DRAFT - DO NOT CITE OR QUOTE EPA/63 5/R-09/003D
www.epa.gov/iris
?/EPA
TOXICOLOGICAL REVIEW
OF
TRICHLOROACETIC ACID
(CAS No. 76-03-9)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
July 2011
NOTICE
This document is a Final Agency Review/Interagency Science Discussion draft. This
information is distributed solely for the purpose of pre-dissemination peer review under
applicable information quality guidelines. It has not been formally disseminated by EPA. It
does not represent and should not be construed to represent any Agency determination or policy.
It is being circulated for review of its technical accuracy and science policy implications.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document is a preliminary draft for review purposes only. This information is
distributed solely for the purpose of pre-dissemination peer review under applicable information
quality guidelines. It has not been formally disseminated by EPA. It does not represent and
should not be construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
DRAFT - DO NOT CITE OR QUOTE
-------
CONTENTS—TOXICOLOGICAL REVIEW OF
TRICHLOROACETIC ACID (CAS No. 76-03-9)
LIST OF TABLES vii
LIST OF FIGURES ix
LIST OF ABBREVIATIONS AND ACRONYMS x
FOREWORD xii
AUTHORS, CONTRIBUTORS, AND REVIEWERS xiii
1. INTRODUCTION 1
2. CHEMICAL AND PHYSICAL INFORMATION 3
3. TOXICOKINETICS 5
3.1. ABSORPTION 5
3.2. DISTRIBUTION 7
3.3. METABOLISM 11
3.4. EXCRETION 17
3.5. PHYSIOLOGICALLY BASED PHARMACOKINETIC MODELS 20
4. HAZARD IDENTIFICATION 21
4.1. STUDIES IN HUMANS 21
4.1.1. Oral Exposure 21
4.1.2. Dermal Exposure 21
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION 22
4.2.1. Short-term and Subchronic Studies 22
4.2.1.1. Oral 22
4.2.1.2. Subchronic Inhalation Studies 39
4.2.2. Chronic Studies and Cancer Assays 39
4.2.2.1. Oral Studies 45
4.2.2.2. Inhalation Studies 58
4.2.2.3. Studies Using Other Routes of Exposure 58
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES - ORAL AND INHALATION ...60
4.3.1. Reproductive Studies 60
4.3.2. Developmental Studies 60
4.3.2.1. Oral Developmental Studies 60
4.3.2.2. Inhalation Developmental Studies 68
4.3.2.3. In Vitro Studies 68
4.4. OTHER DURATION- OR ENDPOINT-SPECIFIC STUDIES 69
4.4.1. Immunological Studies 69
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE
OF ACTION 70
4.5.1. Mechanistic Studies 70
4.5.1.1. Peroxisome Proliferation 70
4.5.1.2. Oncogene Activation 70
4.5.1.3. Cell Proliferation 72
4.5.1.4. DNA Hypomethylation 74
4.5.1.5. Inhibition of Intercellular Communication 79
iii DRAFT - DO NOT CITE OR QUOTE
-------
4.5.1.6. Oxidative Stress 80
4.5.1.7. Histochemical Characteristics of TCA-Induced Tumors 82
4.5.2. Genotoxicity Studies 84
4.5.2.1. In Vitro Studies 84
4.5.2.2. In Vivo Studies 88
4.6. SYNTHESIS OF MAJOR NONCANCER EFFECTS 90
4.6.1. Oral 90
4.6.1.1. Metabolic Alterations 90
4.6.1.2. Liver Toxicity 91
4.6.1.3. Developmental Toxicity 92
4.6.2. Inhalation 93
4.6.3. Mode-of-Action Information - Non-Cancer 93
4.6.3.1. Metabolic Alterations 93
4.6.3.2. Liver Toxicity 94
4.6.3.3. Developmental Toxicity 94
4.7. EVALUATION OF CARCINOGENICITY 95
4.7.1. Summary of Overall Weight of Evidence Error! Bookmark not defined.
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence 95
4.7.3. Mode-of-Action Information - Cancer 99
4.7.3.1. PPARaagonism 100
4.7.3.2. Additional Proposed Hypotheses and Key Events with Limited
Evidence or Inadequate Experimental Support 118
4.7.3.3. Conclusions About the Hypothesized Mode of Action 122
4.8. SUSCEPTIBLE POPULATION AND LIFE STAGES 123
4.8.1. Possible Childhood Susceptibility 123
4.8.2. Possible Gender Differences 124
4.8.3. Other 125
5. DOSE-RESPONSE ASSESSMENTS 126
5.1. ORAL REFERENCE DOSE (RfD) 126
5.1.1. Choice of Principal Study and Critical Effect—Rationale and Justification 126
5.1.2. Methods of Analysis—Including Models (e.g, PBPK and HMD) 131
5.1.2.1. BMD Modeling of Liver and Testicular Effects from DeAngelo et
al. (2008) 131
5.1.2.2. BMD Modeling of Developmental Toxicity Data from Smith et al.
(1989) 134
5.1.2.3. Selection of POD 135
5.1.3. RfD Derivation—Including Application of Uncertainty Factors (UFs) 136
5.1.4. RfD Comparison Information 137
5.1.5. Previous RfD Assessment 137
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 138
5.3. UNCERTAINTIES IN THE RfD 138
5.4. CANCER ASSESSMENT 139
5.4.1. Choice of Study/Data—Rationale and Justification 140
5.4.2. Dose-Response Data 140
5.4.3. Dose Conversion 143
5.4.4. Extrapolation Methods 143
5.4.5. Time-to-tumor Modeling 144
5.4.6. Oral Cancer Slope Factor and Inhalation Unit Risk 146
5.4.7. Previous Cancer Assessment 147
iv DRAFT - DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE 148
6.1. HUMAN HAZARD POTENTIAL 148
6.2. DOSE RESPONSE 151
6.2.1. Noncancer/Oral 151
6.2.2. Noncancer/Inhalation 151
6.2.3. Cancer/Oral and Inhalation 151
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION A-Error! Bookmark not defined.
APPENDIX B. BENCHMARK DOSE MODELING RESULTS FOR LIVER DATA SETS
FROM DeANGELO ET AL. (2008) B-l
B.I. INCIDENCE OF HEPATOCELLULARINFLAMMATION B-l
B.2. INCIDENCE OF HEP ATOCELLULAR NECROSIS B-6
B.3. INCIDENCE OF TESTICULAR TUBULAR DEGENERATION B-9
B.4. CYANIDE-INSENSITIVE PCO ACTIVITY B-12
APPENDIX C. BENCHMARK DOSE MODELING RESULTS FOR DEVELOPMENTAL
DATA SETS FROM SMITH ET AL. (1989) C-l
C.I. FETAL BODY WEIGHT C-l
C.2. FETAL CROWN-RUMP LENGTH (SMITH ETAL., 1989) C-9
APPENDIX D. MODELING OF LIVER TUMOR INCIDENCE DATA FOR MICE
EXPOSED TO TCA IN DRINKING WATER D-l
D. 1. FIFTY-TWO-WEEK STUDY FROM BULL ET AL. (2002) WITH THREE DOSE
GROUPS D-4
D.2. FIFTY-TWO-WEEK STUDY FROM BULL ET AL. (1990) WITH THREE DOSE
GROUPS D-6
D.3. SIXTY-WEEK STUDY FROM DeANGELO ET AL. (2008) WITH FOUR DOSE
GROUPS D-8
D.4. EIGHTY-TWO-WEEK STUDY FROM PEREIRA (1996) WITH FOUR DOSE
GROUPS D-10
D.5. ONE-HUNDRED-FOUR-WEEK STUDY FROM DeANGELO ET AL. (2008)
WITH THREE DOSE GROUPS D-12
APPENDIX E. MULTISTAGE-WEIBULL (MSW) TIME-TO-TUMOR MODELING OF
INDIVIDUAL AND COMBINED LIVER TUMOR INCIDENCE DATA
SETS FROM DeANGELO ET AL. (2008) E-l
E.I. DOSE CONVERSIONS E-l
E.2. DOSE-RESPONSE DATA E-3
E.3. MSW TIME-TO-TUMOR MODELING E-7
E.4. STATISTICAL ANALYSIS FOR DATA COMPATIBILITY E-8
E. 5. EXTRAPOLATION METHOD AND ORAL CANCER SLOPE FACTOR E-9
E.6. OUTPUT FILES AND PLOTS FOR MSW TIME-TO-TUMOR MODELS E-ll
E.6.1. Study 1 from DeAngelo et al. (2008); 60-Week Study with Four Dose
Groups E-ll
E.6.1.1. MSW Time-to-Tumor Model Run E-ll
E.6.1.2. MSW Time-to-Tumor Plots E-12
DRAFT - DO NOT CITE OR QUOTE
-------
E.6.2. Study 2 from DeAngelo et al. (2008); 104-Week Study with Two Dose
Groups E-13
E.6.2.1. MSW Time-to-Tumor Model Run E-13
E.6.2.2. MSW Time-to-Tumor Plots E-15
E.6.3. Study 3 from DeAngelo et al. (2008); 104-Week Study with Three Dose
Groups E-15
E.6.3.1. MSW Time-to-Tumor Model Run E-15
E.6.3.2. MSW Time-to-Tumor Plots E-17
E.6.4. Combined Dataset (Study 1+3) from DeAngelo et al. (2008) E-17
E.6.4.1. MSW Time-to-Tumor Model Run E-17
E.6.4.2. MSW Time-to-Tumor Plots E-19
E.6.5. Other Combined Data Sets from DeAngelo et al. (2008) E-19
E.6.5.1. MSW Time-to-Tumor Model Run for Combining Study 1, Study 2,
and Study 3 E-19
E.6.5.2. MSW Time-to-Tumor Model Run for Combining Study 1 and
Study 2 E-21
E.6.5.3. MSW Time-to-Tumor Model Run for Combining Study 2 and
Study 3 E-23
vi DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF TABLES
2-1. Selected physical and chemical properties of TCA (CASRN 76-03-9) 3
3-1. Binding of TCA to plasma proteins from different species 9
4-1. Summary of acute, short-term, and subchronic studies evaluating effects of TCA after
oral administration in rats and mice 23
4-2. Summary of chronic studies evaluating noncancer effects of TCA after oral
administration in rats and mice 40
4-3. Summary of cancer bioassays and tumor promotion studies of TCA in rats and mice 42
4-4. Incidence and severity of nonneoplastic lesions in male B6C3Fi mice exposed to TCA
in drinking water for 60 weeks 49
4-5. Incidence and severity of hepatocellular necrosis at 30-45 weeks in male B6C3Fi mice
exposed to TCA in drinking water 49
4-6. Prevalence and multiplicity of hepatocellular tumors in male B6C3Fi mice exposed to
TCA in drinking water for 60 weeks 50
4-7. Incidence of hepatocellular tumors in male B6C3Fi mice exposed to TCA in drinking
water for 104 weeks 51
4-8. Mean PCO activity in male B6C3Fi mice exposed to TCA in drinking water for up to
60 weeks 52
4-9. Incidence of adenomas and hepatocellular carcinomas in B6C3Fi mice treated with
ENUandTCA 56
4-10. Summary of developmental studies evaluating effects of TCA after oral administration
in rats 61
4-11. Selected data for fetal anomalies, showing dose-related trends following exposure of
female Long-Evans rats to TCA on GDs 6-15 64
4-12. Summary of available genotoxicity data on TCA 85
5-1. Candidate studies for derivation of theRfD for TCA 127
5-2. Incidence of nonneoplastic lesions in male B6C3Fi mice exposed to TCA in drinking
water for 60 weeks 131
5-3. Mean PCO activity in male B6C3Fi mice exposed to TCA in drinking water for up to
60 weeks 131
vii DRAFT - DO NOT CITE OR QUOTE
-------
5-4. BMD modeling results for data sets from DeAngelo et al. (2008) 133
5-5. Dose-response data for developmental endpoint in TCA-treated Long-Evans rats 134
5-6. BMD modeling results for data sets from Smith et al. (1989) 135
5-7. Incidence of hepatocellular adenomas, carcinomas, or adenomas and carcinomas
combined in male B6C3Fi mice exposed to TCA in drinking water for 52 weeks 141
5-8. Incidence of hepatocellular adenomas, carcinomas, or adenomas and carcinomas
combined in male B6C3Fi mice exposed to TCA in drinking water for 52 weeks 141
5-9. Incidences of hepatocellular adenomas, carcinomas, or adenomas and carcinomas
combined in male B6C3Fi mice exposed to TCA in drinking water for up to
60 weeks 142
5-10. Incidence of hepatocellular adenomas, carcinomas, or adenomas and carcinomas
combined in female B6C3Fi mice exposed to TCA in drinking water for 82 weeks 142
5-11. Incidence of hepatocellular adenomas, carcinomas, or adenomas and carcinomas
combined in male B6C3Fi mice exposed to TCA in drinking water for up to
104 weeks 143
5-12. Candidate oral cancer slope factors derived from cancer bioassays in B6C3Fi mice 144
5-13. Candidate oral cancer slope factors derived from liver tumor data sets in B6C3Fi
male mice using MSW time-to-tumor modeling and comparison to slope factors
derive using the multistage model inBMDS 145
B-l. BMD modeling results based on incidence of hepatocellular inflammation in male
B6C3Fi mice exposed to TCA in drinking water for 60 weeks B-l
B-2. BMD modeling results based on incidence of hepatocellular necrosis in male B6C3Fi
mice exposed to TCA in drinking water for 30-45 weeks B-6
B-3. BMD modeling results based on incidence of testicular tubular degeneration in male
B6C3Fi mice exposed to TCA in drinking water for 60 weeks B-9
B-4. BMD modeling results based on cyanide-insensitive PCO activity in male B6C3Fi
mice exposed to TCA in drinking water for up to 60 weeks B-12
C-l. BMD modeling results based on fetal body weight in Long-Evans rats exposed to
TCA by gavage on GDs 6-15—male fetuses C-l
C-2. BMD modeling results based on fetal body weight in Long-Evans rats exposed to
TCA by gavage on GDs 6-15—female fetuses C-5
C-3. BMD modeling results based on fetal crown-rump length in Long-Evans rats
exposed to TCA by gavage on GDs 6-15—male fetuses C-9
viii DRAFT - DO NOT CITE OR QUOTE
-------
C-4. BMD modeling results based on fetal crown-rump length in Long-Evans rats
exposed to TCA by gavage on GDs 6-15—female fetuses C-15
D-l. Dose conversion for 60-week study D-l
D-2. Dose conversion for 104-week study D-2
D-3. Comparison of average daily dose, sample size, and tumor incidence from DeAngelo
etal. (2008) as reported by study authors and as recalculated by EPA D-3
E-l. Key characteristics of the three drinking water studies E-l
E-2. Dose adjustments for Study 1 E-2
E-3. Dose adjustments for Study 2 E-2
E-4. Dose adjustments for Study 3 E-2
E-5. Study 1 liver tumor incidence data; B6C3Fi male mice exposed to TCA in drinking
water E-4
E-6. Study 2 liver tumor incidence data; B6C3Fi male mice exposed to TCA in drinking
water E-5
E-7. Study 3 liver tumor incidence data; B6C3Fi male mice exposed to TCA in drinking
water E-6
E-8. Summary of the statistical test for compatibility among the individual studies E-9
E-9. Candidate oral cancer slope factors derived from liver tumor data sets in B6C3Fi
male mice using MSW time-to-tumor modeling E-10
LIST OF FIGURES
2-1. Trichloroacetic acid (TCA) 3
3-1. Proposed metabolic scheme for TCA 13
4-1. Possible key events in theMOA(s) for TCA carcinogenesis 100
5-1. PODs (mg/kg-day) with corresponding potential oral reference values that would
result if alternative endpoints were used as the critical effect 137
ix DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
ACO
ACP
AHF
AIC
ALP
ALT
AST
AUC
BMD
BMDL
BMDS
BMR
BrdU
CAR
CASRN
CPK
CYP450
DCA
DEHP
DEN
ECso
ENU
GD
GGT
GJIC
GSH
GST
HPLC
IAP
IGF
IL
i.p.
IRIS
IUR
LD50
LDH
LINE
LOAEL
LTR
MCA
MDA
5MeC
MNU
MOA
MSW
NADPH
acyl-CoA oxidase
acid phosphatase
altered hepatic foci
Akaike's Information Criterion
alkaline phosphatase
alanine aminotransferase
aspartate aminotransferase
area under the curve
benchmark dose
95% lower confidence limit on the BMD
benchmark dose software
benchmark response
bromodeoxyuridine
constitutive activated/androstane receptor
Chemical Abstracts Service Registry Number
creatine phosphokinase
cytochrome P450
dichloroacetic acid
di(2-ethylhexyl)phthalate
diethylnitrosamine
median effective concentration
ethylnitrosourea
gestation day
gamma-glutamyl transferase
gap junctional intercellular communication
glutathione
glutathione S-transferase
high performance liquid chromatography
intracisternal A particle
insulin-like growth factor
interleukin
intraperitoneal(ly)
Integrated Risk Information System
inhalation unit risk
median lethal dose
lactate dehydrogenase
long interspersed nucleotide element
lowest-observed-adverse-effect level
long terminal repeat
monochloroacetic acid
malondialdehyde
5-methylcytosine
N-methyl-N-nitrosourea
mode of action
multistage Weibull
nicotinamide adenine dinucleotide phosphate (reduced)
x
DRAFT - DO NOT CITE OR QUOTE
-------
NCEA National Center for Environmental Assessment
NHEERL National Health and Environmental Effects Research Laboratory
NOAEL no-observed-adverse-effect level
NRC National Research Council
8-OHdG 8-hydroxy-2'-deoxyguanosine
ORD Office of Research and Development
PB phenobarbital
PBPK physiologically based pharmacokinetic
PCO palmitoyl-CoA oxidase
PFOA perfluorooctanoic acid
POD point of departure
PPAR peroxisome proliferator-activated receptor
RfC reference concentration
RfD reference dose
SD standard deviation
SOD superoxide dismutase
TEARS thiobarbituric acid-reactive substances
TCA trichloroacetic acid
TCE trichloroethylene
TNF-a tumor necrosis factor-alpha
UF uncertainty factor
U.S. EPA U.S. Environmental Protection Agency
XI
DRAFT - DO NOT CITE OR QUOTE
-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard and dose-response assessment in IRIS pertaining to chronic exposure to
trichloroacetic acid (TCA). It is not intended to be a comprehensive treatise on the chemical or
toxicological nature of TCA.
The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
Response, is to present the major conclusions reached in the derivation of the reference dose,
reference concentration and cancer assessment, where applicable, and to characterize the overall
confidence in the quantitative and qualitative aspects of hazard and dose response by addressing
the quality of data and related uncertainties. The discussion is intended to convey the limitations
of the assessment and to aid and guide the risk assessor in the ensuing steps of the risk
assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
xii DRAFT - DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGER/AUTHOR
Zheng (Jenny) Li, Ph.D., DABT
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
CO-AUTHORS
Diana Wong, Ph.D., DABT
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
TedBerner, M.S.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Yuyang Christine Cai, M.S., PMP
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
CONTRIBUTORS
Jordan Trecki, Ph.D.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Karen Hogan, M.S.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
CONTRACTOR SUPPORT
Lori Moilanen, Ph.D., DABT
Syracuse Research Corporation
Syracuse, NY
Peter McClure, Ph.D., DABT
Syracuse Research Corporation
Syracuse, NY
xiii DRAFT - DO NOT CITE OR QUOTE
-------
Brian Anderson, M.E.M.
Syracuse Research Corporation
Syracuse, NY
REVIEWERS
This document has been provided for review to EPA scientists, interagency reviewers
from other federal agencies and White House offices, and the public, and peer reviewed by
independent scientists external to EPA. A summary and EPA's disposition of the comments
received from the independent external peer reviewers is included in Appendix A.
INTERNAL EPA REVIEWERS
Jane Caldwell, Ph.D.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Weihsueh Chiu, Ph.D.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Joyce Donohue, Ph.D.
Office of Water
U.S. Environmental Protection Agency
Washington, DC
Kate Guyton, Ph.D.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Robert McGaughy, Ph.D.
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Susan Rieth, MPH
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
EXTERNAL PEER REVIEWERS
Penelope A. Fenner-Crisp, Ph.D., DABT
Independent Consultant
xiv DRAFT - DO NOT CITE OR QUOTE
-------
David W. Gaylor, Ph.D.
Gaylor and Associates, LLC
Ronald L. Melnick, Ph.D.
Ron Melnick Consulting, LLC
Martha M. Moore, Ph.D.
National Center for Toxicological Research (NCTR)
Food and Drug Administration
Michael A. Pereira, Ph.D.
Ohio State University Comprehensive Cancer Center
Ivan Rusyn, M.D., Ph.D.
University of North Carolina
Andrew G. Salmon, D.Phil.
Office of Environmental Health Hazard Assessment (OEHHA)
California EPA
Anthony R. Scialli, M.D.
Tetra Tech Sciences
Alan H. Stern, Dr.P.H., DABT (Chair)
Independent Consultant
xv DRAFT - DO NOT CITE OR QUOTE
-------
1. INTRODUCTION
This document presents background information and justification for the Integrated Risk
Information System (IRIS) Summary of the hazard and dose-response assessment of
trichloroacetic acid (TCA). IRIS Summaries may include oral reference dose (RfD) and
inhalation reference concentration (RfC) values for chronic and other exposure durations, and a
carcinogenicity assessment.
The RfD and RfC, if derived, provide quantitative information for use in risk assessments
for health effects known or assumed to be produced through a nonlinear (presumed threshold)
mode of action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with
uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (<24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous
exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
derived for chronic exposure duration.
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral and inhalation
exposure may be derived. The information includes 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. Quantitative risk estimates may be derived from the application of a
low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
the estimate of risk per mg/kg-day of oral exposure. Similarly, an inhalation unit risk is a
plausible upper bound on the estimate of risk per ug/m3 air breathed.
Development of these hazard identification and dose-response assessments for TCA has
followed the general guidelines for risk assessment as set forth by the National Research Council
(NRC, 1983). U.S. Environmental Protection Agency (U.S. EPA) Guidelines and Risk
Assessment Forum technical panel reports that may have been used in the development of this
assessment include the following: Guidelines for the Health Risk Assessment of Chemical
Mixtures (U.S. EPA, 1986a), Guidelines for Mutagenicity Risk Assessment (U.S. EPA, 1986b),
Recommendations for and Documentation of Biological Values for Use in Risk Assessment (U.S.
EPA, 1988), Guidelines for Developmental Toxicity Risk Assessment (U.S. EPA, 1991), Interim
1 DRAFT - DO NOT CITE OR QUOTE
-------
Policy for Particle Size and Limit Concentration Issues in Inhalation Toxicity (U.S. EPA,
1994a), Methods for Derivation of Inhalation Reference Concentrations and Application of
Inhalation Dosimetry (U.S. EPA, 1994b), Use of the Benchmark Dose Approach in Health Risk
Assessment (U.S. EPA, 1995), Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA,
1996), Guidelines for Neurotoxicity Risk Assessment (U.S. EPA, 1998), Science Policy Council
Handbook. Risk Characterization (U.S. EPA, 2000a), Benchmark Dose Technical Guidance
Document (U.S. EPA, 2000b), Supplementary Guidance for Conducting Health Risk Assessment
of Chemical Mixtures (U.S. EPA, 2000c), A Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA, 2002), Guidelines for Carcinogen Risk Assessment (U.S.
EPA, 2005a), Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to
Carcinogens (U.S. EPA, 2005b), Science Policy Council Handbook: Peer Review (U.S. EPA,
2006a), and A Framework for Assessing Health Risks of Environmental Exposures to Children
(U.S. EPA, 2006b).
The literature search strategy employed for this compound was based on the Chemical
Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
scientific information submitted by the public to the IRIS Submission Desk was also considered
in the development of this document. The relevant literature was reviewed through June 2011.
It should be noted that references have been added to the Toxicological Review after the external
peer review in response to peer reviewers' comments and for the sake of completeness. These
references have not changed the overall qualitative and quantitative conclusions. See Section 7
for a list of references added after peer review.
DRAFT - DO NOT CITE OR QUOTE
-------
2. CHEMICAL AND PHYSICAL INFORMATION
TCA is a colorless to white crystalline solid with a sharp, pungent odor (NIOSH, 2003).
The dissociation constant (pKa) for TCA at 25°C is 0.51. In aqueous solutions, TCA occurs
almost exclusively in the ionized form as trichloroacetate anion. Common synonyms include
trichloroethanoic acid and trichloro-methanecarboxylic acid. The structure of TCA is shown in
Figure 2-1 and selected physical and chemical properties of TCA are provided in Table 2-1.
o
Cl
OH
Figure 2-1. Trichloroacetic acid (TCA).
Table 2-1. Selected physical and chemical properties of TCA (CASRN 76-
03-9)
Chemical formula
Molecular weight
Density
Melting point
Boiling point
Vapor pressure
Log pKa
Log Kow
Water solubility
Other solubilities
Henry's law constant
C2HC13O2
163.39
1.6126g/mLat64°C
57.5°C
196.5°C
0.16mmHgat25°C
0.51at25°C
1.33
1,306 g/lOOg at 25°C
At 25°C, methanol, 2,143 g/100 g; ethyl ether,
617 g/100 g; acetone, 850 g/100 g; benzene,
201 g/100 g; o-xylene, 110 g/100 g
1.35 x l(T8 atm-nrVmol at 25°C
O'Neil, 2001
O'Neil, 2001
Lide, 2000
Lide, 2000
Lide, 2000
Lileyetal., 1984
Serjeant and Dempsey, 1979
Hanschetal., 1995
Morris and Bost, 2002
Morris and Bost, 2002
Bowdenetal., 1998
"Monsanto (1979).
Estimated by different methods: Dow (1971).
Source: NLM(2007).
TCA is used as a soil sterilizer and a laboratory intermediate or reagent in the synthesis of
a variety of medicinal products and organic chemicals (NLM, 2003). Medical applications of
TCA include use as a reagent for the detection of albumin (Lewis, 1997), application as an
DRAFT - DO NOT CITE OR QUOTE
-------
antiseptic (Morris and Bost, 2002), and use as a skin peeling agent (Al-Waiz and Al-Sharqi,
2002; Lee et al., 2002; Coleman, 2001). TCA is also used industrially as an etching and pickling
agent for the surface treatment of metals and (in solution) as a solvent in the plastics industry
(Koenig, 2005).
TCA can be formed as a combustion byproduct of organic compounds in the presence of
chlorine (Juuti and Hoekstra, 1998). Stack gases of municipal waste incinerators have been
reported to contain 0.37-3.7 |ig/m3 TCA (Mower and Nordin, 1987). TCA could be a
photooxidation product of tetrachloroethylene and 1,1,1-trichloroethane in the atmosphere (Juuti
and Hoekstra, 1998; Reimann et al., 1996; Sidebottom and Franklin, 1996). Sidebottom and
Franklin (1996) suggested that atmospheric degradation of chlorinated solvents could contribute
only a minor amount of TCA to the atmosphere, based on the mechanistic and kinetic evidence,
as well as the observed global distribution of TCA in precipitation. However, TCA has been
detected in rainwater at a concentration range of 0.01-1 jig/L (Reimann et al., 1996).
TCA is formed from organic material during water chlorination (IPCS, 2000; Coleman et
al., 1980) and has been detected in groundwater, surface water distribution systems, and
swimming pool water. Human exposure to TCA occurs directly through the consumption and
use of tap water disinfected with chlorine-releasing disinfectants (U.S. EPA, 2005c). TCA was
detected in vegetables, fruits, and grains (Reimann et al., 1996) and can be taken up into
foodstuffs from the cooking water (U.S. EPA, 2005c). Therefore, human exposure to TCA can
also occur via food consumption.
DRAFT - DO NOT CITE OR QUOTE
-------
3. TOXICOKINETICS
3.1. ABSORPTION
Results from studies with rats and mice indicate that TCA is extensively absorbed by the
gastrointestinal tract. In studies of excreta collected for up to 48 hours from male F344 rats and
B6C3Fi mice given single doses of [14C]-labeled TCA in water ranging from 5 to 100 mg/kg,
radioactivity detected in urine and in CC>2 in expired air represented about 57-72 and 4-8%,
respectively, of the administered dose (Larson and Bull, 1992). Most of the urinary radioactivity
was unmetabolized TCA, which accounted for 81-90% of the urinary radioactivity and 48-65%
of the administered radioactivity. Urinary radioactivity in metabolites of TCA represented only
minor amounts of the administered radioactivity: 1-3% for dichloroacetic acid (DCA) and 5-
11% for a high performance liquid chromatography (HPLC) fraction coeluting with standards for
glyoxylic acid, oxalic acid, and glycolic acid (which exist as glyoxylate, oxalate, and glycolate
anions at physiological pH). Radioactivity detected in feces accounted for only about 2-4% of
the administered radioactivity (Larson and Bull, 1992). In another study in which male B6C3Fi
mice were administered single 100 mg/kg doses of uniformly labeled [14C]-TCA by gavage
(constant volume of 10 mL/kg; vehicle not specified), the average distribution of radioactivity
24 hours after dose administration was about 55% in urine, about 5% in CC>2, and about 5% in
feces, with the remainder in the carcasses (Xu et al., 1995). Radioactivity in urinary metabolites,
expressed as percentage of the administered dose, showed the following distribution: 44.5% as
trichloroacetate, 0.2% as dichloroacetate, 0.03% as monochloroacetate, 0.06% as glyoxylate,
0.11% as glycolate, 1.5% as oxalate, and 10.2% as unidentified compounds. Results from both
of these studies are consistent with extensive absorption by the gastrointestinal tract, followed by
rapid elimination in the urine, principally as the nonmetabolized parent compound.
Indicative of rapid absorption, TCA concentrations in the plasma or liver peaked in the
first hour following oral dosing in other short-term studies with mongrel dogs (Hobara et al.,
1988a) and male B6C3Fi mice (administered either as aqueous free acid, neutral aqueous
solution, or free acid in corn oil) (Styles et al., 1991). Likewise, peak blood concentrations of
TCA were attained at a mean time of 1.55 hours after oral administration of single doses of
500 |imol/kg (82 mg/kg) TCA in water (adjusted to pH 7) to male F344 rats (Schultz et al.,
1999). Comparison of the areas under the curve (AUCs) of plasma concentrations of TCA
following oral administration and intravenous administration of TCA at the same dose level
indicated that oral bioavailability of TCA was approximately equivalent to intravenous
bioavailability (Schultz et al., 1999). The average ratio of oral:intravenous AUCs were 1.16.
The 16% higher AUC value for oral exposure likely reflects measurement or statistical
variability and/or differences in clearance rate by the two routes of administration. The mean
absorption time, which was determined as the difference in the mean residence time in blood
5 DRAFT - DO NOT CITE OR QUOTE
-------
following oral and intravenous dosing, was 6 hours for TCA. The mean absorption time is
dependent on clearance from the blood as well as the absorption rate; therefore, the longer mean
absorption time as compared to time-to-peak blood concentration of 1.55 hours may reflect
slower clearance following oral dosing (Schultz et al., 1999).
Results from studies of urinary excretion of TCA by human subjects following 30-minute
sessions in chlorinated swimming pool water indicate that TCA is rapidly absorbed by the skin
(Kim and Weisel, 1998). TCA concentrations in pool water were measured before and after the
subjects (two males and two females) either walked in the pool without submerging their heads
(dermal exposure only) or swam (dermal exposure plus incidental oral exposure) in the pool for
30 minutes. TCA concentrations in the swimming pool water at various sessions varied from
57 to 871 ng/L, with a mean of 420 |ig/L and a median of 278 |ig/L. Entire urine voids were
collected for at least 24 hours before exposure and 20-40 hours following exposure, at
approximately 3-hour intervals. Additional urine samples were collected 5-10 minutes
immediately before and after exposures. During the 24 hours prior to and following exposure,
subjects avoided activities such as drinking chlorinated tap water or visiting the dry cleaner,
which might have resulted in urinary TCA excretion. For each exposure session, the amount of
urinary TCA associated with exposure was calculated for each subject from the amount of TCA
excreted within 3 hours after exposure minus the amount excreted within 3 hours prior to
exposure. Pre-exposure amounts of TCA in urine ranged from 155 to 1,183 ng, whereas
postexposure amounts ranged from 294 to 1,590 ng. The amount of urinary TCA associated
with the 30-minute exposure sessions ranged from 33 to 824 ng, depending on the subject and
exposure session. Urinary excretion rates (ng/minute), calculated for various intervals before
and after exposure, showed peaks at the postexposure 5-10-minute period that were about
threefold higher than pre-exposure period rates. Excretion rates calculated for the first full
3-hour interval after exposure returned to values that were not discernable from pre-exposure
rates. A scatter plot of the amount of urinary TCA per exposed body surface area (ng/m2) in
subjects under the dermal-exposure-alone scenario versus TCA exposure expressed as the TCA
concentration in water multiplied by the exposure duration (|ig/L x hour) indicated that urinary
excretion (and, thus, presumably, dermal absorption) was higher with higher exposure. For
exposures of about 20 and 420 |ig/L x hour TCA, values for urinary TCA per surface area were
about 10-50 ng/m2 and 60-160 ng/m2, respectively. The results from this study indicate that
dermal absorption and subsequent urinary elimination of TCA are rapid, but are inadequate to
provide more quantitative measures of dermal absorption for TCA, such as dermal permeability
coefficients.
No studies were identified on the extent or rate of TCA absorption following inhalation
exposure.
DRAFT - DO NOT CITE OR QUOTE
-------
3.2. DISTRIBUTION
The tissue distribution of TCA following absorption has been most completely
characterized in male F344 rats injected intravenously with radiolabeled [14C]-TCA at doses of 0,
6.1, 61, or 306 |imol/kg (0, 1, 10, or 50 mg/kg) (Yu et al., 2000). TCA equivalent concentrations
in plasma, red blood cells, and eight tissues (based on levels of detected radioactivity) were
determined at various time points for up to 24 hours after injection (1, 3, 6, 9, and 24 hours).
Peak concentrations in plasma and all tissues were observed at the postexposure first sampling.
Levels of radioactivity in urine, feces, and expired air were also measured. Overall kinetic
behavior was similar at all three doses (i.e., TCA equivalent concentrations declined with time in
plasma and tissues, and first-order elimination rate constants were not consistently changed
across tissues with increasing dose level). At early time points, the highest TCA equivalent
concentrations were measured in plasma, followed by kidney, red blood cells, liver, skin, small
intestine, large intestine, muscle, and fat; the relative order of these concentrations remained
unchanged up to 3 hours following dosing. However, at 24 hours following dosing, the
distribution pattern was changed, with the liver showing the highest TCA equivalent
concentration. First-order rate constants for the disappearance of TCA equivalents from plasma
and tissues were calculated and subsequently classified by the study authors into three groups:
(1) fast elimination (rate constants between 0.081 and 0.156 h"1) in plasma, red blood cells,
muscle, and fat; (2) moderate elimination (rate constants between 0.064 and 0.077 h"1) in kidney
and skin; and (3) slow elimination (rate constants between 0.037 and 0.063 h"1) in liver, small
intestine, and large intestine.
To explore a possible explanation for the apparent differences in elimination kinetics of
TCA in the plasma and liver of rats, Yu et al. (2000) compared the time courses of the
distribution of nonextractable TCA equivalents (i.e., radioactivity from TCA metabolically
incorporated into macromolecules) and extractable TCA equivalents in plasma and liver for up to
24 hours after injection. In both plasma and liver, nonextractable TCA equivalents increased to
plateau levels within 6-10 hours after injection. Although the concentrations of nonextractable
TCA equivalents in liver were higher than those in plasma, the total amount of TCA metabolized
in these 24-hour studies (nonextractable TCA equivalents plus radioactivity in CO2 in expired
air) was estimated to be <20% of the administered dose.
The binding of TCA in plasma and liver homogenate was also investigated (Yu et al.,
2000). Data were fitted using a model that assumed binding and consisted of two components:
low-specificity (nonsaturable, linear) anion binding and high-specificity (saturable, nonlinear)
binding. Results from these in vitro binding studies indicated that reversible binding of TCA in
rat plasma (presumably to serum albumin) was more extensive than binding in liver
homogenates. Yu et al. (2000) hypothesized that TCA disappears from the liver more slowly
than from the plasma because of a concentrating transport process in hepatocyte plasma
membranes. In addition, theoretical calculations of cumulative urinary excretion of TCA,
7 DRAFT - DO NOT CITE OR QUOTE
-------
assuming glomerular filtration of free, unbound plasma TCA (the only operable excretory
process), indicated that actual urinary excretion rates of TCA were slower than the theoretical
values (Yu et al., 2000). It was hypothesized that this difference may be due to the occurrence of
reabsorption of TCA into renal tubules and/or from the bladder. Support for this hypothesis,
which provides at least a partial explanation for the relatively high concentrations of TCA
equivalents in the kidney, includes the observation of reabsorption of TCA into the systemic
circulation following injection into the bladder of dogs (Hobara et al., 1988b, 1987).
Reversible binding of trichloroacetate anion to positively charged proteins in plasma has
been hypothesized to play a role in determining the tissue distribution and elimination of TCA
and has been demonstrated in both in vivo and in vitro studies (Lumpkin et al., 2003; Toxopeus
and Frazier, 2002, 1998; Yu et al., 2000; Schultz et al., 1999; Templin et al., 1993).
Unbound TCA accounted for an average of 53 ± 4% (mean ± standard deviation [SD]) of
the total TCA plasma concentration in blood samples collected at 0.25, 1, and 3 hours after
intravenous injection of single doses of 500 |imol/kg (81.7 mg/kg) TCA to male F344 rats
(Schultz et al., 1999). In this in vivo study, gas chromatography and electron capture detection
were used to determine TCA concentrations in plasma samples and ultrafiltrates of plasma
samples from which proteins with molecular weights >10,000-12,000 were removed. The
blood/plasma concentration ratio for TCA was 0.76, indicating some propensity for TCA to
partition to the plasma, and was consistent with the ability of TCA to bind plasma proteins.
Templin et al. (1993) estimated the degree of in vitro TCA binding to plasma proteins by
incubating [14C]-TCA (position of radiolabel not specified) at various concentrations with
plasma obtained from unexposed male B6C3Fi mice. The amounts of unbound and bound
radioactivity were determined in samples removed after various incubation times, using
ultrafiltration to remove proteins from the samples. At TCA concentrations <306 nmol/mL,
approximately 50-57% of the TCA was bound to plasma constituents, while percentage binding
decreased with increasing TCA concentrations. Approximately 41, 34, and 23% of TCA was
bound to plasma constituents at TCA concentrations of 306, 612, and 1,224 nmol/mL,
respectively.
Templin et al. (1995) measured the binding of TCA to plasma proteins in four different
species: dog, rat, mouse, and human. Plasma samples were prepared from whole blood and
incubated with 3-1,224 nmol/mL [14C]-TCA at 37°C for 30 minutes. Binding of TCA to plasma
constituents was analyzed by using a Scatchard plot and is summarized in Table 3-1. Binding of
TCA to plasma proteins was higher in humans than in rats and mice.
DRAFT - DO NOT CITE OR QUOTE
-------
Table 3-1. Binding of TCA to plasma proteins from different species3
Species
Mouse
Rat
Dog
Human
Binding11
6 nmol/mL
55%
53.5%
64.8%
84.3%
61 nmol/mL
52%
48.9%
58.5%
83.3%
612 nmol/mL
34%
38.3%
54.2%
74.8%
Kdc
OiM)
46.1
383.6
No data
174.6
"Values are expressed as percent of [14C]-TCA associated with protein fraction, expressed as mean value for two
replications of pooled samples.
bTemplinetal. (1995).
"Lumpkin et al. (2003).
Kd = Dissociation constant
Toxopeus and Frazier (1998) investigated the kinetics of TCA in isolated perfused rat
liver from male F344 rats. The isolated perfused rat liver system was dosed with either 5 or
50 jimol of TCA, and TCA concentrations were monitored in perfusion medium supplemented
with 4% bovine serum albumin and in bile for 2 hours. Liver viability was assessed by
measuring lactate dehydrogenase (LDH) leakage into perfusion medium and by the rate of bile
production. At the end of the exposure period, the concentration of TCA in liver was measured.
In the study with 50 jimol TCA, the total TCA concentration (free and bound to bovine serum
albumin) in perfusion medium decreased slightly during the first 30 minutes and then remained
constant for the duration of the exposure period; the total TCA concentration in the perfusion
medium was relatively constant in the study with 5 jimol TCA. At the high concentration,
approximately 93% TCA was bound to bovine serum albumin, and the free TCA concentration
averaged 15.4 jiM at 5 minutes of exposure and 14.9 jiM at 120 minutes of exposure. At the low
concentration, 96% of the TCA was bound to protein and the free TCA concentration was
approximately constant at 0.9-1 jiM over the study period. The calculated free-TCA
concentration in the liver intracellular space was higher than the free-TCA concentration in the
perfusion medium. Enzyme leakage and bile flow were similar at both TCA exposure levels to
those in the control liver, indicating the absence of hepatotoxicity. The authors concluded that
the binding of TCA to bovine serum albumin in perfusion medium limits the uptake of TCA by
the liver and that TCA is virtually unmetabolized by the liver. These findings are consistent with
those from in vivo mouse studies (e.g., Templin et al. [1993]) demonstrating TCA binding to
plasma proteins and suggest that TCA kinetics may be influenced by plasma-protein binding. In
a similar study conducted in the same laboratory, using concentrations of 50, 250, or 1,000 jiM
TCA (Toxopeus and Frazier, 2002), >90% of the TCA in the perfusion medium was bound to
albumin, confirming the results for extent of binding obtained by Toxopeus and Frazier (1998).
Lumpkin et al. (2003) measured the in vitro binding of TCA at 13 concentrations ranging
from 0.06 to 6,130 jiM (0.01-1,000 |ig/mL) to plasma proteins in samples of plasma from
DRAFT - DO NOT CITE OR QUOTE
-------
humans, rats, and mice. Pooled plasma for each species was obtained from commercial sources.
Neither donor strain (for rodents) nor donor sex were specified. Binding was determined by
using an equilibrium dialysis technique. Plots of bound versus free TCA concentrations were
compared with simulations from three binding models—a single saturable site model, a two
saturable site model, and a saturable plus unsaturable site model—to explore the mechanistic
basis for species differences. Plots of bound versus free TCA concentration indicated that the
proportion of bound TCA is substantially higher for human plasma than for rodent plasmas.
Decreases in the proportion of bound to free TCA at concentrations >307 jiM were indicative of
saturation of plasma binding. Human plasma showed the most pronounced binding over the
tested range of concentrations, followed by rat, then mouse. Binding to human plasma was
highest (86.8%) at the lowest quantifiable TCA concentration (0.12 jiM). The bound fraction in
human plasma remained relatively constant, with a mean value of 81.6% over a 3.7 order of
magnitude increase in TCA concentration. In comparison, maximum and average quasi-steady-
state bound fractions were 66.6 and 38.6% for the rat and 46.6 and 19.1% for the mouse,
respectively.
Lumpkin et al. (2003) noted that the average value of TCA protein binding for the mouse
was considerably lower than the range of 34-57% determined in vitro in male B6C3Fi mice
reported by Templin et al. (1993). The reason for the disparity is unclear, but Lumpkin et al.
(2003) noted that Templin et al. (1993) used Scatchard analysis over a narrower range of TCA
concentrations to estimate binding parameters. The best fits to the observed data were obtained
by using the single saturable binding process model, but data limitations (inadequate number of
data points at low TCA concentrations) precluded acceptable fits of the two-saturable-process
model. Use of albumin rather than total plasma protein concentration also improved model fit.
The calculated binding capacity (Bmax) values for humans, rats, and mice were 709, 283, and
29 jiM of TCA, respectively. The average number of binding sites per molecule of protein was
2.97, 1.49, and 0.17, respectively. The low number of binding sites observed for mice may
indicate the existence of other ligands competing for TCA binding sites in mouse plasma. The
dissociation constant values for humans, rats, and mice were 174.6, 383.6, and 46.1 jiM,
respectively. The higher binding capacity of human plasma was a product of a higher number of
binding sites per molecule of protein and higher reported plasma concentrations of albumin
(239 jiM for humans versus 190-196 jiM for rodents).
A possible toxicological significance of these findings for binding of TCA to plasma
proteins is that the extent of plasma binding may influence the distribution of TCA from blood to
target tissues to a degree that may influence species differences in susceptibility to TCA toxicity.
Based on the results from these in vitro binding studies and published reports of peak plasma
concentrations of total TCA in mice (580 jiM) and rats (300 jiM) following gavage exposure to
1,200 mg/kg trichloroethylene (TCE), Lumpkin et al. (2003) calculated that plasma levels of free
TCA would be about four- to fivefold higher in mice than in rats at this dose level. Lumpkin et
10 DRAFT - DO NOT CITE OR QUOTE
-------
al. (2003) speculated that this difference was consistent with the apparent relative susceptibility
of mice to TCA-induced liver tumors. The relative susceptibility of rats and mice to
TCA-induced liver tumors awaits confirmation from further research (as discussed in Section
4.7), as does the hypothesis that toxicokinetics of TCA in humans may be more like TCA
toxicokinetics in rats than in mice.
Abbas and Fisher (1997) determined in vitro tissue:blood partition coefficients for TCA
in B6C3Fi mouse tissues by using a closed vial equilibration method. The tissue:blood partition
coefficients were 1.18 for the liver, 0.88 for the muscle, 0.74 for the kidney, and 0.54 for the
lung. Comparable empirical data for TCA tissue:blood partition coefficients in other species
were not located.
No additional studies were identified that might confirm the nature and extent of species
differences in TCA distribution. Indirect evidence, primarily from studies involving exposure to
chlorinated solvents, suggests that TCA is available for systemic distribution in humans, as
determined by appearance of TCA in the blood and urine. TCA is a metabolite of TCE and has
been frequently measured in the urine or blood of humans exposed to TCE as a result of
environmental contamination (Bruning et al., 1998; Skender et al., 1994; Vartiainen et al., 1993;
Ziglio et al., 1983; Ziglio, 1981) and in volunteer studies (Fisher et al., 1998; Brashear et al.,
1997; NIOSH, 1973). TCA is also found in the blood and urine of humans without known
chlorinated solvent exposures (Hajimiragha et al., 1986) and in individuals exposed to low
concentrations of TCA in swimming pool water (Kim and Weisel, 1998) and drinking water
(Calafat et al., 2003; Froese et al., 2002; Kim and Weisel, 1998).
No studies investigating the toxicokinetics or degree of maternal-to-fetus or blood-to-
breast milk transfer of TCA were located, although TCA has been detected in mouse fetuses and
amniotic fluid following 1-hour inhalation exposures of pregnant C57BL mice (gestation day
[GD] 17) to high concentrations of TCE or tetrachloroethylene (presumably 1,100-1,200 ppm)
(Ghantous et al., 1986). In these studies, peak TCA concentrations in fetuses and amniotic fluid
were attained 4 hours after cessation of exposure.
3.3. METABOLISM
As discussed in Sections 3.1 and 3.2, results from studies of rats and mice involving oral
or intravenous administration of radiolabeled TCA indicate that TCA is only metabolized to a
limited extent. Urinary excretion of nonmetabolized TCA accounted for about 48-55% of
administered oral doses ranging from 5 to 100 mg/kg in rats and mice (Xu et al., 1995; Larson
and Bull, 1992). Radioactivity in CC>2 collected in expired air accounted for 5-8% of
administered doses in these studies, and amounts of radioactivity detected in individual
metabolites in urine, such as DCA, monochloroacetic acid (MCA), glyoxylic acid, glycolic acid,
and oxalic acid, were generally small, each accounting for less than 2 or 3% of administered
doses (Xu et al., 1995; Larson and Bull, 1992). In contrast, orally administered radiolabeled
11 DRAFT - DO NOT CITE OR QUOTE
-------
DCA is more extensively metabolized in rats and mice than is TCA (Larson and Bull, 1992).
Based on measurement of radioactivity in expired CC>2 and in nonextractable radioactivity in
plasma and tissues (i.e., radioactivity from metabolized TCA incorporated into macromolecules),
Yu et al. (2000) estimated that <20% of an administered intravenous dose of 50 mg/kg TCA was
metabolized in rats within 24 hours. Within 24 hours after injection of 1 or 50 mg/kg TCA,
urinary excretion accounted for about 48 and 87%1 and total exhaled CC>2 accounted for about
12 and 8% of the administered doses, respectively (Yu et al., 2000). These results are consistent
with the idea that, at the higher dose level, metabolism of TCA may have been saturated, leading
to an increased percentage of dose excreted as TCA in the urine and a decreased percentage of
dose exhaled as metabolized CC>2. However, the distribution of radioactivity among TCA and
potential metabolites in the urine was not quantified in this study (Yu et al., 2000), so
confirmation of this idea awaits further research.
Figure 3-1 presents a proposed metabolic scheme for TCA, which is based on results
from in vivo and in vitro studies in animals. The first proposed step is the reductive
dehalogenation of TCA by cytochrome P450 (CYP450) enzymes, producing DCA via a free
radical intermediate, the dichloroacetate radical. Early evidence in support of this step was
restricted to the detection of radioactivity from TCA in urinary DCA (Xu et al., 1995; Larson and
Bull, 1992) and the formation of lipid peroxidation byproducts following incubations of liver
microsomes with TCA (Ni et al., 1996; Larson and Bull, 1992).
1 These values were extracted from Figure 2 of the Yu et al. (2000) report.
12 DRAFT - DO NOT CITE OR QUOTE
-------
OH O
I II
C—C :
H* in
Glycolic acid
C02*-
Cl O
I II
ci—c-c
I I
CI OH
TCA
O
DCA
, Glyoxylic acid
-C C
I I
OH OH
Oxalic acid
H-C-C
I I
CI OH
MCA
Note: Molecules in brackets are intermediates proposed by Xu et al. (1995).
Sources: Adapted from Bull (2000); Lash et al. (2000); Merdink et al. (2000); Xu
etal. (1995).
Figure 3-1. Proposed metabolic scheme for TCA.
Some uncertainty about the metabolic formation of DCA from TCA has been expressed,
because DCA has been shown to form as an artifact during sample processing (Ketcha et al.,
1996). Using analytical processes and methods to prevent the artifactual conversion of TCA to
DCA, Merdink et al. (1998) reported that DCA was not detected in blood samples from male
B6C3Fi mice given single intravenous doses of 100 mg/kg TCA. Likewise, Yu et al. (2000)
reported that radiolabeled DCA or other radiolabeled metabolites were not detected in plasma,
urine, or other tissues collected from male F344 rats following intravenous injection of
13
DRAFT - DO NOT CITE OR QUOTE
-------
[14C]-labeled TCA, although metabolism of TCA was indicated in this study by the detection of
radioactivity in exhaled CO2 and in nonextractable materials (e.g., incorporated into cellular
macromolecules) in plasma and tissue extracts. However, simulations with a pharmacokinetic
model indicated that the rapid elimination of DC A from blood, relative to its formation, is
consistent with the lack of accumulation of measurable amounts of DCA in the blood following
injection of TCA (Merdink et al., 1998). Studies with a chemical Fenton reaction system and
with suspensions of rat or mouse liver microsomes incubated with TCA detected the
dichloroacetate radical by gas chromatography / mass spectrometry analysis following trapping
of an adduct between the dichloroacetate radical and phenyl-tertiary-butyl nitroxide (Merdink et
al., 2000), providing evidence for the occurrence of the metabolic conversion of TCA to DCA
via reductive dehalogenation.
As shown in Figure 3-1, the reductive dechlorination of DCA to MCA has been proposed
to proceed via a proposed monochloroacetate radical, which has also been proposed to be
transformed to glyoxylic acid via oxidative dechlorination (Xu et al., 1995). Also shown in
Figure 3-1 is a proposed oxidative dechlorination pathway that transforms DCA to oxalic acid
via a proposed monochloroacetaldehyde intermediate (Xu et al., 1995). More direct evidence for
these pathways is not available, and enzymes that may catalyze the reactions are not
characterized. Glyoxylic acid can be metabolically transformed to glycolic acid and oxalic acid,
as well as to CO2j via mainstream carbon metabolic pathways (Figure 3-1).
Although the metabolism of TCA to DCA has been proposed, as shown in Figure 3-1, the
mechanisms of dehalogenation of DCA have not been conclusively determined. The metabolism
of both TCA and DCA to similar downstream metabolites, as shown in Figure 3-1, suggests that
they may be sequential metabolites in the same pathway. For this reason, a brief summary of
DCA metabolism is included in this review. For a more detailed analysis of data on DCA
metabolism, the reader is referred to the IRIS ToxicologicalReview ofDichloroaceticAcid(\J.S.
EPA, 2003). DCA undergoes metabolic conversion via dechlorination and oxygenation to yield
glyoxylate, oxalate, carbon dioxide, and several glycine conjugates, including hippuric acid
(James et al., 1998a; Lin et al., 1993; Evans and Stacpoole, 1982; Crabb et al., 1981). In vitro
experiments have demonstrated that conjugation with glutathione (GSH) also occurs and that this
is the primary metabolic conversion pathway for DCA in the B6C3Fi mouse, F344 rat, and
human-liver cytosol (James et al., 1997; Lipscomb et al., 1995). The GSH-dependent
oxygenation of DCA to form the initial major metabolite, glyoxylic acid, is catalyzed by
glutathione S-transferase-zeta (GST-Q (Tong et al., 1998a, b).
Studies on enzyme pathways that might play a role in the metabolism of TCA are limited
to one that evaluated the toxic effects of DCA and TCA on liver slices from male B6C3Fi mice,
as well as the metabolic capacity of the liver for these two compounds (Pravacek et al., 1996).
To evaluate cytotoxicity (as evidenced by potassium content and liver enzyme leakage), the liver
slices were exposed for up to 8 hours at concentrations of TCA ranging from 0 to 86 mM
14 DRAFT - DO NOT CITE OR QUOTE
-------
(0-14 mg/mL) TCA. To determine if TCA treatments can alter phase I or phase II
biotransformations, the liver slices were exposed to a low or high concentration of DC A or TCA,
and the conversion of 7-ethoxycoumarin to 7-hydroxycoumarin (a measure of phase I
metabolism) and formation of sulfate and glucuronide conjugates of hydroxycoumarin (a
measure of phase II metabolism) were assessed. TCA treatment with 1,000 |ig/mL increased
phase I metabolism but had no effect on phase II metabolism at either 25 or 1,000 jig/mL.
Metabolism of TCA was monitored by the rate of removal of the parent compound. The removal
of TCA was not saturable at non-cytotoxic concentrations over the range of concentrations tested
(0-5,000 |ig/mL); thus, neither the Km (the concentration at which half-maximal metabolic rate is
reached) nor the Vmax (maximum metabolic rate) was estimated. In contrast, DCA metabolism
was saturable. Based on this difference in kinetics, Pravacek et al. (1996) suggested that TCA
and DCA might be metabolized through distinct pathways, a finding consistent with other data
demonstrating that the primary metabolic pathway for DCA is nicotinamide adenine dinucleotide
phosphate (reduced) (NADPH) and GSH dependent (e.g., Cornett et al. [1999, 1997]; Lipscomb
et al. [1995]), whereas the primary metabolic pathway for TCA appears to be mediated by
CYP450 pathways. However, an alternative explanation for these data was noted, namely, that
both TCA and DCA share a metabolic pathway that has a lower capacity for DCA.
TCA may be converted to DCA in situ in the gastrointestinal tract of mice, leading to the
question of whether or not this process may influence levels of DCA in blood following
exposure of mice to TCE (which is metabolically transformed to TCA) or TCA itself
(Moghaddam et al., 1997, 1996). Under in vitro anaerobic conditions, microflora from the
cecum of B6C3Fi mice were clearly shown to convert TCA to DCA (Moghaddam et al., 1996).
In contrast, gavage administration of 1,200 mg/kg TCE to control male B6C3Fi mice and to
mice whose gut was depleted of microflora by antibiotic treatment resulted in equivalent
concentrations of DCA and other TCE metabolites (TCA, chloral hydrate, and trichloroethanol)
in blood and liver (Moghaddam et al., 1997). These results suggest that metabolic formation of
DCA by gut microflora does not influence circulating levels of DCA. In this study, antibiotic
treatment resulted in large increases, compared with control values, in the total cecum content of
TCA (4- and 9.5-fold at 4 and 8 hours after exposure), trichloroethanol (4.4- and 1.8-fold), and
chloral hydrate (96- and 69-fold) but no significant change in total cecum content of DCA
(93 and 74% of control values at 4 and 8 hours) (Moghaddam et al., 1997). The lack of a large
effect of antibiotic treatment on DCA cecum content in situ, even when TCA levels were
increased by this treatment, suggests that some other pathway may exist (other than conversion
of TCA to DCA) for the appearance of DCA in the cecum of mice exposed to TCE.
In order to determine if TC A-induced lipid peroxidation is due to the formation of radical
intermediates following dehalogenation of TCA by CYP450 enzymes, Austin et al. (1995)
evaluated the effects of pretreating mice with TCA. Male B6C3Fi mice were pretreated with
1,000 mg/L (estimated to be 228 mg/kg-day by the study authors) TCA in drinking water for
15 DRAFT - DO NOT CITE OR QUOTE
-------
14 days and then administered 300 mg/kg TCA, DCA, or an equivalent volume of distilled water
(control) by gavage as an acute challenge. Animals were sacrificed 9 hours following the acute
challenge, and lipid peroxidation, peroxisome proliferation, and TCA-induced changes in phase I
metabolism were measured. Measures of phase I metabolism included: (1) changes in
12-hydroxylation of lauric acid (an assay specific for CYP4A isoform activity, which is believed
to be associated with induction of peroxisome proliferation in rats and mice [Gibson, 1989]);
(2) changes in/>-nitrophenol hydroxylation (an assay specific for CYP2E1 activity);
(3) immunoblot analysis for induction of CYP450 isoforms CYP2E1, CYP4A, CYP1A1/2,
CYP2B1/2, and CYP3A1; and (4) total liver CYP450. Pretreatment with TCA increased
12-hydroxylation of lauric acid, demonstrating an increase in CYP4A activity (and apparently
reflecting a peroxisome proliferation response), whereas/>-nitrophenol hydroxylation was
unchanged, indicating no effect on CYP2E1 activity. Immunoblot analysis, a measure of the
amount of a protein, was consistent with the increase in CYP4A activity. Increased band
intensities on the immunoblot appeared to occur at locations corresponding to those that have
been identified as the CYP4A2 and CYP4A3 isoform bands. Similarly, immunoblot analysis
was consistent with the absence of an effect on CYP2E1 activity and also showed no changes in
CYP1A1/2, 2B1/2, or 3A1 protein levels. TCA pretreatment did not alter the overall amount of
total liver microsomal P450. These data demonstrate that pretreatment of mice with TCA
modifies the lipid peroxidation responses following acute challenge. The study authors
suggested that this modification resulted from activities associated with peroxisome proliferation
and might be related to a shift in the expression of P450 isoforms. The increased levels of
CYP4A in TCA-pretreated mice are consistent with results observed in other studies with other
peroxisome proliferators (Okita and Okita, 1992).
Results from another study with B6C3Fi mice indicated that pretreatment with DCA or
TCA in drinking water at concentrations of 2 g/L for 14 days had very little influence on the
metabolism or kinetics of elimination of single 100 mg/kg gavage doses of [14C]-labeled TCA
(Gonzalez-Leon et al., 1999). Pretreated and control mice showed similar TCA blood
concentration-time profiles. No significant differences in elimination kinetic parameters, such as
volume of distribution, AUC, elimination half time, total body clearance, and renal clearance,
were found between pretreated mice and control mice. The amount of radiolabel exhaled as
CC>2, taken as an index of metabolism of TCA, was also not influenced by pretreatment. These
results provide no evidence that pretreatment with TCA may induce levels of enzymes involved
in the metabolism of TCA or inhibit metabolism of TCA or DCA (Gonzalez-Leon et al., 1999).
In summary, the available data on TCA metabolism in animal studies indicate that:
(1) TCA is not as extensively metabolized as other chlorinated acids, such as DCA (Larson and
Bull, 1992); (2) TCA is metabolically converted to DCA, but levels of DCA in blood, liver, and
urine are low or not detectable, presumably due to rapid metabolic transformation of DCA into
other metabolites (Merdink et al., 2000, 1998; Yu et al., 2000; Xu et al., 1995; Larson and Bull,
16 DRAFT - DO NOT CITE OR QUOTE
-------
1992); (3) the metabolic conversion of TCA to DCA via reductive dehalogenation is likely
catalyzed by CYP450 enzymes through the dichloroacetate radical intermediate (Merdink et al.,
2000); (4) enzymes involved in TCA metabolism are poorly characterized; (5) microbial
metabolism of TCA to DCA in the gut does not appear to influence circulating levels of DCA in
the blood (Moghaddam et al., 1997, 1996); and (6) pretreatment of mice with TCA in drinking
water does not markedly influence (e.g., enhance or inhibit) the metabolism or elimination
kinetics of single challenge doses of TCA (Gonzalez-Leon et al., 1999; Austin et al., 1995).
3.4. EXCRETION
As described previously in Section 3.2, TCA in urine has been used as a biomarker for
exposure to chlorinated solvents, which are metabolized to TCA, or exposure to disinfectant
byproducts. This use is consistent with results from studies of rodents, clearly showing that,
following oral or parenteral exposure to [14C]-labeled TCA, TCA is principally eliminated from
the body as the parent compound in the urine and that elimination of metabolites in the urine,
elimination via the feces, and exhalation of completely metabolized TCA as CC>2 represent minor
routes of elimination (Yu et al., 2000; Xu et al., 1995; Larson and Bull, 1992). For example,
during a 48-hour period following administration of single doses of radiolabeled TCA ranging
from 5 to 100 mg/kg to male F344 rats or male B6C3Fi mice, radioactivity in urine, CC>2, and
feces accounted for about 58-72, 4-8, and 2-4% of the administered dose, respectively (Larson
and Bull, 1992). Non-metabolized TCA accounted for 81-90% of the radioactivity detected in
the urine (Larson and Bull, 1992). Similarly, within 24 hours of intravenous injection of single
doses of 1, 10, or 50 mg/kg radiolabeled TCA into male F344 rats, urinary excretion of
radioactivity accounted for 48, 67, and 84% of the administered doses, respectively, whereas
radioactivity in feces and CC>2 in expired air accounted for 4-8 and 8-12% of the administered
doses, respectively (Yu et al., 2000).
Results from studies that monitored TCA concentration in bile from isolated rat livers
perfused with TCA solution (Toxopeus and Frazier, 2002, 1998) or from dogs given intravenous
doses of TCA (Hobara et al., 1986) indicate that rates of biliary excretion of TCA are low. For
example, when isolated rat livers were perfused for 2 hours with medium containing initial TCA
concentrations of 5 or 50 jiM, excretion of TCA in bile was linear over time and cumulative
excretion was 0.1% of the total dose by the end of the experiment (Toxopeus and Frazier, 1998).
These results are consistent with the findings of low amounts of radioactivity in feces in the
studies with radiolabeled TCA (Yu et al., 2000; Xu et al., 1995; Larson and Bull, 1992).
Studies comparing the relative importance of urinary, fecal, and exhalation routes of
elimination in humans are not available.
Although elimination half-lives for TCA in urine were not reported in the available
animal toxicokinetic studies involving direct exposure to TCA (e.g., Yu et al. [2000]; Schultz et
al. [1999]; Xu et al. [1995]; Larson and Bull [1992]), the consistent finding of >50% of
17 DRAFT - DO NOT CITE OR QUOTE
-------
administered doses being excreted in the urine within 24-hours of dose administration is
consistent with the hypothesis that significant portions of absorbed TCA can be rapidly
eliminated from the body. However, the demonstrations of significant reversible binding of
TCA to plasma proteins (e.g., Lumpkin et al. [2003]; Toxopeus and Frazier [2002, 1998];
Templin et al. [1993]) provide indirect evidence that bound TCA may contribute to TCA
eliminated in the urine over periods of time longer than 24 hours after administration.
Limited support for a relatively slow elimination from the human body of at least some
portion of absorbed TCA comes from a study of urinary TCA excretion in three human subjects
during a 2-week period in which they ingested their normal tap water containing TCA, followed
by a 2-week period in which tap water was replaced with bottled water containing no detectable
TCA (Froese et al., 2002). TCA ingestion from tap water averaged 5.6 ± 3.1, 41 ± 27, and
73 ± 47 |ig/day for the three subjects, reflecting substantial intrasubject and intersubject
variability in daily intakes of TCA from tap water. TCA concentration was measured in first
morning urine samples and normalized to creatinine concentration to adjust for differences in
first morning urine volume. The logarithm of the creatinine-normalized TCA concentration was
plotted against time during the bottled-water period and evaluated for a linear fit. The values for
elimination half-life determined in this way ranged from 2.3 to 3.7 days.
In another study, three male volunteers ingested either 10 mg/kg trichloroethanol (in
water), 3 mg/kg sodium trichloroacetate (in water), or 15 mg/kg chloral hydrate (in gelatin
capsules) (Muller et al., 1974). The trichloroethanol and TCA concentrations in blood and urine
were determined. The half-lives of TCA after ingestion of sodium trichloroacetate,
trichloroethanol, and chloral hydrate were 50.6, 65.4, and 62.5 hours, respectively. Muller et al.
(1974) demonstrated that the longer half-lives of TCA after ingestion of trichloroethanol and
chloral hydrate were due to the prolonged formation of TCA from trichloroethanol, and the
storage of trichloroethanol in the tissues, especially the fatty tissues.
When 15 mg/kg chloral hydrate was administered orally to a volunteer, there was a rapid
increase in trichloroethanol and TCA concentrations, while no chloral hydrate could be
measured. The half-life of trichloroethanol was about 7 hours, while the half-life of TCA was 4-
5 days (Breimer et al., 1974).
Following inhalation exposure of five volunteers to 50 ppm TCE for 6 hours, the half-life
of TCA was found to be 100 hours (Muller et al., 1972). Similarly, when five male volunteers
were exposed to 100 ppm TCE for 2 weeks or 50 ppm TCE for 1 week, the half-lives of TCA
were found to be 85.6 and 99 hours, respectively (Muller et al., 1974).
A study of urinary excretion of TCA following inhalation exposure to tetrachloroethylene
(of which TCA is a metabolite) reported similar urinary elimination half-lives for TCA in
humans. Volkel et al. (1998) exposed three male and three female human subjects and three
male and three female Wistar rats to 10, 20, or 40 ppm tetrachloroethylene for 6 hours via
inhalation and measured metabolites in the urine. Urine was collected at intervals before
18 DRAFT - DO NOT CITE OR QUOTE
-------
exposure, during exposure, and up to 79 hours after beginning exposure. Urine was analyzed by
gas chromatography / mass spectrometry for concentrations of DCA, TCA, and N-acetyl-S-
(trichlorovinyl)-L-cysteine. TCA was the major metabolite recovered in the urine of both
humans and rats. Half-lives of elimination of TCA from urine (estimated from the time course
of TCA concentrations in urine following exposure) were 45.6 ± 2.5 hours in humans and 11.0 ±
1.2 hours in rats. It is uncertain if the apparent difference in elimination half-lives between
humans and rats was due to species differences in rates of conversion of tetrachloroethylene to
TCA, species differences in other processes more directly related to the appearance of TCA in
the urine, or some other physiological difference between rats and humans.
In contrast to the relatively slow urinary excretion of TCA after cessation a 2-week
exposure to tap water containing TCA (Froese et al., 2002) or cessation of a 6-hour inhalation
exposure to perchloroethylene (Volkel et al., 1998), rapid urinary elimination kinetics of TCA
were indicated in humans following exposure to TCA in swimming pool water (Kim and Weisel,
1998). In this study, four subjects (two/sex) walked in the pool for one 30-minute period (dermal
exposure only) or swam (dermal exposure and presumed oral exposure from incidental ingestion
of pool water during swimming) during a separate 30-minute period. TCA levels in the urine
void collected 5-10 minutes after each 30-minute exposure period were elevated and generally
returned to pre-exposure levels within 3 hours after exposure (i.e., were indistinguishable from
pre-exposure levels). The relatively rapid return to pre-exposure levels within 3 hours after
cessation of exposure is consistent with fast elimination kinetics in this study. However, as
discussed in Section 3.1, there was large variability in the pre-exposure levels of TCA in urine2,
limiting the ability of this study to detect differences in pre- and postexposure levels of TCA in
urine.
In summary, results from studies with animals indicate that urinary excretion of TCA is
the principal route of elimination of TCA from the body (Yu et al., 2000; Xu et al., 1995; Larson
and Bull, 1992). Other minor routes of elimination include urinary elimination of metabolites,
exhalation of completely metabolized TCA as CC>2, and excretion of TCA in the bile or feces
(Toxopeus and Frazier, 2002, 1998; Yu et al., 2000; Xu et al., 1995; Larson and Bull, 1992;
Hobara et al., 1986). Although data on the kinetics of urinary elimination of TCA are limited,
there are estimates that the half-life of TCA in urine from human subjects may be on the order of
2-3 days (Froese et al., 2002; Volkel et al., 1998). These findings are consistent with the idea
that reversible binding of TCA to plasma proteins may influence the delivery of TCA to target
tissues and prevent faster elimination of absorbed TCA in the urine.
2Pre-exposure amounts of TCA in urine ranged from 155 to 1,183 ng, whereas postexposure amounts ranged from
294 to 15,990 ng (Kim and Weisel, 1998).
19 DRAFT - DO NOT CITE OR QUOTE
-------
3.5. PHYSIOLOGICALLY BASED PHARMACOKINETIC MODELS
Physiologically based pharmacokinetic (PBPK) models have not been developed for
TCA. A PBPK model for TCE in humans (Fisher et al., 1998; Allen and Fisher, 1993) included
a TCA compartment to account for metabolism of TCE. Fisher et al. (1998) concluded that
further research is needed to explain the observed variability in urinary excretion of
trichloroethanol glucuronide and TCA and the metabolic pathway resulting in the formation of
DCA.
20 DRAFT - DO NOT CITE OR QUOTE
-------
4. HAZARD IDENTIFICATION
4.1. STUDIES IN HUMANS
4.1.1. Oral Exposure
No human epidemiology studies that evaluated TCA alone were located. Most of the
human health data for chlorinated acetic acids concern components of complex mixtures of water
disinfectant byproducts. These complex mixtures of disinfectant byproducts have been
associated with increased potential for bladder, rectal, and colon cancer in humans (reviewed by
Boorman et al. [1999]; Mills et al. [1998]) and adverse effects on reproduction (reviewed by
Nieuwenhuijsen et al. [1999]; Mills et al. [1998]).
Most of the studies of human health effects following exposure to water disinfectant
byproducts have used trihalomethanes and haloacetic acid concentrations as the exposure metric
(Hinckley et al., 2005; King et al., 2005; Porter et al., 2005). These studies are not evaluated in
this review as data on exposure to mixtures cannot be applied to the individual components of
the mixture.
No clinical studies of the effects of oral or inhalation exposure of humans to TCA were
located.
4.1.2. Dermal Exposure
Identified case reports demonstrate the corrosive potential of TCA to human skin.
Depending on concentration and duration of contact, TCA can denature and precipitate protein.
This characteristic has been used clinically in chemical skin peeling treatments for many years.
TCA at concentrations ranging from 15 to 35% has been used in skin peeling treatments to treat
conditions such as actinic damage, scars, wrinkles, and dyspigmentation (Cotellessa et al., 2003;
Lee et al., 2002; Coleman, 2001; Kang et al., 1998; Chiarello et al., 1996; Moy et al., 1996; Tse
et al., 1996; Witheiler et al., 1996; Rubin, 1995). Concentrations of >45% have an increased risk
of causing scarring. The skin peeling procedure results in a pink erythema and swelling for the
first few days posttreatment and is followed by exfoliation of the dead skin. Histological studies
(Moy et al., 1996; Tse et al., 1996) indicate that the TCA-induced skin damage is characterized
by epidermal loss, early inflammatory response, and collagen degeneration. Adverse side effects
or complications resulting from these treatments are uncommon (Fung et al., 2002; Coleman,
2001) and are usually mild in severity (Fung et al., 2002). Reported side effects in patients
receiving the skin peel procedure have included infection (Coleman, 2001), persistent
(>1 month) erythema (Al-Waiz and Al-Sharqi, 2002; Coleman, 2001), transient hyper-
pigmentation (Fung et al., 2002; Lee et al., 2002; Coleman, 2001), acne or cyst formation (Lee et
21 DRAFT - DO NOT CITE OR QUOTE
-------
al., 2002; Coleman, 2001), keratoacanthomas3 (Cox, 2003), and fine crusting (Kim et al., 2002).
One case was reported where a 35% TCA solution inadvertently entered the eye of a patient
receiving a dermal peel, resulting in marked conjunctivitis and abrasions that involved 25% of
the cornea (Fung et al., 2002). Complete corneal healing was reported within 72 hours of
initiation of supportive care, and no lasting effects were evident, suggesting that the response to
TCA was reversible under the reported exposure conditions.
Nunns and Mandal (1996) reported two cases of inflammation of the vulva caused by the
use of TCA in topical treatments of genital warts. The surface of each wart was coated with
TCA (concentration not reported). Initially, the patients complained of burning, which was
short-lived. After a second TCA treatment a week later, the patients reported continual soreness
or burning. On clinical examination, marked erythema and tenderness in the vulvar and
vestibular areas were noted. The symptoms in these patients lasted for 2-15 weeks. Wilson et
al. (2001) did not report any adverse side effects in patients (n = 95) treated for genital warts
with either TCA, cryotherapy, or electrocautery (number of patients treated with TCA was not
reported); however, the study was not specifically designed to identify adverse side effects in
treated patients.
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION
4.2.1. Short-term and Subchronic Studies
4.2.1.1. Oral
4.2.1.1.1. Rats. Short-term and subchronic (<90 days) oral exposure studies are summarized in
Table 4-1. Mather et al. (1990) evaluated toxicological effects in male Sprague-Dawley rats
(10/dose group) dosed with neutralized TCA in drinking water at concentrations of 0, 50, 500, or
5,000 ppm (approximately 0, 4.1, 36.5, or 355 mg/kg-day) for 90 days. Animals were weighed
at the beginning of the study and at the time of necropsy. Blood was collected at the time of
sacrifice for clinical chemistry analysis (blood urea nitrogen, creatinine, glucose, alanine
aminotransferase [ALT], alkaline phosphatase [ALP], cholesterol, total protein, albumin,
calcium, phosphorus, creatine phosphokinase [CPK], and gamma-glutamyl transferase [GOT]).
In addition, the following immune function parameters were evaluated: antibody production,
delayed hypersensitivity, natural killer cell cytotoxicity, and production of prostaglandin E2 and
interleukin (IL)-2. Hepatic peroxisomal and microsomal enzyme induction was also assessed.
At sacrifice, a complete necropsy was performed, and the liver, kidneys, and spleen were
weighed.
3Keratoacanthomas are round, firm, usually flesh-colored growths with a central crater that is scaly or crusted.
22 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1. Summary of acute, short-term, and subchronic studies evaluating effects of TCA after oral administration
in rats and mice
Reference
Species
Exposure
route
Exposure
duration
Doses evaluated
Effects3
NOAEL
(mg/kg-d)
LOAEL
(mg/kg-d)
Comments
Rats
Goldsworthy
and Popp,
1987
DeAngelo et
al., 1989
Davis, 1990
Mather et al.,
1990
F344 rats
(males, 5-
6/group)
Sprague-
Dawley,
F344, and
Osborne-
Mendel rats
(males,
6/group/
strain)
Sprague-
Dawley rats
(6/sex/
dose)
Sprague-
Dawley rats
(males,
10/dose)
Oral,
gavage
Oral,
drinking
water
(A) Oral,
drinking
water
(B) Oral,
gavage
Oral,
drinking
water
10 d
14 d
(A) 14 d
(B) Three
doses over
24hrs
90 d
0 or 500 mg/kg-d
in corn oil
0, 212, 327, or
719 mg/kg-d
(A) 5.2-
309 mg/kg-d
(B) 0,0. 15, or
0.4 mg/kg in
water, neutralized
with sodium
hydroxide
0,4.1, 36.5, or
355 mg/kg-d
Hepatic and renal peroxisome
proliferation, increased relative liver
weight
Hepatic peroxisome proliferation
induction (Osborne-Mendel and
F344 rats)
(A) Limited endpoints were
monitored. No effects were
observed on weight gain, urine
volume, and osmolarity, plasma
glucose, and liver lactate levels
(B) Decreased plasma and liver
lactate levels
Increased absolute spleen weight;
increased relative liver and kidney
weights; increased liver, kidney, and
spleen sizes; peroxisome
proliferation
Not
determined
327
(A) Not
determined
(B) Not
determined
36.5
500
719
(A) Not
determined
(B)0.15
355
The cyanide-insensitive
PCO activity assay was
used to measure the
peroxisome proliferative
response. Liverbody
weight ratio was
significantly increased.
Peroxisome proliferation
was observed only in
Osborne-Mendel and
F344 rats. These results
suggest that Sprague-
Dawley rats were the least
sensitive of the three
strains evaluated to
peroxisome proliferation.
(B) At 0.15 mg/kg,
plasma glucose levels
were also decreased in
females. These results are
consistent with effects on
intermediary carbohydrate
metabolism. Similar
effects were not observed
in the 14-d study (A).
23
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1. Summary of acute, short-term, and subchronic studies evaluating effects of TCA after oral administration
in rats and mice
Reference
Bhatetal.,
1991
Acharyaetal.,
1997, 1995
Celik, 2007
Species
Sprague-
Dawley rats
(males,
5/group)
Wistar rats
(males, 5-
6/dose)
Sprague-
Dawley rats
(female)
Exposure
route
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
Exposure
duration
90 d
lOwks
50 d
Doses evaluated
0 or 825 mg/kg-d
0 or 3.8 mg/kg-d
0 or 300 mg/kg-d
Effects3
Decreased body weight gain, minor
changes in liver morphology,
collagen deposition, perivascular
inflammation of the lungs
Decreased terminal body weight,
liver and kidney histopathologic
changes, increased glycogen,
changes in liver lipid and
carbohydrate homeostasis, decreased
kidney GSH
Increase in serum AST, ALT, CPK,
and ACP activities; increase in SOD
and catalase activities in brain, liver,
and kidney tissues
NOAEL
(mg/kg-d)
Not
determined
Not
determined
Not
determined
LOAEL
(mg/kg-d)
825
3.8
300
Comments
/4oftheLD50
(3,300 mg/kg) was
administered daily.
Doses were estimated
based on default drinking
water intake values for
rats. 3.8 mg/kg-d is
judged as an equivocal
LOAEL because the
observed severity of the
observed liver changes
was considered minimal.
Mice
Goldsworthy
and Popp,
1987
DeAngelo et
al., 1989
B6C3FJ
mice
(males,
7-8/group)
B6C3F1;
C3H,
Swiss-
Webster,
C57BL/6
mice
(males,
6/group)
Oral,
gavage
Oral,
drinking
water
10 d
14 d
0 or 500 mg/kg-d
in corn oil
0,261, or
442 mg/kg-d
Induction of hepatic and renal
peroxisome proliferation; increased
relative liver weight
Increased relative liver weight,
peroxisome proliferation (PCO
activity)
Not
determined
Not
determined
500
261
Cyanide-insensitive PCO
activity assay was used to
measure the proliferative
response. Liverbody
weight ratio significantly
increased.
C57BL/6 mice were more
sensitive than the other
strains to peroxisome
proliferation.
24
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1. Summary of acute, short-term, and subchronic studies evaluating effects of TCA after oral administration
in rats and mice
Reference
Sanchez and
Bull, 1990
Dees and
Travis, 1994
Austin etal.,
1995
Austin etal.,
1996
Species
B6C3FJ
mice
(males,
12/group)
B6C3FJ
mice (5/sex/
dose)
B6C3FJ
mice
(males,
6/group)
B6C3FJ
mice
(males,
6/group)
Exposure
route
Oral,
drinking
water
Oral,
gavage
(A) Oral,
drinking
water
(B) Oral,
gavage
Oral,
gavage
Exposure
duration
14 d
lid
(A) 14 d
(B) Single
dose
Single dose
Doses evaluated
0, 75, 250, or
500 mg/kg-d
0, 100, 250, 500,
or 1,000 mg/kg-d
in corn oil
(A) 0 or
250 mg/kg-d
(B) 0 or
300 mg/kg in
distilled water,
pH adjusted to
7.0 with 5 N
NaOH
0, 30, 100, or
300 mg/kg in
water,
pH adjusted to 7
using 5 N NaOH
Effects3
Increased liver weight; hepatocyte
proliferation (DNA labeling)
Increased absolute and relative liver
weight; increased hepatocyte
labeling
(A) Increased relative liver weight
(B) Decreased TBARSb; increased
PCO, catalase, and CYP4A activities
Oxidative stress (increased 8-OHdG°
levels)
NOAEL
(mg/kg-d)
75
Not
determined
Not
determined
Not reported
LOAEL
(mg/kg-d)
250
100
250
Not reported
Comments
Doses were estimated
based on default drinking
water intake values for
male B6C3Fi mice. At
500 mg/kg-d, there was a
slightly increased
hepatocyte diameter
because of increased
glycogen deposition.
(A) Doses were estimated
based on default drinking
water intake values for
male B6C3F! mice.
(B) Acute administration
occurred after a 14-d
pretreatment period.
Doses were estimated
based on default drinking
water intake values for
male B6C3F! mice;
8-OHdGc levels at 30 or
100 mg/kg were not
reported.
25
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1. Summary of acute, short-term, and subchronic studies evaluating effects of TCA after oral administration
in rats and mice
Reference
Parrishetal.,
1996
Kato-
Weinstein et
al., 2001
Laughter et
al., 2004
Species
B6C3FJ
mice
(males,
6/group)
B6C3FJ
mice
(males,
5/dose)
SV129
wild-type
mice;
PPARd
a-null mice
(males,
3-5/group)
Exposure
route
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
Exposure
duration
3 or 10 wks
(A) 4 or
8 wks
(B) 12 wks
7d
Doses evaluated
0, 25, 125, or
500 mg/kg-d
(A) 750 mg/kg-d
(B) 0, 75, 250, or
750 mg/kg-d
0, 62.5, 125, 250,
or 500 mg/kg-d
Effects3
Increased absolute and relative liver
weights; peroxisome proliferation
(increased PCO activity and
increased 12-hydroxylation of lauric
acid)
Increased absolute and relative liver
weights; decreased liver glycogen
content
Induction of markers of peroxisome
proliferation in wild-type but not
PPARd a-null mice at 2.0 g/L;
induction of CYP4A at 1.0 g/L.
Wild-type mice receiving the high
dose exhibited centrilobular
hepatocyte hypertrophy
NOAEL
(mg/kg-d)
25
Not
determined
125
LOAEL
(mg/kg-d)
125
75
250
Comments
Doses were estimated
based on default drinking
water intake values for
male B6C3Fi mice;
results were similar for
the 3- and 10-wk
evaluations; 8-OHdGc
levels were not affected
by TCA.
Doses were estimated
based on default drinking
water intake values for
male B6C3Fi mice.
Doses were estimated
based on default drinking
water intake values for
male B6C3F! mice.
aThe effects listed in this table may have occurred either at the LOAEL or at higher doses.
bTBARS = thiobarbituric acid-reactive substances.
°8-OHdG = 8-hydroxy-2'-deoxyguanosine.
dPPAR = Peroxisome proliferator-activated receptor.
ACP = acid phosphatase; AST = aspartate aminotransferase; LD50 = median lethal dose; LOAEL = lowest-observed-adverse-effect level; NOAEL = no-observed-
adverse-effect level
26
DRAFT - DO NOT CITE OR QUOTE
-------
Histopathologic examination was conducted on the brain, heart, lungs, kidneys, spleen,
thymus, pancreas, adrenals, testes, lymph nodes, gastrointestinal tract, urinary bladder, muscle,
and skin. TCA administration did not affect body weight at any dose. At 355 mg/kg-day,
relative liver and kidney weights were significantly (p < 0.05) increased (7 and 11%,
respectively) compared with controls. At the high dose, hepatic peroxisomal enzyme activity
was significantly (15%, p < 0.05) increased (as measured by palmitoyl-CoA oxidase [PCO]
activity). The liver, spleen, and kidney of high-dose animals were enlarged; however, no
microscopic lesions were observed at any dose. No consistent treatment-related effects were
seen on clinical chemistry or immune function parameters. EPA determined that the no-
observed-adverse-effect level (NOAEL) for this study was 36.5 mg/kg-day and the lowest-
observed-adverse-effect level (LOAEL) was 355 mg/kg-day, based on increased liver size and
weight and peroxisome proliferation as well as statistically significantly increased kidney weight
and size and increased spleen size.
In a subchronic study Bhat et al. (1991) administered 1A of a median lethal dose (LDso) of
TCA, DC A, or MCA in drinking water to male Sprague-Dawley rats (five/dose) for 90 days.
Based on the reported LD50 of 3,300 mg/kg for TCA, 1A of this value would correspond to an
administered dose of approximately 825 mg/kg-day. Body weights were monitored throughout
the study. The animals were sacrificed after 90 days of exposure, and the liver, lung, heart,
spleen, thymus, kidney, testes, and pancreas were removed and weighed. These organs and the
brain were microscopically examined. Liver sections were also stained for collagen deposition.
No other toxicity parameters were evaluated. TCA exposure resulted in a significant depression
(17%, p < 0.0001) of body weight gain throughout the exposure period. Toxicologically
significant changes in liver weight were not observed. Exposure to TCA induced minimal to
moderate collagen deposition (an indication of liver injury) in portal triads and large central
veins in 4/5 animals (minimal collagen deposition was observed in 1/5 controls). Morphologic
changes in the liver included portal vein dilation/extension of minimal to moderate severity in
5/5 TCA-treated animals. Perivascular inflammation of the lungs occurred at unspecified
incidences. EPA determined that the only dose tested in this study, 825 mg/kg-day, was a
LOAEL based on significantly reduced body weight gain.
In a 50-day drinking water study (Celik, 2007), 4-month-old female Sprague-Dawley rats
were administered 2,000 ppm (300 mg/kg-day, assuming a default water intake of 0.15 L/kg-
day) TCA (numbers unknown), while the control group received natural spring water. At the
end of the study, blood samples were collected. Animals were sacrificed, and brain, liver, and
kidney samples were obtained. Serum marker enzymes (aspartate aminotransferase [AST], ALT,
CPK, acid phosphatase [ACP], ALP, and LDH), erythrocytes and tissue antioxidant defense
systems (GSH, GSH reductase, superoxide dismutase [SOD], GST catalase), and
malondialdehyde (MDA) (a product of lipid peroxidation) were measured.
27 DRAFT - DO NOT CITE OR QUOTE
-------
TCA significantly increased serum AST, ALT, CPK, and ACP activity (p < 0.05) in
treated rats. A slight but insignificant increase in MDA was found in the erythrocytes and liver.
The antioxidant enzymes, SOD and catalase, were significantly increased in the brain, liver, and
kidney. However, no changes in GSH, GSH reductase, or GST activities were found in any
tissue. Celik (2007) concluded that elevated serum marker enzymes probably resulted from
damage to liver cells by TCA and subsequent leakage of the enzymes into plasma and that the
increases in SOD and catalase activities in the tissues after TCA treatment were probably due to
increased generation of reactive oxygen species.
Acharya et al. (1995) evaluated liver and kidney toxicity of TCA as part of a study on the
interactive toxicity of tertiary butyl alcohol and TCA. Young male Wistar rats (50 days old)
(five to six/dose) were exposed to water containing 0 or 25 ppm or approximately 0 or
3.8 mg/kg-day, assuming a default water intake of 0.15 L/kg-day TCA (U.S. EPA, 1988) for
10 weeks. Animals were weighed weekly during treatment, and food and water consumption
were recorded daily. Blood was taken from animals after the 10-week exposure, and the
following parameters were evaluated: succinate dehydrogenase, ALP, ACP, AST, ALT, and
serum triglyceride, cholesterol, and glucose levels. In addition, glycogen, triglyceride,
cholesterol, GSH, lipid peroxidation, and diene conjugation were determined in liver
homogenates. Microscopic examination of tissues was not performed.
In animals treated only with TCA, terminal body weight was decreased by approximately
17% in the absence of changes in food consumption (data not shown). Little, if any,
TCA-induced liver toxicity was observed. Relative liver weight did not differ significantly in
TCA-treated animals. No significant changes were detected in AST, ALT, ALP, or ACP. In
contrast to the serum markers of liver necrosis, indicators of lipid and carbohydrate homeostasis
were affected by TCA. Succinate dehydrogenase activity was increased by roughly 30%
compared with controls. Liver triglyceride and cholesterol levels were significantly decreased,
while liver-glycogen levels were increased approximately eightfold. Serum cholesterol levels
were also increased approximately twofold. The study authors suggested that this profile of
carbohydrate and lipid changes was consistent with the onset of hepatomegaly, which would
increase the energy demands of the liver and activate succinate dehydrogenase, leading to
increased oxidative phosphorylation and mobilization of lipids (decreased liver triglyceride and
cholesterol). There was little evidence for induction of oxidative stress in the liver. Kidney, but
not liver, GSH levels were decreased to approximately 66% of control values and no increase in
lipid peroxidation was observed in the liver.
In a follow-up study using the same exposure protocol (Acharya et al., 1997),
histopathologic changes in the liver and kidney were evaluated. The study authors noted that
minimal hepatic alterations were observed in the TCA treatment group, indicating that the
3.8 mg/kg-day dose was marginally toxic. Liver histopathologic changes that were noted
included centrilobular necrosis, hepatocyte vacuolation, loss of hepatic architecture, and
28 DRAFT - DO NOT CITE OR QUOTE
-------
hypertrophy of the periportal region. Incidence and severity data were not reported for these
lesions. Hypertrophy of the periportal region observed in the latter study may have accounted
for the observed marginal increase in liver weight in the former study. The magnitude of the
severity of these changes was reportedly small (the magnitude of the response could not be
accurately quantified from the reported figures) and is consistent with the absence of effects on
serum-liver enzymes in the earlier study (Acharya et al., 1995).
Histopathologic changes were also noted in the kidneys of TCA-treated animals and
included degeneration of renal tubules with syncytial arrangement of the nucleus in the epithelial
cells, degeneration of the basement membrane of Bowman's capsule, diffused glomeruli,
vacuolation of glomeruli, and renal tubular proliferation in certain areas (incidence and severity
not reported). Based on the liver and kidney histopathologic changes at the single dose tested,
the study authors (Acharya et al., 1997) indicated that TCA is a liver and kidney toxicant.
Taken together, the two studies (Acharya et al., 1997, 1995) suggest that the single dose
tested, 3.8 mg/kg-day, is an apparent LOAEL. However, a number of questions regarding these
studies preclude a definitive determination of the LOAEL. First, Acharya et al. (1995) noted a
lack of increase in liver enzyme activity. Although liver histopathologic changes were observed,
they were described as "only marginal" by the authors. The authors did not discuss the severity
of the histopathologic changes in relation to untreated controls, and no incidence data were
provided. Therefore, it is not clear whether the effects observed at the single TCA-only dose that
was evaluated were adverse. Due to this uncertainty, EPA determined that 3.8 mg/kg-day could
be best described as an equivocal LOAEL. It should be noted that Wistar rats were actually
more sensitive than mice to increases in cyanide insensitive acyl-CoA oxidase (AGO) activity by
TCA(Elcombe, 1985).
Davis (1990) investigated the effects of TCA on weight gain, urine volume and
osmolality, and plasma glucose and liver lactate levels in Sprague-Dawley rats in a 14-day study.
Groups of rats (six animals/sex and dose group) received TCA in drinking water at
concentrations of 0, 0.04, 0.16, 0.63, or 2.38 g/L (equivalent to approximate dose levels of 0, 5.2,
20.8, 81.9, or 309 mg/kg-day, based on a water consumption factor of 0.13 L/kg-day for
Sprague-Dawley rats from U.S. EPA [1988]). High-dose rats consumed less food and water and
lost weight during the first few days of exposure. Weight gain was similar to that in controls at
subsequent time points. Urine volume and osmolality were not affected except for a temporary
lesser increase in osmolality to match the decrease in urine volume on day 7 in high-dose
females. No clearly adverse effects or dose-related trends were demonstrated. No effects on
plasma glucose or liver lactate levels occurred after the 14-day exposure period. EPA has not
determined the NOAEL for Davis (1990) since limited endpoints were monitored in this study.
Additional information collected by Davis (1990) suggests that TCA may have transient
effects on plasma glucose and plasma and liver lactate levels in rats. Three gavage doses of
0.92 or 2.45 |imol/kg TCA (approximately 0.15 and 0.40 mg/kg, respectively) were administered
29 DRAFT - DO NOT CITE OR QUOTE
-------
to Sprague-Dawley rats (five/sex/dose) over 24 hours. The rats were killed 3 hours after the last
dose. Significantly reduced plasma (45%) and liver lactate (48%) levels were observed in
females. Plasma lactate level was significantly reduced in males (30%) at the high dose. Plasma
glucose level was significantly reduced (25%) in females given the high dose. These data
suggest that TCA can affect intermediary metabolism, although the absence of effects on plasma
lactate or glucose levels in the 14-day study conducted by Davis (1990) suggests that the effect
may be transient.
The ability of TCA to induce peroxisome proliferation and oxidative stress has been
evaluated in a number of studies. Goldsworthy and Popp (1987) investigated the ability of TCA
to induce hepatic and renal peroxisome proliferation (as assessed by the cyanide-insensitive PCO
activity assay) in adult male F344 rats (five to six/dose) given 0 or 500 mg/kg-day TCA in corn
oil via gavage for 10 consecutive days. Toxicological parameters other than liver and kidney
weights were not evaluated. Hepatic peroxisomal enzyme activity increased significantly
(p < 0.05) in rats receiving TCA, resulting in levels of enzyme activity approximately 2.8-fold
greater than in controls. Liver-to-body-weight ratios were also significantly (41%, p < 0.05)
increased relative to those in controls. Body weight gain was not changed. Renal peroxisomal
enzyme activity was significantly (p < 0.05) increased by approximately 1.8-fold over that in
controls in rats. Kidney weights were not affected by treatment. This study demonstrated that
TCA treatment induced peroxisome proliferation in the livers and kidneys of male F344 rats.
Elcombe (1985) demonstrated species differences in peroxisome proliferation after TCA
treatment in vivo and in vitro. Male Wistar rats and male Swiss mice were administered 10-
200 mg/kg-day TCA in corn oil by gavage for 10 consecutive days. Control animals received
10 mL/kg corn oil vehicle alone. The animals were sacrificed 24 hours following the final dose,
and the livers were excised and homogenized. Liver catalase activity and cyanide insensitive
palmitoyl CoA oxidation (a peroxisomal p-oxidation marker) were determined
spectrophotometrically.
In a separate study, Elcombe (1985) isolated hepatocytes from rats and mice. The
isolated cells were seeded in a tissue culture flask and incubated at 37°C. Human hepatocytes
were prepared from liver obtained from brain-dead renal transplant donors. TCA (up to a
noncytotoxic concentration of 5 mM), dissolved in N,N-dimethylformamide, was added to the
monolayer cultures at each 24-hour medium change. Ninety-six hours after seeding, the
hepatocytes were harvested. Protein content and cyanide insensitive palmitoyl CoA oxidation in
the cell homogenate were determined.
Dose-related increases in cyanide insensitive palmitoyl CoA oxidation were observed in
rats and mice after TCA treatment. At doses of 200 mg/kg-day TCA for 10 days, 6.5-fold
(Wistar rat) and 4.8-fold (Swiss mouse) increases in peroxisomal P-oxidation were observed.
Peroxisome volume densities were increased concomitantly with P-oxidation activity. On the
other hand, TCA had no effect on hepatic catalase activity.
30 DRAFT - DO NOT CITE OR QUOTE
-------
Dose-related increases in cyanide insensitive palmitoyl CoA oxidation were also
observed in cultured rat and mouse hepatocytes exposed to TCA (Elcombe, 1985). No
stimulation of peroxisomal p-oxidation was observed, however, in cultured human hepatocytes
prepared from two human liver samples and treated with TCA.
DeAngelo et al. (1989) conducted a series of experiments in three strains of rats and four
strains of mice to determine relative species and strain sensitivities to the induction of hepatic
peroxisome proliferation by chloroacetic acids (results of the mouse studies are described later in
this section). Male Sprague-Dawley, F344, and Osborne-Mendel rats (six/dose/strain) received
drinking water supplemented with 0, 6, 12, or 31 mM (approximately 0, 212, 327, or 719 mg/kg-
day as calculated by the study authors) for 14 days. Hepatic PCO activity was used to assess
peroxisome proliferation in all three strains. Carnitine acetyl-CoA transferase activity (another
peroxisomal enzyme marker) was determined only in Sprague-Dawley rats, and induction of the
peroxisome proliferation-associated protein was evaluated in high-dose Sprague-Dawley rats.
Morphometric analysis of peroxisome proliferation was conducted by electron microscopy on
liver sections from two high-dose Sprague-Dawley rats. No other toxicological parameters were
evaluated.
TCA treatment did not significantly affect body weights or liver-to-body-weight ratios in
either Osborne-Mendel or F344 rats. The final mean body weight of Sprague-Dawley rats was
significantly reduced at 719 mg/kg-day when compared with controls (16% reduction). No
effects were seen on liver-to-body-weight ratios in any of the strains. PCO activity was elevated
in Osborne-Mendel rats by 2.4-fold and in F344 rats by 1.6-fold over control values at the high
dose. In contrast, PCO activity was not affected in treated Sprague-Dawley rats at any dose.
Carnitine acetyl-CoA transferase activity, however, was increased by 321% above the controls in
Sprague-Dawley rats at the high dose (significant increases were not observed at lower doses),
but the volume fraction of cytoplasm from hepatic tissue occupied by peroxisomes was
decreased to less than half that seen in controls in this strain. The reason for this paradoxical
effect was not addressed. Taken together, these observations suggest that Sprague-Dawley rats
are not sensitive to peroxisome proliferation in response to TCA exposure under the
experimental conditions tested. EPA determined that the NOAEL and LOAEL values for
peroxisome proliferation were 327 and 719 mg/kg-day, respectively, in both Osborne-Mendel
and F344 rats.
Collectively, the data in rats suggest that short-term exposure to TCA primarily affects
the liver, although effects on the kidneys and lungs have also been observed. Liver effects have
included increased size and weight, collagen deposition, indications of altered lipid and
carbohydrate metabolism, and peroxisome proliferation. Effects were observed at doses as low
as 0.45 mg/kg-day (decreased liver and plasma lactate levels) (Davis, 1990). Strain differences
were also evident. An equivocal LOAEL of 3.8 mg/kg-day (liver and kidney pathology) was
identified in 10-week studies in Wistar rats (Acharya et al., 1997, 1995). In a 90-day study
31 DRAFT - DO NOT CITE OR QUOTE
-------
(Mather et al., 1990), a higher LOAEL of 355 mg/kg-day (increase in liver and kidney weight
and peroxisome proliferation) was identified in Sprague-Dawley rats.
4.2.1.1.2. Mice. Short-term and subchronic studies in mice are summarized in Table 4-1. The
available studies in mice have primarily been conducted to evaluate TCA-induced effects on the
liver and the mode of action (MOA) underlying hepatic effects. No toxicity studies that
evaluated a complete suite of toxicological parameters (e.g., body weight, clinical pathology,
gross pathology, and microscopic pathology of a comprehensive set of tissues) in mice were
located.
Goldsworthy and Popp (1987) investigated the ability of TCA to induce hepatic and renal
peroxisome proliferation as assessed by the cyanide-insensitive PCO activity assay in adult male
B6C3Fi mice (seven to eight/dose) given 0 or 500 mg/kg-day in corn oil for 10 days via gavage.
Relative liver and kidney weight were the only other toxicological parameters evaluated.
Hepatic peroxisomal enzyme activity increased significantly (p < 0.05) in mice receiving TCA,
resulting in levels of enzyme activity that were 280% those of the controls. Renal peroxisomal
enzyme activity was significantly (p < 0.05) increased to 305% of control levels in mice. Liver-
to-body weight ratios were also significantly increased (40%; p < 0.05) relative to controls.
DeAngelo et al. (1989) investigated the effects of TCA exposure on hepatic peroxisome
proliferation by using four strains of male mice (B6C3Fi, C3H, Swiss-Webster, and C57BL/6).
Groups of six mice per strain and dose were exposed to TCA in drinking water that contained 0,
12, or 31 mM (approximately 0, 261, or 442 mg/kg-day) TCA for 14 days. No effects were seen
on body weight, but liver-to-body-weight ratios were significantly increased at both dosages in
all four strains. The activity of PCO was elevated in all four strains for all TCA dose groups.
PCO levels were 276, 325, and 456% above controls at 12 mM and 648, 644, and 678% above
controls at 31 mM for Swiss-Webster, C3H, and B6C3Fi mice, respectively. PCO activity in
C57BL/6 mice was increased by 2,100 and 2,500% above control levels at the high and low
doses for TCA, respectively, indicating that this is a particularly sensitive strain of mouse.
In another phase of this study, catalase activity was increased by 461% above controls in
B6C3Fi mice at the high dose, with accompanying increases in the level of peroxisome
proliferation-associated protein and number and size of peroxisomes in liver cytoplasm. The
results indicated that mice, in general, are more sensitive than rats to the effects of TCA on
peroxisome proliferation, as indicated by PCO activity. As described previously, levels of PCO
activity in F344 and Osborne-Mendel rats were increased only by approximately 63 and 138%,
respectively, at an approximate TCA dose of 719 mg/kg-day, and no significant effects on PCO
activity occurred at 327 mg/kg-day in any strain. No effects were seen on this parameter in
Sprague-Dawley rats at any dose (DeAngelo et al., 1989).
Miyagawa et al. (1995) tested for acute toxicity as part of a dose-range finding study on
hepatocyte replicative DNA synthesis for 41 putative Ames-negative mouse hepatocarcinogens.
32 DRAFT - DO NOT CITE OR QUOTE
-------
Groups of male B6C3Fi mice (four or five/dose) were administered a single gavage dose of TCA
to determine the maximum tolerated dose, which was set at about half the LD50. The maximum
tolerated dose for TCA was estimated to be 500 mg/kg.
Several studies have evaluated the ability of TCA to induce oxidative stress in the liver of
treated mice. These studies range from single-dose studies to 10-week studies. In an acute study
by Austin et al. (1996), male B6C3Fi mice (six/group) were treated with a single oral dose of
TCA (0, 30, 100, or 300 mg/kg) in water, adjusted to pH 7 using 5 N NaOH. Mice were
deprived of food for 3 hours prior to dosing. Liver nuclear DNA was extracted to assess
increases in 8-hydroxy-2'-deoxyguanosine (8-OHdG) adducts, a measure of oxidative damage to
DNA resulting from oxidative stress. TCA has been shown to induce lipid peroxidation in
rodents (Larson and Bull, 1992), and compounds that produce oxidative stress also increase
8-OHdG, which is capable of inducing DNA base transversions that might be involved in the
carcinogenic process (Chang et al., 1992). A significant increase in 8-OHdG in nuclear DNA in
the liver was observed in the 300 mg/kg group at 8-10 hours post-dosing. The maximum
8-OHdG level was observed at 8 hours and was an increase of approximately 33% (estimated
from Chang et al. [1992], Figure 3) over controls. The 8-OHdG levels in groups dosed with
30 or 100 mg/kg were not reported.
Austin et al. (1996) reported that the maximum concentration of TCA-induced
thiobarbituric acid-reactive substances (TEARS) (an indicator of lipid peroxidation) occurred in
the liver of mice 9 hours after dosing. In an earlier study, Larson and Bull (1992) also reported
that the maximum concentration of TEARS occurred at 9 hours post-dosing in the livers of mice
given 2,000 mg/kg TCA. The Larson and Bull (1992) study reported that a single oral dose of
TCA 9 hours after dosing induced TEARS levels 1.15-, 1.7-, 2-, and 2.7-fold over controls at
100, 300, 1,000, and 2,000 mg/kg, respectively. Austin et al. (1996) suggested that the ability of
haloacetates to increase both TEARS and 8-OHdG levels indicates that oxidative stress may be
related to their hepatocarcinogenicity. The concordance between TEARS and 8-OHdG levels
also suggested a common mechanism of induction of these two markers. Neither a NOAEL nor
a LOAEL were identified for Austin et al. (1996) because no other measures of liver or systemic
toxicity were reported.
Parrish et al. (1996) evaluated the ability of haloacetic acids to induce oxidative DNA
damage in the livers of mice. Male B6C3Fi mice (six/group) were exposed to 0, 100, 500, or
2,000 mg/L TCA in drinking water for either 3 or 10 weeks. The study authors did not estimate
the average daily doses resulting from exposure to these concentrations. Based on default water-
intake values of 0.25 L/kg-day for male B6C3Fi mice (U.S. EPA, 1988), the corresponding
doses were approximately 0, 25, 125, or 500 mg/kg-day. Body weight and liver weight were
evaluated. Several indicators for peroxisome proliferation were measured, including cyanide-
insensitive PCO activity and increased 12-hydroxylation of lauric acid, which have been
identified in other studies as "classical" responses resulting from exposure to compounds that are
33 DRAFT - DO NOT CITE OR QUOTE
-------
known peroxisome proliferators (Fairish et al., 1996). The level of 8-OHdG in liver nuclear
DNA was also evaluated as an indicator of oxidative DNA damage. No histopathologic
examination or standard clinical chemistry measurements were performed.
No differences in body weight were observed for any of the treatments (Parrish et al.,
1996). The absolute liver weight was increased at the high dose, and relative liver weight was
increased at the mid and high dose (by 13 and 33%, respectively) following exposure for
3 weeks (p < 0.05). After 10 weeks of exposure, the absolute liver weights were significantly
increased at the mid dose and higher, and there were statistically significant increases in relative
liver weight beginning at the mid dose (increases of 12 and 35%, respectively). Significant dose-
related increments in cyanide-insensitive PCO activity were observed in mice treated at all TCA
doses for 3 weeks (indicating peroxisome proliferative changes before liver weight changes);
these increases persisted when treatment was extended to 10 weeks. Significantly increased
12-hydroxylation of lauric acid was also observed after 3 and 10 weeks of TCA exposure (the
response was statistically significant at the high dose), whereas 8-OHdG levels were unchanged
at both time periods. Thus, oxidative damage to genomic DNA as measured by 8-OHdG adducts
did not occur with prolonged TCA treatment, even though peroxisome proliferation was induced,
as indicated by increased PCO activity and 12-hydroxylation of lauric acid. The authors
concluded that the lack of an increase in 8-OHdG indicated that this type of DNA base damage
was not likely to be associated with the initiation of cancer by TCA; either the formation of these
adducts was inhibited or their repair was enhanced with continued TCA treatment. The
increased relative liver weight of approximately 10% at the mid dose (125 mg/kg-day) was
accompanied by a significant increase in PCO activity but not 12-hydroxylation of lauric acid.
The magnitude of these changes at the high dose was greater, with relative liver weight
increasing approximately 35% over controls and significant increases in both indicators of
peroxisome proliferation. Microscopic examination of the liver was not conducted in these
experiments. However, based on significant increases in relative liver weight (p < 0.05)
accompanied by markers of peroxisome proliferation, EPA considered the mid dose of
125 mg/kg-day a LOAEL. The low dose of 25 mg/kg-day is considered a NOAEL.
Austin et al. (1995) tested whether TCA pretreatment would alter the lipid-peroxidation
response of a subsequent acute dose of TCA. They also explored the relationship between
TCA-induced lipid peroxidation and the ability of TCA to induce markers of peroxisome
proliferation or CYP450s following short-term treatments. Male B6C3Fi mice (18/group) were
treated with 0 or 1,000 mg/L TCA for 14 days, which corresponds to estimated average doses of
approximately 0 or 250 mg/kg-day based on the default water intake of 0.25 L/kg-day for male
B6C3Fi mice (U.S. EPA, 1988). For the lipid peroxidation experiments, the water or TCA
pretreated mice were divided into six/group and administered 300 mg/kg of TCA, DC A, or an
equivalent volume of distilled water by gavage (control) as an acute challenge. Animals were
sacrificed 9 hours after the acute challenge. The livers were removed and homogenized, and the
34 DRAFT - DO NOT CITE OR QUOTE
-------
following endpoints were evaluated: (1) lipid peroxidation response, as measured by the
production of TEARS; (2) indicators of peroxisome proliferation, as measured by increased PCO
activity, increased catalase activity, and changes in microsomal 12-(co) hydroxylation of lauric
acid (an indicator for the activity of CYP4A); (3) hydroxylation of />-nitrophenol (as an index of
CYP2E1 activity); and (4) protein levels for a panel of CYP450s, as described in Section 3.3. In
addition to measurements following 14 days of treatment, TEARS levels were also measured for
the acute-challenge experiments.
No changes in water consumption or body weight were observed, although relative liver
weight was increased by 29% after 14 days of TCA treatment. TCA-treated mice had a lower
mean TEARS level as compared with controls, but the difference was not statistically
significant. In the acute challenge experiment, TCA-pretreated mice exhibited a significant
decrement in TEARS in liver homogenates, following acute dosing with either TCA or DCA, as
compared with animals that received the same acute challenge but had not been pretreated. In
contrast to the decrease in TEARS induced by TCA pretreatment, PCO, catalase, and CYP4A
activities were increased by 4.5-, 1.7-, and 2-fold, respectively, with TCA pretreatment. The
TCA-pretreated group showed no increase in CYP2E1 activity and no changes in the overall
amount of total liver microsomal P450. These data demonstrate that treatment of mice with TCA
reduced lipid peroxidation responses but increased other markers that have been associated with
peroxisome proliferation. The study authors suggested that the reduction in the TEARS response
observed in TCA-pretreated animals resulted from activities associated with peroxisome
proliferation, although it was not clear if the modifications were due to altered haloacetate
metabolism or induction of systems that would quench reactions subsequent to lipid peroxidation
initiation. The increased levels of CYP4A in TCA-pretreated mice are consistent with results
observed in other studies with other peroxisome proliferators (Gibson, 1989). Peroxisomes were
not measured directly. However, based on significant increases in relative liver weight and
several indirect markers of peroxisome proliferation (PCO, catalase, and CYP4A activities), the
single dose tested, 250 mg/kg-day, is considered a LOAEL for this study.
In summary, the ability of TCA to induce oxidative stress responses, such as lipid
peroxidation and oxidative DNA damage, and the relationship between these responses and
indicators of peroxisome proliferation or altered CYP450 activities has been tested in a series of
studies following acute or short-term TCA dosing in mice (Austin et al., 1996, 1995; Parrish et
al., 1996; Larson and Bull, 1992). TCA induces both lipid peroxidation (TEARS) and oxidative
DNA damage (8-OHdG) following administration of single oral doses. These increases appear
transient, however, since neither lipid peroxidation (Austin et al., 1995) nor 8-OHdG formation
(Parrish et al., 1996) were increased in multiple-dose studies. In contrast, responses associated
with peroxisome proliferation were induced following TCA dosing for up to 10 weeks (Parrish et
al., 1996; Austin et al., 1995). These results suggest that peroxisome proliferation is more likely
3 5 DRAFT - DO NOT CITE OR QUOTE
-------
than oxidative stress responses to be associated with liver toxicity observed in subchronic
studies.
Sanchez and Bull (1990) investigated the effects of trichloroacetate on reparative
hyperplasia in the livers of male B6C3Fi mice (12 animals/dose group). TCA was administered
in the drinking water for 14 days at concentrations of 0, 300, 1,000, or 2,000 mg/L, which
correspond to estimated average doses of approximately 0, 75, 250, or 500 mg/kg-day based on
the default water intake of 0.25 L/kg-day for male B6C3Fi mice (U.S. EPA, 1988). Food and
water consumption were recorded during the exposure period. After 14 days of exposure,
animals were sacrificed, their livers and kidneys were removed and weighed, hepatocyte
diameter was determined, and cell proliferation in the liver was assessed using [3H]thymidine
labeling after 2-day (n = 4), 5-day (n = 4), or 14-day (n = 12) treatments. Liver weight was
significantly (p < 0.05) increased compared with controls at 250 (23%) and 500 mg/kg-day
(38%). Hepatocyte diameter was significantly increased (13%; p < 0.05) at 500 mg/kg-day.
Periodic acid-Schiff s reagent-positive material (glycogen) was confined to periportal areas.
Necrosis was evident in 2 of 20 sections examined from high-dose animals, but it was not
possible to determine whether this low frequency was treatment related. A significant (p < 0.05)
increase in incorporation of [3H]thymidine into hepatic DNA was seen at 5 and 14 days at the
highest dose. However, this effect was not correlated with replicative synthesis of DNA as
measured by autoradiography. These data suggest that other processes must account for the
increased incorporation of radiolabel. The study authors suggested increased DNA repair
synthesis or alterations in thymidine pool size as possible explanations for the observed results
but noted that the mechanism for [3H]thymidine uptake could not be determined based on the
available data. EPA determined the LOAEL for this study to be 250 mg/kg-day for increased
liver weight, and the NOAEL to be 75 mg/kg-day.
Dees and Travis (1994) evaluated the ability of TCA to induce DNA synthesis in the
livers of male and female B6C3Fi mice. Mice (five/sex/dose) were given gavage doses of 0,
100, 250, 500, or 1,000 mg/kg-day TCA in corn oil for 11 days. Twenty-four hours after the last
dose, [3H]thymidine was administered intraperitoneally (i.p.). Six hours later, the mice were
sacrificed and their livers were removed. Liver samples were subsequently fixed for
histopathologic examination and evaluation of DNA synthesis (based on incorporation of the
radiolabeled thymidine). Final mean body weight and liver weight were also determined. There
were no clinical signs of toxicity at the time of sacrifice, and no significant effects on body
weight or body weight gain were observed. Absolute and relative liver weights were statistically
significantly increased in all male and female treatment groups when compared with controls. In
males, the relative liver weight was increased by 15% (at 500 mg/kg-day) to 28% (at 250 mg/kg-
day), and the increases were not dose related. In contrast, the relative liver weight in females
was increased by <9% at all doses, indicating that males may be more sensitive than females.
36 DRAFT - DO NOT CITE OR QUOTE
-------
Histopathologic changes were observed for both males and females at 1,000 mg/kg-day.
Histopathologic changes included a slight increase in the eosinophilic cytoplasmic staining of
hepatocytes near the central veins (incidence not reported). The increase in eosinophilic staining
was accompanied by a loss of cytoplasmic vacuoles. In the intermediate zone, subtle changes in
cellular architecture were noted, including that the normally parallel pattern of hepatic cords was
in disarray. Dee and Travis (1994) indicated that the appearance resembled areas of nodular
cellular proliferation but did not discuss their criteria for evaluation of this lesion. In
TCA-treated mice, [3H]thymidine incorporation (observed in autoradiographs) was mostly
localized in the intermediate zone in cells that resembled mature hepatocytes, while labeling in
controls occurred primarily in the peri-sinusoidal cells. Similar patterns of labeling were
observed in male and female mice. In addition, mitotic figures (indicative of dividing cells) were
observed in the livers of TCA-treated mice but not in controls, and these dividing cells had often
incorporated the radiolabel into the DNA. The observed mitotic figures and active labeling of
dividing cells suggest the labeling of newly replicated DNA rather than labeling of damaged
DNA as proposed by Sanchez and Bull (1990). The number of mature hepatocytes labeled with
[3H]thymidine appeared to increase with increasing TCA dose, reaching a maximum of
approximately 2.5-fold increase at 1,000 mg/kg-day (no statistical analysis was reported). In
contrast, the proportions of radiolabel incorporated into other cells (principally small peri-
sinusoidal cells) remained relatively constant at all TCA doses.
Incorporation of [3H]thymidine in extracted liver DNA also increased as TCA dose
increased. In female mice, labeling was 1.1-, 2.0-, 2.9-, and 3.3-fold the control value at 100,
250, 500, and 1,000 mg/kg-day, respectively. In male mice, labeling was 1.3-, 1.4-, 1.8-, and
2.0-fold the control value at 100, 250, 500, and 1,000 mg/kg-day, respectively. The increase in
DNA synthesis ([3H]thymidine/|ig DNA) became statistically significant at >250 mg/kg-day for
female mice and >100 mg/kg-day for males. No difference in total liver DNA content (mg
DNA/g liver) was observed. Peroxisome proliferation was not quantified. Dee and Travis
(1994) concluded that their results are consistent with an increase in DNA synthesis and cell
division/proliferation in response to TCA treatment. The authors further suggested that, since
only slight histopathologic effects were observed at the highest dose, it was unlikely that the
increased DNA synthesis and cell division were secondary to tissue repair. Based on the
increased relative liver weight (16%) at 100 mg/kg-day, accompanied by an increase in the
[3H]thymidine incorporation (1.3-fold) in male mice and supported by the histopathologic
evidence of cell proliferation, EPA determined that 100 mg/kg-day was the LOAEL for this
study. A NOAEL was not observed.
Kato-Weinstein et al. (2001) evaluated the ability of several haloacetic acids to affect
liver glycogen content, serum insulin levels, and serum glucose levels in mice. Groups of five
male B6C3Fi mice were exposed daily to neutralized TCA (>98% pure) in the drinking water at
3 g/L for 4 or 8 weeks and at 0.3, 1, or 3 g/L for 12 weeks. The concentrations provided
37 DRAFT - DO NOT CITE OR QUOTE
-------
correspond to estimated average doses of approximately 0, 75, 250, or 750 mg/kg-day,
respectively, based on a reference water intake value of 0.25 L/kg-day for male B6C3Fi mice
(U.S. EPA, 1988). Body and liver weights were recorded, and liver glycogen content and serum
glucose and insulin levels were determined after 4, 8, or 12 weeks of exposure. Localization of
glycogen in the liver was evaluated by periodic acid-Schiff s reagent staining.
TCA treatment did not affect body weight at any tested concentration. Relative liver
weights were significantly (p < 0.05) greater than controls at all exposure groups, and absolute
liver weights were significantly (p < 0.05) greater than controls at all exposure groups except in
mice exposed at 0.3 g/L for 12 weeks. The magnitude of these increases was 20-50% greater
than controls. The time course for liver glycogen content was significantly lower (approximately
25-33% as estimated from Figure 1A in Kato-Weinstein et al. [2001]; p < 0.05) than in controls
after 8 and 12 weeks of treatment at 3 g/L. After 12 weeks of treatment, liver glycogen
concentration was significantly decreased at all tested concentrations. No consistent or dose-
related effects on insulin or glucose levels were observed at any concentration of TCA in this
study. Histopathologic examination of livers from control animals revealed that glycogen-rich
(strong periodic acid-Schiff s reagent staining) and glycogen-poor (low periodic acid-Schiff s
reagent staining) cells were mixed in each hepatic zone, with slightly higher numbers of
glycogen-rich cells in the portal area. In comparison, periodic acid-Schiff s reagent staining was
confined to the periportal region in animals exposed to 0.3 and 1.0 g/L of TCA. This observation
suggests that glycogen depletion occurred in the central lobular area as a result of depletion of
glycogen from cells that appear to concentrate it in the liver of control mice. This result can be
compared with observations made by Bull et al. (1990) and Sanchez and Bull (1990), who
reported that TCA-treated animals displayed less evidence for glycogen accumulation and noted
that when staining occurred, it was more prominent in the periportal portions than in
centrilobular portions of the liver acinus.
Laughter et al. (2004) exposed wild-type SV129 mice (males, 3-5/group) and a mouse
strain lacking a functional form of peroxisome proliferator-activated receptor (PPAR)a (PPARa-
null mice, males, 3-5/group) to TCA at 0, 0.25, 0.5, 1, or 2 g/L in the drinking water
(neutralized) for 7 days. These concentrations correspond to estimated doses of approximately 0,
62.5, 125, 250, or 500 mg/kg-day, respectively, based on a reference water intake value of 0.25
L/kg-day for male B6C3Fi mice (U.S. EPA, 1988). Wy-14,693 at 50 mg/kg was given as the
positive control. Following exposure, the mice were sacrificed, and livers were removed and
weighed. Subsamples of liver were processed for histopathologic examination, analysis of
CYP4A and AGO protein expression, and measurement of PCO activity. Exposure to TCA
increased liver-to-body-weight ratios in wild-type mice, but the response was not statistically
significant. Exposure to TCA induced markers of peroxisome proliferation in wild-type mice but
not PPARa-null mice. Exposure to 1 or 2 g/L TCA significantly increased the level of CYP4A
protein, and exposure to 2 g/L significantly increased PCO and AGO activity in liver
3 8 DRAFT - DO NOT CITE OR QUOTE
-------
homogenates from wild-type mice only, indicating that PPARa is necessary for TCA to induce
lipid metabolism enzymes associated with peroxisome proliferation. Centrilobular hepatocyte
hypertrophy was observed in wild-type mice exposed to 2 g/L TCA, but not in PPARa-null mice
exposed to the same concentration. The results of this study indicate that TCA induces
peroxisome proliferation through activation of PPARa.
4.2.1.2. Subchronic Inhalation Studies
No short-term or subchronic toxicity studies for TCA were identified for exposure by the
inhalation route.
4.2.2. Chronic Studies and Cancer Assays
Long-term oral toxicity studies for TCA are available for rats and mice. The available
data are summarized in Table 4-2 (noncancer data) and Table 4-3 (cancer and tumor promotion
data).
39 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-2. Summary of chronic studies evaluating noncancer effects of TCA after oral administration in rats and mice
Reference"
Species
Exposure
route
Exposure
duration
Doses evaluated
Noncancer effects
evaluated
Effects'1
NOAEL
(mg/kg-d)
LOAEL
(mg/kg-d)
Comments
Rats
DeAngelo et
al., 1997
F344 rats
(males,
50/group)
Oral,
drinking
water
104 wks
0, 3.6, 32.5, or
364 mg/kg-d
Body weight, ALT and
AST activity,
histopathology (liver,
kidneys, spleen, testes,
excised lesions at
interim and terminal
sacrifice;
comprehensive
histopathologic exam in
high-dose group at
terminal sacrifice),
peroxisome
proliferation
Decreased body
weight, increased
serum ALT activity;
mild hepatocellular
necrosis; increased
peroxisome
proliferation
32.5
364
Time-weighted
average daily doses
were calculated by
the study authors; a
comprehensive set
of tissues was
examined
microscopically.
Mice
DeAngelo et
al., 2008
B6C3FJ mice
(males, Study
1: 50/group;
Study 2:
58/group;
Study 3:
72/group; 27-
30/dose at
terminal
sacrifice;
5/dose at
interim
sacrifices)
Oral,
drinking
water
Study 1:
60 wks
Studies 2
and 3:
104 wks
Study 1: 0,8,68,
or 602 mg/kg-d;
Study 2: 0 or
572 mg/kg-d;
Study 3: 0, 6, or
58 mg/kg-d (doses
based on nominal
drinking water
concentrations; see
text)
Body weight, liver
weight, serum LDH
activity, liver PCO
activity, hepatocyte
proliferation,
histopathologic
examination for gross
lesions, liver, kidney,
spleen, and testis at
interim and terminal
necropsies; complete
histopathologic
examination on five
mice from the high-
dose and control groups
LOAEL based on
increased liver weight,
hepatic necrosis, LDH
activity (30 wks), and
testicular degeneration.
Other effects at higher
doses included
decreased body weight
and hepatic
inflammation.
Increased liver PCO
activity and labeling
index for nuclei outside
of hepatic proliferative
lesions were observed
at mid and high doses.
8
68
Time-weighted
average daily doses
were calculated by
the study authors; a
comprehensive set
of tissues was
examined
microscopically.
40
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-2. Summary of chronic studies evaluating noncancer effects of TCA after oral administration in rats and mice
Reference"
Pereira, 1996
Bull et al.,
1990
Herren-
Freundetal.,
1987
Species
B6C3FJ mice
(females, 38-
134/group)
B6C3FJ mice
(5-35 mice/
dose/time
point, see
text)
(B)
(11 males/
dose)
B6C3F: mice
(males, 22-
33/group)
Exposure
route
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
Exposure
duration
51or82wks
(A) 52 wks
(with interim
sacrifices at
15, 24, and
37 wks)
(B) 37 wks +
15-wk
recovery
61 wks
Doses evaluated
0, 78, 262, or
784 mg/kg-d
(A) 0, 164, or
329 mg/kg-d
(B) 0 or
309 mg/kg-d
0, 500, or
1,250 mg/kg-d
Noncancer effects
evaluated
Body and liver weight,
liver histopathology
Liver and kidney
weight and
histopathology
Liver weight and
histopathology
Effects'1
Increased relative liver
weight
Increased absolute and
relative liver weight,
cytomegaly, modest
glycogen accumulation,
accumulation of
lipofuscin in liver
Increased absolute and
relative liver weight
NOAEL
(mg/kg-d)
78
Not
determined
Not
determined
LOAEL
(mg/kg-d)
262
164
500
Comments
Increased liver
weight was
observed after
82 wks at
262 mg/kg-d;
262 mg/kg-d was
judged to be an
equivocal LOAEL
in the absence of
other measures of
liver toxicity.
Only the liver and
kidneys were
evaluated; dose was
estimated by the
authors.
Only the liver was
examined
microscopically.
aCancer studies that evaluated noncancer endpoints are included in this table; data from von Tungeln et al. (2002) were not included in this table because animals were
not dosed by the oral route (i.p. injection).
bThe effects listed in this table may have occurred either at the LOAEL or at higher doses.
Source: adapted from U.S. EPA (2005c).
41
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-3. Summary of cancer bioassays and tumor promotion studies of TCA in rats and mice
Reference
Species
Study type
Exposure route
Exposure duration
Doses
evaluated
Results
Comments
Cancer bioassays
Rats
DeAngelo et
al., 1997
F344 rats
(males,
50/group)
Cancer assay,
multiple
organs
Oral, drinking
water
104 wks
0, 3.6, 32.5, or
364 mg/kg-d
Negative for tumor
induction
A comprehensive set of tissues was
microscopically examined; only about 30
animals/ concentration were exposed for
>60 wks.
Mice
DeAngelo et
al., 2008
Pereira, 1996
Bull et al.,
2002
Bull et al.,
1990
B6C3FJ mice
(males, 25-
42/dose at
terminal
sacrifice;
five/dose at
interim
sacrifices)
B6C3FJ mice
(females, 38-
1347 group)
B6C3FJ mice
(males, 20 or
40/group)
B6C3FJ mice
(5-35 mice/
dose/time
point, see
text)
(B) 11 males/
dose
Cancer
bioassay
Cancer
bioassay
Cancer
bioassay
Chronic
toxicity study
with
microscopic
examination
of the liver
Oral, drinking
water
Oral, drinking
water
Oral, drinking
water
Oral, drinking
water
Study 1: 60 wks;
interim sacrifices at
4, 15, 30, and 45 wks
Studies 2 and 3 :
104 wks (doses based
on nominal drinking
water concentrations;
see text)
5 lor 82 wks
52 wks
(A) 52 wks (interim
sacrifices at 15, 24,
and 37 wks)
(B) 37 wks + 15-wk
recovery
(males only)
Study 1: 0,8,
68, or 602
mg/kg-d;
Study 2: 0 or
572 mg/kg-d;
Study 3: 0,6,
or 58 mg/kg-d
0, 78, 262, and
784 mg/kg-d
0, 120, or
480 mg/kg-d
(A) 0, 164, or
329 mg/kg-d
(females only
treated with
329 mg/kg-d
for 52 wks)
(B) 0 or 309
mg/kg-d
Positive for liver
tumors starting at
45 wks
Positive for liver
tumors at 5 1 and
82 wks
Increased incidence of
liver tumors
Males: dose-related
increase in liver
tumors at 52 wks
Females: no liver
tumors found
Liver, kidneys, spleen, and testes were
evaluated microscopically for tumors;
complete histopathologic evaluation was
conducted on other organs for five mice
from the control and high-dose groups.
Only the liver was evaluated for tumors.
Only the liver was microscopically
examined; doses were estimated based on a
default water intake of 0.24 L/kg-d.
Hepatoproliferative lesions were only
observed in males, but noncancer effects
were reportedly similar in incidence and
severity in males and females; only the liver
and kidneys were evaluated.
42
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-3. Summary of cancer bioassays and tumor promotion studies of TCA in rats and mice
Reference
Von Tungeln
et al., 2002
Species
B6C3FJ mice
(23-24/sex/
dose, males
and females)
Study type
Neonatal
cancer assay
Exposure route
i.p. injection
Exposure duration
Doses administered
at 8 and 15 ds of age;
tumors evaluated at
12 or 20 mo of age
Doses
evaluated
Total dose of
16 or 33 mg/kg
over a 2-d
period (at 8 and
15 d of age)
Results
Negative for tumor
induction
Comments
TCA induced oxidative stress, but there was
no significant increase in tumors in the
neonatal mouse.
Tumor promotion studies
Rats
Parnell etal.,
1988
Sprague-
Dawley rats
(males, 6-
12/dose and
sampling
time)
Promotion,
multiple
organs,
partially
hepat-
ectomized
rats
Oral, drinking
water
Up to 12 mo
0, 6, 60, or
600 mg/kg-d
GGT-positive foci in
liver
TCA promoted GGT-positive foci in
DEN-initiated rats at all doses evaluated,
but only one rat showed a liver carcinoma.
TCA showed no activity as an initiator.
Mice
Herren-
Freundetal.,
1987
Pereira and
Phelps, 1996
Pereira etal.,
2001
B6C3F: mice
(males, 22-
33/group)
B6C3FJ mice
(females, 8-
407 group)
B6C3FJ mice
(males and
females, 14-
16/sex)
Cancer assay
and tumor
promotion,
liver
Cancer assay
and tumor
promotion
Tumor
promotion
Oral, drinking
water
Oral, drinking
water
Oral, drinking
water
61 wks
Up to 52 wks
31 wks
0, 500, or
1,250 mg/kg-d
0, 78, 262, or
784 mg/kg-d
0 or 960
(females) or
1,000 (males)
mg/kg-d
Positive for tumor
production and for
tumor promotion
Positive with or
without MNU
initiation
Males: significant
increase in liver and
kidney tumors in
TCA-treated mice
initiated with MNU
Females: insignificant
increase in liver and
kidney tumors in mice
initiated with MNU
and promoted by TCA
Ethylnitrosourea (ENU) was used as an
initiator. Only the liver was
microscopically examined; liver tumors
were observed either with or without ENU
pretreatment.
MNU was used as an initiator. Only the
liver was examined for tumors.
Only the liver and kidneys were examined
for tumors; MNU was used as an initiator;
statistically significant increases in tumor
yield were only observed in males.
43
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-3. Summary of cancer bioassays and tumor promotion studies of TCA in rats and mice
Reference
Pereiraetal.,
1997
Species
B6C3FJ mice
(females, 20-
457 dose)
Study type
Tumor
promotion
Exposure route
Oral, drinking
water
Exposure duration
44wks
Doses
evaluated
0, 235, or
980 mg/kg-d
Results
Positive for liver
tumors
Comments
MNU was used as an initiator; only the
liver was microscopically examined.
DEN = diethylnitrosamine; MNU = N-methyl-N-nitrosourea
44
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.2.1. Oral Studies
4.2.2.1.1. Rats
4.2.2.1.1.1. Chronic studies. DeAngelo et al. (1997) evaluated the tumorigenicity of TCA in
male F344 rats exposed for 104 weeks via drinking water. Groups of 50 rats received TCA in
drinking water (adjusted to physiological pH) at 0, 50, 500, or 5,000 mg/L, resulting in time-
weighted mean doses of 0, 3.6, 32.5, or 364 mg/kg-day as calculated by the study authors.
Dosing was initiated at 28-30 days of age. Interim sacrifices (18-21 rats/group) were conducted
at 15, 30, 45, and 60 weeks, and gross lesions in the body and internal organs were examined.
The organs examined histologically at the interim and terminal sacrifices were liver, kidney,
spleen, and testes. The survivors were sacrificed at 104 weeks. At study termination, blood
from all treatment groups was analyzed for serum AST and ALT activity, and livers were
analyzed for cyanide-insensitive PCO activity and extent of hepatocyte proliferation
([3H]thymidine incorporation). At sacrifice, all animals were subjected to a complete necropsy.
A comprehensive set of tissues, including all major organs, was examined microscopically in
high-dose rats. The liver, kidney, spleen, and testes were examined in the remaining dose
groups.
Survival in dosed animals was similar to that in controls (79, 75, 59, and 76% in the
control, low-, mid-, and high-dose groups, respectively), and there were no significant
differences in water consumption between exposed and control groups. A maximum tolerated
dose was reached, as indicated by a 10.7% decrease in the final mean body weight of the high-
dose animals relative to controls. Absolute liver weight was decreased by 11% at the high dose.
No significant differences from the control values were observed in the absolute and relative
weights of the kidney, spleen, or testes. AST activity was significantly decreased in the mid-
dose group, but the data did not show a dose-related trend. ALT activity increased in a dose-
related manner, and the response was statistically significant at the high dose. Peroxisome
proliferation in the livers of animals exposed to the high dose (364 mg/kg-day) of TCA was
significantly increased, based on a twofold increase in cyanide-insensitive PCO activity
throughout the exposure period. There was no evidence of a dose-related increase in hepatocyte
proliferation. Most nonneoplastic hepatic lesions were spontaneous and age related. A minimal-
to-mild, treatment-related increase in hepatic cytoplasmic vacuolization was evident at the low
and mid doses, but not at the high dose (data not shown). A mild increase in the severity of
hepatocellular necrosis was observed in high-dose animals (data not shown). No treatment-
related histopathologic changes were noted for the kidney, spleen, or testes. No dose-related
increases in the incidences of neoplasms or hyperplasia were observed in the liver or other
tissues. Animals for interim sacrifices were from the same exposed groups. The number of
animals at final sacrifice ranged from 19 to 24/dose group. Hence, the power of detection of this
bioassay was limited by the relatively small group sizes. DeAngelo et al. (1997) determined the
study NOAEL and LOAEL values to be 32.5 and 364 mg/kg-day, respectively, based on
45 DRAFT - DO NOT CITE OR QUOTE
-------
decreased body weight, increased serum ALT activity, mild hepatocellular necrosis, and
increased peroxisome proliferation.
4.2.2.1.1.2. Tumor initiation and promotion studies. Parnell et al. (1988) investigated the
initiating and promoting effects of TCA by using two short-term tests: the rat hepatic enzyme-
altered foci assay and stimulation of peroxisomal-dependent PCO activity in the liver. In the
initiation protocol, male Sprague-Dawley rats (6-12/treatment/time point) underwent a two-
thirds partial hepatectomy or sham operation as control, followed 24 hours later by a single
gavage dose of 10 mg/kg diethylnitrosamine (DEN) (a known initiator) or 1,500 mg/kg of TCA.
Additional groups of hepatectomized rats began a regimen of exposure to 5,000 mg/L of TCA in
drinking water (about 600 mg/kg-day) for 10, 20, or 30 days to assess the effects of an extended
initiation period. Two weeks following the initiation period, all groups were promoted for the
remainder of the study (up to 12 months after beginning the promotion phase) with 500 mg/L
phenobarbital (PB) in the drinking water. Animals were randomly sampled 24 hours after the
end of the initiation period, 24 hours prior to the start of the promotion phase, and 3, 6, and
12 months after beginning promotion. In the initiation study, the positive control is the group
with partial hepatectomy, treated with DEN as the initiator and PB for promotion.
In the promotion protocol, rats (6-12/treatment/time point) underwent the two-thirds
hepatectomy or sham operation followed 24 hours later by administration of a single 10 mg/kg
oral dose of DEN (the initiator) or distilled water (control). Promotion was begun 2 weeks later
by the addition of 500 mg/L PB (the positive control) or 0, 50, 500, or 5,000 mg/L TCA
(equivalent to doses of about 0, 6, 60, or 600 mg/kg-day as calculated by using the chronic water
intake factor of 0.12 L/kg-day for Sprague-Dawley rats [U.S. EPA, 1988]) to the drinking water.
The test animals were randomly sampled at 2 weeks and 1, 3, 6, and 12 months after beginning
promotion. In the initiation bioassay, only the positive control group showed a statistically
significant induction of GGT-positive foci at the 3-, 6-, and 12-month evaluation intervals. (The
development of GGT-positive foci has been closely linked to the subsequent development of
both neoplastic nodules and hepatomas.) None of the groups that received initiation doses of
TCA or the associated controls exhibited significant induction of GGT-positive foci. Since TCA
did not induce GGT-positive foci (as did the tumor initiator DEN), TCA did not appear to be an
initiator based on the results of this assay.
In the promotion bioassay, GGT-positive foci were induced in the positive control
(partial hepatectomy/DEN/PB) at all evaluation intervals. Exposure of rats to 50, 500, or
5,000 mg/L TCA as a promoter for 6 or 12 months produced a significant increase in the number
and size (mean area) of GGT-positive foci over the negative control groups (partial hepatectomy
alone, partial hepatectomy/DEN, or TCA alone). At 3 months, rats in the 50 and 5,000 mg/L
TCA promotion groups also had significantly greater numbers of GGT-positive foci compared
with the negative controls (data on size of foci were not reported for this time point). The
46 DRAFT - DO NOT CITE OR QUOTE
-------
promotion protocol also resulted in a statistically significant, but weak (10-20% greater than
controls), increase in peroxisomal-specific PCO activity at the 5,000 mg/L drinking water level.
No significant gross or histopathologic lesions, hepatomegaly, or changes in organ-to-body-
weight ratios could be attributed to TCA exposure and only one hepatocellular carcinoma in an
animal from the partial hepatectomy/DEN/5,000 mg/L TCA group was found in this study. The
study authors concluded that TCA has significant, but relatively weak, tumor-promoting activity
in the tested bioassay model. It should be noted that the observed promotion effect was from
both partial hepatectomy and TCA. There was no study group that treated sham-operated rats
with DEN, followed by TCA. Partial hepatectomy can function as a promoter by itself.
4.2.2.1.2. Mice
4.2.2.1.2.1. Chronic studies. DeAngelo et al. (2008) evaluated the induction of hepatocellular
neoplasia in male B6C3Fi mice exposed to TCA in drinking water in three studies. Male
B6C3Fi mice were exposed to nominal concentrations of 0.05, 0.5, or 5 g/L TCA in the drinking
water (50/dose at study initiation) for 60 weeks (Study 1); 0 or 4.5 g/L TCA (58 animals/group)
for 104 weeks (Study 2); or 0, 0.05, or 0.5 g/L TCA (72/group) for 104 weeks (Study 3). The pH
of the dosing solutions was adjusted to 6.0-7.1 by the addition of 10 N sodium hydroxide. Mice
in the control group in Study 1 received 2 g/L sodium chloride in the drinking water, while those
in Study 2 received 1.5 g/L neutralized acetic acid to account for any taste aversion of TCA in
dosing solutions. In Study 3, deionized water served as the control. Body weights and water
consumption were measured twice monthly for the first 2 months and then monthly afterwards.
In Study 1, groups of five animals from each dose group were examined at necropsy at 4, 15, 31,
and 45 weeks. In Study 2, serial necropsies were conducted at 15, 30, 45, and 60 weeks. In
Study 3, serial necropsies were conducted at 26, 52, and 78 weeks.
At interim and terminal necropsies, gross lesions, livers, kidneys, spleens, and testes were
examined by a board-certified veterinary pathologist. For all other tissues, a complete
pathological examination was performed on five mice from the high-dose and control groups. If
the number of any histopathologic lesions in a tissue was significantly increased above that in the
control animals, then that tissue was examined in all TCA dose groups. To determine long-term
hepatocellular damage during TCA treatment, arterial blood was collected at 30 and 60 weeks
(Study 1) and 4, 30, and 104 weeks (Study 2), and serum LDH activity was measured. Portions
of liver tissue from the interim-sacrifice animals (5/group/duration) were frozen and analyzed for
PCO activity, a marker of peroxisome proliferation. Five days prior to each scheduled necropsy,
osmotic pumps containing 200 uL [3H]thymidine (62-64 Ci/mmol) or 20 mg/mL bromo-
deoxyuridine (BrdU) (Study 3) were implanted subcutaneously. Autoradiography using
paraffin-embedded sections of liver was performed to evaluate hepatocyte proliferation, as
measured by the incorporation of [3H]-labeled thymidine or BrdU into nuclear DNA. The
47 DRAFT - DO NOT CITE OR QUOTE
-------
labeling index was calculated by dividing the number of labeled hepatocyte nuclei (S-phase) by
the total number of hepatocyte nuclei scored.
For Study 1, time-weighted mean doses of 8, 68, and 602 mg/kg-day were calculated by
the study authors from nominal TCA concentrations (0.05, 0.5, and 5 g/L, respectively) and
drinking water consumption data for the low-, mid-, and high-dose groups.4 Animals in the mid-
and high-dose groups consumed significantly less water than the controls. No significant
differences in animal survival were noted for any treatment group. The study authors estimated
the mean doses to be 572 mg/kg-day for a nominal drinking water concentration of 4.5 g/L TCA
(Study 2), and 6 and 58 mg/kg-day for nominal concentrations of 0.05 and 0.5 mg/kg-day
(Study 3). With the exception of liver neoplasia, all data presented by DeAngelo et al. (2008)
were from the 60-week study (Study 1).
No decrease in animal survival was found at any TCA dose in any of the three studies.
Exposure to TCA in the drinking water decreased body weight by 15% in the high-dose group
relative to the control. Significant, dose-related increases in absolute and relative liver weights
were observed in the 0.5 and 5 g/L treatment groups at all scheduled sacrifices, with the
exception of the 0.5 g/L dose group at 30 days.
Nonneoplastic alterations in the liver and testes were seen at study termination at
60 weeks and appeared to be dose related (Tables 4-4 and 4-5). The nonneoplastic alterations
observed in the liver included hepatocellular cytoplasmic alteration, necrosis, and inflammation.
Cytoplasmic alterations were observed in all treatment groups; however, the incidence did not
increase monotonically with dose. These lesions were most prominent in the 5 g/L TCA group
throughout the study and were most severe after 60 weeks of treatment. The alterations were
characterized by an intense eosinophilic cytoplasm with deep basophilic granularity and slight
cytomegaly. The distribution ranged from centrilobular to diffuse. Hepatic necrosis was
observed in the middle- and high-dose group at all time points and was reported to be most
severe at 30-45 weeks; the study report provided only combined data for the 30- and 45-week
interim sacrifices (Table 4-5). A significant increase in the severity of inflammation was seen in
the high-dose group at 60 weeks. A dose-related increase in serum LDH activity (a measure of
liver damage) was observed at 30 weeks, and significant increases were measured in the 0.5 and
5.0 g/L dose groups. No change in LDH activity was found in any treatment groups at 60 weeks.
No other hepatic changes showed statistically significant increases in incidence or severity level.
4DeAngelo et al. (2008) also reported measured TCA concentrations in drinking water. Doses calculated by EPA
based on those concentrations and reported drinking water consumption are as follows:
Study 1: time-weighted mean doses were calculated as 7.7, 68.2, and 602.1 mg/kg-day for measured TCA
concentrations of 0.05, 0.48, and 5.06 g/L, respectively.
Study 2: time-weighted mean dose was calculated as 571.5 mg/kg-day for a measured TCA concentration of
4.43 g/L.
Study 3: time-weighted mean doses were calculated as 6.7 and 81.2 mg/kg-day for measured TCA concentrations of
0.06 and 0.70 g/L, respectively.
48 DRAFT - DO NOT CITE OR QUOTE
-------
An increased incidence of testicular tubular degeneration was seen in the 0.5 and 5 g/L treatment
groups (Table 4-4). No treatment-related changes were observed in the spleen or kidney.
Table 4-4. Incidence and severity of nonneoplastic lesions in male B6C3Fi
mice exposed to TCA in drinking water for 60 weeks
Lesion
Hepatocellular
cytoplasmic alteration
Hepatocellular
inflammation
Testicular tubular
degeneration
Treatment
Dose"
Numberb
Incidence0
Severityd
Incidence0
Severityd
Incidence0
Severityd
Control
0
30
7%
0.10 ±0.40
10%
0.13 ±0.40
7%
0.10 ±0.40
0.05 g/L TCA
8
27
48%e
0.70 ±0.82
0
0
0
0
0.5 g/L TCA
68
29
20.6%e
0.34 ±0.72
7%
0.07 ±0.03
14%e
0.17 ±0.47
5 g/L TCA
602
29
93%e
1.60±0.62e
24%e
0.24 ±0.44
21%e
0.21 ±0.41
"Time-weighted mean daily dose in mg/kg-day based on nominal TCA concentration in drinking water as reported
in DeAngelo et al. (2008). Doses calculated by EPA using measured concentrations are 7.7, 68.2, and
602.1 mg/kg-day.
bNumber of animals examined.
Percentage of animals with alteration.
dSeverity: 0 = no lesion, 1 = minimal, 2 = mild, 3 = moderate, 4 = severe (reported as the average severity of all
animals in the dose group).
Statistically significant from the control group, p < 0.05.
Source: DeAngelo et al. (2008).
Table 4-5. Incidence and severity of hepatocellular necrosis at 30-45 weeks
in male B6C3Fi mice exposed to TCA in drinking water
Treatment
Dose3 (mg/kg-d)
Numberb
Incidence0
Severityd
Control
0
10
0
0
0.05 g/L TCA
8
10
0
0
0.5 g/L TCA
68
10
30.0%
0.50 ±0.97
5 g/L TCA
602
10
50.0%
1.30±1.49e
"Time-weighted mean daily dose in mg/kg-day based on nominal TCA concentration in drinking water as reported
in DeAngelo et al. (2008). Doses calculated by EPA using measured concentrations are 7.7, 68.2, and
602.1 mg/kg-day.
bNumber of animals examined.
Percentage of animals with alteration.
dSeverity: 0 = no lesion, 1 = minimal, 2 = mild, 3 = moderate, 4 = severe (reported as the average severity of all
animals in the dose group).
Statistically significant from the control group, p < 0.05.
Source: DeAngelo et al. (2008).
Areas of inflammation (at high dose only) and necrosis (at mid- and high-dose) were
present during the early course of TCA administration, but abated after week 60 in all studies.
49
DRAFT - DO NOT CITE OR QUOTE
-------
Similarly, LDH activity was elevated in the mid- and high-dose groups at week 30 but not at
week 60. Cytoplasmic alterations occurred as early as week 4 and persisted throughout the three
studies at all doses, indicating that this effect did not correlate with other nonneoplastic changes
in the liver. For the 60-week study, EPA determined the LOAEL for effects on the liver
(increased liver weight, hepatic necrosis, and serum LDH activity at 30 weeks) and testes
(testicular tubular degeneration) to be 0.5 g/L (68 mg/kg-day) and the NOAEL to be 0.05 g/L
(8 mg/kg-day).
Exposure to TCA induced tumors in the liver at 60 weeks (Table 4-6). There were
significant dose-related trends for increased prevalence and multiplicity of adenomas and
carcinomas. The prevalence and numbers of hepatocellular carcinomas and hepatocellular
adenomas were significantly increased in the high-dose group. The number of animals with
either lesion was significantly increased in the 0.5 g/L treatment group. Neoplasia was first seen
in all dose groups after 45 weeks of treatment. The prevalence and number of tumors in the
5 g/L group were 60% (3/5 animals with a lesion) and 0.80 lesions/animal. One hepatocellular
carcinoma was found in the 0.5 g/L group and one hepatocellular adenoma was found in the
0.05 g/L group. No induction of tumors was reported in other organs.
Table 4-6. Prevalence and multiplicity of hepatocellular tumors in male
B6C3Fi mice exposed to TCA in drinking water for 60 weeks
Neoplasia
typec
HA
HC
HAorHC
Treatment
Dose3
Numberb
Prevalence"1
Multiplicity"
Prevalence"1
Multiplicity6
Prevalence"1
Multiplicity6
Control
0
30 (30)
7%
0.07±0.05e
7%
0.07 ±0.05
14%
0.13 ±0.06
0.05 g/L TCA
8
27 (30)
15%
0.15 ±0.07
4%
0.04 ± 0.04
15%
0.19 ±0.09
0.5 g/L TCA
68
29 (30)
22%
0.24 ±0.10
21%
0.28 ±0.22
38%f
0.52±0.14f
5 g/L TCA
602
29 (30)
38%f
0.55±0.15f
38%f
0.41±0.11f
55.%f
1.00±0.19f
""Time-weighted mean daily dose in mg/kg-day based on nominal TCA concentration in drinking water as
reported in DeAngelo et al. (2008). Doses calculated by EPA using measured concentrations are 7.7, 68.2, and
602.1 mg/kg-day.
bNumber of animals examined at terminal sacrifice. Parentheses = number of animals/group scheduled for
terminal necropsy. Data for interim-sacrifice animals not included.
°HA = hepatocellular adenoma, HC = hepatocellular carcinoma, HA or HC = either hepatocellular adenoma or
hepatocellular carcinoma.
Percentage of animals with a lesion as reported in the study report.
eNumber of lesions/animal, mean ± standard error of the mean.
Statistically significant from the control group, p < 0.05.
Source: DeAngelo etal. (2008).
50
DRAFT - DO NOT CITE OR QUOTE
-------
Significant increases above the control values were also observed for the prevalence and
multiplicity of adenomas, carcinomas, and either adenomas or carcinomas for mice exposed to
4.5 g/L TCA for 104 weeks (Study 2) or 0.5 g/L TCA for 104 weeks (Study 3) (Table 4-7).
Neoplastic lesions observed at organ sites other than the liver were considered spontaneous for
the male mouse and did not exceed the tumor incidences when compared to a historical control
database.
Table 4-7. Incidence of hepatocellular tumors in male B6C3Fi mice exposed
to TCA in drinking water for 104 weeks
Neoplasia type0
HA
HC
HAorHC
HA
HC
HAorHC
Treatment
Dose3
Numberb
Prevalence"1
Multiplicity6
Prevalence"1
Multiplicity6
Prevalence"1
Multiplicity6
Treatment
Dose3
Numberb
Prevalence"1
Multiplicity6
Prevalence"1
Multiplicity6
Prevalence"1
Multiplicity6
Control
0
25 (32)
0
0
12%
0.20 ±0.12
12%
0.20 ±0.12
Control
0
42 (50)
21%
0.21±0.06
55%
0.74 ±0.12
64%
0.93 ±0.12
0.05 g/L TCA
6
35 (50)
23%
0.34 ±0.12
40%
0.71 ±0.19
57%
1.11±0.21
0.5 g/L TCA
58
37 (50)
51%f
0.78±0.15f
78%f
1.46±0.21f
87%f
2.14±0.26f
4.5 g/L TCA
572
36 (43)
59%f
0.61±0.16f
78%g
1.50±0.22f
89%f
2.11±0.25f
"Time-weighted mean daily dose in mg/kg-day calculated over 104 weeks based on nominal TCA concentration in
drinking water as reported in DeAngelo et al. (2008). Doses calculated by EPA using measured concentrations are
6.7, 81.2, and 571.5 mg/kg-day, respectively.
bAnimals surviving >78 weeks, parentheses = number of animals/group scheduled for terminal necropsy. Data for
interim-sacrifice mice not included.
°HA = hepatocellular adenoma; HC = hepatocellular carcinoma; HA or HC = either hepatocellular adenoma or
hepatocellular carcinoma.
dNumber of animals with a lesion/number of animals examined x 100%.
6Mean number of lesions ± standard error of the mean.
Statistically significant from the control group, p < 0.03.
Source: DeAngelo et al. (2008).
Liver PCO activity was significantly increased at the mid and high doses at intervals of 4,
15, 30, 45, and 60 weeks when compared with control values. The range of PCO activity for
mice exposed to 0.5 and 5 g/L was 129-260 and 326-575%, respectively, above the control
value. The mean PCO activity (averaged over the five intervals) is summarized in Table 4-8.
51
DRAFT - DO NOT CITE OR QUOTE
-------
Autoradiographs of the livers from animals exposed to 5 g/L TCA showed significantly
increased labeling of hepatocyte nuclei at 30 weeks (about 3-fold) and 40 weeks (about 2.5-fold).
Increased nuclear labeling was observed in the mid-dose treatment group at 60 weeks (about
threefold). These data indicate that TCA induced treatment-related tumors in male mice at doses
that also induced peroxisome proliferation and hepatocyte proliferation.
Table 4-8. Mean PCO activity in male B6C3Fi mice exposed to TCA in
drinking water for up to 60 weeks
Mean PCO activity (nmol
NAD reduced/min/mg
protein)3
(% control)
Control
2.59 ±1.04
0.05 g/L TCA
(8 mg/kg-d)b
2.85 ± 0.86
(117%)
0.5 g/L TCA
(68 mg/kg-d)b
4.75 ±1.16
(200%)
5 g/L TCA
(602 mg/kg-d)b
11.99 ±3.04
(475%)
"Mean PCO activity ± SD was calculated as an arithmetic mean of the PCO activity for weeks 4, 15, 30, 45, and 60.
PCO activity for each time point was based on 5 mice/group/time point. The total number of mice for each
concentration was 25 (with the exception of 24 mice for the 5 g/L TCA group).
bTime-weighted mean daily dose in mg/kg-day based on nominal TCA concentration in drinking water as reported
in DeAngelo et al. (2008). Doses calculated by EPA using measured concentrations are 7.7, 68.2, and
602.1 mg/kg-day.
Source: DeAngelo et al. (2008) and email dated March 12, 2010, from Anthony DeAngelo, National Health and
Environmental Effects Research Laboratory (NHEERL), Office of Research and Development (ORD), U.S. EPA,
to Diana Wong, National Center for Environmental Assessment (NCEA), ORD, U.S. EPA.
Bull et al. (1990) examined the induction of tumors in the liver of male and female
B6C3Fi mice given TCA in drinking water (neutralized to pH 6.8-7.2). Groups of mice (males:
24/high dose, 1 I/low dose, 35 controls; females: 10/group) were exposed to neutralized TCA
(males: 0, 1, or 2 g/L; females: 0 or 2 g/L) for 52 weeks. Interim sacrifices were performed at
15, 24, and 37 weeks on separate groups of male mice (five/group). An additional group of
11 males received 2 g/L TCA for 37 weeks, followed by a 15-week recovery period. The 0, 1,
and 2 g/L concentrations used in this study corresponded to estimated average doses of 0, 164,
and 329 mg/kg-day as calculated from data for total dose provided in the study report. The
approximate average dose for the 37-week exposure with recovery was 309 mg/kg-day.
No effects of treatment on survival or body weight were observed. Body weight and food
and water consumption data were recorded but not reported. A significant increase in the
relative liver weight was seen in the 1 g/L males (30% increase from control), 2 g/L males (63%
increase), and 2 g/L females (25% increase) at 52 weeks when compared with controls. No
changes in kidney weights were observed. Mild intracellular swelling and some indication of
glycogen accumulation in the periportal region were observed in the livers of treated male and
female mice at 52 weeks. Male mice in the 2 g/L group had a dose-related accumulation of
lipofuscin near proliferative lesions (no incidence reported) and hyperplastic liver nodules (9/24).
52
DRAFT - DO NOT CITE OR QUOTE
-------
The incidences of hepatocellular adenomas in male mice were 0/35 (0%), 2/11 (18%),
and 1/24 (4%), and the incidences of hepatocellular carcinomas were 0/35 (0%), 2/11 (18%), and
4/24 (17%) in the 0, 1, and 2 g/L exposure groups, respectively. Female mice did not develop
any tumors in response to TCA treatment and might be less sensitive to TCA treatment than
males. However, fewer female mice (52 weeks: 2 g/L, 10 females) were evaluated in this study
than were male mice (37 weeks: 2 g/L, 11 males; 52 weeks: 1 g/L, 11 males; 2 g/L, 24 males),
which may have limited the ability of the study to detect tumors in female mice. Fifteen weeks
into the 37-week study, exposure to 2 g/L resulted in hepatocellular carcinomas in 3/11 (30%)
male mice, but hepatic adenomas had not occurred by that time. Since the maximum exposure
duration in this study was only 52 weeks, this study may not have evaluated mice for an adequate
length of time to observe the full carcinogenic potential of TCA. In addition, the numbers of
animals tested were less than adequate. EPA determined that the LOAEL for noncancer effects
was 164 mg/kg-day based on increase in liver weight, cytomegaly, and modest glycogen
accumulation.
Pereira (1996) administered 0, 2.0, 6.67, or 20.0 mmol/L TCA (0, 327, 1,090, or
3,268 mg/L) (neutralized with sodium hydroxide to pH 6.5-7.5) in drinking water to female
B6C3Fi mice from 7 to 8 weeks of age until sacrifice after 360 days (51 weeks) or 576 days
(82 weeks) of exposure. A control group of 134 mice was administered 20 mmol sodium
chloride. There were 93, 46, and 38 mice in the low-, mid-, and high-dose groups, respectively.
Estimates of daily doses resulting from exposure to treated drinking water were not reported.
Based on the default water intake for female B6C3Fi mice of 0.24 L/kg-day, calculated from the
default body weight in an allometric equation (U.S. EPA, 1988), the estimated doses were 0, 78,
262, and 784 mg/kg-day. Drinking water consumption was monitored during the first 4 weeks of
exposure. Body weights were monitored throughout the study. At sacrifice, livers were
collected, weighed, and processed for histopathologic examination.
Drinking water consumption was decreased only for the first week for the high-dose
group. Body weight was decreased beginning after 51 weeks of treatment with 20 mmol/L TCA.
Body weights were significantly decreased (p < 0.05) by approximately 10% on sporadic
occasions beginning at week 51 until study termination. Relative liver weight increased with
dose (linear regression coefficient, r = 0.991). The relative liver weights of the high-dose group
increased by approximately 40% over controls at 51 weeks, and liver weights for the mid- and
high-dose groups increased by approximately 25 and 60% over controls, respectively, after
82 weeks. EPA determined the NOAEL to be 2.0 mmol/L (78 mg/kg-day) and the LOAEL to be
6.67 mmol/L (262 mg/kg-day) based on increased in liver weight. This study was not designed,
however, to evaluate noncancer effects of TCA. The identification of a LOAEL at 6.67 mmol/L
based on liver weight is supported by short-term studies in B6C3Fi mice that have reported some
evidence for glycogen accumulation (Sanchez and Bull, 1990), increased DNA synthesis in
53 DRAFT - DO NOT CITE OR QUOTE
-------
hepatocytes (Dees and Travis, 1994), and peroxisome proliferation (Parrish et al., 1996) at TCA
doses that increased liver weights.
The incidence of hepatocellular carcinoma was significantly increased (p < 0.05) at
20 mmol/L (784 mg/kg-day) after 51 weeks (control: 0/40, 0%; 2.0 mmol/L [78 mg/kg-day]:
0/40, 0%; 6.67 mmol/L [262 mg/kg-day]: 0/19, 0%; 20.0 mmol/L [784 mg/kg-day]: 5/20, 25%).
At 82 weeks, the incidence of foci of altered hepatocytes was significantly increased at 6.67 and
20.0 mmol/L (10/90, 11.1%; 10/53, 18.9%; 9/27, 33.3%; 11/18, 61.1%, respectively). The
incidence of hepatocellular adenomas was significantly increased at 20.0 mmol/L (2/90, 2.2%;
4/53, 7.6%; 3/27, 11.1%; 7/18, 38.9%, respectively), and the incidence of hepatocellular
carcinomas was significantly increased at 6.67 and 20.0 mmol TCA (2/90, 2.2%; 0/53, 0%; 5/27,
18.5%; 5/18, 27.8%, respectively).
As part of experiments designed to evaluate if TCA alone was responsible for
TCE-induced liver tumors, Bull et al. (2002) exposed 40 male B6C3Fi mice to neutralized TCA
in drinking water at 2 g/L for 52 weeks (Experiment 1) and 20 male mice at 0.5 or 2 g/L for
52 weeks (Experiment 2). Controls (12 in Experiment 1 and 20 in Experiment 2) were given
untreated drinking water. After exposure, animals were sacrificed and livers were removed,
weighed, grossly examined, and processed for histopathologic examination. No other tissues
were examined histologically. The estimated doses resulting from exposure to these
concentrations were not reported. However, based on reference water intake of 0.24 L/kg-day
for male B6C3Fi mice (U.S. EPA, 1988), the estimated doses used in this study were 0, 120, and
480 mg/kg-day. Groups of animals were also exposed to TCE, DC A, and various concentrations
of a mixture of DCA and TCA. Those results are not fully discussed in the context of this
toxicological review.
Random tumor samples were stained with an anti c-jun antibody; all tumors were
analyzed for mutation frequency and spectra of the H-ras codon 61; and these results were
compared with those from DCA- and TCE-induced tumors. Proteins involved in the mitogen-
activated protein kinase-signaling cascade (Ras, MeK, active Erkl/2, and c-fos) were examined
by western blotting in order to determine if the three common codon 61 mutations of ras had
different effects on downstream effectors. Tumor incidence and multiplicity were significantly
(p < 0.05) greater than controls at all TCA exposure concentrations. Tumor incidence in animals
exposed to TCA at 2 g/L for 52 weeks (Experiment 1) was 33/40 compared with 4/12 in
controls; tumor incidences in mice exposed to TCA at 0.5 or 2 g/L for 52 weeks (Experiment 2)
were 11/20 and 9/20, respectively, compared with an incidence of 1/20 in controls. All tumor
cells from TCA-treated mice were nonreactive with the c-jun antibody (c-jun^), which is
consistent with previous reports (Stauber and Bull, 1997). The mutation frequency at H-ras
codon 61 in TCA-induced tumors (44%) was lower than the frequency of codon 61 mutations
(56%) in spontaneous liver tumors in B6C3Fi mice but higher than that in TCE-induced tumors
(21%). The H-ras mutation spectrum of TCA-induced tumors did not differ significantly from
54 DRAFT - DO NOT CITE OR QUOTE
-------
that of historical controls. TCA had no effect on activation of the mitogen-activated protein
kinase cascade.
4.2.2.1.2.2. Tumor promotion studies. Herren-Freund et al. (1987) investigated the
initiation/promotion potential of TCA in male B6C3Fi mice (22-33/group). At 15 days of age,
mice were pretreated with a single i.p. dose of ethylnitrosourea (ENU) as a tumor initiator at
doses of 0 mg/kg (uninitiated control, treated with 2 uL/g sodium acetate and 5 g/L TCA),
2.5 mg/kg (2 and 5 g/L TCA groups), or 10 mg/kg (5 g/L TCA group only). Following
pretreatment, TCA was administered in the drinking water at concentrations of 2 or 5 g/L (500 or
1,250 mg/kg-day) as calculated by using a subchronic water intake factor of 0.25 L/kg-day (U.S.
EPA, 1988) from 4 to 65 weeks of age. The negative control groups for tumor promotion (22-
23 animals/group) received 2 g/L sodium chloride in drinking water and 0, 2.5, or 10 mg/kg
ENU. The mice were sacrificed after 61 weeks of exposure. Survival data were not reported.
Significant decreases of 9-12% in final mean body weight were observed in the 5 g/L
TCA groups relative to the corresponding sodium chloride control. Absolute and relative liver
weights were significantly increased (by 41-73%) in all TCA treatment groups relative to the
corresponding sodium chloride control group. The incidences of hepatocellular adenomas and
hepatocellular carcinomas were significantly increased in the uninitiated group receiving 5 g/L
TCA when compared with the uninitiated sodium chloride control group (see Table 4-9). The
incidences of hepatocellular adenomas and hepatocellular carcinomas were significantly
increased in the TCA groups initiated with 2.5 mg/kg ENU. Mice initiated with 10 mg/kg ENU
and then administered 5 g/L TCA also showed an increase in the incidence of hepatocellular
carcinomas, although the increase was not statistically significant. Thus, TCA enhanced the
incidence of hepatocellular adenomas and carcinomas above control levels, with or without prior
initiation. The study authors concluded that TCA acted as a complete carcinogen in B6C3Fi
mice.
55 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-9. Incidence of adenomas and hepatocellular carcinomas in B6C3Fi
mice treated with ENU and TCA
ENU(jig/g
body wt)
0
2.5
0
2.5
2.5
10.0
TCA
(as promoter)
(g/L)
Sodium chloride
Sodium chloride
5
2
5
5
Incidence of
adenomas
(%)
2/22 (9)
1/22 (5)
8/22 (3)b
ll/33(33)b
6/23 (26)b
11/28(39)
No. adenomas
per animal3
0.09 ±0.06
0.05 ±0.05
0.50±0.16b
0.42±0.12b
0.30±0.12b
0.61 ±0.16
Incidence of
carcinomas
(%)
0/22 (0)
1/22 (5)
7/22 (32)b
16/33 (48)b
ll/23(48)b
15/28 (54)
No. carcinomas
per animal3
0
0.05 ±0.05
0.50±0.17b
0.64±0.14b
0.57±0.21b
0.93 ±0.22
"The number of adenomas per animal and the number of carcinomas per animal are expressed as the
mean ± standard error of the mean.
bSignificantly different from the corresponding sodium chloride control group not treated with TCE, DCA, or PB as
determined by the Fisher's exact test; p < 0.01.
Source: Herren-Freundetal. (1987).
Pereira and Phelps (1996) assessed liver tumor promotion activity by TCA in female
B6C3Fi mice. Test animals were treated with 25 mg/kg of the tumor initiator N-methyl-
N-nitrosourea (MNU) at 15 days of age or given 4 mL/kg sterile saline (vehicle control).
Starting at 7 weeks of age, animals were administered neutralized TCA in drinking water at
concentrations of 0, 2.0, 6.67, or 20.0 mmol/L (0, 327, 1,090, or 3,268 mg/L) for either 31 weeks
(n = 8-15/group) or 52 weeks (n = 39 for MNU controls, 40 for the low-dose TCA-only group,
19 for the mid- and high-dose TCA-only groups, and 6-23 for TCA + MNU groups). Dose
estimates were not reported by the study authors. The drinking water concentrations used
resulted in doses of approximately 0, 78, 262, or 784 mg/kg-day based on the default drinking
water value of 0.24 L/kg-day for female B6C3Fi mice (U.S. EPA, 1988). A recovery group
(n = 11) was removed from treatment after 31 weeks and retained for an additional 21 weeks.
At 31 weeks, treated animals exhibited a slight, dose-related linear increase in relative
liver weights. At 31 and 52 weeks, no significant increase in foci of altered hepatocytes,
adenomas, or carcinomas was observed in mice that received MNU only. In mice administered
TCA but not initiated with MNU, the only tumorigenic response was a slight increase in the
yield of hepatocellular carcinomas/animal (0.50 tumors/mouse) in the high-dose group
(784 mg/kg-day) after 52 weeks of treatment. Animals initiated with MNU and treated with
TCA exhibited an increase in liver tumors following both 31 and 52 weeks of exposure in the
784 mg/kg-day group and following 52 weeks of exposure in the 262 mg/kg-day group. Both
the numbers of adenomas/mouse and carcinomas/mouse were statistically elevated as compared
with controls, and the tumor yield generally increased with exposure duration from 31 to
52 weeks. However, there was no significant increase in the yield of altered hepatocyte foci at
either time point in any dose group. The concentration-response relationships for total
56
DRAFT - DO NOT CITE OR QUOTE
-------
lesions/mouse (foci plus tumors) after both 31 and 52 weeks of treatment were best described by
a linear regression line.
When exposure to 784 mg/kg-day TCA was terminated after 31 weeks and the animals
were held for an additional 21 weeks, the yield of tumors/mouse remained stable (1.50 and
1.64 at 31 and 52 weeks, respectively). However, the yield of hepatocellular carcinomas
increased from 0.20/mouse in mice exposed for 31 weeks to 0.73/mouse in mice held to
52 weeks. When treatment continued between weeks 31 and 52, the yield of tumors/mouse rose
from 1.50 at 31 weeks to 4.21 at study termination. These findings indicate that, although the
occurrence of additional TCA-promoted tumors was dependent on continuous treatment, the
stability and progression to carcinoma appeared to be independent of further treatment.
Histochemical staining indicated that >71% of tumors promoted with either 262 or 784 mg/kg-
day TCA were basophilic and did not contain GST-u, a phase II conjugation enzyme highly
expressed in some tumor types, except for very small areas comprising <5% of the tumor. The
predominantly basophilic nature of the tumors promoted by TCA is consistent with the character
of lesions induced by tumorigenic compounds that are rodent peroxisomal proliferators, but
"spontaneous" liver tumors in mice have also been reported to be predominantly basophilic and
lacking GST-u (Pereira and Phelps, 1996).
Pereira et al. (2001) administered MNU to B6C3Fi mice (16 males and 14 females) via
i.p. injection at 30 mg/kg, then exposed the MNU-initiated mice to TCA at 4 g/L in the drinking
water for 31 weeks. Based on reference drinking water intake values for B6C3Fi mice (0.25 and
0.24 L/kg-day for males and females, respectively), male and female mice received
approximately 1,000 and 960 mg/kg-day, respectively. After the treatment period, the liver and
kidneys were removed, weighed, and microscopically examined. The study was designed to
evaluate the effects of chloroform on TCA-induced tumor promotion, and only the TCA-only
treated groups are discussed in this review. Relative liver weight was significantly (p < 0.001,
75% in males and 35% in females) increased compared with controls. A significant (p < 0.05)
increase in the number of mice with liver tumors (adenomas + adenocarcinomas) was observed
in TCA-treated males initiated with MNU (incidence of 13/16 compared with 2/8 MNU-treated
controls). These tumors were >97% basophilic. Although an increase was also observed in
females (incidence of 6/14 compared with 2/29 controls), the increase was not statistically
significant (p < 0.05). Similarly, an increase in kidney tumors was also observed in male mice
initiated with MNU and promoted by TCA (incidence of 0/8 in MNU-only treated controls
compared with an incidence of 14/16 in MNU + TCA treated mice). Incidences of kidney
tumors in female mice were not significantly increased compared with MNU-treated controls
(incidence not reported). The study authors also investigated hypomethylation of the c-myc gene
in liver and kidney tumors from TCA-treated mice. These results are discussed in Section 4.5.1.
In a study designed to compare the promotion of liver tumors in TCA- and DCA-treated
mice initiated with MNU, Pereira et al. (1997) exposed female B6C3Fi mice (20-45/dose) to
57 DRAFT - DO NOT CITE OR QUOTE
-------
TCA at 6 or 25 mmol/L in drinking water with or without addition of various concentrations of
DCA for 44 weeks. Based on reference water intake for female B6C3Fi mice of 0.24 L/kg-day
(U.S. EPA, 1988), the estimated doses were 0, 235, and 980 mg/kg-day. Body weight was
monitored throughout the study. Livers were removed, weighed, and microscopically examined
for presence of tumors. Liver sections were also stained immunohistochemically for GST-u. A
significant increase in adenomas was observed in TCA-only treated mice at 25 mmol/L
(0.52 tumors/mouse compared with 0.07 tumors/control mouse) but not at 6 mmol/L
(0.15 tumors/mouse). The tumors from TCA-treated mice were exclusively basophilic and were
generally without GST-u (with the exception of four carcinomas at 25 mmol/L TCA), which is
consistent with the results reported by Pereira and Phelps (1996). In contrast, tumors from
DCA-treated mice were primarily eosinophilic and were positive for GST-u. When TCA and
DCA were administered together (25 mmol/L TCA +15.6 mmol/L DCA), the tumor yield
increased synergistically. At the lower concentration, the relationship was at least additive. The
tumors in the livers from mice treated with DCA + TCA were more consistent with the
characteristics of DCA-induced livers (eosinophilic and containing GST-7i). These data suggest
that TCA and DCA both promote tumor formation; however, the different tumor characteristics
are consistent with the conclusion that the mechanisms for the tumor-promoting activity of each
compound are different.
4.2.2.2. Inhalation Studies
No chronic toxicity studies or cancer studies in animals exposed by inhalation to TCA are
available.
4.2.2.3. Studies Using Other Routes of Exposure
Von Tungeln et al. (2002) evaluated the neonatal tumorigenicity of TCA in B6C3Fi mice
(23-24 animals/sex/dose) in two bioassays. For each assay, TCA was dissolved in dimethyl
sulfoxide and administered via i.p. injections at 8 and 15 days of age. In Assay A, neonatal mice
were given a total dose of 2,000 nmol (approximately 33 mg/kg based on a reference body
weight of 0.01 kg for B6C3Fi mice at weaning) (U.S. EPA, 1988) and were sacrificed at
12 months of age. In Assay B, neonatal mice were given a total dose of 1,000 nmol
(approximately 16 mg/kg) and were sacrificed at 20 months of age. 4-Aminobiphenyl was used
as the concurrent positive control (22-24 mice/sex/dose) and total doses of 1,000 and 500 nmol
were given by i.p. injection for Assays A and B, respectively. Dimethyl sulfoxide solvent
control groups (23-24 mice/sex) were included in each assay. Body weight (at 28-day intervals)
and mortality were evaluated in all treatment groups. At sacrifice, all test animals were
necropsied for gross tumor count, microscopic examination of tissues, and histopathologic
diagnoses. No unscheduled deaths occurred in Assay A. In Assay B, one mouse each died in the
male and female solvent control groups and in the female TCA group. A marginal increase (not
58 DRAFT - DO NOT CITE OR QUOTE
-------
statistically significant) in liver tumors was observed in TCA-treated males in Assay A (4/24)
when compared with the control group (1/24). The incidence of liver tumors in TCA-treated
males in Assay B (5/23) was less than in the control group (7/23). No tumors were observed in
dimethyl sulfoxide-treated control females in either assay. The study authors concluded that
TCA did not induce significant tumor incidences when compared with the dimethyl sulfoxide
controls. In contrast, all male mice treated with 4-aminobiphenyl (the positive control substance)
in Assays A and B developed liver tumors, and 9/22 male mice in Assay B also developed lung
tumors. Nine of 23 female mice treated with 4-aminobiphenyl in Assay B developed liver
tumors; no tumors were diagnosed in female mice dosed with 4-aminobiphenyl in Assay A.
In a related mechanistic study, von Tungeln et al. (2002) dosed an additional group of
male neonatal B6C3Fi mice with TCA to evaluate TCA-induced formation of MDA-derived
deoxyguanosine adducts and 8-OHdG in hepatic DNA in relation to TCA tumorigenicity. This
study was conducted because previous results from the same laboratory had shown that in vitro
metabolism of TCA by hepatic microsomes isolated from adult mice results in lipid peroxidation,
with subsequent production of MDA (Ni et al., 1996) (see Section 3.3), and metabolism of TCA
in the presence of calf thymus DNA resulted in the formation of MDA-derived deoxyguanosine
adducts (Ni et al. [1995], as cited in von Tungeln et al. [2002]). In addition, TCA induces
formation of 8-OHdG (see Section 4.2.1.1), and induction of elevated levels of 8-OHdG may
induce tumors (Wagner et al., 1992).
Male neonatal B6C3Fi mice (number of animals treated not stated) were given a total
dose of 2,000 nmol TCA by i.p. injection as described for the neonatal mice cancer assays
summarized above (von Tungeln et al., 2002). The test animals were sacrificed 1, 2, or 7 days
after the final TCA treatment at 15 days of age, and liver tissue was collected for extraction of
DNA and determination of levels of MDA-derived deoxyguanosine and 8-OHdG. TCA induced
a significant (p < 0.05) increase in MDA-derived deoxyguanosine adduct formation in liver DNA
at 24 and 48 hours (but not at 7 days) after the final dose. The increase was approximately 190%
of the control value at each time point. TCA treatment also resulted in a significant (p < 0.05)
increase in 8-OHdG formation in liver DNA at 24 and 48 hours and at 7 days after
administration of the final dose. The magnitude of the increase was approximately 2.5-fold
greater than the control values. Because TCA was not carcinogenic in the neonatal cancer
bioassays conducted by von Tungeln et al. (2002), these results suggest that neonatal B6C3Fi
mice are not sensitive to either TCA-induced lipid peroxidation or oxidative stress as an MOA
for tumor induction under the experimental conditions used in these studies. The study authors
speculated that TCA was negative in their neonatal cancer bioassays because it may act as a cell
proliferator. According to this hypothesis, liver cells were already replicating at a very high rate
in the neonatal mice when TCA was administered; therefore, any additional cell proliferation
induced by TCA may have been negligible in comparison with the existing rate of proliferation.
59 DRAFT - DO NOT CITE OR QUOTE
-------
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES - ORAL AND INHALATION
4.3.1. Reproductive Studies
The effect of TCA on in vitro fertilization was examined in B6D2Fi mouse gametes
(Cosby and Dukelow, 1992). TCA was constituted in a culture medium to yield concentrations
of 100, 250, or 1,000 ppm on a volume/volume basis (approximately 0.98, 2.4, and 9.8 mM) and
incubated with mouse oocytes and sperm for 24 hours. Each culture dish was subsequently
scored for percentage of oocytes fertilized. The percent of oocytes fertilized was significantly
decreased from 82% for controls to 77.3% when oocytes and sperm were placed in a medium
containing 2.4 mM TCA. At 9.8 mM TCA, only 53.1% of the oocytes were fertilized; the
decrease was statistically significant (p < 0.001).
4.3.2. Developmental Studies
4.3.2.1. Oral Developmental Studies
Seven studies have evaluated the potential of TCA to induce developmental toxicity in
rats (Table 4-10). In addition, one study has been conducted to identify embryonic genes, which
undergo changes in expression (up- or down-regulation) in response to maternal TCA exposure.
No studies in other test species (e.g., mice or rabbits) were located.
60 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-10. Summary of developmental studies evaluating effects of TCA after oral administration in rats
Reference
Smith et al.,
1989
Johnson et
al., 1998
Fisher et al.,
2001
Species
Long-Evans rats
(20-2 I/dose)
Sprague-
Dawley rats
(55 controls and 11
TCA treated)
Sprague-Dawley rats
(19/dose)
Exposure
route
Oral, gavage
Oral,
drinking
water
Oral, gavage
Exposure
duration
CDs 6-15
CDs 1-22
CDs 6-15
Doses evaluated
0, 330, 800,
1,200, or
1,800 mg/kg-d
in distilled
water, adjusted
to pH 7 with
NaOH
0 or 291 mg/kg-
d in distilled
water, titrated to
pH 7 with NaOH
0 or 300 mg/kg-
d in water,
adjusted to
pH7.5
Effects3
Decreased fetal weight,
decreased crown-rump
length, increased
incidence of soft-tissue
malformations and
cardiovascular
malformations,
increased maternal
spleen and kidney
weights
Increased cardiac
malformations, number
of implantation
sites/litter, number of
resorption sites/litter,
and total number of
resorptions among
treated dams
Decreased maternal
weight gain, reduced
fetal body weight
NOAEL
(mg/kg-d)
Maternal: not
determined
Developmental:
not determined
Maternal: not
determined
Developmental:
not determined
Maternal: not
determined
Developmental:
not determined
LOAEL
(mg/kg-d)
Maternal: 330
Developmental:
330
Maternal: 291
Developmental:
291
Maternal: 300
Developmental:
300
Comments
Dose estimated by the
authors, based on the
average amount of water
consumed by the animals on
a daily basis. The tested
concentration/dose was also
a maternal LOAEL for
decreased weight gain.
Study was not adequately
designed and/or reported,
and a complete array of
standard developmental
endpoints was not assessed.
Cardiac defects were the
only visceral malformation
evaluated; maternal toxicity
indicated by decreased body
weight gain for GDs 7-15
and 18-21; mean uterine
weight was also
significantly less (p < 0.05)
than controls.
61
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-10. Summary of developmental studies evaluating effects of TCA after oral administration in rats
Reference
Singh,
2005a
Singh,
2005b
Singh, 2006
Warren et
al., 2006
Species
Inbred Charles
Foster rats
(6-12/group)
Inbred Charles
Foster rats
(6-12/group)
Inbred Charles
Foster rats
(6-12/group)
Sprague-Dawley
Crl:CDR (SD) BR
rats
Exposure
route
Oral, gavage
Oral, gavage
Oral, gavage
Oral, gavage
Exposure
duration
CDs 6-15
CDs 6-15
CDs 6-15
CDs 6-15
Doses evaluated
0, 1,000, 1,200,
1,400, 1,600, or
1,800 mg/kg-d
in distilled
water, adjusted
to pH 7.0-7.5
0, 1,000, 1,200,
1,400, 1,600, or
1,800 mg/kg-d
in distilled
water, adjusted
to pH 7.0-7.5
0, 1,000, 1,200,
1,400, 1,600, or
1,800 mg/kg-d
in distilled
water, adjusted
to pH 7.0-7.5
0 or 300 mg/kg-
d in water
Effects3
Increase in post-
implantation loss,
decreased fetal testes
weight, reduction in the
length and diameter of
the seminiferous
tubules, increased
apoptosis of the
gonocytes
Decrease in fetal
ovaries weight with
increasing dose;
decrease in the number
of oocytes and the size
of the ovaries,
apoptosis of oocytes
Decrease in maternal
weight gains; decrease
in fetal weight and fetal
brain weight;
hydrocephalus,
vacuolation, and
hemorrhages in fetal
brains
No significant decrease
in maternal body
weight; significant
decrease in fetal body
weight, no eye
malformation, no
significant reductions
in lens area, globe area,
medial canthus
distance, or interocular
distance
NOAEL
(mg/kg-d)
Developmental:
not determined
Developmental:
1,200
Maternal:
1,000
Developmental:
not determined
Developmental:
not determined
LOAEL
(mg/kg-d)
Developmental
(increase in
post-
implantation
loss): 1,000
Developmental
(effect on fetal
ovary): 1,400
Maternal:
1,200
Developmental
(effect on fetal
brain): 1,000
Developmental:
300
Comments
Only evaluated effects on
fetal testes. The most
sensitive effect was
postimplantation loss.
Maternal toxicity was not
reported.
Only evaluated effects on
fetal ovaries. Maternal
toxicity was not reported.
Focused only on effects of
TCA on fetal brains.
Maternal toxicity was not
reported.
Focused on eye
malformations and
microphthalmia in fetal rats.
aThe effects listed in this table may have occurred either at the LOAEL or at higher doses.
62
DRAFT - DO NOT CITE OR QUOTE
-------
Smith et al. (1989) dosed pregnant Long-Evans rats (20-21/dose) with 0, 330, 800, 1,200,
or 1,800 mg/kg-day TCA (adjusted to pH 7 by NaOH) by gavage on GDs 6-15. Clinical signs of
toxicity and body weight gain were monitored throughout the exposure period. The dams were
sacrificed on GD 20. The liver, spleen, and kidneys were removed and weighed. The uterine
horns were examined for the number and location of fetuses or resorption sites. The fetuses were
subsequently removed and weighed, measured, sexed, and evaluated for external malformations.
Two-thirds of each litter was preserved for evaluation of visceral abnormalities. The remaining
one-third of the fetuses was reserved and processed for evaluation of skeletal abnormalities.
Evidence of maternal toxicity was observed in all TCA treatment groups as indicated by a
significant (p < 0.05) increase in spleen (up to 74% increase) and kidney (up to 24% increase)
weights when compared with the control group. Unadjusted mean terminal (GD 20) body
weights were significantly reduced (5-12%; p < 0.05) at all doses, but no statistically significant
differences were observed in average percent maternal weight gain when adjusted for gravid
uterine weight. Dams exposed to 800, 1,200, or 1,800 mg/kg-day had significantly (p < 0.05)
decreased body weight gains on GDs 6-9 and 15-20 (up to a 54% decrease). The weight change
for GDs 15-20 may have been influenced by reductions in fetal body weight. The number of
litters totally resorbed was significantly increased (5/21 and 12/20, respectively), and the number
of viable litters (14/21 and 8/20, respectively) was significantly decreased at 1,200 and
1,800 mg/kg-day. Developmental effects were observed at all doses (Table 4-11) and included
significant (p < 0.05) decreases in mean fetal weight per fetus (up to a 33% decrease in males
and females); significant decreases in fetal crown-rump length (up to a 15% decrease in males
and females); and increased percentages of fetuses affected per litter with total soft-tissue
malformations. Most of the total soft tissue malformations were cardiovascular malformations,
and in particular levocardia (or extremely left-sided heart). The authors noted that the Long-
Evans strain of rat appears somewhat susceptible to this alteration. They also observed that
levocardia is an ill-defined malformation and of trivial appearance as found in Bouins fixed
sections. The maternal and developmental LOAELs in this study are 330 mg/kg-day. Maternal
and developmental NOAEL values for TCA could not be determined because adverse effects
were observed at all tested doses.
63 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-11. Selected data for fetal anomalies, showing dose-related trends
following exposure of female Long-Evans rats to TCA on GDs 6-15
Type
Maternal body weight on
GD20
Dose (mg/kg-d)
0
344.97 ±19.04a
330
327.12 ±25.42a
800
323.87 ±19.04a
1,200
306.88 ±29.09a
1,800
303.40 ±27.03a
Malformations: mean % fetuses affected per litter ± SD (number of litters affected/number examined)b
Total soft tissue (visceral)
Cardiovascular
3. 50 ±8.7
(4/26)
0.96 ±4.9
(1/26)
9.06 ± 12.9a
(8/19)
5.44±10.0a
(6/19)
30.37 ±28.1a
(15/17)
23.59±28.0a
(12/17)
55.36±36.1a
(12/14)
46.83 ±36.5a
(11/14)
96.88 ±8.8a
(8/18)
94.79 ±9.9a
(8/8)
Levocardia. number of fetuses or litters affected/number examined0
Fetal incidence
Litter incidence
0/196
0/26
9/151
6/19
20/111
12/17
24/69
10/14
17/22
7/8
Intraventricular septal defect, number of fetuses or litters affected/number examined0
Fetal incidence
Litter incidence
0/196
0/26
0/151
0/19
6/111
4/17
3/69
3/14
5/22
5/8
Fetal crown-rump length (cm): mean ± SDd
Male
Female
3.71 ±0.12
3.64 ±0.15
3.58±0.10a
3.53±0.09a
3.46±0.10a
3.38±0.12a
3.36±0.15a
3.33±0.16a
3.16±0.12a
3.15±0.15a
Mean fetal body weight (g): mean ± SD°
Male
Female
3.70 ±0.24
3.54 ±0.20
3.20±0.26a
3.08±0.27a
2.98±0.17a
2.83±0.18a
2.74±0.30a
2.67 ± 0.29a
2.49±0.16a
2.36±0.15a
"Significantly different from control (p < 0.05) as reported by Smith et al. (1989).
bTable 5 of Smith et al. (1989).
"Table 6 of Smith et al. (1989).
dTable 4 of Smith et al. (1989).
Source: Smith etal. (1989).
Johnson et al. (1998) evaluated the teratogenicity of TCA (adjusted to pH 7 by NaOH) by
exposing pregnant Sprague-Dawley rats to 0 (n = 55) or 2,730 (n = 11) mg/L TCA in neutralized
drinking water on GDs 1-22. The authors estimated the doses to be 0 or 291 mg/kg-day, based
on the average amount of water consumed by the animals on a daily basis and measured body
weights. Maternal toxicity was evaluated by clinical observation and maternal weight gain.
Dams were sacrificed on GD 22, and implantation sites, resorption sites, fetal placements, fetal
weights, placental weights, fetal crown-rump lengths, gross fetal abnormalities, and abnormal
fetal abdominal organs were recorded. In addition, the fetal hearts were removed, dissected, and
examined microscopically for abnormalities by using a detailed microdissection cardiac
evaluation technique.
No signs of maternal toxicity were reported. Although the authors reported that the
weight gain during pregnancy of treated females was not significantly different from controls,
the average maternal weight gain for TCA-exposed animals was 84.6 g as compared with 122 g
64
DRAFT - DO NOT CITE OR QUOTE
-------
for control animals, representing a 30% decrease in maternal body weight gain. No measure of
variation around the mean (e.g., SD or standard error) was reported, and it is not clear why this
reduction was not reported as statistically significant. Nonetheless, a decrease of this magnitude
in body weight gain during pregnancy is considered to be lexicologically significant. Average
drinking water consumption was reported as 38 mL/day in treated rats as compared with
46 mL/day in control rats; this difference was not reported as statistically significant, but it was
unclear from the publication whether a statistical analysis was performed.
Statistically significant increases were reported in the average number of resorption sites
(2.7 resorptions/litter in treated animals, compared with 0.7 in the controls), total number of
resorptions (30 resorptions reported among 11 treated females as compared with 40 resorptions
among 55 control females), and average number of implantation sites (defined as sites where the
fetus was implanted but did not mature) (1.1 implantation sites/litter, compared with 0.2 in the
controls). In treated groups, the total number of fetuses reported was 115 in 11 rats, resulting in
an average number of fetuses/litter of 10.5. In the control group, the total number of fetuses was
reported as 605 in 55 rats, with an average number of fetuses/litter of 11.3. These differences
were not reported as statistically significant. The number of maternal rats with abnormal fetuses
was 7 out of 11 for TCA-treated animals as compared with 9 out of 55 for controls. No
significant differences were reported in the numbers of live or dead fetuses, fetal weight,
placental weight, fetal crown-rump length, fetal external morphology, or fetal gross external or
noncardiac internal congenital abnormalities; however, data for these endpoints were not
reported in the paper and could not be independently assessed.
Cardiac abnormalities were evident in 10.5% of the fetuses in the TCA group, compared
with 2.15% of the controls. Although these results were not reported in terms of the more
appropriate measure of number of affected litters, Johnson et al. (1998) stated that the incidence
of cardiac malformations was significantly greater in treated rats as compared with control rats
on both a per-fetus basis (p = 0.0001) and a per-litter basis (p = 0.0004). Complete fetal
examinations for internal or skeletal abnormalities were not conducted, and the study is limited
by the small size of the exposed group and the use of only one dosed group. Based on the
lexicologically significant decrease in maternal body weight, 291 mg/kg-day is considered to be
a maternal LOAEL. Based on an increase in cardiac malformations occurring at a maternally
toxic dose, the developmental LOAEL is 291 mg/kg-day. The maternal and developmental
NOAELs could not be determined because adverse effects were observed at the only dose tested.
In contrast to the results of Smith et al. (1989) and Johnson et al. (1998), Fisher et al.
(2001) did not observe significant differences in the fetal or litter incidence of heart
malformations following administration of neutralized TCA in distilled water to groups of
pregnant Sprague-Dawley rats (n = 19). Doses of 0 or 300 mg/kg-day were given by gavage on
GDs 6-15. Vehicle control animals (n = 19) received distilled water. Positive control animals
(n = 12) received all-trans retinoic acid (15 mg/kg-day) dissolved in soybean oil. On GD 21,
65 DRAFT - DO NOT CITE OR QUOTE
-------
body weight, uterine weight, number and viability of fetuses, and number of implantation and
resorption sites were recorded for each pregnant animal. All treated rats were then sacrificed,
full-term fetuses were removed, and the following parameters were recorded: sex, fetal weight
(per fetus and per litter), percent of dams with an early resorption, and number of fetuses per
dam. The heart of each full-term fetus was thoroughly examined in situ and then removed,
sectioned, and microscopically examined for cardiac malformations by using a detailed cardiac
microdissection technique that included staining of fetal heart tissue for detection of
malformations.
The single dose evaluated produced maternal toxicity as indicated by decreased body
weight gain on GDs 7-15 and 18-21 (p < 0.05, approximately 17% relative to controls). Mean
uterine weight was significantly less than controls (p < 0.05, 9%). The number of implantations,
percent of dams with an early resorption, and number of fetuses per litter were similar to control
values. Mean fetal body weight (per litter and per fetus) on GD 21 was significantly less than
that of controls (p < 0.05, approximately 8%). The heart malformation incidence in the
TCA-treated group was similar to that of controls; 3.3% (9/269) of the fetuses and 42% (8/19) of
the litters from TCA-treated animals were affected compared with 2.9% (8/273) of fetuses and
37% (7/19) of litters from control animals. Maternal exposure to the positive control (all-trans
retinoic acid) significantly increased the incidence of cardiac defects when analyzed on a
per-fetus (32.9%) or per-litter basis (92%) when compared with the corresponding soybean oil
vehicle fetal and litter control incidences (6.5 and 52%, respectively). These data identify a
maternal LOAEL of 300 mg/kg-day based on significantly reduced body weight gain and uterine
weight. A developmental LOAEL of 300 mg/kg-day was identified, based on significantly
reduced mean fetal body weight on a per-litter and per-fetus basis. Maternal and developmental
NOAEL values were not identified in this single dose study because adverse effects were noted
at the only dose tested.
Singh (2006, 2005a, b) treated pregnant inbred Charles Foster rats (6-12 rats/dose group;
control group = 25) with 0, 1,000, 1,200, 1,400, 1,600, or 1,800 mg/kg-day TCAby gavage on
GDs 6-15 and examined the effect of TCA on the developing testis (Singh, 2005a), developing
ovary (Singh, 2005b), and developing brain (Singh, 2006). TCA was neutralized by sodium
hydroxide to pH 7.0-7.5 before administration to rats. Control animals received distilled water
by gavage. The pregnant rats were sacrificed on GD 19, and the fetuses and placenta were
collected for examination. The testes of each pup of different dose groups were dissected out,
weighed, and subjected to histological examination (Singh, 2005a). Other than reporting
maternal weight gain on GD 19, maternal toxicity findings were not reported. Percentage of
postimplantation loss was significantly increased in a dose-related manner (22% at 1,000 mg/kg-
day versus 3% for control group). No external abnormalities were observed. The average
weights of the fetal testes were significantly reduced when compared to the control at
> 1,200 mg/kg-day. Histological examination of fetal rat testes of the 1,200 mg/kg-day dose
66 DRAFT - DO NOT CITE OR QUOTE
-------
group revealed a reduction in the diameter of the seminiferous tubules, which only occupied the
peripheral region. This effect was more pronounced in the higher dosed groups. At the higher
doses, reduction in length of the seminiferous tubules was also reported. Examination of the
testes at higher magnification revealed increased apoptosis of the gonocytes as well as the Sertoli
cells within the seminiferous tubules in comparison to the controls at > 1,200 mg/kg-day.
The rat fetal ovaries of each pup of different dose groups from the above study were also
dissected out, weighed, and subjected to histological examination (Singh, 2005b). The average
weights of the ovaries were significantly reduced for the dose groups >1,400 mg/kg-day.
Histological examination of the fetal ovaries showed small size cells with less prominent nuclei
at the coelomic epithelium with > 1,400 mg/kg-day TCA. The cortical cords proliferating from
the coelomic epithelium traversing the gonads were either shortened or lacking. Oocytes in the
ovarian stroma showed shrinkage in size with distorted cell membrane and indistinct nucleus,
suggestive of cell apoptosis. The number of oocytes and the size of ovary were reduced. Singh
(2005b) suggested that the gonadal changes were due to anoxia and oxidative stress resulting
from TCA exposure.
The rat fetal brains of different dose groups from the above study were evaluated (Singh,
2006). Maternal weight gains were statistically significantly decreased at TCA doses
> 1,200 mg/kg-day (38-46%). Mean fetal weight and fetal brain weight decreased significantly
at TCA doses >1,000 mg/kg-day; while the length of the fetal brain increased significantly at
1,000 and 1,200 mg/kg-day (about 10% at 1,000 mg/kg-day) but decreased significantly (8-
16%) at TCA doses >1,400 mg/kg-day when compared with controls. At doses >1,000 mg/kg-
day, the fetal brains showed hydrocephalus with breech of the ependymal lining, altered choroids
plexus architecture, and increased apoptosis. Vacuolation of the neutrophil was a prominent
feature with TCA exposure, with an incidence of 26% at 1,000 mg/kg-day (0% in controls) and
reached 100% in the 1,600 and 1,800 mg/kg-day dose groups. The incidence of brain
hemorrhages increased to 30% at TCA doses >1,200 mg/kg-day (0% in controls) and reached
100% at 1,800 mg/kg-day. The infarcts were mainly concentrated in the periventricular zone.
Singh (2006) concluded that the rat fetal brain was susceptible to the toxic effects of TCA.
In a study that evaluated if TCE, TCA, and DC A affect eye development in the Sprague-
Dawley rat (Warren et al., 2006), pregnant Sprague-Dawley Crl:CDR (SD) BR rats were
administered 0 or 300 mg/kg-day TCA by gavage on GDs 6-15. Retinoic acid (15 mg/kg-day)
was used as a positive control. A subset of the fetuses evaluated in the Fisher et al. (2001) study
was selected for ocular examination (1,185 fetuses [71%] from 108 dams). The number of
fetuses undergoing ocular examination was reduced further to approximately 30% compared to
the cardiac study. Heads of GD 21 fetuses were fixed in Bouin's solution, examined for gross
external malformations, sectioned, and subjected to computerized morphometry. For detection
of subtle eye anomalies, the following measurements on head sections were determined:
67 DRAFT - DO NOT CITE OR QUOTE
-------
interocular distance, total area of the cut surface, areas of left and right lenses, and areas of left
and right globes.
Mean fetal body weight was statistically significantly reduced in the TCA and retinoic
acid treatment groups. Mean maternal body weight was not reduced in these treatment groups
(Warren et al., 2006). Fetuses with exencephaly, anophthalmia, or microphthalmia were found
only in the retinoic acid treatment group. Mean fetal lens and globe areas were statistically
significantly reduced in the retinoic acid treatment group. However, mean lens and globe areas
and mean medial canthus and interocular distances were not reduced in the TCA-exposed fetuses
when compared with values from the control group. Thus, TCA did not appear to affect eye
development in the Sprague-Dawley rat at 300 mg/kg-day.
Collier et al. (2003) investigated the effects of TCA on gene expression in embryos
collected on GDs 10.5-11 from pregnant Sprague-Dawley rats exposed to 0, 1.63, or
16.3 mg/mL (0, 10, or 100 mM, respectively) TCA in drinking water on GDs 0-11. Estimated
intakes were 0, 200, and 2,000 mg/kg-day, based on a body weight of 0.35 kg and drinking water
rate of 0.046 L/day (U.S. EPA, 1988). The objective of the study was to identify altered
expression of genes (using a subtractive hybridization technique) that might be used as markers
of exposure to TCE or its metabolites (i.e., TCA) in the developing rat heart. Exposure to TCA
down-regulated rat ribosomal protein S10 (a housekeeping gene) and rat chaperonin 10 (a stress
response gene) and up-regulated rat Ca2+-ATPase (a calcium-responsive gene) and rat gClqBP
(function not reported). The expression of the up-regulated genes was found to be strongly heart
specific on embryonic days 10.5-11. However, no correlation between up-regulation of these
genes and occurrence of TCA-mediated cardiac defects has yet been identified.
4.3.2.2. Inhalation Developmental Studies
No studies on the developmental toxicity of TCA were identified for exposure by the
inhalation route.
4.3.2.3. In Vitro Studies
TCA has also been tested in a number of alternative screening assays for assessment of
developmental toxicity. Hunter et al. (1996) conducted a 24-hour exposure of 3-6 somite stage
CD-I mice embryos to 11 haloacetic acids, including TCA. TCA was tested at concentrations of
0, 0.5, 1, 2, 3, 4, or 5 mM. Effects on neural tube development were observed at concentrations
lower than effects on other morphological processes. Other statistically significant
dysmorphology included eye defects, pharyngeal arch defects, and heart defects. TCA produced
abnormal embryonic development at concentrations >2 mM, with a very steep dose-response
slope from 2 to 5 mM. No adverse effects were observed at <1 mM, and defects of the eyes,
arches, and heart were seen only in embryos that also had very high rates of neural tube
development abnormalities. The observed effects did not result from low pH in the culture
68 DRAFT - DO NOT CITE OR QUOTE
-------
medium, since they were not seen when HC1 was added to adjust the culture medium to similar
pH values.
The potential developmental toxicity of TC A was studied in vitro by using a rat whole-
embryo culture system by Saillenfait et al. (1995). Groups of 10-20 explanted embryos from
Sprague-Dawley rats on GD 10 were cultured for 46 hours in 0, 0.5, 1, 2.5, 3.5, 5, or 6 mM TCA.
TCA induced statistically significant, concentration-related decreases in the growth and
development parameters of conceptuses. Yolk sac diameter was significantly decreased,
beginning at a concentration of 1 mM. Other developmental measures, including crown-rump
length, head length, somite (embryonic segment) number, protein content, and DNA content,
were significantly decreased at >2.5 mM. The total number of malformed embryos was
increased beginning at 2.5 mM. At 2.5 mM, 55% of the embryos had brain defects, 50% had eye
defects, 32% had reduced embryonic axes, 55% had reductions in the first branchial arch, and
36% had otic (auditory) system defects.
TCA has also been evaluated in developmental toxicity screening assays in
nonmammalian systems. TCA was evaluated using the FETAX assay in a study that assessed
the developmental toxicity of TCE and its metabolites (Fort et al., 1993). Early Xenopus laevis
embryos were exposed to a range of TCA concentrations for 96 hours. The culture stock
solution was buffered to pH 7.0. The median lethal concentration was 4,060 mg/L (24.8 mM)
and the median effective concentration (ECso) for malformations was 1,740 mg/L (10.6 mM).
Malformations were observed at concentrations >1,500 mg/L (9.2 mM) and included gut
miscoiling, craniofacial defects, microphthalmia, microcephaly, and various types of edema.
The assay was conducted with high TCA concentrations, and is of limited value for evaluating
developmental toxicity of TCA.
4.4. OTHER DURATION- OR ENDPOINT-SPECIFIC STUDIES
4.4.1. Immunological Studies
The available information on the potential for TCA to affect the immune system is
limited. Mather et al. (1990) (described in Section 4.2) did not observe any effects on several
immunotoxicity parameters, including antibody production, delayed hypersensitivity, natural
killer cell cytotoxicity, and production of prostaglandin E2 and IL-2 in male Sprague-Dawley
rats (10 males/dose) exposed to TCA in drinking water at up to 355 mg/kg-day for 90 days.
However, Tang et al. (2002) reported that TCA was positive in the guinea pig maximization test.
A 58% sensitization rate (7/12) was observed in animals given an intradermal injection (2%
solution) and topical application (5% solution), then challenged with a topical application of a
2% TCA solution 21 days after the first intradermal induction. The following scale was used to
grade the reactions: 0 = no reaction, 1 = scattered mild redness, 2 = moderate and diffuse
redness, and 3 = intensive erythema and swelling. The mean score for redness in this study was
1.1, and the mean score for swelling was 0.0. Histological examination of the affected skin
69 DRAFT - DO NOT CITE OR QUOTE
-------
revealed that TCA induced allergenic transformation. These limited data suggest that TCA could
induce a mild allergenic response on exposure to sub-irritating doses.
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION
4.5.1. Mechanistic Studies
Several studies have been conducted for the primary purpose of evaluating the potential
mechanisms by which TCA induces tumors in laboratory animals. These studies can be divided
into five types: peroxisome proliferation, oncogene activation, cell proliferation, DNA
hypomethylation, and inhibition of intercellular communication. Histochemical properties of
TCA-induced tumors have also been characterized in a number of studies, and these properties
have been compared with the same properties in DCA-induced tumors in order to compare the
potential mechanisms of tumor induction.
4.5.1.1. Peroxisome Proliferation
The ability of TCA to induce peroxisome proliferation has been demonstrated in several
studies in rats (DeAngelo et al., 1997, 1989; Mather et al., 1990; Goldsworthy and Popp, 1987;
Elcombe, 1985) and mice (DeAngelo et al., 2008, 1989; Laughter et al., 2004; Parrish et al.,
1996; Austin et al., 1995; Goldsworthy and Popp, 1987). These studies and the evidence for
peroxisome proliferation are summarized in Sections 4.2.1.1.1, 4.2.1.1.2, 4.2.2.1.1.1, and
4.2.2.1.2.1.
4.5.1.2. Oncogene Activation
Ferreira-Gonzalez et al. (1995) studied the K- and H-ras proto-oncogene mutation
patterns in TCA-induced tumors in male B6C3Fi mice. The ras gene encodes a plasma
membrane-bound guanosine triphosphatase. Guanosine triphosphatase activates kinase cascades
that regulate cell proliferation. The ras gene was studied because changes in the rate and
spectrum of mutations in the ras proto-oncogene have been linked to the carcinogenic
mechanism of various liver carcinogens.
Mice (number per group not reported) were exposed to 0 or 4,500 mg/L (1,080 mg/kg-
day based on default water intake values in U.S. EPA [1988]) TCA in drinking water for
104 weeks. The incidence of liver carcinomas was 19% in the untreated mice and 73.3% in the
TCA-exposed group. DNA samples were extracted from 32 spontaneous liver tumors from the
control group and from 11 liver tumors in mice treated with TCA. DNA samples containing
point mutations in exons 1, 2, and 3 of the K- and H-ras genes were detected by the presence of
single-stranded conformation polymorphisms. The single-stranded conformation polymorphism
analysis involved amplification of DNA from the control or tumor tissue to generate DNA
fragments containing normal or mutated ras gene fragments. Since single-stranded DNA
70 DRAFT - DO NOT CITE OR QUOTE
-------
fragments containing base-pair changes have different mobility's when run in polyacrylamide
gels (gel electrophoresis), the pattern of bands observed following gel electrophoresis served to
indicate the presence of a mutated base.
In the spontaneous tumors from control mice, ras mutations were detected only at the
H-61 codon (i.e., the mutation was in the H-ras gene, in the 61st codon, which is in the second
exon); 58% of the spontaneous liver carcinomas showed mutations in H-61, compared with 45%
of the tumors from TCA-treated mice. One TCA-induced tumor showed a mutation in K-61
(i.e., in the K-ras gene, in the second exon). Identification of the specific base-pair change was
done by sequencing of the DNA fragment obtained in the single-stranded conformation
polymorphism analysis. Comparative sequence analysis of exon 2 mutations from spontaneous
and TCA-induced tumors revealed that mutations detected in the TCA tumors matched the
mutation spectrum seen in the spontaneous tumors from control mice. Therefore, TCA changed
neither the rate of ras mutations nor the type of mutations occurring at codon 61.
These results were confirmed in a more recent study. Bull et al. (2002) (described in
Section 4.2) exposed male B6C3Fi mice (20-40/group) at 125-500 mg/kg-day in the drinking
water for 52 weeks. A decrease in the mutation frequency in H-ras codon 61 in TCA-induced
tumors compared with spontaneous tumors from control animals was observed, confirming the
observations of Ferreira-Gonzalez et al. (1995). Also, the type of H-ras codon 61 mutations was
similar to the spectra of mutations observed in spontaneous tumors from control animals.
Based on the absence of an effect on mutation rate, the authors indicated that it was not
clear if TCA was acting through a genotoxic or nongenotoxic mechanism (Ferreira-Gonzalez et
al., 1995). However, the number of tumors with ras mutations was slightly decreased in TCA-
treated animals, consistent with TCA acting through a nongenotoxic mechanism. Because of the
large proportion of tumors carrying a ras mutation, the authors concluded that ras mutations are
important for the development of carcinogen-induced tumors as well as spontaneous tumors.
TCA increased the tumor yield but did not change mutations in ras, leading the study authors to
conclude that TCA might facilitate the growth of preneoplastic lesions that arise from
spontaneously initiated (i.e., ras mutated) hepatocytes.
Tao et al. (1996) investigated whether liver tumors initiated by MNU and promoted by
TCA exhibited loss of heterozygosity in four polymorphic loci on chromosome 6. According to
the authors, inactivation of one or more of the polymorphic alleles at these loci may be related to
the inactivation of an, as yet, unidentified tumor-suppressor gene, resulting in oncogene
activation that may be a key event in the pathogenesis of some liver tumors. This hypothesis is
supported by the results of a study by Davis et al. (1994), in which 20% of hepatic tumors
induced by tetrachloroethylene exhibited loss of heterozygosity on chromosome 6, suggesting
the presence of a tumor suppressor gene at this site. In this study, 15-day-old female B6C3Fi
mice were pretreated with 25 mg/kg MNU via i.p. injection and administered TCA in drinking
water at a concentration of 20.0 mmol/L (3,268 mg/L) for 52 weeks. The authors did not provide
71 DRAFT - DO NOT CITE OR QUOTE
-------
a dose estimate, but the approximate dose is 784 mg/kg-day, based on the default drinking water
intake value for female B6C3Fi mice (U.S. EPA, 1988). Thirty-seven liver tumors promoted by
TCA were examined for loss of heterozygosity by using four polymorphic loci on chromosome
6. Ten of 37 tumors (7/27 carcinomas and 3/10 adenomas) promoted by TCA showed evidence
of loss of heterozygosity for at least two loci on chromosome 6. The C57BL/6J alleles at both
the D6mit9 and D6mit323 loci were lost in all 10 tumors exhibiting loss of heterozygosity, and
2 of these 10 tumors also lost at least one of the C3H/HeJ alleles. No loss of heterozygosity on
chromosome 6 was observed in 24 DCA-promoted liver tumors. The observed loss of
heterozygosity on chromosome 6 in many of the tumors suggests the presence of an unidentified
tumor-suppressor gene on this chromosome. However, as the majority of tumors in TCA-treated
mice did not exhibit loss of heterozygosity on chromosome 6, the authors concluded that other
molecular activity is probably involved in the hepatocarcinogenesis of TCA.
4.5.1.3. Cell Proliferation
Investigations of the effects of TCA on cell growth rates have produced conflicting
results. Miyagawa et al. (1995) examined the effect of TCA (and a battery of putative
nongenotoxic liver carcinogens and noncarcinogens) on replicative DNA synthesis to assess the
utility of measurement of cell proliferation as a screening assay for detecting nongenotoxic
carcinogens. Groups of male B6C3Fi mice (four or five per dose) were administered a single
gavage dose of TCA in an acute toxicity test to determine the maximum tolerated dose. The
maximum tolerated dose for TCA was reported to be approximately one-half of the LD50.
Groups of four or five animals were administered a single gavage dose of one-half of the
maximum tolerated dose (250 mg/kg, as estimated from data provided by the authors) or the
maximum tolerated dose (500 mg/kg, as estimated from data provided by the authors), and
incorporation of [3H]thymidine in harvested hepatocytes was measured 24, 39, or 48 hours after
dosing. For TCA, positive responses were observed at 250 mg/kg at 24 and 39 hours (6.5- and
4.9-fold above controls) and at 500 mg/kg (9.8-fold above controls). Although the mean
increase in replicative DNA synthesis met the criteria for a positive response, the increases did
not appear to be statistically significant based on the SDs supplied in the summary table.
In contrast to the increased cell proliferation observed by Miyagawa et al. (1995),
Channel and Hancock (1993) found that TCA can decrease the rate of progression through
S-phase of the cell cycle. WB344 cells, a non-tumorigenic epithelial rat hepatocyte cell line,
were exposed to TCA-free medium or medium containing 100 |ig/mL TCA. Cell growth rates
were assessed by cell counting, and transition through the cell cycle was monitored by labeling
nascent DNA with BrdU. The resulting labeling data were used to identify fractions of cells in
various stages of the cell cycle and to model transit times through each phase. The transit time
through S-phase was estimated to be 5.20 hours for treated and 5.02 hours for control cells
(p < 0.05). As further support for this effect, cells in S-phase were elevated by approximately 5-
72 DRAFT - DO NOT CITE OR QUOTE
-------
20% for the first 6 hours after release from TCA-treatment but returned to control values after
this initial period. In contrast to these results, indicating slowing of S-phase transit, relative
movement plots (also related to S-phase transit time) did not differ from controls. The authors
suggested, however, that this might reflect the insensitivity of relative movement plots for
detection of small treatment-related changes, such as those observed for TCA. The authors
suggested that the observed pattern of cell cycle perturbation (i.e., a slightly extended period of
S-phase) would be consistent with a sublethal effect of cytotoxicity and would be less serious
than a decrease in transit time through G2M phase (which could potentially increase
chromosomal mismatches and rearrangements, due to an insufficient time spent in mitosis). The
toxicological significance of these results by Miyagawa et al. (1995) and Channel and Hancock
(1993) are difficult to interpret, since they might not reflect the cell growth conditions of normal
hepatocytes in vivo. For this reason, these studies are of limited use in evaluating the effects of
TCA on cell growth in vivo.
Pereira (1996) evaluated cell proliferation in the liver of female B6C3Fi mice (10/group)
treated with 0, 2, 6.67, or 20 mmol/L TCA in drinking water for 5, 12, or 33 days by estimating
hepatocyte BrdU-labeling index. TCA statistically significantly increased the BrdU-labeling
index after 5 days of exposure by approximately 2.5-3-fold at all three concentrations (values
estimated from Figure 4 in Pereira, 1996); however, BrdU-labeling indices were not increased
after 12 or 33 days of exposure. Thus, cell proliferation was enhanced by 5 days exposure to
TCA but not for longer exposures of 12 or more days.
In a cell proliferation study reported by Stauber and Bull (1997), male B6C3Fi mice were
pretreated with 2,000 mg/L of TCA (480 mg/kg-day based on default water-intake values in
U.S. EPA [1988]) in drinking water for 50 weeks. The mice were then given drinking water
containing 0, 20, 100, 500, 1,000, or 2,000 mg/L TCA (estimated doses of 0, 5, 23, 115, 230, or
460 mg/kg-day, based on default water intake values in U.S. EPA [1988]) for 2 additional weeks
to assess whether cell proliferation induced by TCA in either normal liver cells or tumors was
dependent on continued treatment. All dose groups contained 12 animals, except for the
2,000 mg/L group, which consisted of 22 mice. Five days prior to sacrifice, DNA in replicating
hepatocytes was labeled in vivo by administering BrdU via subcutaneously implanted pumps.
Liver tissue was stained, and dividing nuclei were counted. Cell division rates were evaluated
separately in normal hepatocytes, in tumors, and in altered hepatic foci (AHF).
A transient but significant elevation (about twofold) in normal hepatocyte division rates
was evident in mice consuming 2,000 mg/L TCA for 14 or 28 days (apparently as part of the
pretreatment phase), but continued treatment for 52 weeks resulted in a significant decrease
(about 70%) in hepatocyte division rate. In the mice treated for 50 weeks with 2,000 mg/L and
then shifted to the lower concentrations for 2 weeks, the cell division rate in normal liver cells
was elevated (but not statistically significantly so) at 100 and 500 mg/L, but in mice exposed to
1,000 or 2,000 mg/L for 2 weeks, there was a significant decrease (about 1.5-fold) in cell
73 DRAFT - DO NOT CITE OR QUOTE
-------
division. Cell division rates in TCA-induced AHFs and tumors were high at all doses. Rates of
cell division in AHFs and tumors remained high in mice whose exposure was terminated during
the last 2 weeks of the study, indicating that these rates were independent of continued TCA
treatment.
TCA-induced lesions were histochemically stained with anti-c-jun and anti-c-fos
antibodies, component proteins of the AP-1 transcription factor that up-regulates expression of
genes required for DNA synthesis. No differences were observed in the levels of proteins
reacting with c-jun and c-fos antibodies in either liver AHFs or tumors, relative to normal
hepatocytes, indicating that TCA produces little, if any, direct stimulation of the replication of
initiated cells through this pathway. However, three tumors induced by TCA each contained a
nodule that stained heavily for c-fos, and cell-division rates within these nodules were very high,
suggesting a transition to an aggressive tumor. The low frequency of this marker (3/52 tumors)
suggested that its presence in these nodules was not due to a direct effect of TCA.
Based on these results, Stauber and Bull (1997) proposed a mechanism for TCA-induced
hepatocarcinogenesis. They proposed that the initial growth stimulation induced by TCA causes
normal cells to compensate by increasing signals that inhibit cell proliferation, which ultimately
results in the TCA-induced growth inhibition observed with chronic treatment. Pre-initiated
cells refractory to this growth inhibition would then have a selective growth advantage. The
authors noted that the lack of effect on c-jun by TCA was consistent with tumor characteristics of
other peroxisome proliferators in rats, as demonstrated by Rao et al. (1986). Because cell
replication in AHFs was independent of TCA (i.e., discontinued TCA treatment did not alter
AHFs or tumor-cell labeling), it was proposed by Stauber and Bull (1997) and Ferreira-Gonzalez
et al. (1995) that TCA might enhance growth of initiated cells by suppressing apoptosis in such
cells, as has been demonstrated for other peroxisome proliferators and is consistent with agonism
of PPARa receptor playing an important role in TCA-induced carcinogenesis. Cell proliferation
has also been observed in short-term studies (Dees and Travis, 1994; Sanchez and Bull, 1990)
that are described in Section 4.2. The results of these studies were consistent with the results
described by Stauber and Bull (1997).
4.5.1.4. DNA Hypomethylation
The hypomethylation of DNA in response to TCA exposure was investigated by Tao et
al. (1998) as a potential nongenotoxic mechanism involved in TCA-induced tumor promotion
and carcinogenesis. Mammalian DNA naturally contains the methylated base 5-methylcytosine
(5MeC), which plays a role in regulation of gene expression and DNA imprinting (Razin and
Kafri, 1994). An overall decrease in the content of 5MeC in DNA is often found in tumors and
has been considered to represent an important event in the clonal expansion of premalignant cells
during neoplastic progression (Counts and Goodman, 1995, 1994).
74 DRAFT - DO NOT CITE OR QUOTE
-------
In the Tao et al. (1998) study, female B6C3Fi mice were injected i.p. with 25 mg/kg of
MNU at 15 days of age. When the mice were 6 weeks of age, TCA, neutralized to a
concentration of 25 mmol/L (4,085 mg/L), was administered in drinking water for 44 weeks.
This concentration corresponds to approximately 980 mg/kg-day, based on a default water factor
of 0.24 L/kg-day for female B6C3Fi mice for chronic exposure (U.S. EPA, 1988). Control mice
received only MNU.
To test the effects of short-term treatment with TCA on DNA methylation, mice not
administered MNU were given 0 or 25 mmol/L TCA in drinking water for 11 days,
corresponding to approximately 1,062 mg/kg-day, based on the strain-specific water factor for a
short-term study (U.S. EPA, 1988). DNA extracted from liver tissue and tumors were
hydrolyzed, and 5MeC and the four DNA bases were separated and quantified by HPLC.
After 11 days of exposure to TCA (without pretreatment with MNU), the level of 5MeC
in total-liver DNA was decreased (about 60%) relative to untreated controls. After 44 weeks of
TCA treatment, 5MeC levels were not different from controls that had received only MNU. No
difference in DNA methylation was observed between the control groups in the short-term
(drinking water control) and long-term (MNU only control) experiments. These results indicate
that TCA caused only a transient decrease in DNA methylation in the liver.
In TCA-promoted hepatocellular adenomas and carcinomas, the level of 5MeC in DNA
was decreased 40 and 51% when compared with either noninvolved tissue from the same animal
and liver tissue from control animals given only MNU, respectively. Termination of TCA
treatment 1 week prior to sacrifice did not change the levels of 5MeC in either adenomas or
carcinomas; however, they remained lower than in noninvolved tissue. 5MeC levels in DNA
from carcinomas were lower than in DNA from adenomas, suggesting that DNA methylation is
further decreased with tumor progression. DNA hypomethylation tends to favor gene
expression, which may drive cell-proliferation responses. Therefore, based on the change
observed in the adenomas and carcinoma tissue compared with the uninvolved tissue, Tao et al.
(1998) suggested that hypomethylation of DNA, as indicated by decreased 5MeC in tumor DNA,
is involved in the carcinogenic and tumor-promoting activity of TCA.
The marked increase in hypomethylated DNA in mouse liver tumors observed by Tao et
al. (1998) indicated that the methylation of numerous genes was decreased. Tao et al. (2004,
2000a, b) investigated the methylation status and expression of specific genes in mouse liver
tumors and uninvolved liver tissue, as well as in livers of mice initiated with MNU but not
exposed to TCA, in a series of studies described below.
Tao et al. (2000a) evaluated the methylation and expression of c-jun and c-myc
protooncogenes in mouse liver after short-term exposure to TCA. Female B6C3Fi mice
(four/group) were dosed by gavage for 5 days with 500 mg/kg-day TCA in water neutralized
with sodium hydroxide to pH 6.5-7.5. This dose was selected because it was reported to
increase liver growth, cell proliferation, and lipid peroxidation in mice (Dees and Travis, 1994;
75 DRAFT - DO NOT CITE OR QUOTE
-------
Larson and Bull, 1992). Vehicle-control mice received the same volume of water or corn oil. At
30 minutes after each dose of TCA or vehicle, the mice received 0, 30, 100, 300, or 450 mg/kg
methionine by i.p. injection. The mice were sacrificed 100 minutes after the last dose and the
livers were excised. Methylation status in the promoter region for c-jun and c-myc
protooncogenes was evaluated by using methylation-sensitive restriction endonuclease Hpall
digestion, followed by Southern blot analysis of DNA. Hpall does not cut CCGG sites when the
internal cytosine is methylated, and Southern blots, probed for the promoter region of these two
genes, would only contain extra bands in Hpall digested hypomethylated DNA. Expression of
mRNA for c-jun and c-myc protooncogenes and c-jun and c-myc proteins were also analyzed.
Decreased methylation in the promoter regions of the c-jun and c-myc genes and
increased levels of their mRNA and proteins were found in the livers of TCA-treated mice.
Methionine prevented the decreased methylation of the two genes in a dose-dependent manner,
with the effective dose >100 mg/kg. Methionine also prevented the increased levels of the
mRNA and proteins from the two genes at 450 mg/kg. Tao et al. (2000a) concluded that the
prevention of TCA-induced DNA hypomethylation by methionine suggested that the decrease in
the formation of 5MeC in DNA is due to a decrease in the concentration of S-adenosyl-
methionine substrate, and the dose of TCA must be sufficient to decrease the level of S-adeno-
sylmethionine in order for it to be active as a carcinogen.
In another study, Tao et al. (2000b) examined the methylation of c-jun and c-myc genes,
expression of both genes, and activity of DNA methyltransferase in mouse liver tumors initiated
by MNU and promoted by TCA in female B6C3Fi mice. The tumors were obtained from test
animals used in the promotion study described by Pereira and Phelps (1996) (see Section
4.2.2.1). Briefly, the test animals were given either 25 mg/kg MNU or the saline vehicle control
at 15 days of age. Starting at 6 weeks of age, animals were given neutralized TCA in drinking
water at 20 mmol/L (3,268 mg/L) continuously until 52 weeks of age. Dose estimates were not
reported by the study authors, but the concentration provided in drinking water would result in a
dose of approximately 784 mg/kg-day based on the default drinking water value of 0.24 L/kg-
day for female B6C3Fi mice (U.S. EPA, 1988). TCA-promoted liver tumors and noninvolved
liver tissue, as well as liver tissue from MNU-initiated mice not exposed to TCA, were collected
when the animals were euthanized at 52 weeks of age.
Methylation status in the promoter regions of the c-jun and c-myc genes was determined
by Southern blot analysis of DNA extracted from the three types of harvested tissues and
digested with the methylation-sensitive restriction endonuclease Hpall. Expression of the c-jun
and c-myc genes was determined by northern blot analysis of mRNA levels and western blot
analysis of protein levels. DNA methyltransferase activity was determined in nuclear extracts
prepared from the harvested liver tumors or the other two types of liver tissues described
previously. Tao et al. (2000b) concluded that the promoter regions of c-jun and c-myc in tumors
were hypomethylated relative to the promoter regions in noninvolved liver tissue from TCA-
76 DRAFT - DO NOT CITE OR QUOTE
-------
promoted animals. The expression of the mRNA and protein for each of these genes was also
increased in TCA-promoted tumors relative to noninvolved liver tissue. DNA methyltransferase
activity was significantly increased in liver tumors from TCA-promoted mice when compared
with noninvolved liver from the same mice. Collectively, these results suggest that TCA-
promoted carcinogenesis involves decreased methylation and increased expression of the c-jun
and c-myc protooncogenes in the presence of increased DNA methyltrnasferase activity.
In a related study, Tao et al. (2004) investigated DNA hypomethylation and the methylation
status and expression of the insulin-like growth factor (IGF)-II gene^ in TCA-promoted mouse
liver tumors and noninvolved liver tissue, as well as in liver tissue samples from MNU-initiated
mice that were not exposed to TCA. Expression of the IGF-II gene was investigated because
increased hepatic cell proliferation is associated with increased expression of growth-related
genes, such as IGF-II (Furstenberger and Senn, 2002; Werner and Le Roith, 2000). Loss of
imprinting4 and increased expression of IGF-II have been observed in liver tumors (Scharf et al.,
2001; Khandwala et al., 2000).
In the study by Tao et al. (2004), mouse liver tumors and tissues were obtained from
female B6C3Fi mice as described above. At necropsy, no liver tumors were found in mice that
were treated with MNU alone or TCA alone. The levels of 5MeC in DNA extracted from tumors
and liver tissues were quantified by a dot blot analysis procedure that used a mouse monoclonal
primary antibody specific for 5MeC. Methylation status of 28 cytosine-guanine dinucleotide
sites6 in the differentially methylated region-2 of the mouse IGF-II gene was determined by a
bisulfite-modified DNA sequencing procedure. In this procedure, DNA extracted from tumors
and liver tissues was incubated with sodium metabisulfite to convert unmethylated (but not
methylated) cytosine to uracil to enable detection of unmethylated sites in the sequencing
analysis. Bisulfite-modified DNA was recovered, and the differentially methylated region-2 of
the IGF-II gene was amplified by polymerase chain reaction for sequencing. Expression of
IGF-II mRNA was determined by reverse transcription polymerase chain reaction. The level of
5MeC in DNA from noninvolved liver tissue in mice treated with TCA was decreased relative to
that in DNA from mice initiated with MNU but not exposed to TCA. The level of 5MeC in
TCA-promoted tumors was further decreased relative to the noninvolved liver tissue, indicating
5IGF-II is involved in cell division, differentiation, and apoptosis. According to information presented in Tao et al.
(2004), the IGF-II gene is imprinted with the paternal allele being expressed, and the maternal allele is methylated
and silent in normal adult tissue, including the mouse liver, while in tumors, the imprinting is lost. Loss of
imprinting is accompanied by increased expression of its mRNA in tumors.
6Cytosine-guanine dinucleotide sites are regions in DNA where a cytosine nucleotide (C) is situated next to a
guanine nucleotide (G). The "p" denotes the phosphodiester bond that links the nucleotides. Cytosine-guanine
dinucleotide sites are relatively rare in eukaryotic genomes except in regions near the promoter regions of genes.
Methylation of the cytosine nucleotide at cytosine-guanine dinucleotide sites to form 5MeC is believed to play a
critical role in regulation of gene expression. Decreased methylation or hypomethylation is associated with gene
expression, while increased methylation has an inhibitory effect on gene expression. Aberrant promoter methylation
has been proposed as a possible mechanism for increased protooncogene expression in cancer.
77 DRAFT - DO NOT CITE OR QUOTE
-------
hypomethylation. These observations confirm the previous results of Tao et al. (1998) for DNA
hypomethylation obtained by using HPLC analysis.
Sequencing of the differentially methylated region-2 of the IGF-II gene promoter
revealed that 21-24 cytosine-guanine dinucleotide sites were methylated in initiated liver,
compared with 15-17 sites in noninvolved liver tissue from TCA-promoted mice. Thus,
exposure to TCA reduced the percentage of cytosine-guanine dinucleotide sites that were
methylated from approximately 79 to 58%. The number of methylated cytosine-guanine
dinucleotide sites was further reduced to 0-7 (approximately 11%) in liver tumors promoted by
TCA. mRNA expression was significantly increased (5.1-fold) in liver tumors relative to
noninvolved liver tissue from mice treated with TCA. mRNA expression was not increased in
noninvolved liver tissue from TCA-promoted animals when compared to level of expression in
the MNU-initiated control. These results demonstrated that TCA treatment caused
hypomethylation of DNA and increased expression of the IGF-II gene in the TCA-promoted
liver tumors. Thus, the hypothesis that DNA hypomethylation is involved in the mechanism for
tumorigenicity of TCA is supported.
The temporal association of DNA methylation and cell proliferation in mice treated with
TCA has been investigated by Ge et al. (2001a). Female B6C3Fi mice were given gavage doses
of 500 mg/kg-day TCA and sacrificed at 24, 36, 48, 72, and 96 hours after the first dose. (TCA
was neutralized to pH 6-7 with NaOH.) The liver, kidney, and urinary bladder were removed
and weighed, and subsamples were processed for extraction of DNA and determination of
methylation status in the promoter region of the c-myc protooncogene. Methylation status was
determined by Southern blot analysis following digestion of the isolated and purified DNA with
a methyl-sensitive restriction enzyme. Liver and kidney tissues were collected for measurement
of cell proliferation by determination of proliferating cell nuclear antigen labeling and mitotic
indices.
Relative liver weights were significantly increased at the 36-, 72-, and 96-hour time
points; there was no effect of TCA on relative kidney weights. The proliferating cell nuclear
antigen labeling index was significantly increased in liver cells at 72 and 96 hours relative to
controls. The mitotic index was significantly elevated in liver cells at 96 hours after the first
dose. Southern blot analysis indicated that the tumor promoter region of the c-myc
protooncogene in the liver was hypomethylated at the 72- and 96-hour time points. These data
indicate that TCA caused simultaneous enhancement of cell proliferation and decreased
methylation in liver cells starting at 72 hours after exposure. TCA also decreased methylation in
the promoter region of the c-myc gene in the kidney and urinary bladder after 72 and 96 hours of
treatment, but the response was less pronounced than in liver. Cell proliferation data for the
kidney were not reported. The study authors proposed that TCA induces hypomethylation by
inducing DNA replication and preventing the methylation of the newly synthesized strands of
DNA.
78 DRAFT - DO NOT CITE OR QUOTE
-------
Pereira et al. (2001) examined the effect of chloroform (a disinfection byproduct present
as a co-contaminant with TCA in drinking water) on DCA- or TCA-induced hypomethylation
and expression of the c-myc protooncogene in female B6C3Fi mice. Chloroform has been
reported to cause hypomethylation of DNA and of the c-myc gene by preventing the methylation
of hemimethylated DNA formed when DNA is replicated (Coffin et al., 2000). Six mice per
treatment group were exposed to 0, 400, 800, or 1,600 mg/L chloroform in the drinking water for
17 days. A DCA or TCA dose of 500 mg/kg was administered daily by gavage on the last 5 days
of the exposure period. At sacrifice, livers were removed and processed for extraction of DNA.
Methylation of the promoter region was evaluated by using Hpall restriction enzyme digestion
followed by Southern blot analysis. Expression of c-myc mRNA was evaluated using reverse
transcription-polymerase chain reaction followed by northern blot analysis. Both DCA and TCA
decreased methylation in the promoter region of the c-myc gene and increased expression of
c-myc mRNA. Coadministration of chloroform did not affect the extent of TCA-induced
hypomethylation or mRNA expression or the incidence or multiplicity of liver tumors promoted
by TCA. By contrast, Coadministration of chloroform prevented the hypomethylation and
mRNA expression of the c-myc gene and the promotion of liver tumors by DCA. This study
suggests that the ability of chloroform, DCA, and TCA to hypomethylate c-myc and increase
c-myc mRNA expression in the liver was correlated with their effect on liver tumor promotion.
4.5.1.5. Inhibition of Intercellular Communication
Benane et al. (1996) assessed the effects of TCA on gap junctional intercellular
communication (GJIC) in Clone 9 (ATCC CRL 1439), a normal liver epithelial cell line from a
4-week-old Sprague-Dawley male rat. The cells were grown in a nutrient mixture, plated, and
exposed to TCA at concentrations of 0, 0.5, 1.0, 2.5, and 5 mM (1, 4, 6, 24, 48, or 168 hours) for
varying time periods. Lucifer yellow scrape-load dye transfer was used as a measure of GJIC.
At a concentration of 0.5 mM, there were no statistically significant differences in dye transfer
among control and treated cells at any of the time points. At a concentration of 1.0 mM,
statistically significant differences were found for all time periods except 4 and 168 hours. At
concentrations of 2.5 and 5 mM, the level of dye transfer was statistically decreased as compared
with controls for all time points. The lowest concentration and shortest time to reduce dye
transfer was 1 mM over a 1-hour period. The reduction in dye transfer increased with higher
concentrations and longer treatment time. 12-O-tetradecanoylphorbol 13-acetate, a tumor
promoter and a known disrupter of intercellular communication, used as positive control, caused
a rapid reduction in dye transfer.
Klaunig et al. (1989) performed a series of experiments to determine the effects of TCA
on GJIC in primary cultured B6C3Fi mouse and F344 rat hepatocytes. Mouse and rat
hepatocytes were isolated from 6-8-week-old male mice and rats by two-stage collagenase
perfusion and plated in glass Petri dishes or flasks. Following preliminary experiments to
79 DRAFT - DO NOT CITE OR QUOTE
-------
identify cytotoxic concentrations, 24-hour-old hepatocytes were treated with 0, 0.1, 0.5, or 1 mM
TCA dissolved in dimethyl sulfoxide for up to 24 hours. The controls included "no treatment"
and solvent controls in sealed and unsealed culture vessels. PB was used as the positive control.
Effects on GJIC were evaluated by the iontophoretic microinjection of fluorescent Lucifer
yellow CH dye into one hepatocyte and observation of the dye spread to adjacent hepatocytes.
Adjacent cells that fluoresced were designated as dye coupled (i.e., communicating through gap
junctions). The experimental results were expressed as the number of coupled/noncoupled
recipient cells and a percentage of coupled cells. TCA inhibited dye transfer in both 24-hour-old
and freshly plated mouse hepatocytes. The inhibitory effect in 24-hour-old cultures was
transient; dye coupling was significantly reduced at all tested concentrations after 4 hours of
treatment but not after 8 or 24 hours. PB, the positive control, significantly reduced dye transfer
in cells treated with 1 or 2 mM after 4 or 8 hours of treatment but not after 24 hours. In an
experiment to compare the response of freshly plated and 24-hour-old mouse hepatocytes, all
tested concentrations of TCA significantly inhibited dye transfer in both types of culture after
3 and 6 hours of treatment. The inhibitory effect on dye transfer in mouse cells was unaffected
by treatment with SKF-525A, a CYP450 inhibitor.
Dye transfer in 24-hour-old primary rat hepatocytes was unaffected by treatment with
TCA at concentrations up to 1 mM for as long as 24 hours. Dye transfer in freshly plated rat
primary rat hepatocytes was unaffected by treatment with concentrations up to 1 mM TCA for as
long as 6 hours. PB, the positive control, significantly reduced the percentage of coupled cells in
cultures treated with 1 or 2 mM after 4 or 8 hours of treatment but not after 24 hours. The results
obtained for primary F344 rat hepatocytes by Klaunig et al. (1989) differ from those reported in
rat cell cultures by Benane et al. (1996), who observed inhibition of dye transfer in cells from a
Sprague-Dawley rat epithelial cell line treated with 1 mM for durations of 1-168 hours. The
reason for the differential response in rat liver cells is unknown but may be related to differences
in the originating strain or in the type of cultured cell tested (primary cultured hepatocytes versus
established cell line).
4.5.1.6. Oxidative Stress
The ability of TCA to induce oxidative-stress responses, such as lipid peroxidation and
oxidative DNA damage, and the relationship between these responses and indicators of
peroxisome proliferation or altered CYP450 activities have been tested in a series of studies
following acute or short-term TCA dosing in mice (Austin et al., 1996, 1995; Parrish et al., 1996;
Larson and Bull, 1992). TCA induced both lipid peroxidation (TEARS) and oxidative DNA
damage (8-OHdG) following administration of single oral doses. These studies are described in
Section 4.2.
A potential mechanism of TCA-induced oxidative stress via macrophage activation was
investigated by Hassoun and Ray (2003). Studies have shown that macrophages can be activated
80 DRAFT - DO NOT CITE OR QUOTE
-------
and become a source of reactive oxygen species that may produce damage to surrounding tissues
(Karnovsky et al., 1988; Briggs et al., 1986). In this study, the ability of TCA to activate
cultured macrophages (J744A.1 cell line) in vitro to become a source of reactive oxygen species
was evaluated. Oxidative stress was evaluated by time- and concentration-dependent production
of superoxide anion in response to TCA; resulting cytotoxicity, as indicated by effects on SOD
activity and cell viability; and release of LDH by the cells into cultured media. Cells were
exposed to TCA at 8-32 mM for 24-60 hours (pH of TCA solution was adjusted to pH 7.0 by
NaOH).
Incubation with TCA caused a significant decrease in cell viability as assessed by trypan
blue staining at all concentrations tested, although at 8 mM cell viability was only significantly
reduced compared with controls at the 60-hour incubation. Reduced cell viability results
correlated well with increased LDH activity in media. Twenty-four hour incubation with TCA
did not cause increases in superoxide anion levels; however, incubations of 36 and 60 hours
caused significant increases in superoxide anion levels at 16, 24, and 32 mM (p < 0.05). SOD
activity was also affected by TCA treatment. Significant increases in SOD activity occurred at
lower TCA concentrations (8-24 mM) compared with controls, but SOD activity at the highest
concentration (32 mM) for 24-36 hours was similar to that of controls. Incubation of cells with
32 mM TCA for 60 hours resulted in 100% cell death. These results indicate that incubation
with TCA at 8-32 mM for 24-60 hours induces macrophage activation, which resulted in
cytotoxicity due to oxidative stress. Hassoun and Ray (2003) noted that, although TCA exposure
concentrations were high, they were comparable to those used in animal studies (Austin et al.,
1996; Larson and Bull, 1992; Bull et al., 1990; Sanchez and Bull, 1990).
The activation of phagocytic cells was supported by in vivo studies (Hassoun and Dey,
2008). Groups of male B6C3Fi mice (8 animals/group) were administered 300 mg/kg TCA by
gavage and sacrificed after 6 or 12 hours. Because obtaining pure Kupffer cells from liver is
difficult, peritoneal lavage cells were isolated to examine the production of superoxide anion as
the indication of phagocytic activation. Hepatic tissues were isolated to assay superoxide anion,
lipid production, and DNA single-strand breaks (SSBs). At 6 hours, none of the biomarkers was
induced by TCA. At 12 hours, the superoxide anion increased 62.5% in peritoneal lavage cells
and 17.6% in hepatic tissue. Lipid peroxidation and DNA single-strand breaks increased 29.4
and 167%, respectively, in hepatic tissue.
The same group of scientists further conducted subacute (4 weeks) and subchronic
(13 weeks) studies in mice to investigate the possible role of oxidative stress induced by
macrophage activation in TCA hepatocarcinogenicity (Hassoun et al., 2010a, b). Groups of male
B6C3Fi mice (7 animals/group) were administered 7.7, 77, 154, and 410 mg/kg-day TCA by
gavage for 4 and 13 weeks. Hepatic tissues were examined for the production of superoxide
anion, lipid peroxidation, and DNA single-strand breaks. Peritoneal lavage cells, a surrogate for
liver Kupffer cells, were collected and tested for biomarkers of phagocytic activation, including
81 DRAFT - DO NOT CITE OR QUOTE
-------
superoxide anion, tumor necrosis factor-alpha (TNF-a), and myeloperoxidase. The results
showed dose- and time-dependent increases in the production of superoxide anion (increases of
30 and 167% at doses of 154 and 410 kg/mg-day at 4 weeks; 20, 100, 133, and 200% at doses of
7.7, 77, 154, and 410 mg/kg-day at 13 weeks), lipid peroxidation (increases of 67, 80, and 567%
at doses of 77, 154, and 410 mg/kg-day at 4 weeks; 33, 400, 500, and 733% at doses of 7.7, 77,
154, and 410 mg/kg-day at 13 weeks), and DNA single-strand breaks (increases of 75, 125, and
300% at doses of 77, 154, and 410 mg/kg-day at 4 weeks; 125, 200, and 310% at doses of 77,
154, and 410 mg/kg-day) in the liver, indicating the production of reactive oxygen species and
their associated effects on hepatic cellular components. Because the doses administered in these
studies are comparable to the doses inducing hepatocarcinogenicity, but the treatment was a
shorter period, the authors considered the significant increases of oxidative stress as initial events
that may lead to later production of long-term effects.
These studies also showed that TCA induced increases of biomarkers of phagocytic
activation. Interestingly, the biomarkers of phagocytyic activation in peritoneal lavage cells
showed dose-dependent increases after 4 weeks of treatment, but not at 13 weeks of treatment.
The production of superoxide anion at 4 weeks increased 56, 106, and 175% at doses of 77, 154,
and 410 mg/kg-day, whereas at 13 weeks, the production increased significantly (60%) in the
group treated at 77 mg/kg-day only. Similarly, the increase of myeloperoxidase activity was
robust at 4 weeks (increased by 15-, 20-, 29, and 7.5-fold at doses of 7.7, 77, 154, and
410 mg/kg-day) and was modest at 13 weeks (increased by 2.5-, 5-, and 2-fold at doses of 7'.7,
77, and 154 mg/kg-day). TNF-a, released by peritoneal lavage cells, increased dose-dependently
at 4 weeks (increased by 2-, 3.2-, and 9-fold at 77, 154, and 410 mg/kg-day), whereas at
13 weeks, the increase (1.8-fold) was only found at 77 mg/kg-day. The findings, i.e., more
increases in the biomarkers at the 4-week treatment period than at the 13-week treatment period,
and in response to the lower doses to a greater extent than the higher doses (carcinogenic doses),
indicate that TCA-induced phagocytic activation may be an initial adaptive response to protect
against TCA-induced damage.
4.5.1.7. Histochemical Characteristics of TCA-induced Tumors
Biomarkers of cell growth, differentiation, and metabolism in proliferative hepatocellular
lesions promoted by TCA were investigated by Latendresse and Pereira (1997) to further
determine differences in DCA and TCA carcinogenesis. Female B6C3Fi mice were initiated
with an i.p. injection of MNU at 15 days of age and treated with TCA in drinking water at a
concentration of 20 mmol/L from age 49 days to 413 days. The authors did not provide a dose
estimate, but the approximate dose is 784 mg/kg-day, based on the default drinking water intake
value for female B6C3Fi mice (U.S. EPA, 1988). At 413 days of age, the mice were sacrificed
and liver tissues were examined histologically. A panel of histochemical markers was evaluated,
including transforming growth factor-a (a growth factor that stimulates cell proliferation and is
82 DRAFT - DO NOT CITE OR QUOTE
-------
expressed in tumor cells), transforming growth factor-p (a growth factor that is inhibitory to
hepatocyte proliferation), c-jun and c-fos (component proteins of the AP-1 transcription factor
that regulates expression of genes involved in DNA synthesis), c-myc (a regulator of gene
transcription induced during cell proliferation), CYP2E1 (potentially involved in TCA
metabolism) and CYP4A1 (induced by peroxisome proliferation signaling), and GST-Ti (a
marker for certain tumor types).
TCA-induced foci of altered hepatocytes and tumors tended to be predominantly
basophilic and stained variably for the histochemical markers examined. In TCA-treated mice,
none of the markers stained positive in >50% of the cells/tumor, except c-jun, which was
observed in >50% of cells from 9 of the 13 tumors evaluated. This profile of marker expression
contrasts with the tumors from DCA-treated mice for which more than half of the examined
tumors expressed transforming growth factor-a, c-myc, CYP2E1, CYP4A1, and GST-u in >50%
of the cells. The contrasting histochemical-marker profiles induced by DCA and TCA provide
evidence for a different MOA for these two haloacetic acids. In a recent study, Bull et al. (2002)
(described in Section 4.2) observed that TCA-induced tumors were uniformly lacking in c-jun
expression, but DC A-induced tumors often expressed c-jun, providing further evidence of a
different MOA for TCA and DCA induction of liver tumors.
In the case of the TCA-promoted tumors, the minimal immunostaining for most markers
(with the exception of c-jun) suggested that these proteins are not particularly important in
TCA-induced tumor promotion. On the other hand, Latendresse and Pereira (1997) pointed out
that the regional staining variability within the lesions for c-jun and c-myc proteins is consistent
with localized clonal expansion and/or tumor progression. Non-tumor hepatocytes in TCA-
treated animals were generally negative for transforming growth factor-p and GST-u staining and
positive for CYP2E1 (centrilobular region) and CYP4A1 (panlobular region). CYP4A1 is an
enzymatic marker for peroxisome proliferation, since its expression precedes peroxisomal
response, and is coordinated with the transcription of the peroxisomal p-oxidation enzymes. The
expression of CYP4A1 in normal hepatocytes in TCA-treated animals is consistent with
TCA-induced peroxisome proliferation. However, CYP4A1 was not highly expressed in the
tumor cells. This result suggests that, if PPARa agonism is involved in TCA-induced cancer, it
is likely that the effect occurs earlier in the tumorigenic process than was evaluated in this study.
Pereira (1996) studied the characteristics of the lesions in female B6C3Fi mice to
evaluate differences in MOA of DCA and TCA. AHFs and tumors induced by DCA were
reported as being predominantly eosinophilic. AHFs induced by TCA were equally distributed
between basophilic and eosinophilic, whereas hepatic tumors induced by TCA were
predominantly basophilic, including all observed hepatocellular carcinomas (n = 11), and lacked
GST-71 expression. These characteristics for TCA-induced tumors were also reported by Pereira
et al. (1997) (described in Section 4.2). Tumors in control mice were also mostly basophilic or
mixed basophilic and eosinophilic. Since comparable numbers of the foci of TCA-treated
83 DRAFT - DO NOT CITE OR QUOTE
-------
animals were basophilic and eosinophilic, the author suggested that the basophilic foci induced
by TCA treatment may be more likely to progress to tumors. Based on differences in the shape
of the tumor dose-response curves and staining characteristics of tumors, Pereira (1996)
concluded that DC A and TCA act through different mechanisms. The characteristics of the foci
and tumors induced by TCA were described as being consistent with the predominant basophilic
staining observed in tumors induced by peroxisome proliferators, suggesting that this pathway
might be involved in the observed hepatocarcinogenicity of TCA.
Similarly, Bull et al. (1990) (described in Section 4.2) also presented evidence that the
mechanisms of TCA and DCA carcinogenesis are different. In this study, DCA-treated mice
showed marked cytomegaly, substantial glycogen accumulation, and necrosis of the liver. The
dose-response relationship between proliferative liver lesions and DCA treatment followed a
"hockey stick" pattern. In contrast, these effects were either minimal or absent in TCA-treated
mice, and accumulation of lipofuscin (an indication of lipid peroxidation) was observed only in
TCA-treated mice. In contrast to the dose-response curve for DCA, the dose-response curve for
TCA and proliferative lesions was linear. Based on these data, the authors suggested that DCA
may induce tumors by stimulating cell division through cytotoxicity, while TCA may induce
tumors via lipid peroxidation.
4.5.2. Genotoxicity Studies
4.5.2.1. In Vitro Studies
TCA has been evaluated in a number of in vitro test systems (Table 4-12). The
mutagenicity of TCA has been assessed in several variations of the Ames test. Among the
Salmonella typhimurium strains that have been evaluated (i.e., TA98, TA100, TA104, TA1535,
and RSJ100), the available studies have produced mixed results. Rapson et al. (1980) reported
negative results for TCA in strain TA100 in the absence of metabolic activation (S9). Similarly,
Nelson et al. (2001) reported negative results in strain TA104 with or without the addition of S9
or rat cecal homogenate. In an assay designed to investigate the genotoxicity of the volatile
organic solvent tetrachloroethylene and its metabolites, TCA was also negative in S.
typhimurium TA100 at up to cytotoxic concentrations of 600 ppm (without S9) and -80 ppm
(with S9). The assay utilized the vaporization technique, which permits the evaluation of volatile
agents as vapors within a closed system (DeMarini et al., 1994). In this system, agar cultures on
Petri dishes were inserted into a sealed Tedlar bag, and various amounts of the test compound
were injected through a septum on the bag into the inverted top of the Petri dish. In a more
recent study by Kargalioglu et al. (2002), TCA (0.1-100 mM) was not mutagenic when tested in
TA98, TA100, and RSJ100 with or without S9.
84 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-12. Summary of available genotoxicity data on TCA
Endpoint
Test system
Metabolic
activation"
Concentration/dose
Results
Reference
In vitro studies
Reverse
mutations
Prophage
induction
SOS repair
induction
SOS chromotest
Forward
mutations
Chromosomal
aberrations
Chromosomal
damage
DNA strand
breaks
S. typhimurium (TA98)
S. typhimurium (TA100)
S. typhimurium (TA100)
S. typhimurium (TA104)
S. typhimurium TA100
(TCA vapors were tested
in a closed system)
S. typhimurium TA100
(fluctuation assay)
S. typhimurium
RSJ100
Escherichia coli
microscreen assay
S. typhimurium (TA1535)
E. coli (PQ37)
Cultured mammalian
cells (Chinese hamster
ovary Kl cells) HGPRT
assay
Cultured mammalian
cells (L5178Y/TK+/-
mouse lymphoma cells)
Mouse lymphoma cells
Cultured human
peripheral lymphocytes
Chinese hamster ovary
AS52 cells
+/-
+/-
-
+/-
+/-
+/-
+/-
+/-
+
+/-
+/-
+/-
+/-
—
10-80 mM
5-100/0.5-80 mM
0. 1-1,000 ug/plate
1 mg/mL
0-600 mg/L
+ S9: 3,000-
7,500 ug/mL;
-S9: 1,750-
2,250 ug/mL
0. 1-100/5-80 mM
0-10 mg/mL
58.5 ug/mL
10-10,000 ug/mL
200-10,000 uM
+S9: 0-3,400 ug/mL;
-S9: 0-2,150 ug/mL
0-2,500 ug/mL
2,000 and 5,000 ug/mL
0.1-3mM
Negative
Negative
Negative
Negative
Negative
Positive,
addition of S9
decreased
mutagenicity.
Toxic
concentration:
10,000 ug/mL
with S9;
2,500 ug/mL
without S9
Negative
Negative
Positive
Negative
Negative
+ S9: weakly
positive
-S9: equivocal
Weakly positive
TCA as free
acid: positive;
neutralized
TCA: negative
Negative
Kargalioglu et
al., 2002
Kargalioglu et
al., 2002
Rapson et al.,
1980
Nelson et al.,
2001
DeMarini et
al., 1994
Ciller etal.,
1997
Kargalioglu et
al., 2002
DeMarini et
al., 1994
Ono et al.,
1991
Ciller etal.,
1997
Shao-Hui
Zhang etal.,
2010
Harrington-
Brock et al.,
1998
Harrington-
Brock et al.,
1998
Mackay et al.,
1995
Plewa et al.,
2002
85
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-12. Summary of available genotoxicity data on TCA
Endpoint
Test system
Metabolic
activation"
Concentration/dose
Results
Reference
In vivo studies
Chromosomal
aberration
Micronucleus
induction
DNA strand
breaks (alkaline
unwinding
assay)
Oxidative DNA
damage
(8-OHdG
adducts)
Swiss mice, bone marrow
Swiss mice, bone marrow
C57BL mice, bone
marrow evaluated
Newt larvae (Pleurodeles
waltl), erythrocytes
B6C3F! mice and
Sprague-Dawley rats
B6C3FJ mice
B6C3FJ mice and F344
rats
B6C3FJ mice
B6C3FJ mice
NA
NA
NA
NA
NA
NA
NA
NA
NA
0, 125, 250, or
500 mg/kgi.p.;
500 mg/kg orally
(TCA not neutralized
before administration)
0, 125, 250, or
500 mg/kg-day i.p.
(two doses)
(TCA not neutralized
before administration)
337-1,300 mg/kg-day
i.p. (25-80% of LD50)
(neutralized TCA was
administered)
40, 80, or 160 ug/mL
(TCA not neutralized
before treatment)
0.6 mmol/kg oral
(TCA not neutralized)
500 mg/kg p.o. in one,
two, or three daily
doses
(TCA neutralized)
Mice: 10 mmol/kg,
oral
Rats: 5 mmol/kg
(TCA neutralized)
300 mg/kg, single dose
(TCA neutralized)
0-3 g/L TCA oral, for
21or71d
Positive
Positive
Negative
Weakly positive
at 80 ug/mL
Positive
Negative
Negative
Positive
Negative
Bhunya and
Behera, 1987
Bhunya and
Behera, 1987
Mackay etal.,
1995
Ciller etal.,
1997
Nelson and
Bull, 1988
Styles et al.,
1991
Chang et al.,
1992
Austin et al.,
1996
Parrish et al.,
1996
aNA = not applicable; +/- = with and without S9.
In contrast, Giller et al. (1997) reported that TCA demonstrated mutagenic activity in an
Ames fluctuation test in S. typhimurium TA100 in the absence of S9 at noncytotoxic
concentrations ranging from 1,750 to 2,250 |ig/mL. The addition of S9 decreased the mutagenic
response, and genotoxic effects were observed at 3,000-7,500 |ig/mL. Cytotoxic concentrations
in the Ames fluctuation assay were 2,500 and 10,000 |ig/mL without and with microsomal
activation, respectively. Similarly, TCA induced a weak increase in "SOS DNA repair" (an
inducible error-prone repair system) in S. typhimurium strain TA1535 in the presence of S9 (Ono
etal., 1991).
In other bacterial test systems, TCA was negative in the SOS chromotest (which
measures DNA damage and induction of the SOS repair system) in Escherichia coli PQ37, +/-
86
DRAFT - DO NOT CITE OR QUOTE
-------
S9 (Giller et al., 1997). The test evaluated concentrations of TCA ranging from 10 to
10,000 |ig/mL. Similarly, TCA was not genotoxic in the Microscreen prophage-induction assay
in E. coli with TCA concentrations ranging from 0 to 10,000 |ig/mL, with and without S9
activation (DeMarini et al., 1994).
TCA mutagenicity has also been tested in cultured mammalian cells. The potential of
TCA to induce mutations in L5178Y/TK+ -3.7.2C mouse lymphoma cells was examined by
Harrington-Brock et al. (1998). The mouse lymphoma cells were incubated in culture medium
treated with TCA concentrations up to 2,150 |ig/mL without S9 metabolic activation and up to
3,400 |ig/mL with S9. TCA was in free acid form when evaluated without S9. When it was
evaluated with S9, the sodium salt form was used to maintain neutral pH. A positive response
was defined as a doubling of the background mutant frequency7. In the absence of S9, TCA
increased the mutant frequency by twofold or greater in one experiment only at concentrations
resulting in <11% survival (>2,000 |ig/mL). In a repeat experiment, cultures producing the same
level of cytotoxicity were not positive. Therefore, the authors considered the mutagenicity of
TCA without activation to be equivocal. In the presence of S9, a doubling of mutant frequency
was seen at concentrations of >2,250 |ig/mL, including several concentrations with survival
>10%. The response was considered by the study authors to be very weakly positive. Because
of the weak mutagenic response, cytogenetic analysis was not conducted with TCA-treated cells.
However, the study authors noted that the mutants included both large-colony and small-colony
mutants. The small-colony mutants are indicative of chromosomal damage, which cannot be
attributed to low pH, since the authors stated that no pH change was observed in the presence of
S9. Harrington-Brock et al. (1998) noted that TCA (with S9 activation) was one of the least
potent mutagens evaluated in this in vitro system and that the weight of evidence suggested that
TCA was unlikely to be mutagenic. Other mutagenicity/genotoxicity studies support this
conclusion (Shao-Hui Zhang et al., 2010). It is noteworthy that in vitro mutagenicity tests
mentioned in this section were not designed specifically to detect genotoxic endpoints induced
by oxidative DNA damage.
Plewa et al. (2002) evaluated the induction of DNA strand breaks by TCA (0.1-3 mM) in
Chinese hamster ovary cells. TCA was found to be not genotoxic in this assay. Mackay et al.
(1995) investigated the ability of TCA to induce chromosomal DNA damage in an in vitro assay
by using cultured human lymphocytes. Treatment with TCA as free acid, with and without
metabolic activation, induced chromosome damage in cultured human peripheral lymphocytes
7 As an outcome of the Mouse Lymphoma Assay Workgroup of the International Workshop on Genotoxicity Testing
(Moore et al., 2006), the criteria for calling a response positive in the mouse lymphoma assay has changed. A
twofold response is no longer considered to be positive. Rather, there is a requirement that the induced mutant
frequency (i.e., the response above the background mutant frequency) should exceed a global factor of 90 x io~6.
Application of the new criterion does not change the overall determination for TCA in Harrington-Brock et al.
(1998).
87 DRAFT - DO NOT CITE OR QUOTE
-------
only at concentrations (2,000 and 3,500 |ig/mL) that significantly reduced the pH of the medium.
Neutralized TCA had no effect in this assay even at a cytotoxic concentration of 5,000 |ig/mL,
suggesting that reduced pH was responsible for the TCA-induced clastogenicity in this study. To
further evaluate the role of pH changes in the induction of chromosome damage, isolated liver-
cell nuclei from B6C3Fi mice were suspended in a buffer at various pH levels and were stained
with chromatin-reactive (fluorescein isothiocyanate) and DNA-reactive (propidium iodide)
fluorescent dyes. Chromatin staining intensity decreased with decreasing pH, suggesting that pH
changes alone can alter chromatin conformation. Thus, Mackay et al. (1995) concluded that
TCA-induced pH changes were likely to be responsible for the chromosome damage induced by
un-neutralized TCA.
4.5.2.2. In Vivo Studies
TCA has been tested for genotoxicity in several in vivo test systems (Table 4-12).
Bhunya and Behera (1987) treated Swiss mice with 0, 125, 250, or 500 mg/kg unneutralized
TCA i.p. (the highest dose, 500 mg/kg, was also administered orally for the chromosome
aberration assay). Three different cytogenetic assays—bone marrow chromosomal aberrations
and micronucleus and sperm-head abnormalities—were carried out. TCA induced chromosomal
aberrations and micronuclei in bone-marrow and altered sperm morphology of treated mice. In a
later study, Mackay et al. (1995) utilized the study design of Bhunya and Behera (1987),
including an extra sampling time at 24 hours to investigate the ability of TCA to induce
chromosomal DNA damage in the in vivo bone-marrow micronucleus assay in mice. C57BL
mice were given neutralized TCA at i.p. doses of 0, 337, 675, or 1,080 mg/kg-day for males and
0, 405, 810, or 1,300 mg/kg-day for females for 2 consecutive days, and bone-marrow samples
were collected 6 and 24 hours after the last dose. The administered doses represented 25, 50, and
80% of the LDso. No significant treatment-related increase in micronucleated polychromatic
erythrocytes was observed. Mackay et al. (1995) concluded that the positive results previously
observed by Bhunya and Behera (1987) may have been due to a non-genotoxic mechanism,
possibly caused by physicochemically induced stress resulting from i.p. pH changes. In another
study, unneutralized TCA induced a small increase in the frequency of micronucleated
erythrocytes at 80 |ig/mL in a newt (Pleurodeles waltl larvae) micronucleus test (Giller et al.,
1997).
Studies on the ability of TCA to induce single-strand breaks have produced mixed results
(Chang et al., 1992; Styles et al., 1991; Nelson and Bull, 1988). Nelson and Bull (1988)
evaluated the ability of TCA to induce single-strand breaks in vivo in Sprague-Dawley rats and
B6C3Fi mice. Single oral doses of unneutralized TCA in 1% Tween were administered to three
groups of three animals, with an additional group as a vehicle control. Animals were sacrificed
after 4 hours, and 10% liver suspensions were analyzed for single-strand breaks by the alkaline
unwinding assay. Dose-dependent increases in single-strand breaks were induced in both rats
88 DRAFT - DO NOT CITE OR QUOTE
-------
and mice, with mice being more susceptible than rats. The lowest doses of TCA that produced
significant single-strand breaks were 0.6 mmol/kg (98 mg/kg) in rats and 0.006 mmol/kg (0.98
mg/kg) in mice.
Styles et al. (1991) tested TCA for its ability to induce strand breaks in male B6C3Fi
mice in the presence and absence of liver growth induction. The test animals were given one,
two, or three daily doses of neutralized TCA (500 mg/kg) by gavage and killed 1 hour after the
final dose. Additional mice were given a single 500 mg/kg gavage dose and sacrificed 24 hours
after treatment. Liver nuclei DNA were isolated, and the induction of single-strand breaks was
evaluated by using the alkaline unwinding assay. Exposure to TCA did not induce strand breaks
under the conditions tested in this assay. In a study by Chang et al. (1992), administration of
single oral doses of neutralized TCA (1-10 mmol/kg) to B6C3Fi mice did not induce DNA
strand breaks in a dose-related manner as determined by the alkaline unwinding assay. No DNA
damage (as strand breakage) was detected in F344 rats administered by gavage up to 5 mmol/kg
(817 mg/kg) neutralized TCA. In evaluating these studies, the reason for the inconsistent results
among studies may be related to whether TCA was administered as sodium salt (neutralized) or
as free acid (not neutralized). The different results did not appear to be related to the method
chosen to measure strand breakage. Although Chang et al. (1992) used a different unwinding
assay, Nelson and Bull (1988) and Styles et al. (1991) employed the same unwinding assay and
obtained contrasting results.
Two related studies were conducted to evaluate the relationship between TCA-induced
lipid peroxidation and oxidative DNA damage (Austin et al., 1996; Parrish et al., 1996)
(described in detail in Section 4.2.1.1). In the acute study by Austin et al. (1996), male B6C3Fi
mice (six/group) were treated with a single oral dose of TCA (0, 30, 100, or 300 mg/kg), and
8-OHdG adducts were measured in liver DNA. A significant increase of about one-third in
8-OHdG levels was observed in the 300 mg/kg group at 8-10 hours post-dosing. Parrish et al.
(1996) expanded on this study by evaluating TCA-induced oxidative DNA damage following
repeated dosing. Male B6C3Fi mice (six/group) were exposed to 0, 100, 500, or 2,000 mg/L
TCA in drinking water for either 3 or 10 weeks (approximate doses of 0, 25, 125, or 500 mg/kg-
day). The levels of 8-OHdG were unchanged at both time periods. Thus, oxidative damage to
genomic DNA as measured by 8-OHdG adducts did not occur with prolonged TCA treatment.
In summary, these data collectively provide limited evidence regarding the genotoxicity
of TCA. No mutagenicity was reported in S. typhimurium strain TA100 in the absence of
metabolic activation (Rapson et al., 1980) or in an alternative protocol using a closed system
(DeMarini et al., 1994), but a mutagenic response was induced in this same strain in the Ames
fluctuation test reported by Giller et al. (1997). On the other hand, mutagenicity in mouse
lymphoma cells was only induced at cytotoxic concentrations (Harrington-Brock et al., 1998).
Measures of DNA-repair responses in bacterial systems have been similarly inconclusive, with
induction of DNA repair reported in S. typhimurium by Ono et al. (1991) but not by Giller et al.
89 DRAFT - DO NOT CITE OR QUOTE
-------
(1997) in E. coli. Although positive results were reported for unneutralized TCA in three in vivo
cytogenetic assays by Bhunya and Behera (1987), later in vitro studies by Mackay et al. (1995),
using neutralized TCA, reported negative results, suggesting that TCA-induced clastogenicity
may occur secondary to pH changes. TCA-induced hepatic DNA strand breaks and chromosome
damage have been observed in two studies (Giller et al., 1997; Nelson and Bull, 1988) and were
suggested by the results of Harrington-Brock et al. (1998). However, these effects have not been
uniformly reported (Chang et al., 1992; Styles et al., 1991) and may be related to low pH when
TCA was not neutralized. TCA induced oxidative DNA damage in the livers of mice following
a single dose (Austin et al., 1996) but not following repeated dosing over 3 or 10 weeks (Parrish
etal., 1996).
4.6. SYNTHESIS OF MAJOR NONCANCER EFFECTS
No epidemiologic data that evaluate TCA alone for noncancer effects in humans are
available. The experimental database for animals includes acute, short-term, subchronic, and
chronic studies conducted in rats and mice. A major limitation of the experimental database is
that few studies have examined toxic effects in organs other than the liver. Based on the
currently available data, oral exposure of rats and mice to TCA induces systemic, noncancer
effects in animals that can be grouped into three general categories: metabolic alterations, liver
toxicity, and developmental toxicity. These effects are described below.
4.6.1. Oral
4.6.1.1. Metabolic Alterations
Chronic exposure to TCA results in accumulation of lipofuscin in areas surrounding
hepatoproliferative lesions in the liver of mice (Bull et al., 1990). Lipofuscin is a complex of
lipid-protein substances derived from lipid peroxidation of membranes and hence provides
evidence of lipid peroxidation initiated by a free radical species generated from its metabolism.
Alternatively, Bull et al. (1990) suggested that accumulation of lipofuscin could be related to the
ability of TCA to induce peroxisomal oxidative enzymes. TCA also demonstrated its ability to
induce lipid peroxidation by the formation of TEARS in the liver of rats and mice when
administered acutely (Austin et al., 1996; Larson and Bull, 1992). This lipid peroxidation
response was reduced with pretreatment of TCA for 14 days (Austin et al., 1995). Decreased
liver triglyceride and cholesterol levels were observed in Wistar rats treated with 25 ppm TCA in
drinking water for 10 weeks, while serum triglyceride level increased (Acharya et al., 1995).
Exposure to TCA has been reported to alter liver glycogen content secondary to
alterations of lipid and carbohydrate homeostasis (Acharya et al., 1995; Bull et al., 1990;
Sanchez and Bull, 1990). TCA significantly increased glycogen content in the livers of rats
exposed to 25 mg/L in the drinking water (neutralization not reported) for 10 weeks, as assessed
by analysis of liver homogenates (Acharya et al., 1995). Bull et al. (1990) reported that
90 DRAFT - DO NOT CITE OR QUOTE
-------
"TCA-treated animals displayed less evidence of glycogen accumulation [than DCA-treated
animals] and it was more prominent in periportal than centrilobular portions of the liver acinus"
as assessed by periodic acid-Schiff s reagent staining in a 52-week study of mice exposed to 1 or
2 g/L in drinking water. In a study where mice were exposed to 0.3, 1.0, or 2.0 g/L TCA in
neutralized drinking water for 14 days, Sanchez and Bull (1990) reported that glycogen, as
detected by periodic acid-Schiff s reagent-staining in hepatic sections from animals receiving the
highest concentrations of TCA, "displayed a much less intense staining [than DCA-treated mice]
that was confined to periportal areas." In contrast, Kato-Weinstein et al. (2001) reported
significantly decreased glycogen content, especially in the central lobular region in mice treated
with 3.0 g/L in neutralized drinking water for 4 or 8 weeks and in mice treated with 0.3, 1.0, or
3.0 g/L for 12 weeks, as measured chemically in liver preparations and verified histologically by
periodic acid-Schiff s reagent staining. The reason for the discrepancy is unknown, but does not
appear to be related to differences in study duration or administered dose.
4.6.1.2. Liver Toxicity
The liver has consistently been identified as a target organ for TCA toxicity in short-term
(Sanchez and Bull, 1990; DeAngelo et al., 1989; Goldsworthy and Popp, 1987) and longer-term
(Bhat et al., 1991; Bull et al., 1990; Mather et al., 1990) studies. Collective analysis of the
available studies reveals a common spectrum of liver effects that includes changes in lipid and
carbohydrate homeostasis, increased liver weight, increased hepatic DNA labeling, and
hepatocyte necrosis. The ability of TCA to induce peroxisome proliferation has been a primary
endpoint evaluated (DeAngelo et al., 2008, 1997; Parrish et al., 1996).
TCA induced peroxisome proliferation (in the absence of effects on liver weight) in
B6C3Fi mice exposed for 3 or 10 weeks to drinking water concentrations as low as 0.5 g/L
(approximately 125 mg/kg-day) (Parrish et al., 1996). The NOAEL in this study was 25 mg/kg-
day. In rats exposed to TCA for up to 104 weeks (DeAngelo et al., 1997), peroxisome
proliferation was observed at 364 mg/kg-day but not at 32.5 mg/kg-day. Peroxisome
proliferation has also been demonstrated in a number of other short-term and long-term exposure
studies in both rats and mice (DeAngelo et al., 2008, 1989; Austin et al., 1995; Mather et al.,
1990; Parnell et al., 1988; Goldsworthy and Popp, 1987). Increased liver weight and significant
increases in hepatocyte proliferation have been observed in short-term studies in mice at doses as
low as 100 mg/kg-day (Dees and Travis, 1994), but no increase in hepatocyte proliferation was
noted in rats given TCA at up to 364 mg/kg-day (DeAngelo et al., 1997). More clearly adverse
liver toxicity endpoints, including increased serum levels of liver enzymes (indicating leakage
from cells) or histopathologic evidence of necrosis, have been reported in rats but generally only
at high doses. For example, increased hepatocyte necrosis was observed at a dose of 364 mg/kg-
day in a rat chronic drinking water study (DeAngelo et al., 1997). Increased hepatic necrosis
91 DRAFT - DO NOT CITE OR QUOTE
-------
was observed in male B6C3Fi mice treated with >68 mg/kg-day TCA in a chronic drinking
water study (DeAngelo et al., 2008).
Rats are less sensitive than mice to the peroxisome-proliferating effects of TCA. For
example, PCO activity was measured by DeAngelo et al. (1989) (described in Section 4.2) in
four strains of male mice and three strains of male rats exposed to TCA in drinking water for
14 days. PCO activity was increased by 648-2,500% over controls in the mouse strains
compared with increases of up to 138% over controls in rats at the same drinking water
concentrations (31 mM), demonstrating the greater response in exposed mice. The relevance of
TCA effects associated with peroxisome proliferation to human health is presently uncertain.
Further information on this issue is presented in Section 4.7.3.1.1.4.
4.6.1.3. Developmental Toxicity
Six published studies have addressed the developmental toxicity of TCA in rats exposed
via the oral route. Some of these studies were conducted with excessively high TCA
concentrations (Singh, 2006, 2005a, b) or with a single dose of TCA (Fisher et al., 2001; Johnson
et al., 1998), and therefore provide limited information useful for informing the dose-response
relationship for TCA in the low-dose region.
Nevertheless, available data indicate that TCA is a developmental toxicant in the
pregnant rat at doses of >300 mg/kg-day. TCA decreased fetal weight and length and increased
cardiovascular malformations at 330 mg/kg-day in Long-Evans rats (Smith et al., 1989). In a
study focused on cardiac teratogenicity, Fisher et al. (2001) observed significantly reduced fetal
body weights on GD 21 following treatment of Sprague-Dawley rats with 300 mg/kg-day of
TCA. In contrast to the Smith et al. (1989) study, Fisher et al. (2001) did not observe treatment-
related effects on the incidence of cardiac malformations. The reason for the inconsistent
findings is unknown. Smith et al. (1989) considered levocardia to be an ill-defined malformation
and possibly of trivial appearance as found in Bouin's fixed slices. Thus, the finding of possible
cardiac malformations is of uncertain biological relevance. Overall, the available data do not
permit identification of NOAEL values for the developmental or maternal toxicity of TCA, since
in each study, adverse effects were observed at the lowest or only dose tested.
Mouse and rat whole embryo cultures have been used to assess the potential for
developmental toxicity of TCA (Hunter et al., 1996; Saillenfait et al., 1995). TCA induces a
variety of morphologic changes in mouse and rat whole embryo cultures, supporting the
appearance of soft-tissue malformations observed in vivo at maternally toxic doses. The
xenopus assay system (frog embryo teratogenesis assay) (Fort et al., 1993) provided positive
results for developmental toxicity of TCA. Because of the high concentrations used in these
assays, however, in vitro test systems are limited in their utility to predict adverse developmental
effects and associated toxic potencies in intact organisms.
92 DRAFT - DO NOT CITE OR QUOTE
-------
4.6.2. Inhalation
No inhalation studies of TCA are available.
4.6.3. Mode-of-Action Information - Non-Cancer
Target organs for the toxicity of TC A in humans have not been specifically identified.
The experimental database for MOA in animals is limited to studies in rats and mice, and few
studies have evaluated events in organs other than the liver. Based on currently available data,
systemic, noncancer effects induced in animals can be grouped into three general categories:
metabolic alterations, liver toxicity, and developmental toxicity.
4.6.3.1. Metabolic Alterations
Exposure to TCA causes disturbances in lipid homeostasis. TCA is a PPARa agonist.
An associated event with the activation of PPARa receptor by TCA is proliferation of
peroxisomes (reviewed in Bull [2000]; Austin et al. [1996, 1995]; Parrish et al. [1996]).
Peroxisomes contain hydrogen peroxide and fatty acid oxidation systems important in lipid
metabolism. Activation of the peroxisome proliferation pathway induces the transcription of
genes that encode enzymes responsible for fatty acid metabolism (Lapinskas and Gorton, 1999),
suggesting that lipid homeostasis might be affected through this mechanism. Alternatively,
metabolism of TCA might generate free radical species that initiate lipid peroxidation (Bull et
al., 1990). The appearance of DCA in the urine of TCA-exposed animals provided evidence for
a free radical-generating, reductive dechlorination metabolic pathway (Larson and Bull, 1992).
TCA has been reported to induce glycogen accumulation in rats (Acharya et al., 1995)
and possibly in mice (Bull et al., 1990; Sanchez and Bull, 1990). The data are not fully
consistent, however, since Kato-Weinstein et al. (2001) observed decreased glycogen content in
mice treated with TCA. Although TCA-induced changes in glycogen storage have not been well
studied, examination of DCA effects on the same endpoint can be informative. DCA-induced
glycogen accumulation is potentially pathological, because chronic treatment might result in
glycogen stores becoming difficult to mobilize (Kato-Weinstein et al., 1998). The mechanism
for glycogen accumulation is not known, but it may be associated with inhibition of
glycogenolysis, since the observed effects resemble those observed in glycogen storage disease,
an inherited deficiency or alteration in any one of the enzymes involved in glycogen metabolism.
In this regard, the enzymatic basis for increased hepatic glycogen accumulation was studied by
Kato-Weinstein et al. (1998). TCA was not evaluated as part of this study. However, TCA
might act similarly to DCA, since both compounds induce glycogen accumulation (Acharya et
al., 1995), although the degree of accumulation is less with TCA. Kato-Weinstein et al. (1998)
reported that DCA concentrations that induced glycogen accumulation did not alter glycogen
synthase activity and had no effect on glycogen phosphorylase (which degrades glycogen) or the
activity of glucose-6-phosphatase (which converts glucose-6-phosphate to glucose) from liver
93 DRAFT - DO NOT CITE OR QUOTE
-------
homogenates. In an in vitro study using purified enzyme, DCA did not alter the activity of
glycogen synthase kinase-3p (which down-regulates glycogen synthase activity and up-regulates
glycogen phosphorylase activity). Based on the absence of an effect on enzymes that regulate
glycogen synthesis rates and decreased glycogen degradation observed in fasted mice, the
authors concluded that glycogen accumulation was related to a decrease in degradation rate.
There are currently no data on TCA to show that it acts via a similar MO A.
4.6.3.2. Liver Toxicity
Increased liver weight is typically observed concurrently with or at lower doses than
other endpoints following oral dosing with TCA. Changes in liver weight can reflect increases in
cell size, cell number, or both. TCA appears to induce both hepatocellular enlargement (Acharya
et al., 1997; Mather et al., 1990) and cell proliferation as assessed by differences in hepatocyte
DNA labeling (Dees and Travis et al., 1994; Sanchez and Bull, 1990). Increased cell
proliferation in normal cells may, however, be transient, with no change or even decreased
growth observed after chronic exposure (DeAngelo et al., 1997; Pereira, 1996). Both
cytomegaly and increased cell proliferation might be explained by TCA-induced peroxisome
proliferation (Lapinskas and Gorton, 1999). There is little evidence that increased cell
proliferation is secondary to hepatocyte cytotoxicity, as previously discussed in Section 4.5.1.2,
although TCA can induce hepatic necrosis at high doses (DeAngelo et al., 1997).
Oxidative stress may also contribute to the toxicity of TCA in the liver. Several studies
have shown that TCA induces oxidative-stress responses (e.g., lipid peroxidation and oxidative
DNA damage) in the liver in single-dose or short-term studies (Austin et al., 1996, 1995; Parrish
et al., 1996; Larson and Bull, 1992). Oxidative stress may contribute to the short-term toxicity of
TCA; however, the contribution of oxidative stress to the chronic toxicity of TCA is uncertain
because the response is transient and is not observed in longer-term studies (Parrish et al., 1996).
4.6.3.3. Developmental Toxicity
The mechanisms for developmental toxicity are unknown. However, TCA was found to
accumulate in amniotic fluid when pregnant rodents were exposed to TCE or tetrachloroethylene
(Ghantous et al., 1986). Thus, TCA may also have accumulated in amniotic fluid when pregnant
rodents were exposed to this chemical, because most of the parent compound remains
unmetabolized. TCA accumulated in the amniotic fluid may be transported through fetal skin or
swallowed then excreted by the fetus. Singh (2006) suggested that TCA in the amniotic fluid
may be circulated several times, which would contribute to the long-term retention in the fetus.
Since TCA is a strong acid with high protein binding and was reported to cause placental lesion
(Ghantous et al., 1986), developmental toxicities may be related to anoxia resulting from toxic
effect on the placenta and from apoptosis resulting from oxidative stress, as observed in studies
by Singh (2006, 2005a, b). On the other hand, Selmin et al. (2008) reported that TCA disrupted
94 DRAFT - DO NOT CITE OR QUOTE
-------
the expression of genes involved in processes important during embryonic development. A
microarray study conducted on P19 mouse embryonic carcinoma cells treated with TCA
provided evidence that TCA altered the expressions of several genes implicated in calcium
regulation and heart development (Selmin et al., 2008). Real-time polymerase chain reaction
analysis confirmed the effect of TCA on genes involved in calcium regulation (CamK and RyR),
glucose/insulin signaling (Dok3), and ubiquitin-mediated cell proliferation (Ubec2).
4.7. EVALUATION OF CARCINOGENICITY
4.7.1. Summary of Overall Weight of Evidence
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), TCA is "likely
to be carcinogenic to humans" based on statistically significantly increased incidences of
hepatocellular adenomas and carcinomas in male and female B6C3Fi mice in multiple studies
following lifetime and less-than-lifetime oral exposures in drinking water (DeAngelo et al.,
2008; Bull et al., 2002, 1990; Pereira, 1996; Pereira and Phelps, 1996; Herren-Freund et al.,
1987). Additionally, liver tumors were observed in mouse tumor promotion assays with and
without initiation (Pereira et al., 2001, 1997; Pereira and Phelps, 1996; Herren-Freund et al.,
1987). Treatment-related tumors were not observed in a study of male F344/N rats following
lifetime exposure in drinking water (DeAngelo et al., 1997). No information is available on the
carcinogenicity of TCA in humans.
The Cancer Guidelines emphasize the importance of weighing all of the evidence in
reaching conclusions about the human carcinogenic potential of agents. Choosing a descriptor is
a matter of judgment and cannot be reduced to a formula. Each descriptor may be applicable to a
wide variety of potential data sets and weights of evidence. The "likely to be carcinogenic to
humans" descriptor is appropriate when the weight of the evidence is adequate to demonstrate
carcinogenic potential to humans but does not reach the weight of evidence for the descriptor
"carcinogenic to humans." Examples provided in the Cancer Guidelines (U.S. EPA, 2005a)
include "an agent that has tested positive in animal experiments in more than one species, sex,
strain, site, or exposure route, with or without evidence of carcinogenicity in humans" and "a
positive tumor study that raises additional biological concerns beyond that of a statistically
significant result, for example, a high degree of malignancy, or an early age at onset."
TCA induced liver adenomas and carcinomas in male and female B6C3Fi mice in
multiple drinking water bioassays. EPA acknowledges that the mouse, and in particular the
B6C3Fi mouse, is relatively susceptible to liver tumors, thus background incidence of this tumor
is generally high. For these reasons, use of mouse liver tumor data in risk assessment has been a
subject of controversy (King-Herbert and Thayer, 2006). The majority of TCA drinking water
bioassays in the B6C3Fi mouse (DeAngelo et al., 2008; Bull et al., 2002, 1990; Pereira, 1996)
reported relatively low incidences of liver adenomas and carcinomas in control animals (ranging
from 0 to 13%), thereby minimizing the possible confounding of compound-related liver tumors.
95 DRAFT - DO NOT CITE OR QUOTE
-------
Furthermore, it is noteworthy that statistically significant increases in tumor incidence were
induced by TCA following drinking water exposures of only 51-82 weeks in these studies. In
tumor promotion assays, TCA induce liver tumors in mice with and without pre-treatment with
an initiator (Herren-Freund et al., 1987; Pereira and Phelps, 1996; Pereira et al., 2001).
Although there is evidence from multiple drinking water bioassays that TCA can cause
cancer in male and female mice, the characterization of the carcinogenic potential to humans is
complicated by the limited scope of testing of TCA for carcinogenicity (i.e., all but one assay
were conducted in a single mouse strain by a single route of exposure).
As emphasized by EPA in Section 2.5 of the Cancer Guidelines (U.S. EPA, 2005a),
cancer descriptors represent points along a continuum of evidence; consequently, there are
gradations and borderline cases. Therefore, although the tumor data for TCA can be considered
consistent with the descriptor of "likely to be carcinogenic to humans," the evidence supporting
the descriptor of "suggestive evidence of carcinogenic potential" was also considered. This
descriptor is appropriate when a concern for potential carcinogenic effects in humans is raised,
but the data are judged not sufficient for a stronger conclusion or when there are few pertinent
data to form a conclusion about the agent's carcinogenic potential (U.S. EPA, 2005a).
EPA identified several limitations associated with the carcinogenic database for TCA and
with some of the studies in particular. The only lifetime TCA cancer bioassay (104-week study
by DeAngelo et al., 2008) was conducted in male mice only. The interpretation of the positive
liver tumor findings (increases of 44 and 71% at the low and high dose, respectively) in this
study is complicated by a high rate of background tumors (i.e., 55% in the controls). The
remaining mouse cancer bioassays for TCA are limited to the B6C3Fi strain only; other mouse
strains have not been tested for TCA carcinogenicity. Additionally, other than Bull et al. (1990),
the studies only evaluated a single sex and none of the studies performed a comprehensive
histologic evaluation in all of the mice tested.
Cancer bioassay information in other species is limited to a lifetime study in male F344
rats (104-week study by DeAngelo et al., 1997) in which TCA did not induce tumors. GGT-
positive foci (closely linked to the subsequent development of tumors) were observed in DEN-
initiated male Sprague-Dawley rats following promotion with TCA; one rat developed a liver
carcinoma (Parnell et al., 1988). However, the attribution of this effect to TCA is confounded by
the fact that the treated rats also received a partial hepatectomy, which can itself act as a
promoter. Because female rats were not included in the DeAngelo et al. (1997) bioassay and
because other species have not been tested for TCA carcinogenic potential, limited conclusions
can be drawn about the carcinogenic properties of TCA in experimental animals other than the
B6C3Fi mouse.
Based on an in-depth review of the available information related to TCA tumor induction
and after consideration of all pertinent issues, EPA has concluded that despite the limited scope
of testing, the positive findings observed in male and female B6C3Fi mice and the early age of
96 DRAFT - DO NOT CITE OR QUOTE
-------
onset of tumor development in multiple studies is most consistent with a characterization that
TCA is "likely to be carcinogenic to humans."
EPA's Cancer Guidelines (U.S. EPA, 2005a) indicate that for tumors occurring at a site
other than the initial point of contact, the weight of evidence for carcinogenic potential may
apply to all routes of exposure that have not been adequately tested at sufficient doses. An
exception occurs when there is convincing toxicokinetic data that absorption does not occur by
other routes. For TCA, systemic tumors were observed in mice following oral exposure.
Information on carcinogenic effects via the inhalation or dermal routes in humans or animals is
absent. Data evaluating absorption by the inhalation route are unavailable and limited data are
reported for dermal absorption (Kim and Weisel, 1998). However, TCA is highly soluble in
water. Thus, it is reasonable to assume that TCA can be absorbed and taken up into the blood
via the inhalation route. Moreover, based on the observance of systemic tumors following oral
exposure, and in the absence of information to indicate otherwise, it is assumed that an internal
dose will be achieved regardless of the route of exposure. Therefore, TCA is "likely to be
carcinogenic to humans" by all routes of exposure.
The MOA for TCA-induced liver carcinogenesis has not been established. The available
data collectively provide limited evidence regarding the genotoxicity of TCA. Tumor induction
appears to include perturbation of cell growth, both through growth inhibition of normal cells
and proliferation of selected cell populations. Specific mechanisms of altered growth control
that have been investigated for TCA include activation of the PPARa pathway, global DNA
hypomethylation, reduced intercellular communication, and oxidative stress. Of these, PPARa
agonism has been advanced as the most likely MOA contributing to the development of liver
tumors. However, significant gaps in knowledge exist in the hypothesized PPARa MOA (Ito et
al., 2007; Yang et al., 2007), such that the formation of liver tumors cannot be sufficiently
accounted for by this proposed MOA and the existence of other contributing MOA(s) is
assumed.
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence
There are no epidemiologic studies of TCA carcinogenicity in humans. Most of the
human health data for chlorinated acetic acids concern components of complex mixtures of water
disinfectant byproducts. These complex mixtures of disinfectant byproducts have been
associated with increased potential for bladder, rectal, and colon cancer in humans (reviewed by
Boorman et al. [1999]; Mills et al. [1998]).
The experimental database for carcinogenicity of TCA consists of studies in rats and
mice. Studies in mice indicate that TCA is a complete carcinogen that significantly increased the
incidence of liver tumors in male and female B6C3Fi mice exposed via drinking water for 52-
104 weeks (DeAngelo et al., 2008; Bull et al., 2004, 2002, 1990; Pereira et al., 2001, 1997;
Pereira, 1996; Pereira and Phelps, 1996; Herren-Freund et al., 1987). Incidence of tumors
97 DRAFT - DO NOT CITE OR QUOTE
-------
increased with increasing TCA concentrations (DeAngelo et al., 2008; Bull et al., 2002, 1990;
Pereira, 1996). These results were obtained under conditions where the background incidence of
tumors in control animals was generally low. The development of tumors in animals exposed to
TCA progressed rapidly, as evident from the observation of significant numbers of tumors in
less-than-lifetime studies of 82 weeks or less. Positive evidence for tumor promotion by TCA
(following exposure to known tumor initiators) has been reported for liver tumors in B6C3Fi
mice (Pereira et al., 2001, 1997) and for GGT-positive foci in livers of partially hepatectomized
Sprague-Dawley rats (Parnell et al., 1988).
In contrast to the results observed for mice, treatment-related tumors were not observed
in a study of male F344/N rats exposed to TCA via drinking water for 104 weeks (DeAngelo et
al., 1997). The carcinogenicity of TCA has not been evaluated in female rats or in other species
of experimental animals. However, treatment of primary cultures of male Long-Evans rat
hepatocytes with 0.01-1.0 mM TCA for 10-40 hours did not induce proliferation of the cultured
hepatocytes (Walgren et al., 2005).
A significant limitation of the experimental database for carcinogenicity is the limited
number of studies that included microscopic examination of a comprehensive set of organs in
addition to the liver. The most complete evaluations were conducted by DeAngelo et al. (2008),
who examined a comprehensive set of organs in B6C3Fi mice from the high-dose and control
groups. The kidney, liver, spleen, and testes were examined in all dose groups. DeAngelo et al.
(1997) also examined a comprehensive set of organs in F344 rats receiving the highest dose of
TCA and selected tissues (kidney, liver, spleen, testes) in the remainder of the treatment groups.
Evidence for genotoxic activity of TCA is inconclusive. No mutagenicity was reported in
S. typhimurium strain TA100 in the absence of metabolic activation (Rapson et al., 1980) or in an
alternative protocol using a closed system (DeMarini et al., 1994), but a mutagenic response was
induced in this same strain in the Ames fluctuation test reported by Giller et al. (1997).
Mutagenicity in mouse lymphoma cells was only induced at cytotoxic concentrations
(Harrington-Brock et al., 1998). Measures of DNA-repair responses in bacterial systems are
similarly inconclusive, with induction of DNA repair reported in S. typhimurium (Ono et al.,
1991), but not in E. coli (Giller et al., 1997). Although positive results were reported for
unneutralized TCA in three in vivo cytogenetic assays by Bhunya and Behera (1987), later in
vitro studies by Mackay et al. (1995), using neutralized TCA, reported negative results,
suggesting that TCA-induced clastogenicity may occur secondary to pH changes. Some
evidence for TCA-induction of hepatic DNA strand breaks and chromosome damage has been
reported (Harrington-Brock et al., 1998; Giller et al., 1997; Nelson and Bull, 1988); however,
these effects have not been uniformly reported (Chang et al., 1992; Styles et al., 1991) and may
be related to low pH when TCA was not neutralized. TCA induced oxidative DNA damage in
the livers of mice following a single dose (Austin et al., 1996), but not following repeated dosing
over 3 or 10 weeks (Parrish et al., 1996).
98 DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3. Mode-of-Action Information - Cancer
Multiple studies have demonstrated that exposure to TCA in drinking water for periods of
52-104 weeks can produce an increased incidence of liver tumors in B6C3Fi mice (DeAngelo et
al., 2008; Bull et al., 2004, 2002, 1990; Pereira et al., 2001, 1997; Pereira, 1996; Pereira and
Phelps, 1996; Herren-Freund et al., 1987). In the only available chronic study in rats, TCA did
not increase tumor incidence in male F344 rats exposed to TCA for up to 102 weeks (DeAngelo
et al., 1997). The events leading to the development of liver cancer in mice exposed to TCA
have not been fully characterized, although several MO As have been postulated.
Analysis of the available MOA information reveals that the cancer MOA for TCA is
complex, and more than one MOA may be operative in the development of mouse liver tumors.
This section will discuss the evidentiary support for several hypothesized modes of action for
liver carcinogenicity (including peroxisome proliferation, as well as several additional proposed
hypotheses and key events with limited evidence or inadequate experimental support), following
the framework outlined in the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a, b).8
Specific cancer MO As for TCA addressed in the following sections include PPARa-
agonism (Section 4.7.3.1), Kupffer cell activation and subsequent release of cytokines and
oxidants (Section 4.7.3.2), DNA hypomethylation (Section 4.7.3.3), decreased intercellular
communication (Section 4.7.3.4), and genotoxicity (Section 4.7.3.5). Possible key events in
these hypothesized MO As for hepatocarcinogenesis are illustrated in Figure 4-1.
8 As recently reviewed (Guyton et al., 2008) the approach to evaluating mode of action information described in
EPA's Guidelines for Carcinogen Risk Assessment (2005a) considers the issue of human relevance of a
hypothesized mode of action in the context of hazard evaluation. This excludes, for example, consideration of
toxicokinetic differences across species; specifically, the Cancer Guidelines state, "the toxicokinetic processes that
lead to formation or distribution of the active agent to the target tissue are considered in estimating dose but are not
part of the mode of action." In addition, information suggesting quantitative differences in the occurrence of a key
event between test species and humans are noted for consideration in the dose-response assessment, but is not
considered in human relevance determination.
99 DRAFT - DO NOT CITE OR QUOTE
-------
Trichloroaceticacid
Activation of non-
parenchymal cells
(macrophages)
Superoxide
production
Hypomethylation
Inhibition of
intercellular
communication
Cell cycle growth
and apoptosis
gene expression
Activation of
PPARa
Peroxisome
gene
expression
Peroxisome
proliferation
DNA damage, growth
inhibition of normal cells,
and proliferation of
selected cells
Oxidative
stress
Preneoplastic
foci
Figure 4-1. Possible key events in the MOA(s) for TCA carcinogenesis.
4.7.3.1. PPARa agonism9.
The hypothesis is that TCA acts by a PPARa agonism MO A in inducing mouse liver
tumors. Three key events are proposed in this MOA: activation of the receptor, perturbation of
hepatocellular apoptosis and proliferation, and selective clonal expansion.
Peroxisome proliferators are a structurally diverse group of nonmutagenic or weakly
mutagenic chemicals that induce a predictable suite of pleiotropic (multiple) responses, including
the induction of tumors in rodents (Reddy et al., 1979). At one time, peroxisome proliferation,
i.e., an increase in the number and volume fraction of peroxisomes (subcellular organelles) in the
cytoplasm of mammalian and other eukaryotic cells, was proposed as a causative factor in the
development of liver tumors. However, increased knowledge of the molecular events leading to
peroxisome proliferation suggests that it is an associative event, rather than a causal event, in the
development of liver tumors (Klaunig et al., 2003).
The data related to PPARa-agonism is relatively extensive, as this has been well-studied in order to inform
the human relevance of PPARa-agonism in hepatocarcinogenesis (e.g., Klaunig et al., 2003). Therefore, this
hypothesized MOA is discussed in the following section in relatively more detail than other hypothesized MOA
topics.
100
DRAFT - DO NOT CITE OR QUOTE
-------
Current understanding of the events leading to peroxisome proliferation indicates that
peroxisome proliferating chemicals initiate the pleiotropic response by interacting with PPARs.
PPARs are ligand-activated transcription factors that belong to the nuclear receptor
"superfamily." When activated10 by peroxisome proliferators (agonists), PPARs bind to
response elements in the promoter regions of genes and elicit changes in gene expression. Three
PPAR isoforms have been identified to date and are designated PPARa, PPARp/5, and PPARy.
Gene disruption experiments in mice indicate that PPARa is required for the pleiotropic response
(including development of liver tumors) observed following exposure to the prototypical PPARa
agonist Wy-14,643 (Escher and Wahli, 2000; Peters et al., 1997; Issemann and Green, 1990).
However, peroxisome-proliferation-like events have been observed in PPARa-null mice treated
with extremely high doses of ligands specific for other PPAR family members (Klaunig et al.,
2003), suggesting possible cross talk between PPAR isoforms. Moreover, development of liver
tumors was observed in PPARa-null mice treated with the peroxisome proliferator, di(2 ethyl-
hexyl)phthalate (DEHP), suggesting that PPARa may not be a key event in the MOA for liver
tumors of some peroxisome proliferators (Ito et al., 2007).
PPARa is highly expressed in cells that have active fatty acid oxidation capacity,
including hepatocytes, cardiomyocytes, enterocytes, and the proximal tubule cells of the kidney,
and it is well accepted that PPARa plays a central role in lipid metabolism (Dreyer et al., 1992;
Gottlicher et al., 1992). Ligand or pharmaceutical activation of PPARa facilitates increased
mobilization, transport, and oxidation of fatty acids, which serve as energy substrates during
periods of starvation or activation, by hypolipidemic drugs such as clofibrate (Gottlicher et al.,
1992). PPARa is known to interact with other transcription factors (e.g., the retinoic acid
receptor and thyroid hormone receptor), co-activators, and co-repressors to regulate gene
expression (Aranda and Pascual, 2001).
4.7.3.1.1. Identification of key events. Klaunig et al. (2003) have proposed an MOA hypothesis
for induction of liver tumors by PPARa agonists that incorporates the following key events.
PPARa ligands activate PPARa, which subsequently alters the transcription of genes involved in
peroxisome proliferation, cell cycling/apoptosis, and lipid metabolism. The changes in gene
expression lead to perturbations in cell proliferation and apoptosis and to peroxisome
proliferation. Suppression of apoptosis coupled with increased cell proliferation allows
DNA-damaged cells to persist and proliferate, resulting in preneoplastic hepatic foci and
ultimately in tumors via selective clonal expansion.
10The term "activation" refers to an alteration of the three-dimensional structure of the receptor protein or receptor
complex, resulting in altered response element binding potential. The alterations initiated by ligand binding may
include events such as loss of heat shock and chaperone proteins, nuclear translocation, and protein turnover
(Klaunig etal., 2003).
101 DRAFT - DO NOT CITE OR QUOTE
-------
In describing this progression of events, Klaunig et al. (2003) distinguish between what
they consider to be causal events (i.e., required for this MO A) and associative events (i.e.,
markers of PPARa agonism but not shown to be directly involved with formation of liver
tumors). Among the causal events postulated for PPARa-induced hepatocarcinogenesis,
activation of PPARa is highly specific for this MO A. Alterations in cell proliferation and
apoptosis and clonal expansion were also postulated causal events, but are common to other
MO As, and hence not specific to this MOA. Moreover, while it is known that activation of
PPARa leads to an increase in cell proliferation and inhibition of apoptosis, it is uncertain
whether this is due to a direct interaction with an unidentified target gene or occurs through
secondary or tertiary events. Oxidative stress occurring in conjunction with peroxisomal
proliferation is regarded as a general phenomenon and is considered neither a causal event nor a
highly specific marker of PPARa-induced liver carcinogenesis. Peroxisome proliferation may
lead to oxidative stress, which potentially contributes to the proposed MOA by causing indirect
DNA damage and/or by contributing to the stimulation of cell proliferation. This section will
focus on those key events considered required for this MOA (e.g., PPARa activation and
alterations in proliferation, gene expression, and oxidative stress) with associated key events
described in subsequent sections.
PPARa activation. The understanding of the PPARa agonism MOA has been expanded
with recent findings. As reviewed by Guyton et al. (2009), recent data strongly suggest that
PPARa and key events hypothesized by Klaunig et al. (2003) are not sufficient for
carcinogenesis induced by the purported prototypical agonist Wy-14643. Therefore, the
proposed PPARa MOA is likely "incomplete" in the sense that the sequence of key events11
necessary for cancer induction has not been identified. It has been demonstrated in a transgenic
mouse model that activation of PPARa alone in hepatocytes was not sufficient to induce
hepatocellular tumors (Yang et al., 2007). In this mouse model, the potent viral transcriptional
activator VP16 was fused to the mouse PPARa cDNA to create a transcription factor that
constitutively activates PPARa-responsive genes in the absence of ligands. The transgenic mice
demonstrated responses that mimic wild-type mice when treated with peroxisome proliferator
Wy-14,643, a potent PPARa agonist, including significantly decreased serum fatty acids and
marked induction of PPARa target genes encoding fatty acid oxidation enzymes, suggesting that
the transgene functions in the same manner as peroxisome proliferators to regulate fatty acid
metabolism. In addition, while these transgenic mice demonstrated increased hepatocellular
proliferation (Yang et al., 2007), no liver tumors were observed. Therefore, it appeared that
11 As defined by the EPA Guidelines for Carcinogen Risk Assessment (2005a) a "key event" is "an empirically
observable precursor step that is itself a necessary element of the mode of action or is a biologically based marker
for such an element," and the term "mode of action" (MOA) is defined as "a sequence of key events and processes,
starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and
resulting in cancer formation." Therefore, a single key event alone is necessary, but not necessarily sufficient for
carcinogenesis; however, the sequence of key events constituting a MOA needs to be sufficient for carcinogenesis.
102 DRAFT - DO NOT CITE OR QUOTE
-------
many of the hepatocellular responses commonly associated with PPARa agonism—fatty acid
oxidation, peroxisome proliferation, hepatocellular proliferation, and cell-cycle control gene
expression—were not sufficient to induce liver tumors. However, it should be noted that, while
most genes associated with exposure to PPARa agonists were activated in the LAP-VP16
PPARa mice, several genes (e.g., c-myc) were not activated without treatment with Wy-14643.
Thus, it appears that this PPARa agonist regulates genes in addition to those of the LAP-VP16
PPARa fusion protein.
Alterations in proliferation. Several studies have observed hepatocyte proliferation in
response to TCA in mice (e.g., DeAngelo et al., 2008; Stauber and Bull, 1997; Pereira, 1996;
Dees and Travis, 1994; Sanchez and Bull, 1990). For instance, Dees and Travis (1994) observed
relatively small (two- to threefold) but statistically significant increases in [3H]thymidine
incorporation in hepatic DNA in mice exposed for 11 days at TCA doses (100-1,000 mg/kg) that
increased relative liver weight. Increased hepatic DNA labeling was seen at doses lower than
those associated with evidence of necrosis, suggesting that TCA-induced cell proliferation is not
due to regenerative hyperplasia. PPARa-null mice exposed to 2-g/L TCA in drinking water for 7
days do not show the characteristic responses of AGO, PCO, and CYP4A induction associated
with PPARa activation and peroxisome proliferation in wild-type mice (Laughter et al., 2004).
In addition, the livers from wild-type but not PPARa-null mice exposed to TCA developed
centrilobular hepatocyte hypertrophy, although no significant increase in relative liver weight
was observed. Therefore, while there are data associating TCA exposure, PPARa activation, and
cell proliferation, it is not clear the extent to which PPARa activation is the cause of the
observed cell proliferation.
Gene expression alterations. A novel mechanism by which PPARa regulates gene
expression, hepatocellular proliferation, and tumorigenesis was uncovered by Shah et al. (2007).
Activated PPARa was demonstrated to be a major regulator of hepatic microRNA (miRNA)12
expression, especially let-7C, an miRNA found to be a potential tumor suppressor (Zhang et al.,
2007; Lee and Dutta, 2007) and to inhibit the expression of the ras oncogene (Johnson et al.,
2005). Let-7C was inhibited following treatment with 0.1% Wy-14,643 in wild-type mice for 4
hours, 2 weeks, or 11 months. No decrease in let-7C miRNA was observed in the PPARa-null
mice that underwent the same treatment. In addition, expression of the longer primary let-7C
transcript (pri-let-7C) was also decreased following 4-hour and 2-week Wy-14,643 treatments.
Moreover, pri-let-7C, AK033222, and pri-mir-99a were regulated in a PPARa-dependent
manner, since Wy-14,643 had no effect on pri-let-7C, AK033222, or pri-mir-99a in PPARa-null
mice treated for 4 hours or 2 weeks. The chromosomal positional relationship of let-7C was
found to be downstream of mir-99a and EMBL transcript AK033222 (Shah et al., 2007).
12miRNAs are noncoding RNAs that are transcribed in the nucleus as single primary transcripts (pri-miRNAs) or
large polycistronic transcripts encoding several miRNAs. Mature miRNA molecules are partially complementary to
one or more mRNA molecules, and they function to down-regulate gene expression.
103 DRAFT - DO NOT CITE OR QUOTE
-------
Shah et al. (2007) observed that let-7C regulated c-myc gene expression via direct
interaction with the 3'-untranslated region of c-myc mRNA, causing mRNA degradation.
Increasing let-7C expression in the mouse hepatoma cell line Hepa-1 decreased c-myc expression
in a dose-dependent manner. PPARa-mediated induction of c-myc via let-7C subsequently
increased expression of the oncogenic mir-17-92 polycistronic cluster, which has been
implicated in enhanced cell cycle progression, blockade of tumor cell apoptosis, and increased
neovascularization. These events did not occur in PPARa-null mice (Shah et al., 2007). When
Hepa-1 cells were transfected with 5-25 nM let-7C, at 72 hours post-transfection, cell growth
was inhibited in a dose-dependent manner. Let-7C decreased BrdU incorporation in a dose-
dependent manner, but had no effect on cell apoptosis. In addition, co-transfection of let-7C and
c-myc increased cell proliferation in Hepa-1 cells compared with cells transfected with let-7C
alone, suggesting that c-myc is a critical downstream effector of let-7C.
No difference in basal let-7C expression was observed between wild-type mice and the
LAP-VP16 PPARa transgenic mice mentioned previously, even though PPARa was activated in
the hepatocytes of transgenic mice. However, Shah et al. (2007) reported that Wy-14,643
treatment decreased let-7C expression in these transgenic mice (which still possessed native
PPARa), suggesting either that ligand treatment is needed for inhibition of let-7C, indicating that
PPARa agonists may regulate other genes in addition to the gene for the VP16 PPARa fusion
proteins, or that activation of non-parenchymal cells is critical for tumorigenesis and let-7C
expression. Moreover, let-7C was not suppressed in humanized PPARa mice, which were
resistant to Wy-14,643-induced hepatocellular proliferation and liver tumor formation (Shah et
al., 2007). Wy-14,643 treatment of humanized PPARa mice also did not induce c-myc and mir-
17 expression. These findings suggest that the let-7C signaling cascade may be critical for
PPARa agonist-induced liver proliferation and tumorigenesis. Interestingly, however, the LAP-
VP16 PPARa mice described above showed liver proliferation with neither changes in let-7C
expression nor tumorigenesis, thus suggesting that proliferation itself is a poor marker for
tumorigenicity.
Another mechanism, hypomethylation of DNA, has been proposed by Pogribny et al.
(2007) as an important link between hepatocellular proliferation and hepatocarcinogenesis in the
MO A of peroxisome proliferators. Hypomethylation of DNA is an early event to most cancers,
including liver (Yamada et al., 2005; Baylin et al., 1998; Counts and Goodman, 1995; Gama-
Sosa et al., 1983), and has been postulated to be a secondary mechanism involved in
carcinogenesis (Watson and Goodman, 2002). DNA hypomethylation is associated with opening
of the chromatin configuration and transcript onal activation, leading to chromosomal instability
and aberrant gene expression (Dunn, 2003; Baylin et al., 2001, 1998; Jones and Gonzalgo, 1997).
When male SV129 mice were fed a control diet or Wy-14,643-containing diet
(1,000 ppm) for 1 week, 5 weeks, or 5 months, treatment with Wy-14,643 led to progressive
global hypomethylation of liver DNA as determined by Hpall-cytosine extension assay, reaching
104 DRAFT - DO NOT CITE OR QUOTE
-------
the maximum effect of >200% at 5 months. Trimethylation of histone H4 lysine 20 and
H3 lysine 9 was significantly decreased at all time points. Since the majority of cytosine
methylation in mammals resides in repetitive DNA sequences, Pogribny et al. (2007) measured
the effect of Wy-14,643 on the methylation status of major and minor satellites, as well as in the
intracisternal A particle (IAP) of long terminal repeat (LTR) retrotransposone, and long
interspersed nucleotide elements (LINE) 1 and 2 (representing the non-LTR retrotransposons) in
liver DNA and found that exposure to Wy-14,643 resulted in a gradual loss of cytosine
methylation in major and minor satellites, IAP, LINE1, and LINE2 elements. Previously, gavage
of female B6C3Fi mice with 50 mg/kg Wy-14,643 for up to 4 days resulted in hypomethylation
of the c-myc gene in the liver and temporally correlated with an earlier burst of cell proliferation
(Ge et al., 2001b). No effect on c-myc promoter methylation was observed with long-term
treatment (Pogribny et al., 2007). Pogribny et al. (2007) concluded that alterations in the
genome methylation patterns with long-term exposure to Wy-14,643 may not be confined to
specific cell-proliferation-related genes. It has been demonstrated that genome-wide
hypomethylation in cancer, including liver cancer, largely involves repetitive DNA elements
(Schulz et al., 2006; Chalitchagorn et al., 2004).
Pogribny et al. (2007) also found that Wy-14,643 had no effect on DNA or histone
methylation status in PPARa-null mice at any of the evaluated time points. Previously,
treatment of PPARa-null mice with Wy-14,643 for 11 months had produced no liver tumors,
whereas treatment of wild-type mice with 1,000 ppm Wy-14,643 had resulted in 100% incidence
of hepatocellular adenomas and carcinomas (Peters et al., 1997). In addition, Wy-14,643 had no
effect on liver cell proliferation in PPARa-null mice (Woods et al., 2007c; Peters et al., 1997).
Therefore, these epigenetic alterations were PPARa-dependent and may play a key role in
hepatocarcinogenesis of peroxisome proliferators. It was suggested that peroxisome-proliferator-
induced increases in hepatocellular proliferation prevented the methylation of newly synthesized
strands of DNA (Ge et al., 2001b), since a temporal relationship between increased cell
proliferation and DNA hypomethylation of the c-myc gene was observed after a single dose of
Wy-14,643 to mice. Long-term treatment of wild-type mice with Wy-14,643 in Pogribny et al.
(2007) demonstrated gradual worsening dysregulation of normal methylation patterns in genomic
DNA.
In a subsequent study (Pogribny et al., 2008), male F344 rats were treated with 1.2% w/w
DEHP, a peroxisome proliferator, in their diet for 5 months; DNA methylation in liver was
unchanged. In another group of male F344 rats treated with 0.1% w/w Wy-14,643 in their diet
for 5 months, global hypomethylation of DNA in liver occurred, along with a significant
(twofold) increase in DNA single-strand breaks . It is unclear why the effects on DNA
hypomethylation differ between the two PPARa agonists. While these results are not consistent
with hypomethylation of DNA being a key event in the PPARa MO A, such a conclusion is
105 DRAFT - DO NOT CITE OR QUOTE
-------
complicated by the fact that DEHP can induce mouse liver tumors in a PPARa-independent
manner.
4.7.3.1.2. Biological plausibility, consistency, specificity of association. TCA is classifiable as
a peroxisome proliferator based on morphologic and biochemical evidence from multiple studies.
With respect to peroxisome proliferation, microscopic examination of responses consistent with
peroxisome proliferation (e.g., induction of lipid metabolism enzymes such as AGO and PCO,
increased liver weight) has been observed in male F344 rats exposed to TCA by gavage for
14 days (Goldsworthy and Popp, 1987), in male F344 rats exposed to TCA in drinking water for
14 days (DeAngelo et al., 1989) or 104 weeks (DeAngelo et al., 1997), in male Osborne-Mendel
rats exposed to TCA in drinking water for 14 days (DeAngelo et al., 1989), and in male Sprague-
Dawley rats treated with TCA in the drinking water for 90 days (Mather et al., 1990). In mice,
peroxisome proliferation or changes consistent with peroxisome proliferation have been reported
in male B6C3Fi mice exposed to TCA in drinking water for 2-10 weeks (Parrish et al., 1996;
Austin et al., 1995; Sanchez and Bull, 1990; DeAngelo et al., 1989), in male B6C3Fi mice
exposed by gavage for 10 days (Goldsworthy and Popp, 1987), and in male C57BL/6 and Swiss-
Webster mice exposed to TCA in the drinking water for 14 days (DeAngelo et al., 1989).
Furthermore, PPARa-null mice exposed to 2 g/L TCA in drinking water for 7 days do not show
the characteristic responses of AGO, PCO, and CYP4A induction associated with PPARa
activation and peroxisome proliferation in wild-type mice (Laughter et al., 2004). In addition,
the livers from wild-type but not PPARa-null mice exposed to TCA developed centrilobular
hepatocyte hypertrophy, although no significant increase in relative liver weight was observed.
In addition, PPARa agonism in response to treatment with TCA has been demonstrated
in vitro in COS-1 cells transfected with human and mouse PPARa expression plasmids together
with a peroxisome proliferator response element-luciferase reporter (Maloney and Waxman,
1999). Cells were treated for 24 hours with 0.1-5 mM TCA. TCA activated human and mouse
PPARa with no difference between species in receptor sensitivity or maximal responsiveness.
Third, TCA has been shown to increase hepatocyte proliferation in DNA-labeling
experiments in mice (Dees and Travis, 1994). Relatively small (two- to threefold) but
statistically significant increases in [3H]thymidine incorporation in hepatic DNA were observed
in mice exposed to 100-1,000 mg/kg-day TCA for 11 days at doses that increased relative liver
weight. Dees and Travis (1994) observed increased hepatic DNA labeling at doses lower than
those associated with evidence of necrosis, suggesting that TCA-induced cell proliferation is not
due to regenerative hyperplasia. The study authors reached this conclusion based on the pattern
of observed histopathologic changes, which indicated nodular areas of cellular proliferation, and
the results of liver DNA labeling experiments, which showed incorporation of [3H]thymidine in
extracted liver DNA but no difference in total liver DNA content (mg DNA/g liver). Dees and
Travis (1994) concluded that their results were consistent with an increase in DNA synthesis and
106 DRAFT - DO NOT CITE OR QUOTE
-------
cell division in response to TCA treatment. The authors further suggested that the absence of
histopathologic effects makes it unlikely that the increased radiolabel was secondary to tissue
repair.
Hepatocyte proliferation in response to treatment with TCA has also been demonstrated
in studies by Stauber and Bull (1997), Pereira (1996), and Sanchez and Bull (1990). Details of
these studies were provided in Sections 4.5.1.3 and 4.2.2.1. A dose-related increase in
incorporation of [3H]thymidine into hepatic DNA was observed in B6C3Fi mice treated with
0.3-2 g/L TCA for 5 or 14 days (Sanchez and Bull, 1990). This increase was significant at 2 g/L
TCA. No increases in labeled hepatocytes as seen by autoradiography were apparent at 2 or
5 days. Thus, an increase in incorporation of [3H]thymidine did not correlate with replicative
synthesis of DNA measured by autoradiography up to 5 days of treatment. Pereira (1996)
reported that TCA increased the BrdU-labeling index (calculated as the percentage of
hepatocytes with labeled nuclei) in mice exposed to 0.33-3.3 g/L TCA for 5 days but not after
12 or 33 days. Stauber and Bull (1997) reported a statistically significant two- to threefold
elevation in division rate in normal hepatocytes after male B6C3Fi mice were treated for 14 or
28 days with 2 g/L TCA. However, continued treatment for 52 weeks resulted in a decrease in
division rate in normal hepatocytes. Cell division rates in TCA-induced AHFs and tumors were
high at all TCA doses administered in the last 2 weeks of the study.
DeAngelo et al. (2008) reported hepatocyte proliferation in B6C3Fi mice exposed to
5 g/L TCA at 30 and 40 weeks, with mice exposed to 0.5 g/L TCA demonstrating hepatocyte
proliferation at 60 weeks. Therefore, DeAngelo et al. (2008) observed hepatocyte proliferation
in mice after long-term TCA treatment in contrast to Stauber and Bull (1997), who observed it as
a transient event. This result was in agreement with the observation by Woods et al. (2007b) that
the robust proliferative effect of Wy-14,643 in rodent livers extended beyond the short time
frame that was traditionally considered. Hepatocyte proliferation has been demonstrated in
chronic studies with other peroxisome proliferators (Woods et al., 2007a; Yeldandi et al., 1989;
Ward et al., 1988). It should also be noted that TCA did not induce hepatocyte proliferation or
tumors in F344 rats after 104 weeks of exposure (DeAngelo et al., 1997), consistent with the
hypothesis that cell proliferation is a causal event in tumorigenesis under the PPARa MOA.
Moreover, as presented previously, whereas PPARa-null mice treated with 2 g/L TCA in
drinking water for 7 days did not develop centrilobular hepatocyte hypertrophy, treated wild-type
mice did (Laughter et al., 2004). Thus, TCA-induced hepatocyte hypertrophy is PPARa
dependent.
A 2006 report by the NRC of the National Academy of Sciences, Assessing the Human
Health Risks ofTrichloroethylene: Key Scientific Issues (NRC, 2006), stated that "[t]here is
sufficient weight of evidence to conclude that the mode of action of trichloroacetic acid as a
rodent liver carcinogen is principally as a liver peroxisome proliferator in a specific strain of
mouse, B6C3Fi." However, the NRC (2010) panel reviewing EPA's 2008 external review draft
107 DRAFT - DO NOT CITE OR QUOTE
-------
of tetrachloroethylene did not, as a whole, support the view that the only MO A of TCA is
peroxisome proliferation. They judged that the relevance of the peroxisome proliferator MOA to
mouse and human hepatic cancer remains hypothetical and requires further rigorous testing.
Hence, the report concludes that is premature to draw conclusions on the relevance of the
PPARa MOA to human hepatic carcinogenesis (NRC, 2010).
The evidence described above supports the involvement of PPARa agonism in the overall
cancer MOA for TCA; however, some studies of PPARa published since NRC (2006), notably
Ito et al. (2007) (see discussion below), suggest that the mechanism by which TCA induces liver
tumors in mice is more complex than that presented in NRC (2006). Indeed, the 2011 Science
Advisory Board reviewing EPA's draft trichloroethylene IRIS assessment cited studies in
PPARa-null mice (Eveillard et al., 2009; Takashima et al., 2008; Ito et al., 2007), PPARa
humanized transgenic mice (Morimura et al., 2006), and hepatocyte-specific constitutively-
activated PPARa transgenic mice (Yang et al., 2007) in concluding that activation of PPARa is
an important factor, but not a limiting factor for the development of mouse liver tumors and that
additional molecular events may be involved. Inconsistencies and gaps in the data with respect
to consistency and specificity of PPARa agonism as an MOA are also discussed further below.
PPARa-independent tumor induction by DEHP. Ito et al. (2007) recently reported that
the peroxisome proliferator, DEHP, induces hepatic tumorigenesis through a PPARa-
independent pathway. Specifically, the authors administered relatively low doses of DEHP (0,
0.01, and 0.05% in diet) to wild-type and PPARa knockout mice for 22 months and found a
higher incidence of liver tumors in treated PPARa knockout than in treated wild-type mice at the
higher dose. (This was the first published study using PPARa knockout mice that were treated
for over 1 year, allowing for the full expression of tumor development.) DEHP treatment also
dose-dependently increased 8-OHdG levels in mice of both genotypes, although the degree of
increase was higher in PPARa knockout mice. Ito et al. (2007) suggested that increases in
oxidative stress induced by DEHP exposure may lead to induction of inflammation, resulting in a
higher incidence of liver tumors in PPARa knockout mice and a potential PPARa-independent
pathway for DEHP-induced liver tumors. The NRC (2008) report entitled Phthalates and
Cumulative Risk Assessment: The Tasks Ahead states that the Ito et al. (2007) results "suggest
that DEHP might cause hepatic cancer in rodents through a mechanism that is independent of
PPARa, as has been suggested by others (see, for example, Takashima et al. [2008])." A
separate NRC (2009) report entitled Science and Decisions: Advancing Risk Assessment states
that the Ito et al. (2007) study "calls into question" the conclusion regarding DEHP
carcinogenicity that is based on the PPARa activation MOA. Similar conclusions were reached
by NRC (2010) in their review of EPA's external review draft of tetrachloroethylene.
Another possibility is the involvement of other nuclear receptors. A comprehensive
evaluation of the ability of DEHP to activate gene expression through nuclear receptors other
than PPARa (Ren et al., 2010) demonstrated that exposure to DEHP activated multiple nuclear
108 DRAFT - DO NOT CITE OR QUOTE
-------
receptors, including PPARa, constitutive activated/androstane receptor (CAR), and pregnane X
receptor, in the rodent liver. Although direct evidence for TCA is not available, studies have
shown that other PPARa agonists, such as Wy-14,643, perfluorooctanoic acid (PFOA),
ciprofibrate, and clofibrate, regulate gene expression via CAR (Guo et al., 2007; Cheng and
Klaassen, 2008). It should be noted that DEHP induced liver tumors in rats and mice, but TCA
induced liver tumors only in mice. Therefore, the MOA for hepatocarcinogenesis for DEHP and
TCA may not be comparable. However, the above findings for DEHP suggest that
demonstration of many of the key events proposed for a PPARa MOA are insufficient to
preclude existence of a PPARa-independent pathway for tumorigenesis. Previously, Melnick
(2001) suggested PPARa-independent pathways for tumorigenesis by DEHP.
let-7C miRNA mediated signaling cascade. Researchers have explored other possible key
events for a PPARa agonism MOA, including the possible roles of let-7C miRNA on
hepatocarcinogenesis of PPARa agonists in mice. The expression of c-myc mRNA was
increased in TCA-treated female B6C3Fi mice (Pereira et al., 2001). c-myc has been
demonstrated to be a critical downstream effector of let-7C (Shah et al., 2007). Thus, increased
expression of c-myc mRNA in TCA-treated mice is consistent with the proposed let-7C miRNA
mediated signaling cascade in alteration of gene expression, hepatocellular proliferation, and
tumorigenesis in TCA-treated mice. However, it has not been shown that TCA-induced
increases in c-myc expression are PPARa-dependent, since increased expression of c-myc is
common to both carcinogens and noncarcinogenic mitogens (Hasmall et al., 1997).
Differences in species response to hepatocarcinogenicity. While TCA induces
peroxisome proliferation (a marker for PPARa agonism) in both rats and mice, to date, TCA has
been shown to be tumorigenic in B6C3Fi mice but not F344 rats (DeAngelo et al., 1997) (the
only strains tested for carcinogenicity). No complete explanation for this species difference has
been developed, although the NRC (2006) suggested that, at the same doses, rats and mice have
different responsiveness to peroxisome proliferation. For instance, Bull (2000) noted that, under
similar dosing regimens, a two- to threefold increase in peroxisome proliferation was observed in
F344 rats compared with a 10-fold increase over controls in mice (strains not specified).
However, this relationship may not hold for all mouse and rat species and strains and may be
chemical specific. For example, Elcombe (1985) reported that Wistar rats displayed a higher
induction of peroxisome proliferation than mice in response to TCA, as measured by increases in
cyanide insensitive palmitoyl CoA oxidation in both species. Moreover, evidence from other
peroxisome proliferators suggests that the degree of peroxisome proliferation and
hepatocarcinogenic potency are not well correlated (Marsman et al., 1988). The finding that
hepatocyte proliferation only occurred in TCA-treated mice (DeAngelo et al., 2008) but not in
treated rats (DeAngelo et al., 1997), is consistent with it being a key event in tumorigenesis
under the PPARa agonism MOA. However, it still does not provide an explanation as to the
109 DRAFT - DO NOT CITE OR QUOTE
-------
species difference, given that the prototypical PPARa agonist Wy-14,643 is hepatocarcinogic in
both rodent species.
Another possible explanation for the lack of TCA-induced tumors in rats is that the
binding of TCA to total plasma protein may be higher in rats than in mice, reducing its
bioavailability in the liver. However, the extent of these differences in binding is not clear. For
instance, at around 600 jiM, Lumpkin et al. (2003) reported the plasma-bound fraction of TCA in
rats to be about four- to fivefold higher than in mice, while Templin et al. (1995, 1993) reported
this difference to be only about 1.1-fold.
Therefore, overall, the lack of explanation for the absence of liver tumors in TCA-treated
rats that demonstrate peroxisome proliferation raises some questions about PPARa agonism as
the only MOA for liver tumor induction by TCA.
Phenotypic characteristics of tumors. TCA has also been associated with a PPARa-
mediated MOA based on evidence that the phenotypic characteristics of TCA-induced tumors
appear similar to those of tumors induced by other peroxisome proliferators (NRC, 2006).
However, on closer examination, certain characteristics of TCA-induced foci and tumors,
including mutation frequencies and spectra, phenotypic characteristics, and immunostaining
characteristics, are different from those induced by other peroxisome proliferators, and those
characteristics that are similar may be relatively nonspecific to peroxisome proliferators. This
suggests that PPARa agonism may not be the sole MOA for TCA-induced tumors in mice.
Specifically, with respect to mutations in TCA-induced foci and tumors, both Ferreira-
Gonzalez et al. (1995) and Bull et al. (2002) observed that the H-ras codon 61 mutation
frequency and spectrum of TCA-induced tumors were similar to historical controls, while
peroxisome proliferators ciprofibrate (Hegi et al., 1993) and methylclofenapate (Stanley et al.,
1994) have lower H-ras codon 61 mutation frequency than do spontaneous tumors in B6C3Fi
mice (11/46 versus 85/130 for methylclofenapate; 8/39 versus 32/50 for ciprofibrate) and their
mutation spectrums differed from those of spontaneous tumors. The lower frequency and
distinct pattern of H-ras mutation observed in methylclofenapate and ciprofibrate would suggest
that the activation of H-ras protooncogene in spontaneous liver lesions is not involved in
hepatocarcinogenesis by these two peroxisome proliferators. Since the H-ras codon 61 mutation
frequency and the spectrum of TCA-induced tumors were similar to historical controls, a similar
conclusion as to the role of H-ras activation cannot be drawn for TCA-induced tumors. On the
other hand, Ferreira-Gonzalez et al. (1995) reported K-ras codon 61 mutations in 1/11
TCA-induced liver tumors and none in 32 spontaneous tumors from control animals. Both Hegi
et al. (1993) and Stanley et al. (1994) found such a rare mutation in 1/23 ciprofibrate-induced
and one methylclofenapate-induced hepatocarcinoma (the number of samples examined was not
provided), suggesting that such a rare mutation may be caused by indirect DNA damage induced
by treatment (Hegi et al., 1993). Reynolds et al. (1987) reported K-ras mutations in mouse liver
tumors induced by the peroxisome proliferators furfural and furan, but the mutations were not at
110 DRAFT - DO NOT CITE OR QUOTE
-------
codon 61. However, it should be noted that, in all cases, the overall rates of K-ras mutations are
low (<10% of tumors), so their reliability as indicators of MO A is likely to be low.
With respect to tumor phenotype13, although Stauber and Bull (1997) reported
TCA-induced foci and tumors to be predominantly basophilic, Pereira (1996) reported that the
foci of altered hepatocytes in mice treated with TCA were half basophilic and half eosinophilic,
with liver tumors predominantly basophilic. DeAngelo et al. (2008) reported that cytoplasmic
alterations in hepatocytes of TCA-treated mice were characterized by intense eosinophilic
cytoplasm with deep basophilic granularity (microsomes). By contrast, it has been suggested
that peroxisome proliferators selectively promote basophilic foci generally (Cattley et al., 1995).
Furthermore, Weber et al. (1988) and Bannasch et al. (2001) reported that foci of altered
hepatocytes in rats treated with peroxisome proliferators are amphophilic-basophilic
(amphophilic: increased granular acidophilia and randomly scattered cytoplasmic basophilia),
suggesting a phenotype that also has increased mitochondrial proliferation and peroxisome
proliferation. Thus, the phenotype of TCA hepatic preneoplastic lesions may be different than
that induced by peroxisome proliferators.
Kraupp-Grasl et al. (1991, 1990) noted a difference in the ability of a peroxisome
proliferator to promote tigroid foci, which are characterized by large basophilic bodies on a clear
or eosinophilic cytoplasmic background, and weakly basophilic foci, which are characterized by
weak diffuse basophilia and some eosinophilia (equivalent to amphophilic foci described earlier).
In their experiments, using PB or the peroxisome proliferator nafenopin as promoters, only
nafenopin and not PB promoted the weakly basophilic foci. In addition, a substantial number of
spontaneous foci (the number of which were actually decreased by nafenopin) were tigroid.
Both tigroid and weakly basophilic foci may appear to be basophilic at the light microscopic
level; thus, it is not clear from Stauber and Bull (1997) and Pereira (1996) whether the reported
"basophilic" foci from TCA treatment are actually "tigroid" or "weakly basophilic." Moreover,
because of the natural progression of several lineages of preneoplastic lesions, including those
not induced by peroxisome proliferators, to basophilic neoplasms (Bannasch, 1996), basophilic
tumors themselves are nonspecific to peroxisome proliferators.
13According to the extensive published literature (Bannasch et al., 2001; Bannasch, 1996; Weber et al., 1988),
altered hepatic foci in hepatocarcinogenesis generally fall into three types: (1) glycogenotic-basophilic lineage:
glycogenotic clear and acidophilic (smooth endoplasmic reticulum-rich) hepatocytes that progress to glycogen-poor,
homogenously basophilic (ribosome rich) phenotype in undifferentiated hepatocellular carcinomas; (2) tigroid-
basophilic lineage: tigroid foci, a variant of glycogenotic foci (probably occurring at low dose), contain large
basophilic bodies on a clear or eosinophilic cytoplasmic background; (3) amphophilic-basophilic cell lineage:
amphophilic cells consist of glycogen-poor cytoplasm containing both abundant granular-acidophilic (mitochondria
and peroxisomes) and basophilic (ribosomes) components. Amphophilic cells occur when rats are treated with
nongenotoxic peroxisome proliferators. All three types of foci can progress to a basophilic phenotype as tumors
progress.
111 DRAFT - DO NOT CITE OR QUOTE
-------
Immunostaining characteristics. With respect to immunostaining characteristics, the foci
and tumors induced by peroxisome proliferators have been noted to not express GOT and GST-u
(Rao et al., 1986). It has been shown by Parnell et al. (1988) that TCA promotes GGT-positive
foci in partially hepatectomized rats initiated with DEN, which is the opposite of that expected
for peroxisome proliferators. (However, it is not known if TCA promotes GGT-positive foci in
rats that were not partially hepatectomized.) With respect to GST-u, Pereira and Phelps (1996),
Pereira et al. (1997), and Latendresse and Pereira (1997) found most tumors in their initiation-
promotion studies of MNU+TCA to be lacking in GST-u, consistent with that expected from
peroxisome proliferators. However, basophilic foci that are both GGT negative and GST-u
negative are not specific to peroxisome proliferators. For instance, Kraupp-Grasl et al. (1990)
and Grasl-Kraupp et al. (1993) reported that tigroid foci, which display basophilia, were
predominantly GGT negative regardless of whether they were found in control rats or rats given
AfBl only, AfBl plus the peroxisome proliferator nafenopin, or AfBl plus the non-peroxisome
proliferator PB. Ittrich et al. (2003) stated that GST-u is negative in preneoplastic and neoplastic
cell populations with increased basophilic components.
With respect to immunostaining characteristic for c-jun, Stauber and Bull (1997)
suggested that their observation that all TCA-induced tumors were c-jun negative, a
characteristic also found by Bull et al. (2002), was consistent with peroxisome proliferators.
However, tumors promoted by TCA in the experiments of Latendresse and Pereira (1997)
variably stained for c-jun. Furthermore, although spontaneous and some chemically-induced
foci and tumors have been reported to express or stain for c-jun (Sakai et al., 1995; Nakano et al.,
1994; Suzuki et al., 1990), both induction (Tharappel et al., 2003) and suppression (Yokoyama et
al., 1993) of c-jun by short-term exposure to peroxisome proliferators have been reported in the
liver or in vitro, with no studies located that report c-jun immunostaining of peroxisome
proliferator-induced foci or tumors. Therefore, the use of immunostaining characteristic for
c-jun as an indicator for the PPARa MOA is questionable.
Summary. In summary, proposed key events in the hypothesized PPARa agonism MOA
have been shown to occur with TCA treatment, including PPARa activation and hepatocellular
proliferation. The available data are insufficient, however, to confirm the PPARa MOA as a sole
causative factor for TCA hepatocarcinogenesis. Studies of PPARa published since NRC (2006)
indicate that the TCA MOA is more complex than that presented in NRC (2006). Specifically, a
study by Yang et al. (2007) showed that ligand-independent PPARa activation in hepatocytes
evokes the MOA, but not hepatocarcinogenesis in a transgenic mouse model. In addition, while
other data associated PPARa agonism with DEHP hepatocarcinogenesis, a second recent study
found that DEHP induces liver tumors in PPARa-null mice (Ito et al., 2007). Together, these
studies demonstrate that PPARa activation is neither sufficient for carcinogenesis nor necessary
for DEHP-induced liver tumors. While prior reviews (e.g., Klaunig et al. [2003]) have proposed
that PPARa agonism and its sequelae constitute an MOA for hepatocarcinogenesis as a sole
112 DRAFT - DO NOT CITE OR QUOTE
-------
causative factor, these newer data have raised considerable doubt about the validity of this
hypothesis for DEHP14. In addition, effects of TCA, including increased c-myc expression and
hypomethylation of DNA, are not specific to the PPARa activation MO A, and other data also
contribute uncertainty as to whether a PPARa-independent MOA explains TCA-induced tumors
in mice.
4.7.3.1.3. Dose-response concordance. Clear dose-response concordance between proposed
key events and tumor response is lacking. The doses that induce peroxisome proliferation in
mice are similar to tumorigenic doses of TCA (Bull, 2000). B6C3Fi and other strains of mice
treated with 1-5 g/L TCA in drinking water for 14 days showed dose-dependent increases in
hepatic peroxisomal enzyme carnitine acetyl-CoA transferase activity and cyanide-insensitive
PCO activity (DeAngelo et al., 1989). Dose-dependent increases in relative liver weights were
also observed. Similarly, dose-related increases in hepatic cyanide-insensitive AGO activity and
12-hydroxylation of lauric acid were observed in male B6C3Fi mice treated with 0.1-2 g/L TCA
in drinking water for 3 or 10 weeks.
Peroxisome proliferation was evaluated in only one chronic bioassay in mice (DeAngelo
et al., 2008). PCO activity was increased in mice treated with 0.5 g/L (68 mg/kg-day) or 5 g/L
(602 mg/kg-day) of TCA, dose levels that were carcinogenic, providing support that PPARa
agonism is related to tumor formation. As stated above, however, peroxisome proliferation is an
associative event and marker of PPARa agonism and not correlated with carcinogenic potency of
PPARa agonists.
The doses that induce hepatocellular proliferation in mice corresponded to tumorigenic
doses of TCA in DeAngelo et al. (2008). An increase in incidence of hepatocellular adenomas
and carcinomas was observed in male B6C3Fi mice exposed to 0.5 or 5 g/L TCA for 30-
60 weeks but not at 0.05 g/L TCA. A significant increase in hepatocellular proliferation was
found in mice exposed to 5 g/L TCA at 30 and 45 weeks and in 0.5 g/L TCA group at 60 weeks.
A small increase in hepatocyte proliferation was found in the 0.05 g/L TCA group at 78 weeks.
Doses of 0.3-3.3 g/L TCA that caused hepatocellular proliferation in short-term studies (Pereira,
1996; Sanchez and Bull, 1990) were similar to the tumorigenic doses.
4.7.3.1.4. Human relevance. In its framework for making conclusions about human relevance,
EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) outline the following
elements to evaluate: (1) identifying critical similarities and differences between test animals
14The NRC (2008) report entitled Phthalates and Cumulative Risk Assessment: The Tasks Ahead states that the Ito et
al. (2007) results "suggest that DEHP might cause hepatic cancer in rodents through a mechanism that is
independent of PPARa, as has been suggested by others (see, for example, Takashima et al. [2008])." A separate
NRC (2009) report entitled Science and Decisions: Advancing Risk Assessment states that the Ito et al. (2007) study
"calls into question" the conclusion regarding DEHP carcinogenicity that is based on the PPARa activation MOA.
113 DRAFT - DO NOT CITE OR QUOTE
-------
and humans regarding the sequence of key precursor events; (2) flagging quantitative differences
for consideration in dose-response assessment, such as the potential for different internal doses
of the active agent or differential occurrence of a key precursor event; and (3) considering all
populations and life stages, including special attention to whether tumors can arise from
childhood exposure.
With respect to the first element, it was originally believed that there is no evidence for
functional differences between rodents and humans in the key events described above for the
proposed PPARa MO A, and humans possess PPARa at sufficient levels to mediate the human
hypolipidemic response to peroxisome-proliferating fibrate drugs (Klaunig et al., 2003). Klaunig
et al. (2003) reached a conclusion (reiterated in NRC [2006]) that the key events are plausible in
humans in the sense that "a point in the rat/mouse key events cascade where the pathway is
biologically precluded in humans cannot be identified, in principle." This was supported by an
early in vitro study (Maloney and Waxman, 1999), in which the human and mouse forms of
PPARa are comparable in their affinity for TCA. The results from in vitro studies should be
interpreted cautiously because cultured human hepatocytes could lose or gain biological
characteristics in the process of immortalization, and the microenvironment of cultured cells is
different from the in situ hepatocytes in terms of the three dimensional cell-cell contact, cell
heterogenicity, and endocrine feedback in the intact animal. Recent studies suggested that there
might be functional differences between human and mouse PPARa. Studies on PFOA and
ammonium perfluorooctanoate showed that a lower concentration of PFOA and ammonium
perfluorooctanoate was required to activate mouse PPARa than to activate human PPARa
(Nakamura et al., 2009; Takacs and Abbott, 2007). The activation of mouse PPARa by PFOA
and ammonium perfluorooctanoate was generally higher compared to that of human PPARa
(Wolf et al., 2008). Altogether, human PPARa may have a weaker affinity for PFOA than does
mouse PPARa. No direct evidence exists to show that this is also true for TCA.
With respect to the second question, the limited available data suggest that there are
quantitative differences between rodents and humans in the occurrence of events following
PPARa activation. However, these data do not appear sufficient for use in dose-response
analysis. Walgren et al. (2000) found that TCA did not increase palmitoyl CoA oxidation and
caused a decrease in DNA synthesis in primary and long-term human hepatocytes cultures (in
contrast to rodents). Palmer et al. (1998) and Holden and Tugwood (1999) reported about
10-fold less PPARa mRNA in human liver as compared with rat or mouse, but mRNA levels are
not necessarily indicative of protein levels. Walgren et al. (2000) found, on average, lower
levels of PPARa protein in human livers compared with rodents, but expression levels were
highly variable among individuals and, at least in one case, were comparable to rodents' levels.
Moreover, expression levels may not be related to potency, since the hypolipidemic response to
PPARa agonists is similar in humans and rodents. On the other hand, humans and nonhuman
primates appear less sensitive than rodents to the PPARa-mediated peroxisome proliferation
114 DRAFT - DO NOT CITE OR QUOTE
-------
response and its associated changes in regulation of peroxisomal genes and proteins. None of
these effects, however, are thought to be causally related to hepatocarcinogenesis (Klaunig et al.,
2003), and it appears that carcinogenic potency and degree of peroxisomal response are not well
correlated (Marsman et al., 1988).
Lack of induction of cell proliferation or increased apoptosis have been observed in vitro
with human hepatocytes, but no method for quantitative extrapolation of these results to the in
vivo situation is available. Moreover, these assay systems remove the non-parenchymal cells
(e.g., Kupffer cells) during preparation, which has been shown to prevent the proliferative
response to PPARa agonists (Parzefall et al., 2001; Hasmall et al., 2000a). In vivo, no increase
in cell proliferation was observed in nonhuman primates treated with PPARa agonists (Doull et
al., 1999), but no human data are available. Hoivik et al. (2004) noted that fenofibrate and
ciprofibrate induced treatment-related increases in liver weight, hypertrophy, numbers of
peroxisomes, and numbers of mitochondria and smooth endoplasmic reticulum in cynomolgus
monkeys at 15 days of exposure; however, no cell proliferation was found.
While the observed species differences in the occurrence of key events may be explained
partially by differences in expression levels of PPARa in liver, recent studies (Shah et al., 2007;
Morimura et al., 2006; Cheung et al., 2004) using PPARa-humanized (hPPARa) mice fed
Wy-14,643 examine the hypothesis that structural differences in human and mouse PPARa
receptors may be responsible. A hPPARa mouse line in which the human PPARa was expressed
in liver under control of the tetracycline responsive regulatory system was used in these studies.
The hPPARa mice were fed the prototype peroxisome proliferator, Wy-14,643, or the lipid-
lowering drug, fenofibrate. Decreased serum triglycerides were observed in both the wild-type
and hPPARa mice, with no difference in basal serum triglyceride levels between the two types of
mice. In addition, a robust induction of the expression of genes encoding enzymes involved in
peroxisomal, mitochondrial, and microsomal fatty acid catabolism and those involved in fatty
acid synthesis and transport was found in hPPARa mice after 2 weeks of Wy-14,643 or
fenofibrate feeding. Hepatomegaly and increases in hepatocyte size were observed in mice fed
Wy-14,643 for 2 weeks; however, the extent of cell size and hepatomegaly was markedly less in
hPPARa mice when compared with wild-type mice, especially after 8 weeks of Wy-14,643
feeding.
Cheung et al. (2004) also evaluated peroxisome-proliferator-induced replicative DNA
synthesis by measuring BrdU incorporation into hepatocyte nuclei in hPPARa mice and wild-
type mice after 8 weeks of feeding with Wy-14,643. In wild-type mouse livers, Wy-14,643
treatment resulted in a BrdU labeling index of 57.9% compared with 1.6% in untreated controls.
However, in hPPARa mice, Wy-14,643 treatment did not increase the incorporation of BrdU
with average labeling indices of 2.8 and 1.6% in Wy-14,643-treated and control mice,
respectively. In addition, Wy-14,643 treatment resulted in a marked induction in the expression
of various genes involved in cell cycle control (proliferating cell nuclear antigen, c-myc, CDK1,
115 DRAFT - DO NOT CITE OR QUOTE
-------
and CDK4 and cyclins A2, Dl, and E) in the livers of wild-type mice. By contrast, the
expression of these genes was unchanged with Wy-14,643 treatment in hPPARa mice.
However, the lack of induction of these cell cycle regulated genes in hPPARa mice may be due
to differences in binding of activated hPPARa to mouse co-activators or to certain mouse
peroxisome proliferator response elements. On the other hand, genes encoding peroxisomal,
mitochondrial, and microsomal fatty acid oxidation enzymes were still markedly induced in
hPPARa mice following 8 weeks of Wy-14,643 feeding. Therefore, whereas human PPARa in
mice regulates induction of fatty acid catabolism and lipid lowering, it does not stimulate the cell
proliferative response that is thought to contribute to liver carcinogenesis. In addition, as
discussed above, Shah et al. (2007) reported that miRNA let-7C was not suppressed in
Wy-14,643-treated hPPARa mice. Wy-14,643 treatment of hPPARa mice also did not induce
c-myc and mir-17 expression.
Decreased susceptibility of hPPARa mice to Wy-14,643-induced liver tumorigenesis was
shown by Morimura et al. (2006). When the feeding study of 0.1% Wy-14,643 was extended to
44 weeks for hPPARa mice and 38 weeks for wild-type mice, the incidence of liver tumors,
including hepatocellular carcinoma, was 71% in wild-type mice (five adenomas and two
carcinomas out of seven mice; 3/10 treated mice died of toxicity); by contrast, only 5% of
Wy-14,643-treated hPPARa mice developed liver tumors (one adenoma out of 20 mice; the
adenoma resembled spontaneous tumor). In addition, up-regulation of cell cycle regulated
genes, such as cyclin Dl (cdl) and cyclin-dependent kinases (Cdks) 1 and 4, were observed in
non-tumorous liver tissues of Wy-14,643-treated wild-type mice. The c-myc mRNA was also
significantly overexpressed in the Wy-14,643-treated wild-type mice. On the other hand,
expression of the tumor suppressor gene, p53, was increased only in the livers of
Wy-14,643-treated hPPARa mice.
Morimura et al. (2006) concluded that these data in hPPARa mice are consistent with
toxicodynamic differences between humans and mice being due to structural differences between
human and mouse PPARa. It should be noted, however, that only Wy-14,643 has been tested in
hPPARa mice for carcinogenicity to date, and the duration of treatment was <1 year. Therefore,
more studies need to be conducted, especially with TCA, before definitive conclusions can be
made regarding human relevance using data from hPPARa mice.
As discussed previously, toxicokinetic differences also exist between humans and mice.
Binding of TCA to plasma proteins was found to be higher in humans than in mice in two
in vitro studies (Lumpkin et al., 2003; Templin et al., 1995). Thus, plasma levels of free TCA
would be expected to be lower in humans than in mice administered the same dose of TCA.
However, the extent to which administration of the same dose in humans and mice would yield a
relatively smaller tissue dose in humans is not directly related to plasma binding, due to
differences in clearance between species (humans having a longer half-life). Due to lack of the
116 DRAFT - DO NOT CITE OR QUOTE
-------
PBPK model, the extent to which such toxicokinetic differences would impact species
differences has not been quantified.
With respect to the final question, little data on population variability and lifestages,
particularly with respect to childhood exposures and susceptibility, are available either for TCA
or PPARa agonists in general.
A number of other reports have also reached conclusions as to the human relevance of
PPARa-agonist-induced hepatocarcinogenesis, both in general and with respect to specific
chemicals. The NRC (2006) report reiterated the position of Klaunig et al. (2003) that
"[w]hereas the mode of action is plausible in humans, the weight of evidence suggests that this
mode of action is not likely to occur in humans based on differences in several key steps when
taking into consideration kinetic and dynamic factors." NRC (2006) also stated "[i]nduction of
peroxisome proliferation in human liver is not a prominent feature; therefore, this key event
related to trichloroacetic acid liver carcinogenesis is not likely to occur in humans." In the
framework for MOA used here (U.S. EPA, 2005a), human relevance is considered in the context
of hazard characterization. As discussed above, both humans and rodents share the ability for
PPARa receptor activation but with similarities and differences in a number of responses. In
addition, in this analysis (U.S. EPA, 2005a), quantitative differences due to "kinetic and dynamic
factors" are flagged for consideration in dose-response assessment. Toxicokinetics of TCA are
discussed earlier in this document. With respect to toxicodynamics, as discussed above, data
suitable for use in dose-response analysis of TCA hepatocarcinogenic risk are lacking.
In the Science Advisory Board's review of EPA's draft risk assessment of potential
human health effects associated with PFOA and its salts (U.S. EPA, 2006c), it was concluded
that PFOA-induced liver tumors in rats were considered relevant to humans based on the
following considerations: (1) "uncertainties still exist as to whether PPARa agonism constitutes
the sole mode of action for perfluorooctanoic acid effects on liver"; (2) "[uncertainties exist
with respect to the relevance to exposed fetuses, infants and children of the PPARa agonism
mode of action for induction of liver tumors in adults"; and (3) "the interplay between PPARa
agonism and Kupffer cells (resident macrophages in the liver) has not been characterized.
Kupffer cells do not express PPARa, but are activated by peroxisome proliferators. Prevention
of Kupffer cell activation by glycine inhibited, although not completely, the development of liver
tumors by the potent peroxisome proliferator Wy-14,643. There are no data available on the
effects of peroxisome proliferators on human Kupffer cells." These conclusions regarding
human relevance are similar to those reached here for TCA.
4.7.3.1.5. Summary. The data for TCA, while supportive of the involvement of PPARa in
hepatocarcinogenesis, are not sufficient to conclude that it is the sole MOA. Moreover, issues
with respect to biological consistency and specificity of association for this proposed MOA have
been identified. Thus, the current data do not rule out the possibility that TCA could induce
117 DRAFT - DO NOT CITE OR QUOTE
-------
cancer in humans by a MOA not associated with PPARa agonism. To the extent that PPARa is
involved, the key events in the proposed MOA by Klaunig et al. (2003) are biologically plausible
in humans, so this MOA would be considered relevant to humans. On the other hand, data are
consistent with toxicokinetic and toxicodynamic differences between species in the responses to
the prototypical PPARa agonists, Wy-14,643, but data are lacking for TCA specifically. The
available data on such differences are not suitable for use in dose-response analysis of TCA
hepatocarcinogenic risk. Further studies with various types of PPARa agonists need to be
conducted before definitive conclusions can be drawn regarding the relative human sensitivity to
the hepatocarcinogenic effects of PPARa agonists.
4.7.3.2. Additional Proposed Hypotheses and Key Events with Limited Evidence or
Inadequate Experimental Support
Several effects that been hypothesized to be associated with liver cancer induction are
discussed in more detail below, including Kupffer cell activation, DNA hypomethylation,
decreased intercellular communication, and genotoxicity.
4.7.3.2.1. Kupffer cell activation: release ofcytokines and oxidants. The hypothesis is that
Kuppfer cell activation plays a critical role in hepatocarcinogenesis. This MOA entails the
following key events leading to TCA-induced liver tumor formation: following activation of
Kupffer cells, oxidants and cytokines are released; the resultant oxidative stress and cytokines
advance acquisition of the multiple critical traits contributing to carcinogenesis.
The liver consists of the hepatic parenchyma (hepatocytes) and non-parenchymal cells,
including sinusoidal endothelial cells, Ito cells, and Kupffer cells. Kupffer cells are dedicated
hepatic macrophages. Investigation of the role of non-parenchymal cells in mediating
hepatocarcinogenesis has focused mainly on Kupffer cells, which have been proposed as
important mediators of cell proliferation by tumor promoters (Hasmall et al., 2000b; Rusyn et al.,
1998; Rose et al., 1997). The role of Kupffer cell activation in the induction of a proliferative
response has been documented for peroxisome proliferators more generally, although evidence
specific to TCA is limited.
Progress has been made in understanding the involvement of non-parenchymal cells,
specifically Kupffer cells (i.e., liver-specific macrophages), in peroxisome-proliferator-induced
liver tumors, though many questions remain. Yang et al. (2007) suggested that activation of non-
parenchymal cells, which is independent of PPARa activation, plays an important role in
peroxisome-proliferator-induced hepatocarcinogenesis. Specifically, induction of proliferation
of non-parenchymal cells was observed in wild-type mice upon Wy-14,643 treatment, but not in
transgenic mice. Yang et al. (2007) suggested that lack of tumor induction in transgenic mice as
compared to Wy-14,643-treated wild-type mice may be associated with the differences of non-
parenchymal cell activation. To examine the role of Kupffer-cell-derived oxidants in the MOA
118 DRAFT - DO NOT CITE OR QUOTE
-------
for liver carcinogenesis, Woods et al. (2007a) treated NADPH-oxidase-deficient mice (their
Kupffer cells cannot produce oxidants), along with wild-type and PPARa knockout mice, with
Wy-14,643 for 1 week, 5 weeks, or 5 months. Wy-14,643 treatment induced similar levels of
hepatocyte proliferation and DNA damage in NADPH-oxidase-deficient and wild-type mice,
while both were abolished in PPARa knockout mice. By contrast, evidence of suppressed
apoptosis by Wy-14,643 was absent in both NADPH-oxidase-deficient and PPARa knockout
mice. Thus, NADPH oxidase was not required for chronic proliferative response or DNA
damage, although it played a role in the suppression of apoptosis along with PPARa. Woods et
al. (2007a) concluded that Kupffer-cell-derived oxidants may play a limited, if any, role in long-
term effects of peroxisome proliferators, such as hepatocarcinogenesis.
Activation of Kupffer cells by toxic agents can result in the release a wide range of
biologically active products, including reactive oxygen and nitrogen species, cytokines (Decker,
1990), such as TNF-a and ILs, proteases, and lipid metabolites, such as prostaglandins and
thromboxane. The mediators released from Kupffer cells can initiate a variety of downstream
events that may initially stimulate survival and protection but with continued or higher dose
exposure, may ultimately contribute to hepatic injury. In particular, TNF-a has been linked to
the stimulation of hepatocellular growth by tumor promoting compounds (Hasmall et al., 2000b).
Roberts et al. (2007) hypothesized that activation of Kupffer cells caused the release of cellular
growth regulatory signaling molecules that resulted in an increase in the proliferation of
hepatocytes; this is expected to be transient in normal hepatocytes but sustained in preneoplastic,
initiated hepatocytes, ultimately resulting in selective clonal expansion of the preneoplastic
hepatocytes (i.e., hepatic tumor promotion).
Activation of Kupffer cells by peroxisome proliferators is PPARa independent (Peters et
al., 2000), involves generation of reactive oxygen species, and leads to production of mitogenic
cytokines (Rusyn et al., 2000). Peroxisome proliferators appear to directly activate Kupffer cells
through mechanisms involving oxygen radicals, protein kinase C, and the transcription factor
nuclear factor-kappa B (NF-KB) (Rose et al., 1999).
Activation of Kupffer cells resulted in production of super oxide anion via NADPH
oxidase (Decker, 1990). It was suggested that Kupffer cell-derived oxidants play a role in
signaling rapid and robust increases in cell proliferation caused by peroxisome proliferators in
rodent liver via a mechanism that also involves activation of NF-KB and production of TNF-a
(Rose et al., 2000).
Recent studies (Woods et al., 2007b, c) have revealed that NADPH oxidase-dependent
events in the Kupffer cells in response to peroxisome proliferators (Wy-14,643 and DEHP) may
only be transient. As peroxisome proliferator treatment is continued, there appeared to be a shift
of the cellular source of the radicals, from Kupffer cells to hepatocytes. This is in line with the
study findings from Hassoun and colleagues (Hassoun et al., 2010a, b; Hassoun and Dey, 2008),
who showed an increase in biomarkers of phagocytic activation, including superoxide anion and
119 DRAFT - DO NOT CITE OR QUOTE
-------
lipid peroxidation, in mice exposed to TCA once and for 4 weeks, but not in mice exposed for
13 weeks (see Section 4.5.1.6).
In summary, Kupffer cells, the resident macrophages of the liver, are mediators of acute
phase responses to peroxisomal proliferators, including TCA. The release of cellular growth
regulatory signaling molecules and oxidants from Kupffer cells results in an increase in the
proliferation of hepatocytes, which may play a role in TCA-induced hepatocarcinogenesis.
4.7.3.2.2. Hypomethylation ofDNA. The hypothesis is that TCA induces hepatocarcinogenesis
via the induction of epigenetic changes, particularly DNA methylation. Key events in this MOA
comprise the induction of epigenetic alterations that advance acquisition of the multiple critical
traits contributing to carcinogenesis. Experimental evidence supports the hypothesis that
hypomethylation of DNA may be related to the carcinogenicity of TCA in mice. In female
B6C3Fi mice that received an i.p. injection of MNU and were then administered TCA in
drinking water at 25 mmol/L (4,085 mg/L) for 44 weeks, the level of 5MeC in the DNA of
hepatocellular adenomas and carcinomas was decreased by 40 and 51%, respectively, as
compared with noninvolved liver tissue from the same animal and control animals given only
MNU; termination of TCA treatment 1 week prior to sacrifice did not change the levels of 5MeC
in either adenomas or carcinomas (Tao et al., 1998). In another experiment, female B6C3Fi
mice treated with 25 mmol/L (1,062 mg/kg-day) TCA for 11 days in their drinking water also
showed a 60% decrease in the level of 5MeC in total liver DNA (Tao et al., 1998).
The decrease in the level of 5MeC in these studies indicated that many genes may be
hypomethylated. For example, Tao et al. (2000a) reported that the promoter regions of the c-jun
and c-myc genes were hypomethylated in the livers of mice exposed to 500 mg/kg-day TCA for
5 days. Expression of the mRNA and proteins of these two protooncogenes were increased.
This is in line with the studies by Latendresse and Pereira (1997) and Nelson et al. (1990), which
reported increased mRNA and proteins of c-jun and c-myc protooncogenes in TCA-induced foci
of altered hepatocytes and liver tumors. In another study (Tao et al., 2000b), the expression of
the mRNA and proteins of the two protooncogenes were found to be increased in MNU-initiated
and TCA-promoted mouse liver tumors. DNA methyltransferase activity was increased in
tumors and decreased in noninvolved liver tissue. Increased expression of c-jun and c-myc has
been associated with increased cell proliferation (Fausto and Webber, 1993; Saeter and Seglen,
1990). Therefore, increased expression and decreased methylation of the c-jun and c-myc genes
could be involved in the carcinogenic activity of TCA by facilitating cell proliferation.
The same group of scientists (Tao et al., 2004) also demonstrated that a region of the
IGF-II gene was hypomethylated in the livers of mice initiated with MNU and subsequently
exposed to TCA. [The IGF-II gene is growth-related and is associated with hepatic cell
proliferation (Furstenberger and Senn, 2002; Werner and Le Roith, 2000).] In TCA-exposed
mice, the percentage of cytosine-guanine dinucleotide sites that were methylated was reduced
120 DRAFT - DO NOT CITE OR QUOTE
-------
from 79.3 to 58% in noninvolved liver tissue and further reduced to 10.7% in liver tumors.
mRNA expression increased by 5.1-fold in liver tumors relative to noninvolved liver tissue from
mice treated with TCA.
An association between hypomethylation and cell proliferation in liver of TCA-treated
mice was demonstrated by Ge et al. (2001b). An increase in DNA replication (evidenced by
increased proliferating cell nuclear antigen labeling index and mitotic labeling index) was
observed 72 and 96 hours after the first gavage dose of 500 mg/kg-day TCA. Hypomethylation
of the internal cytosine of CCGG sites in the promoter region of the c-myc gene began between
48 and 72 hours from the initiation of treatment with TCA and continued to 96 hours.
These experimental findings suggest that TCA induces global and locus-specific DNA
hypomethylation in mouse liver. Given the recent finding discussed in Section 4.7.3.1.1.1 that
the DNA hypomethylation by the potent PPARa agonist, Wy-14,643, was PPARa-dependent
(Pogribny et al., 2007), but hypomethylation of DNA by DEHP-treated rats was not observed.,
hypomethylation of DNA may not be a key event in PPARa MO A. Moreover, hypomethylation
is a relatively ubiquitous phenomenon in carcinogenesis and it has not been demonstrated that
TCA-induced hypomethylation is PPARa-dependent. Therefore, the possibility of hypomethy-
lation of DNA as a PPARa-independent MO A cannot be discounted.
4.7.3.2.3. Decreased intercellular communication. Inhibition of intercellular communication
has been identified as a contributor to tumor induction by some peroxisome proliferators
(Klaunig et al., 2003, 1988). However, similar inhibition has been reported with nongenotoxic
liver carcinogens that are not peroxisome proliferators. Thus, this hypothesized MOA is not
specific to peroxisome proliferators and PPARa agonism. This MOA has not been well
characterized with respect to the component key events.
From a physiological perspective, the formation of gap junctions with short half-lives in
cell membranes can be considered a regulatory control factor for tumor formation (Benane et al.,
1996). Transfer of molecules from neighboring normal cells to transformed cells via
intercellular communication allows growth suppression of transformed cells. Blocking
intercellular communication on a repetitive basis releases the "initiated" cells from the growth
control constraint exerted by neighboring cells and facilitates tumor formation. Studies by
Benane et al. (1996) and Klaunig et al. (1989) (see Section 4.5.1) suggest that TCA-induced
inhibition of gap junction intercellular communication could potentially play a role in regulation
of cell differentiation, growth and homeostasis, and tumor promotion.
4.7.3.2.4. Genotoxicity. A hypothesized mutagenic MOA entails the following key events
leading to TCA-induced liver tumor formation: TCA alters the genetic material in a manner that
causes changes to be transmitted during cell division through one or more mechanisms (gene
mutations, deletions, translocations, or amplification). TCA has been tested for genotoxicity in a
121 DRAFT - DO NOT CITE OR QUOTE
-------
variety of in vitro and in vivo assays as described in Section 4.5.2. Most, but not all, studies
(Kargalioglu et al., 2002; Nelson et al., 2001; DeMarini et al., 1994; Rapson et al., 1980) report
negative results for mutagenicity in S. typhimurium in the absence of cytotoxicity. Mutagenicity
in mouse lymphoma cells was only induced at cytotoxic concentrations (Harrington-Brock et al.,
1998). Both positive and negative responses have been observed in vivo. TCA-induced DNA
strand breaks and chromosome damage were observed in the liver in several studies (Giller et al.,
1997; Nelson and Bull, 1988; Bhunya and Behera, 1987) and were suggested by the results of
Harrington-Brock et al. (1998), although these effects have not been uniformly reported (Chang
et al., 1992; Styles et al., 1991). However, some evidence indicates that TCA-induced
chromosome damage assayed in vitro and in vivo may be secondary to pH changes rather than a
direct effect of TCA (Mackay et al., 1995), underscoring the need to carefully evaluate assay
conditions.
In other studies of potential genotoxicity, DNA-repair responses to TCA in bacterial
systems have been inconsistent, with induction of DNA repair reported in S. typhimurium (Ono
et al., 1991) but not in E. coli (Giller et al., 1997). TCA induced oxidative DNA damage in the
livers of mice following a single dose (Austin et al., 1996) but not following repeated dosing
over 3 or 10 weeks (Parrish et al., 1996), possibly suggesting either effective DNA repair and/or
adaptation to repeated TCA exposures. Ferreira-Gonzalez et al. (1995) found that the mutation
frequency and mutation spectrum in the H-ras gene were similar in tumors from control and
TCA-treated mice, suggesting that TCA was not inducing tumors through direct DNA damage at
this locus. The pattern of TCA-induced tumors in mice does not support a mutagenic MO A.
Tumors were observed only in livers of TCA-exposed mice and no tumors were found in
TCA-treated rats.
In summary, there is some evidence that TCA is weakly mutagenic. Therefore, the
hypothesis that mutagenicity contributes to the MOA for TCA-induced liver tumors cannot be
ruled out.
4.7.3.3. Conclusions About the Hypothesized Mode of Action
In summary, TCA is carcinogenic in mice (DeAngelo et al., 2008; Bull et al., 2004, 2002,
1990; Bull, 2000; Pereira, 1996). Studies of the mechanism by which TCA induces liver tumors
reveal that the MOA for TCA is complex and that TCA may induce tumors by multiple MOAs
that may not be mutually exclusive. While PPARa-related events represent some of the major
components of the overall mechanism of toxicity and carcinogenicity, it is premature to conclude
that this is the only MOA for TCA-induced carcinogenicity. In addition, in light of new
evidence that challenges the hypothesis that PPARa is absolutely required for hepatocarcino-
genesis of peroxisome proliferators in mice (Ito et al., 2007), the strength of this linkage becomes
more uncertain. Tumor induction by TCA appears to involve perturbation of cell growth,
reduced intercellular communication, release of cytokines and oxidants by activated Kupffer
122 DRAFT - DO NOT CITE OR QUOTE
-------
cells, and hypomethylation of DNA. The data do not support a major role for a mutagenic MOA
(Bull, 2000; Moore and Harrington-Brock, 2000).
4.8. SUSCEPTIBLE POPULATION AND LIFE STAGES
4.8.1. Possible Childhood Susceptibility
Age-dependent differences in susceptibility to TCA have not been investigated in
systemic toxicity studies. The dose spacing in the available developmental toxicity studies
(Table 4-10) is inadequate to determine the relative fetal and maternal toxicity of TCA. The
LOAELs for developmental toxicity range from 291 mg/kg-day (Johnson et al., 1998) to
1,000 mg/kg-day (Singh, 2005a). Most developmental LOAELs occurred at maternally toxic
doses. Therefore, these developmental toxicity data are too limited to draw any conclusions on
whether developing organisms might be a sensitive subpopulation. In subchronic toxicity
studies, LOAEL and NOAEL values of 355 and 36.5 mg/kg-day, respectively, were observed in
male rats exposed to TCA in drinking water for 90 days (Mather et al., 1990). In the Parrish et
al. (1996) 10-week drinking water study with male mice, the LOAEL and NOAEL values were
125 and 50 mg/kg-day, respectively. The LOAELs observed in the subchronic toxicity studies
suggest that systemic effects are observed at doses similar to, or less than, those at which
developmental toxicity has been observed; however, no developmental NOAELs are available
for comparison with the subchronic systemic NOAELs. Given the lack of a developmental
NOAEL, it is uncertain what dose would be protective for developmental toxicity.
The data are also insufficient to determine whether there are age-dependent differences in
the toxicokinetics (e.g., plasma binding and metabolism) of TCA that might lead to differences
in health risk. There are no published comparative data for plasma binding of TCA in young and
old animals. The enzymes responsible for the metabolism of TCA have not been conclusively
identified. Even in the cases where relevant metabolizing enzymes have been identified, no
information on age-dependent changes in the expression or activity of these enzymes has been
identified. The health implications of any differences between children and adults in metabolic
capacity are also difficult to determine for the haloacetic acids, since the toxic form of each
compound has not been identified. The mechanisms involved in haloacetic acid toxicity are not
sufficiently understood to make this determination. The preliminary results of Hunter and
Rogers (1999) in whole embryo culture suggest that, at least for developmental effects, the
parent compound may be involved in the toxicity of MCA, while for TCA, a metabolite may be
involved. However, in vitro studies such as whole embryo culture have limited utility for
predicting the developmental toxicity of chemical agents in intact organisms and are considered
to be useful only for hypothesis generation not for hypothesis testing. Further in vivo studies are
needed to determine whether there are age-related differences in susceptibility to toxic effects of
TCA.
123 DRAFT - DO NOT CITE OR QUOTE
-------
The cancer potency of TCA in very young animals has been investigated in a mouse
neonatal cancer assay (von Tungeln et al., 2002). In this study, neonatal male and female
B6C3Fi mice were given i.p. injections of TCA in dimethyl sulfoxide at 1,000 or 2,000 nmol
(total dose, which corresponds to approximately 16 or 32 mg/kg) in split doses delivered at 8 and
15 days of age. The test animals were sacrificed and evaluated for liver tumors at 12 (high dose)
or 20 (low dose) months of age. The incidence of hepatic tumors in TCA-treated animals did not
differ significantly from tumor incidences observed in the solvent controls.
4.8.2. Possible Gender Differences
No data directly relevant to the evaluation of the effects of gender on TCA toxicity in
humans were located. The available animal data, although limited, suggest that males may be
more sensitive to the carcinogenicity of TCA than females. Only one cancer bioassay was
located that concurrently exposed both male and female mice to TCA (Bull et al., 1990)
(described in Section 4.2). In this study, male and female B6C3Fi mice were exposed to TCA in
the drinking water at concentrations that resulted in doses of up to approximately 329 mg/kg-day
for 52 weeks. A clear dose-related increase in animals with proliferative lesions (hyperplastic
nodules, adenomas, or carcinomas) was observed in males (incidence of up to 19/24, which
occurred at 329 mg/kg-day). In contrast, the incidence of proliferative lesions in females was not
increased (data not reported). Although no other studies were available that evaluated the
carcinogenicity of TCA in males and females concurrently, the available single-sex cancer
bioassays conducted in separate laboratories also suggest that males may be more sensitive than
females to TCA carcinogenicity. For example, Pereira et al. (2001) (described in Section 4.2)
observed a tumor incidence of 25% in female B6C3Fi mice exposed to TCA in the drinking
water at a dose of 784 mg/kg-day for 51 weeks. In contrast, tumor incidences ranging from 55 to
83% have been reported in males exposed to lower TCA doses (309-480 mg/kg-day) in the
drinking water for a comparable duration (Bull, 2000; Bull et al., 1990). These data indicate that
TCA is a more potent carcinogen in male than in female mice.
Although males appear to be more sensitive than females to carcinogenicity of TCA, the
available data suggest that males and females are about equally sensitive to noncancer effects
induced by TCA. For example, Bull et al. (1990) observed that the type and magnitude of the
noncancer liver effects induced by TCA were similar in male and female B6C3Fi mice exposed
to TCA in the drinking water at comparable doses for 52 weeks. Davis (1990) did not observe
marked differences in the susceptibility of males and females to TCA-induced noncancer effects
in a short-term toxicity study. Although both of these studies were limited by the scope of
toxicological parameters evaluated, they suggest that male and female animals are similar in their
sensitivity to TCA-induced noncancer effects.
124 DRAFT - DO NOT CITE OR QUOTE
-------
4.8.3. Other
Limited information was identified regarding other factors (e.g., genetic polymorphisms,
enzyme deficiencies, or altered health states) that might influence susceptibility to TCA. Some
data are available for DCA and may be relevant to TCA. Several genetic polymorphisms have
been identified in GST-(^, a key enzyme involved in DCA metabolism. As noted previously, it is
unclear whether TCA is metabolized to DCA (Bull, 2000; Lash et al., 2000); these
polymorphisms would be relevant to TCA susceptibility only if DCA is a metabolite of TCA.
As noted previously, TCA induces glycogen accumulation. Kato-Weinstein et al. (1998)
suggested that prolonged glycogen accumulation can become irreversible. These data suggest
that individuals with glycogen storage disease (an inherited deficiency or alteration in any one of
the enzymes involved in glycogen metabolism) constitute another group that may be more
susceptible to TCA toxicity.
No quantitative evaluation has been conducted on the health impact of environmental
exposures for individuals harboring polymorphisms in genes related to glycogen storage or
antioxidant response. In each of these cases, a significant background load of the stressor may
be present; thus, the excess risk associated with low doses of TCA is not clear.
125 DRAFT - DO NOT CITE OR QUOTE
-------
5. DOSE-RESPONSE ASSESSMENTS
5.1. ORAL REFERENCE DOSE (RfD)
The RfD15 for TCA was derived through a three-step process: (1) evaluating all toxicity
studies and selecting the critical effects from these studies that occur at the lowest dose;
(2) selecting the dose or point of departure16 (POD) at which the critical effect either is not
observed or would be predicted to occur at a relatively low incidence (e.g., 10%); and
(3) dividing this POD by uncertainty factors (UFs) to reflect uncertainties in extrapolating from
study conditions to conditions of human environmental exposure.
5.1.1. Choice of Principal Study and Critical Effect—Rationale and Justification
Chronic, subchronic, and developmental animal toxicity studies considered for derivation
of the oral RfD are summarized in Table 5-1. Two of the available chronic oral drinking water
studies (DeAngelo et al., 2008, 1997) were identified as potential candidates from which to
derive the RfD. The study in rats by DeAngelo et al. (1997) identified a NOAEL of 32.5 mg/kg-
day and a LOAEL of 364 mg/kg-day based on significantly decreased body weight, a statistically
significant and dose-related increase in serum ALT activity, and histopathologic changes in the
liver. The study in mice by DeAngelo et al. (2008) identified a NOAEL of 8 mg/kg-day and a
LOAEL of 68 mg/kg-day for increased liver weight, liver peroxisome proliferation, hepatic
necrosis, and testicular tubular degeneration. Histopathologic examinations were conducted on
organs other than the liver in both DeAngelo et al. (1997) and DeAngelo et al. (2008); other
chronic mouse studies have only evaluated the liver. In a cancer study in mice by Pereira (1996),
only a limited number of endpoints were evaluated, but a higher NOAEL of 78 mg/kg-day for
liver effects was identified. Two other chronic-duration drinking water studies (Bull et al., 1990;
Herren-Freund et al., 1987) were not further considered for derivation of the RfD because they
examined only a limited number of endpoints in the liver and used higher administered doses
than those employed by DeAngelo et al. (2008, 1997).
15The RfD is an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the
human population (including sensitive subgroups) that is likely to be without appreciable risk of deleterious effects
during a lifetime. It can be derived from a NOAEL, LOAEL, or benchmark dose (BMD), with UFs generally
applied to reflect limitations of the data used. The RfD is expressed in terms of mg/kg-day of exposure to an agent.
16The POD denotes a dose at the lower end of the observed dose-response curve where extrapolation to lower doses
begins. For effects other than cancer, the POD is either a NOAEL, a LOAEL if no NOAEL can be identified, or a
modeled point (for example, a 95% lower bound on exposure dose or concentration at 10% extra risk) if the data are
suitable for dose-response modeling.
126 DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Candidate studies for derivation of the RfD for TCA
Reference
Species
Exposure
route
Exposure
duration
Doses evaluated
(mg/kg-d)
Observed effects
NOAEL
(mg/kg-d)
LOAEL
(mg/kg-d)
Comments
Chronic studies
DeAngelo
et al., 1997
DeAngelo
et al., 2008
Pereira,
1996
Bull et al.,
1990
Herren-
Freund et
al., 1987
F344 rats
(males,
50/group)
B6C3FJ mice
(males,
50/group)
B6C3FJ mice
(females, 38-
134/group)
B6C3FJ mice
(11-35/sex
and dose)
B6C3FJ mice
(males, 22-
33/group)
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
Oral,
drinking
water
104 wks
60wks
51 or 82 wks
(A) 52 wks
(B) 37 wks +
15-wk
recovery
61 wks
0,3.6, 32.5, or
364
0,8,68, or 602
0, 78, 262, or 784
(A) 0, 164, or 329
(B) 0 or 309
0, 500, or 1,250
Decreased body weight,
increased serum ALT activity;
increased peroxisome
proliferation
Increase in liver weight,
increase in liver peroxisome
proliferation, hepatic necrosis,
testicular tubular degeneration
Increased relative liver weight
Increased absolute and relative
liver weight, cytomegaly,
glycogen accumulation
Increased absolute and relative
liver weight
32.5
8
78
Not
determined
Not
determined
364
68
262
164
500
Time-weighted average daily
doses were calculated by the
authors; a comprehensive set
of tissues was
microscopically examined.
Time-weighted average daily
doses were calculated by the
authors; a comprehensive set
of tissues was
microscopically examined for
the control and high-dose
groups.
Increased liver weight was
observed after 82 wks at
262 mg/kg-d; 262 mg/kg-d
was judged to be an
equivocal LOAEL in the
absence of other measures of
liver toxicity.
Only the liver and kidneys
were evaluated; dose was
estimated by the authors.
Only the liver was
microscopically examined.
127
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Candidate studies for derivation of the RfD for TCA
Reference
Species
Exposure
route
Exposure
duration
Doses evaluated
(mg/kg-d)
Observed effects
NOAEL
(mg/kg-d)
LOAEL
(mg/kg-d)
Comments
Subchronic studies
Mather et
al., 1990
Bhatetal.,
1991
Sprague-
Dawley rats
(males,
10/dose)
Sprague-
Dawley rats
(males,
5/group)
Oral,
drinking
water
Oral,
drinking
water
90 d
90 d
0,4.1, 36.5, or
355
0 or 825
Decreased absolute spleen
weight; increased relative liver
and kidney weights; peroxisome
proliferation
Decreased body weight gain;
minor changes in liver
morphology; collagen
deposition; perivascular
inflammation of the lungs
36.5
Not
determined
355
825
1/4 of the LD50 (3,300 mg/kg)
was administered daily.
Developmental studies
Smith et al.,
1989
Long-Evans
rats (20-
2 I/dose)
Oral,
gavage
CDs 6-15
0, 330, 800, 1,200,
or 1,800
Maternal: decreased body
weight; increased spleen and
kidney weights
Developmental:
decreased fetal weight,
decreased crown-rump length,
increased incidence of soft-
tissue and cardiovascular
malformations (mainly
levocardia); increased maternal
spleen and kidney weights
Maternal:
Not
determined
Develop-
mental:
Not
determined
Maternal:
330
Develop-
mental:
330
Critical study for 1994 RfD.
The developmental LOAEL
was also a maternal LOAEL.
Cardiovascular
malformations were not
confirmed by Fisher et al.
(2001).
128
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Candidate studies for derivation of the RfD for TCA
Reference
Fisher et
al., 2001
Johnson et
al. (1998)
Species
Sprague-
Dawley rats
(19/dose)
Sprague-
Dawley rats
(55 controls
and 11 TCA
treated rats)
Exposure
route
Oral,
gavage
Drinking
water
Exposure
duration
CDs 6-15
CDs 1-22
Doses evaluated
(mg/kg-d)
0 or 300
0 or 291
Observed effects
Maternal: decreased body
weight gain on GDs 7-15 and
18-21; decreased uterine weight
Developmental: Decreased
fetal body weight (per litter and
per fetus)
Maternal: decreased body
weight
Developmental: Increase in
cardiac malformations; increase
in number of implantation
sites/litter, number of resorption
sites/litter, and total resorptions
NOAEL
(mg/kg-d)
Maternal:
Not
determined
Develop-
mental: Not
determined
Maternal:
None
Develop-
mental: None
LOAEL
(mg/kg-d)
Maternal:
300
Develop-
mental: 300
Maternal: 291
Develop-
mental: 291
Comments
A limited number of fetal
endpoints were evaluated,
including sex, fetal weight,
and incidence of heart
malformations.
Dose estimated by the
authors, based on the average
amount of water consumed
by the animals on a daily
basis.
Study was not adequately
designed and/or reported, and
a complete array of standard
developmental endpoints was
not assessed.
129
DRAFT - DO NOT CITE OR QUOTE
-------
Subchronic toxicity data were available from studies conducted in rats by Mather et al.
(1990) and Bhat et al. (1991). The 90-day drinking water study by Mather et al. (1990)
established NOAEL and LOAEL values of 36.5 and 355 mg/kg-day, respectively, for effects on
relative liver and kidney weights and peroxisome proliferation. These values are similar to and
support the NOAEL and LOAEL values obtained for hepatic effects in the chronic study of
DeAngelo et al. (1997) in rats. Bhat et al. (1991) observed decreased body weight gain, minor
changes in liver morphology, and inflammation of the lungs in rats administered a dose
equivalent to one-fourth of the LD50 of 3,300 mg/kg (or approximately 825 mg/kg-day).
Three developmental toxicity studies (Fisher et al., 2001; Johnson et al., 1998; Smith et
al., 1989) were also evaluated as potential candidates for use in the derivation of the RfD. Smith
et al. (1989) identified a developmental LOAEL of 330 mg/kg-day (the lowest dose tested) for
increased incidence of fetal cardiac malformations and significantly reduced fetal body weight
and crown-rump length in Long-Evans rats dosed by gavage on GDs 6-15. Johnson et al. (1998)
identified a developmental LOAEL of 291 mg/kg-day for fetal cardiac malformations in a single-
dose study where Sprague-Dawley rats were dosed via drinking water on GDs 1-22. Fisher et al.
(2001) observed decreased fetal body weight, but saw no evidence of cardiac malformations in a
single-dose study where Sprague-Dawley rats were dosed with 300 mg/kg-day by gavage on
GDs 6-15. Because of inconsistent findings of cardiac malformations (in particular levocardia)
across the three developmental toxicity studies and questions of interpretation raised by Smith et
al. (1989) (see Section 4.6.1.3), cardiac malformation was not considered a candidate critical
effect. Furthermore, because the incidence of total soft tissue (visceral) malformations as
reported by Smith et al. (1989) was attributable largely to the incidence of cardiac
malformations, total soft tissue malformations were similarly not considered a candidate critical
effect. Fisher et al. (2001) and Johnson et al. (1998) were single-dose studies, and as such,
provide less useful information for dose-response analysis than does Smith et al. (1989).
Johnson et al. (1998) also suffers from issues related to adequacy of reporting and limited
examination of endpoints. Therefore, Fisher et al. (2001) and Johnson et al. (1998) were not
considered as candidate principal studies.
The chronic drinking water study in mice by DeAngelo et al. (2008) was considered the
most appropriate choice among the available studies for derivation of the RfD. In this study, the
route of exposure was oral, both a LOAEL and NOAEL were identified for liver effects that
were both lower than the corresponding values identified in the chronic drinking water study in
rats (DeAngelo et al., 1997), and the data in this chronic mouse study were consistent with the
findings in both chronic drinking water studies in rats (DeAngelo et al., 1997; Mather et al.,
1990). In addition, complete histopathologic examinations were conducted for all organs for the
control and high-dose groups, whereas other studies in mice only evaluated the liver. Moreover,
the incidence data in DeAngelo et al. (2008) were amenable to benchmark dose (BMD)
modeling.
130 DRAFT - DO NOT CITE OR QUOTE
-------
Selected data sets from the developmental toxicity study conducted by Smith et al.
(1989), specifically data on mean fetal body weight and fetal crown-rump length, were analyzed
by BMD modeling for comparison with the candidate PODs derived for endpoints from the
DeAngelo et al. (2008) study.
5.1.2. Methods of Analysis—Including Models (e.g, PBPK and BMD)
5.1.2.1. BMD Modeling of Liver and Testicular Effects from DeAngelo et al. (2008)
BMD modeling was used to analyze liver and testicular effects in male mice exposed to
TCA in drinking water (DeAngelo et al., 2008). Incidence data for hepatocellular inflammation,
hepatocellular necrosis, and testicular tubular degeneration are summarized in Table 5-2 and
mean PCO activity (a marker of peroxisome proliferation) is summarized in Table 5-3. The
incidence of hepatocellular cytoplasmic alteration was elevated in TCA-exposed mice (see
Table 4-4); however, because the incidence deviated from a monotonic dose-response
relationship, this endpoint was not subject to BMD modeling.
Table 5-2. Incidence of nonneoplastic lesions in male B6C3Fi mice exposed
to TCA in drinking water for 60 weeks
Lesion
Hepatocellular inflammation
Hepatocellular necrosis
Testicular tubular degeneration
Control
3/30
0/10
2/30
0.05 g/L TCA
(8 mg/kg-d)a
0/27
0/10
0/27
0.5 g/L TCA
(68 mg/kg-d)a
2/29
3/10
4/29b
5 g/L TCA
(602 mg/kg-d)a
7/29b
5/10b
6/29b
"Time-weighted mean daily dose in mg/kg-day.
bStatistically significant from the control group,/) < 0.05.
Source: DeAngelo et al. (2008).
Table 5-3. Mean PCO activity in male B6C3Fi mice exposed to TCA in
drinking water for up to 60 weeks
Mean PCO activity (nmol NAD
reduced/min/mg protein)3
Control
2.59 ±1.04
0.05 g/L TCA
(8 mg/kg-d)b
2.85 ±0.86
0.5 g/L TCA
(68 mg/kg-d)b
4.75 ±1.16
5 g/L TCA
(602 mg/kg-d)b
11.99 ±3.04
aMean PCO activity ± SD was calculated as an arithmetic mean of the PCO activity for mice sacrificed at weeks 4,
15, 30, 45, and 60. PCO activity for each time point was based on five mice/group/time point. The total number of
mice for each concentration was 25 (with the exception of 24 mice for the 5 g/L TCA group).
bTime-weighted mean daily dose in mg/kg-day.
Source: DeAngelo et al. (2008) and email dated March 12 2010, from Anthony DeAngelo, NHEERL, ORD, U.S.
EPA, to Diana Wong, NCEA, ORD, U.S. EPA.
131
DRAFT - DO NOT CITE OR QUOTE
-------
All of the available dichotomous models in U.S. EPA's benchmark dose software
(BMDS) (version 1.4.1) were fit to incidence data for hepatocellular inflammation,
hepatocellular necrosis, and testicular tubular degeneration. Doses (i.e., benchmark dose
[BMDio] and 95% lower confidence limit on the BMD [BMDLio]) associated with a benchmark
response (BMR) of 10% extra risk were calculated. A BMR of 10% is generally used in the
absence of information regarding what level of change is considered biologically significant, and
also to facilitate a consistent basis of comparison across assessments (U.S. EPA, 2000b). All of
the continuous models in BMDS (version 2.1.1) were fit to mean PCO activity data. A BMR of
1 SD from the control mean was used to calculate the BMDiso and BMDLiso for mean PCO
activity. A BMR of 1 SD is generally used as the BMR for continuous data in the absence of
knowledge of what level of response to consider as biologically significant, and to facilitate a
consistent basis of comparison across assessments where continuous data are used (U.S. EPA,
2000b).
Details of the BMD modeling conducted for each data set from DeAngelo et al. (2008)
are provided in Appendix B. In general, model fit was assessed by a %2 goodness-of-fit test (i.e.,
models with/? < 0.1 failed to meet the goodness-of-fit criterion) and the Akaike's Information
Criterion (AIC) value (i.e., a measure of the deviance of the model fit that allows for comparison
across models for a particular endpoint). Of the models exhibiting adequate fit, the lowest
BMDL was selected as the POD when the BMDLs estimated from these models varied by more
than threefold; otherwise, the BMDL from the model with the lowest AIC was chosen. If more
than one model shared the lowest AIC, BMDLio values from these models were averaged to
obtain a POD (U.S. EPA, 2000b). The model results for the best fitting model for each data set
from DeAngelo et al. (2008) are summarized in Table 5-4.
132 DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-4. BMD modeling results for data sets from DeAngelo et al. (2008)
Endpoint
Hepatocellular inflammation in male
B6C3F! mice exposed to TCA in
drinking water for 60 wksa
Incidence of hepatocellular necrosis in
male B6C3F! mice exposed to TCA in
drinking water for 30-45 wksa
Incidence of testicular tubular
degeneration in male B6C3F! mice
exposed to TCA in drinking water for
60 wksa
Cyanide-insensitive PCO activity in
male B6C3F! mice exposed to TCA in
drinking water for up to 60 wksb
Best fitting model
Logistic and log-
probit
Log-logistic
Log-logistic
Polynomial (2°)
BMR
Extra risk
10%
Extra risk
10%
Extra risk
10%
1 SD
BMD
(mg/kg-d)
393.0C
40.7
298.2
28.4
BMDL
(mg/kg-d)
260.5C
17.9
127.4
21.1
aAll dichotomous dose-response models were fit using BMDS, version 1.4.1. The best-fit model is presented
here. See Appendix B for the results of all dichotomous models in BMDS.
bAll continuous dose-response models were fit using BMDS, version 2.1.1. The best-fit model is presented
here. See Appendix B for the results of all continuous models in BMDS.
'Because the logistic and log-probit models shared the lowest AIC value (i.e., 74.19), the BMD10 and
BMDL10 values from these two models were averaged.
For hepatocellular inflammation, the logistic, one-stage multistage, probit, and log-probit
models all exhibited adequate fit. Because the logistic and log-probit models shared the lowest
AIC value (i.e., 74.19), the BMDLio values from these two models were averaged to yield a
candidate POD of 260.5 mg/kg-day. Four of the seven dichotomous models in BMDS fit to the
incidence of hepatocellular necrosis in male mice exhibited adequate fit. These four models
were the gamma, log-logistic, one-stage multistage, and Weibull models. Among these four
models, the gamma, one-stage multistage, and Weibull models yielded identical fits, essentially
reducing the number of adequately fitting models to two. The log-logistic model yielded the
lowest AIC value (i.e., 30.42) of the two adequately fitting models. Thus, the BMDLio of
17.9 mg/kg-day estimated by the log-logistic model was selected as a candidate POD. All of the
models fit to the incidence of testicular tubular degeneration exhibited adequate fit. Of these
seven models, the gamma, one-stage multistage, and Weibull models yielded identical fits,
essentially reducing the number of adequately fitting models to five. The log-logistic model
yielded the lowest AIC (i.e., 76.08). Therefore, the BMDLio estimate of 127.4 mg/kg-day from
the log-logistic model was selected as a candidate POD. For mean PCO activity, only the
second-degree polynomial model of the four continuous models in BMDS showed adequate fit.
Thus, the BMDLiso of 21.1 mg/kg-day estimated by the second-degree polynomial model was
selected as a candidate POD for this endpoint.
Of the four endpoints evaluated by DeAngelo et al. (2008) for which dose-response
modeling was performed, hepatocellular necrosis was the most sensitive, as it yielded the lowest
133
DRAFT - DO NOT CITE OR QUOTE
-------
POD of 17.9 mg/kg-day. Therefore, 17.9 mg/kg-day was selected as a candidate POD for use in
derivation of the RfD.
5.1.2.2. BMD Modeling of Developmental Toxicity Data from Smith et al. (1989)
As discussed in Section 5.1.1, selected data from the developmental toxicity study
conducted by Smith et al. (1989)—fetal body weight and fetal crown-rump length—were
analyzed by BMD modeling for comparison with the POD derived from DeAngelo et al. (2008).
These data sets are summarized in Table 5-5.
Table 5-5. Dose-response data for developmental endpoint in TCA-treated
Long-Evans rats
Endpoint
Dose (mg/kg-d)
0
330
800
1,200
1,800
Mean fetal crown-rump length (cm)
Male
Female
3.71 ±0.12
3.64 ±0.15
3.58±0.10a
3.53±0.09a
3.46±0.10a
3.38±0.12a
3.36±0.15a
3.33±0.16a
3.16±0.12a
3.15±0.15a
Mean fetal body weight (g)
Male
Female
3.70 ±0.24
3.54 ±0.20
3.20±0.26a
3.08±0.27a
2.98±0.17a
2.83±0.18a
2.74±0.30a
2.67 ± 0.29a
2.49±0.16a
2.36±0.15a
aMean is significantly different from control mean (p < 0.05) as reported by Smith et al. (1989).
Source: Smith et al. (1989), Table 4.
All of the continuous models in BMDS (version 2.1.1) provided by U.S. EPA's were fit
to the data for fetal body weight and fetal crown rump length data. Doses (i.e., BMD05 and
BMDLos) associated with a BMR of 5% extra risks were calculated. A BMR of 5% extra risk
was selected for developmental endpoints to assure protection of the sensitive developing fetus.
This selection is consistent with the EPA's BMD technical guidance (U.S. EPA, 2000b).
Details of the BMD modeling conducted for each data set from Smith et al. (1989) are
provided in Appendix C. As with the endpoints from DeAngelo et al. (2008), model fit was
assessed by a %2 goodness-of-fit test (i.e., models with/? < 0.1 failed to meet the goodness-of-fit
criterion) and the AIC value. The model results for the best fitting model for each data set from
Smith et al. (1989) are summarized in Table 5-6.
134
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-6. BMD modeling results for data sets from Smith et al. (1989)
Endpoint
Best fitting
model
BMR
BMD
(mg/kg-d)
BMDL
(mg/kg-d)
Fetal body weight
Male3
Female3
Hill
Hill
Relative
deviation 5%
121.4
126.5
84.0
87.7
Fetal crown-rump length
Male3
Female3
Exponential
model 2
Exponential
model 2
Relative
deviation 5%
600.7
650.9
534.4
562.9
3A11 continuous dose-response models were fit using BMDS, version 2.1.1. The best-fit model is presented
here. See Appendix C for the results of all continuous models in BMDS.
For body weight in male and female fetuses, two of the continuous models in BMDS
exhibited adequate fit—the exponential model 4 and the Hill model. For both male and female
data, the Hill model yielded the lowest AIC values (i.e., -151.07 and -158.80) of the adequately
fitting models. The BMDLos of 84.0 mg/kg-day estimated by the Hill model for male fetal body
weight was smaller than the BMDLos of 87.7 mg/kg-day for female fetal body weight and was
therefore selected as a candidate POD. It should be noted that this value is well below the lowest
tested dose of 330 mg/kg-day.
For fetal crown-rump length in male and female fetuses, all continuous models in BMDS
exhibited adequate fit. For both male and female fetal data, exponential model 2 had the lowest
AIC value (i.e., -273.53 and -250.59, respectively). The BMDL05 of 534.4 mg/kg-day estimated
by exponential model 2 for male fetal crown-rump length was smaller than the BMDLos of
562.9 mg/kg-day for females and was therefore selected as a candidate POD.
5.1.2.3. Selection of POD
Comparison of the candidate PODs based on endpoints from the DeAngelo et al. (2008)
study with PODs based on developmental endpoints reported by Smith et al. (1989) reveal that
the liver endpoints are more sensitive than testicular or developmental endpoints. Therefore, the
BMDLio of 17.9 mg/kg-day based on hepatocellular necrosis was selected as the POD for use in
deriving the TCA RfD.
135
DRAFT - DO NOT CITE OR QUOTE
-------
5.1.3. RfD Derivation—Including Application of Uncertainty Factors (UFs)
The chronic mouse drinking water study by DeAngelo et al. (2008) was selected as the
principal study for derivation of the oral RfD as discussed in Section 5.1.1. The RfD for TCA is
calculated by using the POD based on the incidence of hepatocellular necrosis identified in the
principal study as follows:
RfD = POD -H UF
= 17.9 mg/kg-day - 1,000
= 0.0179 mg/kg-day, rounded to 0.02 mg/kg-day
Where 17.9 mg/kg-day = POD for the incidence of hepatocellular necrosis in mice
exposed to TCA via drinking water for 30-45 weeks (DeAngelo et al., 2008) and 1,000 =
composite UF chosen to account for extrapolation from animals to humans, interindividual
variability in humans, and deficiencies in the database. The UFs used in the calculation of the
RfD were selected for the following reasons:
• Human variation. An UF of 10 was selected for interindividual variability to account for
human-to-human variability in susceptibility in the absence of quantitative information to
assess the toxicokinetics and toxicodynamics of TCA in humans.
• Animal-to-human extrapolation. An UF of 10 was selected for interspecies extrapolation
to account for uncertainty in extrapolating from laboratory animals to humans (i.e.,
interspecies variability) because information was unavailable to quantitatively assess
toxicokinetic or toxicodynamic differences between animals and humans for TCA.
• Database deficiences. An UF of 10 was used to account for database deficiences. There
are no TCA-specific systemic toxicity data in humans. Although subchronic and chronic
animal studies of TCA have been conducted in rats and mice, most studies have focused
primarily or exclusively on liver lesions and have not examined other organs for
microscopic lesions. DeAngelo et al. (2008) is the only study in mice that included
histopathological examination of organs other than the liver; however, complete
histopathologic examinations were performed on only five mice from the high-dose and
control groups. Other data gaps include lack of a multigeneration reproductive toxicity
study. Available developmental studies were conducted at high doses, and did not allow
identification of a NOAEL
• Subchronic-to-chronic extrapolation. An UF for study duration was not required in this
assessment because the principal study was of chronic duration.
• LOAEL-to-NOAEL extrapolation. An UF for LOAEL-to-NOAEL extrapolation was not
applied because the current approach is to address this factor as one of the considerations
in selecting a BMR for BMD modeling. In this case, a BMR of 10% increase in the
incidence of hepatocellular necrosis was selected under an assumption that it represents a
minimally biologically significant change.
136 DRAFT - DO NOT CITE OR QUOTE
-------
5.1.4. RfD Comparison Information
The RfD derived using liver as an endpoint (specifically hepatocellular necrosis) based
on data from the DeAngelo et al. (2008) mouse study was compared with potential reference
values that would result from the use of alternative critical effects in target organs other than the
liver, specifically testicular effects identified in the rat (DeAngelo et al., 2008) and
developmental effects (decreased fetal body weight) in the rat (Smith et al., 1989). The potential
reference values derived from these studies are presented in Figure 5-1.
1000
100 -
-S 10
60
60
O
Q
1 --
0.1 -
0.01 -I
POD
JHuman variability
inimal-to-human
^Database deficiencies
A Reference value
Liver (hepatocellular Testes (testicular tubular Developmental (decreased
necrosis, mouse; DeAngelo degeneration, mouse; fetal body weight, rat;
etal., 2008) DeAngelo et al., 2008) Smith etal., 1989)
Figure 5-1. PODs (mg/kg-day) with corresponding potential oral reference
values that would result if alternative endpoints were used as the critical
effect.
5.1.5. Previous RfD Assessment
The previous IRIS assessment for TCA did not provide an RfD.
137
DRAFT - DO NOT CITE OR QUOTE
-------
5.2. INHALATION REFERENCE CONCENTRATION (RfC)
No inhalation studies adequate for the derivation of an RfC were located. The
respiratory tract has not been examined in oral studies of TCA. Because the liver is the critical
target organ for oral toxicity and first-pass effect by the liver is expected following oral
administration, the route of exposure may influence the hepatic response to TCA. PBPK models
that would support route-to-route extrapolation for TCA have not been published. Thus, the
available information is inadequate for extrapolation of oral toxicity data to the inhalation
pathway. For these reasons, an RfC for TCA was not derived.
5.3. UNCERTAINTIES IN THE RfD
The following discussion identifies uncertainties associated with the RfD for TCA. As
presented in Section 5.1.3, the UF approach, following EPA methodology for RfD development
(U.S. EPA, 2002), was applied to a POD. For the RfD, the POD was determined as the BMDLio
for hepatocellular necrosis in treated mice. Factors accounting for uncertainties associated with a
number of steps in the analyses were adopted to account for extrapolating the POD, the starting
point in the analysis, to a diverse population of varying susceptibilities. These extrapolations are
carried out with default approaches instead of from data on TCA, given the limited experimental
TCA data to inform individual steps.
Selection of principal study and critical effect for reference value determination. The
selected principal study (DeAngelo et al., 2008) was the most complete study in mice, with a
well-defined NOAEL/LOAEL and data that were amenable to dose-response modeling.
Complete histopathologic examination was conducted for the high-dose and control groups,
although examination was limited to only five mice from each group, and female mice were not
studied. Most subchronic and chronic animal studies of TCA conducted in rats and mice have
focused primarily or exclusively on liver lesions and, other than DeAngelo et al. (2008), have not
examined other organs for microscopic lesions. Nevertheless, liver toxicity appears to be the
most consistent and most sensitive effect in rats and mice. Liver toxicity, specifically
hepatocellular necrosis, was selected as the critical effect for the RfD. The uncertainty
associated with the relevance of this effect to humans is therefore considered low.
Animal-to-human extrapolation. No human exposure studies are available for derivation
of the RfD. For derivation of the RfD, extrapolating dose-response data from animals to humans
is a source of uncertainty. Uncertainties pertaining to unknown interspecies differences in
toxicokinetics and toxicodynamics were addressed by application of an UF of 10.
17The RfC is an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the
human population (including sensitive subgroups) that is likely to be without appreciable risk of deleterious effects
during a lifetime. It can be derived from a NOAEL, LOAEL, or benchmark concentration, with UFs generally
applied to reflect limitations of the data used.
13 8 DRAFT - DO NOT CITE OR QUOTE
-------
Dose-response modeling. BMD modeling was used to estimate the POD for the RfD.
While models with better biological support may exist, the selected models provided adequate
mathematical fits to the experimental data sets. BMD modeling has advantages over a POD
based on a NOAEL or LOAEL because the latter are a reflection of the particular exposure
concentration or dose at which a study was conducted, they lack characterization of the dose-
response curve, and they do not address the variability of the study population. NOAELs and
LOAELs also are less amenable to quantitative uncertainty analysis.
Interhuman variability. Heterogeneity among humans is another source of uncertainty.
Although male mice appear to be more sensitive than female mice to the carcinogenicity of TCA,
available data suggest that males and females are about equally sensitive to noncancer effects
induced by TCA. Limited information was identified regarding other factors (e.g., genetic
polymorphism) that might influence susceptibility to TCA (see Section 4.8.3). An UF of 10 was
used to account for interhuman variability. A factor of 10 was found to be generally sufficient to
account for human variability (Renwick and Lazarus, 1998).
5.4. CANCER ASSESSMENT
As discussed in Section 4.1.1, no epidemiologic studies currently exist that have
evaluated the carcinogenicity of TCA in humans. The carcinogenicity of TCA has been
evaluated, however, in several studies in both rats and mice. In mice, bioassay results provide
evidence that TCA is a complete carcinogen, as exposure to TCA in drinking water for periods
of 52-104 weeks significantly increased the incidence of liver tumors in male and female
B6C3Fi mice (DeAngelo et al., 2008; Bull et al., 2002, 1990; Pereira, 1996; Pereira and Phelps,
1996; Herren-Freund et al., 1987). In several of these studies, a clear monotonic dose-response
relationship was evident (DeAngelo et al., 2008; Bull et al., 2002, 1990; Pereira, 1996).
Moreover, the development of tumors in animals exposed to TCA progressed rapidly, as evident
from the appearance of significant numbers of tumors in several of the less-than-lifetime studies
(i.e., <82 weeks). Positive evidence for tumor promotion by TCA (following exposure to known
tumor initiators) has been reported for liver tumors in B6C3Fi mice (Pereira et al., 2001, 1997)
and for GGT-positive foci in livers of partially hepatectomized Sprague-Dawley rats (Parnell et
al., 1988). In contrast to the results observed in mice, TCA was not carcinogenic in a study of
male F344/N rats exposed via drinking water for 104 weeks (DeAngelo et al., 1997). The
carcinogenicity of TCA has not been evaluated in female rats or in other species of experimental
animals.
As discussed in Section 4.7.3, data from recent TCA studies that have investigated the
MOA for hepatocarcinogenesis do not support a direct mutagenic mechanism. Instead, tumor
induction appears to result from perturbation of cell growth and/or reduced intercellular
communication that is consistent with a PPARa-mediated MOA (NRC, 2006). Considerable
debate currently exists about the mechanism by which peroxisome proliferators cause liver
139 DRAFT - DO NOT CITE OR QUOTE
-------
tumors in rodents, and whether these chemicals represent a human cancer risk (NRC, 2006).
Other potential MO As (e.g., oxidative stress induced by activated Kupffer cells and
hypomethylation of DNA) may also play a role in TCA tumor induction.
One possible interpretation is that the carcinogenicity of TCA may involve multiple
MO As that are not mutually exclusive. While PPARa-related events represent some of the
major components of the overall mechanism of toxicity and carcinogenicity, certain
inconsistencies in the data exist. Unresolved issues for PPARa as a MOA for TCA-induced liver
tumors include: inconsistencies in experimental results across species, sex, and PPARa agonists;
lack of specificity of some proposed key events to PPARa; lack of clear dose concordance
between proposed key events and tumor response; and the finding that PPARa activation by
itself was insufficient to induce liver tumors (Yang et al., 2007). In addition, in light of recent
evidence that challenges the hypothesis that PPARa is absolutely required for
hepatocarcinogenesis of peroxisome proliferators in mice (Ito et al., 2007), the strengths of this
linkage become more uncertain. Based on these concerns, it is premature to conclude that
PPARa is the sole operative MOA for TCA-induced liver tumors.
Because the MOA for TCA-induced liver carcinogenesis has not been established, the
cancer dose-response modeling is carried out using linear extrapolation (U.S. EPA, 2005a). In
addition, no data were found that were suitable for accounting for interspecies differences in
toxicokinetics or toxicodynamics in dose-response modeling.
5.4.1. Choice of Study/Data—Rationale and Justification
Five bioassays in B6C3Fi mice exposed to TCA in drinking water were selected for
analysis and derivation of an oral slope factor for TCA because they: (1) included adequate
numbers of animals for statistical analyses; (2) showed statistically significant increased
incidences of liver tumors (i.e., combined incidences of adenomas and carcinomas) compared
with control values; and (3) included multiple TCA exposure levels to support characterization
of the dose-response relationship. These five bioassays consisted of two 52-week studies in male
mice (Bull et al., 2002, 1990), a 60-week study in male mice (DeAngelo et al., 2008), an
82-week study in female mice (Pereira, 1996), and a 104-week study in male mice (DeAngelo et
al., 2008).
5.4.2. Dose-Response Data
Dose-response data for the combined incidence of hepatocellular adenomas and
carcinomas from five bioassays of TCA are shown in Tables 5-7 through 5-11.
140 DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-7. Incidence of hepatocellular adenomas, carcinomas, or adenomas
and carcinomas combined in male B6C3Fi mice exposed to TCA in drinking
water for 52 weeks
TCA concentration
(g/L)
0
0.5
2
Estimated intake"
(mg/kg-d)
0
120
480
Human equivalent
lifetime doseb
(mg/kg-d)
0
2.3
9.0
Incidence of
adenomas
0/20
5/20
6/20
Incidence of
carcinomas
0/20
3/20
3/20
Incidence of
adenomas or
carcinomas0
0/20
6/20
8/20
"Doses were calculated using reference water intakes of 0.24 L/kg-day for male B6C3F! mice (U.S. EPA, 1988).
bSee Appendix D for conversion of mouse daily intakes to human equivalent lifetime doses.
°Bull et al. (2002) reported combined incidences of adenomas or carcinomas for each dose group.
Source: Bull et al. (2002).
Table 5-8. Incidence of hepatocellular adenomas, carcinomas, or adenomas
and carcinomas combined in male B6C3Fi mice exposed to TCA in drinking
water for 52 weeks
TCA concentration3
(g/L)
0
1
2
Estimated intake1"
(mg/kg-d)
0
164
329
Human equivalent
lifetime dosec
(mg/kg-d)
0
3.1
6.2
Incidence of
adenomas
0/35
2/11
1/24
Incidence of
carcinomas
0/35
2/11
4/24
Incidence of
adenomas or
carcinomas'1
0/35
4/11
5/24
aAn experimental design that included a control group and one dose group (2 g/L) using female mice was also part
of this study, but the data were deemed inadequate for modeling because a response at a single dose was considered
insufficient for properly characterizing a dose-response relationship.
bCalculated using total doses (g/kg) reported by Bull et al. (1990).
°See Appendix D for conversion of mouse daily intakes to human equivalent lifetime doses.
dBull et al. (1990) did not report combined incidences for adenomas and carcinomas, so this total assumes that each
animal had either adenomas or carcinomas, but not both.
Source: Bull etal. (1990).
141
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-9. Incidences of hepatocellular adenomas, carcinomas, or adenomas
and carcinomas combined in male B6C3Fi mice exposed to TCA in drinking
water for up to 60 weeks
TCA concentration
(g/L)
0
0.05
0.5
5
Estimated
intake"
(mg/kg-d)
0
7.7
68.2
602.1
Human equivalent
lifetime doseb
(mg/kg-d)
0
0.2
2.0
17.4
Incidence of
adenomas0
2/30
1/32
7/34
13/34
Incidence of
carcinomas0
2/30
5/32
6/34
13/34
Incidence of
adenomas or
carcinomas0
4/30
5/32
12/34
19/34
""Estimated daily intakes were calculated with the mean measured TCA concentrations reported by DeAngelo et al.
(2008) where available; if not, the nominal concentration for the dose group was used (see Appendix D, Table D-l
for details).
bSee Appendix D, Table D-l for conversion of mouse estimated daily intake to human equivalent lifetime dose.
Individual animal data were obtained from the study author (email dated April 26, 2010, from Anthony DeAngelo,
NHEERL, ORD, U.S. EPA, to Diana Wong, NCEA, ORD, U.S. EPA). Because the first liver tumor occurred at
45 weeks for 0.05, 0.5, 5 g/L dose groups, adenoma or carcinoma data for all mice examined histopathologically
between weeks 45-60 were included for those dose groups. For the control group, the first tumor occurred at
60 weeks, so adenoma or carcinoma data for all mice examined histopathologically on and after 52 weeks were
included.
Source: DeAngelo et al. (2008).
Table 5-10. Incidence of hepatocellular adenomas, carcinomas, or
adenomas and carcinomas combined in female B6C3Fi mice exposed to
TCA in drinking water for 82 weeks
TCA concentration
(mmol/L)
0
2
6.67
20
Estimated
intake"
(mg/kg-d)
0
78
262
784
Human equivalent
lifetime doseb
(mg/kg-d)
0
5.7
19.3
57.6
Incidence of
adenomas
2/90
4/53
3/27
7/18
Incidence of
carcinomas
2/90
0/53
5/27
5/18
Incidence of
adenomas or
carcinomas0
4/90
4/53
8/27
12/18
Intakes were calculated using reference water intake of 0.24 L/kg-day for female B6C3F! mice (U.S. EPA, 1988).
bSee Appendix D for conversion of mouse daily intakes to human equivalent lifetime doses.
°Pereira (1996) did not report combined incidences for adenomas and carcinomas, so this total assumes that each
animal had either adenomas or carcinomas, but not both.
Source: Pereira (1996).
142
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-11. Incidence of hepatocellular adenomas, carcinomas, or adenomas
and carcinomas combined in male B6C3Fi mice exposed to TCA in drinking
water for up to 104 weeks
TCA concentration
(g/L)
0
0.05
0.5
Estimated
intake"
(mg/kg-d)
0
6.7
81.2
Human equivalent
lifetime doseb
(mg/kg-d)
0
1
12.8
Incidence of
adenomas0
10/56
10/48
20/51
Incidence of
carcinomas0
26/56
15/48
32/51
Incidence of
adenomas or
carcinomas0
31/56
21/48
36/51
"Estimated daily intakes were calculated with the mean measured TCA concentrations reported by DeAngelo et al.
(2008) where available; if not, the nominal concentration for the dose group was used (see Appendix D, Table D-l
for details).
bSee Appendix D, Table D-2 for conversion of mouse estimated daily intake to human equivalent lifetime dose.
Individual animal data were obtained through the study author (email dated February 1, 2010, from Anthony
DeAngelo, NHEERL, ORD, U.S. EPA, to Diana Wong, NCEA, ORD, U.S. EPA). Because the first liver tumor
occurred at 52 weeks or after, adenoma or carcinoma data for all mice examined histopathologically between
weeks 52-104 were included.
Source: DeAngelo et al. (2008).
5.4.3. Dose Conversion
Before fitting the multistage model to the combined incidence data for adenomas and
carcinomas in Tables 5-7 through 5-11, estimated daily intakes of TCA from the mouse studies
were converted to human equivalent doses for continuous lifetime exposure using an interspecies
body weight scaling factor and continuous exposure time adjustment factors (see Appendix D for
the equations and calculations). The human equivalent lifetime doses used in the dose-response
modeling are shown in the third column of Tables 5-7 through 5-11.
5.4.4. Extrapolation Methods
As discussed in Section 4.7.3, studies investigating the MOA for TCA-induced liver
tumors do not provide strong evidence for genotoxicity (Bull, 2000; Moore and Harrington-
Brock, 2000). Rather, tumor induction appears to involve perturbation of cell growth through
activation of the PPARa pathway (Bull, 2000; Austin et al., 1996; Parrish et al., 1996) and
reduced intercellular communication (Benane et al., 1996). The existing evidence is not
sufficient, however, to determine which, if any, of these mechanisms are causally related to the
observed tumor responses. In addition, data to identify dose-response relationships for possible
precursor events for TCA-induced liver tumors are not available. Therefore, data from mouse
studies are too limited for the application of a biologically-based dose-response model.
Moreover, based on dose-response modeling, both Pereira (1996) and Bull et al. (1990)
concluded that the tumorigenic response of TCA exhibited a linear relationship with increasing
dose. Based on the uncertainty regarding the MOA, linear extrapolation from the BMDLio for
liver tumors was used for deriving an oral slope factor for TCA (U.S. EPA, 2005a).
143
DRAFT - DO NOT CITE OR QUOTE
-------
The multistage model in U.S. EPA's BMDS (version 2.1.1) was fit to liver tumor
incidence data for the five data sets described in Section 5.4.2. The multistage model has been
used by EPA in the vast majority of quantitative cancer assessments because it is thought to
reflect the multistage carcinogenic process. Furthermore, this model can accommodate a wide
variety of dose-response shapes and its use provides consistency with previous quantitative dose-
response assessments for cancer.
The multistage model was restricted to two stages or less for the 52-week Bull et al.
(2002, 1990) and the 104-week DeAngelo et al. (2008) data sets employing three dose groups
(including controls), and to three stages or less for the 82-week Pereira (1996) and the 60-week
DeAngelo et al. (2008) data sets employing four dose groups (including controls). For each of
the five data sets, the best-fit model was selected by comparing AIC values, as well as by
examining the visual fit of the model to the data. The BMDLio estimates from the best-fit
models were used as the POD for deriving the candidate oral cancer slope factors (Table 5-12).
Additional model details, including model outputs from BMDS, are provided in Appendix D.
Table 5-12. Candidate oral cancer slope factors derived from cancer
bioassays in B6C3Fi mice
Study reference
(study duration)
BMD10
(mg/kg-d)a
BMDL10
(mg/kg-d)a
X2 goodness-
of-fit
/7-value
Slope of linear
extrapolation
from BMD10b
(mg/kg-d)1
Oral cancer
slope factor0
(mg/kg-d)1
Male mice
Bull etal., 2002 (52 wks)
Bull etal, 1990 (52 wks)
DeAngelo etal., 2008
(60 wks) (Study 1)
DeAngelo etal., 2008
(104 wks) (Study 3)
1.34
1.87
2.67
5.71
0.89
1.13
1.67
1.50
0.17
0.12
0.22
0.23
7.5 x 1Q-2
5.3 x 10'2
3.7 x 10'2
1.8 x lO'2
1.1 x 1Q-1
8.8 x 10'2
6.0 x 10'2
6.7 x lO'2
Female mice
Pereira, 1996 (82 wks)
6.73
4.67
0.51
1.5 x 1Q-2
2.1 x 1Q-2
aBMD10 and BMDL10 were derived from the best-fit multistage model.
bThe slope of a linear extrapolation from the BMD10 is calculated as 0.1/BMD10.
°The oral cancer slope factor is derived by linear extrapolation from the BMDL10 (i.e., 0.1/BMDL10).
5.4.5. Time-to-tumor Modeling
Individual animal data (specifying when liver tumors were detected in each animal) for
the three bioassays conducted by DeAngelo et al. (2008) were obtained from the study author
(emails dated February 1 and April 26, 2010, from Anthony DeAngelo, NHEERL, ORD, U.S.
EPA, to Diana Wong, National Center for Environmental Assessment (NCEA), ORD, U.S.
EPA). The availability of individual animal data permitted the application of more sophisticated
144
DRAFT - DO NOT CITE OR QUOTE
-------
dose-response modeling approaches (i.e., time-to-tumor modeling) to estimate lifetime cancer
risks based on both the TCA dose and the liver tumor appearance time. These bioassays
included the 60- and 104-week studies considered in Section 5.4.2 and a third (104-week) study
that used only one dose group and a control.
Consideration was also given to whether the liver tumor incidence data from these three
bioassays could be combined to derive an oral cancer slope factor. A statistical analysis was
conducted employing a generalized likelihood ratio test (Stiteler et al., 1993), after both
individual and combined data sets were fitted by the multistage Weibull (MSW) time-to-tumor
model (U.S. EPA, 2009). This statistical analysis for data set compatibility is presented in
Appendix E. The analysis revealed that two liver tumor data sets from DeAngelo et al. (2008)
(i.e., the 60-week study and the multi-dose 104-week study) were statistically compatible to be
combined for MSW time-to-tumor modeling.
The results of the MSW time-to-tumor modeling for both individual and combined data
sets from DeAngelo et al. (2008) are presented in Appendix E and are summarized in Table 5-13.
For the individual studies, the cancer slope factors derived using the MSW time-to-tumor model
and those derived with the multistage model in BMDS were similar. In the case of the 60-week
study, the multistage model in BMDS yielded a cancer slope factor fivefold higher than the value
derived from the MSW time-to-tumor model. In the case of the 104-week study, the multistage
model in BMDS yielded a cancer slope factor 21% lower than the MSW time-to-tumor model.
Table 5-13. Candidate oral cancer slope factors derived from liver tumor
data sets in B6C3Fi male mice using MSW time-to-tumor modeling and
comparison to slope factors derive using the multistage model in BMDS
Study 1
Study 3
Study
1+3
Model
MSW time-to-tumor
(Stage 1)
BMDS multistage (Stage
1)
MSW time-to-tumor
(Stage 2)
BMDS multistage (Stage
2)
MSW time-to-tumor
(Stage 1)
AIC
158.9
149.0
226.4
210.0
381.0
BMR
0.1
0.1
0.1
0.1
0.1
BMD10a
13.5
2.7
5.0
5.7
2.2
BMDL10b
8.4
1.7
1.2
1.5
1.4
Slope of linear
extrapolation
from BMD10C
7.4 x 1Q-3
3.7 x 1Q-2
2.0 x 1Q-2
1.8 x 1Q-2
4.5 x 1Q-2
Cancer slope
factor from
BMDL10d
1.2 x 1Q-2
6.0 x 1Q-2
8.5 x 1Q-2
6.7 x 1Q-2
7.2 x 1Q-2
aBMD10 = dose at 10% cancer risk.
bBMDL10 = dose at 95% lower bound with 10% cancer risk.
°Slope of linear extrapolation from BMD10 = 0.1/BMD10.
dCancer slope factor = 0.1/BMDL10.
145
DRAFT - DO NOT CITE OR QUOTE
-------
The cancer slope factor derived from the combined data set was similar to the cancer
slope factors derived from the individual study data sets. As shown in Table 5-13, the cancer
slope factor for the combined data set (7.2 x 10"2 [mg/kg-day]"1) fell between the values based on
the individual study data sets (1.2 x 10"2 [mg/kg-day]"1 and 8.5 x 10"2 [mg/kg-day]"1). Also as
shown in Table 5-13, the cancer slope factors derived using the MSW time-to-tumor modeling
and the multistage model were similar, especially when applied to tumor incidence data from the
104-week DeAngelo et al. (2008) study (i.e., 8.5 x 10"2 [mg/kg-day]"1 and 6.7 x 10"2 [mg/kg-
day]"1).
For consistency with the dose-response analyses conducted for the tumor data sets from
Bull et al. (2002, 1990) and Pereira (1996) and because application of the MSW time-to-tumor
model to the DeAngelo et al. (2008) data yielded cancer slope factors similar to the multistage
model, further evaluation of candidate cancer slope factor derived for all five TC A tumor data
sets was based on model results using the multistage model.
5.4.6. Oral Cancer Slope Factor and Inhalation Unit Risk
The oral cancer slope factor is an upper-bound estimate of risk per increment of dose that
can be used to estimate lifetime cancer risk from different TCA exposure levels. The candidate
oral cancer slope factors derived from the five bioassays in mice with exposure durations of 52-
104 weeks ranged from 2.1 x 10"2 to 1.1 x 10"1 (mg/kg-day)"1 (see Table 5-12).
In the conversion of animal doses to human equivalent doses for continuous lifetime
exposure, exposure time adjustment factors (i.e., [duration of experiment/duration of animal
life]3), were used. For the 104-week study of DeAngelo et al. (2008), this factor was equal to 1.
Because of the uncertainty inherent in applying this adjustment factor, the slope factor derived
from the study of longest duration is generally preferred. Moreover, TCA may be a more potent
carcinogen in male mice than in female mice, as discussed previously in Section 4.8.2. In
addition, the four slope factors derived from the incidence data in male mice varied by about
twofold. In light of these considerations, the slope factor of 6.7 x 10"2 (mg/kg-day)"1 derived
from the study of longest duration (i.e., the 104-week mouse bioassay by DeAngelo et al. [2008])
was selected as the cancer slope factor for TCA.
The slopes of the linear extrapolation from the BMDio, the central estimate of exposure
associated with 10% extra cancer risk, were also derived (Table 5-12) from the same studies used
to derive the oral cancer slope factors (DeAngelo et al., 2008; Bull et al., 2002, 1990; Pereira,
1996). Based on the study of longest duration (the 104-week data from DeAngelo et al., 2008),
the slope of the linear extrapolation from the BMDio is 1.8 x 10"2 (mg/kg-day)"1.
No inhalation unit risk (IUR) for TCA was derived. Cancer bioassays involving
inhalation exposure to TCA are not currently available, and PBPK models that could be used to
support route-to-route extrapolation for TCA have not been published. In the absence of a PBPK
model, route-to-route extrapolation (from oral to inhalation) is not recommended because the
146 DRAFT - DO NOT CITE OR QUOTE
-------
liver is the critical target organ for oral toxicity, and first-pass effect by the liver is expected
following oral administration. Furthermore, the respiratory tract has not been evaluated in oral
exposure studies.
5.4.7. Previous Cancer Assessment
In the previous cancer assessment of TCA posted to the IRIS database in 1996, TCA was
classified as a "C," or "possible human carcinogen." This classification was based on a lack of
human data, limited evidence of an increased incidence of liver neoplasms in both sexes of one
strain of mice, and no evidence of carcinogenicity in rats. The previous IRIS assessment did not
provide quantitative estimates of carcinogenic risk from oral or inhalation exposure to TCA.
147 DRAFT - DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE
6.1. HUMAN HAZARD POTENTIAL
TCA (CASRN 76-03-9) has the chemical formula C2HC13O2 and a molecular weight of
163.39 g/mol. At room temperature, TCA is a colorless to white crystalline solid with a sharp,
pungent odor. It is used as a soil sterilant and as a laboratory reagent in the synthesis of
medicinal products and organic chemicals. TCA is used in industry as an etching and pickling
agent. Medical applications of TCA include use as an antiseptic, as a reagent for detection of
albumin, and as a skin peeling agent. TCA is formed as a combustion byproduct of organic
compounds in the presence of chlorine. TCA is also formed by the interaction of organic
material with chlorine during drinking water disinfection. TCA has been detected in water
distribution systems, tap water used for drinking and household activities, and swimming pools.
Direct human exposure to TCA occurs via ingestion of disinfected tap water, inhalation,
and dermal contact. TCA is also formed as a metabolite in the human body after exposure to the
environmental contaminants TCE, tetrachloroethylene, and chloral hydrate.
TCA is readily absorbed by the oral route in rats and by the dermal and oral routes in
humans. Once absorbed, TCA is available for systemic distribution, based on the appearance of
TCA in blood after oral exposure in rodents. Tissue distribution of TCA appears to be dependent
on the time of measurement following dosing. TCA binds to plasma proteins, which is an
important determinant of the extent to which TCA partitions from plasma into target tissues. No
studies were identified that investigated the tissue distribution of TCA in humans, but the
appearance of TCA in the blood and urine of humans exposed to chlorinated solvents or orally
administered chloral hydrate indicates that it is present in the systemic circulation as a
downstream metabolite. No studies investigating the kinetics or degree of maternal-to-fetus or
blood-to-breast-milk transfer of TCA were located.
TCA is not readily metabolized, as indicated by minimal first-pass metabolism in the
liver following oral dosing with TCA and by limited amounts of radioactivity excreted in
exhaled air or present as non-extractable radioactivity in plasma and liver following intravenous
administration of [1-14C]-TCA. Results from animal studies indicate that TCA is not as
extensively metabolized as other chlorinated acids, such as DCA, and that TCA is metabolically
converted to DCA. However, with exposure to TCA, levels of DCA in blood, liver, and urine
are low or not detectable, presumably due to rapid metabolism of DCA. The metabolic
conversion of TCA to DCA via reductive dehalogenation is likely catalyzed by CYP450
enzymes through the dichloroacetate radical intermediate, but, in general, enzymes involved in
TCA metabolism are poorly characterized. The primary route of excretion of TCA is in the
urine, with exhalation of CC>2 and fecal excretion contributing to a lesser extent.
148 DRAFT - DO NOT CITE OR QUOTE
-------
The available human data do not provide a definitive picture of the possible noncancer
effects of long-term human exposure to TCA. No human epidemiology or occupational studies
of TCA were located. Case reports and accounts of the medical use of TCA for skin treatments
demonstrate its potential for skin corrosion and eye irritation. However, no information on
systemic toxicity following dermal exposure of humans to TCA was identified.
In animals, TCA induces systemic, noncancer effects that can be grouped into three
general categories: liver toxicity, metabolic alterations, and developmental toxicity. Studies in
rats and mice indicate that TCA primarily affects the liver, although effects on the lungs and
kidneys have also been noted in rats. Observed hepatic effects in rodents include increased size
and weight, collagen deposition, indications of altered lipid and carbohydrate metabolism,
histopathologic changes, peroxisome proliferation, evidence of lipid peroxidation, and oxidative
damage to hepatic DNA. TCA may influence intermediary carbohydrate metabolism, as shown
by altered glycogen content in the livers of mice treated with TCA. Administration of TCA to
female rats during pregnancy induced developmental effects in six studies at doses that also
resulted in maternal toxicity. Two of these studies are single-dose studies. The observed effects
include fetal cardiac malformations, decreased crown-rump length, and reduced fetal body
weight. The pattern of observed fetal cardiac malformation effects is not consistent across the
available studies. The reason for this inconsistency is unknown but may be related to factors
such as the differences in susceptibility of the test animal strain and/or the poor definition of the
cardiac malformations.
There appear to be different MOAs for the liver toxicity, metabolic alterations, and
developmental effects induced by TCA. For liver effects, some changes such as cytomegaly and
cell proliferation may be explained by TCA-induced peroxisome proliferation. Oxidative stress
responses such as lipid peroxidation and/or oxidative DNA damage may also contribute to the
hepatotoxicity of TCA. The cellular mechanisms underlying changes in lipid and carbohydrate
homeostasis have not been conclusively identified. It has been proposed that TCA may alter
carbohydrate and lipid homeostasis by activation or inhibition of key liver enzymes; by
activation of the peroxisome proliferation pathway, which in turn induces transcription of genes
that encode enzymes responsible for fatty acid metabolism; and/or by suppression of one or more
steps of the glycogen degradation process. The MOA for developmental toxicity is unknown. It
has been suggested that TCA, as a strong acid, might induce developmental toxicity by causing
lesions in the placenta, resulting in anoxia, oxidative stress, and apoptosis in the developing fetus
or embryo.
The genotoxicity of TCA has been evaluated in assays of mutagenicity, DNA repair,
clastogenicity, micronucleus induction, and DNA strand breaks. The weight of evidence from
these studies suggests that TCA is, at most, weakly genotoxic.
No human oral or inhalation cancer data are available specifically for TCA. In animals,
the carcinogenic potential of TCA has been evaluated in oral bioassays conducted in mice and
149 DRAFT - DO NOT CITE OR QUOTE
-------
rats. TCA induced tumors in the livers of male and female mice in multiple bioassays, but
treatment-related tumors of the liver or other organs were not observed in a chronic drinking
water bioassay in rats.
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), TCA is
determined to be "likely to be carcinogenic to humans" by all routes of exposure. This
descriptor is supported by the fact that: TCA is carcinogenic in the liver in multiple studies
conducted in B6C3Fi mice of both sexes; and liver tumors were induced following less-than-
lifetime exposures (as little as 51 weeks) in studies where tumor rates in control animals were
relatively low.
It is possible that there are segments of the human population that are especially
susceptible to the toxic effects of TCA as a result of age, gender, health status, or genetic factors,
but there are no studies specifically on TCA to fully evaluate this possibility. Age-dependent
differences in susceptibility to noncancer effects of TCA have not been investigated in systemic
toxicity studies. The developmental toxicity data on TCA are too limited to draw any
conclusions on whether developing organisms might be a sensitive subpopulation. The LOAELs
observed in subchronic toxicity studies suggest that systemic effects are observed at doses
similar to or less than those at which developmental toxicity has been observed; however, no
developmental NOAELs are available for comparison with the subchronic systemic NOAELs.
Given the lack of a developmental NOAEL, it is uncertain what dose would be protective for
developmental toxicity. The existing data on TCA are also insufficient to determine whether
there are age-dependent differences (e.g., plasma binding and metabolism) in the toxicokinetics
of TCA that might lead to differences in health risk. There are no published comparative data for
plasma binding of TCA in young and old animals. In the only study to evaluate the cancer
potency of TCA in young animals, the incidence of liver tumors in mice injected with TCA as
neonates did not differ significantly from solvent controls when evaluated at 15 or 20 months of
age.
No data on gender effects of TCA toxicity in humans were located. Studies in mice and
rats where males and females were tested concurrently suggest that both sexes are about equally
susceptible to the noncancer effects of TCA. In contrast, male mice appear to be more
susceptible to the carcinogenic effects of TCA, based on the observation of a dose-related
increase in proliferative lesions in males but not females when both sexes were tested
concurrently. Other factors that might confer greater susceptibility to the toxic effects of TCA
include a medical history of glycogen storage disease or genetic deficiencies in glyoxylate-
metabolizing enzymes or antioxidant response.
150 DRAFT - DO NOT CITE OR QUOTE
-------
6.2. DOSE RESPONSE
6.2.1. Noncancer/Oral
No human data were available for oral dose-response analysis; therefore, the oral RfD is
based on data from laboratory animals. An estimated BMDLio of 18 mg/kg-day derived using
BMD modeling based on the increased incidence of hepatocellular necrosis in male B6C3Fi
mice exposed to TCA via drinking water for 30-45 weeks (DeAngelo et al., 2008) was selected
as the POD for calculation of the RfD. This value was divided by a composite UF of 1,000 that
includes individual factors of 10 each to account for variability among humans, extrapolation
from laboratory animal data to humans, and database limitations. The oral RfD is therefore
18 mg/kg-day/1,000 = 0.02 mg/kg-day.
Confidence in the principal study (DeAngelo et al., 2008) is medium. The study was well
designed, with a study duration of up to 104 weeks, and well conducted. Quantitative data for
the incidence and severity of the various endpoints were included in the published paper.
Complete histopathologic examination was conducted for control and high-dose groups.
Confidence in the database is medium. Human data are limited primarily to case reports of skin
or eye effects associated with medical treatments, and information on systemic toxicity is
lacking. Significant gaps in the animal database include absence of a multigeneration
reproductive toxicity study. Overall confidence in the RfD is medium, reflecting these
considerations.
6.2.2. Noncancer/Inhalation
No inhalation studies adequate for the derivation of an RfC were located. The respiratory
tract has not been examined in oral studies of TCA. Because the liver is the critical target organ
for oral toxicity and first-pass effect by the liver is expected following oral administration, the
route of exposure may influence the hepatic response to TCA. PBPK models that would support
route-to-route extrapolation for TCA have not been published. Thus, the available information is
inadequate for extrapolation of oral toxicity data to the inhalation pathway. For these reasons, an
RfC for TCA was not derived.
6.2.3. Cancer/Oral and Inhalation
In the absence of a well-characterized MOA that could explain dose-response
relationships at doses lower than those leading to observed effects, the cancer dose-response
modeling was carried out using linear extrapolation (U.S. EPA, 2005a). No data were found that
were suitable for accounting for interspecies differences in toxicokinetics or toxicodynamics in
dose-response modeling.
Candidate oral cancer slope factors were derived from liver tumor incidence data from
male B6C3Fi mice exposed to TCA in drinking water for 52 weeks (Bull et al., 2002, 1990),
60 weeks (DeAngelo et al., 2008), or 104 weeks (DeAngelo et al., 2008) and from female
151 DRAFT - DO NOT CITE OR QUOTE
-------
B6C3Fi mice exposed to TCA in drinking water for 82 weeks (Pereira, 1996). The slope factors
derived from these studies were 1.1 x 10"1, 8.8 x 10"2, 6.0 x 10"2, 6.7 x 10"2, and 2.1 x 10"2
(mg/kg-day)"1, respectively. These candidate oral slope factors varied by less than one order of
magnitude. The oral cancer slope factor derived from the 104-week bioassay in male B6C3Fi
mice (DeAngelo et al., 2008), or 6.7 x 10"2 (mg/kg-day)"1, was selected as the oral cancer slope
factor for TCA. This bioassay is the only lifetime study of TCA, and an exposure time
adjustment factor (i.e., duration of experiment/duration of animal life)3 is not required. Because
of the uncertainty inherent in applying this adjustment factor, the slope factor derived from the
study of longest duration is generally preferred.
No IUR for TCA was derived. Cancer bioassays involving inhalation exposure to TCA
are not currently available, and PBPK models that could be used to support route-to-route
extrapolation for TCA have not been published. In the absence of a PBPK model, route-to-route
extrapolation (from oral to inhalation) is not recommended because the liver is the critical target
organ for oral toxicity, and first-pass effect by the liver is expected following oral administration.
Furthermore, the respiratory tract has not been evaluated in oral exposure studies.
152 DRAFT - DO NOT CITE OR QUOTE
-------
7. REFERENCES
Abbas, R; Fisher, JW. (1997) A physiologically based pharmacokinetic model for trichloroethylene and its
metabolites, chloral hydrate, trichloroacetate, dichloroacetate, trichloroethanol, and trichloroethanol glucuronide in
B6C3FJ mice. Toxicol Appl Pharmacol 147:15-30.
Acharya, S; Mehta, K; Rodrigues, S; et al. (1995) Administration of subtoxic doses of t-butyl alcohol and
trichloroacetic acid to male Wistar rats to study the interactive toxicity. Toxicol Lett 80:97-104.
Acharya, S; Mehta, K; Rodrigues, S; et al. (1997) A histopathological study of liver and kidney in male Wistar rats
treated with subtoxic doses of t-butyl alcohol and trichloroacetic acid. Exp Toxic Pathol 49:369-373.
Al-Waiz, MM; Al-Sharqi, AL (2002) Medium-depth chemical peels in the treatment of acne scars in dark-skinned
individuals. Dermatol Surg 28(5):383-387.
Austin, EW; Okita, JR; Okita, RT; et al. (1995) Modification of lipoperoxidative effects of dichloroacetate and
trichloroacetate is associated with peroxisome proliferation. Toxicology 97:59-69.
Austin, EW; Parrish, JM; Kinder, DH; et al. (1996) Lipid peroxidation and formation of 8-hydroxydeoxyguanosine
from acute doses of halogenated acetic acids. Fundam Appl Toxicol 31:77-82.
Bannasch, P. (1996) Pathogenesis of hepatocellular carcinoma: sequential cellular, molecular, and metabolic
changes. Prog Liver Dis 14:161-197.
Bannasch, P; Kopp-Schneider, A; Nehrbass, D. (2001) Significance of hepatic preneoplasia for cancer
chemoprevention. IARC SciPubl 154:223-240.
Baylin, SB; Herman, JG; Graff, JR; et al. (1998) Alterations in DNA methylation: a fundamental aspect of
neoplasia. Adv Cancer Res 72:141-196.
Baylin, SB; Esteller, M; Rountree, MR; et al. (2001) Aberrant patterns of DNA methylation, chromatin formation
and gene expression in cancer. Hum Mol Genet 10(7):687-692.
Benane, SG; Blackman, CF; House, DE. (1996) Effect of perchloroethylene and its metabolites on intercellular
communication in clone 9 rat liver cells. J Toxicol Environ Health 48:327-437.
Bhat, HK; Ahmed, AE; Ansari, GAS. (1991) Toxicokinetics of monochloracetic acid: a whole-body
autoradiography study. Toxicology 63:35-43.
Bhunya, SP; Behera, BC. (1987) Relative genotoxicity of trichloroacetic acid (TCA) as revealed by different
cytogenetic assays: bone marrow chromosome aberration, micronucleus and sperm-head abnormality in the mouse.
MutatRes 188:215-221.
Boorman, GA; Dellarco, V; Dunnick, JK; et al. (1999) Drinking water disinfection byproducts: review and approach
to toxicity evaluation. Environ Health Perspect 107(Suppl. 1):207-217.
Bowden, DJ; Clegg, SL; Brimblecombe, P. (1998) The Henry's law constants of the haloacetic acids. J Atmos
Chem 29:85-107.
Brashear, WT; Bishop, CT; Abbas, R. (1997) Electrospray analysis of biological samples for trace amounts of
trichloroacetic acid, dichloroacetic acid, and monochloroacetic acid. J Anal Toxicol 21:330-334.
Briggs, RT; Robinson, JM; Karnovsky, ML; et al. (1986) Superoxide production by polymorphonuclear leukocytes.
A cytochemical approach. Histochemistry 84:371-378.
Bruning, T; Vamvakas, S; Makropoulos, V; etal. (1998) Acute intoxication with trichloroethene: clinical symptoms,
toxicokinetics, metabolism, and development of biochemical parameters for renal damage. Toxicol Sci 41:157-165.
153 DRAFT - DO NOT CITE OR QUOTE
-------
Bull, RJ. (2000) Mode of action of liver introduction by trichloroethylene and its metabolites, trichloroacetate and
dichloroacetate. Environ Health Perspect 108(Suppl. 2):241-259.
Bull, RJ; Sanchez, IM; Nelson, MA; et al. (1990) Liver tumor induction in B6C3F! mice by dichloroacetate and
trichloroacetate. Toxicology 63:341-359.
Bull, RJ; Orner, GA; Cheng, RS; et al. (2002) Contribution of dichloroacetate and trichloroacetate to liver tumor
induction in mice by trichloroethylene. Toxicol Appl Pharm 182:55-65.
Bull, RJ; Sasser, LB; Lei, XC. (2004) Interactions in the tumor-promoting activity of carbon tetrachloride,
trichloroacetate, and dichloroacetate in the liver of male B6C3F! mice. Toxicology 199:169-183.
Calafat, AM; Kuklenyik, Z; Caudill, SP; et al. (2003) Urinary levels of trichloroacetic acid, a disinfection by-
product in chlorinated drinking water, in a human reference population. Environ Health Perspect 111(2): 151-154.
Cattley, RC; Miller, RT; Gorton, JC. (1995) Peroxisome proliferates: potential role of altered hepatocyte growth
and differentiation in tumor development. Prog Clin Biol Res 391:295-303.
Celik, I. (2007) Determination of toxicity of trichloroacetic acid in rats: 50 days drinking water study. Pest Biochem
Physiol 89:39-15.
Chalitchagorn, K; Shuangshoti, S; Hourpai, N; et al. (2004) Distinctive pattern of LINE-1 methylation level in
normal tissues and the association with carcinogenesis. Oncogene 23:8841-8846.
Chang, LW; Daniel, FB; DeAngelo, AB. (1992) Analysis of DNA strand breaks induced in rodent liver in vivo,
hepatocytes in primary culture, and a human cell line by chloroacetic acids and chloroacetaldehydes. Environ Mol
Mutagen 20:277-288.
Channel, SR; Hancock, BL. (1993) Application of kinetic models to estimate transit time through cell cycle
compartments. Toxicol Lett 68( 1-2) :213-221.
Cheung, C; Aklyama, TE; Ward, JM; et al. (2004) Diminished hepatocellular proliferation in mice humanized for
the nuclear receptor peroxisome proliferator-activated receptor a. Cancer Res 64:3849-3854.
Chiarello, SE; Resnik, BI; Resnik, SS. (1996) The TCA Masque. A new cream formulation used alone and in
combination with Jessner's solution. Dermatol Surg 22(8):687-690.
Coffin, JC; Ge, R; Yang, S; et al. (2000) Effect of trihalomethanes on cell proliferation and DNA methylation in
female B6C3F1 mouse liver. Toxicol Sci 58:243-252.
Coleman, WP. (2001) Dermal peels. Dermatol Clin 19(3):405^H1.
Coleman, WE; Melton, RG; Kopfler, FC; et al. (1980) Identification of organic compounds in a mutagenic extract of
a surface drinking water by a computerized gas chromatography/mass spectrometry system (GS/MS/COM).
Environ Sci Technol 14(5):576-588.
Collier, JM; Selmin, O; Johnson, PD; et al. (2003) Trichloroethylene effects on gene expression during cardiac
development. Clin Mol Teratol 67(7):488-495.
Cornett, R; Yan, Z; Henderson, G; et al. (1997) Cytosolic biotransformation of dichloroacetic acid (DCA) in the
Sprague-Dawley rat. Fundam Appl Toxicol 36(Suppl.):318.
Cornett, R; James, MO; Henderson, GN; et al. (1999) Inhibition of glutathione S-transferase zeta and tyrosine
metabolism by dichloroacetate: a potential unifying mechanism for its altered biotransformation and toxicity.
Biochem Biophys Res Commun 262:752-756.
Cosby, NC; Dukelow, WR. (1992) Toxicology of maternally ingested trichloroethylene (TCE) on embryonal and
fetal development in mice and of TCE metabolites on in vitro fertilization. Fundam Appl Toxicol 19(2):268-274.
154 DRAFT - DO NOT CITE OR QUOTE
-------
Cotellessa, C; Peris, K; Fargnoli, MC; et al. (2003) Microabrasion versus microabrasion followed by 15%
trichloroacetic acid for treatment of cutaneous hyperpigmentations in adult females. Dermatol Surg 29(4):352-356.
Counts, JL; Goodman, JI. (1994) Hypomethylation of DNA: an epigenetic mechanism involved in tumor promotion.
Mol Carcinog 11:185-188.
Counts, JL; Goodman, JI. (1995) Hypomethylation of DNA: a nongenotoxic mechanism involved in tumor
promotion. Toxicol Lett 82/83:663-672.
Cox, S. (2003) Rapid development of keratoacanthomas after a body peel. Dermatol Surg 29(2):201-203.
Crabb, DW; Yount, EA; Harris, RA. (1981) The metabolic effects of dichloroacetate. Metabolism 30:1024-1039.
Davis, ME. (1990) Subacute toxicity of trichloroacetic acid in male and female rats. Toxicology 63(l):63-72.
Davis, LM; Caspary, WJ; Sakallah, SA; et al. (1994) Loss of heterozygosity in spontaneous and chemically induced
tumors of B6C3Fi mice. Carcinogenesis 15:1637-1645.
DeAngelo, AB; Daniel, FB; McMillan, L; et al. (1989) Species and strain sensitivity to the induction of peroxisome
proliferation by chloroacetic acids. Toxicol Appl Pharmacol 101:285-289.
DeAngelo, AB; Daniel, FB; Most, BM; et al. (1997) Failure of monochloroacetic acid and trichloroacetic acid
administered in the drinking water to produce liver cancer in male F344/N rats. J Toxicol Environ Health 52:425-
445.
DeAngelo, AB; Daniel, FB; Wong, D; et al. (2008) The induction of hepatocellular neoplasia by trichloroacetic acid
administered in the drinking water of the male B6C3F1 mouse. J Toxicol Environ Health Part A 71:1056-1068.
Dees, C; Travis, C. (1994) Trichloroacetate stimulation of liver DNA synthesis in male and female mice. Toxicol
Lett 70:343-355.
DeMarini, DM; Perry, E; Sheldon, ML. (1994) Dichloroacetic acid and related compounds: induction of prophage in
E. coli and mutagenicity and mutation spectra in salmonella TA 100. Mutagenesis 9:429-437.
Doull, J; Cattley, R; Elcombe, C; et al. (1999) A cancer risk assessment of di(2-ethylhexyl)phthalate: application of
the new U.S. EPA Risk Assessment Guidelines. Regul Toxicol Pharmacol 29(3):327-357.
Dunn, BK (2003) Hypomethylation: one side of a larger picture. Ann N Y Acad Sci 983:2&-42.
Elcombe, CR. (1985) Species differences in carcinogenicity and peroxisome proliferation due to trichloroethylene: a
biochemical human hazard assessment. Arch Toxicol Suppl. 8:6-17.
Evans, OB; Stacpoole, PW. (1982) Prolonged hypolactatemia and increased total pyruvate dehydrogenase activity
by dicloroacetate. Biochem Pharmacol 31:1295-1300.
Fausto, N; Webber, EM. (1993) Control of liver growth. Crit Rev Eukaryotic Gene Expr 3:117-135.
Ferreira-Gonzalez, A; DeAngelo, AB; Nasim, S; et al. (1995) Ras oncogene activation during hepatocarcinogenesis
in B6C3F! male mice by dichloroacetic and trichloroacetic acids. Carcinogenesis 16(3):495-500.
Fisher, JW; Mahle, D; Abbas, R. (1998) A human physiologically based pharmacokinetic model for
trichloroethylene and its metabolites, trichloroacetic acid and free trichloroethanol. Toxicol Appl Pharmacol
152(2):339-359.
Fisher, JW; Channel, SR; Eggers, JS; et al. (2001) Trichloroethylene, trichloroacetic acid, and dichloroacetic acid:
do they affect fetal rat heart development? Int J Toxicol 20(5):257-267.
155 DRAFT - DO NOT CITE OR QUOTE
-------
Fort, D; Stover, E; Rayburn, J; et al. (1993) Evaluation of the developmental toxicity of trichloroethylene and
detoxification metabolites using xenopus. Teratog Carcinog Mutagen 13:35-45.
Froese, KL; Sinclair, MI; Hrudey, SE. (2002) Trichloroacetic acid as abiomarker of exposure to disinfection by-
products in drinking water: a human exposure trial in Adelaide, Australia. Environ Health Perspect 110(7):679-687.
Fung, JF; Sengelmann, RD; Kenneally, CZ. (2002) Chemical injury to the eye from trichloroacetic acid. Dermatol
Surg28(7):609-610.
Furstenberger, G; Serin, HJ. (2002) Insulin-like growth factors and cancer. Lancet Oncol 3:298-302.
Gama-Sosa, MA; Slagel, VA; Trewyn, RW; et al. (1983) The 5-methylcytosine content of DNA from human
tumors. Nucleic Acids Res 11:6883-6894.
Ge, R; Yang, S; Kramer, PM; et al. (2001a) The effect of dichloroacetic acid and trichloroacetic acid onDNA
methylation and cell proliferation in B6C3Fi mice. J Biochem Mol Toxicol 15(2): 100-106.
Ge, R; Wang, W; Kramer, PM; et al. (200 Ib) Wy-14643-induced hypomethylation of the c-myc gene in mouse liver.
Toxicol Sci 62:28-35.
Ghantous, H; Danielsson, RG; Dencker, L; et al. (1986) Trichloroacetic acid accumulates in murine amniotic fluid
after tri- and tetrachloroethylene inhalation. Acta Pharmacol Toxicol 58:105-114.
Gibson, GG. (1989) Comparative aspects of the mammalian cytochrome P450 IV gene family. Xenobiotica
19(10):1123-1148.
Giller, S; Le Curieux, F; Erb, F; et al. (1997) Comparative genotoxicity of halogenated acetic acids found in
drinking water. Mutagenesis 12(5):321-328.
Goldsworthy, TL; Popp, JA. (1987) Chlorinated hydrocarbon-induced peroxisomal enzyme activity in relation to
species and organ carcinogenicity. Toxicol Appl Pharmacol 88:225-233.
Gonzalez-Leon, A; Merdink, JL; Bull, RJ; et al. (1999) Effect of pre-treatment with dichloroacetic or trichloroacetic
acid in drinking water on the pharmacokinetics of a subsequent challenge dose in B6C3FJ mice. Chemic-Biol Inter
123:239-253.
Grasl-Kraupp, B; Waldhor, T; Huber, W; et al. (1993) Glutathione S-transferase isoenzyme patterns in different
subtypes of enzyme-altered rat liver foci treated with the peroxisome proliferator nafenopin or with phenobarbital.
Carcinogenesis 14(11):2407-2412.
Hajimiragha, H; Ewers, U; Jansen-Rosseck, R; et al. (1986) Human exposure to volatile halogenated hydrocarbons
from the general environment. Int Arch Occup Environ Health 58:141-150.
Hansch, C; Leo, A; Hoekman, D. (1995) Exploring QSAR: hydrophobic, electronic, and steric constants. ACS
professional reference book. Washington, DC: American Chemical Society; p. 4.
Harrington-Brock, K; Doerr, CL; Moore, MM. (1998) Mutagenicity of three disinfection by-products; di- and
trichloroacetic acid and chloral hydrate in L5178Y/TK+/~ - 3.7.2C mouse lymphoma cells. Mutat Res 413:265-276.
Hasmall, SC; Pyrah, IT; Soames, AR; et al. (1997) Expression of the immediate-early genes, c-fos, c-jun, and
c-myc: a comparison in rats of nongenotoxic hepatocarcinogens with noncarcinogenic liver mitogens. Fundam Appl
Toxicol 40(1): 129-137.
Hasmall, SC; West, DA; Olsen, K; et al. (2000a) Role of hepatic non-parenchymal cells in the response of rat
hepatocytes to the peroxisome proliferator nafenopin in vitro. Carcinogenesis 21(12):2159-2165.
Hassoun, EA; Ray, S. (2003) The induction of oxidative stress and cellular death by the drinking water disinfection
by-products, dichloroacetate and trichloroacetate in J774.Al cells. Comp Biochem Physiol C Pharmacol Toxicol
Endocrinol 135:119-128.
156 DRAFT - DO NOT CITE OR QUOTE
-------
Hegi, ME; Fox, RR; Belinsky, SA; et al. (1993) Analysis of activated protooncogenes in B6C3Fi mouse liver
tumours induced by ciprofibrate, a potent peroxisome proliferator. Carcinogenesis 14:145-149.
Herren-Freund, SL; Pereira, MA; Khoury, MD; et al. (1987) The carcinogenicity of trichloroetliylene and its
metabolites, trichloroacetic acid and dichloroacetic acid, in mouse liver. Toxicol Appl Pharmacol 90:183-189.
Hinckley, F; Bachand, AM; Reiff, JS. (2005) Late pregnancy exposures to disinfection by-products and growth-
related birth outcomes. Environ Health Perspect 113:1808-1813.
Hobara, T; Kobayashi, J; Kawamoto, T; et al. (1986) Biliary excretion of trichloroethylene and its metabolites in
dogs. Toxicol Lett 32:119-122.
Hobara, T; Kobayashi, J; Kawamoto, T; et al. (1987) The cholecystohepatic circulation of trichloroethylene and its
metabolites in dogs. Toxicol Lett 44:283-295.
Hobara, T; Kobayashi, J; Kawamoto, T; et al. (1988a) Intestinal absorption of chloral hydrate, free trichloroethanol
and trichloroacetic acid in dogs. Pharmacol Toxicol 62:250-258.
Hobara, T; Kobayashi, J; Kawamoto, T; et al. (1988b) The absorption of trichloroethylene and its metabolites from
the urinary bladder of anesthetized dogs. Toxicology 48(2): 141-153.
Hoivik, DJ; Quails, CW Jr; Mirabile, RC; et al. (2004) Fibrates induce hepatic peroxisome and mitochondria!
proliferation without overt evidence of cellular proliferation and oxidative stress in cynomolgus monkeys.
Carcinogenesis 25(9): 1757-1769.
Holden, PR; Tugwood, JD. (1999) Peroxisome proliferator-activated receptor alpha: role in rodent liver cancer and
species differences. J Mol Endocrinol 22(1): 1-8.
Hunter, ES; Rogers, EH. (1999) Dysmorphogenic effects of three metabolites of haloacetic acids in mouse embryo
culture. Teratology 59(6):402.
Hunter, ES, III; Rogers, EH; Schmid, JE; et al. (1996) Comparative effects of haloacetic acids in whole embryo
culture. Teratology 54:57-64.
IPCS (International Programme on Chemical Safety). (2000) Disinfectants and disinfectant by-products.
Environmental health criteria. Vol. 216. World Health Organization, Geneva, Switzerland. Available online at
http://www.inchem.org/documents/ehc/ehc/ehc216.htm (accessed March 14, 2011).
Ito, Y; Yamanoshita, O; Asaeda, N; et al. (2007) Di(2-ethylhexyl)phthalate induces hepatic tumorigenesis through a
peroxisome proliferator-activated receptor a-independent pathway. J Occup Health 49:172-182.
Ittrich, C; Demi, E; Oesterle, D; et al. (2003) Prevalidation of a rat liver foci bioassay (RLFB) based on results from
1600 rats: a study report. Toxicol Pathol 31(l):60-79.
James, MO; Cornett, R; Yan, Z; et al. (1997) Glutathione-dependent conversion to glyoxylate, a major pathway of
DCA biotransformation in hepatic cytosol from humans and rats, is reduced in DCA-treated rats. Drug Metab Disp
25(11): 1223-1227.
James, MO; Yan, Z; Cornett, R; et al. (1998a) Pharmacokinetics and metabolism of [14C]dichloroacetate in male
Sprague-Dawley rats. Identification of glycine conjugates, including hippurate, as urinary metabolites of
dichloroacetate. Drug Metab Dispos 26(11): 1134-1143.
Johnson, PD; Dawson, BV; Goldberg, SJ. (1998) Cardiac teratogenicity of trichloroethylene metabolites. J Am Coll
Cardiol 32(2):540-545.
Johnson, SM; Grosshans, H; Shingara, J; et al. (2005) RAS is regulated by the let-7 microRNA family. Cell
120(5):635-647.
157 DRAFT - DO NOT CITE OR QUOTE
-------
Jones, PA; Gonzalgo, ML. (1997) Altered DNA methylation and genome instability: a new pathway to cancer? Proc
Natl Acad Sci 94:2103-2105.
Juuti, S; Hoekstra, E. (1998) New directions: the origins and occurrence of trichloroacetic acid. Atmos Environ
32(17):3059-3060.
Kang, WH; Kim, NS; Kim, YB; et al. (1998) A new treatment for syringoma. Combination of carbon dioxide laser
and trichloroacetic acid. Dermatol Surg 24(12): 1370-1374.
Kargalioglu, Y; McMillan, BJ; Minear, RA; et al. (2002) Analysis of the cytotoxicity and mutagenicity of drinking
water disinfection by-products in Salmonella typhimurium. Teratog Carcinog Mutagen 22:113-128.
Karnovsky, M; Badwey, J; Lochner, J; et al. (1988) Trigger phenomena for the release of oxygen radicals by
phagocytic leukocytes. In: Cerutie, P; Fridovich, I; McCord, J; eds. Oxy-radicals in molecular biology and
pathology: proceedings of anUpjohn-UCLA symposium; January 24-30; Park City, UT. New series. Vol. 82. New
York, NY: AlanR. Liss; pp. 61-81.
Kato-Weinstein, J; Lingohr, MK; Orner, GA; et al. (1998) Effects of dichloroacetate on glycogen metabolism in
B6C3FJ mice. Toxicology 130:141-154.
Kato-Weinstein, J; Stauber, AJ; Orner, GA; et al. (2001) Differential effects of dihalogenated and trihalogenated
acetates in the liver of B6C3FJ mice. J Appl Toxicol 21:81-89.
Ketcha, MM; Stevens, DK; Warren, DA; et al. (1996) Conversion of trichloroacetic acid to dichloroacetic acid in
biological samples. J Anal Toxicol 20(4):236-241.
Khandwala, HM; McCutcheon, IE; Flyvbjerg, A; et al. (2000) The effects of insulin-like growth factors on
tumorigenesis and neoplastic growth. Endocrin Rev 21:215-244.
Kim, H; Weisel, CP. (1998) Dermal absorption of dichloro- and trichloroacetic acids from chlorinated water. J
Expo Anal Environ Epid 8(4):555-575.
Kim, YJ; Shin, BS; Chung, BS; et al. (2002) A simple technique for treatment of nasal telangiectasia using
trichloroacetic acid and CO2 laser. Dermatol Surg 28(8):729-731.
King, WD; Dodds, L; Allen, AC; et al. (2005) Haloacetic acids in drinking water and risk for stillbirth. Occup
Environ Med 62:124-127.
Klaunig, J; Ruch, R; DeAngelo, A; et al. (1988) Inhibition of mouse hepatocyte intercellular communication by
phthalate esters. Cancer Lett 43:65-71.
Klaunig, J; Ruch, RJ; Lin, ELC. (1989) Effects of trichloroethylene and its metabolites on rodent hepatocyte
intercellular communication. Toxicol Appl Pharmacol 99:454-465.
Klaunig, J; Babich, MA; Baetcke, KP; et al. (2003). PPARa agonist-induced rodent tumors: modes of action and
human relevance. Crit Rev Toxicol 33(6):655-780.
Koenig, G. (2005) Chloroacetic Acids In: Ullmann's encyclopedia of industrial chemistry. Available online at
http://onlinelibrary.wiley.com/doi/10.1002/14356007.a06_537/pdf (accessed. March 14, 2011).
Kraupp-Grasl, B; Huber W; Putz B; et al. (1990) Tumor promotion by the peroxisome proliferator nafenopin
involving a specific subtype of altered foci in rat liver. Cancer Res 50(12):3701-3708.
Kraupp-Grasl, B; Huber, W; Taper, H; et al. (1991) Increased susceptibility of aged rats to hepatocarcinogenesis by
the peroxisome proliferator nafenopin and the possible involvement of altered liver foci occurring spontaneously.
Cancer Res 51(2):666-671.
Lapinskas, PJ; Gorton, JC. (1999) Molecular mechanisms of hepatocarcinogenic peroxisome proliferators. In: Puga,
A; Wallace, KB; eds. Molecular biology of the toxic response. Philadelphia, PA: Taylor and Francis; pp. 219-253.
158 DRAFT - DO NOT CITE OR QUOTE
-------
Larson, JL; Bull, RJ. (1992) Metabolism and lipoperoxidative activity of trichloroacetate and dichloroacetate in rats
and mice. Toxicol Appl Pharmacol 115:268-277.
Lash, LH; Fisher, JW; Lipscomb, JC; et al. (2000) Metabolism of trichloroethylene. Environ Health Perspect
108(Suppl. 2): 177-200.
Latendresse, JR; Pereira, MA. (1997) Dissimilar characteristics of N-methyl-N-nitrosourea-initiated foci and tumors
promoted by dichloroacetic acid or trichloroacetic acid in the liver of female B6C3F! mice. Toxicol Pathol
25(5):433^40.
Laughter, AR; Dunn, CS; Swanson, CL; et al. (2004) Role of the peroxisome proliferator-activated receptor alpha
(PPARalpha) in responses to trichloroethylene and metabolites, trichloroacetate and dichloroacetate in mouse liver.
Toxicology 203:83-98.
Lee, YS; Dutta, A. (2007) The tumor suppressor microRNA let-7 represses the HMGA2 oncogene. Genes Dev
21(9):1025-1230.
Lee, JB; Chung, WG; Kwahck, H; et al. (2002) Focal treatment of acne scars with trichloroacetic acid: chemical
reconstruction of skin scars method. Dermatol Surg 28(11):1017-1021.
Lewis, RJ, Sr; ed. (1997) Hawley's condensed chemical dictionary. 13th edition. New York, NY: John Wiley &
Sons, Inc.; p. 1124.
Lide, DR; ed. (2000) CRC handbook of chemistry and physics, 2000-2001. 81st edition. Boca Raton, FL: CRC
Press; pp. 3-9.
Liley, PE; Reid, RC; Buck, E. (1984) Physical and chemical data. In: Perry, RH; Green, D; eds. Perry's chemical
engineer's handbook. 6th edition. New York, NY: McGraw Hill.
Lin, EL; Mattox, JK; Daniel, FB. (1993) Tissue distribution, excretion, and urinary metabolites of dichloroacetic
acid in the male Fisher 344 rat. J Toxicol Environ Health 38(1): 19-32.
Lipscomb, JC; Mahle, DA; Brashear, WT; et al. (1995) Dichloroacetic acid: metabolism in cytosol. Drug Metab
Dispos 23(11): 1202-1205.
Lumpkin, MH; Bruckner, JV; Campbell, JL; et al. (2003) Plasma binding of trichloroacetic acid in mice, rats, and
humans under cancer bioassay and environmental exposure conditions. Drug Metab Dispos 31(10): 1203-1207.
Mackay, JM; Fox, V; Griffiths, K; et al. (1995) Trichloroacetic acid: investigation into the mechanism of
chromosomal damage in the in vitro human lymphocyte cytogenetic assay and the mouse bone marrow
micronucleus test. Carcinogenesis 16(5):1127-1133.
Maloney, EK; Waxman, DJ. (1999) Trans-activation of PPAR-alpha and PPAR-gamma by structurally diverse
environmental chemicals. Toxicol Appl Pharmacol 161:209-218.
Marsman, DS; Cattley, RC; Conway, JG; et al. (1988) Relationship of hepatic peroxisome proliferation and
replicative DNA synthesis to the hepatocarcinogenicity of the peroxisome proliferators di(2-ethylhexyl)phthalate
and [4-chloro-6-(2,3-xylidino)-2-pyrimidinylthio]acetic acid (Wy-14,643) in rats. Cancer Res 48(23):6739-6744.
Mather, GG; Exon, JH; Koller, LD. (1990) Subchronic 90-day toxicity of dichloroacetic and trichloroacetic acid in
rats. Toxicology 64:71-80.
Melnick, RL. (2001) Is peroxisome proliferation an obligatory precursor step in the carcinogenicity of
di(2-ethylhexyl)phthalate (DEHP)? Environ Health Perspect 109(5):437-442.
Merdink, JL; Gonzalez-Leon, A; Bull, RJ; et al. (1998) The extent of dichloroacetate formation from
trichloroethylene, chloral hydrate, trichloroacetate, and trichloroethanol in B6C3F! mice. Toxicol Sci 45:33^11.
159 DRAFT - DO NOT CITE OR QUOTE
-------
Merdink, JL; Bull, RJ; Schultz, RJ. (2000) Trapping and identification of the dichloroacetate radical from the
reductive dehalogenation of trichloroacetate by mouse and rat liver microsomes. Free Rad Biol Med 29:125-130.
Mills, CJ; Bull, RJ; Cantor, KP; et al. (1998) Health risks of drinking water chlorination by-products: report of an
expert working group. Chronic Dis Can 19(3):91-101.
Miyagawa, M; Takasawa, H; Sugiyama, A; et al. (1995) The in vivo-in vitro replicative DNA synthesis (RDS) test
with hepatocytes prepared from male B6C3F! mice as an early prediction assay for putative nongenotoxic (Ames-
negative) mouse hepatocarcinogens. MutatRes 343:157-183.
Moghaddam, AP; Abbas, R; Fisher, JW; et al. (1996) Metabolism of trichloroacetic acid to dichloroacetic acid by rat
and mouse gut microflora, and in vitro study. Biochem Biophys Res Commun 228:639-645.
Moghaddam, AP; Abbas, R; Fisher, JW; et al. (1997) Role of mouse intestinal microflora in dichloroacetic acid
formation, an in vivo study. Human Exp Toxicol 16:629-635.
Moore, MM; Harrington-Brock, K. (2000) Mutagenicity of trichloroethylene and its metabolites: implications for
the risk assessment of trichloroethylene. Environ Health Perspect 108(Suppl. 2):215-223.
Morimura, K; Cheung, C; Ward, J; et al. (2006) Differential susceptibility of mice humanized for peroxisome
proliferator-activated receptor a to Wy-14643-induced liver tumorigenesis. Carcinogenesis 27:1074-1080.
Morris, ED; Bost, JC. (2002) Acetic acid, halogenated derivatives. In: Kirk-Othmer encyclopedia of chemical
technology. Available online at
http://onlinelibrary.wiley.com/doi/10.1002/0471238961.0801121513151818.a01.pub2/pdf (accessed March 15,
2011).
Mower, J; Nordin, J. (1987) Characterization of halogenated organic acids in five gases from municipal waste
incinerators. Chemosphere 16(6):1181-1192.
Moy, LS; Peace, S; Moy, RL. (1996) Comparison of the effect of various chemical peeling agents in a mini-pig
model. DermatolSurg22(5):429-432.
Nakano, H; Hatayama, I; Satoh, K; et al. (1994) c-Jun expression in single cells and preneoplastic foci induced by
diethylnitrosamine in B6C3F1 mice: comparison with the expression of pi-class glutathione S-transferase.
Carcinogenesis 15(9): 1853-1857.
Nelson, MA; Bull, RJ. (1988) Induction of strand breaks in DNA by trichloroethylene and metabolites in rat and
mouse liver in vivo. Toxicol Appl Pharmacol 94:45-54.
Nelson, MA; Sanchez, IM; Bull, RJ; et al. (1990) Increased expression of c-myc and c-Ha-ras in dichloroacetate and
trichloroacetate-induced liver tumors in B6C3F1 mice. Toxicology 64:47-57.
Nelson, GM; Swank, AE; Brooks, LR; et al. (2001) Metabolism, microflora effects, and genotoxicity in haloacetic
acid-treated cultures of rat cecal microbiota. Toxicol Sci 60(2):232-241.
Ni, YC; Kadlubar, FF; Fu, PP. (1995) Formation of a malondialdehyde-modified 2"-deoxyguanosinyl adduct from
metabolism of chloral hydrate by mouse liver microsomes. Biochem Res Commun 216:1110-1117. (As cited in
von Tungeln et al. [2002]).
Ni, YC; Wong, TT; Lloyd, RV; et al. (1996) Mouse liver microsomal metabolism of chloral hydrate, trichloroacetic
acid, and trichloroethanol leading to induction of lipid peroxidation via a free radical mechanism. Drug Metab
Dispos 24:81-90.
Nieuwenhuijsen, MJ; Toledano, MB; Eaton, NE; et al. (1999) Chlorination disinfection byproducts in water and
their association with adverse reproductive outcomes: a review. Occup Environ Med 57:73-85.
160 DRAFT - DO NOT CITE OR QUOTE
-------
NIOSH (National Institute for Occupational Safety and Health). (1973) Urinary metabolites from controlled
exposures of humans to trichloroethylene. Centers for Disease Control and Prevention, Public Health Service, U.S.
Department of Health and Human Services, Cincinnati, OH. Available from the National Technical Information
Service, Springfield, VA; PB82-151713.
NIOSH. (2003) NIOSH pocket guide to chemical hazards. Centers for Disease Control and Prevention, Public
Health Service, U.S. Department of Health and Human Services, Cincinnati, OH. Available online at
http://www.cdc.gov/niosh/npg/pgintrod.html. Accessed December 29, 2003.
NLM (National Library of Medicine). (2003) Trichloroacetic acid. HSDB (Hazardous Substances Data Bank).
National Institutes of Health, U.S. Department of Health and Human Services, Bethesda, MD. Available online at
http://toxnet.nlm.nih.gov (accessed December, 2003).
NRC (National Research Council). (1983) Risk assessment in the federal government: managing the process.
Washington, DC: National Academy Press.
NRC. (2006) Assessing the human health risks of trichloroethylene: key scientific issues. Washington, DC:
National Academies Press.
NRC. (2008) Phthalates and cumulative risk assessment: the tasks ahead. Washington, DC: National Academies
Press.
NRC. (2009) Science and decisions: advancing risk assessment. Washington, DC: National Academies Press.
Nunns, D; Mandal, D. (1996) Trichloroacetic acid: a cause of vulvar vestibulitis. Acta Derm Venereol Suppl.
(Stockh) 76:334.
Okita, RT; Okita, R. (1992) Effects of diethyl phthalate and other plasticizers on laurate hydroxylation in rat liver
microsomes. Pharm Res 9:1648-1653.
O'Neil, MJ; ed. (2001) The Merck index: an encyclopedia of chemicals, drugs, and biologicals. 13th edition.
Whitehouse Station, NJ: Merck and Co., Inc.; p. 1716.
Ono, Y; Somiya, I; Kawamura, M. (1991) The evaluation of genotoxicity using DNA repairing test for chemicals
produced in chlorination and ozonation processes. Water Sci Technol 23(l-3):329-338.
Palmer, CN; Hsu, MH; Griffin, KJ; et al. (1998) Peroxisome proliferator-activated receptor-alpha expression in
human liver. Mol Pharmacol 5 3 (1): 14-22.
Parnell, MJ; Exon, JH; Koller, LD. (1988) Assessment of hepatic initiation-promotion properties of trichloroacetic
acid. Arch Environ Contam Toxicol 17(4):429-436.
Parrish, JM; Austin, EW; Stevens, DK; et al. (1996) Haloacetate-induced oxidative damage to DNA in the liver of
maleB6C3F! mice. Toxicology 110:103-111.
Parzefall, W; Berger, W; Kainzbauer, E; et al. (2001) Peroxisome proliferators do not increase DNA synthesis in
purified rat hepatocytes. Carcinogenesis 22(3):519-523.
Pereira, MA. (1996) Carcinogenic activity of dichloroacetic acid and trichloroacetic acid in the liver of female
B6C3FJ mice. Fundam Appl Toxicol 31:192-199.
Pereira, MA; Phelps, JB. (1996) Promotion by dichloroacetic acid and trichloroacetic acid of N-methyl-
N-nitrosourea-initiated cancer in the liver of female B6C3F! mice. Cancer Lett 102:133-141.
Pereira, MA; Li, K; Kramer, PM. (1997) Promotion by mixtures of dichloroacetic acid and trichloroacetic acid of
N-methy 1-N-nitrosourea-initiated cancer in the liver of female B6C3FJ mice. Cancer Lett 115:15-23.
161 DRAFT - DO NOT CITE OR QUOTE
-------
Pereira MA; Kramer, MP; Conran, PB; et al. (2001) Effect of chloroform on dichloroacetic acid and trichloroacetic
acid-induced hypomethylation and expression of the c-myc gene and on their promotion of liver and kidney tumors
in mice. Carcinogenesis 22(9): 1511-1519.
Peters, JM; Cattley, RC; Gonzalez, FJ. (1997) Role of PPAR alpha in the mechanism of action of the nongenotoxic
carcinogen and peroxisome proliferator Wy-14,643. Carcinogenesis 18(11):2029-2033.
Plewa, MJ; Kargalioglu, Y; Vankerk, D; et al. (2002) Mammalian cell cytotoxicity and genotoxicity analysis of
drinking water disinfection by-products. Environ Mol Mutagen 40:134-142.
Pogribny, IP; Tryndyak, VP; Woods, CG; et al. (2007) Epigenetic effects of the continuous exposure to peroxisome
proliferator Wy-14643 in mouse liver are dependent upon peroxisome proliferator activated receptor a. Mutat Res
625:62-71.
Porter, CK; Putnam, SD; Hunting, KL et al. (2005) The effect of trihalomethane and haloacetic acid exposure on
fetal growth in a Maryland County. Am J Epidemiol 162:334-344.
Pravacek, TL; Channel, SR; Schmidt, WJ; et al. (1996) Cytotoxicity and metabolism of dichloroacetic and
trichloroacetic acid in B6C3F! mouse liver tissue. In Vitro Toxicol 9(3):261-269.
Rao, MS; Tatematsu, M; Subbarao, V; et al. (1986) Analysis of peroxisome proliferator-induced preneoplastic and
neoplastic lesions of rat liver for placenta! form of glutathione S-transferase and gamma-glutamyl transpeptidase.
Cancer Res 46:5287-5290.
Rapson, WH; Nazar, MA; Butsky, W. (1980) Mutagenicity produced by aqueous chlorination of organic
compounds. Bull Environ Contam Toxicol 24:590-596.
Razin, A; Kafri, T. (1994) DNA methylation from embryo to adult. Prog Nucleic Acid Res Mol Biol 48:53-81.
Reimann, S; Grob, K; Frank, H. (1996) Environmental chloroacetic acids in foods analyzed by GC-ECD. Mitteil
Aus Dem Geb der Lebensmittel und Hygiene 87(2):212-222.
Renwick, AG; Lazarus, NR. (1998) Human variability and noncancer risk assessment—an analysis of the default
uncertainty factor. Regul Toxicol Pharmacol 27(1, Pt. 1):3-20.
Reynolds, SH; Stowers, SJ; Patterson, RM; et al. (1987) Activated oncogenes in B6C3F1 mouse liver tumors:
implications for risk assessment. Science 237(4820): 1309-1316.
Rubin, MG. (1995) The efficacy of a topical lidocaine/prilocaine anesthetic gel in 35% trichloroacetic acid peels.
Dermatol Surg 21(3):223-225.
Rusyn, I; Yamashina, S; Segal, BH; et al. (2000) Oxidants from nicotinamide adenine dinucleotide phosphate
oxidase are involved in triggering cell proliferation in the liver due to peroxisome proliferators. Cancer Res
60(17):4798-4803.
Saeter, G; Seglen, PO. (1990) Cell biology of hepatocarcinogenesis. CritRev Oncogen 1:437-466.
Saillenfait, AM; Langonne, I; Sabate, JP. (1995) Developmental toxicity of trichloroethylene, tetrachloroethylene
and four of their metabolites in rat whole embryo culture. Arch Toxicol 70:71-82.
Sakai, M; Matsushima-Hibiya, Y; Nishizawa, M; et al. (1995) Suppression of rat glutathione transferase P
expression by peroxisome proliferators: interaction between Jun and peroxisome proliferator-activated receptor
alpha. Cancer Res 55(22):5370-5376.
Sanchez, IM; Bull, RJ. (1990) Early induction of reparative hyperplasia in B6C3F! mice treated with dichloroacetate
and trichloroacetate. Toxicology 64:33^6.
Scharf, JG; Dombrowski, F; Ramadori, G. (2001) The IGF axis and hepatocarcinogenesis. Mol Pathol 54:138-144.
162 DRAFT - DO NOT CITE OR QUOTE
-------
Schultz, IR; Merdink, JL; Gonzalez-Leon, A; et al. (1999) Comparative toxicokinetics of chlorinated and
brominated haloacetates in F344 rats. Toxicol Appl Pharmacol 158(2): 103-114.
Schulz, WA; Steinhoff, C; Flori, AR. (2006) Methylation of endogenous human retroelements in health and disease.
Curr Top Microbiol Immunol 310:211-250.
Selmin, OI; Thorne, PA; Caldwell, PT et al. (2008) Trichloroethylene and trichloroacetic acid regulate calcium
signaling pathways in murine embryonal carcinoma cells P19. Cardiovasc Toxicol 8:47-56.
Serjeant, EP; Dempsey, B. (1979) lonization constants of organic acids in aqueous solution. IUPAC chemical data
series no. 23. New York, NY: Pergamon Press; p. 989.
Shah, YZM; Morimura, K; Yang, Q; et al. (2007) Peroxisome proliferation-activated receptor a regulates a micro-
RNA-mediated signaling cascade responsible for hepatocellular proliferation. Mol Cell Biol 27:4238-4247.
Sidebottom, H; Franklin, J. (1996) The atmospheric fate and impact of hydrochlorofluorocarbons and chlorinated
solvents. Pure Appl Chem 68(9): 1757-1769.
Singh, R. (2005a) Testicular changes in rat exposed to trichloroacetic acid (TCA) during organogenesis. Biomed
Res 16:45-52.
Singh, R. (2005b) Effect of maternal administration of trichloroacetic acid (TCA) on fetal ovary rats. Biomed Res
16:195-200.
Singh, R. (2006) Neuroembryopathic effect of trichloroacetic acid in rats exposed during organogenesis. Birth
Defects Research (Part B) 77:47-52.
Skender, L; Karacic, V; Bosner, B; et al. (1994) Assessment of urban population exposure to trichloroethylene and
tetrachloroethylene by means of biological monitoring. Arch Environ Health 49(6):445-451.
Smith, MK; Randall, JL; Read, EJ; et al. (1989) Teratogenic activity of trichloroacetic acid in the rat. Teratology
40:445-451.
Stanley, LA; Blackburn, DR; Devereaux, S; et al. (1994) Ras mutations in methylclofenapate-induced B6C3F1 and
C57BL/10J mouse liver tumours. Carcinogenesis 15:1125-1131.
Stauber, AJ; Bull, RJ. (1997) Differences in phenotype and cell replicative behavior of hepatic tumors induced by
dichloroacetate (DCA) and trichloroacetate (TCA). Toxicol Appl Pharmacol 144(2):235-246.
Styles, JA; Wyatt, I; Coutts, C. (1991) Trichloroacetic acid: studies on uptake and effects on hepatic DNA and liver
growth in mouse. Carcinogenesis 12(9):1715-1719.
Suzuki, H; Fujita, H; Mullauer, L; et al. (1990) Increased expression of c-jun gene during spontaneous
hepatocarcinogenesis in LEG rats. Cancer Lett 5 3 (2-3): 205-212.
Takashima, K; Ito, Y; Gonzalez, FJ; et al (2008) Different mechanisms of DEHP-induced hepatocellular adenoma
tumorigenesis in wild-type and PPAR-alpha-null mice. J Occup Health 50:169-180.
Tang, XJ; Li, LY; Huang, JX; et al. (2002) Guinea pig maximization test for trichloroethylene and its metabolites.
Biomed Environ Sci 15(2):113-118.
Tao, L; Li, K; Kramer, PM; et al. (1996) Loss of heterozygosity on chromosome 6 in dichloroacetic acid and
trichloroacetic acid-induced liver tumors in female B6C3Fi mice. Cancer Lett 108:257-261.
Tao, L; Kramer, PM; Ge, R; et al. (1998) Effect of dichloroacetic acid and trichloroacetic acid on DNA methylation
in liver and tumors of female B6C3FJ mice. Toxicol Sci 43:139-144.
163 DRAFT - DO NOT CITE OR QUOTE
-------
Tao, L; Yang, S; Xie, M; et al. (2000a) Effect of trichloroethylene and its metabolites, dichloroacetic acid and
trichloroacetic acid, on the methylation and expression of c-jun and c-myc protooncogenes in mouse liver:
prevention by methionine. Toxicol Sci 54:399-407.
Tao, L; Yang, S; Xie, M; et al. (2000b) Hypomethylation and overexpression of c-jun and c-myc protooncogenes
and increased DNA methyltransferase activity in dichloroacetic and trichloroacetic acid-promoted mouse liver
tumors. Cancer Lett 158:185-193.
Tao, L; Li, K; Kramer, PM; et al. (2004) Hypomethylation of DNA and the insulin-like growth factor-II gene in
dichloroacetic acid and trichloroacetic acid-promoted mouse liver tumors. Toxicology 196:127-136.
Templin, MV; Parker, JC; Bull, PJ. (1993) Relative formation of dichloroacetate and trichloroacetate from
trichloroethylene in male B6C3Fi mice. Toxicol Appl Pharmacol 123:1-8.
Templin, MV; Stevens, DK; Stenner, RD; et al. (1995) Factors affecting species differences in the kinetics of
metabolites of trichloroethylene. J Toxicol Environ Health 44:435-447.
Tharappel, JC; Nalca, A; Owens, AB; et al. (2003) Cell proliferation and apoptosis are altered in mice deficient in
the NF-kappaB p50 subunit after treatment with the peroxisome proliferator ciprofibrate. Toxicol Sci 75(2):300-
308.
Tong, Z; Board, PG; Anders, MW. (1998a) Glutathione transferase zeta catalyzes the oxygenation of the carcinogen
DCA to glyoxylic acid. Biochem J 331(2):371-374.
Tong, Z; Board, PG; Anders, MW. (1998b) Glutathione transferase zeta-catalyzed biotransformation of
dichloroacetic acid and other alpha-haloacids. Chem Res Toxicol 11:1332-1338.
Toxopeus, C; Frazier, JM. (1998) Kinetics of trichloroacetic acid and dichloroacetic acid in the isolated perfused rat
liver. Toxicol Appl Pharmacol 152:90-98.
Toxopeus, C; Frazier, JM. (2002) Simulation of trichloroacetic acid kinetics in the isolated perfused rat liver using a
biologically based kinetic model. Toxicol Sci 70(l):27-39.
Tse, Y; Ostad, A; Lee, H; et al. (1996) A clinical and histologic evaluation of two medium-depth peels: glycolic acid
versus Jessner's trichloroacetic acid. Dermatol Surg 22:781-786.
U.S. EPA (Environmental Protection Agency). (1980) Guidelines and methodology used in the preparation of health
effect assessment chapters of the consent decree water criteria documents. Federal Register 45(231):79347-79357.
U.S. EPA. (1986a) Guidelines for the health risk assessment of chemical mixtures. Federal Register
51(185):34014-34025. Available online at http://www.epa.gov/iris/backgrd.html (accessed April 20, 2010).
U.S. EPA. (1986b) Guidelines for mutagenicity risk assessment. Federal Register 51(185):34006-34012. Available
online at http://www.epa.gov/iris/backgrd.html (accessed April 20, 2010).
U.S. EPA. (1988) Recommendations for and documentation of biological values for use in risk assessment.
Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Cincinnati, OH;
EPA/600/6-87/008. PB88-179874/AS. Available online at http://www.epa.gov/iris/backgrd.html (accessed April 20,
2010).
U.S. EPA. (1991) Guidelines for developmental toxicity risk assessment. Federal Register 56(234):63798-63826.
Available online at http://www.epa.gov/iris/backgrd.html (accessed April 20, 2010).
U.S. EPA. (1992) Draft report: a cross-species scaling factor for carcinogen risk assessment based on equivalence of
mg/kg3/4/day. Federal Register 57(109):24152-24173.
U.S. EPA. (1994a) Interim policy for particle size and limit concentration issues in inhalation toxicity: notice of
availability. Federal Register 59(206):53799. Available online at http://www.epa.gov/iris/backgrd.html (accessed
April 20, 2010).
164 DRAFT - DO NOT CITE OR QUOTE
-------
U.S. EPA. (1994b) Methods for derivation of inhalation reference concentrations and application of inhalation
dosimetry. Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment,
Cincinnati, OH; EPA/600/8-90/066F. PB2000-500023. Available ,online at http://www.epa.gov/irisftackgrd.html
(accessed April 20, 2010).
U.S. EPA. (1995) Use of the benchmark dose approach in health risk assessment. Risk Assessment Forum,
Washington, DC; EPA/630/R-94/007. PB95-213765. Available online at
http://www.epa.gov/raf/publications/pdfs/BENCHMARK.PDF (accessed March 11, 2011).
U.S. EPA. (1996) Guidelines for reproductive toxicity risk assessment. Federal Register 61(212):56274-56322.
Available online at http://www.epa.gov/iris/backgrd.html (accessed April 20, 2010).
U.S. EPA. (1998) Guidelines for neurotoxicity risk assessment. Federal Register 63(93):26926-26954. Available
online at http://www.epa.gov/iris/backgrd.html (accessed April 20, 2010).
U.S. EPA. (2000a) Science policy council handbook: risk characterization. Office of Science Policy, Office of
Research and Development, Washington, DC; EPA/100-B-00-002. Available online at
http://www.epa.gov/irisftackgrd.html (accessed April 20, 2010).
U.S. EPA. (2000b) Benchmark dose technical guidance document [external review draft]. Risk Assessment Forum,
Washington, DC; EPA/630/R-00/001. Available online at http://www.epa.gov/irisftackgrd.html (accessed April 20,
2010).
U.S. EPA. (2000c) Supplementary guidance for conducting health risk assessment of chemical mixtures. Risk
Assessment Forum, Washington, DC; EPA/630/R-00/002. Available online at http://www.epa.gov/irisftackgrd.html
(accessed April 20, 2010).
U.S. EPA. (2002) A review of the reference dose concentration and reference concentration processes. Risk
Assessment Forum, Washington, DC; EPA/630/P-02/002F. Available online at
http://www.epa.gov/irisftackgrd.html (accessed March 14, 2011).
U.S. EPA. (2003) lexicological review of dichloroacetic acid. Integrated Risk Information System (IRIS), National
Center for Environmental Assessment, Washington, DC. Available online at http://www.epa.gov/iris (accessed
March 14, 2011).
U.S. EPA. (2005a) Guidelines for carcinogen risk assessment. Federal Register 70(66): 17765-18717. Available
online at http://www.epa.gov/irisftackgrd.html (accessed April 20, 2010).
U.S. EPA. (2005b) Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens. Risk
Assessment Forum, Washington, DC; EPA/630/R-03/003F. Available online at
http://www.epa.gov/irisftackgrd.html (accessed April 20, 2010).
U.S. EPA. (2005c) Drinking water addendum to the criteria document for trichloroacetic acid. Health and
Ecological Criteria Division, Office of Water, Washington, DC; EPA/822/R-05/010. Document ID EPA-HQ-OW-
2002-0043-0727; in docket EPA-HQ-OW-2002-0043. Available online at
http://www.regulations.gov/#!searchResults;rpp=10;po=0;s=EPA-HQ-OW-2002-0043-0727 (accessedMarch 14,
2011);
U.S. EPA. (2006a) Science policy council handbook: peer review. 3rd edition. Office of Science Policy, Office of
Research and Development, Washington, DC; EPA/100/B-06/002. Available online at
http://www.epa.gov/irisftackgrd.html (accessed April 20, 2010).
U.S. EPA. (2006b) A framework for assessing health risks of environmental exposures to children. National Center
for Environmental Assessment, Washington, DC; EPA/600/R-05/093F. Available online at
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=158363 (accessed March 14, 2011).
165 DRAFT - DO NOT CITE OR QUOTE
-------
U.S. EPA. (2006c) SAB review of EPA's draft risk assessment of potential human health effects associated with
perfluorooctanoic acid and its salts. Science Advisory Board , Washington, DC; EPA-SAB-06-006. Available
online at http://www.epa.gov/sab/pdf/2006_0120_fmal_draft_pfoa_report.pdf (accesssed March 14, 2011).
Vartiainen, TE; Pukkala, T; Rienoja, T; et al. (1993) Population exposure to tri- and tetrachloroethene and cancer
risk: two cases of drinking water pollution. Chemosphere 27(7):1171-1181.
Volkel, W; Friedwald, M; Lederer, E; et al. (1998) Biotransformation of perchloroethene: dose-dependent excretion
of trichloroacetic acid, dichloroacetic acid, and N-acetyl-s-(trichlorovinyl)-l-cysteine in rats and humans after
inhalation. Toxicol Appl Pharmacol 153:20-27.
Von Tungeln, LS; Yi, P; Bucci, TJ; et al. (2002) Tumorigenicity of chloral hydrate, trichloroacetic acid,
trichloroethanol, malondialdehyde, 4-hydroxy-2-nonenal, crotonaldehyde, andacroleinintheB6C3F(l) neonatal
mouse. Cancer Lett 185(1): 13-19.
Wagner, JR; Hu, CC; Ames, BN. (1992) Endogenous oxidative damage of deoxycytidine in DNA. Proc Natl Acad
Sci 89:3380-3384.
Walgren, JE; Kurtz, DT; McMillan, JM. (2000) Expression of PPAR(alpha) in human hepatocytes and activation by
trichloroacetate and dichloroacetate. Res Commun Mol Pathol Pharmacol 108(1-2): 116-132.
Walgren, JL; Kurtz, DT; McMillan, JM. (2005) Lack of direct mitogenic activity of dichloroacetate and
trichloroacetate in cultured rat hepatocytes. Toxicology 211:220-230.
Ward, JM; Hagiwara, A; Anderson, LM; et al. (1988) The chronic hepatic or renal toxicity of
di(2-ethylhexyl)phthalate, acetaminophen, sodiumbarbital, and phenobarbital in male B6C3F1 mice:
autoradiographic, immunohistochemical, and biochemical evidence for levels of DNA synthesis not associated with
carcinogenesis or tumor promotion. Toxicol Appl Pharmacol 96:494-506.
Warren, DA; Gracter, LJ; Channel, SR; et al. (2006) Trichloroethylene, trichloroacetic acid, and dichloroacetic acid:
do they affect eye development in the Sprague-Dawley rat? Int J of Toxicology 25:279-284.
Watson, RE; Goodman, JI. (2002) Effects of phenobarbital on DNA methylation in GC-rich regions of hepatic DNA
from mice that exhibit different levels of susceptibility to liver tumorigenesis. Toxicol Sci 68(1):51-58.
Weber, E; Moore, MA; Bannasch, P. (1988) Enzyme histochemical and morphological phenotype of amphophilic
foci and amphophilic/tigroid cell adenomas in rat liver after combined treatment with dehydroepiandrosterone and
N-nitrosomorpholine. Carcinogenesis 9(6): 1049-1054.
Werner, H; Le Roith, D. (2000) New concepts in regulation and function of the insulin-like growth factors:
implications for understanding normal growth and neoplasia. Cell Mol Life Sci 57:932-942.
Wilson, JD; Brown, CB; Walker, PP. (2001) Factors involved in clearance of genital warts. Int J STD AIDS
12:789-792.
Witheiler, D; Lawrence, N; Cox, SL; et al. (1996) Facial actinic keratoses (AK) treated with Jessner's solution (JD)
and 35% trichloroacetic widespread acid (TCA) peel vs 5% fluorouracil (5-FU) cream: long-term. Dermatol Surg
22:807-815.
Woods, CG; Burns, AM; Bradford, BU; et al. (2007a) Wy-14643-induced cell proliferation and oxidative stress in
mouse liver are independent of NADPH oxidase. Toxicol Sci 98(2):366-374.
Xu, G; Stevens, DK; Bull, RJ. (1995) Metabolism of bromodichloroacetate in B6C3F! mice. Drug Metab Disp
23(12):1412-1416.
Yamada, Y; Jackson-Grusby, L; Linhart, H; et al. (2005) Opposing effects of DNA hypomethylation on intestinal
and liver carcinogenesis. Proc Natl Acad Sci U S A 102(38):13580-13585.
166 DRAFT - DO NOT CITE OR QUOTE
-------
Yang, Q; Ito, S; Gonzalez, FJ. (2007) Hepatocyte-restricted constitutive activation of PPARa induces
hepatoproliferationbut not hepatocarcinogenesis. Carcinogenesis 28:1171-1177.
Yeldandi, AV; Milano, M; Subbarao, V; et al. (1989) Evaluation of liver cell proliferation during ciprofibrate-
induced hepatocarcinogenesis. Cancer Lett 47:21-27.
Yokoyama, Y; Tsuchida, S; Hatayama, I; et al. (1993) Lack of peroxisomal enzyme inducibility in rat hepatic
preneoplastic lesions induced by mutagenic carcinogens: contrasted expression of glutathione S-transferase P form
and enoyl CoA hydratase. Carcinogenesis 14(3):393-398.
Yu, KO; Barton, HA; Mahle, DA; et al. (2000) In vivo kinetics of trichloroacetate in male Fisher 344 rats. Toxicol
Sci 54:302-311.
Zhang, W; Dahlberg, JE; Tarn, W. (2007) MicroRNAs in tumorigenesis: a primer. Am J Pathol 171(3):728-738.
Ziglio, GG. (1981) Human exposure to environmental trichloroethylene and tetrachloroethylene: preliminary data on
population groups of Milan, Italy. Bull Environ Contam Toxicol 26:131-136.
Ziglio, GG; Fara, GM; Beltramelli, G; et al. (1983) Human environmental exposure to trichloro- and
tetrachloroethylene from water and air in Milan, Italy. Arch Environ Contam Toxicol 12:57-64.
REFERENCES ADDED AFTER EXTERNAL PEER REVIEW
Allen, BC; Fisher, JW. (1993) Pharmacokinetic modeling of trichloroethylene and trichloroacetic acid in humans.
Risk Anal 13:71-86.
Aranda, A; Pascual, A. (2001) Nuclear hormone receptors and gene expression. Physiol Rev 81 (3): 1269-1304.
Breimer, DD; Ketelaars, HCJ; Van Rossum, JM. (1974) Gas chromatographic determination of chloral hydrate,
trichloroethanol and trichloroacetic acid in blood and in urine employing head-space analysis. J Chromatogr 88:55-
63.
Burnham, KP., and Anderson DR. 2002. Model selection and multimodel inference: a practical information-
theoretic approach. Springer-Verlag, New York, NY.488pp. ISBN 0-387-95364-7.
Cheng, X; Klaassen, CD. (2008). Critical role of PPAR-alpha in perfluorooctanoic acid- and perfluorodecanoic acid-
induced downregulation of Oatp uptake transporters in mouse livers. Toxicol Sci 106:37-45.
Gorton, JC. (2008) Evaluation of the role of peroxisome proliferator-activated receptor a (PPARa) in mouse liver
tumor induction by trichloroethylene and metabolites. Crit Rev Toxicol 38:857-875.
Decker, K. (1990) Biologically active products of stimulated liver macrophages (Kupffer cells). Eur J Biochem
192:245-261.
Dreyer, C; Krey, G; Keller, H; et al. (1992) Control of the peroxisomal beta-oxidative pathway by a novel family of
nuclear hormone receptors. Cell 68:879-887.
Escher, P; Wahli, W. (2000) Peroxisome proliferator-activated receptors: Insights into multiple cellular functions.
Mutat Res 448:121-138.
Eveillard, A; Mselli-Lakhal, L; Mogha, A; et al. (2009) Di-(2-ethylhexyl)-phthalate (DEHP) activates the
constitutive androstane receptor (CAR): a novel signalling pathway sensitive to phthalates. Biochem Pharmacol
77(11): 1735-1746.
Gonzalez, FJ; Shah, YM. (2008) PPARa: mechanism of species differences and hepatocarcinogenesis of peroxisome
proliferators. Toxicology 246:2-8.
167 DRAFT - DO NOT CITE OR QUOTE
-------
Gottlicher, M; Widmark, E; Li, Q; et al. (1992) Fatty acids activate a chimera of the clofibric acid-activated receptor
and the glucocorticoid receptor. Proc Natl Acad Sci 89:4653-4657.
Guo, D; Sarkar, J; Suino-Powell, K; et al. (2007) Induction of nuclear translocation of constitutive androstane
receptor by peroxisome proliferator-activated receptor alpha synthetic ligands in mouse liver. J Biol Chem
14:36766-776.
Guyton KZ, Chiu WA, Bateson TF; et al. (2009) A reexamination of the PPAR-alpha activation mode of action as a
basis for assessing human cancer risks of environmental contaminants. Environ. Health Perspect. 117: 1664-1672
Guyton KZ, Barone S Jr, Brown RC; et al. (2009) Mode of action frameworks: a critical analysis. J Toxicol Environ
Health B Crit Rev. 11(1): 16-31
Hasmall, SC; James, NH; Roberts, RA. (2000b) Suppression of mouse hepatocyte apoptosis by peroxisome
proliferators: Role of PPARa and TNFa. Mutat Res 448:193-200.
Hassoun, EA; Dey, S. (2008) Dichloroacetate- and trichloroacetate-induced phagocytic activation and production of
oxidative stress in the hepatic tissues of mice after acute exposure. J Biochem Mol Toxicol 22(l):27-34.
Hassoun, EA; Spildener, J; Cearfoss, J. (2010a) The induction of tumor necrosis factor-alpha, superoxide anion,
myeloperoxidase, and superoxide dismutase in the peritoneal lavage cells of mice after prolonged exposure to
dichloroacetate and trichloroacetate. J Biochem Mol Toxicol 24(2):136-144.
Hassoun, EA; Cearfoss, J; Spildener, J. (2010b) Dichloroacetate- and trichloroacetate-induced oxidative stress in the
hepatic tissues of mice after long-term exposure. J Appl Toxicol 30(5):450-456.
I ARC (International Agency for Research on Cancer). (2004) Trichloroacetic acid. In: I ARC Monographs on the
evaluation of carcinogenic risks to humans. Vol 84. Some drinking-water disinfectants and contaminants, including
arsenic. Lyon, France: World Health Organization, p. 403-440.
Issemann, I; Green, S. (1990) Activation of a member of the steroid hormone receptor superfamily by peroxisome
proliferators. Nature 347:645-650.
King-Herbert A; Thayer K. (2006) NTP workshop: animal models for the NTP rodent cancer bioassay: stocks and
strains - should we switch? Toxicol Path 34:802-805.
Kohle, C; Schwarz, M; Bock, KW. (2008) Promotion of hepatocarcinogenesis in humans and animal models. Arch
Toxicol 82:623-631.
Moore, MM; Honma, M; Clements, J; et al. (2006) Mouse lymphoma thymidine kinase gene mutation assay:
follow-up meeting of the International Workshop on Genotoxicity Testing-Aberdeen, Scotland, 2003-assay
acceptance criteria, positive controls, and data evaluation. Envir Mol Mutagen 47:1-5.
Muller, G; Spassovski, M; Henschler, D. (1972) Trichloroethylene exposure and trichloroethylene metabolites in
urine and blood. Arch Toxikol 29:335-340.
Muller, G; Spassovaki, M; Henschler, D. (1974) Metabolism of trichloroethylene in man. II. Pharmacokinetics of
metabolites. Arch Toxicol 32:283-295.
Nakamura T; Ito Y; Yanagiba Y et al. (2009) Microgram-order ammonium perfluorooctanoate may activate mouse
peroxisome proliferator-activated receptor a, but not human PPARa. Toxicology 265(1-2): 27-33.
NRC (National Research Council). (2010) Review of the Environmental Protection Agency's draft IRIS assessment
of tetrachloroethylene. Washington, DC. National Academies Press.
Peters, JM; Rusyn, I; Rose, ML; et al. (2000) Peroxisome proliferator-activated receptor alpha is restricted to hepatic
parenchymal cells, not Kupffer cells: Implications for the mechanism of action of peroxisome proliferators in
hepatocarcinogenesis. Carcinogenesis 21:823-826.
168 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION
The Toxicological Review of Trichloroacetic Acid (dated September 2009) has undergone
a formal external peer review performed by scientists in accordance with the EPA guidance on
peer review (U.S. EPA, 2006a, 2000a). An external peer review workshop was held December
10, 2009. The external peer reviewers were tasked with providing written answers to general
questions on the overall assessment and on chemical-specific questions in areas of scientific
controversy or uncertainty. A summary of significant comments made by the external reviewers
and EPA's responses to these comments arranged by charge question follow. In many cases the
comments of the individual reviewers have been synthesized and paraphrased in development of
Appendix A. EPA received no scientific comments on this assessment from the public.
EXTERNAL PEER REVIEW PANEL COMMENTS
The reviewers made several editorial suggestions to clarify specific portions of the text.
These changes were incorporated in the document as appropriate and are not discussed further.
When the external peer reviewers commented on decisions and analyses in the Toxicological
Review under multiple charge questions, these comments were organized under the most
appropriate charge question.
General Comments
1. Is the Toxicological Review logical, clear and concise? Has EPA clearly synthesized the
scientific evidence for noncancer and cancer hazard?
Comments: Reviewers generally considered the Toxicological Review to be logical and clear.
One reviewer observed that the document would benefit from less use of acronyms. Two
reviewers observed that the document was not concise, largely as a function of the complexity of
the subject and the standard structure of the Toxicological Review.
Response: The content of the Toxicological Review of Trichloroacetic Acid \$ consistent with
the current outline for IRIS Toxicological Reviews; however, the Toxicological Review was
revised as much as possible to streamline the document and reduce redundancy. Acronyms used
infrequently in the Toxicological Review were spelled out.
Comments: Three reviewers offered comments on the MOA section. One reviewer considered
the MOA section difficult to follow. A second reviewer considered the issues associated with
the PPARa activation MOA to be adequately described, but suggested that a table showing
A-1 DRAFT - DO NOT CITE OR QUOTE
-------
consistencies/inconsistencies and data gaps regarding this MOA would provide greater clarity for
the current review of TCA as well as for future assessments of other potential peroxisome
proliferators. A third reviewer commented that the conclusions derived from the literature
review were speculative, especially with respect to MO As other than peroxisome proliferation.
Response: These comments regarding the MOA section are addressed in EPA's response to
comments for Charge Questions C2 and C3.
Comments: One reviewer suggested that additional justification be provided for the selection of
dose-response models (when multiple models provided an adequate fit) and a BMR of 10%. The
same reviewer also questioned why human equivalent doses were not estimated in the derivation
of the oral RfD.
Response: These comments are addressed in EPA's response to comments for Charge Question
A3.
2. Please identify any additional studies that should be considered in the assessment of the
noncancer and cancer health effects of TCA.
Comments: Five reviewers did not identify any additional studies. One reviewer recommended
that the following studies be included:
Allen, B; Fisher, J. (1993) Pharmacokinetic modeling of trichloroethylene and trichloroacetic acid in humans. Risk
Anal 13:71-86.
Breimer, DD; Ketelaars, HCJ; Van Rossum, JM. (1974) Gas chromatographic determination of chloral hydrate,
trichloroethanol and trichloroacetic acid in blood and in urine employing head-space analysis. J Chromatography
88:55-63.
Muller, G; Spassovaki, M; Henschler, D. (1972) Trichloroethylene exposure and trichloroethylene metabolites in
urine and blood. Arch Toxikol 29:335-340.
Muller, G; Spassovaki, M; Henschler, D. (1974) Metabolism of trichloroethylene in man. II. Pharmacokinetics of
metabolites. ArchToxicol 32:283-295.
Response: Information from these studies was added to Sections 3.4 and 3.5 of the
Toxicological Review.
Comments: Four reviewers recommended updating the literature on the MOA of PPARa
agonists. Specific papers that the reviewers suggested be added or expanded upon were:
Elcombe, CR. (1985) Species differences in carcinogenicity and peroxisome proliferation due to trichloroethylene: a
biochemical human hazard assessment. Arch Toxicol Suppl 8:6-17.
A-2 DRAFT - DO NOT CITE OR QUOTE
-------
Ren, H; Akeksunes, LM; Wood, C; et al. (2010) Characterization of peroxisome proliferator-activated receptor a
(PPARa) - independent effects of PPARa activators in the rodent liver: Di-(2-ethylhexyl) phthalate also activates
the constitutive activated receptor. Toxicol Sci 113:45-59.
One reviewer observed that the toxicology of TCA and dieldrin were similar; both are PPARa
agonists, peroxisome proliferators, and cause liver tumors in mice but not rats. One reviewer
commented that the document needs to include literature pertaining to the histopathology and
molecular biology of the tumors induced by other PPARa agonists and a discussion of the
similarity between these tumors and those found in TCA-treated mice. This reviewer suggested
that the following review articles be included:
Gorton, JC. (2008) Evaluation of the role of peroxisome proliferator-activated receptor a (PPARa) in mouse liver
tumor induction by trichloroethylene and metabolites. Crit Rev Toxicol 38:857-875.
Gonzalez, FJ; Shah, YM. (2008) PPARa: mechanism of species differences and hepatocarcinogenesis of peroxisome
proliferators. Toxicology 246:2-8.
Kohle, C; Schwarz, M; Bock, KW. (2008) Promotion of hepatocarcinogenesis in humans and animal models. Arch
Toxicol 82:623-631.
International Agency for Research on Cancer (IARC). (2004) IARC Monographs on the evaluation of carcinogenic
risks to humans. Vol 84. Some drinking-water disinfectants and contaminants, including arsenic, p. 403-440.
Response: The description of the study findings of Elcombe (1985) in Section 4.2.1.1.1 was
expanded. Information from the Ren et al. (2010) study related to PPARa agonism was added to
Section 4.7.3.1.1. The studies on dieldrin were not included. Unlike dieldrin, TCA exposure
does not result in mammary gland tumors in rats and mice. The four review papers identified by
the reviewers were not included; however, original key studies from these papers that are
relevant to the assessment of TCA are discussed in the Toxicological Review. Other comments
on revision of the MO A section are addressed in EPA's response to comments for Charge
Questions C2 and C3.
Comments: One reviewer asked that data from Study 2 of DeAngelo et al. (2008) be included in
the cancer assessment.
Response: All three studies presented in DeAngelo et al. (2008) were included in the cancer
assessment. The data from Study 2 of DeAngelo et al. (2008) were included in the time-to-
tumor analysis found in Section 5.4.
Comments: One reviewer recommended that the maximum possible exposure of TCA to
humans be presented in the Toxicological Review. This reviewer stated that if maximum
A-3 DRAFT - DO NOT CITE OR QUOTE
-------
possible exposure to humans is below the calculated cancer risk, then TCA should be classified
as noncarcinogenic to humans following environmental exposure.
Response: In general, the scope of an IRIS assessment is the evaluation of hazard and dose-
response analysis for a chemical substance. An exposure analysis has not been performed as a
part of the development of this Toxicological Review. The recommendation offered by this
reviewer involves risk characterization, which is not within the scope of an IRIS assessment. As
a general rule, the magnitude of risk at a given level of exposure to a chemical carcinogen is not
a factor used in assigning the cancer descriptor according to EPA's Cancer Guidelines (U.S.
EPA, 2005a).
Chemical-Specific Charge Questions
(A) Oral Reference Dose (RfD) for Trichloroacetic Acid
1. A 60-week drinking water study in mice (DeAngelo et al., 2008) was selected as the basis
for derivation of the RfD for TCA. Please comment on whether the selection of DeAngelo
et al. (2008) as the principal study is scientifically justified. Please identify and provide the
rationale for any other studies that should be selected as the principal study.
Comments: The reviewers generally agreed with the selection of the 60-week study in male
B6C3Fi mice by DeAngelo et al. (2008) as the principal study for RfD derivation. One reviewer
commented that a deficiency of DeAngelo et al. (2008) is that complete histopathologic
examinations were reportedly performed on only five mice from the high-dose and control
groups, and recommended that EPA clarify with the study authors the extent of histopathologic
examinations that were performed at the interim sacrifices and at the termination of the 60-week
study. This reviewer also recommended further discussion of study limitations (i.e., that effects
at sites other than those examined microscopically or in female mice might have been missed).
Two reviewers commented that the developmental study by Smith et al. (1989) should also be
considered, and a third reviewer observed that the Smith et al. (1989) study provided sufficient
information for deriving an RfD.
Response: EPA agrees with the majority of the reviewers in retaining DeAngelo et al. (2008) as
the principal study. As stated on page 1058 of DeAngelo et al. (2008), gross lesions, liver,
kidney, spleen, and testis were examined by a board-certified veterinary pathologist at the
interim and terminal necropsies. For all other tissues, a complete pathologic examination was
performed on five mice from the high-dose and control groups. If the number of any
histopathologic lesion in a tissue was significantly increased above that in the control animals,
then that tissue was examined in all TCA dose groups. This information was included in Section
A-4 DRAFT - DO NOT CITE OR QUOTE
-------
4.2.2.1.2.1. A discussion of limitations in the DeAngelo et al. (2008) study was added to
Sections 5.1.3 and 5.3. Data from Smith et al. (1989) were also modeled to compare the results
with that obtained by using the DeAngelo et al. (2008) study as the basis for the RfD (Section
5.1.2.2).
2. Liver toxicity (hepatocellular necrosis) was selected as the critical effect for the
determination of the POD. Please comment on whether the selection of this critical effect is
scientifically justified. Please identify and provide the rationale for any other endpoints
that should be considered in the selection of the critical effect.
Comments: Five reviewers agreed with the selection of hepatocellular necrosis as the critical
effect. One of these reviewers provided qualified support, noting that hepatocellular necrosis is
an appropriate endpoint to be used as a critical effect, assuming that the MOA by which it occurs
is relevant to humans. This reviewer further observed that if necrosis in mice is a consequence
of a PPARa agonism MOA leading to tumors, then one or more of the endpoints identified in the
long-term rat study, testicular effects in mice, or developmental effects in rats would be more
appropriate. A second of these reviewers recommended that both liver and developmental
toxicity endpoints, which yielded similar candidate RfDs, be emphasized throughout the review
to strengthen the confidence in the final RfD. Two reviewers (including one who supported
hepatocellular necrosis as a critical effect) also recommended consideration of increased liver
weight as a candidate critical effect because continuous data often are more sensitive than
quantal data and liver weight is less subjective a measure than pathologist ratings. However, one
of the reviewers determined that the BMDLio based on the liver weight endpoint was 58 mg/kg-
day, which was higher than the BMDLio of 18 mg/kg-day for hepatocellular necrosis.
One reviewer did not consider hepatocellular necrosis to be the best choice for the critical effect
because similar mild centrilobular necrosis has been reported for some PPARa agonists. This
reviewer proposed the use of testicular tubular degeneration in mice as the critical effect with
alternative UFs applied to both hepatocelluar necrosis and testicular tubular degeneration such
that testicular effects would yield a lower (more sensitive) RfD.
Finally, several reviewers suggested that markers of peroxisome proliferation be used as
the critical effect. One of the reviewers commented that evaluation of necrosis was subjective,
and that a more appropriate endpoint would be one with a more dynamic range and relevant to
the MOA (e.g., a marker of peroxisome proliferation). This reviewer suggested cyanide-
insensitive PCO activity as a possible critical effect because it shows consistent dose- and time-
responsive changes and because it is amenable to statistical analysis. A second of these
reviewers suggested a biomarker of peroxisome proliferation as a possible critical effect because
it represents a more upstream effect in the mechanistic chain of responses.
A-5 DRAFT - DO NOT CITE OR QUOTE
-------
Response: EPA agrees with the majority of the reviewers who suggested hepatocellular necrosis
as the critical effect. In response to the reviewers suggestions, developmental endpoints, i.e.,
fetal body weight and crown rump length, were remodeled as continuous data sets (Appendix C,
Sections C.3 and C.4) using 5% extra risk as the BMR. The lowest candidate POD derived from
these two endpoints was 84 mg/kg-day (fetal body weight), indicating that these endpoints
represented less sensitive effects than the hepatocellular necrosis that was used as the critical
endpoint for the derivation of the POD. The BMDLio for testicular tubular degeneration in mice
was calculated to be 127 mg/kg-day. This information has also been added. This value is also
higher than the potential PODs based on any of the observed liver effects. EPA concurs with the
BMDLio of 58 mg/kg-day for liver weight data reported by one of the reviewers, and that a
comparison of the BMDLioS for liver weight and liver necrosis indicate that liver weight is less
sensitive. Of the four endpoints pertaining to liver effects, liver inflammation, liver weight,
cyanide-insensitive PCO activity, and liver necrosis, necrosis remains the most sensitive
endpoint. The comments related to selection of the POD and UFs are addressed in response to
comments on Charge Questions A3 and A4.
Cyanide-insensitive PCO activity was added as a candidate critical effect, and dose-
response modeling was performed using BMD modeling methods (BMDS, version 2.1.1). The
results of this BMD modeling were added to Appendix B and to a summary of the analysis to
Section 5.1.2.1. These continuous data were best fit using a second-degree polynomial model,
which yielded a BMDLiso of 21 mg/kg-day. The candidate POD associated with cyanide-
insensitive PCO is higher than the BMDLio of 18 mg/kg-day for hepatocellular necrosis.
Therefore, hepatocellular necrosis was retained as the critical effect.
3. Benchmark dose (BMD) modeling was conducted on the liver and testicular effects in
male mice exposed to TCA in the drinking water study by DeAngelo et al. (2008) in order
to determine the POD. Has the BMD modeling been appropriately conducted? Is the
benchmark response (BMR) selected for use in deriving the POD (i.e., 10% extra risk of
hepatocellular necrosis) scientifically justified? Please identify and provide the rationale
for any alternative approaches (including the selection of the BMR, model, etc.) for the
determination of the POD and discuss whether such approaches are preferred to EPA's
approach.
Comments: The reviewers generally agreed that the modeling was appropriately conducted.
Two reviewers commented that modeling should also be conducted on one or more of the
endpoints that identified the NOAEL in the long-term rat drinking water study. Three reviewers
recommended conducting BMD modeling of continuous endpoints, including liver weight, from
the principal study in the mouse.
A-6 DRAFT - DO NOT CITE OR QUOTE
-------
Response: EPA considered conducting dose-response modeling for the other endpoints
suggested by the reviewers. Data from the rat drinking water study (DeAngelo et al., 1997) were
not used to derive a candidate POD because the NOAEL (32.5 mg/kg-day) and LOAEL (364
mg/kg-day) identified in this study were higher than those identified in the chronic mouse study
(DeAngelo et al., 2008). Comments on the consideration of continuous endpoints as the basis for
the RfD are addressed in response to Charge Question A2.
Comments: One reviewer specifically agreed with defining the BMR as an extra risk of 10%.
Two reviewers commented that better justification for the selection of a 10% BMR was needed.
Two reviewers recommended the use of a BMR of 5% on the grounds that the BMD is intended
to approximate the NOAEL and that the BMR level should be close to the lower end of the
observed data. In this particular case, a reviewer noted that the BMR0s is within the range of
doses for which observations are reported.
Response: The justification provided for selection of a 10% BMR in Section 5.1.2.1 was
reconsidered. Reference to EPA's BMD technical guidance in support of selection of this BMR
was added to Section 5.1.2.1.
Comments: Three reviewers suggested that the Toxicological Review be revised to provide
clearer documentation of guidance used for model selection. One reviewer commented that
minor differences in AIC are not meaningful and should not be over-interpreted.
Response: The procedures used in the selection of the best-fit model are described in Section
5.1.2.1 and are consistent with the guidance in EP'A's Benchmark Dose Technical Guidance
Document (U.S. EPA, 2000b). With regard to the use of AICs in model selection, the
application of the BMD approach recognizes that more than one model can adequately fit a dose-
response data set. Thus, a reasonable and practical procedure for selecting amongst these models
is needed in order to support estimation of a BMD and BMDL. The recommended procedure
described in Section 5.1.2.1 of selecting the model with the lowest AIC when the BMDLs from
all adequately fitting alternative models are within a factor of 3 of each other provides a clear-cut
decision rule to arrive at a needed BMDL value — under circumstances where the model choice
does not have a large impact on the POD.
Comments: One reviewer observed that DeAngelo et al. (2008) reported severity as well as
incidence data for the various histological endpoints, and suggested that EPA consider the
possibility of treating the overall incidence and severity data as a pseudo-continuous variable.
One reviewer observed that in Study 1 of DeAngelo et al. (2008) at the 60-week time point, the
A-7 DRAFT - DO NOT CITE OR QUOTE
-------
low-dose group (0.05 g/L TCA) showed a considerably higher incidence and greater severity of
hepatocellular cytoplasmic alterations than the mid-dose group (0.5 g/L), but incidence and
severity of hepatocellular inflammation and testicular tubular degeneration were reported as zero.
This raised the question of whether there was something anomalous in the treatment or analysis
of the low-dose group at 60 weeks and, by extension, of the hepatocellular necrosis findings in
this group at 30-45 weeks.
Response: Hepatocellular necrosis was reported in mid- and high-dose mice in the DeAngelo et
al. (2008) study. Average severity in these groups was characterized as minimal to mild; data on
severity of necrosis in individual animals were not provided. In light of the similarity in severity
scores in the two groups and the lack of more detailed severity information, a more rigorous
analysis using both incidence and severity data would not be supported. The lack of a monotonic
dose-related increase in the incidence of centrilobular cytoplasmic alteration is acknowledged
and was added to Section 5.1.2.1. It should be noted that this endpoint was not considered as a
candidate data set for derivation of the RfD. As the peer reviewer observed, anomalous dose-
response findings may simply reflect normal fluctuations in response and, in this instance, did
not necessarily reduce confidence in other data sets from this study.
Comments: One reviewer commented that the BMD modeling in the Toxicological Review
should reflect the latest version of BMDS.
Response: EPA BMDS version 1.4.1 was used in the analysis of dichotomous data sets. BMDS
version 2.0 includes more functions, but the core calculation remains the same, and as such, use
of the newer version would not change the modeling results. Therefore, the modeling based on
BMDS version 1.4.1 was retained.
Comments: One reviewer stated that because hepatocellular effects in the mouse result from
peroxisome proliferation, these effects do not represent a toxic response in humans and as such,
the UF for mouse-to-human extrapolation should be 1 and not 10. This reviewer further
observed that the POD for testicular tubular degeneration divided by an UF of 10 would result in
a lower RfD (0.127 mg/kg-day) than that derived for liver inflammation using an UF of 1
(0.18mg/kg-day).
Response: The conclusion that the available data do not support a determination that TCA
induces all hepatocellular effects solely via peroxisome proliferation has been retained.
Therefore, all liver endpoints have been considered for selection of the critical effect.
Responses related to this issue are also addressed in Charge Question A2 and comments related
to UFs are addressed in response to Charge Question A4.
A-8 DRAFT - DO NOT CITE OR QUOTE
-------
4. Please comment on the rationale for the selection of the uncertainty factors applied to
the POD for the derivation of the RfD. If changes to the selected uncertainty factors are
proposed, please identify and provide a rationale(s).
Comments: Eight reviewers agreed with the UF for interspecies extrapolation of 10. One
reviewer did not consider hepatocellular effects originating from peroxisome proliferation to be
related to a toxic response in humans and that a more appropriate UF for mouse-to-human
extrapolation would be 1 or 3.
Response: EPA agrees with the majority of the reviewers in applying an UF of 10 for
interspecies extrapolation. Insufficient information is currently available to assess mouse-to-
human differences in TCA toxicokinetics or toxicodynamics. The MOA by which TCA induces
liver toxicity is complex; while peroxisome proliferation is likely a component of the overall
mechanism of toxicity, the final conclusion is that it is not established as the sole MOA of TCA
toxicity. Reduction of the UF based on issues of human relevance was considered but was not
thought to be supported.
Comments: Eight reviewers agreed with the UF for human variation of 10. One reviewer
commented that the use of an UF of 3 would be better justified because of limited TCA
metabolism and a peroxisome proliferation MOA that suggests that human susceptibility to TCA
would not vary significantly. Another reviewer commented that a number of publications and
guidance documents suggest that the default value of 10 for human variability is insufficient.
particularly when the range of human metabolic capabilities and the need to protect children and
other sensitive subpopulations are considered; however, this reviewer further observed that
because the extent of TCA metabolism is minor and not a major determinant of TCA toxicity or
clearance, an UF for human diversity of 10 is probably sufficient.
Response: EPA agrees with the majority of the reviewers in applying an UF of 10 for human
variation, including protection of sensitive subpopulations, in light of the lack of information to
support a chemical-specific factor.
Comments: Eight reviewers agreed with the database UF of 10. Two reviewers commented that
the discussion of deficiencies in the database should be expanded to consider limitations of the
available developmental studies, the failure of the principal study (DeAngelo et al., 2008) to
include female mice, and questions about the completeness of the histopathological
examinations. One reviewer stated that a case could be made for the use of a database UF of 1 or
A-9 DRAFT - DO NOT CITE OR QUOTE
-------
3 rather than 10 because the DeAngelo et al. (2008, 1997) studies were conducted in two species
(mice and rats) and developmental data are available.
Response: EPA agrees with the majority of the reviewers in applying a database UF of 10. The
justification for the database UF in Section 5.1.3 was expanded to include the additional database
deficiencies identified by the reviewers.
Comments: Most reviewers agreed with the application of a LOAEL-to-NOAEL UF of 1. Two
reviewers commented on the LOAEL-to-NOAEL UF in the context of selection of the BMR.
One of these reviewers suggested an additional UF of 3-10 be applied with a BMR of 10% extra
risk. The second reviewer recommended the use of a BMDLos instead of a BMDLio, but further
stated that if the BMDLio was retained, a LOAEL-to-NOAEL UF is required.
Response: EPA agrees with the majority of the reviewers and has retained an UF UF of 1 for
LOAEL-to-NOAEL extrapolation since BMD modeling was used to estimate the POD. In
response to the reviewers who made alternative recommendations, it may be helpful to note that
EPA considers biological significance of the endpoint when selecting a BMR and chooses a
BMR that is intended to represent a minimally biologically significant change, if data are
available to inform the choice as per EPA Guidance (U.S. EPA, 2000). When data are not
available or informative, as is generally the case for TCA, a 10% BMR is chosen for consistency
across endpoints.
Comments: One reviewer discussed evidence for inter-strain differences in adsorption,
distribution, metabolism, and excretion and response to chemicals in mice and recommended an
UF for within-mouse (or rat) variability.
Response: EPA recognizes the variability within and among mouse (and rat) strains; however,
information to characterize the magnitude of this variability, particular for TCA, is not available.
The general practice of selecting the most sensitive species and strain tested from among
adequately conducted studies reduces the uncertainty associated with this variability. For these
reasons, an additional UF for inter-strain variability was not included in the derivation of the
RfD.
A-10 DRAFT - DO NOT CITE OR QUOTE
-------
(B) Inhalation Reference Concentration (RfC) for Trichloroacetic Acid
1. An RfC was not derived for TCA. Has the scientific justification for not deriving an
RfC been clearly described in the document? Please identify and provide the rationale for
any studies that should be selected as the principal study.
Comments: Four reviewers agreed that not deriving an RfC for TCA was justified. Three
reviewers commented that the decision for not deriving an RfC should be better justified (e.g., by
using EPA's 1994 Methods for Derivation of Inhalation Reference Concentrations and
Application of Inhalation Dosimetry). Two reviewers acknowledged that there is no sophisticated
PBPK model for TCA exposure by inhalation or other exposure routes. However, these reviewers
suggested that an RfC could be derived by route-to-route extrapolation from oral data and the use
of a simple set of assumptions such as 100% absorption by inhalation, followed by systemic
distribution via the bloodstream. One reviewer commented that given the possibility of
respiratory-specific toxicity and the lack of data that can address respiratory-specific toxicity, it
may be more appropriate to simply provide the exposure relationship that would allow an
estimate of cumulative (ingestion plus inhalation) exposure under standard exposure assumptions.
Response: No inhalation studies of TCA are available. Available information suggests that a
simple set of assumptions might not account for possible differences in route-specific toxicity.
The respiratory tract has not been studied in the available oral studies. In addition, the liver is
the critical target organ for oral toxicity, and a first-pass effect by the liver is expected following
oral administration. Consistent with guidance for route-to-route extrapolation in EPA's RfC
Methodology (p. 4-5) (U.S. EPA, 1994b), which specifies that PBPK modeling is the preferred
methodology for extrapolation, simple route-to-route extrapolation was not performed and an
RfC for TCA was not derived. Justification for not deriving an RfC for TCA in Section 5.2 was
expanded.
(C) Carcinogenicity of Trichloroacetic Acid
1. Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment
(www.epa.gov/iris/backgr-d.htm), the Agency concluded that TCA is "likely to be
carcinogenic to humans" by all routes of exposure. Please comment on the cancer weight
of evidence characterization. Is the weight of evidence characterization scientifically
justified?
Comments: Four reviewers commented that TCA meets the criteria for the characterization of
"likely to be carcinogenic to human." Two of these reviewers, while considering the descriptor
of "likely" to be justified, also considered the case to be relatively weak. One of these reviewers
A-l 1 DRAFT - DO NOT CITE OR QUOTE
-------
observed that TCA is one of the weakest among the chemicals in the IRIS database with this
descriptor, and recommended that EPA provide some perspective on the extent of evidence
relative to other chemicals that share this descriptor. The second of these reviewers observed
that TCA tested positive in only one strain of one species; that the tumor response, while
statistically significant, was not extraordinary; that liver tumors are relatively common in
rodents; and that supporting evidence for events associated with tumor formation were somewhat
speculative. One reviewer observed that the "likely" descriptor is largely based on lack of
evidence to the contrary. Another reviewer commented that the "likely" descriptor appears to be
appropriate for high doses.
Three reviewers disagreed with the characterization of "likely to be carcinogenic to
humans." One of these reviewers commented that the classification for TCA is at best
"suggestive evidence of carcinogenic potential" but should more likely be "not likely to be
carcinogenic to humans." A second of the reviewers commented that TCA is not likely to be
carcinogenic to humans because the increase in liver tumors occurred in a strain of mice with a
high background incidence, TCA did not induce tumors in rats, and TCA induces tumors by a
nongenotoxic MOA related to peroxisome proliferation that is not relevant to humans. A third
reviewer did not consider the characterization of "likely" to be justified, noting that evidence of
carcinogenicity in multiple studies in a single species (mouse) and both sexes, with lack of
concordance in the rat, was not adequate biological support. This reviewer did not suggest an
alternative descriptor.
Response: As noted in EPA's Guidelines for Carcinogen Risk Assessment ("Cancer
Guidelines"), adequate evidence consistent with the descriptor of "likely to be carcinogenic to
humans" covers a broad spectrum, and a broad range of data combinations exists that are covered
by this descriptor. As discussed in Section 4.7, the carcinogenicity of TCA has been
demonstrated in mice of both sexes and in multiple studies after only 51 weeks of exposure.
According to the Cancer Guidelines, data supporting the descriptor "likely to be carcinogenic to
humans" may include positive findings in animal experiments in more than one sex or early age
at the onset of response, with or without evidence of carcinogenicity in humans. While the
carcinogenicity data for TCA fall within the broad spectrum consistent with the descriptor
"likely to be carcinogenic to humans," EPA agrees that the strength of the available data places
TCA at the lower end of this spectrum. Evidence considered in assigning the descriptor of
"likely" included:
Liver tumor data in the mouse. As some peer reviewers observed, TCA induced tumors at only
one site (liver) in a strain of mouse with a high background rate of liver tumors. EPA recognizes
that the interpretation of liver tumor findings from the B6C3Fi mouse has been the subject of
controversy because this strain of mouse is relatively susceptible to liver tumors and the
background incidence of this tumor is relatively high (King-Herbert and Thayer, 2006). Because
A-12 DRAFT - DO NOT CITE OR QUOTE
-------
the majority of TCA drinking water bioassays in the B6C3Fi mouse involved less-than-lifetime
exposures (DeAngelo et al., 2008; Bull et al., 2002, 1990; Pereira, 1996), the control incidence of
liver tumors (adenomas and carcinomas combined) was relatively low (0 to 13%) such that there
was no elevated background incidence of liver tumors in these studies to confound identification
of TCA-related liver tumor induction. A high incidence of liver tumors in the control group was
observed only in the DeAngelo et al. (2008) 104-week study (see Tables 5-7, 5-8, 5-9, 5-10 and
5-11). Furthermore, it is noteworthy that statistically signicant increases in tumor incidence were
induced by TCA following drinking water exposures of only 51-82 weeks. EPA recognizes that
the available literature provides evidence of TCA tumor induction in only one strain of mouse;
however, mouse cancer bioassay data for TCA are limited to the B6C3Fi strain only.
Cancer data in the rat. TCA did not induce tumors in male F344 mice in a 104-week bioassay by
DeAngelo et al. (1997); however, female rats were not included in the study by DeAngelo et al.
(1997) and other experimental animal species have not been tested for TCA carcinogenic
potential. Therefore, the absence of cancer data for species other than the mouse is a limitation
of the TCA database, but does not support conclusions about lack of response in other species.
With regard to the comment regarding a lack of concordance in tumor response between the
mouse and rat, EPA notes that site concordance is not always assumed between animals and
humans (U.S. EPA, 2005a), nor is site concordance between different animal species assumed.
Therefore, the fact that TCA did not induce liver tumors in the male F344 rat does not
substantially reduce the strength of the positive mouse liver tumor findings.
MOA considerations. In response to the reviewer who disagreed with the descriptor of "likely"
because this reviewer did not consider the cancer MOA related to peroxisome proliferation
relevant to humans, EPA points to the response to Carcinogenicity Charge Question 3 (i.e., that
while PPARa-related events represent some of the major components of the overall MOA, the
MOA of TCA carcinogenicity is complex and more than one MOA that are not mutually
exclusive may be involved). As noted in the Cancer Guidelines, MOA information can influence
the choice of descriptor. Accordingly, a summary of cancer MOA information was added to
Section 4.7.1.
The cancer descriptor for TCA of "likely to be carcinogenic to humans" by all routes of
exposure was retained as the descriptor consistent with the Cancer Guidelines (U.S. EPA,
2005a). The descriptor was revised, however, to better reflect the nature of the available
carcinogenicity data and mode of action information considered in assigning this descriptor.
Because assignment of a descriptor is based on the evaluation of the data set for the chemical
under assessment and is not established relative to other chemical data sets, information
pertaining to other chemicals sharing the same descriptor was not added to the Toxicological
Review.
A-13 DRAFT - DO NOT CITE OR QUOTE
-------
Comments: One reviewer concurred that the descriptor is applicable to all routes of exposure
because absorbed TCA from any route of administration will likely be systemically distributed, a
major portion of inhaled TCA would likely be absorbed given its high water solubility, and
because TCA undergoes minimal metabolism, it is likely that carcinogenic effects observed with
oral exposure would also occur with dermal or inhalation exposure. Another reviewer disagreed
with EPA's conclusion of "likely to be carcinogenic" by all routes of exposure, and stated that
only ".. .by the oral and dermal routes of exposure" is supported. This reviewer further observed
that there are no inhalation studies to determine whether or not TCA is absorbed following
inhalation exposure, and no scientific argument is offered to support the conclusion that uptake
into the blood of intact parent TCA could occur by the inhalation route.
Response: EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) indicate that
for tumors occurring at a site other than the initial point of contact, the weight of evidence for
carcinogenic potential may apply to all routes of exposure that have not been adequately tested at
sufficient doses. An exception occurs when there is convincing toxicokinetic data that
absorption does not occur by other routes. As stated by one reviewer, data evaluating absorption
by the inhalation route are unavailable. However, TCA is highly soluble in water and, as such, it
is reasonable to assume that it can be absorbed and taken up into the blood via the inhalation
route. Moreover, based on the observance of systemic tumors following oral exposure, and in
the absence of information to indicate otherwise, it is assumed that an internal dose will be
achieved regardless of the route of exposure. The cancer descriptor narrative in Section 4.7.1
was revised to better describe this decision.
Comments: One reviewer expressed reservations about the characterization of the genetic
toxicity information for TCA. The reviewer agreed with the conclusion that TCA is, at best, a
weak mutagen, but did not consider the case for mutagenesis related to oxidative stress to have
been made. This reviewer suggested that for the most part, the tests utilized, particularly those in
vitro, either lacked the ability to detect oxidative effects or, even if they had such sensitivity, the
likely metabolic or cellular mechanisms for generating activated oxygen species were absent
from the test systems.
Response: EPA agrees that in vitro tests of TCA genotoxicity do not include tests designed
specifically to detect genotoxic endpoints induced by oxidative DNA damage. This information
has been added to the text. Studies investigating the ability of TCA to induce oxidative stress,
including DNA damage, are included in Sections 4.2 and 4.5.1.6.
A-14 DRAFT - DO NOT CITE OR QUOTE
-------
2. Have the studies supporting the discussion of the mode(s) of carcinogenic action been
clearly described?
Comments: Six reviewers generally concurred with the cancer MOA discussion as written;
comments included statements that the available studies were adequately described and that the
review was extensive and comprehensive. One reviewer commented that the document suffered
from inconsistencies in listing the components of the MOA in different sections, which detracted
from driving the message home that the MOA is complex, and suggested refocusing such
discussions to strengthen the MOA analysis. Two reviewers did not consider the cancer MOA
section to be clearly described or complete. One of these reviewers commented that the MOA
discussion needed to be restructured in order to achieve adequate clarity. This reviewer
commented that the analysis did not do a good job of applying EPA's framework from the
Cancer Guidelines for evaluating the hypothesized MOA. This reviewer also suggested that a
second possible MOA (direct damage to DNA) undergo a "formal" MOA analysis, and that there
should be a separate section for other possible MO As (including the role of nonparenchymal
cells such as Kupffer cells, the role of other nuclear receptors, and GJIC-intercellular
communication) that currently do not have sufficient information to be subjected to a formal
MOA analysis. Two reviewers suggested the use of tables in the MOA section. One of these
reviewers suggested adding a table that identifies consistencies/inconsistencies in a series of
experimental observations and data gaps for other well studied peroxisome proliferators (e.g.,
DEHP and Wy-14643) and for TCA.
Response: Section 4.7.3 was revised to improve the clarity of the MOA discussion and to make
the discussion more concise. Included in the revisions is a paragraph that was added at the start
of the MOA section to lay out the potential MO As to be discussed. Discussions of other MO As
(Kupffer cell activation, hypomethylation of DNA, and reduced intercellular communication)
were added as a new Section 4.7.3.2. Data gaps related to consistency and specificity of PPARa
agonism as the sole MOA for liver tumor induction were discussed from multiple perspectives in
Section 4.7.3.1.1. An analysis of the data for other well-studied peroxisome proliferators and
comparison to TCA was considered, but was determined to be outside the scope of this
Toxicological Review.
Comments: One reviewer provided the following references for discussion in the MOA section:
Preston, RJ; Williams, GM. (2005) DNA-reactive carcinogens: mode of action and human cancer hazard. Crit Rev
Toxicol 3 3:673-683.
Roberts, RA; et al. (2007) Role of the Kupffer cell in mediating hepatic toxicity and carcinogenesis. Toxicol Sci
96(1):2-15.
A-l 5 DRAFT - DO NOT CITE OR QUOTE
-------
Woods, CG; et al. (2007a) Sustained formation of POBN radical adducts in mouse liver by peroxisome proliferators
is dependent upon PPARa but not NADPH oxidase. Free Radic Biol Med 42(3):335-342.
Woods, CG; et al. (2007b) Time-course investigation of PPARa- and Kupffer cell-dependent effects of Wy-14643
in mouse liver using microarray gene expression. Toxicol Appl Pharmacol 225(3):267-277.
Guo, D; et al. (2007) Induction of nuclear translocation of constitutive androstane receptor by peroxisome
proliferator-activated receptor a synthetic ligands in mouse liver. J Biol Chem 282(50): 36766-36776.
Zhen, Y; et al. (2007) Metabolomic and genetic analysis of biomarkers for peroxisome proliferator-activated
receptor a expression and activation. Mol Epidemiol 21(9):2136-2151.
The reviewer also provided six references on PFOA, another PPARa agonist, that this reviewer
thought could be helpful.
Response: The references were added to the Toxicological Review, with the exception of
Preston and Williams (2005), Zhen et al. (2007), and the papers on PFOA, which were of interest
but did not directly relate to TCA toxicity.
Comments: One reviewer recommended that a more complete description be provided of the
early report by Elcombe (1985), and the studies by Ito et al. (2007) and Ren et al. (2010).
Another reviewer suggested that EPA: (1) strengthen the quantitative assessment of the relative
potency of PPARa activation by TCA in comparison to other chlorinated solvents; and
(2) compare the potency indicators for mouse hepatocarcinogenicity of various peroxisome
proliferators, including chloroacetic acids, with common short-term markers of PPARa
activation and in vitro transactivation of PPARa. One reviewer commented that the short section
on decreased cell-cell communication seemed speculative.
Response: The study description for Elcombe (1985) was expanded in the Toxicological Review
as requested. A summary of Ren et al. (2010) was added to support the possibility of activation
of other nuclear receptors by PPARa agonists, but a detailed description of this study was not
included as the study investigated DEHP and not TCA. For the same reason, the Ito et al. (2007)
study description was not expanded. As indicated above, an analysis of the data for other well-
studied peroxisome proliferators and comparison to TCA was considered, but was determined to
be outside the scope of this Toxicological Review. The MOA of decreased cell-cell
communication has not been well-characterized in the scientific literature, and is characterized as
such in Section 4.7.3.1.4.
Comments: One reviewer commented that potential carcinogenic MO As other than PPARa
agonist-induced peroxisome proliferation are speculative and should either be identified as such
or not discussed. This reviewer believed that "TCA could not have produced any tumors by a
MOA similar to a non-PPARa agonist" because: (1) foci and tumors found in mouse liver
A-16 DRAFT - DO NOT CITE OR QUOTE
-------
tumors produced by chemicals that are not PPARa agonists are eosinophilic and GST-u positive,
which is not the characteristic of TCA-induced tumors; and (2) the histopathology, biology, and
molecular biology of the liver tumors in TCA-treated mice are completely consistent with those
found in mice treated with other inducers of peroxisome proliferation.
Response: Hypothesized MO As other than PPARa agonism were based on studies of TCA as
well as other classic PPARa agonists such as DEHP and Wy-14643. Section 4.7.3 was revised
to clarify when data for PPARa agonists other than TCA provided the major support for
alternative potential MO As for TCA-induced hepatocarcinogenesis.
3. EPA has concluded that the available data do not support any specific MOA. In
addition, EPA has determined that the data are not supportive of PPARa agonist-induced
peroxisome proliferation as the sole MOA leading to tumor formation. Please comment on
whether these determinations are scientifically justified.
Comments: Six reviewers generally agreed with EPA's conclusions regarding cancer MOA.
Several of these reviewers further observed that PPARa-related events represent some major
components of the MOA, but that there are likely multiple MO As that may not be mutually
exclusive. One reviewer suggested that the statement that the data are not supportive of PPARa
induced peroxisome proliferation as the sole MOA was too strong; this reviewer preferred the
statement that the data do not identify any specific MOA (including peroxisome proliferation) as
the sole MOA.
Three reviewers disagreed with the conclusion that there are insufficient data to establish
PPARa agonism as a MOA. It was the judgment of two of these reviewers that the data
supported a PPARa agonist MOA as the sole MOA; one of these reviewers commented that
MO As other than PPARa were speculative and without support. One reviewer commented that
if more than one MOA is involved, then it is necessary to describe which MOA could occur
under what circumstances/conditions.
Response: EPA agrees that the MOA for TCA carcinogenicity is complex, that multiple MO As
that are not mutually exclusive may be involved, and that while PPARa-related events represent
some of the major components of the overall MOA, it is premature to conclude that this is the
only MOA for TCA. Discussion of data gaps with respect to consistency and specificity of
PPARa agonism as the sole MOA for TCA carcinogenicity in Section 4.7.3.1.1 was re-structured
to improve the clarity. The MO As other than PPARa agonism are hypotheses based on
experimental evidence. Discussions of these MO As in Sections 4.7.3.1.2-4.7.3.1.4 were revised
to clarify the nature and extent of scientific support.
A-17 DRAFT - DO NOT CITE OR QUOTE
-------
Comments: Two reviewers indicated that the discussion of human relevance of the PPARa
agonist MOA was incomplete. One of these reviewers stated that the growing body of evidence
that reveals qualitative differences between the rodent and human PPARa cascade needed to be
summarized. The second reviewer recommended that differences in mouse and human binding
of activated hPPARa and the reliability of studies of primary human hepatocyte cultures be
included in the discussion of human relevance.
Response: The discussion of human relevance has been expanded. New studies related to the
possible differences between human and mouse PPARa and a discussion of the weakness of in
vitro studies were added to Section 4.7.3.1.1.4.
4. A 104-week drinking water study in mice (DeAngelo et al., 2008) was selected as the
basis for quantification of the oral cancer slope factor. Please comment on whether the
selection of this study is scientifically justified.
Comments: Five reviewers considered the selection of DeAngelo et al. (2008) as the basis for
the oral cancer slope factor to be justified. One reviewer disagreed with the application of an
exposure duration adjustment factor for the 82-week study (Pereira, 1996). This reviewer
suggested that the slope factor from this study be recalculated without the exposure duration
factor, and re-compared to the slope factor based on the 104-week study before reaching a final
decision as to which of the two studies provides the best data set for quantitative assessment.
This reviewer also suggested excluding oral slope factors calculated from the 52- and 60-week
studies because of the uncertainty introduced by the exposure duration factor. One reviewer
expressed concern that there were a large number of animals identified as unscheduled deaths in
the DeAngelo et al. (2008) study that were not examined for hepatocellular neoplasia and
suggested obtaining an explanation from the study author as to why these animals were not
examined and the impact of the missing data on the cancer potency estimate. This reviewer also
asked for an explanation for the difference in the liver tumor incidence in the control group from
Study 2 (12%) versus the control group from Study 3 (64%). One reviewer commented that the
104-week study does not appear to be a good choice for modeling of the cancer slope factor
because: (1) the incidence of combined adenomas and carcinomas at the lower of the two doses
in this study is less than the incidence in the control mice (so that the POD is based essentially on
the observation of a single dose group); and (2) the 104-week duration study had a very high
incidence of combined adenomas and carcinomas in the control group (64%) when compared
with other studies in the male B6C3Fi mouse. Two reviewers commented that the development
of an oral cancer slope factor is not justified since the liver tumors induced by TCA are not
sufficient to classify TCA as a potential human carcinogen.
A-l 8 DRAFT - DO NOT CITE OR QUOTE
-------
Response: Because the 82-week study by Pereira (1996) was not a lifetime study for the mouse,
applying an exposure time adjustment factor as (82/104)°'25 = 0.49 is considered to be
appropriate. Without this adjustment, the candidate cancer slope factor derived from this study
would be approximately half of the current value (i.e., 2.1 x 10"2), which is lower than the oral
cancer slope factor derived from the 104-week study by DeAngelo et al. (2008). The cancer
slope factors from the 52- and 60-week studies were calculated for purpose of comparison. The
final slope factor was based on the 104-week lifetime study.
In the 104-week study from DeAngelo et al. (2008), the animals identified as
unscheduled deaths without pathology vs. unscheduled deaths in the 0, 0.05, and 0.5 g/L TCA
dose groups were 9/17, 17/24, and 14/24, respectively. According to the study author, Dr.
Anthony DeAngelo (email dated April 6, 2010, from Anthony DeAngelo, NHEERL, ORD, U.S.
EPA, to Diana Wong, NCEA, ORD, U.S. EPA), most of the pathology data were missing
because the tissues of the animals found dead were autolyzed or the remains were eaten by cage
mates. One or two animals might also have escaped over the 2-year period of study. The
possible impact of missing pathology could be an underestimation of the liver tumor incidence
and a lower oral cancer slope factor; however, the difference among the five candidate cancer
slope factors derived from five independent studies was approximately fivefold, which lends
some confidence to the oral slope factor derived from the DeAngelo et al. (2008) study (see
Table 5-12).
For B6C3Fi mice, the higher rate of adenomas and carcinomas in the control group in the
104-week study is not unexpected. For the cancer quantification, percent increase in tumors
compared with the control was used in BMD modeling, not absolute percent tumors.
The study 2 and 3 control groups were treated with different control vehicles—one was
neutralized acetic acid and the second was deionized water. The experiments were conducted in
different labs at different times. These factors might contribute to the differences in tumor
incidence in the two control groups.
Comments related to the choice of studies for BMD modeling are addressed in responses
to Charge Question C5.
5. The oral cancer slope factor was calculated by linear extrapolation from the POD (lower
95% confidence limit on the dose associated with 10% extra risk for liver tumors). Has the
modeling approach been appropriately conducted? Please identify and provide the
rationale for any alternative approaches for the determination of the slope factor and
discuss whether such approaches are preferred to EPA's approach.
Comments: Four reviewers generally agreed that modeling of the cancer dose response and
derivation of the oral cancer slope factor was appropriately undertaken. One of these reviewers
observed that the assumption used in the exposure duration scaling of a mouse lifetime of
A-19 DRAFT - DO NOT CITE OR QUOTE
-------
104 weeks underestimates the true lifetime of a B6C3Fi mouse and that longer lifetimes should
be used for this scaling. This reviewer questioned why the multistage model was the only model
in BMDS used to fit liver tumor data and recommended that other models be fit and evaluated
for goodness-of-fit. This reviewer also suggested that male mouse liver tumor data from the
60- and 104-week studies by DeAngelo et al. (2008) be combined and used for the determination
of the BMLDio and the oral cancer slope factor. A second of these reviewers observed that the
response at the low dose in the 104-week DeAngelo et al. (2008) study was essentially the same
as the control, so that there was only one positive value for tumor incidence. In this case, the
reviewer argued that the use of BMD modeling did not make sense.
One reviewer commented that the margin-of-exposure approach would be more
appropriate for a chemical with a "suggestive" weight-of-evidence characterization, and that no
quantitative assessment is necessary for an "unlikely" characterization. Three reviewers did not
comment on this charge question.
Response: EPA considers 104 weeks to be the standard lifetime for rats and mice. Therefore,
exposure duration scaling for the 104-week mouse study is not warranted. The rationale for use
of the multistage model is provided in Section 5.4.4. As noted in this section, the multistage
model has been used by EPA in the vast majority of quantitative cancer assessments because it is
thought to reflect the multistage carcinogenic process. This model can accommodate a wide
variety of dose-response shapes and its use provides consistency with previous quantitative dose-
response assessments for cancer.
An analysis using combined tumor data from the DeAngelo et al. (2008) study was
conducted. Because the 60- and 104-week studies used different study time frames, dose-
response analysis of combined data sets based on the summary incidence data in the published
paper could not be performed with the models in BMDS. EPA obtained the individual animal
data from Dr. Anthony DeAngelo, including when the liver tumor was identified in individual
animals. The MSW time-to-tumor model, which models both the exposure dose and appearance
time of the tumor, was used to model combined tumor data sets. A statistical analysis
(generalized likelihood ratio test) was first applied to determine which of the three studies
reported in DeAngelo et al. (2008) were compatible for combined analysis. A summary of the
MSW time-to-tumor modeling was added to Section 5.4.5, and a detailed discussion of the
modeling, including model outputs, was provided in a new Appendix E. This analysis revealed
that oral cancer slope factors for the individual study and combined data sets were not
substantially different, nor were there substantial differences in the slope factors derived with the
MSW time-to-tumor model or the multistage model in BMDS. The 104-week study in
DeAngelo et al. (2008) was selected for deriving the cancer slope factor because it is the only
lifetime study of TCA. The BMDL derived from this study was within 2.5-fold of BMDL values
from other TCA cancer bioassays, including 52-week studies in male mice (Bull et al., 2002,
A-20 DRAFT - DO NOT CITE OR QUOTE
-------
1990), a 60-week study in male mice (DeAngelo et al., 2008), and an 82-week study in female
mice (Pereira, 1996). This consistency in BMDL values supports the use of data from the
104-week study for BMD modeling.
6. An inhalation unit risk (IUR) for cancer was not derived for TCA. Is the determination
that the available data for TCA do not support derivation of an IUR scientifically justified?
Comments: Most reviewers agreed that the decision not to derive an IUR for TCA was justified.
One reviewer considered the justification for not deriving an IUR to be inadequate. Another
reviewer observed that, in view of the relatively minor importance of metabolism in TCA
toxicity and the water-soluble nature of the chemical, an IUR could be derived by route-to-route
extrapolation by assuming 100% absorption by the inhalation route followed by systemic
distribution via the bloodstream. One reviewer questioned how the Toxicological Review could
support the conclusion that inhaled TCA is carcinogenic in laboratory animals and humans if no
inhalation data exist.
Response: In the absence of inhalation studies of TCA and given the lack of a PBPK model for
TCA to support route-to-route extrapolation, an IUR for TCA was not derived. Available
information suggests that a simple set of assumptions might not account for possible differences
in the magnitude of response across routes.The liver is the critical target organ for oral toxicity,
and a first-pass effect by the liver is expected following oral administration. Justification for not
deriving an IUR in Section 5.4.6 was expanded. The rationale for characterizing TCA as likely
to be carcinogenic by all routes of exposure is addressed in response to comments on Charge
Question Cl.
A-21 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX B. BENCHMARK DOSE MODELING RESULTS FOR LIVER DATA SETS
FROM DeANGELO ET AL. (2008)
B.I. INCIDENCE OF HEPATOCELLULAR INFLAMMATION
Table B-l. BMD modeling results based on incidence of hepatocellular
inflammation in male B6C3Fi mice exposed to TCA in drinking water for 60
weeks
Fitted dichotomous model3
Gamma
Logistic
Log-logistic
Multistage (1°)
Probit
Log-probit
Weibull
%2 goodness-of-fit test
/7-valueb
0.096
0.24
0.096
0.22
0.24
0.26
0.096
AICC
76.15
74.19
76.16
74.29
74.20
74.19
76.16
BMD10d
(mg/kg-d)
354.2
391.9
351.0
292.0
376.1
394.1
361.9
BMDL10e
(mg/kg-d)
151.6
276.6
132.1
149.4
257.1
244.4
151.6
aAll dichotomous dose-response models were fit using BMDS, version 1.4.1. The best-fit models are
indicated in boldface type.
V-Value from the %2 goodness-of-fit test for the selected model. Values <0.1 suggest that the model exhibits
a significant lack of fit, and a different model should be chosen.
°AIC is a value useful for evaluating model fit. For those models exhibiting adequate fit, lower values of the
AIC suggest better model fit.
dBMD10 = BMD at 10% extra risk.
eBMDL10 = 95% lower confidence limit on the BMD at 10% extra risk.
Source: DeAngelo et al. (2008).
Of the seven models fit, four (i.e., logistic, one-stage multistage, probit, and log-probit)
showed adequate fit. The BMDS outputs from the two models with the best fit (based on lowest
AIC value), the logistic and log-probit models, are provided below.
B-l
DRAFT - DO NOT CITE OR QUOTE
-------
Logistic Model with 0.95 Confidence Level
0.4
I 0.3
I
"
CD
0.2
0.1
Logistic
BMD
.BMP.
100
200
300
dose
400
500
600
12:0709/052008
Logistic Model. (Version: 2.10; Date: 09/23/2007)
Input Data File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_HEPATOCELLULAR_INFLAMMATION_60_WKS_DEANGELO_2008.(d)
Gnuplot Plotting File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_HEPATOCELLULAR_INFLAMMATION_60_WKS_DEANGELO_2008.plt
Fri Sep 05 12:07:17 2C
The form of the probability function is:
P[response] = I/[1+EXP(-intercept-slope*dose)]
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le
Parameter Convergence has been set to: le-008
B-2
DRAFT - DO NOT CITE OR QUOTE
-------
intercept
1
-0.76
Variable
intercept
slope
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit
0.482625 -3.80523
0.00109927 0.000690752
Model
Full model
Fitted model
Reduced model
Analysis of Deviance Table
Log(likelihood) # Param's Deviance Test d.f.
-33.0575 4
-35.0966 2 4.07833 2
-38.4712 1 10.8276 3
Est. Prob.
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 391. 918
BMDL = 276.646
B-3
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
T3
-------
( *** The model parameter(s) -slope
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
intercept
Variable
background
intercept
slope
Analysis of Deviance Table
Log(likelihood) # Param's Deviance Test d.f.
-33.0575 4
-35.0974 2 4.07991 2
-38.4712 1 10.8276 3
P-value
Est. Prob.
B-5
DRAFT - DO NOT CITE OR QUOTE
-------
B.2. INCIDENCE OF HEPATOCELLULAR NECROSIS
Table B-2. BMD modeling results based on incidence of hepatocellular
necrosis in male B6C3Fi mice exposed to TCA in drinking water for 30-45
weeks
Fitted dichotomous model3
Gamma, multistage (1°), and Weibull
Logistic
Log-logistic
Probit
Log-probit
%2 goodness-of-fit
test/7-valueb
0.18
0.058
0.49
0.060
0.036
AICC
31.85
36.39
30.42
36.26
36.84
BMD10d
(mg/kg-d)
64.9
205.1
40.7
188.0
158.7
BMDL10e
(mg/kg-d)
37.6
128.4
17.9
120.0
54.3
aAll dichotomous dose-response models were fit using BMDS, version 1.4.1. The best-fit models are
indicated in boldface type.
V-Value from the %2 goodness-of-fit test for the selected model. Values <0.1 suggest that the model exhibits
a significant lack of fit, and a different model should be chosen.
°AIC is a value useful for evaluating model fit. For those models exhibiting adequate fit, lower values of the
AIC suggest better model fit.
dBMD10 = BMD at 10% extra risk.
eBMDL10 = 95% lower confidence limit on the BMD at 10% extra risk.
Source: DeAngelo et al. (2008).
Of the seven models fit, four (i.e., gamma, log-logistic, one-stage multistage, and
Weibull) showed adequate fit. The BMDS output for the best fitting model of the four (based on
lowest AIC value), the log-logistic model, is provided below.
B-6
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
T3
CD
0
CO
0.8
0.7
0.6
0.4
0.3
0.2
0.1
0
Log-Logistic
BMPLI
BMD
0
100
200
300
dose
400
500
600
14:21 09/052008
Logistic Model. (Version: 2.10; Date: 09/23/2007)
Input Data File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_HEPATOCELLULAR_NECROSIS_DEANGELO_2008.(d)
Gnuplot Plotting File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_HEPATOCELLULAR_NECROSIS_DEANGELO_2008.plt
Fri Sep 05 14:21:36 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = background+(1-background)/[1 + EXP(-intercept-slope*Log(dose) ) ;
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0
intercept = -5.96722
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
B-7
DRAFT - DO NOT CITE OR QUOTE
-------
intercept
intercept 1
Parameter Estimates
Std. Err.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
Model
Full model
Fitted model
Reduced model
Analysis of Deviance Table
# Param's Deviance
4
1
1
Dose
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 40.6639
BMDL = 17.8767
B-8
DRAFT - DO NOT CITE OR QUOTE
-------
B.3. INCIDENCE OF TESTICULAR TUBULAR DEGENERATION
Table B-3. BMD modeling results based on incidence of testicular tubular
degeneration in male B6C3Fi mice exposed to TCA in drinking water for
60 weeks
Fitted dichotomous model3
Gamma, multistage (1°), and Weibull
Logistic
Log-logistic
Probit
Log-probit
X2 goodness-of-fit test
p-v&lueb
0.19
0.16
0.19
0.17
0.13
AICC
76.16
76.59
76.08
76.54
77.06
BMD10d
(mg/kg-d)
321.9
439.7
298.2
425.3
471.6
BMDL10e
(mg/kg-d)
153.3
290.3
127.4
271.2
276.8
aAll dichotomous dose-response models were fit using BMDS, version 1.4.1. The best-fit models are indicated in
boldface type.
kp-Value from the %2 goodness-of-fit test for the selected model. Values <0.1 suggest that the model exhibits a
significant lack of fit, and a different model should be chosen.
°AIC is a value useful for evaluating model fit. For those models exhibiting adequate fit, lower values of the AIC
suggest better model fit.
dBMD10 = BMD at 10% extra risk.
eBMDL10 = 95% lower confidence limit on the BMD at 10% extra risk.
Source: DeAngelo etal. (2008).
All seven models showed adequate fit. The BMDS output from the model that provided
the best fit of the seven (based on lowest AIC value), the log-logistic model, is provided below.
B-9
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
0.4
0.35
0.3
T3
0
I °'25
c 0.2
0
1§ 0.15
ul
0.1
0.05
0
Log- Logistic
-
1
<
J
'
<
^
V^
(
0
_
r-^^^
BMDL
100
'
^-—^^^
T ]
^ — ""^ <
;
\
\
I- \
\
L ;
^
BMD ;
200 300 400 500 600
dose
13:5009/052008
Logistic Model. (Version: 2.10; Date: 09/23/2007)
Input Data File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_TESTICULAR_DEGENERATION_60_WKS_DEANGELO_2008.(d)
Gnuplot Plotting File: M:\TCA DOSE-RESPONSE
MODELING\MALE_MOUSE_TESTICULAR_DEGENERATION_60_WKS_DEANGELO_2008.plt
Fri Sep 05 13:50:29 2008
The form of the probability function is:
P[response] = background+(1-background)/[1 + EXP(-intercept-slope*Log(dose) ) ;
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.0666667
intercept = -7.67626
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
B-10
DRAFT - DO NOT CITE OR QUOTE
-------
background intercept
background 1 -0.47
intercept -0.47 1
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
Log(likelihood) # Param's Deviance Test d.f.
-33.7671 4
-36.0406 2 4.54705 2
-38.4712 1 9.40833 3
Est. Prob.
d.f. =2
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 298.169
BMDL = 127.35
B-ll
DRAFT - DO NOT CITE OR QUOTE
-------
B.4. CYANIDE-INSENSITIVE PCO ACTIVITY
Table B-4. BMD modeling results based on cyanide-insensitive PCO activity
in male B6C3Fi mice exposed to TCA in drinking water for up to 60 weeks
Fitted
continuous
model3
Hill
Linear
Polynomial (2°)
Power
Test of homogeneity
of variances /7-value
0.18
0.18
0.18
0.18
%2 goodness-of-fit test
/7-valueb
NA
0.003
0.66
0.003
AICC
165.52
173.39
163.71
173.39
BMD1SDd
(mg/kg-d)
33.5
61.7
28.4
61.7
BMDL1SDe
(mg/kg-d)
20.0
50.2
21.1
50.2
"All continuous dose-response models were fit using BMDS, version 2.1.1. The variances were determined
to be non-homogeneous across dose groups, so, in each case, the non-constant variance version of the model
was fit to the data. The best-fit model is indicated in boldface type.
V-Value from the %2 goodness-of-fit test for the selected model. Values <0.1 suggest that the model exhibits
a significant lack of fit, and a different model should be chosen.
°AIC is a value useful for evaluating model fit. For those models exhibiting adequate fit, lower values of the
AIC suggest better model fit.
dBMD1SD = BMD estimated at 1 SD from the control mean.
eBMDL1SD = estimated 95% lower confidence limit on the BMD at 1 SD from the control mean.
Source: DeAngelo et al. (2008).
Of the four models fit to the data, only the second-degree polynomial model showed
adequate fit, and thus, the BMDS output from this model is provided below.
B-12
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
14
12
10
o
Q.
in
-------
lalpha
rho
beta_0
beta_l
beta 2
Parameter Estimates
Lower Conf. Limit
-2.45612
1.09045
2.35095
0.0213061
-4.34447e-005
Upper Conf. Limit
-1.02835
1.9675
2.88985
0.0418251
-9.36686e-006
Model Descriptions for likelihoods calculated
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
B-14
DRAFT - DO NOT CITE OR QUOTE
-------
Likelihoods of Interest
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (Al vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Test
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 28.3615
BMDL = 21.0843
B-15 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX C. BENCHMARK DOSE MODELING RESULTS FOR
DEVELOPMENTAL DATA SETS FROM SMITH ET AL. (1989)
C.I. FETAL BODY WEIGHT
Table C-l. BMD modeling results based on fetal body weight in Long-Evans
rats exposed to TCA by gavage on GDs 6-15—male fetuses
Model
Exponential (model 2)
Exponential (model 3)
Exponential (model 4)
Exponential (model 5)
Hill (constant variance)
Polynomial-linear
Polynomial (degree >2)
Power
^j-value
0.0061
0.0061
0.1050
0.1050
0.1861
0.0005
0.0363
0.0005
AIC
-144.03
-144.03
-149.93
-149.93
-151.07
-138.89
-147.80
-138.89
Largest
residual
(mg/kg-d)
-2.6
-2.6
-1.5
-1.5
-1.1
-2.7
-2.1
-2.7
BMD1SD
(mg/kg-d)
307.2
307.2
182.4
182.4
160.0
373.0
224.2
373.0
BMDL1SD
(mg/kg-d)
258.5
258.5
133.0
133.0
109.0
317.9
180.4
317.9
BMD05
(mg/kg-d)
221.3
221.3
138.3
138.3
121.4
261.0
168.4
261.0
BMDLos
(mg/kg-d)
197.8
197.8
103.4
103.4
84.0
236.7
141.9
236.7
BMD05 = BMD at 5% change in mean relative to the control mean; BMDiSD = BMD at 1 SD change in mean from
the control mean
Source: Smith etal. (1989).
C-l
DRAFT - DO NOT CITE OR QUOTE
-------
Hill Model with 0.95 Confidence Level
c
o
8-
-------
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -rho -n
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
alpha intercept v k
alpha 1 4.5e-008 l.le-007 -9.6e-008
intercept 4.5e-008 1 0.2
v l.le-007 0.2 1
k -9.6e-008 -0.45 -0.95
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.0553685 0.00854354 0.0386234 0.0721135
intercept 3.68734 0.0465166 3.59617 3.77851
v -1.81631 0.35395 -2.51004 -1.12258
n 1 NA
k 1074.7 449.461 193.777 1955.63
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma~2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma~2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e (i ) } = Sigma~2
Log(likelihood) # Param's AIC
81.219029 6 -150.438057
84.930660 10 -149.861321
C-3 DRAFT - DO NOT CITE OR QUOTE
-------
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (Al vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 127.619 8 <.0001
Test 2 7.42326 4 0.1151
Test 3 7.42326 4 0.1151
Test 4 3.36346 2 0.1861
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears
to be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 0.05
Risk Type = Relative risk
Confidence level = 0.95
BMD = 121.414
BMDL = 84.0299
C-4 DRAFT - DO NOT CITE OR QUOTE
-------
Table C-2. BMD modeling results based on fetal body weight in Long-Evans
rats exposed to TCA by gavage on GDs 6-15—female fetuses
Model
Exponential (model 2)
Exponential (model 3)
Exponential (model 4)
Exponential (model 5)
Hill (constant variance)
Polymomial-linear
Polynomial (degree >2)
Power
/7-value
0.0106
0.0106
0.1113
0.1113
0.1970
0.0011
0.0372
0.0011
AIC
-152.82
-152.82
-157.66
-157.66
-158.80
-148.04
-155.46
-148.04
Largest
residual
(mg/kg-d)
-2.359
-2.359
-1.328
-1.328
-1.000
-2.470
-1.880
-2.470
BMD1SD
(mg/kg-d)
305.9
305.9
188.2
188.2
166.1
369.7
230.7
369.7
BMDL1SD
(mg/kg-d)
257.5
257.5
136.9
136.9
113.4
315.2
185.0
315.2
BMD05
(mg/kg-d)
222.5
222.5
143.1
143.1
126.5
261.8
173.6
261.8
BMDLos
(mg/kg-d)
199.0
199.0
106.7
106.7
87.7
237.5
145.7
237.5
BMD05 = BMD at 5% change in mean relative to the control mean; BMD1SD = BMD at 1 SD change in mean from
the control mean
Source: Smith etal. (1989).
C-5
DRAFT - DO NOT CITE OR QUOTE
-------
Hill Model with 0.95 Confidence Level
3.6
3.4
3.2
c
o
8-
-------
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho -n
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
alpha intercept v
alpha 1 3.3e-009 -4.8e-009
intercept 3.3e-009 1 0.22
v -4.8e-009 0.22 1
k l.le-008 -0.45 -0.95
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.0505044 0.00779299 0.0352304 0.0657783
intercept 3.52807 0.0444292 3.44099 3.61515
v -1.78616 0.368915 -2.50922 -1.0631
n 1 NA
k 1154.27 497.126 179.921
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
14
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma~2
Model A3: Yij = Mu ( i ) + e(ij)
Var{e(ij)} = Sigma~2
Model A3 uses any fixed variance parameters that
were specified by the user
Likelihoods of Interest
Model Log ( likelihood ) # Param's AIC
Al 85.023823 6 -158.047645
A2 88.973191 10 -157.946383
A3 85.023823 6 -158.047645
fitted 83.399219 4
C-7 DRAFT - DO NOT CITE OR QUOTE
-------
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (Al vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Test -2*log(Likelihood Ratio) Test df p-value
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. Consider running a
non-homogeneous variance model
The p-value for Test 3 is less than .1. You may want to consider a
different variance model
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 0.05
Risk Type = Relative risk
Confidence level = 0.95
BMD = 126.489
BMDL = 87.7222
C-8 DRAFT - DO NOT CITE OR QUOTE
-------
C.2. FETAL CROWN-RUMP LENGTH (SMITH ET AL., 1989)
Table C-3. BMD modeling results based on fetal crown-rump length in
Long-Evans rats exposed to TCA by gavage on GDs 6-15—male fetuses
Model
Exponential (model 2)
Exponential (model 3)
Exponential (model 4)
Exponential (model 5)
Hill (constant variance)
Polymomial-linear
Polynomial (degree >2)
Power
/7-value
0.762
0.762
0.762
0.559
0.562
0.751
0.560
0.751
AIC
-273.53
-273.53
-273.53
-271.53
-271.54
-273.49
-271.53
-273.49
Largest
residual
(mg/kg-d)
-0.740
-0.740
-0.740
-0.740
-0.756
-0.816
-0.766
-0.816
BMD1SD
(mg/kg-d)
369.1
369.1
369.1
369.1
373.0
388.5
375.2
388.5
BMDL1SD
(mg/kg-d)
311.7
311.7
271.8
271.8
265.2
330.2
282.2
330.2
BMDos
(mg/kg-d)
600.7
600.7
600.7
600.7
605.7
625.4
608.4
625.4
BMDLos
(mg/kg-d)
534.4
534.4
468.4
468.4
460.3
560.7
482.6
560.7
BMD05 = BMD at 5% change in mean relative to the control mean; BMD1SD = BMD at 1 SD change in mean from
the control mean
Source: Smith etal. (1989).
C-9
DRAFT - DO NOT CITE OR QUOTE
-------
Exponential Model 2 with 0.95 Confidence Level
&
-------
MLE solution provided: Exact
Initial Parameter Values
Model
Inalpha
rho(S)
-4.34162
0
3.8955
0.000919238
0.772564
-4.34162
0
3.8955
0.000919238
0.772564
1
Parameter Estimates by Model
Model 2 Model 3 Model 4
Model
Inalpha
rho
Table of Stats From Input Data
Dose
Model
Estimated Values of Interest
Dose Est Mean Est Std
Scaled Residual
2
3
4
5
0
330
800
1200
1800
0
330
800
1200
1800
0
330
800
1200
1800
0
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
.702
. 599
. 458
.342
.175
.702
.599
. 458
.342
.175
.702
. 599
. 458
.342
.175
.702
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
.1149
0
-0
0.
0
-0
0
-0
0.
o
-0
0
-0
0.
0
-0
0
.3393
.7395
07542
.5931
. 3669
.3393
.7395
07542
.5931
. 3669
.3393
.7395
07542
.5931
. 3669
.3393
C-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Other models for which likelihoods are calculated:
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma^2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(1)^2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(1alpha + log(mean(i)) * rho)
Model R: Yij = Mu + e(i)
Var{e(ij)} = Sigma^2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al
A2
A3
R
2
3
4
5
140
142
140
90.
139
139
139
139
.3479
.1334
.3479
80382
. 7669
.7669
. 7669
.7669
6
10
6
2
3
3
3
4
-268.
-264 .
-268.
-177 .
-273.
-273.
-273.
-271.
. 6958
. 2668
. 6958
. 6076
.5337
.5337
.5337
.5337
Additive constant for all log-likelihoods = -77.19. This constant added to the
above values gives the log-likelihood including the term that does not
Explanation of Tests
Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)
Test 2: Are Variances Homogeneous? (A2 vs. Al)
Test 3: Are variances adeguately modeled? (A2 vs. A3)
Test 4: Does Model 2 fit the data? (A3 vs. 2)
Test 6a: Does Model 4 fit the data? (A3 vs 4)
Test 6b: Is Model 4 better than Model 2? (4 vs. 2)
Test 7a: Does Model 5 fit the data? (A3 vs 5)
Test 7b: Is Model 5 better than Model 3? (5 vs. 3)
Test 7c: Is Model 5 better than Model 4? (5 vs. 4)
Tests of Interest
Test -2*log(Likelihood Ratio) D. F. p-value
C-12 DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.
The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.
The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.
The p-value for Test 4 is greater than .1. Model 2 seems
to adeguately describe the data.
The p-value for Test 5a is greater than .1. Model 3 seems
to adeguately describe the data.
The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.
The p-value for Test 7a is greater than .1. Model 5 seems
to adeguately describe the data.
The p-value for Test 7b is greater than .05. Model 5 does
not seem to fit the data better than Model 3.
The p-value for Test 7c is greater than .05. Model 5 does
not seem to fit the data better than Model 4.
Benchmark Dose Computations:
Specified Effect = 0.050000
Risk Type = Relative deviation
Confidence Level = 0.950000
BMD and BMDL by Model
Model BMD BMDL
C-13 DRAFT - DO NOT CITE OR QUOTE
-------
C-14 DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4. BMD modeling results based on fetal crown-rump length in
Long-Evans rats exposed to TCA by gavage on GDs 6-15—female fetuses
Model
Exponential (model 2)
Exponential (model 3)
Exponential (model 4)
Exponential (model 5)
Hill (constant variance)
Polymomial-linear
Polynomial (degree >2)
Power
/7-value
0.658
0.658
0.515
0.515
0.521
0.595
0.503
0.595
AIC
-250.59
-250.59
-248.88
-248.88
-248.90
-250.31
-248.83
-250.31
Largest
residual
(mg/kg-d)
-0.897
-0.897
0.887
0.887
0.883
0.702
0.891
0.702
BMD1SD
(mg/kg-d)
468.9
468.9
421.5
421.5
417.2
491.4
429.6
491.4
BMDL1SD
(mg/kg-d)
387.8
387.8
287.8
287.8
274.2
409.6
312.7
409.6
BMD05
(mg/kg-d)
650.9
650.9
596.7
596.7
592.4
675.7
605.2
675.7
BMDLos
(mg/kg-d)
562.9
562.9
428.6
428.6
414.2
589.5
458.9
589.5
BMD05 = BMD at 5% change in mean relative to the control mean; BMD1SD = BMD at 1 SD change in mean from
the control mean
Source: Smith etal. (1989).
C-15
DRAFT - DO NOT CITE OR QUOTE
-------
Exponential Model 2 with 0.95 Confidence Level
-------
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
MLE solution provided: Exact
Variable
Model 2
Initial Parameter Values
Model 3 Model 4
Model 5
Inalpha
rho(S)
Variable
Parameter Estimates by Model
Model 2 Model 3 Model 4
Model 5
Inalpha
rho
-4.05805
0
3.63687
0.000271529
0.665739
1
Table of Stats From Input Data
Model
Estimated Values of Interest
Dose Est Mean Est Std
2
3
4
0
330
800
1200
1800
0
330
800
1200
1800
0
330
800
1200
1800
3
3
3
3
3
3
3
3
3
3
3. 63
.537
. 409
.303
3.15
3. 63
.537
.409
.303
3.15
. 637
533
3.4
299
.167
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0 .
0.
0.
0.
0.
0.
.131
.131
.131
.131
.131
.131
.131
.131
.131
.131
.131
. 131
.131
.131
.131
7
7
7
7
7
7
7
7
7
7
5
5
5
5
5
0
-0
-0
o
.3701
.2401
. 8969
.7706
-0. 007621
0
-0
-0
0
.3701
.2401
. 8969
.7706
-0. 007621
0
-0.
0
-0
.1213
08889
0.612
. 8873
.3633
C-17
DRAFT - DO NOT CITE OR QUOTE
-------
Other models for which likelihoods are calculated:
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma^2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(1)^2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(1alpha + log(mean(i)) * rho)
Model R: Yij = Mu + e(i)
Var{e(ij)} = Sigma^2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al
A2
A3
R
2
3
4
5
129.
132.
129.
91.7
128.
128.
128.
128.
1014
5454
1014
5225
2973
2973
4379
4379
6
10
6
2
3
3
4
4
-246.
-245.
-246.
-179.
-250.
-250.
-248.
-248.
.2027
. 0907
.2027
. 5045
. 5946
.5946
. 8759
. 8759
Additive constant for all log-likelihoods = -77.19. This constant added to the
above values gives the log-likelihood including the term that does not
depend on the model parameters.
Explanation of Tests
Test 1: Does response and/or variances differ among Dose levels? (A2 vs. R)
Test 2: Are Variances Homogeneous? (A2 vs. Al)
Test 3: Are variances adeguately modeled? (A2 vs. A3)
Test 4: Does Model 2 fit the data? (A3 vs. 2)
Test 6a: Does Model 4 fit the data? (A3 vs 4)
Test 6b: Is Model 4 better than Model 2? (4 vs. 2)
Test 7a: Does Model 5 fit the data? (A3 vs 5)
Test 7b: Is Model 5 better than Model 3? (5 vs. 3)
Test 7c: Is Model 5 better than Model 4? (5 vs. 4)
Tests of Interest
Test -2*log(Likelihood Ratio) D. F. p-value
C-18 DRAFT - DO NOT CITE OR QUOTE
-------
Test 3
Test 4
Test 5a
Test 5b
Test 6a
Test 6b
Test 7a
Test 7b
Test 7c
6. 888
1. 608
1. 608
684e-014
1.327
0.2813
1.327
0.2813
0
4
3
3
0
2
1
2
1
0
0 .
0 .
0 .
0 .
0 .
0 .
0 .
.1419
. 6575
.6575
N/A
.5151
. 5958
.5151
. 5958
N/A
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose
levels, it seems appropriate to model the data.
The p-value for Test 2 is greater than .1. A homogeneous
variance model appears to be appropriate here.
The p-value for Test 3 is greater than .1. The modeled
variance appears to be appropriate here.
to adeguately describe the data.
The p-value for Test 5a is greater than .1. Model 3 seems
to adeguately describe the data.
Degrees of freedom for Test 5b are less than or egual to 0.
The Chi-Sguare test for fit is not valid.
The p-value for Test 6a is greater than .1. Model 4 seems
to adeguately describe the data.
The p-value for Test 6b is greater than .05. Model 4 d
not seem to fit the data better than Model 2.
.oes
The p-value for Test 7a is greater than .1. Model 5 seems
to adeguately describe the data.
The p-value for Test 7b is greater than .05. Model 5 does
not seem to fit the data better than Model 3.
Degrees of freedom for Test 7c are less than or egual to 0.
The Chi-Sguare test for fit is not valid.
Benchmark Dose Computations:
Specified Effect = 0.050000
Risk Type = Relative deviation
Confidence Level = 0.950000
BMD and BMDL by Model
Model BMD BMDL
C-19 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX D. MODELING OF LIVER TUMOR INCIDENCE DATA FOR MICE
EXPOSED TO TCA IN DRINKING WATER
Five tumor data sets for combined incidence of hepatocellular adenomas and carcinomas
in B6C3Fi mice are shown in Tables 5-7 to 5-11 in Section 5.4.2. The estimated daily intakes of
TCA from the mouse studies were converted to human equivalent lifetime doses using an
interspecies body weight scaling factor and exposure time adjustment factor, which was based on
the assumption that the age-specific rate for cancer in humans will increase by at least the third
power of age (U.S. EPA, 1980).
For studies by Bull et al. (2002, 1990) and Pereira (1996), the body weight scaling factor
was calculated as [male B6C3Fi mouse reference body weight/human reference body weight]0'25
= [0.0373/70]0'25 = 0.15 (U.S. EPA, 1992, 1988). The exposure time adjustment factor was
calculated as: [duration of experiment/duration of animal lifetime (i.e., 104 weeks)]3. For the
52-week study by Bull et al. (2002, 1990), the exposure time adjust factor was calculated as
(52/104)3 = 0.125; for the 82-week study by Pereira (1996), the factor was calculated as
(82/104)3 = 0.49.
For studies by DeAngelo et al. (2008), the dose conversion is detailed in Tables D-l and
D-2.
Table D-l. Dose conversion for 60-week study
TCA dose
group
(g/L)
0.00
0.05
0.50
5.00
Mean water
consumption
(mL/kg-d)a
171
153
142
119
Mean measured
TCA
concentration
(mg/mL)3
NA
NA
0.48
5.06
Estimated
mean intake
(mg/kg-d)b
0
7.70
68.16
602.14
Continuous
exposure
time
adjustment
factor0
0.19
0.19
0.19
0.19
Average
animal
lifetime
weight (g)d
38.0
38.0
37.7
36.0
Body
weight
scaling
factor6
0.15
0.15
0.15
0.15
Human
equivalent
lifetime dose
(mg/kg-d)f
0
0.2
2.0
17.4
"Reported by DeAngelo et al. (2008).
Estimated mean daily intake = mean water consumption x mean measured TCA concentration; where the mean
measured TCA concentration was not reported, the nominal concentration for the dose group was used to calculate
the estimated mean daily intake.
Continuous exposure time adjustment factor = [duration of experiment/duration of animal lifetime]3, or [60/104]3
dCalculated using animal body weights at different weeks reported by DeAngelo et al. (2008).
eBody weight scaling factor = [average animal lifetime weight (kg)/human reference body weight (i.e., 70 kg)]025.
fHuman equivalent lifetime dose = estimated mean daily intake x exposure time adjustment factor x body weight
scaling factor.
Source: DeAngelo et al. (2008).
D-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-2. Dose conversion for 104-week study
TCA dose
group
(g/L)
0.00
0.05
0.50
Mean water
consumption
(mL/kg-d)a
112
111
116
Mean measured
TCA
concentration
(mg/niL)3
Not applicable
0.06
0.70
Estimated
mean intake
(mg/kg-d)b
0
6.66
81.20
Continuous
exposure
time
adjustment
factor0
1
1
1
Average
animal
lifetime
weight (g)d
42.2
42.5
42.6
Body
weight
scaling
factor"
0.16
0.16
0.16
Human
equivalent
lifetime dose
(mg/kg-d)f
0
1.0
12.8
"Reported by DeAngelo et al. (2008).
Estimated mean daily intake = mean water consumption x mean measured TCA concentration; where the mean
measured TCA concentration was not reported, the nominal concentration for the dose group was used to calculate
the estimated mean daily intake.
Continuous exposure time adjustment factor = [duration of experiment/duration of animal lifetime]3, or [104/104]3.
dCalculated using animal body weights at different weeks reported by DeAngelo et al. (2008).
eBody weight scaling factor = [average animal lifetime weight (kg)/human reference body weight (i.e., 70 kg)]025.
fHuman equivalent lifetime dose = estimated mean daily intake x exposure time adjustment factor x body weight
scaling factor.
Source: DeAngelo et al. (2008).
In the cancer dose-response analysis, EPA used tumor incidence values that differed from
those presented in the publication by DeAngelo et al. (2008). Tumor incidence values were
derived by EPA from individual animal data obtained from the study author (emails dated
February 1 and April 26, 2010, from Anthony DeAngelo, NHEERL, ORD, U.S. EPA, to Diana
Wong, NCEA, ORD, U.S. EPA). DeAngelo et al. (2008) based their tumor incidence values on
terminal sacrifice animals only. In EPA's analysis, the sample sizes represented the number of
animals at risk for tumor development (i.e., the number of mice included in the study when the
first tumor was discovered, which was week 45 in the 60-week study [Study 1] and week 52 in
the 104-week study [Study 3]). Animals that died before the first tumor was discovered were
excluded. The estimated mean daily intakes used in EPA's BMD analysis were calculated as a
product of mean daily water consumption data and measured TCA concentrations as reported in
DeAngelo et al. (2008), and not the nominal concentration presented DeAngelo et al. (2008).
Table D-3 compares the nominal doses and tumor incidence values presented in DeAngelo et al.
(2008) and the values used in the cancer dose-response analysis for TCA.
D-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-3. Comparison of average daily dose, sample size, and tumor
incidence from DeAngelo et al. (2008) as reported by study authors and as
recalculated by EPA
Endpoints
Average dose
(mg/kg-d)
Sample size
Tumor
incidence
Study duration
60 wks (study 1)
104 wks (study 3)
60 wks (study 1)
104 wks (study 3)
60 wks (study 1)
104 wks (study 3)
Nominal drinking
water concentration
(g/L)
0
0.05
0.5
5
0
0.05
0.5
0
0.05
0.5
5
0
0.05
0.5
0
0.05
0.5
5
0
0.05
0.5
DeAngelo et al. (2008)
0
8
68
602
0
6
58
30
27
29
29
42
35
37
4/30a
4/27a
11/293
16/293
27/42a
20/3 5a
32/37a
EPA re-calculation
0
7.7
68.2
602.1
0
6.7
81.2
35
32
34
34
56
48
51
4/35
5/32
12/34
19/34
31/56
21/48
36/51
"Tumor incidence was estimated based on data from DeAngelo et al. (2008), which reported only the number of
animals examined and percent of animals with tumors.
Using the EPA BMDS (version 2.1.1), the multistage model was fit to the combined
incidence of hepatocellular adenomas and carcinomas. Output files from BMDS are provided in
Sections D.l-D.5.
D-3
DRAFT - DO NOT CITE OR QUOTE
-------
D.I. FIFTY-TWO-WEEK STUDY FROM BULL ET AL. (2002) WITH THREE DOSE
GROUPS
Multistage Cancer Model with 0.95 Confidence Level
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMD Lower Bound
BMDL
BMD
4 5
dose
Note: Doses on x-axis are human equivalent doses for lifetime exposure in units of mg/kg-day.
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l)]
The parameter betas are restricted to be positive
Dependent variable = Incidence
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
D-4
DRAFT - DO NOT CITE OR QUOTE
-------
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -25.6775 3
Fitted model -27.2494 1 3.14381 2 0.2076
Reduced model -32.5964 1 13.8377 2 0.000989
Chi~2 = 3.55 d.f. = 2 P-value = 0.1695
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.34483
BMDL = 0.887265
BMDU = 2.61396
D-5 DRAFT - DO NOT CITE OR QUOTE
-------
D.2. FIFTY-TWO-WEEK STUDY FROM BULL ET AL. (1990) WITH THREE DOSE
GROUPS
Multistage Cancer Model with 0.95 Confidence Level
0.5
0.4
Multistage Cancer
Linear extrapolation
BMD Lower Bound
3
dose
Note: Doses on x-axis are human equivalent doses for lifetime exposure in units of mg/kg-day.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_52wkBulll990_stl.(d)
Gnuplot Plotting File: C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_52wkBulll990_stl.plt
The form of the probability function is:
Total number of observations = 3
Total number of records with missing value
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
D-6
DRAFT - DO NOT CITE OR QUOTE
-------
Asymptotic Correlation Matrix of Parameter Estimates
Beta(1)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background
Beta(l)
- Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -19.4921 3
Fitted model -21.2903 1 3.59649 2 0.1656
Reduced model -26.8563 1 14.7286 2 0.0006335
Scaled
Est. Prob. Expected Observed Size Residual
11
24
Chi~2 = 4.25 d.f. = 2 P-value = 0.1193
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level =
BMD =
BMDL =
BMDU =
D-7 DRAFT - DO NOT CITE OR QUOTE
-------
D.3. SIXTY-WEEK STUDY FROM DeANGELO ET AL. (2008) WITH FOUR DOSE
GROUPS
Multistage Cancer Model with 0.95 Confidence Level
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMD Lower Bound
6 8 10 12 14 16 18
13:43 10/292010
Note: Doses on x-axis are human equivalent doses for lifetime exposure in units of mg/kg-day.
Multistage Cancer Model. (Version: 1.9; Date: 05/26/2010)
Input Data File:
C:/USEPA/BMDS212/Data/TCA/TCA_Studyl/DeAngelo_2008_Tumor_MultiCancl_ExtraRisklO%.(d)
Gnuplot Plotting File:
C:/USEPA/BMDS212/Data/TCA/TCA_Studyl/DeAngelo_2008_Tumor_MultiCancl_ExtraRisklO%.plt
Total number of observations = 4
Total number of records with missing values
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
D-8
DRAFT - DO NOT CITE OR QUOTE
-------
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.51
Beta(l) -0.51 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf.
Limit
Background 0.182037 * * *
Beta(l) 0.0393995 * * *
^ - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -71.0546 4
Fitted model -72.4916 2 2.87397 2 0.2376
Reduced model -80.2414 1 18.3736 3 0.0003683
AIC: 148.983
Dose
0.1820 5.461 4.000 30 -0.691
0.1885 6.031 5.000 32 -0.466
0.2440 8.297 12.000 34 1.479
0.5879 19.989 19.000 34 -0.344
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 2.67416
BMDL = 1.6767
BMDU = 5.1239
Taken together, (1.6767 , 5.1239 ) is a 90 % two-sided confidence
interval for the BM
Multistage Cancer Slope Factor = 0.0596408
D-9 DRAFT - DO NOT CITE OR QUOTE
-------
D.4. EIGHTY-TWO-WEEK STUDY FROM PEREIRA (1996) WITH FOUR DOSE
GROUPS
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMD Lower Bound
0.8
0.6
0.4
0.2
Note: Doses on x-axis are human equivalent doses for lifetime exposure in units of mg/kg-day.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_82wkPerel996_st4-l.(d)
Gnuplot Plotting File: C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_82wkPerel996_st4-l.plt
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l)]
The parameter betas are restricted to be positive
Dependent variable = Incidence
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Default Initial Parameter Values
Background = 0.00436735
Beta(l) = 0.0188431
Asymptotic Correlation Matrix of Parameter Estimates
D-10
DRAFT - DO NOT CITE OR QUOTE
-------
Background
Background 1
Beta(l) -0.43
Variable
Background
Beta(1)
Std. Err.
Indicates that this value is not calculated.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
Model
Full model
Fitted model
Reduced model
Log(likelihood) # Param's Deviance Test d.f.
-58.4099 4
-59.1538 2 1.48782 2
-79.1216 1 41.4233 3
Goodness of Fit
Est._Prob. Expected Observed Size
0.0000 0.0373 3.361 4.000 90
5.7000 0.1196 6.337 4.000 53 -0.989
19.3000 0.2885 7.789 8.000 27
57.6000 0.6095 10.971 12.000 18
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 6.72658
BMDL = 4.67475
BMDU = 10.3673
D-ll
DRAFT - DO NOT CITE OR QUOTE
-------
D.5. ONE-HUNDRED-FOUR-WEEK STUDY FROM DeANGELO ET AL. (2008) WITH
THREE DOSE GROUPS
Multistage Cancer Model with 0.95 Confidence Level
0.8
0.7
0.6
0.5
0.4
0.3
Multistage Cancer
Linear extrapolation
BMD Lower Bound
BMDL
BMD
6
dose
10
12
Note: Doses on x-axis are human equivalent doses for lifetime exposure in units of mg/kg-day.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_TCALiverTumorStudy3_LiverCacerPol2.(d)
Gnuplot Plotting File:
C:\USEPA\BMDS21\Data\TCA\LiverCancer\msc_TCALiverTumorStudy3_LiverCacerPol2.pit
BMDS Model Run
The form of the probability function is:
Dependent variable = Incidence
Independent variable = Dose
Total number of observations = 3
Total number of records with missing value
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Default Initial Parameter Values
D-12
DRAFT - DO NOT CITE OR QUOTE
-------
Asymptotic Correlation Matrix of Parameter Estimates
Background
Background 1
Beta(2) -0.48
Std. Err.
Indicates that this value is not calculated.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
Model
Full model
Fitted model
Reduced model
Log(likelihood) # Param's Deviance Test d.f.
-102.285 3
-103.003 2 1.4352 1
-106.011 1 7.4518 2
Dose
Chi'
d.f. = 1
P-value = 0.2314
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0.1
Extra risk
0. 95
5.70799
1.50402
10.2515
D-13
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX E. MULTISTAGE-WEIBULL (MSW) TIME-TO-TUMOR MODELING OF
INDIVIDUAL AND COMBINED LIVER TUMOR INCIDENCE DATA SETS FROM
DeANGELO ET AL. (2008)
The findings of three chronic drinking water bioassays of TCA were reported by
DeAngelo et al. (2008) in male B6C3Fi mice. Key characteristics of the three bioassays are
presented in Table E-l.
Table E-l. Key characteristics of the three drinking water studies
Study number
Study 1
Study 2
Study 3
Study duration (wks)
60
104
104
Dose groups
0, 0.05, 0.5, 5 g/L TCA
0, 4.5 g/L TCA
0, 0.05, 0.5 g/L TCA
Source: DeAngelo et al. (2008).
Consideration was given to combining the liver tumor incidence data from the three
bioassays in order to derive an oral cancer slope factor based on a wider range of doses and a
larger number of mice. To determine whether the incidence data from these three studies could
be combined, a statistical analysis was conducted employing a generalized likelihood ratio test
(Stiteler et al., 1993) after both individual and combined data sets were fitted by the MSW time-
to-tumor model.
The results of the generalized likelihood ratio test of statistical compatibility are
presented in Section E.4 and the results of the dose-response analysis based on combined data
sets are presented in Section E.5.
E.I. DOSE CONVERSIONS
Before fitting the MSW time-to-tumor model to the liver tumor incidence data, estimated
mean daily intakes of TCA from the mouse studies were converted to human equivalent lifetime
doses by adjusting for continuous exposure time (in this case, the adjustment factor was 1) and
applying body weight scaling factors (see Tables E-2 to E-4).
E-l DRAFT - DO NOT CITE OR QUOTE
-------
Table E-2. Dose adjustments for Study 1
TCA dose
group (g/L)
0
0.05
0.5
5.0
Mean water
consumption
(mL/kg-d)a
171
153
141
119
Mean measured
TCA
concentration
(mg/mL)a
Not applicable
Not applicable
0.48
5.06
Estimated
mean intake
(mg/kg-d)b
0.00
7.70
68.16
602.14
Average
animal
lifetime
weight (g)c
38.0
38.0
37.7
36.0
Body weight
scaling
factord
0.15
0.15
0.15
0.15
Human
equivalent
lifetime dose
(mg/kg-d)e
0.0
1.2
10.4
90.7
"Reported by DeAngelo et al. (2008).
Estimated mean daily intake = mean water consumption x mean measured TCA concentration; where the mean
measured TCA concentration was not reported, the nominal concentration for the dose group was used to calculate
the estimated mean daily intake.
Calculated using animal body weights at different weeks reported by DeAngelo et al. (2008).
dBody weight scaling factor = [average animal lifetime weight (kg)/human reference body weight (i.e., 70 kg)]0 25.
eHuman equivalent lifetime dose = estimated mean daily intake x body weight scaling factor.
Table E-3. Dose adjustments for Study 2
TCA dose
group (g/L)
0
4.5
Mean water
consumption
(mL/kg-d)a
132
129
Mean measured
TCA
concentration
(mg/mL)a
Not applicable
4.43
Estimated
mean intake
(mg/kg-d)b
0.00
571.47
Average
animal
lifetime
weight (g)c
37.6
63.1
Body weight
scaling
factor"1
0.15
0.15
Human
equivalent
lifetime dose
(mg/kg-d)e
0
86.1
"Reported by DeAngelo et al. (2008).
Estimated mean daily intake = mean water consumption x mean measured TCA concentration.
Calculated using animal body weights at different weeks reported by DeAngelo et al. (2008).
dBody weight scaling factor = [average animal lifetime weight (kg)/human reference body weight (i.e., 70 kg)]0
eHuman equivalent lifetime dose = estimated mean daily intake x body weight scaling factor.
Table E-4. Dose adjustments for Study 3
TCA dose
group (g/L)
0
0.05
0.5
Mean water
consumption
(mL/kg-d)a
112
111
116
Mean measured
TCA
concentration
(mg/mL)a
Not applicable
0.06
0.70
Estimated
mean intake
(mg/kg-d)b
0
6.66
81.20
Average
animal
lifetime
weight (g)c
42.2
42.5
42.6
Body weight
scaling
factor"1
0.16
0.16
0.16
Human
equivalent
lifetime dose
(mg/kg-d)e
0.0
1.0
12.8
"Reported by DeAngelo et al. (2008).
Estimated mean daily intake = mean water consumption x mean measured TCA concentration.
Calculated using animal body weights at different weeks reported by DeAngelo et al. (2008).
dBody weight scaling factor = [average animal lifetime weight (kg)/human reference body weight (i.e., 70 kg)]0 25.
eHuman equivalent lifetime dose = estimated mean daily intake x body weight scaling factor.
E-2
DRAFT - DO NOT CITE OR QUOTE
-------
E.2. DOSE-RESPONSE DATA
Individual animal data for the three bioassays reported in DeAngelo et al. (2008) were
obtained from the study author (emails dated February 1 and April 26, 2010, from Anthony
DeAngelo, NHEERL, ORD, U.S. EPA, to Diana Wong, NCEA, ORD, U.S. EPA). Before fitting
MSW time-to-tumor models, each animal was classified into one of three response categories:
"I" (hepatocellular carcinoma and/or adenoma were detected when the mouse was removed from
the study due to scheduled sacrifice or unscheduled death), "U" (the presence or absence of
hepatocellular carcinoma and/or adenoma could not be determined when the mouse was removed
from the study due to scheduled sacrifice or unscheduled death or other reasons), and "C"
(neither hepatocellular carcinoma nor adenoma was detected when the mouse was removed from
the study due to scheduled sacrifice or unscheduled death). See Tables E-5 to E-7 for details.
E-3 DRAFT - DO NOT CITE OR QUOTE
-------
Table E-5. Study 1 liver tumor incidence data; B6C3Fi male mice exposed
to TCA in drinking water
Human lifetime
equivalent dose
(mg/kg-d)
0
1.2
10.4
90.7
Wk of death
5
15
30
45
60
60
1
2
5
15
30
45
45
60
60
60
5
15
30
45
45
60
60
5
15
30
45
45
60
60
60
Response category for
hepatocellular carcinoma and/or
adenoma"
C
C
C
C
C
I
u
u
C
C
C
C
I
C
I
u
C
C
C
C
I
I
C
C
C
C
C
I
C
I
u
Number of animals
5
5
5
5
26
4
1
1
5
5
5
4
1
23
4
1
5
5
5
4
1
11
18
5
5
5
2
3
13
16
1
"Response categories:
C: Neither hepatocellular carcinoma nor adenoma was detected when the mouse was removed from the study
due to scheduled sacrifice or unscheduled death.
U: The presence or absence of hepatocellular carcinoma and/or adenoma could not be determined when the
mouse was removed from the study due to scheduled sacrifice or unscheduled death or other reasons.
I: Hepatocellular carcinoma and/or adenoma were detected when the mouse was removed from the study due to
scheduled sacrifice or unscheduled death.
E-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-6. Study 2 liver tumor incidence data; B6C3Fi male mice exposed
to TCA in drinking water
Human lifetime
equivalent dose
(mg/kg-d)
0
86.1
Wk of death
16
29
31
45
60
71
95
98
102
105
105
106
106
15
23
27
30
41
45
45
72
82
89
89
92
94
94
104
104
Response category for
hepatocellular carcinoma and/or
adenoma"
C
u
C
C
C
u
u
u
u
C
I
C
I
C
u
u
C
I
C
I
I
I
I
u
u
u
I
C
I
Number of animals
5
2
5
5
10
1
1
1
1
4
1
18
2
5
1
1
5
1
4
1
1
1
1
1
2
1
1
4
26
"Response categories:
C: Neither hepatocellular carcinoma nor adenoma was detected when the mouse was removed from the study due
to scheduled sacrifice or unscheduled death.
U: The presence or absence of hepatocellular carcinoma and/or adenoma could not be determined when the
mouse was removed from the study due to scheduled sacrifice or unscheduled death or other reasons.
I: Hepatocellular carcinoma and/or adenoma were detected when the mouse was removed from the study due to
scheduled sacrifice or unscheduled death.
E-5
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-7. Study 3 liver tumor incidence data; B6C3Fi male mice exposed
to TCA in drinking water
Human lifetime
equivalent dose
(mg/kg-d)
0
1.0
Wk of death
17
26
43
52
52
72
76
78
78
79
80
81
84
86
86
89
91
95
98
99
101
101
104
104
26
38
39
52
52
59
70
72
73
75
76
77
78
78
79
79
80
83
86
89
90
Response category for
hepatocellular carcinoma and/or
adenoma"
U
c
U
c
I
U
U
I
c
U
c
I
I
c
U
I
c
I
U
U
U
I
I
c
c
U
U
c
I
U
U
c
U
U
U
U
c
I
U
c
c
I
c
U
U
Number of animals
1
7
1
5
2
1
1
2
5
22
12
7
1
1
6
1
2
2
1
1
1
1
2
4
E-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-7. Study 3 liver tumor incidence data; B6C3Fi male mice exposed
to TCA in drinking water
Human lifetime
equivalent dose
(mg/kg-d)
12.8
Wk of death
91
94
100
104
104
104
26
50
52
55
56
61
69
72
78
78
79
80
82
84
86
91
92
92
94
97
99
100
103
104
104
104
Response category for
hepatocellular carcinoma and/or
adenoma"
I
U
U
c
I
U
c
U
c
U
U
U
U
U
c
I
I
U
U
I
I
U
U
I
U
c
I
I
U
I
c
U
Number of animals
1
1
1
13
17
1
7
1
7
4
4
1
1
1
1
1
1
1
2
1
1
2
1
1
24
3
2
"Response categories:
C: Neither hepatocellular carcinoma nor adenoma was detected when the mouse was removed from the study due
to scheduled sacrifice or unscheduled death.
U: The presence or absence of hepatocellular carcinoma and/or adenoma could not be determined when the
mouse was removed from the study due to scheduled sacrifice or unscheduled death or other reasons.
I: Hepatocellular carcinoma and/or adenoma were detected when the mouse was removed from the study due to
scheduled sacrifice or unscheduled death.
E.3. MSW TIME-TO-TUMOR MODELING
MSW time-to-tumor modeling is used to model both the dose and the time of appearance
of a detectable tumor. With this model, the probability of observing a tumor prior to some
E-7
DRAFT - DO NOT CITE OR QUOTE
-------
specific observation time, t, upon exposure to a carcinogen at dose level, d, is given by the
function:
MSW time-to-tumor models were fit to three individual liver tumor data sets (i.e.,
Study 1, Study 2, and Study 3) and four combined data sets (i.e., Study 1+2+3, Study 1+2,
Study 1+3, and Study 2+3). For each individual or combined data set, specific n-stage model
was selected because of lowest log-likelihoods and/or AIC. A tumor incidental risk of 10% was
used for low-dose extrapolation. Critical outputs and plots are provided in Section E.6.
The MSW time-to-tumor modeling software program, available for download from the
EPA's HMDS website (U.S. EPA, 2009), was used to conduct the MSW time-to-tumor analysis;
EPA's gofplot_msw() was used to produce plots to assess goodness-of-fit for the MSW time-to-
tumor models.
E.4. STATISTICAL ANALYSIS FOR DATA COMPATIBILITY
To evaluate whether the three independent studies from DeAngelo et al. (2008) were
compatible to be combined for MSW time-to-tumor modeling, a generalized likelihood ratio test
described by Stiteler et al. (1993) was used, which has an asymptotic %2 distribution:
-2LnA= 2[max In L(H0 UH})-max In L(H0)]
Two hypotheses were tested: (1) the null hypothesis (H0), i.e., that the data sets from
individual studies are compatible to be combined for MSW time-to-tumor modeling; and (2) the
alternative hypothesis (Hi), i.e., that the data sets from individual studies are not compatible to
be combined for MSW time-to-tumor modeling.
The determination to either to accept or reject the null hypothesis was made by
comparing the calculated value of-2LnA against the tabulated %2 at the level of significance a =
0.05 and a = 0.025 for one-sided distributions (see Table E-8). Based on this statistical analysis,
only Study 1 and Study 3 were determined to be statistically compatible to be combined for
MSW time-to-tumor modeling.
E-8 DRAFT - DO NOT CITE OR QUOTE
-------
Table E-8. Summary of the statistical test for compatibility among the
individual studies
Study 1, log (likelihood)
Study 2, log (likelihood)
Study 3, log (likelihood)
Combined data sets, log
(likelihood)
-21n/l
Degree of freedom
Critical %2i,0.o5, one sided
Critical %2i, 0.025, one sided
Critical y?2,o .05, one sided
Critical ^2,0.025, one sided
Conclusion
Study 1+2+3
-76.44
-33.71
-109.20
-238.35
37.99
2
5.99
7.38
Reject H0, not
compatible
Study 1+3
-76.44
-109.20
-187.49
3.71
1
3.84
5.02
Accept H0,
compatible
Study 1+2
-76.44
-33.71
-117.36
14.41
1
3.84
5.02
Reject H0, not
compatible
Study 2+3
-33.71
-109.20
-158.52
31.21
1
3.84
5.02
Reject H0, not
compatible
Example for calculation:
Study 1, Log(likelihood) = -76.44
Study 3, Log(likelihood) = -109.20
Study 1+3, Log(likelihood) = -187.49
-2LnA = 2[187.49 - (76.44 + 109.20)] = 3.7, degree of freedom (df) = 1
X i, o.os = 3.84 (df = 1, a = 0.05),
0.025
= 5.02 (df = 1, a = 0.025)
The calculated -2LnA for combining data sets from Study 1 and Study 3 equaled 3.7,
which is smaller than the right-sided critical value of $ distribution for a = 0.05 or a = 0.025 at
df = 1. Therefore, the null hypothesis was accepted, indicating that Study 1 and Study 3 were
statistically compatible to be combined for MSW time-to-tumor modeling.
E.5. EXTRAPOLATION METHOD AND ORAL CANCER SLOPE FACTOR
As discussed in Section 5.4.4, linear extrapolation was applied in this assessment and
BMDLio was used as POD for linear extrapolation. The oral cancer slope factor, the upper-
bound estimation of risk, was calculated as O.I/ BMDLio. The cancer slope factors derived using
MSW time-to-tumor model software are provided in Table E-9.
E-9
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-9. Candidate oral cancer slope factors derived from liver tumor
data sets in B6C3Fi male mice using MSW time-to-tumor modeling
Study 1
Study 2
Study 3
Study 1+3
Best time-to-
tumor model
for the study
Stage 1
polynomial
Stage 1
polynomial
Stage 2
polynomial
Stage 1
polynomial
Log
(likelihood)
-76.4
-33.7
-109.2
-187.5
AIC
158.9
73.4
226.4
381.0
BMR
0.1
0.1
0.1
0.1
BMD10a
13.5
5.4
5.0
2.2
BMDL10b
8.4
3.8
1.2
1.4
Slope of linear
extrapolation
from BMD10C
7.4 x 10'3
1.9 x 10'2
2.0 x 10'2
4.5 x 10'2
Cancer slope
factor from
BMDL10d
1.2 x 10'2
2.6 x 10'2
8.5 x 10'2
7.2 x 10'2
aBMD10 = dose at 10% cancer risk.
bBMDL10 = dose at 95% lower bound with 10% cancer risk.
°Slope of linear extrapolation from BMD10 = 0.1/BMD10.
dCancer slope factor = 0.1/BMDL10.
E-10
DRAFT - DO NOT CITE OR QUOTE
-------
E.6. OUTPUT FILES AND PLOTS FOR MSW TIME-TO-TUMOR MODELS
E.6.1. Study 1 from DeAngelo et al. (2008); 60-Week Study with Four Dose Groups
E.6.1.1. MSW Time-to-Tumor Model Run
Timer to Tumor Model, TCA, DeAngelo et al, Study 1, Poly 1
Total number of observations = 199
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
User specifies the following parameters:
t_0 = 0
Maximum number of iterations = 64
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t_0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
c beta_0 beta_l
1 -1 -1
-1
beta_l -1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
c 3.23626 1.63392 0.0338329 6.43868
3.53142e-007 2.34711e-006 -4.2471e-006 4.95339e-006
E-11 DRAFT - DO NOT CITE OR QUOTE
-------
beta 1
Log(likelihood) # Param
Fitted Model -76.4417 3
Data Summary
CONTEXT
C F I
DOSE
0 46 0 4
1.2 42 0 5
10 37 0 12
91 30 0 19
U Total Expected Response
0 50 5.95
3 50 5.83
0 49 7.82
1 50 19.96
Benchmark Dose Computation
Risk Response =
Risk Type
Specified effect =
Confidence level =
Time
BMD =
BMDL =
BMDU =
E.6.1.2. MSW Time-to-Tumor Plots
Incidental Risk: Study1_P1_TCA_DeAngelo
Dose = 0.00 Dose = 1.20
i I I I I I
0 10 20 30 40 50 60
Time
I I I I I \
0 10 20 30 40 50 60
Time
n i i i i i
0 10 20 30 40 50 60
Time
\ I I I I I T
0 10 20 30 40 50
Time
E-12
DRAFT - DO NOT CITE OR QUOTE
-------
E.6.2. Study 2 from DeAngelo et al. (2008); 104-Week Study with Two Dose Groups
E.6.2.1. MSW Time-to-Tumor Model Run
Multistage Weibull Model. (Version: 1.6.1; Date: 11/24/2009)
Solutions are obtained using donlp2-intv, (c) by P. Spellucci
Input Data File: TCA-DeAngelo-St2-Pl.(d)
Timer to Tumor Model, TCA, DeAngelo et al, Study 2, Poly 1
The form of the probability function is:
P[response] = l-EXP{-(t - t_0)~c *
(beta 0+beta l*dose^l)}
Dependent variable = CONTEXT
Independent variables = DOSE, TIME
Total number of observations = 112
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
Default Initial Parameter Values
c = 2.57143
t_0 = 0 Specified
beta_0 = 6.55001e-007
beta 1 = 1.26153e-007
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t 0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
-1
-1
E-13 DRAFT - DO NOT CITE OR QUOTE
-------
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
c 2.65637 1.11083 0.479187 4.83356
beta_0 4.44615e-007 2.28279e-006 -4.02958e-006 4.91881e-006
beta 1 8.57058e-008 4.34281e-007 -7.65469e-007 9.36881e-007
Log(likelihood) # Param AIC
Fitted Model -33.7142 3 73.4284
Data Summary
CONTEXT
F I
3
32
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
Time = 104
BMD
BMDL
BMDU
E-14 DRAFT - DO NOT CITE OR QUOTE
-------
E.6.2.2. MSW Time-to-Tumor Plots
Incidental Risk: Study2_P1_TCA_DeAngelo
Dose = 0.00 Dose = 86.20
CO
ci
\ i r
0 20 40 60 80 100
Time
CO
ci
q
ci
I I I I T
0 20 40 60 80 100
Time
E.6.3. Study 3 from DeAngelo et al. (2008); 104-Week Study with Three Dose Groups
E.6.3.1. MSW Time-to-Tumor Model Run
Timer to Tumor Model, TCA, DeAngelo et al, Study 3, Poly2
The parameter betas are restricted to be positive
Dependent variable = CONTEXT
Independent variables = DOSE, TIME
Total number of observations = 216
Total number of records with missing values = 0
Total number of parameters in model = 5
Total number of specified parameters = 1
Degree of polynomial = 2
User specifies the following parameters:
t 0 0
Default Initial Parameter Values
c 3
t_0 = 0 Specified
beta_0 = 7.0609e-007
beta_l = 2.30386e-028
beta 2 = 3.14871e-009
E-15
DRAFT - DO NOT CITE OR QUOTE
-------
( *** The model parameter (s) -t 0 -beta 1
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
-1
-1
Variable
Parameter Estimates
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
4.11251 1.07308 2.00931 6.21571
4.40074e-009 2.16313e-008 -3.79957e-008 4.67972e-008
0 NA
2.09521e-011 1.02607e-010 -1.80154e-010 2.22058e-010
NA - Indicates that this parameter has hit a
bound implied by some inequality constraint
and thus has no standard error.
Log(likelihood) # Param
Fitted Model -109.195 4
1
13
32
34
22
Data Summary
CONTEXT
F I
U Total Expected Response
72
72
72
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
Time = 104
BMD
BMDL
BMDU
E-16
DRAFT - DO NOT CITE OR QUOTE
-------
E.6.3.2. MSW Time-to-Tumor Plots
Incidental Risk: Study3_P2_TCA_DeAngelo
Dose = 0.00 Dose = 1.00
\ i i i ir
0 20 40 60 80 100
Time
>. o
i i'—i 1 1 r
0 20 40 60 80 100
Time
Dose= 12.80
\ i i i i r
0 20 40 60 80 100
Time
E.6.4. Combined Dataset (Study 1+3) from DeAngelo et al. (2008)
E.6.4.1. MSW Time-to-Tumor Model Run
Timer to Tumor Model, TCA, DeAngelo et al. Combine Study 1 and 3, Exact Adj. Doses, Polyl
The form of the probability function is:
P[response] = l-EXP{-(t - t_0)^c *
(beta_0+beta_l*dose^l) }
The parameter betas are restricted to be positive
Dependent variable = CONTEXT
Independent variables = DOSE, TIME
Total number of observations = 415
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
User specifies the following parameters:
t 0 0
E-17
DRAFT - DO NOT CITE OR QUOTE
-------
Default Initial Parameter Values
c = 3.6
t_0 = 0 Specified
beta_0 = 4.78773e-008
beta 1 = 3.31927e-009
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t 0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
-1
-1
Parameter Estimates
/ariable
beta_0
beta 1
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
2.12983 4.19636
-1.29018e-006 1.98786e-006
-6.50113e-008 1.04871e-007
Log(likelihood) # Param
Fitted Model -187.491 3
Data Summary
CONTEXT
F I
U Total Expected Response
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
BMD
BMDL
E-18
DRAFT - DO NOT CITE OR QUOTE
-------
BMDU =
E.6.4.2. MSW Time-to-Tumor Plots
Incidental Risk: Combine_1and3_P1_TCA_DeAngelo
Dose = 0.00 Dose = 1.00
e
D_
rn i —i 1—r
0 20 40 60 80 100
Time
e
o_
^ i 1 1 r
0 20 40 60 80 100
Time
Dose= 12.80
i r—i—^ 1—r~
0 20 40 60 80 100
Time
CO
e
Q_
i\ \ \
0 20 40 60 80 100
Time
Dose= 10.40
e
D_
I I I I
0 20 40 60 80 100
Time
Dose = 90.70
0 I I I I I T
0 20 40
100
Time
E.6.5. Other Combined Data Sets from DeAngelo et al. (2008)
E.6.5.1. MSWTime-to-Tumor Model Run for Combining Study 1, Study 2, and Study 3
E-19
DRAFT - DO NOT CITE OR QUOTE
-------
Total number of observations = 527
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
Maximum number of iterations = 64
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
Default Initial Parameter Values
c = 2.57143
t_0 = 0 Specified
beta_0 = 4.20208e-006
beta 1 = 1.4128e-007
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t 0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
-1
-1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
c 2.51012 0.393072 1.73971 3.28052
beta_0 5.52994e-006 9.77318e-006 -1.36251e-005 2.4685e-005
beta 1 1.84135e-007 3.1736e-007 -4.37878e-007 8.06149e-007
Log(likelihood) # Param
Fitted Model -238.347 3
Data Summary
CONTEXT
38
21
14
E-20 DRAFT - DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
Time = 104
BMD
BMDL
BMDU
E.6.5.2. MSW Time-to-Tumor Model Run for Combining Study 1 and Study 2
Multistage Weibull Model. (Version: 1.6.1; Date: 11/24/2009)
Solutions are obtained using donlp2-intv, (c) by P. Spellucci
Input Data File: Combineland2-Pl-TCA-DeAngelo.(d)
Timer to Tumor Model, TCA, DeAngelo et al, Combine Study 1 and 2, Polyl
The form of the probability function is:
P[response] = l-EXP{-(t - t_0)~c *
(beta 0+beta l*dose^l)}
Dependent variable = CONTEXT
Independent variables = DOSE, TIME
Total number of observations = 311
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
User specifies the following parameters:
t_0 = 0
Maximum number of iterations = 64
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
Default Initial Parameter Values
c = 1.8
t_0 = 0 Specified
beta 0 = 6.38638e-005
E-21 DRAFT - DO NOT CITE OR QUOTE
-------
beta 1 = 4.37061e-00£
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t 0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Variable
beta_0
beta 1
95.0% Wald Confidence Interval
Lower Conf . Limit Upper Conf . Limit
0.898849 2.56759
-0.000221989 0.000391616
-1 . 51641e-005 2.67997e-005
Log(likelihood) # Param
Fitted Model -117.363 3
Data Summary
CONTEXT
F I
6 106 11.58
3 50 3.49
0 49 5.52
6 56 34.78
1 50 18.86
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
BMD
BMDL
BMDU
E-22
DRAFT - DO NOT CITE OR QUOTE
-------
E.6.5.3. MSW Time-to-Tumor Model Run for Combining Study 2 and Study 3
Dependent variable = CONTEXT
Independent variables = DOSE, TIME
Total number of observations = 328
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 1
Degree of polynomial = 1
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -t_0
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
_ beta_l
1 -1
_ -1 1
beta 1
E-23 DRAFT - DO NOT CITE OR QUOTE
-------
Parameter Estimates
/ariable
beta_0
beta 1
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
1.8207 4.82262
-7.89212e-007 1.05866e-006
-2.11176e-008 2.82629e-008
Log(likelihood) # Param
Fitted Model -158.516
1
13
Data Summary
CONTEXT
F I
U Total Expected Response
15
17
14
Benchmark Dose Computation
Risk Response = Incidental
Risk Type = Extra
Specified effect = 0.1
Confidence level = 0.9
BMD
BMDL
BMDU
E-24
DRAFT - DO NOT CITE OR QUOTE
------- |