in TSH over
work shift. 2.2 -> 2.5. In
workforce, T4 declines and TSH
increases from low to high exposure
but also from low exposure to
unexposed; see inconsistent TSH
trends using two lab groups; for
both thy and liv outc, SDs increased
in high dose group- for thy and liv
fen, averages for low vs high AP
very different but %iles up to 75th
are not Implies big excursion at
high exposure end.
18% of total airborne Mb is
rcspirable (range 8-25), urinary
excretion of P shows much higher
absorbed dose in unexposed
workers than expected from air
samples: (mg): .88. 4.0, 10 9, 33.6
(assuming 8 hr halflife) Thy,
hematol by current exp group:
no association (T3, T4]; absorb dose
greatly exceeds resp total inhaled
dose [F51 See aberrant clearance in
I of 2 6-day subjects fF2], Authors
conclude no AP health effects.
Recommend in high risk patients
(low iodide areas and elderly) a
combination of perchlorate and
contrast agent.
Possibly half of cligibles did not
participate in slnfi study, possibly
confounded by shift duration.
Did not evaluate ITR.
Suggestion of inappropriate
unexposed comparison group
In this steady state and cross-
sectional population, difficult to
assess thyroid regulatory status
SDs suggest heterogeneity of
effect. Indications qf chronic
effects.
Some misclassification apparent
among exposure groups based on
absorbed dose; non-inhalablc
contribution may constitute
important deficit in air sampling
results Steady-state, cross-
sectional population difficult to
interpret Thy, hematol results
based on current, non-cumulative
AP exposure are uninterpretable
for chronic effects. Possible
increasing trend for max(T3) with
exposure group
Not relevant to and uninformative
on chronic exposure effects in
adults and acute effects in infants
-------
3
e
P5
ON
CO
O
o
TABLE 4-5 (cont'd). SUMMARY OF HUMAN POPULATION STUDIES (Park, 2001)
Publication
Study Population
CIO/ Source and
Levels
Duration
Outcomes
studied
Findings
Problems/Comment
OJ
o
o
2
o
H
/O
G
O
H
m
o
n
Li FX, Squartsoff L,
Lamm Sl l. JOEM
2001; 43 630-634,
Prevalence of thyroid
diseases in Nevada
counties with respect to
perchlorate in drinking
water.
Lawrence JE, Lamm
SH, Pino S, Richman
K, Braverman LE.
Thyroid 2000; 10:659-
663. The effect of
short-term low-dose
perchlorate on various
aspects of thyroid
function.
Medicaid population
at risk for thyroid
disease in Nevada in
1997-98.
Crump C, Michaud P,
Tellez R et al. and
Crump KS, Gibbs JP.
JOEM 2000; 42:603-
612. Does perchlorate
in drinking water affect
thyroid function in
newborns or school-
age children?
9 healthy, male
volunteers K-
perchlorate -
1 Omg/day
Perchlorate in
drinking water in one
county (P= 8.9-
11.6 /^g/L) versus all
others
Lifetime
1CD 240-246;
ICD 193: thy
cancer
School children from
1 or 2 schools in three
cities in Chile
(n=53,49,60 in 0, low
and high P cities); all
newborns 2/96-1/99 in
same cities
(n=8888,468,428)
Geological Na-P in
drinking water
(0, 5.5, 1116 ^g/L )
Recent and T3, T4, free T4,
lifetime for 6-8 FT1, TSH,
yr-olds; hematol, liver,
gestation kidney,
prevgoiter,
prev: family H,
thy disease
T4:IO.O/^g/dL
TSH: 3.0/^IU/mL
Potassium
perchlorate
10 mg/day
14 days
T3, T4, FTI,
TSH, THBR,
RAIU, liver,
hematology
T4. 7.0 /ig/dL
TSH 1 0 vl\U/mL
Exposed county (Clark) with
Las Vegas compared to another
county with a city (Reno/Washoe)
as well as with all other counties.
No significant excesses found for
exposed county for the 8 outcomes
studied. Actually, the comparison
counties (one with a city, and all
others) for all important outcomes
differed more between them than
with the exposed county. For the
6 more prevalent outcomes
(n=3069) the exposed county had
higher rates than the unexposed
(Washoe) county.
Did comparisons across cities.
Urinary l/creatine low in city-2
lifetime residents: (1,092, 862,
963); goiter high in city-2 recent
residents (17.7, 26.5, 23 3%) and
high in city-3 lifetime residents:
(22.2, 19.5, 26.0 based on 8, 8, 13
cases); family H, of the disease high
in city-3: OR=4.9 (11.1, 9.8, 30.0);
highly significant increase in T4
with increased P(1 25, 1 34, 1.50).
Highly significant decrease in log
(TSH+1) in newborns in
city-3-high P (.91, .91, .66) [T9],
which is in the unexpected
direction. There was a diverse age-
at-screen distribution across cities.
Assumed identical P doses.
Upward trend for T3 at BL, 7-, 14-,
and 28-days (136, 140, 151, 157;
trend not tested) See depressed
l-uptake at 14 days (40%) with
rebound at 28 days; non-24 hour
urinary- and scrum-l was unchanged
throughout. Authors conclude:
no thyroid impact because of large
I-storage.
Based on period-prevalence rates.
Two outcomes with small
numbers are not informative:
congenital hypothyroidism (n=22)
and thyroid cancer (n=44). The
difference in the comparison
counties suggests that
uncontrolled confounders or
uncertain estimates are
affecting this analysis and that
the study is uninterpretable for
all but large effects.
Confounders might include age,
gender, body mass, diet, iodine
intake, ethnicity, occupational
exposures.
Dietary, ethnic, birthwt, SES
confounders of thy fen
uncontrolled; observe trends in
unexpected directions; suggesting
confounding. Unknown if some
Chileans boil drinking water.
Significant paradoxical effects
indicate uncontrolled
confounding and inappropriate
thy fen model in relation to P in
this population Possible role of
ambient temperatures.
Hematol, liver test results
clinically "normal" but no data
presented. Inappropriate
assessment: clinical rather than
epidemiological. T3 effect not
addressed; dietary I not controlled
or reported. Suggests long term
iodine depletion.
-------
CO
3
C
ON
K)
O
O
K)
TABLE 4-5 (cont'd). SUMMARY OF HUMAN POPULATION STUDIES (Park, 2001)
Publication
Study Population
CI04" Source and
Levels
Duration
Outcomes
studied
Findings
Problems/Comment
Lawrence JE, Lamm S,
Braverman LE. Thyroid
2001. 11 295 (letter)
Low dose perch/orate
(3 mg daily) and
thyroid function
8 healthy volunteers
Potassium
perchlorate
3 mg/day
14 days T3, T4, FTI, No signif changes (data not
TSI I, THBR, presented) except for depressed
RAW, liver, l-uptakc at 14 days (10%) with
hematol significant rebound (22%) at
28 days;
Implies some I depiction over
2 weeks at 3 mg/day (seen by
other investigators at 1.4 mg/day)
U>
4^
o
o
z
o
H
o
C!
o
H
W
o
n
M
H
m
Lamm Sl-I, Doemland
M JOEM 1999;
41 -409-411 Has
perchlorate in drinking
water increased the
rate vf congenital
hypothyroidism 7
Li Z, Li FX, Byrd D,
et al and Lamm JOEM
2000; 42 200-205
Neonatal thyroxine
level and perchlorate
m drinking water.
10 Li FX, Byrd DM,
Deyhle GM et al. and
Lamm. Teratology
2000, 62-429-431.
Neonatal thyroid-
stimulating hormone
level and perchlorate
in drinking water.
Newborns in CA and
NV in 1996-97 in
7 counties
Newborns in Reno
and Las Vegas NV
12/98- 10/99 with
birthwt 2 5-4 5 kg
Perchlorate in
drinking water:
4-16 £
-------
3
P
e>
p\
K)
O
O
KJ
TABLE 4-5 (cont'd). SUMMARY OF HUMAN POPULATION STUDIES (Park, 2001)
Publication
Study Population
CIO/ Source and
Levels
Duration
Outcomes
studied
Findings
Problems/Comment
u>
Lfx
o
o
•z
o
H
o
c
o
H
tfl
o
p
n
Z
m
12
Brechner RJ, Parkhurst
GD, Humble WO et al.
JOEM 2000; 42:777-
782. Ammonium
perchlorale contamin-
ation of Colorado
River drinking water is
associated with
abnormal thyroid
function in newborns in
Arizona.
Schwartz
J.Dissertation, UC
Berkeley, 2001.
Gestational exposure to
perchlorale is
associated with
measures of decreased
thyroid function in a
population of
California neonates
Newborns 10/94-
12/97 in two Arizona
cities whose T4 screen
was below state-wide
daily 10%ile
Perchlorate in
drinking water
<16 /ig/L
Gestation
TSH
TSH:
13.4 /ilU/mL
99% of California
newborns screened for
thy disease in 1996
Perchlorate in
drinking water
classified in 3 levels
and assigned by zip
code: 1-2,3-12,
13+ //g/L
Gestation
Compared cities. TSH higher in
newborns from exposed city
(median: 19.9 vs 13.4); age at
screen distribution very different
between two cities exposed
screened sooner. Stratifying on age
at screen (0, 1-4, 5+ days) and
Hispanicity, see signif increase
(p=.017); adj effect not reported
T4, TSH,
presumptive
positive;
congenital
hypothyroidm
T4: 160 mg/dL
TSH- 7.6 ^IU/mL
Compared across four levels of
estimated exposure Has detailed
covariates: birthweight, age at
screen in hours, ethnicity in
20 groups; birth multiplicity;
ANCOVA model with extensive
control of most confounders finds
highly significant decrease in T4
(mean=l66) with P level (0, -9.7,
-11.2,-18.2) and large effects for
birthweight (-72 for birthweight
1500-2500), age (-50 for hours
7-18) and ethnic groups (- 10 to
-30); sec initial T4 fall followed by
surge by 12 hours and stays elevated
until 36 hours; initial onset of TSH
surge unresolvablc in time; stays
elevated till 18 hours Significant P
effect on TSH (0, .029, .03, .128)
but birthweight effects models (-.09
for < 1.5 kg). Model for
presumptive positives shows strong
age at screen and ethnicity effects;
for congenital hypothyroidism,
insignificant effect.
TSH levels (13-20) higher than
reported for other newborns (7-
10).] Selection on T4 level is
problematic due to strong age
dependence of T4 surge al birth
thus causing variable percentile
discrimination with age (8-40%
were screened depending on age).
This effect could increase TSH of
the exposed cily relative to
unexposed city but the cfTect of
the bias is difficult to predict.
Uncontrolled other confounding
e.g., birthwt, gest. age, iodine
intake, SES.
[T4 is reported at levels
10,000-fold higher than in other
studies.] presumptive positive
criterion not clear (all at or below
9 mg/dl plus lowest 5%
immediately above 9 mg/dl?).
NO P-ITR reported, e.g., P * age
(especially on surge amplitude), P
* birthweight, possible selection
bias in identification of TSH
subjects. Age at screen was not
included in logistic regression
model of congenital
hypothyroidemia This study
presents strong evidence of
perchlorate health effects in
neonates from drinking water
contamination with perchlorate.
-------
(u
D
C
P
ON
K)
O
O
K>
TABLE 4-5 (cont'd). SUMMARY OF HUMAN POPULATION STUDIES (Park, 2001)
Publication
Study Population
CIO/ Source and
Levels
Duration
Outcomes
studied
Findings
Problems/Comment
13 Soldin OP, Braverman Review of animal and
This review, co-authorcd by two
Not considered in this review are
LE, Lamm SH human evidence
major participants in industry
issues such as (1) short term
Therapeutic Drug
funded perchlorate research, argues
effects of variable exposure
Monitoring 2001;
that there is now sufficient evidence
during pregnancy, (2) the effects
23:316-331.
to recommend safe levels for
of maternal iodine depletion on
Perchlorate clinical
regulatory purposes, i.c , at this time
T4 or TSI1 surge response at birth,
pharmacology and
there is no need for further
(3) the equilibration of this system
human health • a
refinement of the physiological
under chronic exposure and Ihe
review.
issues underlying the existing
masking of potential deficiency
epidemiologic study designs or for
states, and (4) the special
new initiatives in evaluating such
situation of populations with
issues in human populations.
inadequate iodine intake.
U>
OS
O
o
z
o
H
o
c
o
H
tn
o
50
n
3
tn
1 = iodine; P = perchlorate, AP — ammonium perchlorate; exp — exposure, thy :
[Tn] = table in paper; [Fn] = figure in paper
thyroid; liv = liver; hematol = hematologic; ITR = interaction; outc = outcomes, SD = standard deviation; = history;
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5. TOXICOLOGICAL EFFECTS IN LABORATORY
ANIMAL STUDIES
This chapter provides a review of the relevant laboratory animal toxicity data for
quantitative dose-response analysis of the toxic effects of perchlorate exposure. Evidence that
both the neoplastic and non-neoplastic effects of perchlorate derive from its anti-thyroid effects
at the sodium (Na+)-iodine (I ) symporter (NIS) should be appreciated. Studies completed before
the initiation of the perchlorate testing strategy described in Chapter 3 are included here, but the
major emphasis is on these newer studies given their contemporary design and integrated
approach to evaluating perchlorate's mode of action. This introduction provides a brief review of
the status of issues after the previous external peer review and a summary of studies
recommended and performed since that time. In response to the 1999 external peer review, the
EPA committed to a second external peer review to address these recommendations and to
evaluate the data from new analyses and studies (Noonan, 1999).
At the external peer review in February 1999, it was noted by the EPA that the thyroid
histopathology that had made a significant contribution to the risk assessment had never
undergone an independent peer review by a second pathologist in any of the studies. In addition,
these studies had been performed at several different laboratories with several different study
pathologists using different lesion grading systems. The external peer review panel agreed that
these inconsistencies between study reports made it difficult to compare studies and could
contribute to variability in the resultant dose-response estimate (Research Triangle Institute,
1999).
In response, the National Center for Environmental Assessment (NCEA) committed to a
Pathology Working Group (PWG) process in collaboration with the NIEHS. The purpose of the
independent peer review and PWG was to decrease variability in response across the studies by
providing a common nomenclature for lesions and a consistent pathology review. Determination
of No-Observed-Adverse-Effect-Levels (NOAELs) or designation of adversity was not the
objective of this review. NCEA asked Dr. Douglas C. Wolf in the EPA's National Health and
Environmental Effects Research Laboratory (NHEERL) to conduct the requisite independent
January 16, 2002
5-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
peer review (second pathology review) using one consistent lesion grading system on the
materials. Dr. Wolf was chosen because he had not been involved in any of the work performed
with ammonium perchlorate and because he had developed a thyroid grading scheme (Hooth
et al., 2001) to analyze a similar thyroid response in rodents exposed to sodium chlorate that
would be useful to the perchlorate review.
After the initial pathology review of 100% of the thyroid slides by Dr. Wolf, Dr. Peter
Mann of Experimental Pathology Laboratories, Inc. (EPL), reviewed 100% of the slides for
quality assurance/quality control (QA/QC) and consistency. Subsequent to this QA/QC review
of the independent peer review, a NIEHS-sponsored PWG of 5 experienced veterinary
pathologists was conducted on a subset of the slides. Recommendations of that PWG
(Experimental Pathology Laboratories, 2000) were then incorporated into the final report on the
independent review of 100% of the slides conducted and reported by Dr. Wolf (Wolf, 2000).
Both of these reports were made available almost immediately to the public on the NCEA
website. During subsequent analyses it was appreciated that the slides provided for the
two-generation study (Argus Research Laboratories, Inc. 1999) were from animals not on test
and some of the mean severity scores were miscalculated. These minor changes are provided in
Wolf (2001).
The thyroid slides that underwent the PWG review included materials from the following
studies: Argus Research Laboratories, Inc. (1998a,b,c); Caldwell, et al. (1995); Keil et al.
(1998); and Springborn Laboratories, Inc. (1998). It should be noted that the two-generation
reproduction study performed by Argus Research Laboratories (1999) was completed at the time
of the PWG, and the review included all final thyroid tissue slides despite its listing in the PWG
and Wolf (2000; 2001) reports as 1998c. The newest study, that of Argus Research Laboratories,
Inc. (2001) described below in Section 5.3.3, was also performed with the new nomenclature and
grading system. The study pathologist had been a member of the PWG; therefore, the pathology
results can be considered consistent with the results of Wolf (2000, 2001). However, a second
independent review of the pathology in that study has not been performed.
All analyses performed on thyroid histopathology in this revised risk assessment rely on
either the PWG data (Wolf, 2000; 2001) or the new 2001 study (Argus Research Laboratories,
Inc., 2001). The revised benchmark dose (BMD) analyses for thyroid colloid depletion,
hypertrophy, and hyperplasia diagnosed in the studies reviewed by the PWG are presented in
January 16, 2002
5-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Table 5-1 (Geller, 2001a). Figures 5-1 and 5-2 present these estimates and their distributions
graphically in comparison to the previous 1998 assessment values. It is worthwhile to note that
while hyperplasia occurs at slightly higher concentrations in the analysis of the overall data array,
there is considerable overlap with the distributions of the other two thyroid histopathology
indices (colloid depletion and hypertrophy). This overlap is especially evident when evaluating
BMD or benchmark dose lower confidence level (BMDL) values within individual studies.
The potential for variability due to inconsistent handling of the radioimmunoassay (RIA)
kits used for serum thyroid and pituitary hormone levels was also noted at the external peer
review (Research Triangle Institute, 1999). In response, the Air Force Research Laboratory
(AFRL) conducted a study to compare serum thyroid hormone and TSH data obtained by RAI
procedures for three different research laboratories that participated in perchlorate toxicity
studies involving hormone analysis (Narayanan, 2000). The purpose was to statistically
investigate the reproducibility (i.e., variability across laboratories) and the repeatability (i.e.,
variability within a laboratory) of the hormone measurements expressed as counts per minute
(CPM). RIA kits from the same batch number and with the same expiration date were used for
all the hormone measurements for all the standard and unknown samples. For unknown samples,
six rat serum samples plus six samples obtained from different species (dog, guinea pig, rabbit
and mouse) were used. Assays were performed using the RIA kits according to the
manufacturers' recommended procedures and each laboratories' standard operating procedures.
Reproducibility limits (RL) for each sample and for each hormone were determined. The
RL was defined as approximately 95% of all pairs of means from the same hormone and same
sample; different laboratories should differ in absolute value by less than the RL. The difference
in means between any two laboratories is a normally distributed random variable with a mean of
zero. The range ± RL is then the middle 95% for this distribution (i.e., 2.5% in each tail). The
reproducibility varied for each hormone with T3 showing the best reproducibility and TSH the
least. Three replicates ensured a more reproducible sample even when repeatability was not as
consistent. The results suggest that the variability in the RIA determination should be considered
when determining effect levels.
It was also recommended at the external peer review, by the biostatistician Dr. Joseph
Haseman, that different approaches to the thyroid and pituitary hormone analyses be explored
(Research Triangle Institute, 1999). EPA complied with this request and developed two new
January 16,2002
5-3
DRAFT-DO NOT QUOTE OR CITE
-------
D
g
ON
K>
O
O
KJ
TABLE 5-1. BENCHMARK DOSE (BMD)a AND BENCHMARK DOSE LOWER CONFIDENCE LIMIT (BMDL)a
ESTIMATES CALCULATED FROM THE WOLF (2000, 2001) THYROID HISTOPATHOLOGY DATA (Geller, 2001a)
Study Name, Time Point
Wolf (2000; 2001)
Table Number
Ammonium perchlorate dose
levels test
(mg/kg-day)
Colloid Depletion
Hypertrophy
Hyperplasia
BMD
BMDL
xJb
Expc
BMD
BMDL x"
Exp°
BMD
BMDL x2"
Exp'
1. Caldwell
Tbls. 1 and 2
0, 1.25, 5, 12.5, 25, 50, 125,250
13 29
0.72
0.97
4 37
Not done"1
35.29
0.78 0.20
0.88
2. Subchonic, 14-day
Tbls. 3 and 6
0,0.01,0.05,0.2, 1.0, 10.0
2.55
0.28
0.20
0.74
0.75
0 017 0.54
0.78
NOEe
3. Subchronic, 90-day
Tbls. 4 and 7
0, 0.01,0.05,0.2, 1.0, 10 0
0 13
0.03
0.70
0.50
0 21
0.008 0.74
0.55
8.36
2.09 1 00
7.87
4. Subchronic, 120-day
Tbls. 5 and 8
0, 0.05, 1.0, 10.0
NOE
NOE
NOE
5 Neurobehav., FO Fem
Tbl 9
0,0 1, 1,3, 10
NOE
NOE
NOE
6. Neurobehav , PND5
Tbls. 10 and 11
0,0.1, 1,3, 10
0.45
0.53
0.009
0.33
0.46
0.67f
0.94
1.0
0.92
1.27
0.24 024
0.88 0.26'
0.81
1 0
15 18
11 02
1 86 0.70
3.62 0 32f
0.36
1.0
7. Neurobehav, adult
Tbls. 12 and 13
0,0.1, 1,3, 10
0.72
0.029
0.23
0.89
3.48
NC 0.72
0.29
NOE
8. 2-gen., PI
Tbls. 14 and 15
0, 0.3, 3, 30
1.97
0.11
0.68
3.84
Poor fit8
7.89
2.44 0.41
0.72
9. 2-gen., P2
Tbls. 16 and 17
0, 0 3, 3, 30
2.16
0 90
0.06
1.16
0 99
0.15 0.67
0.70
4 62
0.0004 0.14
0.31
10. 2-gen., F1-weanling
Tbls 18 and 19
0, 0.3,3,30
2.51
0.80
0.17
1.2
0.21
0.057 0 40
0.79
2 74
0.66 0.85
0.52
11 2-gen , F2-weanhng
Tbls. 20 and 21
0, 0.3,3,30
Poor fit
1.19
0 32 0.25
0 52
NOE
BMDL Range: Rat Studies
0.009 - 0.90
0.008 - 0.74
0.0004 - 3.62
12. Dev tox., rabbit dams
Tbl. 22
0, 0.1, 1, 10, 30, 100
0.12
0.008
0.19
0.36
Poor fit
1.53
0.42 0.13
0.61
13. Immunotox. Mice,
combined studies
Tbl 23
0, 0.1, 1, 3, 30
26.07
5.15
1 00
7.88
1.62
0 97 0 58
0.84
24.92
4.48 1.00
7 86
Lfl
I
H
I
o
o
2
O
H
/O
c
o
H
W
O
o
H
tfl
1 Units of mg/kg-day
b x2 p-value.
c Exponent in Weibull model fit not restrained to 2 1.0 unless indicated.
d Not done: Because of non-routine staining, cytological characteristics were not adequate to make determination of hypertrophy on these samples
(Wolf, personal communication).
1 No observed effect (NOE): Either no incidence of endpoint noted in animals tested or no notable difference between dosed and controls.
' Exponent in Weibull model fit restrained to i I.
' Poor fit. p < 0.05 for x2 test
-------
100
10 -
1 -
0.1
0.01 -
0.001 -
0.0001
—I—
f
V
~
t
~
o
a
+
~
o
7
V
Solid symbols - BMD10
Open symbols - BMDL10
O Colloid Depletion
V Hypertrophy
~ Hyperplasia
-r
_u
~
~
~
-i 1 r
9
»
~
Figure 5-1. Benchmark dose (BMD) and benchmark dose lower limit (BMDL) estimates
recalculated for thyroid histopathology based on 2000 Pathology Working
Group review (Wolf, 2000; 2001). Data on incidence of colloid depletion,
thyroid hypertrophy and thyroid hyperplasia were submitted to the EPA for
the perchlorate risk characterization. Values used are presented in Table 5-1
(Geller, 2001a). Greater value represents the BMD and lesser value represents
the BMDL. The + denotes BMD and BMDL from previous EPA risk
characterization (U.S. Environmental Protection Agency, 1998d; Geller,
1998a). Values to the left of the vertical solid line are from the rat studies.
Values to the right are from the developmental study in rabbits (Argus
Research Laboratories, 1998c) and the mouse immunotoxicity studies (Keil
et al., 1998). Study denoted by "Caldwell" refers to Caldwell et al. (1995);
"Subchronic" to Springborn Laboratories, Inc. (1998); "Neurobeh" to the 1998
developmental neurobehavioral study (Argus Research Laboratories, 1998a);
and "2-gen" to the completed 2-generation reproductive toxicity study in rats
(Argus Research Laboratories, 1999).
January 16, 2002
5-5
DRAFT-DO NOT QUOTE OR CITE
-------
100
0.0001
0.01 -
0.001 -
. y,y y ,-«v
9?
Figure 5-2. Distribution of BMD and BMDL estimates shown by "box and whisker" plots
of colloid depletion (colloid), hypertrophy (hyptry), and hyperplasia (hyppls)
from rat studies recalculated for thyroid histopathology based on 2000
Pathology Working Group review (Wolf, 2000; 2001). Values are presented in
Table 5-1. Study #4 was excluded since it was a 30-day recovery experiment
and #5 was excluded due to lack of monotonicity. The boundary of the box
closest to zero indicates the 25th percentile, a line within the box denotes the
median, and the boundary of the box farthest from zero indicates the 75th
percentile. Whiskers above and below the box indicate the 90th and 10th
percentiles. The two rightmost boxes plot values from the combined rat
studies from the 1998 EPA risk characterization (U.S. Environmental
Protection Agency, 1998d; Geller, 1998a).
January 16, 2002
5-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
approaches to the analyses that address these comments (Crofton and Marcus, 2001; Marcus,
2001; Crofton, 2001a). All thyroid and pituitary hormone analyses presented will utilize these
new approaches. The reanalyses of the hormone data for the previous set of studies can be found
in Table 5-2.
Finally, a number of additional new toxicology studies were recommended by the EPA and
the external review panel in 1999. These included a study of the developmental effects of
perchlorate (Section 5.4.3); a re-evaluation of the effects of perchlorate on neurodevelopmental
motor activity (Section 5.3.2); refinement of the evaluation of immunotoxicity concerns with a
repeat of the sheep red blood cell (SRBC) response using the established plaque-forming cell
(PFC) assay for humoral effects and an additional test for contact hypersensitivity (Section 5.6);
and what has become known as the "Effects Study" (Section 5.3.3). The objective of the
"Effects Study" (Argus Research Laboratories, Inc., 2001) was to reevaluate brain morphometry
effects and to evaluate thyroid histopathology and thyroid and pituitary hormones at various
stages of development, including during gestation and post-natal days 5, 10 and 22.
5.1 CHRONIC STUDIES AND GENOTOXICITY ASSAYS
This section discusses the data establishing perchlorate as a carcinogen. A few long-term
studies at comparatively high doses performed before the 1997 perchlorate testing strategy
showed that perchlorate causes thyroid tumors. These studies are discussed in Section 5.1.1. In
order to invoke the conceptual mode-of-action framework for the anti-thyroid effects of
perchlorate causing thyroid neoplasia via a non-genotoxic mechanism, the testing strategy had to
determine whether or not perchlorate acts directly with DNA. This evidence is discussed in
Section 5.2.2. The completed genotoxicity data were presented at the 1999 external peer review
as Attachment A to the February 1, 1999 submission provided by NCEA to the peer review panel
(Zeiger, 1999a,b; Dellarco, 1999; BioReliance, 1999; Moore, 1999). Dr. David Brusick, the
genetic toxicologist on the previous external peer review panel, agreed with the EPA conclusions
(Research Triangle Institute, 1999) that perchlorate's ability to cause thyroid tumors was not
likely to be via a directly genotoxic mechanism.
It should be noted that perchlorate exposure also caused a statistically-significant increase
in tumors at the 30 mg/kg-day dose in the F1-generation pups of the two-generation rat
January 16,2002 5-7 DRAFT-DO NOT QUOTE OR CITE
-------
p
3
g
On
N>
O
O
K>
TABLE 5-2. A COMPARISON OF NOAELs AND LOAELs FROM THE ORIGINAL 1998 ANALYSES AND THE
2001 RE-ANALYSES FOR HORMONE AND MORPHOMETRY ON THYROID FOLLICULAR LUMEN SIZE
(Crofton and Marcus, 2001; Marcus, 2001; Crofton, 2001a)
Time Point, Age
(Doses, mg/kg-day)
Original Analyses
Rc-Analyscs°,h
Spccics/Study
Endpoint
Sex
NOAEL
LOAEL
NOAEL
LOAEL
Rat
14-Day
(Caldwell et al., 1995)
14-Day
(males - 0.0, 0.11, 0.44,
1.11, 2.26, 4.32, 11.44,
T3
M
F
0.11
0.44
0.11
0.11
0.44
0.12
22.16)
(females - 0.0, 0.12, 0.47,
1.23, 3.06, 4.91, 11.47,
T4
M
F
—
0.11
0.12
0.11
0.12
Zh.oO)
TSH
M
0.44
1.11
0.44
1.11
F
0.12
0.47
—
0.12
hTg
M
—
0.11
—
0.11
F
—
0.12
—
0.12
rT3
M
0.44
1.11
0.11
0.44
F
0.47
1.23
0.12
0.47
Rat
Subchronic Study
(Springborn, 1998)
14-Day
(0, 0.01, 0.05, 0.2, 1.0,
10.0)
73
M
F
10.0
0.01
10.0
0.01
T4
M
F
1.0
10.0
—
0.05
TSH
M
0.05
0.2
0.01
0.05
F
0.01
0.05
—
0.01
I
00
H
t
o
o
2
o
H
o
c|
O
H
m
o
O
H
m
-------
g
CT\
K>
O
O
TABLE 5-2 (cont'd). A COMPARISON OF NOAELs AND LOAELs FROM THE ORIGINAL 1998 ANALYSES AND THE
2001 RE-ANALYSES FOR HORMONE AND MORPHOMETRY ON THYROID FOLLICULAR LUMEN SIZE
(Crofton and Marcus, 2001; Marcus, 2001; Crofton, 2001a)
Species/Study
Time Point, Age
(Doses, mg/kg-day)
Original Analyses
Endpoint
Sex
Re-Analyses"
NOAEL LOAEL
NOAEL
LOAEL
Rat
Subchronic Study
(Springborn, 1998)
(cont'd)
90-Day
(0, 0.01, 0.05, 0.2, 1.0,
10.0)
T3
T4
TSH
M
F
M
F
M
F
0.05
0.01
0.01
0.01
0.2
0.05
0.01
0.01
0.01
0.01
0.2
Rat
Subchronic Study
(Springborn, 1998)
120-Day
(0, 0.05,1.0, 10.0)
T3
T4
TSH
M
F
M
F
M
F
1.0
10.0
10.0
0.05
1.0
1.0
10.0
0.05
10.0
0.05
o
o
z
o
H
/O
c
o
H
m
o
*
O
3
m
Rat
Developmental
Neurotoxicity Study
(Argus, 1998a)
PND5
(0, 0.1, 1.0, 3.0, 10.0)
PND90
(0, 0.1, 1.0, 3.0, 10.0)
PND5
(0, 0.1, 1.0, 3.0, 10.0)
PND90
Lumen size
Lumen size
T4
T3
TSH
M
F
M
F
Data not available for original
analyses
1.0
3.0
0.1
1.0
3.0
10.0
0.3
10
10
0.1
0.1
3.0
T4, T3, and TSH
No data available
1.0
1.0
10.0
-------
3
C
pa
ON
K>
O
o
K)
TABLE 5-2 (cont'd). A COMPARISON OF NOAELs AND LOAELs FROM THE ORIGINAL 1998 ANALYSES AND THE
2001 RE-ANALYSES FOR HORMONE AND MORPHOMETRY ON THYROID FOLLICULAR LUMEN SIZE
(Crofton and Marcus, 2001; Marcus, 2001; Crofton, 2001a)
Time Point, Age
Original Analyses
Rc-Analyscs",b
Spccics/Study
(Doses, mg/kg-day)
Endpoint
Sex
NOAEL LOAEL
NOAEL
LOAEL
Mouse
14-Day
T4
M
3.0 30.0
—
0.1
Hormone and
Immunotoxicity
(Keil ct al., 1998)
(0.0, 0.1, 1.0, 3.0, 30)
T3
M
Data not available at time of
1998 analysis
—
0.1c
TSH
M
No data
90-Day
T4
M
0.1 3.0
—
o.r
(0.0, 0.1, 1.0, 3.0, 30)
T3
M
Data not available at time of
1998 analysis
—
0.1d
TSH
M
30.0 —
—
0.1d
120-Day
T4
M
30.0 —
30.0
—
(0.0, 0.1, 1.0, 3.0, 30)
T3
M
Data not available at time of
1998 analysis
30.0
—
TSH
M
30.0
—
—
Rabbit Developmental
Gestation Day 28
T4
F
0.1 1.0
0.1
1.0
Toxicity
(Argus, 1998b)
(0.0, 0.1, 1.0, 10.0,
30.0, 100.0)
T3
F
100
100
—
TSH
F
100 —
100
—
i
©
H
i
d
o
2
o
H
/O
c
o
H
tfl
o
50
o
M
H
tn
"Bold indicates where 2001 analyses is different than 1998 analyses.
bRcsults from the liberal and conservative statistical approaches were the same.
cNo dose response - 0.1 and 1.0 differ from control; 0.3 and 30.0 do not differ from control.
dNo dose response - 0.1 and 1.0 differ from control; 0.3 and 30.0 do not differ from control.
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
reproductive study (Argus Research Laboratories, 1999). These pups were used as the parents of
the second generation (F2) pups in the study. When these F1 animals were sacrificed after only
19-weeks, tumors were observed (Wolf, 2000). The type was the expected benign thyroid
adenoma consistent with the anti-thyroid effect at the NIS (iodine uptake inhibition) with thyroid
hormone disruption followed by TSH upregulation. The early onset at 19 weeks is remarkable
and suggests the potential for in utero imprinting, a phenomenon beginning to be appreciated
with other endocrine-disrupting compounds (Prins et al., 2001; Phillips et al., 1998; Seckl, 1997).
These tumor results will be discussed in Section 5.5.
5.1.1 Cancer Studies
Kessler and Kriiskemper (1966) provided potassium perchlorate in drinking water at a
concentration of 0 or 1% to male Wistar rats for 2 years. Body weights and thyroid weights were
reported for groups of 6 to 8 rats sacrificed after 0, 40, 120, 220, and 730 days of treatment, and
thyroid glands from the animals were examined histologically. Using body weight data provided
in the report to calculate a time-weighted average body weight of 0.336 kg and using an
estimated water consumption of 0.045 L/day (calculated with the allometric equation
recommended by U.S. Environmental Protection Agency [1987]), a dose of 1,339 mg/kg-day can
be derived. Body weights of control and treated animals were comparable throughout the
experiment. In contrast, thyroid weights, both relative and absolute, were increased markedly in
treated rats compared to controls at each examination interval. Histological examination of
thyroids from treated rats at 40 days revealed follicular cell hyperplasia. The authors
characterized these changes as typical for a thyroid gland stimulated by TSH during a relatively
short period of time. After 200 days of perchlorate treatment, diffusely degenerative changes
with fibrosis and increased colloid were observed. The authors commented that the course of the
histological changes in the thyroid was similar to that produced by long-term administration of
thiouracil, another antithyroid agent. The authors further reported that 4 of 11 rats treated with
potassium perchlorate for 2 years developed benign tumors of the thyroid gland and that
20 untreated Wistar control rats displayed no thyroid gland tumors. The 1,339 mg/kg-day dose
suggested a free-standing LOAEL because no other doses were tested.
Pajer and Kalisnik (1991) administered 0 or 1.2% sodium perchlorate in drinking water to
groups of 36 female BALB/c mice (12/group) for up to 46 weeks. Eight or 12 weeks after the
January 16, 2002
5-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
beginning of the experiment, one group of treated and control mice were totally irradiated with
0.8 Gy on 5 consecutive days at a dose rate of 1.45 Gy/min so that each mouse received a total of
4 Gy. Assuming a body weight of 0.0353 kg and a water consumption rate of 0.0063 L/day (U.S.
Environmental Protection Agency, 1987), a dose of 2,147 mg/kg-day can be calculated. Thirty
animals died during the experimental period; however, details about the cause of death were not
provided. Forty-two animals were sacrificed at 46 weeks for histological examination of the
thyroid and pituitary gland. No other tissues were examined. Obvious treatment-related
histological changes were observed in the thyroid and pituitary gland, including thyroid follicular
cell carcinoma. Immunoperoxidase staining of pituitary thyrotropic cells and antihuman TSH
serum provided qualitative evidence of increased TSH production in the pituitary gland.
Perchlorate treatment was associated with an increased total volume of the thyroid and of the
distal parts of the anterior pituitary gland (adenohypophysis). In addition, increased average
volume and numbers of epithelial, thyrotropic, and parafollicular cells were observed. Irradiation
appeared to enhance the effects of perchlorate treatment. This study suggested a free-standing
LOAEL of 2,147 mg/kg-day for thyroid effects.
5.1.2 Genotoxicity Assays
ManTech Environmental Technology, Inc. (1998) performed a battery of three genotoxicity
assays {Salmonella typhimuriumlmicrosome mutagenesis assay [Ames assay], the mouse
lymphoma cell mutagenesis assay [L5178Y-TK test], and the in vivo mouse bone marrow
micronucleus induction assay) with ammonium perchlorate to help determine its potential for
various interactions with DNA and to gain insight into its possible carcinogenicity. To confirm
the findings of ManTech Environmental Technology, Inc., the EPA requested that the National
Toxicology Program (NTP) also evaluate ammonium perchlorate in the Ames assay and the
mouse bone marrow micronucleus test (Zeiger, 1999a). The sponsor (PSG) also had the mouse
lymphoma assay repeated (BioReliance, 1999).
Ammonium perchlorate was evaluated in the Ames assay {Salmonella typhimurium!
microsome mutagenesis assay), which is a well-defined assay for detection of mutagens.
It measures the reversion from a histidine-independent state (his ) induced by chemicals that
cause base-pair changes or frameshift mutations in the genome of the organism (i.e., it measures
for point mutations [e.g., substitution, addition, or deletion of one or a few DNA base pairs
January 16,2002 5-12 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
within a gene]). In this assay, bacteria are exposed to the test chemical with and without a
metabolic activation system (Arochlor 1254-induced rat liver S9 with co-factors).
The mutagenicity is evaluated by the increase in the number of revertant colonies. The L5178Y
mouse-lymphoma assay is another short-term in vitro assay to detect both point mutations and
structural chromosomal changes. The in vivo mammalian micronucleus test detects the damage
of chromosomes or of the mitotic apparatus caused by a clastogenic chemical in bone marrow
cells (polychromatic erythrocyte [PCE] stem cells) of treated animals. Micronuclei are believed
to be formed from chromosomes or chromosome fragments left behind during anaphase of
mitosis. The induction of micronuclei indicates changes in either chromosome structure or
number in bone marrow cells. ManTech Environmental Technology, Inc. (1998) performed this
assay in Swiss-CD-I mice and the NTP used B6C3F1 mice (Zieger, 1999a). The micronucleus
assay also was performed as part of the 90-day bioassay in Spraque-Dawley rats (Springborn
Laboratories, Inc., 1998). This is considered an adequate series of tests to determine the
mutagenic and clastogenic (chromosomal breaking) potential of an agent. It should be noted that
perchlorate is not likely to be mutagenic, given its physical and chemical properties (i.e., it is
simply an anion). Although perchlorate is an oxidizing agent, it is not expected to produce
oxidative DNA damage because of the kinetic considerations discussed in Chapter 2.
5.1.2.1 In Vitro Assays
Ammonium perchlorate was not found to be mutagenic in the Salmonella typhimurium
(Ames assay) with and without Arochlor 1254-induced rat liver S9 activation by two separate
laboratories (ManTech Environmental Technology, Inc., 1998; Zeiger, 1999a). In the ManTech
study, ammonium perchlorate was dissolved in distilled water and tested at five concentrations
(5,000, 2,500, 1,250, 625, and 312.5 /^g/plate) in tester strains TA98, TA100, TA1535, and
TA1537, with and without Arochlor 1254-induced rat liver S9 using the plate incorporation
assay. Although this study was regarded as adequate, the EPA requested that the Ames assay be
repeated by the NTP to confirm the negative findings and to include additional tester strains (i.e.,
TA102, and TA104) that are able to detect a variety of oxidative mutagens. Therefore, the NTP
evaluated ammonium perchlorate in the Salmonella/Ames assay in tester strains TA98, TA100,
TA1535, TA97, TA102, and TA104 (Zeiger, 1999b). Ammonium perchlorate was dissolved in
distilled water and tested using the preincubation procedure at doses of 10,000, 3,333, 1,000,
January 16, 2002
5-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
333, and 100 /Ug/plate, with and without metabolic activation from Arochlor-induced rat and
hamster livers. Ammonium perchlorate was neither toxic nor mutagenic under the conditions of
the NTP assay.
The L5178Y//F'" mouse lymphoma assay also was used to evaluate the mutagenic and
chromosomal breaking potential of ammonium perchlorate in vitro. Ammonium perchlorate was
reported to be negative both in the absence and presence of rat Arochlor-induced S9 liver
activation (ManTech Environmental Technology, Inc., 1998). Ammonium perchlorate was
evaluated at 5.0, 2.5, 0.5, 0.25, 0.05, and 0.025 mg/mL without S9 activation, and at 2.5, 0.5,
0.25, 0.05, and 0.025 mg/mL with S9 activation. Although a small increase in mutation
frequency was found in the absence of S9 activation at 2.5 mg/mL, which appeared to be
statistically significant (p < 0.05) by the two-tail Student's t-test, a repeat assay found no increase
in mutation frequency at this concentration compared with controls. Therefore, ammonium
perchlorate is considered to be negative in the absence of S9 activation. Confidence in the
negative findings without S9 activation is reinforced by the wide range of ammonium perchlorate
concentrations evaluated. Although ammonium perchlorate also was reported as negative in the
presence of S9 activation, the response of the positive control, 3-methyl cholanthrene (MCA), in
the actual experiment was too low (182.6 x 10"6) to be acceptable. The highest dose of
ammonium perchlorate produced a mutation frequency of 194 x 10"6. The MCA at 2.5 /ig/mL
should induce a mutation frequency of 300 to 350 x 10"6 or higher. Such a low positive control
response weakens the confidence for the negative finding with S9 activation. In addition, the
cloning efficiencies for the S9 test appear to be too high (143%), further reducing the confidence
in a negative finding. Therefore, only the assays on ammonium perchlorate without S9 are
considered unequivocally to be negative. Although perchlorate is not expected to be metabolized
to a mutagenic intermediate, these S9 data were not of sufficient quality to support a
negative-response conclusion.
Because of the problems described above, the sponsor (PSG) had the mouse lymphoma
assay repeated. In this recent mouse lymphoma assay, ammonium perchlorate was evaluated at
concentrations of 1000, 2000, 3000, 4000, and 5000 ^m/ml without and with Arochlor
1254-induced rat liver S9 activation (BioReliance, 1999). No increase in mutant frequencies
were found after treatment with perchlorate. The data were judged to be of sufficient quality to
determine perchlorate to be nonmutagenic both with and without S9 activation. Although the
January 16, 2002
5-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
background mutant frequency was low, particularly in the S9 experiment, the data set still is
considered to be very good overall, as well as internally consistent. The problems that were
observed in the data generated by the first laboratory (ManTech Environmental Technology, Inc.,
1998) were not present in the data form the BioReliance (1999) study.
5.1.2.2 In Vivo Assays
The potential for ammonium perchlorate to induce micronuclei was evaluated in mice and
rats. Ammonium perchlorate was administered by drinking water gavage for 3 consecutive days
to Swiss CD-I mice (5 females and 5 males per dose group) at 1,000, 500, 250, 125, and
62.5 mg/kg-day (ManTech Environmental Technology, Inc., 1998). Twenty-four hours after the
last dose, the mice were sacrificed, and the frequency of micronucleated cells were evaluated by
counting 1,000 PCEs per animal. The assay was conducted in accordance with existing EPA
Federal Insecticide, Fungicide, and Rodenticide Act/Toxic Substances Control Act
(FIFRA/TSCA) testing guidelines. No increase in the frequency of micronuclei were found for
any dose group. There is some uncertainty whether a maximum tolerated dose (MTD) was
reached in this study. The study authors reported that at 2,000 mg/kg, 4 out of 6 animals died
after one dosing of ammonium perchlorate. Typically, the assay is performed at 85% of the
MTD, and the 1,000 mg/kg-day represents approximately 50% of the LD50. There was no
indication of toxicity to the bone marrow cells because the polychromatic erythrocyte to
normochromatic erythrocyte (PCE/NCE) ratio was not different from negative controls.
Furthermore, the study authors did not report any indication of clinical signs of toxicity in the
highest dose group. Despite a rebuttal submitted by Dourson (1998) on behalf of the sponsor
(PSG), EPA remained concerned because of the importance of this test in the overall
determination of the approach to be taken for the carcinogenicity assessment (i.e., to rule out
direct genotoxicity).
The NTP agreed to expedite and repeat this test in response to an EPA request. The assay
was performed by ip injection to ensure the greatest delivery to the bone marrow. Male B6C3F1
mice were treated with 125, 250, 500, 1,000, 1,500, and 2,000 mg/kg ammonium perchlorate in
buffered saline, plus solvent and positive (cyclophosphamide) controls. Note that this study uses
two dose groups higher than those used in the previous study (i.e., 1,500 and 2,000 mg/kg).
Furthermore, the use of ip injection as the route of administration would result in a direct
January 16, 2002
5-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
delivery of the compound to the bone marrow cells versus delivery from drinking water gavage.
Five mice per group were injected daily for 3 consecutive days and were sacrificed 24 h after the
last injection; 2,000 PCEs were scored per animal for micronuclei. All animals in the 1,500- and
2,000-mg/kg groups died after the first ip injection, and 4/5 animals died in the 1,000-mg/kg
group after the second ip injection. No increases in percent PCE were observed in any of the
remaining test groups (125, 250, and 500 mg/kg). No bone marrow toxicity was seen as
indicated by the percent of PCE (Zeiger, 1999a,b). These results are interpreted to be consistent
with those of the ManTech Environmental Technology, Inc. (1998) study that used gavage
drinking water administration, and confirm that perchlorate does not induce micronuclei in
rodents.
The 90-day subchronic bioassay using Spraque-Dawley rats also evaluated micronuclei
induction (Springborn Laboratories, Inc., 1998). The frequency of micronuclei induction was
examined in both the males and females after the 90-day sacrifice in the 10-mg/kg-day dose
group of ammonium perchlorate administered by drinking water. Although there was no
induction of micronuclei at this dose, 10 mg/kg-day does not appear to reach a MTD because
there were no overt signs of toxicity. However, the definition of MTD may be somewhat moot,
given the changes in thyroid hormone economy and histopathology seen in the thyroids at that
dose. There was significant reduction in the PCE/NCE ratio (i.e, an indicator of toxicity to the
bone marrow cells).
5.1.2.3 Summary of Genotoxicity Battery Results
Negative results were reported in all genotoxicity assays conducted on ammonium
perchlorate when evaluated by two independent laboratories. Ammonium perchlorate was not
mutagenic in the Ames assay (with or without S9 activation). Negative results were also found
in the mouse lymphoma gene mutation assay without and with S9 activation. Ammonium
perchlorate did not induce chromosomal anomalies when evaluated for micronuclei induction in
the bone marrow of mice when administered via drinking water gavage or ip injection.
No increases in micronuclei were found in Spraque-Dawley rats when evaluated from the 90-day
study at the highest dose, which produced both thyroid hormone perturbations and follicular cell
hyperplasia.
January 16, 2002
5-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
In conclusion, ammonium perchlorate does not have the potential to be mutagenic or
clastogenic. The in vitro and in vivo studies discussed above provide support for that conclusion.
Therefore, mutagenicity is not considered a possible mode of carcinogenic action for this
chemical.
5.2 GENERAL TOXICITY: SHORT-TERM AND SUBCHRONIC
TESTING
The majority of the data on perchlorate toxicity available from previous studies or as a
result of the current perchlorate testing strategy involved either short-term or subchronic
exposures and are presented in this section. As discussed in Chapter 3, the testing strategy
included targeted studies to evaluate different endpoints, e.g., developmental neurotoxicity
(Section 5.3), developmental studies (Section 5.4) reproductive studies (Section 5.5) and
immunotoxicity assays (Section 5.6). The rationale behind the 90-day study (Section 5.2) with
satellite examination of thyroid and pituitary hormones and a 30-day recovery period was to
evaluate anti-thyroid effects as possible precursor lesions. If a NOAEL could be established for
these precursor lesions, it was thought that a two-year bioassay would not be required. This
assumption is now more tenuous due to the tumors observed in the F1-generation at 19 weeks.
The integration of these results with the available human data to arrive at a risk assessment will
be discussed in Chapter 7.
5.2.1 Historical Data
Mannisto et al. (1979) measured serum levels of TSH, T3, and T4 by RIA in groups of 5 to
6 male Sprague-Dawley rats weighing 180 to 220 g that were exposed to potassium perchlorate
in their drinking water at concentrations of 0, 10, 50, 100, or 500 mg/L for 4 days. Potassium
perchlorate doses of 0, 1.5, 7.6, 15.3, or 76.3 mg/kg-day, respectively, were calculated assuming
a body weight of 0.2 kg and a water consumption rate of 0.0305 L/day (U.S. Environmental
Protection Agency, 1987). Perchlorate produced statistically significant increases in serum TSH
levels and decreases in serum T3 and T4 levels. Significant changes in all three parameters were
measured in the 100 and 500 mg/L (15.3 and 76.3 mg/kg-day, respectively) dose groups. In the
50 mg/L (7.6 mg/kg-day) dose group, levels of T3 and T4 were decreased significantly; TSH
January 16,2002 5-17 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
levels were increased slightly, but the increase was not significant. At the low dose, T3, T4, and
TSH levels were unchanged from controls. This study suggested a NOAEL of 1.5 mg/kg-day
and a LOAEL of 7.6 mg/kg-day for short-term exposures to potassium perchlorate.
Shigan (1963) administered 190 mg/kg-day of potassium perchlorate in water to rabbits and
white rats (number, sex, and strain not identified) for 3 mo. The author did not indicate whether
the compound was administered in drinking water or by gavage with water. The animals were
examined for cardiac function; liver function, based on changes in serum proteins; immune
function, based on leukocyte phagocytosis; and adrenal function. Perchlorate at the dose tested
caused a change in the electrocardiogram and a decrease in serum proteins, indicating a
disruption of the glycogen-forming function of the liver. Shigan (1963) did not indicate whether
these changes were observed in both rabbits and rats. Perchlorate had no effect in the remaining
tests. This study suggested a LOAEL of 190 mg/kg-day although the study translation is reported
incompletely, limiting its usefulness for risk assessment.
In a second set of experiments, Shigan (1963) also treated rabbits and white rats (number,
sex, and strain not identified) with 0, 0.25, 2.0, and 40 mg/kg-day of potassium perchlorate for
9 mo. The medium for dosing was not reported. The animals were examined for cardiac and
liver function, for conditioned reflexes, and for uptake and discharge of iodide by the thyroid. In
the two highest dose groups, there was a statistically significant increase in the amount of iodide
excreted from the thyroid; this increase was not observed in the 0.25-mg/kg-day dose group. The
study does not indicate if the effect was seen in one or both species tested. This study suggested
a NOAEL of 0.25 mg/kg-day and a LOAEL of 2 mg/kg-day for thyroid effects.
Hiasa et al. (1987) measured serum levels of T3, T4, and TSH by radioimmunassay in
groups of 20 male Wistar rats administered 0 or 1,000 ppm potassium perchlorate in the diet for
20 weeks. Assuming a body weight of 0.34 kg (the average final body weight of rats treated with
perchlorate) and a food consumption rate of 27.4 g/day (U.S. Environmental Protection Agency,
1987), an estimated dose of 80.7 mg/kg-day was calculated. Absolute and relative thyroid
weights were significantly increased compared to controls in perchlorate-treated rats. No effects
were seen on liver weights. The T4 levels decreased slightly, but the decrease was
not statistically significant. The T3 levels were unchanged compared to controls. The TSH
levels were increased statistically significantly compared to controls. Histological examination
January 16, 2002
5-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of the thyroid revealed diffused small follicles in perchlorate-treated rats and one case of
follicular hyperplasia. Thus, the 80.7-mg/kg-day dose could be considered a LOAEL.
Gauss (1972) fed female NMRI mice a diet containing 0 or 1% potassium perchlorate for
up to 160 days. Mice were between 50 and 60 days old at the beginning of treatment and
weighed between 19 and 28 g (average, 23.23 g). During the first 2 mo of treatment, body
weights increased about 12%; body weight data for longer treatment periods were not reported.
Assuming a body weight of 23 g and a food consumption value of 4.625 g/day (U.S.
Environmental Protection Agency, 1987), a dose of 2,011 mg/kg-day was calculated. Thyroid
glands were examined histologically at 10- to 20-day intervals throughout the 160-day study
period. Thyroid and nuclei volumes and height of epithelial follicles were increased in treated
mice throughout the treatment period compared to controls. The histological examinations
showed a progressive change in the histological appearance of the thyroid of treated mice,
beginning with colloid loss, nuclei volume expansion, and rising epithelium height, followed by
the appearance of hypertrophy and hyperplasia of the thyroid parenchyma. At later stages of the
treatment period, hyperplastic follicles, areas of adenomatic tissue, adenoma complexes, and
secreting cystadenomas were observed; however, no progression to malignancy was apparent.
The 2,011 mg/kg-day dose suggested a free-standing LOAEL because no other doses were tested.
5.2.2 Caldwell et al. (1995) 14-Day Study
Caldwell et al. (1995) administered ammonium perchlorate in drinking water at
concentrations of 0, 1.25, 5.0, 12.5, 25, 50, 125, or 250 mg/L to Sprague-Dawley rats
(6/sex/group) for 14 days. The actual dose administered to each animal was calculated by
multiplying the concentration of ammonium perchlorate administered in the drinking water by
each rat's average water consumption over the 14-day period and dividing this number by each
animal's average body weight for the same period, resulting in doses (male/female) of 0,
0.11/0.12, 0.44/0.47, 1.11/1.23, 2.26/3.06, 4.32/4.91, 11.44/11.47, and 22.16/24.86 mg/kg-day,
respectively (Caldwell et al., 1995). Caution must be used when interpreting these reports
because the conversion is sometimes not included (e.g., the Channel [1998b] consultative letter
reports results in units of the test concentrations rather than the dose converted to milligrams per
kilogram per day). Thyroids were weighed, histopathology and morphometry performed, and
thyroid hormone levels were measured with a radioimmune assay technique.
January 16,2002 5-19 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The consultative letter of Channel (1998b) provides results and comments on a
histopathological analysis of the rat thyroids from the Caldwell et al. (1995) 14-day study that
was performed by the Air Force Research Laboratory/Human Effectiveness Directorate
(AFRL/HEST) and never officially published (Eggers, 1996, as cited in Channel, 1998b).
As part of the previous assessment, EPA requested from the AFRL/HEST the previously
unpublished histopathology data from the 14-day oral dosing study performed by Caldwell et al.
(1995). The histopathology was discussed in the paper on the study design (Caldwell and Mattie,
1995) but had not been published in either Caldwell et al. (1995) or King (1995). The
histopathology data discussed herein were provided in a consultative letter from the AFRL/HEST
(Channel, 1998b). The EPA also performed a reanalysis of the thyroid hormone data (T4, T3,
rT3, TSH, and thyroglobulin [hTg]) found in the Caldwell et al. (1995) and King (1995) reports
(Crofton, 1998a). Because these individual data were supplied only electronically on Microsoft
Excel® spreadsheets and not submitted formally to EPA, Crofton, (1998a) represents the official
publication of these data. These histopathology data and reanalyses of effect levels using the
PWG results and new hormone analyses are found in the following sections.
5.2.2.1 Thyroid Histology Data
Channel (1998a) submitted that the incidence of thyroid follicular cell hypertrophy
determined by standard histology was significantly different from control at a lower dose
(0.44 0.47 mg/kg-day) than for the incidence of decrease in follicular lumen size (2.26
3.06 mg/kg-day), but the statistics indicate a NOAEL at 0.11 0.12 mg/kg-day. However, the
documentation of the statistics was not provided, and Eggers (1996) apparently combined both
sexes for the analyses. It is recommended in the report (Channel, 1998a), and EPA concurred,
that a re-analysis was warranted for a number of reasons. First, there was a gender-by-treatment
interaction observed in the thyroid hormone analyses (see Section 5.2.2.2). Secondly, there was
an apparent dose trend, despite the limited sample size, in the incidence of response: male and
female combined was 7/12, 6/11, 11/12, 10/12, 12/12, 12/12, 12/12, and 12/12; male only was
3/6, 4/6, 5/6, 5/6, 6/6, 6/6, 6/6, and 6/6; and female only was 4/6, 2/5, 6/6, 5/6, 6/6, 6/6, 6/6, and
6/6 for the 0, 0.1, 1.0, 5.0, 10, 20, 50, and 100 mg/kg-day groups, respectively. Finally, the
analysis did not combine severity and incidence data for the decrease in lumen size, but the mean
severity scores alone were statistically significant from control above the 0.44/0.47 mg/kg-day
January 16, 2002
5-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
group. A separate computerized morphometric analysis of follicular lumen size was performed
by AFRL/HEST for the 0, 0.11/0.12, 1.11/1.23, 4.32/4.91, and 22.16/24.86 mg/kg-day groups,
and a statistically significant difference in the incidence of decrease in lumen size was evident in
the males at the 1.11 mg/kg-day dose and, in females, at the 4.91 mg/kg-day dose; however, the
gender-by-treatment effect was not taken into account. Relative thyroid weights were
significantly increased in 11.44/11.47 and the 22.16/24.86 mg/kg-day dose groups compared to
controls.
Results of the PWG analysis can be found in Wolf (2000; 2001; Tables 1 and 2). Female
rats appeared to be slightly more sensitive in this study with a NOAEL designated at 1.23 mg/kg-
day; whereas, in males it was somewhat difficult to ascertain. This may be due to the difficulty
that the PWG had in reading the slides from this study due to the non-routine staining method
(periodic acid shift [PAS] reaction with a green counterstain) as noted in Wolf (2000). BMD
analysis (Table 5-1) for the combined female and male data results in BMDL values for a 10%
increase in incidence at 0.72 mg/kg-day for colloid depletion and 0.78 mg/kg-day for hyperplasia.
The difficulty noted above with the staining for this study was most prominent in evaluating
hypertrophy (Wolf, personal communication), so that these estimates were not calculated.
Re-analysis of the morphometry on thyroid follicular lumen size identified a NOAEL at the
0.44/0.47 mg/kg-day dose.
5.2.2.2 Thyroid and Pituitary Hormone Analyses
The thyroid and pituitary hormone data were reanalyzed using five two-way analysis of
variance (ANOVA) tests, one each for all of the hormones (Crofton, 1998a). Data from
dependent measures (T3, T4, rT3, TSH, and hTg) were subjected to separate two-way ANOVAs,
with gender (male and female), and treatment (dose) as independent, between-subject variables.
Step-down ANOVA tests were conducted as indicated by significant interactions and discussed
in Crofton and Marcus (2001) and Marcus (2001). Mean contrasts were performed using
Duncan's Multiple Range Test. Results of these reanalyses are similar to those stated in the
Caldwell et al. (1995) and King (1995) reports with some notable exceptions. Figure 5-3 shows
the dose-dependent effects on T3, T4, and TSH.
There was a significant gender-by-treatment interaction on total serum T3, and subsequent
step-down ANOVA tests showed significant treatment effects for both genders. Figure 5-3(A)
January 16, 2002
5-21
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 5-3. Effects in the Caldwell et al. (1995) study of 14-day drinking water
administration of ammonium perchlorate to SD rats on serum total T3 (A), T4
(B), and TSH (C) concentrations (ng/mL; mean ± SE) as recalculated in
Table 5-2 (Crofton and Marcus, 2001). Means with different letters were
significantly different (p < 0.05). Data of Channel (1998b) and Crofton
(1998a). Daily dose was estimated from water consumption data.
January 16, 2002
5-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
illustrates dose-dependent decreases in T3 for both genders while females were slightly more
sensitive compared to males. The overall gender-by-treatment interaction was not significant for
T4, but there was a significant main effect of treatment (Figure 5-3(B)). Perchlorate also
decreased T4 in a dose-dependent mariner. There was a significant gender-by-treatment
interaction on total serum TSH, and subsequent step-down ANOVA tests showed significant
treatment effects for both genders. Dose-dependent increases in TSH were observed for both
genders; however, females were slightly more sensitive compared to males.
The Caldwell et al. (1995) study is the only one in which an additional thyroid hormone,
rT3, and hTg were assayed (Tg in rats was assayed with a human RIA kit, thus the notation "h").
There was no significant gender-by-treatment interaction for rT3. Figure 5-4(A) clearly indicates
that perchlorate increases rT3 in a dose-dependent manner. There was a significant gender-by-
treatment interaction on hTg, and subsequent step-down ANOVA tests showed significant
treatment effects for both genders. Figure 5-4(B) illustrates the dose-dependent increases in hTg
for both genders. Both genders were equally sensitive, with males exhibiting a slightly greater
response to the lowest dosage.
Perchlorate exposure decreased circulating T3 and T4 and increased TSH. This report also
provides evidence that rT3, formed mostly in extrathyroidal tissues, was increased by this
exposure. Thyroglobulin also was increased. The NOAELs and LOAELs are summarized in
Table 5-2. A NOAEL for TSH was observed in males only at 0.44 mg/kg-day and at 0.11/0.12
for rT3. Note that free-standing LOAELs (i.e., effects at the lowest dosage tested) were found at
0.11/0.12 mg/kg-day for T3 in females, for T4 and hTg in both sexes, and for TSH in females.
5.2.3 The 90-Day Testing Strategy Bioassay in Rats
The 90-day study that was part of the testing strategy consisted of oral administration of
ammonium perchlorate via drinking water to male and female Sprague-Dawley rats at doses of
0, 0.01, 0.05, 0.2, 1.0, and 10 mg/kg-day (Springborn Laboratories, Inc., 1998). This study has
also been reported in the literature (Siglin et al., 2000), but because that manuscript did not use
the thyroid histopathology as reported by the PWG (Wolf, 2000) it will not be discussed further
in this document. A 14-day sacrifice also was included in the study for comparison with the
Caldwell et al. (1995) study of that same duration. Ten rats/sex/dose were used, and an
additional 10 rats/sex/dose were sacrificed after the 30-day recovery period following cessation
January 16,2002 5-23 DRAFT-DO NOT QUOTE OR CITE
-------
40
35
UJ
CO
+1
c
(6
a;
£ 30
u>
£
c
o
(V
c
0)
u
c
o
o
co
25
20
15
10
0.1 0.4 1.1 2.5 4.5 11 23
Dose (mg/kg-day)
m
CO
+i
c
re
0)
E
o>
c
o
*3
<0
c
0)
o
c
o
o
D)
I-
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.1 0.4 1.1 2.5 4.5 11 23
Dose (mg/kg-day)
Figure 5-4. Effects in the Caldwell et al. (1995) study of 14-day drinking water
administration of ammonium perchlorate to SD rats on serum rT3 (A) and
hTg (B) concentrations (ng/mL; mean ± SE) as recalculated in Table 5-2
(Crofton and Marcus, 2001). Data of Channel (1998b) and Crofton (1998a).
Means with different letters were significantly different (p < 0.05). Daily dose
was estimated from water consumption data.
January 16, 2002
5-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of the 90-day exposure at doses of 0, 0.05, 1.0, and 10 mg/kg-day to evaluate reversibility of any
observed lesions.
The stock solution of the test article was diluted with reverse osmosis (RO) water and
prepared fresh five times during the study (at least once every 5 weeks). Stability analyses were
performed by the sponsor (AFRL/HEST) and showed that ammonium perchlorate solutions were
stable for 109 days (Tsui et al., 1998). The sponsor also confirmed that the stock and dosing
solutions were within an acceptable concentration range (Springborn Laboratories, Inc., 1998;
Appendix B). Control drinking water solutions were analyzed by the sponsor to confirm no
contamination of detectable nitrate, an ion that could cause possible interference to estimating the
dose of test article. Dosing solutions were prepared fresh for each week, and the administered
concentrations were adjusted based on measured body weights and water intake.
The parameters evaluated included clinical observations, body and organ weights, food and
water consumption, hematology, clinical chemistry, ophthalmology, and gross necropsy.
Histopathology was performed on all tissues from the control and high-dose groups. The liver,
kidneys, lungs, thyroid/parathyroid, and gross lesions from all intermediate dose groups and for
the recovery groups also were examined microscopically. Evaluation of additional reproductive
parameters, i.e., estrous cyclicity in females and sperm motility and morphology in males, also
was performed. Thyroid hormone analyses were performed at the 14-, 90-, and 120-day
sacrifices. Only the 0, 0.05, 1.0 and 10.0 mg/kg-day groups were continued until the 120-day
time point. All hormone and tissue collection was balanced over time-of-day to control for the
circadian rhythms of hormones.
5.2.3.1 General Toxicity, Thyroid Histopathology Results, and Satellite
Reproductive Assay
There were no clinical signs of toxicity observed during the treatment or recovery periods.
All rats survived to scheduled sacrifice except one female rat in the 0.05 mg/kg-day group that
was found dead during the recovery period. However, this death was considered unrelated to
treatment because no deaths occurred in any of the higher dose groups, and the histopathologic
evaluation for cause of death was inconclusive. No statistically significant or remarkable
findings were observed among the groups with respect to clinical observations, body weights,
food or water consumption, ophthalmology, hematology, or clinical chemistry. Miscellaneous
January 16, 2002
5-25
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
lesions that occurred with equal incidence and severity in all dose groups and controls included
extramedullary hematopoiesis in the livers, inflammation in the lungs, minimal nephropathy in
the kidneys and inflammation of the heart. Because none of these lesions demonstrated a dose
response, and some are commonly seen in young rats, they were not considered treatment-related.
The only treatment-related lesions observed at gross necropsy were reddened thyroids, attributed
to minimal congestion of the blood vessels.
Absolute thyroid weight and thyroid weight relative to both final body weight and brain
weight were increased significantly in males of the 10 mg/kg-day dose group after 14 and
90 days of treatment and in females at the 10 mg/kg-day dose group after 90 days indicating
LOAEL at 10 and a NOAEL at 1 mg/kg-day. These thyroid weight measures were comparable to
control values in both males and females of the 10 mg/kg-day group at the end of the 30-day
recovery period. Histopathology was evaluated on Days 14, 90, and 30 postexposure (120 days).
The corresponding PWG review results can be found in Wolf (2000; 2001, Tables 3 through 8).
Male rats appeared to be slightly more sensitive, exhibiting follicular cell hyperplasia by Day 14
and not recovering fully for any of the thyroid histopathological indices by 30 days post
exposure. On Day 14, females showed decreased colloid and follicular cell hypertrophy at
10 mg/kg-day. Males also showed a significant increase in these two thyroid response measures
at this dose but also exhibited changes at lower doses and in addition showed hyperplasia.
By 90 days, all three response measures (colloid depletion, follicular cell hypertrophy, and
follicular cell hyperplasia) in both sexes were significant at 10.0 mg/kg-day, again indicating a
LOAEL at 10 and a NOAEL at 1 mg/kg-day. Recovery of the thyroid histopathological changes
was essentially complete by 30 days post-exposure although the males did have some indication
of residual toxicity.
The BMD analyses for these data are found in Table 5-1 and Figures 5-1 and 5-2. Data for
females and males were combined. The BMDL for colloid depletion and hypertrophy at 14 days
were 0.28 and 0.017 mg/kg-day, with no estimate for hyperplasia. By 90-days, the BMDL values
decreased for colloid depletion and hypertrophy to 0.03 and 0.008 mg/kg-day. The BMDL value
for hyperplasia was 2.09 mg/kg-day. No observed effect was estimated for the 120 day value.
Estrous cyclicity was evaluated for 3 weeks prior to sacrifice in all females of the 90- and
120-day termination groups by examining daily vaginal smears. The number and percentage of
females cycling and the mean cycle length were determined for each group. There is an apparent
January 16, 2002
5-26
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
dose-related response for the absolute number and proportion of females with an abnormal
estrous cycle (defined as less than 3 or more than 5 days). The number and percentage of
females with at least one abnormal cycle in those females cycling was 1/10 (10%), 1/10 (10%),
5/9 (56%), 6/9 (67%), 0/8 (0%), and 0/10 (0%) at the 0, 0.01, 0.05, 0.2, 1.0, and 10-mg/kg-day
doses. The proportion began to increase at the 0.05 mg/kg-day dose level, peaked at the
0.2 mg/kg-day dose level, and then declined at the two higher doses. This suggests the
possibility of an inverted U-shaped dose-response pattern. Examination of the 120-day data
(after 30-day recovery) also revealed changes in cyclicity with 1/5 (20%), 1/7 (14%), 1/6 (16%),
and 4/6 (67%) females not cycling in the 0.0, 0.05, 1.0, and 10-mg/kg-day groups, respectively.
Because the number of rats in the add-on groups (n = 10) did not provide the level of statistical
power that would be desired, this indication of an effect in a study with limited power was of
concern in 1998, but the results of the two-generation reproductive study completed in 1999 did
not indicate any effects on this endpoint (Section 5.5.1).
Sperm samples were obtained from all male rats terminated after 90 or 120 days for
evaluation of sperm count, concentration, motility, and morphology. The mean percentage of
normal sperm was calculated for each group. There were no treatment-related effects on sperm
parameters noted although again the number tested is small. The effects on the percentage of
normal sperm appear to be artifacts because of a single outlier in each of the two groups with
lower means. These occurred at different dose levels in the exposure versus recovery phases.
5.2.3.2 Thyroid and Pituitary Hormone Analyses
The assays for T4, T3, and TSH were performed using RLA kits according to the
manufacturer's standard procedures. Assay kits from the same batch number and with the same
expiration date were used for each animal termination period (Study Days 14, 90, or 120).
Samples and standards were run in triplicate. The Springborn Laboratories report included an
appendix (Springborn Laboratories, Inc., 1998; Appendix I) containing the results of these
thyroid hormone assays. The Springborn report used a series of individual ANOVA tests to
determine main effects of treatment for all three hormones in both genders and at three time
points during the study (Day 14, Day 90, and Day 120 a [30-day recovery time]). As part of its
1998 assessment, EPA reanalyzed these thyroid hormone data using three-way ANOVA tests,
one for each of the three hormones, to allow for a statistical comparison of the interaction
January 16, 2002
5-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
between gender, time, and treatment (Crofton, 1998b). The Crofton (1998b) analysis also
contains a printout of all of the individual animal data, an omission from Springborn
Laboratories, Inc. (1998). As suggested in the external peer review (Research Triangle Institute,
1999), EPA reanalyzed these data from each hormone at each time point (Day 14, Day 90, and
Day 120) with two-way ANOVA tests. Gender and treatment (dose) were used as independent
between-subject variables. Dependent variables were T3, T4, and TSH. Step-down ANOVA
tests were conducted as indicated by significant interactions (Crofton and Marcus, 2001; Marcus,
2001). Mean contrasts were performed using Duncan's Multiple Range Test.
Results of the EPA reanalyses, shown in Table 5-2 and illustrated in Figures 5-5 through
5-7, are similar to those stated in the contract report (Springborn Laboratories, Inc., 1998) with a
few notable exceptions. First, there is only a marginal interaction between gender and treatment,
resulting from a slight difference in magnitude of effects between genders. However, no
differences in LOAELs between genders were observed (with minor exceptions likely caused by
small changes in variance between groups, which are probably not biologically significant [see
below]). Results of the analyses for each thyroid hormone and TSH are discussed individually.
There were significant day-by-gender-by-treatment interactions for T3 on Day 14 and
Day 90. Therefore, separate ANOVA tests were conducted on each gender to test for a main
effect of treatment. Lack of a significant gender-by-treatment interaction on the 120-day data led
to one subsequent ANOVA to test for a main effect of treatment. Data from Day 14 revealed a
LOAEL of 0.01 mg/kg-day for males (see Figure 5-5). There was a NOAEL of 10 mg/kg-day for
T3 in females. The low potency of perchlorate on T3 in females at the 14-day time point may be
artifactual. Not plotted on the figure for Day 14 are all the available data from control female
rats from this laboratory, including the Day 90 and Day 120 time points, and the data from two
other studies. These historical data show that the group mean for females in Figure 5-5 for the
14-day time point may be artificially low relative to some of the other data from the AFRL/HEST
laboratory. Thus, the biological significance of this gender-dependent effect of perchlorate after
14-days of exposure is suspect. Consistent with this conclusion is the significant dose-dependent
decrease in T3 concentrations in female rats exposed to 0.125 to 250 mg/kg-day perchlorate in a
previous 14-day exposure study by this same laboratory (Caldwell et al., 1995). The LOAEL for
effects on T3 for both males and females was 0.01 on Day 90. The NOAEL for effects on T3 at
January 16, 2002
5-28
DRAFT-DO NOT QUOTE OR CITE
-------
Ill
in
+i
c
(U
-------
LD
CO
+1
(U
a>
E
O)
6.0
5.5.-
5.0
4.5
4.0
3.5
3.0
6.0
Day 14
5 5
5 0
4.5 -
4.0 -
3.5
3.0 -
2.5 -
2.0
6 0
5 5 -
5 0 -
4 5 -
4.0 -
3.5 -
3.0 -
2.5 -
2.0
Day 90
Male
Female
Day 120
Male
Female
0 0.01 0.05 0.2 1 10
Perchlorate (mg/kg-day)
Figure 5-6. Effects from 90-day drinking water administration of ammonium perchlorate
to SD rats on serum total T4 concentrations as recalculated in Table 5-2
(Crofton and Marcus, 2001). Means with different letters were significantly
different (p < 0.05). The 120-day time point is 30 days after cessation of
exposure.
January 16, 2002
5-30
DRAFT-DO NOT QUOTE OR CITE
-------
CO
+1
as
Cl)
£
~OJ
CO
24 -
22 -
20
18 -I
16
14
12 -
10
b
bf
a
'r i
Day 120
0 0.01 0.05 0.2 1
Perchlorate (mg/kg-day)
10
Figure 5-7. Effects from 90-day drinking water administration of ammonium perchlorate
to SD rats on serum total TSH as recalculated in Table 5-2 (Crofton and
Marcus, 2001). Data of Springborn Laboratories, Inc. (1998). A main gender-
by-treatment interaction was observed for Day 14, but not Days 90 and 120;
therefore, data are presented separately for males and females on Day 14 and
collapsed across gender for Days 90 and 120. Means with different letters were
significantly different (p < 0.05). The 120-day time point is 30 days after
cessation of exposure.
January 16, 2002
5-31
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Day 120 was 10 mg/kg-day, indicative of a recovery of T3 concentrations after cessation of
treatment.
There were significant day-by-treatment interactions for effects on T4 at the 90- and
120-day time points but not at the 14-day time point. Mean contrast tests for Day 14 data
revealed a free-standing LOAEL of 0.01 mg/kg-day for effects on T4 in both sexes. The
0.01 mg/kg-day dosage was also a free-standing LOAEL on Day 90 for effects on T4 in both
sexes. Analysis of the data from the 30-day recovery period (the Day 120 time point) revealed a
free-standing LOAEL of 0.05 mg/kg-day in males and a NOAEL of 1.0 mg/kg-day in females for
effects on T4.
There was a significant day-by-gender-by-treatment interaction for TSH only on Day 14.
Therefore, separate ANOVA tests were conducted on each gender to test for a main effect of
treatment for the Day 14 time point. Lack of a significant gender-by-treatment interaction for the
90- and 120-day data led to subsequent one-way ANOVA tests at each time point to test for a
main effect of treatment. Perchlorate caused a dose-dependent increase in TSH that was apparent
at the Day 14 and Day 90 time points (see Figure 5-7). The NOAEL for effects on TSH at
Day 14 data was 0.01 mg/kg-day in the males. The 0.01 mg/kg-day dose was a free-standing
LOAEL in the females. This small difference between males and females likely is caused by
small changes in variance between groups rather than by a biologically significant difference (the
absolute increase relative to the control mean in the 0.05-mg/kg-day female group is actually
smaller than the same comparison in the males). The TSH concentrations did not recover to
control values 30 days after cessation of treatment with a free-standing LOAEL at 0.05 mg/kg-
day in both sexes.
The data demonstrate a dose- and time-dependent effect of perchlorate on thyroid hormones
and TSH. There was no LOAEL established in this data set due to multiple effects at the lowest
dose of 0.01 mg/kg-day. There was some evidence of recovery at the Day 120-evaluation
(30 days after cessation of treatment). The NOAEL for effects on T3 increased to 1.0 mg/kg-day.
However, the omission of the 0.01 mg/kg-day dose group at the 120-day time point make it
difficult to conclude about a recovery of effects on T4 and TSH.
January 16, 2002
5-32
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
5.3 DEVELOPMENTAL NEUROTOXICITY STUDIES
Concern for potential neurodevelopmental sequelae was warranted given the established
mode of action for perchlorate, and the original 1997 testing strategy included a developmental
neurotoxicity study (Argus Research Laboratories, Inc., 1998a). Results of that study raised
additional issues and concerns so that the external peer review convened in 1999 recommended
additional testing. This section describes results of the available studies that tested
neurodevelopmental indices per se. The 1998 neurodevelopmental study is reviewed in
Section 5.3.1. Results of the new study on motor activity are reviewed in Section 5.3.2. The
"Effects Study" repeated the study of brain morphometry as a measure of neurodevelopmental
toxicity and is reviewed in Section 5.3.3.
5.3.1 The 1998 Developmental Neurotoxicity Study
The neurobehavioral developmental study of ammonium perchlorate that was part of the
original 1997 testing strategy was performed by drinking water administration in Sprague-
Dawley rats (Argus Research Laboratories, Inc., 1998a). A schematic of this study design is
provided as Figure A-l (Appendix A) of this document to aid understanding of terminology and
the protocol. It should be noted that Argus Laboratories identifies the day of birth as PND1;
therefore, the age of PND10 and PND22 actually correspond to PND9 and PND21 in this study.
The description of the study design will use the Argus nomenclature in order to readily compare
with the contract report. Subsequent supplemental data submittals and additional analyses
pertaining to this were requested by EPA and provided by Argus Laboratories study (York,
1998a,b,c,d,e).
Female rats (25/dosage group) were administered target doses of 0, 0.1, 1.0, 3.0, and
10 mg/kg-day by continual access to ammonium perchlorate in nonchlorinated RO deionized
water beginning on gestation day zero (GD0) and ending at scheduled sacrifice. Test substance
concentrations were evaluated weekly, based on actual water consumption levels recorded the
previous week and adjusted as necessary to more closely achieve the target dose levels. Test
solutions were prepared weekly. The stability of the stock solution and that concentrations
agreed well with nominal concentrations were determined by AFRL/HEST (Argus Research
January 16, 2002
5-33
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Laboratories, Inc., 1998a; Appendix J). Feed and water consumption were recorded daily during
exposure.
After acclimation for 14 days, virgin female rats were cohabited with breeder male rats
(one male rat per female rat) for a maximum of 7 days. Female rats with spermatozoa observed
in a vaginal smear or a copulatory plug observed in situ were considered to be at GDO. The
FO-generation dams were examined at approximately the same time each day during the exposure
period for signs of maternal behavior, autonomic dysfunction, abnormal postures, abnormal
movements or behavior patterns, and unusual appearance. Pregnancy outcome measures
evaluated at birth included pregnancy rate, duration of gestation, number of implantation sites,
gestation index (number with live pups/number pregnant), number of pups/litter, sex ratio of
pups, and viability and lactation indices. Maternal body weight was recorded on GDO, daily
during the exposure period, weekly during the post-weaning period, and at sacrifice. The same
set of signs as examined during exposure were evaluated on a weekly basis during post-weaning.
Thyroids from all FO-generation rats were weighed and evaluated histologically. Five dams per
group were selected for sacrifice and blood collection on post-natal day 10 (PND10) from those
with no surviving pups or with litters of less than eight pups. Thyroid and pituitary hormone
analyses (T3, T4, and TSH) were done on the blood (see Section 5.3.1.3). All dams not selected
for continued observation were sacrificed on PND22.
Pups (F1-generation) were counted and clinical signs were recorded once daily during
pre-and post-weaning. Body weight was recorded on PNDs 1, 5, 8, 12, 14, 18, and 22 and then
weekly during post-weaning. Feed consumption values were recorded weekly during
post-weaning. Pups that appeared stillborn and those that died before initial examination on
PND1 were examined for vital status, and the gross lesions were preserved. Pups that were not
selected for continued observation were sacrificed and necropsied on PND5. Blood was sampled
for thyroid and pituitary hormone analysis, and the thyroids were examined histologically. The
F1-generation pups not selected for continued observation on PND10 (n = 102) were sacrificed
and examined for gross lesions. Post-weaning pups that were selected for continued observation
were given ammonium perchlorate in RO deionized water with chlorine (added at a maximum of
1.2 ppm as a bacteriostat).
Other pups (F1-generation) were assigned to four different subsets for additional
evaluations. The first male and female pup (1/sex/dose; total of 97 male and 100 female pups)
January 16, 2002
5-34
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
were assigned randomly to Subset 1 for brain weight and neurohistological examination
(including morphometric measurements). All pups were selected for fixed brain weights on
PND12; 6/sex/dose (total of 30 male and 30 female pups) were selected for neurohistological
examination. The second male and female pup (1/sex/dose; total of 100 male and 100 female
pups) were assigned randomly to Subset 2 for passive avoidance testing on PNDs 23 to 25 and
PNDs 30 to 32; water maze testing on PNDs 59 to 63 and PNDs 66 to70; and scheduled sacrifice
at PNDs 90 to 92, with blood collection for thyroid and pituitary hormone analysis. The third
male and female pup (1/sex/dose; total of 100 male and 100 female pups) were assigned
randomly to Subset 3 for motor activity evaluation on PNDs 14, 18, 22, and 59; auditory startle
habituation on PNDs 23 and 60; and scheduled sacrifice on PNDs 67 to 69. The fourth male and
female pup (1/sex/dose; total of 100 male and 100 female pups) were assigned randomly to
Subset 4 for regional brain weight evaluation on PNDs 81 to 86 (6/sex/dose; total of 30 male and
30 female rats) and neurohistological examination on PNDs 82 to 85 (6/sex/dose; total of
30 male and 30 female rats). Female pups also were evaluated for the age of vaginal patency
beginning on PND28, and male pups were evaluated for the age of preputial separation beginning
on PND39. A few of these measurements inadvertently went unrecorded, but the laboratory
asserted that this did not affect the results because a sufficient amount of data on other rats was
recorded.
5.3.1.1 Results of General Toxicity Measures, Neurohistology, and Morphology
Results in the dams (FO-Generation) revealed no treatment-related effects on food or water
consumption (Argus Research Laboratories, Inc., 1998a; Appendix B, Tables B7 through B14),
mortality (Appendix B, Tables B2 and B18), clinical signs (Appendix B, Table B2), necropsy
(Appendix B, Table B18), body weight (Appendixes A and B, Figure A1 and Tables B3 through
B6), or pregnancy outcome measures (Appendix B, Tables B15 through B16). Effects on thyroid
weight, histopathology, and thyroid and pituitary hormone analyses will be discussed below in
Sections 5.3.1.2 and 5.3.1.3.
Results in the pups (F1 -generation) revealed no treatment-related effects on feed
consumption (Argus Research Laboratories, Inc., 1998a; Appendix C, Tables C18 and C19),
mortality (Appendix C, Tables CI and C2), clinical signs (Appendix C, Tables CI and C2), body
weight (Appendixes A and C, Figures A2 and A3 and Tables C3 through C6), or sexual
January 16, 2002
5-35
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
development landmarks (Appendix C, Table CI 1). No treatment-related effects were observed
on mortality, brain weight, or body weight in the pups of Subset 1 at PND12 (Argus Research
Laboratories, Inc., 1998a; Tables D1 and D2), Subset 2 at PNDs 90 to 92 (Tables E3 and E4), or
Subset 3 at PNDs 67 to 69 (Tables F5 and F6). Results of the neurobehavioral tests from
Subsets 2 and 3 will be discussed in Section 5.3.1.4.
In the Subset 1 subgroup subjected to neurohistological examination (the F1 pups sacrificed
on PND12), morphometric analyses revealed a 23.4% increase in the size of the corpus callosum
in females and a 30.2% increase in males (not significant) at the high dose (10 mg/kg-day).
Slight decreases in brain weight also were noted at the highest dose in females. In Subset 4 (the
F1 pups sacrificed on PND82), there was a continued effect on the size of the corpus callosum
(20.9% increase) in males, but no effect in females at the highest dose. There was also a 3.4%
increase in the brain weight in males and increases in the size of the frontal cortex (9.2%) and the
caudate putamen (10.2%). The EPA concluded that the effects may be significant and that
analyses of the next lower dose (or, at least, historical control data for the affected endpoints)
were warranted and requested additional analyses from the sponsor (PSG). York (1998d)
responded with morphometry analyses of the next lower dose (3.0 mg/kg-day) of the Subset 1
F1 pups at PND12. The new analysis noted, in addition to previous findings, a statistically
significant increase in the anterior/posterior cerebellum size, a statistically significant decrease in
the caudate putamen for the F1 PND12 female pups, and a statistical significant decrease in the
hippocampal gyrus size for the F1 PND12 male pups. These effects were not considered
treatment-related by the Primedica/Argus pathologist because they were not dose dependent.
A preliminary reanalysis by EPA (Crofton, 1998c) of the control, 3- and 10-mg/kg-day
groups (York 1998d) was restricted to the corpus callosum because this was the area with the
largest effect. The analysis revealed no interaction of gender and treatment; however, there was a
significant effect of treatment (F[2,30] = 7.65, p < 0.0021). There was a significant increase in
the size of the corpus callosum only in the 10-mg/kg-day group. Group means were 288, 278,
and 366 /um for the controls and 3- and 10-mg/kg-day groups, respectively. Incorporation of
historical control data from both PND10 and PND12 (mean for controls = 264 /u.m for PNDs 10
and 265 fu.m for PND12; York, 1998a) supports the conclusion that the control values for corpus
callosum size in the data set are within the "normal" range (York 1998a; see also Argus Research
Laboratories, Inc., 1998a).
January 16, 2002
5-36
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
EPA did not agree with the argument put forth by Argus Research Laboratories, Inc.
(1998a) that these effects were "not suggestive of a neurotoxic effect" because of "an unknown
biological significance." EPA considers a 27% increase in the size of any brain region to be a
potentially adverse effect (U.S. Environmental Protection Agency, 1998e), and designated
10 mg/kg-day as the LOAEL and the NOAEL at 3 mg/kg-day for these changes in brain
histology. No additional evaluation of the brains from the neurohistological examination of
Subset 4 pups (PND82 to PND85) were ever submitted to EPA although it was suggested again
that the next lower dose group be analyzed because of the significant increases in brain weights
and in the frontal cortex and corpus callosum measurements for the males in the high-dose group.
Additional analyses of the brain morphometry were provided by the EPA at the 1999
external peer review (Geller, 1999a) that corroborated the preliminary finding of Crofton
(1998c). The data were analyzed using a 2-way ANOVA, with dose and sex as independent
variables. To correct for multiple comparisons, the acceptable alpha for significance (for all
interaction main effects) was corrected to 0.016 (alpha of 0.05 divided by the square root of the
number of ANOVA tests).
Significant effects of dose were found in corpus collosum, hippoacampal gyrus, anterior
and posterior cerebellum, and caudate putamen. An effect of sex was also found in caudate
putamen. The effect on corpus callosum was confirmed and showed an increase in size at the
10 mg/kg-day dose. Hippocampal gyrus (12% less than control) and caudate putamen (7.3% less
than control) showed a decrease in size at the 3 mg/kg-day dose, with no significant difference
between control and high dose, yielding a U-shaped dose response. The anterior and posterior
cerebellum showed a significant increase in size at the 3 mg/kg-day group (13%).
Because of concern for this effect voiced at the 1999 external peer review, the blocks of
brain tissue were evaluated to determine if they could be refaced and additional sections
evaluated. It was determined that the remaining materials were of insufficient quality for
additional sectioning and histological evaluation (Harry, 2001). As an alternative, brain
morphometry measurements were included in the "Effects Study", described below in
Section 5.3.3, to determine if the alteration in brain morphometry could be repeated.
January 16, 2002
5-37
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
5.3.1.2 Evaluation of Thyroid Histopathology
Appendix O of the Argus Research Laboratories, Inc. (1998a) neurodevelopmental study
presents thyroid histopathology data provided by the sponsor (AFRL/HEST). Note that the data
analyzed by EPA in the 1998 document for PND5 F1 -generation rat pups are from the final
report for the PND5 time point (Channel, 1998c). Channel (1998c) reported that the decrease in
follicular lumen area in these pups at PND90 to PND92 showed no significant differences
between dose groups and controls for either females or males based on t-test or Mann-Whitney
Rank Sum Test (M-W RST). These data suggest a recovery from the effects observed in the
thyroids of the pups at PND5.
The report also contained measurements, performed by Dr. William Baker of AFRL/HEST,
of both follicular epithelial cell height and the follicular lumen diameter. These data were
subsequently formally transmitted to EPA by consultative letter (Channel, 1998c) in Microsoft
Excel® spreadsheets. For the final morphometric study (Channel, 1998c), the arbitrary decision
based on ease of detection of this region in digitized images was made by Dr. William Baker to
focus on only a lumen area measurement because of time constraints (Jarabek, 1998). The mean
follicular lumen area represents the mean area of all follicular lumens measured from the three
histological sections sampled from each rat and is expressed in microns. In the opinion of
Dr. Charles Capen of Ohio State University (Crofton, 1998d), the measurement of follicular
height is usually more sensitive than those of follicle diameter and lumen area. In support of this
opinion, data collected by Dr. Baker (Argus Research Laboratories, Inc., 1998a; Appendix O)
demonstrated significant increases in males rats in the incidence of follicular epithelial cell
hypertrophy at doses much lower than those doses that increased the incidence of decreased
lumen area. The difference observed between standard histopathology as originally reported by
Argus Research Laboratories, Inc. (1998a) and the thyroid morphometry performed by Dr. Baker
was analyzed extensively by the EPA in its 1998 assessment. The results indicated that the
morphometry performed on lumen size was a less sensitive measure of thyroid histopathology.
The analyses of the thyroid morphometry are retained in this reassessment; whereas, the PWG
review results will be presented below for the histopathology.
Data from the dependent measure (follicle lumen size) based on the morphometric analyses
(Channel, 1998c) were available for pups sacrificed at ages PND5 and PND90. These data were
reanalyzed by EPA (Crofton and Marcus, 2001). Because there was only one block of animals at
January 16, 2002
5-38
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
PND90 compared to two blocks of data at PND5, and because the slides for PND90 were
processed at a much later time, the data for the two ages were analyzed separately. Data from
PND5 pups were subjected to three-way ANOVA tests with gender, treatment (dose), and block
(two separate analyses of separate blocks of data) as independent between-subjects variables.
Data from PND90 were subjects to a two-way ANOVA with gender and treatment (dose) as
independent between-subjects variables. Step-down ANOVA tests were conducted as indicated
by significant interactions and recalculated by Crofton and Marcus (2001) and Marcus (2001).
Mean contrasts were performed using Duncan's Multiple Range Test. Note that in the Crofton
and Marcus (2001) memorandum the 0.1 mg/kg-day dose is incorrectly labeled as 0.3 mg/kg-day.
There was a significant main effect of treatment on lumen size for all doses at PND5, resulting in
a free-standing LOAEL of 0.1 mg/kg-day. The data are plotted in Figure 5-8. There was no
significant effect of perchlorate on lumen size at PND90.
The thyroid histopathology as reviewed and reported by the PWG can be found in Wolf
(2001; Tables 9 through 13). This report includes corrections for slides sent to EPA that
contained animals with autolysis and those necropsied at different times than indicated for the
study protocol or to exclude dams that did not have litters.
The F0 generation dams (Wolf, 2001: Table 9) exhibited decreased colloid and increases
in both hypertrophy and hyperplasia. A clear dose-response was not evident, however, with the
possible exception of colloid depletion at levels above 0.1 mg/kg-day.
Thyroid histopathology in the pups on PND4 (Wolf, 2001: Tables 10 and 11) was more
pronounced, with colloid depletion and increases in hypertrophy at 0.1 and 3 mg/kg-day.
Hyperplasia appeared to be effected at 3 mg/kg-day. The BMD analyses presented in Table 5-1
support these levels with BMDL estimates for colloid depletion at 0.33, increased hypertrophy at
0.88, and increased hyperplasia at 3.62 mg/kg-day. These results were obtained with a
constrained model, but an adequate fit is obtained by fitting the model without restricting the
exponent on dose to be £ 1 and results in a BMDL for pups on PND4 in this study at 0.009 for
colloid depletion (Geller, 2001a).
The argument for the lack of biological plausibility of unrestricted functions is based on
cancer modeling theory from the early 1960s (Mantel and Bryan, 1961) that attempted to derive a
default procedure for modeling tumor data at the time when cancer was thought to be a one-stage
process and many bioassays used only 1 dose and control. Given the increased sophistication of
January 16, 2002
5-39
DRAFT-DO NOT QUOTE OR CITE
-------
340
111
(f)
+1
C
ro
v
E,
0)
N
'55
c
d>
UJ
to
+1
c
ro
a>
E_
«
N
'<75
c
0)
E
3
320
300
280
260
240
220
200
180
PND5
o 1
10
0.1 1 3 10
Perchlorate (mg/kg-day)
Figure 5-8. Effects from maternal drinking water administration of ammonium
perchlorate to SD rats on thyroid gland follicular lumen size in Fl-generation
offspring on PND5 as recalculated in Crofton and Marcus (2001). Data of
Channel (1998c) and Argus Research Laboratories, Inc. (1998a). Means with
different letters were significantly different (p < 0.05). Daily dose was
estimated from water consumption data.
January 16, 2002
5-40
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
contemporary bioassays and the level of organization at which effects are now being identified
(i.e., precursor events at the cellular and molecular levels), Hasselblad et al. (1995) have argued
that restricting the slopes of fits to the data prioritizes mathematical convenience over fitting the
data. The thyroid hormone data show exquisite sensitivity to very low doses of perchlorate. This
suggests that models fit with nonsupralinear slopes and lower doses need to be tested. It is
interesting to note that PWG results for colloid depletion are very similar to the 1998 EPA
analysis on the previous histopathological read by Argus Laboratories, Inc. (1998a) for
hypertrophy/hyperplasia that resulted in a BMDL of 0.1 mg/kg-day.
Histopathology in the animals from PND90 and PND92 (Wolf, 2001: Tables 12 and 13)
indicated variable effects on colloid depletion, hypertrophy, and hyperplasia. As indicated in
Table 5-1, a BMDL was only calculated with confidence for colloid depletion with a resultant
estimate of 0.03 mg/kg-day.
Evaluation of the histopathology in this study indicate that the pups are the most sensitive
with a BMDL between 0.009 and 0.33 mg/kg-day.
5.3.1.3 Thyroid and Pituitary Hormone Analyses
Serum was collected and thyroid hormone analyses performed as part of the
neurodevelopmental study (Argus Research Laboratories, Inc., 1998a; Crofton, 1998f)). The
following is a statistical analysis of the thyroid and pituitary hormone data (T4, T3, and TSH)
found in that report (Crofton and Marcus, 2001). At the time of this assessment, individual
animal data were available from both the F1-generation pups (male and female samples were
pooled for each litter) on PND5 and the F0 generation (parents) on post-partum Day 10 (PP10).
Only the F1 data were reanalyzed because of the very limited (n = 2 to 5/group) data for the
parental F0 PP10 group.
All data were supplied in Microsoft Excel® spreadsheets via E-mail by Dr. David Mattie
(AFRL/HEST). Data for dependent measures (T4, T3, and TSH) were subjected to separate one-
way ANOVA tests. Treatment (dose) was used as the independent, between-subjects variable.
Mean contrasts were performed using Duncan's Multiple Range Test.
There were significant main effects of treatment for all the hormones. The data are plotted
in Figure 5-9. Results of these reanalyses are similar to those stated in the report (Argus
Research Laboratories, Inc., 1998a). There was a significant decrease in both T3 and T4, as well
January 16, 2002
5-41
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 5-9. Effects from maternal drinking water administration of ammonium
perchlorate to SD rat Fl-generation pups on serum total T3 (A), T4 (B) and
TSH (C) concentrations (ng/dL; mean ± SE) as recalculated in Table 5-2
(Crofton and Marcus, 2001). Data of Argus Research Laboratories, Inc.
(1998a). Means with different letters were significantly different (p<0.05).
Daily dose was estimated from water consumption data.
January 16, 2002
5-42
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
as the expected increase in TSH. The NOAEL for the effects of perchlorate on T3, T4, and TSH
are 0.1, 0.1, and 3.0 mg/kg-day, respectively. These results are consistent with the known
mechanism-of-action of perchlorate (inhibition of thyroid hormones). The increased TSH is
likely a result of the activation of the pituitary-thyroid feedback mechanism.
5.3.1.4 Behavioral Evaluations
The 1998 EPA review of the behavioral evaluations performed on Subset 3 pups agreed
with the Argus Research Laboratories, Inc. (1998a) report with one exception regarding an
increase in motor activity in male rats on PND14 that no perchlorate-induced changes were
detected in any of the other behavioral indices (i.e., passive avoidance, water maze, auditory
startle). The EPA disagreed with the Argus Research Laboratories, Inc. (1998a) report and
subsequent submissions (York, 1998a,b,c,d,e) with regard to the significance of the motor
activity changes.
The data originally were analyzed using two separate three-way ANOVA tests (age,
treatment, and habituation block), one for each gender (Argus Research Laboratories, Inc.,
1998a). This analysis demonstrated a significant decrease in the amount of habituation in the
two highest dose groups on PND14 in the male pups. There were no changes detected at any
other ages (i.e., PND18, PND22, PND59). On initial review by EPA, it was recommended to the
sponsor (PSG) that an additional analysis of the data be conducted using gender as a
within-subject variable, or alternatively, to use a nested design with gender nested under litter
(see Holson and Pearce [1992] and Cox [1994], for a review of statistical methods used in
developmental studies and the importance of using litter as the unit of measure). The EPA also
questioned why the method or statistics did not detect significance for the dose-dependent
increase in total session counts that amounted to a 95% increase over controls in the highest
dosage group (see Figure 5-10). The response from Argus Laboratory (York, 1998b) included a
new analysis in which gender was used as a between-subjects variable. No interactions with, or
main effects of, treatment were found in this analysis.
EPA remained concerned that Argus Research Laboratory and the sponsor (PSG) failed to
respond adequately to the request for an explanation of why the analysis failed to detect
significance in the PND14 motor activity for the male rats. Figure 5-10 illustrates the clear
dose-dependent increase in two different measurements of motor activity: (1) time-spent-in-
January 16, 2002
5-43
DRAFT-DO NOT QUOTE OR CITE
-------
500
LD
(/)
+1
C
cs
a>
E
o
<
450
400
350
300
250
200
150
100
50
0
-Q~ Movements
Time
0 0.1 13
Dosage (mg/kg-day)
10
Figure 5-10. The effects of developmental exposure to perchlorate on motor activity in
male rats on PND14. Data of Argus Research Laboratories, Inc. (1998a).
The dose-dependent increases in both number of movements and time spent
in movement were not statistically different, even though the increases were
substantial at the higher dosages.
1 movement ("time") and (2) total number of movements ("movements"). The time variable
2 increased over 95% at the highest dose relative to controls (group means of 363 and 186,
3 respectively). The number-of-movements variable increased approximately 65% relative to
4 controls. Expert opinion of EPA neurotoxicologists was sought, and it was their opinion that
5 increases in motor activity over 50%, especially in developing animals, were clearly of concern
6 from a biological perspective (Crofton et. al., 1998). The critical issue for evaluation of these
7 motor activity data was how to resolve the difference between what is a clearly a biologically
8 significant alteration in behavior with a lack of statistical significance. In an attempt to resolve
9 the issue, EPA also requested positive control data from the testing laboratory for this device that
January 16, 2002
5-44
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
was not provided in the original report, as well as any available historical control data. York
(1998a) replied with a number of positive control studies and a limited amount of historical
control data from PND14 pups.
The positive control data were requested to help understand the sensitivity of the device in
detecting increases in motor activity (i.e., what is the smallest increase in motor activity that has
been detected by this device). Unfortunately, the positive control data were of limited use in
interpreting the sensitivity of the device. The submission (York, 1998a) contained data from
experiments with amphetamine and triadimefon in adult rats. The smallest increase in activity
that was induced by either chemical was a 109% increase relative to controls. Although these
effects were statistically significant, they are greater than the effects produced by the highest
dosage of perchlorate in the PND14 pups. There were also positive control data from
chlorpromazine-treated animals that showed significant decreases (s: 32%) in activity. However,
ability to detect decreases does not necessarily translate to the detection of increases.
The historical control data from PND14 rats were requested to help understand the
variability normally found in control animals. Unfortunately, the historical control data
submitted were only useful in that the data raised more suspicion that the degree of experimental
control over this behavior by the testing facility was inadequate. For the time data, the control
mean for the perchlorate data set was 186 sec. For the three relevant historical control data sets,
the means were 1026, 965, and 458 sec. Either the lab had very little control over the behavior,
or the data were from a different test apparatus or from a different usage of the same apparatus.
In any case, the data were of no use in helping EPA determine the historical profile of control
animal behavior in this test apparatus.
In lieu of the absence of useful positive control and historical control data, EPA was left
with the issue of ascertaining statistical versus biological significance. There were a number of
reasons for the lack of statistical significance. The first reason was the extremely large within-
group variability exemplified by coefficients of variation (CV) greater than 100%. It was the
opinion of Crofton et al. (1998) that this was likely caused by the inability of the testing
laboratory to gain adequate control over the behavior being tested. This large variability results
in very little statistical power and increases the potential for Type II errors. Normally, an
increase in sample size (by additional testing) allows for adequate power to refute or support the
conclusion of an effect. Given the CVs of about 100%, simple power calculations (see Cohen,
January 16, 2002
5-45
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1987) for detecting a 40% change in one group out of five results in needed group sizes of about
70 to 90 animals per group. The second reason was that the effect, a 95% increase, while rather
large from a biological perspective, occurs in only one gender on only 1 day out of 4 test days.
The large variability coupled with the complicated design (treatment, age, gender, and block)
would tend to mask anything other than extremely large effects. This conclusion is consistent
with the content of a phone conversation (Crofton, 1998g) with Dr. Simon Mats. Dr. Mats was
the statistician from the contract laboratory (Primedica/Argus) who conducted the revised
statistical analysis of these data. Lastly, the effect seen in the males on PND14 may indeed be a
Type I error and may not be found again if this experiment was repeated.
The assignment of biological significance to the effect seen was supported by both the
underlying mode of action of perchlorate and the effects of other chemical and physical insults on
the motor activity of post-natal rats. The hypothesis that a thyrotoxic chemical would induce a
delay in any aspect of nervous system development is highly plausible. A delay in the onset of
habituation would be evidenced by an increase in overall counts, as well as a decrease in the rate
of a habituation (Ruppert et al., 1985a,b). This delay could be quite transient. Other agents that
interfere with thyroid hormones during development are known to induce delays of a few days
magnitude in developmental landmarks such as eye opening (Goldey et al., 1995a,b). This is the
type of effect seen on PND14 in the Argus Research Laboratories, Inc. (1998a) report.
Developmental exposure to numerous hypothyroid-inducing agents (e.g., propylthiouracil,
methimazole) are known to result in delays in the ontogeny in many behaviors (cf., Comer and
Norton, 1982; Goldey et al., 1995a,b; Schneider and Golden, 1986; Tamasy et al., 1986),
including the development of habituation. However, effects of these chemicals on total motor
activity counts vary from increased to decreased, depending on the chemical and age of testing.
Rice (2000) has noted parallels between the features of attention deficit hyperactivity disorder
(ADHD) and the behavior of monkeys exposed to polychlorinated biphenyls (PCBs). The
mechanism for the gender-dependent nature of the effect of perchlorate also remains to be
determined. In addition, there are numerous reports from the literature that support the biological
significance of a 40 to 50% increase in motor activity in postnatal rats (cf., Campbell et al., 1969;
Ruppert et al., 1985a,b).
In summary, EPA maintained that the increase in activity should be considered biologically
significant until additional data could be marshaled to suggest or prove otherwise. The
January 16, 2002
5-46
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
inadequacy of standard parametric statistics to detect a significant difference suggested that
alternative analyses should be used on these data, such as the benchmark approach. This type of
statistical approach may be useful because of the inverse relationship between the data variability
and the benchmark dose (BMD). The BMDL estimates were calculated for data on the
movement (number of movements) and time (time spent moving) measures from the motor
activity test from PND14 pups. These data were fit by a linear function with fairly shallow slope,
yielding BMD estimates for movement and time of 1.94 and 1.33 mg/kg-day and BMDL
estimates of 1.04 and 0.66 mg/kg-day, respectively. These BMD and BMDL estimates could
serve as estimates of LOAEL and NOAEL for this data set. The estimates are in accord with
doses with activity values that may have emerged as significantly different from control had the
data set not had its unusually high variability. These BMD analyses bring the motor activity
NOAEL more within the range of the T3 and T4 NOAEL and below that for TSH.
5.3.2 Motor Activity Study (Bekkedal et al., 2000)
In response to recommendations at the 1999 peer review for an additional study, the United
States Navy (USN) performed a study that included evaluation of motor activity in Sprague
Dawley rats of both sexes (Bekkedal et al., 2000). Female Sprague-Dawley rats were dosed with
ammonium perchlorate for two weeks at 0, 0.1, 1.0, 3.0 or 10.0 mg/kg-day prior to mating with
the breeder males and through PND10. PND1 was counted as the day when the first pup was
observed in the cage. All pups within a litter were weighed on PND5 when the litters were
culled to eight pups of 4 males and 4 females or as close as possible to that combination. Pups
and dams from any litters with less than 8 pups were eliminated. On PND14, one male and one
female were randomly selected from each litter to be used in the motor activity testing. These
same animals were tested on PND14, PND18 and PND22. Nine different measures of motor
activity were automatically recorded using Opto-Varimex activity meters at ten minute intervals.
The measures included: frequency and time of ambulatory movements, frequency and time of
sterotypic movements, frequency of movements in the horizontal plane, distance traveled in the
horizontal plane, frequency of rears, total number of horizontal movements made while in the
rearing position (vertical plane movements), and time spent resting.
Bekkedal et al. (2000) analyzed each of the nine measures of motor activity separately
using a univariate repeated-measures ANOVA. The between subjects variable was perchlorate
January 16,2002 5-47 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
dose, with 5 levels. The three within-subject variable were sex (2 levels), age (3 levels), and time
block (9 levels). Due to violation of the sphericity assumption, the Greenhouse-Geisser test was
employed with a fiducial limit set at p < 0.05. No statistically significant differences were found
for the main effect of perchlorate exposure for any of the 9 measures nor any reliable interactions
related to dose. The authors do note, however, a general pattern of dose-dependent changes in
the later sessions (90-minute). They also note that this pattern, as in the previous Argus
Laboratories, Inc. (1998a), suggest that exposed pups have a slightly slower rate of habituation
and thus maintain a higher level of activity as compared to untreated pups. Additional follow-up
tests were suggested.
5.3.2.1 EPA and NIEHS Statistical Analyses of Motor Activity Effects
Because EPA was concerned about effect on motor activity in the original study and it
appeared that a similar pattern of effects was again emerging in the study repeated by Bekkedal
et al. (2000), EPA requested that NIEHS perform a statistical evaluation that could formally
integrate the various measures together as well as statistically compare the two studies with each
other (Dunson, 2001a). A Bayesian hierarchical model (Gelfand et al., 1990) was chosen to
assess the weight of evidence of a dose-response trend in motor activity. A linear mixed-effects
regression model (Laird and Ware, 1982) related dose, sex, age, habituation time and a
habituation time x dose interaction term to the expected number of ambulatory movements, with
an animal-specific intercept included to account for within-animal dependency. To complete a
Bayesian specification of the model, a vague (or uninformative) but proper prior distributions for
each of the unknown parameters was chosen. In particular, the prior for the parameters that
related dose to motor activity was centered on a value corresponding to the null hypothesis of no
effect of perchlorate. The model was fit using BUGS, a widely-used software package for
Bayesian analyses (Gilks et al., 1994).
The analyses were conducted under a variety of different choices of prior variance for the
dose parameters and prior means and variances for the other parameters in the model. The dose
level associated with a 10% increase in the number of ambulatory movements by inverse
estimation (refer to Appendix A in Dunson, 2001a). The choice of 10% as the benchmark level
is consistent with standard practice for dichotomous outcomes. The 5% level often used for
January 16, 2002
5-48
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
continuous outcomes was judged to be too low for measuring a biologically significant increase
in motor activity. Conclusions were consistent across the analyses.
As noted by Bekkedal et al. (2000), the effect of ammonium perchlorate on the number of
ambulatory movements was found to increase significantly with habituation time (posterior
probability = 0.98). In the first habituation interval there was modest evidence of an increase in
motor activity with dose (posterior probability = 0.79), while in the final interval there was clear
evidence of an increase in motor activity with dose (posterior probability > 0.99). The posterior
density for the expected increase in the logarithm of the number of ambulatory movements at the
final habituation time per unit (mg/kg-day) increase in dose of ammonium perchlorate is plotted
in Figure 5-11 for the USN study (Bekkedal et al., 2000). The posterior density is centered on a
positive slope and assigns low probability to a negative slope, suggesting a clear increase in
motor activity with dose. The dose estimated to increase the mean number of ambulatory
movements at the final habituation time by 10% is 1.62 with a 95% credible interval of (0.90,
7.87). There was no evidence of an interaction between age and dose, nor of any effect of
gender.
The previous study of Argus Laboratories, Inc. (1998a) was also analyzed in this fashion
and results were very similar (Figure 5-11). In contrast to the Bekkedal et al. (2000) study,
dosage began at the first day of gestation and continued through parturition and up to lactation
day 10 (PND10). Dams were dosed at 0, 0.1, 1.0, 3.0 and 10.0 mg/kg-day. Movements of each
pup were monitored by a passive infrared sensor. Each test session was 90 minutes in duration.
The number and time spent in movement was tabulated at each five-minute interval. In order to
be comparable with the USN analysis, every two of the five-minute intervals were combined into
a ten-minute interval. However, the Bekkedal et al. (2000) study did not have data for PND59,
so the results are not entirely comparable. Again, there was evidence of an increase in the effect
of ammonium perchlorate on motor activity at the later habituation times (posterior probability =
0.93). In the first habituation interval there was no evidence of an increase in motor activity with
dose (posterior probability = 0.58), while in the final interval there was moderate evidence
(posterior probability = 0.94). The dose estimated to increase the average of ambulatory
movements in the final habituation time by 10% is 4.60 with a credible interval of (2.18,
infinity). This interval was wider than the interval observed in the Bekkedal et al. (2000) study;
January 16, 2002
5-49
DRAFT-DO NOT QUOTE OR CITE
-------
Rate of increase in log {# movements} with dose (mg/kg/day)
Figure 5-11. Bayesian estimates of the posterior densities for the expected increase in the
logarithm of the number of ambulatory movements at the final habituation
time per unit dose (mg/kg-day) increase of ammonium perchlorate (Dunson,
2001a). A separate analysis for the Argus Research Laboratories, Inc.
(1998a) and United States Navy (Bekkedal et al., 2000) was performed.
1 possibly due to greater variability in the Argus data as noted in 1998 by EPA. This result is
2 slightly higher than the BMD analysis (Section 5.3.1.4) estimate of 1.04 mg/kg-day.
3 One of the advantages of Bayesian analysis is that it provides for formal combination of
4 data from different studies. To perform a combined analysis of data from the USN Study
5 (Bekkedal et al., 2000) and the Argus (1998) study, a modification of the model described above
6 was used (Dunson, 2001a). The number of ambulatory movements was first standardized by
7 subtracting the overall mean and dividing by the standard deviation. A linear mixed-effects
8 regression model that incorporated distinct baseline parameters (i.e., intercept, age-effects,
January 16, 2002
5-50
DRAFT-DO NOT QUOTE OR CITE
-------
1 habituation time effects, error variances) for the two studies was then fit, assuming common
2 slope parameters. This approach allowed the different studies to have distinct baseline
3 parameters, including aging effects.
4 Figure 5-12 shows the posterior density from the combined analysis of the Argus Research
5 Laboratories, Inc. (1998a) study and the Bekkedal et al. (2000) study. In this combined analysis,
6 the posterior probability of an increase in motor activity with dose was 0.99. For rats that
7 averages 34.09 ambulatory movements at the final habituation time in the absence of exposure
8 (the average value in the Argus study), the estimated dose needed to increase this average by
9 10% is 3.33 [95% credible interval = (1.91,12.78)].
10
Rate of increase in log {# movements} with dose (mg/kg/day)
Figure 5-12. Bayesian estimate of the posterior density for the expected increase in the
logarithm of the number of ambulatory movements at the final habituation
time per unit dose (mg/kg-day) increase of ammonium perchlorate for the
combined data from the two studies of motor activity effects shown in Figure
5-12 (Dunson, 2001a).
January 16, 2002
5-51
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
There was evidence of an increasing dose-response trend in motor activity in both the
Argus Research Laboratories, Inc. (1998a) and Bekkedal et al. (2000) studies, although the effect
in the Argus study was less pronounced, likely due to the variability in the data previously noted.
Given this, it is remarkable that the two studies showed such similar results. The Bayesian
analysis can be applied to risk assessment in an analogous fashion to the benchmark dose
analysis (Hasselblad and Jarabek, 1996). The lower limit on the estimated dose corresponding to
a 10% increase in motor activity relative to control can be used as a surrogate for the NOAEL for
the point of departure for reference dose derivation. For the Argus Research study, the lower
limit of the 95% credible interval for the dose was 2.18, while for the Bekkedal et al. (2000)
study the corresponding estimate was 0.90. In the combined analysis, the lower limit was 1.91.
Because of the variability in the Argus Research Laboratories, Inc. (1998a) study, a NOAEL that
relied on the Bekkedal et al. (2000) was chosen at 1.0 mg/kg-day to represent effects on motor
activity from these combined data.
5.3.3 The 2001 "Effects Study"
The Argus Research Laboratories, Inc. (2001) study was performed in response to
recommendations made at the 1999 external peer review (Research Triangle Institute, 1999) for
additional analyses of the thyroid and brain effects during gestation and post-natal days. Because
Argus Laboratories identified the day of birth as PND1, the age nomenclature of PND5, PND10,
and PND22 (Argus, 2001) is off by one day as referenced by EPA definition. These ages are
therefore referred to as PND4, PND9, and PND21.
It should be noted that exposure in this study started two weeks prior to the start of
cohabitation. The rationale was to ensure a hypothyroid state, but given the response of the rat
system to perturbation, it is more likely that this resulted in the dams already compensating for
the effect of perchlorate prior to pregnancy by upregulation of the NIS, making comparison with
the 1998 developmental neurotoxicity study (Section 5.3.1) more difficult.
The thyroid and brain from one male and one female pup per litter were selected for
histological and morphometric evaluation, with one set evaluated on PND4, PND9, and PND21.
January 16,2002 5-52 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
5.3.3.1 Results of General Toxicity Measures
There were no remarkable clinical or necropsy observations. Average body weights and
body weight changes for female rats were comparable among the five exposure groups through
the pre-cohabitation and gestation periods. Body weight gains for female rats in the 1.0 and
30.0 mg/kg-day target dosage groups were significantly increased on PND12 to PND15
compared to the carrier group. These increases were not considered treatment-related because
they were a singular occurrence and were transient.
5.3.3.2 Evaluation of Thyroid Histopathology
The thyroid histopathology in this study was evaluated using the same scoring system as
developed for the PWG review and was performed by one of the pathologists who served on the
PWG. A second read of these slides has not occurred. The data will be discussed individually
for each of the time points. Benchmark dose analyses conducted by EPA will be presented in
Section 5.3.3.2.1.
Absolute thyroid weights were increased significantly in the 30.0 mg/kg-day group in the
dams on GD21 and decreased colloid; increased hypertrophy and increased hyperplasia were also
noted at this dose. Thyroid weights were not collected for fetuses on GD21, but colloid depletion
was noted in both male and female fetuses at both the 1.0 and 30.0 mg/kg-day doses.
Thyroid weight in pups was measured on PND4, and the absolute weight was significantly
effected at 30 mg/kg-day, suggesting a NOAEL at 1.0 mg/kg-day. Histopathology was evident at
lower doses, suggesting a NOAEL at 0.1 for colloid depletion; however, no real dose-related
trend in either hypertrophy or hyperplasia was evident.
Thyroid weight in dams on PND9 continued to be effected significantly at 30 mg/kg-day,
with histopathology noted at lower doses. The pups on PND9 were more sensitive than the
dams, exhibiting statistically increased absolute thyroid weights at 0.1 mg/kg-day and higher
doses and suggesting a NOAEL at 0.01 mg/kg-day. A dose-related trend in histopathology in
this same range of doses was noted in the pups, especially for colloid depletion.
Thyroid weight in dams on PND21 remained significantly effected at 30 mg/kg-day, with a
clear dose-related trend in colloid depletion, hypertrophy and hyperplasia. All three
histopathological indices were increased at 30 mg/kg-day, and hyperplasia was also significantly
increased at the 1 mg/kg-day dose. It is interesting to note that hyperplasia was more sensitive
January 16, 2002
5-53
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
than both hypertrophy and colloid depletion in the dams at this time point, perhaps indicating a
system coming into the chronic phase of compensation described in Chapter 6.
Pups on PND21 also continued to exhibit increased thyroid weights at both 1 and
30 mg/kg-day (females only at 1.0 mg/kg-day). Colloid depletion was clearly significant at
30 mg/kg-day, and hyperplasia was noted although not designated as significant. Despite the
assertion by Argus Research Laboratories, Inc. (2001) that there was no dose-related trend in
hyperplasia, a BMD analysis indicated otherwise (see below). Hypertrophy was not noted, again
indicating an overlap among the three diagnostic indices of thyroid effects used by the PWG.
Benchmark dose analyses performed by EPA are presented in Table 5-3 (Geller, 2001b).
A benchmark response level of a 10% increase in incidence over controls, i.e., BMD 10 and
BMDL10, was adopted for all studies. Data were fit with a log-logistic function constrained such
that the slope was £ 1.
5.3.3.2.1 Benchmark Dose Analyses of Thyroid Histopathology
BMDL values in the dams on GD21 were 1.01, 1.19, and 8.51 mg/kg-day for colloid
depletion, hypertrophy, and hyperplasia. By PND9, these values decreased to 0.13, 1.01, and
0.92 mg/kg-day. Similar values for dams on PND21 were 0.62, 1.24, and 0.99 mg/kg-day for
colloid depletion, hypertrophy, and hyperplasia. Of note is the overlap between the estimates for
hypertrophy and hyperplasia.
The effects of ammonium perchlorate on the pups' thyroid glands are largely limited to
colloid depletion. The dams show additional dose-related effects on thyroid histopathology that
were evaluated as thyroid hypertrophy and hyperplasia. The low incidence of these latter two
endpoints in pups may be related to the duration of exposure compared to the dams and the adult
rats examined in earlier studies (Geller, 2001a). Alternatively, hyperplasia and hypertrophy may
be have been difficult to detect in the smaller thyroid glands from the young pups.
The BMDL 10 is lowest in the GD21 pups and is estimated at 0.12 mg/kg-day for the male
and female pups combined, or for male pups alone, and for female pups alone at 0.04 mg/kg-day.
The BMDL10 increases with age (Figure 5-13), suggesting that the thyroid gland may be most
susceptible to the effects of perchlorate during gestation or at the time of parturition (Geller,
2001b). This is likely due to the double effects of perchlorate inhibition of thyroid function in
January 16, 2002
5-54
DRAFT-DO NOT QUOTE OR CITE
-------
p
3
g
•5
ON
to
o
o
NJ
TABLE 5-3. BENCHMARK DOSE (BMD)a AND BENCHMARK DOSE LOWER CONFIDENCE LIMIT (BMDL)a
ESTIMATES FROM THYROID HISTOPATHOLOGY IN THE "EFFECTS STUDY"
(Argus Laboratories, Inc., 2001; Geller, 2001b)
Study Population
"Effects" Study
(Argus, 2001)
Colloid Depletion
Hypertrophy
Hyperplasia
BMD
BMDL
x2b
Expc
BMD
BMDL
x2b
Expc
BMD
BMDL
x2b
Expc
GD 21 Dams
5.10
1.01
1.00
17.90
15.46
1.19
1.00
6.25
28.54
8.51
1.0
5.03
GD 21 Male pups
0.69
0.12
1.00
8.82
NOE4
NOE
NOE
NOE
NOE
NOE
NOE
NOE
GD 21 Female pups
0.18
0.04
0.60
2.08
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
GD 21 M + F pups
0.65
0.12
0.16
7.80
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND4 Male pups
0.88
0.29
0.12
7.37
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND4 Female pups
0.82
0.18
0.12
7.78
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND4 M + F pups
0.84
0.33
0.02
7.50
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND9 Dams
0.62
0.13
0.59
2.65
2.65
1.01
0.22
17.86
2.24
0.92
0.49
1.0
PND9 Male pups
1.29
0.71
0.59
6.40
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND9 Female pups
0.33
0.13
0.61
1.30
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND9 M + F pups
0.93
0.48
0.36
3.77
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND21 Dams
1.21
0.62
0.34
4.90
15.60
1.24
1.0
6.34
3.59
0.99
0.66
1.0
PND21 Male pups
17.33
1.36
1.0
5.85
NOE
NOE
NOE
NOE
26.97
5.45
0.58
5.06
PND21 Female pups
16.42
1.24
1.00
5.94
NOE
NOE
NOE
NOE
NOE
NOE
NOE
NOE
PND21 M + Fpups
17.32
2.17
1.0
5.92
NOE
NOE
NOE
NOE
54.17
13.70
0.24
1.0
LT)
H
6
o
x
o
H
O
c!
O
a
g
o
3
m
a Units of mg/kg-day.
b x2 goodness of fit criterion
c Exponent in log-logistic function restricted to be k 1
d NOE = No observed effect.
0.
-------
1
2
3
4
5
6
7
8
9
10
11
PND21
Figure 5-13. Lower confidence limit on the dose of ammonium perchlorate in drinking
water that produced a 10% increase in the incidence of colloid depletion in
the thyroid gland as a function of post-natal age of rat pups. Data of Argus
Laboratories, Inc. (2001). Male and female data combined (Geller, 2001b).
the pup and the lack of protection of the pup by the dam because of her own compromised
thyroid function. After 21 days of post-natal exposure, the male pups also show follicular cell
hyperplasia.
The BMD and BMDL estimates of 0.84 and 0.33 mg/kg-day for the PND4 male and female
pups in this study (Table 5-3) do corroborate the BMD and BMDL for colloid depletion for the
PND4 pups from the 1998 Neurobehavioral Developmental study of 0.53 and 0.33 mg/kg-day
(Table 5-1). However, it should be noted that an unrestricted model also fits those data
adequately and results in a BMD and BMDL estimate of 0.45 and 0.009 mg/kg-day, suggesting
variability in those analyses (Geller, 2001b). Again, the lower estimates based on the 1998 data
at this time point (PND4) may be due to differences in the dosing of the dams between the two
studies.
January 16, 2002
5-56
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The BMD and BMDL estimates of 17.32 and 2.17 mg/kg-day for the PND21 male and
female pups in this "Effects Study" (Table 5-3) are somewhat higher than the previous 1999
two-generation reproductive toxicity study estimates of 2.51 and 0.80 mg/kg-day (Table 5-1).
However, comparison of the results of the two-generation reproductive toxicity study to the
current results may be difficult because of differences in the spacing of doses tested.
5.3.3.3 Thyroid and Pituitary Hormone Analyses
Thyroid (T3 and T4) and pituitary (TSH) hormones were also analyzed in the "Effects
Study" at various time points. Thyroid hormones and TSH were evaluated in the dams and fetus
on GD21, in the dams on PND10 and PND22, and in neonates on PND5, PND10, and PND22
(corresponding to PND4, PND9, and PND21 according to EPA nomenclature as explained
earlier). Table 5-4 presents the results of ANOVA analyses performed by EPA (Crofton, 2001b).
Maternal serum measures of the hormones were subjected to separate two-way ANOVA.
Treatment (dose) and age (GD21 or PND5, PND10 or PND22) were the independent between-
subjects variables. Two separate approaches were used to address the offspring data due to
differences in experimental design. The data from GD21, PND5 and PND10 were obtained from
litter-pooled samples due to the small volumes of blood and no gender analyses were possible.
These data were subjected to separate two-way ANOVA with age (GD21, PND5, or PND10) and
treatment (dose) as between-subjects variables. Blood samples from PND22 were not pooled so
that the data from this age were subjected to separate two-way ANOVA with gender and
treatment (dose) as independent variables. Mean contrasts were performed using Duncan's
Multiple range test. Significant two-way ANOVA were followed by step-down one-way
ANOVA to determine the main effects of treatment. If the interaction term was not significant,
then the model was refit if main effects were found. A reduced model was then fitted to the data
retaining only the main effects found significant previously, described as the "liberal" approach
in Crofton and Marcus (2001) and Marcus (2001).
EPA benchmark dose analyses (Geller, 2001c) of these results will also be discussed. The
benchmark estimates were generated using the Bench Mark Dose Software version 1.30, and fit a
Hill equation constrained such that the exponent on dose was £ 1.0 (Geller, 2001c). The BMDL
estimates indicate that the thyroid and pituitary hormones are exquisitely sensitive to the effects
of perchlorate.
January 16, 2002
5-57
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-4. NOAELs AND LOAELs FOR EFFECTS ON THYROID AND
PITUITARY HORMONES FROM THE ARGUS 2001 "EFFECTS STUDY"
(Crofton, 2001b)
Effect Level Designation
Generation Hormone Age Sex NOAEL LOAEL
Dams T3 GD21 F
PND10 F 1.0 30.0
PND22 F
T4 GD21 F — 0.01
PND10
F
0.1
1
PND22
F
1.0
30.0
GD21
F
—
0.01
PND10
F
—
0.01
PND22 F 0.01 0.1
Fetus and Offspring T3 GD21 Pooled
PND5 Pooled — 0.01
PND10 Pooled
PND22 F
0.01 1.0
M
T4 GD21 Pooled
PND5 Pooled 0.01 0.1
PND10 Pooled
PND22 F no significant effects
M — 0.01
TSH GD21 Pooled 0.1 1.0
PND5
Pooled
no significant effects
PND10
Pooled
— 0.01
PND22
F
0.01 0.1
M
— 0.01
"Dosages of 0, 0.01, 0.1, 1.0, and 30 mg/kg-day.
January 16, 2002
5-58
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
5.3.3.3.1 Maternal Hormone Analyses
Exposure to perchlorate produced significant decreases in thyroid hormones and an
increase in TSH in the dams at the various ages tested. For effects on maternal T3, there was no
age-by-treatment interaction and the NOAEL at all time points was 1.0 mg/kg-day. There was a
significant age-by-treatment interaction for effects on maternal T4. Step-down analyses resulted
in a LOAEL at 0.01, 1.0 and 30.0 mg/kg-day at GD21, PND9 and PND21. The 0.01 mg/kg-day
level is a LOAEL for the dams at GD21. There was also a significant age-by-treatment
interaction for the effects on maternal TSH. Step-down analyses resulted in a LOAEL at 0.01,
0.01 and 0.1 mg/kg-day at GD21, PND9 and PND21. As for the effects on T4, there was no
NOAEL at GD21 for the effects on TSH. There was no NOAEL for the effects on TSH at PND9
as well. These effects on T4 and TSH at GD21 are consistent with the Argus Laboratories Inc.
(2001) analyses. Benchmark dose analyses resulted in BMD estimates of 1.63, 0.006 and
2.38 mg/kg-day for the effects on T3, T4, and TSH at GD21. BMDL estimates were only
calculable for T4 in the dams and resulted in an estimate of 0.004 mg/kg-day. Benchmark dose
calculations were not performed for the dams on PND9. At PND21, a BMDL estimate was
calculable only for TSH in the dams with a resultant estimate of 0.53 mg/kg-day.
5.3.3.3.2 Fetal and Neonatal Hormone Analyses
Maternal exposure to perchlorate resulted in hypothyroidism in the offspring. There were
significant dose-related decreases in thyroid hormones and increases in TSH at all time points
evaluated.
There were no age-by-treatment interactions for the effects on T3 at any age tested. The
LOAEL for GD21, and post-natal days 4 and 9 was 0.01 mg/kg-day. This value is lower than
that reported in the Argus Laboratories, Inc. (2001) analyses. The specified benchmark dose
analysis were not computable for T3 at PND4 or PND21. There was no significant gender-by-
treatment interaction for the effects on T3. The NOAEL for effects on T3 at PND21 was
0.1 mg/kg-day. A BMDL was calculable only for the male pups and resulted in an estimate of
0.13 mg/kg-day.
There were also no age-by-treatment for the effects on T4. The LOAEL was 0.1 mg/kg-day
and the NOAEL was 0.01 mg/kg-day for GD21 and PND4 and PND9. At PND21, there was a
significant gender-by-treatment interaction for the effects on T4. There was no NOAEL
January 16, 2002
5-59
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
established for the male pups and 0.01 mg/kg-day was a LOAEL, whereas 0.01 was suggested as
a NOAEL in the Argus Laboratories, Inc. (2001) analyses. The females did not show significant
effects in either the EPA or Argus Laboratories, Inc. (2001) analyses. BMDL estimates were
extremely sensitive for changes in T4 at PND21 in the males with a BMD and BMDL at
0.001 and 2.86 x 10"7 mg/kg-day. Benchmark analyses did not converge for the data from the
female pups alone or for the combined data.
There was a significant age-by-treatment interaction for the effects on TSH. Step-down
analyses revealed a NOAEL at 0.1 mg/kg-day for GD21. There was no significant effect on TSH
at PND5, but then no NOAEL on PND9 with a LOAEL at 0.01 mg/kg-day. The LOAEL was
also 0.01 mg/kg-day in male pups at PND21. The females were slightly less sensitive as
suggested by the significant gender-by-treatment interaction. The NOAEL in female pups on
PND21 was 0.01 mg/kg-day. Benchmark analyses on the combined data resulted in a BMD and
BMDL of 0.06 and 0.02 mg/kg-day for the effects on TSH.
5.3.3.4 Brain Morphometry Effects
Due to the deficiencies of the remaining tissue blocks from the previous developmental
neurotoxicity study (Argus Research Laboratories, Inc. 1998a), it was determined that the
recommendation of the external peer review panel to evaluate more sections could not be
accomplished unless a new study was performed (Harry, 2001). Thus, one major objective of the
Argus Laboratories, Inc. (2001) "Effects Study" was replication of brain morphometric
measurements in order to address concerns raised by the US EPA, the NIEHS, and the external
peer review panel regarding results observed in the 1998 developmental neurotoxicity study
(Argus, Protocol Number 1613-002, 1998a; U.S. Environmental Protection Agency, 1998d). The
purpose was to evaluate, under more rigorous experimental conditions and according to the EPA
developmental neurotoxicity guidelines (U.S. Environmental Protection Agency, 1998b),
whether the effect in the corpus callosum identified by the EPA in the previous assessment
(Section 5.3.1) would be replicated.
In addition, another objective was to identify effects that may occur in other brain regions.
Details with respect to the rationale motivating the experimental design can be found in Harry
(2001). A brief summary of important points will be provided here, but the reader is referred to
Harry (2001) for specifics on this protocol and to other review articles (Garman et al., 2001;
January 16, 2002
5-60
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Adams et al., 2000; Rice and Barone, 2000; U. S. Environmental Protection Agency, 1998b,g,h)
for a fuller appreciation of the state-of-the-science supporting the use of these measures as
developmental neurotoxicity indices in risk assessment. The use of the rodent and not a non-
human primate was based on the degree of difficulty and the ethical issues involved with
conducting such screening studies in addition to the need to replicate previous findings. The
work, to document the process of normal development and alterations in the rat cited in these
reviews, supports the use of rodent models for determining potential adverse effects on the
developing brain.
It should be noted that Argus Laboratories identifies the day of birth as PND1; therefore,
the age nomenclature as recommended in the EPA guidelines for PND10 and PND22 actually
corresponds to PND9 and PND21 in this study. Likewise, in the previous 1998 Argus Research
Laboratories, Inc. Study (Section 5.3.1), the morphometry performed on PND12 was actually
done on PND11. While the actual ages were slightly different between the two studies, the
concept of capturing an active process of development with brain morphometry remains in effect
(Harry, 2001).
The motivation for evaluation of brain morphometry was based on the fact that the
formation and maturation of the nervous system is critically dependent upon both a temporal and
spatial organization pattern (U.S. EPA, 1998b; Harry, 2001). Within this framework, an
interdependency between the various cell types in the brain and a precise spatial relationship of
one cell type to one cell type another has been demonstrated. During this time, the developing
system is undergoing rapid maturation of organizational and regulatory processes. Thus, the
disruption of the developmental profile of one cell type may significantly influence critical events
in later development, resulting in an alteration of the normal formation of the brain and its
functional connections. Many toxic agents have been shown to interfere with one or more of the
developmental processes of the brain (i.e., cell division of neuronal and glia precursor cells, cell
interaction with the immediate environment through surface receptors or cell adhesion
molecules, regulation of cytoskeletal processes that control proliferation and migration, cell-cell
interactions that underlie synaptogenesis, development of the cerebral circulation and the blood-
brain barrier, myelination, and programmed cell death). Such perturbations may not be evident
by standard histological assessments as often there is little, if any, evidence of cell death. Rather
January 16, 2002
5-61
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
what is seen is a delay or disruption in the normal development and maturation of specific neural
regions (Harry, 2001).
Immersion fixation was the tissue processing method of choice and was both recommended
and agreed upon by both the EPA and the PSG for the study. While the tissue fixation method of
choice in adult rodents is via cardiac perfusion, even this procedure is not without problems that
can compromise tissue integrity. It has been documented that immersion fixation artifacts can
influence histological and morphometric evaluations of adult brains; however, a less than optimal
cardiac perfusion can also result in morphological artifacts. For the younger animal, there is less
of a consensus on the proper manner of fixation. With the decreasing size and blood volume of
the younger animal (PND4 and PND9) used in the protocol, the difficulty of ensuring a good
fixation via cardiac perfusion is significantly increased over that in the adult. Further, because
comparisons were to be made between the 1998 and the 2001 study, consistency in method of
fixation was considered to be a critically important variable to maintain as constant across
studies.
Following the review of the previous developmental neurotoxicity study (Argus Research
Laboratories, Inc., 1998a), and in considering design considerations for the subsequent study, the
plane of cut for the brain was discussed (Garman, 2001 a,b). While sagittal sections for analysis
were recommended for some aspects of morphometric analysis, coronal sections were ultimately
adopted since comparisons were to be made between the 1998 and the 2001 study. This final
design of the study also adhered to the EPA developmental neurotoxicity testing guidelines that
call for coronal sections (U.S. Environmental Protection Agency, 1998g, h). It was originally
recommended by the NIEHS that measurements of the corpus callosum in coronal sections
should not be conducted at the midline due to possible edema artifacts that can occur from the
close proximity of the ventricle. Three sites were recommended for measurement that would
have been consistent with the evaluation conducted by NIEHS on the sections from the Argus
Research Laboratories, Inc. (1998a) study (Section 5.3.1). It was agreed upon in the final design
meeting with PSG contractors that, given the time constraints and need for comparison to the
1998 study, one measurement per hemisphere would be recorded at the same site as used in this
previous study (Garman, 2001 a,b). This was a site just off of the midline of the two
hemispheres.
January 16, 2002
5-62
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Finally, a question raised in the PSG-contracted review (Toxicology Excellence for Risk
Assessment, 2001) with regard to age of sampling as it relates to myelin formation should be
addressed. The process of myelination is a "developmental landmark" for the maturation of the
brain, that is initiated upon the presence of the axon and continues over an extended period of
time. It is a structure that matures over time with the accumulation of protein and structural
lamella. One major period of myelin protein and lipid synthesis occurs approximately between
PND19 and PND35. Thus, while examination at PND21 would not capture the final
accumulation of myelin, it would capture events occurring at a time during which myelin
processing and lamella wrapping of the axon is actively occurring. Therefore, this may represent
a period of critical development of the myelin sheath. Examination of animals with a mature
myelin sheath (e.g., ages greater than PND40) may offer information regarding whether any of
the changes seen at earlier time points represent a permanent structural alteration. The majority
of studies that have examined myelin development and/or alterations in this developmental
process have employed biochemical, molecular, as well as, morphological evaluations to make
such determinations regarding delay or hypomyelination. From such studies, the time most
appropriate for examination appears to be between the ages of PND15 and PND35. Thus,
examination of the corpus callosum at PND9 is probably at the limit of early development for an
evaluation of the myelin sheath. However, it should be noted again that this study was intended
to determine if the effects seen previously (Argus Research Laboratories, Inc., 1998a) could be
repeated. Effects in the corpus callosum in that previous study occurred at the early (PND11)
and remained at the late (PND82) time points. Brain weight and the size of the frontal cortex and
caudate putamen also were effected at the PND82 sacrifice (Section 5.3.1.1).
In addition, the development of the axonal pathways connecting the two hemispheres via
the corpus callosum also continues to develop during this time period. While the study design
allowed for the collection of tissue at PND4, it is felt that any measurements recorded at such age
would be very limited in their contribution to the interpretation of the currently available data set.
In addition, given the variability of the plane of cut and the difficulty in examining brains of
young animals, EPA and NIEHS agree that examination of the corpus callosum in younger
animals (the remaining materials available for PND4) would present an even greater problem.
Figure 5-14 illustrates where the section levels were taken for the brain morphometry
measurements and shows the anatomical landmarks on the ventral and dorsal surfaces of the
January 16, 2002
5-63
DRAFT-DO NOT QUOTE OR CITE
-------
Level 1 Level 2
Bregma
Figure 5-14. Topograph of the approximate anatomical landmarks on the ventral and
dorsal surfaces of the brain used for making the morphometry measurements
(Garman, 2001c). The topograph provided is for an adult brain, but the same
landmarks are used for PND9 and PND21 brains although the sections at
these two other ages would differ due to the rapid growth during this period.
1 brain. The veterinary pathologist who performed the work has noted that while the landmarks
2 were the same for both the PND9 and PND21 brains, it must be appreciated that the sections
3 from one age versus the other would not look precisely similar (Garman, 2001c) due to the fact
4 that the brain is rapidly growing at this time.
5 Overall, the images of the brain sections from the PND9 and PND21 time points
6 demonstrated that the processing of the brain was adequate for conducting limited morphometric
7 measurements as outlined in the protocol. As mentioned by the PSG-contracted reviewers
8 (Toxicology Excellence for Risk Assessment, 2001) and stated in the study and additional reports
9 (Argus Research Laboratories, Inc., 2001; Consultants in Veterinary Pathology, 2001; Garman,
January 16, 2002
5-64
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
200 Id), there was a greater degree of variation in the PND9 sections than in the PND21 brain
sections (Harry, 2001). Many sections in the PND9 brains also showed signs of disruption or
damage that may have compromised the measurements. For these reasons the EPA relied upon
the PND21 measurements, despite corroborating effects from the materials at PND9.
There were no significant effects of treatment or sex on brain weight, anterior-posterior
cerebrum length, or anterior-posterior cerebellar size at either age tested. As discussed in the
Argus Research Laboratories, Inc. (2001) report, statistical analyses consisted of Students' t-test
comparisons between the control and the corresponding group of each sex at each separate dose
level. For example, PND9 male control striatum measurements were compared to measurements
for the PND9 male 30 mg/kg-day dose group, then PND9 male control striatum measurements
were compared to the PND9 1 mg/kg-day-dose group. These analyses were run separately for
both sexes and ages and all brain areas, right and left sides. The Argus Laboratories, Inc. (2001)
analyses found a large number of significant effects on brain morphometry at doses of 0.1 and
0.01 mg/kg-day ammonium perchlorate in drinking water.
Guidelines on the assessment of neurotoxicity (U.S. Environmental Protection Agency,
1998b) specify that alterations in brain structure should be considered adverse and relevant to
human health risk assessment. Alterations in brain structure are consistent with the mode-of-
action for perchlorate, i.e., transient decrements in T4 and T3 during development can result in
neurodevelopmental effects. The significant findings reported in the Argus Laboratories, Inc.
(2001) report strongly argue, therefore, that adverse effects of ammonium perchlorate are present
at the lowest dose tested and that this data set contains only LOAELs, no NOAELs.
While the analysis in the Argus report was provocative, the number of t-tests run increases
the risk of introducing Type I error into this analysis. To address this, a more conservative
multivariate analysis, profile analysis (Johnson and Wichern, 1988; Tabachnick and Fidell,
2001), was run by the EPA (Geller, 200Id). Profile analysis is more conservative than the
analysis described above because a multiple analysis of variance (MANOVA) takes into account
any correlations between the independent variables; whereas, the multiple t-tests assume
complete independence. This analysis also reduced the number of main effects tests by nesting
gender within litter and by constructing a vector composed of all of the morphometric data from
each animal, then comparing these vectors. The approach is explained in more detail below.
January 16, 2002
5-65
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
5.3.3.4.1 Description of EPA Profile Analysis of Brain Morphometry Effects
When a series of measurements are made from a single animal, i.e., within-subjects
measurements, they can be used to build a profile or vector of scores across the measurement
variables. Profile analysis makes between-groups comparisons using a vector composed of all of
the (within-subject) measurements taken from each animal. Its primary test, for parallelism of
the vectors, establishes whether the pattern of results between treatment groups is the same or
different. It is a much more rigorous and conservative test, requiring that all of the measurements
(i.e., all brain regions) show a dependence on dose with the same pattern. This determination
also allowed examination of the entire set of data without an a priori expectation of effect in one
brain region or another or the direction of the effect (i.e., decrease or increase). While there is
indication that certain areas of the brain are likely susceptible to the effects on thyroid hormones
of perchlorate (e.g., Madeira et al., 1991, 1992, 1993), and the previous study performed by
Argus indicated that the corpus callosum was affected (U.S. Environmental Protection Agency,
1998d; Crofton, 1998c), definitive gestational windows for specific brain areas are unknown.
Profile analysis determines whether there were dose-related changes in the pattern of brain
growth, i.e., brain growth in one region relative to another while precluding prior expectations
about specific areas of the brain or the direction and magnitude of these changes.
The profile analysis was run on the data from the PND9 and PND21 animals separately
with gender nested within litter (PROC GLM, SAS Institute, Inc, Cary, NC). The data were
provided in electronic form from Argus Research Laboratories, Inc. (2001) and in an additional
report (Garman, 200Id). Profile analysis requires data from each endpoint for each animal. Data
from individual brain regions, both right and left sides, were missing from 8 animals in the PND9
cohort and 3 animals in the PND21 cohort, eliminating these animals from the analysis (Geller,
200 Id: Table 1). If a sex by treatment interaction was found, separate analyses were run on
males and females. Treatment effects within a brain region were examined with univariate
analyses of variance with gender nested within litter. Dunnett's two-tailed t-test was used to
compare each dose group to controls at a = 0.05 for step-down tests of treatment effects within a
brain region as guided by the overall (univariate) treatment or sex by treatment effects.
Right and left side measures of the same brain structures were examined with profile
analyses (whole set of data) and repeated measures analyses of variance (univariate analysis on
each brain region). While there was no a priori reason to expect other than a bilateral effect, the
January 16, 2002
5-66
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
presence of this kind of bias could reflect either anisometries in brain regions (i.e., lateralization)
or sectioning that was not perfectly perpendicular to the anterior-posterior axis of the brain and
that would have resulted in sampling brain regions at different depths on right and left side.
These analyses, together with examination of the images of the brain sections (Harry, 2001)
demonstrated some systematic variability in the sectioning resulting in differences in right versus
left measurements in different brain regions. The magnitude of the variability was small and not
always in the same direction, even within a brain region (varying with the dose group sampled).
The small magnitude of difference relative to the dose-related changes found in this study, the
fact that different brain regions varied in their laterality bias in different directions, and that
different dose groups varied in different directions all argue for simply averaging the right and
left brain region measurements for each animal rather than tailoring different analyses for
different brain regions. In addition, averaging could help to reduce variability in the data due to
sampling only one histological section/brain region/animal. Therefore, data from right and left
sides of the brain were averaged before the analysis of dose effects. Where data were missing
from only one side of the brain, the existing measurement was used for the analysis.
Two additional analyses were run with adjustments to the raw morphometry data in
response to suggestions made by reviewers hired by the PSG (Toxicology Excellence for Risk
Assessment, 2001) designed to subtract variability due to variation in brain size and focus on
changes in the sizes of brain areas relative to one another. As suggested by the PSG review, one
analysis was run dividing all of the linear dimensions through by the post-fixation brain weight
from each brain. However, EPA and NIEHS note that there are little historical data for
normalizing data with post-fixation brain weight (Harry, 2001) and that fixation results in the
loss of any evidence of hydration-related changes such as edema or other swelling.
The second additional analysis was suggested by the NIEHS and also adjusted for brain
size using the anterior-posterior (a-p) measurements of cerebrum and cerebellum and the full
width measure of hippocampus to adjust the linear dimensions. In this analysis, frontal, parietal,
and corpus collosum dimensions were divided by a-p cerebrum size; dentate, CA1, and CA3
were divided by hippocampal width; and the cerebellar linear measurement was divided by the
a-p cerebellum measurement. Hippocampus, a-p cerebrum, and a-p cerebellum were not
included in the analysis as separate measures. The striatum and external germinal layer
measurements were not adjusted by these other linear dimensions.
January 16, 2002
5-67
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Ail additional two analyses were run on the PND21 data. These analyses omitted (1) the
posterior corpus callosum measurement, or (2) the posterior corpus collosum and all
hippocampal measures; i.e., all measures that came from the Level II section since there was
some indication that there may have been a systematic difference in the plane of sectioning with
dose (Harry, 2001).
5.3.3.4.2 Results of EPA Profile Analysis of Brain Morphometry Effects
The brain morphometry profiles were not parallel across treatment groups for PND9 pups
(Geller, 200 Id: Table 2). The absence of parallel profiles obviates further analysis for equal
profiles. This means that the effects of developmental dosing with ammonium perchlorate were
different on different brain regions. Planned contrasts show that the 0.01 and 1.0 mg/kg-day
doses were significantly different than controls (Geller, 200Id: Table 2A). Adjusting for brain
weight had little effect on these results (Geller, 200Id: Table 2B), though the adjustment for the
linear size of the different brain regions made the effect at the highest dose (30 mg/kg-day) also
significantly different from control (Geller, 2001d: Table 2C).
The brain morphometry profiles were also not parallel across treatment groups for the
PND21 pups (Geller, 200 Id: Table 2 A). Contrasts between each of the dose groups and controls
showed that the controls differed from all other dose groups at better than p < 0.0001, including
at the lowest dose used, 0.01 mg/kg-day ammonium perchlorate in drinking water. The absence
of parallel profiles obviates further analysis for equal profiles. The analysis adjusting for brain
weight or regional size yielded similar, highly significant effects (Geller, 200Id: Tables 2B, 2C).
Sex by dose interactions were significant in the parallel profiles analysis of the raw data and with
the data adjusted by brain region size. The parallel profile MANOVA remained significant at
p < 0.0001 in the overall and contrast tests with the posterior corpus callosum or posterior corpus
callosum and all hippocampal measurements (i.e., all measurements taken at section Level II
removed from the analysis) decreasing concern for confounding introduced by potential bias in
sectioning at this level suggested for the males (Harry, 2001).
The profile analysis was done using the raw (right-left averaged) data values. Because the
brain structures measured yield a range of measurements varying 10-fold, it is difficult to plot the
raw data vector in a meaningful way in order to see the differences driving the findings of
significant differences between dose groups. Figure 5-15 plots the (unadjusted) region-by-region
January 16, 2002
5-68
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 5-15. Profile analysis of brain morphometry measurements for PND21 rat pup
brain regions. The male and female data on linear thickness measurements
were combined and normalized by the control mean of each region. The
control data are represented by the horizontal line at 1.0. Profile analysis
determines whether the vectors of measurements from each treatment group
differ from each other and control in a dose-dependent fashion. The heavy
line represents the ± 99% confidence interval around the mean control
values. Note that while this plot uses the normalized data to more easily
illustrate the data vectors, the actual analysis was performed using raw data
values (Geller, 2001d). A similar analysis showed effects in PND9 brains
(data not shown).
1 size of each brain structure normalized by the mean size of that brain structure in the controls,
2 male, and female combined for the PND21 pup data. The control group is therefore represented
3 by a horizontal line at 1.0 with associated variability. The other dose groups differ from this
4 horizontal line to different extents, and the parallel profiles analysis tests, in essence, whether
5 these departures make the other dose groups significantly "non-horizontal". Note that the
January 16, 2002
5-69
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
analysis was not done on the normalized data; the control values were divided through to aid in
visualizing the data vectors used in this analysis. The 99% confidence intervals around the
control means represent an envelope inside of which comparable values ± standard error of the
mean (SEM) are not significantly different from controls.
5.3.3.4.2.1 Univariate analyses of brain morphometry
While the main reason to use profile analysis was to benefit from the power it brings to an
analysis by its conservative constraint that requires the entire vector of measurements depend on
dose with a consistent pattern, univariate analyses also were evaluated to gain insights into
effects on specific regions.
PND9 brains. Univariate tests yielded significant effects of treatment with ammonium
perchlorate in the frontal and parietal regions of the cerebral cortex, the striatum, region CA1 of
the hippocampus, the corpus callosum, and the external germinal layer of PND9 pup brains
(Geller, 2001d: Table 3A). There is an increase in size at the 1.0 mg/kg-day dose in the frontal,
parietal, and striatum measurements, and decreases in size in CA1 and the external germinal
layer. There were also treatment-by-sex interactions in the corpus callosum and CA1 regions
(Geller, 200 Id: Table 3 A). Both of these brain regions showed a treatment-related decrease in
linear extent in females while showing an increase in size in males. While most of the changes
in linear extent measured in the sampled brain regions were ±5 to 11%, the male corpus callosum
was increased 23% at both the 0.1 and 1.0 mg/kg-day doses.
The adjustment for brain size reduced the significance of treatment effects in the striatum,
CA1, and external germinal layer (Geller, 2001 d: Table 3A, center). The analysis using
adjustment for regional size (Geller, 2001 d: Table 3 A, right) was nearly identical to the raw data
analysis, with the addition of significant effects being noted on cerebellum.
A comparison of the profile analysis and the analysis presented in Argus Research
Laboratories, Inc. (2001) shows similar results were obtained on the PND9 brain morphometry
with one exception. Both analyses found an increase in linear extent of frontal, parietal, and
striatum at 1.0 mg/kg-day ammonium perchlorate and in the corpus callosum at the 0.1 and
1.0 mg/kg-day dose, with the corpus callosum increase limited to males. There was a decrease in
the linear extent of the striatum at 0.1 mg/kg-day dose and decreases in the size of region CA1 of
January 16, 2002
5-70
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
females at the 0.01, 0.1, and 1.0 mg/kg-day doses. The Argus Laboratories, Inc. (2001) analysis
did not detect a significant difference in female CA1 at the 0.01 mg/kg-day dose.
A post-hoc analysis of the plane of cut of the PND9 brain sections suggested that the
0.1 and 1.0 mg/kg-day dose groups were sectioned at a different depth than were the other dose
groups (Harry, 2001). This likely contributed to the small but significant increase in size of the
frontal, parietal, and striatum sections in the 1.0 mg/kg-day dose groups and may have
contributed to the large increase in size of the anterior corpus callosum seen in the PND9 males.
PND21 brains. The striatum, cerebellum, and corpus callosum II (posterior sample) all
showed significant changes with the lowest administered dose of ammonium perchlorate, 0.01
mg/kg-day (Geller, 200Id: Table 3B, left). The striatum was significantly reduced in size at all
but the highest dose. Region CA3 of the hippocampus similarly showed a U-shaped dose
response. The cerebellum and the posterior corpus callosum increased in size with dose in an
inverted U-shape. There were sex-by-treatment interactions in striatum and frontal cortex such
that the female rats showed a stronger dose-related decrease in linear measurement than males.
Both males and females show a complex dose response in the anterior corpus callosum
measurement. As in the PND9 animals, the changes in linear extent were generally in the ± 5 to
11% range with the exception of the posterior portion of the corpus callosum, which showed an
increase in size of 24% in the 0.01 and 1.0 mg/kg dose groups, and a 39% increase in the
0.1 mg/kg dose group.
The adjustments for brain size had little effect on the region by region results at PND21
(Geller, 2001 d: Table 3B, center, right). Dividing through by the a-p or hippocampal
measurements resulted in additional significant dose effects noted on CA1 and a sex by dose
effect on cerebellum.
The Argus Research Laboratories Inc. (2001) and current EPA analyses agreed. Both
analyses found a significant decrease in size of the striatum at 0.01, 0.1, and 1.0 mg/kg doses and
increases in size of the corpus callosum EI (posterior) and cerebellum at the same doses. Both
analyses noted the decrease in size of C A3 at the 0.1 mg/kg dose, the decreased anterior corpus
callosum in females at 0.01 mg/kg, and the increased size of the frontal region in males at 0.1 and
30 mg/kg.
January 16, 2002
5-71
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5.3.3.4.3 Conclusions of EPA Brain Morphometry Analyses of Brain Morphometry Effects
There were significant differences in brain morphometry due to treatment with ammonium
perchlorate at both PND9 and PND21 in this study. Tables 2 and 3 in Geller (2001 d) enumerate
strong effects of developmental exposure to ammonium perchlorate on brain morphometry
considered across all regions tested and in the analysis of individual brain regions. These effects
were present at PND9 and PND21, with the latter age group showing stronger effects. Many of
these effects represent an increase or decrease of ± 10% in the size of a brain region, similar to
the range of morphometric alteration noted in a recent study of fetal alcohol syndrome
(Bookstein, et al., 2001). The corpus callosum showed a notable increase of 24% or more in
linear extent at PND21 in the 0.01,0.1, and 1.0 mg/kg ammonium perchlorate dosing groups.
Adjusting the raw morphometric determinations by either brain weight or measurements of larger
brain areas (i.e., cerebrum, cerebellum, and hippocampus) had no strong effect on the results of
the analysis.
The significant differences in the parallel-profiles test demonstrate exposure-related
changes in relative growth of different brain areas even at the lowest administered dose (Geller,
200Id: Table 2). Univariate analyses to further investigate these effects showed effects on a
number of different brain regions at both ages tested. The most sensitive endpoints were the
linear dimensions of the striatum, corpus callosum, and cerebellum at the 0.01 mg/kg-day dose
when males and females were considered together at PND21. Thus, these analyses ultimately
agree with those submitted in Argus Laboratories, Inc. (2001): exposure to 0.01 mg/kg-day
ammonium perchlorate during gestational and post-partum (weanling) development resulted in
measurable changes in brain structures.
The increase in the size of the corpus callosum in this study replicates that seen in the
previous morphometric analysis of rats developmentally exposed to ammonium perchlorate (U.S.
Environmental Protection Agency, 1998d, Crofton, 1998c). This is notable given the differences
between the two studies. The previous data were obtained from tissues from rats aged PND11
rather than PND9 and PND21, and dose spacing included high doses of 3 and 10 mg/kg rather
than 1 and 30 mg/kg as in this study. Fewer animals were used in the previous study (6/dose/sex)
than in the current study (approximately 15/dose/sex), and litter identity was considered in the
current analysis. It also has been noted by Garman (2001c), a principal investigator with
January 16, 2002
5-72
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
established experience in performing brain morphometry on a substantial number of studies, that
such a treatment-related pattern has not been observed in other studies.
It should be noted that changes in thyroid hormone levels effect different brain regions
differently during development. For example, developmental hypothyroidism prolongs the
expansion of the external granular layer and increases fissure formation in the cerebellum
(Lauder, et. al., 1974). Different brain regions show an inverted U or U-shape dose response;
this is not uncommon in biological systems as compensatory or other mechanisms may be
triggered at high doses.
Fixation artifacts are not a concern in the study because all brains were fixed and embedded
at the same time. In addition, dose-related effects were seen as both increases and decreases in
brain region size. EPA concludes from this that whatever artifacts may be present were not large
enough to obviate alterations of the magnitude observed. There is some concern over sectioning
artifacts because the brains from the different dose groups were sectioned at different intervals
after sacrifice (Argus Research Laboratories, Inc., 2001) and post-hoc analysis of the brain
sections did reveal some systematic differences in the PND9 animals and in a limited sample of
sections examined from the PND21 animals (Harry, 2001). Additional sectioning is being
performed by EPA to address whether the anterior to posterior bias selection suggested in the
males (Harry, 2001) is a true confounder because normative data for brain measurements at these
ages are not available. These new data will be made available to the external peer review panel
as soon as possible. Because the analyses conducted without sections from this level still
resulted in a significant effect at the 0.01 mg/kg-day dose and the dose-related changes noted in
this study have not been noted in other studies with tissue sampler treated similarly (Garman,
2001c), this concern is somewhat mitigated. Certainly to be protective of public health, these
effects should be viewed as adverse until additional data either confirm or contradict that
conclusion.
In summary, two different analyses of the brain morphometry data from the 2001 "Effects
Study" (Argus Research Laboratories, Inc., 2001) yielded significant effects (i.e., alteration of
brain structures) of developmental exposure to ammonium perchlorate in drinking water at doses
of 0.01 mg/kg-day and higher in a mammalian (rat) model of neurodevelopment. These
alterations included a 23-39% increase in the size of the corpus callosum over controls in the
progeny of dams dosed with 0.01 to 1.0 mg/kg of ammonium perchlorate in drinking water.
January 16, 2002
5-73
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Alteration of brain structures in a laboratory animal model is considered to be an adverse
neurotoxic effect (U.S. Environmental Protection Agency, 1998b). One of the analyses used a
series of t-tests; the other a more conservative multivariate analysis employing a nested model
profile analysis followed by univariate analysis of specific brain regions. The latter method is
more likely to be considered a valid analytic method because it better incorporates the design
elements of the study and reduces the likelihood of Type I statistical error. These effects on brain
morphometry dictate a designation of 0.01 mg/kg-day as a LOAEL.
5.4 DEVELOPMENTAL STUDIES
The 1997 testing strategy included a developmental study in rabbits to evaluate both a
potential critical effect and to characterize the toxic effects of perchlorate in a species other than
rats. Testing guidance for developmental toxicity typically requires data in two different species.
A new study of developmental toxicity in rats was recommended at the 1999 external peer
review. This section reviews the historical data on the developmental effects of perchlorate
(5.4.1), the 1998 study in rabbits (5.4.2), and the new 2000 study in rats (5.4.3).
5.4.1 Historical Studies
Brown-Grant (1966) examined the effects of perchlorate on implantation and pregnancy
outcome in Wistar rats. Potassium perchlorate or potassium chloride (control) was administered
at 1.0% (w/v) in drinking water from GD2 through GD8. The daily calculated intake rates were
237 and 371 mg/rat for potassium perchlorate and potassium chloride, respectively. Rats were
administered methythiouracil 45 min before injection of 5 fu.Ci sodium radioiodide (13lr) and
sacrificed 2 h later. Rats clearly not pregnant were sacrificed on Day 20; whereas, pregnant rats
were allowed to deliver prior to sacrifice. Pregnancy was successful in 7/11 control rats and
8/11 perchlorate-treated rats. Among nonpregnant animals, implantation sites were not found.
Litter size, number of pups, and pregnancy were not affected.
In the same study, false pregnancy was induced by mating females with vasectomized
males. Females were dosed as before on GD2 through GD8 to 0.25 or 1.0% potassium
perchlorate or potassium chloride (control). These doses correspond to 63 and 246 mg potassium
January 16, 2002
5-74
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
perchlorate/rat and 82 and 308 mg potassium chloride per rat, respectively. Deciduoma
formation was induced through traumatizing one uterine horn while under anesthesia. Rats
exposed to the 0.25% dose were traumatized on GD3 and sacrificed on GD7. Trauma and
sacrifice occurred on GD4 and GD8, respectively, in the 1.0%-dose group. Methylthiouracil and
sodium radioiodide (131I ) were administered prior to sacrifice as before. Deciduoma formation
was not different between dosed and control rats. Thyroid weights were increased significantly
in the rats of the 1.0% potassium perchlorate-dose group.
A related study was performed by Brown-Grant and Sherwood (1971). Wistar rats were
mated shortly post-partum, and the present litter was culled to nine. The dams were then
administered 0.1% potassium iodide or 1.0% potassium chloride, potassium perchlorate, or
potassium iodide in the drinking water until sacrifice. The average daily intake of potassium
perchlorate and potassium chloride was 615 and 655 mg/rat, respectively; calculated daily doses
were approximately 2,440 and 2,660 mg/kg body weight. The litters were sacrificed on GD9 or
GD10. The dams then were sacrificed on GD12 or GD13, allowing time for the new blastocysts
to implant. Potassium perchlorate again did not affect blastocyst ability to survive prior to
implantation or implantation rate after lactation ceased. Relative thyroid weights of the dams and
litters were increased significantly compared with potassium-chloride-dosed controls. The high
dose of potassium iodide (average daily intake of 234 mg/rat [approximately 1,150 mg/kg]) was
maternally toxic.
All dams were sacrificed on Day 12 or 13 and examined for the number of implantation
sites. There was 100% incidence of dams with implantation sites for all groups except the
perchlorate-treated group in which only 70% of the dams had implantation sites. The number of
implantation sites per dam was comparable for all groups. Thyroid weights in the perchlorate-
treated dams appeared to be increased compared with the chloride- or iodide-treated dams. Also,
thyroid weights of the offspring of perchlorate-treated dams were increased compared with
offspring from iodide-treated dams. The authors concluded that treatment with potassium
perchlorate had no significant effect on blastocyst survival or the ability to implant under
conditions delaying implantation (i.e., concurrent lactation).
Postel (1957) reported administration of 1% potassium perchlorate in drinking water to
pregnant guinea pigs (n=16) and a control group (n = 3) receiving a diet of 0.48 /^g iodine per
gram. Dosing with perchlorate during GD21 through GD48 produced enlarged thyroids in the
January 16, 2002
5-75
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
fetuses compared to the thyroids of control fetuses. In contrast, perchlorate treatment did not
have any effect on the thyroids in dams. Enlarged fetal thyroids also occurred when perchlorate
treatment was accompanied by daily subcutaneous treatment with T3 doses as high as
32 /ig/kg/day. From water intake and body weight data, the author calculated an average daily
dose to the dams of 740 mg/kg-day. The fetuses were not examined for other developmental
effects. This study suggested a free-standing LOAEL of 740 mg/kg-day for fetal thyroid
enlargement because no other doses were tested. In a separate experiment to test effects on adult
guinea pigs, 0 or 1% potassium perchlorate was administered to nonpregnant female guinea pigs
for 30, 60, or 90 days. Thyroid enlargement and hyperplasia were apparent in treated animals
after 60 or 90 days of treatment.
Similar results in rabbits were described by Lampe et al. (1967). Dams were dosed with
100 mg potassium perchlorate/kg by weight daily, mixed with feed. Dosing occurred from
conception through GD21 or GD28. Maternal thyroid weights in treated animals were three
times higher than control thyroids; fetal thyroids were nearly four times the control weights. The
number of epithelial cells were increased, and the amount of colloid decreased in treated animals.
The relative volume of the stroma, the supporting matrix, was increased because of the reduced
follicle sizes. Likewise, maternal thyroids showed decreased luminal size and increased
epithelial cells. The authors asserted that these results demonstrated that the placenta is
permeable to perchlorate. Because fetal thyroids were more enlarged relative to maternal thyroid
glands, the fetal thyroid system is independent of the maternal regulatory system and more
sensitive to changes in iodine availability.
5.4.2 Segment II Developmental Toxicity Study in Rabbits
A developmental toxicity study was performed in New Zealand White (Hra:[NZW]SPF)
rabbits as part of the overall perchlorate testing strategy (Argus Research Laboratories, Inc.,
1998c). This study has also appeared in the literature (York et al., 2001a); however, because that
manuscript did not use the PWG review of thyroid histopathology and its conclusions on other
endpoints are the same as the contract report, the manuscript will not be discussed further in this
document. To aid understanding of terminology and the protocol, a schematic of the study
design is provided in Figure A-3 of Appendix A to this document. The study design meets the
requirements of the 1998 EPA Office of Pollution Prevention and Toxic Substances (OPPTS)
January 16,2002 5-76 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
870.3700 guideline. A deviation from the use of double staining was noted in Appendix D of the
Argus report, but EPA determined that this should not have had an effect on the overall outcome
of this study.
The dose groups tested were 0, 0.1, 1.0, 10, 30, and 100 mg/kg-day of ammonium
perchlorate in RO water provided by continual access on presumed GD6 to GD28. Each group
was comprised of 25 time-mated does assigned on a randomized basis stratified by weight.
Doses were selected on the basis of a dose range-finding study (Argus Research Laboratories,
Inc., 1998d) in which thyroid histopathology was evident in the does at 20, 50, and 100 mg/kg-
day; thyroid hormone levels (T3, T4, and TSH) in the does were reduced at all doses; and three
malformed fetuses from three litters in the 20-mg/kg-day group were observed upon gross
external examination. EPA was concerned about these pilot study results, particularly because
the original target doses of 0.1 and 10 mg/kg-day were changed on GDI3 to 50 and 100 mg/kg-
day based on the lack of clinical toxicity at these doses. The fact that these were the doses at
which effects were observed, together with the fact that a low number of animals (n = 5) was
used in this range-finding study caused EPA to counsel the sponsor (PSG) to examine an
expanded range of doses in the definitive study. The dose groups chosen for the definitive
developmental study were thus aimed to bracket the dose levels in the range-finding study and to
go below the doses causing thyroid hormone perturbations and above those associated with the
fetal malformations.
Dosing solutions of ammonium perchlorate were prepared at least weekly from stock
solution, and the results of the concentration analyses were within acceptable ranges. Stability of
solutions was assumed based on determinations by AFRL/HEST for the 90-day bioassay as
discussed in Section 5.2.3. Rabbits were observed for viability at least twice daily, and body
weight, food and water consumption, clinical observations, deaths, abortions, and premature
deliveries were evaluated daily. On GD29, rabbits were terminated and cesarean sections were
performed. Blood samples from the does were taken for evaluation of thyroid and pituitary
hormones (T3, T4, and TSH). Gross necropsy was performed on the thoracic, abdominal, and
pelvic viscera of each doe. Parameters evaluated in the does included pregnancy status, gravid
uterine weight, number of corpora lutea in each ovary, number and distribution of implantations,
early and late resorptions, and live and dead fetuses. The thyroids/parathyroids were evaluated
histologically. Weight, gross external alterations, sex, in situ brain status (in one-half of the
January 16,2002
5-77
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
fetuses in each litter), brain histology (in the other one-half of all fetuses in each litter), cavitated
organs, and skeletal and cartilaginous alterations were examined in the fetuses. No
measurements of thyroid structure or function were made in the fetuses.
5.4.2.1 Results of Maternal Examinations and Thyroid Histopathology
Two does in the 1.0-mg/kg-day group aborted either dead pups or had late resorptions on
GD28. Both of these abortions were considered unrelated to treatment because the incidences
were not dose-dependent and were consistent with historical control data for rabbits in that
laboratory (Argus Research Laboratories, Inc., 1998c; Appendix J). One doe in the 100-mg/kg-
day group delivered prematurely on GD27 (normal delivery in rabbits occurs on GD31), but it
was assumed that this rabbit had been identified and shipped incorrectly by the supplier because
the pups appeared to be full-term (i.e., they had fur and were nursing). There were no treatment-
related effects on maternal clinical signs, body weight, body weight change, gravid uterine
weight, or food and water consumption. It is interesting to note that there were decreases (not
statistically significant) in several of these endpoints, at the 1.0-mg/kg-day group-the same at
which the abortions occurred-as did one adverse necropsy observation of a mottled liver.
However, none of these responses showed a dose-response with the current treatment regimen,
and none were out of the range of normal occurrence.
The only remarkable histopathology in the does was observed in the thyroids. There was
an apparent dose-related but not statistically significant decrease in thyroid weight). The
histopathology in the dams as reviewed by the PWG can be found in Wolf (2000; 2001,
Table 22). There was a clear dose-response for colloid depletion, hypertrophy, and hyperplasia,
indicating that another species has conserved the hypothalamic-pituitary-thyroid feedback
regulation. All three indices appeared to be significantly increased at 1.0 mg/kg-day and above.
Benchmark dose analyses resulted in BMDL estimates of 0.008 for colloid depletion and 0.42 for
hyperplasia. A poor fit prevented BMDL estimation for hypertrophy.
5.4.2.2 Developmental Endpoints
There were no treatment-related effects on gross external endpoints (Argus Research
Laboratories, Inc., 1998c, Table 16). With regard to soft tissue anomalies (Argus Research
Laboratories, Inc.,1998c, Table 17), there were several occurrences of lung lobe and gallbladder
January 16, 2002
5-78
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
absence, but their incidence was not treatment related. The statistically significant decrease in
folded retina was attributed to an artifact of tissue processing. There were no treatment-related
effects in skeletal or ossification alterations (Argus Research Laboratories, Inc.,1998c, Tables 18
and 19), and no indication of an increased incidence of the more apical endpoint (i.e., any
skeletal change). The fetal NOAEL thus is identified as greater than 100 mg/kg-day for embryo-
fetal developmental toxicity, other than that which may have occurred in the thyroid.
5.4.2.3 Maternal Thyroid and Pituitary Hormone Analyses
The thyroid and pituitary hormone (T3, T4, and TSH) analyses were performed by
AniLytics, Inc., for the does in the developmental rabbit study (Argus Research Laboratories,
Inc., 1998c). Assays for T3 and T4 were performed using RIA kits according to manufacturer's
standard procedures. Assay kits from the same batch number and with the same expiration date
were used for the T3 and T4 measurements for each rabbit. The TSH assay was a
double-antibody, RIA procedure developed for rabbits and performed by AniLytics, Inc. The
analyses discussed in the Argus Research Laboratories, Inc. (1998c) report contain data from
both pregnant and nonpregnant rabbits, with both groups combined in the analyses. Because of
the known effects of pregnancy on thyroid hormones, EPA decided to reanalyze separately the
data from the pregnant and nonpregnant animals. However, EPA determined that the analyses
for nonpregnant animals were not useful because of the very limited number of subjects per
group (final number of does: n = 3, 1,0, 1, 1, and 1 nonpregnant does/group, and n = 22, 24, 25,
24, and 23 pregnant does/group for the 0.0, 0.1, 1.0, 10, 30, and 100 mg/kg-day groups,
respectively). Therefore, EPA conducted reanalyses for these two groups separately (Crofton,
1998h). All data were taken from Appendix I of the report (Argus Research Laboratories,
Inc., 1998c). The analyses used the pregnancy status data subsequently submitted (York, 1998e).
Data from dependent measures (T3, T4, and TSH) were subjected to separate one-way ANOVA
tests with treatment (dose) as the independent between-subjects variable as calculated in Crofton
and Marcus (2001) and Marcus (2001). Mean contrasts were performed using Duncan's
Multiple Range Test.
The main effect of treatment was not significant for T3. The T3 data are plotted in
Figure 5-16A. There was a main effect of treatment and a significant difference between group
means for the control versus 1.0, 10, 30, and 100 mg/kg-day groups on T4. These data are
January 16, 2002
5-79
DRAFT-DO NOT QUOTE OR CITE
-------
2.5
W 20
(0
+1
c
re
Q)
E 15
o>
" 1.0 H
0.5
0.1
10
30
100
ID
V)
+l
C
03
0)
E
5
O)
c
UJ
CO
+l
c
flj
0)
E
O)
c
X
so
1 10 30 100
Dose, mg/kg-day
Figure 5-16. Effects from ammonium perchlorate in drinking water administration in
pregnant New Zealand rabbits during GD6 to GD28 on T3 (A), T4 (B) and
TSH (C) concentrations (ng/dL; mean ± SE) as recalculated in Table 5-2
(Crofton and Marcus, 2001). Data of Argus Research Laboratories, Inc.
(1998c). Means with different letters were significantly different (p<0.05).
Daily dose was estimated from water consumption data.
January 16, 2002
5-80
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
plotted in Figure 5-16B. The main effect of treatment was not significant for TSH
(Figure 5-16C). Results of these EPA reanalyses are different from those stated in the report.
The report (Argus Research Laboratories, Inc., 1998c) states that the NOAEL for T4 was
10 mg/kg-day. The current EPA analyses excluding nonpregnant animals, demonstrate a
NOAEL at 0.1 mg/kg-day for T4. There was no statistical significance of any dose on T3 or
TSH.
The lack of effect of any dose of perchlorate on T3 and TSH is difficult to explain. One
must note that these data are from rabbits (the majority of other data are from rats) and that the
data were collected 1 day prior to birth from the maternal compartment (whereas, all other data
were collected in adults or from postnatal day time points). In a previous study in guinea pigs
(Postel, 1957), enlarged thyroids were found in fetuses; whereas, there was no change in maternal
weight or histology. Lampe et al. (1967) demonstrated a larger effect on fetal thyroid weight
compared to maternal thyroid weights during late gestational exposure to perchlorate in rabbits.
These data warrant caution when comparing effects of perchlorate in the maternal with the
fetal/post-natal compartments.
5.4.3 Segment II Developmental Study in Rats
As recommended at the 1999 external peer review, a developmental study in addition to the
one in rabbits was performed in rats (Argus Research Laboratories, Inc., 2000). The EPA review
(Kimmel, 2000) was first performed on the audited final report (June 2000) and then on
clarifications provided by the principal investigator (York, 2000) that do not appear in the final
report.
Rats were given continuous access to target dosages of 0.01, 0.1, 1.0, and 30 mg/kg-day
ammonium perchlorate in deionized drinking water beginning at least 15 days before
cohabitation and continuing through the day of sacrifice. Each dosage group was comprised of
24 females, assigned on a random basis, stratified by weight. There were no maternal deaths.
Of these females, 20 were selected for evaluation; of these, 19, 19, 17, 20, and 20 were pregnant
in the 0, 0.01, 0.1, 1.0, and 30 mg/kg-day groups. The EPA OPPTS 870.3700 testing guidelines
recommend 20 pregnant animals per group at necropsy so that the power of the study to detect an
exposure-related response was somewhat lower.
January 16, 2002
5-81
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
All rats were sacrificed on day 21 of presumed gestation (GD21), and a gross necropsy of
the thoracic, abdominal, and pelvic visera was performed. Gravid uterine weights were recorded,
and the uterus then excised and examined for pregnancy, number and distribution of
implantations, live and dead fetuses, and early and late resorptions. The number of corpora lutea
in each ovary was recorded. Placentae were examined for abnormalities (size, color or shape).
Each fetus was identified, weighed and examined for sex and gross external alterations.
Approximately one-half of the fetuses in each litter were examined for soft tissue alterations.
The heads of these fetuses were examined by free-hand sectioning. The remaining fetuses in
each litter were examined for skeletal alterations and cartilage development.
5.4.3.1 Results of Maternal Examinations
Three dams in the 30 mg/kg-day group showed an increase in localized alopecia that was
statistically significant and was observed over 9-11 days during mid-late gestation. EPA feels
that this should be considered biologically significant and exposure-related despite the claim by
Argus Research Laboratories, Inc. (2000) and the study director (York, 2000) that such incidence
is within the range observed historically at their testing facility.
There were no other maternal parameters that were clearly supportive of exposure-related
effects. There was a statistically significant increase in corrected maternal body weight gain over
gestation in the 0.1 and 30.0 mg/kg-day groups, and an increase (not statistically significant) in
the 1.0 mg/kg-day group. There was also a reduction, again not statistically significant, in gravid
uterine weight in three of the four exposure groups. These latter changes may be associated with
reduced number of implants in the exposed groups (see below).
5.4.3.2 Developmental Endpoints
The Argus Research Laboratories, Inc. (2000) report (Table B17) did not record
preimplantation loss as an endpoint. EPA notes that there is an increase in this parameter over
control (12%) at each dose level: 0.01 (18% ), 0.1 (20%), 1.0 (16%), and 30.0 (25%) mg/kg-day.
Whether this is statistically significant or biologically significant is unclear; although a decrease
in live fetuses in three of the four exposure groups that was significant at the highest dosage was
reported. Given the reduced power of this study to detect an effect, consideration was paid to
this finding. The lack of an effect on live fetuses at the 1.0 mg/kg-day level is not clear, and
January 16, 2002
5-82
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
these results by themselves are insufficient to establish an effect level below 30 mg/kg-day. EPA
recommends that preimplantation loss and embryo/fetal viability should be evaluated in any other
study reports on this chemical.
Ossification sites per litter for sternal centers and forelimb phalanges were significantly
reduced at 30 mg/kg-day, but Argus Laboratories, Inc. (2000) dismissed them as "reversible
developmental delays." EPA disagrees and contends that developmental delays, be they
permanent or reversible, are not to be discounted as potential indicators of developmental
toxicity. EPA additionally had some concern over the staining technique used for cartilage
(Kimmel, 2000) which was not accepted by Argus Research Laboratories, Inc. (York, 2000) as an
issue.
5.4.3.3 Conclusions Regarding Developmental Toxicity in Rats
Based on the review of the maternal and fetal data, EPA concludes that there are signs of
maternal and developmental toxicity at the 30.0 mg/kg-day level suggesting it as a LOAEL with
a NOAEL then at 3.0 mg/kg-day. While none of the results were so clear that a definitive
assessment can be made, the suggestive results are important to consider in light of the overall
data base and mode of action for the toxicity of perchlorate.
5.5 TWO-GENERATION REPRODUCTIVE TOXICITY STUDY
The 1997 recommendation to characterize the potential perchlorate toxicity on reproductive
parameters in a two-generation study was completed in 1999 (Argus Research Laboratories, Inc.,
1999). This study has also been reported in the literature (York et al., 2001b), but since that
manuscript did not use the PWG review of thyroid histopathology and its conclusions on other
endpoints are the same as in the contract report, the manuscript will not be discussed further in
this document. A schematic of the study design is provided as Figure A-2 of this document
(Appendix A) to aid understanding of terminology and the protocol.
The target doses (30 rats/sex/group) were 0, 0.3, 3.0, and 30 mg/kg-day of ammonium
perchlorate in RO water provided by continual access. Concentrations were adjusted based upon
actual water consumption and body weights recorded the previous week. Dosing solutions of
ammonium perchlorate were prepared weekly, and the results of concentration analyses were
January 16,2002 5-83 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
within acceptable ranges (±10%) with one exception in the 3.0-mg/kg-day target group on May 5,
1998 (15.8%). The stock solution was prepared at least once, but the exact number of times was
not reported. Stability of solutions was assumed based on determinations by AFRL/HEST for
the 90-day bioassay, as discussed in Section 5.2.3.
On arrival, Spraque-Dawley rats were assigned randomly to individual housing, and
consecutive order was used to assign the PI generation rats to cohabitation (one male rat per
female rat). The cohabitation period lasted a maximum of 14 days. Females with spermatozoa
observed in a vaginal smear or with a copulatory plug observed in situ were considered to be at
GDO and assigned to individual housing. Estrous cycling was evaluated daily by examination of
vaginal cytology beginning 21 days before the scheduled cohabitation period and continuing until
GDO. The rats were observed for viability at least twice each day of the study and daily for
clinical signs. Body weights were recorded weekly during acclimation, on the first day of
dosage, weekly thereafter, and at scheduled sacrifice. Feed consumption and water consumption
values were recorded at least three times per week. Females were evaluated for duration of
gestation (GDO to the day the first pup was delivered). Day 1 of lactation (LD1, post-partum)
was defined as the day of birth and was the first day on which all pups in a litter were weighed
individually. Maternal behavior was observed on LD1, 4, 7, 14, and 21. Rats that did not deliver
a litter were sacrificed on GD25 and examined for pregnancy status. Each litter was evaluated
for litter size (live and dead pups versus live pups only) and pup viability at least twice each day
of the 21 -day post-partum period, and pups were counted daily. Deviations from expected
nursing behavior also were recorded. All F1-generation rats were weaned at the same age based
on observed growth and viability at LD21, unless required to be extended to LD28.
At the end of the 21-day post-partum period, all surviving PI rats were sacrificed. Gross
necropsy was performed on all animals, and all gross lesions were examined histologically.
Organ weights were obtained for the thyroid, adrenal glands, brain, epididymides, heart, kidneys,
liver, ovaries, pituitary, prostate, seminal vesicles, spleen, and testes. The thyroids and
parathyroids were submitted for histopathological examination. Histopathology of other organs
was performed for the control and high-dose groups. Blood was collected for determination of
hormone levels (T3, T4, and TSH). Portions of the epididymides were used either for evaluation
of sperm count or motility. The left testis was homogenized after weighing for analysis of
spermatid concentration (spermatids per gram of tissue).
January 16, 2002
5-84
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Pups not selected for continued evaluation in the study also were sacrificed on LD21.
Blood was pooled by sex per litter for analysis of T3, T4, and TSH. At least 3 pups/sex/litter
were necropsied and examined for gross lesions, including a single cross-section of the head at
the level of the frontal-parietal suture and examination of the head for apparent hydrocephaly.
Brain, thymus, spleen, and thyroid/parathyroid organ weights were obtained prior to fixation.
The adrenal glands, thyroid/parathyroid, kidneys, and liver were retained in formalin.
5.5.1 General Toxicity Results and Evaluation of Reproductive Parameters
There was a statistically significant decrease in water consumption by males, but not by
females. The decrease with males and a smaller decrease with females were sufficiently small
that they are not considered to be biologically significant (Argus Research Laboratories, Inc.,
1999; Tables B5 and B6). There was a significant increase in ovarian weight at the 0.3-mg/kg-
day dose level only (Argus Research Laboratories, Inc., 1999; Table C26). There also was
slightly increased (not statistically significant) pituitary weight in females at the 0.3- and
3.0-mg/kg-day dose levels.
The fertility results are potentially of concern, but the statistical analyses did not show any
significant differences between groups for any of the tested parameters (Argus Research
Laboratories, Inc., 1999; Table C21 through C23). However, at 0.3 mg/kg-day, there were four
pairs that did not mate compared with one or two pairs in the other groups. Also at
0.3 mg/kg-day, there were three females that showed at least one signal of persistent diestrus and
one with persistent estrus (Argus Research Laboratories, Inc., 1999; Table C40). Incidences
were lower in all other groups. Only one of those females did not have evidence of mating, but
there were also four females that did not have evidence of mating in the 0.3 mg/kg-day group.
When mating and conception failures are combined, pregnancy rates were 28/30, 22/30, 26/30,
and 24/30 for the 0-, 0.3-, 3.0-, and 30-mg/kg-day groups, respectively. Of the females that were
pregnant, litter size was slightly lower at the 3.0- and 30-mg/kg-day dose levels, with values of
15.0, 14.9, 14.1, and 14.0 with increasing dose level. A similar trend was seen in the number of
implantation sites (15.8, 15.8, 15.0, and 15.0). None of these results were statistically significant
for the PI generation, and the effect was not seen in the F1 generation. Consequently, this was
not considered a significant finding (Clegg, 1999; Rogers, 2000). Note should be made that
female intake of perchlorate during the last week of gestation was higher (Argus Research
January 16,2002 5-85 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Laboratories, Inc., 1999; Table CI). Additionally, in many of the perchlorate intake and feed
consumption summary data, observations were reported for a low numbers of rats, apparently
because of spillage.
In the F1 matings, all three perchlorate-dosed groups had a slightly higher fertility index
than did the vehicle controls, but this appears to be due to a control value that was low (Clegg,
1999; Rogers, 2000). These findings, the high dosage level of 30 mg/kg-day is designated as a
NOAEL for reproductive parameters (Rogers, 2000), a finding that is consistent with the
preliminary evaluation presented by EPA in 1999 (Clegg, 1999).
5.5.2 Evaluation of Thyroid Histology
The histopathology from the completed Argus Laboratories, Inc. (1999) two-generation
reproductive study was limited to the thyroid gland and can be found in Wolf (2001; Tables 14
through 21). In addition to the precursor lesion data (colloid depletion, hypertrophy, hyperplasia)
discussed in Section 5.5.2.1, Wolf (2000) noted that two animals from the high dose group (30
mg/kg-day) in the F1 generation (second parental generation, P2) in the study had adenomas and
one of these animals had two adenomas for a total of three. Although statistically significant
decreases in colloid were reported at both the 3.0 and 30.0 mg/kg-day dose levels (Argus, 1999),
none of the rats in the other groups (0, 0.3, 3.0 mg/kg-day) developed thyroid follicular cell
adenomas (0/30, 0/30, 0/30, respectively). These animals were dosed from conception to 19
weeks of age (adult male F1 rats). The tumors were considered to be treatment related (Wolf,
2000). Compared to the background incidence of thyroid follicular cell adenomas in male F344
rats after 2 years on study at 38/3419 from 67 NTP studies or 1.1% incidence at the 2-year end
sacrifice date, this study showed an incidence of 2/30 or 6.7% at 19 weeks. The tumors that
occurred in the F1 generation male rat pups at 19 weeks were considered particularly remarkable
(Wolf, 2000), and the EPA asked the NIEHS to review this incidence in context with the data
from the National Testing Program (NTP). The finding is especially of concern since three of the
F1 males in this high dosage group died of unknown causes (Rogers, 2000). This NIEHS
analysis of the tumor incidence is described below (Dunson, 2001b) in Section 5.5.2.2.
January 16, 2002
5-86
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5.5.2.1 Thyroid Weight, Colloid Depletion, Hypertrophy, and Hyperplasia
Absolute thyroid weight was increased significantly in the PI males at the 3.0- and
30-mg/kg-day dose levels. An increase was significant in females at 30 mg/kg-day.
A significant increase in thyroid weight relative to both body weight and brain weight also
occurred at 30 mg in both sexes (Argus Research Laboratories, Inc., 1999; Tables B11 through
B13 and C26 through C28). The histopathology for the PI generation as reported by the PWG
can be found in Wolf (2000; 2001, Tables 14 and 15). All three indices (colloid depletion,
hypertrophy, and hyperplasia) were present with a clear suggestion of an increase in females for
colloid depletion and hypertrophy at 3 and 30 mg/kg-day that supported the thyroid weight
changes. Hyperplasia was more prominent at 30 mg/kg-day. Benchmark dose analyses using the
male and female data for the PI generation combined (Table 5-1; Geller, 2001a) result in BMDL
estimates of 0.11 mg/kg-day for colloid depletion and 2.44 mg/kg-day for hyperplasia. The data
for hypertrophy resulted in inadequate model fit.
The F1-generation (second parental, P2 generation) rats also exhibited all three thyroid
histopathological indices in a dose-related fashion with 3 and 30 mg/kg-day as effect levels
(Wolf, 2000; 2001, Tables 16 and 17). Benchmark dose analyses (Table 5-1; Geller, 2001a)
using the male and female data combined for the P2 generation estimate 0.90, 0.15, and
0.0004 mg/kg-day as the BMDL for colloid depletion, hypertrophy, and hyperplasia. Of note is
the dramatic overlap between colloid depletion and hypertrophy in this generation. It was the
males in these rats, exposed in utero and then sacrificed at 19 weeks, that showed the
3 adenomas.
The Fl-generation weanling rat data are presented in Tables 18 and 19 (Wolf, 2000; 2001)
and also exhibit the three thyroid histopathology indices increased at 3 and 30 mg/kg-day.
Benchmark dose analyses (Table 5-1; Geller, 2001a) using the male and female data combined
result in BMDL estimates of 0.80, 0.057, and 0.66 mg/kg-day for colloid depletion, hypertrophy,
and hyperplasia. Again, the overlap among indices is present.
Data for the second weanling generation (F2) rats are presented in Wolf (2000, 2001;
Tables 20 and 21). Decreased colloid and hypertrophy remain increased at 3 and 30 mg/kg-day,
but hyperplasia was not remarkable. Benchmark dose analyses (Table 5-1; Geller, 2001a) only
provided adequate fit to the hypertrophy data and resulted in a BMDL of 0.32 mg/kg-day.
January 16, 2002
5-87
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Across the generations, this study results in a range of BMDL estimates (mg/kg-day) for
colloid depletion of 0.11 to 0.90, for hypertrophy of 0.057 to 0.32, and for hyperplasia of
0.0004 to 2.44. Of note is the low BMDL value for hyperplasia (0.0004 mg/kg-day) in the P2
generation, the same animals that exhibited tumors.
5.5.2.2 Bayesian Analysis of Tumor Incidence
In order to properly interpret the results from a given toxicological study, it is often
necessary to consider the data in light of additional information from outside of the study such as
the variability and average level of response for positive and negative controls in past studies that
are similar to the current study. It is also necessary to account for confounding effects that an
exposure may have on variables that are associated with the outcome of interest. For example, it
is important to adjust for animal survival to avoid bias in analyses of animal tumorigenicity
(McRnight and Crowley, 1984) and reproductive toxicity (Dunson and Perreault, 2001).
Typically, expert knowledge and information from related studies are accounted for only
informally in the interpretation of a statistically significant or non-significant result. However,
there are clear advantages to formally incorporating such extra information into the statistical
analysis because it can be very difficult to interpret statistical significance when some aspect of
the data is inconsistent with outside information (e.g., the control response is higher or lower
than typically seen in related studies). In addition, the formal incorporation of outside
information can improve sensitivity and limit bias when assessing toxicological effects. The
advantages of including historical control data, in particular, has been well documented in the
toxicological and statistical literature (Dunson and Dinse, 2001; Haseman, Huff, and Boorman,
1984; Ibrahim, Ryan and Chen, 1998; Tarone, 1982).
Although frequentist (i.e., non-Bayesian) hypothesis tests can sometimes incorporate
historical control data (see, for example, Tarone, 1982), outside information can be incorporated
more naturally and flexibly within a Bayesian analysis. In Bayesian analyses, the unknown
parameters in a statistical model are assigned prior probability distributions quantifying
uncertainty prior to observing data from a current study. For example, based on experience with
an assay system, a toxicologist may be 95% certain that the average level of response among
vehicle control animals is between bounds A and B with C being the most likely value. This
information can be formally incorporated into a Bayesian analysis through a prior distribution,
January 16, 2002
5-88
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
for a parameter measuring expected control response, which is centered on C and assigns 95%
probability to values between A and B. Alternatively, the prior distribution can be estimated
using data or summary statistics for control animals in historical studies if such information is
available (Ibrahim, Ryan and Chen, 1998; Dunson and Dinse, 2001). For parameters about
which little is known, noninformative or vague prior distributions that assign equal prior
probability to a wide range of plausible values can be chosen.
Bayesian inferences about toxicological effects can be based on the posterior distribution
for the parameters in the statistical model. The posterior distribution, which quantifies the
current state of knowledge about the unknown quantities in the statistical model, is obtained by
updating the prior distribution with the information in the data from the current study using
Bayes theorem (refer to Gelman et al., 1995 for an overview). One can use the posterior
distribution as a basis for conclusions about effects of interest by using posterior means, 95%
credible intervals, and posterior probabilities as Bayesian alternatives to the maximum likelihood
estimates, 95% confidence intervals, and p-values used in frequentist analyses. For example, as
an alternative to a p-value, one could calculate the posterior probability of an increase in the
proportion of animals with an adverse response in a treated group relative to the control.
Bayesian approaches have been developed for a wide variety of toxicological applications,
including risk assessment (e.g., Hill, 1996; Hasselblad and Jarabek, 1996), toxicokinetic
modeling (e.g., Bernillon and Bois, 2000), and analysis of skin papilloma data (Dunson et al.,
2000).
Without incorporating historical data on spontaneous neoplasms in Sprague-Dawley rats,
the difference between 0/30 in the vehicle control and 2/30 in the 30 mg/kg-day group is
non-significant by standard tests (e.g., Fisher's exact). However, the reported historical control
incidence of thyroid follicular adenomas for male Sprague-Dawley rats in two-year studies is
approximately 3-4% (Chandra et al., 1992; McMartin et al., 1992), suggesting that these tumors
should be extremely rare among 19-week old animals in the absence of a treatment effect.
Without formally incorporating this historical information into the statistical analysis through a
prior distribution, it is very difficult to assess the weight of evidence in favor of a treatment-
related increase in thyroid follicular adenoma incidence. A Bayesian approach was used to
assess the effect of ammonium perchlorate in drinking water on thyroid follicular cell adenoma
January 16, 2002
5-89
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
incidence in male Sprague-Dawley rats from the two-generation study (Argus Research
Laboratories, Inc., 1999).
5.5.2.2.1 Choosing prior distributions based on historical controls
The proportion of control male Sprague-Dawley rats developing thyroid follicular cell
adenomas in two-year carcinogenicity studies has been reported in the literature. Chandra et al.
(1992) reported a rate of 48/1340 (3.6%), and McMartin et al. (1992) reported a rate of 23/583
(3.9%). In order to incorporate this historical control data into our analysis of the effect of
ammonium perchlorate on thyroid incidence at 19 weeks of age, we follow a Bayesian approach.
The historical data can be summarized using a Beta (71,1852) prior distribution for the
probability of a male Sprague-Dawley rat developing a thyroid follicular cell adenoma (in the
absence of treatment with a test agent) by the time of natural death or sacrifice at two years. The
Beta prior is the standard choice for a prior distribution on a probability (c.f., Dunson and
Tindall, 2000 and Gelman et al., 1996 for further discussion of the Beta prior). The values
71 and 1923 are simply the numbers of control male Sprague-Dawley that did and that did not
develop thyroid follicular cell adenomas, respectively, from the Chandra et al. (1992) and
McMartin et al. (1992) articles.
To account for the fact that the Argus (1999) study recorded thyroid incidence at 19 weeks
and not at the time of natural death or at sacrifice at two years, a prior distribution for the ratio of
the probability of thyroid follicular cell adenomas at 19 weeks to the lifetime probability in a
two-year study was chosen. Portier, Hedges, and Hoel (1986) suggest that the probability of a
control male Fischer 344/N rat developing a thyroid follicular cell adenoma increases
approximately in proportion to age4 78. Based on this estimate and on the average survival time
for male Fischer 344/N rats in the NTP historical control database (95.2 weeks), the prior
expectation for the ratio is (19/95.2)4 78 = 5e-04. Allowing for a high degree of uncertainty in this
prior expectation due to uncertainty in the Portier, Hedges, and Hoel (1986) estimate and in
extrapolation from Fischer 344/N rats to Sprague-Dawley rats, a Beta (0.11, 2.6) for the ratio was
chosen. This prior has median 5e-04 and 95% interval (0,0.379).
January 16, 2002
5-90
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
5.5.2.2.2 Results of NIEHS analysis
Using the prior described in the previous subsection and "updating" the prior with control
data from the Argus study (i.e., 0 tumors out of 30 control male rats), the NIEHS analysis
estimated that a control rat has a 0.15% chance of developing a thyroid follicular cell adenoma by
19 weeks (Dunson, 2001b). In addition, had perchlorate had no effect on the incidence of thyroid
follicular cell adenomas, the probability of observing two or more rats with these tumors out of
30 would be approximately 0.005. Thus, the data strongly support the hypothesis that
ammonium perchlorate in the drinking water at 30 mg/kg-day causes an increase in the incidence
of thyroid follicular cell adenomas.
5.5.2.2.3 Summary of NIEHS analysis of tumor data
Incorporating historical control data in a Bayesian analysis, a significant increase in thyroid
follicular cell adenoma incidence at 19 weeks in male Sprague-Dawley rats exposed to 30 mg/kg-
day relative to controls was found (Dunson, 2001b). There was no evidence of an increase at low
dose levels. This finding raises concern for in utero imprinting (i.e., that pups exposed in utero
are subsequently more susceptible to thyroid hormone perturbation during post-natal
development and adulthood), a phenomenon that is now appreciated in the endocrine disruption
arena (Prins et al., 2001; Phillips et al., 1998; Seckl, 1997).
5.5.3 Thyroid and Pituitary Hormone Analyses
Thyroid and pituitary hormones were assayed in the PI-generation (both sexes), the
F1-generation adults, the F1-generation pups (PND21) and the F2-generation pups.
In the PI-generation, there was an unexpected and unexplained increase in T3 levels.
Effects on T4 and TSH were as expected, with a significant decrease in T4 and increase in TSH
at the 30 mg/kg-day level.
An anomalous increase in T3 was also reported in the F1-generation adults. Significant
(p ^ 0.01) decreases in T4 of the F1-generation adult males occurred at the high dosage but
increases (p ^ 0.05) at the mid-doses are unexplained; TSH in the adult males was significantly
increased (p ^ 0.01) at the 30 mg/kg-day level. Similar results were reported for the
F1 -generation adult females.
January 16, 2002
5-91
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
In the F1-generation pups, the only statistically significant effects was an unexpected
decrease in TSH at the mid doses in the males and an increase in the females at the lowest.
Similarly seemingly spurious results were observed for the F2-generation pups.
Geller (1999b) presented the EPA analysis of thyroid hormones for this study for the
PI-and F1 -generation using separate repeated-measures ANOVAs with treatment as the
independent between-subjects variable and sex as a within-litter repeated-measures variable.
Mean contrasts were performed using Tukey's Studentized Range (HSD) test. In order to correct
for multiple comparisons, the alpha for significance (for all interaction main effect tests) was
adjusted to 0.029 (alpha of 0.05 divided by the square root of the number of ANOVA tests).
In the PI-generation rats, there was a significant dose effect and dose by sex interaction for
both T4 and TSH. A NOAEL was identified for males only for T4 and TSH at a dose of
3.0 mg/kg-day.
In the F1-generation (weanling pups) on PND21, the contract laboratories reported a
decrease in TSH and an increase in T4. This effect was discounted by Argus Research
Laboratories, Inc. (1999) because the decrease was not dosage-dependent and because TSH
would be expected to increase and T4 to decrease. EPA found similar results with its analyses,
noting that the significant dose effect on female T4 data was due to an elevated level in the
0.3 mg/kg-day group relative to the high dose and also noting that the results were inconsistent
with the mode of action for perchlorate (Geller, 1999b).
A significant increase in TSH was found in the adult F1 (P2 generation) rats at 30 mg/kg-
day; a finding consistent with the tumors observed at this dosage, but T4 and T3 appeared to
have increased in a dose-dependent fashion. Again the reason for this disparity is not clear.
5.6 IMMUNOTOXICITY STUDIES
As discussed in Chapter 3, immunotoxicity studies were included in the perchlorate testing
strategy due to indications in humans and laboratory animals that perchlorate may affect immune
and hematological function. For example, a study by Weetman et al. (1984) that appeared as a
Letter to the Editor in The Lancet, investigated the effect of potassium perchlorate on human
T- and B-cell responses to mitogens in vitro. Perchlorate at concentrations of 0, 0.01, 0.1, and
10 mmol/L (1.17 g/L) were tested in cultures of human peripheral blood lymphocytes. IgG and
January 16,2002 5-92 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
IgM in culture superhatants were measured by ELISA after culture for 10 days with pokeweed
mitogen (PWM). Perchlorate at 0.1 to 10 mmol/L resulted in inhibition of PWM-induced LgG
production at 10 mmol perchlorate/L inhibited IgM production. Lymphocyte proliferation as
measured by 3H-thymidine incorporation was reduced by 33 to 35% in cultures from five of six
individuals in the presence of the T-lymphocyte mitogen phytohemagglutinin (PHA). Weetman
et al. (1984) concluded that perchlorate had significant immunosuppressive activity on
lymphocytes at pharmacologically-relevant concentrations in the absence of cytotoxicity, the
latter of which was assessed by ethidium bromide/acridine orange fluorescence. Unfortunately,
no details were provided as to when viability was determined during the 10 days of lymphocyte
culture with perchlorate and PWM. While these and other results were not sufficient to infer that
perchlorate was immunosuppressive or had other immunotoxic effects, there was uncertainty
with respect to its potential to do so. It was not known whether this could be a direct effect of
perchlorate but could plausibly also be due to its anti-thyroid effects.
An array of 14- and 90-day experiments, to evaluate the effects of drinking water
administration of ammonium perchlorate on immunotoxicological and hematological parameters
were performed using female B6C3F1 (Keil et al., 1998; Kiel et al., 1999; BRT-Burleson
Research Technologies, Inc., 2000a,b,c,) or CBA/J Hsd mice (BRT-Burleson Research
Technologies, Inc., 2000a,b,c). Parameters also were evaluated 30 days after one 90-day study to
assess the reversibility on several observed effects. The mouse was chosen for these studies
because it is the typical experimental species for immunotoxicological studies. In addition, data
were collected on thyroid and pituitary hormones and thyroid histology to provide additional
insight on interspecies variability by comparison with results of the rabbit and rat studies
included in the testing strategy. The mice (8 to 10 weeks of age) were acclimated for 1 week
prior to initiation of any study. Ammonium perchlorate was obtained from the sponsor
(AFRL/HEST), and different lots were used for each of the major study groups (i.e., Keil, et al.,
1998; Keil et al., 1999; BRT-Burleson Research Technologies, Inc., 2000a,b,c,). Primary stock
solutions were prepared approximately every 1 to 2 months, and dosing solutions were prepared
weekly. In the Keil et al. (1998) studies, there was an indication of a trend that mice exposed at
the 30 mg/kg-day level consumed slightly less water on a weekly basis (=3 mL/week less than
control). Consequently, differences were noted in the actual exposure for the high-dose group in
the 14-day studies. This difference was not as marked in the 90-day studies. Concentration of
January 16, 2002
5-93
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
dosing solutions was verified by the sponsor (AFRL/HEST; data not shown). The one apparent
disparity in dose level (0.1 mg/kg-day; experiment not specified) was rectified after
reexamination of calculations (data not shown) (Keil et al., 1998). The mice were exposed to
levels of 0, 0.1, 1.0, 3.0, or 30 mg/kg-day in the Keil et al., (1998, 1999) studies; while in the
BRT-Burleson Research Technologies, Inc. (2000a,b,c) studies, the mice were exposed to levels
of 0, 0.02, 0.06, 0.2, 2.0 or 50 mg/kg-day. The doses were established based on the mean body
weight for each treatment group per week. Each dose group generally consisted of 6 to 10 mice,
with the exception that some control groups in the BRT-Burleson Research Technologies, Inc.
(2000a,b,c) studies had a group size of 20.
A number of 14-day experiments were conducted. In Experiments "C", "G", "I", "J", "T",
and "K" (Keil et al., 1998), the mice were sacrificed at Day 14; and body weight, organ weight
and cellularity (thymus, spleen, liver, and kidney), a number of immunotoxicology and
hematological parameters, thyroid histology, and thyroid and pituitary hormone levels were
measured. These data are summarized in Tables 3, 6, 9, 12, 14, 16, 18, and 21 of the "Final
Report" (Keil et al., 1999). In Experiments "U" and "V" (Keil et al., 1998), mice were
challenged with sublethal amounts (2,300 or 2,700 colony-forming units [CFU]) of Listeria
monocytogenes on Day 7 and then sacrificed on Day 14. The spleens were removed for a
delayed-type hypersensitivity (DTH) assay (Keil et al. 1999: Table 31). In experiments "H",
"F", and "M" (Keil et al., 1998), mice were challenged with P815 tumor cells by ip injection.
At the 14-day terminal sacrifice, spleens were removed for the cytotoxic T lymphocyte (CTL)
activity assay (Keil et al., 1999: Table 23).
A series of 90-day experiments also were conducted. In Experiments "A", "D", and "N"
(Keil et al., 1998), mice were sacrificed after 90 days; and body weight, organ weight and
cellularity (bone marrow, thymus, spleen, liver, and kidney), a number of immunotoxicology and
hematological parameters, thyroid histology, and thyroid and pituitary hormone levels were
measured (Keil et al., 1999: Tables 4, 7, 10, 13, 15, 17, 19, 20, and 22). In Experiments "B" and
"E" (Keil et al., 1998), these same endpoints were measured after a 30-day recovery period (Keil
et al., 1999: Tables 5, 8, 11, and 22,). In Experiment "P" (Keil et al., 1998), mice were
challenged with P815 tumor cells by ip injection on Day 76. Spleens were removed at terminal
sacrifice for the CTL activity assay (Keil et al., 1999: Table 24).
January 16, 2002
5-94
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Two host resistance models, one a bacteria and the other a tumor, were used in 90-day
experiments. Mice in Experiment "L" (Keil et al., 1998) were challenged with Listeria
monocytogenes by iv injection. At terminal (90-day) sacrifice, spleens and livers were removed
and cultured for L. monocytogenes growth. Unfortunately, the challenge concentration (i.e.,
5360 CFU) of bacteria used was excessive, thereby prohibiting enumeration of the bacteria in the
spleens of these mice. A second 90-day L. monocytogenes-chaUsnge experiment (Keil et al.,
1999) was subsequently undertaken using a lower (i.e., 2700 CFU) challenge concentration (see
Keil et al., 1999: Table 34). For the tumor model, in Experiments "Q" and "O" (Keil et al.,
1998), mice were challenged with B16F10 tumor cells by iv injection on Day 76. At the 90-day
sacrifice, the lungs were removed, and the number of tumor nodules in both lungs were
enumerated (Keil et al. 1999: Table 33).
The IgM and IgG antibody responses to sheep red blood cells (SRBCs) of mice exposed to
ammonium perchlorate for 90 days and the IgM anti-SRBC response of mice exposed for 14 days
was determined using an enzyme linked immunosorbent assay (ELISA) (figures on page 59, Keil
et al., 1999). Based on EPA comments in 1998 and external peer review recommendation
(Research Triangle Institute, 1999), a second contract was let to determine the antibody response
to SRBCs using the more traditional antibody plaque-forming cell (PFC) assay (BRT-Burleson
Research Technologies, Inc., 2000a,b,c). Unlike the ELISA, which measures SRBC-specific
IgM antibody in serum, the PFC assay quantifies the number of plasma cells in the spleen which
produce SRBC-specific IgM. The potent immunosuppressant cyclophosphamide (CP) was used
as a positive control in these latter studies. In both the 14- and 90-day studies, mice were
immunized iv with SRBCs 4 days prior to assay. The positive control mice were injected ip with
15 mg/kg-day CP on the last 4 days of dosing prior to assay.
Concern about potential effects of ammonium perchlorate on contact hypersensitivity, also
raised at the 1999 external peer review, were addressed in studies performed by Burleson et al.
(2000). Eight-week-old female CBA/J Hsd mice that had been acclimated one week prior to
dosing were exposed to 0, 0.02, 0.06, 0.2, 2.0, or 50.0 mg/kg-day for 14 or 90 days. The contact
sensitizer, 2,4-dinitrochlorobenzene (DNCB), was applied to the surface of both ears on days 9,
10 and 11 in the 14-day study, and on days 92, 93, and 94 in the 90-day study. Mice were
assayed using the local lymph node assay (LLNA) on day 14 and 97 for the 14-day and 90-day
January 16, 2002
5-95
DRAFT-DO NOT QUOTE OR CITE
-------
]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
studies respectively. A CP-positive control group was included in each study, with
administration of 15 mg/kg-day CP ip for 5 consecutive days prior to assay.
Data from the Keil et al. (1998, 1999) studies were analyzed as follows. Initially, analysis
of variance was performed using Tukey's multicomparison (p < 0.05) for the various parameters
measured. A Fisher's multicomparison test was used in previous interim reports but not in the
final one. The previous analyses reported effects. Consequent to criticisms of the analyses
performed, as stated in the previous external review Draft Toxicological Review Document on
Perchlorate (U.S. Environmental Protection Agency, 1998d), and reinforced by the comments of
Dr. Kimber White at the previous external peer review (Research Triangle Institute, 1999), these
and new data (i.e., the 14-day antibody response to SRBCs and the 90-day host resistance to
L. monocytogenes) were analyzed as indicated in the "Final Report" (Keil et al., 1999). That is,
data were combined from two or three experiments and evaluated by the Kolmogorv-Smimov
test for normality and Bartlett's test for homogeneity of variance. If data displayed a normal
distribution and equal variance, two-way ANOVA, with experiments and treatments as factors,
was performed. Tukey's pairwise comparison was performed to determine differences (p<0.05)
between control and treatments if no interaction was identified due to combining multiple
experiments. If a significant interaction was identified in the ANOVA, data from each
experiment were analyzed using one-way ANOVA and Tukey's pair wise analysis. The Kruskal-
Wallis test was used if data were not normally distributed or variances were not equal; and if
significant, the Mann-Whitney test was employed to determine differences (p<0.05) between
control and treatments.
The results of the BRT-Burleson Research Technologies, Inc. (2000a,b,c) studies were
analyzed as follows. Data from each treatment group were compared by first performing a
Bartlett's Chi-Square test for variance of homogeneity. If found to be non-significant, ANOVA
was employed using dose. If significant, then Dunnett's t-test was performed, with p<0.05 being
significant. On the other hand, if Bartlett's Chi-square was significant, the non-parametric
Kruskal-Wallis test was performed, which if significant was followed by a Jonckheere's-Terpera
test for dose-dependent trends. The parametric ANOVA and the non-parametric extended
Cochran-Mantel-Haenszel test were performed to determine whether the data could be pooled.
Results for the general toxicity and organ weight measures will be discussed in
Section 5.6.1. Thyroid histopathology evaluations will be reported in Section 5.6.2, and analyses
January 16, 2002
5-96
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
of T3, T4, and TSH in Section 5.6.3. Results for the immunotoxicological and hematological
parameters are discussed in Sections 5.6.4 and 5.6.5. A summary of the results and their
potential significance is presented in Section 5.6.6.
5.6.1 Results for General Toxicity, Organ Weight, and Cellularity Measures
There were no effects observed on body, thymus, spleen, liver, or kidney weights in the
14-, 90-, or 120-day studies (Keil et al., 1999: Tables 6-8). Earlier interim reports indicated
considerable variability in the splenic and thymic cellularity of ammonium perchlorate-exposed
mice. This variability was due, in large part, to technical errors. Recognizing this, the contractor
performed additional studies (i.e., "on at least two or more occasions") in which "no significant
changes in cellularity were observed." (Keil et al., 1999). As such, in the "Final Report" no
consistent alteration in splenic or thymic cellularity was observed in the 14-, 90-, or 120-day
studies (Keil et al., 1999: Tables 9-11), nor in splenic lymphocyte CD4/CD8 subsets (Keil et al.,
1999: Tables 14 and 15). With the exception that CD4-CD8+ thymic lymphocytes were
increased in mice exposed to 0.1- and 1.0-mg/kg-day doses in the 14-day experiment, there were
no other alterations in thymocyte subsets observed in the 14- or 90-day studies (Keil et al., 1999:
Table 12). Furthermore, there were no alterations in the number of peritoneal macrophages
obtained from mice exposed to any doses of ammonium perchlorate in the 14-, 90-, and 120-day
studies (Keil et al., 1999: Tables 9-11), nor in bone marrow cellularity in the 14- and 90-day
studies (Keil et al., 1999: Tables 9 and 10). Due to the absence of effects in the latter studies, no
120-day study was performed.
5.6.2 Evaluation of Thyroid Histology
Thyroid histopathology evaluation was performed for two experiments (A and D) in the
Keil et al. (1998) study and eventually published in the fmal report (Keil et al., 1999). These data
were transmitted by Warren (1999), and a preliminary review by EPA was presented at the 1999
external peer review (Jarabek, 1999). The materials were provided to the PWG review, and the
results are found in Wolf (2000, 2001; Table 23). These results corroborate the preliminary
analyses that showed decreased colloid, follicular hypertrophy and hyperplasia to occur at the
30 mg/kg-day dose. Congestion in the intrafollicular capillaries and the nuclear to cytoplasmic
January 16, 2002
5-97
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
ratio of the follicular cells were not recorded by the PWG but were both noted in the Warren
(1999) report at 30 mg/kg-day (Jarabek, 1999). Hypertrophy was additionally observed in the
lower doses of experiment "A", and the reason for the disparity between the two studies is
unclear. These results support the assertion that the hypothalamic-pituitary-thyroid feedback
regulatory mechanism is conserved across species (rats, rabbits, mice and humans) and suggest a
NOAEL of 3 mg/kg-day in this strain of mouse.
5.6.3 Thyroid and Pituitary Hormone Analyses
The report (Keil et al., 1998) contains thyroid hormone and thyrotrophin (TSH) data from
14- and 90-day exposures to ammonium perchlorate in B6C3F1 mice. The following is a
statistical analysis of the thyroid and pituitary hormone data (T4 and TSH) found in that report.
There were no data for T3 reported in the original study submitted to EPA (Keil et al., 1998).
The EPA reanalyzed the data that were supplied in Excel® spreadsheets to EPA by Dr. Deborah
Keil, and the data are published therein (Crofton, 1998i). Subsequent submission of additional
data files also containing data for T3 were included in reanalyses (Crofton, 2001a). Data for
dependent measures (T4 and TSH) were subjected to separate analyses. The T4 and TSH data
were analyzed with a two-way ANOVA, with duration (14, 90, and 120 days) and treatment
(dose) as the independent between-subjects variables as reanalyzed by Crofton and Marcus
(2001) and Crofton (2001a) as reported in Table 5-2. Mean contrasts were performed using
Duncan's Multiple Range Test.
Results of these EPA reanalyses are different from those stated in the Keil et al. (1998)
report. The EPA reanalysis of the T3 data (Crofton, 2001a) found main effects of time and
treatment, but no time-by-treatment interaction. Mean contrast testing showed a LOAEL of
0.1 mg/kg-day; however, the dose-related decrease was not linear. The 0.1 and 3.0 mg/kg-day
doses differed from controls but the 1.0 and 30.0 mg/kg-day doses did not. There was a
significant time-by-treatment interaction for T4. After 14 days of exposure there was no effect
with a NOAEL at 30 mg/kg-day; whereas, after 90 days of exposure the LOAEL was 0.1 mg/kg-
day. T4 recovered after 30 days postexposure. There was no effect of perchlorate on TSH
concentration contrary to the changes in histopathology discussed in Section 5.6.2.
January 16, 2002
5-98
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
These effects are of interest in that they demonstrate effects in mice comparable in nature to
that in rats and indicate that the hypothalamic-pituitary-thyroid feedback system is conserved
across species.
5.6.4 Results of Immune Function Assays
No consistent alteration in CTL activity was observed in three 14-day studies ("M", "H",
and "F", Keil et al., 1998). No effects were observed on CTL activity in Experiments "M" and
"H". However, in Experiment "F", increases in CTL activity were observed at the 0.1-mg/kg-day
ammonium perchlorate dose for effector to target cell (E:T) ratios of 100:1, 30:1, and 10:1, and,
at the 1- and 3-mg/kg-day doses, for an E:T ratio of 10:1. In a 90-day study ("P", Keil et al.,
1998) there were no alterations in CTL activity at any dosages or E:T ratios. The variability and
inconsistencies observed in the early interim reports were ascribed to "technical issues" that were
consequently "corrected". In fact, the data presented in Tables 23 and 24 (Keil et al., 1999)
which includes data for dexamethasone, a potent immunosuppressant and positive control,
indicates that there were no effects of ammonium perchlorate AP exposure on CTL activity.
There was also no consistent alteration in the DTH response, as measured by the
lymphoproliferation (LP) of splenic lymphocytes from L. monocytogenes-challenged mice
incubated with soluble Listeria antigen (SLA) in two 14-day studies ("U" and "V", Keil et al.,
1998). The LP response was increased only in cultured splenic lymphocytes from mice in the
30-mg/kg-day group stimulated with 0.1 //g/mL SLA in Experiment "U" and in cultures of
splenic lymphocytes from mice in the 3-mg/kg-day group stimulated with 8 /^g/mL SLA in
Experiment "V" (Keil et al., 1998). The "Results Summary and Status" page of Keil et al. (1998)
indicates that a 90-day DTH study was planned. These 90-day data and a summary of the 14-day
data are presented in Tables 32 and 31 respectively, of the "Final Report" (Keil et al., 1999). The
data indicated an enhanced LP response in mice dosed at 30-mg/kg-day in both the 14-day and
90-day studies.
No alteration in splenic natural killer (NK) cell activity was observed in two 14-day studies
("G" and "T", Keil et al., 1998). The 14-day Experiment "T" data are presented in a table;
however, the raw data and statistics for this study were not found in the submission. Inconsistent
results were obtained in two 90-day studies ("D" and "N", Keil et al., 1998) in which NK cell
activity was increased at the 30-mg/kg-day ammonium perchlorate in Experiment "N"; however,
January 16,2002 5-99 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
no effects were observed at any doses in Experiment "D". A similar increase in NK cell activity
at the 30-mg/kg-day dose was observed in the 120-day Experiment "E" (see also the data in
Tables 21-22, Keil et al., 1999, in which the positive control dexamethasone was employed).
The lack of any change in the number of B16F10 tumor nodules in the lungs of mice from the
90-day "Q" study (Keil et al., 1998; see also Table 33, Keil et al., 1999), particularly at
30 mg/kg-day, suggests that the increased NK-activity does not reflect a significant biological
effect (see below). The EPA notes that there is a good deal of variation in NK activity data for
the controls in the 14-day "G" study, the 90-day "D" and "N" studies, and the 120-day "E" study,
which were 34, 6.4, 13.6, and 18.4 lytic units/107 splenic lymphocytes, respectively. Also, the
14-day "G" study was not included in Table 33 (Keil et al., 1999).
Decreased in vitro phagocytosis of L. monocytogenes was observed at 3 and 30 mg/kg-day
of ammonium perchlorate in the 14-day "C" and 90-day "A" studies (Keil et al., 1998). In the
90-day "N" study, macrophage phagocytosis was decreased in all dose groups. However, in the
14-day "G" and 90-day "D" studies and in two 120-day studies ("B" and "E"), no effect on
macrophage phagocytosis was observed (Keil et al., 1998). In the "Final Report" (Keil et al.,
1999), these alterations were confirmed (i.e., decreased phagocytosis at 1.0 and 30.0 mg/kg-day
in the 14-day study (Keil et al., 1999: Table 27) and decreased phagocytosis at 0.1, 1.0, 3.0, and
30.0 mg/kg-day in the 90-day study (Keil et al., 1999: Table 28). However, after a 30 day
recovery period (i.e., 120-day study, Keil et al., 1999: Table 29) phagocytic function was
comparable across control and treated mice. These data suggest that ammonium perchlorate
suppresses the phagocytic capacity of peritoneal macrophages in vitro, but that this suppression
may be reversed after a 30-day recovery period. Criticism of the use of an in vitro rather than an
in vivo assessment of macrophage function was raised in the 1998 EPA ERD document and at
the 1999 external peer review by Dr. Kimber White (Research Triangle Institute, 1999).
No consistent alteration in peritoneal macrophage nitrite production was observed in 14-,
90-, and 120-day studies. Increased nitrite production from macrophages cultured with interferon
(IFN) occurred at doses of 3 and 30 mg/kg-day and from macrophages cultured with IFN and
lipopolysaccharide for the 30-mg/kg-day dose in the 90-day "D" study (Keil et al., 1998). Also,
increased nitrite production from macrophages cultured with IFN was observed at 3 mg/kg-day in
the 90-day "N" study (Keil et al., 1998). An increase in nitrite production for macrophages
cultured with IFN or LPS alone also occurred for the 30-mg/kg-day group in the 120-day "B"
January 16, 2002
5-100
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
study (Keil et al., 1998). These data suggest a "trend" toward increased nitrite production at the
higher doses of ammonium perchlorate.
A subsequent analysis of these data, as presented in Tables 25 and 26 (Keil et al., 1999),
demonstrates "no significant difference in nitrite production of peritoneal macrophages" (Keil
et al., 1999).
A 90-day study ( "L", Keil et al., 1998) was performed to determine if exposure of mice to
ammonium perchlorate results in alterations in resistance to infection with L. monocytogenes.
A trend toward increased resistance was suggested by the data; however, technical difficulties
were encountered. For example, there was variability in the number of L. monocytogenes CFU/g
liver recovered from control mice. It was not possible to enumerate the number of CFU/g spleen
in mice due to the high concentration of bacteria injected and also to an inadequate dilution of
spleen cell suspensions. In a subsequent 90-day study, mice were challenged with a lower
concentration of bacteria such that both the CFU/g liver and spleen could be determined. These
results, presented in Table 34 (Keil et. al., 1999), indicate that ammonium perchlorate exposure
does not alter resistance to infectious challenge to L. monocytogenes.
No effects were observed in an initial 90-day B6F10 tumor challenge host-resistance model
experiment ("Q", Keil et al., 1998). Another 90-day B6F10 tumor challenge experiment (i.e.,
"O") was performed, and the combined results of these two experiments are presented in
Table 33 (Keil et al., 1999). These data indicate that there were no differences in the number of
tumors present in the lungs of ammonium perchlorate-exposed mice compared with control mice.
Two separate groups of studies examining the effect that ammonium perchlorate has on the
antibody response to SRBCs were performed by independent contractors (Keil et al, 1999;
BRT-Burleson Research Technologies, Inc, 2000a,b,c). Initial studies were performed by Keil
et al. (1999), in which the IgM and IgG antibody responses were determined using ELISAs.
As indicated in the figures on page 59 (Keil et al., 1999), no change in the IgM levels in a 14-day
study, nor in the IgM and IgG levels in a 90-day study, was observed between control and any
ammonium perchlorate treated mice .
In the second set of studies, the anti-IgM SRBC PFC assay was employed (BRT-Burleson
Research Technologies, Inc, 2000a,b,c), using CP as a positive immunosuppressant control.
In the 14-day study there were no differences in the PFC response between control and treated
mice when expressed either as the number of PFC/spleen or PFC/106 spleen cells (BRT-Burleson
January 16, 2002
5-101
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Research Technologies, Inc, 2000a,b,c: Figures 3 and 4). On the other hand, in the 90-day study
the PFC response was increased in the 2.0 and 50.0 mg/kg-day groups when expressed as the
PFC/spleen and increased only in the 50.0 mg/kg-day group expressed as PFC/106 spleen cells
(BRT-Burleson Research Technologies, Inc., 2000a,b,c: Figures 5 and 6). This disparity was not
due to any difference in splenic cellularity between the control and treated mice. In both the
14- and 90-day studies, CP significantly inhibited the PFC response, expressed either as
PFC/spleen or PFC/106 spleen cells compared to the controls.
The results of the effect that 14- and 90-day exposure to ammonium perchlorate has on the
development of a contact hypersensitivity response to DNCB, as determined by the LLNA,
indicate that an ammonium perchlorate dose as low as 0.06 mg/kg-day enhances this response.
In the 14-day study, the LLNA was increased at doses of 0.06, 0.2, and 50.0 mg/kg-day, but not
2.0 mg/kg-day (BRT-Research Technologies, Inc., 2000a,b,c: Figure 8). The results of the
90-day study were somewhat different in that, while the LLNA was enhanced at 0.06 and
0.2 mg/kg-day, it was suppressed at 50 mg/kg-day (BRT-Research Technologies, Inc., 2000a,b,c:
Figure 9). Another disparity between these two studies was that while CP suppressed the LLNA
in the 14-day study, it did not suppress this response in the 90-day study.
5.6.5 Results for Evaluations of Hematological Parameters
There were no differences observed between control and dosed mice in 14- or 90-day
experiments for erythrocyte cell count, hemoglobin, hematocrit, mean corpuscular volume, mean
corpuscular hemoglobin, and mean corpuscular hemoglobin concentration; nor in leukocyte
differential counts of neutrophils, monocytes, and lymphocytes. Because of the absence of
effects in these studies, no 120-day study was performed. No effects were observed in a single
14-day study (Experiment "T", Keil et al., 1998) on platelet counts. An increase in the
percentage of reticulocytes was observed in the peripheral blood of mice exposed to 3 mg/kg-day
of ammonium perchlorate in a 90-day study ("N", Keil et al., 1998). No other reticulocyte data
are available because of "the minimal availability of blood obtained from each mouse" in other
studies (Keil et al., 1998). In a subsequent 14-day study, there were no changes in the
hematological parameters examined between control and ammonium-perchlorate-treated mice
(Keil, et al., 1999: Table 16).
January 16, 2002
5-102
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
No consistent alteration in the bone marrow stem cell assay was observed. An increase in
the number of colony-forming units was observed in bone marrow cell cultures from mice dosed
at 30 mg/kg-day in a 14-day study ("K", Keil et al., 1998). However, there was no effect of
ammonium perchlorate exposure on the stem cell assay in a 90-day study ("D", Keil et al., 1998).
In a subsequent 90-day study, while no alteration in the stem cell assay was observed between
control and ammonium perchlorate-treated mice, exposure to the positive control dexamethasone
resulted in suppression of the stem cell assay (Keil, et al., 1999: Table 20).
5.6.6 Results Summary
The results of the various studies of immue function are summarized in Table 5-5.
Although innate (i.e., macrophage and NK cell function) and cell-mediated (i.e., cytotoxic
T lymphocytes [CTL], CD4, and CD8) immune functions were evaluated in the initial studies by
Keil et al, (1998), EPA noted that humoral immunity (i.e., B cells and antibody response) was not
(Smialowicz, 1999). The EPA suggested strongly that the antibody response to SRBCs is one of
the most commonly effected functional parameters in animals exposed to chemical
immunosuppressants (Luster et al., 1988). In fact, it is one of the assays required by EPA for test
rules. The EPA also requested that an additional 90-day L. monocytogenes host-resistance study
be undertaken consequent to technical problems associated with the initial 90-day study (Keil
et al., 1998). As such, the EPA felt that these data would provide a more comprehensive
evaluation of the potential for immunosuppression by ammonium perchlorate. In addition, the
EPA requested that thyroid histology and thyroid and pituitary hormone data be obtained in order
to provide additional insights on interspecies variability for this effect.
Consequently, the sponsor and investigators, Keil et al. (1998), agreed to perform these
assays, the results of which are presented in the "Final Report" (Keil et al., 1999).
Subsequent to receipt of the results of the antibody response to SRBCs (Keil et al., 1999),
in which antibody titers were expressed as the O.D. 50 or midpoint titer, rather than the more
conventional titer to achieve a 0.5 O.D., a second request to determine the potential effects of
ammonium perchlorate on the response to SRBCs was issued. In this same solicitation, the EPA
also requested that a sensitization test be performed. The results of these studies are found in
BRT-Burleson Research Technologies, Inc. (2000a, b, c).
January 16, 2002
5-103
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-5. SUMMARY OF IMMUNOTOXICITY TEST RESULTS
Scrics/Strain/Scx (Study)
Exposures
Period and
Doses (mg/kg/d)
Endpoint
NOAEL/LOAEL
Designations
Mouse/B6C3Fl/Female
14-days 0, 0.1,
Weights: body, thymus,
None
(Keil et al., 1998; 1999)
1.0,3.0, or 30
spleen, liver, kidney
Cellularity: spleen, thymus,
None
bone marrow
Splenic CD4CD8 cells
None
NK cell activity/B 16F10
None/Not Done
tumor challenge
CTL to P815 cells (in vitro)
Increased at 0.1, 1.0 and 3.0; no
effect in subsequent "corrected"
study.
L. monocytogenes challenge
Not Done
DTH to L. monocytogenes
Increased at 30.
antigen
NOAEL = 3.0, LOAEL = 30
Macrophage phagocytosis
Decreased at 1.0 and 30.
(in vitro)
NOAEL = 0.1, LOAEL = 1.0
Macrophage nitrate (in vitro
None
+ IFN or LPS)
IgM ELISA to SRBCs
None
90-days 0, 0.1,
Weights: body, thymus,
None
1.0,3.0, or 30
spleen, liver, kidney
Cellularity: spleen, thymus,
None
bone marrow
Splenic CD4CD8 cells
None
Mousc/B6C3Fl/Female
90-days 0, 0.1,
NK cell activity/B 16F10
Increase NK activity at 30 in
(Kiel ct al., 1998; 1999)
1.0, 3.0, or 30
tumor challenge
one experiment and at 30 in
120-day study: NOAEL = 3,
LOAEL = 30. No effect on
B16F10 tumor challenge.
CTL to P815 cells (in vitro)
None
L. monocytogenes challenge
None
DTH to L. monocytogenes
Increase at 30.
antigen
NOAEL = 3.0, LOAEL = 30
Macrophage phagocytosis
Decreased at 0.1, 1.0, 3.0
(in vitro)
and 30, LOAEL = 0.1
January 16, 2002
5-104
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 5-5 (cont'd). SUMMARY OF IMMUNOTOXICITY TEST RESULTS
Exposures
Period and NOAEL/LOAEL
Series/Strain/Sex (Study) Doses (mg/kg/d) Endpoint Designations
Macrophage nitrate
None
(in vitro + IFN or LPS)
IgM EL1SA to SRBCs
None
Mouse/CBA/JHsd/Female
14-days 0, 0.02,
anti-SRBC PFC/106 cells
None
(BRT-Burleson Research
0.06, 0.2, 2.0,
Technologies, Inc.,
or 50
2000a,b,c)
anti-PFC/spleen
None
LLNA to DNCB
Increased at 0.06, 0.2, and 50,
but not at 2.0. NOAEL = .02,
LOAEL = 0.06
90-days 0, 0.02,
anti-SRBC PFC/106 cells
Increased at 50.
0.06, 0.2, 2.0,
NOAEL = 2.0, LOAEL = 50
or 50
anti-PFC/spleen
Increased at 2.0 and 50.
NOAEL = 0.2, LOAEL = 2.0
LLNA to DNCB
Increased at 0.06 and 0.2, but
not at 2.0; decreased at 50.
NOAEL = 0.02, LOAEL = 0.06
NK = natural killer; CTL = cytotoxic lymphocyte; DTH = delayed type hypersensitivity; IFN = interferon;
SRBC = sheep red blood cell; PFC = plaque forming colony; LLNA = local lymph node assay; DNCB =
2,4-Dinitrochlorobenzene.
1 The three immune function parameters altered by ammonium perchlorate exposure were
2 the following: suppression of in vitro peritoneal macrophage phagocytosis of L. monocytogenes,
3 enhancement of the PFC response to SRBCs, and enhancement of the LLNA to DNCB. These
4 results are summarized and discussed below.
5 Decreased in vitro phagocytosis of L. monocytogenes by peritoneal macrophages obtained
6 from mice dosed for 14 days at 1- or 3- and 30-mg ammonium perchlorate/kg-day was observed
7 (Keil et al, 1998, 1999). In mice exposed for 90-days, phagocytosis was decreased in all dosage
8 groups (Keil, 1998, 1999). However, in the 120-day (i.e., 90-day ammonium perchlorate
9 exposure followed by 30-day recovery) studies, no effect on macrophage phagocytosis of
10 L. monocytogenes was observed (Keil et al., 1998,1999). Taken together, these data suggest that
January 16, 2002
5-105
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
ammonium perchlorate suppresses the in vitro phagocytic capacity of peritoneal macrophages,
but that this suppression is reversed after a 30 day recovery period.
This decrease in macrophage phagocytic activity could be expected to be reflected in the
results of the L. monocytogenes infectivity data because, along with other immune system
components, macrophages play a pivotal role in resistance to infection by this bacterium.
For example, the pathogenesis of L. monocytogenes is associated with its ability to grow within
mononuclear phagocytes. Complement (C') plays an important role in L. monocytogenes
infections, as demonstrated by the fact that C'-deficient mice have impaired host resistance to
this bacterium. This impairment in C'-deficient mice is caused by the absence of macrophage-
associated C'. The T-lymphocytes also play a major role in defense against L. monocytogenes
because complete elimination of bacteria from infected tissue is accomplished by macrophages
activated by T-cell dependent mechanisms.
However, the L. monocytogenes host-resistance studies indicate that ammonium
perchlorate exposure of mice does not alter the ability to combat this bacterial infection. With
the exception that clearance of L. monocytogenes from the liver of mice given a 5360 CFU
challenge following dosing at 3.0 mg AP/kg/day for 90 days was reduced, no other effect was
observed (Keil et a., 1999: Table 43). These data imply that while in vitro phagocytosis by
peritoneal macrophages of this bacterium was reduced following ammonium perchlorate
exposure, the ability of macrophages from other in situ sites (e.g., spleen, liver) to clear
L. monocytogenes was not altered.
Exposure of mice to 2.0 or 50.0 mg ammonium perchlorate/kg/day for 90, but not 14, days
resulted in enhancement of the antibody response to SRBCs as determined by the PFC assay
(BRI-Burleson Research Technologies, Inc., 2000a,b,c). In both the 14- and 90-day studies, the
PFC response was suppressed by dosing mice with the immunosuppressive positive control CP.
The PFC assay is routinely used for identifying chemicals that are immunosuppressive. The
reason why the highest dose(s) of ammonium perchlorate, given over 90 days, enhanced this
response is not known. It is possible that under these dosing conditions ammonium perchlorate
may have an adjuvant-like or enhancing effect on the antibody response to SRBCs. The ELISA
data for mice exposed to up to 30.0 mg ammonium perchlorate/kg/day, for 14 or 90 days (Keil
et al., 1999), do not corroborate this enhanced response to SRBCs as determined by the PFC
January 16, 2002
5-106
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
assay. However, taken together, the PFC and ELISA data indicate that ammonium perchlorate
does not suppress the immune response to SRBCs.
The LLNA is an accepted approach for identifying chemicals with the potential of causing
dermal contact hypersensitivity (CHS) reactions in humans. In this assay the test substance,
2,4-dinitrochlorobenzene (DNCB) was topically applied on three consecutive days to both ears of
the mouse. Two days later the mice were injected iv with radioactive uridine (e.g., 125IUDR).
Five hours later, the lymph nodes draining the ears, referred to as the "auricular" lymph nodes,
were removed and 125IUDR incorporation by the lymph node cells determined. Since the nodes
draining the ear (i.e., "auricular" nodes) have no standard anatomical nomenclature, experience
in identifying these nodes as well as appropriate and consistent excision of these nodes from
control and test mice is critical. The LLNA evaluates the induction phase of the CHS reaction by
assessing the influx of lymphoid cells and the differential argumentation of lymphocyte
proliferation elicited by exposure to the test chemical relative to that of a vehicle control.
The data from BRT-Burleson Research Technologies, Inc. (2000a,b,c) report that exposure
to ammonium perchlorate enhances/exacerbates the LLNA response to DNCB at doses of 0.06,
and 0.2 mg/kg/d in both the 14- and 90-day. While a dose of 50.0 mg/kg-day for 14 days also
enhanced this response, a dose of 2.0 mg/kg-day did not. Similarly, a dose of 2.0 mg/kg-day in
the 90-day study did not enhance the LLNA response to DNCB. In contrast to the 14-day study,
exposure of mice to 50.0 mg ammonium perchlorate/kg/day in the 90-day study resulted in
suppression of the LLNA response. In the 90-day study, the positive control CP did not suppress
the LLNA response to DNCB. The failure of CP to suppress this response in the 90-day vs.
14-day study is disquieting because CP was administered similarly (i.e., 15 mg/kg-day for
5 consecutive days prior to the LLNA) in both studies. The only difference between these two
studies was that the mice in the 90-day study were 11 weeks older. This difference in age,
however, should not influence the ability of CP to suppress this response. The fact that CP did
not suppress the LLNA response in the 90-day study calls into question the performance of this
and perhaps the 14-day study.
Application of the LLNA for identification of chemicals that are contact sensitizers
routinely involves a demonstration of a dose-related increase in the LLNA using, at a minimum,
three increasing concentrations of the test agent. Neither the 14- nor 90-day ammonium
perchlorate LLNA data demonstrate a dose-response relationship, which would be expected if
January 16, 2002
5-107
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
ammonium perchlorate was acting additively or synergistically with DNCB to increase the
LLNA response. While higher concentrations of a contact sensitizing agent will increase the
LLNA response, there is no information in the literature that indicates such an increase results in
a more serious or potentially detrimental effect on the host. Consequently, the physiologic
significance of the observed increase in the LLNA response to DNCB in ammonium perchlorate-
exposed mice is unknown. This is unlike the situation with immunosuppressive agents where
suppression of specific immune function(s) can be linked to a biological detrimental effect (i.e.,
decreased host resistence to an infectious agent or tumor).
It is interesting to note that there are published reports in which non-sensitizing agents have
been employed to improve the sensitivity of the LLNA to detect sensitizers. For example,
Vitamin A acetate dietary supplementation enhances the detection of weak sensitizers, and at low
concentrations of moderate sensitizers, assessed by the LLNA. The mechanism(s) for this
increased detection of contact sensitizers is not known. However, topically applied retinol causes
epidermal hyperplasia which may lead to increased numbers of antigen-presenting cells in the
epidermis. Retinoids also up regulate the sensitization phase of DTH induction through direct or
indirect stimulation of T cells. Non-sensitized mice, fed a diet supplemented with retinol,
display somewhat higher LLNA responses compared to control mice on a normal diet. This
suggests that dietary retinol itself causes cellular infiltration and/or proliferation in the absence of
a contact sensitizer as measured by the LLNA. It may be that ammonium perchlorate, in the
absence of DNCB, has the capability of raising the baseline LLNA response compared to water
control mice. Unfortunately, there were no negative controls in the Burleson et al. (2000)
studies. Appropriate negative controls would have included the following: (1) ammonium
perchlorate-dosed non-sensitized mice; (2) ammonium perchlorate-dosed and ammonium
perchlorate-challenged mice; and (3) water control mice dermally exposed to ammonium
perchlorate on the ear pinna. Another group of appropriate and informative studies would
involve ammonium perchlorate-dosed mice that would be challenged with a series of low to
moderate concentrations of DNCB, for comparison with the current LLNA "optimal DNCB"
response concentration data.
Enhancement of the LLNA to DNCB in mice exposed to 0.06 mg ammonium
perchlorate/kg-day for 14 or 90 days represents the Lowest Observed Effect Level (LOEL) for all
of the immune function tests performed. While this is the LOEL it is unknown if this is the
January 16, 2002
5-108
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Lowest-Observed-Adverse-Effect- Level (LOAEL) because it is not clear that enhancement of
the LLNA is a physiologically relevant adverse effect. Studies are needed to determine if
ammonium perchlorate itself is a contact sensitizer as determined by the LLNA, as described
above, and whether the degree of the LLNA response to ammonium perchlorate itself or to a
known contact sensitizer can be linked to a quantifiable adverse outcome.
It is important to note that clinical studies in the 1960s reported that some patients suffering
from Graves' disease and treated with potassium perchlorate presented with agranulocytosis
and/or skin rashes. While the studies reported by Keil et al. (1998, 1999) indicated that there
were no alterations in the proportion of peripheral blood leukocytes of mice dosed with
ammonium perchlorate for 14- or 90-days, the work of BRI-Burleson Research Technologies,
Inc. (2000a,b,c) suggests that ammonium perchlorate appears to exacerbate the contact
sensitizing potential of the known skin sensitizer DNCB. However, due to the uncertainties
associated with any attempt to extrapolate from the incomplete database of the mouse LLNA
performed by BRI-Burleson Research Technologies, Inc. (2000a,b,c) to the clinical observations
of skin rash and agranulocytosis in Graves' disease patients treated with potassium perchlorate,
an uncertainty factor based on deficiencies in the database is recommended to be applied to this
risk assessment.
January 16, 2002
5-109
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
CHAPTER 6. CONSTRUCTION OF PBPK MODELS
TO ADDRESS PERCHLORATE'S MODE-OF-ACTION
The purpose of this chapter is to describe the progress that has been made in developing
physiologically-based pharmacokinetic (PBPK) models to aid interspecies extrapolation of
effects observed in the toxicity studies. The models describe perchlorate and iodide kinetics in
rats and humans. Because of the complex challenge posed in arriving at a representation of the
regulation system for hypothalamic-pituitary-thyroid feedback, the modeling effort was not able
to satisfactorily develop models that linked the observed effects of perchlorate inhibition of
iodine uptake at the NIS with the resultant hormone perturbations and available toxicological
information in the proposed mode-of-action framework.
Because of their potential role in the risk assessment and regulatory applications, the EPA
required that all human clinical data utilized in this modeling effort undergo a quality
assurance/quality check (QA/QC). The QA/QC report is presented in Merrill (2001 a,b). These
QA/QC data represent the most contemporary, comprehensive, and consistent set of human
pharmacokinetic data available for perchlorate.
The PBPK models discussed herein (Merrill, 2001c,d; Clewell, 2001 a,b) were developed
by the AFRL/HEST to provide more accurate descriptions of the kinetics of iodide and
perchlorate with respect to perchlorate's inhibition of iodide uptake at the NIS and their serum
and tissue time courses as well as to aid evaluation of subsequent perturbations in thyroid
hormones and TSH. A general discussion of the model development for the various PBPK
model structures of perchlorate distribution will be provided in this chapter to aid appreciation of
their attributes and applications. Because of the mode of action for perchlorate, an accurate
description of iodide kinetics is critical to the description of perchlorate effects on iodide uptake
at the NIS so that each of these models also includes iodide-specific parameters and accounts for
iodide disposition.
A similar model was developed for each of the various life stages of importance to
interspecies extrapolation of the laboratory animal data: adult, pregnant rat and fetus, and the
lactating rat and neonate. The adult male rat model was developed using data from the ADME
January 16, 2002
6-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
studies in the perchlorate testing strategy, together with experimental data and parameter values
available in the existing literature. The subsequent model structures for the human and various
life stages of the rat were similarly developed based largely on the adult male rat structure
through scaling and optimization of parameters to available data.
It should be noted that the original motivation for including human studies (as discussed in
Chapter 3) in the perchlorate testing strategy was to support such interspecies extrapolation and
not to derive NOAEL estimates for thyroid effects in the human population. As discussed in
Chapter 4, the EPA feels that both the observational epidemiological and the human clinical
studies have significant scientific and technical limitations that preclude their use as the basis for
a quantitative dose-response assessment. In addition, some of the clinical studies contained in
this database fall in the category of studies not to be considered under EPA's Dec. 14, 2001
interim policy on the use of third-party human studies (U.S. Environmental Protection Agency,
2001c). However, the scientific and technical strengths and weaknesses of these studies were
described before this Agency policy was articulated. Therefore, because of the scientific
shortcomings of these studies, they will not be used as "principal studies" in the derivation of
an RfD. The clinical study subject attributes (euthyroid adults) and study design issues (sample
size, RAIU time points, etc.) made these data less reliable than the laboratory animal toxicological
data to ascertain effect levels for the basis of an RfD derivation. Models of perchlorate distribution
for human pregnancy and lactation have not been developed.
More detailed discussion can be found for each model structure in the accompanying
references provided for each in the sections that follow. The adult male rat and human model
(Merrill, 2001c,d) will be discussed in Section 6.2. Section 6.3 discusses the pregnant dam and
fetal rat PBPK model (Clewell, 2001a), and the lactating dam and neonate model (Clewell,
2001b) is discussed in Section 6.4. The purpose of providing these model descriptions and a
discussion of the data used to develop and validate their structures is to provide the external peer
reviewers an opportunity to critically evaluate the model structures, the use of the data in model
development or validation exercises, and the model applications.
The simultaneous ordinary differential equations used in the proposed PBPK models to
simulate radioiodide and perchlorate distribution were written and solved using advanced
continuous simulation language (ACSL) software (AEqis Technologies, Austin, TX).
January 16, 2002
6-2
DRAFT-DO NOT QUOTE OR CITE
-------
1 6.1 MODE-OF-ACTION FRAMEWORK AND UNDERLYING
2 MODELING APPROACH
3 The mode-of-action model proposal by the EPA for the previous perchlorate assessment
4 and discussed in Chapter 3 served as the conceptual construct for the development of the PBPK
5 models. Shown again in Figure 6-1, the model lays out the biomarkers of exposure and effect in
6 a continuum from ingestion of perchlorate in drinking water and uptake into the blood, the key
7 event of iodide uptake inhibition at the NIS in the thyroid gland, and subsequent effects on
8 thyroid hormone economy leading to neurodevelopmental and neoplastic sequelae.
9
• III til
• ¦ • ¦ » • •
I ill lla
bbSI* iwpM
* ! ' # *
* » * *
Susceptibility
Figure 6-1. Mode-of-action model for perchlorate toxicity proposed by the U.S. EPA (U.S.
Environmental Protection Agency, 1998d). Schematic shows the exposure-
dose-response continuum considered in the context of biomarkers (classified as
measures of exposure, effect, and susceptibility) and level of organization at
which toxicity is observed (U.S. Environmental Protection Agency, 1994;
Schulte, 1989). The model maps the toxicity of perchlorate on this basis by
establishing casual linkage or prognostic correlations of precursor lesions.
January 16, 2002
6-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The temporal pathological and serum hormone changes that accompany exposure to
perchlorate corresponding to this continuum are represented in Figure 6-2. The inhibition of
iodide uptake at the NIS results in a transient decrease in serum T4 and T3. This transient phase
of thyroid hormone deficit is of concern during pregnancy and development due to the critical
role that these hormones play in preventing adverse neurodevelopmental sequelae as described in
Chapters 3 and 5. The hypothalamic-pituitary-thyroid feedback system is designed to regulate
the circulating levels of thyroid hormone and will respond to the thyroid hormone decreases by
upregulating TSH production in order to stimulate the thyroid to increase its production of
thyroid hormones to compensate. Represented as the "chronic phase" in Figure 6-2, the
upregulation of TSH would bring the system back into apparent homeostasis. As depicted in the
figure, however, this apparent homeostasis may actually represent subclinical disease in that the
system is only maintaining homeostasis by upregulation and can be considered a stressed system
with respect to its ability to compensate for additional insults caused by other chemicals or
diseases that might impact the thyroid. Further, it should be emphasized that recent
epidemiological investigations have indicated concern about decrements in T4, i.e., thyroxinemia
without concomitant upregulation of TSH that would constitute hypothyroidism (Morreal de
Escobar, 2000; Haddow et al., 1999; Pop et al., 1999).
In order to adequately characterize the transient phase of events, evaluation of the initial
effect of perchlorate at the NIS is necessary. This can be accomplished by determining
perchlorate inhibition with radioactive iodide uptake (RAIU) studies. The timing and route of
administration are important considerations in evaluating these types of studies. Studies of
RAIU that occur during the chronic phase, such as longer-term studies of hormones, offer little
insight to the critical decrements in T4 that may occur during the transient phase due to iodine
inhibition. Likewise, longer-term studies of hormones often represent the upregulated system
and may not be especially informative.
6.1.1 Parallelogram Approach to Interspecies Extrapolation
PBPK models have proven to be very useful tools for performing interspecies extrapolation
of dose for applications in risk analysis. Interspecies extrapolation is often necessary because, as
in this case of perchlorate, critical effects at levels of organization below that of the population
(e.g., thyroid histopathology or brain morphometry) can not be evaluated easily or ethically in
January 16,2002 6-4 DRAFT-DO NOT QUOTE OR CITE
-------
(D
>
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Human
Effective Dose
Human
Health Effect
Rat
Effective Dose
Rat
Adverse Effect
Figure 6-3. Schematic of parallelogram approach used for interspecies extrapolation (U.S.
Environmental Protection Agency, 1994). Dose and adverse effect in rat can be
used to predict human effective dose and response.
effects cannot be accurately measured in humans, the dose associated with an observed critical
effect in the laboratory animal is scaled to the human by adjusting the PBPK model with human
physiological parameters and variables. The human model is typically constructed by
allometrically scaling some parameters in the laboratory animal model based on body weight, and
some parameters such as partition coefficients can be measured in vitro. An administered dose
associated with the critical effect is determined based on an appropriate internal dose metric.
The internal dose is scaled to an equivalent exposure (HEE) in humans by exercising the human
model with human parameters and exposure assumptions. Thus, the HEE represents the human
exposure that would result in the same amount of internal dose metric in a human as that which
caused the effect in the laboratory animal.
The dose-response relationship is considered to be the same as that in the laboratory animal
as the default or more biologically-based models may contain additional parameters that also
account for species-specific determinants of toxicant-target interaction. Figure 6-4 illustrates the
use of the laboratory animal and human PBPK models to arrive at the HEE. Simulations used to
January 16. 2002
6-6
DRAFT-DO NOT QUOTE OR CITE
-------
Laboratory
Animal
Exposure
(mg/kg-day)
Rat
PBPK Model
*
>
Rat
Effective
Dose Metric
(mg/kg-day)
Human
Equivalent
Exposure
(mg/kg-day)
<¦
Human
PBPK Model
Human
Effective
Dose Metric
(mg/kg-day)
Figure 6-4. Illustration of how human equivalent exposure (HEE) is calculated using
PBPK models. An effective internal dose associated with a critical health effect
at an administered dose (mg/kg-day) is calculated by simulating the
experimental exposure regimen (e.g., 5 days/week) for a relevant metric (e.g.,
area under the curve in blood, [AUCB]). The human PBPK model is then used
to simulate an exposure that achieves the same effective internal dose metric
level using human parameters.
1 arrive at HEE for different internal dose metrics and a sensitivity analysis of the adult model
2 structure will be discussed in Section 6.5.
3 The parallelogram approach has also been used to predict effective doses for structurally
4 related chemicals (Jarabek et al., 1994). Disposition of one chemical associated with an effect
5 can be predicted for another after appropriate adjustments for chemical structure and activity are
6 made. In the case of these models, it should be appreciated that the accurate modeling of iodide
7 in addition to that of perchlorate represents such a validation.
8
9
January 16, 2002
6-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6.1.2 Extending the Parallelogram Approach to Various Experimental
Life Stages
Because effects at various life stages (adult, pregnant dam, fetus, lactating dam, and
neonate) were evaluated in the perchlorate laboratory animal studies, the parallelogram approach
had to be extended as shown in Figure 6-5. There are no human models of perchlorate
disposition for pregnant women, lactating women, fetuses, or children, so the relationships to the
adult human HEE had to rely on the relationships determined in the laboratory animal species.
This approach assumes that the relationships, expressed as ratios between one life stage and
another, will be comparable in humans.
Adult Human
Effective Dose
Pregnant Human II
Effective Dose "
Lactating Human
Effective Dose
Adult Rat
Effective Dose
Pregnant Rat
Effective Dose
Lactating Rat
Effective Dose
Figure 6-5. Schematic of extended parallelogram approach used for perchlorate due to
effects at different life stages. Doses in the pregnant rat and fetus are related
back to the adult male rat, likewise, the effects in lactating rats and neonates.
The various PBPK models are used to predict equivalent effective doses at the various
administered doses used in the experiments; e.g., 1.0 mg/kg-day ammonium perchlorate given in
drinking water to both the adult male rat and the pregnant dam. Each PBPK model is exercised
(adult rat and pregnant rat) to predict the amount of internal dose metric achieved at each life
stage. The ratio of the effective internal dose metrics of the life stage in question is then used to
January 16, 2002
6-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
adjust the HEE based on the adult male rat. For example, the HEE for the pregnant dam would
be found by adjusting the HEE for the adult male rat by the ratio of the male rat:pregnant rat as:
Pregnant HEE (mg/kg-day) =
Adult male rat internal effective dose metric (6-1)
Adult male rat HEE (mg/kg-day) x ; ———; ;—
Pregnant rat internal effective dose metric
This ratio is unitless and accounts for the differences between the two life stages in
question in an analogous fashion to the dosimetric adjustment factor (DAF) used in the EPA's
inhalation reference concentration methods to extrapolate respiratory tract doses in different
regions of the laboratory animal to human equivalent concentrations (U.S. Environmental
Protection Agency, 1994). The same ratio approach is used to extend the model predictions to
HEE estimates for the fetus, lactating dam, and neonate. Development of the ratios for two
internal effective dose metrics, perchlorate area-under-curve (AUC) concentrations in serum and
iodide uptake inhibition, will be discussed in Section 6.5.
6.2 ADULT RAT AND HUMAN MODEL STRUCTURES
Because the same model structure is used to describe perchlorate and iodide disposition
(absorption, distribution, and elimination) for both the adult male rat and human, this section will
describe the development of both of these models together. Data supporting development and
validation of the structures will be summarized in this section while additional detail, including
some of the governing equations, can be found in the consultative letters from the AFRL/HEST
(Merrill, 2001c,d).
As discussed in Chapter 2, the perchlorate anion (C104) is very similar in ionic size, shape,
and charge to that of iodide (I ). These shared properties allow perchlorate to interfere with the
first stage of thyroid hormone synthesis by competitively inhibiting the active transfer of iodide
into the thyroid by the sodium (Na+)-iodide (I ) symporter or NIS. The NIS is a protein that
resides in the basolateral membrane of thyroid epithelial cells (Spitweg et al., 2000). NIS
simultaneously transports both sodium and iodide ions from extracellular plasma into the thyroid
epithelial cell via an active process. Energy is provided by the electrochemical gradient across
January 16, 2002
6-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
the cell membrane. The low intracellular concentration of sodium is maintained by sodium-
potassium pumps (Ajjan et al., 1998). The kinetics of perchlorate and iodide anions differ
mainly in that iodide is organified in the thyroid (thyroid hormone production); whereas,
perchlorate is thought to be unreactive and eventually diffuses from the thyroid into systemic
circulation.
The proposed PBPK model structure for the adult male rat (Merrill, 2001c) and human
(Merrill, 2001 d) describes active uptake of iodide and perchlorate in gastric juice, thyroid, and
skin, and competitive inhibition of iodide uptake by perchlorate in NIS-containing tissues, as
well as venous equilibration with slowly and richly perfused tissues as shown in Figure 6-6.
Tissues that exhibited evidence of sodium iodide symporter and were found to concentrate either
anion were depicted as compartments with nonlinear uptake (Merrill, 2001c,d). Tissues with
active uptake include the thyroid, skin, and gastric mucosa (Wolff, 1998; Chow et al., 1969;
Kotani et al., 1998). Although other tissues have been known to sequester iodide and similar
anions (e.g., salivary glands, choroid plexus, ovaries, mammary glands, placenta) (Brown-Grant,
1961, Honour et al., 1952; Spitzweg et al., 1998), the iodide and perchlorate pools of these
tissues was expected to be too small to significantly affect plasma levels. These tissues were
lumped with slowly and richly perfused tissues.
The model also includes separate compartments for plasma, kidney, liver, and fat. These
compartments do not maintain concentrations greater than the plasma at steady state, and
therefore, were not described with terms for active uptake. The rapid urinary clearance of
perchlorate (Yu, 2000) mandated the inclusion of a kidney compartment in the model. A liver
compartment was also utilized due to its significant impact on iodide homeostasis. The majority
of extrathyroidal deiodination takes place within the liver. Fat was primarily added as an
exclusionary compartment. Due to its significant percentage of body weight, the skin represents
an important pool for slow iodide turnover.
The modelers at AFRL/HEST found that a separate skin compartment was necessary.
Experiments performed with radioiodide in rats resulted in skin:serum iodide ratios of close to
one (Yu, 2000). Other researchers have reported higher ratios in rats, but results have not been
consistent. Similar observations during dialysis with pertechnate of slow uptake and retention in
human skin was observed by Hays and Green (1973) and the skin was therefore maintained as a
separate compartment in the model. The skin contains two sub-compartments representing the
January 16, 2002
6-10
DRAFT-DO NOT QUOTE OR CITE
-------
Iodide
Perch lo rate
Ivc i
\
pDoscj
Plasma
|A
!~!
RBCs
Ivc_p
QC
Stomach
.^.Contents
Stomach
ATis_suc__
Stomach
Blood
t
QC,
QG
Liver
QL
Richlv
Perfused
QR
Kidney
vQK
H Colloid
~T fhvroid
L4...F.QjJMfe.
Stroma
QT
lA Skin
Skin
Blood
t
QSK
Fat
OF
Slowly
Perfused
CA i
QS
pDosc_p
'Urine
IA
ItH
bound
Plasma^
free
RBCs
OC
Stomach
lAContcnts
Stomach
A..Ii»JlS„
Stomach
Blood
t
or.
QG
Liver
QL
Richly
Perfused
QR
Kidney
H Colloid ~
*Urine
Thyroid 1
IA Follicle ~
Stroma 1
QT
|A Skin
Skin
Blood
t
QSK
Fat
QF
Slowly
Perfused
CA_p
QS
Figure 6-6. Schematic for the adult male rat and human PBPK models of perchlorate and
iodide distribution (Merrill, 2001c,d). Bold arrows indicate active uptake
(except for plasma binding) at N1S sites in thyroid, gut, and skin. Plasma
binding was also described with Michaelis-Menten terms for the association of
perchlorate anion to binding sites with first-order clearance rates for
dissociation. Small arrows indicate passive diffusion. Boxes represent specific
compartments in the model structure. The thyroid consists of the stroma, the
follicle, and the colloid; and the stomach consists of the capillary bed, stomach
wall, and stomach contents. The skin contains two subcompartments: the
capillary bed and skin tissue. Permeability area cross products and partition
coefficients were used to describe the first-order movement of the perchlorate
(C104~) and iodide (I') anions into deeper subcompartments.
January 16, 2002
6-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
capillary bed and the skin tissue. The thyroid and stomach consist of three sub-compartments:
the stroma, the follicle, and the colloid in the thyroid and the capillary bed, stomach wall, and
contents in the case of the stomach.
Active uptake into the thyroid colloid, stomach contents, and skin were described using
Michaelis-Menten kinetics for nonlinear processes (Figure 6-6, bold arrows). Permeability area
cross products and partition coefficients were used to describe the first order movement of the
anions (C104" and I") between the capillary bed, tissue, and inner (deep) compartments
(Figure 6-6, small arrows) that results from the inherent electrochemical gradient within the
tissues. Passive diffusion through the kidney, liver, and fat compartments were described with
partitions and blood flows. Plasma binding of perchlorate was described with Michaelis-Menten
terms for the association of the perchlorate anions to plasma binding sites and a first order
clearance rate for the dissociation. First-order clearance rates from the kidney were also used to
describe urinary clearance of the anions.
The blood compartment differs between the perchlorate and iodide models. The
perchlorate blood compartment is composed of plasma and plasma proteins to simulate binding.
Plasma binding was required to simulate serum perchlorate concentrations at lower doses.
Iodinated hormones bind to plasma proteins, but free iodide apparently does not. Therefore, a
single compartment for plasma iodide was used. The free anions in plasma are available for
diffusion and active uptake into tissues.
The presence of NIS is an indicator of active uptake for iodide. NIS is highly expressed in
thyroid epithelial cells. Lower levels of expression have been detected in the mammary gland,
salivary gland, skin, stomach, and colon (Ajjan et al., 1998; Spitzweg et al., 1998). However,
only the thyroid has been found to organify iodide (Ajjan et al., 1998). The most important
regulator of symporter gene and protein expression is thyroid-stimulating hormone (TSH). This
is also the case for other important thyroid proteins such as thyroglobulin and thyroid peroxidase
(Spitzweg et al., 1998).
The parameters used in the adult male rat and human model for the various compartments
are provided in Table 6-1. The parameters were based on literature values or fitted to data using
the model as described in the table. It is important to note that the model structure for both
species is the same. The difference, per typical for PBPK models, is that there are species- and
chemical-specific parameters for each. For example, the volume of the thyroid (as percent of
January 16, 2002
6-12
DRAFT-DO NOT QUOTE OR CITE
-------
g
ON
N>
O
o
K)
TABLE 6-1. PHYSIOLOGICAL PARAMETERS FOR THE ADULT MALE RAT AND HUMAN PBPK MODELS
(Merrill, 2001 c,d)
Physiological Parameters
Tissue Volumes
Male Rat
Source
Human
Source
Body Weight BW (kg)
0.3
Measured (rat specific)
-70.0
Subject-specific
Slowly Perfused VSc (%BW)
74.6
Brown ct al., 1997
65.1
Brown et al., 1997
Richly Perfused VRc (%BW)
11.0
Brown ct al., 1997
12.4
Brown et al., 1997
Fat VFc (%BW)
7.4
Brown et al., 1997
d" 21.0
? 2.7
Brown et al., 1997
Kidney VKc (%BW)
1.7
Brown et al., 1997
0.44
Brown et al., 1997
Liver VLc (%BW)
5.5
Brown et al., 1997
2.6
Brown et al., 1997
Stomach Tissue VGc (%BW)
0.54
In house male rat C104" kinetics
(Yu et al., 2000)
1.7
Brown et al., 1997
Gastric Juice VGJc (%BW)
1.68
In house male rat C104' kinetics
(Yu et al., 2000)
0.071
Licht and Deen, 1988
Stomach Blood VGBc (%VG)
4.1
Altman & Dittmer, 1971b
4.1
Altman & Dittmer, 1971a
Skin Tissue VSkc (%BW)
19.0
Brown et al., 1997
3.7
Brown et al., 1997
Skin Blood VSkBc (%VSk)
2.0
Brown et al., 1997
8.0
Brown et al., 1997
Thyroid Vttotc (%BW)
0.0077
Malendowicz, 1977
0.03
Yokoyama ct al., 1986
Thyroid Follicle VTc (%Vttot)
59.9
Malendowicz, 1977
57.3
Brown ct al., 1986
Thyroid Colloid VDTc (%VTtot)
24.4
Malendowicz, 1977
15.0
Brown et al., 1986
Thyroid Blood VTBc (%VTtot)
15.7
Malendowicz, 1977
27.6
Brown et al., 1986
Plasma Vplasc (%BW)
4.1
Brown et al., 1997, Altman & Dittmer,
1971a
4.4
Marieb, 1992; Altman & Dittmer, 1971b
Red Blood Cells VRBCc (%BW)
3.3
Brown et al., 1997, Altman & Dittmer,
1971a
3.5
Marieb, 1992; Altman & Dittmer, 1971b
Adjusted Slowly Perfused VS (L)
0.138
Calculated from model
28.0
Calculated from model
Adjusted Richly Perfused VR (L)
0.01
Calculated from model
5.34
Calculated from model
o\
I
u>
H
I
o
o
2
o
H
o
c
o
H
tT)
o
n
3
M
-------
TABLE 6-1 (cont'd). PHYSIOLOGICAL PARAMETERS FOR THE ADULT MALE RAT AND HUMAN PBPK
MODELS (Merrill, 2001 c,d)
Physiological Parameters
Tissue Volumes
Male Rat
Source
Human
Source
Blood Flows
Cardiac Output QCc (L/hr-kg)
14.0
Brown et al., 1997, Hanwcll & Linzcll,
1973
16.5
Brown et al., 1997; Hanwcll & Linzell,
1973
Slowly Perfused QSc (%QC)
24.0
Brown et al., 1997
5.2
Brown ct al., 1997
Richly Perfused QRc (%QC)
76.0
Brown et al., 1997
17.5
Brown ct al., 1997
Fat QFc (%QC)
6.9
Brown et al., 1997
22.0
Brown et al., 1997
Kidney QK.c (%QC)
14.0
Brown et al., 1997
1.0
Lcggett & Williams, 1995; Malik ct al.,
1976
Liver QLc (%QC)
17.0
Brown et al., 1997
1.6
Brown ct al., 1997
Stomach QGc (%QC)
1.61
Malik et al., 1976
13.0
Calculated, using 24% QC as flow to all
slowly perfused tissues (Brown et al.,
1997)
Skin QSkc (%QC)
5.8
Brown et al., 1997
33.0
Calculated, using 76% QC as flow to all
richly perfused tissues (Brown ct al.,
1997)
Thyroid QTc (%QC)
1.6
Brown et al., 1997
Adjusted Slowly Perfused QS (%QC)
11.3
Calculated from model
Adjusted Richly Perfused QR (%QC)
41.8
Calculated from model
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
body weight), the maximum capacity of thyroid iodide or perchlorate uptake, and plasma binding
of perchlorate. The chemical-specific parameter for each model for both perchlorate and iodide
are provided in Table 6-2.
In order to simulate the daily dosing regimen of the drinking water experiment, the rats
were assumed to drink at constant rate for 12 of the 24 hours per day (1800 to 0600 hours).
A pulse function in ACSL was used to introduce drinking water to the gastrointestinal (GI)
compartment of the rat for the first 12 hours of each 24-hour period and to stop dosing while the
rat was presumably sleeping. Intravenous (iv) dosing was introduced into the venous blood
compartment of the model. Intraperitoneal (ip) injection was introduced into the model in the
same manner as the iv dosing.
6.2.1 Data and Methods
This section summarizes the AFRL/HEST data and data available in the literature that were
used for model development. Details on experimental methods, including protocol design,
exposure regimen, chemical source and purity, animals (housing, feeding, surgical procedures,
etc), and the analytical methods for measurement of RAIU; of perchlorate in plasma, urine and
tissues; and of thyroid hormones and TSH can be found in the associated consultative letters
from AFRL/HEST (Merrill, 2001c,d; Yu, 2000, 2001, 2002; Yu et al., 2000).
6.2.1.1 Studies in Laboratory Rats
The studies performed at AFRL/HEST included both "acute" iv experiments to measure
radiolabled iodide or perchlorate as well as measurements of the same after drinking water
administration. These two different regimens provided a better characterization of the transient
("acute") and chronic behavior necessary for an accurate description of the disposition of the
anions. Adult male Sprague-Dawley rats (330 ± 35 g; n = 6 rats per group) that were purchased
from Charles River Laboratory (Raleigh, NC) were used in the experiments.
In these experiments, the term total iodine includes bound iodine plus fee inorganic iodide.
Carrier doses included tracer doses of carrier free radiolabled iodide (125I ) along with non-
radiolabeled iodide. Free 125I" radioactivity was determined by subtracting the bound from total
measurements (Merrill, 2001c; Yu, 2000, 2001, 2002; Yu et al., 2000).
January 16, 2002
6-15
DRAFT-DO NOT QUOTE OR CITE
-------
£>
D
g
•5
OS
K)
O
O
NJ
TABLE 6-2. CHEMICAL-SPECIFIC PARAMETERS FOR THE ADULT MALE RAT AND HUMAN PBPK MODELS
(Merrill, 2001c,dV
Male Rat
Human
(unitlcss)
Pcrchloratc
Iodide
Source
Pcrchloratc
Iodide
Source
Slowly Perfused/Plasma PS_
0.31
0.21
Yu et al., 2000; Halmi ct al., 1956
0.31
0.21
Halmi et al., 1956;Yu ct al.,
2000
Richly Perfused/Plasma PR_
0.56
0.40
Yu et al., 2000; Halmi et al., 1956
0.56
0.40
Halmi el al., 1956;Yu et al.,
2000
Fat/ Plasma PF_
0.05
0.05
Pena et al., 1976
0.05
0.05
Pena et al., 1976
Kidney/Plasma PK._
0.99
1.09
Pcrlman et al., 1941
0.99
1.09
Pcrlman ct al., 1941; Yu ct al.,
2000
Liver/Plasma PL_
0.56
0.44
Pcrlman et al., 1941
0.56
0.44
Pcrlman ct al., 1941; Yu ct al.,
2000
Gastric Tissue/Gastric Blood
PG
1.80
1.40
Yu et al., 2000; Yu, 2000
1.80
0.50
Yu ct al., 2000;Yu, 2000
Gastric Juice/Gastric Tissue
PGJ_
2.30
3.00
Yu et al., 2000; Yu, 2000
2.30
3.50
Yu ct al., 2000; Yu, 2000
Skin Tissue/Skin Blood PSk_
1.15
0.70
Yu, 2000, Perlman et al., 1941
1.15
0.70
Pcrlman ct al., 1941; Yu, 2000
Thyroid Tissue/Thyroid Blood
PT
0.13
0.15
Chow & Woodbury (1970)
0.13
0.15
Chow & Woodbury (1970)
Thyroid Lumen/Thyroid Tissue
PDT_
7.00
7.00
Chow & Woodbury (1970)
7.00
7.00
Chow & Woodbury (1970)
Red Blood Cells/Plasma
0.80
1.00
Yu ct al., 2000; Rail et al., 1950
0.80
1.00
Rail et al., 1950; Yu ct al., 2000
Max Capacity, Vmaxc (ng/hr-
kg)
Thyroid Colloid Vmaxc_DT
1.0E+04
4.0E+07
Fitted
2.5E+5
1.0E+8
Fitted
Thyroid Follicle Vmaxc_T
2.2E+03
5.5E+04
Fitted
5.0E+4
-1.5E+5
Fitted
Skin Vmaxc_S
6.2E+05
5.0E+05
Fitted
1.0E+6
7.0E+5
Fitted
Gut Vmaxc_G
3.0E+05
1.0E+06
Fitted
1.0E+5
9.0E+5
Fitted
Plasma Binding Vmaxc_Bp
9.5E+03
—
Fitted
5.0E+2
—
Fitted
CT\
I
C\
H
I
a
o
z
o
H
o
d
o
H
m
o
n
H
tn
-------
g
g
Os
N>
O
O
K>
TABLE 6-2 (cont'd). CHEMICAL-SPECIFIC PARAMETERS FOR THE ADULT MALE RAT AND HUMAN PBPK
MODELS (Merrill, 2001c,d
lQ
Male Rat
Human
(unitlcss)
Perchlorate
Iodide
Source
Perchlorate
Iodide
Source
Affinity Constants, Km (ng/L)
Thyroid Lumen Km_DT
1.0E+08
1.0E+09
Golstein et al., 1992
1.0E+8
1.0E+9
Golstein et al., 1992
Thyroid Kjn_T
2.5E+05
4.0E+06
Gluzman & Nicpomniszcze,
1983; Wolff, 1998
1.8E+5
4.0E+6
Gluzman & Nicpomniszcze,
1983; Wolff, 1998
Skin Km_S
2.0E+05
4.0E+06
Gluzman & Niepomniszcze,
1983; Wolff, 1998
2.0E+5
4.0E+6
Gluzman & Nicpomniszcze,
1983; Wolff 1998
Gut Km_G
2.0E+05
4.0E+06
Gluzman & Niepomniszcze,
1983; Wolff, 1998
2.0E+5
4.0E+6
Gluzman & Niepomniszcze,
1983; Wolff, 1998
Plasma binding Km_B
1.1E+04
—
Fitted
1.8E+4
—
Fitted
Permeability Area Cross Products (L/hr-kg)
Gastric Blood to Gastric Tissue
PAGc_
0.80
0.10
Fitted
0.6
0.2
Fitted
Gastric Tissue to Gastric Juice
PAGJc_
0.80
0.10
Fitted
0.8
2.0
Fitted
Skin Blood to Skin Tissue
PASkc_
1.0
0.10
Fitted
1.0
0.06
Fitted
Plasma to Red Blood Cells
PARBCc_
0.10
1.00
Fitted
1.0
1.0
Fitted
Follicle to thyroid blood
PATc_
4.0E-05
1.0E-04
Fitted
1.0E-4
1.0E-4
Fitted
Lumen to Thyroid Follicle
PADTc_
0.01
1.0E-04
Fitted
0.01
1.0E-4
Fitted
Clearance Values (L/hr-kg)
Urinary excretion CLUc_
0.07
0.05
Fitted
0.126
0.1
Fitted
Plasma unbinding Clunbc
0.1
—
Fitted
0.025
—
Fitted
C\
I
-J
H
6
o
o
H
O
c
o
H
ffl
o
n
"All parameters listed are notated in the model by either an i (for iodide) or p (for perchlorate) following an underscore in the parameter name (e.g., PR_/,
PR_p, Vmaxc_T/', Vmaxc_T/?, etc.).
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.2.1.1.1 Acute iv Experiments in Rats
Radiolabeled iodide (125I") kinetics. Male rats were administered a single iv tail-vein
injection with physiological saline (control group) or 33 mg/kg l25I" (with carrier) in physiological
saline. Rats were euthanized by C02 asphyxiation at 5, 15, and 30 minutes (min), 1, 2, 6, 9, 24,
32, 48, and 96 hours (hr) post dosing to collect thyroid and blood from the vena cava. Rats for
the 24 hour time point were placed individually in metabolism cages to collect urine.
In an additional study, male rats were intravenously dosed with 33 mg/kg l25I" (with carrier)
and euthanized at 0.5, 2 and 6 hours post dosing. Total, bound, and free 125I' were analyzed in
thyroid and serum, and total 125I" was measured in skin and gastric contents (Yu, 2001).
Radiolabeled 36C104" kinetics. Naive adult male rats (300 ± 20 g) were dosed once by iv
tail-vein injection with 3.3 mg/kg radiolabeled perchlorate. Due to the low specific activity, a
smaller dosing level could not be achieved. Each rat received less than 6 /xCi. Rats were
euthanized by C02 asphyxiation at 0.5, 6, 12, 24, 32, and 48 hours after dosing. The thyroid,
intestinal tract, intestinal tract contents, muscle, skin, liver, kidney, spleen, bladder, plasma, and
red blood cells were harvested from the rats and stored at -20°C until analysis of 36C104\ Rats
for 12, 24, 32, and 48 hours time points were placed individually in metabolism cages for urine
collection. Metabolism cages were washed with 500 mL de-ionized water. Urine and cage wash
samples were stored under the same conditions until analysis.
125I Kinetics and Inhibition from Acute iv Dosing with C104". Rats were injected with
one of five doses of perchlorate (0.0, 0.01, 0.1, 1.0, and 3.0 mg/kg). At 2 hours post dosing, they
were challenged with 125I' with carrier (33 mg/kg) by intravenous injection and euthanized at 5,
15, and 30 min, 1, 2, 6, 9, and 24 hours post dosing of iodide. This corresponds to 2.08, 2.25,
2.5, 3, 4, 8, 11, and 26 hours, respectively, after dosing with perchlorate. Blood and thyroid were
harvested from all time point groups; urine was collected from rats in the 24 hours dose group.
Perchlorate and iodide levels were determined in the thyroid, serum and urine.
In an additional study, three rats were intravenously dosed with 0.0, 0.1, and 1.0 mg/kg
perchlorate and challenged two hours later with 33 mg/kg 125I". Rats were euthanized at 15 min,
1, 2, and 4 hours after they were dosed with iodide. Levels of perchlorate and 125I" were
determined in thyroid, serum, skin and gastric contents (Yu, 2001).
January 16, 2002
6-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.2.1.1.2 Drinking Water Studies in Rats
Three drinking water studies (1,5, and 14 days) were performed with target perchlorate
concentrations of 0.0, 1.0, 3.0, 10.0, and 30.0 mg/kg-day with adult male rats continually
exposed via drinking water. At the end of day 1, 5, or 14, rats (n=6 per group) were challenged
once with 33 mg/kg l25I" with carrier and euthanized at 2 hours post iodide dosing. Blood and
thyroid gland were collected for C104" and 125I" analyses in serum. For the 10 and 30 mg/kg dose
groups, perchlorate was measured in serum and thyroid on day 5; however, the iodide inhibition
study for these dose groups was conducted on Day 14.
6.2.1.2 Human Studies
The data used in development of the Merrill (200Id) human model were obtained from
Hays and Solomon (1965) or recent data, both published and unpublished, that underwent the
QA/QC check described in the introduction of this chapter (Merrill, 2001 a,b). These data
included the published and unpublished data from a human study of drinking water exposure to
perchlorate that measured RAIU in the thyroid (Greer et al., 2000).
Data supporting model validation were obtained from another unpublished drinking water
study conducted under contract to AFRL/HEST by Drs. Holger Leitolf and Georg Brabant of the
Medizinische Hoschschule, Hanover, Germany. Urinary perchlorate clearance data by Eichler
(1929), Kamm and Drescher (1973), and Durand (1938) were also used to validate model
predictions.
6.2.1.2.1 Human Iodide Kinetic Data (Hays and Solomon, 1965)
A comprehensive human kinetic study on early iodide distribution was reported in 1965 by
Hays and Solomon. The authors studied the effect of gastrointestinal cycling on iodide kinetics
in nine healthy males after an iv dose of 10 //Ci radiolabeled iodide (131I), approximately
3.44 x 10'3 ng 131I"/kg body weight. Frequent measurements of radioiodide uptake in the thyroid,
gastric secretions, plasma, and cumulative urine samples were taken during the three hours
following injection. Gastric secretions were collected using a nasogastric tube with constant
suction while the subjects remained in a resting position (only standing to urinate). Saliva was
not collected separately and therefore pooled, to some extent, with gastric juices. To account for
the removal of gastric iodide from circulation and to determine its impact on free iodide
January 16, 2002
6-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
distribution, the authors ran a control session on the same subjects without aspirating gastric
secretions. Aspirated gastric secretions accounted for 23% of the 131I" administered.
6.2.1.2.2 Perchlorate Kinetics and Inhibition of Thyroid Iodide Uptake (Greer et al., 2000)
Perchlorate data. As described in Chapter 4, Greer et al. (2000) recently studied the
effects of repeated low level exposure to perchlorate on humans. Subjects received 0.5, 0.1,
0.02, or 0.007 mg/kg-day perchlorate in drinking water over a two week period. Each dose group
consisted of eight healthy volunteers (four males and four females) with no signs or symptoms of
thyroid disorders (euthyroid). The daily dose was dissolved in 400 mL water and divided into
four 100 mL servings that were ingested at approximately 0800, 1200, 1600, and 2000 hours.
Baseline serum and urine samples were collected before the first perchlorate treatment.
During perchlorate exposure, serum samples were collected at the following approximate times:
day 1 at 1200 and 1600, day 2 at 0800, 1200, and 1700, day 3 at 0900, day 4 at 0800 and 1200,
day 8 between 0800 and 0900 and day 14 at 0800 and 1700. Serum samples were also collected
on post-exposure days 1, 2, 3, and 14. Twenty-four hour urine collections were taken on
exposure days 1,2, 14 and post-exposure days 1 through 3. Serum and 24-hour urine samples
from the study were provided to AFRL/HEST compliments of Dr. Monte Greer of Oregon
Health Science University (OHSU), Portland, OR, and Dr. Gay Goodman of Intertox, Seattle,
WA. The samples were analyzed for perchlorate at the Operational Toxicology Branch, Human
Effectiveness Directorate at the Air Force Research Laboratory (AFRL/HEST), Wright Patterson
Air Force Base (WPAFB), OH, using the analytical methods described in Merrill (200Id).
Iodide Inhibition Data. Eight and 24 hour thyroid 123I" uptakes (radioiodine uptake or
RAIU) were measured one to two days prior to perchlorate treatment (baseline) on days 2 and
14 of perchlorate exposure and 14 days after perchlorate exposure was discontinued. A gelatin
capsule containing 100 mCi of 123I" was administered orally at 0800, before the first perchlorate
solution for that day was drunk. Thyroid scans were then taken 8 and 24 hours later.
Thyroid and Pituitary Hormone Data. The serum samples were also analyzed for TSH,
T4, T3, and free T4 at OHSU. However, these hormone data were not used in the PBPK model
described below. Statistical analysis of the data is described in Attachment 2 of Merrill (200 Id).
In summary, there was little effect of perchlorate on levels of T4, free T4, or T3. TSH
decreased significantly from baseline by Exposure Day 3. On Post-Exposure Day 1, the TSH
January 16, 2002
6-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
levels of the subjects in the 0.5 mg/kg-day group had decreased by an average of 35% from
baseline (ranging from 17% to 52%). Therefore, it appears that TSH was dropping while
inhibition remained the same. It is possible that there is an increase in thyroid sensitivity to TSH
as an early response to inhibition (Brabant et al., 1992). This increased sensitivity (possibly an
increase affinity of the TSH receptor) could possibly decrease circulating TSH levels while T4
has not decreased sufficiently yet to stimulate the hypothalamus to increased TRH secretions.
After perchlorate was discontinued, between Post-Exposure Days 1 and 15, the mean TSH level
increased significantly over baseline (23% greater than baseline), with TSH of one subject
remaining below baseline. The drop in TSH during perchlorate exposure and the rise above
baseline measurements after perchlorate seem counter-intuitive to the TSH regulation expected
but may be part of a rebound phenomenon as the NIS begins to upregulate.
In addition, the data by Greer et al. (2000) showed an increase in radioiodide uptake in
excess of baseline measurements 14 days after perchlorate exposure. An increase in radioiodide
uptake is expected due to the rise in TSH mentioned above. This rebound effect has been noted
in other human inhibition studies (using both iodide and perchlorate as inhibitors). Saxena et al.
(1962) evaluated the prophylactic doses of iodide required to suppress thyroid uptake of 131I" in
euthyroid mentally defective children. They found a minimal effective oral dose of 1500 to
2000 ng iodide per square meter of body surface per day was required to completely suppress
mr uptake. Within a week after iodide administration was stopped, a rebound of uptake was
noted. In some instances these uptakes were even higher in subsequent weeks.
6.2.1.2.3 Supporting Kinetic Studies
Both urine and serum perchlorate concentrations for a validation exercise were provided
from a recent unpublished study by Drs. Brabant and Leitolf of Medizinische Hochschule,
Hanover, Germany. In their study, seven healthy males ingested 12.0 mg/kg perchlorate
dissolved in 1 liter of drinking water every day for two weeks. The daily perchlorate dose was
divided equally in three portions and ingested three times per day (approximately between 0600
and 0800, 1100 and 1300 and 1800, and 2000 hours). Blood specimens were collected on days 1,
7, and 14 of perchlorate treatment and on the two mornings after perchlorate administration was
discontinued. Samples were analyzed for perchlorate at AFRL/HEST.
January 16, 2002
6-21
DRAJT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Three published studies reported cumulative urine concentrations collected from healthy
males after receiving a high oral dose of perchlorate (Durand, 1938; Kamm and Drescher, 1973;
Eichler, 1929). Oral doses administered in these studies were 784 mg NaC104 (635 mg C104";
Durand, 1938); 1000 mg NaC104 (765 mg C104'; Kamm and Drescher, 1973), and 2000 mg
KC104 (1400 mg C104'; Eichler, 1929). The studies did not report serum perchlorate levels but
could be used to validate the model.
Stanbury and Wyngaarden (1952) measured radioiodide uptake in a patient with Grave's
disease. The patient received a tracer dose of13'I" as a control before perchlorate dosing and
again one hour after administration of 100 mg KC104. Thyroid scans of radioiodide uptake were
performed both after the control and perchlorate sessions to determine the level of inhibition.
6.2.2 Adult Male Rat Model Development
This section summarizes some key features necessary to the development of the adult male
rat model structure and shows results of predictions made with simulations against experimental
data used to parameterize and validate the model.
6.2.2.1 Physiologic Parameters and Tissue Partition Coefficients
The adult male rat volumes and blood flows were obtained from the literature or the
AFRL/HEST studies as described in Table 6-1. Allometric scaling was used to account for
parameter differences due differences in body weights between rats and humans. Because no
steady-state values from infusion studies were available, the partition coefficients for iodide and
perchlorate were estimated from the various studies listed in Table 6-2. The liveriserum and
muscle:serum ratios of 0.56 and 0.31 were obtained in the AFRL/HEST radiolabled perchlorate
(36C104 ) iv study described above. The liver:serum partition value was used to represent
partitioning to the liver and richly perfused compartments and the muscle:serum value to
represent the slowly perfused compartment.
For compartments with nonlinear uptake of the anions, effective partition coefficients were
used that represented either approximate tissue:serum concentration ratios or electrical potential
gradients. Chow and Woodbury (1970) measured electrochemical potentials within the thyroid
stroma, follicular membrane, and colloid at three different doses of perchlorate. The measured
difference in electrical potential between the thyroid stroma and follicle was interpreted by
January 16,2002 6-22 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Merrill (2001c) as an effective partition coefficient for the perchlorate and iodide anions,
hindering the entry of negatively charged ions into the follicle. The equal and opposite potential
from the follicle to the colloid enhances passage of negatively charged species into the colloid
and indicates an effective partition coefficient of greater than one. The equivalence between
electrical potential differences f and effective partition coefficients for the thyroid
subcompartments (stroma:follicle and follicle:colloid) were estimated in the manner of Kotyk
and Janacek (1977) based on the Chow and Woodbury (1970) data as described in Merrill
(2001c).
6.2.2.2 Chemical-Specific Parameters
The various active transport processes, tissue permeabilities, and clearance rates (excretion)
are described in PBPK models for each species on a chemical-specific basis. This section
outlines how the values for perchlorate and iodide used in the adult male rat model were derived.
The values can be found in Table 6-2 and details on derivation in Merrill (2001c).
6.2.2.2.1 Affinity Constants and Maximum Velocities for Active Transport Processes
Kinetic values for the saturable (Michaelis-Menten) active uptake process of perchlorate,
the affinity constant and maximum velocity capacity (Km_p and Vmaxc_p), were not available
in the literature nor were they determined experimentally at AFRL/HEST. Only the affinity of
iodide for NIS was available in the literature. The Merrill (2001c) adult rat model uses a
Michaelis-Menten affinity constant (Km) value of 4.0 x 106 ng/L to describe the affinity of iodide
(Km_/) across compartments involving active transport by NIS (e.g., in the thyroid and gastric
juices). This was based on the mean value of 3.96 x 106 ng/L for iodide derived by Gluzman and
Niepomniszcze (1983) from thyroid slices of 5 normal individuals. The thyroid slices were
incubated with several medium iodide concentrations. The experimentally determined Km
values for iodide are similar across species (Gluzman and Niepomniszcze, 1983) and across
different tissues (Wolff, 1998). This average literature value was therefore used for iodide in
tissues described with active uptake.
The values for perchlorate affinity were originally assumed to be the same as those for the
Km of iodide, due to the similar mechanism in which the two anions are transported into the
tissues. Thus, the iodide values were adjusted for the difference in mass to give an estimated
January 16, 2002
6-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
value for the affinity of perchlorate. The molar equivalent of iodide's Km for perchlorate is
3.1 x 106 ng/L. However, these values were not adequate for use in the models. Several
literature sources suggest that perchlorate may have a significantly higher affinity for NIS than
iodide. In his 1963 paper (Wolff and Maurey, 1963) and his 1998 review, Wolff concluded that
perchlorate has a greater affinity than iodide for the NIS. This assumption was based upon his
work with iodide, perchlorate, and several other anions actively sequestered in the thyroid.
Wolff measured the Km of a few of the anions and inhibition constants (Ki's) for several ions,
including perchlorate. As noted in Chapter 2, Wolff found that the relative potency of inhibition
by the various anions could be described with the following series: TcCV^lO^ReO^SCN"
>BF4"> l>N03>Br">Cr. Wolff reported that the measured Km values for several of these
inhibiting anions were not the same as those measured for iodide. In fact, measured values for
Km and Ki for several of the inhibiting anions revealed that affinity increased with increased
inhibitory potency.
Several studies suggest perchlorate is a more potent inhibitor than iodide. In the rat
thyroid, Wyngaarden et al. (1952) have shown that perchlorate was a more powerful inhibitor of
the iodide trap than thiocyanate. Halmi and Stuelke (1959) showed that perchlorate was ten
times as effective as iodide in depressing tissue to blood ratios in the rat thyroid and gut.
Similarly, Harden et al. (1968) compared human saliva to plasma radioiodide concentration
ratios after equimolar doses of perchlorate and iodide. The saliva:plasma iodide ratios during
resting conditions were approximately seven times lower after a molar equivalent dose of
perchlorate versus iodide. Lazarus et al. (1974) also demonstrated that perchlorate was taken up
to greater extent in mice salivary glands than iodide. These studies, in addition to the work of
Chow et al. (1969), support the use of a lower Km for perchlorate uptake in the tissues with
sodium iodide symporter. Based on this information, a value of 2.5 x 105 ng/L for the thyroid
(Km_Tp) and 2.0 x 105 ng/L for skin (Km_Sp) or gut (Km_G/?), approximately 10 times lower
than that of iodide, was estimated by Merrill (2001 c,d) to represent perchlorate's affinity for
transport by the NIS.
The apical follicular membrane (between the thyroid follicle and colloid) also exhibits a
selective iodide uptake mechanism. Golstein et al. (1992) measured a Km value of
approximately 4.0 x 109 ng/L for the transport of iodide between the thyroid follicle and colloid
(Km_DT/?) in bovine thyroid. This iodide channel also appears to be very sensitive to
January 16,2002
6-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
perchlorate inhibition and shares a similar permeability to perchlorate as to iodide. The ability of
perchlorate to inhibit iodide uptake at the apical follicular membrane suggests that the Km of
perchlorate at the apical follicular membrane (Km_Dtp) is also lower than that of iodide. Model
simulations of thyroid inhibition supported a value of 1.0 x 108 ng/L, approximately ten times
less than that of iodide.
Whereas the Km is similar across tissues containing NIS, the maximum velocity term
(Vmaxc) does vary between tissues and species (Wolff, 1998), being lower in humans than other
species (Gluzman and Niepomniszcze, 1983; Wolff and Maurey, 1961). Maximum velocities or
capacities (Vmaxc) were not found in the literature and were estimated for a given compartment
by fitting the simulation to the data at varying doses.
6.2.2.2.1 Effective Partitions, Permeability Area Cross Products and Clearance Values
Permeability area cross products and partition coefficients were used to describe diffusion
limited uptake in tissues requiring subcompartments. The permeability area values in the Merrill
(2001c) model were fitted by setting the partition coefficients to the literature values in
Table 6-2. Fitted clearance values were used to describe first-order urinary excretion rates and
reversible plasma binding to serum. Equations for these representations are provided in Merrill
(2001c).
6.2.2.3 Adult Male Rat Model Simulation Results and Validation
The simulations shown in this section result from exercising the model with the
physiological and chemical-specific parameters provided in Tables 6-1 and 6-2. Figure 6-7
illustrates the model predictions versus data time course for the iv radiolabeled perchlorate study
described in Section 6.2.1.1.1. The model produced good simulations for the trend of the data
but slightly over predicts the thyroid concentrations at later time points (Panel A). Model
predictions fit the data well for perchlorate concentrations in the serum (Panel B) and kidney
(Panel C), as well as the amount excreted in the urine (Panel D). Other tissue concentrations not
shown herein also were predicted well by the model (Merrill, 2001c).
Figure 6-8 shows that plasma binding of perchlorate was necessary to provide adequate
model predictions. Thyroid, serum, and urine were collected from the iv studies described in
Section 6.2.1.1.1 using cold (i.e., not radiolabeled) perchlorate at 0.01, 0.1, 1.0, and 3.0 mg/kg.
January 16, 2002
6-25
DRAFT-DO NOT QUOTE OR CITE
-------
•g
*
O
o
2 A 0
1 00-
0 10-
294
Time (hours)
49 0
o
o
$
O
O
49 0
Time (hours)
19 6 29 4
Time (hours)
39 2
49 0
Figure 6-7. Adult male rat PBPK model predictions after an acute iv dosing with
radiolabeled perchlorate (36C104~). Panels A and B show model predictions
(lines) versus data time course (mean ± SD) of labeled perchlorate (mg/L) in
the thyroid and serum. Panel C shows model predictions versus data time
course of labeled perchlorate (mg/L) in the kidney. Panel D shows cumulative
excretion (nig) of labeled perchlorate in the urine (Merrill, 2001c).
1 Model predictions without plasma binding (Panel A, left) resulted in an underestimation of
2 serum perchlorate concentrations at the 1 mg/kg-day dosage level and below. Low serum
3 predictions suggested either greater uptake into other tissues or protein binding. To provide
4 better estimates of perchlorate serum concentrations at the 0.01 and 0.1 mg/kg doses, Merrill
5 (2001c) added protein binding to the venous blood compartment of the model. An affinity
January 16, 2002
6-26
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Time (hours) Time (hours)
Figure 6-8. Simulations illustrating the necessity of including plasma binding in the adult
male rat PBPK model structure (Merrill, 2001c). Model predictions (lines)
versus data time course (mean ± SD) of perchlorate concentration (mg/L) in
serum after doses of 3.0,1.0, 0.1 and 0.01 mg/kg-day are shown in Panel A
without and in Panel B with plasma binding. Only part of the simulation for
the 0.01 dose in Panel A can be seen in the lower left corner. Data of Yu
(2000).
constant for this binding of perchlorate in the blood (Km_B/?) of 1.1E6 ng/L and a maximum
velocity capacity for this blood binding (Vmaxc_B/?) of 9.3E3 ng/h/kg was fitted to serum levels
from doses ranging 0.01 to 3.0 mg/kg (Panel B, right). The model underpredicts serum
perchlorate from the 0.1 mg/kg dose group; but it fits serum at 0.01 mg/kg and cumulative urine
across the doses. Interestingly, the urinary excretion at 0.01 mg/kg was lower than the other
doses, accounting for elevated serum concentrations. Mean 24 hour urinary excretions (± SD) of
perchlorate were approximately 97% (± 2), 72% (± 1), 87% (± 17), and 91% (± 13) of the
administered iv dose for the 0.01, 0.1, 1.0, and 3.0 mg/kg dose groups, respectively.
The literature discussed in Chapter 3 and in Merrill (2001c) suggests that serum albumin is
the major binding protein; however, it does not confirm that albumin is the only binding site.
Merrill (2001c) notes that no studies were found that evaluated whether perchlorate or similar
anions bind to thyroglobulin. However, Yamada (1967) studied the effects of perchlorate and
January 16, 2002
6-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
other anions on T4 metabolism and noted a significant decrease in serum protein-bound iodide
(PBI) in thyroidectomized T4-maintained perchlorate-fed rats. In a 1968 in vitro study, Yamada
and Jones reported that T4 was displaced from plasma protein as indicated by an uptake of T4 by
muscle in the presence of plasma taken from perchlorate-fed rats. This suggested, but did not
demonstrate directly, that perchlorate interferes with T4 binding with plasma proteins.
Pertechnetate is known to bind to plasma proteins. Hays and Green (1973) studied the
blocking of pertechnetate binding with human serum proteins by other anions. Perchlorate was
found to be one of the most effective, while iodide was ineffective. In dialysis studies, inorganic
iodide did not bind to plasma proteins. The pertechnetate binding appeared to be reversible in
serum.
Simulations of thyroid perchlorate concentrations and of the amount of perchlorate excreted
in the urine from the four dose groups are shown in Figure 6-9. It was noted that the thyroid
concentrations resulting from the 3.0 mg/kg cold perchlorate study were slightly higher than
those from the radiolabeled perchlorate (36C104 ) study at 3.3 mg/kg (Figures 6-9A and 6-7A,
respectively). This may reflect the analytical differences in measuring cold versus radiolabeled
perchlorate. The model slightly underpredicts the thyroid concentrations at 3.0 mg/kg, based on
the cold perchlorate data (Figure 6-9A), and slightly overpredicts the 36C104" thyroid
concentration at 3.3 mg/kg (Figure 6-7A).
The model is able to adequately predict data from studies that were not used in the
development process. Figure 6-10 shows the model predictions versus the data of Chow and
Woodbury (1970) and Eichler (1929). Model predictions fit the data well for radiolabeled
perchlorate concentration in the thyroid (A); whereas, the serum (B) is underpredicted. Merrill
(2001c) notes the difference and provides some plausible explanations. The rats in the Chow and
Woodbury (1970) study were functionally nephrectomized by ligating the renal pedicle of both
kidneys and given the radiolabeled perchlorate ip. Analytical differences between AFRL/HEST
and Chow and Woodbury could exist, and it is also possible that the nephrectomization creates
physiological changes that can not be accounted for sufficiently by "turning off" urinary
excretion in the model simulations. One hypothesis is that saturation in NIS-containing tissues
occurs to a lesser extent as a result of increased extracellular sodium cation (Na+) and possibly
other competitive anions when renal clearance is blocked, thereby increasing the arterial
radiolabeled perchlorate. While the underprediction in serum would suggest the need for an
January 16, 2002
6-28
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
100 3
o>
£
¦g
o
>.
.c
I-
c
0)
13
o
n
o
0)
Q.
1 0
o>
E.
0
C
- 0 1
0
Q.
0)
>
ro
D
E
D
o
0 01
Time (hours)
Time (hours)
Figure 6-9. Adult male rat PBPK model predictions (lines) versus data time course (mean
± SD) of perchlorate concentrations in the thyroid (mg/L) in Panel A or
cumulative excreted perchlorate in the urine (mg) in Panel B (Merrill, 2001c).
Male rats were dosed iv with 3.0,1.0, 0.1 or 0.01 mg/kg-day perchlorate
(Yu, 2000).
increased binding constant for perchlorate, this was not consistent with the data from
AFRL/HEST for studies at lower doses (Merrill, 2001c). Panel C in Figure 6-10 shows the
model predictions versus the data of Eichler (1929) for cumulative perchlorate excreted in the
urine. These rats were given perchlorate subcutaneously (sc) at doses of 1.6, 8.0, and 49 mg/kg.
The adult male rat model (Merrill, 2001c) is also able to predict iodide distribution.
Figure 6-11 shows the model predictions versus a time course for radiolabeled iodide data from
the AFRL/HEST experiments outlined in Section 6.2.1.1.1. Adequate fit is demonstrated for
both the thyroid and serum concentrations at doses of radiolabeled iodide differing by an order of
magnitude (0.033 and 0.33 mg/kg).
Figure 6-12 demonstrates the fit of the model simulations of perchlorate thyroid
concentration (mg/L) after drinking water exposures to perchlorate. The model was coded to
simulate oral dosing for 12 hours per day, assuming that rats drink fairly continuously during
their waking hours. The same perchlorate parameters used to describe the "acute" (iv) kinetics
also adequately described serum concentrations from these "chronic" drinking water exposures
January 16, 2002
6-29
DRAFT-DO NOT QUOTE OR CITE
-------
o
o
s
Time (hours)
Time (hours)
26
Time (hours)
Figure 6-10. Validation for male rat PBPK model of perchlorate disposition (Merrill,
2001c). Model predictions (lines) versus data time course for concentrations
(mg/L) in the thyroid (A) and serum (B) for ip administration in rats of 200,
10, and 0.5 mg/kg 36C104" (data of Chow and Woodbury, 1970). Panel C
shows model predictions (lines) and data time course for cumulative
perchlorate in the urine (mg) of male rats after subcutaneous doses of 1.6, 8.0,
and 49 mg/kg (data of Eichler, 1929).
January 16, 2002
6-30
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 6-11. Male rat PBPK model (Merrill, 2001c) predictions (lines) versus data time
course (mean ± SD) of iodide concentrations (mg/L) at two doses of 1Z5I~
with carrier, 0.033 mg/kg or 0.33 mg/kg, in the thyroid (A) or (B) and in the
serum (C) or (D). Data of Yu (2001).
1 (data shown in Merrill, 2001c) but failed to predict thyroid concentrations from the 3.0 mg/kg-
2 day dose and higher. TSH in these same studies was increased during drinking water exposure
3 across all doses so that Merrill (2001c) accounted for the TSH-induced upregulation in the NIS
4 by fitting an increased effective thyroid follicle:stroma partition coefficient (PT_p) at these
5 higher doses. Merrill (2001c) noted that TSH is not expected to increase NIS in tissues other than
6 the thyroid (Brown-Grant, 1961) and that these simulations agree. Given the small size of the
7 thyroid, its upregulation would not decrease serum concentrations significantly. This explains
January 16, 2002
6-31
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
100
CD
E
10 0 mg/kg, 1.25
3.0 mg/kg, 0.4
1 0 mg/kg. 0.13
67.2
134 4 201.6
Time (hours)
268 8
336
Figure 6-12. Male rat PBPK model predictions (lines) versus data time course (mean ± SD)
of thyroid perchlorate concentrations (mg/L) in male rats during ingestion of
30,10, 3.0,1.0, 0.1, or 0.01 mg/kg-day in drinking water for 14 days (Merrill,
2001c). Data across the doses were fit by increasing the thyroid
follicle:stroma effective partitioning for perchlorate (PT_p) from 0.13 to 0.4,
1.25, and 2.0 at the 3,10, and 30 mg/kg-day doses.
why the model successfully predicted serum perchlorate concentrations across drinking water
doses with the same parameters used to describe acute exposures and why it could not predict
thyroid concentrations above 3 mg/kg-day.
It could be expected that other parameters (e.g., follicle size and follicular Vmaxc) would
also increase with TSH stimulation. There is an increase in percent of thyroid volume attributed
to the follicle cells (Conde et al., 1991; Ginda et al., 2000), total protein, RNA and DNA content,
and the incorporation of labeled amino acids into protein (Pisarev and Kleiman de Pisarev,
1980). However, Merrill (2001c) notes that adequate predictions could be achieved by adjusting
additional parameters; although, without incorporation of regulation by the hypothalamic-
pituitary-thyroid axis, such adjustments provide little additional insight.
The ability of the adult male rat model to predict iodide uptake inhibition in the thyroid is
demonstrated in Figure 6-13 for a single iv dose of perchlorate (right) or for a 14-day drinking
January 16, 2002
6-32
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
100 q
325 0
325 6
326 2 326 8
Time (hours)
327.4
328 0
CD
E,
•g
o
>.
.C
a>
c
•a
o
"D
0)
0)
.Q
TO
o
"D
CD
q:
10
JP 10 •
0 1
0 01 mg/kg
0.01 mg/kg
0.1 mg/kg
2 3
Time (hours)
Figure 6-13. Male rat PBPK model predictions (lines) versus data time course (mean ± SD)
of iodide uptake inhibition in male rats administered perchlorate either by a
single iv dose (right) or in drinking water for 14 days (left), followed by an iv
dose of 33 ug/kg ,25I'with carrier (Merrill, 2001c). Perchlorate doses were 3.0,
1.0, 0.1, and 0.01 mg/kg-day. Inhibition at the 0.01 and 0.1 mg/kg-day doses
overlaps for the iv dose (right).
water exposure (left). Perchlorate-induced inhibition of 125I" uptake in the thyroid was 13, 24, 70,
and 88% at 2 hours and 11, 29, 55, and 82% at 9 hours after iv dosing with 125I" with carrier for
the 0.01, 0.1, 1.0, and 3.0 mg/kg dose groups. Good simulations were achieved across doses.
However, at 3.0 mg/kg, the model slightly overpredicts inhibition 6 hrs after the perchlorate dose
(4 hours after 125I" administration). TSH was measured from the highest dose level (3.0 mg/kg)
between 8 and 48 hours post dosing and was found to increase between 8 and 12 hrs. It is
possible that TSH was already elevated at 6 hrs, allowing upregulation of the thyroid to
compensate for inhibition at that time point, which the model would not predict. Yu (2000)
provides greater details on hormone fluctuations resulting from the AFRL/HEST experiments.
With respect to iodide inhibition after 14 days of drinking water exposure to perchlorate at
0.01, 0.1, 1.0, 3.0, 10.0, and 30.0 mg/kg-day (Figure 6-13, left), the model overpredicts inhibition
at the 1.0 mg/kg-day dosage and greater. TSH-induced upregulation of the thyroid compensates
for competitive inhibition, resulting in little or no inhibition of radioiodide uptake on Day 14 of
January 16, 2002
6-33
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
exposure in all dose groups except 30 mg/kg-day. In all treated groups, TSH levels were already
increased after the first day. Serum T4 initially decreased in all dose groups except the
0.01 mg/kg-day group. By day 14, T4 levels had increased to control values in the 0.1 and
1.0 mg/kg-day dose groups. FT4 increased in all dose groups on day 1, returned to normal values
by day 5, and were significantly elevated across all dose groups by day 14 (except the 0.1 mg/kg-
day group).
6.2.3 Human Model Development
The adult human PBPK model (Merrill, 2001 d) was developed concurrently with that for
the adult male rate (Merrill, 2001c) and updates the preliminary structure provided to EPA
(Merrill, 2000). Much of the early development was based upon generalizations from previous
AFRL/HEST work on perchlorate (Fisher, 1998a; 2000) and the work of Hays and Wegner
(1965) describing iodide kinetics. As discussed above and shown in Figure 6-1, a nearly
identical model structure to that of the adult male rat was used for the adult human. The human
physiological parameters will of course be different as these should be species-specific. This
section will only highlight notable differences in parameter sources between the two models.
6.2.3.1 Physiologic Parameters and Tissue Partition Coefficients
Human tissue volumes and blood flows were obtained from the literature as shown in
Table 6-1. Merrill (200Id) notes that considerable variability was reported for some parameters.
For example, blood flow to the gastrointestinal (GI) tract can increase ten-fold in response to
enhanced functional activity (secretion and digestion) (Granger et al., 1985). Blood flows used
in the model represent estimates of resting values. Human data on the volume of the gut
capillary bed (VGBc) were not found in the published literature. Therefore, Merrill (200Id) used
a value derived from rat stomach data (Altman and Dittmer, 1971a) for the volume of the
gastrointestinal blood (VGBc) in the human model.
Thyroid volume was obtained from ultrasound measurements on 57 healthy volunteers with
no thyroid disorders (37 to 74 years of age) in a study conducted by Yokoyama et al. (1986). The
mean thyroid volume was 13.4 ± 4.1 mL and mean thyroid volume to body weight ratio was
0.251 ± 0.074 mL/kg (mean ± SD), approximately 0.03% of body weight. Yokoyama et al.
(1986) found a positive correlation between thyroid volume and both body weight and age, with
January 16,2002 6-34 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
weight having the most pronounced influence. The percent of total thyroid volume attributed to
the thyroid follicular epithelium, colloid, and stroma were estimated from histometric
measurements of patients at necropsy by Brown et al. (1986). Their findings on the histological
features of thyroids of men and women showed overlapping distributions without evidence of a
significant difference between sexes. However, a significant sex difference in total fat mass is
reported in humans, with women having approximately 10% more fat than men (Brown et al.,
1997). Based on these data, Merrill (200Id) used a gender-specific value for this parameter.
6.2.3.2 Chemical-Specific Parameters
The various active transport processes, tissue permeabilities, and clearance rates (excretion)
are described in PBPK models for each species on a chemical-specific basis. This section
outlines how the values for perchlorate and iodide used in the human model were derived. The
values can be found in Table 6-2, and the details on derivation are in Merrill (2001d).
6.2.3.2.1 Affinity Constants and Maximum Velocities
The Michaelis-Menten affinity constant (Km) estimates for perchlorate and iodide in the
various tissues with active transport were developed in the human in an analogous fashion to that
in the rat, as described above in Section 6.2.2.2., based on Golstein et al. (1992), Gluzman and
Niepomniszcze (1983), and Wolff (1998). The maximum velocity capacity (Vmaxc) values were
estimated for the various compartments by fitting the simulations to available data at various
doses (Merrill, 200Id).
6.2.3.2.2 Effective Partitions, Permeability Area Cross Products, and Clearance Values
Permeability area cross products and clearance values for perchlorate and iodide were
developed by fitting to literature values in an analogous fashion to that for the rat described in
Section 6.2.2.3 (Merrill, 2001d).
6.2.3.3 Adult Human Model Parameterization and Validation
The human PBPK model for iodide was developed based on the data of Hays and Solomon
(1965) described in Section 6.2.1.2.1. Model predictions versus the data are shown in
Figure 6-14 for iodide concentrations (ng/L) in the serum (A), thyroid (B), and gastric juice (C);
January 16, 2002
6-35
DRAFT-DO NOT QUOTE OR CITE
-------
*5b 0016 <
"¦5 0 004 -
Control Session
16 24
Time (hours)
"oh 0 032
B
*5b
¦g
£
E
a
U
D
Figure 6-14. Human PBPK model (Merrill, 2001d) predictions (lines) versus mean 131I"
concentration (mg/L) time course (asterisks) in serum (A), thyroid (B), gastric
juice (C), and urine (D). Data of Hays and Solomon (1965) are for nine
healthy males dosed with 10 fid 131I" (approximately 3.44 ng/kg).
1 cumulative iodide excreted in the urine (ng) is shown in D. In this study, aspirated gastric juice
2 accounted for an average of 23% of the iv dose within 3 hours after iv injection with radiolabeled
3 iodide (131I") (Merrill, 2001d). Simulation of the gastric juice removed during the aspiration
4 session (Figure 6-14, C) required mathematically removing the amount of l31I" reabsorbed by the
5 stomach wall. This was accomplished by adjusting the rate of reabsorption of 131I" from gastric
6 juice to gastric tissue during the aspiration session as described in Merrill (200Id). The Vmaxc
7 values for the gut and thyroid were then obtained by fitting values of 131I" uptake into gastric juice
January 16, 2002
6-36
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
from the aspiration session (lower lines in Figures 6-14; B and C). The urinary clearance value
was fitted to simulate both cumulative urine content and serum iodide concentration from the
aspiration session data (lower lines in Figures 6-14; A and D). Once parameters were established
using the aspiration session, the rate of change in the gastric juice and partitioning back into the
gastric juice from the systemic circulation was fitted to predict the corresponding increase in
13'I" in plasma, thyroid, and urine seen in the control session versus the aspiration session (upper
lines in Figures 6-14; A, B and D).
Figure 6-15 illustrates that, as for the adult male rat model, plasma binding of perchlorate
was necessary to fit the serum concentration data of the 14-day study by Greer et al. (2000). The
model indicates that humans have a lower binding capacity for perchlorate than rats.
For example, the Vmaxc value for perchlorate is 9.3103 ng/hr-kg in the male rat versus 5.0 x 102
ng/hr-kg in the human. Merrill (200 Id) noted that while the effect of the plasma binding is
subtle at 0.5 mg/kg-day dose, including the plasma binding improved the fit for uptake and
clearance at the 0.1 and 0.02 mg/kg-day dosage levels.
(eiuori) smiT (ziuorl) smiT
Figure 6-15. Simulations illustrating the necessity of including plasma binding in the human
PBPK model structure (Merrill, 2001d). Model predictions (lines) versus data
time course (mean ± SD) are shown with (A) and without (B) plasma binding
for serum concentrations (mg/L) from 4 male subjects dosed with perchlorate
at 0.5, 0.1, or 0.02 mg/kg-day for 14 days (data of Greer et al., 2000).
January 16, 2002
6-37
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
1
2
Cumulative urinary perchlorate excretion (mg) predictions versus the data (mean ± SD) at
each dosage level are shown in Figure 6-16. Merrill (2001d) also simulated serum concentration
(mg/L) and cumulative urinary perchlorate levels (mg) for each individual in the 0.5, 0.1, and
0.02 mg/kg-day dose groups of the Greer et al. (2000) study. An average value for urinary
clearance of perchlorate (ClUc_p) of 0.126 L/hr-kg (± 0.050) was calculated from the
individually fitted values. Figures 6-17 and 6-18 show a representative plot of model prediction
versus individual subject data at the 0.5 and 0.1 mg/kg-day dosage. Additional plots provided in
Merrill (200 Id) provide an appreciation for the high degree of variability in the data.
W>
E
V
c
X
D
«
I—
O
u
0-
u
>
3
u
10.0-
5 10 15
Time (hours)
20
25
Figure 6-16. Human PBPK model predictions (lines) versus data (mean ± SD) of the
observed cumulative urine excretion (mg) in male subjects dosed with
perchlorate 0.5, 0.1, or 0.02 mg/kg-day for 14 days. Model of Merrill (2001d)
and data of Greer et al. (2000).
Serum perchlorate levels were not available for the 0.02 mg/kg-day dose group, but
cumulative urinary excretion amounts (mg) for this group were fitted using the average
January 16, 2002
6-38
DRAFT-DO NOT QUOTE OR CITE
-------
1.20
0.96 -
0.72 "
"3b
E,
7
2
u
CO
c
3 0.48
0.24
0.00
160 240
Time (hours)
400
00
E
3
E
a
U
19.6 29.4
Time (hours)
39.2
49.0
Figure 6-17. Human PBPK model predictions (lines) versus data of one subject's serum
perchlorate concentration (mg/L) shown in (A) and corresponding 48-hour
cumulative urine perchlorate (mg) shown in (B). Subject consumed
0.5 mg/kg-day perchlorate in drinking water, 4 times per day, for 14 days.
Model predictions for the individual obtained by using study average value of
all subjects for urinary clearance of perchlorate (ClUc_p). Model of Merrill
(2001d) and data of Greer et al. (2000).
0.20
160 240
Time (hours)
400
00
E
¦c
D
E
3
U
19.6 29.4
Time (hours)
49.0
Figure 6-18. Human PBPK model predictions (lines) versus data of one subject's serum
perchlorate concentration (mg/L) shown in (A) and corresponding 48-hour
cumulative urine perchlorate (mg) shown in (B). Subject consumed
0.1 mg/kg-day perchlorate in drinking water, 4 times per day, for 14 days.
Model predictions for the individual obtained by using study average value of
all subjects for urinary clearance of perchlorate (ClUc_p). Model of Merrill
(2001 d) and data of Greer et al. (2000).
January 16,2002 6-39 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
1
2
3
4
5
6
7
8
9
perchlorate urinary clearance (C1UC_p) value of 0.126 L/hr-kg calculated from the individual fits
for the 0.1 and 0.5 mg/kg-day groups. Figure 6-19 shows the model predictions versus 48-hour
cumulative urine perchlorate (mg) for two different subjects.
Time (hours) Time (hours)
Figure 6-19. Human PBPK model predictions (lines) versus data of 48-hour cumulative
urine perchlorate (mg) shown for two different subjects. Subject consumed
0.02 mg/kg-day perchlorate in drinking water, 4 times per day, for 14 days.
Model predictions for the individual obtained by using study average value of
all subjects for urinary clearance of perchlorate (ClUc_p). Model of Merrill
(2001d) and data of Greer et al. (2000).
Due to its small size, variations in the thyroid parameters have little effect on serum
concentrations of both iodide and perchlorate. As described for Figure 6-14, Merrill (2001d)
estimated parameters for iodide disposition, including those of the thyroid, from fits to the data
of Hays and Solomon (1965). Using these same iodide parameters, baseline thyroid RAIU
measurements performed by Greer et al. (2000) were fit with the model by adjusting the Vmaxc
for the thyroid follicular epithelium (Vmaxc_T/). Figures 6-20, 6-21, 6-22, and 6-23 illustrate
the model predictions of thyroid RAIU versus data for subjects in the 0.5, 0.1, 0.02, and
0.007 mg/kg-day dosage groups, using either the individual's Vmaxc_T/ (left) or an average
value (right). The average Vmaxc_Tz (1.5 x 105 ng/hr-kg) was obtained from fitting baseline
January 16, 2002
6-40
DRAFT-DO NOT QUOTE OR CITE
-------
0.0180
317 322 327 332
Time (hours)
ob 0.0144 -
I
e
0 0108 -
- 0.0072
£ 0.0036
0.0000
312 317 322 327
Time (hours)
Time (hours)
312 317
322 327 332
Time (hours)
Figure 6-20. Human PBPK model predictions (lines) versus data (asterisks) for thyroid
RAIU (ng/L) on day 14 of perchlorate exposure at 0.5 mg/kg-day for a
healthy female (top panel) and male (bottom panel). Prediction on left for
female (A) obtained by using individually fitted maximum capacity (ng/hr-
kg) for active transport of iodide into the thyroid follicular epithelium
(Vmaxc_T/) of 1.3 x 10s and on right (B) by using an average Vmaxc_Ti.
Prediction on left for male (C) obtained by using individually fitted
Vmaxc_Ti of 1.24 x 10s and on right (D) by using an average Vmaxc_Ti of
1.5 x 10s. Model of Merrill (2001d) and data of Greer et al. (2000).
January 16,2002 6-41 DRAFT-DO NOT QUOTE OR CITE
-------
00180
3
c
00144 -
."2
o
.£
f—
00108 -
C
u
1
0.0072 -
"S
J
J2
o
*5
0 0036 -
oc
0.0180
eb 00144
.c
H
¦2 0 0072
* 0 0036 -
322 327
Time (hours)
0.0160
322 327
Time (hours)
Figure 6-21. Human PBPK model predictions (lines) versus data (asterisks) for thyroid
RATU (ng/L) on day 14 of perchlorate exposure at 0.1 mg/kg-day for a
healthy female (top panel) and male (bottom panel). Prediction on left for
female (A) obtained by using individually fitted maximum capacity (ng/hr-
kg) for active transport of iodide into the thyroid follicular epithelium
(Vmaxc_Ti) of 1.65 x 10s and on right (B) by using an average Vmaxc_Tt".
Prediction on left for male (C) obtained by using individually fitted
Vmaxc_Ti of 1.2 x 10s and on right (D) by using an average Vmaxc_T/ of
1.5 x 10s. Model of Merrill (2001d) and data of Greer et al. (2000).
January 16, 2002
6-42
DRAFT-DO NOT QUOTE OR CITE
-------
0.0180
"8b 0.0144
322 327 332
Time (hours)
Figure 6-22. Human PBPK model predictions (lines) versus data (asterisks) for thyroid
RATU (ng/L) on day 14 of perchlorate exposure at 0.02 mg/kg-day for a
healthy female (top panel) and male (bottom panel). Prediction on left for
female (A) obtained by using individually fitted maximum capacity (ng/hr-
kg) for active transport of iodide into the thyroid follicular epithelium
(Vmaxc_T/) of 1.4 x 10s and on right (B) by using an average Vmaxc_T/.
Prediction on left for male (C) obtained by using individually fitted
Vmaxc_Ti of 1.5 * 10s and on right (D) by using an average Vmaxc_Ti of
1.5 x 10s. Model of Merrill (2001d) and data of Greer et al. (2000).
January 16, 2002
6-43
DRAFT-DO NOT QUOTE OR CITE
-------
00 00144
312 317 322 327
Time (hours)
332 337
312 317 322 327
Time (hours)
312 317 322 327
Time (hours)
332 337
Figure 6-23. Human PBPK model predictions (lines) versus data (asterisks) for thyroid
RAIU (ng/L) on day 14 of perchlorate exposure at 0.007 mg/kg-day for a
healthy female (top panel) and male (bottom panel). Prediction on left for
female (A) obtained by using individually fitted maximum capacity (ng/hr-
kg) for active transport of iodide into the thyroid follicular epithelium
(Vmaxc_T7) of 2.8 x 10s and on right (B) by using an average Vmaxc_Ti.
Prediction on left for male (C) obtained by using individually fitted
Vmaxc_Ti of 1.24 x 10s and on right (D) by using an average Vmaxc_T/ of
1.35 x 10s. Model of Merrill (2001d) and data of Greer et al. (2000).
January 16, 2002
6-44
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
radioiodide uptake measurements provided by Greer et al. (2000) across doses (see Merrill,
2001d; Table 3). Merrill (2001d) hypothesized that the large variability in Vmaxc_T/, ranging
from 5.0 x 104 to 5.0 * 105 ng/hr-kg, may be attributed to variability in endogenous iodide levels,
as dietary iodide was not controlled. Merrill (2001 d) estimated these values from best visual fits
of baseline 8- and 24-hour thyroid RAIU data. Inhibition data restricted to each time point (i.e.,
8- versus 24-hour time points) and from higher dose groups would be useful to test the
robustness of the model to predict inhibition of uptake of iodide in the thyroid.
The ability of the human model to predict data from other independent experiments not
used to develop the model is illustrated in Figure 6-24. The model adequately predicts
cumulative perchlorate in urine (mg) reported in three published studies using therapeutic
perchlorate dose levels (Merrill, 200Id). Oral doses administered in these studies were
approximately 9.07 mg/kg (Durand, 1938), 9.56 mg/kg (Kamm and Drescher, 1973), and
20 mg/kg (Eichler, 1929). It is worth noting that the previously determined urinary clearance
value (ClUc_p) of 0.126 L/hr-kg was used with all validation data and that an adequate fit was
observed.
The ability of the model to predict cumulative perchlorate in urine from three different
studies at three different doses with the same set of parameters, established from the studies by
Hays and Solomon (1965) and Greer et al. (2000), demonstrates the usefulness of the model and
provides validation for the model structure and the physiological and chemical parameters used.
The model also predicts serum perchlorate concentrations at 12 mg/kg-day from an
unpublished study performed by Dr. Georg Brabant at the Medizinische Hochschule, Hanover,
Germany (Figure 6-25). Subjects received 12 mg/kg-day perchlorate in drinking water near meal
times. Variability in the observed serum measurements is believed to reflect variability in the
dosing regimen, as the experimental protocol was less fixed than that used in Greer et al. (2000).
Again the usefulness of the model is demonstrated by its ability to successfully predict serum
concentrations from a dose 24 times higher than the high dose used to establish perchlorate
parameters (0.5 mg/kg-day).
The model is also able to successfully predict the thyroidal iodide uptake in a subject from
the Stanbury and Wyngaarden (1952) study with patients with Grave's disease. The maximum
velocity capacity in the follicular epithelium (Vmaxc_T/) had to be increased to 5.0E6 ng/hr-kg,
a factor of ten times higher than in normal subjects, in order to achieve this fit (upper line in
January 16, 2002
6-45
DRAFT-DO NOT QUOTE OR CITE
-------
Time (hours)
Time (hours)
Time (hours)
Figure 6-24. Validation for human PBPK model (Merrill, 2001d). Model predictions
(lines) versus data (asterisks) for cumulative perchlorate excretion in urine
(mg) in a healthy male after an oral dose of 9.56 mg (A), 20 mg (B) or 9.07 mg
(C). Data are from three different studies. Data of Kamm and Drescher
(1973) for (A), Eichler (1929) for (B) and Durand (1938) for (C).
January 16, 2002
6-46
DRAFT-DO NOT QUOTE OR CITE
-------
Time (hours)
Figure 6-25. Validation for human PBPK model (Merrill, 2001d). Model predictions
(lines) versus data (asterisks) for serum perchlorate concentrations (mg/L) in
5 subjects received 12 mg/kg-day in drinking water (data of Brabant and
Letiolf, 2000 as cited in Merrill, 2001d). Subjects were instructed to ingest the
solution 3 times/day for 14 days. Serum samples were collected 2 hours after
the first dose, after 12 pm on day two, the morning of day 14 and post
exposure days 1 and 2. Usefulness of the model is demonstrated by its ability
to successfully predict serum concentrations at a dose 24 times higher than the
dose used to develop parameters in the model.
1 Figure 6-26). This increase in Vmaxc_T/' is supported in the literature, as Gluzman and
2 Niepomniszcze (1983) measured elevated Vmaxc(s) in thyroid specimens from subjects with
3 Grave's disease. However, the model underpredicts the degree of inhibition caused by
4 perchlorate in this subject (Figure 6-26, lower line). It would appear that the increased inhibition
5 could be attributed to a lower Km value. However, Gluzman and Niepomniszcze (1983) noted
6 that the Km did not differ greatly between thyroid specimens from hyperthyroid subjects and
January 16, 2002
6-47
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
0.060
T3
CQ
OS
0.048 -
0.036 -
J
"5ft
c
I
JO
f-
fC
O)
T3
— 0.024 H
T3
JJ
JS
jg
5 0.012 H
Before Perchlorate
After Perchlorate
0.000 A
0.0
1.2
2.4 3.6
Time (hours)
4.8
6.0
Figure 6-26. Validation for human PBPK model (Merrill, 2001 d). Model predictions
(lines) versus data (asterisks) for RAIU in the thyroid (131I" ng/L) of a male
with Graves' disease after an iv dose of 10 txCi 131I" before and after a 100 mg
dose of potassium perchlorate. Data of Stanbury and Wyngaarden (1952).
normal subject. This suggests that the increased inhibition by perchlorate seen in Grave's disease
may be attributed to a mechanism other than NIS affinity (Merrill, 200Id).
6.2.4 Summary
The proposed model structures for the adult male rat (Merrill, 2001c) and adult human
(Merrill, 200Id) have been shown to adequately describe both perchlorate and iodide disposition
by demonstrating good correspondence between predicted tissue compartment concentrations
and measured values in the thyroid, serum, red blood cells, urine, liver, muscle, skin, and
stomach in the rat and by adequately predicting serum concentrations and cumulative urine after
drinking water exposure to perchlorate spanning four orders of magnitude (0.02 to 12.0 mg/kg-
day) in the human. Serum perchlorate levels for human subjects were not available at
January 16, 2002
6-48
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
0.02 mg/kg-day; however, the model did predict the cumulative urine from that dose group
(Figure 6-19).
The model structure of the thyroid requires three compartments (stroma, follicle, and
colloid) to quantify rapid organification in the gland. Differences in model parameters between
iodide and perchlorate indicate that iodide kinetics are very similar to perchlorate kinetics, but
cannot be applied directly. The main differences involve the saturable kinetics of the thyroid,
skin, and stomach, with perchlorate exhibiting higher Vmaxc's except in the skin. Because
organification of iodide occurs in both the thyroid follicle and colloid, their respective Vmaxc's
are over 1,000 and 10 times higher than those for perchlorate, which is discharged unchanged.
Perchlorate affinity for the symporters into the follicle and colloid were approximately an order
of magnitude greater (lower Km) than those of iodide.
The thyroid perchlorate concentrations from high drinking water exposures in the rat were
fitted by increasing the effective follicle: stroma partition coefficient (PT_p) to account for TSH
stimulation and upregulation of N1S. Since these values were not supported by additional data,
thyroid concentrations may not be as reliable. Further, the toxic effects of perchlorate are most
likely due to secondary effects on thyroid hormones due to its action at the NIS.
The model, however, could simulate serum concentrations from drinking water exposures
using parameters established from the acute data. The thyroid, given it's small size, would not be
expected to significantly alter serum concentrations, even during hyperstimulation. Although
TSH has not been shown to increase the NIS in other tissues, NIS-containing tissues were not
obtained from the AFRL/HEST studies to support this.
The models support plasma protein binding of perchlorate in both species; a saturable term
is required to simulate serum concentrations at lower doses. It is possible that perchlorate
competes with thyroxine for the same binding sites of plasma proteins, as the work of Yamada
and Jones (1968) suggests. Urinary clearance values of 0.05 L/hr for iodide and 0.07 L/hr for
perchlorate were used across data sets in the rats, and average urinary clearance values were
found to be 0.1 L/hr-kg for iodide and 0.126 L/hr-kg in humans. Excretion constants were
highest among the 0.1 mg/kg-day group. With the urinary excretion rates fitted to cumulative
urine data, the model tends to slightly underestimate serum perchlorate levels at repeated low
doses. Elevated serum concentrations may indicate plasma binding of perchlorate. Yamada and
Jones (1967) studied effects of different anions on plasma binding to thyroxine and noted that
January 16, 2002
6-49
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
some of the thyroxine had been displaced after perchlorate was introduced. Thus, it is possible
that perchlorate competes with thyroxine for the same binding sites of plasma proteins (Merrill,
2001c,d; Clewell, 2001a).
While there are limited data suggesting iodide and perchlorate uptake through the skin, the
models and the kinetic studies required this assumption in the models for both rats and humans.
Without the skin compartment, the models overestimated circulating plasma inorganic iodide and
perchlorate in both species. Due to its large size, skin appears to be an important pool for slow
turnover of these anions. Brown-Grant (1961) noted that the uptake of iodide was higher in the
male rat and pup than in the female. The findings of Merrill (2001c) agree, with the rat model
requiring a higher Vmaxc in the skin for the male rat than that reported for the pregnant rat
(Clewell, 2001a) discussed in the next section. Cutaneous uptake of iodide and perchlorate in
mice and rats has been reported (Brown-Grant and Pethes, 1959; Zeghal et al., 1995). The lack
of reported iodide in human skin from clinical radioiodide scans may be due to the difficulty in
differentiating it from background radioactivity.
Merrill (200Id) notes that GI clearance of iodide is rapid and plays an important role in
radioiodide conservation. Further, Merrill (200Id) suggests that the appearance of time-course
radioiodine in stomach contents of any species is complicated by the fact that it reflects more
than sequestration of radioiodide by NIS. Its appearance also reflects radioiodide contributed
through the gradual accumulation of iodide in saliva that is swallowed involuntarily throughout
the study. Several studies that examined sequestration of these anions in digestive juices have all
shown high variability in the concentrations measured over time (Honour et al., 1952; Hays and
Solomon, 1965; Merrill, 2001d). There is a tendency for the gastric juice to plasma ratio to be
low when the rate of secretion of juice is high (Honour et al., 1952). Fluctuations in the secretion
rate are probably the most important factor in determining the pattern of the concentration ratios
in individuals. Therefore, variability in stomach or GI tract parameters between models is
expected. However, the early rise in the gastric juice:plasma ratio mentioned earlier is a constant
feature across these data sets, whether or not an attempt was made to eliminate contamination of
gastric juices by dietary contents or saliva. The human model successfully predicted this same
trend.
Merrill (2001 d) also noted dietary iodine and endogenous inorganic iodide levels to be
clearly important in modeling iodide and perchlorate kinetics, because excessive iodide levels
January 16, 2002
6-50
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
cause the ion to inhibit its own uptake. Plasma inorganic iodide (PII) is rarely reported in the
literature due to analytical difficulties, and it was not available in any of the studies presented in
this paper. While measurements of tracer radioiodide can be fitted to predict transfer rates, its
use is limited when attempting to predict the saturation of nonlinear compartments, such as the
thyroid that are dependent upon the existing amount of iodide already present. Subsequent
modeling efforts on predicting subsequent effects of iodide inhibition on thyroid hormone
synthesis and regulation in humans will require the capability of the model to predict PH.
6.3 PREGNANT RAT AND FETAL MODEL STRUCTURE
This section describes the model developed by AFRL/HEST in response to concerns about
interspecies extrapolation of effects due to perchlorate exposure during gestation (Clewell,
2001a). The model predicts the distribution of perchlorate within the pregnant and fetal rat
through gestation and at birth and predicts the short-term effect of acute perchlorate exposure on
iodide kinetics, including iodide uptake into the maternal thyroid. The general model structure
relied on the adult male rat model (Merrill, 2001c) described in Section 6.2 and approaches to
gestational growth of the dam and fetus were based on the work of O'Flaherty et al. (1992) and
Fisher et al. (1989) with weak acids.
The model structure is shown in Figure 6-27. Table 6-3 provides the physiological
parameters for the pregnant rat and fetus PBPK models. Table 6-4 provides the perchlorate-
specific parameters, and Table 6-5 provides the iodide-specific parameters for each.
The compartments shared with the adult male rat were developed as described in
Section 6.2. The pregnant rat model also includes a mammary gland and placenta compartment.
The mammary gland consists of two subcompartments that represent the capillary bed and the
tissue. The mammary gland has been shown to concentrate both perchlorate and iodide during
lactation. However, the mammary NTS is regulated by hormones produced during lactation and
has been found to increase at the onset of lactation (Tazebay et al., 2000). This concentrating
mechanism does not appear to be as established during pregnancy. Studies reported by Yu
(2000) showed mammary gland:plasma ratios of less than one for perchlorate. However,
mammary gland perchlorate levels are slowly built up and remain high well into the clearance
phase of the serum. This behavior suggested a very slow diffusion between the mammary gland
January 16, 2002
6-51
DRAFT-DO NOT QUOTE OR CITE
-------
Ivc_p
\
pDose_
Pregnant Dam
bound C104"
Plasma
free Cl(
/iv/4
RBCs
GI ^
C ojitents. A
^ Tract
GI
Blood
QG
Liver
Richly
Perfused
Kidney
Wi
Mammary
Gland.ifc
Mammary I
Blood
^ Thyroid
F.QiUgte A
Stroma I
Colloid
Skin
Skin
Blood
Fat
Slowly
Perfused
Placenta
*
Fetus
Ktransl
QC
QG
QL
QR
QM
QT
QSK
QF
QS
CA_p
1 Urine
QPln
Ktrans2
bound CIO4"
Plasma 1, |
|A freeC104"
^ RBCs
QC
GI
^.Contents.
JA Tract
l*r gi
Blood
QG
QG
Liver
QL
Richly
Perfused
QR
Kidney
QK
Colloid
Thyroid
"—Follicle.
Stroma
Skin
Skin
Blood
QT
Slowly
Perfused
QSK.
QS
CA_p
Figure 6-27. Schematic for the pregnant dam and fetal rat PBFK model of perchlorate and
iodide distribution (Clewell, 2001a). Bold arrows indicate (except for plasma
binding) active uptake at NIS sites into the thyroid, GI contents, and skin.
Plasma binding was also described with Michaelis-Menten terms for the
association of perchlorate anion to binding sites with first-order clearance
rates for dissociation. Small arrows indicate passive diffusion. Boxes
represent specific compartments in the model structure. The thyroid consists
of the stroma, the follicle, and the colloid; and the stomach consists of the
capillary bed, GI wall, and contents. The skin and mammary gland each
contain two subcompartments representing the capillary bed and tissue.
Permeability area cross products and partition coefficients were used to
describe the first-order movement of the perchlorate (C104 ) and iodide (I )
anions into deeper subcompartments. Placental-fetal transfer and urinary
clearance were represented by first order clearance rates.
January 16, 2002
6-52
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-3. PHYSIOLOGICAL PARAMETERS FOR THE PREGNANT RAT AND
FETUS PBPK MODEL (Clewell, 2001a)
Physiological Parameters Pregnancy
Tissue Volumes (%BW)
Dam
Fetus
Source
Body Weight BWand Vlfet (kg)
0.280-0.361
0.0 - .0045
O'Flahertyet al., 1992
Slowly Perfused VSc (%BW)
74.6
74.6
Brown et al., 1997
Richly Perfused VRc (%BW)
11
11
Brown et al., 1997
Fat VFc (%BW)
10.0- 11.0
0.0
Naismith et al., 1982
Kidney VKc (%BW)
1.7
1.7
Brown et al., 1997
Liver VLc (%BW)
3.4
3.4
Brown et al., 1997
GI Tract VGc (%BW)
3.60
3.60
Brown et al., 1997
GI Contents VGJc (%BW)
7.20
7.20
Yu et al., 2000
GI Blood VGBc (%VG)
2.9
2.9
Altman and Dittmer, 1971
Skin Tissue VSkc (%BW)
19.0
19.0
Brown et al., 1997
Skin Blood VSkBc (%VSk)
2.0
2.0
Brown et al., 1997
Thyroid Total VTtotc (%BW)
0.0105
0.0234
Malendowicz and Bednarek, 1986;
Florsheim et al., 1966
Thyroid Follicle VTc (%BW)
45.9
61.4
Malendowicz and Bednarek, 1986;
Conde et al., 1991
Thyroid Colloid VDTc (%BW)
45
18.3
Malendowicz and Bednarek, 1986;
Conde et al., 1991
Thyroid Blood VTBc (%VT)
9.1
20.3
Malendowicz and Bednarek, 1986;
Conde et al., 1991
Plasma VPlasc (%BW)
4.7
4.7
Brown et al., 1997;
Altman and Dittmer, 1971
Red Blood Cells VRBCc (%BW)
2.74
2.74
Brown et al., 1997;
Altman and Dittmer, 1971
Placenta VPl ac (%BW)
0.0-2.57
—
O'Flaherty et al., 1992
Mammary Tissue VMc (%BW)
1.0-5.5
—
Knight et al., 1984; O'Flaherty et al., 1992
Blood Flows (%QC)
Cardiac Output QCc (L/hr-kg)
14
14.0
Buelke-Sam, 1982a & b;
O'Flaherty et al., 1992
Slowly Perfused QSc (%QC)
24.0
24.0
Brown et al., 1997
Richly Perfused QRc (%QC)
76.0
76.0
Brown et al., 1997
Fat QFc (%QC)
7-8.1
0.0
Brown et al., 1997
Kidney QKc (%QC)
14.0
14.0
Brown et al., 1997
Liver QLc (%QC)
18.0
18.0
Brown et al., 1997
GI QGc (%QC)
13.60
13.60
Brown et al., 1997
Thyroid QTc (%QC)
1.6
1.6
Brown et al., 1997
Mammary QMc (%QC)
0.2-1.2
—
Hanwell and Linzell, 1973
Placenta QPlc (%QC)
0.0-12.3
—
O'Flaherty et al., 1992
January 16, 2002
6-53
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-4. PERCHLORATE-SPECIFIC PARAMETERS FOR THE PREGNANT RAT
AND FETUS PBPK MODEL (Clewell, 2001 a)a
Pregnancy Parameters
Pcrchloratc Values
Partition Coefficients (unitless)
Dam
Fetus
Source
Slowly Perfused/Plasma PS_
0.31
0.31
Yu et al., 2000
Rapidly Perfused/Plasma PR_
0.56
0 56
Yu et al., 2000
Fat/Plasma PF_
0 05
—
Pena et al., 1976
Kidney/Plasma PK_
0.99
0.99
Yu et al , 2000
Liver/Plasma PL_
0.56
0.56
Yu et al., 2000
Gastric Tissue/Gastric Blood PG_
0.50
1.80
Yu et al., 2000
G1 Contents/GI Tissue PGJ_
1.30
2.30
Yu, 2000
Skin Tissue/Skin Blood PSk_
1.15
1 15
Yu, 2000
Thyroid Tissue/Thyroid Blood PT_
0.13/225
0 13/2.25
Chow and Woodbury, 1970b
Thyroid Lumen/Thyroid Tissue PDT_
7.00
7 00
Chow and Woodbury, 1970
Red Blood Cells/Plasma
0.73
0 73
Yu et al., 2000
Placenta/ Plasma PPL_
0.56
—
Assume same as richly perfused
Mammary/Plasma PMam_p
0.66
—
Anbar et al., 1959
Max Capacity, Vmaxc (ng/hr-kg)
Thyroid Follicle Vmaxc_T
1.80E+03
1.80E+03
Fitted'
Thyroid Colloid Vmaxc_DT
1.00E+04
1.00E+04
Fitted'
Skin Vmaxc_S
6.00E+05
4.00E+05
Fitted
Gut Vmaxc_G
8.00E+05
8.00E+05
Fitted
Mammary Gland Vmaxc_M
3.90E+04
...
Molar equivalent to Vmaxc_Mi
Plasma Binding Vmaxc_Bp
5.00E+03
1.50E+03
Fitted
Affinity Constants, Km (mg/L)
Thyroid Follicle Km_T
1.00E+05
1.OOE+05
Wolff, 1998
Thyroid Colloid Km_DT
1,00E-H)8
1.00E+08
Golstein et al., 1992; Wolff, 1998
Skin Km_S
1 OOE+05
1.00E+05
Wolff, 1998
Gut lCm_G
1.00E+05
1.00E+05
Wolff, 1998
Mammary Gland
1.00E+5
—
Wolff, 1998
Plasma Binding Km_Bp
1 00E+05
1 00E+05
Fitted
Permeability Area Cross Products, (L/hr-kg)
GI Blood to GI Tissue PAGc_
1.00
1.00
Fitted
GI Tissue to GI Contents PAGJc_
1.00
1 00
Fitted
Thyroid Blood to Thyroid Tissue PATc_
4.0E-5 / 6.0E-4
4.0E-5 / 6 0E-4
Fittedb
Thyroid Tissue to Thyroid Lumen PADTc_
0.01
0.01
Fitted
Skin Blood to Skin Tissue PASkc_
1 00
1.00
Fitted
Plasma to Red Blood Cells PRBCc_
1 00
1.00
Fitted
Clearance Values, (L/hr-kg)
Urinary Excretion CLUc_
0.07
—
Yu et al., 2000
Transfer from Placenta to Fetus Cltranslc_
0.10
0.10
Yu, 2000
Transfer from Fetus to Placenta Cltrans2c_
0 19
0.19
Yu, 2000
Dissociation from Plasma Binding Sites
Clunbc_p
0.034
0010
Yu, 2000
"All parameters listed are notated in the model by either an i (for iodide) or p (for perchlorate) following an underscore in the parameter
name (e.g., PR_z, PR_p, Vmaxc_T/, etc.)
'Parameters with two values indicate acute and drinking water parameters, respectively.
'Fetus was given maternal values for Vmax (scaled by fetal body weight) in the absence of data.
January 16, 2002
6-54
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-5. IODIDE-SPECIFIC PARAMETERS FOR THE PREGNANT RAT AND
FETUS PBPK MODEL (Clewell, 2001a)a
Pregnancy Parameters
Iodide Values
Partition Coefficients (unitless)
Dam
Fetus
Iodide Source
Slowly Perfused/Plasma PS_
0.21
0.21
Halmi et al., 1956
Rapidly Perfused/Plasma PR_
0.40
0.40
Halmi et al., 1956
Fat/Plasma PF_
0.05
—
Pena et al., 1976
Kidney/Plasma PK_
1.09
1.09
Yu et al., 2000
Liver/Plasma PL_
0.44
0.44
Yu et al., 2000
GI Tissue/GI Blood PG_
1.0
1.0
Yu, 2000
GI Contents/GI Tissue PGJ_
2.0
2.0
Yu, 2000
Skin Tissue/Skin Blood PSk_
0.70
0.70
Perlman et al., 1941
Thyroid Tissue/Thyroid Blood PT_
0.15
0.15
Chow and Woodbury, 1970
Thyroid Lumen/Thyroid Tissue PDT_
7.00
7.00
Chow and Woodbury, 1970
Red Blood Cells/Plasma
1.00
1.00
Yu et al., 2000
Placenta/Plasma PPL_
0.99
—
Unpublished GD20 data
Mammary/Plasma PMam_p
0.66
—
Anbar et al., 1959 (for C104)
Max Capacity, Vmaxc (ng/hr-kg)
Thyroid Follicle Vmaxc_T
4.00E+04
0.0 - 7.5E+04
Fitted
Thyroid Colloid Vmaxc_DT
6.00E+07
6.00E+07
Fitted
Skin Vmaxc_S
6.00E+04
3.00E+05
Fitted
Gut Vmaxc_G
1.00E+06
2.00E+05
Fitted
Mammary Gland Vmaxc_M
5.00E+04
—
Fitted
Affinity Constants, Km (mg/L)
Thyroid Follicle Km_T
4.00E+06
4.00E+06
Gluzman and Niepomniszcze, 1983
Thyroid Colloid Km_DT
1.00E+09
1.00E+09
Golstein et al., 1992
Skin Km_S
4.00E+06
4.00E+06
Gluzman and Niepomniszcze, 1983
Gut Km_G
4.00E+06
4.00E+06
Gluzman and Niepomniszcze, 1983
Mammary Gland Km_M
4.00E+06
—
Gluzman and Niepomniszcze, 1983
Permeability Area Cross Products, (L/hr-kg)
GI Blood to GI Tissue PAGc_
0.80
0.10
Fitted
GI Tissue to GI Contents PAGJc_
0.60
0.30
Fitted
Thyroid Blood to Thyroid Tissue PATc_
1.000E-04
1.000E-04
Fitted
Thyroid Tissue to Thyroid Lumen PADTc_
1.00E-04
1.00E -04
Fitted
Skin Blood to Skin Tissue PASkc_
0.10
0.02
Fitted
Plasma to Red Blood Cells PRBCc_
1.00
1.00
Fitted
Clearance Values, (L/hr-kg)
Urinary excretion CLUc_
0.03
—
Fitted
Transfer from Placenta to Fetus Cltranslc_
0.06
0.06
Unpublished GD 20 Iodide iv Data
Transfer from Fetus to Placenta Cltrans2c_
0.12
0.12
Unpublished GD 20 Iodide iv Data
"All parameters listed are notated in the model by either an i (for iodide) or p (for perchlorate) following an
underscore in the parameter name (e.g., PR_«, PR_p, Vmaxc_T/, etc.)
January 16, 2002
6-55
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
and blood, so that Clewell (2001a) described the two-subpartment mammary gland with both
diffusion of iodide and active uptake by the NIS.
Although it has been suggested that the placenta may contain the capability for active
uptake in the rat, AFRL/HEST data did not indicate placenta:plasma levels greater than one for
perchlorate or iodide (Yu, 2000), and unpublished iodide time course data indicate that the
behavior of iodide in the placenta closely mirrors that of the plasma (Clewell, 2001a). Thus, the
placenta was simulated with a single, flow-limited compartment.
Partitioning into the mammary, placenta, and other diffusion-limited compartments was
based on effective partitioning. This effective partitioning is probably very similar to that in the
thyroid where an electrochemical gradient is responsible for allowing the C104" anion to move
between the serum and the tissue (Chow and Woodbury, 1970). Urinary clearance and placental-
fetal transfer of the anions were represented by first order clearance rates.
The structure of the fetal perchlorate model is similar to that of the pregnant rat, with the
exception of the mammary gland and placenta compartments. In order to simplify the model, all
of the fetuses from a single litter were combined in the structure of the model, essentially
viewing the individual fetuses as one entity, or one large fetus. The dose to the fetus is based on
the transfer of perchlorate from the maternal placenta to the serum of the fetus, rather than
through direct exposure to the drinking water. Though a kidney is included in the fetal model,
urinary excretion is not used to identify the loss of perchlorate for the fetus. Since the ability to
produce urine is not well developed until after parturition, the loss from the fetus is described as
first order clearance from the fetal arterial blood to the placenta (Clewell, 2001a).
The pregnancy model attempts to describe perchlorate distribution in a highly dynamic
system. In addition to total body weight changes in the dam and fetus, maternal mammary tissue
and blood flow, cardiac output, fractional body fat, placenta and fetus body weight, and fractional
body fat are also changing with respect to time. Growth equations, based on O'Flaherty et al.
(1992) were used to account for these changes (Clewell, 2001a). All tissue volume and blood
flow values were adjusted with respect to the changing parameters.
6.3.1 Data and Methods
This section summarizes the data that Clewell (2001a) used for development and validation
of the pregnant and fetal rat model structures. Details on experimental methods, including:
January 16,2002 6-56 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
protocol design, exposure regimen, chemical source and purity, animals (housing, feeding,
surgical procedures, etc.), and the analytical methods can be found in the consultative letter and
associated reports from AFRL/HEST or cited papers therein.
6.3.1.1 AFRL/HEST Experiments in Laboratory Rats
These studies are described in the consultative letters and reports of Clewell (2001a),
Yu (2000, 2001, 2002) and Yu et al. (2000).
6.3.1.1.1 Drinking Water Study
Perchlorate drinking water experiments used in model development were performed at
AFRL/HEST and described in detail in the report Yu (2000). Pregnant dams of the Sprague-
Dawley strain were exposed to drinking water treated with perchlorate from gestational day (GD)
2 through 20, at perchlorate doses of 0.0, 0.01, 0.1, 1.0 and 10.0 mg/kg-day. GD0 was
determined by the presence of a vaginal plug. Both dams and fetuses were sacrificed on GD20
and maternal and fetal serum analyzed for free and total thyroxine (fT4 and tT4), triiodothyronine
(T3), and TSH. Maternal serum, thyroid, skin, GI contents, placenta, and amniotic fluid were
analyzed for perchlorate at all of the above doses. Fetal serum, skin and GI tract were also
analyzed for perchlorate at all of the doses. Two hours before sacrifice, the dams were given iv
doses of 33 mg/kg radiolabeled iodide (125I~) with carrier. Tissue concentrations of iodide were
measured in order to determine the inhibition in the various tissues after long-term exposure to
perchlorate.
6.3.1.1.2 Preliminary Iodide Kinetics Study
A preliminary study of radiolabeled (125I ) kinetics was performed by AFRL/HEST in which
timed-pregnant dams of the Sprague-Dawley strain were exposed via tail-vein injection to a
tracer dose (average dose = 2.19 ng/kg body weight) of the radiolabeled anion on GD20. Dams
(n=6) were sacrificed at 0.5, 2, 4, and 8 hours post-dosing. Maternal serum, thyroids, skin, GI
contents, placenta and mammary gland tissue, as well as fetal serum, skin, and GI tract were
collected and analyzed for iodide content at each time point. Serum was pooled for all fetuses
within a litter, due to limited sample volume. Fetal skin and GI tract were analyzed individually.
January 16, 2002
6-57
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.3.1.1.3 Iodide Inhibition Kinetics Study
A more in-depth study was performed by AFRL/HEST, in which Sprague-Dawley
timed-pregnant dams were given 1.0 mg/kg body weight perchlorate via tail-vein injection on
GD20; control rats were given saline. The perchlorate or saline dose was followed two hours
post dosing with a tail-vein injection of carrier free 125I" at an average dose of 1.87 ng/kg BW.
Dams (n=6) were sacrificed after 0.5, 1, 2, 4, 8, 12, and 24 hours. Maternal serum, thyroids,
skin, GI contents, placenta, mammary gland tissue, and fetal serum, skin, and GI tract were
collected and analyzed for iodide content at each time point. Serum was again pooled for all
fetuses within a litter. Fetal skin and GI tract were analyzed individually. At this time, only the
maternal serum, maternal thyroids and fetal serum from this study were available for use with the
model. Clewell (2001a) states that further validation of the model structure will be performed at
a later time with the remaining data, but no further work has been provided to the EPA.
Additional data were provided by Yu (2002).
6.3.1.2 Data Published in the Literature
Data available in the literature used in a validation exercise of the model are described
briefly in this section.
6.3.1.2.1 Versloot et al., 1997
Versloot and coauthors measured 125I" as percent of dose in maternal and fetal thyroid,
mammary gland, placenta, and fetal carcass without the thyroid. Pregnant Wistar rats (body
weight [BW] = 300 ± 5 g) were given an injection of 10 yuCi carrier free 125I" into the right vena
jugularis on GDI 9. Measurements of the maternal thyroid were taken at 4 and 24 hours post
dosing. Mammary gland, placenta, fetal thyroid, and fetal carcass minus the thyroid were taken
only 24 hours post dosing.
6.3.1.2.2 Sztanyik and Turai 1988
Sztanyik and Turai measured the uptake of iodide into the placenta and fetal whole body
24 hours post dosing. Five groups of CFY albino rats (BW = 200 to 250 g) were dosed ip with
370 kBq (0.081 ng) carrier free radiolabeled iodide (131I ) on GDs 17, 18, 19, 20, and 22.
Although this is a different strain of rat, the GD20 fetal weights (average BW = 4.088 g) compare
January 16, 2002
6-58
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
favorably with those seen on GD20 in the Sprague-Dawley fetus. As a result, Clewell (2001a)
used the GD20 time point as a means of validating GD20 parameters for iodide across different
data sets and doses. Placental and whole body fetal13'I" were measured in a well-type
scintillation detector.
6.3.1.2.3 Feldman et al., 1961
Feldman and coauthors measured the uptake of iodide into the fetal thyroid and rest of body
carcass on GDs 16, 17, 18, and 19 in pregnant female Holtzman rats. A single subcutaneous
injection was given to the dam, containing 50 iuCi of 131I" on each of the days mentioned above.
Fetal thyroid and carcasses were measured at 24 hours post dosing.
6.3.2 Pregnant Rat and Fetus Model Development
This section summarizes only the key features that were different than the adult male rat
model previously described in Section 6.2.
6.3.2.1 Physiological Parameters and Tissue Partition Coefficients
Maternal parameters were scaled allometrically based on body weight as previously
described for the male rat. Fetal values were scaled in the same manner as the maternal
parameters. However, since the model actually represents several fetuses, it was necessary to
first scale the values for the individual fetus and then adjust for the total number of fetuses in the
litter (Clewell, 2001a).
Clewell (2001a) based the physiological description of the maternal and fetal rat during
gestation on O'Flaherty et al. (1992). However, growth descriptions, body weights, and organ
descriptions were optimized for use within this particular model structure. The model is able to
account for differences in gestation time, pup birth weight, and litter size between experiments
and strains of rats. Growth equations and parameters that change over time were described with
mathematical descriptions of available literature and experimental data. Details and equations
are provided in the consultative letter (Clewell, 2001a).
January 16, 2002
6-59
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.3.2.1.1 Maternal Tissues
The body weight of the dam is known to change significantly throughout the relatively
short gestation time in the rat (21 days). However, the traditional approach utilizing allometric
scaling to describe tissue growth in relation to the change in body weight is not a sufficient
description for the changes taking place during pregnancy. As opposed to the typical growth
scenario, organs and tissues cannot be assumed to increase at the same rate in this dynamic
system (Clewell, 2001a). The placenta, fetal volume, and mammary tissue grow at an
accelerated rate in comparison to the other organs. These require additional descriptions for their
growth beyond the previously described allometric scaling by body weight.
Since the growth of the other tissues is negligible in comparison to the change in the
placenta, mammary gland, fat and fetal volume, Clewell (2001a) described the total change in the
maternal body weight as simply the change in these four volumes added to the initial (pre-
pregnancy) body weight (BWimt). All other maternal organs were assumed to remain constant
and were scaled allometrically relative to the initial body weight (see Table 6-5).
Mammary tissue growth during gestation was described by Knight and Peaker (1982).
Based on this work, Clewell (2001a) described mammary tissue growth as a linear process during
which the mammary gland reaches a maximum volume for gestation on GD21 of 4.6% of the
maternal body weight.
Clewell (2001a) also described the growth of maternal fat as a linear process throughout
gestation based on the work of Naismith et al. (1982). Naismith reported a 40% increase in body
fat throughout gestation. Thus, in the model a linear equation was employed to describe a 40%
increase in body fat during the length of gestation with an initial (non-pregnant) value of 7.0%
body weight for Sprague-Dawley rats (Brown et al., 1997).
Placental volume was described in the model as a sum of three stages of growth, based on
the data of Buelke-Sam et al. (1982a), Sikov and Thomas (1970), and the mathematical
description of data provided in O'Flaherty et al. (1992). The placenta volume is negligible
during gestational days 1 through 5. Individual yolk sac placenta enter a stage of rapid growth
between days 6 and 10 of gestation, and was described by an equation that accounted for yolk sac
placenta, the total volume of the placenta during this time period, and the number of fetuses
present. Placental growth during gestational days 6 through 10 is defined solely by this equation.
Total placenta volume changes during gestational days 10 through 21 (parturition) were defined
January 16, 2002
6-60
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
by two separate processes: the exponential decline in yolk sac volume and the increase in
chorioallantoic placenta (Clewell, 2001a).
O'Flaherty et al. (1992) also described the growth of the uterus and liver during gestation.
However, Clewell (2001a) did not include a specific description of growth in these organs
because the liver is not believed to have a major role in perchlorate kinetics. Further, because the
iodide model does not describe deiodination, the description of liver growth was deemed
unnecessary. The use of a uterine compartment was also not included in the Clewell (2001a)
model due to the lack of available perchlorate and iodide data. The uterus was considered to be
part of the lumped richly perfused tissue. EPA agrees that adding a description of liver growth
would only bring additional complexity to the model structure without providing a real benefit to
the description of perchlorate and total iodide kinetics and that the uterine compartment would be
purely hypothetical and could not be validated without pertinent data.
6.3.2.1.2 Maternal Blood Flow
Clewell (2001a) described temporal changes in maternal cardiac output during gestation as
the sum of the initial cardiac output, given in Brown et al. (1997) for a non-pregnant rat, and the
change in blood flow to the placenta, mammary, and fat tissues. The approach of O'Flaherty
et al. (1992) to changing blood flows was utilized in placental, mammary, and fat blood flows.
The fraction of cardiac output to the mammary gland and fat tissues are described as proportional
to the change in volume of the tissue. The change in blood flow to the yolk sac placenta is
approximately proportional to the change in volume of the yolk sac. However, the blood flow to
the chorioallantoic placenta increases at a faster rate than the change in volume, so three different
equations were used to describe the blood flow for each different stage of placental growth (GDI
to GD6, GD7 to GD10, GDI 1 to GD12, and GD13 to GD21).
6.3.2.1.3 Fetal Tissues
A three stage description of fetal growth was also described in O'Flaherty et al. (1992) in
order to mathematically reproduce data obtained from Beaton et al. (1954), Sikov and Thomas
(1970), Goedbloed (1972), and Buelke-Sam et al. (1982a). Because data are not available for
fetal volume between gestational days 1 through 11, an exponential growth curve was used as a
reasonable approximation of fetal growth and was fit to the first available data for fetal volume
January 16, 2002
6-61
DRAFT-DO NOT QUOTE OR CITE
-------
]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
(Clewell, 2001a). The second stage of growth describes a slower increase in fetal volume,
beginning on GDI 1, based on the same data. Clewell (2001a) described the third stage of fetal
growth as a linear increase between days 18 and the day of parturition. The equation is
dependent on the weight of the pup at the time of birth so that the model can account for the
differences in birth weight encountered when simulating different data sets.
Individual fetal organ weights were assumed to increase linearly with respect to change in
fetal body weight and were therefore scaled allometrically to account for changes in tissue
volumes. Values for tissue volumes were taken from the literature and from experimental data
for the fetus when available. However, most volumes were taken from adult rat data and scaled
allometrically for the fetus due to the lack of tissue data in fetuses.
Florsheim et al. (1966) measured thyroid and body weight of the rat fetus and pup from
GDI8 through PND22 and reported a linear relationship between the thyroid weight and body
weight throughout the time period. The value given for the thyroid of the fetus in %fetal body
weight for GDI9 was used in the Clewell (2001a) model. On the other hand, the physiology of
the developing thyroid was found by Conde et al. (1991) to change significantly between birth
and PND120. Conde reported a decrease in follicle volume from 61.4% to 37.2% of the total
volume of the thyroid from birth to 120 days. An increase in colloid volume from 18.3% of the
total thyroid volume at birth to 32.5% at 120 days was also reported. In the absence of
histometric data in the fetal thyroid, the follicle, colloid, and stroma volumes for the fetus were
described using the thyroid fractions measured at birth. The value for thyroid stroma was
calculated within the model by subtracting the colloid and follicle volumes from the total thyroid
volume.
The fetal body fat content was assumed to be zero in the Clewell (2001a) model. This
assumption is reasonable in light of the data given in Naismith et al. (1982). Naismith et al.
(1982) measured values for the body fat of PND2 and 16 rat pups, corresponding to 0.16% and
3.7% of the body weight. Given that body fat quickly increases in the neonatal period, it is not
unreasonable to assume that body fat in the fetus is negligible. The volume is certainly not large
enough to interfere with iodide or perchlorate kinetics. All other parameters were scaled
allometrically by fetal weight from the adult male rat. The male rat physiological parameters
were used rather than female parameters for several reasons. First, the male rat pups have been
shown to be more sensitive to perturbation of hormone homeostasis by perchlorate, and therefore
January 16, 2002
6-62
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
are considered the sensitive endpoint (Yu, 2000). Additionally, Clewell (2001a) asserts that
sufficient evidence was not found to indicate that physiological parameters between male and
female rats were present in the fetus.
6.3.2.1.4 Fetal Blood Flow
Fetal blood flow was assumed to operate independently from the mother. The transfer of
the chemical was accomplished via diffusion between the placenta and fetal blood. Therefore,
the fetal cardiac output and blood flow to organs (as % cardiac output) were scaled allometrically
from the male rat values relative to the fetal volume.
6.2.2.2 Chemical-Specific Parameters
The various active transport processes, tissue permeabilities, and clearance rates (excretion)
are described in PBPK models for each species on a chemical-specific basis. This section
outlines how the values for perchlorate and iodide used in the pregnant and fetal rat model were
derived. The values can be found in Tables 6-4 and 6-5; details on the derivation can be found in
Clewell (2001a).
6.3.2.2.1 Affinity Constants and Maximum Velocities for Active Uptake Processes
These were developed as described previously for the adult male rat model (Merrill, 2001c)
in Section 6.2. The chemical specific parameters were kept the same in male, female, neonatal
and fetal rats, and humans whenever possible. However, it was necessary to change a few of the
parameters, including the maximum velocities (Vmaxc's) in the Clewell (2001a) model for
pregnant rat and fetus. The Km values were similar between tissues and between female and
male rat and human models. However, the maximum velocity or capacity differs between tissues
(Wolff and Maurey, 1961). Since Vmaxc values for perchlorate were not given in literature, the
values were estimated with the model. In order to determine Vmaxc using the model, the
simulation for the tissue of interest was compared to various data sets with several different
perchlorate dose levels. The value for Vmaxc within a given compartment was then determined
by the best fit of the simulation to the data.
January 16, 2002
6-63
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.3.2.2.2 Effective Partitioning Permeability Area Cross Products and Clearance Values
These were developed as described previously for the adult male rat model (Merrill, 2001c)
in Section 6.2. The value of 0.05 was used to represent the partitioning of perchlorate into the fat
for the pregnant rat and fetus (Clewell, 2001a). This value was based on the data of Pena et al.
(1976) who measured tissue:blood ratios in the laying hen after intra-muscular dosing with either
a single injection of 10 fj.Ci or 3 sequential doses of 10 jaCi radiolabeled perchlorate. Although
the hen is a very different species, several other tissues were reported to have values comparable
to those found by Yu (2000) and Yu et al. (2000) in the male and female rat (0.3 vs. 0.31 in
muscle, 0.1 vs. 0.1 in brain, 0.8 vs. 0.99 in the kidney). Clewell (2001a) noted that the use of this
value is supported by the fact that the polarity of the perchlorate anion would severely limit the
movement of perchlorate into fatty lipophilic tissue. Anbar et al. (1959) measured the mammary
gland:blood ratios in the rat four hours after ip injection of radiolabeled perchlorate (100 mg
KC104), and they reported an effective partition of 0.66 for the rat mammary gland. This value is
in general agreement with that chosen by Clewell (2001a).
Maternal and fetal skin were described using the value Perlman et al. (1941) determined
after a sc tracer dose of iodide for the partition coefficient in this compartment. Iodide partition
coefficients were calculated from the tissue:blood ratios measured during the clearance phase of
iodide data in the tissue of interest. The preliminary iodide kinetics study described in the
supporting experiments was utilized for the determination of the placenta partition coefficients.
For example, values for the GI tract and its contents were determined from the clearance portion
of the iodide kinetic study in the adult male rat (Yu et al., 2000).
For all tissues in which a clearance was described (urinary clearance, transfer between
placenta and fetal serum, and dissociation of perchlorate from the binding sites), a clearance
value was determined. Since perchlorate is quickly excreted in urine and binding has little effect
on serum levels at high doses, the simulation for the 10 mg/kg-day dose group was primarily
dependent on the urinary clearance value (ClUc_p). The urinary clearance value for perchlorate
was therefore based on the fit of the model to the serum data at the high dose. Iodide is
incorporated into many of the constituents in plasma. However, it is not bound to the plasma
proteins (i.e., albumin) in the same manner as perchlorate. Additionally, the iodide model is
currently simplified to account for the distribution of total iodine. Therefore, the urinary
clearance value (ClUc_/) was determined primarily by fitting the model simulation to the iodide
January 16, 2002
6-64
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
serum data, as blood levels were more dependent on excretion than on the amount of iodide in
other tissues. The clearance of both iodide and perchlorate between the fetal serum and maternal
placenta were based on the fit of the model simulation to the fetal and maternal blood levels and
to the placenta concentration.
6.3.2.3 Pregnant Rat and Fetus Model Parameterization and Validation
This section summarizes how Clewell (2001a) used the various data sets to parameterize
the model and the validation exercises performed.
6.3.2.3.1 Perchlorate Model Parameterization
Clewell (2001a) performed model parameterization for perchlorate using the data obtained
from the AFRL/HEST drinking water studies on GD20. Optimized kinetic parameters (Vmax
and permeability area values) were determined by fitting the model simulation to the
experimental data. As for the adult male rat and human, it was necessary to account for the
serum binding of perchlorate in order to adequately describe the blood perchlorate concentrations
at the lower doses (0.01 and 0.1 mg/kg-day). Figure 6-28 illustrates the importance of binding in
the model simulations of both maternal (A) and fetal (B) serum at 0.01 (left) versus the 10.0
(right) mg/kg-day dose. Binding does not have a noticeable effect on the plasma concentrations
in the highest dose. However, as the perchlorate dose decreases, the effect of binding is more
pronounced. Therefore, at lower levels, a larger percent of the injected dose will be bound.
As the amount consumed is increased, the binding process is saturated and eventually the amount
of perchlorate that is bound is negligible in contrast to the large amount of free perchlorate in the
plasma. This is to be expected because the number of binding sites is limited.
Figure 6-29 shows the fit of the model to the maternal serum (left) and thyroid (right)
perchlorate concentration (mg/L) in the dam on GD20. Since saturation of the symporter occurs
between the 1.0 and 10.0 mg/kg-day dose groups, the influence of Vmaxc in the tissues was
primarily in the 0.01 to 1.0 mg/kg-day doses. Thus, the fit of the model simulation to the data in
the lower three doses was used to determine the values for Vmaxc in the tissues. On the other
hand, the Vmaxc did not have a significant effect on the highest dose. The model fits to the
10 mg/kg-day dose group were primarily affected by the partition coefficients and permeability
area values. Clewell (2001a) obtained the permeability area values in the tissues by fitting the
January 16, 2002
6-65
DRAFT-DO NOT QUOTE OR CITE
-------
0.070
10
i
E
3
b
C/>
a
b
a
5
E
u
G
.Q
1
o.
0.0S6
0042
0028
0 014
0000
withbinding
wiihoutbbidmg
|(N/\/\AAAAAAAAAAA/VVVvA/V
A
106
212 318
Time (hours)
424
530
0.20
-fc 016
E
|
$ 0.12
008
B
witb binding
Iwwwwwwvwi
without binding
212 318
Time (hours)
530
on
a
i_
o
e3
2
.£
B
C
o
.e
a
o
C-
106
wah binding
without binding
212
318
Time (hours)
530
4
£
fc i
C/5 3
*5
o
u.
.5
a 2
G
wiih binding
without binding
212 318
Time (hours)
llllli
1
Figure 6-28. Simulations illustrating the necessity of including plasma binding in the
pregnant dam and fetal rat PBPK model structure (Clewell, 2001a). Model
predictions (lines) versus data time course (mean ± SD) are shown with and
without plasma binding for maternal (A) and fetal (B) serum concentrations
(mg/L) at two different doses, 0.01 mg/kg-day (left) and 10.0 mg/kg-day
(right).
January 16, 2002
6-66
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
m
UN
/vwwwwwwwnM
yWWWWWWWvwV
0 0 134.4 268.8 403 2 537.6 672.0
Time (hours)
fVVVVVVWVVVVVVVVVW
aAAAAAAAAAAAAAAAM
ii
0.0 134.4 268.8 403.2 537 6 672.0
Time (hours)
Figure 6-29. Pregnant dam and fetal rat PBPK model predictions (lines) versus data time
course (mean ± SD) of perchlorate concentrations (mg/L) in maternal serum
(left) and thyroid (right) on GD20 (Clewell, 2001a). Pregnant rats were dosed
in drinking water with 10.0,1.0, 0.1, or 0.01 mg/kg-day perchlorate. Data of
Yu (2000).
highest dose to the 10 mg/kg-day data in the tissues. Maternal placenta, mammary gland, and GI
tract concentrations were available at the 10 mg/kg dose only. These tissues were used to verify
the applicability of the assigned partition coefficients to the model. Since mammary glands were
not available for the 0.01 through 1.0 mg/kg-day dose groups, it was not possible to fit the
Vmaxc value to data at which the symporter has a significant effect. Therefore, the Vmaxc in the
mammary gland was assigned the molar equivalent of the iodide Vmaxc. This is probably a
reasonable value in the non-lactating gland. Clewell (2001a) provides additional figures that
demonstrate the fit of the model to the GI tract, mammary glands, and placenta in the pregnant
dam.
Fewer data were available for perchlorate distribution in the fetus than in the dam due to
the experimental difficulty involved in sampling the small fetal tissues. Figure 6-30 depicts the
model simulation of the fetal serum concentration (mg/L) compared to the data obtained in the
drinking water study. Fetal serum and skin were pooled by litter. Fits to additional
compartments are provided in Clewell (2001a).
January 16, 2002
6-67
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
10
~5a>
E
0)
w
a
0>
u.
G
o
-------
3
-------
1
2
3
4
5
6
7
8
9
10
(Clewell, 2001a). The mammary:plasma ratios of greater than one were fit with the Vmaxc for
mammary NTS.
Clearance values for the transfer of iodide between the placenta and fetal blood were
determined by optimizing the fit of the fetal serum to the data points while maintaining the fit of
the simulations of the maternal blood and fetal tissue data. Figure 6-32 shows the model
simulation versus the fetal data in the preliminary iodide time course study for radiolabeled
iodide in fetal serum (ng/L). Clewell (2001a) shows additional simulations for fetal skin and
fetal GI tract.
2.5 -
J
£ 2.0 -
E
u
on
1
J3 0.5 -
.2
T3
0.0 -J H 1 1 "
478.0 480.2 482.4 484.6 486.8 489.0
Time (hours)
Figure 6-32. Pregnant dam and fetal rat PBPK model predictions (lines) versus data time
course (mean ± SD) of 125I' radiolabeled iodide concentrations (ng/L) in fetal
serum on GD20 after an iv injection to the dam with 2.19 ng/kg ,25I" (Clewell,
2001a). Data of Yu (2002).
January 16, 2002
6-70
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The data of Feldman et al. (1961) were used by Clewell (2001a) to determine the values for
maximum velocity of iodide uptake in the fetal thyroid. An exponential function was fit to the
experimental values and time points where time was gestation in hours. This equation was then
used in the model to account for the increasing ability of the fetal thyroid to incorporate iodide.
Iodide levels were negligible on GDI6 but increased dramatically from GDI 7 to GDI9 (see
Clewell, 2001a; Figure 25).
6.3.3 Model Validation
The Clewell (2001a) model predictions for the inhibition of iodide uptake into the thyroid
and the resulting effect on the maternal and fetal serum was validated against the data collected
by AFRL/HEST during the inhibition study on GD20. The kinetic parameters derived from the
perchlorate drinking water and preliminary iodide data sets were used in the model. Because the
inhibition study was performed with an acute perchlorate dose, it was necessary to make some
slight changes in the parameters describing thyroid perchlorate kinetics. The long-term exposure
to perchlorate in the drinking water studies (18 days) that were used to determine the perchlorate
parameters is sufficient to induce up-regulation in the thyroid (Yu, 2000). Thus, it was
determined that the thyroid parameters in the dam at this point would be different from those
seen in an acute situation. The only parameters altered in order to model the acute perchlorate
were the partition coefficient (from 2.25 to 0.13) and permeability area value (from 6.0E-4 to
4.0E-5) into the thyroid at the basolateral membrane (thyroid follicle). The value for the
partitioning into the follicle in a naive thyroid was calculated as described previously from Chow
and Woodbury (1970). The permeability area value in the naive thyroid follicle was determined
with the lactation model, which is described in another consultative letter describing model
development for the lactating rat (Clewell, 2001b).
The model simulation was fit to the available kinetic data in the thyroid while keeping all
other thyroid parameters identical to those in the pregnancy model. Figure 6-33 illustrates the
model prediction of thyroidal iodide uptake with and without perchlorate inhibition, utilizing
these pre-set parameters. The model prediction of inhibition in the thyroid gland at 0.5, 1., 2, 4,
8, 12, and 24 hours after dosing with iodine shows an excellent fit to the data. The use of
parameters derived from the drinking water perchlorate data for acute iodide uptake kinetics is
well supported by the inhibition of iodide because inhibition is highly dependent on the
January 16, 2002 6-71 DRAFT-DO NOT QUOTE OR CITE
-------
6000
"5b
e
w
72 4800 -
o
£
JS
H
2 3600 -
In
B
(8
s
U
"O
• mm
T3
o
•a
o
"o
x>
J3
.9
T3
£
iv dose perchlorate
iv dose 1251"
perchlorate,.
2400 -
1200
478.0
~i r
483.4 488.8
i r
494.2 499.6
505.0
Time (hours)
Figure 6-33. Validation for pregnant dam and fetal rat PBPK model (Clewell, 2001a).
Model predictions (lines) versus data time course (mean ± SD) of 125I"
radiolabeled iodide concentrations (ng/L) in maternal thyroid with and
without 1.0 mg/kg perchlorate administered by iv injection to the dam 2 hours
prior to an iv injection with 1.87 ng/kg 125I" (Clewell, 2001a). The top
simulation represents the control thyroid and the lower indicates the inhibited
thyroid. Data of Yu (2000, 2002).
1 perchlorate concentration in the thyroid and the perchlorate affinity constants in the apical and
2 basolateral membranes of the thyroid. Figure 6-34 illustrates the effect of perchlorate thyroid
3 inhibition on the maternal (top) and fetal (bottom) blood iodide levels. Significant differences
4 were found in the maternal serum iodide concentrations collected at the 1, 4, and 24 hour time
5 points. Fetal serum, however, did not show any significant differences in the total serum iodide
6 between the control and inhibited groups. Additional statistical analysis of these data are
7 provided as Attachment #2 in Clewell (2001a).
8 Clewell (2001a) performed a model simulation of data presented by Versloot et al. (1997)
9 in order to test the ability of the model to predict diverse data sets collected under different
January 16, 2002
6-72
DRAFT-DO NOT QUOTE OR CITE
-------
20
*£b
c
¥
3
km
u
CO
0>
3
2
.c
0
-o
1
1
0)
-8
.2
-o
16
12 -
478.0
iv dose perchlorate
iv dose 1251'
483.4
488.8 494.2
Time (hours)
499.6
505.0
2.0
.J
~sd
c
u
00
8
t>
u-
.6
V
-a
1.6
1.2 -
| 08
5
« 0.4
0.0
478.0 483.4 488.8 494.2
Time (hours)
499.6 505.0
Figure 6-34. Validation for pregnant dam and fetal rat PBPK model (Clewell, 2001a).
Model predictions (lines) versus data time course (mean ± SD) of 125I'
radiolabeled iodide concentrations (ng/L) in maternal (A) and fetal (B) serum
with and without a 1.0 mg/kg perchlorate dose administered by iv injection to
the dam 2 hours prior to an iv injection with 1.87 ng/kg 12SI" (Clewell, 2001a).
The top simulations in each represents the serum during thyroid inhibition
and the lower represents the control serum. Data of Yu (2000, 2002).
January 16,2002 6-73 DRAFT-DO NOT QUOTE OR CITE
-------
1 conditions. This data set provided an additional time point for the iodide model validation
2 (GDI9). Dams were exposed by iv injection to 10 ^Ci (1.74 ng/kg) carrier-free radiolabeled
3 iodide (125I ) on GDI9. Figure 6-35 shows the model predictions versus data (mean ± SD) for the
4 amount (ng) of iodide taken up in maternal thyroid (A), mammary gland (B), and placenta (C), or
5 fetal thyroid (D). The model is able to accurately describe these tissues of interest and fits other
6 compartments (data shown in Clewell, 2001a) within a two-fold factor without changing any
7 parameters. This illustrates its predictive power and usefulness to the extrapolations required.
5
t
4611 467 1 47)4
Time (houre)
Figure 6-35. Validation for pregnant dam and fetal rat PBPK model (Clewell, 2001a).
Model predictions (lines) versus data time course (mean ± SD) of total 125I~
radiolabeled iodide in the maternal thyroid (A), mammary gland (B),
placenta (C), or fetal thyroid (D) at 24 hours afer exposure to the dam by iv
injection of 10 /id (1.74 ng/kg carrier-free) 125I" in GD19 dams. Data of
Versloot et al. (1997).
January 16, 2002
6-74
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Model predictions were also shown to be in good agreement with another unrelated data
set, that of Sztanyik and Turai (1988), who measured carrier-free radiolabeled iodide (l3lI ) in
GD20 dams and in the total (whole body) fetuses afer an iv injection (Clewell, 2001a). This
validation illustrated adequate model fit to another time point and radioactive species of iodide.
The model was additionally validated against AFRL/HEST data for dams and fetuses after
administration of radiolabeled iodide (!25I") with carrier at doses four orders of magnitude greater
than that used to parameterize the model (33000 ng/kg versus 2.19 ng/kg). These validation
simulations are shown in Clewell (2001a).
As a final validation exercise, the model was used to predict radiolabeled iodide uptake
inhibition after perchlorate exposures in drinking water for 18 days at 0.0, 0.01, 1.0, and 10
mg/kg-day (Yu, 2000). It was apparent that even at the lowest dose, the hormonal system had
experienced a perturbation and was attempting to compensate for the interruption caused by the
perchlorate exposure (Clewell, 2001a). Maternal T4 decreased in a dose-dependent manner,
while TSH increased. The maternal total T4 and TSH changes were statistically significant at all
doses. Free T4 was significantly increased at the 0.1,1.0, and 1.0 mg/kg-day doses and total T3
was significantly decreased at the 1.0 and 10.0 mg/kg-day doses. The fetus appeared to follow
the same trends as those seen in the dam. However, only the 1.0 and 10.0 mg/kg-day dose
groups show significant decreases in total T4 and the 0.01, 1.0, and 10.0 mg/kg-day doses
resulted in significantly increases in fetal free T4 and TSH. No significant decrease was seen in
fetal T3. The statistical analysis of the hormone data is provided as Attachment #3 in Clewell
(2001a).
From the perspective of iodide kinetics, these hormone changes are important indicators of
thyroid up-regulation. When TSH is increased, the thyroid is stimulated to increase iodide
uptake. It is evident, then, that after exposure to perchlorate in drinking water for 18 days, the
thyroid of the pregnant dam has experienced both inhibition and up-regulation and has
successfully compensated for the competition of perchlorate for binding sites of NIS. Therefore,
it is not surprising that no inhibition was reported on GD20. It is not that the inhibition is not
taking place, but rather that the system has compensated for the effect.
None of the models is currently equipped with the capability to account for up-regulation of
the thyroid. Therefore, when a simulation of the inhibition is performed with the model, the
concentration of iodide is under-predicted in a perchlorate-dose dependent manner (Clewell,
January 16, 2002
6-75
DRAFT-DO NOT QUOTE OR CITE
-------
1 2001a). Figure 6-36 shows the model prediction of iodide in the thyroid of the dam at drinking
2 water doses of 0.0, 0.1, 1.0, and 10.0 mg/kg-day. The Vmaxc for iodide was decreased to
3 2.5 x 104 to fit the mean from the control data with the control simulation in order to make the
4 comparison of the inhibition data and simulations clearer. All experimental data were actually
5 taken two hours post dosing. However, the data points were separated slightly by time on the
6 plot in order to make them more visible. The prediction of thyroid perchlorate levels from this
7 same study can be seen in Figure 6-29 (right).
8
9
"Sb
£ 10
I.
.S
a>
T3
¦5
O l.o
M 1
T3
U
U
ja
cq
.2
-a
2
0.1
480.0 480.8 481.6 482.4
Time (hours)
483.2
484.0
Figure 6-36. Validation for pregnant dam and fetal rat PBPK model (Clewell, 2001a).
Model predictions (lines) versus data time course (mean ± SD) of radiolabeled
iodide in the maternal thyroid of the dam at doses of perchlorate in drinking
water at 0.0, 0.01,1.0, and 10.0 mg/kg-day for 18 days. Data of Yu (2000).
January 16, 2002
6-76 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.3.4 Summary
The proposed model for the pregnant rat and fetus developed by Clewell (2001a) appears to
adequately describe perchlorate and iodide distribution in a highly dynamic, changing system, by
accounting for growth with age-specific functions. The model predicts the transfer of perchlorate
to the fetus and is also able to describe the uptake into fetal tissues of interest, such as the serum
and thyroid. Fetal and dam tissues were predicted well by fitting data that spans three orders of
magnitude (i.e., 0.01 to 10.0 mg/kg-day).
In addition to the requisite compartments for pregnancy (mammary gland, placenta, fetus),
some differences exist that affect the kinetics of both perchlorate and iodide. The thyroidal
maximum capacities are lower in the pregnant dam than in the male rat. Model parameterization
in the male rat indicated the need for Vmax values for uptake into the follicle of the thyroid of
2.2 x 103 L/hr-kgr for perchlorate and 5.5 x 104 L/hr-kg for iodide, while the gestation model
required values of 1.8 x 103 L/hr-kg and 4.0 x 104 L/hr-kg for the same parameters. This
difference is supported in the literature. Versloot et al. (1997) suggest that the pregnant rat may
have a lowered reserve of iodide in the thyroid toward the end of pregnancy, causing increased
activity in the thyroid. The increased response of the pregnant rat was also seen in the studies
performed by Yu (2000) and Yu et al. (2000) that reported a greater than average inhibition in the
thyroid of the pregnant dam than in the male rat at the same perchlorate dose (78% vs. 70% over
8 hours). The skin of the pregnant dam also required a smaller value for Vmaxc than the male
rat. This is supported by the work of Brown-Grant and Petes (1959), which reported higher
levels of iodide in the skin male rats than in female rats. Skin, therefore, appears to be a more
important iodide reserve in the male rat than the female. It is reassuring that the model is able to
account for the majority of differences in the uptake, distribution, and excretion between the
male and pregnant female by incorporating known differences in physiology.
Clewell (2001a) notes that at this time the amount of data concerning perchlorate kinetics
in the pregnant rat is very limited. Although perchlorate has been used extensively in literature
to study the thyroidal uptake of iodide, it has not been commonly used in rat gestation studies.
As such, the perchlorate model was limited to utilizing the drinking water studies for
parameterization. However, acute kinetic data were available for perchlorate in the lactating dam
and were utilized in the development of the rat lactation model (Clewell, 2001b; see Section 6.4).
This system is similar to that of the pregnant dam. Consequently, it was possible to simulate the
January 16,2002 6-77 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
perchlorate kinetics of the dam with the same general model structure, changing only the
physiological parameters. Therefore, it seemed reasonable to use the acute perchlorate
parameters from the lactation model. The use of the described parameters for acute perchlorate
kinetics is also supported by the ability of the model to predict inhibition in the pregnant dam.
Clewell (2001a) discusses that acute perchlorate kinetic data to further verify the model are
currently being analyzed by AFRL/HEST, and these were provided to the EPA too late for
evaluation (Yu, 2002). In these studies, tissues were collected from pregnant dams and fetuses at
various time points after iv injections of perchlorate. The use of these data in the modeling effort
may be described in draft manuscripts provided to the external peer review.
The kinetic behavior of iodide was also accurately simulated with a range of doses that
spans nearly five orders of magnitude (0.36 to 33,000 ng/kg). The active sequestration of iodide
in maternal and fetal tissues and the transfer of iodide between mother and fetus was described
kinetically with the model, and data have been simulated at a variety of doses and at various time
points up to 24 hours post exposure. The fact that the model was able to simulate data from
other laboratories under a variety of different conditions attests to the validity of the model
structure and its applicability to other studies. The ability of the model to predict iodide was
indicative of the usefulness of the model for predictive purposes. It was possible to predict
inhibition out to 24 hours while simulating the serum and thyroid perchlorate and iodide levels
with satisfactory accuracy. This provides support for the chosen model structure, as well as
validation for the physiological and chemical descriptions used.
Clewell (2001a) notes that the inability of the model to respond to this auto-regulation
presents a considerable need for further model development since drinking water scenarios
would allow time for the hypothalamic-pituitary-feedback system to upregulate. Given that the
temporal windows of developmental susceptibility are not well established across species, this
issue may have to wait for further fundamental neurodevelopmental research.
The EPA has also become aware of a recent human biokinetic model for iodine and
radionuclides at various ages (fetus, children, mothers) that may provide some additional
information with which to validate the iodide kinetic components of the proposed models from
AFRL/HEST scientists (International Commission on Radiological Protection, 2001, 1989).
January 16, 2002
6-78
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.4 LACTATING AND NEONATAL RAT MODEL STRUCTURE
This section describes the model developed by AFRL/HEST in response to concerns about
interspecies extrapolation of effects observed in laboratory rats immediately after parturition up
to about PND22 (Clewell, 2001b) and updates the preliminary structure provided to EPA
(Clewell, 2000). The model predicts the distribution of perchlorate within the lactating dam and
neonatal rat during these first few weeks of life, and also predicts the short-term effect of acute
perchlorate exposure on iodide kinetics, including iodide uptake in the maternal thyroid.
Concern regarding the kinetics of perchlorate in lactating dams and neonates was motivated
by the knowledge that the mammary gland is another tissue with active transport via the NIS, as
described in Section 6.3. Perchlorate can thus competitively inhibit the uptake of iodide into the
mammary gland in a manner reminiscent of the thyroid, and reduce the amount of available
iodide to the infant. Studies utilizing radiolabeled iodide in lactating rats have shown perchlorate
to be an effective inhibitor of iodide secretion of into breast milk (Potter et al., 1959, Brown-
Grant, 1961). The fact that perchlorate not only inhibits the uptake of iodide, but is also taken up
itself into the mammary tissue by way of the NIS, results in an additional risk to the neonate.
The perchlorate is then concentrated in the milk and transferred to the litter through suckling.
Although early papers suggest that perchlorate is not transferred in milk (Zeghal et al.,
1992), newer technology with better analytical sensitivity has detected perchlorate in the milk of
rats dosed with as little as 0.01 mg/kg-day perchlorate in drinking water at the AFRL/HEST. The
perchlorate levels in 5- and 10-day old neonate serum are comparable to those of the mother (Yu
et al., 2000), indicating that the pups are in fact exposed to significant levels of perchlorate
through the maternal milk. This information highlighted the need for more information
regarding the effect of perchlorate exposure on the neonate.
The model structure is shown in Figure 6-37. Table 6-6 provides the physiological
parameters used in the lactating and neonatal rat PBPK models. Table 6-7 provides the
perchlorate-specific parameters, and Table 6-8 provides the iodide-specific parameters for each.
The model structure was developed to be consistent with the previously discussed
structures for the adult male rat, pregnant rat, and fetus. In fact, an important linking to the
pregnancy model was required. Since the experimental data used to develop the lactation model
were taken from drinking water studies in which the dosing began on GD2, it was necessary to
include initial perchlorate concentrations in the tissues at the time of birth (0 hours). In order to
January 16,2002 6-79 DRAFT-DO NOT QUOTE OR CITE
-------
Lactating Dam
Neonate
I vc
\
pDose_
bound CIO,,'
Plasma
I A lxcc_CIOL
" RBCs
QC
1
Stomach
11 A
Contents ^
ITI
Stomach I
ll A
..Tissue
|TI
Stomach
|
Blood
_ __.T_
Li vcr
Richly
Perfused
1.
Kidncv
|±A Milk
1*1 Mammary
I J_4 Ti_s_sjic _ _ i
!*' Mammarv
1 Blood '
QM
Colloid
Thyroid 1
14... Fo.llic.le__ ~
T! SLroma I
Skin
~ I
Skin
Blood
t.
OT
OSK
Fat
Slowly
Perfused
.Urine
Ktrans
Unnc
(lo Stomach
Contents of
Dam)
bound ClOy
Plasma
Ij A tjcc.CJO.
|TI RBCs
QC
1
Stomach
11 A
Contents ^
jTI
Stomach 1
11 A
..ItfiSJJC
|TI
Stomach
|
Blood
QO
Liver
Richly
Perfused
QC,
OI.
QR
|A Colloid a
Thyroid 1
IA Follicle ~
^ Stroma 1
Skin
TtT-
Skin
Blood
Fat
Slowly
Perfused
QSK
QF
QS
Figure 6-37. Schematic for the lactating dam and neonatal rat PBPK model of perchlorate
and iodide distribution (Clewell, 2001b). Boxes represent specific
compartments in the model structure. The thyroid consists of the stroma, the
follicle, and the colloid, and the stomach consists of the capillary bed,
stomach wall, and contents. The skin contains two subcompartments
representing the capillary bed and skin tissue. Bold arrows indicate active
uptake at NIS sites in the thyroid, skin, mammary gland and GI tract.
Plasma binding was also described with Michaelis-Menten terms for the
association of perchlorate anion to binding sites with first-order clearance
rates for dissociation. Sequestration of the perchlorate (C104~) and iodide (I")
anions into milk was also described with Michaelis-Menten kinetics.
Permeability area cross products and partition coefficients were used to
describe the first-order movement of the perchlorate (C104~) anion into
deeper subcompartments which results from the inherent electrochemical
gradient within the tissues. Urinary clearance and transfer of the anions
through suckling were represented by first order clearance rates.
January 16, 2002
6-80
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-6. PHYSIOLOGICAL PARAMETERS FOR LACTATING DAM AND
NEONATE PBPK MODEL (Clewell, 2001b)
Physiological Parameters
Lactation
Tissue Volumes"
Dam
Neonate
Source
Body Weight B W (kg)
0.277 -0.310
0.0075 -0.1985
Yu, 2000
Slowly Perfused VSc (%BW)
37.07-40.42
53.92-49.31
Brown et al., 1997
Richly Perfused VRc (%BW)
5.35
5.36
Brown et al., 1997
Fat VFc (%BW)
12.45-6.9
0.0-4.61
Naismith et al., 1982
Kidney VKc (%BW)
1.7
1.7
Brown et al., 1997
Liver VLc (%BW)
3.4
3.4
Brown et al., 1997
Stomach Tissue VGc (%BW)
0.54
0.54
male rat C104" kinetics
Gastric Juice VGJc (%BW)
1.68
1.68
Yu, 2000
Stomach Blood VGBc (%VG)
2.9
2.9
Altman & Dittmer, 1971
Skin Tissue VSkc (%BW)
19.0
19.0
Brown et al., 1997
Skin Blood VSkBc (%VSkc)
2.0
2.0
Brown et al., 1997
Thyroid Total VTtotc (%BW)
0.0105
0.0125
Malendowicz & Bednarek, 1986;
Florsheim et al., 1966
Thyroid Follicle VTc (%Vttot)
45.89
37.2
Malendowicz & Bednarek, 1986;
Conde et al.,1991
Thyroid Colloid VDTc
(%VTtot)
45
13.8
Malendowicz & Bednarek, 1986;
Conde et al.,1991
Thyroid Blood VTBc (%VTtot)
9.1
49.0
Malendowicz & Bednarek, 1986;
Conde et al.,1991
Plasma VPlasc (%BW)
4.7
4.7
Brown et al., 1997, Altman & Dittmer, 1971
Red Blood Cells VRBCc (%BW)
2.74
2.74
Brown et al., 1997, Altman & Dittmer, 1971
Mammary Tissue VMc (%BW)
4.4 - 6.6
—
Knight et al., 1984
Mammary Blood VMBc (%VM)
18.1
—
Assume same % as Thyroid Blood
Milk VMk (L)
0.002
—
Fisher et al., 1990
Blood Flows
Cardiac Output QCc (L/hr-kg)
14.0-21.0
14.0
Hanwell & Linzell, 1973; Brown et al., 1997
Slowly Perfused QSc (%QC)
7.9-1.9
16.9
Brown et al., 1997
Richly Perfused QRc (%QC)
40.8
40.8
Brown et al., 1997
Fat QFc (%QC)
7.0
7.0
Brown et al., 1997
Kidney QKc (%QC)
14.0
14.0
Brown et al., 1997
Liver QLc (%QC)
18.0
18.0
Brown et al., 1997
GI QGc (%QC)
1.61
1.61
Brown et al., 1997
Skin QSkc (%QC)
0.058
0.058
Brown et al., 1997
Thyroid QTc (%QC)
1.6
1.6
Brown et al., 1997
Mammary QMc (%QC)
9.0-15.0
—
Hanwell & Linzell, 1973
"For calculation of volumes from body weight, a density of 1.0 g/mL was assumed.
January 16, 2002 6-81 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-7. PERCHLORATE-SPECIFIC PARAMETERS FOR LACTATING DAM AND
NEONATE PBPK MODEL (Clewell. 2001 b)a
Pcrchlorate Parameters
Lactation Values
Partition Coefficients (unitlcss)
Dam
Neonate
Source
Slowly Perfused/Plasma PS_
0.31
0.31
Yu et al., 2000
Rapidly Perfused/Plasma PR_
0.56
0.56
Yu ct al., 2000
Fat/ Plasma PF_
0.05
0.05
Pena et al., 1976
Kidney/ Plasma PK_
0.99
0.99
Yu et al., 2000
Liver/Plasma PL_
0.56
0.56
Yu et al., 2000
Gastric Tissue/Gastric Blood PG_
1.80
3.21
Yu, 2000; Yu et al., 2000
Gastric Juice/Gastric Tissue PGJ_
2.30
5.64
Yu, 2000; Yu et al., 2000
Skin Tissue/Skin Blood PSk_
1.15
1.15
Yu et al., 2000
Thyroid Tissue/Thyroid Blood PT_
0.13/2.0
0.13/2.0
Chow and Woodbury, 1970;
Yu, 2000"
Thyroid Lumen/Thyroid Tissue PDT_
7.0
7.0
Chow and Woodbury, 1970;
Yu, 2000
Red Blood Cells/Plasma PRBC_
0.73
0.73
Yu et al, 2000
Mammary Tissue/Mammary Blood PM_
0.66
—
Anbar et al, 1959
Milk/Mammary Tissue PMk_
2.39
—
Yu, 2000
Max Capacity, Vmaxc (ng/hr-kg BW)
Thyroid Follicle Vmaxc_T
1.50E+03
1.50E+03
Fitted'
Thyroid Colloid Vmaxc_DT
1.00E+04
1.00E+04
Fitted"
Skin Vmaxc_S
8.00E+05
8.00E+05
Fitted
Gut Vmaxc_G
1.00E+06
1.00+06
Fitted
Mammary Tissue Vmaxc_M
2.0E+5/2.0E+4
—
Fitted1"
Milk Vmaxc_Mk
2.00E+04
Fitted
Plasma Binding Vmaxc_B
9.00E+03
1.00E+03
Fitted
Affinity Constants, Km (ng/L)
Thyroid Follicle Km_T
1.00E+05
1.00E+05
Gluzman & Niepomniszcze, 1983;
Wolff, 1998
Thyroid Colloid Km_DT
1.0E+09
1.0E+09
Golstein et al, 1992;
Wolff, 1998
Skin Km_S
1 .OOE+05
1.00E+05
Gluzman & Niepomniszcze, 1983;
Wolff, 1998
Gut Km_G
1.00E+05
1.OOE+05
Gluzman & Niepomniszcze, 1983;
Wolff, 1998
Mammary Km_M
1.0E+05
—
Gluzman & Niepomniszcze, 1983;
Wolff, 1998
Milk Km_Mk
1.00E+06
Fitted
Plasma Binding Km_B
1.00E+04
1.0 01:-04
Fitted
January 16, 2002
6-82
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
TABLE 6-7 (cont'd). PERCHLORATE-SPECIFIC PARAMETERS FOR LACTATING
DAM AND NEONATE PBPK MODEL (Clewell, 2001b)a
Lactation Values
Perchlorate Parameters
Dam
Neonate
Source
Permeability Area Cross Products, (L/hr-kg)
Gastric Blood to Tissue PAGc_
1.00
1.00
Fitted
Gastric Tissue to Juice PAGJc_
1.00
1.00
Fitted
Thyroid Blood to Tissue PATc_
4.0E-05/6.0E-04
4.0E-05/6.0E-05
Fitted111
Thyroid Tissue to Colloid PADTc_
0.01
0.01
Fitted
Skin Blood to Tissue PASkc_
0.50
1.00
Fitted
Mammary Blood to Tissue PAMc_
0.01
—
Fitted
Mammary Tissue to Milk PAMkc_
0.001/1.0
Fitted
Plasma to Red Blood Cells PRBCc_
1.00
1.00
Fitted
Clearance Values, (L/hr-kg)
Urinary excretion CLUc_
0.07
0.005
Fitted
Dissociation from Binding Sites Clunbc_
0.034
0.034
Fitted
Transfer from Milk to Pup Ktransc
6.4E-04/1.04E-03
6.4E-04/1.04E-03
Sampson & Jansen, 1984
"All parameters listed are notated in the model either by an /' (for iodide) or p (for perchlorate) following an underscore
in the parameter name (e.g., PR_/, PR_p, Vmaxc_T/, Vmaxc_Tp, etc.).
''Neonate was given maternal values for Vmax (scaled by body weight) in the absence of data.
cParameters with two values indicate acute and drinking water parameters, respectively.
obtain these initial values for tissue loading at birth, the pregnancy model had to include all of
the compartments contained in the lactation model (Clewell, 2001a). The pregnancy model was
then allowed to run until the day of birth (GD22), and the average tissue concentrations of
perchlorate or iodide for the final day of gestation were used as the starting values for the
respective tissues in the lactation model (Clewell, 2001b).
As discussed, the mammary tissue has been shown to concentrate both perchlorate and
iodide during lactation via the NIS symporter. Additionally, hormones produced during lactation
such as prolactin which stimulates milk production, have been shown to regulate the mammary
NIS. Suckling of the neonatal rats has also been shown to stimulate mammary NIS activity
(Tazebay et al., 2000). An additional symporter has been identified in the experiments of
Shennan (2001). In vitro studies of iodide transport into the mammary gland and the resulting
efflux of sulfate from the cells in the absence of sodium cation (Na+), indicates that another form
of transport exists for iodide in the mammary gland in addition to the NIS. Shennan suggests
January 16, 2002
6-83
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-8. IODIDE-SPECIFIC PARAMETERS FOR LACTATING DAM AND
NEONATE PBPK MODEL (Clewell, 2001bV
Iodide Parameters
Lactation Values
Partition Coefficients (unitless)
Dam
Neonate
Source
Slowly Perfused/Plasma PS_
021
021
Halmi et al , 1956
Rapidly Perfused/Plasma PR_
0.40
0.40
Halmi et al., 1956
Fat/Plasma PF_
0.05
0.05
Pena et al., 1976
Kidney/Plasma PK._
1.09
1.09
Perlman et al., 1941
Liver/Plasma PL_
0.44
0.44
Perlman et al., 1941
Gastric Tissue/Gastric Blood PG_
1.00
1.00
Unpublished Lactation Inhibition Study
Gastric Juice/Gastric Tissue PGJ_
1 00
3.50
Unpublished Lactation Inhibition Study
Skin Tissue/Skin Blood PSk_
0 70
0 70
Perlman et al., 1941
Thyroid Tissue/Thyroid Blood PT_
0.15
0.15
Chow and Woodbury, 1970
Thyroid Lumen/Thyroid Tissue PDT_
7.00
7.00
Chow and Woodbury, 1970
Red Blood Cells/Plasma
1.00
1.00
Rail et al., 1950
Mammary Tissue/Mammary Blood PM_
0.66
—
Anbar et al., 1959
Milk/Mammary Tissue PMk_
4.00
—
Yu, 2000
Max Capacity, Vmaxc (ng/hr-kg BW)
Thyroid Follicle Vmaxc_T
4.00E+04
4.00E+04
Fitted"
Thyroid Colloid Vmaxc_DT
6.00E+07
6.00E+07
Fitted''
Skin Vmaxc_S
6.00E+04
2.50E+05
Fitted
Gut Vmaxc_G
1.00E+06
2.00E+05
Fitted
Mammary Tissue Vmaxc_M
8.00E+05
—
Fitted
Milk Vmaxc_Mk
5.00E+06
—
Fitted
Affinity Constants, Km (ng/L)
Thyroid Follicle Kjti_T
4 00E+O6
4.00E+06
Gluzman and Niepomniszcze, 1983
Thyroid Colloid Km_DT
1.00E+09
1.00E+O9
Golstein et al., 1992
Skin Kjti_S
4.00E+06
4.00E+06
Gluzman and Niepomniszcze, 1983
Gut Km_G
4.00E+06
4.00E+06
Gluzman and Niepomniszcze, 1983
Mammary Km_M
4.00E+06
—
Gluzman and Niepomniszcze, 1983
Milk Km_Mk
1.00E+06
—
Fitted
Permeability Area Cross Products, (L/hr-kg)
Gastnc Blood to Gastric Tissue PAGc_
0.80
0.05
Fitted
Gastric Tissue to Gastric Juice PAGJc_
0.60
0 06
Fitted
Thyroid Blood to Thyroid Tissue PATc_
1.00E-O4
1 00E-O4
Fitted"
Thyroid Tissue to Thyroid Colloid PADTc_
1.00E-04
1.00E-04
Fined"
Skin Blood to Skin Tissue PASkc_
0 50
0.02
Fitted
Mammary Blood to Tissue PAMc_
0.02
—
Fitted
Mammary Tissue to Milk PAMkc_
1.00
—
Fitted
Plasma to Red Blood Cells PRBCc_
1.00
1.00
Fitted
Clearance Values, (L/hr-kg)
Urinary excretion CLUc_
0.03
0 02
Fitted
Transfer from Milk to Pup Ktransc
6.4E-04 -
1 04E-03
Sampson & Jansen, 1984
"All parameters listed are notated in the model either by an i (for iodide) or p (for perchlorate) following an underscore in the
parameter name (e.g., PR_i, PR_p, Vmaxc_Ti, Vmaxc_Tp, etc.).
'Neonate was given maternal values for Vmax (scaled by body weight) in the absence of data.
January 16, 2002
6-84
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
that this anion transport mechanism is able to transfer perchlorate and iodide into the secretory
cells against a concentration gradient. Since the secretory cells are responsible for secreting their
contents into the milk, the anion transport mechanism was included in the milk compartment of
the Clewell (2001b) model.
The structure of the Clewell (2001b) neonatal model is similar to that of the pregnant and
fetal rat model, with the exception of the mammary gland compartment as will be described in
6.4.2.1.1. In order to simplify the model, all neonates from a single litter were combined in the
structure of the model, essentially viewing the entire litter as one entity, or one large neonate.
The dose to the neonate is based on the transfer of perchlorate from the maternal milk to the GI
contents of the neonate rather than through direct exposure to the drinking water. The 60% of
urinary excretion of the neonate is then entered back into the GI contents of the dam in order to
account for maternal ingestion of the pup's urine during cleaning, based on the work of Samuel
and Caputa (1965).
The same challenge posed by the pregnancy model (i.e., to describe perchlorate and iodide
distribution in a highly dynamic system) was the objective of the lactating and neonatal rat model
(Clewell, 2001b). In addition to total body weight changes in the dam and neonate, maternal
mammary tissue and blood flow, cardiac output, fractional body fat and neonatal body weight,
and fractional body fat change with respect to time. All tissue volume and blood flow values
were adjusted with respect to the changing parameters.
Clewell (2001b) assumed the neonate to be nursing at a constant rate, 24 hours a day. This
assumption is based on the fact that young nursing rats are unable to go for long periods of time
without suckling. The loss through suckling was then described with a first order clearance rate
from the mother's milk to the gastric juice of the neonate, based on the experiments of Sampson
and Jansen (1984). The milk production rate was assumed to be equal to the amount of milk
ingested by the litter.
6.4.1 Data and Methods
This section summarizes the data that Clewell (2001b) used for development and validation
of the lactating and neonatal rat model structures. Details on experimental methods, including:
protocol design, exposure regimen, chemical source and purity, animals (housing, feeding,
January 16, 2002
6-85
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
surgical procedures, etc.), and the analytical methods can be found in the consultative letter and
associated reports from AFRL/HEST or papers cited therein.
6.4.1.1 AFRL/HEST Experiments in Laboratory Rats
These studies are described in the consultative letter and reports of Clewell (2001b), Yu
(2000, 2002), Yu et al. (2000), and Mahle (2000; 2001).
6.4.1.1.1 Drinking Water Study
Perchlorate drinking water experiments used in development of the Clewell (2001b) model
included this study in which pregnant Sprague-Dawley dams were exposed to drinking water
treated with perchlorate from GD 2 through PND5 or PND10 at perchlorate doses of 0.0, 0.01,
0.1, 1.0, and 10.0 mg/kg-day. GD0 was determined by the presence of a vaginal plug. Litters
were standardized to eight pups (four male and four female, when possible) on PND2. Dams and
their litters were euthanized on either PND5 or PND10; maternal and neonatal serum was
analyzed for fT4, tT4, T3, and TSH. Maternal serum, thyroid, skin, and gastric contents were
analyzed for perchlorate at all doses. Neonatal serum, skin, and GI contents were also analyzed
for perchlorate at all doses. Milk was analyzed only on PND10 at all doses. Perchlorate analysis
was performed only on maternal gastric tract, mammary tissue, and neonatal gastric tract samples
from the PND5 study at the 10.0 mg/kg-day dose. Two hours before euthanization, the dams
were given iv doses of 33 mg/kg radiolabeled iodide ( 125I") with carrier. Tissue concentrations of
iodide were measured in order to determine the inhibition in the various tissues after long-term
exposure to perchlorate. This study is described in detail in the consultative letter (Yu, 2000).
6.4.1.1.2 Cross-fostering Study
The cross-fostering study involved four groups of rats with varied experimental conditions:
true control, control, exposed, and true exposed. True control rats were never dosed with
perchlorate. Neonates remained with the dam after birth. In the control group, dams were never
exposed to perchlorate in drinking water. However, at the time of birth, the neonates were
replaced with pups (less than 24 hours old) that had been exposed to perchlorate throughout
gestation (1.0 mg/kg-day to mother through drinking water). In the exposed group, the dams
were dosed with 1.0 mg/kg-day perchlorate in drinking water from GD2 to PND10. At the time
January 16, 2002
6-86
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of birth, the neonates were replaced with pups (less than 24 hours old) that had never been
exposed to perchlorate. The true exposed dams were dosed with 1.0 mg/kg-day perchlorate from
GD2 to PND10. Neonates remained with their mother after birth. All dams and pups were
euthanized on PND10. The skin, GI contents, and serum from the neonates and dam were
analyzed for perchlorate. Results indicated that both true control and control (exposed neonates
with control dams) showed no perchlorate present on PND10. True exposed and exposed
(exposed dams with control litters) showed comparable perchlorate levels on PND10. This study
is described in detail in the consultative letters (Mahle, 2000; 2001).
6.4.1.1.3 Perchlorate Kinetics Study
In order to evaluate the acute kinetics of perchlorate in the lactating dam and neonate,
AFRL/HEST performed a study of the kinetic behavior of perchlorate after the administration of
an acute dose. PND10 Sprague-Dawley dams were given 0.1 mg/kg perchlorate by tail-vein
injection. The dams were left with their neonates until the time of euthanization at 0.5, 1, 2, 4, 8,
or 12 hours post-dosing. Maternal serum, thyroid, stomach contents, skin, and mammary gland
were collected and analyzed for perchlorate content at all time points. Neonate serum, stomach
contents, and skin were also collected for perchlorate analysis at all time points. Fat, liver,
kidney and bladder tissues were also collected from the dam at the eight hour time point.
Perchlorate analysis was performed on the serum of the dam and neonates and the maternal
thyroid, mammary gland, GI contents, and skin.
6.4.1.1.4 Iodide Inhibition Kinetics Study
A study of iodide time course and inhibition kinetics was performed by AFRL/HEST in
which Sprague-Dawley timed-pregnant dams were given 1.0 mg/kg body weight perchlorate via
tail-vein injection on PND10. The perchlorate dose was followed at two hours post-dosing with
a tail-vein injection of carrier free radiolabeled iodide (125I ) at an average dose of 2.10 ng/kg.
Dams (n=6) were euthanized after 0.5, 1, 2, 4, 8, and 24 hours. Maternal and neonatal serum,
skin, GI contents and tract, as well as the maternal thyroid and mammary gland tissue, were
collected and analyzed for total iodide content at each time point. Neonatal serum was pooled by
sex in each litter. Neonatal skin and GI contents and tract were analyzed individually.
January 16, 2002
6-87
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.4.1.2 Data Published in the Literature
Data available in the literature and used in development and validation of the model are
described briefly in this section.
6.4.1.2.1 Sztanyik and Turai, 1988
Five groups of CFY albino rats (BW = 200 to 250 g) were dosed ip with either 370 kBq
(0.081 ng) or 740 kBq (1.61 ng) carrier-free radiolabeled iodide (131I ) on PND1 (after 24 hours).
Sztanyik and Turai measured the total iodide burden of each litter at 29 hours and on PNDs 2, 5,
7, 9, and 14. Since the litters were not standardized, the number of pups in each litter varied.
6.4.1.2.2 Potter et al., 1959
Four dams of the Long-Evans strain (PND 17-18) were dosed ip with 500 /J,Ci of carrier-
free radiolabeled iodide (13!I ). Iodide uptake was measured in the milk and plasma of the dam 3,
6, and 24 hours postdosing and in the maternal thyroids 24 hours postdosing.
6.4.2 Lactating and Neonatal Rat Model Development
This section summarizes only the key features that were different than the preceding model
structures described in Sections 6.2 and 6.3.
6.4.2.1 Physiological Parameters and Partition Coefficients
Maternal parameters were scaled allometrically based on body weight as previously
described for the male rat. Neonatal values were scaled in the same manner as the maternal
parameters. However, since the model actually represents several neonates, it was necessary to
scale the values for the individual pup first, then to adjust for the total number of pups in the
litter as was done in an analogous fashion as for the fetuses in the pregnant rat model (Clewell,
2001 a,b).
6.4.2.1.1 Maternal Tissues
During lactation, the mammary gland grows in response to the increased need for milk
production by the growing neonates. Knight et al. (1984) measured the mammary gland on
several days during lactation. They found the mammary tissue to be 4.4, 5.6, 6.3, and 6.6% of
January 16,2002 6-88 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
the maternal body weight on days 2, 7, 14, and 21, respectively. The residual milk was assumed
to be 0.002 L based on the model of Fisher et al. (1990). Naismith et al. (1982) examined the
change in body fat content of the lactating rat. They reported values for the volume of maternal
body fat of 15.2 and 6.9% of the body weight on PND 2 and 16, respectively. The body fat
composition of the dam on PND1 was calculated to be 12.4% from the PBPK model for
perchlorate and iodide kinetics in the pregnant rat model described in Section 6.3 (Clewell,
2001a).
In order to describe the changes in the physiology of the lactating rat, it was not sufficient
to simply scale some of the parameters allometrically. As opposed to the typical growth
scenario, some of the tissues in the lactating rat cannot be assumed to increase at the same rate in
this dynamic system. Rather, a few tissues, such as the mammary gland and fat, are changing at
an accelerated rate in comparison to the other organs. These parameters required additional
descriptions for their growth beyond the previously described allometric scaling by body weight.
Clewell (2001b) based the approach to modeling these changing parameters on the work of
Fisher et al. (1990) with trichloroethylene.
Additionally, the thyroid of the female rat was found by investigators to be significantly
larger than that of the male rat (Malendowicz and Bednarek, 1986). Clewell (2001b) assigned
values to these parameters based on these data and relevant to the gender and condition (i.e.,
lactation) of the animal. A value of 1.05% of the maternal body weight was used for the thyroid
in the lactation model. The volume fractions of the colloid, follicle, and stroma were given
values of 45, 46, and 9% of the thyroid volume. These are significantly different from the values
given for the male rat. The volume of the colloid in particular is much greater in the female than
the male rat (46 vs. 24% of the thyroid volume). Parameters that were not available specifically
for the female were described by adjusting the values for the male rat by body weight.
In the PND 10 drinking water study performed by AFRL/HEST (see Attachment #2;
Clewell, 2001b), the body weight of the dam showed an average increase of 12% between PND1
and PND 10, but did not show a significant difference in weight between dose groups. As a
result, Clewell (2001b) calculated the average body weight of the dams for all dose groups for
each day of the study and then programmed this changing body weight into the model as a table
function.
January 16, 2002
6-89
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.4.2.1.2 Neonatal Tissues
As for the lactating rats, Clewell (2001b) programmed the overall average body weights of
the neonates measured on PNDs 3, 5, 7, 9 and 10 into the model is a table function, in order to
estimate growth. Naismith et al. (1982) reported the body fat in the pup at PND2 and PND16 to
be 0.167 and 3.65% of the neonatal body weight. The amount of body fat in a 41-day old rat was
given in Brown et al. (1997) as 4.61% of the body weight.
The volume of the thyroid was studied by Florsheim et al. (1966). The volume of the
thyroid was found to increase in a fairly linear relationship with body weight between PND1 and
PND22. These investigators reported thyroid volumes of 0.0125, 0.0146, 0.0120, 0.0137,
0.0130, 0.0130, and 0.0131% body weight for neonates on PND1 through 5, 7, and 11. These
values were used in a table function in the model to describe the growth of the neonatal thyroid
(Clewell, 2001b). The histometry of the thyroid in the neonate was examined by Conde et al.
(1991). The authors found a significant difference between the volume fractions of the colloid,
follicle and stroma in the neonatal rat versus those in the adult. The reported values of 18.3,
61.4, and 20.3% thyroid volume were used to describe the colloid, follicle, and stroma fractions
in the neonatal rat (Clewell, 2001b).
The suckling rate of the neonatal rat has been examined in more than one literature study
and has been shown to change over time in response to the growth of the neonatal rats. As the
pups grow, they require larger amounts of milk. Sampson and Jansen (1984) measured the
amount of milk suckled in rats by removing neonates from the dams for two hours and then
allowing the pups to suckle for two hours. This process was repeated throughout the day on
several days of lactation. By assuming that the weight gained by the neonates during the suckling
period was due to the milk intake and the weight lost while separated from the dam was through
excretion, Sampson and Jansen were able to develop an equation that describes the suckling rate
of the neonatal rat. Since this equation is dependent on the body weight and growth rate of the
neonates, it is able to account for the change over time and the difference between strains and
studies. The equation was used in the Clewell (2001b) model which assumed the milk yield of
the dam was equal to the suckling rate of the neonate.
January 16, 2002
6-90
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.4.2.1.3 Blood Flows
All maternal and neonatal blood flows that were not directly affected by the changes
induced by lactation were scaled by weight from the adult male rat parameters. For those blood
flow parameters that change in response to lactation, some additional description was required
(Clewell, 2001b). Cardiac output has been shown to increase during lactation (Hanwell and
Linzell, 1973). The values given by Hanwell and Linzell (1973) of 14.0, 18.6, 19.0, and
21.0 L/hr-kg for days 3, 8, 13, and 23 of lactation were used in the model as a table function to
describe the change in cardiac output over time (Clewell, 2001b). Additionally, the blood flow
to the mammary tissue was also found to increase during lactation. Reported fractional blood
flows to the mammary tissue of 9, 10, 11, 14, 14, and 15% of the cardiac output on PNDs 1,5,
10, 15, 17, and 21, again from Hanwell and Linzell (1973), were used.
6.4.2.2 Chemical-Specific Parameters
The various active transport processes, tissue permeabilities and clearance rate (excretion)
are described in PBPK models for each species on a chemical-specific basis. This section
outlines how the values for perchlorate and iodide used in the lactating and neonatal rat model
were derived. The values can be found in Tables 6-7 and 6-8. Details on the derivation can be
found in Clewell (2001b).
6.4.2.2.1 Affinity Constants and Maximum Velocities for Active Uptake Processes
Whenever possible, chemical specific parameters were kept the same in human and in
male, female, neonatal, and fetal rats. However, it was necessary to change a few of the
parameters, including the maximum velocity capacity (Vmaxc). The Km values were similar
between tissues and between female and male rat and human models. However, the maximum
velocity capacity differs between tissues (Wolff and Maurey, 1961). Since values for the tissue
maximum velocity capacity for perchlorate (Vmaxc-/?) were not given in literature, the values
were estimated with the model. In order to determine Vmax with the model, the simulation for
the tissue of interest was compared to various data sets with several different perchlorate dose
levels. The value for Vmaxc within a given compartment was then determined by the best fit of
the simulation to the data.
January 16,2002
6-91
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.4.2.2.2 Effective Partitions, Permeability Area Cross Products and Clearance Values
Anbar et al. (1959) measured the mammary gland:blood ratios in the rat four hours after an
intra-peritoneal injection of 100 mg radiolabeled perchlorate (36C104~) as potassium perchlorate.
They reported an effective partition of 0.66 for the rat mammary gland. Clewell (2001b) used
this value in the model. Since the partition for iodide into the mammary gland was not available
in the literature, Clewell (2001b) assigned the same effective partition coefficient as used for
perchlorate.
When available, iodide partition coefficients were calculated from the tissue:blood ratios
measured during the clearance phase of iodide data in the tissue of interest. For example, GI
tract and contents were determined from the clearance portion of the data from the iodide kinetic
study in the lactating rat.
For tissues in which a clearance was described (urinary clearance and dissociation of
perchlorate from the binding sites), a clearance value was determined by fitting the model
simulation to the appropriate tissue data. Since perchlorate is quickly excreted in urine and
binding has little effect on serum levels at high doses, the simulation for the 10 mg/kg-day dose
group was primarily dependent on the urinary clearance value (ClUc_p). The urinary clearance
value for perchlorate was therefore based on the fit of the model to the serum data at the high
dose. The value obtained in this manner was similar to that determined by fitting the male rat
PBPK simulation to urinary perchlorate at several doses (Merrill, 2001a) and to the high dose in
the pregnant rat (Clewell, 2001a). The rate of dissociation of perchlorate from the binding sites
was fit to the serum data across doses.
6.4.2.3 Lactating Rat and Neonate Model Parameterization and Validation
This section summarizes how Clewell (2001b) used the various data sets to parameterize
the model and how the validation exercises were performed.
6.4.2.3.1 Perchlorate Model Parameterization
Clewell (2001b) performed model parameterization for perchlorate using the data obtained
for the tissues from the AFRL/HEST drinking water studies on PND5 and PND10. Optimized
kinetic parameters (Vmaxc and permeability area) were determined by visually fitting the model
simulation to the experimental data. As for the previous model structures (adult male rat, human,
January 16, 2002
6-92
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
1
2
3
4
5
6
7
8
pregnant rat and fetus), it was necessary to account for the serum binding of perchlorate in order
to adequately describe the serum perchlorate concentrations at the lower doses (0.01 and
0.1 mg/kg-day). Figure 6-38 illustrates the importance of binding in the model simulations in the
dam on these days.
" Time (hours) D Time (hours)
Figure 6-38. Simulations illustrating the necessity of including plasma binding in the
lactating dam and neonatal rat PBPK model structure (Clewell, 2001b).
Model predictions (lines) versus data time course (mean ± SD) for maternal
serum perchlorate concentrations (mg/L) on PND5 and PND10 at doses to the
dam of 10.0,1.0, 0.1, and 0.01 mg/kg-day are shown with (A) and without
(B) plasma binding.
Figure 6-39 shows the perchlorate tissue concentrations (mg/L) in the lactating dam thyroid
(A) and in maternal milk (B) at PND5 and PND10 for the 0.01, 0.1, 1.0 and 10.0 mg/kg-day
doses. It was noticed that during the drinking water studies, the daily dose to the dams varied
somewhat due to their changing water intake. Therefore, all of the model simulations of the
drinking water studies reflect the actual daily dose to the dam, which Clewell (2001b) calculated
from the daily water consumption and body weight measurements.
Figure 6-40 shows the model simulations of the male and female neonate plasma levels
compared to the data obtained in the AFRL/HEST drinking water study. Plasma concentrations
January 16,2002 6-93 DRAFT-DO NOT QUOTE OR CITE
-------
£
£
*T3
I
e
s
.a
too
10
0.1
V'ViV'VrV>vrVV^V
A
64 128 192 2S6 320
Time (hours)
10.0
"a.
£
=fi
!£ 1.0
.s
0.1
0.01
|,A,|'/\/A\/AvAvA\/Av/[vA'4
/VVvVvVVVyv
F
B
r
—i—i—i i—
64 128 192 236 320
Time (hours)
Figure 6-39. Lactating dam and neonatal rat PBPK model predictions (lines) versus data
time course (mean ± SD) of perchlorate concentrations (mg/L) in the
maternal thyroid (A) and milk (B) on PND5 and PND10 at doses in drinking
water to the dam of 10.0,1.0, 0.1, or 0.01 mg/kg-day perchlorate (Clewell,
2001b).
Figure 6-40. Lactating dam and neonatal rat PBPK model predictions (lines) versus data
time course (mean ± SD) of perchlorate concentrations (mg/L) in the serum of
male (A) and female (B) neonates on PND5 and PND10 at doses in drinking
water to the dam of 10.0,1.0, 0.1, or 0.01 mg/kg-day perchlorate (Clewell,
2001b).
January 16, 2002
6-94
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
varied significantly between the male and female neonates, and Clewell (2001b) noted that the
difference appears to be a function of age. At PND5, the male neonatal plasma concentrations
were nearly 4 times higher than those of the female neonates in the 0.1 mg/kg-day dose group.
By PND10, however, no significant sex difference was found in the plasma perchlorate
concentrations at the same dose.
Clewell (2001b) fit the male neonatal serum data because the male pups showed higher
perchlorate concentrations in the serum than the female pups (Yu, 2000). The neonatal serum
was under-predicted by the model in the 0.01 mg/kg-day dose group. Clewell (2001b) strongly
asserts that this was due to the fact that the milk concentration was also under-predicted in that
same dose group. The three higher doses are well described in the male neonate. The female
pups also show acceptable fits at PND10. However, since the PND5 data were much lower in
the female than male neonates, the model over-predicts the PND5 time-points in the 0.1 and
1.0 mg/kg-day doses. Fits of the model to neonatal skin and GI tract are discussed in Clewell
(2001b).
As in the maternal model, the clearance value for urinary excretion was determined by the
fit of the model to the serum from the 10 mg/kg-day dose, while the lower doses were used to
determine the kinetic parameters for the binding in the neonate. Both binding and urinary
clearance were considerably lower in the pup than in the dam (Table 6-7).
6.4.2.3.2 Iodide Model Parameterization
Clewell (2001b) developed the iodide aspect of the model by visually fitting the model to
measured tissue concentrations in the dam and neonate from the control group of the inhibition
kinetic study. Only the values for Vmax and permeability area needed to be fit with the model.
As shown in Figure 6-41, the model simulations of iodide concentrations (ng/L) after an iv
injection of 2.10 ng/kg radioalabeled iodide (125I ) on PND10 versus the experimental data in the
lactating dam are shown in the dam serum (A) and thyroid (B) and in male (C) and fetal (D)
neonatal serum.
The model simulations describe the data well with the exception of the longest time point
in the neonates. The clearance value for urinary excretion was determined by fitting the maternal
serum prediction to the above data while keeping good fits in the other tissues, such as maternal
skin, GI, and mammary gland (Clewell, 2001b). Permeability area values were adjusted to
January 16, 2002
6-95
DRAFT-DO NOT QUOTE OR CITE
-------
-1
~ob
E
a
"8
I
J6
MOO
1*
o 2400
1800
JS 600
<2
Time (hours)
240.0 245.2 2S0.4 2S5.6 260 8 266.0
Time (hours)
238.0
244.4
250.8 257.2
Time (hours)
263.6
270.0
238.0 244.4 250.8 257.2 263.6 270.0
Time (boun)
Figure 6-41. Lactating dam and neonatal rat PBPK model predictions (lines) versus data
time course (mean ± SD) of iodide concentrations (mg/L) in the maternal
serum (A) or thyroid (B) and in male (C) or female (D) neonatal pups on
PND10 after an iv dose to the lactating dams of 2.10 ng/kg 125I" (Clewell,
2001b). Data ofYu (2000, 2002).
describe the behavior of the iodide data; varying the permeability area values toward 1.0 L/hr-kg
generally increased the rate at which uptake and clearance in a particular tissue occurred;
decreasing permeability area slowed the uptake and clearance.
January 16, 2002
6-96
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The behavior of the iodide in the neonatal skin and GI tract and contents appeared to be
different from the dam. The iodide tended to stay in the tissue of the neonate longer, requiring a
slower clearance in the fetal tissues than was used in the corresponding maternal tissue. As a
result, permeability area values used for the GI and skin in the neonate were lower than those
used in the dam (Table 6-8). For example, the permeability area value in the skin was
determined to be 0.5 L/hr-kg in the dam, but was decreased to 0.02 L/hr-kg in the neonate.
However, these values correspond well to the values used for the fetus in the pregnancy model
(Clewell, 2001a).
The neonatal urinary clearance value was determined to be 0.02 L/hr-kg in the neonate,
which is very similar to the maternal value (0.03 L/hr-kg of the dam). This was a surprise,
because the neonate was expected to have a much lower rate of excretion than the more mature
dam; however, Clewell (2001b) notes that this trend is supported in the literature. Capek and
Jelinek (1956) measured the amount excreted by pups at various ages. The neonates required
external stimulation by the mother in order to release the urine from their bladders. However,
when that stimulation was supplied, the neonates were able to excrete urine at the same rate as an
adult rat. Therefore, it is reasonable that the urinary excretion rate is similar between the pup and
adult. The amount of iodide lost to urine is then dependent on both the urinary clearance value
and the concentration of the ion in the kidney (Clewell, 2001b).
6.4.3 Model Validation
The ability of the model to simulate the kinetics of perchlorate in the lactating dam and
neonate was tested against the perchlorate time course data collected in vivo by AFRL/HEST.
Since the study was performed with an acute perchlorate dose, it was necessary to make minor
changes in the thyroid perchlorate parameters. The long-term exposure to perchlorate in the
drinking water studies that were used to determine the perchlorate parameters is sufficient to
induce up-regulation in the thyroid (Yu, 2000). Therefore, the thyroid parameters in the dam at
this point would be different from those seen in an acute situation. Clewell (2001b) achieved the
model fits to the acute data by altering the partition coefficient (from 2.25 in the drinking water
to 0.13 in the acute exposure) and permeability area value (from 6.0E-4 to 4.0E-5) into the
thyroid at the basolateral membrane (thyroid follicle). The value for the partitioning into the
follicle in a naive thyroid was calculated as described previously from Chow and Woodbury
January 16,2002 6-97 DRAFT-DO NOT QUOTE OR CITE
-------
1 (1970). The permeability area value in the naive thyroid follicle was determined by fitting the
2 model prediction to the thyroid data, while keeping good fits in the serum and other tissues.
3 Figure 6-42 shows the model predictions versus the data time course of perchlorate
4 concentrations in maternal serum (A), thyroid (B), or mammary gland (C) and in neonatal serum.
5
6
0.30
242.8
245 6 248.4
Time (hours)
251 2
254 0
242.8
245.6 248.4
Time (hours)
251.2
240 0
245 6 248.4
Time (hours)
g» 0 04 -
¥
2
S? 0-03 -
3
y 0.01
240.0
242.8
245 6 248.4
Time (hours)
254.0
Figure 6-42. Validation for lactating dam and neonatal rat PBPK model (Clewell, 2001b).
Model predictions (lines) versus data time course (mean ± SD) of perchlorate
in the maternal serum (A), thyroid (B), or mammary gland (C) and in
neonatal serum (D) after an iv dose of 1.0 x 106 mg/kg perchlorate on PND10.
Data of Yu (2000, 2002) and Yu et al. (2000).
January 16, 2002
6-98
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
The maternal serum is not fit particularly well and the neonatal serum fit could also be
improved. Clewell (2001b) notes the difficulties may be due to the use of the iv kinetic data as
well as some additional challenges not yet met by the model with respect to the mammary gland.
Clewell (2001 b) increased the transfer of perchlorate through the milk in the acute studies in
order to fit the model derived on drinking water studies to these acute (iv) data. That is, the
value for the Vmax into the mammary tissue was increased in order to allow more perchlorate
into the mammary compartment, and the permeability area into the milk was decreased in order
to minimize the back flow of perchlorate into the mammary from the milk. This essentially
forced the perchlorate in the milk to be passed to the neonate rather than return to the mammary
tissue of the mother. The Vmaxc for the binding in the neonate was decreased slightly from the
value used in the drinking water simulations. This may have been due to increased transfer of
iodide in the acute simulations. When the same parameters were used in the mammary
compartment that were determined with the drinking water studies, the amount in the mammary
tissue was low and the clearance of the mammary was too slow. As a result, acute neonatal
serum levels were under-predicted. By adjusting the Vmaxc, the model was able to achieve
reasonable fits to the available data in the maternal and neonatal tissues. Clewell (2001b)
suggests that different fractions of the dose are transferred through the milk during an acute (iv)
exposure versus a drinking water scenario.
Figure 6-43 shows the model predictions against the data obtained in the AFRL/HEST
cross-fostering study described in Section 6.4.1.1.2. Perchlorate concentrations (mg/L) in the
maternal thyroid of dams exposed during gestation (A) or only during lactation (B) show similar
results. Perchlorate concentrations (mg/L) in neonatal serum exposed only during gestation (C)
or only during lactation (D) also contained similar levels. Because the data were taken on
PND10, the sex difference seen at the earlier time points was not present and the simulation is
shown for the average of all pups.
The model is able to predict the data from the cross-fostering study very well. It is apparent
from the data and from the model prediction of the cross-fostering data that the gestational
exposure to perchlorate does not affect the perchlorate concentrations of the maternal serum and
thyroid or the neonatal serum. This is in agreement with other studies that indicate the rapid
clearance of perchlorate in the urine (Yu et al., 2000), but not in agreement with the toxicological
January 16, 2002
6-99
DRAFT-DO NOT QUOTE OR CITE
-------
64 128 192 2S6
Time after Parturition (hours)
64 128 192 256
Time after Parturition (hours)
320
64 128 192 256
Time after Parturition (hours)
320
~o4
B
B
5
£
z
c
V
2
64 128 192 256
Time after Parturition (hours)
Figure 6-43. Validation for lactating dam and neonatal rat PBPK model (Clewell, 2001b).
Model predictions (lines) versus data time course (mean ± SD) of perchlorate
in the maternal thyroid during gestation (A) or during lactation only (B) and
in the neonatal serum during gestation (C) or during lactation only (D) after
an iv dose of 1.0 x 106 mg/kg perchlorate on PND10. Data of Mahle (2001).
observations between the 1998 and 2001 developmental neurotoxicological studies performed by
Argus Research Laboratories, Inc. (1998; 2001). Differences in the hormone data are discussed
in Clewell (2001c) and other differences may be due to strain differences (Fail et al., 1999).
From the model, even though the neonatal urinary excretion is much lower than that of the dam
January 16, 2002 6-100 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
(0.005 vs. 0.07 L/hr-kg), the prenatal exposure does not affect the serum levels of the neonate
past PND2. This is in accord with the observations made of the BMDL estimates for the post-
natal thyroid discussed in Chapter 5.
Additional validation exercises were performed by Clewell (2001b), showing reasonably
adequate model fits to the data of Potter et al. (1959) and that of Sztanyik and Turai (1988) as
shown in Clewell (2001b). Maternal radiolabeled iodide concentrations were overpredicted in
the thyroid on PND18. The maternal milk concentrations were also overpredicted for the earlier
time point, but were within the range at the later. The model predicted the radiolabeled iodide
data obtained in the litters of Sztanyik and Turai (1988) quite well. This indicates that the
lactation and neonatal kinetics are characterized accurately.
Figure 6-44 shows that the Clewell (2001b) model is able to predict the radiolabeled iodide
(125r) uptake-inhibition data in maternal thyroids on PND10 from the AFRL/HEST "acute" (iv)
studies with perchlorate. The inhibition was described well by the model across the range of
time points from 0.5 to 24 hours postdosing. The top line indicates the prediction for the control
thyroid, and the bottom line shows the effect of perchlorate. The model is able to describe the
kinetics of iodide under both conditions.
The Clewell (2001b) model is also able to predict the radiolabeled iodide uptake inhibition
data from AFRL/HEST obtained after "chronic" drinking water exposures. Figure 6-45 shows
the radiolabeled iodide (125I ) concentrations (mg/L) in the maternal thyroids at PND5 after
23 days of dosing with perchlorate at 0.0, 0.01, 1.0, and 10.0 mg/kg-day.
6.4.4 Summary
Clewell (2001b) highlights some important differences in the lactating dam and neonatal rat
model structure that were necessary in order to adequately describe the distribution kinetics of
perchlorate and iodide. The loss of iodide and perchlorate in the milk results in much faster
clearance rates of the anions from the dam. Studies also suggest that the loss of iodide to the
mammary gland and milk decreases the iodide available for the maternal thyroid (Brown-Grant,
1961; Yu, 2000; Yu et al., 2000). The thyroidal maximum capacities are lower in the lactating
and pregnant dam than in the male rat. Model parameterization in the male rat indicated the need
for Vmaxc values for uptake into the follicle of the thyroid of 2.2 x 103 L/hr-kg for perchlorate
and 5.5 x 104 L/hr-kg for iodide while the gestation model required values of 1.5 x 103 L/hr-kg
January 16,2002 6-101 DRAFT-DO NOT QUOTE OR CITE
-------
3000
J
^ob
a
o
&
4>
"O
*3
o
T3
oJ
pei
iv dose perchlorate
2400 H I iv dose 1251'
1800 -
~ 1200
1>
JO
CO
600 -
0
Control
Perchlorate
238.0 244.4 250.8 257.2
Time (hours)
263.6
270.0
Figure 6-44. Validation for lactating dam and neonatal rat PBPK model (Clewell, 2001b).
Model predictions (lines) versus data time course (mean ± SD) of 125I"
radiolabeled iodide in the maternal thyroid with and without an iv dose of
perchlorate at 1.0 mg/kg perchlorate 2 hours prior to an iv dose of 2.10 ng/kg
l25I" to the dam on PND10. The top simulation indicates the control thyroid
and the lower indicates the inhibited thyroid. Data of Yu (2000) and Yu et al.
(2000).
1 and 4.0 x 104 L/hr-kg for the same parameters. This difference is supported in the literature.
2 Versloot et al. (1997) suggest that the pregnant rat may have a lowered reserve of iodide in the
3 thyroid toward the end of pregnancy, causing increased activity in the thyroid. This may also be
4 true in the lactating rat. The skin of the lactating dam also required a smaller value for Vmaxc
January 16, 2002
6-102
DRAFT-DO NOT QUOTE OR CITE
-------
120.0 120.8 121.6 122.4 123.2 124.0
Time (hours)
Figure 6-45. Validation for lactating dam and neonatal rat PBPK model (Clewell, 2001b).
Model predictions (lines) versus data time course (mean ± SD) of 12SI"
radiolabeled iodide in the maternal thyroid on PND5 after 23 days dosing
with perchlorate in drinking water at 0.0, 0.1,1.0, and 10.0 mg/kg-day. All
experimental data were taken two hours post-dosing. Data of Yu et al. (2000).
1 than the male rat. This is supported by the work of Brown-Grant and Pethes (1959), who
2 reported higher levels of iodide in the skin of male rats than in female rats. Skin, therefore,
3 appears to be a more important iodide reserve in the male rat than the female.
4 The described PBPK lactation model is able to predict the distribution of perchlorate in the
5 tissues of active uptake and serum of the lactating dam and neonate on PND5 and PND10 after
6 exposure to perchlorate in drinking water. Perchlorate distribution in this dynamic system is
January 16, 2002 6-103 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
described utilizing a pharmacokinetic approach to the modeling and accounting mathematically
or physiological changes, such as changing tissue volumes and maternal and neonatal growth.
The model predicts the transfer of perchlorate to the neonate and is also able to describe the
uptake into tissues of interest in the neonate, such as the GI contents and skin; however, the EPA
believes that both the maternal and neonatal serum fits could be improved. This may already be
accomplished with the additional data to which Clewell (2001b) alludes or, as noted previously,
the radionuclide modeling efforts of the ICRP (2001, 1989) may be informative.
The kinetic behavior of iodide is well described with the existing model, in spite of the
physiological complexity of the described system. The dam and neonate were accurately
simulated at a range of doses that spans four orders of magnitude (2.10 to 33,000 ng/kg) between
days 1 and 18 of lactation. The active sequestration of iodide in maternal and neonatal tissues
and the transfer of iodide between mother and neonate was described kinetically with the model;
data have been simulated at a variety of doses and at various time points up to 14 days after
exposure. The fact that the model was able to simulate data from other laboratories under a
variety of different conditions attests to the validity of the model structure and its applicability to
other studies. This also provides greater confidence in the model structure.
The clear differences between the perchlorate data from iv and drinking water studies draw
attention to unresolved issues in the transfer kinetics of perchlorate. Although lactational transfer
has long been studied, the transport mechanisms of this ion have yet to be elucidated in the
literature. A second transporter has been identified in the mammary gland, which actively
transports anions against the chemical gradient. However, the relationship of this transporter and
the anion concentration resulting from prolonged exposure to the high doses of perchlorate used
in these studies is not known. Clewell (2001b) suggests that it is possible that the high anion
load resulting from the long-term exposure to perchlorate may have resulted in decreased
transport of the ion. It is feasible that the movement of iodide may be regulated in the mammary
tissue, because the ion is vital to the development of the newborn. The data obtained between
the acute and drinking water studies suggest that a feedback mechanism is in place, because the
model over-predicts the milk transfer in the drinking water data when the acute parameters are
used. Clewell (2001b) notes in-house experiments that may help resolve these issues are
currently underway. Additional data were provided by Yu (2002), but is not clear that all these
data have been provided to the Agency or how these will be used to improve the modeling effort.
January 16, 2002
6-104
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
6.5 APPLICATION OF PBPK MODEL STRUCTURES TO
INTERSPECIES EXTRAPOLATION
As discussed in the introduction to this chapter, the purpose of developing the proposed
PBPK model structures was to aid interspecies extrapolation. All of the proposed model
structures adequately describe both perchlorate and iodide distributions as evidenced by the fit of
the model predictions against the experimental data shown in the preceding sections of this
chapter. The degree of confidence in the model descriptions differed for the acute (iv) versus
chronic (drinking water) data to some degree in the laboratory animals. A rather large degree of
intersubject variability was evident among the human subjects, but in general the structures are
accepted as quite sound and informative to the task.
The models do not link the perchlorate and iodide kinetics to perturbations in thyroid
hormone. The existing data and current structures were not designed to address the complex
issues involved with hormone homeostasis of the hypothalamic-pituitary-thyroid feedback axis as
described in Chapter 3 or illustrated in the beginning of this chapter. Such a model would need
to incorporate the hormone levels in tissues and serum and processes such as hormone
production, storage, and secretion in the thyroid; conversion of T4 to T3 in the tissues;
deiodination of T4 and T3 to less active forms and a feedback mechanism between the hormone
levels, TSH, and the thyroid N1S. Kohn et al. (1996) developed a PBPK model that attempts to
describe the effect of dioxin on thyroid hormones. Although perchlorate and dioxin act on the
endocrine system through different modes of action, it is likely that a similar approach to that of
Kohn et al. (1996) would be required to begin to address the hormone feedback system in the
case of perchlorate. Parameterization and validation of such a model system would take a
significant number of additional studies.
Nevertheless, the model structures as they exist currently are useful, particularly when
employed in the conceptual framework proposed in Section 6.1. Because the models predict
perchlorate and iodide kinetics, two relevant dose metrics to the mode of action can be evaluated:
(1) the area under the curve (AUC) of perchlorate in the serum and (2) the degree (expressed as a
% of baseline) of iodide uptake inhibition in the thyroid.
Because developmental effects are of concern, an argument could be made that peak and
not AUC is the appropriate dose metric-the rationale being that any transient dose could be
responsible for permanent deficits. However, the AUC values, as opposed to peak
January 16, 2002
6-105
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
concentrations, were used based on the assumption that these dose metrics would represent an
averaging of the serum and thyroid perchlorate concentrations and would be better correlated
with the inhibition effect on iodide uptake. The correlation was shown to be good between the
AUC and the degree of inhibition (see Section 6.5.2). Further, due to the rapid phase of
distribution after an iv dose, measurements of concentrations are very difficult to attain
experimentally and are more variable. Using simulated peak concentrations after iv injections is
potentially problematic due to the inexact modeling of the actual distribution of dose in the tail-
vein volume and the exact time of mixing in the whole blood compartment (Merrill, 200 le).
It was also observed by EPA that the ratios for peak perchlorate serum values (Merrill, 200le;
Table 6) were in good agreement with those for the perchlorate serum AUC and that the serum
AUC were slightly more conservative if really different at all at the lower doses of concern to the
risk assessment.
The perchlorate AUC concentration in the thyroid was also considered, but the EPA and
AFRL/HEST agreed that this was a less satisfactory dose metric based on a number of
considerations. These included the following: that the thyroid Vmaxc estimates had to be
adjusted to account for upregulation of the NIS, but that this adjustment was more an empirical
exercise than a true biological model (since the hormone changes discussed above regulate the
NIS); that the thyroid concentrations were not actually measured in the fetus and neonate so that
verification of the parameters was not possible; and that the effects of perchlorate are related to
its effects on the NIS and secondary impact on thyroid hormone economy rather than to the
concentrations in the gland itself. Results of a sensitivity analysis on the adult male rat model
structure supported these conclusions (Merrill, 2001 e). The results of the sensitivity analysis will
be discussed in Section 6.5.1. Thus, the models were exercised to develop human equivalent
exposure (HEE) estimates based on internal perchlorate concentration and iodide uptake
inhibition, both components of exposure in the proposed EPA model (Merrill, 2001e). The
purpose of Section 6.5.2 is to describe the modeling exercises underlying the HEE estimates that
are used in Chapter 7.
6.5.1 Sensitivity Analysis of Proposed Adult Male Rat Model
A sensitivity analysis was performed on the adult male rat model of Merrill (2001c) in
order to determine which parameters had the most significant impact on serum and thyroid AUC
January 16, 2002 6-106 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
perchlorate concentrations. All chemical specific kinetic parameters were increased individually
by 1% from the original, optimized values. The model-predicted dosimetrics were recalculated
after each change to determine the effect on the AUC estimates. This exercise was performed at
the four-hour time point after iv dosing for the 0.1 and 1.0 mg/kg-day doses. The equation
describing the calculation of the Sensitivity Coefficient value for each PBPK perchlorate
parameter tested is (Merrill, 200le):
Sensitivity Coefficient = (A - BVB. (6-2)
(C - D)/D
where:
A = AUC for either serum or thyroid perchlorate with 1% increased parameter value,
B = AUC for either serum or thyroid perchlorate at initial parameter value,
C = Parameter value increase 1% over initial parameter value, and
D = Original initial starting parameter value.
Results are presented for the physiological parameters and chemical specific parameters
separately. Tables 6-9 and 6-10 provide the results for the 0.1 mg/kg-day dose, and
Tables 6-11 and 6-12 provide the results for the 1.0 mg/kg-day dose. The sensitivity coefficients
for the AUC estimates in both the thyroid and serum are provided and the changes in predicted
AUC estimates for the thyroid and serum are presented in the final two columns (Merrill, 200le).
The sensitivity of serum and thyroid concentrations to model parameters is not linear.
At an iv dose level of 1.0 mg/kg, the model prediction of the AUC for serum C104" concentration
is most sensitive to urinary clearance (ClUc_p). A one percent increase in this value, from
0.07 to 0.0707 ng/hr-kg, causes a decrease in AUC serum C104" concentration from 4.69 x 105 to
4.63 x 105 ng, with a sensitivity coefficient of -1.271 (Table 6-12). Serum concentration is next
most sensitive to the rate C104" unbinds from plasma proteins (Clunbc_p), with a sensitivity
coefficient of -0.869 (Table 6-12).
January 16, 2002
6-107
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-9. SENSITIVITY ANALYSIS FOR PHYSIOLOGICAL PARAMETERS IN
THE ADULT MALE RAT MODEL AT 0.1 mg/kg PERCHLORATE (C104) DOSE
(Merrill, 2001e)
Parameter"
Original
Parameter
Value
1% Increase
in Parameter
Value
AUC Thyroid
Sensitivity
Coefficient
AUC Serum
Sensitivity
Coefficient
Increase in
AUC Thyroid
C104 (ng)b
Increase in
AUC Serum
CIO/ (ng)c
BW
3.00E-01
3.03E-01
0.315
0.182
1.88E+06
9.95E+04
Blood Flows (fraction of cardiac output, QCc [L/hr])
QCc
1.40E+01
1.41E+01
-0.005
-0.006
1.88E+06
9.94E+04
QTc
1.60E-02
] .62E-02
NSd
NS
1.88E+06
9.94E+04
QSKc
5.80E-02
5.86E-02
NS
-0.003
1.88E+06
9.94E+04
QGc
1.60E-02
1.62E-02
0.011
0.008
1.88E+06
9.94E+04
QLc
1.70E-01
1.72E-01
NS
NS
1.88E+06
9.94E+04
QKc
1.40E-01
1.41E-01
-0.016
-0.010
1.88E+06
9.93E+04
QFc
6.90E-02
6.97E-02
NS
NS
1.88E+06
9.94E+04
Tissue Volumes (fraction of body weight)
Vplasc
4.10E-02
4.14E-02
0.155
0.079
1.88E+06
9.94E+04
VRBCc
3.30E-02
3.33E-02
0.192
0.109
1.88E+06
9.95E+04
Vttotc
7.70E-05
7.78E-05
0.187
0.113
1.88E+06
9.95E+04
VDTc
2.44E-01
2.46E-01
0.928
0.114
1.89E+06
9.95E+04
VTBc
1.57E-01
1.58E-01
0.203
0.114
1.88E+06
9.95E+04
VTc
6.00E-01
6.05E-01
0.453
0.114
1.88E+06
9.95E+04
VGc
5.40E-03
5.45E-03
0.197
0.112
1.88E+06
9.95E+04
VGJc
1.68E-02
1.70E-02
0.165
0.091
1.88E+06
9.94E+04
VGBc
4.10E-02
4.14E-02
0.197
0.114
1.88E+06
9.95E+04
VSkc
1.90E-01
1.92E-01
-0.053
-0.023
1.87E+06
9.93E+04
VSkBc
2.00E-02
2.02E-02
0.203
0.117
1.88E+06
9.95E+04
VLc
5.50E-02
5.56E-02
0.197
0.114
1.88E+06
9.95E+04
VKc
1.70E-02
1.72E-02
0.197
0.113
1.88E+06
9.95E+04
VFc
7.40E-02
7.47E-02
0.208
0.118
1.88E+06
9.95E+04
"Parameters as defined in Tables 6-1 and 6-2.
bAUC Thyroid Concentration using original parameters = 1.88E+06 ng C104'.
CAUC Serum Concentration using original parameters = 9.94E+04 ng CIO.
''NS = sensitivity coefficient less than 0.001.
January 16, 2002
6-108 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-10. SENSITIVITY ANALYSIS FOR CHEMICAL SPECIFIC PARAMETERS
IN THE ADULT MALE RAT MODEL AT 0.1 mg/kg PERCHLORATE (C104) DOSE
(Merrill, 2001e)
Parameter"
Original
Parameter
Value
1% Increase
in Parameter
Value
AUC Thyroid
Sensitivity
Coefficient
AUC Serum
Sensitivity
Coefficient
Increase in
AUC Thyroid
C1CV (ng)b
Increase in
AUC Serum
C104 (ng)c
Iodide Tissue/Blood Partition Coefficients
PS _p
3.10E-01
3.13E-01
0.149
0.085
1.88E+06
9.94E+04
PR _p
5.60E-01
5.66E-01
0.192
0.111
1.88E+06
9.95E+04
PK_P
9.90E-01
1.00E+00
0.192
0.111
1.88E+06
9.95E+04
PL_p
5.60E-01
5.66E-01
0.187
0.108
1.88E+06
9.95E+04
PG _p
1.80E+00
1.82E+00
0.160
0.088
1.88E+06
9.94E+04
PGJ _p
2.30E+00
2.32E+00
0.165
0.090
1.88E+06
9.94E+04
PT_p
1.30E-01
1.31E-01
1.184
0.113
1.90E+06
9.95E+04
PDT_p
7.00E+00
7.07E+00
0.928
0.114
1.89E+06
9.95E+04
PF _p
5.00E-02
5.05E-02
0.197
0.114
1.88E+06
9.95E+04
PSk_p
7.00E-01
7.07E-01
11.154
6.024
2.08E+06
1.05E+05
PRBC _p
8.00E-01
8.08E-01
11.324
6.112
2.09E+06
1.05E+05
PS _p
3.10E-01
3.13E-01
0.149
0.085
1.88E+06
9.94E+04
PR_p
5.60E-01
5.66E-01
0.192
0.111
1.88E+06
9.95E+04
PK_P
9.90E-01
1.00E+00
0.192
0.111
1.88E+06
9.95E+04
PL _p
5.60E-01
5.66E-01
0.187
0.108
1.88E+06
9.95E+04
PG _p
1.80E+00
1.82E+00
0.160
0.088
1.88E+06
9.94E+04
PGJ _p
2.30E+00
2.32E+00
0.165
0.090
1.88E+06
9.94E+04
PT_p
1.30E-01
1.31E-01
1.184
0.113
1.90E+06
9.95E+04
PDT_p
7.00E+00
7.07E+00
0.928
0.114
1.89E+06
9.95E+04
PF _p
5.00E-02
5.05E-02
0.197
0.114
1.88E+06
9.95E+04
PSk_p
7.00E-01
7.07E-01
11.154
6.024
2.08E+06
1.05E+05
PRBC _p
8.00E-01
8.08E-01
11.324
6.112
2.09E+06
1.05E+05
Perchlorate Active Uptake Parameters - Vmaxc (ng/hr-kg BW) Km (ng/L)
Vmaxc_Tp
2.90E+03
2.93E+03
47.830
6.088
2.77E+06
1.05E+05
Km_Tp
2.50E+05
2.53E+05
45.154
6.090
2.72E+06
1.05E+05
Vmaxc_DT/>
1.00E+05
1.01E+05
55.875
6.081
2.92E+06
1.05E+05
Km_DT/7
1.00E+08
1.01E+08
55.673
6.081
2.92E+06
1.05E+05
Vmaxc_Gp
1.00E+04
1.01E+04
55.769
6.080
2.92E+06
1.05E+05
Km_Gp
2.00E+05
2.02E+05
55.774
6.081
2.92E+06
1.05E+05
Vmaxc_S/?
6.50E+05
6.57E+05
54.713
5.678
2.90E+06
1.05E+05
Km_Sp
2.00E+05
2.02E+05
55.060
5.811
2.91E+06
1.05E+05
January 16, 2002
6-109
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
TABLE 6-10 (cont'd). SENSITIVITY ANALYSIS FOR CHEMICAL SPECIFIC
PARAMETERS IN THE ADULT MALE RAT MODEL AT 0.1 mg/kg PERCHLORATE
(C104 ) DOSE (Merrill, 2001e)
Parameter"
Original
Parameter
Value
1% Increase
in Parameter
Value
AUC Thyroid
Sensitivity
Coefficient
AUC Scrum
Sensitivity
Coefficient
Increase in
AUC Thyroid
CIO/ (ng)b
Increase in
AUC Serum
C104" (ng)<
Perchlorate Plasma Binding Parameters
Vmaxc_B/?
9.50E+03
9.60E+03
54.857
6.417
2.90E+06
1.06E+05
km_B/?
1.10E+04
1.11E+04
54.916
5.590
2.91E+06
1.05E+05
K.unbc _p
1.00E-01
1.01E-01
54.948
5.096
2.91E+06
1.04E+05
Perchlorate Urinary Clearance and Permeability Area Cross Products (L/hr
-kg)
ClUc _p
7.00E-02
7.07E-02
54.047
5.399
2.89E+06
1.05E+05
PAGc _p
8.00E-01
8.08E-01
54.905
5.752
2.91E+06
1.05E+05
PAGJc _p
8.00E-01
8.08E-01
54.905
5.752
2.91E+06
1.05E+05
PATc _p
5.00E-05
5.05E-05
23.273
5.776
2.31E+06
1.05E+05
PADTc _p
1.00E-02
1.01E-02
24.398
5.775
2.33E+06
1.05E+05
PASKc _p
4.00E-01
4.04E-01
3.759
-4.354
1.95E+06
9.50E+04
PARBCc _p
1.00E-01
1.01E-01
3.455
-4.508
1.94E+06
9.49E+04
"Parameters as defined in Tables 6-1 and 6-2.
bAUC Thyroid concentration using original parameters = 1.88E+06 ng C104".
CAUC Serum concentration using original parameters = 9.94E+04 ng C104'.
The predicted AUC for total thyroid concentration at a dose level of 1.0 mg/kg-day is most
sensitive to changes in the maximum capacity of the thyroid colloid (Vmaxc_DTp). A one
percent increase in this value from 1.00 x 105 to 1.01105ng/hr-kg results in an increase in AUC
thyroid concentration from 9.84 x 106 to 1.04 x 107 ng (Table 6-12). However, the AUC thyroid
concentration is almost equally sensitive to other parameters of saturable processes, including
Vmaxc, Km, and the permeability area cross product values of other saturable tissues.
With a lower iv dose of 0.1 mg/kg, the blood serum concentration remains sensitive to
changes in urinary clearance, but demonstrates increased sensitivity to the parameters of
saturable compartments and effective partitioning with skin (PSk_p) and red blood cells
(PRBC_j)). Serum concentration is most sensitive to the maximum capacity for plasma binding
(Vmaxc_B/?) at this dose level (Table 6-10).
At the lower dose level of 0.1 mg/kg, thyroid concentrations show a similar sensitivity to
parameters of saturable processes, including plasma binding, permeability area cross products,
January 16, 2002
6-110
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-11. SENSITIVITY ANALYSIS FOR PHYSIOLOGICAL
PARAMETERS IN THE ADULT MALE RAT MODEL
AT 1.0 mg/kg PERCHLORATE (C104 ) DOSE (Merrill, 2001e)
Parameter*
Original
Parameter
Value
1 % Increase
in Parameter
Value
AUC Thyroid
Sensitivity
Coefficient
AUC Serum
Sensitivity
Coefficient
Increase in
AUC Thyroid
CKV(ng)b
Increase in
AUC Serum
C104 (ng)<
BW
3.00E-01
3.03E-01
-5.944
-0.534
9.81E+06
4.67E+05
Blood Flows [fraction of cardiac output, QCc (L/hr)]
QCc
1.40E+01
1.41E+01
-0.192
-0.014
9.84E+06
4.69E+05
QTc
1.60E-02
1.62E-02
0.021
NSb
9.84E+06
4.69E+05
QSKc
5.80E-02
5.86E-02
-0.085
0.001
9.84E+06
4.69E+05
QGc
1.60E-02
1.62E-02
0.128
0.005
9.84E+06
4.69E+05
QLc
1.70E-01
1.72E-01
0.021
NS
9.84E+06
4.69E+05
QKc
1.40E-01
1.41E-01
-0.234
-0.021
9.84E+06
4.69E+05
QFc
6.90E-02
6.97E-02
0.021
NS
9.84E+06
4.69E+05
Tissue Volumes (fraction of bodyweight)
Vplasc
4.10E-02
4.14E-02
-7.734
-0.701
9.80E+06
4.66E+05
VRBCc
3.30E-02
3.33E-02
-7.649
-0.691
9.80E+06
4.66E+05
VTtotc
7.70E-05
7.78E-05
-7.841
-0.683
9.80E+06
4.66E+05
VDTc
2.44E-01
2.46E-01
7.606
-0.683
9.87E+06
4.66E+05
VTBc
1.57E-01
1.58E-01
-7.500
-0.682
9.80E+06
4.66E+05
VTc
6.00E-01
6.05E-01
-2.322
-0.683
9.83E+06
4.66E+05
VGc
5.40E-03
5.45E-03
-7.649
-0.685
9.80E+06
4.66E+05
VGJc
1.68E-02
1.70E-02
-7.883
-0.710
9.80E+06
4.66E+05
VGBc
4.10E-02
4.14E-02
-7.628
-0.682
9.80E+06
4.66E+05
VSkc
1.90E-01
1.92E-01
-8.799
-0.829
9.80E+06
4.65E+05
VSkBc
2.00E-02
2.02E-02
-7.606
-0.680
9.80E+06
4.66E+05
VLc
5.50E-02
5.56E-02
-7.628
-0.683
9.80E+06
4.66E+05
VKc
1.70E-02
1.72E-02
-7.628
-0.685
9.80E+06
4.66E+05
VFc
7.40E-02
7.47E-02
-7.585
-0.676
9.80E+06
4.66E+05
"Parameters as defined in Tables 6-1 and 6-2.
bOriginal AUC Thyroid concentration = 9.84E+06 ng C104".
'Original AUC Serum concentration = 4.69E+05 ng C104\
''NS = sensitivity coefficient less than 0.001.
January 16, 2002
6-111
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-12. SENSITIVITY ANALYSIS FOR CHEMICAL-SPECIFIC
PARAMETERS IN THE MALE RAT MODEL
AT 1.0 mg/kg PERCHLORATE (C104) DOSE (Merrill, 2001 e)
Parameter*
Original
Parameter
Value
1% Increase
in Parameter
Value
AUC Thyroid
Sensitivity
Coefficient
AUC Scrum
Sensitivity
Coefficient
Increase in
AUC Thyroid
CIO/ (ng)b
Increase in
AUC Scrum
CIO; (ng)c
Pcrchloratc Tissue/Blood Partition Coefficients
PS _p
3.10E-01
3.13E-01
-7.862
-0.728
9.80E+06
4.66E+05
PR_p
5.60E-01
5.66E-01
-7.649
-0.688
9.80E+06
4.66E+05
PK_P
9.90E-01
1.00E+00
-7.649
-0.688
9.80E+06
4.66E+05
PL _p
5.60E-01
5.66E-01
-7.670
-0.692
9.80E+06
4.66E+05
PG _p
1.80E+00
1.82E+00
-7.905
-0.714
9.80E+06
4.66E+05
PGJ _p
2.30E+00
2.32E+00
-7.883
-0.711
9.80E+06
4.66E+05
PT_p
1.30E-01
1.31E-01
12.911
-0.683
9.90E+06
4.66E+05
PDT _p
7.00E+00
7.07E+00
7.606
-0.683
9.87E+06
4.66E+05
PF _p
5.00E-02
5.05E-02
-7.628
-0.684
9.80E+06
4.66E+05
PSk _p
7.00E-01
7.07E-01
-8.885
-0.846
9.80E+06
4.65E+05
PRBC _p
8.00E-01
8.08E-01
-7.649
-0.691
9.80E+06
4.66E+05
Perchlorate Active Uptake Parameters - Vmaxc (ng/hr-kg BW), Km (ng/L)
Vmaxc_Tp
2.90E+03
2.93E+03
12.890
-0.683
9.90E+06
4.66E+05
Km_Tp
2.50E+05
2.53E+05
-15.745
-0.682
9.76E+06
4.66E+05
Vmaxc_DT/7
1.00E+05
1.01E+05
123.554
-0.687
1.04E+07
4.66E+05
Km_DTp
1.00E+08
1.01E+08
120.784
-0.687
1.04E+07
4.66E+05
Vmaxc_G/7
1.00E+04
1.01E+04
122.062
-0.687
1.04E+07
4.66E+05
Km_G/?
2.00E+05
2.02E+05
122.062
-0.687
1.04E+07
4.66E+05
. Vmaxc_Sp
6.50E+05
6.57E+05
120.997
-0.806
1.04E+07
4.66E+05
Km_Sp
2.00E+05
2.02E+05
122.914
-0.641
1.04E+07
4.66E+05
Perchlorate Plasma Binding Parameters - Vmaxc (ng/hr-kg BW), Km (ng/L)
Vmaxc_Bp
9.50E+03
9.60E+03
122.062
-0.500
1.04E+07
4.67E+05
km_B/7
1.10E+04
1.11E+04
122.062
-0.694
1.04E+07
4.66E+05
Kunbc _p
1.00E-01
1.01E-01
122.275
-0.869
1.04E+07
4.65E+05
Perchlorate Urinary Clearance and Permeability Area Cross Products (L/hr-
kg)
ClUc _p
7.00E-02
7.07E-02
115.031
-1.271
1.04E+07
4.63E+05
PAGc _p
8.00E-01
8.08E-01
122.275
-0.685
1.04E+07
4.66E+05
PAGJc _p
8.00E-01
8.08E-01
122.275
-0.686
1.04E+07
4.66E+05
PATc _p
5.00E-05
5.05E-05
100.969
-0.686
1.03E+07
4.66E+05
PADTc _p
1.00E-02
1.01E-02
120.784
-0.687
1.04E+07
4.66E+05
PASKc _p
4.00E-01
4.04E-01
123.341
-0.567
1.04E+07
4.67E+05
PARJBCc _p
1.00E-01
1.01E-01
122.062
-0.687
1.04E+07
4.66E+05
"Parameters as defined in Tables 6-1 and 6-2.
bOriginal AUC Thyroid concentration = 9.84E+06 ng C104'.
'Original AUC Serum concentration = 4.69E+05 ng C104".
January 16, 2002
6-112
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
and urinary clearance. However, the predicted thyroid concentrations at both dose levels (1.0 and
0.1 mg/kg) are most sensitive to a change in Vmax_DT/?. The Vmax values of the thyroid were
established by empirically fitting thyroid radioiodide and perchlorate uptake from several data
sets ranging in three orders of magnitude
6.5.2 Derivation of Human Equivalent Exposure Estimates
As discussed, the following internal dosimetrics were chosen to represent output from each
of the PBPK models: area under the curve (AUC) perchlorate concentrations in serum and
thyroid; peak serum and thyroid perchlorate concentrations; the total amount of perchlorate
excreted in the urine; the AUC for the lactational and placental transfer of perchlorate; and the
percent inhibition of iodide uptake into the thyroid. In order to explore the dose-response
elationship of these values, the target dosimetrics were evaluated across several doses in both
acute and sub-chronic exposure scenarios using previously developed PBPK models at the
AFRL/HEST; i.e., the models for the adult male rat (Merrill, 2001c) and human (Merrill, 2001d)
described in Section 6.2, the pregnant and fetal rat model (Clewell, 2001a) and the lactating and
neonatal rat model (Clewell, 2001b).
Acute (iv) pharmacokinetic studies in the adult male rat were used as the basis for this
dose-response analysis because iodide uptake inhibition could be correlated to perchlorate levels.
Further, as discussed in Section 6.1, the initial inhibition of iodide is viewed in the conceptual
model as the important step in the transient phase (Figure 6-2). Transient decrements in T4 can
result in permanent neurodevelopmental sequelae. In drinking water studies, upregulation of NIS
in the rat is so rapid that it resulted in no measurable thyroid iodide inhibition, so the iv doses
were used to estimate this initial insult. The target internal dosimetrics were first calculated in
each of the rat models for acute exposure to perchlorate (single iv administration) at doses of
0.01, 0.1, 1.0, 3.0, 5.0, 10.0, 30.0, and 100.0 mg/kg. In order to correlate perchlorate parameters
to data-validated inhibition, the 2 to 4 hr time-frame was used for all acute calculations. The
AUC for thyroid and serum were calculated by integrating predicted tissue concentrations from
2 to 4 hrs post dosing.
These same dosimetrics calculated for acute exposures were also determined for subchronic
(drinking water) perchlorate exposures at doses of 0.01, 0.1, 1.0, 3.0, 5.0, 10.0, 30.0, and
100.0 mg/kg-day. In order to achieve steady state concentrations, the models were run until the
January 16,2002 6-113 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
predicted peak and trough heights did not change from one day to the next (Merrill, 200 le).
Serum and thyroid perchlorate AUC concentrations were then determined over a 24 hr period
(240-264 hrs in male, lactating, and neonatal rats; 480-504 hrs in pregnant and fetal rats).
Although the tissues reach steady state perchlorate concentrations within one week, the above
time-points were chosen in the lactation and gestation models for their ability to be verified with
data (Clewell, 2001 a,b). The male rat model was run at the same time as lactation for the sake of
consistency with the other models. The total perchlorate AUC in the serum and thyroid were
determined from each the models at 240 and 264 hrs (or 480 and 504 hrs). The difference in the
two values was then divided by 24 hrs to give the AUC in units of ng/L-hr.
The AFRL/HEST experiments (Yu, 2000; Yu et al., 2000) have shown upregulation of the
NIS to be both time and dose-dependent. Thus, at lower doses, the rat thyroid was completely
upregulated after only a few days of drinking water exposure. Iodide uptake in the thyroid at
higher perchlorate doses (>10 mg/kg-day) was completely restored by the 18th day of exposure,
the time of data collection in the pregnant and fetal rats (Clewell, 2001).
Drinking water studies in the adult male rats showed elevated perchlorate uptake in the
thyroid at drinking water doses of 3.0 mg/kg-day and higher (Yu et al., 2000; Merrill et al.,
2001c). Increased perchlorate uptake also results from upregulation of NIS. Since perchlorate is
transferred into the thyroid via NIS, the inhibiting anion is "upregulated" along with iodide.
In order to simulate increased perchlorate concentrations in thyroids of the 3.0, 10.0, and
30.0 mg/kg-day dose groups, the original value for follicular Vmaxc (Vmaxc_Tp) was adjusted
to obtain the best fit of the model simulation to experimental data (Table 6-13). Since there were
no pharmacokinetic data available for the 5.0 and 100.0 mg/kg-day dose groups, values for
Vmaxc_T/? were estimated from a Michaelis-Menten fit to the adjusted Vmaxc's at 3.0, 10.0, and
30.0 mg/kg-day doses (Figure 6-46). Target dosimetrics in the male rat were calculated for both
originally optimized parameters and these adjusted ("upregulated") parameters.
This process of adjusting ("upregulating") the Vmaxc_T/? values was not necessary in the
gestation, lactation, or human models, as they were able to successfully describe perchlorate
concentrations in serum and thyroid at all measured doses (0.01 - 10.0 mg/kg-day in gestation
and lactation; 0.02 - 12 mg/kg-day in human) using one set of model parameters (Clewell,
2001 a,b; Merrill, 2000). Merrill (200le) posits that it was not necessary because it is likely that a
loss of maternal iodide to the fetus and neonate causes dams to exist in a chronic state of
January 16, 2002
6-114
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 6-13. "UP-REGULATED" VALUES OF VMAXC_T/>a AFTER
PERCHLORATE DRINKING WATER EXPOSURE IN THE ADULT MALE RAT
MODEL (Merrill, 2001e)
Drinking Water Dose (mg/kg-day)
Adjusted Vmaxc_Tp (ng/hr-kg)
0.01
2900
0.1
2900
1
2900
3
9000
5
17500"
10
32000
30
55000
100
79000"
"Maximum velocity capacity of active transport in the thyroid follicle.
bData not available for these dose levels.
1e+5 i
3
n
O 8e+4
w
®
CO
£
CD
_c
C
6e+4 -
O 4S+4 '
Q.
h|
O
x
CO
E
>
2e+4 -
0.0 2.0e+7 4.0e+7 6.0e+7 8.0e+7 1.0e+8
Drinking Water Dose in Rats (mg/kg/day)
1.2e+8
Figure 6-46. Upregulation of maximal capacity (ng/kg-hr) of active transport into the
thyroid follicle for perchlorate (Vmaxc_Tp) optimized by fitting to drinking
water data in the rat. Upregulation is first noted in the 3.0 mg/kg-day dose
group.
January 16, 2002
6-115
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
thyroidal up-regulation. As a result, the effect of perchlorate on the thyroid was less dramatic
than in the male rat where a completely naive system is perturbed by an inhibiting chemical.
Thus, the PBPK models for gestation and lactation were able to describe thyroid perchlorate
levels at drinking water doses from 0.01 to 10.0 mg/kg-day without adjusting the follicular
Vmaxc (Vmaxc_T/7) values with dose.
Increased follicular Vmaxc values were not needed to fit the human data likely due to the
larger size of the human thyroid colloid versus that of the rat and to the differences in plasma
protein binding discussed in Chapter 3.
The human PBPK model (Merrill, 200Id) was used to calculate all target dose metrics in
both acute and two-week drinking water perchlorate exposures in a 70 kg adult at doses of 0.01,
0.1, 1.0, 3.0, 5.0, 10.0, 30.0, and 100.0 mg/kg-day. Acute serum and thyroid perchlorate AUC
concentration estimates were calculated with the model over an eight hr time period (from 24 to
32 hrs post-exposure) in order to correlate perchlorate parameters to data-validated iodide
inhibition. For two-week drinking water exposures, the thyroid and serum perchlorate AUC
concentration estimates were calculated over a 24 hr period after serum and thyroid
concentrations reached steady state. The 240 to 264 hr time period was chosen for consistency
with the male rat model (Merrill, 2001c).
The adult human model (Merrill, 200Id) was also used to predict dosimetry in a 15 kg
child. The same dosimetrics were run in the model for the child and adult. However, since an
average child drinks less water than an adult (approximately 1 L/d as opposed to 2 L/d in the
adult), the actual exposures of a child and adult from the same water source would be different.
For example, a 15kg child consuming 1 L of contaminated water would receive a daily dose (per
kg bodyweight) that was 2.3 times that of a 70 kg adult consuming 2 L of water. Table 2 shows
the concentration of the drinking water required to deliver the same dose to a 15 kg child and a
70 kg adult. For the purpose of this paper, dosimetric comparisons were calculated using the
same dose (mg/kg-day) in the adult and child.
Figure 6-47 shows the curve generated from plotting the experimentally-determined percent
inhibition versus the corresponding PBPK-derived serum (A) and thyroid (B) perchlorate AUC
concentration estimates after acute (iv) exposure in rats. Thyroidal radiolabeled iodide (125I )
uptake measurements were taken two hours after iv administration of perchlorate. The solid line
January 16, 2002
6-116
DRAFT-DO NOT QUOTE OR CITE
-------
• - represents experimental data
~ - represents PBPK model simulation values
Figure 6-47. Michaelis-Menten fit of the "acute" male rat area under the curve (AUC) for
serum (A) and thyroidal perchlorate (AUCTtot_p) in ng/L-hr. Model
predictions and actual data shown for percent radiolabeled iodide uptake
inhibition after iv injection of perchlorate.
January 16, 2002
6-117
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
represents a fit (not a PBPK model simulation) using the Michaelis-Menten type equation given
below:
Y = (A x AUCdosc)/(AUCdosc + B) (6-3)
Where 'Y' represents the predicted percent inhibition of radioiodide uptake, 'A' represents the
maximal percent inhibition of radioiodide uptake, 'B' is related to the affinity of iodide uptake
based on serum concentration, and AUCdose represents the AUC at each specific dose of
perchlorate. The above equation was also used to derive the dose-response relationship in
subsequent figures. The correlation coefficient (r2) greater than 0.91 in all cases indicated
excellent fit for all (see Merrill, 200 le; Table 3).
Figure 6-48 shows the PBPK-derived AUC perchlorate concentration estimates for
drinking water exposure to the adult male rat versus the calculated percent inhibition of
radioiodide in the serum (A) and thyroid (B). The values for AUC of perchlorate concentration
in the serum were determined by running the adult male rat model (Merrill, 2001c) across doses.
Corresponding percent inhibitions were calculated by putting serum AUC perchlorate
concentration values into the equation from Figure 6-47. Human response (thyroid inhibition) to
subchronic exposure is similar to that of an acute exposure in the rat. This approach allows the
sub-chronic serum levels in the rat be related to iodide uptake in the native thyroid. The values
for AUC of thyroid perchlorate concentration (B) were determined by running the male rat model
(Merrill, 2001c) at steady state (between 240 and 264 hours of drinking water exposure) across
the doses shown. Corresponding percent inhibitions were calculated by putting thyroid AUC
values in the equation from Figure 6-47.
The actual human iodide inhibition data (Greer et al., 2000) were plotted as a function of
the perchlorate AUC concentration estimates for serum and thyroid calculated with the PBPK
model in Figure 6-49. The measured percent inhibition of radiolabeled iodide uptake in the
serum and thyroid on Day 2 of drinking water exposure to perchlorate is shown versus the
PBPK-derived estimates for human volunteers (both male and female). Inhibition data from time
points earlier than Day 2 of perchlorate in the human drinking water (Greer et al., 2000) and
inhibition data from acute perchlorate dosing in humans were not available. Therefore, the
inhibition measurements on Day 2 of perchlorate drinking water exposure were the closest-
January 16, 2002
6-118
DRAFT-DO NOT QUOTE OR CITE
-------
Figure 6-48. Michaelis-Menten fit of the "chronic" male rat area under the curve (AUC)
for serum (A) and thyroidal (B) perchlorate (ng/L-hr). Model predictions and
actual data shown for percent radiolabeled iodide uptake inhibition after
drinking water exposure of perchlorate. Fit for serum calculated percent
inhibition of radioiodide uptake calculated from equation used in Figure 6-47
(A) and for thyroid from Figure 6-47(B).
January 16,2002
6-119
DRAFT-DO NOT QUOTE OR CITE
-------
Serum CI04" AUC (ng/L-hr)
Figure 6-49. Michaelis-Menten fit of the human area under the curve (AUC) for serum (A)
and thyroidal (B) perchlorate (ng/L-hr) on exposure Day 2. Model
predictions and actual data shown for percent radiolabeled iodide uptake
inhibition after drinking water exposure of perchlorate.
January 16, 2002
6-120
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
available representation of an acute human dose. Measured serum TSH and thyroid hormones
indicated that thyroids were in normal homoeostatic state in human volunteers during the entire
two week study (Merrill, 200Id).
The HEE estimates were calculated using the models as described in Section 6.1
(Figure 6-4). The HEE that would result in the same perchlorate AUC concentration estimates
for serum (A) and thyroid (B) in the human and rat and the corresponding percent inhibition of
iodide uptake is presented in Figure 6-50. Values for percent inhibition were determined from
the rat serum AUC during drinking water exposures to perchlorate using the Michaelis-Menten
equations from Figure 6-47. The correlation coefficient for both the serum and thyroid AUC
versus percent iodide uptake inhibition relationship was 0.99.
6.5.3 Summary
The correlation coefficients for the dose-response relationships using the PBPK-model
generated HEE estimates between serum and thyroid perchlorate AUC concentration versus
iodide inhibition indicated good fits. Tables of the actual estimates and their ratios can be found
in Merrill (200 le).
The rat serum ratios (AUC and peak concentrations) change significantly between 0.1 and
3.0 mg/kg-day due to binding of perchlorate by plasma proteins. Plasma binding is saturated at
doses greater than 1.0 mg/kg-day. Male rat to human ratios are notably lower than those ratios
between rats, as plasma binding of perchlorate occurs to a much lesser extent in humans.
HEE estimates were calculated for both a 15 and 70 kg human. The differences between
the 15 and 70 kg human HEE estimates were never greater than 75%, indicating that body weight
doesn't significantly affect the target dose metrics. Interestingly, the HEE estimates were greater
in the 15 kg child. One might expect the adult and child HEE estimates to be nearly equal, given
no parameters were changed in the human model except body weight. However, physiological
parameters within the model are linearly scaled by body weight; whereas, chemical-specific
parameters are scaled nonlinearly (e.g., as a multiple of body weight to a power of 3A).
As indicated later in the sensitivity analysis, the internal dose metrics presented are more
sensitive to chemical-specific parameters, especially those describing saturable kinetics.
Therefore, the chemical-specific parameter values for the 15 kg child are proportionally greater
January 16, 2002
6-121
DRAFT-DO NOT QUOTE OR CITE
-------
100 -I
80 -
60
40 -
20 -
O.O 2.0e*7 4.0e*7 6.0e*7 8.0e+7 1.0e-*-8 l.2e*8
Drinking Water Dose (ug/kg-day)
1,4e+8
100
0.0
S.Oe+8 1.0e+9 1.5e+9 2.0e+9
Drinking Water Dose (ug/kg-day)
2.5e+9
Figure 6-50. Michaelis-Menten fit of the human equivalent exposure (HEE) of perchlorate
in drinking water derived from the area under the curve (AUC) for serum
(A) or thyroid (B) versus percent predicted inhibition in the rat after an
"acute" (iv) dose.
1
2
(in terms of body weight) than those of the adult. As a result, a slightly higher dose is required to
saturate these tissues in a child.
January 16, 2002 6-122 DRAFT-DO NOT QUOTE OR CITE
-------
1 When comparing the dose metrics for serum versus thyroid, the HEE estimates calculated
2 from the thyroid were less than the HEE estimates calculated from the serum by a factor of 100 at
3 a 0.01 mg/kg-day dose level. This difference became a factor of 10 starting at the 5.0 mg/kg-day
4 concentration for the 15 kg child and at 10.0 mg/kg-day for the adult.
5 These considerations will be explored in Chapter 7 to develop dosimetric adjustment
6 factors for the observed effect levels.
7
January 16, 2002
6-123
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
7. BOSE-RESPONSE ASSESSMENTS FOR
HUMAN HEALTH
The available database prior to initiation of the perchlorate testing strategy in 1997 (see
Chapter 3) on the health effects and toxicology of perchlorate or its salts was very limited. The
majority of human data were clinical reports of patients treated with potassium perchlorate for
hyperthyroidism resulting from an autoimmune condition known as Graves' disease. Potassium
perchlorate still is used diagnostically to test TSH, T3, and T4 production in some clinical
settings. The primary effect of perchlorate is to decrease the production of thyroid hormones by
competitively inhibiting iodide anion uptake into the thyroid at the sodium (NaT)-iodide (I)
symporter (NIS) and by causing a discharge of stored iodide from the thyroid gland.
It was difficult to establish a dose-response for the effects on thyroid function from daily or
repeated exposures in healthy humans based on the data in patients with Graves' disease because
of a variety of confounding factors, including that the disease itself has effects; that often only a
single exposure and not repeated exposures were tested; that only one or two doses were
employed; and that often the only effect monitored was iodide release from the thyroid or control
of the hyperthyroid state. There were limited data in normal human subjects and laboratory
animals that support the effect of perchlorate on thyroid hormones, but the majority of these
studies suffer from the same limitations as those with the Graves' disease patients, with respect
to the number of doses and exposures. These limitations prevent establishment of a quantitative
dose-response estimate for the effects on thyroid hormones after long-term repeated exposures to
perchlorate in healthy human subjects.
In addition, on December 14, 2001, after internal peer review of this document, the Agency
articulated its interim policy on the use of third-party studies submitted by regulated entities
(U.S. Environmental Protection Agency, 2001c). For these purposes, EPA is considering "third
party studies" as studies that have not been conducted or funded by a federal agency pursuant to
regulations that protect human subjects. Under the interim policy, the Agency will not consider
or rely on any such human studies (third-party studies involving deliberate exposure of human
subjects when used to identify or quantify toxic endpoints such as those submitted to establish a
January 16, 2002
7-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
NOAEL or NOEL for systemic toxicity of pesticides) in its regulatory decision making, whether
previously or newly submitted. Some of the clinical studies contained in this database fall in this
category of studies not to be considered. However, the scientific and technical strengths and
weaknesses of these studies were described before this Agency policy was articulated.
Therefore, because of the scientific shortcomings of these studies, they will not be used as
"principal studies" in the derivation of an RfD. The ethical issues surrounding the conduct of
these studies or their use for regulatory purposes in light of the Agency's interim policy will not
be discussed in this document. The Agency is requesting that the National Academy of Sciences
conduct an expeditious review of the complex scientific and ethical issues posed by EPA's
possible use of third-party studies which intentionally dose human subjects with toxicants to
identify or quantify their effects.
Thyroid hormone deficiencies, such as those induced by perchlorate, can affect normal
metabolism, growth, and development. However, no robust data existed previously with which
to evaluate potential target tissues or effects other than those in the thyroid. The data on the
thyroid effects were also insufficient for quantitative dose-response assessment. Additionally,
there were no data with which to evaluate the effects of perchlorate in potentially susceptible
populations, such as developing fetuses; nor were there data on the effects of perchlorate on the
reproductive capacity of male or female laboratory animals.
Benign tumors had been reported in the thyroids of male Wistar rats and female BALB/c
mice treated with repeated, high-dose exposures (2 years at 1,339 mg/kg-day and 46 weeks at
2,147 mg/kg-day, respectively) of potassium perchlorate in drinking water, establishing
perchlorate as a carcinogen. Benign tumors in the thyroid have been established to be the result
of a series of progressive changes that occur in the thyroid in response to interference with
thyroid-pituitary homeostasis (i.e., perturbation of the normal stable state of the hormones and
functions shared between these two related glands). This progression is similar regardless of the
cause of the thyroid hormone interference (Hill et al., 1989; Capen, 1997; Hurley et al., 1998).
EPA has adopted the policy that for the dose-response of chemicals that cause disruption in the
thyroid but that do not have genotoxic activity (i.e., cause damage to DNA or show other genetic
disruption) a threshold for carcinogenicity is to be based on precursor lesions (U.S.
Environmental Protection Agency, 1998e).
January 16, 2002
7-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
1
2
3
In the case of perchlorate, an overall model based on its mode of action has been developed
as shown in Figure 7-1. The model supports iodide inhibition as the key event that precedes the
hormone and thyroid changes with subsequent neurodevelopmental and neoplastic sequelae.
Focusing on the key event of iodide uptake inhibition allows a harmonized approach to both the
"noncancer" and "cancer" toxicity that occurs downstream along the continuum. Thus, one
harmonized risk estimate is derived for both sequelae based on their common mode of action.
¦ ¦ ¦ ¦ ¦ ¦ •
• iii ¦ • ¦
¦ •
Susceptibility
Figure 7-1. Mode-of-action model for perchlorate toxicity proposed by the U.S. EPA
(U.S. Environmental Protection Agency, 1998d). Schematic shows the
exposure-dose-response continuum considered in the context of biomarkers
(classified as measures of exposure, effect, and susceptibility) and level of
organization at which toxicity is observed (U. S. Environmental Protection
Agency, 1994a; Schulte, 1989). The model maps the toxicity of perchlorate on
this basis by establishing casual linkage or prognostic correlations of
precursor lesions.
This chapter presents the synthesis of the most relevant data for deriving a revised
quantitative assessment of human health risk for perchlorate. The new data were consistent with
the limited historical characterization and the 1998 EPA assessment in that the anti-thyroid
January 16,2002 7-3 DRAFT-DO NOT QUOTE OR CITE
-------
]
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
effects remain the focus of concern and the key event of its mode of action remained identified as
the inhibition of iodide uptake at the NIS. However, data from the testing strategy allowed a
more comprehensive evaluation of the possible sequelae of the iodide uptake inhibition and its
thyroid-pituitary axis perturbations with respect to other endpoints, notably effects in dams and
their offspring and on nerurodevelopmental, reproductive, and immunotoxicity parameters.
The key event is defined as an empirically observable precursor step that is a necessary
element of the mode of action or is a marker for such an element. This will be discussed in
Section 7.1.1. Section 7.1.2 discusses dosimetric adjustment of effect levels observed in the
laboratory animals to human equivalent exposures (HEE). Choice of the point of departure for
the assessment based on a quantitative consideration of the key event, observed effects, and
weight of the evidence is discussed in Section 7.1.3. Application of factors to account for
uncertainty and variability in the extrapolations required to use the data is discussed in Section
7.1.4. The overall operational derivation is presented in Section 7.1.5, and the assignment of
confidence levels is discussed in Section 7.1.6. Section 7.1.5 also presents a discussion of the
cancer assessment in the context of the RfD. Section 7.2 discusses the inhalation reference
concentration. Susceptible population considerations are discussed in Section 7.1.5.3. Section
7.3 presents a brief summary of the findings.
7.1.1 Key Events and Weight of the Evidence
Results of the testing strategy have established that the critical target tissue for perchlorate
is the thyroid gland, with some remaining concern for adequate characterization of its potential
for immunotoxicity, notably contact hypersensitivity. Changes in thyroid weights, three response
indices of thyroid histopathology (colloid depletion, hypertrophy and hyperplasia), and thyroid
and pituitary hormones were consistently altered across the array of experimental designs
represented by the data base. The developmental and reproductive NOAEL and LOAEL values
were higher than those associated with thyroid toxicity per se.
Figure 7-2 highlights the temporal considerations that have to be superimposed on
evaluation of the data from the various studies in laboratory animals and humans in order to
characterize the anti-thyroid effects from perchlorate exposure. Conceptually, competitive
inhibition of iodide uptake at the NIS by perchlorate is the key event leading to both potential
neurodevelopmental and neoplastic sequelae. The decrement in iodide uptake leads to
January 16, 2002
7-4
DRAFT-DO NOT QUOTE OR CITE
-------
Transient Chronic
Phase Phase
a.
ir -i
Serum TSH (Upregulated)
Thyroidal Iodide
/
10 15
Time (Days)
Serum T4
Subclinical
Disease
Figure 7-2. Schematic of thyroid and pituitary hormone levels with associated
pathology after acute versus chronic dosing with perchlorate. The transient
phase is represented by decreases in thyroidal iodide due to the inhibition
by perchlorate at the NIS with subsequent drop in T4. The transient drops
in T4 can lead to permanent neurodevelopmental sequelae. Once TSH is
upregulated via the hypothalamic-pituitary-thyroid feedback, T4 appears to
be in normal homeostasis but actually can represent subclinical or
undiagnosed disease (hypothyroxinemia). The upregulation of TSH can
result in neoplasia. Normal thyroid tissue is represented in Panel A. Panel
B shows lace-like colloid depletion w hich is more pronounced in subsequent
panels C, D and E. Panels D and E represent hypertrophy and hyperplasia.
1 subsequent drops in T4 (and T3) that can lead to permanent neurodevelopmental deficits.
2 Corroborating evidence for this likely outcome given the mode of action of perchlorate comes
3 from the iodide deficiency literature and recent studies showing that maternal hypothyroxinemia
4 (i.e.. decrements in T4 with or without concomitant increases in TSH) is linked to poor
5 developmental, neuropsychological and cognitive outcomes (Haddow, et al., 1999; Pop et al.,
January 16, 2002
7-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1999; Morreale de Escobar, et al., 2000). It should be noted that medical concern for
hypothyroxinemia remains in the "chronic phase"; i.e., once TSH upregulates to attempt to
regulate the hypothalamic-pituitary-thyroid feedback system back to an apparent homeostasis,
because this stress on the system essentially represents a "subclinical" disease state. Indeed,
adverse outcome in women with hypothyroxinemia per se has been demonstrated because
adversity includes the inability of an organism to respond to additional stressors. The system in
this case, particularly when considered on a population level, would present a diminished
capacity to compensate for other anti-thyroid insults. Since a large percentage of women are
believed to already be hypothyroid, the importance of this effect to women in general, pregnant
women, and fetuses on a population level can not be discounted. Weiss (2000) has noted that
even if the magnitude of effect may be relatively small for most environmental levels, such
neurotoxicity is extremely significant for public health.
Of notable concern, as previously discussed in Chapter 3, is that the developing fetus is
dependent on the mother for its T4 and T3 through parturition, as illustrated in Figure 7-3 for
humans with a similar pattern in rats. During the period illustrated in Figure 7-3, a number of
critical stages in neural development take place, some of which depend on thyroid hormones.
The cell precursors of the brain and spinal cord which compose the central nervous system
(CNS) begin to develop early in embryogenesis through the process called neurulation.
Beginning early in the second week of gestation in rodents (GD9.5 in rats) and the first month of
gestation in humans, specific areas of the CNS begin to form with the neurogenesis and
migration of cells in the forebrain, midbrain, and hindbrain. This sequence of developmental
processes includes proliferation, migration, differentiation, synaptogenesis, apoptosis, and
myelination (Rice and Barone, 2000). As discussed in Chapter 3, thyroid hormones play a role
throughout this process, regulating proliferation, migration, and differentiation. Alterations in
these processes can result in abnormalities of the brain and developmental delays.
The upregulation in TSH in the "chronic phase" (see Figure 6-2) also presents an increased
potential for neoplasia because stimulation of the thyroid to produce more T4 and T3 can result
in hyperplasia. Both the decrement in T4 and T3 and increase in TSH is shown in Figure 7-1 at
the same step along the continuum. Which of these thyroid responses is the most sensitive to
hormone changes has not specifically been studied in the perchlorate testing strategy. As noted
in the analyses of the studies in Chapter 5, there is a considerable degree of overlap among the
January 16, 2002
7-6
DRAFT-DO NOT QUOTE OR CITE
-------
300-.
250-
200-
O)
c
co
t 150
CO
H-
100-
50-
10-
8-
-i 6-
x
l- 4H
2-
-14
-12
-10
3
-8 5
O)
i
-6 ?
-4
-2
0 20
Gestational
Age (weeks)
I : I I i I
30 Birth 2 4 6
Postnatal
Age (weeks)
Figure 7-3. Pattern of change in fetal and neonatal thyroid function parameters during
pregnancy and extrauterine adaptation in the human (from Fisher, 1996).
A similar pattern is thought to exist in the rat (see text for further details).
1 three different diagnoses of thyroid histopathology: colloid depletion, hypertrophy, and
2 hyperplasia.
3 Colloid depletion does appear to be slightly more sensitive across the perchlorate studies.
4 The fact that thyroid follicular colloid depletion is a consistent finding not only across this study,
5 but in rodents in general, would suggest that it is a good indicator of sufficient exposure to inhibit
6 thyroid hormone synthesis. From a physiologic point of view this may be logical and supports
7 the mode-of-action model. If there is any reserve thyroid hormone in the colloid, it is depleted
January 16, 2002
7-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
before serum hormones are altered. Once serum levels are altered, TSH is upregulated and
hypertrophy and hyperplasia are initiated in an attempt by the gland to restore circulating levels
of T4 and T3. The diagnosis of colloid depletion has been reported with a similar compound,
sodium chlorate, in the rat (Hooth et al., 2001), with many other chemicals in the rat, and with
numerous goitrogens and pharmaceutical agents in the mouse. Colloid depletion in association
with hypertrophy and hyperplasia suggests sufficient dose of the compound to inhibit colloid
synthesis and decreases of circulating serum thyroid hormone levels sufficient to stimulate TSH.
Colloid depletion as the most sensitive indicator is most notable in the pups of the 2001
"Effects Study" on GD21 and then immediately post parturition on PND4. Alternatively, as
discussed in Chapter 5, it may have been harder to diagnose hypertrophy and hyperplasia in the
younger (smaller) and growing glands. The BMDL for colloid depletion increased with post-
natal age and by PND21, hyperplasia was also present. In contrast, all three thyroid indices were
present in the PND4 pups of the previous Argus Laboratories, Inc. (1998a) study. This may be
due to the difference in dosing of the dams. The dams in the 1998 study were only dosed during
gestation and, therefore, likely had a greater decrement in thyroid hormones. The dams in the
2001 study were dosed for two weeks during cohabitation, sufficient time as evidenced in the
data described in Chapter 6, for upregulation of NIS to compensate.
Other studies indicate that whichever index is most sensitive could be dependent on dose
spacing in the study, age of animals on test, and sacrifice time point. For example, hyperplasia
was the most sensitive of the three in the P2-generation adults (19 week F1 -generation pups) and
these same pups developed thyroid adenomas.
The proposed mode of action mapped in Figure 7-1 is supported by correlations between
thyroid hormones and TSH and between thyroid hormones or TSH and an objective measure of
lumen size from laboratory animals exposed to ammonium perchlorate. There were positive
correlations between T3 and T4, and negative correlations between either T3 and T4 and TSH, as
expected based on the mode of action model (Appendix 7A). The positive correlation between
TSH and decreased follicular lumen size and negative correlation between T4 or T3 and
decreased follicular lumen size similarly support the proposed model (Appendix 7A). Some of
the correlations used in the 1998 assessment were precluded due to the limited severity scoring
system used by the PWG.
January 16, 2002
7-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Additional support for the mode of action comes from data that now allow the linkage of
both neurodevelopmental and neoplastic sequelae into the model. These definitive data were not
available prior to the 1997 perchlorate testing strategy and especially not before the most recent
studies recommended by the 1999 external peer review. The repeat of observed effects on the
motor activity and brain morphometry results by new studies allowed definitive determination
that perchlorate exposure poses a neurodevelopmental hazard.
Repeatability and variability in statistics, sometimes a concern for evaluation of behavioral
assays (Cory-Slechta et al., 2001) were addressed by the Bayesian approach employed for the
motor activity analysis (Dunson, 2001a) that showed remarkable reproducibility between the two
studies despite the deficits previously noted for the Argus Research Laboratories, Inc. (1998a)
study. The effects on the size of the corpus callosum measurements were also repeated, and
effects on additional brain regions identified. The new data were subject to a more rigorous
statistical analysis than in 1998. The profile analysis described in Chapter 5 required that all
areas of the brain measured were altered in a dose-dependent fashion and effects were again
demonstrated not only in the corpus callosum but other brain regions as well (Geller, 200Id).
Likewise the neoplastic potential for perchlorate that had been demonstrated only at high
doses in historical studies was confirmed at lower doses by the thyroid adenomas reported by the
PWG (Wolf, 2000; 2001) for the F1-generation pups at 19 weeks (P2 parents) from the
two-generation reproductive study (Argus Laboratories, Inc., 1999). Consistent with the
proposed mode-of-action model, the anti-thyroid effects leading to neoplasia are likely to be via
the non-linear mechanism described above. The genotoxicity battery established that perchlorate
is not directly damaging to DNA.
Thus, the key event for the anti-thyroid effects of perchlorate is its perturbation of the
hypothalamic-pituitary-thyroid axis by competitive inhibition of iodide uptake at the NIS. The
evidence for this effect is built upon the observation of consistent changes across a range of
experimental designs, including various species. These changes demonstrate effects on thyroid
and pituitary hormones, increases in thyroid weight, and increases in three different diagnoses of
thyroid histopathology (colloid depletion, hypertrophy, and hyperplasia). In addition,
corresponding neurodevelopmental (motor activity and brain morphometry) and neoplastic
outcomes were observed in special assays; these outcomes are also consistent with the proposed
January 16, 2002
7-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
mode of action and provide further evidence to confirm that the perturbation of the thyroid
hormone economy should be viewed as adverse.
Due to the age and time-dependent nature of the critical effect, no one principal study is
being chosen for this derivation. Instead, a weight-of-the-evidence approach will be taken to
arrive at a point of departure in Section 7.1.3.
7.1.2 Dosimetric Adjustment of Exposures Associated with Effect Levels
Adjustments for interspecies differences in the internal dose delivered to target tissues
should be made before an evaluation of the data array for valid comparisons across endpoints
(U.S. Environmental Protection Agency, 1994). Based on the mode of action and the available
PBPK model structures, two dose metrics were considered to describe the biologically effective
dose for perchlorate: (1) the area under the curve (AUC) for perchlorate in the serum associated
with drinking water exposures and (2) the percent of iodide uptake inhibition in the thyroid.
These correspond to the different exposure components along the exposure-dose-response
continuum in the mode-of-action model (Figure 7-1).
As described in Chapter 6, the serum perchlorate AUC was developed as the first dose
metric based on data in rats and humans after drinking water exposures. To predict the
"transient" phase of initial iodide inhibition in the rat, i.e., before upregulation of the NIS or
increases in TSH, the second dose metric was based on RAIU measurements made in adult male
rats dosed with perchlorate by iv two hours prior to an iv dose of radiolabled iodide. Table 7-1
presents the human equivalent exposures (HEE) estimates calculated using the PBPK models for
serum perchlorate AUC as the dose metric. Table 7-2 shows the ratios for this same dose metric
that can be applied in the parallelogram approach to arrive at estimates for different life stages
used to observe effects in the different experimental endpoints. Fetal rat predictions were based
on data developed for GD21. Neonatal rat predictions were based on data for PND10. This
approach was taken since PBPK models for human pregnancy and lactation do not exist for
perchlorate distribution. The calculation using the ratios approach is described in Chapter 6.
The resultant adult HEE values for the different life stages of the rat experiments are shown in
Table 7-3.
It can be observed in the tables in Merrill (200 le) that the pregnant and lactating rats have
significantly higher average serum perchlorate concentrations at the lowest drinking water dose
January 16, 2002 7-10 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-1. PBPK-MODEL CALCULATED HUMAN EQUIVALENT EXPOSURES
(HEE) TO VARIOUS EXPERIMENTAL DOSES IN THE MALE RAT FOR 15 AND
70 KG HUMAN BASED ON PERCHLORATE AREA UNDER THE CURVE (AUC) IN
SERUM OR THYROID AS THE DOSE METRIC (Merrill, 2001e)
Adult Male Rat
DW* Dose
(mg/kg-day)
Human IS kg HEE
(mg/kg-day) based
on serumb
AUC
Human 70 kg HEE
(mg/kg-day) based
on serumb
AUC
Human 15 kg HEE
(mg/kg-day) based
thyroidb
AUC
Human 70 kg HEE
(mg/kg-day) based
on thyroidb
AUC
0.010
0.030
0.021
0.0002
0.0001
0.1
0.145
0.100
0.002
0.001
1.0
0.745
0.505
0.008
0.006
3.0
2.05
1.35
0.052
0.035
5.0
3.35
2.25
0.145
0.098
10.0
6.75
4.45
0.725
0.460
30.0
20.3
13.2
163.0
110.0
100.0
65.0
43.8
490.0
330.0
'DW = drinking water.
bCalculated from PBPK-derived rat AUC(s) at steady state between 240 and 264 hrs during DW exposure, using
upregulated Vmaxv_TP values from (Merrill, 2001e: Table 1).
TABLE 7-2. RATIO OF PBPK-DERIVED PERCHLORATE AREA UNDER THE
CURVE (AUC) SERUM CONCENTRATIONS IN DRINKING WATER FOR
VARIOUS EXPERIMENTAL LIFE STAGES (Merrill, 2001e)
Rat DW*
Dose
(mg/kg-day)
Male Rat:
Pregnant
Rat
Male Rat:
Lactating
Rat
Male Rat:
Fetal Rat
Male Rat:
Neonate Rat
Pregnant
Rat: Fetal
Rat
Lactating
Rat: Neonate
Rat
0.01
0.63
0.58
1.44
1.16
2.28
1.99
0.1
0.73
0.54
1.06
0.85
1.46
1.56
1.0
0.90
0.84
1.44
1.01
1.61
1.20
3.0
0.94
0.95
1.67
1.71
1.77
1.80
5.0
0.95
0.98
1.74
2.14
1.82
2.18
10.0
0.96
1.01
1.80
2.70
1.87
2.69
30.0
0.97
1.02
1.84
3.33
1.90
3.26
100.0
0.97
1.03
1.85
3.65
1.92
3.55
"DW = drinking water.
January 16, 2002 7-11 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
TABLE 7-3. PBPK-MODEL CALCULATED HUMAN EQUIVALENT EXPOSURES
(HEE) TO VARIOUS EXPERIMENTAL LIFE STAGES IN THE RAT USING SERUM
PERCHLORATE AREA UNDER THE CURVE (AUC) AS THE DOSE METRIC
Dose
(mg/kg-day)
Human Equivalent Exposure* (mg/kg-day)
Adult Male Rat
Pregnant Rat
Fetal Rat
Lactating Rat
Neonate Rat
0.01
0.02
0.01
0.03
0.01
0.02
0.1
0.10
0.07
0.10
0.05
0.08
1.0
0.51
0.46
0.73
0.43
0.52
3.0
1.35
1.3
2.3
1.3
2.4
5.0
2.25
2.14
3.92
2.20
4.82
10.0
4.4
4.22
7.9
4.4
11.9
30.0
13.2
12.8
24.3
13.5
43.95
100.0
43.8
42.5
81.0
45.11
160.0
"Based on predicting the area under the curve in the blood (AUCB) using the human PBPK model that achieves
an equivalent degree to that simulated for the rat experimental regimen associated at different life stages. See
Tables 7-1 and 7-2 and Chapter 6 for explanation of calculation.
(0.01 mg/kg-day). This is likely due to increased binding in the serum (Merrill, 200 le). It has
been shown that the estrus cycle affects the concentration of binding proteins within the blood.
Thyroxine, which is displaced from plasma proteins by perchlorate, is bound to a greater extent
in the pregnant rat (lino and Greer, 1960). It follows then that perchlorate would also be bound
to a greater extent during pregnancy and possibly lactation. Since serum binding affects only the
low doses, it is reasonable that the higher doses (1.0 through 100 mg/kg-day) would be similar
across the male, pregnant and lactating rats (Merrill, 2001 e).
Tables 7-4 through 7-7 are a comparable set of tables but are based on using thyroid uptake
inhibition as the dose metric. Table 7-5 shows the percent of iodide uptake inhibition predicted
at each dose for the various life stages used in the various laboratory rat experiments.
7.1.2.1 Choice of Dose Metric
Because developmental effects are of concern, an argument could be made that peak—and
not AUC—is the appropriate dose metric with the rationale that any transient dose could be
responsible for permanent deficits. However, the AUC values, as opposed to peak
January 16, 2002
7-12
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-4. PBPK-MODEL CALCULATED HUMAN EQUIVALENT
EXPOSURES (HEE) TO VARIOUS EXPERIMENTAL DOSES IN THE ADULT
MALE RAT FOR 15 AND 70 KG HUMAN BASED ON % IODIDE UPTAKE
INHIBITION IN THE THYROID
Rat iv
Dose (mg/kg)
Adult male rat inhibition
at 2-hr post iv dose
Human 15 kg HEE
(mg/kg-day)
Human 70 kg HEE
(mg/kg-day)
0.01
1.5%
0.006
0.004
0.1
16.3%
0.075
0.048
1.0
74.5%
1.5
0.9
3.0
90.0%
4.8
2.7
5.0
93.5%
8.0
4.9
10.0
96.2%
16.0
9.0
30.0
98.1%
35.0
19.3
100.0
98.7%
50.0
33.0
TABLE 7-5. PBPK-MODEL PREDICTED % INHIBITION OF IODIDE UPTAKE IN
THE THYROID"
Rat DW b Dose
(mg/kg-day)
Adult
Male Rat
Pregnant Rat
Fetal Ratc
Lactating
Rat"
Neonate
Ratc,d
70 kg
Human
0.01
1.5%
3.2%
-129.1%
0.5%
0.4%
2.8%
0.1
16.3%
30.1%
27.9%
5.3%
1.3%
23.7%
1.0
74.5%
88.7%
81.2%
62.9%
3.0%
80.2%
3.0
90.0%
93.8%
90.3%
92.8%
3.3%
92.3%
5.0
93.5%
97.0%
90.4%
95.8%
3.1%
95.2%
10.0
96.2%
97.9%
97.9%
97.6%
3.8%
97.4%
30.0
98.1%
98.6%
98.9%
98.5%
6.1%
98.9%
100.0
98.7%
98.8%
99.2%
98.8%
13.4%
99.4%
"Based on iv administration to rat and drinking water in human.
bDW = drinking water
Values for these tissues not validated versus data.
dAll calculations are for PND10 in lactating and neonatal rat.
January 16, 2002
7-13
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7-6. RATIOS OF PBPK-DERIVED % IODIDE UPTAKE INHIBITION IN
DRINKING WATER FOR VARIOUS EXPERIMENTAL LIFE STAGES"
Rat DWb Dose Male Rat:
(mg/kg-day) Pregnant Rat
Male Rat:
Lactating Rat
Male Rat:
Fetal Ratc
Male Rat:
Neonate Ratc
Pregnant Rat:
Fetal Rat
Lactating Rat
Neonate Rat"
0.01
0.48
3.24
-0.01
4.02
-0.02
1.2
0.1
0.54
3.06
0.59
12.75
1.08
4.2
1.0
0.84
1.18
0.92
24.53
1.09
20.7
3.0
0.96
0.97
1.00
27.49
1.04
28.4
5.0
0.96
0.98
1.03
30.45
1.07
31.2
10.0
0.98
0.99
0.98
25.61
1.00
26.0
30.0
0.99
1.00
0.99
16.06
1.00
16.1
100.0
1.00
1.00
1.00
1.37
1.00
7.4
"Inhibition in human was PBPK-derived from 2 wks C104"-exposure in drinking water (DW); all rat values
simulated from an iv dose.
bDW = drinking water
cModel predicted in fetal and neonate rats not validated with data.
dAll calculations are for PND10 in lactating and neonatal rat.
TABLE 7-7. PBPK-MODEL CALCULATED HUMAN EQUIVALENT EXPOSURES
(HEE) TO VARIOUS EXPERIMENTAL LIFE STAGES IN THE RAT USING %
IODIDE UPTAKE INHIBITION IN THE THYROID AS THE DOSE METRIC
Human Equivalent Exposure" (mg/kg-day)
Dose
(mg/kg-day)
Adult Male Rat
Pregnant Rat
Fetal Rat
Lactating Rat
Neonate Rat
0.01
0.004
0.002
—
0.01
0.02
0.1
0.048
0.026
0.03
0.15
0.61
1.0
0.90
0.756
0.83
1.06
22.05
3.0
2.7
0.259
2.70
2.62
74.2
5.0
4.9
4.70
5.05
4.80
149.2
10.0
9.0
8.82
8.82
8.91
230.5
30.0
19.3
19.1
19.1
19.3
309.96
100.0
33.0
33.0
33.0
33.0
33.0
"Based on predicting the % iodide uptake in the thyroid using the human PBPK model that achieves an
equivalent degree to that simulated for the rat experimental regimen associated at different life stages.
See Tables 7-4 and 7-6 and text for explanation of calculation.
January 16, 2002
7-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
concentrations, were used based on the assumption that these dose metrics would represent an
averaging of the serum and thyroid perchlorate concentrations and would be better correlated
with the inhibition effect on iodide uptake. The correlation was shown to be good between the
AUC and the degree of inhibition (Figures 6-47 through 6-50). Further, due to the rapid phase of
distribution after an iv dose, measurement of concentrations are very difficult to attain
experimentally and are more variable. Using simulated peak concentrations after iv injections is
potentially problematic due to the inexact modeling of the actual distribution of dose in the
tail-vein volume and the exact time of mixing in the whole blood compartment (Merrill, 200 le).
It was also observed by EPA that the ratios for peak perchlorate serum values (Merrill, 200le:
Table 6) were in good agreement with those for the perchlorate serum AUC and that the serum
AUC were slightly more conservative if different at all.
Merrill (200le) expressed concern regarding the thyroid values in neonates and fetuses
because these values were not validated against experimental data. Fetal and neonatal thyroid
were never actually analyzed for perchlorate concentration. In the case of the fetus, kinetic
parameters were determined by fitting model simulations of fetal thyroid concentration to
available iodide data and assuming that the perchlorate:iodide ratio would be similar to that of
the mother. In the case of the neonatal rat, no data were available for thyroid concentrations for
either perchlorate or iodide. Thus, model predictions were based on allometrically scaling
maternal parameters for thyroid uptake. It was the opinion of the AFRL/HEST authors that while
the thyroid parameters in the fetus and neonatal rat were highly informative, they should not be
used in the formal risk assessment (Merrill, 200le). EPA concurs with these considerations and
recommendation.
In general, the models were believed to provide a good description of perchlorate and
iodide disposition in the blood. Using the models to describe dose metrics in the thyroid was
viewed as less reliable due to assumptions regarding parameters and the lack of experimental
data for validation. The models were able to successfully describe serum perchlorate and iodide
concentrations for both acute (based on iv doses) and chronic drinking water in the adult male,
pregnant, neonatal and fetal rat, and greater confidence can be afforded these predictions
(Merrill, 200le).
Tables 7-3 and 7-7 demonstrate good correspondence in the HEE estimates predicted for
both dose metrics at the lower doses for the lactating and neonatal rats, but not for the male adult,
January 16, 2002
7-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
pregnant or fetal rats where there is an order of magnitude difference. The iodide inhibition
metric predicts a 10-fold lower HEE in both the adult male and pregnant dam when compared to
the HEE estimated based on the serum AUC. The fetal rat value for iodide inhibition was
viewed as unreliable for the reasons stated above. All of the factors influencing this disparity are
not fully appreciated at this time but can reasonably be ascribed to uncertainty in the thyroid
descriptions that were not validated with experimental data, and will require additional studies to
characterize accurately. For these reasons, the adjustment factor to arrive at an HEE estimate
was based on perchlorate serum AUC as the dose metric.
7.1.3 Point-of-Departure Analysis
Various statistical procedures were used for each of the different outcome measures for the
various endpoints described in Chapter 5. The weight-of-evidence approach herein relies on the
results, and the details on the statistical analyses are provided in Chapter 5 and associated
memoranda from EPA and NTEHS scientists. In general, benchmark dose analysis was used for
the thyroid histopathology because the EPA advocates the use of quantitative dose-response
modeling to diminish the influence of dose-spacing, sample size, and variability on the NOAEL
designation (Crump et al., 1995). Likewise, ANOVA was used to evaluate the thyroid and
pituitary hormone data (Crofton and Marcus, 2001) although benchmark analyses were also
performed as a comparison (Geller, 2001c). The 1998 benchmark analyses for the hormone data
from the previous set of studies (Geller, 1998a) is provided in Appendix 7B.
Specific Bayesian statistical analyses were employed for the motor activity data and for
evaluating the significance of the tumors in the 19-week old F1-generation adult rats (Dunson,
2001 a,b). Another specific statistical approach, profile analysis, was used to evaluate the brain
morphometry effects (Geller, 200Id).
Several studies suggest 0.01 mg/kg-day as the exposure dose that is a level of concern for
the adverse effects of perchlorate. The first is the profile analysis on brain morphometry effects
in PND21 pups in the "Effects Study" (Argus Research Laboratories, Inc., 2001) which
demonstrated a dose-dependent and significant effect on the size of the corpus callosum and
other brain regions. Statistically significant changes were also demonstrated in the PND9 pups.
This effect repeated effects on brain morphometry observed in the previous neurodevelopmental
study (Argus Research Laboratories, Inc. 1998a) that were a noted concern to the EPA in the
January 16, 2002
7-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1998 risk assessment. Changes in the corpus callosum at a later time point on PND82 were also
observed in that previous study.
An increase in the corpus callosum plausibly represents a delay in developing brain
structures since this area is known to increase in size and then decrease later during development.
Neurodevelopmental toxicity suggestive of delays was also demonstrated by effects on motor
activity in both the Argus Research Laboratories, Inc. (1998a) and repeated in the Bekkedal et al.
(2000) study. The increases in motor activity represent activity that should have subsided by
these test dates. A type of hyperactivity has been noted in monkeys exposed to PCBs (Rice,
2000).
These effects on brain morphometry and motor activity are of particular concern because
the relative sensitivity of laboratory animal assays to adequately characterize the types of deficits
related to maternal hypothyroxinemia in large population studies is unknown (Morreale de
Escobar, 2000; Haddow et al., 1999; Pop, 1999). Screening neurodevelopmental studies may not
have the power to ascertain neurological effects that might result from small changes in the
thyroid-pituitary hormone economy. As pointed out by Crofton (1998j), the sensitivity of animal
models used to explore the role of thyroid hormones in neural development is currently
equivocal. Most of the data collected and published to date were with high doses of thyrotoxic
chemicals (e.a., methimazole, propylthiouracil) or with thyroidectomy. It is not known whether
the available tests are capable of detecting more subtle changes in nervous system development.
An analysis presented by Crofton (1998j) suggested that measurements of nervous system
development are less sensitive than measurements of T4. Two reasons for this relationship were
presented. First, the brain may be protected from perturbations in circulating concentrations of
T4, as demonstrated by upregulation of deiodinases in brain tissue that compensate for very large
decreases in circulating T4. The second reason, and one for concern in the context development
of this model, is that currently available testing methods, particularly screening methods, may not
be sufficiently sensitive. Recent data suggest that the battery is insensitive to alterations in
thyroid hormones during development (Goldey, 1995a,b).
The 0.01 mg/kg-day dosage as a level for concern was also supported by thyroid
histopathology in the database. Changes in colloid depletion observed on PND4 in both the 1998
neurodevelopmental study (Argus Research Laboratories, Inc., 1998a) and the newer 2001
"Effects Study" (Argus Research Laboratories, Inc. 2001) were demonstrated. The BMDL
January 16, 2002
7-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
estimated for those studies on PND4 was 0.33 mg/kg-day, but an estimate of 0.009 mg/kg-day is
also obtained with a model demonstrating adequate fit to the data. The BMDL for colloid
depletion in pups on GD21 was 0.12 mg/kg-day, but for female pups alone on GD21 was 0.04
mg/kg-day. The BMDL estimated for thyroid hypertrophy in weanling pups from the two-
generation study (Argus Research Laboratories, Inc., 1999) was 0.06 mg/kg-day. Of notable
concern to this consideration was that the BMDL estimates decreased with duration in the 90-day
study. The BMDL estimates for colloid depletion were 0.28 and 0.03 mg/kg-day at the 14-day
and 90-day time points in the Springbom Laboratories, Inc. (1998) study. The BMDL estimates
for hypertrophy were 0.017 and 0.008 mg/kg-day at the 14-day and 90-day time points. This
effect of duration was of concern as it was also evident by the observation of tumors in the
F1-generation adults at 19 weeks. Both observations suggest concern that duration may
recalibrate either the homeostatic interactions of the hypothalamic-pituitary-feedback system or
the cellular sensitivity and demand for the thyroid hormones.
The thyroid hormone data in a number of studies also designated 0.01 mg/kg-day as a
LOAEL. Levels of T4 were significantly decreased and TSH levels statistically increased at this
dosage in the dams on GD21 in the same study as the significant brain morphometry
measurements in the PND21 pups (Argus Research Laboratories, Inc. 2001), revealing no
NOAEL for hypothyroidism in the dams. The pups in that study were also affected at
0.01 mg/kg-day. Effects on T3 occurred at GD21, PND5, and PND9 at this dosage. The
0.01 mg/kg-day dose was the LOAEL for effects on T4 and TSH at PND21 in the male pups and
for TSH in both sexes at PND9 as well. This same dose (0.01 mg/kg-day) was also the LOAEL
for decreases in T4 and increases in TSH at the 14-day and 90-day time points in the 90-day
study (Springbom Laboratories, Inc., 1998).
The ANOVA estimates for hormone data were used to characterize this effect after serious
consideration. While in clinical studies a normal range typically is defined by a control healthy
population, the ANOVA approach is an equally valid approach in that a statistically significant
value represents a shift in the mean for the population. The control group defines the range for
the unexposed, presumably healthy population, and statistically significant differences indicate
that the mean for an exposed group is outside of that normal range. Circadian fluctuations are
addressed because the same fluctuations in the control population occur as in the exposed
population. A small shift in the mean of a population can have significant consequences to
January 16, 2002
7-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
individuals in the tails of the distributions of those populations. Indeed, such an evaluation
underlies the basis for the blood lead level used in the National Ambient Air Quality Standard
(Davis and Elias, 1996) and has been noted as an important consideration for neurotoxicity
(Weiss, 2000).
The notion that continuous data should be considered in the context of the specific dose-
response rather than to a prioro categories defined outside of the data under analysis is supported
in the benchmark dose literature. Murrell et al. (1998) point out that a continuous quantity
measurement such as the hormone data should be scaled by the range from background response
level to maximum response level (for increasing response functions). The authors go on to note
that it is a biological reality that, whatever the mechanism of effect of the toxicant, there is some
dose level beyond which no further change in response is seen or is theoretically feasible.
In general, there is some type of limitation or saturation phenomenon that occurs at high enough
doses (e.g., in the saturation of the symporter capacity, as suggested by the modeling effort in
Chapter 6 and the data of Chow and Woodbury [1970] and of Meyer [1998]).
An analogy to the case of quantal data for which an effect is defined as a probability metric
in which the response reaches a maximum at one, is, that for continuous measures, the extra
effect can be defined as the change in effect from background standardized by the total range of
response (Murrell et al., 1998). The total response range is not necessarily the response range of
the observed responses in a study; rather, it is defined by a determination of the minimum and
maximum possible responses according to, for example, a model equation fitted to the data as in
the case of benchmark analyses. In all BMD analyses, however, the hormone BMDL estimates
were shown to be extremely low (Geller, 1998a; Geller, 2001c). This may not necessarily be
surprising given that hormones are operative at low doses by definition, but corresponding
changes in thyroid histopathology were more consistent with the ANOVA estimates.
Finally, the NOAEL for immunotoxicity suggested by the dermal contact hypersensitivity
assay at 0.02 mg/kg-day can be viewed as supportive, especially since deficiencies in this study
raise concern for the characterization and because a LOAEL for the effect was demonstrated at
0.06 mg/kg-day.
January 16, 2002
7-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
7.1.4 Application of Uncertainty Factors
The types of uncertainty factors (UF) applied for various extrapolations required to arrive at
a reference dose were discussed in Chapter 3. Figure 7-4 illustrates schematically that the
interspecies and intraspecies UFs embody attributes of both uncertainty and variability. A factor
for variability across humans typically is applied to account for potentially susceptible portions of
the population. As shown in Figure 7-5 (Jarabek, 1995b), both of these factors typically are
broken into components of approximately three each for pharmacokinetics (toxicokinetics) and
pharmacodynamic (toxicodynamic) processes. This scheme is consistent with that used by the
World Health Organization (WHO) (Jarabek, 1995b).
Interspecies
III
Rat to Human
Intrahuman
II
Variability Across
Humans
Figure 7-4. Consideration of uncertainty and variability influence interspecies and
intrahuman extrapolation.
There were a total of four (4) uncertainty factors applied in this derivation, resulting in a
composite factor of 300. The partial factors of 3 represent "halving" of each UF that is believed
to be an upper bound on a lognormal distribution; i.e., 10°5, so that multiplication of the various
partial factors results in a composite of 100 (U.S. Environmental Protection Agency, 1994).
A 3-fold factor for intraspecies variability was retained due to the variability observed in
the data and PBPK modeling for the adult humans and because these subjects do not represent
kinetic data for the potentially susceptible populations of the hypothyroid or hypothyroxinemic
pregnant women and their fetuses. There was also uncertainty in the parallelogram approach to
extending the adult structure to address different life stages. These uncertainties might be
January 16, 2002
7-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Pharmacokinetic Pharmacodynamic
Parameters and Processes Parameters and Processes
Figure 7-5. Schematic of uncertainty factor components incorporated into exposure-dose-
response characterization for interspecies and intrahuman extrapolations
(Jarabek, 1995b).
mitigated by further development of pregnancy and lactation models or the models might be
further validated with radionucleide data using a parallelogram approach between perchlorate
and iodide as described in Chapter 6. This reduced factor was a point of considerable debate,
especially given the concern over the animal neurodevelopmental assays for adequately
characterizing neuropsychological development deficits in susceptible populations. However, it
was also discussed that the UF values are not entirely independent; e.g., aspects underlying the
duration extrapolation also might underlay the intrahuman UF (Jarabek, 1995b).
The interspecies factor was omitted due to general confidence that the extrapolation based
on perchlorate distribution (and on iodide inhibition by perchlorate at lower doses) was
accurately characterized by the PBPK modeling effort described in Chapter 6. Concern for
eliminating this factor was again considered in the context of the lack of independence with other
applied UF. The concern that the HEE was not based on iodide inhibition but rather the serum
January 16, 2002
7-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
perchlorate AUC was assuaged somewhat by the correlations that demonstrated a close
relationship between these two measures.
A full 10-fold factor was applied to extrapolate the LOAEL for the brain morphometry,
thyroid histopathology, and hormone changes observed at the 0.01 mg/kg-day level. Designating
these changes to be adverse is consistent with the proposed mode of action and existing Agency
guidance and procedures. The neurotoxicity assessment guidelines (U.S. EPA, 1998a) specify
changes in brain structure as adverse. The OPPTS has used thyroid hormone changes to
designate effect levels. Finally, the shallow slope of the response curve at these lower levels
suggested that a full factor should be applied.
A 3-fold factor for duration was applied due to the concern for the biological importance of
the statistically significant increase in tumors in the F1-generation pups at 19 weeks (P2, second
parental generation). The occurrence of these tumors with a dramatically reduced latency and
with a significance in incidence greater than the NTP historical data (Dunson, 2001b) for thyroid
tumors in this strain of rat was reason for concern. As discussed earlier, the concerns were that
this observation represented the potential for in utero programming; and that the decrease in the
NOAEL/LOAEL estimates for hormone perturbations and histopathology between the 14-day
and 90-day time points represented a recalibration of the regulatory feedback system or changes
in cellular sensitivity and demand for thyroid hormones with extended exposures. This factor
can also be viewed as part of a data base deficiency because there are no long-term bioassays of
perchlorate with contemporary design and data quality. While the original strategy aimed at
determining a NOAEL for thyroid histopathology as a precursor lesion to tumors in the 90-day
study, this finding in the F1-generation cannot be ignored, especially in light of an emerging
appreciation of findings suggesting a phenomenon known as in utero imprinting with endocrine
disruption (Prins et al., 2001; Phillips et al., 1998; Seckl, 1997). Thus, in utero disruption of
thyroid hormones in the developing fetus may predispose the developing neonate and adult to
future environmental insults to the thyroid gland by making the fetus more sensitive. Weiss
(2000) has noted that changes in brain functions occur throughout life and some consequences of
early damage may not even emerge until advanced age. This could be exacerbated if
environmental insults to the thyroid were to be continued throughout life.
The potential for perchlorate to cause immunotoxicity remains a concern so that a 3-fold
factor was applied for the database insufficiency. New studies based on recommendations at the
January 16, 2002
7-22
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1999 external peer review had some deficiencies and reinforced concern about the lack of an
accurate characterization of this endpoint.
7.1.5 Operational Derivation of the Reference Dose
The HEE for the neonatal rat corresponding to brain morphometry and hormone changes
observed in the PND21 pups (also the PND9 pups) at the 0.01 mg/kg-day dosage would be 0.02
mg/kg-day (Table 7-3). However, because the dams on GD21 were shown to be hypothyroid
(with statistically-significant decreases in T4 and increases in TSH) at this same dosage, and
because the temporal windows underlying the neonatal brain morphometry effects are unknown,
and because the brain morphometry effects may have occurred in utero due to the dams'
hormone deficiency, the HEE estimate for dams of 0.01 mg/kg-day was chosen as the operational
derivation. It was noted that this choice was not as conservative as using the HEE for iodide
inhibition in the dams (0.002 mg/kg-day), but it was viewed as more accurate given the concerns
for the reliability of the thyroid estimates.
According to Dollarhide (1998), who spoke with Argus laboratory on behalf of the sponsor
(PSG), the reported doses were of ammonium perchlorate and not the anion itself. Thus, an
adjustment for percent of the molecular weight of the salt from ammonium (15.35%) must also
be made. Further, because the analytical methods measure the anion concentration in
environmental samples, this is the appropriate expression for the RfD to use while making valid
comparisons for risk characterization. Thus, the derivation for an RfD for the perchlorate anion
as itself is as follows:
Note that the appropriate adjustment for any salt of perchlorate (e.g., adjustment by a factor of
0.72 for potassium perchlorate) should be made when evaluating toxicity data for similar
assessment activities.
It is critically important to distinguish the proposed RfD from any guidance value that may
result. An RfD would be only one step in the future regulatory process of determining, based on
a variety of elements, whether a drinking water standard for perchlorate is appropriate. As with
any draft EPA assessment containing a quantitative risk value, that risk estimate is also draft and
0.01 mg/kg-day x 0.85 / 300 = 0.00003 mg/kg-day.
(7-1)
January 16, 2002
7-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
should be construed at this stage to represent Agency policy. The units for an RfD are mg/kg-
day. Conversion of an RfD to a drinking water equivalent level (DWEL) is based on adjusting
by body weight (kg) and drinking water consumption (L) to arrive at a level expressed in units of
mg/L (ppb). Derivation of a maximum contaminant level goal (MCLG) from a DWEL by the
OW typically involves the use of a relative source contribution (RSC) factor to account for non-
water sources of exposures such as those discussed in Chapters 8 and 9.
Because the effect is viewed to be the result of neurodevelopmental deficits resulting from
the hypothyroid or hypothyroxinemic state induced by the mother's exposure, and because
developmental neurotoxicity may emerge later in the life or be exacerbated later in life,
conversion factors for the adult of 70 kg body weight and 2 L of water per day are considered
appropriate. Recent guidance from the OW in its Methodology for Deriving Ambient Water
Quality Criteria for the Protection of Human Health (U.S. Environmental Protection Agency,
2000) provides a decision flow chart for derivation of the RSC and recommends 80% as a ceiling
and 20% as the floor for this factor when data are adequate to estimate sources of exposure.
When data are not adequate to estimate other anticipated exposures, OW recommends a default
RSC of 20%. (U.S. Environmental Protection Agency, 2000: Chapter 4, Section 4.2.2.4 on
apportionment decisions). EPA does not recommend that high-end intakes be assumed for every
exposure source since the combination may not be representative of any actually exposed
population or individual.
A hypothetical adjustment of the 0.00003 mg/kg-day RfD by 70 kg and 2 L would thereby
result in a DWEL of 1 ug/L (ppb) and application of an RSC between 0.2 to 0.8 would thereby
result in an MCLG in the range of 0.2 to 0.8 ug/L (ppb). These values are in the range of current
analytical capabilities. As discussed in Chapter 1, improvements to the analytical methods on the
near horizon or expected to be published this spring could result in minimum reporting limits in
this range and lower (Yates, 2001).
Concern is often expressed in the regulatory arena for the potential added susceptibility of
children in developing DWEL estimates based on different conversion factors (15 kg and 1 L).
Consequently, the EPA asked for additional PBPK simulations to help inform this dialogue.
As shown in Table 7-1, the HEE estimates for a 15 kg human for serum perchlorate AUC can be
as great as two-fold higher than those predicted for the 70 kg human due to differences in
distribution volumes and excretion. Thus, if the 15 kg and 1 L values are used to convert this 2-
January 16, 2002
7-24
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
fold higher HEE value in an analogous derivation to the adult RfD derivation and DWEL
calculation above, an estimate of 1 ppb that is equivalent to the adult conversion results.
7.1.5.1 Comparison with Derivation Considering Human Data
It is important to evaluate this derivation in context with the evidence from the available
and relevant human data. As described in Chapter 4, the EPA felt that both the observational
epidemiological and the human clinical studies have significant scientific and technical
limitations that preclude their use as the basis for a quantitative dose-response assessment. The
clinical study subject attributes (euthyroid adults) and study design issues (sample size, RAIU
time points, etc.) made these data less reliable than the laboratory animal toxicological data to
ascertain effect levels for the basis of an RfD derivation. In addition, on December 14, 2001,
after internal peer review of this document, the Agency articulated its interim policy on the use of
third-party studies submitted by regulated entities (U.S. Environmental Protection Agency,
2001c). For these purposes, EPA is considering "third party studies" as studies that have not
been conducted or funded by a federal agency pursuant to regulations that protect human
subjects. Under the interim policy, the Agency will not consider or rely on any such human
studies (third-party studies involving deliberate exposure of human subjects when used to
identify or quantify toxic endpoints such as those submitted to establish a NOAEL or NOEL for
systemic toxicity of pesticides) in its regulatory decision making, whether previously or newly
submitted. Some of the clinical studies contained in this database fall in this category of studies
not to be considered. However, the scientific and technical strengths and weaknesses of these
studies were described before this Agency policy was articulated. Therefore, because of the
scientific shortcomings of these studies, they will not be used as "principal studies" in the
derivation of an RfD. The ethical issues surrounding the conduct of these studies or their use for
regulatory purposes in light of the Agency's interim policy will not be discussed in this
document. The Agency is requesting that the National Academy of Sciences conduct an
expeditious review of the complex scientific and ethical issues posed by EPA's possible use of
third-party studies which intentionally dose human subjects with toxicants to identify or quantify
their effects.
These issues not withstanding, a dose of 0.007 mg/kg-day has been suggested by some
authors in an abstract (Greer et al., 2000) to be a NOAEL estimate. This was based on an
January 16, 2002
7-25
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
average 6.2 % decrease relative to baseline of RAIU measured on Day 14 of exposure to seven
subjects at the 8-hour time point (unpublished data presented in Merrill, 2001a; Attachment #7).
The values for RAIU ranged from a 38.6% decrease in a 34-year old female to a 27.9% increase
in a 49-year old female at that dosage.
Prior to the articulation of the Agency's interim policy, the Agency had conducted a
comparison of its reference dose derivation considering the results of the study described above,
which falls within the category of a "third-party study" described by the authors as demonstrating
a NOAEL in humans. If this study were to be considered in lieu of the laboratory animal data
and PBPK modeling, the following would need to be considered. The seven subjects (six
females and one male) were euthyroid and ranged in age from 18 to 49. Because this is a limited
data set (small sample size), with noted variability and because of relevance to the elderly
woman, cardiac risk patient, hypothyroid or hypothyroxinemic pregnant woman, or fetus as the
susceptible population is difficult to ascertain, an uncertainty factor of 3-fold for this iodide
uptake inhibition level as a minimal LOAEL as well as a 3-fold factor for intrahuman variability
would be warranted. This is particularly relevant if this value is viewed in context with the
neurodevelopmental effects in laboratory animal data. At a minimum each factor should be
3-fold, and discussion with respect to the meaning of these factors with respect to population
effects again entertained. None of the human studies of perchlorate reviewed in Chapter 4 have
adequately investigated neurodevelopmental outcomes. The concern for duration of exposure
was at least a 3-fold factor per the above laboratory animal data discussion and should also be
applied, as well as the 3-fold factor for database deficiencies because these considerations and
deficiencies are not obviated by the use of human data.
Thus, a derivation based on the available human data would estimate the RfD at a
maximum of 0.00007 mg/kg-day, an estimate in rather good agreement with that proposed based
on the laboratory animal data (0.00003 mg/kg-day). If a larger UF were to be applied to the
human data, as could be justified for the intrahuman factor, the resultant estimate would be
essentially equivalent to that proposed using the laboratory animal data.
The consistency between the estimates based on the laboratory animal versus the human
data is likely due, at least in apart, to the use of AFRL/HEST PBPK modeling (Merrill, 2001c,d;
Clewell, 2001 a,b) to perform the interspecies extrapolation rather than the use of default factors.
It should be noted that the original motivation for performing these human studies (as discussed
January 16, 2002
7-26
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
in Chapter 3) in the perchlorate testing strategy was to support such interspecies pharmacokinetic
extrapolation and not to derive NOAEL estimates for thyroid effects in the human population. In
addition, as noted in Chapter 4, the EPA felt that both the observational epidemiological and the
human clinical studies have significant scientific and technical limitations that precluded their
use as the basis for a quantitative dose-response assessment. As mentioned previously, under the
interim policy articulated on December 14, the Agency will not consider or rely on any such
human studies (third-party studies involving deliberate exposure of human subjects when used to
identify or quantify toxic endpoints such as those submitted to establish a NOAEL or NOEL for
systemic toxicity of pesticides) in its regulatory decision making, whether previously or newly
submitted. Nonetheless, the use of both previously published and newly-derived human data by
the Air Force in its modeling efforts was important. The AFRL/HEST PBPK model approach
allowed EPA to confirm that humans were as sensitive as rats to the iodide uptake inhibition
effects of perchlorate at the NIS, the key event for the proposed mode-of action of perchlorate on
the thyroid. In addition, the PBPK models increased the accuracy of interspecies extrapolation
by allowing the incorporation and integration of ADME data to describe perchlorate and iodide
disposition relative to the key event. These two outcomes from the integration of human and
animal data in the AFRL/HEST models provide greater confidence than would the laboratory
animal data alone that the reference dose that is derived will be protective of human health.
7.1.5.2 Comparison with Derivation Based on Tumor Data
To address neoplasia as the other potential adverse endpoint, this section will discuss how
an estimate could be derived based on the recently acquired tumor data.
7.1.5.2.1 Choice of Dose-Response Procedure
As discussed in Chapter 5, the genotoxicity assays included in the testing strategy
determined that perchlorate was not likely to be mutagenic. This was one of the critical
determinants in deciding on a dose-response approach for a cancer derivation. The EPA
guidance on assessment of thyroid follicular cell tumors (U.S. Environmental Protection Agency,
1998a) sets forth data needs to establish the default dose-response procedure that should be used
to establish that a chemical has antithyroid activity (i.e., that it is disrupting the thyroid-pituitary
hormone status). Table 7-8 lists the default procedures for thyroid carcinogens that would be
January 16, 2002
7-27
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
1
2
3
4
5
6
7
8
TABLE 7-8. DEFAULT DOSE-RESPONSE PROCEDURES FOR
THYROID CARCINOGENS (U.S. Environmental Protection Agency, 1998a)
Array of Effects
Example Mutagenic Antithyroid Dose-Response Methodology
1 Either or both unknown Linear
2 Yes No Linear
3 No Yes Margin of exposure
4 Yes Yes Linear and margin of exposure
used. The thyroid lesions observed (colloid depletion, hypertrophy, and hyperplasia) are among
the required lesions to demonstrate antithyroid activity. Table 7-9 shows the types of data
required.
TABLE 7-9. DATA DEMONSTRATING ANTITHYROID ACTIVITY
(U.S. Environmental Protection Agency (1998a)
Required Desirable
1. Increases in cellular growth 6. Lesion progression
2. Hormone changes 7. Structure-activity relationships
3. Site of action 8. Other studies
4. Dose correlations
5. Reversibility
What has been proposed in this assessment is the harmonization of the "noncancer" and
"cancer" assessment approaches because the target tissue is the thyroid and the mode of action is
the same for both the neurodevelopmental and neoplastic sequelae. The proposed RfD based on
precursor lesions is analogous to a nonlinear approach and viewed as a protective for thyroid
tumors.
Perchlorate has clearly demonstrated an effect in both adult, fetal, and neonatal stages in
thyroid histopathology, as well as a decrease in lumen size in a dose-dependent fashion. Thyroid
and pituitary hormone changes and expected correlations all have been demonstrated for T3, T4,
January 16,2002 7-28 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
and TSH across an array of studies at different time points. The site of action has been
established as competitive inhibition of the iodide symporter although there remains some
uncertainty as to whether that is the only locus for the effect (e.g., evidence for intrathyroidal
activity) because of the efflux (discharge) phenomenon. Dose-correlations in this case were not
with tumors, but rather for precursor lesions (colloid depletion, hypertrophy, hyperplasia, and
decreased follicular lumen size). Reversibility has been demonstrated in thyroid weight, colloid
depletion, hypertrophy, hyperplasia, and thyroid and pituitary hormones in the 30-day recovery
period after the 90-day study in rats and in T4 levels of the various immunotoxicity experiments
in mice.
Lesion progression was difficult to determine because of dose-spacing and differences in
sample size and histological methods among the studies. However, there was a progression
within the 90-day study between the 14- and 90-day time points.
Analyses of other anions have fairly well established that the mode of action of perchlorate
arises from it being an anion that is recognized by the NIS (see Chapter 3).
Thus, the appropriate dose-response procedure for perchlorate would be a nonlinear
margin-of exposure approach based on demonstration that it is not genotoxic and that its
anti-thyroid effects are consistent with a mode of action leading from inhibition of iodide uptake
at the NIS through precursor lesions of perturbation of thyroid hormone economy and resultant
histopathological changes in the thyroid gland.
7.1.5.2.2 Dose-response Assessment for Thyroid Neoplasia
Thyroid adenomas were statistically increased in the high dose (30 mg/kg-day) group of
F1-generation animals sacrificed as adults (P2-generation) at 19 weeks in the Argus Research
Laboratories, Inc. (1999) two-generation reproductive study. Both the latency and incidence of
these tumors were remarkable relative to the entirety of the NTP data base for this type of tumor
in this strain of rat (Dunson, 2001b). Colloid depletion, hypertrophy, and hyperplasia were all
observed at dosages of 0.3 mg/kg-day and above with BMDL estimates of 0.9, 0.15, and
0.0004 mg/kg-day. This last estimate is outside the range of possible dosimetric adjustment so it
will not be carried forward, but consideration of the overlap among colloid depletion,
hypertrophy, and hyperplasia should be superimposed on the derivation. The HEE values for
adult versus neonatal rats are comparable at these dosages. Using the adult male rat dosimetric
January 16, 2002
7-29
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
adjustment factor to correspond to sacrifice date results in HEE estimates of 0.45 and 0.02 for
colloid depletion and hypertrophy.
Using the nonlinear approach and applying a composite factor of 100 to the HEE estimates
to account as above for uncertainty in intrahuman variability, duration, and database deficiencies;
and with factor for a minimal LOAEL of 3 to account for the fact that hyperplasia occurred at
over an order of magnitude lower than these two thyroid histopathology estimates, results in an
RfD derivation in the range of 0.005 to 0.0002 mg/kg-day. Applying a larger uncertainty factor
for intrahuman variability would result in a range of 0.002 to 0.00007 mg/kg-day. Thus, the
derivation based on tumor outcome data supports the mode-of-action concept and corroborates
that the proposed RfD that as derived would be protective of both neurodevelopmental and
neoplastic sequelae.
7.1.5.3 Possible Susceptibility
Based on the mode-of-action for perchlorate, the competitive inhibition of iodide uptake,
and the subsequent perturbation of thyroid hormone homeostasis, a number of factors potentially
could cause an increase in susceptibility of a population to perchlorate toxicity. As already
indicated by the choice of critical effect, the fetus, and perhaps the developing child, may
represent susceptible populations. However, critical data on the steady-state pharmacokinetics
and placental dosimetry are lacking to definitively state whether or not there is an inherent
pharmacodynamic component to the apparent sensitivity of pups versus dams in the laboratory
animal models. Individuals that are iodine deficient may be another susceptible population. The
elderly, especially women, and hypothyroid and hypothyroxinemic individuals or those treated
with anti-thyroid drugs, may be others more susceptible than the general population to the effects
of perchlorate. Patients with cardiac dysfunction or elevated levels of cholesterol may also be at
increased risk.
7.1.6 Designation of Confidence Levels
Confidence in the principal study is medium. The dose level of 0.01 mg/kg-day was the
lowest tested, and it was determined to be a LOAEL (not NOAEL). The small sample size for
the critical effect also reduces confidence in the study. Despite the new data, the confidence in
the database at this time remains medium because the sensitivity of these animal assays versus
January 16,2002 7-30 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
evaluation of neuropsychological development in human population studies is not known, and
because a concern for potential immunotoxicity remains. Based on confidence in the study and
on the database together in setting the overall confidence in the RfD, the confidence in the RfD
currently is also medium.
7.2 INHALATION REFERENCE CONCENTRATION
Derivation of an inhalation reference concentration is precluded because there are no
inhalation data available with which to characterize dose-response or the portal-of-entry
modulation of internal dose. However, the EPA has been questioned as to whether the potential
for inhalation exposure of perchlorate from showering with contaminated water poses a health
risk. Given the low vapor pressure of perchlorate, it is not likely that it would come out of
solution. Further, Giardino et al. (1992) characterized shower particle droplet size as ranging
from 200 to 3,000 iiva. Thus, there is minimal chance for inhalation or deposition of perchlorate -
laden droplets in the respiratory tract.
7.3 SUMMARY
The model based on mode of action for perchlorate served as a useful construct for the
integration of a diverse set of data. Results of studies in the testing strategy confirmed that the
target tissue for perchlorate is the thyroid and that the key event for its antithyroid effects is the
inhibition of iodide uptake at the NIS with corresponding perturbations of thyroid hormone
economy. Disturbances in thyroid hormone economy were confirmed to result in thyroid
histology as diagnosed by decreases in colloid depletion or follicular lumen size and increases in
hypertrophy and hyperplasia. Effects on both neurodevelopmental indices (brain morphometry
and motor activity) and neoplasia that could be expected based on the mode of action were also
demonstrated. Other developmental and reproductive effects were not observed to be as
sensitive as the neurodevelopmental and thyroid histopathological changes. Accurate
characterization of the immunotoxicity of perchlorate, notably its potential to cause contact
January 16, 2002
7-31
DRAFT-DO NOT QUOTE OR CITE
-------
1 hypersensitivity, either secondarily to these hormone effects or possibly via a direct effect of the
2 anion itself, remains a concern.
January 16, 2002
7-32
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
APPENDIX 7 A
CORRELATION ANALYSES
The correlation analyses were of two types. Hormone levels are continuous, ratio-scaled
values, so correlations were computed using the conventional Pearson's r statistic. Correlations
between ratio-scaled hormone levels and ordinally-scaled standard histology ratings must be
computed using nonparametric correlations. To compare variables from the different scales, it is
simplest to recode the data by converting the variable values into rank scores. Spearman's rank
order (rs) was used to compute the correlation between the rankings of two variables. When there
were ties in the ranks, as there were in this data set, each value was assigned the mean of the
ranks that they would otherwise occupy. A correlation coefficient was then computed for the
rankings of the variables of interest.
An alternative statistic used for comparing the data sets was Kendall's tau, best thought of
as a measure of agreement or concordance between two sets of ranked data. It searches for the
number of inversions in two sets of ranked data (i.e., observations are ranked according to the
first variable, then reranked according to the second, and the number of interchanges that occur is
used to compute the statistic). The Spearman and Kendall statistics produced nearly identical
results. Statistics were computed using SAS® software (PROC RANK and PROC CORR,
SAS Institute, Cary, NC). All statistics corresponding to Figures 7A-1 through 7A-7 can be
found in Tables 7A-1 through 7A-6.
TABLE 7A-1. PEARSON'S r CORRELATIONS (n = 96) BETWEEN THYROID
HORMONES AND TSH IN RATS OF THE CALDWELL et al. (1995) 14-DAY STUDY
T3
T4
TSH
T3
1.00
0.81
-0.65
p = 0.00
p = 0.0001
p = 0.0001
T4
1.00
-0.67
p = 0.00
p = 0.0001
TSH
1.00
p = 0.00
January 16, 2002
7A-1
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7A-2. SPEARMAN'S rs CORRELATIONS (n = 95) BETWEEN THE RANK
ORDER OF HORMONE LEVELS AND HISTOLOGICAL SEVERITY RATING
DECREASE IN FOLLICULAR LUMEN SIZE (LS) IN RATS OF
THE CALDWELL et al. (1995) 14-DAY STUDY
LS
T3
-0.74
p = 0.0001
T4
-0.70
p = 0.0001
TSH
0.79
p = 0.0001
FH
0.75
p = 0.0001
TABLE 7A-3. PEARSON'S r CORRELATIONS (n = 223) BETWEEN THYROID
HORMONES AND TSH IN RATS FOR THE COMBINED 14- AND 90-DAY DATA OF
THE SPRINGBORN LABORATORIES, INC. (1998) SUBCHRONIC RAT STUDY
T3
T4
TSH
T3
1.00
0.42
-0.18
p = 0.00
p = 0.0001
p = 0.007
T4
1.00
-0.20
p = 0.00
p = 0.0027
TSH
1.00
p = 0.00
TABLE 7A-4. PEARSON'S r CORRELATIONS (n = 104) BETWEEN
THYROID HORMONES AND TSH FOR THE 14-DAY DATA OF THE
SPRINGBORN LABORATORIES, INC. (1998) SUBCHRONIC RAT STUDY
T3
T4
TSH
T3
1.00
0.36
-0.11
p = 0.00
p = 0.0001
p = 0.27
T4
1.00
0.20
p = 0.00
p = 0.04
TSH
1.00
p = 0.00
January 16, 2002
7A-2
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7A-5. PEARSON'S r CORRELATIONS (n = 119) BETWEEN THYROID
HORMONES AND TSH OF THE 90-DAY DATA OF THE SPRINGBORN
LABORATORIES, INC. (1998) SUBCHRONIC RAT STUDY
T3
T4
TSH
T3
1.00
p = 0.00
0.66
p = 0.0001
-0.40
p = 0.0001
T4
1.00
p = 0.00
-0.38
p - 0.0001
TSH
1.00
p = 0.00
TABLE 7A-6. PEARSON'S r CORRELATIONS (n = 22 to 27) BETWEEN THYROID
HORMONES AND TSH FOR THE F1 RAT PUPS ON PND5 IN THE
DEVELOPMENTAL NEUROTOXICITY STUDY
(Argus Research Laboratories, Inc., 1998a)
T3
T4
TSH
T3
1.00
p = 0.00
0.87
p = 0.0001
-0.43
p = 0.03
T4
1.00
p = 0.00
-0.57
p = 0.0046
TSH
1.00
p = 0.00
1 In general, positive correlations were expected between T3 and T4 and between TSH and
2 the histopathology rating. Negative correlations were expected between T4 and TSH and
3 between T4 and histopathology.
4 Figure 7A-1 shows the correlations between T3 and T4 and between T4 and TSH levels
5 from the 14-day Caldwell et al. (1995) study in rats. Robust relationships are illustrated:
6 a positive correlation is shown between T3 and T4; whereas, the T4 and TSH varied inversely.
7 Hormone levels also correlated highly with decrease in follicular lumen size. Figure 7A-2 shows
8 the rank of T4 level and TSH level versus the severity rating for follicular lumen size to be highly
9 correlated inversely. Figure 7A-3 shows the correlations for the combined 14-day and 90-day
10 time points (male and female) from the subchronic study performed in rats (Springborn
January 16, 2002
7A-3
DRAFT-DO NOT QUOTE OR CITE
-------
160
140 H
120
O)
? iooH
>
CD
P 80-
60 -
40
Pearson's r = 0.81, p < 0.0001
CP
o
CP o CP
„ o
K £
,•.*» *
d"
0 00 mg/kg-day
0 11 mg/kg-day
0 25 mg/kg-day
1 11 mg/kg-day
2 63 mg/kg-day
4 62 mg/kg-day
11 45 mg/kg-day
22 51 mg/kg-day
T4 Level (pg/dL)
TJ
o>
o 4
CD
3-
Pearson's r = - 0.67, p < 0.0001
8P
<5*>
<8>,
- ;£•
~<
•••
%
» * o fi +
+ * o,
T
10
T
20 30 40
TSH Level (ng/mL)
50
Figure 7A-1. Correlations between T3 versus T4 (top panel) and T4 versus TSH
(bottom panel) in rats of the 14-day Caldwell et al. (1995) study (Geller,
1998a). Data of Channel (1998a) and Crofton (1998a).
January 16, 2002
7A-4
DRAFT-DO NOT QUOTE OR CITE
-------
120
100 -
Spearman's rs = - 0.70, p < 0.0001
Kendall's tau-b = - 0.56, p < 0.0001
a?
>
0)
o
c
CD
a:
80
60
40
20
~
T
B
+
!
*
$
o-
T
0
1
Standard Histopathological Severity Rating of
Decrease In Follicular Lumen Size
120 -r
£ 100
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Laboratories, Inc., 1998). As shown in Figure 7A-3 (top panel), T3 and T4 were highly
significantly correlated, with low levels of T3 and T4 associated with high doses. Both T4 and
TSH were significantly negatively correlated (bottom panel). After 14-days of dosing
(Figure 7A-4), T3 and T4 are highly associated (top panel), but there is an unexpected positive
relation between T4 and TSH (bottom panel). At the 90-day time point, there are the expected
strong correlations between T3 and T4 (Figure 7A-5, top panel) and between T4 and TSH
(bottom panel).
Correlations also were performed on the data from the neurodevelopmental study for the
PND5 pups (Argus Research Laboratories, Inc., 1998a). T3 and T4 were strongly positively
correlated, and T4 and TSH were negatively correlated (Figure 7A-6). Figure 7A-7 (top panel)
shows that T4 is negatively associated with a significant decrease in lumen area. Figure 7A-7
(bottom panel) also shows that TSH is positively correlated with a decrease in lumen size.
In total, these correlations lent strong support to the mapping model proposed. Strong
correlations were observed between T3 and T4 levels, T3 or T4, and TSH levels, and hormone
levels and a decrease in thyroid lumen size. These relationships were most definitive in the
Caldwell et al. (1995) study, in which strong correlations existed between the elements of the
thyroid hormone homeostasis feedback loop and between hormone levels and severity ratings for
lumen size decrease as a measure of thyroid histopathology. In the subchronic (Springborn
Laboratories, Inc., 1998) study, correlations were established between hormone levels across
both the 14- and 90-day dosing points and for each time point individually. At 14 days of dosing,
the expected inverse relationship between T4 and TSH was not found. At the 90-day dosing
point, the inverse relationships between T3 or T4 and TSH were found.
Similar relationships were observed in pups on PND5 of the developmental neurotoxicity
study (Argus Research Laboratories, Inc., 1998a; York, 1998c). The T4 and TSH were
significantly correlated negatively, as expected. The T3, T4, and TSH were all significantly
correlated with decrease in lumen size. The correlations in the rat studies support the model that
manipulations resulting in decreased levels of circulating thyroid hormone are linked to thyroid
histopathological changes that are thought to result directly from elevation of TSH.
January 16, 2002
7A-6
DRAFT-DO NOT QUOTE OR CITE
-------
CD
C
260-
240-
220-
200-
180-
(D
5> 160 -
140-
120-
100-
80-
Pearson's r = 0 42, p = 0.0001
O
O
O O
^ d ^ n
°CD2)
T4 Level (ng/dL)
o
0 00 mg/kg-day
~
0 01 mg/kg-day
A
1.00 mg/kg-day
V
10 mg/kg-day
+
30 mg/kg-day
*
100 mg/kg-day
5 5 ¦
O)
0)
> .
(D 4 -
0 Pearson's r = -0.20, p = 0.0027
O
c±
10 15 20
TSH Level (ng/dL)
25
30
Figure 7A-3. Correlations between T3 versus T4 (top panel) and T4 versus TSH
(bottom panel) for the combined male and female data of the 14-day
and 90-day time points from the Springborn Laboratories Inc. (1998)
subchronic study (Geller, 1998a).
January 16, 2002
7A-7
DRAFT-DO NOT QUOTE OR CITE
-------
260
240 -
220 -
200 -
—I
F
ra
180 -
a
a3
>
160 -
<1)
CO
h-
140 -
120 -
100 -
80-
Pearson's r = 0 36, p = 0.0001
O
O
o
o
o
o A
v-r:
*
VD
v. «7 ^7 D
* Astn*8 ^ ^
+
4 5
T4 Level (pg/dL)
O
O
0 00 mg/kg-day
~
0 01 mg/kg-day
A
0 05 mg/kg-day
V
0 2 mg/kg-day
+
1 0 mg/kg-day
~
10 mg/kg-day
15 20
TSH Level (ng/mL)
Figure 7A-4. Correlations between T3 versus T4 (top panel) and T4 versus TSH
(bottom panel) for the combined male and female data of the 14-day
time point from the Springborn Laboratories Inc. (1998) subchronic
study in rats (Geller, 1998b).
January 16, 2002
7A-8
DRAFT-DO NOT QUOTE OR CITE
-------
240
220
200
E 180
o>
c
o 160
>
<1)
—I
« 140
120
100
80
2 3 4 5 6 7
T4 Level (pg/dL)
o
0 00 mg/kg-day
~
0.01 mg/kg-day
A
0 05 mg/kg-day
V
0 2 mg/kg-day
+
1 0 mg/kg-day
~
10 mg/kg-day
7
6
3 5
O)
0)
>
0> 4
_i H
I-
3
2
12 14 16 18 20 22 22
TSH Level (ng/mL)
Figure 7A-5. Correlations between T3 versus T4 (top panel) and T4 versus TSH
(bottom panel) for the combined male and female data of the 90-day
time point from the Springborn Laboratories Inc. (1998) subchronic
study in rats (Geller, 1998b).
Pearson's r = 0.66, p = 0 0001
o
° o
D
~DO
O
o
° CcO
O
4^' ^
O
* sHAv ~ 1=1
*+*±ir& +£A Av
i i r
O
O O O
Pearson's r = - 0 38, p = 0.0001
O
°o
° D ~
D°^ °
J? *>' A * V+
~
A"
, * * *V + *
+ * * + **
•***
January 16, 2002
7A-9
DRAFT-DO NOT QUOTE OR CITE
-------
100 -
90-
80-
E 70 H
oi
c
aj SO
>
0)
o 50 H
H
40 -
30-
Pearson's r3 = 0 87, p = 0.0001
A
~a
%
V ^
yX V
V
O
~
A
1 1 1—
2 3 4
T4 Level (pg/dL)
O 0 0 mg/kg-day
~ 0 1 mg/kg-day
A 1 0 mg/kg-day
x 3 0 mg/kg-day
v 10 mg/kg-day
4 5 6
TSH Level (ng/mL)
Figure 7A-6. Correlations between T3 versus T4 (top panel) and T4 versus TSH
(bottom panel) for the Fl-generation rat pups on PND5 in the
developmental neurotoxicity study (Geller, 1998b). Data of Argus
Research Laboratories, Inc. (1998a), York (1998c), Channel (1998c),
and Crofton (1998f).
January 16, 2002
7 A-10
DRAFT-DO NOT QUOTE OR CITE
-------
30-
25-
8
c
<0
IX
20-
15-
10-
5-
0-
A
8
V
6
o
V
V
Pearson's rB = - 0.51, p = 0.007
Kendall's tau-b = - 0 39, p = 0 008
1 1
40 •
30 ¦
a>
>
a>
_i
X
CO
20 -
a>
"2 10 •
ro
QL
0 -
0 12 3 4
Standard Histopathological Severity Rating of
Decrease in Follicular Lumen Size
o o 0 mg/kg-day
~ 0 1 mg/kg-day
A 1 0 mg/kg-day
* 3.0 mg/kg-day
v 10 mg/kg-day
2
V
A
O
k
V
V
0
* x '
~
&
~
x ~
Pearson's r, = 0.28, p = 0 10
Kendall's tau-b = 0.21, p = 0 09
0 12 3
Standard Histopathological Severity Rating of
Decrease in Follicular Lumen Size
Figure 7A-7. Correlations between the rank order of T4 (top panel) and TSH
(bottom panel) versus histopathology severity rating of the decrease in
follicular lumen size for the postnatal day 5 (PND5) pups in the 1998
neurodevelopmental study (Geller, 1998b). Data of Argus Research
Laboratories, Inc. (1998b), Channel (1998c), and Crofton (1998e, f).
January 16, 2002
7A-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
1
2
Appendix 7B
Benchmark Dose Statistics for Hormone Analyses
As mentioned in Chapter 5, benchmark dose analyses were performed in addition to the
ANOVA for all hormone data. Benchmark analysis of the 2001 "Effects Study" is presented in
Geller (2001c). This appendix presents analyses performed on the other data sets provided in the
1998 assessment.
For the continuous hormone data, the BMD and BMDL estimates were calculated using a
variety of benchmark response (BMR) values. Generally, the BMR was equal to a response 10%
less than the control mean (i.e., 10% of the actual control response was subtracted from the
estimate of the control value generated by the fit to the data). This is a less rigorous standard
than the (control minus 5% of control) BMR that provided a close match to NOAELs in the
evaluation of BMD for developmental toxicity by Kavlock et al. (1995) although this may be
warranted because other endpoints (thyroid hormone and histopathology) are being evaluated.
For the natural log (In) transformed data, this means subtracting the constant 0.1053 from the
control value, which is equivalent to multiplying the control value by 0.90. The BMD and
BMDLs at 20 and 30% less than control and control standard deviations also are provided as a
yardstick for evaluating how other clinical criteria may affect the estimates. Hormone data were
fit with polynomial (linear or quadratic) or power functions (Table 7B-1).
TABLE 7B-1. CONTINUOUS FUNCTIONS USED IN
BENCHMARK DOSE (BMD) MODELING
Power function f(dose) = control + slope * dosepowCT
Polynomial function f(dose) = P0 + pi * dose + P2 * dose2 +...
(includes linear and quadratic)
Adequacy of fit for continuous data was evaluated by the statistical goodness-of-fit
(-2 x log likelihood ratio) test provided by the EPA BMD program output, visual comparison,
January 16, 2002 7B-1 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
and whether the fit was biologically plausible. The latter criterion in most cases,
non-monotonicities in the function fit to the data, precluded a fit from consideration. In general,
the second order quadratic fits suffered from minima or maxima between the data points from the
two highest data points in a given experiment. This consideration also precluded the use of
polynomials of higher than second order because these higher order polynomials generally had a
local maxima or minima between data points (dose levels) and did not model the data plausibly.
It should be noted that the interpretation of the test for constant variance included in the output of
the version of the BMD software (version 0.96) was not reliable.
7B.1 Benchmark Dose Estimates Submitted to U.S. Environmental
Protection Agency
Two sets of BMD calculations were derived from the Caldwell et al. (1995) 14-day study
and submitted to the EPA (Dollarhide and Dourson, 1997). One set was calculated for TSH and
T4 levels for males and females separately using the THC (polynomial fit) module of the Crump
software, and the model coefficients were restricted to be nonnegative to prevent
non-monotonicity. This resulted in linear fits to curvilinear data, and the fits were judged to be
poor by both visual inspection and statistical goodness-of-fit criteria (Geller, 1998a).
An alternative approach to calculating BMD estimates based on additional risk also was
derived using the Kodell-West algorithm (Kodell-West, 1993). This model generates a quadratic
fit to the dose-response data using a maximum likelihood estimator, defines an adverse effect
level based on the variability present in the data, and then calculates additional risk. The EPA
recalculated these fits using Kodell's SAS® program (Geller, 1998a). The EPA estimates
correspond to those previously reported, as shown in Table 7B-2 of Appendix 7B. The
coefficients of the fits are provided in Table 7B-3. None of the fits to the data reached statistical
significance, and all contain minima (T3 and T4) or maxima (TSH) within the dose range tested.
Again, the lack of fit raises difficulties with interpretation and suggests that these estimates
should not be used as the basis for risk assessment. The EPA also calculated BMD estimates on
In-transformed data because the Kodell-West algorithm assumes constant variance, and the
transformed data is more likely to fit this assumption. The BMD estimates calculated with the In
transform, however, were virtually identical to those of the previous estimates.
January 16, 2002
7B-2
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-2. BENCHMARK DOSE (BMD) ESTIMATES FOR MALE HORMONE
DATA OF CALDWELL et al. (1995) 14-DAY RAT STUDY, USING
KODELL-WEST ALGORITHM
BMD Associated with 1% BMD Associated with 10% BMD:N(L)OAEL
Responders Additional Risk (mg/kg-day) Additional Risk (mg/kg-day) 1%; 10%
TSH
EPA0
D&D, 1997b
EPA"
D&D, 1997b
1.11
k = 3
0.832
0.823
2.078
2.074
0.75; 1.87
k = 2
0.176
0.172
0.972
0.970
0.16; 0.88
In TSH
1.11
k = 3
0.845
2.115
0.76; 1.91
k = 2
0.181
0.987
0.16; 0.89
T3
EPA"
D&D, 1997"
EPA"
D&D, 1997b
0.1 lcd
k = 3
0.980
0.983
2.485
2.495
8.1; 22.59
k = 2
0.209
0.207
1.146
1.151
1.9; 10.42
lnT3
0.1lcd
k = 3
0.891
2.244
8.1; 20.4
(N
II
0.190
1.042
1.73; 9.47
T4
EPA"
D&D, 1997"
EPA"
D&D, 1997b
0.1 lcd
k = 3
0.797
0.658
1.969
1.639
7.25; 17.9
k = 2
0.172
0.136
0.927
0.774
1.56; 8.43
In (T4)
0.1 lcd
k = 3
1.002
2.490
9.11; 22.64
k = 2
0.215
1.169
1.95; 10.63
"EPA refers to BMD estimates calculated using SAS® software received from Dr. Ralph Kodell for Kodell-West
calculations (Geller, 1998a).
bD&D refers to BMDs included in Dollarhide and Dourson (1997).
cLOAEL; otherwise, value indicates NOAEL.
dLOAEL from combined male and female.
January 16, 2002
7B-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
TABLE 7B-3. COEFFICIENTS AND GOODNESS-OF-FIT STATISTICS OF
KODELL-WEST (QUADRATIC POLYNOMIAL) MODEL FITS TO MALE
HORMONE DATA OF CALDWELL et al. (1995) 14-DAY RAT STUDY"
Responders
B0
B1
B2
Dose (mg/kg-day) of
Global Max/Min
p of Fit2b
TSH
17.182
2.895
-0.0914
max: 15.84
<0.00001
In TSH
2.825
0.1269
-0.004202
max: 15.11
<0.00001
T3
112.871
-8.987
0.3169
min: 14.18
<0.00001
lnT3
4.7114
-0.09702
0.0034
min: 14.27
<0.00001
T4
4.7712
-0.1791
0.00445
min: 20.11
<0.00001
In (T4)
1.563
-0.0414
0.0009
min: 23.00
0.00012
"Coefficients generated by using SAS software received from Dr. Ralph Kodell (Geller, 1998a). Identical
coefficients were generated by using EPA BMD software.
bp > 0.05 denotes significant fit. Goodness-of-fit derived using -2 log (likelihood ratio) test from EPA BMD
software (see Geller, 1998a).
7B.2 U.S. Environmental Protection Agency Benchmark Dose Estimates for
Thyroid and Pituitary Hormones
The hormone data from the Caldwell et al. (1995) subchronic (Springborn Laboratories,
Inc., 1998) and rabbit developmental studies (Argus Research Laboratories, Inc., 1998c) were
best fit by unrestricted power functions. The hormone data from the developmental neurotoxicity
study (Argus Research Laboratories, Inc., 1998a; York, a,b,c,d,e) and mouse immunotoxicity
study (Keil et al., 1998) were fit by either unrestricted power or polynomial functions. It is noted
that the unrestricted power function fits generally have an extremely high slope as dose
approaches zero. Tables 7B-4 through 7B-14 provide the statistics for each study.
Many of the BMDL estimates derived from these studies were lower than the NOAEL or
LOAEL values derived by ANOVA, particularly those derived from power function fits. Murrell
et al. (1998) suggested that this occurs when sampling statistics (i.e., small group sample sizes
and few dose groups) play a large role in inflating NOAELs while depressing BMDL estimates.
This may be the case for some of the data examined herein. Murrell et al. (1995) suggested that
under such conditions using the BMD point estimate, rather than the lower confidence limit,
would be a more accurate representation of the dose-response behavior.
January 16, 2002
7B-4
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-4. BENCHMARK DOSE (BMD) ESTIMATES USING POWER
FUNCTION FIT TO COMBINED MALE AND FEMALE HORMONE DATA OF
CALDWELL et al. (1995) 14-DAY RAT STUDY
(Benchmark response based on 10% change from control value.)
Endpoint
p of Fit
BMD
BMDL
NOAEL/
LOAEL
BMD:
N(L)OAEL
BMDL:
N(L)OAEL
BMR:
10% control SD
TSH"
0.272
0.014
0.0002
0.44
0.032
4.55e-4
1.29
1.88
In TSH"
0.099
0.017
0.002
0.44
0.039
4.55e-3
-0.1053
Female TSHb
0.077
0.19
0.032
0.1
1.90
0.32
1.125
0.48
Female
ln(TSH)"
0.50
0.078
0.035
0.1
0.78
0.35
-0.1053
Male TSH
No significant fits to male TSH
or male ln(TSH) data
T3a
0.107
0.00035
0.00
0.1c
0.0035
NA
13.07
10.21
lnT3"
0.091
0.0004
2e-6
0.1c
0.004
2.00e-5
-0.1053
T4°
0.303
0.243
0.096
0.1c
2.43
0.96c
0.506
0.321
In (T4)d
0.172
0.340
0.0997
0.1c
3.40
1.00c
-0.1053
"Unrestricted quadratic: fit nonmonotonic, not significant. Restricted polynomial (linear): fit not significant.
bUnrestricted quadratic: fit monotonic but not significant. Restricted polynomial (linear): fit not significant.
cLOAEL; otherwise, value is NOAEL.
dUnrestricted quadratic: fit not significant, global minimum at approximate high dose. Restricted polynomial
(linear): fit not significant.
1 The BMD estimates calculated with a benchmark response of 10% less than control on the
2 TSH hormone dose-response data are spread over 2.5 orders of magnitude, a similar range to that
3 seen in the distribution of NOAELs calculated for TSH. The BMDL estimates are distributed
4 more widely, over 5 orders of magnitude. These reflect the steepness of the confidence limits on
5 the slope at low doses.
6 The T3 BMD estimates are spread over approximately two orders of magnitude, similar to
7 the variability seen across studies in the LOAEL and NOAEL estimates. The T3 BMD estimates
8 are 100-fold lower than the NOAEL/LOAEL estimates, however. A BMDL could be calculated
January 16, 2002
7B-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
TABLE 7B-5. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT (BMDL)
ESTIMATES USING POWER FUNCTION FIT TO COMBINED MALE AND FEMALE
HORMONE DATA OF CALDWELL et al. (1995) 14-DAY RAT STUDY
(Benchmark response based on 10, 20, and 40% changes from control value.)
Endpomt
p of Fit
BMD
BMDL
(10%)
BMD
BMDL
(20%)
BMD
BMDL
(40%)
Mean
NOAEL
TSH
0.272
0.014
0.083
0.507
12.861
0.44
0.0002
0.0038
0.0604
ln(TSH)°
0.099
0.002
0.043
1.11
0.44
T3
0.0108
0.00035
0.0338
3.27
130.69
0.10b
0.00
0.000036
0.042°
ln(T3)a
0.091
0.000002
0.000642
0.478
0.10"
T4
0.303
0.243
2.28
21.44
5.06
0.10b
0.096
1.299
16.78
ln(T4)°
0.172
0.100
1.213
16.89
0.10b
"For In-transformed data, only BMDL estimates are displayed.
bLOAEL, not NOAEL.
CBMDL calculation failed at some values. This means BMDL value may not be accurate.
for only one of the data sets, and this value was approximately 10,000 times lower than the
LOAEL. The BMD estimates comprising the 25th to 75th percentiles for T4 cover the same
2.5 orders of magnitude as those covered by the NOAEL and LOAEL estimates for T4. The
BMDL estimates for this same percentile range are distributed a little more widely, but do
include the range of T4 NOAEL and LOAEL estimates.
7B.3 Summary of U.S. Environmental Protection Agency Benchmark
Dose Analyses
The BMD analyses of previously reported estimates for the hormone data of Caldwell et al.,
(1995) 14-day study in rats (Dollarhide and Dourson, 1997) were shown to be based on
inadequate model fits. The EPA was able to successfully model the hormone data. However,
these estimates raised a number of issues with respect to approaches for these types of data.
An alternative may be to pursue a model form of the Hill equation which recently has been used
for endocrine disruption data (Barton et al., 1998).
January 16, 2002
7B-6
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-6. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT (BMDL)
ESTIMATES FOR COMBINED MALE AND FEMALE HORMONE DATA OF
14-DAY TIME POINT IN THE SPRINGBORN LABORATORIES, INC. (1998)
SUBCHRONIC STUDY (Benchmark response based on 10% change from
control value.)
BMR:
p of
NOAEL/
BMDL:
BMD:
10% control
Endpoint
Model
Fit
BMD BMDL
LOAEL
N(L)OAEL
N(L)OAEL
SD
TSH
Power
0.45
0.037 0.000075
0.01
0.0075
3.7
1.26
Quadratic
0.069
Fit significant, but not
monotonic
0.01
2.52
In TSH
Power
0.43
0.043 Could not
calculate
0.01
NA
4.3
-0.1053
Quadratic
Fit not significant, nonmonotonic
0.01
T3
Power
0.41
0.000033 Lower limit
includes 0
0.01*
NA
0.0033
16.65 38.51
Quadratic
Fit not significant, nonmonotonic
0.01'
lnT3
Power
0.35
0.000168 Lower limit
includes 0
0.01°
NA
0.0168
-0.1053
Quadratic
Fit not significant, nonmonotonic
0.01°
T4
Power
0.203
1.16 0.0035
1.0
0.0035
1.16
0.506
Quadraticb
0.12
3.27 1.09
1.0
1.09
3.27
0.603
In (T4)
Power
0.22
1.64 0.04
1.0
0.04
1.64
-0.1053
Quadraticb
0.16
3.25 1.06
1.0
1.06
3.25
"LOAEL; otherwise, value is NOAEL.
bGlobal minimum of quadratic function is at dose =9.50 mg/kg-day.
January 16, 2002
7B-7
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-7. BENCHMARK DOSE (BMD) AND BMD 95% LOWER
LIMIT (BMDL) ESTIMATES FOR COMBINED MALE AND FEMALE
HORMONE DATA OF 14-DAY TIME POINT IN THE SPRINGBORN
LABORATORIES, INC. (1998) SUBCHRONIC STUDY
(Benchmark response based on 10, 20, and 40% changes from control value.)
Endpoint
Model
p of
Fit
BMD
BMDL
(10%)
BMD
BMDL
(20%)
BMD
BMDL
(40%)
Mean
NOAEL
T4
Power
0.203
1.16
0.0035
12.73
1.21
138.94
38.33
5.066
1.0
ln(T4)
Power
0.22
0.037
3.899
36.48
1.0
T3
Power
0.41
0.000033
0.207
129.39
0.129°
166.5
0.01b
ln(T3)
Power
0.35
Lower limit
includes 0
0.000054°
43.16°
0.01b
TSH
Power
0.45
0.037
0.000076
0.326
0.005
2.89
0.36
12.616
0.01
ln(TSH)
Power
0.43
0.0015
0.098
6.587
0.01
"BMDL calculation failed at a number of values. This means BMDL value may not be accurate.
bLOAEL, not NOAEL.
TABLE 7B-8. BENCHMARK DOSE (BMD) ESTIMATES FOR COMBINED MALE
AND FEMALE HORMONE DATA OF 90-DAY TIME POINT IN THE SPRINGBORN
LABORATORIES, INC. (1998) SUBCHRONIC STUDY
(Benchmark response based on 10% change from control value.)
Endpoint
p of Fit
BMD
BMDL
NOAEL/
LOAEL
BMD:
N(L)OAEL
BMDL:
N(L)OAEL
BMR:
10% Control SD
TSH"
0.42
0.269
0.018
0.05
5.38
0.36
1.633
1.464
In TSH"
0.40
0.492
0.0796
0.05
9.84
1.6
-0.1053
T3°
0.01
No fit
No fit
0.01b
NA
NA
17.50
18.924
lnT3°
0.01
No fit
No fit
0.01b
NA
NA
NA
T4°
0.14
6e-6
Lower limit
includes 0
0.01b
6e-4
NA
0.475
0.576
In (T4)°
0.17
1.10e-5
0.00
0.01b
l.le-3
oo
-0.1053
"Unrestricted quadratic: fit nonmonotonic, not significant. Restricted polynomial (linear): fit not significant.
bLOAEL; otherwise, value is NOAEL.
January 16, 2002
7B-8
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-9. BENCHMARK DOSE (BMD) AND BMD 95% LOWER
LIMIT (BMDL) ESTIMATES FOR COMBINED MALE AND FEMALE
HORMONE DATA OF 90-DAY TIME POINT IN THE SPRINGBORN
LABORATORIES, INC. (1998) SUBCHRONIC STUDY
(Benchmark response based on 10, 20, and 40% changes from control value.)
Model
p of
Fit
BMD
BMDL
(10%)
BMD
BMDL
(20%)
BMD
BMDL
(40%)
Mean
NOAEL
T4
Power
0.14
0.000006
0.01
15.09
4.75
0.01b
—
0.000001
0.52s
ln(T4)
Power
0.165
0.00
0.004
4.87
0.01b
T3
Power
0.01
No significant fit
174.96
0.01b
ln(T3)
Power
0.01
No significant fit
0.01b
TSH
Power
0.43
0.272
8.808
285.52
16.33
0.05
0.019
2.404
73.80
ln(TSH)
Power
0.40
0.082
7.94
405.14
0.05
"BMDL calculation failed at a number of values. This means BMDL value may not be accurate.
hLOAEL not NOAEL.
January 16, 2002
7B-9
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-10. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT
(BMDL) ESTIMATES FOR HORMONE AND THYROID MORPHOMETRY
DATA OF Fl-GENERATION PUPS AT PND5 IN THE DEVELOPMENTAL
NEUROTOXICITY STUDY
(Argus Research Laboratories, Inc., 1998a, and Channel, 1998c)a
(Benchmark response based on 10% change from control value.)
BMR
NOAELor BMD: BMDL: 10%
Endpoint
Model
p of Fit
BMD
BMDL
LOAEL
N(L)OAEL
N(L)OAEL
Control SD
TSH
Linear
Power
0 50
031
4.64
4.48
3.77
1.43
3.0
3.0
1.55
1.49
1.26
0.48
0 45
0 465
lnTSH
Linear
0 48
5.51
4.43
3.0
1.84
0 54
-0.1054
Power
0 30
5.03
2.11
3.0
1.68
0.70
T3
Neither linear,
quadratic, or power
FCNS fit data
<0 00001 forall
No fit
No fit
0 1
NA
NA
lnT3
Neither linear,
quadratic, or power
FCNS fit data
<0.00001 forall
No fit
No fit
0 1
NA
NA
T4
Nonmonotonic
quadratic significant fit
0.50 mtn =
7.45 mg/kg
1 26
0 98
1.0
1.26
0 98
0.341
0.370
In (T4)
Nonmonotonic
quadratic significant fit
0.50 mtn =
7.14 mg/kg
1.18
0 92
1.0
1.18
0 92
Morphometry
Control-10% Control
(=31.78), SD = 0.37
Nonmonotonic
quadratic significant fit
Power FCN BMDL
interval includes 0 00
0.19 global min =
6.81 mg/kg
1.053
0 644
1.00
1.053
0.644
In (morph)
Control-10% Control
(= 0.341); SD = 0.37
Nonmonotonic
quadratic significant fit
Power FCN BMDL
computational failures
0 19 global min =
7.01 mg/kg
0 822
0 538
1 00
0.822
0 538
"italics denote estimates derived from nonmonotonic fits to data. FCN = function and SD = standard deviation.
January 16, 2002
7B-10
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-11. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT
(BMDL) ESTIMATES FOR HORMONE DATA OF F1-GENERATION PUPS AT
PND5 IN THE DEVELOPMENTAL NEUROTOXICITY STUDY
(Argus Research Laboratories, Inc., 1998a, and Channel, 1998c)
(Benchmark response based on 10, 20, and 40% changes from control value.)
p of Fit
BMD
BMDL
(10%)
BMD
BMDL
(20%)
BMD
BMDL
(40%)
Mean
NOAEL
T4
0.50"
1.26"
2.89°
BMD set to°
3.41
1.0
0.973°
2.16°
1,000°
ln(T4)
0.50°
0.92"
NC"
NC°
1.0
T3
<0.00001
NC
NC
NC
87.97
0.1
ln(T3)
<0.00001
NC
NC
0.1
TSH
0.50
4.64
9.30
18.61
4.51
3.0
3.77
7.55
15.10
ln(TSH)
0.48
NC
NC
NC
3.0
"Underlined values from nonmonotonic fits to data. (NC = not computed.) The BMDL calculation failed at a
number of values. This means BMDL value may not be accurate.
TABLE 7B-12. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT
(BMDL) ESTIMATES USING THE LINEAR MODEL FIT TO THE MOTOR
ACTIVITY DATA OF Fl-GENERATION PUPS AT PND14 IN THE
DEVELOPMENTAL NEUROTOXICITY STUDY
(Argus Research Laboratories, Inc., 1998a)
(Benchmark response based on 10% change from control value.)
Endpoint
p of Fit
BMD
BMDL
NOAEL/
LOAEL
BMD:
N(L)OAEL
BMDL:
N(L)OAEL
BMR:
10% control
SD
Movement"
0.72
1.94
1.04
None
NA
NA
24.45
162.75
Timeb
0.69
1.33
0.66
None
NA
NA
18.60
184.78
"Number of movements.
"Time spent in activity.
January 16, 2002
7B-11
DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 7B-13. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT
(BMDL) ESTIMATES USING THE POWER MODEL FIT TO THE HORMONE
DATA OF FEMALE RABBITS ON GESTATION DAY 29 IN THE
DEVELOPMENTAL STUDY (Argus Research Laboratories, Inc., 1998c)
(Benchmark response based on 10% change from control value.)
NOAEL/ BMD: BMDL:
Endpoint p of Fit BMD BMDL LOAEL N(L)OAEL N(L)OAEL BMR.
TSH, In TSH NA No effect of
dose
T3, In T3 NA No effect of
dose
T4
0.06
0.54
Lower limit
includes 0
0.1
5.4
NA
0.187
In (T4)
0.0503 1.69
0.002
0.1
16.9
0.02
0.1053
TABLE 7B-14. BENCHMARK DOSE (BMD) AND BMD 95% LOWER LIMIT
(BMDL) ESTIMATES USING THE POWER MODEL FIT TO THE HORMONE
DATA OF FEMALE RABBITS ON GESTATION DAY 29 IN THE
DEVELOPMENTAL STUDY
(Argus Research Laboratories, Inc., 1998c)
(Benchmark response based on 10, 20, and 40% changes from control value.)
p of Fit (10%) (20%) (40%) Mean NOAEL
T4 0.06 0.54 7.05 91.76 1.874 0.1
— — 0.63
ln(T4) 0.05 1.69 10.97 86.19 0.1
0.0018 0.033 7.278
T3 No effect
ln(T3) No effect
TSH No effect
ln(TSH) No effect
January 16, 2002
7B-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
8. SCREENING ECOLOGICAL RISK ASSESSMENT
FOR PERCHLORATE
8.1 INTRODUCTION
As discussed in Section 1.1, perchlorate salts including ammonium, potassium, sodium,
and magnesium perchlorate, are manufactured as oxidizer components for propellants and
explosives. The manufacture or use of perchlorate salts has been reported in most of the states of
the continental United States (Figure 1 -3). In some areas involved with the manufacture, use, or
disposal of perchlorate salts, perchlorate, as the anion dissociated from these salts, has
contaminated soils or ground or surface waters (Figure 1-4). These releases of perchlorate into
the environment have been confirmed to have occurred in 20 states, clustered primarily in the
southwestern United States where most sampling has occurred (Figures 1-3 and 1-4). Currently,
there is a research need to determine whether perchlorate ion is causing any potential effects on
ecosystems or ecosystem components. This chapter presents a screening-level ecological risk
assessment of environmental contamination by perchlorate. In organization, it follows the
Guidelines for Ecological Risk Assessment (U.S. Environmental Protection Agency, 1998c).
8.1.1 Management Goals and Decisions
The discovery that perchlorate release in some sites has contaminated ground and surface
waters in certain locations has raised public and regulatory agency concerns. Much concern has
focused on potential public exposures through drinking water and on the possible needs to
improve analytical and treatment methods and to develop drinking water regulations
(Section 1.4). Consequently, an extensive scientific assessment effort is underway to address
those concerns (Section 1.5). A balanced approach requires assessing ecological effects as well.
The goal of this screening-level ecological risk assessment is, therefore, to indicate the likelihood
that adverse ecological effects (i.e., toxicity to specific organisms or effects on aquatic or
terrestrial ecosystems) will result from observed levels of environmental contamination by
perchlorate. The results of this assessment may be used to address the following questions:
January 16, 2002
8-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
• Are ecological risks best characterized as de minimis (exposures clearly are below levels of
concern), de manifestis (risks are clearly significant and require management action to reduce
exposures); or somewhere in between and requiring further characterization?
• Are analytical detection methods for determining levels of perchlorate in the environment
sufficient, or is it likely that adverse ecological effects occur at levels below current detection
limits?
• Is the available ecotoxicological information on perchlorate sufficient, or are additional studies
needed?
8.1.2 Scope, Complexity, and Focus
In the previous ERD version of this document (U.S. EPA, 1998d), the available
information for this ecological risk assessment was characterized as "very limited" and the
assessment was characterized as "screening-level." Information about the environmental levels
of perchlorate to which organisms were exposed and about its effects on diverse taxonomic
groups was practically nonexistent. Since then, additional information has become available that
improves the database in some respects. Most significantly, additional data are available on
effect levels in aquatic animals, an aquatic plant, a terrestrial plant, and a soil invertebrate; some
of these data are for chronic exposures. Effect levels in rodents have been reevaluated as part of
the human health risk assessment for perchlorate, and the ecological implications of those
changes are reflected herein. In addition, surveys have been conducted at several sites of known
or suspected perchlorate contamination, and environmental and biological materials have been
analyzed for perchlorate. Nonetheless, the level of knowledge on this issue must still be
characterized as limited because the number of species tested is still quite minimal, and the site
surveys focused only on the range of exposures at those sites. This ecological risk assessment is
therefore still a screening-level, rather than definitive, assessment. The materials used in the
1998 ERD and those that are new to this present draft, are described in this section.
Interagency Perchlorate Steering Committee Report. Perchlorate Ecological Risk
Studies is a report of the IPSC's Ecological Risk/Transport and Transformation Subcommittee,
dated November 13, 1998 (Interagency Perchlorate Steering Committee, 1998). This report
presents a literature review on perchlorate toxicity to nonmammalian organisms, recognizing that
few published studies exist, and a rationale for the selection of a battery of ecotoxicology tests
January 16,2002 8-2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
conducted for the USAF Armstrong Laboratory by EA Engineering, Science and Technology,
Inc. It then summarizes those test results, discusses the findings in the context of observed
exposures, discusses uncertainties, and makes recommendations for further study. The present
report constitutes a reevaluation of much of the same information from EPA's perspective,
except that EPA did not examine the open literature studies reviewed by the IPSC subcommittee.
Test Battery Reports. The EA Engineering, Science and Technology, Inc. (1998) final
report, Results of Acute and Chronic Toxicity Testing with Sodium Perchlorate, dated November
1998, details the test methods and results of the ecotoxicology battery. A follow-up report (EA
Engineering, Science and Technology, Inc., 2000) details the test methods and results from
additional chronic toxicity testing with the freshwater amphipod Hyalella azteca and the fathead
minnow Pimephales promelas.
Block Environmental Services, Inc., Report. The report, LC50 Aquatic Toxicity Test
Results for Ammonium Perchlorate—A Two-Species Chronic Definitive Bioassay (Block
Environmental Services, Inc., 1998) presents additional bioassay results that were not included in
the IPSC report.
Algal Toxicity Testing. The EA Engineering, Science and Technology, Inc. (1999) final
report, Results of Algal Toxicity Testing with Sodium Perchlorate, dated September 1999, details
the test methods and results of the ecotoxicological testing with the algae, Selenastrum
capricornutum.
Frog Embryo Teratogenesis Assay: Xenopus (FETAX) Study. The report, FETAX
Analysis of Ammonium Perchlorate (Dumont and Bantle, 1998), prepared by the Department of
Zoology, Oklahoma State University, and dated May 22, 1998, presents results of the Frog
Embryo Teratogenesis Assay: Xenopus (FETAX) conducted with ammonium perchlorate.
Recent data received by the EPA that the Agency has not yet fully reviewed indicate effects on
thyroid function, metamorphosis and sex ratio in developing Xenopus laevis (Goleman et al.,
2002). These data are made available with this document to the external peers for their review.
Phytotransformation Study. Two sets of studies report on the accumulation and potential
degradation of perchlorate by plants. The study, Laboratory Characterization of
Phyto-transformation Products of Perchloroethylene (PCE), Trichloroethylene (TCE) and
Perchlorate (Nzengung, n.d.; Nzengung et al., 1999), examined perchlorate distribution and
degradation in experimental systems containing sand, aqueous perchlorate solution, and rooted
January 16, 2002
8-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
cuttings of woody plant species. This study also examined systems containing chopped leaves or
microbial mats and aqueous perchlorate solution. A second study, Potential Species for
Phytoremediation of Perchlorate (Susarla et al., 1999a; Susarla et al., 2000a), reported
perchlorate depletion from test media over a ten day period by 13 vascular plant species and their
potential for phytoremediation of perchlorate contaminated sites.
Biotransport Investigation Studies. These studies assess the potential for
bioaccumulation of perchlorate in food webs by answering the question of whether perchlorate is
present in biological receptors. The report Scientific and Technical Report for Perchlorate
Biotransport Investigation: A Study of Perchlorate Occurrence in Selected Ecosystems (Parsons,
2001) examined perchlorate concentrations in site media and in various ecological receptors at
six sites with known or suspected perchlorate contamination: (1) sites associated with withdraw
of irrigation water from the Colorado River in the vicinity of Yuma, Arizona; (2) Las Vegas
Wash and Lake Mead near Las Vegas, Nevada; (3) Allegany Ballistics Laboratory, Rocket
Center, West Virginia; (4) Holloman Air Force Base in Otero County, New Mexico; (5) Naval
Surface Warfare Center, Indian Head, Maryland; and (6) Longhorn Army Ammunition Plant,
Karnack, Texas. Additional data are available for one of these sites, Longhorn Army
Ammunition Plant (LHAAP), Texas, in a paper published by Smith et al. (2001). In both studies,
ion chromatography with an AS-16 analytical column was used to measure for perchlorate
concentrations. Analyses with this analytical column have been shown to be superior than other
columns for detecting and quantifying perchlorate (Ellington and Evans, 2000; Susarla et al.,
2000b).
All these sites, except for those in the vicinity of Yuma, are associated with localized
contamination related to the manufacture, handling, or use of perchlorate in solid propellants.
The Yuma sites are approximately 250 miles downstream along the Colorado River from the Las
Vegas Wash and Lake Mead sites; the report suggests that there is no localized source of the
perchlorate; therefore, the most likely potential source of any perchlorate contamination in these
soils is believed to be Colorado River irrigation water. However, portions of the Yuma Proving
Grounds are drained by washes that pass near some of the agricultural locations sampled, and the
information provided in the report was not sufficient for ruling out the possibility of
contamination from the Yuma Proving Grounds.
January 16, 2002
8-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
8.2 PROBLEM FORMULATION
The characteristics of perchlorate and its sources are described earlier in this document
(Chapters 1 and 2). Because this assessment is site independent, this problem formulation
focuses on the selection of assessment endpoints, derivation of the conceptual model, and the
analysis plan.
8.2.1 Assessment Endpoints
In ecological risk assessment, assessment endpoints are operational definitions of the
environmental values to be protected. They are chosen based on policy goals and societal values,
their ecological relevance, and their susceptibility to the stressor and are defined in terms of an
entity and a property of that entity. The assessment endpoints for this ecological risk assessment
are described in the following five subsections.
8.2.1.1 Fish Community Richness and Productivity
Fish communities are valued societally and are ecologically important. The productivity of
these communities is important in terms of the support of fisheries. Species richness is important
in terms of maintaining biodiversity. This importance is reflected by the use of species
sensitivity distributions in the derivation of national ambient water quality criteria and the use of
fish species richness as an important component of bioassessment procedures for enforcement of
the Clean Water Act.
8.2.1.2 Aquatic Invertebrate Community Richness and Productivity
Aquatic invertebrate communities have little direct societal value but are important to
energy and nutrient dynamics in aquatic ecosystems. The productivity of these communities is
important in terms of trophic support of fisheries, of other groups of aquatic species, and of some
terrestrial insectivores. Species richness is important in terms of maintaining biodiversity. This
importance is reflected by the use of species sensitivity distributions in the derivation of national
ambient water quality criteria and the use of invertebrate species richness as an important
component of bioassessment procedures for enforcement of the Clean Water Act.
January 16, 2002
8-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
8.2.1.3 Aquatic Plant Richness and Productivity
Algae and other aquatic plants have little direct societal value but are important to energy
and nutrient dynamics in aquatic ecosystems. Species richness is important in terms of
maintaining biodiversity. Because of their importance to the trophic support of fisheries and
other aquatic consumers, productivity is an important endpoint for this assemblage.
8.2.1.4 Soil Invertebrate Community Richness and Productivity
Soil invertebrate communities have little direct societal value, but, in nearly all terrestrial
ecosystems, they are important to energy and nutrient dynamics and to maintenance of soil
structure. The productivity of these communities is also important in terms of trophic support of
some terrestrial insectivores. Species richness is important in terms of the policy of maintaining
biodiversity.
8.2.1.5 Terrestrial Plant Richness and Productivity
Terrestrial plants are valued highly by society for production of food, fiber, and timber, as
well as their aesthetic value. The primary valued property of terrestrial plants is their
productivity. As autotrophs, plants are the basis of energy and nutrient dynamics in most
terrestrial or aquatic food webs. Moreover, species richness is important in terms of the policy of
maintaining biodiversity.
8.2.1.6 Population Productivity of Herbivorous Wildlife
Herbivorous wildlife are included as an endpoint entity because of the apparent
bioconcentration of perchlorate in plant foliage. The meadow vole (Microtus pennsylvanicus) is
used as a representative species for this group. Population productivity is used as the endpoint
property because growth and reproduction are generally sensitive properties and because
herbivores are valued for their production of food for human and nonhuman carnivores.
8.2.2 Conceptual Models
The conceptual model describes the relationships between sources of perchlorate and the
endpoint receptors (Figure 8-1). Sources include spills during the flushing of rockets; the
combustion of rocket fuel; the improper disposal of rocket fuel, open burn or open detonation
January 16,2002 8-6 DRAFT-DO NOT QUOTE OR CITE
-------
p
3
g
ca
Os
K)
O
o
N)
00
1
H
6
o
2
o
H
o
C!
o
H
W
o
o
3
M
Perchlorate
Discharge
i
Surface
Water
-I ,
f
1
f
Aquatic
Invertebrates
Figure 8-1.
A conceptual model of exposure of ecological endpoint receptors to perchlorate. Specific endpoint taxa are
identified in italics; all other endpoints are defined at the community level. Processes are designated by
hexagonal boxes, compartments by rectangular boxes.
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
operations, explosives, or manufacturing wastes; and the aqueous discharge of waste water from
manufacturing of perchlorate. The most recent information on perchlorate content in fertilizers
demonstrates that fertilizer use is unlikely to constitute an environmentally significant source of
perchlorate contamination, and ecological risks from this source are not considered further (see
Chapter 9). Spills contaminate the soil at the site and, through leaching and run-off, contaminate
the surface water and groundwater. The discharge of groundwater to surface water may result in
locally high levels of perchlorate in surface waters. Aquatic communities are exposed directly to
contaminated surface water; soil invertebrate and plant communities are exposed to perchlorate
in soil at the spill site and through irrigation with either surface or groundwater; and herbivorous
terrestrial wildlife are exposed through their consumption of plants that have bioconcentrated
perchlorate. However, the potential for transfer of perchlorate further up the terrestrial food web
is currently unknown.
This conceptual model is relatively simple because it excludes some potential routes and
receptors. Dietary exposures are excluded from aquatic systems because, as of this writing,
available data have not shown perchlorate to bioconcentrate to any significant extent.
Information newly received form the U.S. Army Corps of Engineers (Condike, 2001) report on
the analysis of environmental samples from perchlorate-contaminated water bodies near
McGregor Naval Weapons Industrial Reserve Plant (NWIRP), TX, and purports to show fish
tissue concentrations that exceed comparable water concentrations. These data suggest that
perchlorate not only accumulates but is bioconcentrated. This information, which has not yet
been fully reviewed by the U.S. EPA, is herewith made available with this document to external
peers for their review.
Wildlife are assumed to have negligible exposure from air or from direct exposure to soil.
Exposures of wetlands to groundwater or surface water are not included explicitly because their
exposures and effects are assumed to be equivalent to irrigation exposures. That is, plants and
invertebrates are assumed to be exposed to pore-water concentrations equal to surface or
groundwater concentrations. Exposures to contaminated sediments also are not included
explicitly because they are believed to be equivalent to surface water exposures. Perchlorate salts
are highly water soluble and the anion is unlikely to adsorb to anionic particles, such as soils or
humic substances, to a significant extent. Therefore, sediment exposures are expected to be
dominated by exposure to pore water, which is assumed to be equal to surface water.
January 16, 2002
8-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
8.2.3 Analysis Plan
This screening assessment uses existing information to determine whether the existing
environmental contamination by perchlorate poses a clearly significant risk, insignificant risk, or
an ambiguous risk. The analysis of effects will consist of the derivation of screening benchmarks
through the application of conservative extrapolation models. The analysis of exposure for
ecological endpoints consists of measured concentrations reported in Chapter 1 or derived from
Parsons (2001) or Smith et al. (2001). Soil exposure estimates are based on exposure to
perchlorate in irrigation water.
8.3 ANALYSIS
8.3.1 Characterization of Exposure
8.3.1.1 Water Concentrations
As previously described, fishes, aquatic invertebrates, and aquatic plants may be exposed
directly to concentrations of perchlorate in surface waters. These concentrations may result from
surface run-off from perchlorate-contaminated soil, from leaching of perchlorate from
contaminated soil via shallow groundwater, or from direct discharge of aqueous wastes. Surface
or groundwater may be used for irrigation, resulting in direct exposure of soil invertebrates or
plants (Figure 8-1).
Perchlorate salts are dissolved readily given the conditions under which the contamination
has occurred, releasing the perchlorate anion and the associated cation. Sorption is not expected
to attenuate perchlorate because it absorbs weakly to most soil minerals, and natural chemical
reduction in the environment is not expected to be significant. Consequently, perchlorate is both
very mobile in aqueous systems and persistent for many decades under typical ground and
surface water conditions (Section 1.1).
Limited information is available on perchlorate concentrations in surface waters.
Perchlorate from an ammonium perchlorate manufacturing area has been detected at 4 to 16 jj.g/L
downstream in Lake Mead and the Colorado River (Section 1.2). Information on the frequency
or central tendency (mean or median) of perchlorate detection in those water bodies was not
available for this review, but it is assumed that some aquatic organisms are exposed chronically
January 16, 2002
8-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
to concentrations as high as 16 /ugfL. On the other hand, perchlorate concentrations have been
measured as high as 0.37% (37 * 106 /ig/L) in groundwater-monitoring wells at facilities that
manufacture or test rocket motors and at 280 //g/L in public water supply wells (Section 1.2)
Smaller surface water bodies, including some that are supplied primarily by groundwater, are
likely to exist near sites of soil contamination and to have perchlorate concentrations much
higher than those reported for Lake Mead and the Colorado River. A spring associated with the
Las Vegas Wash site had concentrations of 1.0 to 1.3 x io5 //g/L in surface water (Parsons,
2001). Perchlorate concentrations in a pond (INF Pond) that receives water from the pump and
treat system at the Longhorn Army Ammunition Plant near Karnack, TX ranged from 30,776 to
31,438 (J-gfL in November 1999 (Smith et al., 2001) and ranged from 3500 to 3800 pcgfL in
September 2000 (Parsons, 2001). It is also possible that, within large water bodies, there are
locally elevated concentrations at sites of groundwater discharge. In the vicinity of a sediment
delta created by the Las Vegas Wash in Las Vegas Bay of Lake Mead, Parsons (2001) documents
a maximum perchlorate concentration of 68 //g/L in surface water. At the Allegany Ballistics
Laboratory in Rocket Center, WV, discharge water from a Comprehensive Environmental
Response Compensation, and Liability Act (CERCLA) groundwater pump and treat facility to
the North Branch Potomac River contained 250 to 280 /j-g/L perchlorate (Parsons 2001). Surface
water concentrations in Town Gut Marsh adjacent to the Naval Surface Warfare Center at Indian
Head, MD ranged from not detected (reporting limit = 4.0 u-gfL) to 25 /ig/L. It should be noted
that the groundwater pump and treat facilities either at Longhorn Army Ammunition Plant or
Allegany Ballistics Laboratory were not equipped with facilities to treat perchlorate in water.
Surface water concentrations in Harrison Bayou below the discharge point for the INF pond
at LHAAP also ranged from undetectable (reporting limit = 4.0 /ugfL) to 4.0 //g/L (Parsons,
2001; Smith et al., 2001). However, Smith et al. (2001) point out that water from the pond is
discharged to Harrison Bayou only during periods when Harrison Bayou is flowing, and neither
study apparently sampled Harrison Bayou when water was being discharged from the pond.
Therefore, higher concentrations of perchlorate in surface water of Harrison Bayou are likely to
be measured at other times.
It is assumed that irrigation waters pumped from Lake Mead or the Colorado River are in
the range of downstream concentrations given above (i.e., 4-16 //g/L). Groundwater irrigation
January 16, 2002
8-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
may be contaminated at levels similar to those observed in public water supplies (^280 ng/L),
unless the well is appreciably nearer a perchlorate-contaminated site.
8.3.1.2 Aquatic Bioaccumulation
As discussed above, little information has been previously available on the potential for the
perchlorate ion to accumulate in animal tissues. The studies outlined in the Parsons (2001)
report sought to answer the question whether perchlorate is present in ecological receptors.
In these studies, concentrations of perchlorate in aquatic vegetation, fish, amphibians, aquatic
invertebrates, and birds were compared to surface-water, pore-water, and sediment
concentrations from the same water body. This information is supplemented by the additional
studies conducted at LHAAP by Smith et al. (2001).
When perchlorate concentrations in physical media (i.e., surface water or sediment) were
greater than the reporting limits for biological media (£300 ppb [//g/L or /ig/kg] in Parsons
[2001]), concentrations in aquatic vegetation were similar to or greater than the concentrations in
surface water or pore water; but concentrations in fish, amphibians, or invertebrates were less.
In Smith et al. (2001) reported the detection of high concentrations of perchlorate in the INF
Pond and lower concentrations in aquatic vegetation and in animals than in surface water or
sediments.
In Parsons (2001), when perchlorate concentrations in the physical media were lower,
concentrations in aquatic vegetation or amphibians were in a few cases greater than the
concentrations in surface water or sediment; but in most cases, perchlorate was not detected in
aquatic receptors. However, our understanding of bioaccumulation of perchlorate in this lower
concentration range is limited because the reporting limits in the Parsons (2001) studies for
perchlorate in animal tissues (i.e., 300-400 //g/kg) were greater then the reporting limits for
surface water or pore water (i.e., = 4 fx g/L) or for sediments (i.e., = 80 yUg/L).
Although Smith et al. (2001) do not identify their reporting limits, their reporting limits for
biological tissues appear to be less (i.e., =70 /ugfkg based on their lowest detected concentration)
than those of Parsons (2001). In the Smith et al. (2001) study of LHAAP, detected perchlorate
concentrations were similar in surface water (44-85 //g/L), sediments (78 /Ug/kg), and fish tissues
(83-131 Aig/kg) at Goose Prairie Creek. In Harrison Bayou, the single detected concentration in
surface water (4 //g/L) was less than detected concentrations in animal tissues (86-356 /u.gfkg).
January 16, 2002
8-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
However, as the authors discuss, the measured concentration in surface water in Harrison Bayou
is likely less than when water is being discharged from the INF Pond (Smith et al., 2001). In
addition, the study did not collect sufficient samples from any one site and medium or species for
any significant statistical comparisons to be made.
Information newly received form the U.S. Army Corps of Engineers (Condike, 2001) report
on the analysis of environmental samples from perchlorate-contaminated water bodies near
McGregor Naval Weapons Industrial Reserve plant (NWIRP), TX, and purports to show fish
tissue concentrations that exceed comparable water concentrations. These data suggest that
perchlorate not only accumulates but is bioconcentrated. This information, which has not yet
been fully reviewed by the U.S. EPA, is herewith made available with this document to external
peers for their review.
The above information indicates that perchlorate may bioaccumulate in aquatic organisms
living in contaminated waters, but it does not resolve the question of whether perchlorate may
bioconcentrate in the tissues of aquatic organisms to levels exceeding the surface water
concentrations. The existing data are also insufficient to determine whether there is further
trophic transfer of perchlorate within aquatic food webs.
8.3.1.3 Soil Levels
On-site soils may be contaminated by direct spills of perchlorate solutions from flushing
rockets, combustion of rocket fuel, improper disposal of rocket fuel, open burn/open detonation
operations, explosives, or manufacturing wastes. Perchlorate concentration measurements at
disposal sites range from less than 1 to 1470 mg/kg (Parsons, 2001). Off-site soils may be
contaminated via irrigation (Figure 8-1). Because of the high water solubility of perchlorate
salts, perchlorate is unlikely to accumulate via adsorption to irrigated soils, and aqueous
perchlorate was not found to adsorb to sand in laboratory reactors (Nzengung, n.d.). By gross
approximation, then, soil concentrations (expressed as milligrams per kilogram) would be
unlikely to exceed the concentrations (expressed as milligrams per liter) in irrigation water.
Similarly, concentrations of perchlorate in soil pore water may be assumed to be equal to the
concentration in irrigation water, both in the field and in soil toxicity tests. However, the
concentration of perchlorate salts in irrigated soils with high evaporation rates cannot be ruled
out. At the Yuma site, soils are irrigated with water from the Colorado River, and concentrations
January 16, 2002
8-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of perchlorate in surface-water samples collected near the irrigation intake locations ranged from
0.003 to 0.006 mg/L. In surface soil, the single detection (0.090 mg/kg) was well above the
concurrently-measured water concentrations, as were the perchlorate detection limits in soil
(0.079 to 0.080 mg/kg). The relatively higher detection limits in soil, the limited nature of the
sampling in soil and water, and the lack of information about potential sources other than
irrigation water (see Section 8.1.2) complicate the interpretation of the presence and fate of
perchlorate in irrigated soils.
8.3.1.4 Uptake by Vegetation
Several laboratory experiments have examined plant uptake of perchlorate from solution
culture. Experiments with candidate plants for use in the phytoremediation of perchlorate-
contaminated sites showed that perchlorate may concentrate in vegetation (Nzengung, n.d.;
Susarla et al., 2000a). Nzengung (n.d.) used rooted cuttings of woody plants, willow (Salix spp.),
Eastern Cottonwood (further identified only as "poplar"), and eucalyptus (Eucalyptus cineria)
planted in sand with nutrient solution containing perchlorate at 20 or 100 mg/L for 24 to 42 days.
In each case, perchlorate was taken up and concentrated in aerial plant parts, especially leaves.
Concentration factors, expressed as the ratio of leaf concentration (mg/kg wet weight) to initial
solution concentration (mg/L), ranged from 7.5 to 25.
Susarla et al. (2000a) used seedlings or rooted cuttings of 13 vascular plant species, planted
in sand with nutrients, and exposed for ten days to 0.2, 2.0 or 20 mg/L perchlorate. These
researchers also reported depletion of perchlorate from test media. Qualitative analyses
suggested accumulation of perchlorate in the aerial tissues of most of the species analyzed.
Using their data and the approach reported by Nzengung above, we calculated concentration
factors ranging from 0 to 330.
Nzengung (n.d.) and Susarla et al. (2000a) reported that perchlorate accumulated primarily
in the leaves, followed by stems, then roots. Predicted perchlorate breakdown products, chlorate,
chlorite, and chloride were detected in plant tissues in both studies, but quantitative evidence was
not presented. In addition to this lack of quantitative data, there are other concerns related to the
potential for plants to degrade perchlorate. First, information concerning accumulation and
potential transformation is limited to a few studies by these two laboratories. Second, the
method used for perchlorate analysis yielded estimates of perchlorate in fertilizer that were
January 16, 2002
8-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
subsequently found to be overestimated by 30 to 150% (Susarla et al., 2000b). Third, no
physiological explanation has been suggested for why plants should accumulate this salt far in
excess of concentrations in water or groundwater, though it appears this may be simply a
function of water uptake rates to meet transpirational losses. Fourth, these two studies were
short-term, material depletion studies, a type of study we believe will overestimate long-term
accumulation rates because some of the "response" is likely the result of factors not related to the
chemical in question. There is ample evidence from salt accumulation studies of plants to
suggest that the initial increases in perchlorate accumulation by plants may be due to a salt effect;
that is, nutrient salts are initially indistinguishable from perchlorate salts in that they all represent
an ionic imbalance across the cell wall. One of the confounding issues that can only be
determined with longer-term studies is the effect of increased cell sap salinity on additional
perchlorate uptake. As sap salinity increases, there should be an increase in H20 uptake, further
increasing the perchlorate concentrations. This will continue only until a certain concentration of
salts, including perchlorate, is reached, at which time the plant will close its stomata and shunt
sap salts to vacuoles.
In addition to the above stated concerns, there is no reason to expect that these are steady-
state concentration factors. These experiments were designed to quantify phytotransformation of
an initially introduced perchlorate quantity, rather than bioconcentration, with respect to an
ambient perchlorate concentration. As the perchlorate-amended nutrient solution was transpired,
and some perchlorate was taken up, it was replenished by solution, without added perchlorate;
thus, perchlorate in the test chamber diminished throughout the experiment. Concentration
factors that would be observed at steady state, such as may result from continual irrigation with
perchlorate-contaminated water, cannot be estimated from this study. Because of the
uncertainties associated with both perchlorate accumulation and degradation by plants, a simple,
conservative, screening-level assumption that concentrations in leaves can exceed water
concentrations by a factor of 100 was made.
If irrigation is from surface water sources similar to the Colorado River or Lake Mead, with
concentrations as high as 16 /Ug/L, then plant concentrations are assumed to be as high as
1.6 mg/kg. If irrigation is from groundwater sources similar to potable water supplies, with
concentrations as high as 280 //g/L, then plant concentrations will be assumed to be as high as
28 mg/kg.
January 16, 2002
8-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Concentrations in plant tissues and soils also have been measured in the field. Ellington
et al. (2001) measured perchlorate concentrations in leaves of tobacco, Nicotiana tabacum var.
K326, field-grown in soil amended with Chilean saltpeter, which is naturally high in perchlorate.
Perchlorate concentrations (± SD) in leaf lamina from the 1999 crop were 96.0 ± 0.6 mg/kg dry
weight and 14.6 ± 0.1 mg/kg wet weight; concentration in a composite soil sample collected in
December 1999 was 0.34 ± <0.01 mg/kg dry weight. The concentration factor in this study was
approximately 43, on the basis of wet weight in leaf lamina and dry weight in soil.
The field studies by Parsons (2001) found that, for various sites, wet-weight perchlorate
concentrations in terrestrial vegetation samples were 1.5 to 80 times the wet-weight
concentrations in soil samples. The data from one site (i.e., Building 25C) at LHAAP (Smith
et al., 2001) seem to indicate greater concentration factors, but the soil and plant samples were
taken at different times of the year (i.e., January and October, respectively) and only one sample
each of three plant species was analyzed.
Soil-to-vegetation concentration factors derived from the above field studies were similar
in magnitude, but when using them for risk assessment care should be taken to note the different
bases; exposure concentration was variously reported as mg/kg wet weight in soil or mg/kg dry
weight in soil. The maximum measured concentration in vegetation at irrigated sites in the
vicinity of Yuma, Arizona was 1.0 mg/kg wet weight. At sites with soil contamination related to
the manufacture, handling, or use of perchlorate in solid propellants, maximum plant
concentrations were 428 mg/kg wet weight at a spring; 99.2 mg/kg wet weight at a site upstream
from Lake Las Vegas in the Las Vegas Wash area of the Lake Mead Recreational Area, Nevada;
and 300 mg/kg wet weight at the Burn Area of Allegany Ballistics Laboratory, West Virginia.
In most cases, detection limits were ~0.4 mg/kg wet weight.
8.3.1.5 Herbivore Exposure
The representative herbivore selected for this assessment, M. pennsylvanicus, has a diet
consisting mainly of monocot and dicot shoots, has an estimated food consumption rate of
0.005 kg/day wet weight, and a body weight of 0.044 kg (Sample and Suter, 1994). Using the
assumptions stated above, daily exposures resulting from surface water and groundwater
irrigation may be as high as 0.18 mg/kg-day and 3.2 mg/kg-day, respectively. Daily exposures
resulting from maximum measured concentrations in plants range from 0.11 mg/kg-day at the
January 16, 2002
8-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
irrigated sites in the vicinity of Yuma to 49 mg/kg-day for the sites with direct soil
contamination.
In the Parsons (2001) studies, except when concentrations in surface soils were high (i.e.,
2:9000 //g/kg), perchlorate was not detected in terrestrial birds, mammals, or insects with
reporting limits of 300 to 400 //g/kg. The vertebrates collected varied substantially between
sites, but the birds collected include the mourning dove (Zenaida macroura), tree swallow
(Tachycineta bicolor), roughwing swallow (Stelgidopteryx serripennis), lesser nighthawk
(Chordeiles acutipennis), nighthawk (C. minor)-, Gambel's quail (Callipepla gambelii), starling
(Sturnus vulgaris)', American crow (Corvus brachyrhynchus), eastern bluebird (Sialia sialis),
eastern phoebe (Sayornis phoebe), and blue grosbeak (Guiraca caerulea). The mammals
collected include the cactus mouse (Peromyscus eremicus), rock pocket mouse (Chaetodipus
intermeius), Audubon's cottontail (Sylvilagus audubonii), deer mouse (P. maniculatus), long-
tailed pocket mouse (Perognathus formosus), western pipestrelle (Pipistrellus hesperus), house
mouse (Mus musculus), white-footed mouse (Peromyscus leucopus) meadow vole (Microtus
pennsylvanicus), Merriam's kangaroo rat (Dipodomys merriami), desert pocket mouse
(C. penicillatus), hispid cotton rat (Sigmodon hispidus), western harvest mouse (Reithrodontomys
megalotis), marsh rice rat {Oryzomys palustris); northern short-tailed shrew (Blarina
brevicauda), racoon (Procyon lotor), eastern harvest mouse {R. fulvescens), little brown bat
(Myotis lucifugus), eastern cottontail (S. floridanus). At those sites where perchlorate
concentrations in surface soils were high, perchlorate concentrations in potential herbivore
tissues were generally an order of magnitude or more less than that in vegetation. At one site,
Yuma, with lower perchlorate concentrations in soil (i.e., mean of all results = 81 //g/kg),
perchlorate was detected in a single terrestrial reptile sample (brush lizard, Urosaurus graciosus),
but this detection was lower than the mean perchlorate concentration in vegetation. Although
detected soil concentrations were lower (i.e., 50 to 322 /ig/kg) in Smith et al. (2001), the
concentrations of perchlorate in two composite samples of livers from harvest mice
(Reithrodontomys fulvescens) were several orders of magnitude less than the detected
concentrations in their potential food, plant leaves or seeds.
January 16, 2002
8-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
8.3.2 Characterization of Effects
8.3.2.1 Aquatic Organisms
Effects on the richness and productivity of fish, aquatic invertebrate, and aquatic plant
communities are jointly characterized using the procedures for deriving Tier II water quality
values (U.S. Environmental Protection Agency, 1993; Suter and Tsao, 1996). Tier II values are
derived where data are not sufficient for deriving ambient water quality criteria (AWQC). The
Tier II value derivation procedures account for missing information with approximately 80%
confidence.
Test results potentially useful for deriving Tier II values were available for five aquatic
species (Table 8-1). In acute tests (48 and 96 h) with sodium perchlorate, using the water flea
Daphnia magna, the amphipod Hyalella azteca, and the fathead minnow Pimephales promelas,
the endpoints lethality and inhibition were studied. In seven-day tests with a different water flea
(Ceriodaphnia dubia) and with P. promelas, acute lethality was studied in addition to more
sensitive endpoints, including the number of offspring per female (C. dubia) and growth rate
(i.e., body weight; P. promelas). A seven-day test with C. dubia generally is here used in place
of a chronic (i.e., twenty-one day) test because test organisms produce three broods during the
test; a seven-day test with P. promelas is arguably subchronic because of the test's short duration
relative to the organism's lifespan (Suter, 1990; Norberg-King, 1990). A 35-day, early-life-stage
(ELS) test with Pimephales, here used in place of a chronic test, showed no significant effects on
any standard endpoint (survival, growth or biomass) at the highest concentration tested.
However, all larvae exposed to perchlorate concentrations, including the lowest concentration of
28 mg/L, exhibited redness and swelling that was not observed in the larvae exposed to the
control water. This finding suggests the presence of subtle effects that could be ecologically
significant and raises doubt about whether a chronic No-Observed-Effect-Concentration (NOEC)
has been adequately determined for this species.
Steps followed in the derivation of the Tier II value for sodium perchlorate are presented in
Table 8-2. The secondary acute value (SAV), 5 mg/L (as C104"), is derived to be protective of
95% of species during short-term exposures with 80% confidence. The secondary chronic value
(SCV), 0.6 mg/L (as C104 ), likewise is derived to be protective of 95% of species during
long-term exposures with 80% confidence. A sodium chloride control test showed that some of
the toxicity to P. promelas was potentially attributable to the sodium cation. These tests suggest
January 16, 2002
8-17
DRAFT-DO NOT QUOTE OR CITE
-------
£>
3
C
CT\
K>
O
O
K)
TABLE 8-1. RESULTS OF PERCHLORATE TOXICITY TESTS IN AQUATIC AND TERRESTRIAL SPECIES
Test Description Endpoints (as mg/L pcrchlorate in water or mg/kg in soil or sand)*
Test Species
Age
Duration
Acute LC50
(95% CL)
NOEC
LOEC
ChV
IC2S
(95% CL)
Sodium pcrchlorate (NaC104)b tests (EA Engineering, Science and Technology, Inc., 1998)
Daphnia magna
<24 hr
Acute (48-hr)
490
(406-591)
—
—
—
—
Pimephales promelas
12-13 days
Acute (96-hr)
1,655
(1,507 - 1,817)
—
—
—
—
Ceriodaphnia dubia
<24 hr
Chronic (7-day)
66
(40-144)
[48-h]
10
33
18.2
17
(8.1 -20.5)
Pimephales promelas
<24 hr
Subchronic (7-day)
614
(540-714)
[96-h]
155
280c
208c
212°
(175 - 231 )c
Lactuca saliva
<24 hr
Subchronic (7-day)
614
(540-714)
[96-h]
155
280c
208c
212c
(175 - 23 l)c
Lactuca saliva
Chronic definitive
(28-d, sand)
<80
80
<80
41
Lactuca saliva
Chronic definitive
(28-d, soil)
40
40
56.6
30
Lactuca sativa
Chronic definitive
(28-d, sand)
20
40
28.3
34.3
Eisenia foetida
Acute definitive
(7 day/14 day, soil)
4,450/4,450
—
—
—
—
Sodium pcrchlorate (NaC10„)b tests (EA Engineering, Science and Technology, I
nc., 2000)
Pimephales promelas
Embryo
Chronic
(35-day, early life
stage)
>490
[96-hr]
> 490d
<28c
> 490d
28c
> 490d
>490d
<28c
Hyalella azteca
7-14 days
Chronic definitive
(28-day)
> 1000
> 1000
> 1000
> 1000
00
1
00
H
I
o
o
o
H
o
c
o
H
m
o
fit
o
M
H
ffl
-------
fa
3
e
p
ON
K)
©
©
N)
TABLE 8-1 (cont'd). RESULTS OF PERCHLORATE TOXICITY TESTS IN AQUATIC AND TERRESTRIAL SPECIES
00
I
vo
Test Description Endpoints (as mg/L perchlorate in water or mg/kg in soil or sand'"
Test Species
Acute LCS0
Age Duration (95% CL) NOEC LOEC
ChV
ic25
(95% CL)
Sodium perchlorate (NaC104)b tests (EA Engineering, Science, and Technology, 1999)
Selenastrum capriconutum
7 days Acute (96-hr) — 500 1,200
775
615
(149-1,126)
Ammonium perchlorate (NH4C104)f tests (Block Environmental Services, Inc., 1998)
Ceriodaphnia dubia
<24 hr8 Chronic (6-day) 77.8 9.6 24
[6-days]
15
24
Pimephales promelas
<24 hr8 Subchronic (7-day) 270 9.6 96
[7-days]
30
114
Ammonium perchlorate (NH4CI04)f tests (Dumont and Bantlc, 1998)
Xenopus
Embryo 96-hr 420h — —
—
—
Xenopus
Embryo 96-hr 336h — —
—
—
H
i
O
o
o
H
O
C!
O
H
W
O
*
o
M
H
m
"Notation: LC50 = Concentration lethal to 50% of individuals; NOEC = No-observed-effect concentration; LOEC = Lowest-observed-effect concentration;
ChV = Chronic value; IC25 = Concentration inhibiting a process (e.g., growth, reproduction) by 25%; CL = confidence limits.
bSodium chloride control showed no adverse effects of sodium ion except as indicated. Reported values are based on nominal concentrations.
cSodium chloride control showed significant adverse effects attributable to sodium cation at highest test concentration. Effects observed at this perchloratc
concentration may have been caused in part by sodium ion toxicity.
dStandard endpoints: survival, growth, biomass
"Although there were no effects on standard endpoints at any tested concentration, the investigators reported that all larvae exposed to perchlorate
concentrations, including the lowest concentration of 28 mg/L, exhibited redness and swelling, which was not observed in the larvae exposed to the control
water.
fAmmonium control was not used; adverse effects of ammonium ion cannot be ruled out at all effect concentrations. C. dubia and P. promelas results are
based on measured concentrations. Xenopus results arc based on nominal concentrations. Confidence limits are not reported.
Wot reported; assumed based on standard protocols.
hlC50 for malformations.
-------
(a
3
C
fa
On
to
o
o
KJ
TABLE 8-2. PROCEDURE FOR DERIVING TIER II WATER QUALITY VALUES FOR SODIUM PERCHLORATE
Step
Value
(mg/L CIO/)
Rationale
oo
¦
O
O
H
I
0
o
o
H
o
c
o
H
tn
o
*3
O
HH
H
Identify the lowest genus mean acute value (GMAV)
Determine the final acute value factor (FAVF), a factor that
compensates for lack of data on a sufficient number of
taxonomic groups
Calculate the secondary acute value (SAV)
Identify three or more acute-chronic ratios (ACRs), which
are ratios of acute value (AV) to chronic value (CV) for a
given species (but ratios must be geometrically averaged
within any single genus)
Derive the secondary acute-chronic ratio (SACR)
Derive the secondary chronic value (SCV)
66 Lowest GMAV is for genus Ceriodaphnia (based on C. dubia)
13.2 The FAVF varies according to the number of specified taxonomic groups
(unitless) for which GMAVs were available. In this case, two specified values were
available (a nonsalmonid fish and a planktonic crustacean), of which one is
a daphnid; the value selected from the FAVF table (U.S. Environmental
Protection Agency, 1993; Sutcr and Tsao, 1996) is 13.2.
5.0 SAV=GMAV-FAVF = 66-13.2
3.6,8.0 ACRs can be derived for two species in different genera. For C. dubia:
(range, <3.4 - ACR=AV - CV = 66 - 18.2 = 3.6
>59), 17.9 For/5 promelas, two AVs arc available. The lower (614) is thrown out
because the larval stage is not standard for acute tests; the higher (1,655)
is used. Three CVs are available: >490 for standard endpoints, and <28 for
redness and swelling, in the 35-d ELS test; and 208 for survival in the 7-d
test. There is uncertainty as to the interpretation of the ELS test results; the
7-d result is used and the two results from the ELS arc used to determine a
range, shown in parentheses:
ACR= 1,655 - 208 (range, >490 - <28) = 8.0 (range, <3.4 - >59)
No ratio is possible for H. azteca because we are unable to calculate CV
due to no acute toxicity. Because a third value is not available, a default
value of 17.9 (which provides 80% confidence based on other toxicants) is
substituted, according to the Tier II method.
8.0 The SACR is the geometric mean of the ACRs. (The uncertainty range
(range, <6.0 - associated with the P. promelas value is carried through and shown in
>15.6) parentheses.)
0.60 SCV=SAV - SACR, 5.0 - 8.0
(range, <0.32 (The uncertainty range associated with the P. promelas value is carried
- >0.83) through and shown in parentheses.)
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
the possibility that if perchlorate were associated with a less toxic cation, the SCV may be lower
than is necessary to protect against perchlorate ion toxicity. Further tests with perchlorate may be
needed to assess potentially less toxic cations.
Similar chronic (or subchronic) tests were conducted with ammonium perchlorate
(Table 8-1). Results, expressed as C104", were very similar for C. dubia, but P. promelas was
more sensitive to ammonium perchlorate than to sodium perchlorate. However, Tier II values for
ammonium perchlorate are not presented for several reasons, including the lack of ammonium
controls which makes it difficult to determine whether the observed effects were caused by the
perchlorate anion; the lack of acute values for C. dubia and P. Pimephales; and the fact that the
FETAX (Xenopus) test is designed to detect teratogenic potential, and the embryo is not a
particularly sensitive life stage for toxicity. When perchlorate is administered as the ammonium
salt, the ammonium ion concentration expressed on an ammonia-nitrogen (in milligrams of
NH3-N/L) basis is 14% of the respective perchlorate ion concentration. A Lowest-Observed-
Effect-Concentration (LOEC) for C. dubia of 24 mg/L perchlorate (Table 8-1) thus corresponds
to 3.4 mg NH3-N/L. Based on a species mean chronic value (SMCV) of 13 mg NH3-N/L for
C. dubia exposed to ammonia alone (U.S. Environmental Protection Agency, 1998h), the former
value is probably too low to be responsible for the observed effects'. On the other hand, the
LOEC observed with P. promelas at 96 mg/L (Table 8-1) corresponds to 14 mg NH3-N/L, which
exceeds a SMCV of 3.09 mg NH3-N/L (U.S. Environmental Protection Agency, 1998h).
Therefore, ammonium exposure alone could have been responsible for the effects of ammonium
perchlorate observed in P. promelas.
The SAV and SCV derived above based on sodium perchlorate are probably protective
even if ammonium perchlorate is the form released, however. Calculated NH3-N concentrations
corresponding to those values are below the acute and chronic ambient water quality criteria for
ammonia, regardless of pH (U.S. Environmental Protection Agency, 1998h). While SAV and
SCV are not calculated for plants, it appears that there is little perchlorate or ammonium toxicity
to the alga Selenastrum in toxicity studies (Table 8-1).
'Ammonia/ammonium toxicity increases as test-water pH increases (U.S. Environmental Protection
Agency, 1998e). The value of 13 mg NH3-N/L corresponds to a pH of 8.0; however, unless the test water pH had
exceeded 8.8, it is doubtful that 3.4 mg NH3-N/L was responsible for the observed effects.
January 16, 2002
8-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
8.3.2.2 Terrestrial Organisms
Plants. The only available phytotoxicity information comes from 28-day seedling growth
tests of lettuce (Lactuca sativa) performed in soil and sand cultures with sodium perchlorate (EA
Engineering, Science and Technology, Inc., 1998). Although the exposure was to sodium
perchlorate solution added to the solid media, the results may be expressed as milligrams per
kilogram soil or sand, wet weight, or as milligrams per liter of irrigation solution. Growth was a
more sensitive response than germination or survival. The quartile inhibitory wet-weight
concentrations (IC25s) for growth in soil and sand were 78 mg/kg (293 mg/L) and 41mg/kg
(160 mg/L), respectively. Survival was reduced 26% at 420 mg/kg (2,520 mg/L) in soil and 39%
at 180 mg/kg (840 mg/L) in sand. To account for interspecies variance, a factor of 10 is applied
to the lowest IC25 to obtain a screening benchmark of 4 mg/kg as a wet-weight concentration in
soil (or 16 mg/L as a concentration in irrigation solution).
Soil Invertebrates. The only available toxicity data for soil invertebrates is a 14-day acute
lethality test of the earthworm (Eisenia foetida) performed in artificial soil irrigated with sodium
perchlorate. The LC50 at both 7 and 14 days was 4,450 mg/kg as a wet-weight concentration in
soil. No factors or other models are available to extrapolate from that LC50 to chronic effects on
survival, growth, or fecundity or to extrapolate from this species to the soil invertebrate
community as a whole. Therefore, the factors applied for aquatic communities in cases where
there is only one LC50 (see Section 8.3.2.1) to obtain a conservative estimate of a soil screening
benchmark for soil community effects, are as follows:
Threshold = LCS0 - (factor for interspecies variance * acute-chronic ratio)
4,450 mg/kg-(242 x 18)
= 1 mg/kg as a wet-weight concentration in soil.
The equivalent aqueous phase benchmark is 2.8 mg/L. This approach requires the assumptions
that the variance among soil species is approximately the same as among aquatic species, and
that the distribution of acute-chronic ratios across chemicals is approximately the same for both
communities. The interspecies variance factor is the one for a test species that has not been
demonstrated to be highly sensitive.
Herbivores. The human health risk assessment for perchlorate uses 0.01 mg/kg-day as the
LOAEL from which the RiD is derived (Chapter 7). That value is based on perturbations in
thyroid and pituitary hormones, thyroid histopathology, and changes in brain morphometry in P0
January 16, 2002
8-22
DRAFT-DO NOT QUOTE OR CITE
-------
1 dams on GD21 and F1-generation rat pups on PND5, PND10, and PND22. Because the
2 representative species for the herbivore endpoint (meadow vole) is a rodent, that value is used
3 here as well. The population-level implications of this effect are unknown; however, it seems
4 likely that such effects on the thyroid, pituitary, and brain could diminish survivorship and
5 fecundity and diminish population production. To account for interspecies variance and LOAEL
6 to NOAEL extrapolation, an uncertainty factor of 10 is applied to obtain a dietary screening
7 benchmark for herbivores of 0.001 mg/kg body weight-day, or ~0.01 mg/kg as a wet-weight
8 concentration in plant tissue (see exposure assumptions in Section 8.3.1.5).
9
January 16, 2002
8-23
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
CHAPTER 9. EVALUATION OF EVIDENCE FOR
INDIRECT EXPOSURES
The primary purpose of this document is to derive human and ecological risk estimates for
perchlorate. As indicated in Chapter 1, pollution of drinking water supplies is the major concern.
Most perchlorate salts are used as solid oxidants or energy boosters in rockets or ordnance;
therefore, much of the perchlorate-tainted waterways in the U.S. can be traced to military
operations, defense contracting, or associated manufacturing facilities. Figure 1-5 shows that the
perchlorate anion could potentially be found in many natural waterways that are used for
irrigation or consumed by livestock or wildlife. Thus, it is logical to question whether there are
means through which humans might consume perchlorate other than drinking water. This
question is compounded by the chemical nature of perchlorate, which grants it long life under
typical environmental conditions (Urbansky, 1998; Urbansky and Schock, 1999; Espenson,
2000).
As discussed in Section 7.1.5, once a reference dose for perchlorate is established, any
burden posed by exposure routes other than potable water necessarily requires that the
contaminant's concentration in a water supply be lowered by an equivalent amount if it is
determined to calculate a maximum contaminant level goal (MCLG). A relative source
contribution (RSC) of between 20% to 80% is used to adjust the RfD according to the decision
framework presented in the EPA's methodology for deriving ambient water quality criteria (U.S.
Environmental Protection Agency, 2000).
Because polluted waters are used for irrigation, there are also questions concerning
absorption, elimination, and retention in food plants. However, this issue becomes considerably
less important if it can be demonstrated that the irrigation water is perchlorate-free. Likewise,
there are concerns that animals raised for food would consume plants that had received
perchlorate-tainted water. As described in Chapter 8, studies are being conducted to assess the
occurrence of perchlorate in biological fluids and tissues of animals and plants in affected
regions in recognition of the inter-connectedness of the food chain/food web continuum.
January 16, 2002
9-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
While much of the perchlorate problem can be traced to specific sites, a few reports have
suggested that fertilizers could represent another source of perchlorate in the environment (TRC
Environmental Corporation, 1998). These will be addressed in further detail in Section 9.1.1.
Sporadic detection of perchlorate in fertilizers was initially alarming because of the widespread
use of fertilizers in production farming. In addition to the ecological impact, this raised the issue
of assigning responsibility for clean-up costs. Because of the dependence of U.S. agriculture on
chemical commodity fertilizers, it was clear that assessment of any possible role of fertilizers
would require investigation.
This chapter summarizes the available data on the potential for exposure through runoff,
erosion, fertilizer, and groundwater movement. Evidence concerning the potential of perchlorate
to contaminate soil, sediment, vegetation, livestock and wildlife is also evaluated.
9.1 FERTILIZERS AS SOURCES OF PERCHLORATE SALTS
9.1.1 The Potential Role of Fertilizers
Recently, attention has been drawn to the possible roles of fertilizers as a source of
perchlorate contamination for two reasons. First, perchlorate-tainted agricultural runoff could
lead to pollution of natural waterways used as drinking water sources. Second, there is a
potential for food plants to take up and retain any soluble compounds—including perchlorate
salts—and thus provide an alternate route of exposure. It has long been known that Chile
possesses caliche ores rich in sodium nitrate (NaN03) that coincidentally are also a natural source
of perchlorate (Schilt, 1979; Ericksen, 1983). The origin of the sodium perchlorate (NaClOJ in
the caliche deposits remains an area of debate, but perchlorate is present and can be incorporated
into any products made from the caliche.
An examination of two manufacturing lots found perchlorate concentrations below 2 mg/g,
(i.e., < 0.2% w/w) with some lot-to-lot variability (Urbansky et al., 2001). Presently, the caliche-
derived products are sold in the U.S. only by Sociedad Quimica y Minera (SQM), but other
companies have mineral rights to some Chilean deposits and mines (U.S. Environmental
Protection Agency 2001b) and are potential sources of caliche-derived products. SQM has now
modified its refining process to produce a fertilizer that contains less than 0.1 mg/g (<100 (ig/g)
January 16, 2002
9-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of perchlorate, further reducing any environmental release (Lauterbach, 2001). Because nitrate
salts (saltpeters) find use as fertilizers, these natural resources have been mined and refined to
produce commercial fertilizers for domestic use or for export. Chilean nitrates make up about
0.1% of the U.S. market. Most U.S. fertilizers are derived from other raw materials other than
sodium nitrate and ammonium nitrate (NH4N03), which is often used for purposes similar to
NaN03, is manufactured from methane, nitrogen, and oxygen. There is no evidence that any
ammonium nitrate is derived from Chilean caliche. On account of its low usage, perchlorate
from Chilean nitrates cannot represent a continuing, significant anthropogenic source of
perchlorate nationwide, especially with its lowered perchlorate content.
9.1.2 Raw Material Use
As with many commodity chemicals, large scale purchases are dictated by cost of raw
materials, which are in turn influenced by transportation costs. Consequently, proximate (rather
than distant) sources of agricultural chemicals are likely to play the greatest roles in production
farming. Additionally, processing aids (e.g., clays) are likely to be derived from the nearest
sources.
Commodity chemicals used as agricultural fertilizers contain fairly high concentrations of
one, or sometimes two, of the primary plant nutrients, expressed as nitrogen (N), phosphorous (as
the oxide P205), or potassium (as the oxide K20). Trace metals (e.g., copper) can be applied
separately or along with these primary nutrients on a farm site. The primary phosphorus sources
are ammonium phosphates and triplesuperphosphate (a hydrous calcium phosphate). The
primary potassium source is potassium chloride. A mixture of synthetic and natural components
are used in fertilizer manufacture, described in detail elsewhere (U.S. Environmental Protection
Agency, 2001b).
Fertilizer application in production farming is highly dependent on the crop and the native
soil. Agriculture is influenced by climate, weather, topography, soil type, and other factors that
are generally similar within a geographical region; therefore, crops and fertilizer use are also
similar within such a region. For example, the Corn Belt relies heavily on urea and anhydrous
ammonia as nitrogen sources. Ammonium nitrate finds greater use in tobacco farming, and
potassium magnesium sulfate finds more use in milk-producing states. Because all plants require
the same primary nutrients, there is some fertilizer usage to provide these regardless of crop.
January 16,2002 9-3 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Local soil conditions also dictate what nutrients should be augmented, causing there to be large
regional variations.
Consumer fertilizer (specialty) products can be distributed over large geographical regions
because of the nature of the market. For example, major manufacturers have a limited number of
sites dedicated to blending multiple-nutrient formulations. These products are often sold as
bagged fertilizers through home-improvement centers, nurseries, florists, horticulturists, and
department (or other retail) stores. Unlike agricultural fertilizers, consumer products are usually
multi-nutrient formulations. In addition, trace metals are sometimes incorporated directly into
them. Because the average user will apply only a very small amount of trace metals (or even
primary nutrients) relative to a production farm, it is more economical, more practical, and more
convenient to use multiple-nutrient formulations. Moreover, the average consumer does not have
the wherewithal to disperse careful doses of several single-component fertilizers at the
appropriate times of the growing season.
Because fertilizer application on production farms is geographically delimited, there is
considerable interest in knowing which commodity chemicals might contain perchlorate—at
least in terms of dosing. Such information might suggest regions which should be investigated
for perchlorate contamination. Moreover, it will be important to know what crops might be
affected—if any.
9.1.3 Fertilizer Analysis Studies
Aside from the analyses of Chilean caliche, there were no studies to suggest that any other
processed fertilizer or raw material might contain perchlorate prior to 1998. That year, the
Ecosystems Research Division of the EPA's National Exposure Research Laboratory (NERL-
ERD) found perchlorate in several samples that were not derived from Chile saltpeter (Susarla,
1999a). This finding was later duplicated by other investigators from the North Carolina State
University College of Agriculture. However, the presence of perchlorate could only be
confirmed in consumer products, not in agricultural fertilizers. Moreover, subsequent analyses
of bags of the same materials acquired at a later date (likely from different manufacturing lots)
did not show perchlorate (Susarla et al., 2000). The choice of fertilizers did not account for the
possibility that the same raw materials must have been used in a variety of products at a point in
time. Additionally, a few major companies are responsible for making a large number of
January 16,2002 9-4 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
products under several brand names. Furthermore, some companies rely on toll manufacturing
so that the products are actually made by another company to meet a specific formulation.
Accordingly, an error or contamination associated with one raw material could affect a variety of
products without regard to company or application.
Perchlorate was found in six of eight lawn and garden fertilizers tested, according to a
report provided to the EPA by the U.S. Air Force Materiel Command (TRC, 1998). However,
the report's authors were careful to point out that the results were obtained from a single
sampling event and that raw material usage was variable; therefore, no general conclusions could
be drawn. These qualifiers are consistent with the limitations enumerated above, but they do
point towards a temporal contamination of some products.
This study helped bring to light a number of important issues for trace analysis of
fertilizers. First, most of the research on determining perchlorate to that time had been focused
on either finished potable water or raw source water (Urbansky, 2000). Second, fertilizers are
considerably more complicated matrixes than dilute water solutions. Third, a solid fertilizer is
not a homogeneous substance. In particular, multi-component formulations used as lawn and
garden fertilizers are macroscopically heterogeneous and it is possible to sort out the particles
visually. Thus, representative subsampling becomes a key issue. Fourth, the effectiveness of
leaching out any perchlorate ion into an aqueous phase was unknown. Fifth, the products chosen
did not reflect the chemical fertilizers used for production farming, but rather the ingredients
used for lawn and garden fertilizers during a specific time period.
Around the same time, the U.S. Air Force Research Laboratories (AFRL) performed a
study to assess interlaboratory corroboration; that is, the ability of different labs to analyze the
same sample and get the same result (AFRL, 1999; Eldridge, 2000). A variety of techniques
performed by multiple laboratories showed acceptable agreement on the concentrations of
perchlorate in solutions prepared from the purchased products. Several limitations (such as
product choice and sampling difficulties with heterogeneous solid products) made it impossible
to gain an understanding of agricultural fertilizer use, and the AFRL intentionally restricted its
use of the data to evaluating interlaboratory agreement. However, data from the AFRL study was
sufficient to confirm independently that some lawn and garden fertilizer products did contain
perchlorate during a certain period of time.
January 16, 2002
9-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Subsequently, the Water Supply and Water Resources Division of the EPA's National Risk
Management Research Laboratory (NRMRL) conducted its own survey of fertilizers in a
collaboration with the Oak Ridge National Laboratory (Urbansky et al., 2000a; Urbansky et al.,
2000b). In addition to a variety of products purchased from retailers, products were purchased
from farming supply stores (e.g., 50-lb bags of urea or ammonium nitrate) in Indiana, Ohio,
Kentucky, Pennsylvania, and Tennessee. In addition, commodity chemical samples were
collected from local distributors in Ohio and Indiana. These included urea, potassium chloride,
ammonium monohydrogen phosphate, and granular triplesuperphosphate, among others.
Samples were leached or dissolved and subjected to complexation electrospray ionization mass
spectrometry (cESI-MS) or ion chromatography (IC). Of 45 tested products, the only ones that
were found to contain any perchlorate were those based on Chile saltpeter. While this study was
the first to include the same products used on agricultural production farms, it did not address the
issues of sampling, product inhomogeneity, or geographical source variation.
In an effort to better address sampling, raw material usage, and other issues, the EPA
undertook an additional study of fertilizers. The project was divided into two phases, the first
part of which evaluated the testing laboratories for their ability to identify and quantitate
perchlorate in a fertilizer matrix. In the second phase, samples gathered under the supervision of
state agricultural agents were homogenized and sent to the laboratories for analysis using a
method established by the EPA (U.S. Environmental Protection Agency, 2001a). This
investigation was the most thorough in terms of including agriculturally relevant materials used
to manufacture a wide variety of specialty products or sold directly to farmers. It also spanned all
major national suppliers of these products. Although it reflected only a temporal snapshot, as
had all of the other studies, the survey of fertilizers incoporated the greatest number of unique
samples, quality control tests, and standardized practices, as well as other design improvements.
Four laboratories analyzed all of the materials, and some samples were analyzed by additional
laboratories. No other materials were found to contain perchlorate at measurable concentrations,
and the EPA concluded that the only clearly identifiable fertilizer source of perchlorate was
caliche. The data collected in this endeavor were additionally used to evaluate laboratory
performance and further validate the method (Urbansky and Collette, 2001). A set of archived
samples of all the Phase 2 materials was analyzed while evaluating an alternate ion
January 16, 2002
9-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
chromatographic column and independently verified all of the results reported in U.S.
Environmental Protection Agency (2001a) (DeBorba and Urbansky, 2001).
The findings reported in U.S. Environmental Protection Agency (2001a) are the most
comprehensive in terms of the types of materials tested, the manufacturers, the number of
laboratories analyzing each field sample of material, and the quality control checks. In these
regards, it represents our best understanding of fertilizers in terms of perchlorate content. While
the presence of perchlorate in the materials gathered in late 1998 through early 1999 remains
enigmatic, there is no evidence to support the concern that there is ongoing or routine perchlorate
contamination in the U.S. fertilizer supply. Reports in 1999 may have reflected the temporal
contamination of one or more raw materials or merely an error in manufacture. Based on the
studies reported to date (Collette and Williams, 2000; Gu et al., 2000; Urbansky et al., 2000a;
Urbansky et al., 2000b; Robarge et al., 2000; EPA, 2001b; Williams et al., 2001; DeBorba and
Urbansky, 2001), there is a consensus among researchers from the EPA, the fertilizer industry,
and other federal and state laboratories that currently used fertilizers are negligible contributors
to environmental perchlorate contamination. Even imported Chile saltpeter or products derived
from it contribute minimally due to their low use and low perchlorate content. Consequently, the
EPA has concluded that further investigation is unwarranted (U.S. Environmental Protection
Agency, 2001b).
IMC-Agrico, a major North American fertilizer manufacturer, has instituted its own
monitoring program for its raw materials and products as a result of continuing interest among
the scientific, industrial, and regulatory communities. These products include various potassium
ores (langbeinite, sylvinite), potash-based products (potassium chloride, potassium sulfate and
potassium magnesium sulfate), and phosphate products (ammonium monohydrogen phosphate,
ammonium dihydrogen phosphate and granular triplesuperphosphate). After more than 100
analyses using the latest method (EPA, 2001a), IMC reported to the EPA that no perchlorate was
detected in any of the materials it tested during a period spanning nearly three years. In addition,
IMC states that it has analyzed Magruder check samples for perchlorate. The Magruder check
sample program is jointly administered by the Association of American Plant Food Control
Officials and The Fertilizer Institute; it bears the name of a chemist from the F. S. Royster Guano
Company named E. W. Magruder, who initiated the program in 1922. The program selects,
prepares, and distributes samples of various materials and finished products to subscribing
January 16, 2002
9-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
laboratories and then collects and analyzes the data. Magruder samples reflect monthly
snapshots taken from the entire fertilizer industry. Perchlorate has not been detected in any IMC
product or any of 16 Magruder samples, according to IMC (personal communication from
William L. Hall).
9.1.4 Complicating Factors
It is worth pointing out at the U.S. Geological Survey (USGS) and Air Force Research
Laboratories have found perchlorate in isolated samples of sylvite taken from New Mexico
(Harvey et al., 1999). The USGS is engaged in additional sampling of North American mining
sites in order to assess whether there are natural mineral deposits of potassium perchlorate in
sylvite or sylvinite. Because little is known about the mechanisms of perchlorate formation in
the natural environment (which are assumed to be meteorological in nature), it is not clear
whether these findings represent a low-level background to be expected in evaporite mineral
deposits or not. Nonetheless, perchlorate has not been detected in any samples of agricultural
grade potassium chloride (0-0-62 or 0-0-60) taken under the direction of the EPA or by IMC-
Agrico. Accordingly, it appears that this mineral commodity does not suffer from inclusions of
perchlorate salts to any environmentally relevant extent.
Decades ago, ammonium nitrate was prepared from Chilean sodium nitrate by ion
exchange rather than by gaseous reactants. It appears that cost began to prohibit this practice for
fertilizer-grade ammonium nitrate. Nonetheless, some facilities appear to have continued the
practice for explosives-grade ammonium nitrate that was used for blasting in mining operations
throughout the American Southwest. It is unlikely that reliable data can be obtained from more
than the past 10 years or so. Prior to the establishment of nitric acid and ammonia factories,
natural saltpeters played significant roles in American agriculture. Thus, there may be
contamination of groundwater in regions where these materials were used historically. The lack
of information concerning natural attenuation, as well as a limited knowledge of hydrogeology,
makes it difficult to determine where and how such problem sites might be found. For this
reason, monitoring for perchlorate under the EPA's Unregulated Contaminant Monitoring Rule
can be expected to provide some of the most useful information.
Even though perchlorate was identified in some fertilizer products and was presumably
introduced through a contaminated raw material, this incident appears to have been entirely
January 16,2002 9-8 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
isolated. Furthermore, awareness within the fertilizer industry and the environmental community
is now heightened to the point that it appears unlikely to happen again.
9.2 MONITORING FATE AND TRANSPORT IN LIVING PLANTS
Due to the reported occurrence of perchlorate in certain water resources and in certain
fertilizer products, several groups have begun to address the extent and significance of
perchlorate uptake by plants. For example, if produce is grown using irrigation water tainted
with perchlorate, or if agricultural soil is amended with perchlorate-tainted fertilizer, this might
constitute a route of human exposure if perchlorate is taken up and retained in the edible parts of
produce plants. The possibility of a relevant exposure route would be increased if perchlorate
was found to bioaccumulate and if it was shown to survive the various processes that edible
plants undergo before being consumed. Unfortunately, experimental results that definitively
gauge the extent of risk from this route of exposure have not yet been published. However, some
progress toward this goal has been made.
9.2.1 Difficulties in Analyzing Plant Tissues and Other Environmental
Samples for Perchlorate
One problem that has delayed accurate and definitive studies of perchlorate uptake by
edible plants is the difficulty of analyzing for perchlorate in plant materials. Ion chromatography
is currently the recommended method for routine analysis of inorganic ions such as perchlorate.
It is a sensitive, reliable, and easily-implemented technique when perchlorate occurs in a matrix
that has a relatively low level of total dissolved solids (TDS). Unfortunately, extracts of plant
materials contain high concentrations of TDS, inorganic ions, amino acids, sugars, fatty acids,
and nucleotides—all of which contribute to the ionic strength of the sample (Ellington and
Evans, 2000). In such matrices with high TDS/ionic strength, other ions can overwhelm the
conductivity detector and effectively mask the signal from perchlorate. Ion chromatography is
not unique in this regard. Other techniques and methods suitable for reasonably dilute drinking
water matrices (Urbansky et al., 2000c; Magnuson et al., 2000a, b; Urbansky et al., 1999;
Urbansky and Magunson, 2000) cannot be readily applied to fertilizers or botanical and
physiological fluids. The problems of trace ionic analysis have led to the development of other
January 16, 2002
9-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
methods that rely on expensive instrumentation, but are not generally available, such as
asymmetric waveform ion mobility mass spectrometry (Handy et al., 2000; Ells et al., 2000) or
tandem mass-spec (MS-MS) systems (Koester et al., 2000).
Recently Ellington and Evans (2000) have reported an IC-based method using an enhanced
clean-up procedure for the quantitation of perchlorate in plant materials that greatly reduces
interferences from dissolved matter. The minimum reporting level (MRL) of perchlorate in
lettuce and tomato was found to be approximately 250 fxg/g on a wet mass basis. Lettuce and
tomato were chosen as representative plants because they are considered high priority candidates
for screening foodstuffs (Ellington and Evans, 2000). Perchlorate was spiked into the extraction
water for one half of the duplicate freeze-dried samples, while one half were extracted with pure
water. In the absence of other ions, some perchlorate is lost to the alumina used for the clean-up;
however, this should not impact application of the method to plant material because most
extracts have sufficient ionic strength. Note that perchlorate was not detected in any produce,
nor was the method applied to any edible plants that were grown with intentional exposure to
perchlorate.
9.2.2 Ecological Transport
In the laboratory setting, some plant species will absorb perchlorate when exposed to
contaminated irrigation water. Uptake by plants has been explored for possible use in
phytoremediation (Nzengung, 1999; 2000). Some investigators have speculated that bacteria are
responsible for this phenomenon in plants. Perchlorate-reducing monera have been identified by
several laboratories, and cultured from a variety of sources (including Las Vegas Wash
sediments, food processing sludge, soils, and sewage sludge); (Logan, 1998; Coates et al., 1999;
Coates et al., 2000; Kim and Logan, 2001; Wu et al., 2001; Logan, 2001). Recent work showing
perchlorate reduction in saline solution suggests that attenuation may be possible even in briny
locations (e.g., the Las Vegas Wash) or in fertilizer-laden farm runoff (Logan et al., 2001; Okeke
et al., 2001). This suggests that perchlorate-reducing bacteria are present at significant levels in
the environment. On the other hand, the bacteria isolated thus far prefer oxygen over nitrate over
perchlorate. In order to for perchlorate reduction to occur, the water must be anoxic and all of
the nitrate must have been consumed. Moreover, these bacterial cultures require a suitably moist
environment; arid soils or regions with low rainfall may not sustain their growth. Natural
January 16, 2002 9-10 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
attenuation probably varies around the nation, depending on local factors. Accordingly, it is not
possible to draw any meaningful conclusions about the ecological impact of fertilizers that
contain perchlorate, for they may or may not be applied in areas where this type of bacterial
degradation can occur.
Another factor that has prevented the early materialization of definitive data on risk from
perchlorate in edible plants is that many researchers who have addressed plant uptake of
perchlorate are primarily interested in other aspects of the problem. For example, Ellington et al.
(2001) have applied the optimized IC-based method described above first to the analysis of
perchlorate in tobacco plants and tobacco products. Tobacco was chosen because it is grown in
some locations in soils amended with Chile saltpeter.
Ellington and Evans (2000) obtained green (uncured) tobacco leaves from the Coastal Plain
Experiment Station (CPES) in Tifton, GA in late July 1999. The plants grew in soil that had
been amended with two fertilizer products, one of which was Chile saltpeter. The perchlorate
level in the Chile saltpeter was 1.5 mg/g, consistent with contemporaneous reports (Urbansky
et al., 2001; personal communication from W.P. Robarge). Perchlorate was also found in a
6-6-18 plant food that had been applied to the same soil. While 3% of the nitrogen was from
nitrate, the perchlorate concentration was only 36 >ug/g; whereas, based on the typical perchlorate
content in Chile saltpeter, it should have been about eight times larger if all of the nitrate were
from Chile saltpeter. This suggests that synthetic nitrates were also part of the fertilizer's
constitution. Perchlorate concentrations in the dried tissue varied from 12.5 to 165 //g/g,
depending on the portion of the leaf examined and the curing process employed. Soil samples
leached with deionized water contained 0.3 //g/g on a dry weight basis. EPA researchers also
analyzed several off-the-shelf cigarettes (2 brands), cigars (1 brand), and chewing tobacco
(7 brands) and found perchlorate concentrations ranging from 0.4 to 21.5 Mg/g (undried), and
only one product that contained none (Wolfe et al., 1999; Ellington et al., 2001). They confirmed
the IC results by chlorine NMR spectrometry and capillary electrophoresis. Collectively, these
observations argue that tobacco plants can take up perchlorate from perchlorate-contaminated
fertilizers via the soil. Furthermore, they indicate the importance of investigating whether crop
plants can accumulate perchlorate in their edible portions and whether any contamination can
persist through the processing that precedes consumption.
January 16,2002
9-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Several groups have looked at the accumulation of perchlorate in various inedible plants as
a potential means of fate and remediation. Perchlorate-tainted water from the Las Vegas Wash
enters Lake Mead and the Colorado River and therefore has the potential to affect the potable
water of many people as well as the irrigation water used for much of the lettuce produced in the
U.S. Salt cedar (Tamarix ramosissima) is an invasive woody plant that grows prolifically in and
around the Las Vegas Wash. Salt cedar consumes and transpires an enormous amount of water
when it is actively growing. Furthermore, it accumulates and secretes salt. For these reasons,
Urbansky et al., (2000d) have analyzed samples of salt cedar that were taken from the Las Vegas
Wash. They found perchlorate at 5-6 jug/g in dry twigs extending above the water and 300 /ug/g
in stalks immersed in the water from a plant growing in a contaminated stream, suggesting that
salt cedar plays a role in the ecological distribution of perchlorate.
Still others have investigated plant uptake with the specific goal of identifying remediation
strategies for perchlorate. The biodegradation of perchlorate in woody plants has been
investigated as a means of phytoremediation (Nzengung et al., 1999; Nzengung and Wang,
2000). Nzengung et al. (1999) and Nzengung and Wang (2000) found that willow trees (genus
Salix) were able to decontaminate aqueous solutions containing 10-100 mg/L of perchlorate to
below the method detection limit of 2 /^g/L and suggest that two distinct phytoprocesses were at
work in their studies. Specifically, they observe evidence for rhizodegradation from the exudates
released from the plant, and—more importantly from the standpoint of relevance for food safety
issues—they see accumulation in branches and leaves. Only about 11% of the perchlorate spiked
into the water in which the trees were grown was found to phytodegrade in 26 days. The
majority of perchlorate that was removed from solution after 26 days was found in the leafs.
Longer term experiments suggest that the perchlorate did not accumulate in the leaves, but was
very slowly transformed there as well. Generally, the perchlorate level in the leaves increased to
a maximum before decreasing to undetectable levels after perchlorate was completely removed
from solution. Nzengung et al. assumed that the phytodegradation pathway of perchlorate leads
to chloride. Moreover, Nzengung et al. explored the role of other anions in the removal of
perchlorate in solution. They found that the perchlorate removal rate was decreased as the N03"
level was increased. This was attributed to competing reactions in which both anions were
utilized as electron acceptors. Clearly this has relevance for the food safety issue and should be
investigated further. For example, the type of fertilizer used in food crop production may have
January 16, 2002
9-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
an effect on the degree to which perchlorate is taken up, depending on the major components of
the fertilizer.
Susarla and coworkers have published results of their investigations on transformation of
perchlorate by a wide range of plant types. For example, Susarla et al. (1999b,c) have performed
screening studies to determine what species might show potential for further investigations of
perchlorate phytoremediation. Thirteen vascular plant species were selected for evaluation in
these preliminary experiments. This included four tree species, four herbaceous wetland species,
four aquatic species, and one herbaceous upland species. Laboratory-scale experiments were
conducted in order to, among other things, evaluate the ability of these plants to remove
perchlorate from solution, evaluate the role of nutrients on perchlorate removal, and determine
the fate of perchlorate removed form solution (e.g., plant tissue distribution, accumulation versus
breakdown). Each of these topics is indirectly relevant to the issue of uptake by edible plants.
For all of these experiments, perchlorate concentrations of 0.2, 2.0 and 20 mg/L were tested
in aqueous and sand treatments for ten-day periods. Perchlorate was found to be depleted from
solution in the presence of all but two species. Susarla et al. (1999a,c) used a system of five
categories to classify the performance of the species based on the degree to which they depleted
the solution. None of the trees tested were included in the highest category of performance, but
some of the wetland and aquatic plants were. Plant tissue (e.g., roots, stems, leaves) were
analyzed from samples that demonstrated the maximum drop in perchlorate concentration.
Susarla et al. (1999a,c) report perchlorate, or some transformation metabolite (chlorate, chlorite,
chloride), in all tissues analyzed. Results of these studies suggested significant influences on
depletion of perchlorate from, among other things, growth substrate (sand versus aqueous
treatment), the level of nutrients, stage of plant maturity, and the presence of other ions. All of
these influences should prove to be valuable insights when considering the uptake of perchlorate
by edible plants. Based on screening studies, additional studies focused on the
phytotransformation of perchlorate by the aquatic plant parrot-feather (Myriophyllum
aquaticum)\ (Susarla et al., 1999b; Susarla et al., 1999c).
Tobacco is one crop for which the use of Chilean nitrate salts can be documented in some
locations. In northern Kentucky, these products are primarily used for seedling beddings rather
than fertilizing fields; for various reasons, ammonium nitrate is preferred by many farmers in
Kentucky. Such preferences vary throughout tobacco-producing states and regions, however.
January 16, 2002
9-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Data on application of Chile saltpeter is sparse, and it is not possible to estimate the ecological
impact in any meaningful way. There can be no question that at least some vascular plants
absorb perchlorate from their local environments. Furthermore, perchlorate has been found in a
number of plants and animals living in contaminated environs (Smith et al., 2001). An obvious
concern raised by finding measurable perchlorate concentrations in plant tissues is whether this
ion can affect food crops and what factors might influence its uptake and accumulations. These
issues shall be considered next.
9.2.3 Extrapolating to Food Plants
Because so much U.S. produce is fertilized with perchlorate-free chemical commodities,
the risk from exposures via fertilizers is small. Some crops (e.g., corn, wheat, and rice) are
fertilized with materials that are unquestionably perchlorate-free. Additionally, there is no reason
to suspect any perchlorate associated with growing grains. However, the risk of exposure
resulting from irrigation with perchlorate-tainted water in the American Southwest is unknown.
At present, there are no efforts to test fruits and vegetables for perchlorate. Many of the studies
on uptake by plants have been based on concentrations higher than those encountered in
irrigation water. Furthermore, some products derived from Chile saltpeter are known to be
among those used on California citrus crops.
One of the few studies of perchlorate uptake by edible plants is the ongoing work of
Hutchinson and coworkers with lettuce grown in a greenhouse with perchlorate-tainted irrigation
water. Lettuce is of particular importance for assessing the risk of perchlorate to the food supply
since much of the lettuce produced in the U.S. is irrigated by water that is fed by the Las Vegas
Wash, which is contaminated with perchlorate. Also, lettuce has a high water content and
virtually the entire above-ground plant is consumed without cooking or processing. These
characteristics would present a potential risk if lettuce efficiently accumulates perchlorate.
Hutchinson and coworkers are irrigating lettuce plants with five different concentrations of
perchlorate (0.1, 0.5, 1.0, 5.0, and 10.0 /Ug/L) for a period of 90 days following planting.
At various intervals of time they divide the plants into green tissue and root samples and analyze
each sample for perchlorate using an analytical method adapted from Ellington and Evans
(2000). Their results show an accumulation of perchlorate into the green tissue. The level of
perchlorate built up steadily over the first 50-60 days of the experiments, then generally leveled
January 16,2002 9-14 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
off. At about 50 days into the experiment, the lettuce irrigated with 10.0 ppm perchlorate
exhibits a perchlorate content of about 3 mg/g on a lettuce dry matter basis. Since lettuce is
about 90% water, this would amount to about 0.300 mg/g on a wet weight basis. The amount of
perchlorate detected in the leaves is generally linear with dosing level for a given day.
Experiments are underway to determine whether lettuce has the capability to degrade perchlorate
if the supply of the contaminant is stopped. However, this determination is frustrated by the fact
that lettuce continues to grow. Therefore, a decline in concentration (e.g., expressed in mg/g)
does not adequately reflect the situation. The preliminary results from these studies (Hutchinson
et al., 2000) suggest that, when complete, they will constitute considerable progress on the issue
of exposure to perchlorate from edible plants.
Even if many food plants can be shown to absorb and retain perchlorate, the primary source
of the contaminant is irrigation water polluted from defense-related activities. Because these
activities are reasonably localized geographically, most of the country's agricultural products
should be perchlorate-free, e.g., corn, wheat, rice, milk. On the other hand, some types of
produce are supplied almost entirely by regions dependent on contaminated irrigation water.
Therefore, these sites represent possible exposure routes for most of the nation via foods such as
lettuce.
Historically, much of the emphasis on fertilizer pollution from agricultural runoff has been
on fertilizers applied to the soil. However, potassium nitrate is usually applied to the leaves of
citrus trees when a potassium deficiency is found by analyzing leaf tissue. Such foliar
application would not necessarily contribute significantly to runoff type pollution of waterways,
but could lead to the absorption of contaminants through the leaves and wood. There are no
reliable data on the sources of potassium nitrate used for citrus crops. While it is known that
absorption of anions similar to perchlorate (e.g., pertechnetate) are affected by the ionic strength
and compostion of the surrounding solution, little is known about the factors that influence
perchlorate influx via roots or leaves. In addition, the fate of absorbed perchlorate in the plants is
also unknown. It may be that xylem-supplied tissues, such as leaves, are the final repository
rather than phloem-supplied tissues, such as fruits.
These issues and more have begun to be examined by the EPA, but there are many
unknowns (U.S. Environmental Protection Agency, 2001b). Until such time as quantitative
studies are performed on various species to determine what factors influence the absorption,
January 16, 2002
9-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
accumulation, and distribution of perchlorate in plants, it is not possible to estimate whether
foods can serve as meaningful contribution to the body burden or to the risk posed to humans
from perchlorate contamination. Even if they do, there is considerable peace of mind in knowing
that fertilizers and water supplies are generally not providing any perchlorate to the plants in the
first place. Consequently, only a small number of foods are worth considering for further study.
On the other hand, it is not known to what extent other countries rely on natural saltpeters to
fertilize food crops. Moreover, it is not known whether fruits and vegetables absorb and retain
the perchlorate ion. Therefore, it is not possible to say whether fruits and vegetables grown
outside the U.S. serve as a possible exposure routes at this time. Depending on the season,
imported oranges, apples, and grapes and their juices are consumed throughout the U.S.
Because there are no data on perchlorate in imported produce, no data on perchlorate in
U.S. produce, and no data from controlled laboratory experiments on uptake in fruit crops, it is
impossible to assess whether these foods can contribute to perchlorate consumption in humans or
whether drinking water constitutes the entire body burden. However, the available information
on fertilizers and irrigation water suggests that foods do not contribute to the body burden. At
the present time, the available data point towards drinking water as the principal exposure
pathway for humans.
9.3 SUMMARY
Despite some initial findings implicating fertilizers as a source of perchlorate, more
thorough and better designed studies that were conducted subsequently have not found this to be
the case. Current fertilizer manufacturing practices and raw material sources make it unlikely
that perchlorate contamination could occur widely and without discovery. While some plants
may absorb or even accumulate perchlorate in specific tissues, there are many unknowns with
regard to the edible portions of nutritionally and agriculturally important crops. Many factors
influence transport of ions, and current understandings of plant physiology and botany suggest
perchlorate uptake would be reduced as a result of such factors. Even if perchlorate uptake
occurred in some food crops, perchlorate contamination is localized geographically outside of
major agricultural regions, minimizing possibility of uptake in edible produce. While
perchlorate-tainted irrigation water may be a source available for uptake of perchlorate by plants,
January 16, 2002
9-16
DRAFT-DO NOT QUOTE OR CITE
-------
this is again localized, and has not been proven to occur at the concentrations of perchlorate that
are observed environmentally. Difficulties in analyzing many plant or animal tissues originally
were obstacles to executing appropriate studies, but these problems have generally been solved.
Ideally, more data would be available on food plants, but current evidence suggests that drinking
water is the primary exposure pathway to perchlorate for humans.
January 16, 2002
9-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
10. MAJOR RISK CHARACTERIZATION
CONCLUSIONS
10.1 HUMAN HEALTH
This section summarizes major findings regarding human health presented in Chapters 1, 2,
3, 4, 5, and 7.
10.1.1 Hazard Potential
Perchlorate is an anion that originates as a contaminant in ground and surface waters from
the dissolution of ammonium, potassium, magnesium, or sodium salts. Ammonium perchlorate
is the oxidizer and primary ingredient in solid propellant for rocket motors. Perchlorate salts also
are used on a large scale as a component of air bag inflators and in the manufacture of
pyrotechnics and explosives. Solid rocket inventories are growing at a significant rate as systems
reach the end of their service life: the solid rocket disposal inventory is expected to be over 164
million lb by the year 2005. Because the accepted method for removal and recovery of solid
rocket propellant is high-pressure water washout, a large amount of aqueous solution containing
ammonium perchlorate is generated. A number of locations where perchlorate has been detected
in groundwater or surface waters are in areas associated with the development, testing, or
manufacturing of aerospace materials. Perchlorate contamination also occurs when explosives
are used extensively, e.g., open burn/open detonation operations and some mining activities.
Perchlorate is rapidly absorbed from the gastrointestinal tract, whereas dermal and
inhalation exposures are not expected to be significant exposure routes for the general public.
The known mode of action for perchlorate is that it acts as a competitive inhibitor of active
iodide uptake by the sodium (Na+)-iodide (T) symporter (NIS) in most mammals, including
humans, laboratory test species, and wildlife. This decrease in intrathyroidal iodide results in a
decreased production of T3 and T4 thyroid hormones. Decrements in thyroid hormones can
cause permanent neurodevelopmental deficits and impair adult organisms as well. A decrease in
thyroid hormones can also potentially perturb the hypothalamic-pituitary-thyroid axis to increase
the pituitary's production of TSH and, consequently, stimulate the thyroid to increase production
January 16, 2002
10-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
of thyroid hormone in an attempt to compensate. Prolonged stimulation of the thyroid by TSH
may result in thyroid neoplasia, particularly in rodents known to be sensitive. Tumors have
occurred in rats dosed with high levels of perchlorate for long periods and at much lower doses in
relatively young adult animals (19 weeks) dosed in utero and during development. These
findings have raised concerns about the in utero imprinting of the regulatory system responsible
for controlling thyroid hormone economy.
The target tissue for systemic effects of perchlorate has been identified as the thyroid. The
key event of its mode of action is iodide uptake inhibition at the NIS. Changes in the thyroid
hormone homeostasis result in histopathological changes in the thyroid, including: colloid
depletion, follicular hypertrophy, follicular hyperplasia, and decrease in follicular lumen size.
If perchlorate exposure is stopped, the thyroid histopathological effects have been shown to be
reversible after exposures as long as 90-days in rats, but incomplete recovery of thyroid
hormones occurs in this same time period. There are also some case studies in humans treated
therapeutically with perchlorate that indicate reversibility of thyroid hormone changes after years
of exposure.
Other potentially adverse and permanent effects from decreased thyroid hormone include
effects during development in utero and early growth, particularly effects on the nervous system
if the pregnant mother was hypothyroxinemic or hypothyroid. Laboratory animal assays
performed in response to recommendations made at the peer review in 1999 and as part of the
perchlorate testing strategy confirmed neurodevelopmental effects observed in previous studies.
Changes in brain morphometry and motor activity were observed. The potential for major
disturbances in thyroid hormone homeostasis to disturb reproductive capacity or to induce
immune effects also exists. The ability of perchlorate to cause contact hypersensitivity is
suggested but remains not well characterized. Finally, a remarkable conservation of the thyroid
hormone regulatory system has been demonstrated across species. Inhibition of iodide uptake by
the NIS has been shown in pharmacokinetic studies to be very similar across species, including
humans.
10.1.2 Dose Response
The revised RfD is based on an assessment that reviewed a set of studies that were
developed to explicitly evaluate these potential toxicities. The quantitative estimate of risk is
January 16, 2002 10-2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
based on laboratory animal data because there are no good observational epidemiological data
concerning human subjects representative of the critical sensitive populations (hypothyroxinemic
pregnant women or children) or that have evaluated neurodevelopmental outcomes; nor have
adequate clinical studies been performed. A harmonized approach was proposed based on the
key event of iodide inhibition and its relationship to disturbances in the hypothalamic-pituitary-
thyroid axis as evidenced by effects on thyroid and pituitary hormones, thyroid histopathology,
and brain morphometry. Using these precursor lesions as the basis for the point-of-departure is
considered to be protective for cancer development as well as for neurodevelopmental sequelae.
The database supported a point-of-departure for the RfD deviation at 0.01 mg/kg-day based
on changes in maternal thyroid and pituitary hormones and on changes in the brain morphometry
and thyroid and pituitary hormones of fetal and neonatal pups. A composite uncertainty factor of
300 was applied in the derivation. An adjustment also was made for administration of
perchlorate as ammonium perchlorate. The RfD is for perchlorate as the anion because that is
what is sampled and analyzed in environmental media and because the salts of perchlorate
readily dissolve. Uncertainty factors were applied for intrahuman extrapolation, the use of a
LOAEL, concern regarding the lack of studies of longer duration and database deficiencies.
Confidence in the study, the database, and the RfD is rated as medium. A major uncertainty is
the sensitivity that the screening neurodevelopmental studies provide to protect against
neuropsychological deficits of exposures that might occur within critical developmental windows
or in susceptible human populations.
The daily perchlorate exposure to the human population that is likely to be without
appreciable risk of either cancer or noncancer toxicity during a lifetime is 0.00003 mg
perchlorate/kg-day. It again is noted that this RfD is specific for the anion because that is what is
detected in most environmental samples and because most salts of perchlorate readily dissolve.
Because of the application of uncertainty factors, this dose is approximately 1/300 of the dose
that resulted in brain morphometry and thyroid changes in pups and hypothyroid status
(decreased T4 and increased TSH) in rat mothers (Argus Research Laboratories Inc, 2001) and in
their pups both during gestation (GD21) and in the post-natal period (PND4 through PND21).
January 16, 2002
10-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
10.1.3 Risk Characterization
Comprehensive risk characterization for the perchlorate contamination issue, as discussed
in Chapter 1 (see Figure 1-5), requires accurate information on exposure levels determined by a
validated analytical method. Dose-response estimates such as the value derived herein can then
be used to gauge the potential toxicity of those exposures. Exposure can be either direct, most
likely by ingestion, or indirect, such as by consumption of contaminated food. When using the
dose-response assessment derived herein to compare with exposure estimates, one should remain
keenly aware that many of these exposure aspects have not yet been characterized accurately for
perchlorate. Fate and transport information do not exist to track the spatial and temporal
distribution of perchlorate; the potential for evaporative concentration in soils has not been
characterized, nor has its uptake in plants or herbivores. In addition, there are uncertainties
remaining in the dose-response estimate itself. These concerns also should be considered
whenever attempting to characterize the risk to a specific human population exposed to a
particular scenario.
10.1.3.1 Direct Exposures
Typically the RiD is used as a comparison for oral ingestion, such as by drinking water.
The RfD is compared with an exposure estimate of the drinking water concentration to
characterize potential toxicity. When making this comparison, the assumptions underlying
derivation of the RfD must be kept in mind. The RfD is intended to be protective of susceptible
populations exposed daily. The frequency and magnitude of exposure is a key attribute of
accurate dose-response characterization (Jarabek, 1995c) and an equally important component of
risk characterization. Transient decreases in T4 can cause permanent neurodevelopmental
deficits. Thus, the degree to which the particular suspected population at risk fits with the
underlying assumptions of the RfD derivation should be kept in mind. Finally, the degree of
imprecision in the derivation of an RiD should be taken into account. The RfD estimates are not
intended to serve as "bright line" estimates. By definition, there is an order of magnitude
uncertainty around the estimate. This generally translates into a range of approximately
three-fold below to three-fold above the RfD, but also depends on the nature of the effects used
as the basis.
January 16, 2002
10-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
10.1.3.2 Indirect Exposures
Where crops are irrigated with perchlorate-contaminated water, indirect human exposures
may result. A number of factors need to be considered in estimating human exposure through
crops.
Concentration in plant parts as a result of root uptake normally is calculated using a soil-to-
plant transfer factor that is expressed as the ratio of plant to soil concentration. If perchlorate is
subject to evaporative concentration in irrigated soils, then soil concentration, and therefore
uptake, may be higher than that expected simply based on concentration in irrigation water. If a
leaf crop such as lettuce is spray-irrigated, perchlorate could be concentrated evaporatively on
external leaf surfaces. Because perchlorate salts have high water solubility, this contamination
probably would be removed largely by washing. On the other hand, if perchlorate is
phytodegraded, as one study has suggested (Nzengung, n.d.), soil or plant concentrations may be
lower than otherwise expected. Studies are needed to determine the behavior and fate of
perchlorate in plant-soil-water systems, including studies that simulate leaf crop irrigation and
that account for full life cycles of crops.
Besides estimates of perchlorate concentrations in crops, the calculation of human daily
intake depends on the number of crop types that are contaminated, the extent to which a
particular individual obtains the crops from a contaminated source, and the individual's daily
consumption of the crops. These factors may vary widely in the exposed population, and
methods for accounting for the combined variability should be used in characterizing these
exposures.
Methods for estimating human exposures resulting from crop uptake of soil-deposited
contaminants are presented in Chapters 6 (Determining Exposure Through the Terrestrial Food
Chain) and 10 (Risk Assessment) of the EPA document, "Methodology for Assessing Health
Risks Associated with Multiple Pathways of Exposure to Combustor Emissions (EPA 600/
R-98/137)." That document currently is undergoing revision and is scheduled for final release in
January 2002. If the needed information can be obtained on perchlorate behavior and fate, the
methods described therein can be used to develop estimates of human exposure and risk.
January 16, 2002
10-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
10.1.4 Major Uncertainties and Research Needs
Reliable exposure estimates are required to accurately and comprehensively characterize
the risk of perchlorate contamination. This section will briefly summarize research needs
associated with aspects of uncertainty about the human health risk dose-response estimate that
were highlighted in Chapter 7.
The greatest need for continued improvement in the dose-response assessment is a more
accurate characterization of the linkage between the key event of the mode of action (i.e.,
inhibition of iodide uptake in the thyroid gland), subsequent changes in thyroid hormones, and
the correlation to outcome measures in hypothyroxinemic pregnant animals and their pups.
Because this need must be addressed in the fetal compartment as well, accurate characterization
of toxicokinetics during pregnancy and lactation also are required. More definitive studies of the
degree of change in perturbation of the hypothalamic-pituitary-thyroid axis (i.e., change in
hormone levels) that is associated with thyroid histology, and with neurobehavioral deficits
especially, would improve the confidence in the accuracy of the exposure-dose-response
continuum. The current studies may need to be repeated with larger sample sizes and lower
doses, and new studies may be needed to evaluate effects on fetal hormone levels and
neurodevelopmental measures both in the laboratory and in a survey of the human population.
Research on potential factors influencing sensitivity is also critically requisite. Animal models of
thyroid impairment such as iodide deficiency and "womb to tomb" exposure designs should be
explored. Finally, mechanistic determinants of these toxicokinetic and toxicodynamic
parameters and processes should be further characterized.
10.2 ECOTOXICOLOGY
10.2.1 Aquatic Life
Procedures for deriving Tier II water quality values were used in Section 8.3.2.1 to jointly
characterize the potential effects of the perchlorate ion on the richness and productivity of fish,
aquatic invertebrate, and plant communities. Tier II values are derived when data are not
sufficient for deriving ambient water quality criteria. The Tier II value derivation procedures
account for missing information with approximately 80% confidence. In this case, the Tier II
January 16, 2002
10-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
values derived, termed secondary acute and chronic values, were 5 and 0.6 mg/L (i.e., 5,000 and
600 Mg/L), respectively; difficulties associated with the interpretation of one test result in an
uncertainty range for the secondary chronic value of < 0.32 to > 0.83 mg/L (< 320 to
> 830 ^g/L). Perchlorate levels reported for large surface waters (as high as 16 Mg/L) and ground
waters (as high as 280 Mg/L in public supply wells) are well below the secondary acute and
chronic values. Thus, at these exposure levels, the likelihood of effects on the richness and
productivity of fish, aquatic invertebrate, and plant communities appears to be low. However,
because much higher perchlorate concentrations have been reported in monitoring wells at rocket
motor manufacturing or testing sites (37 x 106 Mg/L) and in groundwater-dominated surface
water systems close to sites of contamination (3500 to 1.3 x 105 Mg/L), sites clearly exist that
have perchlorate concentrations high enough to cause toxicity to aquatic life. These sites include
springs, such as that sampled along Las Vegas Wash in Nevada (Parsons, 2001) and the INF
Pond at Longhorn Army Ammunition Plant in Texas (Parsons, 2001; Smith et al., 2001). On the
other hand, concentrations below the Tier II values were detected in larger water bodies
immediately adjacent to sites of contaminations, such as in Lake Mead immediately adjacent to
the mouth of the Las Vegas Wash (less than 4 to 68 Mg/L). Water discharged from a CERCLA
groundwater pump-and-treat facility that was not equipped to treat perchlorate at Allegany
Ballistics Laboratory to the North Branch Potomac River contained 250 to 280 Mg/L perchlorate
(Parsons, 2001).
Where high levels of contamination exist, sensitive aquatic organisms such as daphnids
may be the most likely to experience effects; in the reported tests, effects were seen on both
survival and reproduction (neonates per organism). A teratogenicity assay, FETAX, showed
malformations in frog embryos occurring at only slightly lower concentrations than lethality,
indicating that perchlorate is probably not a potent developmental toxicant. Tier II values are not
estimated for plants, but results from algal toxicity tests suggest that even at the higher
perchlorate concentrations associated with rocket motor manufacturing, risk of toxicity to aquatic
plants is low.
The perchlorate anion can be associated with various cations including sodium,
ammonium, and potassium. When sodium perchlorate was tested, the sodium cation was not
toxic to daphnids in sodium chloride control tests but did show toxicity to minnows.
Ammonium controls were not used in tests with ammonium perchlorate, but ammonium ion is a
January 16, 2002
10-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
known toxicant with toxicity that varies according to water temperature and pH. In any aquatic
system where perchlorate is present, attention should be given to determining the concentrations
of potentially toxic cations that may contribute to ecological effects.
Based on a secondary chronic value of 600 //g/L (uncertainty range, < 320 to > 830 //g/L)
for perchlorate, the analytical detection methods for perchlorate in water are sufficient. The
detection limit achieved for perchlorate in water was 4 /ugfL (Parsons, 2001; Smith et al., 2001),
which is much less than the secondary chronic value. Thus, the likelihood that adverse
ecological effects will occur below detection limits is low.
10.2.2 Risks to Consumers of Aquatic Life
Information from Parsons (2001) and Smith et al. (2001) indicate that perchlorate may
bioaccumulate in aquatic invertebrates and fish in contaminated waters, but perchlorate is not
expected to bioconcentrate in these organisms to levels exceeding the surface water
concentrations. Therefore, there currently is no indication that consumers of aquatic
invertebrates or fish are at increased risk of effects from bioconcentration in areas where
perchlorate concentrations in surface water occur. However, there is some uncertainty about the
potential for bioaccumulation of perchlorate at low concentrations (i.e., 4 to 300 /^g/L in water)
because of the higher detection limits for perchlorate in animal tissues, which were 300 to 400
Atg/kg in Parsons (2001) and about 70 /^g/kg in Smith et al. (2001). Furthermore, perchlorate
may bioconcentrate (i.e., to levels exceeding those in water) in aquatic plants; therefore,
consumers of aquatic plants may be at greater risk than consumers of aquatic invertebrates or
fish, but information is not available concerning effect levels in aquatic herbivores.
10.2.3 Terrestrial Life
10.2.3.1 Plants
Terrestrial plants may be exposed to perchlorate in soil at disposal sites and at sites
irrigated with contaminated surface water or groundwater. Perchlorate concentrations in soil at
disposal sites range from less than 1 to 1470 mg/kg (Parsons, 2001) and can be higher than the
screening benchmark of 4 mg/kg and even higher than the lethal concentrations (^ 180 mg/kg;
EA Engineering, Science and Technology, Inc., 1998).
January 16, 2002
10-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
In the absence of reliable information concerning the accumulation of perchlorate in
irrigated soils, it may be assumed that soil concentrations equal irrigation-water concentrations
(Section 8.3.1.3). Reported surface-water concentrations in the Colorado River, 4 to 16 /ugfL,
would translate to 0.004 to 0.016 mg/kg. At the Yuma site, there was a single detection in
surface soil of 0.090 mg/kg; all other measurements were below the detection limits of 0.079 to
0.080 mg/kg (Parsons, 2001). Even the single detected concentration is a factor of 44 lower than
the benchmark value. The reported groundwater concentration in public wells of 280 u-gfL
would translate to 0.28 mg/kg, which is a factor of 14 lower than the benchmark value. Hence,
perchlorate does not appear to constitute a hazard to plants irrigated with surface water.
However, given the large uncertainties concerning exposure, a hazard from groundwater
irrigation cannot be precluded.
Based on this screening benchmark of 4 mg/kg for perchlorate, the analytical detection
methods for perchlorate in soil are sufficient for determining whether soils will cause toxicity to
plants, and there is little likelihood of adverse ecological effects occurring at levels below
detection limits. The detection limit achieved for perchlorate in soils was generally 75-80 i^g/kg
(Parsons, 2001), but there was at least one soil sample where the reporting limit was 803 fxg/kg.
However, all of these limits are less than the screening benchmark.
10.2.3.2 Soil Invertebrates
Soil invertebrates may be exposed to perchlorate in soil at disposal sites and at sites
irrigated with contaminated surface water or groundwater. Perchlorate concentration
measurements at disposal sites range from less than 1 to 1470 mg/kg (Parsons, 2001) and,
therefore, can exceed the soil screening benchmark of 1 mg/kg. In the absence of reliable
information concerning the accumulation of perchlorate in irrigated soils, it may be assumed that
soil concentrations equal irrigation water concentrations (Section 8.3.1.3). Reported surface
water concentrations in the Colorado River, 4 to 16 £ig/L, would translate to 0.004 to
0.016 mg/kg in soils. At the Yuma site, the single detection in surface soil was 0.090 mg/kg with
detection limits of 0.079 to 0.080 mg/kg. This detected concentration is a factor of 11 lower than
the soil screening benchmark value (1 mg/kg). The reported groundwater concentration in public
wells of 280 //g/L would translate to 0.28 mg/kg, which is a factor of 4 lower than the
benchmark value. Hence, perchlorate does not appear to constitute a hazard to soil invertebrates
January 16, 2002
10-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
in soil irrigated with surface water. However, given the large uncertainties concerning exposure,
a hazard from groundwater irrigation cannot be precluded.
Based on this screening benchmark of 1 mg/kg for perchlorate, the analytical detection
methods for perchlorate in soil are sufficient, and there is little likelihood of adverse ecological
effects occurring at levels below detection limits. The detection limit achieved for perchlorate in
soils was generally 75-80 //g/kg (Parsons, 2001), but there was at least one soil sample where the
reporting limit was 803 £ig/kg. However, all of these limits are less than this screening
benchmark.
10.2.3.3 Herbivores
Exposures of voles to perchlorate based on measured plant concentrations at rocket motor
manufacturing or testing sites (0.11 mg/kg day to a maximum of 49 mg/kg day) exceed both the
LOAEL of 0.01 mg/kg/day and the screening benchmark of 0.001 mg/kg day. Estimated
exposures of voles consuming plants on sites irrigated with surface water (0.18 mg/kg day) and
groundwater (3.2 mg/kg day) also exceed the LOAEL and the screening benchmark. Hence,
there is a potential hazard to all herbivorous wildlife living in areas that may be irrigated with
contaminated water. At disposal sites, wildlife would be at risk from the effects of loss of food
and habitat from toxic effects on plants, as well as the potential for direct toxic effects via
consumption of perchlorate-tainted food or water.
Assuming a water ingestion rate of 0.21 g/g-day (U.S. EPA, 1993a,b), the screening
benchmark for herbivores is equivalent to a water concentration of 4.8 /ig/L. Perchlorate levels
reported for large surface waters (as high as 16 //g/L) are greater than this concentration. Much
higher perchlorate concentrations have been reported in monitoring wells at rocket motor
manufacturing or testing sites (37 x 106 /ug/L) and in ground water-dominated surface water
systems close to sites of contamination (3500 to 1.3 x 105 lug/L), and rodent exposures via
drinking water at these sites would exceed the rodent NOAEL.
Based on screening level benchmarks for herbivores, the analytical detection methods for
perchlorate in plant tissues may not be sufficient for the detection of concentrations potentially
toxic to herbivores even though the analytical detection methods for perchlorate in water are
sufficient. The detection limits achieved for perchlorate in water and in plant tissues were 4 yUg/L
and 0.4 mg/kg, respectively (Parsons, 2001; Smith et al., 2001).
January 16, 2002
10-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
10.2.3.4 Carnivores
Available evidence indicates that concentrations in terrestrial invertebrates are less than the
concentrations in plants and similar to that in soils. As a result, there currently is no indication
that terrestrial carnivores are at additional risk from perchlorate. Risks of direct toxic effects are
therefore lower for carnivores than herbivores. In locations where perchlorate levels are
sufficient to significantly affect herbivores, carnivores are more likely to be affected by loss of
prey than by perclorate toxicity. Therefore toxic effects are not quantified.
10.2.4 Uncertainties
This discussion of uncertainties is limited to qualitative uncertainties associated with major
gaps in the data available for ecological risk assessment of perchlorate. This is because, as with
other screening assessments, quantitative uncertainties are treated through the use of conservative
assumptions. It is also because the data gaps are the major sources of uncertainty, not
imprecision or inaccuracy of the available data.
10.2.4.1 Uncertainties Concerning Aquatic Risks
Aquatic Exposures. The primary uncertainty associated with this assessment of aquatic
risks is the paucity of data on perchlorate occurrence in surface waters. For lack of systematic
sampling and analysis, the spatial and temporal distribution of perchlorate in water is unknown.
It is not certain whether the reported concentrations in water represent the highest existing levels.
This is not a large source of uncertainty for this screening assessment if it is assumed that
sampling has been biased to areas of highest likely contamination. However, it would be a major
source of uncertainty in any subsequent definitive assessment.
Aquatic Effects. While the effects of perchlorate on some species of algae are known, the
effects on aquatic macrophytes are unknown. As a result, risks to aquatic primary producers are
estimated using only the chronic toxicity test results for the alga Selenastrum. Because of
physiological differences between algae and vascular plants, effects on aquatic primary producers
are not adequately assessed. In addition, it us unknown how or if physiological variations among
various species of algae or plants may affect their susceptibility to perchlorate.
Algae, aquatic macrophytes, and terrestrial leaf litter are the bases of food chains in many
aquatic ecosystems. Because perchlorate has been shown to concentrate in leaves of terrestrial
January 16,2002 10-11 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
plants and aquatic plants, the potential for direct impacts to primary consumers (i.e., planktonic
and benthic invertebrate communities) is a concern that could not be addressed in this
assessment.
A 35-day, early-life stage (ELS) test with Pimephales, generally regarded as a chronic test
but short of a full-life-cycle test, showed no significant effects on any standard endpoint
(survival, growth or biomass) at the highest concentration tested (490 mg/L). However, all
larvae exposed to perchlorate concentrations, including the lowest concentration of 28 mg/L,
exhibited redness and swelling, which was not observed in the larvae exposed to the control
water. This finding suggests the presence of subtle effects that could be ecologically significant
and raises doubt about whether a chronic NOEC has been adequately determined for this species.
This uncertainty is displayed as a range surrounding the secondary chronic value (i.e., < 0.32 to
> 0.83 mg/L). Because of the inequality signs, even the width of the range is uncertain. For this
reason, and because of the potential for chronic effects caused by thyroid dysfunction, chronic
effects should be investigated in a full life cycle test.
The uncertainty factors in the secondary chronic value are high because of the lack of test
results for aquatic organisms other than fathead minnows, amphipods, and daphnids.
10.2.4.2 Uncertainties Concerning Terrestrial Risks
Terrestrial Exposure. The available data concerning aqueous perchlorate levels is sparse
and has not been collected systematically. As a result, the spatial and temporal distribution of
perchlorate in irrigation water is unknown. It is not clear that the reported concentrations in
water represent the highest existing levels. This is not a major source of uncertainty for this
screening assessment if it is assumed that sampling has been biased to areas of highest likely
contamination. However, it would be a major source of uncertainty in any subsequent definitive
assessment.
The fate of perchlorate in soil, including its tendency for evaporative concentration, is not
well characterized. As a result, soil concentrations were assumed to be equal to irrigation water
concentrations. This assumption could be low by multiple orders of magnitude if evaporative
concentration occurs with perchlorate, as it does with metals. The limited data for irrigated soils
near Yuma (Parsons, 2001) do not support the occurrence of such a high degree of evaporative
January 16, 2002
10-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
concentration, but neither are they sufficient to rule out concentration by up to a factor of 10 or
so. More information on the fate of perchlorate in irrigated soils is needed.
The bioconcentration of perchlorate by plants suggests that perchlorate may be elevated in
leaves and leaf litter to levels that may affect invertebrate herbivores and soil invertebrate
communities. For lack of data concerning dietary toxicity, risks to invertebrates by this route
were not assessed.
Available toxicity data for rodents suggest that vertebrate herbivores may be sensitive to
low levels of perchlorate in plant tissues; concentrations potentially causing toxicity are
calculated to be lower than those currently detectable by chemical analyses of plants. In Parsons
(2001), detection limits for plants were generally about 0.4 mg/kg wet weight; similar detection
limits were achieved by Ellington and Ellis (2000) and Ellington et al. (2001), as compared to an
exposure benchmark of 0.01 mg/kg in plant tissue for a representative herbivore (see Section
8.3.2.2). Therefore, lower detection limits for perchlorate in plant tissues may be needed to
completely assess the risks to vertebrate herbivores.
Terrestrial Effects. The toxicity of perchlorate to nonmammalian vertebrate wildlife is
unknown. As a result, risks to birds, reptiles, and amphibians could not be assessed.
The toxicity of perchlorate to terrestrial invertebrates, other than acute lethality to
earthworms, is unknown. As a result, risks to other terrestrial invertebrates were inadequately
assessed.
10.2.5 Research Needs
Three questions were asked of the screening ecological risk assessment for perchlorate:
• Are ecological risks best characterized as de minimis (exposures clearly are below levels of
concern), de manifestis (risks are clearly significant and require management action to reduce
exposures); or somewhere in between and requiring further characterization?
• Are analytical detection methods for determining levels of perchlorate in the environment
sufficient, or is there a likelihood of adverse ecological effects occurring at levels below current
detection limits?
• Is the available ecotoxicological information on perchlorate sufficient, or are additional studies
needed?
January 16, 2002
10-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
In the immediate vicinity of facilities that were involved in the manufacture, use, or
disposal of perchlorate salts, particularly facilities involved in handling of solid rocket
propellents, ecological exposure can exceed levels of concern and management actions may be
needed to reduce these exposures. Site-specific risk assessments would be needed to guide
remediation of such locally contaminated sites. Farther from such facilities, ecological exposures
appear to be below levels of concern.
The analytical detection methods for perchlorate are generally sufficient, and there appears
to be no indication of adverse ecological effects occurring at levels below detection limits, except
that detection limits in plant tissues are not low enough to ensure that risks to herbivores are
detected. Additionally, there is some uncertainty about the potential for bioaccumulation at low
concentrations of perchlorate in surface water, because of differences in the analytical detection
limits between water and animal tissues.
The available ecotoxicological information on perchlorate is sufficient for this screening-
level ecological risk assessment. However, additional ecotoxicological studies could reduce the
uncertainties about the toxicity of perchlorate to other potential ecological receptors.
While the available information may yield an adequate screening level ecological risk
assessment, the following research needs for exposure and effects analysis deserve mention.
10.2.5.1 Exposure
Concerning exposure, at least three important issues remain unresolved:
• Because the available data on accumulation in terrestrial and aquatic vascular plants are from
studies that were not designed to quantify accumulation factors, the accumulation of
perchlorate in terrestrial and aquatic plants should be further investigated.
• Because of the potential for evaporative concentration, the fate of perchlorate in irrigated soils
should be investigated.
• Because the concentrations that have potential for dietary toxicity to vertebrate herbivores are
less than the limits of detection currently achievable by chemical analysis of plants, analytical
methods for plant tissues that could lower the limits of detection should be investigated.
January 16, 2002
10-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
10.2.5.2 Effects
Also requiring further attention are issues related to the effects of potential perchlorate
exposure:
• The effects of exposure of aquatic plants should be determined.
• The effects of exposure of noncrustacean invertebrates should be determined.
• The effects of dietary exposure to perchlorate should be determined in birds and in herbivorous
or litter-feeding invertebrates.
• The effects of dietary and cutaneous exposure to perchlorate should be determined for adult
amphibians and aquatic reptiles.
• If perchlorate occurs at significant levels in estuarine systems, its toxicity in saline waters
should be determined.
10.2.5.3 Site-Specific Investigations
Some of the research needs that were listed in the previous ERD of this document have
been met by the research conducted by the US Air Force IERA (Parsons, 2001) in which
perchlorate concentrations in environmental media (i.e., surface soils, surface water, sediments,
and pore water) and biological tissues (i.e., terrestrial plants, invertebrates, reptiles, birds, and
mammals and aquatic vegetation, invertebrates, fish, amphibians, reptiles, and birds) were
surveyed at six sites with known perchlorate contamination. These data are supplemented by
additional sampling at one of the sites, Longhorn Army Ammunition Plant in Texas, by Smith
et al. (2001). These studies do address some questions about exposure that were expressed in the
previous ERD of this document (U.S. EPA, 1998d), i.e:
• Because concentrations of perchlorate in water are poorly known, and
concentrations in soil and biota are unknown, a survey of perchlorate contamination
should be conducted.
• Because, contrary to expectations, perchlorate accumulates to high concentrations in
terrestrial vascular plants, the accumulation of perchlorate in aquatic plants and in
animals should be investigated.
However, these studies were screening-level surveys that took small numbers of samples during
limited periods of time. In addition, the studies were not designed to address questions about the
effects of exposure. In some locations, concentrations in environmental media were high enough
January 16, 2002
10-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
that toxicity to ecological receptors was highly likely (i.e., the risks were de manifestis), and in
other locations toxicity could not be ruled out (i.e., the risks could not be termed de minimus).
Therefore, systematic sampling is needed in these locations to more definitively quantify
exposures and effects, so that the likelihood, nature and extent of ecological risks may be
quantified, appropriate remedial alternatives may be designed, and effectiveness of site cleanup
may be judged. In addition, site surveys may be required in other locations where perchlorate
contamination is suspected.
10.3 CHARACTERIZATION PROGRESS SUMMARY
Despite the fact that the appreciation of widespread perchlorate contamination emerged
only five years ago, considerable progress has been made in hazard identification and
quantitative dose-response characterization for both the human health and ecotoxicological risks
of potential perchlorate exposures. The thyroid has been confirmed as the target tissue in
humans, laboratory animals, and wildlife. The key event of the mode of action for perchlorate is
iodide uptake inhibition at the NIS with the potential for both subsequent neurodevelopmental
and neoplastic sequelae. A harmonized human health reference dose has been proposed to be
protective for both sequelae based on a mode of action model. Data insufficiencies for various
ecotoxicological receptors and for accurate exposure estimates precludes other than a screening-
level assessment at this time. Additional research is needed to determine the contribution of
exposure sources other than drinking water. This requires more progress in the area of analytical
methods to extend current approaches to other media.
As with any risk assessment, additional insights and new research will continue to change
our understanding as the knowledge base is informed with new data and as the scientific and
technical areas relevant to the particular risk characterization mature and evolve. Work
dedicated to the areas defined in this chapter should allow continued improvement of the risk
characterizations for perchlorate in the future.
January 16, 2002
10-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
11. REFERENCES
Abu-Omar, M. M.; Appelman, E. H.; Espenson, J. H. (1996) Oxygen-transfer reactions of methylrhenium oxides.
Inorg. Chem. 35: 7751-7757.
Abu-Omar, M. M.; McPherson, L. D.; Arias, J.; Bereau, V. M. (2000) Clean and efficient catalytic reduction of
perchlorate. Angew. Chem. Int. Ed. Engl. 39: 4310-4313.
Adams, J.; Barone, S., Jr.; LaMantia, A.; Philen, R.; Rice, D. C.; Spear, L.; Susser, E. (2000) Workshop to identify
critical windows of exposure for children's health: Neurobehavioral Work Group summary. Environ. Health
Perspect. 108(suppl. 3): 535-544.
Ajjan, R. A.; Kamaruddin, N. A.; Crisp, M.; Watson, P. F.; Ludgate, M.; Weetman, A. P. (1998) Regulation and
tissue distribution of the human sodium iodide symporter gene. Clin. Endocrinol. 49: 517-523.
Allred, M. (1998) Chemical specific consultation for perchlorate. Atlanta, GA: U.S. Department of Health and
Human Services, Agency for Toxic Substances and Disease Registry; October 8.
Altman, P. L.; Dittmer, D. S. (1971a) Volume of blood in tissue: vertebrates. In: Respiration and circulation.
Bethesda, Md. Federation of American Societies for Experimental Biology; pp. 383-387.
Altman, P. L.; Dittmer, D. S. (1971b) Blood volumes. In: Respiration and circulation. Bethesda, Md. Federation of
American Societies for Experimental Biology; pp. 376-383.
American Pacific Corporation. (1998) [Letter with attachments to Mr. John Kemmerer, U.S. Environmental
Protection Agency, Region 9, concerning perchlorate at American Pacific Corporation]. Las Vegas, NV;
April 14.
Anbar, M.; Guttmann, S.; Lewitus, Z. (1959) The mode of action of perchlorate ions on the iodine uptake of the
thyroid gland. Int. J. Appl. Radiat. Isot. 7: 87-96.
Andersen, M. E.; Krishnan, K.; Conolly, R. B.; McClellan, R. O. (1992) Mechanistic toxicology reserach and
biologically-based modeling: partners for improving quantitative risk assessments. CIIT Activities
12(1): 1-7.
Argus Research Laboratories, Inc. (1998a) A neurobehavioral developmental study of ammonium perchlorate
administered orally in drinking water to rats [report amendment: July 27], Horsham, PA: Argus Research
Laboratories, Inc.; protocol no. 1613-002.
Argus Research Laboratories, Inc. (1998b) Oral (drinking water) two-generation (one litter per generation)
reproduction study of ammonium perchlorate in rats. Horsham, PA: Argus Research Laboratories, Inc.;
protocol no. 1416-001.
Argus Research Laboratories, Inc. (1998c) Oral (drinking water) developmental toxicity study of ammonium
perchlorate in rabbits [report amendment: September 10]. Horsham, PA: Argus Research Laboratories, Inc.;
protocol no. 1416-002.
Argus Research Laboratories, Inc. (1998d) Oral (drinking water) dosage-range developmental toxicity study of
ammonium perchlorate in rabbits. Final pilot report. Horsham, PA: Argus Research Laboratories, Inc.;
protocol no. 1416-002P; December 10.
Argus Research Laboratories, Inc. (1999) Oral (drinking water) two-generation (one litter per generation)
reproduction study of ammonium perchlorate in rats. Horsham, PA: Argus Research Laboratories, Inc.;
protocol no. 1416-001.
Argus Research Laboratories, Inc. (2000) Oral (drinking water) developmental toxicity study of ammonium
perchlorate in rats. Horsham, PA: Argus Research Laboratories, Inc.; protocol no. 1416-003D.
Argus Research Laboratories, Inc. (2001) Hormone, thyroid and neurohistological effects of oral (drinking water)
exposure to ammonium perchlorate in pregnant and lactating rats and in fetuses and nursing pups exposed
to ammonium perchlorate during gestation or via maternal milk. Horsham, PA: Protocol no. ARGUS
1416-003.
Atterwill, C. K.; Collins, P.; Brown, C. G.; Harland, R. F. (1987) The perchlorate discharge test for examining
thyroid function in rats. J. Pharmacol. Methods 18: 199-203.
Barton, H. A.; Andersen, M. E.; Allen, B. (1998) Dose response characteristics of uterine responses in rats exposed
to estrogen agonists. Regul. Toxicol. Pharmacol. 28: 133-149.
Barzilai, D.; Sheinfeld, M. (1966) Fatal complications following use of potassium perchlorate in thyrotoxicosis:
report of two cases and a review of the literature. Isr. J. Med. Sci. 2: 453-456.
January 16, 2002
11-1
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Beals, J. A. J.; Funk, L. M.; Fountain, R.; Sedman, R. (1996) Quantifying the distribution of inhalation exposure in
human populations: distribution of minute volumes in adults and children. Environ. Health Perspect.
104: 974-979.
Beck, B. (2001) Protocol and time line status of Loma Linda study funded by Lockheed Martin [personal
communication to Annie M. Jarabek]. Boston, MA: Gradient Corporation; December 19.
Bekkedal, M. Y. V.; Carpenter, T.; Smith, J.; Ademujohn, C.; Maken, D.; Mattie, D. R. (2000) A
neurodevelopmental study of the effects of oral ammonium perchlorate exposure on the motor activity of
pre-weanling rat pups. Wright-Patterson Air Force Base, OH: Naval Health Research Center Detachment,
Neurobehavioral Effects Laboratory; report no. TOXDET-00-03.
Bemillon, P.; Bois, F. Y. (2000) Statistical issues in toxicokinetic modeling: a bayesian perspective. Environ. Health
Perspect. 108(suppl 5): 883-893.
BioReliance. (1999) In vitro mammalian cell gene mutation test (L5178Y/TK+/"mouse lymphoma assay). Final
report. Rockville, MD: laboratory study number G98BA06.702. Available:
http://www.tera.org/Perchlorate/welcome.htm [20 December, 2001],
Block Environmental Services, Inc. (1998) LC50 aquatic toxicity test results for ammonium perchlorate—
a two-species chronic definitive bioassay. San Jose, CA: Santa Clara Valley Water District; August 21.
Bookstein, F. L.; Sampson, P. D.; Streissguth, A. P.; Connor, P. D. (2001) Geometric morphometries of corpus
callosum and subcortical structures in the fetal-alcohol-affected brain. Teratology 64: 4-32.
Brabant, G. (1994) Personal communication with Dr. G. Brabant concerning ongoing perchlorate work in humans
by Drs. Donald Tocco and Bruce Mulholt in March and April 1994 [as cited in Toxicology Excellence for
Risk Assessment, 1997].
Brabant and Letiolf (2000) Unpublished study. [As cited in Merrill (2001a)].
Brabant, G.; Bergmann, P.; Kirsch, C. M.; Kohrle, J.; Hesch, R. D.; Von Zur Muhlen, A. (1992) Early adaptation of
thyrotropin and thyroglobulin secretion to experimentally decreased iodine supply in man. Metabolism
41: 1093-1096.
Braverman, L. E.; Utiger, R. D. (1991) Introduction to thyrotoxicosis. In: Braverman, L. E.; Utiger, R. D., eds.
Werner and Ingbar's the thyroid: a fundamental and clinical text. 6th ed. Philadelphia, PA: J.B. Lippincott
Company; pp. 645-647.
Brechner, R. J.; Parkhurst, G. D.; Humble, W. O.; Brown, M. B.; Herman, W. H. (2000) Ammonium perchlorate
contamination of Colorado River drinking water is associated with abnormal thyroid function in newborns
in Arizona. J. Occup. Environ. Med. 42: 777-782.
Brown, R. A.; Al-Moussa, M.; Beck, J. (1986) Histometry of normal thyroid in man. J. Clin. Pathol. 39: 475-482.
Brown, R. P.; Delp, M. D.; Lindstedt, S. L.; Rhomberg, L. R.; Beliles, R. P. (1997) Physiological parameter values
for physiologically based pharmacokinetic models. Toxicol. Ind. Health 13: 407-484.
Brown-Grant, K. (1961) Extrathyroidal iodide concentrating mechanisms. Physiol. Rev. 41: 189-213.
Brown-Grant, K. (1966) Failure of orally administered perchlorate to affect deciduoma formation or pregnancy in
the rat. J. Reprod. Fertil. 12: 353-357.
Brown-Grant, K.; Pethes, G. (1959) Concentration of radioiodine in the skin of the rat. J. Physiol. 148: 683-693.
Brown-Grant, K.; Sherwood, M. R. (1971) Viability of the rat blastocyst following the oral administration of
potassium perchlorate or potassium iodide to the mother. J. Reprod. Fertil. 27: 265-267.
BRT-Burleson Research Technologies, Inc. (2000a) Ammonium perchlorate: effect on immune function. Quality
assurance audit: study no. BRT 19990524 — plaque-forming cell (PFC) assay; study no. BRT 19990525-
local lymph node assay (LLNA) in mice. Raleigh, NC.
BRT-Burleson Research Technologies, Inc. (2000b) Addendum to study report: ammonium perchlorate: effect on
immune function [with cover letter dated August 31 from G. R. Burleson]. Raleigh NC.
BRT-Burleson Research Technologies, Inc. (2000c) Ammonium perchlorate: effect on immune function.
Raleigh, NC: BRT 19990524 study protocol: plaque-forming cell (PFC) assay; BRT 19990525 study
protocol: local lymph node assay (LLNA) in mice.
Buelke-Sam, J.; Nelson, C. J.; Byrd, R. A.; Holson, J. F. (1982a) Blood flow during pregnancy in the rat: I. Flow
patterns to maternal organs. Teratology 26: 269-277.
Buelke-Sam, J.; Holson, J. F.; Nelson, C. J. (1982b) Blood flow during pregnancy in the rat: II. Dynamics of and
litter variability in uterine flow. Teratology 26: 279-288.
Burgi, H.; Benguerel, M.; Knopp, J.; Kohler, H.; Studer, H. (1974) Influence of perchlorate on the secretion on
non-thyroxine iodine by the normal human thyroid gland. Eur. J. Clin. Invest. 4: 65-69.
January 16, 2002
11-2
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Burke, P. J.; Henrick, K.; McMillin, D. R. (1982) Crystal and molecular structures of
bis(4,4',6,6'-tetramethyl-2,2'-bipyridyl)copper(I) perchlorate,
bis(4,4',6,6'-tetramethyl-2,2'-bipyridyl)copper(II) diperchlorate, and
bis(4,4',6,6'-tetramethyl-2,2'-bipyridyl)copper(II) diperchlorate dihydrate. A search for copper(II) and
copper(I) complexes with a common ligand environment. Inorg. Chem. 21: 1881-1886.
Caldwell, D. J.; Mattie, D. R. (1995) Study design for the toxicity evaluation of ammonium perchlorate.
In: Proceedings of the 1995 JANNAF safety and environmental protection subcommittee joint workshop,
environmentally sound processing technology; July; Tampa, FL. Columbia, MD: Chemical Propulsion
Information Agency; Joint Army, Navy, NASA, Air Force (JANNAF) interagency propulsion committee
publication 626.
Caldwell, D. J.; King, J. H., Jr.; Kinkead, E. R.; Wolfe, R. E.; Narayanan, L.; Mattie, D. R. (1995) Results of a
fourteen day oral-dosing toxicity study of ammonium perchlorate. In: Proceedings of the 1995 JANNAF
safety and environmental protection subcommittee meeting: volume 1; December; Tampa, FL. Columbia,
MD: Chemical Propulsion Information Agency; Joint Army, Navy, NASA, Air Force (JANNAF)
interagency propulsion committee publication 634.
California Department of Health Services. (1997) Preliminary health reviews in Rancho Cordova, Sacramento
County, California [health consultation of the Aerojet General Corporation superfund site under the
Comprehensive Environmental Response, Compensation, and Liability Act of 1980], Atlanta, GA: U.S.
Department of Health and Human Services, Agency for Toxic Substances and Disease Registry; CERCLIS
no. CAD980358832; October 16.
California Department of Health Services. (1998a) Perchlorate contamination in the Citizens Utilities' suburban and
Security Park water service areas [health consultation of the Aerojet General Corporation superfund site
under the Comprehensive Environmental Response, Compensation, and Liability Act of 1980]. Atlanta,
GA: U.S. Department of Health and Human Services, Agency for Toxic Substances and Disease Registry;
CERCLIS no. CAD980358832; March 18.
California Department of Health Services. (1998b) Perchlorate contamination in the Arden Cordova water service
area [health consultation of the Aerojet General Corporation superfund site under the Comprehensive
Environmental Response, Compensation, and Liability Act of 1980], Atlanta, GA: U.S. Department of
Health and Human Services, Agency for Toxic Substances and Disease Registry; CERCLIS no.
CAD980358832; April 21.
California Department of Health Services. (1998c) Perchlorate contamination in the Mather Air Force Base water
service area [health consultation of the Aerojet General Corporation superfund site under the
Comprehensive Environmental Response, Compensation, and Liability Act of 1980]. Atlanta, GA: U.S.
Department of Health and Human Services, Agency for Toxic Substances and Disease Registry; CERCLIS
no. CAD980358832; May 6.
California Department of Health Services. (1998d) Perchlorate contamination in the Fair Oaks water district [health
consultation of the Aerojet General Corporation superfund site under the Comprehensive Environmental
Response, Compensation, and Liability Act of 1980], Atlanta, GA: U.S. Department of Health and Human
Services, Agency for Toxic Substances and Disease Registry; CERCLIS no. CAD980358832; June 5.
California Department of Health Services. (1998e) Perchlorate contamination in the Sunrise district of the
Sacramento County water service [health consultation of the Aerojet General Corporation superfund site
under the Comprehensive Environmental Response, Compensation, and Liability Act of 1980], Atlanta,
GA: U.S. Department of Health and Human Services, Agency for Toxic Substances and Disease Registry;
CERCLIS no. CAD980358832; June 18.
California State Senate. (1998) An act to amend Section 116275 of, and to add Section 116611 to, the Health and
Safety Code, relating to drinking water. Sacramento, CA: Senate bill 1033. Available:
http://www.sen.ca.gov/ [2 January, 2002].
Campbell, B. A.; Lytle, L. D.; Fibiger, H. C. (1969) Ontogeny of adrenergic arousal and cholinergic inhibitory
mechanisms in the rat. Science (Washington, DC) 166: 635-637.
Canaris, G. J.; Manowitz, N. R.; Mayor, G.; Ridgway, E. C. (2000) The Colorado thyroid disease prevalence study.
Arch. Intern. Med. 160: 526-534.
Cao, X. Y.; Jiang, X. M.; Dou, Z. H.; Rakeman, M. A.; Zhang, M. L.; O'Donnel, K.; Ma, T.; Amette, K.;
DeLong, N.; DeLong, G. R. (1994) Timing of vulnerability of the brain to iodine deficiency in endemic
cretinism. N. Engl. J. Med. 331: 1739-1744.
January 16, 2002
11-3
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Capek, K.; Jelinek, J. (1956) The development of water metabolism. I. The excretion of urine in young rats.
Physiol. Bohemoslov. 5: 91-96.
Capen, C. C. (1997) Mechanistic data and risk assessment of selected toxic end points of the thyroid gland.
Toxicol. Pathol. 25: 39-48.
Caron, P.; Hoff, M.; Bazzi, S.; Dufor, A.; Faure, G.; Ghandour, I.; Lauzu, P.; Lucas, Y.; Maraval, D.; Mignot, F.;
Ressigeac, P.; Vertongen, F.; Grange, V. (1997) Urinary iodine excretion during normal pregnancy in
healthy women living in the southwest of France: correlation with maternal thyroid parameters. Thyroid
7: 749-754.
Cavalien, R. R. (1997) Iodine metabolism and thyroid physiology: current concepts. Thyroid 7: 177-181.
Chandra, M.; Riley, M. G.; Johnson, D. E. (1992) Spontaneous neoplasms in aged Sprague-Dawley rats. Arch.
Toxicol. 66: 496-502.
Channel, S. R., Maj. (1998a) Consultative letter, AFRL-HE-WP-CL-1998-0031, pharmacokinetic study of
perchlorate administered orally to humans [memorandum to Annie Jarabek], Wright-Patterson Air Force
Base, OH: Air Force Research Laboratory, Human Effectiveness Directorate; December 18.
Channel, S. R., Maj. (1998b) Consultative letter, AFRL-HE-WP-CL-1998-0027, morphometric analysis report -
thyroid: a neurobehavioral developmental study of ammonium perchlorate administered orally in drinking
water to rats [memorandum with attachments to Annie Jarabek]. Wright-Patterson Air Force Base, OH:
Human Effectiveness Directorate, Operational Toxicology Branch; October 26.
Channel, S. R., Maj. (1998c) Consultative letter, AFRL-HE-WP-CL-1998-0026, histopathology report for thyroids
from a fourteen-day oral dosing toxicity study of ammonium perchlorate [memorandum with attachments to
Annie Jarabek]. Wright-Patterson Air Force Base, OH: Human Effectiveness Directorate, Operational
Toxicology Branch; October 27.
Channel, S. R., Maj. (1999a) Consultative letter, AFRL-HE-WP-CL-1999-0005, kinetic data for iodide uptake
inhibition in the thyroid by perchlorate (2-week drinking water study) [memorandum to Annie Jarabek],
Wright-Patterson Air Force Base, OH: Air Force Research Laboratory; February 5.
Channel, S. R., Lt. Col. (1999b) Consultative letter, AFRL-HE-WP-CL-1999-0032, summary of human kinetic data
on perchlorate [memorandum with attachments for NASA HQ/EM], Wright-Patterson Air Force Base, OH:
Air Force Research Laboratory; December 20.
Chow, S. Y.; Woodbury, D. M. (1970) Kinetics of distribution of radioactive perchlorate in rat and guinea-pig
thyroid glands. J. Endocrinol. 47: 207-218.
Chow, S. Y.; Chang, L. R.; Yen, M. S. (1969) A comparison between the uptakes of radioactive perchlorate and
iodide by rat and guinea-pig thyroid glands. J. Endocrinol. 45: 1-8.
Clegg, E. D. (1999) Assimilation of F1 mating, estrous cyclicity and sperm measure results with P results
[memorandum to Annie M. Jarabek], Washington, DC: U.S. Environmental Protection Agency, National
Center for Environmental Assessment; January 28.
Clewell, R. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0037, preliminary physiological model for
perchlorate in the lactating rat [memorandum with attachment to Annie Jarabek]. Wright-Patterson Air
Force Base, OH: Air Force Research Laboratory; June 28.
Clewell, R. A. (2001a) Consultative letter, AFRL-HE-WP-CL-2001-0006, physiologically-based pharmacokinetic
model for the kinetics of perchlorate-induced inhibition of iodide in the pregnant rat and fetus
[memorandum with attachments to Annie Jarabek]. Wright-Patterson AFB, OH: Air Force Research
Laboratory; May 10.
Clewell, R. A. (2001b) Consultative letter, AFRL-HE-WP-CL-2001-0007, physiologically-based pharmacokinetic
model for the kinetics of perchlorate-induced inhibition of iodide in the lactating and neonatal rat
[memorandum with attachments to Annie Jarabek]. Wright-Patterson AFB, OH: Air Force Research
Laboratory; May 24.
Clewell, R. (2001c) Consultative letter, AFRL-HE-WP-CL-2001-0009, comparison of AFRL/HEST study protocol
for effects of perchlorate exposure via drinking water during pre- and postnatal development versus a
similar study conducted by Argus Research Laboratories, Inc. [memorandum with attachments to Annie
Jarabek]. Wright-Patterson, AFB, OH: Air Force Research Laboratory; June 7.
Coates, J. D.; Michaelidou, U.; Bruce, R. A.; O'Connor, S. M.; Crespi, J. N.; Achenbach, L. A. (1999) The ubiquity
and diversity of dissimilatory perchlorate-reducing bacteria. Appl. Environ. Microbiol. 65: 5234-5241.
Coates, J. D.; Michaelidou, U.; O'Connor, S. M.; Bruce, R. A.; Achenbach, L. A. (2000) The diverse microbiology
of perchlorate reduction. In: Urbansky, E. T., ed. Perchlorate in the environment. New York, NY: Kluwer
Academic Publishers. (Environmental science research: v. 57).
January 16, 2002
11-4
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Cohen, J. (1987) Statistical power analysis for the behavioral sciences. Revised edition. Hillsdale, NJ: Lawrence
Erlbaum Associates, Publishers.
Collette, T. W.; Williams, T. L. (2000) Perchlorate in fertilizers? Analysis by Raman spectrometry. Paper presented
before the Division of Fertilizer and Soil Chemistry at the American Chemical Society 220th national
meeting; August; Washington, DC.
Comer, C. P.; Norton, S. (1982) Effects of perinatal methimazole exposure on a developmental test battery for
neurobehavioral toxicity in rats. Toxicol. Appl. Pharmacol. 63: 133-141.
Conde, E.; Martin-Lacave, 1.; Godalez-Campora, R.; Galera-Davidson, H. (1991) Histometry of normal thyroid
glands in neonatal and adult rats. Am. J. Anat. 191: 384-390.
Condike, B. J. (2001) Perchlorate data in fish and plants [letter with attachments to Annie M. Jarabek]. Fort Worth,
TX: Department of the Army, Fort Worth District, Corps of Engineers; December 21.
Connell, J. M. (1981) Long-term use of potassium perchlorate. Postgrad. Med. J. 57: 516-517.
Consultants in Veterinary Pathology. (2001) Abbreviated morphometry report with appended thumbnails of scanned
sections. Protocol 1416-003. Hormone, thyroid and neurohistological effects of oral (drinking water)
exposure to ammonium perchlorate in pregnant and lactating rats and in fetuses and nursing pups exposed
to ammonium perchlorate during gestation or via maternal milk. Murrysville, PA. (Supplement to
neuropathology report on F1 generation day 10 and day 22 postpartum rats).
Cory-Slechta, D. A.; Crofton, K. M.; Foran, J. A.; Ross, J. F.; Sheets, L. P.; Weiss, B.; Mileson, B. (2001) Methods
to identify and characterize developmental neurotoxicity for human health risk assessment. I: Behavioral
effects. Environ. Health Perspect. 109(suppl. 1): 79-91.
Cox, C. (1994) Statistical issues for animal studies of developmental neurotoxicity. In: Weiss, B.; O'Donoghue,
J. L., eds. Neurobehavioral toxicity: analysis and interpretation. New York, NY: Raven Press; pp. 93-101.
Crofton, K. M. (1998a) Analysis and graphics of thyroid hormone data from the rat 14-day "Caldwell" perchlorate
study [memorandum with attachment to Annie Jarabek]. Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health and Environmental Effects Research Laboratory; October 18 (revised
November 21).
Crofton, K. M. (1998b) Re-analysis of thyroid hormone data from the subchronic perchlorate study submitted by
Springbom Laboratories (SLI study no. 3455.1) [memorandum with attachment to Annie Jarabek].
Research Triangle Park, NC: U.S. Environmental Protection Agency, National Health and Environmental
Effects Research Laboratory; July 21 (revised October 12 and November 18).
Crofton, K. M. (1998c) Preliminary analysis of the postnatal day 12 neuromorphology data from the rat
developmental neurotoxicology study [memorandum to Annie Jarabek]. Research Triangle Park, NC:
U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
December 29.
Crofton, K. M. (1998d) Personal communication [with Charles Capen of Ohio State University concerning the
comparative sensitivity of standard histology versus morphometric measurements and of follicular epithelial
cell height versus lumen area regarding chemically-induced changes in thyroid gland structure]. Research
Triangle Park, NC: U.S. Environmental Protection Agency, National Health and Environmental Effects
Research Laboratory; November 6.
Crofton, K. M. (1998e) Analysis of the thyroid histology data from the rat developmental neurotoxicology study
[memorandum with attachment to Annie Jarabek], Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health and Environmental Effects Research Laboratory; September 24
(revised November 6).
Crofton, K. M. (1998f) Analysis of the thyroid hormone data from the rat developmental neurotoxicology study
[memorandum with attachment to Annie Jarabek]. Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health and Environmental Effects Research Laboratory; October 22 (revised
November 6).
Crofton, K. M. (1998g) Personal communication. [With Dr. Simon Mats, Primedica, Worcester, MA, concerning
the lack of use of litter as dependent variable in the reanalysis of the motor activity data from the Argus
developmental neurotoxicity study]. Research Triangle Park, NC: U.S. Environmental Protection Agency,
National Health and Environmental Effects Research Laboratory; November 4.
Crofton, K. M. (1998h) Analysis and graphics of thyroid hormone data from the rabbit developmental perchlorate
study [memorandum with attachment to Annie Jarabek], Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health and Environmental Effects Research Laboratory; October 10 (revised
October 28).
January 16, 2002
11-5
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Crofton, K. M. (1998i) Analysis and graphics of thyroid hormone data from the mouse immunotoxicology study
[memorandum with attachment to Annie Jarabek]. Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health and Environmental Effects Research Laboratory; October 18 (revised
November 10 and 23).
Crofton, K. M. (1998j) Sensitivity of neurodevelopmental tests versus thyroid hormone concentrations
[memorandum to Annie Jarabek], Research Triangle Park, NC: U.S. Environmental Protection Agency,
National Health and Environmental Effects Research Laboratory; December 29.
Crofton, K. M. (2001a) Revised analysis of the thyroid hormone data from the mouse immunotoxicology study
[memorandum with attachments to Annie M. Jarabek], Research Triangle Park, NC: U.S. Environmental
Protection Agency, National Health Effects and Environmental Research Laboratory; December 30.
Crofton, K. M. (2001b) Revised analysis of the thyroid hormone data from the rat developmental "effects" study -
Argus protocol 1416-003 [memorandum with attachments to Annie M. Jarabek], Research Triangle Park,
NC: U.S. Environmental Protection Agency, National Health Effects and Environmental Research
Laboratory; December 14 (revised December 28).
Crofton, K. M.; Marcus, A. (2001) Re-analyses of perchlorate hormone data from the 1998 ERD [memorandum
with attachments to Annie Jarabek], Research Triangle Park, NC: U.S. Environmental Protection Agency,
Office of Research and Development; October 15.
Crofton, K. M.; MacPhail, R. C.; Tilson, H. A. (1998) Analysis of the motor activity data from the rat
developmental neurotoxicology study [memorandum to Annie Jarabek]. Research Triangle Park, NC:
U.S. Environmental Protection Agency, Office of Research and Development; November 5.
Crooks, J.; Wayne, E. J. (1960) A comparison of potassium perchlorate, methylthiouracil, and carbimazole in the
treatment of thyrotoxicosis. Lancet (7121): 401-404.
Crump, K. S.; Allen, B. C.; Faustman, E. M. (1995) The use of the benchmark dose approach in health risk
assessment. Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum; report no.
EPA/630/R-94/007. Available from: NTIS, Springfield, VA; PB95-213765.
Crump, C.; Michaud, P.; Tellez, R.; Reyes, C.; Gonzalez, G.; Montgomery, E. L.; Crump, K. S.; Lobo, G.;
Becerra, C.; Gibbs, J. P. (2000) Does perchlorate in drinking water affect thyroid function in newborns or
school-age children? J. Occup. Environ. Med. 42: 603-612.
Davis, J. M.; Elias, R. W. (1996) Risk assessment of metals. In: Chang, L. W.; Magos, L.; Suzuki, T., eds.
Toxicology of metals. Boca Raton, FL: CRC Lewis Publishers; pp. 55-67.
Dean, J. A., ed. (1985) Lange's handbook of chemistry. 13th ed. New York, NY: McGraw Hill Book Company.
DeBorba, B. M.; Urbansky, E. T. (2001) Performance of poly(vinyl alcohol) gel columns on the ion
chromatographic determination of perchlorate in fertilizers. J. Environ. Monit.: in press.
Dellarco, V. (1999) Review of the NTP mutagenicity studies on ammonium perchlorate [memorandum to Annie M.
Jarabek]. Washington, DC: U.S. Environmental Protection Agency, Office of Pollution, Prevention and
Toxic Substances; January 28.
Dionex Corporation. (2000) Determination of low concentrations of perchlorate in drinking and ground waters
using ion chromatograpy. Sunnyvale, CA; application note 134.
Dollarhide, J. S. (1992) Provisional non-cancer and cancer toxicity values for potassium perchlorate (CASRN
7778-74-7) (Aerojet General Corp./CA) [memorandum with attachment to Dan Strakla]. Cincinnati, OH:
U.S. Environmental Protection Agency, Environmental Criteria and Assessment Office; December 2.
Dollarhide, J. S. (1995) Review of proposed RfD for perchlorate [memorandum with attachment to Mike Girard].
Cincinnati, OH: U.S. Environmental Protection Agency, National Center for Environmental Assessment;
October 23.
Dollarhide, J. S. (1998) Personal communication [regarding dosing solutions of ammonium perchlorate versus anion
only]. Cincinnati, OH: Toxicology Excellence for Risk Assessment; November 30.
Dollarhide, J.; Dourson, M. (1997) Reference dose for perchlorate. Abstract from "Conference on issues and
applications in toxicology and risk assessment; April; Wright-Patterson Air Force Base, OH.
Dourson, M. (1998) Mouse rrucronucleus test for perchlorate & literature review of relevant mouse studies [report
with memorandum from Tara M. Anderson attached]. Cincinnati, OH: Toxicology Excellence for Risk
Assessment; October 12.
Dumont, J. N.; Bantle, J. A. (1998) FETAX analysis of ammonium perchlorate. Stillwater, OK: Oklahoma State
University, Department of Zoology; May 22.
January 16,2002
11-6
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Dunson, D. B. (2001a) Statistical analysis of the effects of perchlorate on neurobehavior (motor activity) in SD rats
[memorandum to Annie Jarabek]. Research Triangle Park, NC: National Institute of Environmental Health
Scienes, Biostatistics Branch; November 1.
Dunson, D. B. (2001b) Statistical analysis of the tumors observed in male F1 rats (second parental generation, P2)
at week 19 in the Argus (1999) two-generation reproduction study of ammonium perchlorate [memorandum
to Annie Jarabek]. Research Triangle Park, NC: National Institute of Environmental Health Sciences,
Biostatistics Branch; November 1.
Dunson, D. B.; Dinse, G. E. (2001) Bayesian incidence analysis of animal tumongenicity data. J. R. Stat. Soc. Ser.
C 50: 125-141.
Dunson, D. B.; Perreault, S. D. (2001) Factor analytic models of clustered multivariate data with informative
censoring. Biometrics 57: 302-308.
Dunson, D. B.; Tindall, K. R. (2000) Bayesian analysis of mutational spectra. Genetics 156: 1411-1418.
Dunson, D. B.; Haseman, J. K.; Van Birgelen, A. P. J. M.; Stasiewicz, S.; Tennant, R. W. (2000) Statistical analysis
of skin tumor data from Tg.AC mouse bioassays. Toxicol. Sci. 55: 293-302.
Durand, M. J. (1938) Recherches sur l'elimination des perchlorates, sur leur repartition dans les organes et sur leur
toxicite [Research on the elimination, distribution in organs, and toxicity of perchlorate]. Bull. Soc. Chim.
Biol. 20: 423-433.
EA Engineering, Science, and Technology, Inc. (1998) Results of acute and chronic toxicity testing with sodium
perchlorate. Brooks Air Force Base, TX: Armstrong Laboratory; report no. 2900.
EA Engineering, Science, and Technology, Inc. (1999) Results of algal toxicity testing with sodium perchlorate.
Sparks, MD: EA Engineering, Science, and Technology, Inc.
EA Engineering, Science, and Technology, Inc. (2000) Results of chronic toxicity testing with sodium perchlorate
using Hyalella azteca and Pimephales promelas. Sparks, MD: report number 3505.
Earley, J. E.; Tofan, D. C.; Amadei, G. A. (2000) Reduction of perchlorate ion by titanous ions in ethanolic
solution. In: Urbansky, E. T., ed. Perchlorate in the environment. New York, NY: Kluwer Academic
Publishers. (Environmental science research: v. 57).
Eggers, J. (1996) Toxicity evaluation of ammonium perchlorate administered in the drinking water of
Sprague-Dawley rats. Study A10. A narrative pathology report - pilot study (as cited in and attached to
Channel, 1998).
Eichler, O. (1929) Zur Pharmakologie der Perchloratwikung [The pharmacology of the perchlorate effect],
Naunyn-Schmiedebergs Arch. Exp. Pathol. Pharmakol. 144: 251-260.
Eichler, O.; Hackenthal, E. (1962) Uber Ausscheidung und Stoffwechsel von Perchlorat gemessen mit 36C104
[Secretion and metabolism of perchlorate measured with 36C104]. Naunyn-Schmidedberg's Arch. Exp.
Pathol. Pharmakol. 243: 554-565.
Eldridge, J. E.; Tsui, D. T.; Mattie, D. R.; Crown, J.; Scott, R.; Blackman, T. (1999) Perchlorate in fertilizers.
Wright-Patterson AFB, OH: Air Force Research Laboratory, Human Effectiveness Directorate; report no.
AFRL-HE-WP-TR-2000-0037.
Ellington, J. J.; Evans, J. J. (2000) Determination of perchlorate at parts-per-billion levels in plants by ion
chromatography. J. Chromatogr. A 898: 193-199.
Ellington, J. J.; Wolfe, N. L.; Garrison, A. W.; Evans, J. J.; Avants, J. K.; Teng, Q. (2001) Determination of
perchlorate in tobacco plants and tobacco products. Environ. Sci. Technol. 35: 3213-3218.
Ells, B.; Bamett, D. A.; Purves, R. W.; Guevremont, R. (2000) Trace level determination of perchlorate in water
matrices and human urine using ESI-FAIMS-MS. J. Environ. Monit. 2: 393-397.
Environmental Resources Management, Inc. (1995) Extended literature review concerning NOAEL and LOAEL
values for perchlorate. Exton, PA: Environmental Resources Management, Inc.; June 6.
Ericksen, G. E. (1983) The Chilean nitrate Deposits. Am. Sci. 71: 366-374.
Espenson, J. H. (2000) The problem and perversity of perchlorate. In: Urbansky, E. T., ed. Perchlorate in the
environment. New York, NY: Kluwer Academic Publishers. (Environmental Science Research: v. 57).
Experimental Pathology Laboratories, Inc. (2000) The effects of ammonium perchlorate on thyroids. Pathology
Working Group report. Research Triangle Park, NC: EPL project number 480-025, 480-027-480-034.
Available: http://www.epa.gov/ncea/perch.htm [2001, 17 October].
Fail, P. A.; Anderson, S. A.; Friedman, M. A. (1999) Response of the pituitary and thyroid to tropic hormones in
Sprague-Dawley versus Fischer 344 male rats. Toxicol. Sci. 52: 107-121.
Farid, N. R.; Shi, Y.; Zou, M. (1994) Molecular basis of thyroid cancer. Endocr. Rev. 15: 202-232.
January 16, 2002
11-7
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Fawcett, J. W.; Clarke, C. W. F. (1961) Aplastic anaemia due to potassium perchlorate [letter], Br. Med. J.
(5238): 1537.
Federal Register. (1991) Guidelines for developmental toxicity risk assessment. F. R. (December 5)
56: 63798-63826.
Federal Register. (1996) Proposed guidelines for carcinogen risk assessment: notice of availability and opportunity
to comment. F. R. (April 23) 61: 17960-18011.
Federal Register. (1999) Revisions to the unregulated contaminant monitoring regulation for public water systems;
final rule. F. R. (September 17) 64: 50,555-50,620.
Federal Register. (2000) Unregulated contaminant monitoring regulation for public water systems: analytical
methods for perchlorate and acetochlor; announcement of laboratory approval and performance testing (PT)
program for the analysis of perchlorate; final rule and proposed rule. F. R. (March 2) 42: 11,371-11,385.
Fisher, D. A. (1996) Disorders of the thyroid in the newborn and infant. In: Sperling, M. A., ed. Pediatric
endocrinology. Philadelphia, PA: W. B. Saunders Company; pp. 51-70.
Fisher, J. W. (1998a) Consultative letter, AFRL-HE-WP-CL-1998-0022, pharmacokinetics of iodide uptake
inhibition in the thyroid by perchlorate [memorandum with attachments to Annie Jarabek], Wright-Patterson
Air Force Base, OH: Air Force Research Laboratory, Human Effectiveness Directorate, Operational
Toxicology Branch (AFRL/HEST); October 1.
Fisher, J. W. (1998b) Personal communication to Annie Jarabek [concerning percentage of perchlorate likely to
transfer to pups during lactation], Wright-Patterson Air Force Base, OH: Air Force Research Laboratory,
Human Effectiveness Directorate, Operational Toxicology (AFRL/HEST) Branch; December 22.
Fisher, J. W. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0035, physiological model for inhibition of
thyroidal uptake of iodide by perchlorate in the rat [memorandum with attachment to Annie Jarabek].
Wright-Patterson Air Force Base, OH: Air Force Research Laboratory; June 28.
Fisher, J. W.; Whittaker, T. A.; Taylor, D. H.; Clewell, H. J., Ill; Andersen, M. E. (1989) Physiologically based
pharmacokinetic modeling of the pregnant rat: a multiroute exposure model for trichloroethylene and its
metabolite, trichloroacetic acid. Toxicol. Appl. Pharmacol. 99: 395-414.
Fisher, J. W.; Whittaker, T. A.; Taylor, D. H.; Clewell, H. J., Ill; Andersen, M. E. (1990) Physiologically based
pharmacokinetic modeling of the lactating rat and nursing pup: a multiroute exposure model for
trichloroethylene and its metabolite, trichloroacetic acid. Toxicol. Appl. Pharmacol. 102: 497-513.
Florsheim, W. H.; Faircloth, M. A.; Corcorran, N. L.; Rudko, P. (1966) Perinatal thyroid function in the rat.
Acta Endocrinol. 32: 375-332.
Garman, R. H. (2001c) Perchlorate study [letter to Annie Jarabek regarding two topographs], Murrysville, PA:
Consultants in Veterinary Pathology, Inc.
Garman, R. H. (200Id) [Letter to Annie Jarabek concerning faxed image of the sagittal brain view], Murrysville,
PA: Consultants in Veterinary Pathology, Inc.; July 17.
Garman, R. (2001a) Ammonium perchlorate study [memorandum to Ray York, Joanne Fairlie, John Bamett, Jr.,
Joan Dollarhide, Michael Dourson, Jean Harry, and Jerry Hardisty], Murrysville, PA: Consultants in
Veterinary Pathology, Inc.; August 13.
Garman, R. (2001b) The morphometry component of the perchlorate effects protocol [memorandum to Jean Harry].
Murrysville, PA: Consultants in Veterinary Pathology, Inc.; August 14.
Garman, R. H.; Fix, A. S.; Jortner, B. S.; Jensen, K. F.; Hardisty, J. F.; Claudio, L.; Ferenc, S. (2001) Methods to
identify and characterize developmental neurotoxicity for human health risk assessment. II:
Neuropathology. Environ. Health Perspect. 109(suppl 1): 93-100.
Gauss, W. (1972) Das verhalten einiger physiologischer und histologischer kriterien der schilddrusenfunktion
bei einmaliger oder langerer verabreichung von kaliumperchlorat an adulte mause I. Langzeitversuche.
Z. Mikrosk. Anat. Forsch. 4: 469-500.
Gelfand, A. E.; Hills, S. E.; Racine-Poon, A.; Smith, A. F. M. (1990) Illustration of Bayesian inference in normal
data models using Gibbs sampling. J. Am. Stat. Assoc. 85: 972-985.
Geller, A. M. (1998a) Benchmark dose calculations on thyroid data from studies submitted for evaluation of
perchlorate [memorandum with attachments to Annie Jarabek], Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
December 28.
January 16, 2002
11-8
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Geller, A. M. (1998b) Non-parametric correlations run on thyroid data from studies submitted for evaluation of
perchlorate [memorandum with attached report to Annie Jarabek]. Research Triangle Park, NC:
U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
December 21.
Geller, A. M. (1999a) Analysis of the brain morphometry data from the neurobehavioral developmental study of
ammonium perchlorate (Argus, 1998a) [memorandum with attachments to Annie M. Jarabek]. Research
Triangle Park, NC: U.S. Environmental Protection Agency, National Health and Environmental Effects
Research Laboratory; January 27.
Geller, A. M. (1999b) Analysis of the thyroid hormone data from the rat two generation reproduction study
(Argus, 1998b) [memorandum with attachments to Annie M. Jarabek]. Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
February 1.
Geller, A. M. (2001a) Benchmark dose calculations on thyroid histopathology from 1998 data reported after review
by Pathology Working Group (Wolf, 2000) [memorandum to Annie Jarabek], Research Triangle Park, NC:
U.S. Environmental Protection Agency, National Health Effects and Environmental Research Laboratory;
November 13.
Geller, A. M. (2001b) Benchmark dose calculations on thyroid histopathology from the "effects" protocol, Argus
1416-003 (Argus, 2001) [memorandum to Annie Jarabek], Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
November 13.
Geller, A. M. (2001c) Benchmark dose calculations on TSH and thyroid hormones from effects protocol, Argus
1416-003 (Argus, 2001) [memorandum to Annie Jarabek], Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
November 13.
Geller, A. M. (200 Id) Profile analysis of brain morphometry data from Argus/Primedica "effects" protocol
1416-003 [memorandum to Annie Jarabek]. Research Triangle Park, NC: U.S. Environmental Protection
Agency, National Health and Environmental Effects Research Laboratory; November 1.
Gelman, A.; Carlin, J. B.; Stem, H. S.; Rubin, D. B. (1996) Bayesian data analysis. New York, NY: Chapman and
Hall. (Texts in statistical science).
Giardino, N. J.; Eamen, N. A.; Andelman, J. B. (1992) Modeling volatilization of trichloroethylene from a domestic
shower spray: the role of drop-size distribution. Environ. Sci. Technol. 26: 1602-1606.
Gibbs, J. P.; Ahmad, R.; Crump, K. S.; Houck, D. P.; Leveille, T. S.; Findley, J. E.; Francis, M. (1998) Evaluation
of a population with occupational exposure to airborne ammonium perchlorate for possible acute or chronic
effects on thyroid function. J. Occup. Environ. Med. 40:1072-1082.
Gilks, W. R.; Thomas, A.; Spielgelhalter, D. J. (1994) A language and program for complex Bayesian modeling.
Statistician 43: 169-177.
Ginda, W. J.; Nowak. K. W.; Malendowicz. L. K. (2000) Decrease of TSH levels and epithelium/colloid ratio in rat
thyroid glands following administration of proadrenomedullin N-terminal peptide (12-20). Hormone Metab.
Res. 32: 10-14.
Gjemdal, N. (1963) Fatal aplastic anaemia following use of potassium perchlorate in thyrotoxicosis. Acta Med.
Scand. 174: 129-131.
Glinoer, D. (1998) The systematic screening and management of hypothyroidism and hyperthyroidism during
pregnancy. Trends Endocrinol. Metab. 10: 403-411.
Gluzman, B. E.; Niepomniszcze, H. (1983) Kinetics of the iodide trapping mechanism in normal and pathological
human thyroid slices. Acta Endocrinol. 103: 34-39.
Godley, A. F.; Stanbury, J. B. (1954) Preliminary experience in the treatment of hyperthyroidism with potassium
perchlorate. J. Clin. Endocrinol. 14: 70-78.
Goldey, E. S.; Kehn, L. S.; Rehnberg, G. L.; Crofton, K. M. (1995a) Effects of developmental hypothyroidism on
auditory and motor function in the rat. Toxicol. Appl. Pharmacol. 135: 67-76.
Goldey, E. S.; Kehn, L. S.; Lau, C.; Rhenberg, G. L.; Crofton, K. M. (1995b) Developmental exposure to
polychlorinated biphenyls (Arochlor 1254) reduces circulating thyroid hormone concentrations and causes
hearing deficits in rats. Toxicol. Appl. Pharmacol. 135: 77-88.
Goldman, S. J.; Stanbury, J. B. (1973) The metabolism of perchlorate in the rat. Endocrinology 92: 1536-1538.
January 16,2002
11-9
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Goleman, W. L.; Carr, J. A.; Anderson, T. A. (2002) Environmentally relevant concentrations of ammonium
perchlorate inhibit thyroid function and alter sex ratios in developing Xenopus laevis. Environ. Toxicol.
Chem. 31: in press.
Golstein, P.; Abramow, M.; Dumont, J. E.; Beauwens, R. (1992) The iodide channel of the thyroid: a membrane
vesicle study. Am. J. Physiol. 263: C590-C597.
Granger, D. N.; Barrowman, J. A.; Kvietys, P. R. (1985) Clinical gastrointestinal physiology. Philadelphia, PA:
W. B. Saunders. (Saunders monographs in physiology).
Greer, M. A.; Goodman, G.; Pleus, R. C.; Greer, S. E. (2000) Does environmental perchlorate exposure alter human
thyroid function? Determination of the dose-response for inhibition of radioiodine uptake. In: Abstracts of
the 12th International Thyroid Congress; October; Kyoto, Japan. Endocrine J. 47(suppl.): 146.
Ground-Water Remediation Technologies Analysis Center. (2001) Technology status report: perchlorate treatment
technologies, first edition. Pittsburgh, PA: Technology status report TS-01-01. Available:
http://www.gwrtac.org/html/tech_status.html [2001, October 17].
Gu, B.; Brown, G. M.; Urbansky, E. T.; Magnuson, M. L. (2000) Determination of perchlorate in fertilizers by ion
chromatography. Paper presented before the Division of Fertilizer and Soil Chemistry at the American
Chemical Society 220th national meeting; August; Washington, DC.
Haddow, J. E.; Palomaki, G. E.; Allan, W. C.; Williams, J. R.; Knight, G. J.; Gagnon, J.; O'Heir, C. E.; Mitchell,
M. L.; Hermos, R. J.; Waisbren, S. E.; Faix, J. D.; Klein, R. Z. (1999) Maternal thyroid deficiency during
pregnancy and subsequent neuropsychological development of the child. N. Engl. J. Med. 341: 549-555.
Halmi, N. S.; Stuelke, R. G. (1959) Comparison of thyroidal and gastric iodide pumps in rats. Endocrinology
64: 103-109.
Halmi, N. S.; Stuelke, R. G.; Schnell, M. D. (1956) Radioiodide in the thyroid and in other organs of rats treated
with large doses of perchlorate. Endocrinology 58: 634-650.
Hancock, P. V. (1998) Information on ammonium perchlorate [memorandum to Annie Jarabek concerning size of
ammonium perchlorate in processing and manufacturing plants].
Handy, R.; Bamett, D.; Guevremont, R. (2000) Determination of nanomolar levels of perchlorate in water by
ESI-FAIMS-MS. J. Anal. At. Spectrom. 15: 907-911.
Hanwell, A.; Linzell, J. L. (1973) The time course of cardiovascular changes in the rat. J. Physiol. 233: 99-109.
Hard, G. C. (1998) Recent developments in the investigation of thyroid regulation and thyroid carcinogenesis.
Environ. Health Perspect. 106: 427-436.
Harden, R. M.; Alexander, W. D.; Shimmins, J.; Chisholm, D. (1968) A comparison between the gastric and
salivary concentration of iodide, pertechnetate, and bromide in man. Gut 10: 928-930.
Harry, J. (2001) Comments on original experimental design, study performance, and brain morphometry results of
Argus Research Laboratories, Inc. 14 March 2001 study (protocol number 1416-003) and supplemental
materials provided by Dr. Robert Garman, Consultants in Veterinary Pathology, Inc. [letter to Annie
Jarabek]. Research Triangle Park, NC: National Institute of Environmental Health Sciences, Laboratory of
Toxicology; October 11.
Harvey, G. J.; Tsui, D. T.; Eldridge, J. E.; Orris, G. J. (1999) Society of Environmental and Toxicological Chemists;
abstract PHA015; p. 277.
Haseman, J. K.; Huff, J.; Boorman, G. A. (1984) Use of historical control data in carcinogenicity studies in rodents.
Toxicol. Pathol. 12: 126-135.
Hasselblad, V.; Jarabek, A. M. (1996) Dose-response analysis of toxic chemicals. In: Berry, D. A.; Stangl, D. K.,
eds. Bayesian biostatistics. New York, NY: Marcel Dekker, Inc.; pp. 235-259.
Hasselblad, V.; Jarabek, A. M.; Svendsgaard, D. J.; Davis, J. M. (1995) Restricted versus unrestricted models:
Why is one the magic number? In: Abstracts of the 34th annual meeting of the Society of Toxicology;
March; Baltimore, MD. Toxicologist 15: 178.
Hays, M. T.; Green, F. A. (1973) In vitro studies of 99mTc-pertechnetate binding by human serum and tissues.
J. Nucl. Med. 14: 149-158.
Hays, M. T.; Solomon, D. H. (1965) Influence of the gastrointestinal iodide cycle on the early distribution of
radioactive iodide in man. J. Clin. Invest. 44: 117-127.
Hays, M. T.; Wegner, L. H. (1965) A mathematical and physical model for early distribution of radioiodide in man.
J. Appl. Physiol. 20: 1319-1328.
January 16, 2002
11-10
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Hiasa, Y.; Kitahori, Y.; Kato, Y.; Ohshima, M.; Konishi, N.; Shimoyama, T.; Sakaguchi, Y.; Hsahimoto, H.;
Minami, S.; Murata, Y. (1987) Potassium perchlorate, potassium iodide, and propylthiouracil: promoting
effect on the development of thyroid tumors in rats treated with N-bis(2-hydrowypropyl)-nitrosamine. Jpn.
J. Cancer Res. 78: 1335-1340.
Hill, R. A. (1996) From science to decision-making: the applicability of Bayesian methods to risk asssessment.
Hum. Ecol. Risk Assess. 2: 636-642.
Hill, R. N.; Erdreich, L. S.; Paynter, O.; Roberts, P. A.; Rosenthal, S. L.; Wilkinson, C. F. (1989) Thyroid follicular
cell carcinogenesis. Fundam. Appl. Toxicol. 12: 629-697.
Hill, R. N.; Crisp, T. M.; Hurley, P. M.; Rosenthal, S. L.; Singh, D. V. (1998) Risk assessment of thyroid follicular
cell tumors. Environ. Health Perspect. 106: 447-457.
Hobson, Q. J. G. (1961) Aplastic anaemia due to treatment with potassium perchlorate [letter]. Br. Med. J.
(5236): 1368-1369.
Holson, R. R.; Pearce, B. (1992) Principles and pitfalls in the analysis of prenatal treatment effects in multiparous
species. Neurotoxicol. Teratol. 14: 221-228.
Honour, A. J.; Myant, N. B.; Rowlands, E. N. (1952) Secretion of radioiodine in digestive juices and milk in man.
Clin. Sci. 11:447-463.
Hooth, M. J.; DeAngelo, A. B.; George, M. H.; Gaillard, E. T.; Travlos, G. S.; Boorman, G. A.; Wolf, D. C. (2001)
Subchronic sodium chlorate exposure in drinking water results in a concentration-dependent increase in rat
thyroid follicular cell hyperplasia. Toxicol. Pathol. 29: 250-259.
Hurley, P. M.; Hill, R. N.; Whiting, R. J. (1998) Mode of carcinogenic action of pesticides inducing thyroid
follicular cell tumors in rodents. Environ. Health Perspect. 106: 437-445.
Hutchinson, S. L.; Susarla, S.; Wolfe, N. L.; McCutcheon, S. C. (2000) Perchlorate accumulation from
contaminated irrigation water and fertilizer in leafy vegetation. Presented at: Second international
conference on remediation of chlorinated and recalcitrant compounds; May; Monterey, CA.
Ibrahim, J. G.; Ryan, L. M.; Chen, M. H. (1998) Using historical controls to adjust for covariates in trend tests for
binary data. J. Am. Stat. Assoc. 93: 1282-1293.
lino, S.; Greer, M. A. (1961) Thyroid function in the rat during pregnancy and lactation. Endocrinology
68: 253-262.
Interagency Perchlorate Steering Committee. (1998) Perchlorate ecological risk studies. Washington, DC:
Interagency Perchlorate Steering Committee, Ecological Risk/Transport and Transformation Subcommittee;
November 13.
International Commission on Radiological Protection. (1989) Age-dependent doses to members of the public from
intake of radionuclides. Oxford, NY: Pergamon Press. (ICRP publication 56; Annals of the ICRP: v. 20,
no. 2).
International Commission on Radiological Protection. (1994) Human respiratory tract model for radiological
protection: a report of a task group of the International Commission on Radiological Protection. Oxford,
United Kingdom: Elsevier Science Ltd. (ICRP publication 66; Annals of the ICRP: v. 24, nos. 1-3).
International Commission on Radiological Protection. (2001) Doses to the embryo and fetus from intakes of
radionuclides by the mother. Oxford, United Kingdom: Elsevier Science Ltd. (ICRP publication 88;
Annals of the ICRP: v. 31, nos. 1-3).
Jackson, P. E.; Laikhtman, M.; Rohrer, J. S. (1999) Determination of trace level perchlorate in drinking water and
ground water by ion chromatography. J. Chromatogr. A 850: 131-135.
Jackson, P. E.; Gokhale, S.; Rohrer, J. S. (2000) Recent developments in the analysis of perchlorate using ion
chromatography. In: Urbansky, E. T., ed. Perchlorate in the environment. New York, NY: Kluwer
Academic Publishers. (Environmental science research: v. 57).
Jannini, E. A.; Ulisse, S.; D'Armiento, M. (1995) Thyroid hormone and male gonadal function. Endocr. Rev.
16: 443-459.
Jarabek, A. M. (1995a) The application of dosimetry models to identify key processes and parameters for default
dose-response assessment approaches. Toxicol. Lett. 79: 171-184.
Jarabek, A. M. (1995b) Interspecies extrapolation based on mechanistic determinants of chemical disposition.
Hum. Ecol. Risk Assess. 1: 641-662.
Jarabek, A. M. (1995c) Consideration of temporal toxicity challenges current default assumptions. Inhalation
Toxicol. 7: 927-946.
January 16, 2002
11-11
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Jarabek, A. M. (1998) Final consult letter. Re: morphology for neurodevelopmental study [specifically (1)
measurement of mean follicular lumen area, and (2) arbitrary choice of this measure versus follicular
height]. Research Triangle Park, NC: U.S. Environmental Protection Agency, National Center for
Environmental Assessment; October 23.
Jarabek, A. M. (1999) Review of interim pathology report in mice from 90-day immunotoxicity studies
[memorandum to EPA Perchlorate Health Assessment Team]. Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Center for Environmental Assessment; February 1.
Jarabek, A. M.; Fisher, J. W.; Rubenstein, R.; Lipscomb, J. C.; Williams, R. J.; Vinegar, A.; McDougal, J. N.
(1994) Mechanistic insights aid the search for CFC substitutes: risk assessment of HCFC-123 as an
example. Risk Anal. 14: 231-250.
Johnson, R. S.; Moore, W. G. (1961) Fatal aplastic anaemia after treatment of thyrotoxicosis with potassium
perchlorate. Br. Med. J. (5236): 1369-1371.
Johnson, R. A.; Wichern, D. W. (1988) Applied multivariate statistical analysis. 2nd ed. Englewood Cliffs, NJ:
Prentice-Hall.
Kamm, G.; Drescher, G. (1973) Der Nachweis von Perchlorat im Urin [Demonstration of perchlorate in urine].
Beitr. Genchtl. Med. 30: 206-210.
Kammuller, M. E. (1995) Drug-induced autoimmunity. Hum. Exp. Toxicol. 14: 117-119.
Kavlock, R. J.; Allen, B. C.; Faustman, E. M.; Kimmel, C. A. (1995) Dose-response assessments for developmental
toxicity: IV. Benchmark doses for fetal weight changes. Fundam. Appl. Toxicol. 26: 211-222.
Keil, D.; Warren, A.; Bullard-Dillard, R.; Jenny, M.; EuDaly, J. (1998) Effects of ammonium perchlorate on
immunological, hematological, and thyroid parameters. Charleston, SC: Medical University of South
Carolina, Department of Medical Laboratory Sciences; report no. DSWA01-97-1-008.
Keil, D.; Warren, D. A.; Jenny, M.; EuDaly, J.; Dillard, R. (1999) Effects of ammonium perchlorate on
immunotoxicological, hematological, and thyroid parameters in B6C3F1 female mice. Final report.
Charleston, SC: Medical University of South Carolina, Department of Medical Laboratory Sciences;
report no. DSWA01-97-0008.
Kerr-McGee Chemical LLC. (1998) [Letter to Mr. John Kemmerer, U.S. Environmental Protection Agency, Region
9, concerning perchlorate production at Kerr-McGee]. Oklahoma City, OK; April 17.
Kessler, F. J.; Kriiskemper, H. J. (1966) Experimentelle Schilddrusentumoren durch mehijahrige Zufuhr von
Kaliumperchlorat [Experimental thyroid tumors caused by long-term potassium perchlorate administration].
Klin. Wochenschr. 44: 1154-1156.
Kim, K.; Logan, B. E. (2001) Microbial reduction of perchlorate in pure and mixed culture packed-bed bioreactors.
Water Res. 35: 3071-3076.
Kimmel, G. (2000) Review of the oral (drinking water) developmental toxicity study of ammonium perchlorate in
rats (ARGUS 1416-003) [memorandum to Annie Jarabek]. Washington, DC: U.S. Environmental
Protection Agency, Office of Research and Development; October 17.
King, J. H., Jr. (1995) Effects of ammonium perchlorate on the thyroid hormone levels of the Sprague-Dawley rat
[thesis]. Air Force Institute of Technology; AFIT/GEE/ENV/95D-09.
Knight, C. H.; Docherty, A. H.; Peaker, M. (1984) Milk yield in the rat in relation to activity and size of the
mammary secretory cell population. J. Dairy Res. 51: 29-35.
Kodell, R. L.; West, R. W. (1993) Upper confidence limits on excess risk for quantitative responses. Risk Anal.
13: 177-182.
Koester, C. J.; Beller, H. R.; Halden, R. U. (2000) Analysis of perchlorate in groundwater by electrospray ionization
mass spectrometry/mass spectrometry. Environ. Sci. Technol. 34: 1862-1864.
Kohn, M. C.; Sewall, C. H.; Lucier, G. W.; Portier, C. J. (1996) A mechanistic model of effects of dioxin on thyroid
hormones in the rat. Toxicol. Appl. Pharmacol. 165: 29-48.
Kotani, T.; Ogata, Y.; Yamamoto, I.; Aratake, Y.; Kawano, J. I.; Suganuma, T.; Ohtaki, S. (1998) Characterization
of gastric Na+/1- symporter of the rat. Clin. Immunol. Immunopathol. 89: 271-278.
Kotyk, A.; Janacek, K. (1977) Membrane transport: an interdisciplinary approach. New York, NY: Plenum Press.
(Biomembranes: v. 9).
Krevans, J. R.; Asper, S. P., Jr.; Rienhoff, W. F, Jr. (1962) Fatal aplastic anemia following use of potassium
perchlorate in thyrotoxicosis. JAMA J. Am. Med. Assoc. 181: 162-164.
Laird, N. M.; Ware, J. H. (1982) Random-effects models for longitudinal data. Biometrics 38: 963-974.
January 16, 2002
11-12
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Lamm, S. H. (2001) Errors in the Lawrence (Braverman) 2000 (10 mg) and 2001 (3 mg) data sets [memorandum
with attachments to Annie M. Jarabek]. Washington, DC: Consultants in Epidemiology and Occupational
Health, Inc.; November 29.
Lamm, S. H.; Doemland, M. (1999) Has pcrchlorate in drinking water increased the rate of congenital
hypothyroidism? J. Occup. Environ. Med. 41: 409-411.
Lamm, S. H.; Braverman, L. E.; Li, F. X.; Richman, K.; Pino, S.; Howearth, G. (1999) Thyroid health status of
ammonium perchlorate workers: a cross-sectional occupational health study. J. Occup. Environ. Med.
41: 248-260.
Lampe, L.; Modis, L.; Gehl, A. (1967) Effect of potassium perchlorate on the foetal rabbit thyroid. Acta Med. Acad.
Sci. Hung. 23: 223-232.
Lauterbach, A. R. (2001) Reduction of perchlorate levels of sodium and potassium nitrates derived from natural
caliche ore. Presented at: 222nd American Chemical Society national meeting; August; Chicago, IL; no.
AGRO-47.
Lawrence, J. E.; Lamm, S. H.; Braverman, L. E. (1999) The use of perchlorate for the prevention of thyrotoxicosis
in patients given iodine rich contrast agents. J. Endocrinol. Invest. 22: 405-407.
Lawrence, J. E.; Lamm, S. H.; Pino, S.; Richman, K.; Braverman, L. E. (2000) The effect of short-term low-dose
perchlorate on various aspects of thyroid function. Thyroid 10: 659-663.
Lawrence, J.; Lamm, S.; Braverman, L. E. (2001) Low dose perchlorate (3 mg daily) and thyroid function [letter].
Thyroid 11: 295.
Lazarus, J. H.; Harden, R. M.; Robertson, J. W. K. (1974) Quantitative studies of the inhibitory effect of perchlorate
on the concentration of 36C104', I25I", and 99mTc04' in salivary glands of male and female mice. Arch. Oral.
Biol. 19: 493-498.
Leggett, R. W.; Williams, L. R. (1995) A proposed blood circulation model for Reference Man. Health Phys.
69: 187-201.
Li, Z.; Li, F. X.; Byrd, D.; Deyhle, G. M.; Sesser, D. E.; Skeels, M. R.; Lamm, S. H. (2000a) Neonatal thyroxine
level and perchlorate in drinking water. J. Occup. Environ. Med. 42: 200-205.
Li, F. X.; Byrd, D. M.; Deyhle, G. M.; Sesser, D. E.; Skeels, M. R.; Katkowsky, S. R.; Lamm, S. H. (2000b)
Neonatal thyroid-stimulating hormone level and perchlorate in drinking water. Teratology 62: 429-431.
Li, F. X.; Squartsoff, L.; Lamm, S. H. (2001) Prevalence of thyroid diseases in Nevada counties with respect to
perchlorate in drinking water. J. Occup. Environ. Med. 43: 630-634.
Licht, W. R.; Deen, W. M. (1988) Theoretical model for predicting rates of nitrosamine and nitrosamide formation
in the human stomach. Carcinogenesis (London) 9: 2227-2237.
Logan, B. E. (1998) A review of chlorate- and perchlorate-respiring microorganisms. Biorem. J. 2: 69-79.
Logan, B. E. (2001) Assessing the outlook for perchlorate remediation: promising technologies for treating this
widespread water contaminant are emerging. Environ. Sci. Technol. 35: 483A-487A.
Logan, B. E.; Wu, J.; Unz, R. F. (2001) Biological perchlorate reduction in high-salinity solutions. Water Res.
35: 3034-3038.
Luster, M. I.; Munson, A. E.; Thomas, P. T.; Holsapple, M. P.; Fenters, J. D.; White, K. L., Jr.; Lauer, L. D.;
Germolec, D. R.; Rosenthal, G. J.; Dean, J. H. (1988) Development of a testing battery to assess
chemical-induced immunotoxicity: National Toxicology Program's guidelines for immunotoxicity
evaluation in mice. Fundam. Appl. Toxicol. 10: 2-19.
Madeira, M. D.; Paula-Barbosa, M. M. (1993) Reorganization of mossy fiber synapses in male and female
hypothyroid rats: a stereological study. J. Comp. Neurol. 337: 334-352.
Madeira, M. D.; Cadete-Leite, A.; Andrade, J. P.; Paula-Barbosa, M. M. (1991) Effects of hypothyroidism upon the
granular layer of the dentate gyrus in male and female adult rats: a morphometric study. J. Comp. Neurol.
314: 171-186.
Madeira, M. D.; Sousa, N.; Lima-Andrade, M. T.; Calheiros, F.; Cadete-Leite, A.; Paula-Barbosa, M. M. (1992)
Selective vulnerability of the hippocampal pyramidal neurons to hypothyroidism in male and female rats.
J. Comp. Neurol. 322: 501-518.
Magnuson, M. L.; Urbansky, E. T.; Kelty, C. A. (2000a) Determination of perchlorate at trace levels in drinking
water by ion-pair extraction with electrospray ionization mass spectrometry. Anal. Chem. 72: 25-29.
Magnuson, M. L.; Urbansky, E. T.; Kelty, C. A. (2000b) Microscale extraction of perchlorate in drinking water with
low level detection by electrospray-mass spectrometry. Talanta 52: 285-291.
January 16, 2002
11-13
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Mahle, D. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0043, hormone and perchlorate data from
cross-fostering study [memorandum with attachments to Annie Jarabek]. Wright-Patterson Air Force Base,
OH: Air Force Research Laboratory; October 11.
Mahle, D. (2001) Consultative letter, AFRL-HE-WP-CL-2001-0001, hormone and perchlorate data from
cross-fostering study [memorandum with attachments to Annie Jarabek]. Wright-Patterson AFB, OH: Air
Force Research Laboratory; May 1.
Malendowicz, L. K. (1977) Sex dimorphism in the thyroid gland. I. Morphometric studies on the thyroid gland of
intact adult male and female rats. Endokrinologic 69: 326-328.
Malendowicz, L. K.; Bednarek, J. (1986) Sex dimorphism in the thyroid gland. Acta Anat. 127: 115-118.
Malik, A. B.; Kaplan, J. E.; Saba, T. M. (1976) Reference sample methods for cardiac output and regional blood
flow determinations in the rat. J. Appl. Physiol. 40: 472-475.
ManTech Environmental Technology, Inc. (1998) Genotoxicity assays for ammonium perchlorate. I.
Salmonella/microsome mutagenesis. II. Mouse lymphoma cell mutagenesis. III. In vivo mouse bone marrow
micronucleus test. Final report. Cincinnati, OH: Toxicology Excellence for Risk Assessment, Perchlorate
Study Group; study no. 6100-001.
Mannisto, P. T.; Ranta, T.; Leppaluoto, J. (1979) Effects of methylmercaptoimidazole (MMI), propylthiouracil
(PTU), potassium perchlorate (KC104) and potassium iodide (KI) on the serum concentrations of
thyrotrophin (TSH) and thyroid hormones in the rat. Acta Endocrinol. 91: 271-281.
Mantel, N.; Bryan, W. R. (1961) "Safety" testing of carcinogenic agents. J. Natl. Cancer Inst. 27: 455-470.
Marcus, A. H. (2001) Explanation of re-analyses of perchlorate hormone data from the 1998 external review draft
[letter to Annie Jarabek], Research Triangle Park, NC: U.S. Environmental Protection Agency, National
Center for Environmental Assessment; October 19.
Mareib, E. N. (1992) Human anatomy and physiology. 2nd ed. Redwood City, CA: Benjamin/Cummings Publishing
Company. (Benjamin/Cummings series in the life sciences).
Margolis, S. (1986) Health assessment, San Gabriel Valley, Los Angeles, California [memorandum to Donald W.
Hawkins concerning detection of perchlorate in residential wells]. Washington, DC: U.S. Department of
Health and Human Services, Agency for Toxic Substances and Disease Registry; January 21.
Mattie, D. R. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0039, hormone data from Brabant human
perchlorate (1.0 and 12.0 mg/kg-day) kinetics drinking water study [memorandum with attachments to
Annie Jarabek]. Wright-Patterson Air Force Base, OH: Air Force Research Laboratory; June 30.
Mayer, K. (1998) Perchlorate: occurrence data to replace Table 1 from Siddiqui et al. [letter with attachment to
Annie Jarabek]. San Francisco, CA: U.S. Environmental Protection Agency, Region 9; November 24.
Mayer, K. P. (2001) Transmittal of updated perchlorate maps and table [memorandum to Annie M. Jarabek].
San Francisco, CA: U.S. Environmental Protection Agency; November 30.
McClain, R. M. (1992) Thyroid gland neoplasia: non-genotoxic mechanisms. Toxicol. Lett. 64/65: 397-408.
McKnight, B.; Crowley, J. (1984) Tests for differences in tumor incidence based on animal carcinogenicity
experiments. J. Am. Stat. Assoc. 79: 639-648.
McMartin, D. N.; Sahota, P. S.; Gunson, D. E.; Hsu, H. H.; Spaet, R. H. (1992) Neoplasms and related proliferative
lesions in control Sprague-Dawley rats from carcinogenicity studies. Historical data and diagnostic
considerations. Toxicol. Pathol. 20: 212-225.
Meier, C.; Trittibach, P.; Kunz, M.; Guglielmetti, M.; Muller, B.; Staub, J. (2001) TSH is a poor test to assess the
severity of tissue hypothyroidism as compared to metabolic evaluation at the peripheral target organs. Basel,
Switzerland: University Hospital, Division of Endocrinology; clinical science poster P3: Thyroid II.
Merrill, E. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0036, human PBPK model for perchlorate inhibition
of iodide uptake in the thyroid [memorandum with attachments to Annie Jarabek]. Wright-Patterson Air
Force Base, OH: Air Force Research Laboratory; June 28.
Merrill, E. (2001a) Consultative letter, AFRL-HE-WP-CL-2001-0004, QA/QC audit report for the study of
perchlorate pharmacokinetics and inhibition of radioactive iodine uptake (RAIU) by the thyroid in humans
(CRC protocol #628) [memorandum with attachments to Annie M. Jarabek]. Wright-Patterson AFB, OH:
Air Force Research Laboratory; May 10.
Merrill, E. (2001b) Critique of Lawrence and Braverman data [personal communication to Annie M. Jarabek],
Wright-Patterson Air Force Base, OH: Air Force Research Laboratory, Toxicology Division; November 6.
Merrill, E. (2001c) Consultative letter, AFRL-HE-WP-CL-2001-0005, PBPK model for iodide kinetics and
perchlorate-induced inhibition in the male rat [memorandum with attachments to Annie M. Jarabek].
Wright-Patterson AFB, OH: Air Force Research Laboratory; May 8.
January 16, 2002
11-14
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Merrill, E. A. (200Id) Consultative letter, AFRL-HE-WP-CL-2001-0008, PBPK model for perchlorate-induced
inhibition of radioiodide uptake in humans [memorandum with attachments to Annie M. Jarabek],
Wright-Patterson AFB, OH: Air Force Research Laboratory; June 5.
Merrill, E. A. (200le) Consultative letter, AFRL-HE-WP-CL-2002-0003, revision to
AFRL-HE-WP-CL-2001-0010, comparison of internal dosimetrics using PBPK models for perchlorate
induced inhibition of thyroid iodide uptake and sensitivity analysis for male rat model [memorandum with
attachments to Annie M. Jarabek]. Wright-Patterson Air Force Base, OH: Air Force Research Laboratory;
December 20.
Meyer, G. D., Maj. (1998) Consultative letter, AFRL-HE-WP-CL-1998-0035, pharmacokinetic data for iodide
uptake inhibition in the thyroid by perchlorate [memorandum to Annie Jarabek]. Wright-Patterson Air
Force Base, OH: Air Force Research Laboratory, Human Effectiveness Directorate; December 23.
Moore, M. M. (1999) Analysis of perchlorate [memorandum to Vicki Dellarco]. Research Triangle Park, NC: U.S.
Environmental Protection Agency, National Health and Environmental Effects Research Laboratory;
January 29.
Morgans, M. E.; Trotter, W. R. (1960) Potassium perchlorate in thyrotoxicosis [letter]. Br. Med. J.
(October 8): 1086-1087.
Morreale de Escobar, G.; Obregon, M. J.; Escobar del Ray, F. (2000) Is neuropsychological development related to
maternal hypothyroidism or to maternal hypothyroxinemia? J. Clin. Endocrinol. Metab. 85: 3975-3987.
Murrell, J. A.; Portier, C. J.; Morris, R. W. (1998) Characterizing dose-response I: critical assessment of the
benchmark dose concept. Risk Anal. 18: 13-26.
Naismith, D. J.; Richardson, D. P.; Pritchard, A. E. (1982) The utilization of protein and energy during lactation in
the rat, with particular regard to the use of fat accumulated during pregnancy. Br. J. Nutr. 48: 433-441.
Narayanan, L. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0034, thyroid hormone and TSH co-laboratory
study report [memorandum with attachments to Annie Jarabek]. Wright-Patterson Air Force Base, OH:
Air Force Research Laboratory; June 15.
National Academy of Sciences, Institute of Medicine. (2001) Dietary reference intakes for vitamin A, vitamin K,
arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and
zinc. Washington, DC: National Academy Press. Available:
http://books.nap.edu/books/0309072794/htmiyindex.html [11 December, 2001].
National Research Council. (1983) Risk assessment in the federal government: managing the process. Washington,
DC: National Academy Press.
Noonan, N. (1999) Interim assessment guidance for perchlorate [memorandum with attachment to Regional
Administrators, Regional Waste Management Division Directors, and Regional Water Management
Division Directors], Washington, DC: U.S. Environmental Protection Agency, Office of Research and
Development; June 18.
Norberg-King, T. J. (1990) Seven-day tests and chronic tests [author's reply]. Environ. Toxicol. Chem.
9: 1435-1436.
Nzengung, V. A. (n.d.) Laboratory characterization of phyto-transformation products of perchloroethylene (PCE),
trichloroethylene (TCE) and perchlorate. Final report. Athens, GA: University of Georgia, Department of
Geology.
Nzengung, V. A.; Wang, C. (2000) Influences on phytoremediation of perchlorate-contaminated water. In:
Urbansky, E. T., ed. Perchlorate in the environment. New York, NY: Kluwer Academic Publishers.
(Environmental science research: v. 57).
Nzengung, V. A.; Wang, C.; Harvey, G. (1999) Plant-mediated transformation of perchlorate into chloride.
Environ. Sci. Technol. 33: 1470-1478.
O'Flaherty, E. J.; Scott, W.; Schreiner, C.; Beliles, R. P. (1992) A physiologically based kinetic model of rat and
mouse gestation: disposition of a weak acid. Toxicol. Appl. Pharmacol. 112: 245-256.
Okeke, B. C.; Giblin, T.; Frankenberger, W. F. (2001) Reduction of perchlorate and nitrate by salt tolerant bacteria.
Environ. Pollut.: in press.
Pajer, Z.; Kalisnik, M. (1991) The effect of sodium perchlorate and ionizing irradiation on the thyroid parenchymal
and pituitary thyrotropic cells. Oncology 48: 317-320.
Park, R. M. (2001) Perchlorate health effects in the epidemiological literature [with cover letter]. Cincinnati, OH:
National Institute for Occupational Safety and Health, Education and Information Division - Risk
Evaluation Branch.
January 16, 2002
11-15
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Parsons Engineering Science, Inc. (2001) Scientific and technical report for perchlorate biotransport investigation:
a study of perchlorate occurrence in selected ecosystems. Interim final. Austin, TX; contract no.
F41624-95-D-9018.
Pena, H. G.; Kessler, W. V.; Christian, J. E.; Cline, T. R.; Plumlee, M. P. (1976) A comparative study of iodine and
potassium perchlorate metabolism in the laying hen. 2. Uptake, distribution, and excretion of potassium
perchlorate. Poult. Sci. 55: 188-201.
Perlman, I.; Chaikoff, I. L.; Morton, M. E. (1941) Radioactive iodine as an indicator of the metabolism of iodine. 1.
The turnover of iodine in the tissues of the normal animal, with particular reference to the thyroid. J. Biol.
Chem. 149: 433-447.
Phillips, D. I. W.; Barker, D. J. P.; Fall, C. H. D.; Seckl, J. R.; Whorwood, C. B.; Wood, P. J.; Walker, B. R. (1998)
Elevated plasma Cortisol concentrations: a link between low birth weight and the insulin resistance
syndrome? J. Clin. Endocrinol. Metab. 83: 757-760.
Pisarev, M. A.; KJeinman de Pisarev, D. L. (1980) Biochemistry of thyroid regulation under normal and abnormal
conditions. J. Endocrinol. Invest. 3: 317-329.
Pop, V. J.; De Vries, E.; Van Baar, A. L.; Waelkcns, J. J.; De Rooy, H. A.; Horsten, M.; Donkers, M. M.; Komproe,
I. H.; Van Son, M. M.; Vader, H. L. (1995) Maternal thyroid peroxidase antibodies during pregnancy:
a marker of impaired child development? J. Clin. Endocrinol. Metab. 80: 3561-3566.
Pop, V. J.; Kuijpens, J. L.; Van Baar, A. L.; Verkerk, G.; Van Son, M. M.; De Vijlder, J. J.; Vulsma, T.; Wiersinga,
W. M.; Drexhage, H. A.; Vader, H. L. (1999) Low maternal free thyroxine concentrations during early
pregnancy are associated with impaired psychomotor development in infancy. Clin. Endocrinol.
50: 147-148.
Porterfield, S. P. (1994) Vulnerability of the developing brain to thyroid abnormalities: environmental insults to the
thyroid system. Environ. Health Perspect. Suppl. 102(2): 125-130.
Porterfield, S. P. (2000) Thyroidal dysfunction and environmental chemicals - potential impact on brain
development. Environ. Health Perspect. 108(suppl. 3): 433-438.
Portier, C.; Hedges, J. C.; Hoel, D. G. (1986) Age-specific models of mortality and tumor onset for historical
control animals in the National Toxicology Program's carcinogenicity experiments. Cancer Res. 46:
4372-4378.
Postel, S. (1957) Placental transfer of perchlorate and triiodothyronine in the guinea pig. Endocrinology 60: 53-66.
Potter, G. D.; Tong, W.; Chaikoff, I. L. (1959) The metabolism of 1131 -labeled iodine, thyroxine, and
triiodothyronine in the the mammary gland of the lactating rat. J. Biol. Chem. 243: 125-130.
Prins, G. S.; Birch, L.; Couse, J. F.; Choi, 1.; Katzenellenbogen, B.; Korach, K. S. (2001) Estrogen imprinting of the
developing prostate gland is mediated through stromal estrogen receptor a: studies with aERKO and
(JERKO mice. Cancer Res. 61: 6089-6097.
Rail, J. E.; Power, M. H.; Albert, A. (1950) Distribution of radioiodine in erythrocytes and plasma of man.
Proc. Soc. Exp. Biol. Med. 74: 460-461.
Renner, R. (1999) Study finding perchlorate in fertilizer rattles industry. Environ. Sci. Technol. 33: 394A-395A.
Research Triangle Institute. (1999) Perchlorate peer review workshop report. Washington, DC: U.S. Environmental
Protection Agency, Office of Solid Waste and Emergency Response; contract no. 68-W98-085.
Rice, D.; Barone, S., Jr. (2000) Critical periods of vulnerability for the developing nervous system: evidence from
humans and animal models. Environ. Health Perspect. 108(suppl. 3): 511-533.
Rice, D. C. (2000) Parallels between attention deficit hyperactivity disorder and behavioral deficits induced by
neurotoxic exposure in monkeys. Environ. Health Perspect. 108(suppl. 3): 405-408.
Robarge, W. P.; Duffera, M.; Ramirez, G. (2000) Evaluation of the composition of fertilizer source materials:
perchlorate. Paper presented before the Division of Fertilizer and Soil Chemistry at the American Chemical
Society 220th national meeting; August; Washington, DC.
Rockette, H. E.; Arena, V. C. (1983) Mortality patterns of workers in the Niagra plant. Niagra Falls, NY: Occidental
Chemical Corporation; microfiche no. OTS0537001.
Rogers, D. E., Lt. Col. (1998) Perchlorate user and production information [memorandum with attachments to
Annie M. Jarabek]. Wright-Patterson Air Force Base, OH: Department of the Air Force, Air Force Materiel
Command Law Office; October 30.
Ruppert, P. H.; Dean, K. F.; Reiter, L. W. (1985a) Development of locomotor activity of rat pups in figure-eight
mazes. Dev. Psychobiol. 18: 247-260.
Ruppert, P. H.; Dean, K. F.; Reiter, L. W. (1985b) Development of locomotor activity of rat pups exposed to heavy
metals. Toxicol. Appl. Pharmacol. 78: 69-77.
January 16, 2002
11-16
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Said, S.; Schlumberger, M.; Suarez, H. G. (1994) Oncogenes and anti-oncogenes in human epithelial thyroid
tumors. J. Endocrinol. Invest. 17: 371-379.
Sample, B. E.; Suter, G. W., II. (1994) Estimating exposure of terrestrial wildlife to contaminants. Oak Ridge, TN:
Oak Ridge National Laboratories; ES/ER/TM-125.
Sampson, D. A.; Jansen, G. R. (1984) Measurement of milk yield in the lactating rat from pup weight and weight
gain. J. Pediatr. Gastroenterol. Nutr. 3: 613-617.
Samuel, M.; Caputa, A. (1965) The role of the mother in 1131 metabolism of suckling and weaning rats. Can. J.
Physiol. Pharmacol. 43: 431-436.
Saxena, K. M.; Chapman, E. M.; Pryles, C. V. (1962) Minimal dosage of iodide required to suppress uptake of
iodine-131 by normal thyroid. Science (Washington, DC) 138: 430-431.
Scheuplein, R. J.; Bronaugh, R. L. (1983) Percutaneous absorption. In: Goldsmith, L. A., ed. Biochemistry and
physiology of the skin. Volume 2. New York, NY: Oxford University Press; p. 1279.
Schilt, A. A. (1979a) Perchloric acid and perchlorates. Columbus, OH: G. Frederick Smith Chemical Company;
p. 35.
Schilt, A. A. (1979b) Introduction. In: Perchloric acid and perchlorates. Columbus, OH: G. Frederick Smith
Chemical Company; pp. 1-8.
Schneider, B. F.; Golden, W. L. (1986) Acquisition of acoustic startle shows a dose-response to serum free T„.
Int. J. Dev. Neurosci. 4: 397-400.
Schroder-van der Elst, J. P.; Van der Heide, D.; Kastelijn, J.; Rousset, B.; Obregon, M.J. (2001) The expression of
the sodium/iodide symporter is up-regulated in the thyroid of fetuses of iodine-deficient rats. Endocrinology
142: 3736-3741.
Schulte, P. A. (1989) A conceptual framework for the validation and use of biologic markers. Environ. Res.
48: 129-144.
Schwartz, J. (2001) Gestational exposure to perchlorate is associated with measures of decreased thyroid function in
a population of California neonates [thesis]. Berkeley, CA: University of California.
Seckl, J. R. (1997) Glucocorticoids, feto-placental 11 P-hydroxysteroid dehydrogenase type 2, and the early life
origins of adult disease. Steroids 62: 89-94.
Selivanova, L. N.; Boltromeyuk, L. P.; Arefeva, Z. S.; Vavilova, L. N. (1986) Dynamics of the absorption and
elimination of perchloric acid salts in laboratory animals and livestock. Khim. Sel'sk. Khoz. (1963-1987)
(5): 43-45.
Shennan, D. B. (2001) Iodide transport in lactating rat mammary tissue via a pathway independent from the Na+/I-
cotransporter: evidence for sulfate/iodide exchange. Biochem. Biophys. Res. Commun. 280: 1359-1363.
Sher, E. S.; Xu, X. M.; Adams, P. M.; Craft, C. M.; Stem, S. A. (1998) The effects of thyroid hormone level and
action in developing brain: are these targets for the actions of polychlorinated biphenyls and dioxins?
Toxicol. Ind. Health 14: 121-158.
Shigan, S. A. (1963) Substantiation of the maxximum permissable concentration of ammonium perchlorate in water
of reservoirs. Gig. Sanit. 28: 8-14.
Siddiqui, M.; LeChevallier, M. W.; Ban, J.; Phillips, T.; Pivinski, J. (1998) Occurrence of perchlorate and methyl
tertiary butyl ether (MTBE) in groundwater of the American water system. Vorhees, NJ: American Water
Works Service Company, Inc.; September 30.
Siglin, J. C.; Mattie, D. R.; Dodd, D. E.; Hildebrandt, P. K.; Baker, W. H. (2000) A 90-day drinking water toxicity
study in rats of the environmental contaminant ammonium perchlorate. Toxicol. Sci. 57: 61-74.
Smialowicz, R. J. (1999) Review of 90-day ammonium perchlorate exposure on the antibody response to SRBC in
mice [memorandum to Annie M. Jarabek]. Research Triangle Park, NC: U.S. Environmental Protection
Agency, National Health and Environmental Effects Research Laboratory; January 28.
Smit, B. J.; Kok, J. H.; Vulsma, T.; Briet, J. M.; Boer, K.; Wiersinga, W. M. (2000) Neurologic development of the
newborn and young child in relation to maternal thyroid function. Acta Paediatr. 89: 291-295.
Smith, P. N.; Theodorakis, C. W.; Anderson, T. A.; Kendall, R. J. (2001) Preliminary asseessment of perchlorate in
ecological receptors at the Longhom Army Ammunition Plant (LHAAP), Karnack, Texas. Ecotoxicology
10: 305-313.
Snipes, M. B.; James, A. C.; Jarabek, A. M. (1997) The 1994 ICRP66 human respiratory tract dosimetry model as a
tool for predicting lung burdens from exposures to environmental aerosols. Appl. Occup. Environ. Hyg.
12: 547-554.
Soldin, O. P.; Braverman, L. E.; Lamm, S. H. (2001) Perchlorate clinical pharmacology and human health: a review.
Ther. Drug Monit. 23: 316-331.
January 16, 2002
11-17
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Southwell, N.; Randall, K. (1960) Potassium perchlorate in thyrotoxicosis [letter]. Lancet (7125): 653-654.
Spitzweg, C.; Hcufelder, A. E.; Moms, J. C. (2000) Thyroid iodine transport. Thyroid 10: 321-330.
Springbom Laboratories, Inc. (1998) A 90-day drinking water toxicity study in rats with ammonium perchlorate:
amended final report [amended study completion date: June 3], Spencerville, OH: Springbom Laboratories,
Inc.; study no. 3455.1.
Stanbury, J. B.; Wyngaardcn, J. B. (1952) Effect of perchlorate on the human thyroid gland. Metab. Clin. Exp.
1: 533-539.
Stralka, D. (1992) Toxicity assessment request on perchlorate salts [memorandum with attachment to Mindy
Henson, Superfund Technical Support Center], San Francisco, CA: U.S. Environmental Protection Agency;
October 9.
Suffolk County Department of Health Services. (2001a) Investigation summary: perchlorate contamination in
Yaphank, Suffolk County, NY. Hauppage, NY: Division of Environmental Quality.
Suffolk County Department of Health Services. (2001b) Perchlorate monitoring of water supplies in Suffolk
County, NY. Summary. Hauppage, NY: Division of Environmental Quality.
Sunar, O. (1963) Case report—agranulocytosis associated with potassium perchlorate treatment. J. Laryngol. Otol.
77: 353-355.
Susarla, S.; Bacchus, S. T.; Wolfe, N. L.; McCutcheon, S. C. (1999a) Phytotransformation of perchlorate and
identification of metabolic products in myriophyllum aquaticum. Int. J. Phytoremed. 1: 97-107.
Susarla, S.; Collette, T. W.; Garrison, A. W.; Wolfe, N. L.; McCutcheon, S. C. (1999b) Perchlorate identification in
fertilizers. Environ. Sci. Technol. 33: 3469-3472.
Susarla, S.; Bacchus, S. T.; Wolfe, N. L.; McCutcheon, S. C. (1999c) Phytotransformation of perchlorate using
parrot-feather. Soil and Groundwater Cleanup. February/March: 20.
Susarla, S.; Bacchus, S. T.; Harvey, G.; McCutcheon, S. C. (2000a) Phytotransformations of perchlorate
contaminated waters. Environ. Technol. 21: 1055-1065.
Susarla, S.; Collette, T. W.; Garrison, A. W.; Wolfe, N. L.; McCutcheon, S. C. (2000b) Additions and corrections to
"Perchlorate identification in fertilizers" (Environmental Science and Technology 33: 3469-3472). Environ.
Sci. Technol. 34: 224.
Suter, G. W., II. (1990) Seven-day tests and chronic tests [letter]. Environ. Toxicol. Chem. 9: 1435.
Suter, G. W., II; Tsao, C. L. (1996) Toxicological benchmarks for screening potential contaminants of concern for
effects on aquatic biota: 1996 revision. Oak Ridge, TN: Oak Ridge National Laboratory; ES/ER/TM-96/R2.
Available: www.hsrd.ornl.gov/ecorisk/reports.html [1999, November 24].
Sztanyik, L. B.; Turai, I. (1988) Modification of radioiodine incorporation in the fetus and newborn rats by the
thyroid blocking agents. Acta Phys. Hung. 72: 343-354.
Tabachnick, B. G.; Fidell, L. S. (2001) Profile analysis: the multivariate approach to repeated measures. In:
Using multivariate statistics. 4th ed. Boston, MA: Allyn and Bacon; pp. 391-455.
Takata, K.. (1985) Request for CDC assistance regarding potential health effects of perchlorate contamination at the
San Gabriel Valley superfund sites [memorandum with attachments to Don Hawkins]. San Francisco, CA:
U.S. Environmental Protection Agency, Superfund Records Center, Region 9; December 23.
Tamasy, V.; Meisami, E.; Vallerga, A.; Timiras, P. S. (1986) Rehabilitation from neonatal hypothyroidism:
spontaneous motor activity, exploratory behavior, avoidance learning and responses of pituitary-thyroid axis
to stress in male rats. Psychoneuroendocrinology 11: 91-103.
Tarone, R. E. (1982) The use of historical control information in testing for a trend in Poisson means. Biometrics
38: 457-462.
Tazebay, U. H.; Wapnir, I. L.; Levy, O.; Dohan, O.; Zuckier, L. S.; Zhao, Q. H.; Deng, H. F.; Amenta, P. S.;
Fineberg, S.; Pestell, R. G.; Carrasco, N. (2000) The mammary gland iodide transporter is expressed during
lactation and in breast cancer. Nat. Med. (NY) 6: 871-878.
Toxicology Excellence for Risk Assessment. (1997) Proposed perchlorate reference dose (RfD). Cincinnati, OH:
Toxicology Excellence for Risk Assessment; prepared for the Perchlorate Study Group.
Toxicology Excellence for Risk Assessment. (1998a) Notes from the March 1997 ITER peer review meeting.
Cincinnati, OH: Toxicology Excellence for Risk Assessment. Available: www.tera.org/news/eight.htm
[1999, November 24],
Toxicology Excellence for Risk Assessment. (1998b) Results of the perchlorate study protocol review meeting:
perchlorate study protocol peer review May 20, 1997 summary meeting notes. Cincinnati, OH: Toxicology
Excellence for Risk Assessment. Available: www.tera.org/news/eleven.htm [1999, November 24],
January 16, 2002
11-18
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Toxicology Excellence for Risk Assessment. (2001) Report on five expert reviews of the Primedica 2001 Study
Report (hormone, thyroid and neurohistological effects of oral (drinking water) exposure to ammonium
perchlorate in pregnant and lactating rats and in fetuses and nursing pups exposed to ammonium perchlorate
during gestation or via maternal milk, March 2001). Cincinnati, OH: Toxicology Excellence for Risk
Assessment (TERA); prepared for the Perchlorate Study Group; May.
TRC Environmental Corporation. (1998) Chemical fertilizer as a potential source of perchlorate. Burbank, CA:
Lockheed Martin Corporation; November.
Tsui, D. (1998) Oxidizing potential of perchlorate under physiological conditions [letter to Annie Jarabek].
Wright-Patterson Air Force Base, OH: Air Force Research Laboratory, Human Effectiveness Directorate,
Operational Toxicology Branch; July 17.
Tsui, D. T.; Mattie, D. R.; Narayanan, L. (1998) Stability and concentration verification of ammonium perchlorate
dosing solutions. Wright-Patterson Air Force Base, OH: Air Force Research Laboratory, Human
Effectiveness Directorate; AFRL-HE-WP-TR-1998-0068.
U.S. Code. (1996) Safe Drinking Water Act, as amended by PL 104-208, September 30. U. S. C. 42: § 300f et seq.
U.S. Environmental Protection Agency. (1987) The risk assessment guidelines of 1986. Washington, DC: Office of
Health and Environmental Assessment; report no. EPA/600/8-87/045. Available from: NTIS, Springfield,
VA;PB88-123997.
U.S. Environmental Protection Agency. (1988) Recommendations for and documentation of biological values for
use in risk assessment. Cincinnati, OH: Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office; EPA report no. EPA-600/6-87-008. Available from: NTIS, Springfield,
VA; PB88-179874.
U.S. Environmental Protection Agency. (1993a) Water quality guidance for the Great Lakes System and correction;
proposed rules. F. R. 58: 20,802-21,047.
U.S. Environmental Protection Agency. (1993b) Wildlife exposure factors handbook. Volume I. Washington, DC:
Office of Research and Development; report no. EPA/600/R-93/187a. Available:
http://cfpub.epa.gov/ncea/cfm/wefh.cfm [14 November, 2001].
U.S. Environmental Protection Agency. (1994) Methods for derivation of inhalation reference concentrations and
application of inhalation dosimetry [draft final]. Research Triangle Park, NC: Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office; report no. EPA/600/8-88/066F.
U.S. Environmental Protection Agency. (1995) Policy for risk characterization at the U.S. Environmental Protection
Agency [memorandum of Administrator, Carol M. Browner]. Washington, DC: Office of Research and
Development, Science Policy Council; March 21. Available: www.epa.gov/ORD/spc/rcpolicy.htm [1999,
November 24],
U.S. Environmental Protection Agency. (1996a) Guidelines for reproductive toxicity risk assessment. Washington,
DC: Office of Research and Development; report no. EPA/630/R-96/009. Available online at:
www.epa.gov/ORD/WebPubs/repro/.
U.S. Environmental Protection Agency. (1996b) Air quality criteria for particulate matter. Research Triangle Park,
NC: National Center for Environmental Assessment-RTP Office; report nos. EPA/600/P-95/001aF-cF. 3v.
Available from: NTIS, Springfield, VA; PB96-168224.
U.S. Environmental Protection Agency. (1998a) Assessment of thyroid follicular cell tumors. Washington, DC:
Office of Research and Development; report no. EPA/630/R-97/002. Available from: NTIS, Springfield,
VA; PB98-133119. Available online at: www.epa.gov/ncea/thyroid.htm.
U.S. Environmental Protection Agency. (1998b) Guidelines for neurotoxicity risk assessment. Washington, DC:
National Center for Environmental Assessment; report no. EPA/630/R-95/001Fa. Available online at:
http://www.epa.gov/ncea/nurotox.htm.
U.S. Environmental Protection Agency. (1998c) Guidelines for ecological risk assessment. Washington, DC: Risk
Assessment Forum; report no. EPA/630/R-95/002F. Available online at: www.epa.gov/ncea/ecorsk.htm.
U.S. Environmental Protection Agency. (1998d) Perchlorate environmental contamination: toxicological review and
risk characterization based on emerging information. Review draft. Washington, DC: Office of Research
and Development; report no. NCEA-1-0503.
U.S. Environmental Protection Agency. (1998e) Perchlorate: assessment of the state of the science. Report to the
Congress of the United States. Washington, DC: Office of Research and Development; September.
U.S. Environmental Protection Agency. (1998f) Health effects test guidelines: OPPTS 870.6200: neurotoxicity
screening battery. Washington, DC: Office of Prevention, Pesticides and Toxic Substances; report no. EPA
712-C-98-238.
January 16, 2002
11-19
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
U.S. Environmental Protection Agency. (1998g) Health effects test guidelines: OPPTS 870.6300: developmental
neurotoxicity study. Washington, DC: Office of Prevention, Pesticides and Toxic Substances; report no.
EPA 712-C-98-239.
U.S. Environmental Protection Agency. (1998h) 1998 update of ambient water quality criteria for ammonia.
Washington, DC: Office of Water; report no. EPA 822-R-98-008. Available online at:
www.epa.gov/ost/standards/amonsub.html.
U.S. Environmental Protection Agency. (1999) Potential species for phytoremediation of perchlorate. Athens, GA:
National Exposure Research Laboratory; report no. EPA/600/R-99/069.
U.S. Environmental Protection Agency. (2000) Methodology for deriving ambient water quality criteria for the
protection of human health (2000). Washington, DC: Office of Water, Office of Science and Technology;
report no. EPA-822-B-00-004. Available: http://www.epa.gov/ost/humanhealth/method/ [14 November,
2001],
U.S. Environmental Protection Agency. (2001a) Ion chromatographic determination of perchlorate: analysis of
fertilizers and related materials. Cincinnati, OH: Office of Research and Development; report no.
EPA/600/R-01/026.
U.S. Environmental Protection Agency. (2001b) Survey of fertilizers and related materials for perchlorate (CI04 ).
Final report. Cincinnati, OH: U.S. Office of Research and Development; report no. EPA/600/R-01/049.
Available: http://www.epa.gov/ORD/htm/ordpubs.htm [30 October, 2001],
U.S. Environmental Protection Agency. (2001c) Agency requests National Academy of Sciences input on
consideration of certain human toxicity studies; announces interim policy [press release with attached
memorandum]. Washington, DC: December 14.
Underwood, M. (1998) Exposure concerns and health implications of perchlorate. Presented at: The southeast
focused ground water conference: discussing the issue of MTBE and perchlorate in groundwater; June;
Anaheim, CA. Westerville, OH: National Ground Water Association.
Urbansky, E. T. (1998) Perchlorate chemistry: implications for analysis and remediation. Biorem. J. 2: 81-95.
Urbansky, E. T. (2000) Quantitation of perchlorate ion: practices and advances applied to the analysis of common
matrices. Crit. Rev. Anal. Chem. 30: 311-343.
Urbansky, E. T.; Schock, M. R. (1999) Issues associated with managing the risks of perchlorate in the drinking
water. J. Environ. Manage. 56: 79-95.
Urbansky, E. T.; Magnuson, M. L. (2000) Sensitivity and selectivity enhancement in perchlorate anion quantitation
using complexation-electrospray lonization-mass spectrometry. In: Urbansky, E. T., ed. Perchlorate in the
environment. New York, NY: KJuwer Academic Publishers. (Environmental science research: v. 57).
Urbansky, E. T.; Collette, T. W. (2001) Comparison and evaluation of laboratory performance on a method for the
determination of perchlorate in fertilizers. J. Environ. Monit. 3: 454-462.
Urbansky, E.T.; Magnuson, M. L.; Freeman, D.; Jelks, C. (1999) Quantitation of perchlorate ion by electrospray
ionization mass spectrometry (ESI-MS) using stable association complexes with organic cations and bases
to enhance selectivity. J. Anal. At. Spectrom. 14: 1861-1866.
Urbansky, E. T.; Magnuson, M. L.; Kelty, C. A.; Gu, B.; Brown, G. M. (2000a) Comment on "perchlorate
identification in fertilizers" and the subsequent addition/correction. Environ. Sci. Technol. 34: 4452-4454.
Urbansky, E. T.; Magnuson, M. L.; Gu, B.; Brown, G. M.; Kelty, C. A. (2000b) Development of an electrospray
mass spectrometric method for determining perchlorate in fertilizers. Paper presented before the Division of
Fertilizer and Soil Chemistry at the American Chemical Society 220th national meeting; August;
Washington, DC.
Urbansky, E. T.; Gu, B.; Magnuson, M. L.; Brown, G. M.; Kelty, C. A. (2000c) Survey of bottled waters for
perchlorate by electrospray ionization mass spectrometry (ESI-MS) and ion chromatography (IC). J. Sci.
Food Agric. 80: 1798-1804.
Urbansky, E. T.; Magnuson, M. L.; Kelty, C. A.; Brown, S. K. (2000d) Perchlorate uptake by salt cedar (Tamarix
ramosissima) in the Las Vegas Wash riparian ecosystem. Sci. Total Environ. 256: 227-232.
Urbansky, E. T.; Brown, S. K.; Magnuson, M. L.; Kelty, C. A. (2001) Perchlorate levels in samples of sodium
nitrate fertilizer derived from Chilean caliche. Environ. Pollut. 112: 299-302.
Versloot, P. M.; Schroder-Van Der Elst, J. P.; Van Der Heide, D.; Boogerd, L. (1997) Effects of marginal iodine
deficiency during pregnancy: iodide uptake by the maternal and fetal thyroid. Am. J. Physiol. 273:
El 121-El 126.
Von Burg, R. (1995) Toxicology update: perchlorates. J. Appl. Toxicol. 15: 237-241.
January 16, 2002
11-20
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Wall, J. R.; Fang, S. L.; Kuroki, T.; Ingbar, S. H.; Braverman, L. E. (1984) In vitro immunoreactivity to
propylthiouracil, methimazole, and carbimazole in patients with Graves' disease: a possible cause of
antithyroid drug-induced agranulocytosis. J. Clin. Endocrinol. Metab. 58: 868-872.
Warren, A. (1999) Interim pathology report from John R. Latanddresse, DVM, Ph.D.: histopathologic effects of
ammonium perchlorate in thyroid gland of mice [memorandum with attachment to Annie M. Jarabek].
Tallahassee, FL: Toxicology, Ecology, Research and Risk Assessment (TERA); January 13.
Weast, R. C., ed. (1989) CRC handbook of chemistry and physics. 69th ed. Boca Raton, FL: CRC Press, Inc.
Weetman, A. P.; Gunn, C.; Hall, R.; McGregor, A. (1984) Immunosuppression by perchlorate [letter]. Lancet
(8382): 906.
Weetman, A. P. (1997) Hypothyroidism: screening and subclinical disease. Br. Med. J. 314: 1175-1178.
Weiss, B. (2000) Vulnerability of children and the developing brain to neurotoxic hazards. Environ. Health
Perspect. 108(suppl. 3): 375-381.
Wenzel, K. W.; Lente, J. R. (1984) Similar effects of thionamide drugs and perchlorate on thyroid-stimulating
immunoglobulins in Graves' disease: evidence against an immunosuppressive action on thionamide drugs.
J. Clin. Endocrinol. Metab. 58: 62-69.
Whalan, J. E.; Redden, J. C. (1994) Interim policy for particle size and limit concentration issues in inhalation
toxicity studies. U.S. Environmental Protection Agency, Office of Pesticide Products; February 1.
Williams, T. L.; Martin, R. B.; Collette, T. W. (2001) Raman spectroscopic analysis of fertilizers and plant tissue
for perchlorate. Appl. Spectrosc. 55: 967-983.
Williams, V. (2000) Analytical results summary for Region 7 Perchlorate Occurrence Study through October 23,
2000 [memorandum to Kevin Mayer]. Kansas City, KS: U.S. Environmental Protection Agency, Superfund
Technical Assistance & Reuse Branch (STAR).
Wiltse, J.; Dellarco, V. L. (1996) U.S. Environmental Protection Agency guidelines for carcinogen risk assessment:
past and future. Mutat. Res. 365: 3-15.
Wing, S. S.; Fantus, I. G. (1987) Adverse immunologic effects of antithyroid drugs. Can. Med. Assoc. J.
136: 121-125, 127.
Winkler, P. C. (2001) Analysis of perchlorates by LC/MS/MS [memorandum to Kevin Mayer]. Golden, CO:
Acculabs, Inc.; October 26.
Wolf, D. C. (2000) Report of the peer review of the thyroid histopathology from rodents and rabbits exposed to
ammonium perchlorate in the drinking water [momorandum to Annie Jarabek]. Research Triangle Park,
NC: U.S. Environmental Protection Agency, National Health and Environmental Effects Research
Laboratory; May 5. Available: http://www.epa.gov/ncea/perch.htm [2001, 17 October].
Wolf, D. C. (2001) Erratum to the report of the peer review of the thyroid histopathology from rodents and rabbits
exposed to ammonium perchlorate in drinking water [memorandum to Annie Jarabek]. Research Triangle
Park, NC: U.S. Environmental Protection Agency, National Health and Environmental Effects Research
Laboratory; October 26.
Wolfe, N. L.; Ellington, J. J.; Garrison, A. W.; Evans, J. J.; Avants, J. K.; Teng, Q. (2000) Accumulation of
perchlorate in tobacco plants and tobacco products. Paper presented before the Division of Fertilizer and
Soil Chemistry at the American Chemical Society 220th national meeting; August; Washington, DC.
Wolff, J. (1989) Excess iodide inhibits the thyroid by multiple mechanisms. Adv. Exp. Med. Biol. 261: 211-244.
Wolff, J. (1998) Perchlorate and the thyroid gland. Pharmacol. Rev. 50: 89-105.
Wolff, J.; Maurey, J. R. (1961) Thyroidal iodide transport: II. Comparison with non-thyroid iodide-concentrating
tissues. Biochim. Biophys. Acta 47: 467-474.
Wolff, J.; Maurey, J. R. (1962) Thyroidal iodide transport. III. Comparison of iodide with anions of periodic group
VILA. Biochim. Biophys. Acta 57: 422-426.
Wolff, J.; Maurey, J. R. (1963) Thyroidal iodide transport: IV. The role of ion size. Biochim. Biophys. Acta 69:
48-58.
Wu, J.; Unz, R. F.; Zhang, H.; Logan, B. E. (2001) Persistence of perchlorate and the relative numbers of
perchlorate- and chlorate-respiring microorganisms in natural waters, soils, and wastewater. Biorem. J.
5: 119-130.
Wyngaarden, J. B.; Wright, B. M.; Ways, P. (1952) The effect of certain anions upon the accumulation and
retention of iodide by the thyroid gland. Endocrinology 50: 537-549.
Yamada, T. (1967) Effects of perchlorate and other anions on thyroxin metabolism in the rat. Endocrinology
81: 1285-1290.
January 16, 2002
11-21
DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Yamada, T.; Jones, A. E. (1968) Effect of thioeyanate, perehlorate and other anions on plasma protein-thyroid
hormone interaction in vitro. Endocrinology 82: 47-53.
Yates, C. (2001) Personal communication [with Kevin Mayer, U.S. EPA Region 9], U.S. Navy perehlorate
analytical method under development with sub-ppb detection limit. Indian Head, MD: United States Navy;
September 6.
Yokoyama, N.; Nagayama, Y.; Kakezono, F.; Kiriyama, T.; Morita, S.; Ohtakara, S.; Okamoto, S.; Monmoto, I.;
Izumi, M.; Ishikawa, N.; et al. (1986) Determination of the volume of the thyroid gland by a high
rcsolutional ultrasonic scanner. J. Nucl. Med. 27: 1475-1479.
York, R. G. (1998a) Update on additional items for 1613-002 [letter to Darol E. Dodd]. Horsham, PA: Argus
Research Laboratories, Inc.; September 25.
York, R. G. (1998b) Protocol 1613-002 - A neurobehavioral developmental study of ammonium perehlorate
administered orally in drinking water to rats. Sponsor's study number: 7757A210-1096-25F [letter to Annie
Jarabck], Horsham, PA: Argus Research Laboratories, Inc.; October 2.
York, R. G. (1998c) [Letter to Darol E. Dodd concerning geneology of the F1 generation rats], Horsham, PA:
Argus Research Laboratories, Inc.; October 26.
York, R. G. (1998d) Study 1613-002 - A neurobehavioral developmental study of ammonium perehlorate
administered orally in drinking water to rats. Sponsor's study number: 7757A210-1096-25F [letter with
attachments to Annie Jarabek]. Horsham, PA: Argus Research Laboratories, Inc.; November 5.
York, R. G. (1998e) 1416-002 - Oral (drinking water) developmental toxicity study of ammonium perehlorate in
rabbits [letter with attachments to Annie Jarabek]. Horsham, PA: Argus Research Laboratories, Inc.;
October 7.
York, R. G. (2000) Protocol 1416-003 - oral (drinking water) developmental toxicity study of ammonium
perehlorate in rats [letter to Annie Jarabek], Horshan, PA: Primedica, Argus Division; November 21.
York, R. G.; Brown, W. R.; Girard, M. F.; Dollarhide, J. S. (2001a) Oral (drinking water) developmental toxicity
study of ammonium perehlorate in New Zealand white rabbits. Int. J. Toxicol. 20: 199-205.
York, R. G.; Brown, W. R.; Girard, M. F.; Dollarhide, J. S. (2001b) Two-generation reproduction study of
ammonium perehlorate in drinking water in rats evaluates thyroid toxicity. Int. J. Toxicol. 20: 183-197.
Yu, K. O. (2000) Consultative letter, AFRL-HE-WP-CL-2000-0038, tissue distribution and inhibition of iodide
uptake in the thyroid by perehlorate with corresponding hormonal changes in pregnant and lactating rats
(drinking water study) [memorandum with attachment to Annie Jarabek]. Wright-Patterson Air Force Base,
OH: Air Force Research Laboratory; June 28.
Yu, K.. O. (2001) Consultative letter, AFRL-HE-WP-CL-2002-0001, intravenous kinetics of radiolabeled iodide in
tissues of adult male Sprague Dawley rat dosed wth l25l" plus earner [memorandum with attachments to
Annie M. Jarabek]. Wright-Patterson Air Force Base, OH: Air Force Research Laboratory; December 21.
Yu, K.. O. (2002) Consultative letter, AFRL-HE-WP-CL-2002-0002, intravenous kinetics of radiolabeled iodide and
perehlorate in tissues of pregnant and lactating Sprague Dawley female rats dosed with perehlorate and/or
carrier free 125I- [memorandum with attachment to Annie M. Jarabek]. Wright-Patterson Air Force Base,
OH: Air Force Research Laboratory; January 7.
Yu, K. O.; Todd, P. N.; Young, S. M.; Mattie, D. R.; Fisher, J. W.; Narayanan, L.; Godfrey, R. J.; Sterner, T. R.;
Goodyear, C. (2000) Effects of perehlorate on thyroidal uptake of iodide with corresponding hormonal
changes. Wright-Patterson AFB, OH: Air Force Research Laboratory; report no.
AFRL-HE-WP-TR-2000-0076.
Zeghal, N.; Redjem, M.; Gondran, F.; Vigouroux, E. (1995) [Analysis of iodine compounds in young rat skin in the
period of suckling and in the adult. Effect of perehlorate]. Arch. Physiol. Biochem. 103: 502-511.
Zeiger, E. (1998a) Salmonella mutagenicity testing of ammonium perehlorate [memorandum with attachment to
Annie Jarabek], Research Triangle Park, NC: U.S. Department of Health and Human Services, National
Institute of Environmental Health Sciences; September 29.
Zeiger, E. (1998b) Ammonium perehlorate MN test results [memorandum to Annie Jarabek]. Research Triangle
Park, NC: U.S. Department of Health and Human Services, National Institute of Environmental Health
Sciences; December 23.
Zhao, J. (2001) [Letter with attached statistical analysis documents to Annie Jarabek], Cincinnati, OH: Toxicology
Excellence for Risk Assessment; March 30.
January 16, 2002
11-22
DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX A
Schematics of Study Designs for Neurodevelopmental,
Two-Generation Reproductive and Developmental Studies
January 16, 2002
A-l
DRAFT - DO NOT QUOTE OR CITE
-------
OJ
3
e
p
3
On
N>
O
O
K)
| Dosage Period
1
Acclimation
14 Days
Cohabitation 1 DG 0 to DG 31 or DL 10
Maximum 7 Days B
1
FO Female Rats
Arrive nt
Testing Facility
OGO
Dosage
Begins
DG 21
Firsl
Possible
Oehveiy
DG 25 DG 31 or DL 10 DL 10
Last End of Dosage Scheduled
Possible Sacrifice
Delivery (thyroid
evaluations)
PARTURITION
DP 5 Litters Culled to 8 Pups and
Blood Collection
DP 10 5 Dams and Lilters Selected
for Blood Collection
DG 31
Scheduled
Sacrifice
(rats that did
not deliver)
DL 22
Scheduled
Sacrifice
(weaning)
F1 GENERATION
(daily clinical observations and sexual
maturation, weekly body weights, feed
consumption, and blind observations)
>
I
K)
H
i
u
o
2
o
H
/O
c
o
H
ID
o
&
o
3
m
SUBSET 1
100 Male and Female Pups
(one male/female
pup/litter/dosage group)
DP 12 SCHEDULED SACRIFICE,
BRAIN WEIGHTS, AND
NEU ROH1STOLOGICAL
EXAMINATION
6 Male and 6 Female Pups/Dosage Level
(total of 30 male and 30 female pups)
Abbreviations
DG = Day of (Presumed) Gestation
DP = Day Postpartum
DL = Day of Lactation
SUBSET 2
100 Male and Female Pups
(one male/female
pufVlitter/dosage group)
DPs 23-25 AND 30-32 PASSIVE
AVOIDANCE TESTING
DPs 59-63 AND 66-70
WATER MAZE TESTING
DPs 90-92 SCHEDULED SACRIFICE
AND BLOOD COLLECTION
SUBSET 3
100 Male and 100 Female Pups
(one male/female
pup/lrtter/dosage group)
DPs 14, 18, 22, AND 59" MOTOR
ACTIVITY
DPs 23 AND 60 AUDITORY
STARTLE HABITUATION
DPs 67-69 SCHEDULED
SACRIFICE
SUBSET 4
100 Male and 100 Female Pups
(one male/female pup/litter/dosage
group)
DPs 80-86 SCHEDULED
SACRIFICE,
NEUROHISTOLOGICAL
EXAMINATION.
REGIONAL BRAIN WEIGHTS
6 Male and 6 Female/Dosage Group
(30 male and 30 female total)
Figure A-l. Schematic of the neurobehavioral developmental study of ammonium perchlorate administered
orally in drinking water to SD rats (Argus Research Laboratories, Inc., 1998a).
-------
First Day
Postpartum
Weaning c
F1
Postweaning
Cohabitation
Last Day
of Exposure [F1 ]'
Lactation
Generation ¦
Rats
10 weeks
I 3 weeks
3 weeks
Postpartum
F2 Generation I
3 weeks
Day 21
= Dosage Period
a = Male rats sacrificed after determination of sufficient number
of pregnancies
b = P1 generation female rats sacrificed
c = F1 generation dams and F2 generation litters sacrificed
Figure A-2. Schematic of the oral (drinking water), two-generation (one litter per
generation) reproduction study of ammonium perchlorate in SD rats (Argus
Research Laboratories, Inc., 1998b).
January 16, 2002
A-3
DRAFT - DO NOT QUOTE OR CITE
-------
Arrival of
Timed-Pregnant
Female Rabbits
Start of
Exposure
Day 6 of
Presumed
Gestation
End of
Exposure
Caesarean-
Sectioninga.b
Day 28 of Day 29 of
Presumed Presumed
Gestation Gestation
= Exposure Period
a = Blood samples taken from does for thyroid and pituitary hormone (T3, T4, TSH) analyses,
b = Fetal evaluations (external examinations and soft tissue and skeletal examinations).
Figure A-3. Schematic of the oral (drinking water) developmental toxicity study of
ammonium perchlorate in New Zealand rabbits (Argus Research
Laboratories, Inc., 1998c).
January 16, 2002
A-4
DRAFT - DO NOT QUOTE OR CITE
-------
APPENDIX B
List of Acronyms and Abbreviations
Acronym
Definition
AA°
<-Ii * rxn
Helmholtz free energy of reaction
AG°f
Gibbs free energy of formation
AG°
rxn
Gibbs free energy of reaction
AS°
umv
net entropy of the universe
a-p
anterior-posterior
Ab
antibody
ACSL
advanced continuous simulation language
ADHD
attention deficit hyperactivity disorder
ADME
absorption, distribution, metabolism, and elimination
AFB
air force base
AFRL
U.S. Air Force Research Laboratories
AFRL/HEST
Air Force Research Laboratory/Human Effectiveness Directorate
AIDS
acquired immunodeficiency syndrome
AITD
autoimmune thyroid disease
ANCOVA
analysis of covariance
AP
ammonium perchlorate
ATP
adenosine triphosphate
AUC
area-under-the-curve
AV
acute value
AWQC
ambient water quality criteria
bf4-
tetrafluoroborate
BMD
benchmark dose
BMDL
benchmark dose lower confidence limit
BMR
benchmark response
January 16, 2002
B-l
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym Definition
BW body weight
C' complement
CA DHS California Department of Health Services
cAMP cyclic adenosine monophosphate
CBC complete blood count
CCL Contaminant Candidate List
CD4/CD8 cluster of differentiation — cellular markers 4 and 8
CDC Centers for Disease Control and Prevention
CERCLA Comprehensive Environmental Response Compensation Liability
Act
cESI-MS complexation electrospray ionization mass spectrometry
CFU colony-forming units
CHS contact hypersensitivity
ChV chronic value
Cl2 chlorine
CI confidence interval
CIO" hypochlorite
C104" perchlorate
ClUC-p perchlorate urinary clearance
CNS central nervous system
CP cyclophosphamide
CPES Coastal Plain Experiment Station
CPM counts per minute
Cs+ cesium
CsCl cesium chloride
CTL cytotoxic T-lymphocyte
CV coefficient of variation
DAF dosimetric adjustment factor
January 16, 2002
B-2
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym Definition
DEQ Department of Environmental Quality
DIT diiodotyrosine
DNA deoxyribonucleic acid
DNCB dinitrochlorobenzene
DoD Department of Defense
DoE Department of Energy
DTH delayed-type hypersensitivity
DWEL drinking water equivalent level
E:T effector to target cell
EAR estimated average requirement
EGF epidermal growth factor
ELISA enzyme linked immunosorbant assay
ELS early-life stage
EPA U.S. Environmental Protection Agency
EPL Experimental Pathology Laboratories, Inc.
ER endoplasmic reticulum
E° standard electric potential
F Faraday constant
F1 first generation
F2 second generation
FAVF Final acute value factor
FCN function
FETAX Frog Embryo Teratogenesis Assay: Xenopus
FGF fibroblast growth factor
FH follicular epithelial cell hypertrophy or hyperplasia
FIFRA Federal Insecticide, Fungicide, and Rodenticide Act
fT4 free thyroxine
January 16, 2002
B-3
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym
Definition
GA
Golgi apparatus
GD
gestation day
GGTP
g-glutamyl transpeptidase
GI
gastrointestinal
GMAV
Genus mean acute value
gsp
GTP-binding protein mutation
Gy
Gray (equal to 100 rads)
H+
hydrogen
h2o2
hydrogen peroxide
hCG
human chorionic gonadotropin
HC104
perchloric acid
HEE
human equivalent exposure
HOC1
hypochlorous
I"
iodide
IC
ion chromatography
IC25
quartile inhibitory concentration
ICD-9
International Classification of Diseases, 9th Revision
ID
iodine deficiency
IFN
interferon
IGF-1
insulin-like growth factor
IgG
immunoglobulin G
IgM
immunoglobulin M
ip
intraperitoneally
IPSC
Interagency Perchlorate Steering Committee
IRIS
Integrated Risk Information System
IU
international unit
IUDR
uridine
January 16, 2002
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym
Definition
iv
intravenously
ie
potassium
K20
potassium oxide
Km
Michaelis-Menten affinity constant
KN03
potassium nitrate
LC50
concentration lethal to 50% of population
LD
lactation day
LHAAP
Longhorn Army Ammunition Plant
Li+
lithium
LLNA
local lymph node assay
In
natural log
LOAEL
lowest-observed-adverse-effect level
LOEC
lowest-observed-effect concentration
LOEL
lowest-observed effect level
LP
lymphoprol iferation
LS
Lumen size
LY
lysosomes
M-W RST
Mann-Whitney Rank Sum Test
MCA
3-methyl cholanthrene
MCL
maximum contaminant level
MDL
minimum detection limit
MF
modifying factor
Mg(C104)2
magnesium perchlorate
MIT
monoiodotyrosine
MMIA
1 -methyl-2-mercaptoimidazole
MANOVA
multiple analysis of variance
MCLG
maximum contaminant level goal
January 16, 2002
B-5
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym Definition
MRL minimum reporting limit
mRNA messenger ribonucleic acid
MS-MS mass spec — mass spec
MTD maximum tolerated dose
n number of electrons or number of moles
n.d. no date
N-P-K ratio nitrogen-phosphorous-potassium ratio
Na+ sodium
NaC104 sodium perchlorate
NaN03 sodium nitrate
NAS National Academy of Sciences
NASA National Aeronautics and Space Administration
NCE Normochromatic erythrocyte
NCEA National Center for Environmental Assessment
NDEP Nevada Division of Environmental Protection
NERL-ERD Natural Exposure Research
Laboratory's Ecosystems Research Division
NH4+ ammonium
NH4C104 ammonium perchlorate
NH4N03 ammonium nitrate
NHEERL National Health and Environmental Effects Research Laboratory
NIEHS National Institute for Environmental Health Sciences
NIS sodium iodide symporter
NK natural killer
NMR nuclear magnetic resinance
N03" nitrate
NOAEL No-Observed-Adverse-Effect Level
NOEC No-Observed-Effect Concentration
January 16, 2002
B-6
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym Definition
NPDWR National Primary Drinking Water Regulations
NRMRL National Risk Management Research Laboratory
NTP National Toxicology Program
02 oxygen
OEHHA Office of Environmental Health Hazard Assessment
OEPP Office of Emergency Response and Remediation
OPPTS Office of Prevention, Pesticides, and Toxic Substances
ORD Office of Research and Development
OSWER Office of Solid Waste and Emergency Response
OW Office of Water
p probability
P pressure
PI parental generation
P205 disphosphorus pentoxide
p53 p53 tumor suppressor gene
PA prealbumin
PAS periodic acid shift
PBI protein-bound iodide
PBPK physiologically based pharmacokinetic
PCE polychromatic erythrocyte
PCB polychlorinated biphenyl
PFC plaque-forming cell
PHG public health goal
PII plasma inorganic iodide
PND post-natal day
PP post partum
PP-TH plasma protein-thyroid hormone
January 16, 2002
B-7
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym
Definition
PPb
parts per billion
PPm
parts per million
PQL
practical quantitation limit
PSG
Perchlorate Study Group — consortium of defense contractors
PT-p
thyroid follicle:stroma partition coefficient
PTU
propylthiouracil
PWG
Pathology Work Group
QA/QC
quality assurance/quality control
R
ideal gas constant
RAIU
radioactive iodine uptake
ras
ras protooncogene
Rb+
rubidium
RDA
recommended dietary allowance
RfC
inhalation reference concentration
RfD
oral reference dose
RIA
radioimmunoassay
RL
reproducibility limits
RO
reverse osmosis
rs
Spearman's rank order
RSC
relative source contribution
rT3
reverse triiodothyronine
SACR
secondary acute-chronic ratio
SAV
secondary acute value
sc
subcutaneously
SCN
thiocyanate
scv
secondary chronic value
SD
standard deviation
January 16, 2002
B-8
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym
Definition
SD rats
Spraque-Dawley strain
SDWA
Safe Drinking Water Act
SE
standard error of the mean
SGOT
serum glutamyl oxacetic transaminase
SGPT
serum glutamyl pyruvic transaminase
SLA
soluble Listeria antigen
SMCV
species mean chronic value
SNK
Student Newman Keuls
SRBC
sheep red blood cell
SRLB
Sanitation and Radiation Laboratory Branch
T
temperature
T2
diiodothyronine
T3
triiodothyronine
T4
thyroxine or tetraiodothyronine
T4 GLUC
T4-glucuronide conjugate
TBG
thyroid-binding globulin
TCE
trichloroethylene
TDS
total dissolved solids
Tg
thyroglobulin
TH
thyroid hormone
TPO
thyroid peroxidase
TRH
thyrotropin-releasing hormone
TSCA
Toxic Substances Control Act
TSH
thyroid-stimulating hormone
tT4
total thyroxine
UCMR
Unregulated Contaminant Monitoring Rule
UDPGTs
uridine diphosphyl glucuronosyl transferases
January 16, 2002
B-9
DRAFT - DO NOT QUOTE OR CITE
-------
Acronym Definition
UF uncertainty factor
USAF United States Air Force
USGS United States Geological Survey
USN United States Navy
V volume
Vmaxc Michaelis-Menten maximum velocity capacity
Wcxp expansion work
WHO World Health Organization
WPAFB Wright Patterson Air Force Base
WSWRD Water Supply and Water Resources Division
January 16, 2002
B-10
DRAFT - DO NOT QUOTE OR CITE
-------
Federal Register/Vol. 67, No. 1/Wednesday, January 2, 2002/Notices
75
SUMMARY: As required by the Federal
Advisory Committee Act, 5 U.S.C., App.
2 section 9(c), EPA's Office of Pesticide
Programs (OPP) is giving notice of the
renewal of the Pesticide Program
Dialogue Committee (PPDC) and its
Charter and the appointment of new
members.
DATES: The PPDC Charter, which was
filed with Congress on November 9,
2001, will be in effect for 2 years, until
November 9, 2003.
FOR FURTHER INFORMATION CONTACT:
Margie Fehrenbach (7501C), Office of
Pesticide Programs, Environmental
Protection Agency, 1200 Pennsylvania
Ave., NW., Washington, DC 20460;
telephone number: (703) 308-4775 or
(703) 305-7093; fax number: (703) 308-
4776; e-mail address:
Fehrenbach.Margie@epa.gov.
SUPPLEMENTARY INFORMATION:
I. General Information
A. Does this Action Apply to Me?
This action is directed to the public
in general. This action may, however, it
may be of interest to persons who are
concerned about implementation of the
Federal Insecticide, Fungicide, and
Rodenticide Act; the Federal Food,
Drug, and Cosmetic Act; and the
amendments to both of these major
pesticide laws by the Food Quality
Protection Act (Public Law 104-170) of
1996. Since other entities may also be
interested, the Agency has not
attempted to describe all the specific
entities that may be affected by this
action. If you have any questions
regarding the applicability of this action
to a particular entity, consult the person
listed under FOR FURTHER INFORMATION
CONTACT.
B. How Can I Get Additional
Information, Including Copies of this
Document and Other Related
Documents?
1. Electronically. You may obtain
electronic copies of this document, and
certain other related documents that
might be available electronically, from
the EPA Internet Home Page at http://
www.epa.gov/. To access this
document, on the Home Page select
"Laws and Regulations," "Regulations
and Proposed Rules," and then look up
the entry for this document under the
"Federal Register—Environmental
Documents." You can also go directly to
the Federal Register listings at http://
www.epa.gov/fedrgstr/. To access
information about PPDC, go directly to
the Home Page for EPA's Office of
Pesticide Programs at http://
www.epa.gov/pesticides/ppdc.
2. In person. The Agency has
established an administrative record for
this meeting under docket control
number OPP-00439M. The
administrative record consists of the
documents specifically referenced in
this notice, any public comments
received during an applicable comment
period, and other information related to
the Pesticide Program Dialogue
Committee (PPDC). This administrative
record includes the documents that are
physically located in the docket, as well
as the documents that are referenced in
those documents. The public version of
the administrative record, which
includes printed, paper versions of any
electronic comments that may be
submitted during an applicable
comment period, is available for
inspection in the Public Information
and Records Integrity Branch (PIRIB),
Rm. 119, Crystal Mall #2, 1921 Jefferson
Davis Hwy., Arlington, VA, from 8:30
a.m. to 4 p.m., Monday through Friday,
excluding legal holidays. The PIRIB
telephone number is (703) 305-5805.
C. How Can I Participate in PPDC
Meetings?
PPDC meetings and workshops will
be open to the public under section
10(a)(2) of the Federal Advisory
Committee Act, Public Law 92—463.
Outside statements by observers will be
welcome. Oral statements will be
limited to 3-5 minutes, and it is
preferred that only one person per
organization present the statement. Any
person who wishes to file a written
statement may do so before or after the
meeting. These statements will become
part of the permanent record and will be
available for public inspection at the
address in Unit n.2.
II. Background
The PPDC is composed of 42 members
appointed by the EPA Deputy
Administrator. Committee members
were selected from a balanced group of
participants from the following sectors:
Pesticide users, grower and commodity
groups; industry and trade associations;
environmental/public interest and
farmworker groups; Federal, State and
tribal governments; public health
organizations; animal welfare; and
academia. PPDC was established to
provide a public forum to discuss a
wide variety of pesticide regulatory
development and reform initiatives,
evolving public policy and program
implementation issues, and science
policy issues associated with evaluating
and reducing risks from use of
pesticides.
List of Subjects
Environmental protection,
Agriculture, Chemicals, Drinking water,
Foods, Pesticides, Pests.
Dated: December 21, 2001.
Marcia E. Mulkey,
Director, Office of Pesticide Programs.
[FR Doc. 01-32214 Filed 12-31-01; 8:45 am]
BILLING CODE 6560-S
ENVIRONMENTAL PROTECTION
AGENCY
[FRL-7124—1]
Peer Review of EPA Draft Human
Health and Ecological Risk
Assessment of Perchlorate
AGENCY: Environmental Protection
Agency.
ACTION: Notice of Peer Review
Workshop and public comment period.
SUMMARY: The U.S. Environmental
Protection Agency (EPA) Office of
Research and Development is
announcing an external peer review
workshop to review the revised draft
document entitled, "Perchlorate
Environmental Contamination:
Toxicological Review and Risk
Characterization" (NCEA-I-0503). The
EPA is also announcing a public
comment period for this draft
document. The workshop is being
organized and convened by the Eastern
Research Group, Inc. (ERG), an EPA
contractor.
DATES: The two-day peer review
workshop will begin on Tuesday, March
5, 2002, at 9 a.m. and will end on
Wednesday, March 6, 2002, at 4:30 p.m.
The 30-day public review and comment
period will begin January 9, 2002, and
will end February 11, 2002.
ADDRESSES: The external peer review
meeting will be held at a facility in
Sacramento, California. To attend the
meeting as an observer, please register
with ERG via the Internet by visiting
www.meetings@erg.com. You may also
register by calling ERG's conference
registration line at 781-674-7374 or by
faxing a registration request to 781-674-
2906. Upon registering, you will be sent
an agenda and a logistical fact sheet
containing information on the meeting
site, overnight accommodations, and
ground transportation. The deadline for
pre-registration is February 25, 2002.
Space is limited, and reservations will
be accepted on a first-come, first-served
basis. There will be a limited time for
oral comments on the revised draft
document during the meeting. When
registering, please let ERG know if you
-------
76
Federal Register/Vol. 67, No. 1/Wednesday, January 2, 2002/Notices
wish to make a brief statement not to
exceed five minutes.
Document Availability: The external
review draft of the perchlorate
document will be available by January
9, 2002, on EPA's National Center for
Environmental Assessment (NCEA) Web
site at http://www.epa.gov/ncea. In
addition, a compact disk (CD)
containing documents cited in the
"Perchlorate Environmental
Contamination: Toxicological Review
and Risk Characterization" report that
cannot be readily obtained from the
open literature will be available by
request as of January 9, 2002. To obtain
a copy of the CD, you may contact the
EPA Superfund Records Center in San
Francisco, California. A shipping and
handling fee may apply. The circulation
desk phone number for the Superfund
Records Center is 415-536-2000. Copies
of the perchlorate document and CD are
not available from ERG.
Comment Submission: Written
comments should be submitted to ERG,
Inc., 110 Hartwell Avenue, Lexington,
Massachusetts 02421. Comments under
50 pages may be sent via e-mail
attachment (in Word, Word Perfect, or
PDF) to www.meetings@erg.com.
Written comments must be postmarked
by the end of the public comment
period (February 11, 2002). Please note
that all technical comments received in
response to this notice will be placed in
a public record. For that reason,
commentors should not submit personal
information (such as medical data or
home address), Confidential Business
Information, or information protected by
copyright. Due to limited resources,
acknowledgments will not be sent.
FOR FURTHER INFORMATION CONTACT:
Questions regarding registration and
logistics should be directed to EPA's
contractor, ERG, Inc., at 781-674-7374.
For technical inquiries, please contact:
Annie Jarabek, U.S. Environmental
Protection Agency (MD 52), USEPA
Mailroom, Research Triangle Park, NC
27711; telephone 919-541-4847;
facsimile 919-541-1818; e-mail
jarabek.annie@epa.gov.
SUPPLEMENTARY INFORMATION:
Perchlorate (C104) is an anion that
originates as a contaminant in
groundwater and surface waters from
the dissolution of ammonium,
potassium, magnesium, or sodium salts.
Perchlorate is exceedingly mobile in
aqueous systems and can persist for
many decades under typical
groundwater and surface water
conditions. A major source of
perchlorate contamination is the
manufacture of ammonium perchlorate
for use as the oxidizer component and
primary ingredient in solid propellant
for rockets, missiles, and fireworks.
EPA's Superfund Technical Support
Center issued a provisional reference
dose (RfD) for perchlorate in 1992 and
a revised provisional RfD in 1995 based
on the effects of potassium perchlorate
in patients with Graves' disease (an
autoimmune disease that results in
hyperthyroidism). (An RfD is an
estimate of a daily oral human exposure
that is anticipated to be without adverse
noncancer health effects over a lifetime.)
In March 1997, the existing toxicologic
database on perchlorate was determined
to be inadequate for quantitative human
health risk assessment by an external
peer review panel. A lack of data on the
ecotoxicological effects was also noted.
In May 1997, a testing strategy was
developed based on the known mode-of-
action for perchlorate toxicity (the
inhibition of iodide uptake in the
thyroid and subsequent perturbations of
thyroid hormone homeostasis), and an
accelerated research program was
initiated to gain a better understanding
of the human health effects of
perchlorate, examine possible ecological
impacts, refine analytical methods,
develop treatment technologies, and
better characterize the occurrence of
perchlorate in groundwater and surface
waters.
In December 1998, the National
Center for Environmental Assessment
(NCEA) developed an external peer
review draft document that assessed the
human health and ecological risk of
perchlorate ("Perchlorate
Environmental Contamination:
Toxicology Review and Risk
Characterization Based on Emerging
Information," NCEA-I-0503). This
document presented an updated human
health risk assessment that incorporated
results of the newly performed health
effects studies available as of November
1998 and a screening-level ecological
assessment. The human health risk
assessment model utilized a mode-of-
action approach that harmonized
noncancer and cancer approaches to
derive a single oral risk benchmark
based on precursor effects for both
neurodevelopmental and thyroid
neoplasia. A workshop was convened in
February 1999 in San Bernardino,
California, to provide external peer
review of that document. Peer reviewers
endorsed the conceptual approach
proposed by NCEA, but recommended
that new analyses be conducted and that
several additional studies be planned
and performed. NCEA has prepared a
revised perchlorate assessment that
addresses comments from the 1999
external peer review workshop and
incorporates data from additional
studies that were either nearing
completion at the time of the 1999
review or were recommended at that
time. This revised draft document is the
subject of the external peer review
workshop announced in today's Federal
Register notice.
The external peer review panel will
consist of a panel of independent
scientists selected by EPA's contractor,
ERG, from the fields of developmental
toxicology, reproductive toxicology,
neurotoxicology, immunotoxicology,
pharmacokinetics, genetic toxicology,
endocrinology, pathology,
epidemiology, statistics, ecotoxicology,
and environmental transport and
biotransformation. Peer reviewers will
review the revised human health and
ecological risk assessment for
perchlorate as well as new studies
performed since the 1999 external peer
review. Following the external peer
review workshop, ERG will prepare a
report summarizing the workshop. EPA
will address the comments of the
external peer reviewers in finalizing the
perchlorate risk assessment document
and in developing revised toxicity
values. The human health and
ecological risk assessment may be used
in the future to support development of
a health advisory or possible drinking
water regulations and cleanup decisions
at hazardous waste sites. However, any
such future decisions would be subject
to all applicable statutory and regulatory
requirements and policy considerations
for use of the assessments under those
programs.
Dated: December 20, 2001.
George W. Alapas,
Acting Director, National Center for
Environmental Assessment.
[FR Doc. 01-32088 Filed 12-31-01; 8:45 am]
BILLING CODE 6560-50-P
FEDERAL COMMUNICATIONS
COMMISSION
Notice of Public Information
Collection(s) being Reviewed by the
Federal Communications Commission
December 19, 2001.
SUMMARY: The Federal Communications
Commission, as part of its continuing
effort to reduce paperwork burden
invites the general public and other
Federal agencies to take this
opportunity to comment on the
following information collection(s), as
required by the Paperwork Reduction
Act of 1995, Public Law 104-13. An
agency may not conduct or sponsor a
collection of information unless it
displays a currently valid control
-------
Federal Register/Vol. 67, No. 9/Monday, January 14, 2002/Notices
1759
DEPARTMENT OF ENERGY
Federal Energy Regulatory
Commission
[Docket Nos. RT01-100-000, RT01-77-000,
and RT01-75-000]
Regional Transmission Organizations,
Southern Company Services, Inc.,
Entergy Services, Inc., et al.; Notice of
Meeting on SeTrans RTO
January 7, 2002.
The SeTrans Sponsors1 have invited
the Commission to participate in a
meeting that will be held on January 14,
2002 from 1 p.m. to 5 p.m. and January
15, 2002, from 8:30 a.m. to 4 p.m. The
meeting will be held at the Sheraton
Gateway Hotel Atlanta Airport, 1900
Sullivan Road, Atlanta, Georgia 30337.
The purpose of the meeting is to form
a Stakeholder Advisory Committee to
assist the development of the SeTrans
RTO. Representatives of the
Commission's staff will attend the
meeting. Members of the public may
attend. Further information about the
meeting and a copy of the registration
form is available at
www. setransgrid. com.
During the course of the meeting, it is
possible that discussions may overlap
with issues pending in the above-
captioned dockets. A summary of any
such discussion will be placed in each
of the listed dockets, if appropriate.
C.B. Spencer,
Acting Secretary.
[FR Doc. 02-839 Filed 1-11-02; 8:45 am]
BILLING CODE 6717-01-P
ENVIRONMENTAL PROTECTION
AGENCY
[FRL-7128-6]
Peer Review of EPA Draft Human
Health and Ecological Risk
Assessment of Perchlorate
AGENCY: Environmental Protection
Agency.
ACTION: Notice; correction to Notice of
Peer Review Workshop and public
comment period.
SUMMARY: On January 2, 2002, the U.S.
Environmental Protection Agency
(EPA), Office of Research and
Development (ORD) announced an
external peer review workshop to
1 "SeTrans Sponsors" consists of Georgia
Transmission Corporation, MEAG Power, Dalton
Utilities, Entergy Corporation, South Mississippi
Electric Power Association, City of Tallahassee,
Jacksonville Electric Authority, South Carolina
Public Service Authority, and Southern Companies
review the revised draft document
entitled, "Perchlorate Environmental
Contamination: Toxicological Review
and Risk Characterization" (NCEA-1-
0503) and a public comment period for
this draft document (67 FR 75). The peer
review workshop will take place on
March 5 and 6, 2002, in Sacramento,
California. The public comment period
is January 9, 2002, to February 11, 2002.
The deadline for registration is February
25, 2002. This notice corrects the
address for electronic registration and
electronic submission of comments
provided in the January 2 Federal
Register notice.
Correction to Addresses
To attend the meeting as an observer,
please register with the Eastern
Research Group (ERG), an EPA
contractor. Please note that the
registration Internet site provided in the
January 2, 2002, Federal Register notice
is incorrect. To register, send an e-mail
request to ERG at meetings@erg.com
(include name, affiliation, full address,
phone/fax number, and e-mail address)
or by calling the conference registration
line at 781-674-7374 between the hours
of 9 a.m. and 5 p.m. EST or via fax at
781-674—2906. You may also mail a
registration request to ERG, Attn:
Meetings, 110 Hartwell Avenue,
Lexington, MA 02421. Please indicate
when registering whether you plan to
make observer comments.
Correction for Comment Submission
Please note that the e-mail address
provided in the January 2, 2002, Federal
Register notice is incorrect. Written
comments should be submitted to ERG,
Attn: Meetings, 110 Hartwell Avenue,
Lexington, MA 02421. Comments under
50 pages may be sent via e-mail
attachment (in Word, Word Perfect, or
PDF) to meetings@erg.com.
Dated: January 4, 2002.
Arthur Payne,
Acting Director, National Center for
Environmental Assessment.
[FR Doc. 02-877 Filed 1-11-02; 8:45 am]
BILLING CODE 6560-50-P
FEDERAL COMMUNICATIONS
COMMISSION
[DA 02-16]
Public Safety National Coordination
Committee
AGENCY: Federal Communications
Commission.
ACTION: Notice.
SUMMARY: This document advises
interested persons of a meeting of the
Public Safety National Coordination
Committee ("NCC"), which will be held
in Washington, DC. The Federal
Advisory Committee Act, Public Law
92—463, as amended, requires public
notice of all meetings of the NCC. This
notice advises interested persons of the
fifteenth meeting of the Public Safety
National Coordination Committee.
DATES: February 1, 2002 at 9:30 a.m.—
12:30 p.m.
ADDRESSES: Federal Communications
Commission, 445 12th Street, SW.,
Washington, DC 20554.
FOR FURTHER INFORMATION CONTACT:
Designated Federal Officer, Michael J.
Wilhelm, (202) 418-0680, e-mail
mwilhelm@fcc.gov. Press Contact,
Meribeth McCarrick, Wireless
Telecommunications Bureau, 202-418-
0600, or e-mail mmccarri@fcc.gov.
SUPPLEMENTARY INFORMATION: Following
is the complete text of the Public Notice:
This Public Notice advises interested
persons of the fifteenth meeting of the
Public Safety National Coordination
Committee ("NCC"), which will be held
in Washington, DC. The Federal
Advisory Committee Act, Public Law
92-463, as amended, requires public
notice of all meetings of the NCC.
Date: February 1, 2002.
Meeting Time: General Membership
Meeting—9:30 a.m.-12:30 p.m.
Address: Federal Communications
Commission, 445 12th Street, SW.,
Washington, DC 20554.
The NCC Subcommittees will meet
from 9 a.m. to 5:30 p.m. the previous
day. The NCC General Membership
Meeting will commence at 9:30 a.m. and
continue until 12:30 p.m. The agenda
for the NCC membership meeting is as
follows:
1. Introduction and Welcoming
Remarks.
2. Administrative Matters.
3. Report from the Interoperability
Subcommittee.
4. Report from the Technology
Subcommittee.
5. Report from the Implementation
Subcommittee.
6. Public Discussion.
7. Other Business.
8. Upcoming Meeting Dates and
Locations.
9. Closing Remarks.
The FCC has established the Public
Safety National Coordination
Committee, pursuant to the provisions
of the Federal Advisory Committee Act,
to advise the Commission on a variety
of issues relating to the use of the 24
MHz of spectrum in the 764-776/794-
806 MHz frequency bands (collectively,
-------
------- |