EPA/635/R-09/011F

www.epa.gov/iris





TOXICOLOGICAL REVIEW

OF

TRICHLOROETHYLENE

(CAS No. 79-01-6)


In Support of Summary Information on the
Integrated Risk Information System (IRIS)


September 2011















U.S. Environmental Protection Agency
Washington, DC
ii
DISCLAIMER


This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
iii
GUIDE TO READERS OF THIS DOCUMENT


Due to the length of the TCE toxicological review, it is recommended that
Chapters 1 and 6 be read prior to Chapters 25.
Chapter 1 is the standard introduction to an IRIS Toxicological Review, describing the
purpose of the assessment and the guidelines used in its development.
Chapter 2 is an exposure characterization that summarizes information about TCE
sources, releases, media levels, and exposure pathways for the general population (occupational
exposure is also discussed to a lesser extent).
Chapter 3 describes the toxicokinetics and physiologically based pharmacokinetic
(PBPK) modeling of TCE and metabolites (PBPK modeling details are in Appendix A).
Chapter 4 is the hazard characterization of TCE. Section 4.1 summarizes the evaluation
of epidemiologic studies of cancer and TCE (qualitative details in Appendix B; meta-analyses in
Appendix C). Each of the Sections 4.24.9 provides a self-contained summary and syntheses of
the epidemiologic and laboratory studies on TCE and metabolites, organized by tissue/type of
effects, in the following order: genetic toxicity, central nervous system (CNS), kidney, liver,
immune system, respiratory tract, reproduction and development, and other cancers. Additional
details are provided in Appendix D for CNS effects and in Appendix E for liver effects.
Section 4.10 summarizes the available data on susceptible lifestages and populations.
Section 4.11 describes the overall hazard characterization, including the weight of evidence for
noncancer effects and for carcinogenicity.
Chapter 5 is the dose-response assessment of TCE. Section 5.1 describes the dose-
response analyses for noncancer effects, and Section 5.2 describes the dose-response analyses for
cancer. Additional computational details are described in Appendix F for noncancer dose-
response analyses, Appendix G for cancer dose-response analyses based on rodent bioassays, and
Appendix H for cancer dose-response analyses based on human epidemiologic data.
Chapter 6 is the summary of the major conclusions in the characterization of TCE hazard
and dose response.
Appendix I contains the summary of EPAs response to major external peer review and
public comments.
iv
CONTENTS of TOXICOLOGICAL REVIEW for TRICHLOROETHYLENE
(CAS No. 79-01-6)


TOXICOLOGICAL REVIEW OF TRICHLOROETHYLENE ................................................... i
DISCLAIMER ................................................................................................................................ ii
GUIDE TO READERS OF THIS DOCUMENT .......................................................................... iii
CONTENTS of TOXICOLOGICAL REVIEW for TRICHLOROETHYLENE .......................... iv
LIST OF TABLES ....................................................................................................................... xiv
LIST OF FIGURES .....................................................................................................................xxv
LIST OF ABBREVIATIONS AND ACRONYMS .................................................................. xxix
FOREWORD .............................................................................................................................xxxv
AUTHORS, CONTRIBUTORS, AND REVIEWERS ........................................................... xxxvi
EXECUTIVE SUMMARY ......................................................................................................... xlii
1. INTRODUCTION ................................................................................................................... 1-1
2. EXPOSURE CHARACTERIZATION ................................................................................... 2-1
2.1. ENVIRONMENTAL SOURCES .............................................................................. 2-1
2.2. ENVIRONMENTAL FATE ...................................................................................... 2-6
2.2.1. Fate in Terrestrial Environments ................................................................... 2-6
2.2.2. Fate in the Atmosphere .................................................................................. 2-6
2.2.3. Fate in Aquatic Environments ....................................................................... 2-6
2.3. EXPOSURE CONCENTRATIONS .......................................................................... 2-6
2.3.1. Outdoor AirMeasured Levels .................................................................... 2-6
2.3.2. Outdoor AirModeled Levels ...................................................................... 2-8
2.3.3. Indoor Air..................................................................................................... 2-10
2.3.4. Water ............................................................................................................ 2-12
2.3.5. Other Media ................................................................................................. 2-14
2.3.6. Biological Monitoring .................................................................................. 2-15
2.4. EXPOSURE PATHWAYS AND LEVELS ............................................................ 2-16
2.4.1. General Population....................................................................................... 2-16
2.4.1.1. Inhalation ...................................................................................... 2-16
2.4.1.2. Ingestion ........................................................................................ 2-17
v
2.4.1.3. Dermal........................................................................................... 2-18
2.4.1.4. Exposure to TCE Related Compounds ......................................... 2-19
2.4.2. Potentially Highly Exposed Populations ..................................................... 2-20
2.4.2.1. Occupational Exposure ................................................................. 2-20
2.4.2.2. Consumer Exposure ...................................................................... 2-22
2.4.3. Exposure Standards ...................................................................................... 2-22
2.5. EXPOSURE SUMMARY ....................................................................................... 2-22
3. TOXICOKINETICS ................................................................................................................ 3-1
3.1. ABSORPTION .......................................................................................................... 3-2
3.1.1. Oral ................................................................................................................ 3-2
3.1.2. Inhalation ....................................................................................................... 3-4
3.1.3. Dermal............................................................................................................ 3-9
3.2. DISTRIBUTION AND BODY BURDEN ................................................................ 3-9
3.3. METABOLISM ....................................................................................................... 3-16
3.3.1. Introduction .................................................................................................. 3-16
3.3.2. Extent of Metabolism ................................................................................... 3-16
3.3.3. Pathways of Metabolism .............................................................................. 3-20
3.3.3.1. CYP-Dependent Oxidation ........................................................... 3-20
3.3.3.2. GSH Conjugation Pathway ........................................................... 3-35
3.3.3.3. Relative roles of the CYP and GSH pathways.............................. 3-50
3.4. TCE EXCRETION .................................................................................................. 3-53
3.4.1. Exhaled Air .................................................................................................. 3-53
3.4.2. Urine ............................................................................................................ 3-54
3.4.3. Feces ............................................................................................................ 3-56
3.5. PBPK MODELING OF TCE AND ITS METABOLITES ..................................... 3-57
3.5.1. Introduction .................................................................................................. 3-57
3.5.2. Previous PBPK Modeling of TCE for Risk Assessment
Application ................................................................................................... 3-57
3.5.3. Development and Evaluation of an Interim Harmonized TCE
PBPK Model ................................................................................................ 3-59
3.5.4. PBPK Model for TCE and Metabolites Used for This
Assessment ................................................................................................... 3-60
3.5.4.1. Introduction ................................................................................... 3-60
3.5.4.2. Updated PBPK Model Structure ................................................... 3-63
3.5.4.3. Specification of Baseline PBPK Model Parameter ....................... 3-65
3.5.4.4. Dose-Metric Predictions ............................................................... 3-67
3.5.5. Bayesian Estimation of PBPK Model Parameters, and Their
Uncertainty and Variability.......................................................................... 3-68
3.5.5.1. Updated Pharmacokinetic Database ............................................. 3-68
3.5.5.2. Updated Hierarchical Population Statistical Model
and Prior Distributions .................................................................. 3-75
3.5.5.3. Use of Interspecies Scaling to Update Prior
Distributions in the Absence of Other Data .................................. 3-77
vi
3.5.5.4. Implementation ............................................................................. 3-80
3.5.6. Evaluation of Updated PBPK Model ........................................................... 3-80
3.5.6.1. Convergence ................................................................................. 3-80
3.5.6.2. Evaluation of Posterior Parameter Distributions .......................... 3-82
3.5.6.3. Comparison of Model Predictions With Data ............................. 3-103
3.5.6.4. Sensitivity Analysis With Respect to Calibration Data .............. 3-126
3.5.6.5. Summary Evaluation of Updated PBPK Model ......................... 3-132
3.5.7. PBPK Model Dose-Metric Predictions ...................................................... 3-133
3.5.7.1. Characterization of Uncertainty and Variability ......................... 3-133
3.5.7.2. Local Sensitivity Analysis With Respect to Dose-
Metric Predictions ....................................................................... 3-142
3.5.7.3. Implications for the Population Pharmacokinetics of
TCE ............................................................................................. 3-155
3.5.7.4. Key Limitations and Potential Implications of
Violating Key Assumptions ........................................................ 3-162
3.5.7.5. Overall Evaluation of PBPK Model-Based Internal
Dose Predictions ......................................................................... 3-163
4. HAZARD CHARACTERIZATION ....................................................................................... 4-1
4.1. EPIDEMIOLOGIC STUDIES ON CANCER AND TCE
METHODOLOGICAL OVERVIEW ........................................................................ 4-1
4.2. GENETIC TOXICITY............................................................................................. 4-29
4.2.1. TCE .............................................................................................................. 4-30
4.2.1.1. DNA Binding Studies ................................................................... 4-30
4.2.1.2. Bacterial SystemsGene Mutations ............................................ 4-31
4.2.1.3. Fungal and Yeast SystemsGene Mutations,
Conversions, and Recombination ................................................. 4-35
4.2.1.4. Mammalian Systems Including Human Studies ........................... 4-37
4.2.1.5. Summary ....................................................................................... 4-48
4.2.2. TCA.............................................................................................................. 4-49
4.2.2.1. Bacterial SystemsGene Mutations ............................................ 4-49
4.2.2.2. Mammalian Systems ..................................................................... 4-51
4.2.2.3. Summary ....................................................................................... 4-55
4.2.3. DCA ............................................................................................................. 4-56
4.2.3.1. Bacterial and Fungal SystemsGene Mutations ......................... 4-56
4.2.3.2. Mammalian Systems ..................................................................... 4-60
4.2.3.3. Summary ....................................................................................... 4-62
4.2.4. CH ................................................................................................................ 4-62
4.2.4.1. DNA Binding Studies ................................................................... 4-62
4.2.4.2. Bacterial and Fungal SystemsGene Mutations ......................... 4-69
4.2.4.3. Mammalian Systems ..................................................................... 4-70
4.2.4.4. Summary ....................................................................................... 4-72
4.2.5. DCVC and DCVG ....................................................................................... 4-73
4.2.6. TCOH ........................................................................................................... 4-78
4.2.7. Synthesis and Overall Summary .................................................................. 4-79
4.3. CENTRAL NERVOUS SYSTEM (CNS) TOXICITY ........................................... 4-83
vii
4.3.1. Alterations in Nerve Conduction ................................................................. 4-83
4.3.1.1. Trigeminal Nerve Function: Human Studies ................................ 4-83
4.3.1.2. Nerve Conduction VelocityHuman Studies .............................. 4-87
4.3.1.3. Trigeminal Nerve Function: Laboratory Animal
Studies ........................................................................................... 4-88
4.3.1.4. Discussion and Conclusions: TCE-Induced
Trigeminal Nerve Impairment ...................................................... 4-89
4.3.2. Auditory Effects ........................................................................................... 4-91
4.3.2.1. Auditory Function: Human Studies .............................................. 4-91
4.3.2.2. Auditory Function: Laboratory Animal Studies ........................... 4-93
4.3.2.3. Summary and Conclusion of Auditory Effects ............................. 4-97
4.3.3. Vestibular Function ...................................................................................... 4-99
4.3.3.1. Vestibular Function: Human Studies ............................................ 4-99
4.3.3.2. Vestibular Function: Laboratory Animal Data ............................. 4-99
4.3.3.3. Summary and Conclusions for the Vestibular
Function Studies ......................................................................... 4-100
4.3.4. Visual Effects ............................................................................................. 4-101
4.3.4.1. Visual Effects: Human Studies ................................................... 4-101
4.3.4.2. Visual Effects: Laboratory Animal Data .................................... 4-103
4.3.4.3. Summary and Conclusion of Visual Effects ............................... 4-105
4.3.5. Cognitive Function..................................................................................... 4-106
4.3.5.1. Cognitive Effects: Human Studies .............................................. 4-106
4.3.5.2. Cognitive Effects: Laboratory Animal Studies ........................... 4-109
4.3.5.3. Summary and Conclusions of Cognitive Function
Studies ......................................................................................... 4-110
4.3.6. Psychomotor Effects .................................................................................. 4-111
4.3.6.1. Psychomotor Effects: Human Studies ........................................ 4-111
4.3.6.2. Psychomotor Effects: Laboratory Animal Data .......................... 4-114
4.3.6.3. Summary and Conclusions for Psychomotor Effects ................. 4-118
4.3.7. Mood Effects and Sleep Disorders ............................................................ 4-119
4.3.7.1. Effects on Mood: Human Studies ............................................... 4-119
4.3.7.2. Effects on Mood: Laboratory Animal Findings .......................... 4-120
4.3.7.3. Sleep Disturbances ...................................................................... 4-120
4.3.8. Developmental Neurotoxicity .................................................................... 4-120
4.3.8.1. Human Studies ............................................................................ 4-120
4.3.8.2. Animal Studies ............................................................................ 4-121
4.3.8.3. Summary and Conclusions for the Developmental
Neurotoxicity Studies ................................................................. 4-125
4.3.9. Mechanistic Studies of TCE Neurotoxicity ............................................... 4-125
4.3.9.1. Dopamine Neuron Disruption ..................................................... 4-125
4.3.9.2. Neurochemical and Molecular Changes ..................................... 4-127
4.3.10. Potential Mechanisms for TCE-Mediated Neurotoxicity .......................... 4-131
4.3.11. Overall Summary and ConclusionsWeight of Evidence ....................... 4-133
4.4. KIDNEY TOXICITY AND CANCER ................................................................. 4-137
4.4.1. Human Studies of Kidney .......................................................................... 4-137
4.4.1.1. Nonspecific Markers of Nephrotoxicity ..................................... 4-137
4.4.1.2. End-Stage Renal Disease (ESRD) .............................................. 4-143
4.4.2. Human Studies of Kidney Cancer.............................................................. 4-143
viii
4.4.2.1. Studies of Job Titles and Occupations with Historical
TCE Usage .................................................................................. 4-154
4.4.2.2. Cohort and Case-Controls Studies of TCE Exposure ................. 4-155
4.4.2.3. Examination of Possible Confounding Factors........................... 4-160
4.4.2.4. Susceptible PopulationsKidney Cancer and TCE
Exposure ..................................................................................... 4-164
4.4.2.5. Meta-Analysis for Kidney Cancer .............................................. 4-166
4.4.3. Human Studies of Somatic Mutation of VHL Gene................................... 4-172
4.4.4. Kidney Noncancer Toxicity in Laboratory Animals ................................. 4-177
4.4.5. Kidney Cancer in Laboratory Animals ...................................................... 4-185
4.4.5.1. Inhalation Studies of TCE ........................................................... 4-185
4.4.5.2. Gavage and Drinking Water Studies of TCE .............................. 4-187
4.4.5.3. Conclusions: Kidney Cancer in Laboratory Animals ................. 4-188
4.4.6. Role of Metabolism in TCE Kidney Toxicity............................................ 4-188
4.4.6.1. In Vivo Studies of the Kidney Toxicity of TCE
Metabolites.................................................................................. 4-189
4.4.6.2. In Vitro Studies of Kidney Toxicity of TCE and
Metabolites.................................................................................. 4-197
4.4.6.3. Conclusions as to the Active Agents of TCE-Induced
Nephrotoxicity ............................................................................ 4-199
4.4.7. Mode(s) of Action for Kidney Carcinogenicity ......................................... 4-199
4.4.7.1. Hypothesized Mode of Action: Mutagenicity............................. 4-200
4.4.7.2. Hypothesized Mode of Action: Cytotoxicity and
Regenerative Proliferation .......................................................... 4-209
4.4.7.3. Additional Hypothesized Modes of Action with
Limited Evidence or Inadequate Experimental
Support ........................................................................................ 4-212
4.4.7.4. Conclusions About the Hypothesized Modes of
Action.......................................................................................... 4-213
4.4.8. Summary: TCE Kidney Toxicity, Carcinogenicity, and Mode
of Action .................................................................................................... 4-216
4.5. LIVER TOXICITY AND CANCER ..................................................................... 4-218
4.5.1. Liver Noncancer Toxicity in Humans........................................................ 4-218
4.5.2. Liver Cancer in Humans ............................................................................ 4-224
4.5.3. Experimental Studies of TCE in RodentsIntroduction .......................... 4-239
4.5.4. TCE-Induced Liver Noncancer Effects ..................................................... 4-241
4.5.4.1. Liver Weight ............................................................................... 4-247
4.5.4.2. Cytotoxicity and Histopathology ................................................ 4-251
4.5.4.3. Measures of DNA Synthesis, Cellular Proliferation,
and Apoptosis ............................................................................. 4-258
4.5.4.4. Peroxisomal Proliferation and Related Effects ........................... 4-261
4.5.4.5. Oxidative Stress .......................................................................... 4-263
4.5.4.6. Bile Production ........................................................................... 4-264
4.5.4.7. Summary: TCE-Induced Noncancer Effects in
Laboratory Animals .................................................................... 4-265
4.5.5. TCE-Induced Liver Cancer in Laboratory Animals .................................. 4-266
4.5.5.1. Negative or Inconclusive Studies of Mice and Rats ................... 4-266
4.5.5.2. Positive TCE Studies of Mice ..................................................... 4-273
ix
4.5.5.3. Summary: TCE-Induced Cancer in Laboratory
Animals ....................................................................................... 4-275
4.5.6. Role of Metabolism in Liver Toxicity and Cancer .................................... 4-275
4.5.6.1. Pharmacokinetics of CH, TCA, and DCA from TCE
Exposure ..................................................................................... 4-276
4.5.6.2. Comparisons Between TCE and TCA, DCA, and CH
Noncancer Effects ....................................................................... 4-277
4.5.6.3. Comparisons of TCE-Induced Carcinogenic
Responses with TCA, DCA, and CH Studies ............................. 4-296
4.5.6.4. Conclusions Regarding the Role of TCA, DCA, and
CH in TCE-Induced Effects in the Liver .................................... 4-320
4.5.7. Mode of Action for TCE Liver Carcinogenicity ........................................ 4-321
4.5.7.1. Mutagenicity ............................................................................... 4-321
4.5.7.2. PPAR Receptor Activation ....................................................... 4-323
4.5.7.3. Additional Proposed Hypotheses and Key Events
with Limited Evidence or Inadequate Experimental
Support ........................................................................................ 4-332
4.5.7.4. Mode-of-Action Conclusions...................................................... 4-342
4.6. IMMUNOTOXICITY AND CANCERS OF THE IMMUNE
SYSTEM ................................................................................................................ 4-354
4.6.1. Human Studies ........................................................................................... 4-354
4.6.1.1. Noncancer Immune-Related Effects ........................................... 4-354
4.6.1.2. Cancers of the Immune System, Including Childhood
Leukemia .................................................................................... 4-366
4.6.2. Animal Studies ........................................................................................... 4-401
4.6.2.1. Immunosuppression .................................................................... 4-401
4.6.2.2. Hypersensitivity .......................................................................... 4-409
4.6.2.3. Autoimmunity ............................................................................. 4-412
4.6.2.4. Cancers of the Immune System .................................................. 4-424
4.6.3. Summary .................................................................................................... 4-427
4.6.3.1. Noncancer Effects ....................................................................... 4-427
4.6.3.2. Cancer ......................................................................................... 4-429
4.7. RESPIRATORY TRACT TOXICITY AND CANCER ....................................... 4-431
4.7.1. Epidemiologic Evidence ............................................................................ 4-431
4.7.1.1. Chronic Effects: Inhalation ......................................................... 4-431
4.7.1.2. Cancer ......................................................................................... 4-431
4.7.2. Laboratory Animal Studies ........................................................................ 4-443
4.7.2.1. Respiratory Tract Animal Toxicity ............................................. 4-443
4.7.2.2. Respiratory Tract Cancer ............................................................ 4-451
4.7.3. Role of Metabolism in Pulmonary Toxicity .............................................. 4-454
4.7.4. Mode of Action for Pulmonary Carcinogenicity ....................................... 4-459
4.7.4.1. Mutagenicity via Oxidative Metabolism .................................... 4-459
4.7.4.2. Cytotoxicity Leading to Increased Cell Proliferation ................. 4-461
4.7.4.3. Additional Hypothesized Modes of Action with
Limited Evidence or Inadequate Experimental
Support ........................................................................................ 4-462
x
4.7.4.4. Conclusions About the Hypothesized Modes of
Action.......................................................................................... 4-463
4.7.5. Summary and Conclusions ........................................................................ 4-465
4.8. REPRODUCTIVE AND DEVELOPMENTAL TOXICITY ................................ 4-467
4.8.1. Reproductive Toxicity ............................................................................... 4-467
4.8.1.1. Human Reproductive Outcome Data .......................................... 4-467
4.8.1.2. Animal Reproductive Toxicity Studies ....................................... 4-474
4.8.1.3. Discussion/Synthesis of Noncancer Reproductive
Toxicity Findings ........................................................................ 4-487
4.8.2. Cancers of the Reproductive System ......................................................... 4-493
4.8.2.1. Human Data ................................................................................ 4-493
4.8.2.2. Animal Studies ............................................................................ 4-509
4.8.2.3. Mode of Action for Testicular Tumors ....................................... 4-510
4.8.3. Developmental Toxicity............................................................................. 4-511
4.8.3.1. Human Developmental Data ....................................................... 4-511
4.8.3.2. Animal Developmental Toxicology Studies ............................... 4-534
4.8.3.3. Discussion/Synthesis of Developmental Data ............................ 4-556
4.9. OTHER SITE-SPECIFIC CANCERS ................................................................... 4-572
4.9.1. Esophageal Cancer ..................................................................................... 4-572
4.9.2. Bladder Cancer........................................................................................... 4-583
4.9.3. CNS and Brain Cancers ............................................................................. 4-590
4.10. SUSCEPTIBLE LIFESTAGES AND POPULATIONS ....................................... 4-595
4.10.1. Lifestages ................................................................................................... 4-595
4.10.1.1. Early Lifestages .......................................................................... 4-595
4.10.1.2. Later Lifestages ........................................................................... 4-605
4.10.2. Other Susceptibility Factors ....................................................................... 4-606
4.10.2.1. Gender ......................................................................................... 4-606
4.10.2.2. Genetic Variability ...................................................................... 4-611
4.10.2.3. Race/Ethnicity ............................................................................. 4-613
4.10.2.4. Preexisting Health Status ............................................................ 4-613
4.10.2.5. Lifestyle Factors and Nutrition Status ........................................ 4-614
4.10.2.6. Mixtures ...................................................................................... 4-617
4.10.3. Uncertainty of Database and Research Needs for Susceptible
Populations ................................................................................................. 4-617
4.11. HAZARD CHARACTERIZATION ..................................................................... 4-619
4.11.1. Characterization of Noncancer Effects ...................................................... 4-619
4.11.1.1. Neurotoxicity .............................................................................. 4-619
4.11.1.2. Kidney Toxicity .......................................................................... 4-623
4.11.1.3. Liver Toxicity ............................................................................. 4-624
4.11.1.4. Immunotoxicity ........................................................................... 4-626
4.11.1.5. Respiratory Tract Toxicity .......................................................... 4-627
4.11.1.6. Reproductive Toxicity ................................................................ 4-628
4.11.1.7. Developmental Toxicity.............................................................. 4-629
4.11.2. Characterization of Carcinogenicity .......................................................... 4-632
4.11.2.1. Summary Evaluation of Epidemiologic Evidence of
TCE and Cancer .......................................................................... 4-633
xi
4.11.2.2. Summary of Evidence for TCE Carcinogenicity in
Rodents ....................................................................................... 4-641
4.11.2.3. Summary of Additional Evidence on Biological
Plausibility .................................................................................. 4-644
4.11.3. Characterization of Factors Impacting Susceptibility ................................ 4-649
5. DOSE-RESPONSE ASSESSMENT ....................................................................................... 5-1
5.1. DOSE-RESPONSE ANALYSES FOR NONCANCER ENDPOINTS .................... 5-1
5.1.1. Modeling Approaches and UFs for Developing Candidate
Reference Values Based on Applied Dose .................................................... 5-3
5.1.2. Candidate Critical Effects by Effect Domain ................................................ 5-7
5.1.2.1. Candidate Critical Neurological Effects on the Basis
of Applied Dose .............................................................................. 5-7
5.1.2.2. Candidate Critical Kidney Effects on the Basis of
Applied Dose ................................................................................ 5-14
5.1.2.3. Candidate Critical Liver Effects on the Basis of
Applied Dose ................................................................................ 5-21
5.1.2.4. Candidate Critical Body Weight Effects on the Basis
of Applied Dose ............................................................................ 5-22
5.1.2.5. Candidate Critical Immunological Effects on the
Basis of Applied Dose .................................................................. 5-22
5.1.2.6. Candidate Critical Respiratory Tract Effects on the
Basis of Applied Dose .................................................................. 5-27
5.1.2.7. Candidate Critical Reproductive Effects on the Basis
of Applied Dose ............................................................................ 5-27
5.1.2.8. Candidate Critical Developmental Effects on the
Basis of Applied Dose .................................................................. 5-38
5.1.2.9. Summary of cRfCs, cRfDs, and Candidate Critical
Effects ........................................................................................... 5-46
5.1.3. Application of PBPK Model to Inter- and Intraspecies
Extrapolation for Candidate Critical Effects................................................ 5-49
5.1.3.1. Selection of Dose-metrics for Different Endpoints ...................... 5-49
5.1.3.2. Methods for Inter- and Intraspecies Extrapolation
Using Internal Doses ..................................................................... 5-57
5.1.3.3. Results and Discussion of p-RfCs and p-RfDs for
Candidate Critical Effects ............................................................. 5-75
5.1.4. Uncertainties in cRfCs and cRfDs ............................................................... 5-76
5.1.4.1. Qualitative Uncertainties .............................................................. 5-76
5.1.4.2. Quantitative Uncertainty Analysis of PBPK Model-
Based Dose-metrics for LOAEL- or NOAEL-Based
PODs ............................................................................................. 5-79
5.1.5. Summary of Noncancer Reference Values .................................................. 5-89
5.1.5.1. Preferred Candidate Reference Values (cRfCs, cRfD,
p-cRfCs, and p-cRfDs) for Candidate Critical Effects ................. 5-89
5.1.5.2. RfC ................................................................................................ 5-95
5.1.5.3. RfD ................................................................................................ 5-98
5.2. DOSE-RESPONSE ANALYSIS FOR CANCER ENDPOINTS .......................... 5-101
xii
5.2.1. Dose-Response Analyses: Rodent Bioassays ............................................ 5-101
5.2.1.1. Rodent Dose-Response Analyses: Studies and
Modeling Approaches ................................................................. 5-102
5.2.1.2. Rodent Dose-Response Analyses: Dosimetry ............................ 5-110
5.2.1.3. Rodent Dose-Response Analyses: Results.................................. 5-118
5.2.1.4. Uncertainties in Dose-Response Analyses of Rodent
Bioassays .................................................................................... 5-129
5.2.2. Dose-Response Analyses: Human Epidemiologic Data ............................ 5-139
5.2.2.1. Inhalation Unit Risk Estimate for RCC Derived from
Charbotel et al. (2006) Data ........................................................ 5-139
5.2.2.2. Adjustment of the Inhalation Unit Risk Estimate for
Multiple Sites .............................................................................. 5-147
5.2.2.3. Route-to-Route Extrapolation Using PBPK Model .................... 5-150
5.2.3. Summary of Unit Risk Estimates ............................................................... 5-153
5.2.3.1. Inhalation Unit Risk Estimate ..................................................... 5-153
5.2.3.2. Oral Slope Factor Estimate ......................................................... 5-154
5.2.3.3. Application of ADAFs ................................................................ 5-155
5.3. KEY RESEARCH NEEDS FOR TCE DOSE-RESPONSE
ANALYSES ........................................................................................................... 5-164
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD
AND DOSE RESPONSE ............................................................................................................ 6-1
6.1. HUMAN HAZARD POTENTIAL ............................................................................ 6-1
6.1.1. Exposure (see Chapter 2) ............................................................................... 6-1
6.1.2. Toxicokinetics and PBPK Modeling (see Chapter 3 and
Appendix A)................................................................................................... 6-2
6.1.3. Noncancer Toxicity ........................................................................................ 6-4
6.1.3.1. Neurological Effects (see Sections 4.3 and 4.11.1.1
and Appendix D) ............................................................................. 6-4
6.1.3.2. Kidney Effects (see Sections 4.4.1, 4.4.4, 4.4.6, and
4.11.1.2) .......................................................................................... 6-5
6.1.3.3. Liver Effects (see Sections 4.5.1, 4.5.3, 4.5.4, 4.5.6,
and 4.11.1.3, and Appendix E) ....................................................... 6-6
6.1.3.4. Immunological Effects (see Sections 4.6.1.1, 4.6.2,
and 4.11.1.4) ................................................................................... 6-7
6.1.3.5. Respiratory Tract Effects (see Sections 4.7.1.1,
4.7.2.1, 4.7.3, and 4.11.1.5) ............................................................ 6-8
6.1.3.6. Reproductive Effects (see Sections 4.8.1 and 4.11.1.6).................. 6-8
6.1.3.7. Developmental Effects (see Sections 4.8.3 and
4.11.1.7) .......................................................................................... 6-9
6.1.4. Carcinogenicity (see Sections 4.1, 4.2, 4.4.2, 4.4.5, 4.4.7, 4.5.2,
4.5.5, 4.5.6, 4.5.7, 4.6.1.2, 4.6.2.4, 4.7.1.2, 4.7.2.2, 4.7.4, 4.8.2,
4.9, and 4.11.2, and Appendices B and C) ................................................... 6-11
6.1.5. Susceptibility (see Sections 4.10 and 4.11.3) .............................................. 6-17
6.2. DOSE-RESPONSE ASSESSMENT ....................................................................... 6-18
6.2.1. Noncancer Effects (see Section 5.1) ............................................................ 6-18
xiii
6.2.1.1. Background and Methods ............................................................. 6-18
6.2.1.2. Uncertainties and Application of UFs (see Sections
5.1.1 and 5.1.4) ............................................................................. 6-19
6.2.1.3. Noncancer Reference Values (see Section 5.1.5) ......................... 6-28
6.2.2. Cancer (see Section 5.2) .............................................................................. 6-31
6.2.2.1. Background and Methods (rodent: see Section
5.2.1.1; human: see Section 5.2.2.1) ............................................. 6-31
6.2.2.2. Inhalation Unit Risk Estimate (rodent: see Section
5.2.1.3; human: see Sections 5.2.2.1 and 5.2.2.2)......................... 6-33
6.2.2.3. Oral Slope Factor Estimate (rodent: see Section
5.2.1.3; human: see Section 5.2.2.3) ............................................. 6-35
6.2.2.4. Uncertainties in Cancer Dose-Response Assessment ................... 6-36
6.2.2.5. Application of ADAFs (see Section 5.2.3.3) ................................ 6-41
6.3. OVERALL CHARACTERIZATION OF TCE HAZARD AND DOSE
RESPONSE.............................................................................................................. 6-42
7. References ................................................................................................................................ 7-1

APPENDIX A: PBPK MODELING OF TCE AND METABOLITESDETAILED
METHODS AND RESULTS ....................................................................... A-1

APPENDIX B: SYSTEMATIC REVIEW OF EPIDEMIOLOGIC STUDIES ON
CANCER AND TRICHLOROETHYLENE (TCE) EXPOSURE ................B-1

APPENDIX C: META-ANALYSIS OF CANCER RESULTS FROM
EPIDEMIOLOGICAL STUDIES .................................................................C-1

APPENDIX D: NEUROLOGICAL EFFECTS OF TRICHLOROETHYLENE ................... D-1

APPENDIX E: ANALYSIS OF LIVER AND COEXPOSURE ISSUES FOR
THE TCE TOXICOLOGICAL REVIEW ..................................................... E-1

APPENDIX F: TCE NONCANCER DOSE-RESPONSE ANALYSES ............................... F-1

APPENDIX G: TCE CANCER DOSE-RESPONSE ANALYSES WITH RODENT
CANCER BIOASSAY DATA ..................................................................... G-1

APPENDIX H: LIFETABLE ANALYSIS AND WEIGHTED LINEAR
REGRESSION BASED ON RESULTS FROM CHARBOTEL ET
AL. ................................................................................................................ H-1

APPENDIX I: EPA RESPONSE TO MAJOR PEER REVIEW AND PUBLIC
COMMENTS .................................................................................................. I-1
xiv
LIST OF TABLES


Table 2-1. TCE metabolites and related parent compounds
a
...................................................... 2-1
Table 2-2. Chemical properties of TCE ...................................................................................... 2-2
Table 2-3. Properties and uses of TCE related compounds ........................................................ 2-3
Table 2-4. TRI releases of TCE (pounds/year) ........................................................................... 2-4
Table 2-5. Concentrations of TCE in ambient air ....................................................................... 2-7
Table 2-6. TCE ambient air monitoring data (g/m
3
) ................................................................. 2-8
Table 2-7. Mean TCE air levels across monitors by land setting and use (1985
1998) ................................................................................................................................ 2-8
Table 2-8. Concentrations of TCE in water based on pre-1990 studies ................................... 2-12
Table 2-9. Levels in food .......................................................................................................... 2-15
Table 2-10. TCE levels in whole blood by population percentile ............................................ 2-16
Table 2-11. Modeled 1999 annual exposure concentrations (g/m
3
) for TCE ......................... 2-16
Table 2-12. Preliminary estimates of TCE intake from food ingestion .................................... 2-18
Table 2-13. Preliminary intake estimates of TCE and TCE-related chemicals ........................ 2-19
Table 2-14. Years of solvent use in industrial degreasing and cleaning operations ................. 2-21
Table 2-15. TCE standards ....................................................................................................... 2-22
Table 3-1. Blood:air partition coefficient values for humans ..................................................... 3-4
Table 3-2. Blood:air partition coefficient values for rats and mice ............................................ 3-5
Table 3-3. Air and blood concentrations during exposure to TCE in humans ........................... 3-6
Table 3-4. Retention of inhaled TCE vapor in humans .............................................................. 3-6
Table 3-5. Uptake of TCE in volunteers following 4 hour exposure to 70 ppm ........................ 3-7
Table 3-6. Concentrations of TCE in maternal and fetal blood at birth ................................... 3-11
Table 3-7. Distribution of TCE to rat tissues
a
following inhalation exposure .......................... 3-12
Table 3-8. Tissue:blood partition coefficient values for TCE .................................................. 3-13
Table 3-9. Age-dependence of tissue:air partition coefficients in rats ..................................... 3-14
Table 3-10. Predicted maximal concentrations of TCE in rat blood following a
6-hour inhalation exposure ............................................................................................ 3-14
Table 3-11. Tissue distribution of TCE metabolites following inhalation exposure ................ 3-15
Table 3-12. Binding of [
14
C] from [
14
C]-TCE in rat liver and kidney at 72 hours
after oral administration of 200 mg/kg [
14
C]-TCE ........................................................ 3-16
Table 3-13. In vitro TCE oxidative metabolism in hepatocytes and microsomal
fractions.......................................................................................................................... 3-22
Table 3-14. In vitro kinetics of TCOH and TCA formation from CH in rat,
mouse, and human liver homogenates ........................................................................... 3-25
xv
Table 3-15. In vitro kinetics of DCA metabolism in hepatic cytosol of mice, rats,
and humans .................................................................................................................... 3-27
Table 3-17. Reported TCA plasma binding parameters
a
.......................................................... 3-29
Table 3-18. Partition coefficients for TCE oxidative metabolites ............................................ 3-30
Table 3-19. Urinary excretion of TCA by various species exposed to TCE (based
on data reviewed in Fisher et al., 1991) ......................................................................... 3-32
Table 3-20. P450 isoform kinetics for metabolism of TCE to CH in human, rat,
and mouse recombinant P450s....................................................................................... 3-33
Table 3-21. P450 isoform activities in human liver microsomes exhibiting
different affinities for TCE ............................................................................................ 3-34
Table 3-22. Comparison of peak blood concentrations in humans exposed to 100
ppm (537 mg/m
3
) TCE for 4 hours ................................................................................ 3-38
Table 3-23. GSH conjugation of TCE (at 12 mM) in liver and kidney cellular
fractions in humans, male F344 rats, and male B6C3F
1
mice from Lash
laboratory ....................................................................................................................... 3-39
Table 3-24. Kinetics of TCE metabolism via GSH conjugation in male F344 rat
kidney and human liver and kidney cellular and subcellular fractions from
Lash laboratory .............................................................................................................. 3-40
Table 3-25. GSH conjugation of TCE (at 1.44 mM) in liver and kidney cellular
fractions in humans, male F344 rats, and male B6C3F
1
mice from Green
and Dekant laboratories ................................................................................................. 3-41
Table 3-26. GGT activity in liver and kidney subcellular fractions of mice, rats,
and humans .................................................................................................................... 3-47
Table 3-27. Multispecies comparison of whole-organ activity levels of GGT and
dipeptidase ..................................................................................................................... 3-47
Table 3-28. Comparison of hepatic in vitro oxidation and conjugation of TCE
a
..................... 3-51
Table 3-29. Estimates of DCVG in blood relative to inhaled TCE dose in humans
exposed to 50 and 100 ppm (269 and 537 mg/m
3
) ........................................................ 3-52
Table 3-30. Concentrations of TCE in expired breath from inhalation-exposed
humans (Astrand and Ovrum, 1976) .............................................................................. 3-53
Table 3-31. Conclusions from evaluation of Hack et al. (2006), and implications
for PBPK model development ....................................................................................... 3-61
Table 3-32. Discussion of changes to the Hack et al. (2006) PBPK model
implemented for this assessment ................................................................................... 3-65
Table 3-33. PBPK model-based dose-metrics .......................................................................... 3-67
Table 3-34. Rodent studies with pharmacokinetic data considered for analysis ...................... 3-69
Table 3-35. Human studies with pharmacokinetic data considered for analysis ...................... 3-73
xvi
Table 3-36. Parameters for which scaling from mouse to rat, or from mouse and
rat to human, was used to update the prior distributions ............................................... 3-78
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model
parameters ...................................................................................................................... 3-84
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model
parameters ...................................................................................................................... 3-89
Table 3-39. Prior and posterior uncertainty and variability in human PBPK model
parameters ...................................................................................................................... 3-94
Table 3-40. CI widths (ratio of 97.52.5% estimates) and fold-shift in median
estimate for the PBPK model population median parameters, sorted in
order of decreasing CI width
a
........................................................................................ 3-98
Table 3-41. Estimates of the residual-error ............................................................................. 3-104
Table 3-42. Summary comparison of updated PBPK model predictions and in
vivo data in mice .......................................................................................................... 3-108
Table 3-43. Summary comparison of updated PBPK model predictions and in
vivo data used for calibration in rats ........................................................................ 3-114
Table 3-44. Summary comparison of updated PBPK model predictions and in
vivo data used for out-of-sample evaluation in rats ................................................. 3-116
Table 3-45. Summary comparison of updated PBPK model predictions and in
vivo data used for calibration in humans .................................................................. 3-121
Table 3-46. Summary comparison of updated PBPK model predictions and in
vivo data used for out-of-sample evaluation in humans........................................... 3-123
Table 3-47. Summary of scaling parameters ordered by fraction of calibration
data of moderate or high sensitivity ............................................................................. 3-131
Table 3-48. Posterior predictions for representative internal doses: mouse
a
.......................... 3-144
Table 3-49. Posterior predictions for representative internal doses: rat
a
................................ 3-145
Table 3-50. Posterior predictions for representative internal doses: human
a
......................... 3-146
Table 3-51. Degree of variance in dose-metric predictions due to incomplete
convergence (columns 24), combined uncertainty and population
variability (columns 57), uncertainty in particular human population
percentiles (columns 810), model fits to in vivo data (column 11); the
GSD is a fold-change from the central tendency ..................................................... 3-164
Table 4-1. Description of epidemiologic cohort and proportionate mortality ratio
(PMR) studies assessing cancer and TCE exposure ........................................................ 4-2
Table 4-2. Case-control epidemiologic studies examining cancer and TCE
exposure ........................................................................................................................... 4-8
Table 4-3. Geographic-based studies assessing cancer and TCE exposure .............................. 4-19
xvii
Table 4-4. Standards of epidemiologic study design and analysis use for
identifying cancer hazard and TCE exposure ................................................................ 4-21
Table 4-5. Summary of criteria for meta-analysis study selection ........................................... 4-24
Table 4-6. TCE genotoxicity: bacterial assays ......................................................................... 4-33
Table 4-7. TCE genotoxicity: fungal and yeast systems........................................................... 4-36
Table 4-8. TCE genotoxicity: mammalian systemsgene mutations and
chromosome aberrations ................................................................................................ 4-39
Table 4-9. TCE genotoxicity: mammalian systemsmicronucleus, sister
chromatic exchanges ...................................................................................................... 4-43
Table 4-10. TCE genotoxicity: mammalian systemsUDS, DNA strand
breaks/protein crosslinks, and cell transformation......................................................... 4-46
Table 4-11. Genotoxicity of TCAbacterial systems ............................................................. 4-50
Table 4-12. TCA Genotoxicitymammalian systems (both in vitro and in vivo) .................. 4-52
Table 4-13. Genotoxicity of DCA (bacterial systems) ............................................................. 4-57
Table 4-14. Genotoxicity of DCAmammalian systems ........................................................ 4-58
Table 4-15. CH genotoxicity: bacterial, yeast, and fungal systems .......................................... 4-63
Table 4-16. CH genotoxicity: mammalian systemsall genetic endpoints, in
vitro ................................................................................................................................ 4-65
Table 4-17. CH genotoxicity: mammalian systemsall genetic damage, in vivo ................... 4-67
Table 4-18. TCE GSH conjugation metabolites genotoxicity .................................................. 4-74
Table 4-19. Genotoxicity of TCOH .......................................................................................... 4-79
Table 4-20. Summary of human trigeminal nerve and nerve conduction velocity
studies ............................................................................................................................ 4-84
Table 4-21. Summary of animal trigeminal nerve studies ........................................................ 4-88
Table 4-22. Summary of human auditory function studies....................................................... 4-92
Table 4-23. Summary of animal auditory function studies....................................................... 4-94
Table 4-24. Summary of vestibular system studies ................................................................ 4-100
Table 4-25. Summary of human visual function studies ........................................................ 4-102
Table 4-26. Summary of animal visual system studies........................................................... 4-104
Table 4-27. Summary of human cognition effect studies ....................................................... 4-107
Table 4-28. Summary of animal cognition effect studies ....................................................... 4-109
Table 4-29. Summary of human CRT studies ........................................................................ 4-112
Table 4-30. Summary of animal psychomotor function and RT studies ................................ 4-114
Table 4-31. Summary of animal locomotor activity studies ................................................... 4-116
Table 4-32. Summary of animal mood effect and sleep disorder studies ............................... 4-120
Table 4-33. Summary of human developmental neurotoxicity associated with
TCE exposures ............................................................................................................. 4-121
xviii
Table 4-34. Summary of mammalian in vivo developmental neurotoxicity
studiesoral exposures ............................................................................................... 4-122
Table 4-35. Summary of animal dopamine neuronal studies ................................................. 4-127
Table 4-36. Summary of neurophysiological, neurochemical, and
neuropathological effects with TCE exposure ............................................................. 4-128
Table 4-37. Summary of in vitro ion channel effects with TCE exposure ............................. 4-130
Table 4-38. Summary of human kidney toxicity studies ........................................................ 4-139
Table 4-39. Summary of human studies on TCE exposure and kidney cancer ...................... 4-144
Table 4-40. Summary of case-control studies on kidney cancer and occupation or
job title ......................................................................................................................... 4-152
Table 4-41. Summary of lung and kidney cancer risks in active smokers ............................. 4-162
Table 4-42. Summary of human studies on somatic mutations of the VHL gene
a
................. 4-174
Table 4-43. Inhalation studies of kidney noncancer toxicity in laboratory animals ............... 4-178
Table 4-44. Oral and i.p. studies of kidney noncancer toxicity in laboratory
animals ......................................................................................................................... 4-179
Table 4-45. Summary of renal toxicity and tumor findings in gavage studies of
TCE by NTP (1990)
a
.................................................................................................... 4-181
Table 4-46. Summary of renal toxicity and tumor findings in gavage studies of
TCE by NCI (1976)
a
.................................................................................................... 4-182
Table 4-47. Summary of renal toxicity findings in gavage studies of TCE by
Maltoni et al. (1988, 1986) .......................................................................................... 4-183
Table 4-48. Summary of renal toxicity and tumor incidence in gavage studies of
TCE by NTP (1988)
a
.................................................................................................... 4-183
Table 4-49. Summary of renal toxicity and tumor findings in inhalation studies of
TCE by Maltoni et al. (1988, 1986)
a
............................................................................ 4-184
Table 4-50. Summary of renal tumor findings in inhalation studies of TCE by
Henschler et al. (1980)
a
and Fukuda et al. (1983)
b
...................................................... 4-186
Table 4-51. Summary of renal tumor findings in gavage studies of TCE by
Henschler et al. (1984)
a
and Van Duuren et al. (1979)
b
.............................................. 4-188
Table 4-52. Laboratory animal studies of kidney noncancer toxicity of TCE
metabolites ................................................................................................................... 4-190
Table 4-53. Summary of histological changes in renal proximal tubular cells
induced by chronic exposure to TCE, DCVC, and TCOH .......................................... 4-192
Table 4-54. Summary of major mode-of-action conclusions for TCE kidney
carcinogenesis .............................................................................................................. 4-201
Table 4-55. Summary of human liver toxicity studies ............................................................ 4-220
Table 4-56. Selected results from epidemiologic studies of TCE exposure and
cirrhosis ........................................................................................................................ 4-222
xix
Table 4-57. Selected results from epidemiologic studies of TCE exposure and
liver cancer ................................................................................................................... 4-226
Table 4-58. Oral studies of TCE-induced liver effects in mice and rats ................................. 4-242
Table 4-59. Inhalation and i.p. studies of TCE-induced liver effects in mice and
rats ................................................................................................................................ 4-245
Table 4-60. Summary of liver tumor findings in gavage studies of TCE by NTP
(1990)
a
.......................................................................................................................... 4-267
Table 4-61. Summary of liver tumor findings in gavage studies of TCE by NCI
(1976) ........................................................................................................................... 4-267
Table 4-62. Summary of liver tumor incidence in gavage studies of TCE by NTP
(1988) ........................................................................................................................... 4-268
Table 4-63. Summary of liver tumor findings in inhalation studies of TCE by
Maltoni et al. (1988, 1986)
a
......................................................................................... 4-269
Table 4-64. Summary of liver tumor findings in inhalation studies of TCE by
Henschler et al. (1980)
a
and Fukuda et al. (1983) ....................................................... 4-270
Table 4-65. Summary of liver tumor findings in gavage studies of TCE by
Henschler et al. (1984)
a
................................................................................................ 4-271
Table 4-66. Potency indicators for mouse hepatocarcinogenicity and in vitro
transactivation of mouse PPAR for four PPAR agonists ......................................... 4-331
Table 4-67. Potency indicators for rat hepatocarcinogenicity and common
short-term markers of PPAR activation for four PPAR agonists ............................ 4-332
Table 4-68. Summary of mode-of-action conclusions for TCE-induced liver
carcinogenesis .............................................................................................................. 4-343
Table 4-69. Studies of immune parameters (IgE antibodies and cytokines) and
TCE in humans ............................................................................................................ 4-356
Table 4-70. Case-control studies of autoimmune diseases with measures of TCE
exposure ....................................................................................................................... 4-365
Table 4-71. Incidence cohort studies of TCE exposure and lymphopoietic and
hematopoietic cancer risk ............................................................................................ 4-369
Table 4-72. Mortality cohort and PMR studies of TCE exposure and
lymphopoietic and hematopoietic cancer risk .............................................................. 4-374
Table 4-73. Case-control studies of TCE exposure and lymphopoietic cancer,
leukemia or multiple myeloma .................................................................................... 4-383
Table 4-74. Geographic-based studies of TCE and NHL or leukemia in adults .................... 4-389
Table 4-75. Selected results from epidemiologic studies of TCE exposure and
childhood leukemia ...................................................................................................... 4-392
Table 4-76. Summary of TCE immunosuppression studies ................................................... 4-402
Table 4-77. Summary of TCE hypersensitivity studies
a
......................................................... 4-410
xx
Table 4-78. Summary of autoimmune-related studies of TCE and metabolites in
mice and rats (by sex, strain, and route of exposure)
a
................................................. 4-414
Table 4-79. Malignant lymphomas incidence in mice exposed to TCE in gavage
and inhalation exposure studies ................................................................................... 4-425
Table 4-80. Leukemia incidence in rats exposed to TCE in gavage and inhalation
exposure studies ........................................................................................................... 4-426
Table 4-81. Selected results from epidemiologic studies of TCE exposure and
lung cancer ................................................................................................................... 4-432
Table 4-82. Selected results from epidemiologic studies of TCE exposure and
laryngeal cancer ........................................................................................................... 4-439
Table 4-83. Animal toxicity studies of TCE ........................................................................... 4-444
Table 4-84. Animal carcinogenicity studies of TCE .............................................................. 4-452
Table 4-85. Human reproductive effects ................................................................................ 4-468
Table 4-86. Summary of mammalian in vivo reproductive toxicity studies
inhalation exposures..................................................................................................... 4-474
Table 4-87. Summary of mammalian in vivo reproductive toxicity studiesoral
exposures...................................................................................................................... 4-476
Table 4-88. Summary of adverse female reproductive outcomes associated with
TCE exposures ............................................................................................................. 4-488
Table 4-89. Summary of adverse male reproductive outcomes associated with
TCE exposures ............................................................................................................. 4-489
Table 4-90. Summary of human studies on TCE exposure and prostate cancer .................... 4-494
Table 4-91. Summary of human studies on TCE exposure and breast cancer ....................... 4-497
Table 4-92. Summary of human studies on TCE exposure and cervical cancer .................... 4-501
Table 4-93. Histopathology findings in reproductive organs ................................................. 4-509
Table 4-94. Testicular tumors in male rats exposed to TCE, adjusted for reduced
survival
a
....................................................................................................................... 4-510
Table 4-95. Developmental studies in humans ....................................................................... 4-512
Table 4-96. Summary of mammalian in vivo developmental toxicity studies
inhalation exposures..................................................................................................... 4-534
Table 4-97. Ocular defects observed (Narotsky et al., 1995) ................................................. 4-535
Table 4-98. Summary of mammalian in vivo developmental toxicity studies
oral exposures .............................................................................................................. 4-536
Table 4-99. Types of congenital cardiac defects observed in TCE-exposed fetuses .............. 4-544
Table 4-100. Types of heart malformations per 100 fetuses .................................................. 4-545
Table 4-101. Congenital cardiac malformations ..................................................................... 4-547
Table 4-102. Summary of adverse fetal and early neonatal outcomes associated
with TCE exposures ..................................................................................................... 4-557
xxi
Table 4-103. Summary of studies that identified cardiac malformations
associated with TCE exposures ................................................................................... 4-558
Table 4-104. Events in cardiac valve formation in mammals and birds
a
............................... 4-562
Table 4-105. Summary of other structural developmental outcomes associated
with TCE exposures ..................................................................................................... 4-566
Table 4-106. Summary of developmental neurotoxicity associated with TCE
exposures...................................................................................................................... 4-568
Table 4-107. Summary of developmental immunotoxicity associated with TCE
exposures...................................................................................................................... 4-570
Table 4-108. Summary of childhood cancers associated with TCE exposures ...................... 4-571
Table 4-109. Selected observations from case-control studies of TCE exposure
and esophageal cancer.................................................................................................. 4-574
Table 4-110. Summary of human studies on TCE exposure and esophageal
cancer ........................................................................................................................... 4-577
Table 4-111. Summary of human studies on TCE exposure and bladder cancer ................... 4-585
Table 4-112. Summary of human studies on TCE exposure and brain cancer ....................... 4-591
Table 4-113. Estimated lifestage-specific daily doses for TCE in water
a
.............................. 4-597
Table 5-1. Summary of studies of neurological effects suitable for dose-response
assessment ........................................................................................................................ 5-9
Table 5-2. Neurological effects in studies suitable for dose-response assessment,
and corresponding cRfCs and cRfDs ............................................................................. 5-12
Table 5-3. Summary of studies of kidney, liver, and body weight effects suitable
for dose-response assessment ........................................................................................ 5-16
Table 5-4. Kidney, liver, and body weight effects in studies suitable for
dose-response assessment, and corresponding cRfCs and cRfDs ................................. 5-18
Table 5-5. Summary of studies of immunological effects suitable for
dose-response assessment .............................................................................................. 5-23
Table 5-6. Immunological effects in studies suitable for dose-response
assessment, and corresponding cRfCs and cRfDs ......................................................... 5-25
Table 5-7. Summary of studies of reproductive effects suitable for dose-response
assessment ...................................................................................................................... 5-28
Table 5-8. Reproductive effects in studies suitable for dose-response assessment,
and corresponding cRfCs and cRfDs ............................................................................. 5-33
Table 5-9. Summary of studies of developmental effects suitable for
dose-response assessment .............................................................................................. 5-39
Table 5-10. Developmental effects in studies suitable for dose-response
assessment, and corresponding cRfCs and cRfDs ......................................................... 5-43
xxii
Table 5-11. Ranges of cRfCs based on applied dose for various noncancer effects
associated with inhalation TCE exposure
a
..................................................................... 5-47
Table 5-12. Ranges of cRfDs based on applied dose for various noncancer effects
associated with oral TCE exposure
a
............................................................................... 5-48
Table 5-13. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical
neurological effects ........................................................................................................ 5-61
Table 5-14. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical
kidney effects ................................................................................................................. 5-63
Table 5-15. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical liver
effects ............................................................................................................................. 5-66
Table 5-16. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical
immunological effects .................................................................................................... 5-67
Table 5-17. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical
reproductive effects ........................................................................................................ 5-69
Table 5-18. cRfCs and cRfDs (based on applied dose) and p-cRfCs and p-cRfDs
(based on PBPK modeled internal dose-metrics) for candidate critical
developmental effects .................................................................................................... 5-72
Table 5-19. Comparison of sensitive individual HECs or HEDs for
neurological effects based on PBPK modeled internal dose-metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL ..................... 5-82
Table 5-20. Comparison of sensitive individual HECs or HEDs for kidney and
liver effects based on PBPK modeled internal dose-metrics at different
levels of confidence and sensitivity, at the NOAEL or LOAEL ................................... 5-83
Table 5-21. Comparison of sensitive individual HECs or HEDs for
immunological effects based on PBPK modeled internal dose-metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL ..................... 5-85
Table 5-22. Comparison of sensitive individual HECs or HEDs for
reproductive effects based on PBPK modeled internal dose-metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL ..................... 5-86
Table 5-23. Comparison of sensitive individual HECs or HEDs for
developmental effects based on PBPK modeled internal dose-metrics at
different levels of confidence and sensitivity, at the NOAEL or LOAEL ..................... 5-88
Table 5-24. Lowest p-cRfCs or cRfCs for different effect domains ......................................... 5-90
xxiii
Table 5-25. Lowest p-cRfDs or cRfDs for different effect domains ........................................ 5-92
Table 5-26. Lowest p-cRfCs for candidate critical effects for different types of
effect based on primary dose-metric .............................................................................. 5-94
Table 5-27. Lowest p-cRfDs for candidate critical effects for different types of
effect based on primary dose-metric .............................................................................. 5-94
Table 5-28. Summary of critical studies, effects, PODs, and UFs used to derive
the RfC ........................................................................................................................... 5-96
Table 5-29. Summary of supporting studies, effects, PODs, and UFs for the RfC .................. 5-96
Table 5-30. Summary of critical studies, effects, PODs, and UFs used to derive
the RfD ........................................................................................................................... 5-99
Table 5-31. Summary of supporting studies, effects, PODs, and UFs for the RfD ................ 5-100
Table 5-32. Inhalation bioassays............................................................................................. 5-103
Table 5-33. Oral bioassays ...................................................................................................... 5-104
Table 5-34. Specific dose-response analyses performed and dose-metrics used .................... 5-107
Table 5-35. Mean PBPK model predictions for weekly internal dose in humans
exposed continuously to low levels of TCE via inhalation (ppm) or orally
(mg/kg/day) .................................................................................................................. 5-118
Table 5-36. Summary of PODs and unit risk estimates for each
sex/species/bioassay/tumor type (inhalation) .............................................................. 5-119
Table 5-37. Summary of PODs and slope factor estimates for each
sex/species/bioassay/tumor type (oral) ........................................................................ 5-121
Table 5-38. Comparison of survival-adjusted results for three oral male rat data
sets
a
.............................................................................................................................. 5-124
Table 5-39. Inhalation: most sensitive bioassay for each sex/species combination
a
.............. 5-128
Table 5-40. Oral: most sensitive bioassay for each sex/species combination
a
....................... 5-128
Table 5-41. Summary of PBPK model-based uncertainty analysis of unit risk
estimates for each sex/species/bioassay/tumor type (inhalation)................................. 5-136
Table 5-42. Summary of PBPK model-based uncertainty analysis of slope factor
estimates for each sex/species/bioassay/tumor type (oral) .......................................... 5-137
Table 5-43. Results from Charbotel et al. (2006) on relationship between TCE
exposure and RCC ....................................................................................................... 5-140
Table 5-44. Extra risk estimates for RCC incidence from various levels of
lifetime exposure to TCE, using linear cumulative exposure model ........................... 5-142
Table 5-45. EC
01
, LEC
01
, and unit risk estimates for RCC incidence, using linear
cumulative exposure model ......................................................................................... 5-143
Table 5-46. Relative contributions to extra risk for cancer incidence from TCE
exposure for multiple cancer types .............................................................................. 5-149
xxiv
Table 5-47. Route-to-route extrapolation of site-specific inhalation unit risks to
oral slope factors .......................................................................................................... 5-152
Table 5-48. Sample calculation for total lifetime cancer risk based on the kidney
unit risk estimate, potential risk for NHL and liver cancer, and potential
increased early-life susceptibility, assuming a constant lifetime exposure
to 1 g/m
3
of TCE in air .............................................................................................. 5-159
Table 5-49. Sample calculation for total lifetime cancer risk based on the kidney
cancer slope factor estimate, potential risk for NHL and liver cancer, and
potential increased early-life susceptibility, assuming a constant lifetime
exposure to 1 g/L of TCE in drinking water .............................................................. 5-162


xxv

LIST OF FIGURES


Figure 2-1. Molecular structure of TCE. .................................................................................... 2-2
Figure 2-2. Source contribution to TCE emissions. .................................................................... 2-5
Figure 2-3. Annual emissions of TCE. ....................................................................................... 2-5
Figure 2-4. Modeled ambient air concentrations of TCE. ........................................................ 2-10
Figure 3-1. Gas uptake data from closed-chamber exposure of rats to TCE. ............................. 3-8
Figure 3-2. Disposition of [
14
C]-TCE administered by gavage in mice. .................................. 3-18
Figure 3-3. Disposition of [
14
C]-TCE administered by gavage in rats. .................................... 3-19
Figure 3-4. Scheme for the oxidative metabolism of TCE. ...................................................... 3-21
Figure 3-5. Scheme for GSH-dependent metabolism of TCE. ................................................. 3-36
Figure 3-6. Interorgan TCE transport and metabolism via the GSH pathway. ......................... 3-45
Figure 3-7. Overall structure of PBPK model for TCE and metabolites used in this
assessment. ..................................................................................................................... 3-64
Figure 3-8. Schematic of how posterior predictions were generated for comparison with
experimental data. ........................................................................................................ 3-104
Figure 3-9. Comparison of mouse data and PBPK model predictions from a random
posterior sample. .......................................................................................................... 3-106
Figure 3-10. Comparison of rat data and PBPK model predictions from a random
posterior sample. .......................................................................................................... 3-111
Figure 3-11. Comparison of urinary excretion data for NAcDCVC and predictions from
the Hack et al. (2006) and the updated PBPK models. ................................................ 3-119
Figure 3-12. Comparison of human data and PBPK model predictions from a random
posterior sample. .......................................................................................................... 3-120
Figure 3-13. Comparison of DCVG concentrations in human blood and predictions from
the updated model. ....................................................................................................... 3-126
Figure 3-14. Sensitivity analysis results: Number of mouse calibration data points with
SC in various categories for each scaling parameter. .................................................. 3-128
Figure 3-15. Sensitivity analysis results: Number of rat calibration data points with SC in
various categories for each scaling parameter. ............................................................ 3-129
Figure 3-16. Sensitivity analysis results: Number of human calibration data points with
SC in various categories for each scaling parameter. .................................................. 3-130
Figure 3-17. PBPK model predictions for the fraction of intake that is metabolized under
continuous inhalation (A) and oral (B) exposure conditions in mice (white), rats
(diagonal hashing), and humans (horizontal hashing). ................................................ 3-134
Figure 3-18. PBPK model predictions for the fraction of intake that is metabolized by
oxidation (in the liver and lung) under continuous inhalation (A) and oral (B)
xxvi
exposure conditions in mice (white), rats (diagonal hashing), and humans
(horizontal hashing). .................................................................................................... 3-135
Figure 3-19. PBPK model predictions for the fraction of intake that is metabolized by
GSH conjugation (in the liver and kidney) under continuous inhalation (A) and
oral (B) exposure conditions in mice (dotted line), rats (dashed line), and humans
(solid line). ................................................................................................................... 3-136
Figure 3-20. PBPK model predictions for the fraction of intake that is bioactivated
DCVC in the kidney under continuous inhalation (A) and oral (B) exposure
conditions in rats (dashed line) and humans (solid line).............................................. 3-137
Figure 3-21. PBPK model predictions for fraction of intake that is oxidized in the
respiratory tract under continuous inhalation (A) and oral (B) exposure conditions
in mice (dotted line), rats (dashed line), and humans (solid line). ............................... 3-138
Figure 3-22. PBPK model predictions for the fraction of intake that is untracked
oxidation of TCE in the liver under continuous inhalation (A) and oral (B)
exposure conditions in mice (dotted line), rats (dashed line), and humans (solid
line). ............................................................................................................................. 3-139
Figure 3-23. PBPK model predictions for the weekly AUC of TCE in venous blood
(mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats (dashed
line), and humans (solid line). ...................................................................................... 3-140
Figure 3-24. PBPK model predictions for the weekly AUC of TCOH in blood
(mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats (dashed
line), and humans (solid line). ...................................................................................... 3-141
Figure 3-25. PBPK model predictions for the weekly AUC of TCA in the liver
(mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats (dashed
line), and humans (solid line). ...................................................................................... 3-142
Figure 3-26. Sensitivity analysis results: SC for mouse scaling parameters with respect to
dose-metrics following 100 ppm (light bars) and 600 ppm (dark bars), 7
hours/day, 5 days/week inhalation exposures. ............................................................. 3-149
Figure 3-27. Sensitivity analysis results: SC for mouse scaling parameters with respect to
dose-metrics following 300 mg/kg-day (light bars) and 1,000 mg/kg-day (dark
bars), 5 days/week gavage exposures. ......................................................................... 3-150
Figure 3-28. Sensitivity analysis results: SC for rat scaling parameters with respect to
dose-metrics following 100 ppm (light bars) and 600 ppm (dark bars), 7
hours/day, 5 days/week inhalation exposures. ............................................................. 3-151
xxvii
Figure 3-29. Sensitivity analysis results: SC for rat scaling parameters with respect to
dose-metrics following 300 mg/kg-day (light bars) and 1,000 mg/kg-day (dark
bars), 5 days/week gavage exposures. ......................................................................... 3-152
Figure 3-30. Sensitivity analysis results: SC for female (light bars) and male (dark bars)
human scaling parameters with respect to dose-metrics following 0.001 ppm
continuous inhalation exposures. ................................................................................. 3-153
Figure 3-31. Sensitivity analysis results: SC for female (light bars) and male (dark bars)
human scaling parameters with respect to dose-metrics following 0.001
mg/kg-day continuous oral exposures. ........................................................................ 3-154
Figure 4-1. Meta-analysis of kidney cancer and overall TCE exposure. ................................ 4-168
Figure 4-2. Meta-analysis of kidney cancer and TCE exposurehighest exposure
groups. .......................................................................................................................... 4-170
Figure 4-3. Meta-analysis of liver and biliary tract cancer and overall TCE exposure. ......... 4-236
Figure 4-4. Meta-analysis of liver cancer and TCE exposurehighest exposure groups. ..... 4-237
Figure 4-5. Comparison of average fold-changes in relative liver weight to control and
exposure concentrations of 2 g/L or less in drinking water for TCA and DCA in
male B6C3F
1
mice for 1430 days. ............................................................................. 4-279
Figure 4-6. Comparisons of fold-changes in average relative liver weight and gavage
dose of (top panel) male B6C3F
1
mice for 1028 days of exposure and (bottom
panel) in male B6C3F
1
and Swiss mice. ...................................................................... 4-281
Figure 4-7. Comparison of fold-changes in relative liver weight for data sets in male
B6C3F
1
, Swiss, and NRMI mice between TCE studies [duration 2842 days]) and
studies of direct oral TCA administration to B6C3F
1
mice [duration 1428 days]). .. 4-283
Figure 4-8. Comparison of hepatomegaly as a function of AUC of TCA in liver, using
values for the TCA drinking water fractional absorption (Fabs). ................................ 4-285
Figure 4-9. Fold-changes in relative liver weight for data sets in male B6C3F
1
, Swiss,
and NRMI mice reported by TCE studies of duration 2842 days using internal
dose-metrics predicted by the PBPK model described in Section 3.5:
(A) dose-metric is the median estimate of the daily AUC of TCE in blood,
(B) dose-metric is the median estimate of the total daily rate of TCE oxidation. ....... 4-286
Figure 4-11. Dose-response relationship, expressed as (A) incidence and
(B) fold-increase over controls, for TCE hepatocarcinogenicity in Maltoni et al.
(1988, 1986). ................................................................................................................ 4-301
Figure 4-12. Dose-response data for HCCs (A) incidence and (B) multiplicity, induced
by DCA from DeAngelo et al. (1999). ........................................................................ 4-302
Figure 4-14. Reported incidence of HCCs induced by DCA and TCA in 104-week
studies. ......................................................................................................................... 4-306
xxviii
Figure 4-15. Effects of dietary control on the dose-response curves for changes in liver
tumor incidences induced by CH in diet. ..................................................................... 4-309
Figure 4-16. Meta-analysis of NHL and overall TCE exposure. ............................................ 4-397
Figure 4-17. Meta-analysis of NHL and TCE exposurehighest exposure groups. ............. 4-398
Figure 5-1. Flow-chart of the process used to derive the RfD and RfC for noncancer
effects. .............................................................................................................................. 5-3
Figure 5-2. Flow-chart for dose-response analyses of rodent noncancer effects using
PBPK model-based dose-metrics. .................................................................................. 5-58
Figure 5-3. Schematic of combined interspecies, intraspecies, and route-to-route
extrapolation from a rodent study LOAEL or NOAEL. ................................................ 5-59
Figure 5-4. Flow-chart for uncertainty analysis of HECs and HEDs derived using PBPK
model-based dose-metrics. ............................................................................................. 5-80
Figure 5-5. Flow-chart for dose-response analyses of rodent bioassays using PBPK
model-based dose-metrics. ........................................................................................... 5-117
Figure 5-6. Flow-chart for uncertainty analysis of dose-response analyses of rodent
bioassays using PBPK model-based dose-metrics. ...................................................... 5-134
Figure 5-7. Flow-chart for route-to-route extrapolation of human site-specific cancer
inhalation unit risks to oral slope factors. .................................................................... 5-151

xxix
LIST OF ABBREVIATIONS AND ACRONYMS


[
14
C]TCE [
14
C]-radiolabeled TCE
1,2-DCVC S-(1,2-dichlorovinyl)-L-cysteine
17--HSD 17--hydroxy steroid dehydrogenase
8-OHdG 8-hydroxy-2 deoxyguanosine
ACO acyl CoA oxidase
ADAF age-dependent adjustment factor
ADME absorption, distribution, metabolism, and excretion
AIC Akaikes Information Criteria
ALL acute lymphoblastic leukemia
ALP alkaline phosphatase
ALT alanine aminotransferase
ANA antinuclear antibodies
ANCA antineutrophil-cytoplasmic antibody
ANOVA analysis of variance
AOAA a beta-lyase inhibitor
ASD autism spectrum disorder
ASPEN Assessment System for Population Exposure Nationwide
AST aspartate aminotransferase
ATF-2 activating transcription factor 2
ATSDR Agency for Toxic Substances and Disease Registry
AUC area-under-the-curve
AV atrioventricular
AVC atrioventricular canal
AZ DHS Arizona Department of Health Services
BAER brainstem auditory-evoked response
BAL bronchoalveolar lavage
BMD benchmark dose
BMDL benchmark dose lower bound
BMDS BenchMark Dose Software
BMI body mass index
BMR benchmark response
BUN blood urea nitrogen
CA DHS California Department of Health Services
CH chloral hydrate
CI confidence interval
CLL chronic lymphocytic leukemia
CNS central nervous system
CO
2
carbon dioxide
CoA coenzyme A
cRfC candidate RfC
cRfD candidate RfD
CRT choice reaction time
CYP cytochrome P450
xxx
LIST OF ABBREVIATIONS AND ACRONYMS (continued)


DAL dichloroacetyl lysine
DASO
2
diallyl sulfone

DBP dibutyl phthalate
DCA dichloroacetic acid
DCAA dichloroacetic anhydride
DCAC dichloroacetyl chloride
DCE dichloroethylene
DCVC S-dichlorovinyl-L-cysteine (collectively, the 1,2- and 2,2- isomers)
DCVG S-dichlorovinyl-L-glutathione (collectively, the 1,2- and 2,2- isomers)
DEHP di(2-ethylhexyl) phthalate
DHEAS dehydroepiandrosterone sulphate
DNA deoxyribonucleic acid
DNP dinitrophenol
DPM disintegrations per minute
dsDNA double-stranded DNA
EC
x
concentration of the chemical at which x% of the maximal effect is
produced
EEG electroencephalograph
EPA U.S. Environmental Protection Agency
ERG electroretinogram
ESRD end stage renal disease
FAA fumarylacetoacetate
FDVE fluoromethyl-2,2-difluoro-1-(trifluoromethyl)vinyl ether
FMO flavin mono-oxygenase
FOB functional observational battery
FSH follicle-stimulating hormone
G6PDH glucose 6-p dehydrogenase
GABA gamma-amino butyric acid
G-CSF granulocyte colony stimulating factor
GD gestation day
GGT -glutamyl transpeptidase or -transpeptidase
GI gastrointestinal
GIS geographic information system
GSD geometric standard deviation
GSH glutathione
GSSG oxidized GSH
GST glutathione-S-transferase
GT glutamyl transferase
H&E hematoxylin and eosin
H
2
O water
HCC hepatocellular carcinoma
hCG human chorionic gonadotropin
HCl hydrochloric acid
HDL-C high density lipoprotein-cholesterol
HEC human equivalent concentration
xxxi
LIST OF ABBREVIATIONS AND ACRONYMS (continued)


HED human equivalent dose
HgCl
2
mercuric chloride

HH Hamberger and Hamilton
HPLC high-performance liquid chromatography
HPT hypothalamic-pituitary-testis
i.a. intra-arterial
i.p. intraperitoneal
i.v. intravenous
IARC International Agency for Research on Cancer
ICC intrahepatic cholangiocarcinoma
ICD International Classification of Disease
ICRP The International Commission on Radiological Protection
idPOD internal dose points of departure
IDR incidence density ratio
IFN interferon
IgE immunoglobulin E
IGF-II insulin-like growth factor-II (gene)
IL interleukin
IPCS International Programme on Chemical Safety
IUGR intrauterine growth restriction
JEM job-exposure matrix
JTEM job-task-exposure matrix
LC lethal concentration
LCL lower confidence limit
LDH lactate dehydrogenase
LEC
x
lowest effective concentration corresponding to an extra risk of x%
LH luteinizing hormone
lnPBC blood-air partition coefficient
lnQCC cardiac output
lnVMAXC VMAX for oxidation
lnVPRC ventilation-perfusion ratio
LOAEL lowest-observed adverse effect level
LOH loss of heterozygosity
LORR loss of righting reflex
MA maleylacetone
MA DPH Massachusetts Department of Public Health
MAA maleylacetoacetate
MCA monochloroacetic acid
MCMC Markov chain Monte Carlo
MCP methylclofenapate
MDA malondialdehyde
MLE maximum likelihood estimate
MNU methyl nitrosourea
xxxii
LIST OF ABBREVIATIONS AND ACRONYMS (continued)


MS mass spectrometry
MSW multistage Weibull
NAcDCVC N-acetyl-S-(1,2-dichlorovinyl)-L-cysteine
NADH nicotinamide adenine dinucleotide
NADPH nicotinamide adenine dinucleotide phosphate-oxidase
NAG N-acetyl--D-glucosaminidase
NAS National Academy of Sciences
NAT N-acetyl transferase
NCI National Cancer Institute
NF-B nuclear factor kappa-light-chain enhancer of activated B cells
NHL non-Hodgkin lymphoma
NK natural killer
NOAEL no-observed-adverse-effect level
NOEC no-observed-effect concentration
NOEL no-observed-effect level
NPMC nonpurified rat peritoneal mast cells
NRC National Research Council
NSATA National-Scale Air Toxics Assessment
NTP National Toxicology Program
NYS DOH New York State Department of Health
ODE ordinary differential equation
OECD Organization for Economic Co-operation and Development
OFT outflow tract
OP oscillatory potential
OR odds ratio
OR
adj
adjusted odds ratio
PAS periodic acid-Schiff
PBPK physiologically based pharmacokinetics
PCEs polychromatic erythrocytes
PCNA proliferating cell nuclear antigen
PCO palmitoyl-CoA oxidase
PCR polymerase chain reaction
p-cRfC PBPK model-based candidate RfCs
p-cRfD PBPK model-based candidate RfDs
PEG 400 polyethylene glycol 400
PFC plaque-forming cell
PFU plaque-forming units
PMR proportionate mortality ratio
PND postnatal day
PO
2
partial pressure oxygen
POD point of departure
PPAR peroxisome proliferator activated receptor
RBL-2H3 rat basophilic leukemia
xxxiii
LIST OF ABBREVIATIONS AND ACRONYMS (continued)


RCC renal cell carcinoma
RfC inhalation reference concentration
RfD oral reference dose
RNA ribonucleic acid
RR relative risk
RRm summary RR
RT reaction time
S9 metabolic activation system
SBA serum bile acids
SC sensitivity coefficient
SCE sister chromatid exchange
SD standard deviation
SDH sorbitol dehydrogenase
SE standard error
SEER Surveillance, Epidemiology, and End Results
SES socioeconomic status
SGA small for gestational age
SHBG sex-hormone binding globulin
SIR standardized incidence ratio
SMR standardized mortality ratio
SNP single nucleotide polymorphism
SRBC sheep red blood cells
SRT simple reaction time
SSB single-strand breaks
SSCP single stand conformation polymorphism
ssDNA single-stranded DNA
TaClo tetrahydro-beta-carbolines
TBARS thiobarbiturate acid-reactive substances
TCA trichloroacetic acid
TCAA trichloroacetaldehyde
TCAH trichloroacetaldehyde hydrate
TCE trichloroethylene
TCOG trichloroethanol-glucuronide conjugate
TCOH trichloroethanol
ThX T-helper Type X
TNF tumor necrosis factor
TRI Toxics Release Inventory
TSEP trigeminal somatosensory evoked potential
TTC total trichloro compounds
TWA time-weighted average
xxxiv
LIST OF ABBREVIATIONS AND ACRONYMS (continued)


U.S. EPA U.S. Environmental Protection Agency
UCL upper confidence limit
UDS unscheduled DNA synthesis
UF uncertainty factor
USGS United States Geological Survey
U-TCA urinary-TCA
U-TTC urinary total trichloro-compounds
VEGF vascular endothelial growth factor
VEP visual evoked potential
VHL von Hippel-Lindau
VLivC liver volume
VOC volatile organic compound
VSCC voltage sensitive calcium channel
W wakefulness
WHO World Health Organization
YFF fluorescent Y-bodies
xxxv
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
trichloroethylene. It is not intended to be a comprehensive treatise on the chemical or
toxicological nature of trichloroethylene.
The intent of Chapter 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 the 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 EPAs IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
[email protected] (email address).
xxxvi
AUTHORS, CONTRIBUTORS, AND REVIEWERS


CHEMICAL MANAGER

Weihsueh A. Chiu
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC


AUTHORS AND CONTRIBUTORS

Ambuja Bale
National Center for Environmental AssessmentImmediate Office
U.S. Environmental Protection Agency
Washington, DC

Stanley Barone
National Center for Environmental AssessmentImmediate Office
U.S. Environmental Protection Agency
Washington, DC

Rebecca Brown
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Jane C. Caldwell
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Chao Chen
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Weihsueh A. Chiu
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Glinda Cooper
National Center for Environmental AssessmentImmediate Office
U.S. Environmental Protection Agency
Washington, DC


xxxvii
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)


Ghazi Dannan
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Marina Evans
National Health and Environmental Effects Research Laboratory
(on detail to National Center for Environmental AssessmentWashington Office)
U.S. Environmental Protection Agency
Research Triangle Park, NC

John Fox
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Kathryn Z. Guyton
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Maureen R. Gwinn
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Jennifer Jinot
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Nagalakshmi Keshava
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

John Lipscomb
National Center for Environmental AssessmentCincinnati Office
U.S. Environmental Protection Agency
Cincinnati, OH



xxxviii
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)


Susan Makris
National Center for Environmental AssessmentWashington Office
U.S. Environmental Protection Agency
Washington, DC

Miles Okino
National Exposure Research LaboratoryLas Vegas Office
U.S. Environmental Protection Agency
Las Vegas, NV

Fred Power
National Exposure Research LaboratoryLas Vegas Office
U.S. Environmental Protection Agency
Las Vegas, NV

John Schaum
National Center for Environmental AssessmentWashington Office
Office of Research and Development
Washington, DC

Cheryl Siegel Scott
National Center for Environmental AssessmentWashington Office
Office of Research and Development
Washington, DC


REVIEWERS
This document has been reviewed by U.S. EPA scientists, reviewers from other Federal
agencies and White House offices, and the public, and peer reviewed by independent scientists
external to U.S. EPA. A summary and U.S. EPAs disposition of the comments received from
the independent external peer reviewers and from the public is included in Appendix I.


INTERNAL EPA REVIEWERS

Daniel Axelrad
National Center for Environmental Economics

Robert Benson
U.S. EPA Region 8


xxxix
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)


Ted Birner
National Center for Environmental AssessmentImmediate Office

Nancy Chiu
Office of Water

Joyce Donohue
Office of Water

David Farrar
National Center for Environmental AssessmentCincinnati Office

Lynn Flowers
National Center for Environmental AssessmentImmediate Office

Brenda Foos
Office of Children's Health Protection and Environmental Education

Stiven Foster
Office of Solid Waste and Emergency Response

Susan Griffin
U.S. EPA Region 8

Samantha Jones
National Center for Environmental AssessmentImmediate Office

Leonid Kopylev
National Center for Environmental AssessmentWashington Office

Allan Marcus
National Center for Environmental AssessmentImmediate Office

Margaret McDonough
U.S. EPA Region 1

Gregory Miller
Office of Children's Health Protection and Environmental Education

Deirdre Murphy
Office of Air Quality Planning and Standards

Marian Olsen
U.S. EPA Region 2


xl

AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)

Peter Preuss
(formerly) National Center for Environmental AssessmentImmediate Office

Kathleen Raffaele
(formerly) National Center for Environmental AssessmentWashington Office

Nancy Rios-Jafolla
U.S. EPA Region 3

William Sette
Office of Solid Waste and Emergency Response

Bob Sonawane
National Center for Environmental AssessmentWashington Office

Suryanarayana Vulimiri
National Center for Environmental AssessmentWashington Office

Nina Ching Y. Wang
National Center for Environmental AssessmentCincinnati Office

Paul White
National Center for Environmental AssessmentWashington Office

Marcia Bailey
U.S. EPA Region 10


xli

ACKNOWLEDGMENTS
Drafts of Section 3.3 (TCE metabolism) were prepared for the U.S. EPA by Syracuse
Research Corporation under contract. Additional support, including literature searches and
retrievals and drafts of Appendix D were prepared for the U.S. EPA by the Oak Ridge Institute
for Science and Education (ORISE) through interagency agreement number DW-89939822-01-0
with the U.S. Department of Energy (DOE). ORISE is managed by Oak Ridge Associated
Universities under a contract with DOE. The PBPK modeling sections of this report are
dedicated to the memory of Fred Power (19382007). His keen analytical mind will be greatly
missed, but his gentle heart and big smile will be missed even more.
Additionally, we gratefully acknowledge Terri Konoza and Ellen Lorang of NCEA for
their management of the document production and reference/citation management processes.
Technical editing support was provided by ICF International; IntelliTech Systems, Inc.; ECFlex,
Inc.; and Syracuse Research Corporation.

xlii
EXECUTIVE SUMMARY


There is substantial potential for human exposure to trichloroethylene (TCE), as it has a
widespread presence in ambient air, indoor air, soil, and groundwater. At the same time, humans
are likely to be exposed to a variety of compounds that are either metabolites of TCE or which
have common metabolites or targets of toxicity. Once exposed, humans, as well as laboratory
animal species, rapidly absorb TCE, which is then distributed to tissues via systemic circulation,
extensively metabolized, and then excreted primarily in breath as unchanged TCE or carbon
dioxide, or in urine as metabolites.
Based on the available human epidemiologic data and experimental and mechanistic
studies, it is concluded that TCE poses a potential human health hazard for noncancer toxicity to
the central nervous system, kidney, liver, immune system, male reproductive system, and the
developing fetus. The evidence is more limited for TCE toxicity to the respiratory tract and
female reproductive system. Following U.S. Environmental Protection Agency (U.S. EPA,
2005b) Guidelines for Carcinogen Risk Assessment, TCE is characterized as carcinogenic in
humans by all routes of exposure. This conclusion is based on convincing evidence of a causal
association between TCE exposure in humans and kidney cancer. The human evidence of
carcinogenicity from epidemiologic studies of TCE exposure is strong for non-Hodgkin
Lymphoma but less convincing than for kidney cancer, and more limited for liver and biliary
tract cancer. Less human evidence is found for an association between TCE exposure and other
types of cancer, including bladder, esophageal, prostate, cervical, breast, and childhood
leukemia, breast. Further support for the characterization of TCE as carcinogenic in humans by
all routes of exposure is derived from positive results in multiple rodent cancer bioassays in rats
and mice of both sexes, similar toxicokinetics between rodents and humans, mechanistic data
supporting a mutagenic mode of action for kidney tumors, and the lack of mechanistic data
supporting the conclusion that any of the mode(s) of action for TCE-induced rodent tumors are
irrelevant to humans.
As TCE toxicity and carcinogenicity are generally associated with TCE metabolism,
susceptibility to TCE health effects may be modulated by factors affecting toxicokinetics,
including lifestage, gender, genetic polymorphisms, race/ethnicity, preexisting health status,
lifestyle, and nutrition status. In addition, while these some of these factors are known risk
factors for effects associated with TCE exposure, it is not known how TCE interacts with known
risk factors for human diseases.
For noncancer effects, the most sensitive types of effects, based either on human
equivalent concentrations/doses or on candidate inhalation reference concentrations (RfCs)/oral
reference doses (RfDs), appear to be developmental, kidney, and immunological (adult and
developmental) effects. The neurological and reproductive effects appear to be about an order of
xliii
magnitude less sensitive, with liver effects another two orders of magnitude less sensitive. The
RfC of 0.0004 ppm (0.4 ppb or 2 g/m
3
) is based on route-to-route extrapolated results from oral
studies for the critical effects of heart malformations (rats) and immunotoxicity (mice). This
RfC value is further supported by route-to-route extrapolated results from an oral study of toxic
nephropathy (rats). Similarly, the RfD for noncancer effects of 0.0005 mg/kg/day is based on
the critical effects of heart malformations (rats), adult immunological effects (mice), and
developmental immunotoxicity (mice), all from oral studies. This RfD value is further supported
by results from an oral study for the effect of toxic nephropathy (rats) and route-to-route
extrapolated results from an inhalation study for the effect of increased kidney weight (rats).
There is high confidence in these noncancer reference values, as they are supported by moderate-
to-high confidence estimates for multiple effects from multiple studies.
For cancer, the inhalation unit risk is 2 10
-2
per ppm [4 10
-6
per g/m
3
], based on
human kidney cancer risks reported by Charbotel et al. (2006) and adjusted, using human
epidemiologic data, for potential risk for NHL and liver cancer. The oral unit risk for cancer is
5 10
-2
per mg/kg/day, resulting from physiologically based pharmacokinetic model-based
route-to-route extrapolation of the inhalation unit risk based on the human kidney cancer risks
reported in Charbotel et al. (2006) and adjusted, using human epidemiologic data, for potential
risk for NHL and liver cancer. There is high confidence in these unit risks for cancer, as they are
based on good quality human data, as well as being similar to unit risk estimates based on
multiple rodent bioassays. There is both sufficient weight of evidence to conclude that TCE
operates through a mutagenic mode of action for kidney tumors and a lack of TCE-specific
quantitative data on early-life susceptibility. Generally, the application of age-dependent
adjustment factors (ADAFs) is recommended when assessing cancer risks for a carcinogen with
a mutagenic mode of action. However, because the ADAF adjustment applies only to the kidney
cancer component of the total risk, it is likely to have a minimal impact on the total cancer risk
except when exposures are primarily during early life.

1-1
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
trichloroethylene (TCE). 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 ppm or g/m
3
) 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 ppm or g/m
3
in air breathed.
Development of these hazard identification and dose-response assessments for TCE has
followed the general guidelines for risk assessment as set forth by the National Research Council
(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: 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, 1986b), Guidelines for
Mutagenicity Risk Assessment (U.S. EPA, 1986a), Recommendations for and Documentation of
Biological Values for Use in Risk Assessment (U.S. EPA, 1988), Guidelines for Developmental
1-2
Toxicity Risk Assessment (U.S. EPA, 1991), Interim Policy for Particle Size and Limit
Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994b), Methods for Derivation of
Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA,
1994a), Use of the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995a),
Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996), Guidelines for
Neurotoxicity Risk Assessment (U.S. EPA, 1998a), 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, 2002b), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005b),
Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
(U.S. EPA, 2005e), Science Policy Council Handbook: Peer Review (U.S. EPA, 2006b), and A
Framework for Assessing Health Risks of Environmental Exposures to Children (U.S. EPA,
2006a).
The literature search strategy employed for this compound was based on the chemical
name, Chemical Abstracts Service Registry Number (CASRN), and multiple common
synonyms. Any pertinent scientific information submitted by the public to the IRIS Submission
Desk was also considered in the development of this document. Primary, peer-reviewed
literature identified through December 2010 was included where that literature was determined
to be critical to the assessment. The relevant literature included publications on trichloroethylene
which were identified through Toxicology Literature Online (TOXLINE), the U.S. National
Library of Medicine's MEDLINE, the Toxic Substance Control Act Test Submission Database
(TSCATS), the Registry of Toxic Effects of Chemical Substances (RTECS), the Chemical
Carcinogenesis Research Information System (CCRIS), the Developmental and Reproductive
Toxicology/Environmental Teratology Information Center (DART/ETIC), the Environmental
Mutagens Information Center (EMIC) and Environmental Mutagen Information Center Backfile
(EMICBACK) databases, the Hazardous Substances Data Bank (HSDB), the Genetic Toxicology
Data Bank (GENE-TOX), Chemical abstracts, and Current Contents. Other information,
including health assessments developed by other organizations, review articles, and independent
analyses of the health effects data were retrieved and may be included in the assessment where
appropriate. 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.
In addition to using peer-reviewed, published scientific literature, the preparation of this
toxicological review considered the advice to EPA from a 2002 SAB peer review report (SAB,
2002), a 2006 NRC consultation report (NRC, 2006), and a 2011 SAB peer review report (SAB,
2011), as well as comments from the public and other federal Agencies (weblinks).
2-1
2. EXPOSURE CHARACTERIZATION


The purpose of this exposure characterization is to summarize information about TCE
sources, releases, media levels, and exposure pathways for the general population (occupational
exposure is also discussed to a lesser extent). It is not meant as a substitute for a detailed
exposure assessment for a particular risk assessment application. While this section primarily
addresses TCE, it also includes some information on a number of related compounds. These
related compounds include metabolites of TCE and other parent compounds that produce similar
metabolites as shown in Table 2-1. The first column in this table lists the principal TCE
metabolites in humans (trichloroethanol, trichloroethanol-glucuronide, and trichloroacetic acid)
as well as a number of minor metabolites (ATSDR, 1997c). The subsequent columns list parent
compounds that can produce some of the same metabolites. The metabolic reaction pathways
are much more complicated than implied here and it should be understood that this table is
intended only to provide a general understanding of which parent compounds lead to which TCE
metabolites. Exposure to the TCE-related compounds can alter or enhance TCEs metabolism
and toxicity by generating higher internal metabolite concentrations than would result from TCE
exposure by itself. This characterization is based largely on earlier work by Wu and Schaum
(2001, 2000), but also provides updates in a number of areas.

Table 2-1. TCE metabolites and related parent compounds
a

TCE metabolites
Parent compounds
Tetrachloro-
ethylene
1,1-Dichloro-
ethane
1,1,1-Tri-
chloroethane
1,1,1,2-Tetra-
chloroethane
1,2-Dichloro-
ethylene
Oxalic acid X X
Chloral X
Chloral hydrate X
Monochloroacetic acid X X X X X
Dichloroacetic acid X X X
Trichloroacetic acid X X X
Trichloroethanol X X X
Trichloroethanol-
glucuronide
X X X

a
X indicates that the parent compound can produce the corresponding metabolite (Hazardous Substances Data Bank,
http://toxnet.nlm.nih.gov./cgi-bin/sis/htmlgen?HSDB).

2.1. ENVIRONMENTAL SOURCES
TCE is a stable, colorless liquid with a chloroform-like odor and chemical formula
C
2
Cl
3
H as diagrammed in Figure 2-1 (Hawley and Lewis, 2001). Its chemical properties are
listed in Table 2-2.

2-2


C
Cl
Cl
H
C
Cl


Figure 2-1. Molecular structure of TCE.
Table 2-2. Chemical properties of TCE

Property Value Reference
Molecular weight 131.39 Lide et al. (1998)
Boiling point 87.2C Lide et al. (1998)
Melting point 84.7C Lide et al. (1998)
Density 1.4642 at 20C Budavari (1996)
Solubility 1,280 mg/L water at 25C Horvath et al. (1999)
Vapor pressure 69.8 mmHG @ 25C Boublik et al. (1984)
Vapor density 4.53 (air = 1) Budavari (1996)
Henrys law constant 9.85 10
3
atm-cu m/mol @ 25C Leighton and Calo (1981)
Octanol/water partition coefficient log K
ow
= 2.61 Hansch et al. (1995)
Air concentration conversion 1 ppb = 5.38 g/m
3
HSDB (2002)

TCE has been produced commercially since the 1920s in many countries by chlorination
of ethylene or acetylene. Its use in vapor degreasing began in the 1920s. In the 1930s, it was
introduced for use in dry cleaning. This use was largely discontinued in the 1950s and was
replaced with tetrachloroethylene (ATSDR, 1997c). More recently, 8090% of TCE production
worldwide is used for degreasing metals (IARC, 1995a). It is also used in adhesives, paint-
stripping formulations, paints, lacquers, and varnishes (SRI, 1992). A number of past uses in
cosmetics, drugs, foods, and pesticides have now been discontinued including use as an
extractant for spice oleoresins, natural fats and oils, hops, and decaffeination of coffee (IARC,
1995a), and as a carrier solvent for the active ingredients of insecticides and fungicides, and for
spotting fluids (ATSDR, 1997c; WHO, 1985). The production of TCE in the United States
peaked at 280 million kg (616 million pounds) in 1970 and declined to 60 million kg
(132 million pounds) in 1998 (USGS, 2006). In 1996, the United States imported 4.5 million kg
(10 million pounds) and exported 29.5 million kg (65 million pounds) (Chemical Marketing
Reporter, 1997). Table 2-3 summarizes the basic properties and principal uses of the TCE
related compounds.

2-3
Table 2-3. Properties and uses of TCE related compounds


Water
solubility
(mg/L)
Vapor pressure
(mmHG) Uses References
Tetrachloroethylene 150 18.5 @25C Dry cleaning, degreasing, solvent Wu and
Schaum
(2001)
1,1,1-Trichloroethane 4,400 124 @25C Solvents, degreasing Wu and
Schaum
(2001)
1,2-Dichloroethylene 3,0006,000 273395 @30C Solvents, chemical intermediates Wu and
Schaum
(2001)
1,1,1,2-
Tetrachloroethane
1,100 14 @25C Solvents, but currently not produced in
United States
HSDB,
2002; Wu
and Schaum
(2001)
1,1-Dichloroethane 5,500 234 @25C Solvents, chemical intermediates Wu and
Schaum
(2001)
Chloral High 35 @20C Herbicide production Wu and
Schaum
(2001)
Chloral hydrate High NA Pharmaceutical production Wu and
Schaum
(2001)
Monochloroacetic acid High 1 @43C Pharmaceutical production Wu and
Schaum
(2001)
Dichloroacetic acid High <1 @20C Pharmaceuticals, not widely used Wu and
Schaum
(2001)
Trichloroacetic acid High 1 @50C Herbicide production Wu and
Schaum
(2001)
Oxalic acid 220,000 0.54 @105C Scouring/cleaning agent, degreasing HSDB
(2002)
Dichlorovinyl cysteine Not available Not available Not available
Trichloroethanol Low NA Anesthetics and chemical intermediate Hawley and
Lewis
(2001)

Releases of TCE from nonanthropogenic activities are negligible (HSDB, 2002). Most of
the TCE used in the United States is released to the atmosphere, primarily from vapor degreasing
operations (ATSDR, 1997c). Releases to air also occur at treatment and disposal facilities, water
treatment facilities, and landfills (ATSDR, 1997c). TCE has also been detected in stack
emissions from municipal and hazardous waste incineration (ATSDR, 1997c). TCE is on the list
for reporting to U.S. EPAs Toxics Release Inventory (TRI). Reported releases into air
predominate over other types and have declined over the period 19942004 (see Table 2-4).
2-4

Table 2-4. TRI releases of TCE (pounds/year)

Yr
On-site
fugitive air
On-site
stack air
Total on-site
air emissions
On-site
surface
water
discharges
Total on-site
underground
injection
Total on-
site
releases to
land
Total off-
site
disposal or
other
releases
Total on-
and off-
site
disposal or
other
releases
1994 15,018,818 15,929,943 30,948,761 1,671 288 4,070 96,312 31,051,102
1995 12,498,086 13,784,853 26,282,939 1,477 550 3,577 74,145 26,362,688
1996 10,891,223 10,995,228 21,886,451 541 1,291 9,740 89,527 21,987,550
1997 9,276,150 8,947,909 18,224,059 568 986 3,975 182,423 18,412,011
1998 6,769,810 6,504,289 13,274,099 882 593 800 136,766 13,413,140
1999 5,861,635 4,784,057 10,645,692 1,034 0 148,867 192,385 10,987,978
2000 5,485,493 4,375,516 9,861,009 593 47,877 9,607 171,952 10,091,038
2001 4,968,282 3,453,451 8,421,733 406 98,220 12,609 133,531 8,666,499
2002 4,761,104 3,436,289 8,197,393 579 140,190 230 139,398 8,477,790
2003 3,963,054 3,121,718 7,084,772 595 90,971 150,642 66,894 7,393,873
2004 3,040,460 3,144,980 6,185,440 216 123,637 2 71,780 6,381,075
2005 2,733,983 2,893,168 5,627,152 533 86,817 4,711 60,074 5,779,287
2006 2,816,241 2,795,184 5,611,425 482 0 77,339 90,758 5,780,004

Source: EPA TRI Explorer, http://www.epa.gov/triexplorer/trends.htm.

Under the National-Scale Air Toxics Assessment (NSATA) program, EPA has developed
an emissions inventory for TCE (U.S. EPA, 2007a). The inventory includes sources in the
United States plus the Commonwealth of Puerto Rico and the U.S. Virgin Islands. The types of
emission sources in the inventory include large facilities, such as waste incinerators and factories
and smaller sources, such as dry cleaners and small manufacturers. Figures 2-2 and 2-3 show the
results of the 1999 emissions inventory for TCE. Figure 2-2 shows the percent contribution to
total emissions by source category. A variety of sources have TCE emissions with the largest
ones identified as halogenated solvent cleaners and metal parts and products. Figure 2-3 shows a
national map of the emission density (tons/square miles/year) for TCE. This map shows the
highest densities in the far west and northeastern regions of the United States. Emissions range
from 0 to 4.12 tons/square miles/year.

2-5



Figure 2-2. Source contribution to TCE emissions.
Figure 2-3. Annual emissions of TCE.
2-6

2.2. ENVIRONMENTAL FATE
2.2.1. Fate in Terrestrial Environments
The dominant fate of TCE released to surface soils is volatilization. Because of its
moderate water solubility, TCE introduced into soil (e.g., landfills) also has the potential to
migrate through the soil into groundwater; this is confirmed by the relatively frequent detection
of TCE in groundwater. Biodegradation in soil and groundwater may occur at a relatively slow
rate (half-lives on the order of months to years) (Howard et al., 1991).

2.2.2. Fate in the Atmosphere
In the atmosphere, TCE is expected to be present primarily in the vapor phase, rather than
sorbed to particulate, because of its high vapor pressure. Some removal by scavenging during
wet precipitation is expected because of its moderate water solubility. The major degradation
process affecting vapor-phase TCE is photo-oxidation by hydroxyl radicals. Photolysis in the
atmosphere proceeds very slowly, if at all. TCE does not absorb ultraviolet light at wavelengths
of <290 nm and thus, will not directly photolyze. Based on measured rate data for the vapor
phase photo-oxidation reaction with hydroxyl radicals, the estimated half-life of TCE in the
atmosphere is on the order of 111 days with production of phosgene, dichloroacetyl chloride
(DCAC), and formyl chloride. Under smog conditions, degradation is more rapid (half-life on
the order of hours) (HSDB, 2002; Howard et al., 1991).

2.2.3. Fate in Aquatic Environments
The dominant fate of TCE released to surface waters is volatilization (predicted half-life
of minutes to hours). Bioconcentration, biodegradation, and sorption to sediments and
suspended solids are not thought to be significant (HSDB, 2002). TCE is not hydrolyzed under
normal environmental conditions. However, slow photo-oxidation in water (half-life of
10.7 months) has been reported (HSDB, 2002; Howard et al., 1991).
2.3. EXPOSURE CONCENTRATIONS
TCE levels in the various environmental media result from the releases and fate processes
discussed in Sections 2.1 and 2.2. No statistically based national sampling programs have been
conducted that would allow estimates of true national means for any environmental medium. A
substantial amount of air and groundwater data, however, has been collected as well as some
data in other media, as described below.

2.3.1. Outdoor AirMeasured Levels
TCE has been detected in the air throughout the United States. According to ATSDR
(1997c), atmospheric levels are highest in areas concentrated with industry and population, and
2-7
lower in remote and rural regions. Table 2-5 shows levels of TCE measured in the ambient air at
a variety of locations in the United States.

Table 2-5. Concentrations of TCE in ambient air

Area Yr
Concentration (g/m
3
)
Mean Range
Rural
Whiteface Mountain, New York
a

Badger Pass, California
a

Reese River, Nevada
a

Jetmar, Kansas
a

All rural sites

1974
1977
1977
1978
19741978

0.5
0.06
0.06
0.07

<0.31.9
0.0050.09
0.0050.09
0.040.11
0.0051.9
Urban and suburban
New Jersey
a

New York City, New York
a

Los Angeles, California
a

Lake Charles, Louisiana
a

Phoenix, Arizone
a

Denver, Colorado
a

St. Louis, Missouri
a

Portland, Oregon
a

Philadelphia, Pennsylvania
a

Southeast Chicago, Illinois
b

East St. Louis, Illinois
b

District of Columbia
c

Urban Chicago, Illinois
d

Suburban Chicago, Illinois
d

300 cities in 42 states
e

Several Canadian Cities
f

Several United States Cities
f

Phoenix, Arizona
g

Tucson, Arizona
g

All urban/suburban sites

19731979
1974
1976
19761978
1979
1980
1980
1984
19831984
19861990
19861990
19901991
pre1993
pre1993
pre1986
1990
1990
19941996
19941996
19731996

9.1
3.8
1.7
8.6
2.6
1.07
0.6
1.5
1.9
1.0
2.1
1.94
0.821.16
0.52
2.65
0.28
6.0
0.29
0.23

ND97
0.65.9
0.149.5
0.411.3
0.0616.7
0.152.2
0.11.3
0.63.9
1.62.1


116.65





01.53
01.47
097

a
IARC (1995a).
b
Sweet (1992).
c
Hendler (1992).
d
Scheff (1993).
e
Shah (1988).
f
Bunce (1994).
g
Zielinska-Psuja (1998).

ND = nondetect

More recent ambient air measurement data for TCE were obtained from EPAs Air
Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html
(2007b). These data were collected from a variety of sources including state and local
environmental agencies. The data are not from a statistically based survey and cannot be
assumed to provide nationally representative values. The most recent data (2006) come from
258 monitors located in 37 states. The means for these monitors range from 0.03 to 7.73 g/m
3

2-8
and have an overall average of 0.23 g/m
3
. Table 2-6 summarizes the data for the years
19992006. The data suggest that levels have remained fairly constant since 1999 at about
0.3 g/m
3
. Table 2-7 shows the monitoring data organized by land setting (rural, suburban, or
urban) and land use (agricultural, commercial, forest, industrial, mobile, and residential). Urban
air levels are almost 4 times higher than rural areas. Among the land use categories, TCE levels
are highest in commercial/industrial areas and lowest in forest areas.

Table 2-6. TCE ambient air monitoring data (g/m
3
)

Yr
Number of
monitors Number of states Mean
Standard
deviation Median Range
1999 162 20 0.30 0.53 0.16 0.014.38
2000 187 28 0.34 0.75 0.16 0.017.39
2001 204 31 0.25 0.92 0.13 0.0112.90
2002 259 41 0.37 1.26 0.13 0.0118.44
2003 248 41 0.35 0.64 0.16 0.026.92
2004 256 37 0.32 0.75 0.13 0.005.78
2005 313 38 0.43 1.05 0.14 0.006.64
2006 258 37 0.23 0.55 0.13 0.037.73

Source: EPAs Air Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html.

Table 2-7. Mean TCE air levels across monitors by land setting and use
(19851998)

Rural Suburban Urban
Agricul-
tural
Com-
mercial Forest
Indus-
trial Mobile
Resi-
dential
Mean
concentration
(g/m
3
)
0.42 1.26 1.61 1.08 1.84 0.1 1.54 1.5 0.89
n 93 500 558 31 430 17 186 39 450

Source: EPAs Air Quality System database at the AirData Web site: http://www.epa.gov/air/data/index.html.

2.3.2. Outdoor AirModeled Levels
Under the National-Scale Air Toxics Assessment program, EPA has compiled emissions
data and modeled air concentrations/exposures for the Criteria Pollutants and Hazardous Air
Pollutants (U.S. EPA, 2007a). The results of the 1999 emissions inventory for TCE were
discussed earlier and results presented in Figures 2-2 and 2-3. A computer simulation model
known as the Assessment System for Population Exposure Nationwide (ASPEN) is used to
estimate toxic air pollutant concentrations (http://www.epa.gov/ttnatw01/nata/aspen.html). This
model is based on the EPAs Industrial Source Complex Long Term model which simulates the
behavior of the pollutants after they are emitted into the atmosphere. ASPEN uses estimates of
toxic air pollutant emissions and meteorological data from National Weather Service Stations to
2-9
estimate air toxics concentrations nationwide. The ASPEN model takes into account important
determinants of pollutant concentrations, such as:

- rate of release;
- location of release;
- the height from which the pollutants are released;
- wind speeds and directions from the meteorological stations nearest to the release;
- breakdown of the pollutants in the atmosphere after being released (i.e., reactive decay);
- settling of pollutants out of the atmosphere (i.e., deposition); and
- transformation of one pollutant into another (i.e., secondary formation).

The model estimates toxic air pollutant concentrations for every census tract in the
continental United States, the Commonwealth of Puerto Rico and the U.S. Virgin Islands.
Census tracts are land areas defined by the U.S. Bureau of the Census and typically contain about
4,000 residents each. Census tracts are usually smaller than 2 square miles in size in cities but
much larger in rural areas.
Figure 2-4 shows the results of the 1999 ambient air concentration modeling for TCE.
The county median air levels range from 0 to 3.79 g/m
3
and an overall median of 0.054 g/m
3
.
They have a pattern similar to the emission densities shown in Figure 2-3. These NSATA
modeled levels appear lower than the monitoring results presented above. For example, the 1999
air monitoring data (see Table 2-6) indicates a median outdoor air level of 0.16 g/m
3
which is
about 3 times as high as the modeled 1999 county median (0.054 g/m
3
). However, it should be
understood that the results from these two efforts are not perfectly comparable. The modeled
value is a median of county levels for the entire United States which includes many rural areas.
The monitors cover many fewer areas (n = 162 for 1999) and most are in nonrural locations. A
better analysis is provided by EPA (2007a) which presents a comparison of modeling results
from NSATA to measured values at the same locations. For 1999, it was found that
formaldehyde levels were underestimated at 79% of the sites (n = 92). Thus, while the NSATA
modeling results are useful for understanding geographic distributions, they may frequently
underestimate ambient levels.

2-10

Figure 2-4. Modeled ambient air concentrations of TCE.

2.3.3. Indoor Air
TCE can be released to indoor air from use of consumer products that contain it (i.e.,
adhesives and tapes), vapor intrusion (migration of volatile chemicals from the subsurface into
overlying buildings) and volatilization from the water supply. Where such sources are present, it
is likely that indoor levels will be higher than outdoor levels. A number of studies have
measured indoor levels of TCE:

- The 1987 EPA Total Exposure Assessment Methodology study (Wallace, 1987) showed
that the ratio of indoor to outdoor TCE concentrations for residences in Greensboro, NC,
was about 5:1.
- In two homes using well water with TCE levels averaging 22128 g/L, the TCE levels
in bathroom air ranged from <50040,000 g/m
3
when the shower ran <30 minutes
(Andelman, 1985).
- Shah and Singh (1988) report an average indoor level of 7.2 g/m
3
based on over
2,000 measurements made in residences and workplaces during 19811984 from various
locations across the United States.
- Hers et al. (2001) provides a summary of indoor air TCE measurements at locations in
United States, Canada, and Europe with a range of <1165 g/m
3
.
2-11
- Sapkota et al. (2005) measured TCE levels inside and outside of the Baltimore Harbor
Tunnel toll booths during the summer of 2001. Mean TCE levels were 3.11 g/m
3

indoors and 0.08 g/m
3
outdoors based on measurements on 7 days. The authors
speculated that indoor sources, possibly dry cleaning residues on uniforms, were the
primary source of the indoor TCE.
- Sexton et al. (2005) measured TCE levels inside and outside residences in
Minneapolis/St. Paul metropolitan area. Two day samples were collected over
three seasons in 1999. Mean TCE levels were 0.5 g/m
3
indoors (n = 292), 0.2 g/m
3

outdoors (n = 132) and 1.0 g/m
3
based on personal sampling (n = 288).
- Zhu et al. (2005) measured TCE levels inside and outside of residences in Ottawa,
Canada. Seventy-five homes were randomly selected and measurements were made
during the winter of 2002/2003. TCE was above detection limits in the indoor air of
33% of the residences and in the outdoor air of 19% of the residences. The mean levels
were 0.06 g/m
3
indoors and 0.08 g/m
3
outdoors. Given the high frequency of
nondetects, a more meaningful comparison can be made on basis of the 75
th
percentiles:
0.08 g/m
3
indoors and 0.01 g/m
3
outdoors.


TCE levels measured indoors have been directly linked to vapor intrusion at two sites in New
York:


- TCE vapor intrusion has occurred in buildings/residences near a former Smith Corona
manufacturing facility located in Cortlandville, New York. An extensive sampling
program conducted in 2006-2007 has detected TCE in groundwater (up to 22 g/L),
subslab gas (up to 1,000 g/m
3
), and indoor air (up to 34 g/m
3
) (NYSDEC, 2007).
- Evidence of vapor intrusion of TCE has also been reported in buildings and residences in
Endicott, New York. Sampling in 2003 showed total volatile organic compounds
(VOCs) in soil gas exceeding 10,000 g/m
3
in some areas. Indoor air sampling detected
TCE levels ranging from 1 to 140 g/m
3
(Meyers, 2003).

Little et al. (1992) developed attenuation coefficients relating contaminants in soil gas
(assumed to be in chemical equilibrium with the groundwater) to possible indoor levels as a
result of vapor intrusion. On this basis they estimated that TCE groundwater levels of 540 g/L,
(a high contamination level) could produce indoor air levels of 5500 g/m
3
. Vapor intrusion
can be an important contributor to indoor levels in situations where residences are located near
soils or groundwater with high contamination levels. EPA (2002c) recommends considering
vapor intrusion when volatiles are suspected to be present in groundwater or soil at a depth of
<100 feet. Hers et al. (2001) concluded that the contribution of VOCs from subsurface sources
relative to indoor sources is small for most chemicals and sites.

2-12
2.3.4. Water
A number of early (pre-1990) studies measured TCE levels in natural water bodies
(levels in drinking water are discussed later in this section) as summarized in Table 2-8.

Table 2-8. Concentrations of TCE in water based on pre-1990 studies

Water type Location Yr
Mean
(g/L)
Median
(g/L)
Range
(g/L)
Number of
samples Reference
Industrial effluent United States 1983 0.5 NR IARC (1995a)
Surface waters United States 1983 0.1 NR IARC (1995a)
Rainwater Portland,
Oregon
1984 0.006 0.0020.02 NR Ligocki et al. (1985)
Groundwater Minnesota
New Jersey
New York
Pennsylvania
Massachusetts
Arizona
1983
1976
1980
1980
1976

0.2144
1,530
3,800
27,300
900
8.929
NR
NR
NR
NR
NR
NR
Sabel and Clark (1984)
Burmaster et al. (1982)
Burmaster et al. (1982)
Burmaster et al. (1982)
Burmaster et al. (1982)
IARC (1995a)
Drinking water United States
United States
United States
Massachusetts
New Jersey
California
California
North Carolina
North Dakota
1976
1977
1978
1984
1984
1985
1984
1984
1984




23.4

66
5
5
0.249
053
0.5210
max. 267
max. 67
812




1130
486
486
48
48
IARC (1995a)
IARC (1995a)
IARC (1995a)
IARC (1995a)
Cohn et al. (1994b)
EPA, (1987)
EPA, (1987)
EPA, (1987)
EPA, (1987)

NR = not reported

According to IARC (1995a), the reported median concentrations of TCE in 19831984
were 0.5 g/L in industrial effluents and 0.1 g/L in ambient water. Results from an analysis of
the EPA STORET Data Base (19801982) showed that TCE was detected in 28% of
9,295 surface water reporting stations nationwide (ATSDR, 1997c). A more recent search of the
STORET database for TCE measurements nationwide during 2008 in streams, rivers and lakes
indicated three detects (0.030.04 g/L) out of 150 samples (STORET Database,
http://www.epa.gov/storet/dbtop.html).
ATSDR (1997c) has reported that TCE is the most frequently reported organic
contaminant in groundwater and the one present in the highest concentration in a summary of
ground water analyses reported in 1982. It has been estimated that between 9 and 34% of the
drinking water supply sources tested in the United States may have some TCE contamination.
This estimate is based on available Federal and State surveys (ATSDR, 1997c).
Squillace et al. (2004) reported TCE levels in shallow groundwater based on data from
the National Water Quality Assessment Program managed by United States Geological Survey
(USGS). Samples from 518 wells were collected from 1996 to 2002. All wells were located in
2-13
residential or commercial areas and had a median depth of 10 m. The authors reported that
approximately 8.3% of the well levels were above the detection limit (level not specified), 2.3%
were above 0.1 g/L and 1.7% were above 0.2 g/L.
As part of the Agencys first Six-Year Review, EPA obtained analytical results for over
200,000 monitoring samples reported at 23,035 public water systems (PWS) in 16 states (U.S.
EPA, 2003c). Approximately 2.6% of the systems had at least one sample exceed a minimum
reporting level of 0.5 g/L; almost 0.65% had at least one sample that exceeds the maximum
contaminant level of 5 g/L. Based on average system concentrations estimated by EPA,
54 systems (0.23%) had an average concentration that exceeded the maximum contaminant level.
EPAs statistical analysis to extrapolate the sample result to all systems regulated for TCE
resulted in an estimate of 154 systems with average TCE concentrations that exceed the
maximum contaminant level.
TCE concentrations in ground water have been measured extensively in California. The
data were derived from a survey of water utilities with more than 200 service connections. The
survey was conducted by the California Department of Health Services (CDHS, 1986). From
January 1984 through December 1985, untreated water from wells in 819 water systems were
sampled for organic chemical contamination. The water systems use a total of 5,550 wells,
2,947 of which were sampled. TCE was found in 187 wells at concentrations up to 440 g/L,
with a median concentration among the detects of 3.0 g/L. Generally, the wells with the highest
concentrations were found in the heavily urbanized areas of the state. Los Angeles County
registered the greatest number of contaminated wells (149).
A second California study collected data on TCE levels in public drinking water
(Williams et al., 2002). The data were obtained from the CA DHS. The data spanned the years
19952001 and the number of samples for each year ranged from 3,447 to 4,226. The percent of
sources that were above the detection limit ranged from 9.6 to 11.7 per year (detection limits not
specified). The annual average detected concentrations ranged from 14.2 to 21.6 g/L.
Although not reported, the overall average concentration of the samples (assuming an average of
20 g/L among the samples above the detection limit, 10% detection rate and 0 for the
nondetects) would be about 2 g/L.
The USGS (2006) conducted a national assessment of 55 VOCs, including TCE, in
ground water. A total of 3,500 water samples were collected during 19852001. Samples were
collected at the well head prior to any form of treatment. The types of wells sampled included
2,400 domestic wells and 1,100 public wells. Almost 20% of the samples contained one or more
of the VOCs above the assessment level of 0.2 g/L. The detection frequency increased to over
50% when a subset of samples was analyzed with a low level method that had an assessment
level of 0.02 g/L. The largest detection frequencies were observed in California, Nevada,
Florida, the New England States, and Mid-Atlantic states. The most frequently detected VOCs
2-14
(>1% of samples) include TCE, tetrachloroethylene, 1,1,1-trichloroethane (methyl chloroform),
1,2 dichloroethylene, and 1,1-dichloroethane. Findings specific to TCE include the following:


- Detection frequency was 2.6% at 0.2 g/L and was 3.8% at 0.02 g/L.
- The median concentration was 0.15 g/L with a range of 0.02100 g/L.
- The number of samples exceeding the maximum contaminant level (5 g/L) was six at
domestic wells and nine at public wells.


USGS (2006) also reported that four solvents (TCE, tetrachloroethylene, 1,1,1-trichloro-
ethane and methylene chloride) occurred together in 5% of the samples. The most frequently
occurring two-solvent mixture was TCE and tetrachloroethylene. The report stated that the most
likely reason for this co-occurrence is the reductive dechlorination of tetrachloroethylene to
TCE.

2.3.5. Other Media
Levels of TCE were found in the sediment and marine animal tissue collected in
19801981 near the discharge zone of a Los Angeles County waste treatment plant.
Concentrations were 17 g/L in the effluent, <0.5 g/kg in dry weight in sediment, and
0.37 g/kg wet weight in various marine animal tissue (IARC, 1995a). TCE has also been
found in a variety of foods. U.S. Food and Drug Administration (FDA) has limits on TCE use as
a food additive in decaffeinated coffee and extract spice oleoresins (see Table 2-15). Table 2-9
summarizes data from two sources:

- IARC (1995a) reports average concentrations of TCE in limited food samples collected in
the United States.
- Jones and Smith (2003) measured VOC levels in over 70 foods collected from 1996 to
2000 as part of the FDAs Total Diet Program. All foods were collected directly from
supermarkets. Analysis was done on foods in a ready-to-eat form. Sample sizes for most
foods were in the 25 range.

2-15
Table 2-9. Levels in food

IARC (1995a) Fleming-Jones and Smith (2003)
Cheese 3.8 g/kg
Butter and margarine 73.6 g/kg
Cheese 23 g/kg
Butter 79 g/kg
Margarine 221 g/kg
Cheese pizza 2 g/kg
Peanut butter 0.5 g/kg Nuts 25 g/kg
Peanut butter 470 g/kg
Ground beef 36 g/kg
Beef frankfurters 2105 g/kg
Hamburger 59 g/kg
Cheeseburger 7 g/kg
Chicken nuggets 25 g/kg
Bologna 220 g/kg
Pepperoni pizza 2 g/kg
Banana 2 g/kg
Avocado 275 g/kg
Orange 2 g/kg
Chocolate cake 357 g/kg
Blueberry muffin 34 g/kg
Sweet roll 3 g/kg
Chocolate chip cookies 24 g/kg
Apple pie 24 g/kg
Doughnuts 3 g/kg
Tuna 911 g/kg
Cereals 3 g/kg
Grainbased foods 0.9 g/kg
Cereal 3 g/kg
Popcorn 48 g/kg
French fries 3 g/kg
Potato chips 4140 g/kg
Coleslaw 3 g/kg

2.3.6. Biological Monitoring
Biological monitoring studies have detected TCE in human blood and urine in the United
States and other countries such as Croatia, China, Switzerland, and Germany (IARC, 1995a).
Concentrations of TCE in persons exposed through occupational degreasing operations were
most likely to have detectable levels (IARC, 1995a). In 1982, eight of eight human breastmilk
samples from four United States urban areas had detectable levels of TCE. The levels of TCE
detected, however, are not specified (HSDB, 2002; ATSDR, 1997c).
The Third National Health and Nutrition Examination Survey (NHANES III) examined
TCE concentrations in blood in 677 nonoccupationally exposed individuals. The individuals
were drawn from the general U.S. population and selected on the basis of age, race, gender and
region of residence (IARC, 1995a; Ashley et al., 1994). The samples were collected during
19881994. TCE levels in whole blood were below the detection limit of 0.01 g/L for about
90% of the people sampled (see Table 2-10). Assuming that nondetects equal half of the
detection limit, the mean concentration was about 0.017 g/L.
2-16

Table 2-10. TCE levels in whole blood by population percentile

Percentiles 10 20 30 40 50 60 70 80 90
Concentration (g/L) ND ND ND ND ND ND ND ND 0.012

ND = Nondetect, i.e., below detection limit of 0.01 g/L.

Sources: IARC (1995a); Ashley et al. (1994).

2.4. EXPOSURE PATHWAYS AND LEVELS
2.4.1. General Population
Because of the pervasiveness of TCE in the environment, most people are likely to have
some exposure via one or more of the following pathways: ingestion of drinking water,
inhalation of outdoor/indoor air, or ingestion of food (ATSDR, 1997c). As noted earlier, the
NHANES survey suggests that about 10% of the population has detectable levels of TCE in
blood. Each pathway is discussed below.

2.4.1.1. Inhalation
As discussed earlier, EPA has estimated emissions and modeled air concentrations for the
Criteria Pollutants and Hazardous Air Pollutants under the National-Scale Air Toxics
Assessment program (U.S. EPA, 2007a). This program has also estimated inhalation exposures
on a nationwide basis. The exposure estimates are based on the modeled concentrations from
outdoor sources and human activity patterns (U.S. EPA, 2005a). Table 2-11 shows the 1999
results for TCE.

Table 2-11. Modeled 1999 annual exposure concentrations (g/m
3
) for TCE

Percentile
Exposure concentration (g/m
3
)
Rural areas Urban areas Nationwide
5 0.030 0.048 0.038
10 0.034 0.054 0.043
25 0.038 0.065 0.056
50 0.044 0.086 0.076
75 0.053 0.122 0.113
90 0.070 0.189 0.172
95 0.097 0.295 0.262
Mean 0.058 0.130 0.116

Percentiles and mean are based on census tract values.

Source: http://www.epa.gov/ttn/atw/nata/ted/exporisk.html#indb.

2-17
These modeled inhalation exposures would have a geographic distribution similar to that
of the modeled air concentrations as shown in Figure 2-4. Table 2-11 indicates that TCE
inhalation exposures in urban areas are generally about twice as high as rural areas. While these
modeling results are useful for understanding the geographic distribution of exposures, they
appear to underestimate actual exposures. This is based on the fact that, as discussed earlier, the
modeled ambient air levels are generally lower than measured values. Also, the modeled
exposures do not consider indoor sources. Indoor sources of TCE make the indoor levels higher
than ambient levels. This is particularly important to consider since people spend about 90% of
their time indoors (U.S. EPA, 1997). A number of measurement studies were presented earlier
that showed higher TCE levels indoors than outdoors. Sexton et al. (2005) measured TCE levels
in Minneapolis/St. Paul area and found means of 0.5 g/m
3
indoors (n = 292) and 1.0 g/m
3

based on personal sampling (n = 288). Using 1.0 g/m
3
and an average adult inhalation rate of
13 m
3
air/day (U.S. EPA, 1997) yields an estimated intake of 13 g/day. This is consistent with
ATSDR (1997c), which reported an average daily air intake for the general population of
1133 g/day.

2.4.1.2. Ingestion
The median value from the nationwide survey of domestic and public wells by USGS for
19852001 is 0.15 g/L. This value was selected for exposure estimation purposes because it
was the most current and most representative of the national population. Using this value and an
average adult water consumption rate of 1.4 L/d yields an estimated intake of 0.2 g/day. [This is
from U.S. EPA (1997), but note that U.S. EPA (2004) indicates a mean per capita daily average
total water ingestion from all sources of 1.233 L]. This is lower than the ATSDR (1997c)
estimate water intake for the general population of 220 g/day. The use of the USGS survey to
represent drinking water is uncertain in two ways. First, the USGS survey measured only
groundwater and some drinking water supplies use surface water. Second, the USGS measured
TCE levels at the well head, not the drinking water tap. Further discussion about the possible
extent and magnitude of TCE exposure via drinking water is presented below.
According to ATSDR (1997c), TCE is the most frequently reported organic contaminant
in ground water (1997c), and between 9 and 34% of the drinking water supply sources tested in
the United States may have some TCE contamination. Approximately 90% of the
155,000 public drinking water systems
1
in the United States are ground water systems. The
drinking water standard for TCE only applies to community water systems (CWSs) and
approximately 78% of the 51,972 CWSs in the United States are ground water systems (U.S.
EPA, 2008a). Although commonly detected in water supplies, the levels are generally low

1
PWSs are defined as systems which provide water for human consumption through pipes or other constructed
conveyances to at least 15 service connections or serves an average of at least 25 people for at least 60 days a year.
EPA further specifies three types of PWSs, including CWS)a PWS that supplies water to the same population
year-round.
2-18
because, as discussed earlier, maximum contaminant level violations for TCE in public water
supplies are relatively rare for any extended period (U.S. EPA, 1998b). The USGS (2006)
survey found that the number of samples exceeding the maximum contaminant level (5 g/L)
was six at domestic wells (n = 2,400) and nine at public wells (n = 1,100). Private wells,
however, are often not closely monitored and if located near TCE disposal/contamination sites
where leaching occurs, may have undetected contamination levels. About 10% of Americans
(27 million people) obtain water from sources other than public water systems, primarily private
wells (U.S. EPA, 1995b). TCE is a common contaminant at Superfund sites. It has been
identified in at least 861 of the 1,428 hazardous waste sites proposed for inclusion on the EPA
National Priorities List (NPL) (ATSDR, 1997c). Studies have shown that many people live near
these sites: 41 million people live <4 miles from one or more of the nations NPL sites, and on
average 3,325 people live within 1 mile of any given NPL site (ATSDR, 1996b).
Table 2-12 presents preliminary estimates of TCE intake from food. They are based on
average adult food ingestion rates and food data from Table 2-9. This approach suggests a total
ingestion intake of about 5 g/d. It is important to consider this estimate as preliminary because
it is derived by applying data from very limited food samples to broad classes of food.

Table 2-12. Preliminary estimates of TCE intake from food ingestion


Consumption rate
(g/kg-d)
Consumption rate
(g/d)
Concentration in
food (g/kg)
Intake
(g/d)
Fruit 3.4 238 2 0.48
Vegetables 4.3 301 3 0.90
Fish 20 10 0.20
Meat 2.1 147 5 0.73
Dairy products 8 560 3 1.68
Grains 4.1 287 3 0.86
Sweets 0.5 35 3 0.10
Total 4.96

a
Consumption rates are per capita averages from EPA (1997).
b
Consumption rates in g/d assume 70 kg body weight.

2.4.1.3. Dermal
TCE in bathing water and consumer products can result in dermal exposure. A modeling
study has suggested that a significant fraction of the total dose associated with exposure to
volatile organics in drinking water results from dermal absorption (Brown et al., 1984). EPA
(2004) used a prediction model based on octanol-water partitioning and molecular weight to
derive a dermal permeability coefficient for TCE in water of 0.012 cm/hour. EPA used this
value to compute the dermally absorbed dose from a 35 minute shower and compared it to the
dose from drinking 2 L of water at the same concentration. This comparison indicated that the
2-19
dermal dose would be 17% of the oral dose. Much higher dermal permeabilities were reported
by Nakai et al. (1999) based on human skin in vitro testing. For dilute aqueous solutions of
TCE, they measured a permeability coefficient of 0.12 cm/hour (26C). Nakai et al. (1999) also
measured a permeability coefficient of 0.018 cm/hour for tetrachloroethylene in water. Poet
et al. (2000) measured dermal absorption of TCE in humans from both water and soil matrices.
The absorbed dose was estimated by applying a physiologically based pharmacokinetic model to
TCE levels in breath. The permeability coefficient was estimated to be 0.015 cm/hour for TCE
in water and 0.007 cm/hour for TCE in soil (Poet et al., 2000).

2.4.1.4. Exposure to TCE Related Compounds
Table 2-13 presents adult exposure estimates that have been reported for the TCE related
compounds. This table was originally compiled by Wu and Schaum (2001). The exposure/dose
estimates are taken directly from the listed sources or derived based on monitoring data
presented in the source documents. They are considered preliminary because they are
generally based on very limited monitoring data. These preliminary estimates suggest that
exposures to most of the TCE related compounds are comparable to or greater than TCE itself.

Table 2-13. Preliminary intake estimates of TCE and TCE-related chemicals

Chemical Population Media
Range of estimated
adult exposures
(g/d)
Range of adult doses
(mg/kg-d) Data sources
a

Trichloroethylene General Air 1133 1.57 10
-4
4.71 10
-4
ATSDR (1997c)
General Water 220
b
2.86 10
-5
2.86 10
-4
ATSDR (1997c)
Occupational Air 2,2329,489 3.19 10
-2
1.36 10
-1
ATSDR (1997c)
Tetrachloroethylene General Air 80200 1.14 10
-3
2.86 10
-3
ATSDR (1997a)
General Water 0.10.2 1.43 10
-6
2.86 10
-6
ATSDR (1997a)
Occupational Air 5,897219,685 8.43 10
-2
3.14 ATSDR (1997a)
1,1,1-Trichloroethane General Air 10.8108 1.54 10
-4
1.54 10
-3
ATSDR (1995)
General Water 0.384.2 5.5 10
-6
6.0 10
-5
ATSDR (1995)
1,2-Dichloroethylene General Air 16 1.43 10
-5
8.57 10
-5
ATSDR (1996a)
General Water 2.2 3.14 10
-5
ATSDR (1996a)
Cis-1,2-Dichloroethylene General Air 5.4 7.71 10
-5
HSDB (1996)
General Water 0.55.4 7.14 10
-6
7.71 10
-5
HSDB (1996)
1,1,1,2-Tetrachloroethane General Air 142 2.03 10
-3
HSDB (2002)

1,1-Dichloroethane General Air 4 5.71 10
-5
ATSDR (1990)
General Water 2.47469.38 3.53 10
-5
6.71 10
-3
ATSDR (1990)
Chloral General Water 0.0236.4 2.86 10
-7
5.20 10
-4
HSDB (1996)
Monochloroacetic acid General Water 22.4 2.86 10
-5
3.43 10
-5
EPA (1994c)
Dichloroacetic acid General Water 10266 1.43 10
-4
3.80 10
-3
IARC (1995a)
Trichloroacetic acid General Water 8.56322 1.22 10
-3
4.60 10
-3
IARC (1995a)

a
Originally compiled in Wu and Schaum (2001).
b
New data from USGS (2006) suggests much lower water intakes, i.e., 0.2 g/d.

2-20
2.4.2. Potentially Highly Exposed Populations
Some members of the general population may have elevated TCE exposures. ATSDR
(1997c) has reported that TCE exposures may be elevated for people living near waste facilities
where TCE may be released, residents of some urban or industrialized areas, people exposed at
work (discussed further below) and individuals using certain products (also discussed further
below). Because TCE has been detected in breast milk samples of the general population,
infants who ingest breast milk may be exposed, as well. Increased TCE exposure is also a
possible concern for bottle-fed infants because they ingest more water on a bodyweight basis
than adults (the average water ingestion rate for adults is 21 mL/kg-day and for infants under one
year old it is 44 mL/kg-day) (U.S. EPA, 1997). Also, because TCE can be present in soil,
children may be exposed through activities such as playing in or ingesting soil.

2.4.2.1. Occupational Exposure
Occupational exposure to TCE in the United States has been identified in various
degreasing operations, silk screening, taxidermy, and electronics cleaning (IARC, 1995a). The
major use of TCE is for metal cleaning or degreasing (IARC, 1995a). Degreasing is used to
remove oils, greases, waxes, tars, and moisture before galvanizing, electroplating, painting,
anodizing, and coating. The five primary industries using TCE degreasing are furniture and
fixtures; electronic and electric equipment; transport equipment; fabricated metal products; and
miscellaneous manufacturing industries (IARC, 1995a). Additionally, TCE is used in the
manufacture of plastics, appliances, jewelry, plumbing fixtures, automobile, textiles, paper, and
glass (IARC, 1995a).
Table 2-14 lists the primary types of industrial degreasing procedures and the years that
the associated solvents were used. Vapor degreasing has the highest potential for exposure
because vapors can escape into the work place. Hot dip tanks, where TCE is heated to close to
its boiling point of 87C, are also major sources of vapor that can create exposures as high as
vapor degreasers. Cold dip tanks have a lower exposure potential, but they have a large surface
area which enhances volatilization. Small bench-top cleaning operations with a rag or brush and
open bucket have the lowest exposure potential. In combination with the vapor source, the size
and ventilation of the workroom are the main determinants of exposure intensity (NRC, 2006).

2-21
Table 2-14. Years of solvent use in industrial degreasing and cleaning
operations

Years Vapor degreasers Cold dip tanks Rag or brush and bucket on bench top
~19341954 Trichloroethylene
(poorly controlled)
Stoddard solvent
a
Stoddard solvent (general use), alcohols
(electronics shop), carbon tetrachloride
(instrument shop).
~19551968 TCE (poorly controlled,
tightened in 1960s)
TCE (replaced some
Stoddard solvent)
Stoddard solvent, TCE (replaced some
Stoddard solvent), perchloroethylene,
1,1,1-trichloroethane (replaced carbon
tetrachloride, alcohols, ketones).
~19691978 TCE, (better controlled) TCE, Stoddard solvent TCE, perchloroethylene, 1,1,1-trichloro-
ethane, alcohols, ketones, Stoddard solvent.
~19791990s 1,1,1-Trichloroethane
(replaced TCE)
1,1,1-Trichloroethane
(replaced TCE),
Stoddard solvent
1,1,1-Trichloroethane, perchloroethylene,
alcohols, ketones, Stoddard solvent.

a
A mixture of straight and branched chain paraffins (48%), naphthenes (38%), and aromatic hydrocarbons (14%).

Sources: Stewart and Dosemeci (2005); Bakke et al. (2007).

Occupational exposure to TCE has been assessed in a number of epidemiologic and
industrial hygiene studies. Bakke et al. (2007) estimated that the arithmetic mean of TCE
occupational exposures across all industries and decades (mostly 1950s, 1970s, and 1980s) was
38.2 ppm (210 mg/m
3
). They also reported that the highest personal and area air levels were
found in vapor degreasing operations (arithmetic mean of 44.6 ppm or 240 mg/m
3
). Hein et al.
(2010) developed and evaluated statistical models to estimate the intensity of occupational
exposure to TCE (and other solvents) using a database of air measurement data and associated
exposure determinants. The measurement database was compiled from the published literature
and National Institute for Occupational Safety and Health (NIOSH) reports from 1940 to 1998
(n = 484) and were split between personal (47%)and area (53%) measurements. The predicted
arithmetic mean exposure intensity levels for the evaluated exposure scenarios ranged from
0.21 to 3,700 ppm (1.120,000 mg/m
3
) with a median of 30 ppm (160 mg/m
3
). Landrigan et al.
(1987) used air and biomonitoring techniques to quantify the exposure of degreasing workers
who worked around a heated, open bath of TRI. Exposures were found to be between 22 and
66 ppm (117357 mg/m
3
) on average, with short-term peaks between 76 and 370 ppm
(4132,000 mg/m
3
). High peak exposures have also been reported for cardboard workers who
were involved with degreasing using a heated and open process (Henschler et al., 1995).
Lacking industrial hygiene data and making some assumptions about plant environment and TCE
usage, Cherrie et al. (2001) estimated that cardboard workers at a plant in Germany had peak
exposures in the range of 2004,000 ppm (1,10022,000 mg/m
3
) and long-term average
exposures of 10225 ppm (541,200 mg/m
3
). ATSDR (1997c) reports that the majority of
published worker exposure data show time-weighted average concentrations ranging from
<50 ppm100 ppm (<270540 mg/m
3
). NIOSH conducted a survey of various industries from
2-22
1981 to 1983 and estimated that approximately 401,000 U.S. employees in 23,225 plants in the
United States were potentially exposed to TCE during this timeframe (ATSDR, 1997c; IARC,
1995a). Occupational exposure to TCE has likely declined since the 1950s and 1960s due to
decreased usage, better release controls, and improvements in worker protection. Reductions in
TCE use are illustrated in Table 2-14, which shows that by about 1980, common degreasing
operations had substituted other solvents for TCE.

2.4.2.2. Consumer Exposure
Consumer products reported to contain TCE include wood stains, varnishes, and finishes;
lubricants; adhesives; typewriter correction fluids; paint removers; and cleaners (ATSDR,
1997c). Use of TCE has been discontinued in some consumer products (i.e., as an inhalation
anesthetic, fumigant, and an extractant for decaffeinating coffee) (ATSDR, 1997c).

2.4.3. Exposure Standards
Table 2-15 summarizes the federal regulations limiting TCE exposure.

Table 2-15. TCE standards

Standard Value Reference
OSHA Permissible Exposure Limit: Table Z-2 8-hr
time-weighted average.
100 ppm
(538 mg/m
3
)
29 CFR 1910.1000 (7/1/2000)
OSHA Permissible Exposure Limit: Table Z-2
Acceptable ceiling concentration (this cannot be
exceeded for any time period during an 8-hr shift
except as allowed in the maximum peak standard
below).
200 ppm
(1,076 mg/m
3
)
29 CFR 1910.1000 (7/1/2000)
OSHA Permissible Exposure Limit: Table Z-2
Acceptable maximum peak above the acceptable
ceiling concentration for an 8-hr shift. Maximum
Duration: 5 minutes in any 2 hrs.
300 ppm
(1,614 mg/m
3
)
29 CFR 1910.1000 (7/1/2000)
Maximum contaminant level under the Safe
Drinking Water Act.
5 ppb (5 g/L) 40 CFR 141.161
FDA Tolerances for
decaffeinated ground coffee
decaffeinated soluble (instant) coffee
extract spice oleoresins.

25 ppm (25 g/g)
10 ppm (10 g/g)
30 ppm (30 g/g)
21 CFR 173.290 (4/1/2000)

OSHA = Occupational Safety and Health Administration

2.5. EXPOSURE SUMMARY
TCE is a volatile compound with moderate water solubility. Most TCE produced today
is used for metal degreasing. The highest environmental releases are to the air. Ambient air
monitoring data suggests that levels have remained fairly constant since 1999 at about 0.3 g/m
3
.
Indoor levels are commonly three or more times higher than outdoor levels due to releases from
2-23
building materials and consumer products. TCE is among the most common groundwater
contaminants and the median level based on a large survey by USGS for 19852001 is
0.15 g/L. It has also been detected in a wide variety of foods in the 1100 g/kg range. None
of the environmental sampling has been done using statistically based national surveys.
However, a substantial amount of air and groundwater data have been collected allowing
reasonably well supported estimates of typical daily intakes by the general population:
inhalation13 g/day and water ingestion0.2 g/day. The limited food data suggests an
intake of about 5 g/day, but this must be considered preliminary.
Much higher exposures have occurred to various occupational groups. For example, past
studies of aircraft workers have shown short term peak exposures in the hundreds of ppm
(>540,000 g/m
3
) and long term exposures in the low tens of ppm (>54,000 g/m
3
).
Occupational exposures have likely decreased in recent years due to better release controls and
improvements in worker protection.
Preliminary exposure estimates were presented for a variety of TCE related compounds
which include metabolites of TCE and other parent compounds that produce similar metabolites.
Exposure to the TCE related compounds can alter or enhance TCEs metabolism and toxicity by
generating higher internal metabolite concentrations than would result from TCE exposure by
itself. The preliminary estimates suggest that exposures to most of the TCE related compounds
are comparable to or greater than TCE itself.
3-1
3. TOXICOKINETICS


TCE is a lipophilic compound that readily crosses biological membranes. Exposures may
occur via the oral, dermal, and inhalation routes, with evidence for systemic availability from
each route. TCE is rapidly and nearly completely absorbed from the gut following oral
administration, and studies with animals indicate that exposure vehicle may impact the time-
course of absorption: oily vehicles may delay absorption, whereas aqueous vehicles result in a
more rapid increase in blood concentrations.
Following absorption to the systemic circulation, TCE distributes from blood to solid
tissues by each organs solubility. This process is mainly determined by the blood:tissue
partition coefficients, which are largely established by tissue lipid content. Adipose partitioning
is high, adipose tissue may serve as a reservoir for TCE, and accumulation into adipose tissue
may prolong internal exposures. TCE attains high concentrations relative to blood in the brain,
kidney, and liverall of which are important target organs of toxicity. TCE is cleared via
metabolism mainly in three organs: the kidney, liver, and lungs.
The metabolism of TCE is an important determinant of its toxicity. Metabolites are
generally thought to be responsible for toxicityespecially for the liver and kidney. Initially,
TCE may be oxidized via cytochrome P450 (CYP) xenobiotic metabolizing isozymes or
conjugated with glutathione (GSH) by glutathione-S-transferase (GST) enzymes. While
CYP2E1 is generally accepted to be the CYP form most responsible for TCE oxidation at low
concentrations, other forms may also contribute, though their contributions may be more
important at higher, rather than lower, environmentally-relevant exposures.
Once absorbed, TCE is excreted primarily either in breath as unchanged TCE or carbon
dioxide (CO
2
), or in urine as metabolites. Minor routes of elimination include excretion of
metabolites in saliva, sweat, and feces. Following oral administration or upon cessation of
inhalation exposure, exhalation of unmetabolized TCE is a major elimination pathway. Initially,
elimination of TCE upon cessation of inhalation exposure demonstrates a steep concentration-
time profile: TCE is rapidly eliminated in the minutes and hours postexposure, and then the rate
of elimination via exhalation decreases. Following oral or inhalation exposure, urinary
elimination of parent TCE is minimal, with urinary elimination of the metabolites TCA and
TCOH accounting for the bulk of the absorbed dose of TCE.
Sections 3.13.4 below describe the absorption, distribution, metabolism, and excretion
(ADME) of TCE and its metabolites in greater detail. Section 3.5 then discusses PBPK
modeling of TCE and its metabolites.

3-2
3.1. ABSORPTION
TCE is a low-molecular-weight lipophilic solvent; these properties explain its rapid
transfer from environmental media into the systemic circulation after exposure. As discussed
below, it is readily absorbed into the bloodstream following exposure via oral ingestion and
inhalation, with more limited data indicating dermal penetration.

3.1.1. Oral
Available reports on human exposure to TCE via the oral route are largely restricted to
case reports of occupational or intentional (suicidal) ingestions and suggest significant gastric
absorption (e.g., Brning et al., 1998; Yoshida et al., 1996; Perbellini et al., 1991). Clinical
symptoms attributable to TCE or metabolites were observed in these individuals within a few
hours of ingestion (such as lack of consciousness), indicating absorption of TCE. In addition,
TCE and metabolites were measured in blood or urine at the earliest times possible after
ingestion, typically upon hospital admission, while urinary excretion of TCE metabolites was
followed for several days following exposure. Therefore, based on these reports, it is likely that
TCE is readily absorbed in the gastrointestinal (GI) tract; however, the degree of absorption
cannot be confidently quantified because the ingested amounts are not known.
Experimental evidence in mice and rats supports rapid and extensive absorption of TCE,
although variables such as stomach contents, vehicle, and dose may affect the degree of gastric
absorption. DSouza et al. (1985) reported on bioavailability and blood kinetics in fasted and
nonfasted male Sprague-Dawley rats following intragastric administration of TCE at 525 mg/kg
in 50% polyethylene glycol (PEG 400) in water. TCE rapidly appeared in peripheral blood (at
the initial 0.5 minutes sampling) of fasted and nonfasted rats with peak levels being attained
shortly thereafter (610 minutes), suggesting that absorption is not diffusion limited, especially
in fasted animals. The presence of food in the GI tract, however, seems to influence TCE
absorption based on findings in the nonfasted animals of lesser bioavailability (6080 vs. 90% in
fasted rats), smaller peak blood levels (two- to threefold lower than nonfasted animals), and a
somewhat longer terminal half-life (t
1/2
)

(174 vs. 112 minutes in fasted rats).
Studies by Prout et al. (1985) and Dekant et al. (1986b) have shown that up to 98% of
administered radiolabel was found in expired air and urine of rats and mice following gavage
administration of [
14
C]-radiolabeled TCE ([
14
C]-TCE). Prout et al. (1985) and Green and Prout
(1985) compared the degree of absorption, metabolites, and routes of elimination among
two strains each of male rats (Osborne-Mendel and Park Wistar) and male mice (B6C3F
1
and
Swiss-Webster) following a single oral administration of 10, 500, or 1,000 [
14
C]-TCE.
Additional dose groups of Osborne-Mendel male rats and B6C3F
1
male mice also received a
single oral dose of 2,000 mg/kg [
14
C]-TCE. At the lowest dose of 10 mg/kg, there were no major
differences between rats and mice in routes of excretion, with most of the administered
radiolabel (nearly 6070%) being in the urine. At this dose, the expired air from all groups
3-3
contained 14% of unchanged TCE and 914% CO
2
. Fecal elimination of the radiolabel ranged
from 8.3% in Osborne-Mendel rats to 24.1% in Park Wistar rats. However, at doses between
500 and 2,000 mg/kg, the rat progressively excreted a higher proportion of the radiolabel as
unchanged TCE in expired air, such that 78% of the administered high dose was found in expired
air (as unchanged TCE) while only 13% was excreted in the urine.
Following exposure to a chemical by the oral route, distribution is determined by delivery
to the first organ encountered in the circulatory pathwaythe liver (i.e., the first-pass effect),
where metabolism and elimination may limit the proportion that may reach extrahepatic organs.
Lee et al. (1996) evaluated the efficiency and dose-dependency of presystemic elimination of
TCE in male Sprague-Dawley rats following administration into the carotid artery, jugular vein,
hepatic portal vein, or the stomach of TCE (0.17, 0.33, 0.71, 2, 8, 16, or 64 mg/kg) in a
5% aqueous Alkamus emulsion (polyethoxylated vegetable oil) in 0.9% saline. The first-pass
elimination, decreased from 57.5 to <1% with increasing dose (0.1716 mg/kg), which implied
that hepatic TCE metabolism may be saturated at doses >16 mg/kg in the male rat. At doses of
16 mg/kg, hepatic first-pass elimination was almost nonexistent indicating that, at relatively
large doses, virtually all of TCE passes through the liver without being extracted (Lee et al.,
1996). In addition to the hepatic first-pass elimination findings, pulmonary extraction, which
was relatively constant (at nearly 58% of dose) over the dose range, also played a role in
eliminating TCE.
In addition, oral absorption appears to be affected by both dose and vehicle used. The
majority of oral TCE studies have used either aqueous solution or corn oil as the dosing vehicle.
Most studies that relied on an aqueous vehicle delivered TCE as an emulsified suspension in
Tween 80

or PEG 400 in order to circumvent the water solubility problems. Lee et al. (2000a;
2000b) used Alkamus (a polyethoxylated vegetable oil emulsion) to prepare a 5% aqueous
emulsion of TCE that was administered by gavage to male Sprague-Dawley rats. The findings
confirmed rapid TCE absorption, but reported decreasing absorption rate constants (i.e., slower
absorption) with increasing gavage dose (2432 mg/kg). The time to reach blood peak
concentrations increased with dose and ranged between 2 and 26 minutes postdosing. Other
pharmacokinetics data, including area under the blood concentration time curve (AUC) and
prolonged elevation of blood TCE levels at the high doses, indicated prolonged GI absorption
and delayed elimination due to metabolic saturation occurring at the higher TCE doses.
A study by Withey et al. (1983) evaluated the effect of dosing TCE with corn oil vs. pure
water as a vehicle by administering four VOCs separately in each dosing vehicle to male Wistar
rats. Based on its limited solubility in pure water, the dose for TCE was selected at 18 mg/kg
(administered in 5 mL/kg). Times to peak in blood reported for TCE averaged 5.6 minutes when
water was used. In comparison, the time to peak in blood was much longer (approximately
100 minutes) when the oil vehicle was used and the peaks were smaller, below the level of
detection, and not reportable.
3-4
Time-course studies reporting times to peak in blood or other tissues have been
performed using both vehicles (Larson and Bull, 1992a, b; D'Souza et al., 1985; Green and
Prout, 1985; Dekant et al., 1984; Withey et al., 1983). Related data for other solvents (Dix et al.,
1997; Lilly et al., 1994; Kim et al., 1990a; Kim et al., 1990b; Chieco et al., 1981) confirmed
differences in TCE absorption and peak height between the two administered vehicles. One
study has also evaluated the absorption of TCE from soil in rats (Kadry et al., 1991) and reported
absorption within 16 hours for clay and 24 hours for sandy soil. In summary, these studies
confirm that TCE is relatively quickly absorbed from the stomach, and that absorption is
dependent on the vehicle used.

3.1.2. Inhalation
TCE is a lipophilic volatile compound that is readily absorbed from inspired air. Uptake
from inhalation is rapid and the absorbed dose is proportional to exposure concentration and
duration, and pulmonary ventilation rate. Distribution into the body via arterial blood leaving the
lungs is determined by the net dose absorbed and eliminated by metabolism in the lungs.
Metabolic clearance in the lungs will be further discussed in Section 3.3, below. In addition to
metabolism, solubility in blood is the major determinant of the TCE concentration in blood
entering the heart and being distributed to the each body organ via the arterial blood. The
measure of TCE solubility in each organ is the partition coefficient, or the concentration ratio
between both organ phases of interest. The blood-to-air partition coefficient quantifies the
resulting concentration in blood leaving the lungs at equilibrium with alveolar air. The value of
the blood-to-air partition coefficient is used in PBPK modeling (see Section 3.5). The blood-to-
air partition has been measured in vitro using the same principles in different studies and found
to range between 8.1 and 11.7 in humans with somewhat higher values in mice and rats (13.3
25.8) (see Tables 3-13-2, and references therein).

Table 3-1. Blood:air partition coefficient values for humans

Blood:air partition
coefficient Reference/notes
8.1 1.8 Fiserova-Bergerova et al. (1984); mean SD (SD converted from SE based on n = 5)
8.11 Gargas et al. (1989); (n = 315)
9.13 1.73 [6.4711] Fisher et al. (1998); mean SD [range] of females (n = 6)
9.5 Sato and Nakajima (1979); (n = 1)
9.77 Koizumi (1989)
9.92 Sato et al. (1977); (n = 1)
11.15 0.74 [10.112.1] Fisher et al. (1998); mean SD [range] of males (n = 7)
11.2 1.8 [7.915] Mahle et al. (2007); mean SD; 20 male pediatric patients aged 37 yrs (range; USAF,
2004)
11.0 1.6 [6.613.5] Mahle et al. (2007); mean SD; 18 female pediatric patients aged 317 yrs (range;
USAF, 2004)
11.7 1.9 [6.716.8] Mahle et al. (2007); mean SD; 32 male patients aged 2382 yrs (range; USAF, 2004)
10.6 2.3 [314.4] Mahle et al. (2007); mean SD; 27 female patients aged 2382 yrs (range; USAF, 2004)

SE = standard error
3-5
Table 3-2. Blood:air partition coefficient values for rats and mice

Blood:air partition
coefficient Reference/notes
Rat
15 0.5 Fisher et al. (1998); mean SD (SD converted from SE based on n = 3)
17.5 Rodriguez et al. (2007)
20.5 2.4 Barton et al. (1995); mean SD (SD converted from SE based on n = 4)
20.69 3.3 Simmons et al. (2002); mean SD (n = 710)
21.9 Gargas et al. (1989) (n = 315)
25.8 Koizumi (1989) (pooled n = 3)
25.82 1.7 Sato et al. (1977); mean SD (n = 5)
13.3 0.8 [11.615] Mahle et al. (2007); mean SD; 10 PND 10 male rat pups (range; USAF, 2004)
13.4 1.8 [11.817.2] Mahle et al. (2007); mean SD; 10 PND 10 female rat pups (range; USAF, 2004)
17.5 3.6 [11.723.1] Mahle et al. (2007); mean SD; 9 adult male rats (range; USAF, 2004)
21.8 1.9 [16.923.5] Mahle et al. (2007); mean SD; 11 aged male rats (range; USAF, 2004)
Mouse
13.4 Fisher et al. (1991); male
14.3 Fisher et al. (1991); female
15.91 Abbas and Fisher (1997)

PND = postnatal day

TCE enters the human body quickly by inhalation, and, at high concentrations, it may
lead to death (Coopman et al., 2003), narcosis, unconsciousness, and acute kidney damage
(Carrieri et al., 2007). Controlled exposure studies in humans have shown absorption of TCE to
approach a steady state within a few hours after the start of inhalation exposure (Fernandez et al.,
1977; Monster et al., 1976; Vesterberg and Astrand, 1976; Vesterberg et al., 1976). Several
studies have calculated the net dose absorbed by measuring the difference between the inhaled
concentration and the exhaled air concentration. Soucek and Vlachova (1960) reported 5870%
absorption of the amount inhaled for 5-hour exposures of 93158 ppm. Bartonicek (1962)
obtained an average retention value of 58% after 5 hours of exposure to 186 ppm. Monster et al.
(1976) also took into account minute ventilation measured for each exposure, and calculated of
3749% absorption in subjects exposed to 70 and 140 ppm. The impact of exercise, the increase
in workload, and its effect on breathing has also been measured in controlled inhalation
exposures. Astrand and Ovrum (1976) reported 5058% uptake at rest and 2546% uptake
during exercise from exposure to 100 or 200 ppm (540 or 1,080 mg/m
3
, respectively) of TCE for
30 minutes (see Table 3-3). These authors also monitored heart rate and pulmonary ventilation.
In contrast, Jakubowski and Wieczorek (1988) calculated about 40% retention in volunteers
exposed to TCE at 9 ppm (mean inspired concentration of 4849 mg/m
3
) for 2 hours at rest, with
no change in retention during increased workload due to exercise (see Table 3-4).

3-6
Table 3-3. Air and blood concentrations during exposure to TCE in humans

TCE
concentration
(mg/m
3
)
Work
load
(watt)
Exposure
series
a

TCE concentration in
Uptake as %
of amount
available
Amount
taken up
(mg)
Alveolar air
(mg/m
3
)
Arterial
blood
(mg/kg)
Venous
blood
(mg/kg)
540 0 I 124 9 1.1 0.1 0.6 0.1 53 2 79 4
540 0 II 127 11 1.3 0.1 0.5 0.1 52 2 81 7
540 50 I 245 12 2.7 0.2 1.7 0.4 40 2 160 5
540 50 II 218 7 2.8 0.1 1.8 0.3 46 1 179 2
540 50 II 234 12 3.1 0.3 2.2 0.4 39 2 157 2
540 50 II 244 16 3.3 0.3 2.2 0.4 37 2 147 9
1,080 0 I 280 18 2.6 0.0 1.4 0.3 50 2 156 9
1,080 0 III 212 7 2.1 0.2 1.2 0.1 58 2 186 7
1,080 50 I 459 44 6.0 0.2 3.3 0.8 45 2 702 31
1,080 50 III 407 30 5.2 0.5 2.9 0.7 51 3 378 18
1,080 100 III 542 33 7.5 0.7 4.8 1.1 36 3 418 39
1,080 150 III 651 53 9.0 1.0 7.4 1.1 25 5 419 84

a
Series I consisted of 30-minute exposure periods of rest, rest, 50 watts, and 50 watts; Series II consisted of
30-minute exposure periods of rest, 50 watts, 50 watts, 50 watts; and Series III consisted of 30-minute
exposure periods of rest, 50 watts, 100 watts, 150 watts.

Source: Astrand and Ovrum (1976)

Table 3-4. Retention of inhaled TCE vapor in humans

Workload
Inspired concentration
(mg/m
3
)
Pulmonary ventilation
(m
3
/hr) Retention Uptake (mg/hr)
Rest 48 3
a
0.65 0.07 0.40 0.05 12 1.1
25 Watts 49 1.3 1.30 0.14 0.40 0.05 25 2.9
50 Watts 49 1.6 1.53 0.13 0.42 0.06 31 2.8
75 Watts 48 1.9 1.87 0.14 0.41 0.06 37 4.8

a
Mean SD, n = 6 adult males.

Source: Jakubowski and Wieczorek (1988)

Environmental or occupational settings may result from a pattern of repeated exposure to
TCE. Monster et al. (1979a) reported 70-ppm TCE exposures in volunteers for 4 hours for
5 consecutive days, averaging a total uptake of 450 mg per 4 hours of exposure (see Table 3-5).
In dry-cleaning workers, Skender et al. (1991) reported initial blood concentrations of
0.38 mol/L, increasing to 3.4 mol/L 2 days after. Results of these studies support rapid
absorption of TCE via inhalation.

3-7
Table 3-5. Uptake of TCE in volunteers following 4 hour exposure to 70 ppm


Body
weight
(kg)
Minute-volume
(L/min)
Percentage
retained Uptake (mg/d) Uptake (mg/kg-d)
A 80 9.8 0.4 45 0.8 404 23 5.1
B 82 12.0 0.7 44 0.9 485 35 5.9
C 82 10.9 0.8 49 1.2 493 28 6.0
D 67 11.8 0.8 35 2.6 385 38 5.7
E 90 11.0 0.7 46 1.1 481 25 5.3
Mean 5.6 0.4

Source: Monster et al. (1979b).

Direct measurement of retention after inhalation exposure in rodents is more difficult
because exhaled breath concentrations are challenging to obtain. The only available data are
from Dallas et al. (1991), who designed a nose-only exposure system for rats using a facemask
equipped with one-way breathing valves to obtain measurements of TCE in inspired and exhaled
air. In addition, indwelling carotid artery cannulae were surgically implanted to facilitate the
simultaneous collection of blood. After a 1-hour acclimatization period, rats were exposed to
50 or 500 ppm TCE for 2 hours, and the time course of TCE in blood and expired air was
measured during and for 3 hours following exposure. When air concentration data were
analyzed to reveal absorbed dose (minute volume multiplied by the concentration difference
between inspired and exhaled breath), it was demonstrated that the fractional absorption of either
concentration was >90% during the initial 5 minutes of exposure. Fractional absorption then
decreased to 69 and 71% at 50 and 500 ppm during the second hour of exposure. Cumulative
uptake appeared linear with respect to time over the 2-hour exposure, resulting in absorbed doses
of 8.4 and 73.3 mg/kg in rats exposed to 50 and 500 ppm, respectively. Given the 10-fold
difference in inspired concentration and the 8.7-fold difference in uptake, the authors interpreted
this information to indicate that metabolic saturation occurred at some concentration <500 ppm.
In comparing the absorbed doses to those developed for the 70-ppm-exposed human [see
Monster et al. (1979a)], Dallas et al. (1991) concluded that on a systemic dose (mg/kg) basis, rats
receive a much higher TCE dose from a given inhalation exposure than do humans. In
particular, using the results cited above, the absorption per ppm-hour was 0.084 and
0.073 mg/kg-ppm-hour at 50 and 500 ppm in rats (Dallas et al., 1991) and 0.019 mg/kg-ppm-
hour at 70 ppm in humans (Monster et al., 1979a)a difference of around fourfold. However,
rats have about a 10-fold higher alveolar ventilation rate per unit body weight than humans
(Brown et al., 1997), which more than accounts for the observed increase in absorption.
Other experiments, such as closed-chamber gas uptake experiments or blood
concentration measurements following open-chamber (fixed concentration) experiments,
measure absorption indirectly but are consistent with significant retention. Closed-chamber gas-
3-8
uptake methods (Gargas et al., 1988) place laboratory animals or in vitro preparations into sealed
systems in which a known amount of TCE is injected to produce a predetermined chamber
concentration. As the animal retains a quantity of TCE inside its body, due to metabolism, the
closed-chamber concentration decreases with time when compared to the start of exposure.
Many different studies have made use of this technique in both rats and mice to calculate total
TCE metabolism (i.e., Simmons et al., 2002; Fisher et al., 1991; Andersen et al., 1987a). This
inhalation technique is combined with PBPK modeling to calculate metabolic parameters, and
the results of these studies are consistent with rapid absorption of TCE via the respiratory tract.
Figure 3-1 shows an example from Simmons et al. (2002), in Long-Evans rats, that demonstrates
an immediate decline in chamber concentrations of TCE indicating absorption, with multiple
initial concentrations needed for each metabolic calculation. At concentrations below metabolic
saturation, a secondary phase of uptake appears, after 1 hour from starting the exposure,
indicative of metabolism. At concentrations >1,000 ppm, metabolism appears saturated, with
time-course curves having a flat phase after absorption. At intermediate concentrations, between
100 and 1,000 ppm, the secondary phase of uptake appears after distribution as continued
decreases in chamber concentration as metabolism proceeds. Using a combination of
experiments that include both metabolic linear decline and saturation obtained by using different
initial concentrations, both components of metabolism can be estimated from the gas uptake
curves, as shown in Figure 3-1.

10
100
1000
10000
0 1 2 3 4 5 6 7
Time (hr)
T
C
E

c
h
a
m
b
e
r

c
o
n
c
e
n
t
r
a
t
i
o
n

(
p
p
m
)
3000 ppm
1000 ppm
500 ppm
100 ppm

Symbols represent measured chamber concentrations. Source: Simmons et al.
(2002).

Figure 3-1. Gas uptake data from closed-chamber exposure of rats to TCE.

3-9
Several other studies in humans and rodents have measured blood concentrations of TCE
or metabolites and urinary excretion of metabolites during and after inhalation exposure (e.g.,
Fisher et al., 1998; 1991; 1990; Filser and Bolt, 1979). While qualitatively indicative of
absorption, blood concentrations are also determined by metabolism, distribution, and excretion;
thus, comparisons between species may reflect similarities or differences in any of the
absorption, distribution, metabolism, and excretion processes.

3.1.3. Dermal
Skin membrane is believed to present a diffusional barrier for entrance of the chemical
into the body, and TCE absorption can be quantified using a permeability rate or permeability
constant, though not all studies performed such a calculation. Absorption through the skin has
been shown to be rapid by both vapor and liquid TCE contact with the skin. Human dermal
absorption of TCE vapors was investigated by Kezic et al. (2000). Volunteers were exposed to
3.18 10
4
ppm around each enclosed arm for 20 minutes. Adsorption was found to be rapid
(within 5 minutes), reaching a peak in exhaled breath around 30 minutes, with a calculated
dermal penetration rate averaging 0.049 cm/hour for TCE vapors.
With respect to dermal penetration of liquid TCE, Nakai et al. (1999) used surgically
removed skin samples exposed to TCE in aqueous solution in a chamber designed to measure the
difference between incoming and outgoing [
14
C]-TCE. The in vitro permeability constant
calculated by these researchers averaged 0.12 cm/hour. In vivo, Sato and Nakajima (1978)
exposed adult male volunteers dermally to liquid TCE for 30 minutes, with exhaled TCE
appearing at the initial sampling time of 5 minutes after start of exposure, with a maximum
observed at 15 minutes. In Kezic et al. (2001), volunteers were exposed dermally for 3 minutes
to neat liquid TCE, with TCE detected in exhaled breath at the first sampling point of 3 minutes,
and maximal concentrations observed at 5 minutes. Skin irritancy was reported in all subjects,
which may have increased absorption. A dermal flux of 430 295 (mean standard error [SE])
nmol/cm
2
/minute was reported in these subjects, suggesting high interindividual variability.
Another species where dermal absorption for TCE has been reported is in guinea pigs.
Jakobson et al. (1982) applied liquid TCE to the shaved backs of guinea pigs and reported peak
blood TCE levels at 20 minutes after initiation of exposure. Bogen et al. (1992) estimated
permeability constants for dermal absorption of TCE in hairless guinea pigs of 0.16
0.47 mL/cm
2
/hour across a range of concentrations (19100,000 ppm).

3.2. DISTRIBUTION AND BODY BURDEN
TCE crosses biological membranes and quickly results in rapid systemic distribution to
tissuesregardless of the route of exposure. In humans, in vivo studies of tissue distribution are
limited to tissues taken from autopsies following accidental poisonings or from surgical patients
exposed environmentally, so the level of exposure is typically unknown. Tissue levels reported
3-10
after autopsy show wide systemic distribution across all tested tissues, including the brain,
muscle, heart, kidney, lung, and liver (Coopman et al., 2003; Dehon et al., 2000; De Baere et al.,
1997; Ford et al., 1995). However, the reported levels themselves are difficult to interpret
because of the high exposures and differences in sampling protocols. In addition, human
populations exposed environmentally show detectable levels of TCE across different tissues,
including the liver, brain, kidney, and adipose tissues (Kroneld, 1989; Pellizzari et al., 1982;
McConnell et al., 1975).
In addition, TCE vapors have been shown to cross the human placenta during childbirth
(Laham, 1970), with experiments in rats confirming this finding (Withey and Karpinski, 1985).
In particular, Laham (1970) reported determinations of TCE concentrations in maternal and fetal
blood following administration of TCE vapors (concentration unreported) intermittently and at
birth (see Table 3-6). TCE was present in all samples of fetal blood, with ratios of
concentrations in fetal:maternal blood ranging from approximately 0.5 to approximately 2. The
concentration ratio was <1.0 in six pairs, >1 in three pairs, and approximately 1 in one pair; in
general, higher ratios were observed at maternal concentrations <2.25 mg/100 mL. Because no
details of exposure concentration, duration, or time postexposure were given for samples taken,
these results are not suitable for use in PBPK modeling, but they do demonstrate the placental
transfer of TCE in humans. Withey and Karpinski (1985) exposed pregnant rats to TCE vapors
(302, 1,040, 1,559, or 2,088 ppm for 5 hours) on gestation day (GD) 17 and concentrations of
TCE in maternal and fetal blood were determined. At all concentrations, TCE concentration in
fetal blood was approximately one-third of the concentration in corresponding maternal blood.
Maternal blood concentrations approximated 15, 60, 80, and 110 g/g blood. When the position
along the uterine horn was examined, TCE concentrations in fetal blood decreased toward the tip
of the uterine horn. TCE appears to also distribute to mammary tissues and is excreted in milk.
Pellizzari et al. (1982) conducted a survey of environmental contaminants in human milk using
samples from cities in the northeastern region of the United States and one in the southern
region. No details of times postpartum, milk lipid content, or TCE concentration in milk or
blood were reported, but TCE was detected in 8 milk samples taken from 42 lactating women.
Fisher et al. (1990) exposed lactating rats to 600 ppm TCE for 4 hours and collected milk
immediately following the cessation of exposure. TCE was clearly detectable in milk, and, from
a visual interpretation of the graphic display of their results, concentrations of TCE in milk
approximated 110 g/mL milk.

3-11
Table 3-6. Concentrations of TCE in maternal and fetal blood at birth

TCE concentration in blood (mg/100 mL)
Ratio of concentrations fetal:maternal Maternal Fetal
4.6 2.4 0.52
3.8 2.2 0.58
8 5 0.63
5.4 3.6 0.67
7.6 5.2 0.68
3.8 3.3 0.87
2 1.9 0.95
2.25 3 1.33
0.67 1 1.49
1.05 2 1.90

Source: Laham (1970).

In rodents, detailed tissue distribution experiments have been performed using different
routes of administration (Keys et al., 2003; Simmons et al., 2002; Greenberg et al., 1999; Abbas
and Fisher, 1997; Pfaffenberger et al., 1980; Savolainen et al., 1977). Savolainen et al. (1977)
exposed adult male rats to 200 ppm TCE for 6 hours/day for a total of 5 days. Concentrations of
TCE in the blood, brain, liver, lung, and perirenal fat were measured 17 hours after cessation of
exposure on the fourth day and after 2, 3, 4, and 6 hours of exposure on the fifth day (see
Table 3-7). TCE appeared to be rapidly absorbed into blood and distributed to brain, liver, lungs,
and perirenal fat. TCE concentrations in these tissues reached near-maximal values within
2 hours of initiation of exposure on the fifth day. Pfaffenberger et al. (1980) dosed rats by
gavage with 1 or 10 mg TCE/kg/day in corn oil for 25 days to evaluate the distribution from
serum to adipose tissue. During the exposure period, concentrations of TCE in serum were
below the limit of detection (1 g/L) and were 280 and 20,000 ng/g fat in the 1 and 10 mg/day
dose groups, respectively. Abbas and Fisher (1997) and Greenberg et al. (1999) measured tissue
concentrations in the liver, lung, kidney, and fat of mice administered TCE by gavage (300
2,000 mg/kg) and by inhalation exposure (100 or 600 ppm for 4 hours). In a study to investigate
the effects of TCE on neurological function, Simmons et al. (2002) conducted pharmacokinetic
experiments in rats exposed to 200, 2,000, or 4,000 ppm TCE vapors for 1 hour. Time-course
data were collected on blood, liver, brain, and fat. The data were used to develop a PBPK model
to explore the relationship between internal dose and neurological effect. Keys et al. (2003),
exposed groups of rats to TCE vapors of 50 or 500 ppm for 2 hours and sacrificed at different
time points during exposure. In addition to inhalation, this study also includes gavage and intra-
arterial (i.a.) dosing, with the following time course measured: liver, fat, muscle, blood, GI,
brain, kidney, heart, lung, and spleen. These pharmacokinetic data were presented with an
updated PBPK model for all routes.
3-12

Table 3-7. Distribution of TCE to rat tissues
a
following inhalation exposure

Exposure on
5
th
d
Tissue (concentration in nmol/g tissue)
Cerebrum Cerebellum Lung Liver Perirenal fat Blood
0
b
0 0 0.08 0.04 0.23 0.09 0.35 0.1
2 9.9 2.7 11.7 4.2 4.9 0.3 3.6 65.9 1.2 7.5 1.6
3 7.3 2.2 8.8 2.1 5.5 1.4 5.5 1.7 69.3 3.3 6.6 0.9
4 7.2 1.7 7.6 0.5 5.8 1.1 2.5 1.4 69.5 6.3 6.0 0.2
6 7.4 2.1 9.5 2.5 5.6 0.5 2.4 0.2 75.4 14.9 6.8 1.2

a
Data presented as mean of two determinations range.
b
Sample taken 17 hours following cessation of exposure on day 4.

Source: Savolainen et al. (1977).

Besides the route of administration, another important factor contributing to body
distribution is the individual solubility of the chemical in each organ, as measured by a partition
coefficient. For volatile compounds, partition coefficients are measured in vitro using the vial
equilibration technique to determine the ratio of concentrations between organ and air at
equilibrium. Table 3-8 reports values developed by several investigators from mouse, rat, and
human tissues. In humans, partition coefficients in the following tissues have been measured:
brain, fat, kidney, liver, lung, and muscle; the organ having the highest TCE partition coefficient
is fat (6370), while the lowest is the lung (0.51.7). The adipose tissue also has the highest
measured value in rodents, and is one of the considerations needed to be accounted for when
extrapolating across species. However, the rat adipose partition coefficient value is smaller (23
36), when compared to humans (i.e., TCE is less lipophilic in rats than humans). For the mouse,
the measured fat partition coefficient averages 36, ranging between rats and humans. The value
of the partition coefficient plays a role in distribution for each organ and is computationally
described in computer simulations using a PBPK model. Due to its high lipophilicity in fat, as
compared to blood, the adipose tissue behaves as a storage compartment for this chemical,
affecting the slower component of the chemicals distribution. For example Monster et al.
(1979a) reported that, following repeated inhalation exposures to TCE, TCE concentrations in
expired breath postexposure were highest for the subject with the greatest amount of adipose
tissue (adipose tissue mass ranged 3.5-fold among subjects). The intersubject range in TCE
concentration in exhaled breath increased from approximately 2-fold at 20 hours to
approximately 10-fold 140 hours postexposure. Notably, they reported that this difference was
not due to differences in uptake, as body weight and lean body mass were most closely
associated with TCE retention. Thus, adipose tissue may play an important role in postexposure
distribution, but does not affect its rapid absorption.

3-13
Table 3-8. Tissue:blood partition coefficient values for TCE

Species/
tissue
TCE partition coefficient
References Tissue:blood Tissue:air
Human
Brain 2.62 21.2 Fiserova-Bergerova et al. (1984)
Fat 63.870.2 583674.4 Sato et al. (1977); Fiserova-Bergerova et al. (1984); Fisher et al.
(1998)
Kidney 1.31.8 1214.7 Fiserova-Bergerova et al. (1984); Fisher et al. (1998)
Liver 3.65.9 29.454 Fiserova-Bergerova et al. (1984); Fisher et al. (1998)
Lung 0.481.7 4.413.6 Fiserova-Bergerova et al. (1984); Fisher et al. (1998)
Muscle 1.72.4 15.319.2 Fiserova-Bergerova et al. (1984); Fisher et al. (1998)
Rat
Brain 0.711.29 14.633.3 Sato et al. (1977); Simmons et al. (2002); Rodriguez et al. (2007)
Fat 22.736.1 447661 Gargas et al. (1989); Sato et al. (1977); Simmons et al. (2002);
Rodriguez et al. (2007); Fisher et al. (1989); Koizumi (1989); Barton
et al. (1995)
Heart 1.1 28.4 Sato et al. (1977)
Kidney 1.01.55 17.740 Sato et al., (1977); Barton et al., (1995); Rodriguez et al., (2007)
Liver 1.032.43 20.562.7 Gargas et al. (1989); Sato et al. (1977); Simmons et al. (2002);
Rodriguez et al. (2007); Fisher et al. (1989); Koizumi, (1989); Barton
et al. (1995)
Lung 1.03 26.6 Sato et al. (1977)
Muscle 0.460.84 6.921.6 Gargas et al. (1989); Sato et al. (1977); Simmons et al. (2002);
Rodriguez et al. (2007); Fisher et al. (1989); Koizumi, (1989); Barton
et al. (1995)
Spleen 1.15 29.7 Sato et al. (1977)
Testis 0.71 18.3 Sato et al. (1977)
Milk 7.10 Not reported Fisher et al. (1990)
Mouse
Fat 36.4 578.8 Abbas and Fisher (1997)
Kidney 2.1 32.9 Abbas and Fisher (1997)
Liver 1.62 23.2 Fisher et al. (1991)
Lung 2.6 41.5 Abbas and Fisher (1997)
Muscle 2.36 37.5 Abbas and Fisher (1997)

Mahle et al. (2007) reported age-dependent differences in partition coefficients in rats,
(see Table 3-9) that can have implications as to life-stage-dependent differences in tissue TCE
distribution. To investigate the potential impact of these differences, Rodriguez et al. (2007)
developed models for the postnatal day (PND) 10 rat pup; the adult and the aged rat, including
age-specific tissue volumes and blood flows; and age-scaled metabolic constants. The models
predict similar uptake profiles for the adult and the aged rat during a 6-hour exposure to
500 ppm; uptake by the PND 10 rat was higher (see Table 3-10). The effect was heavily
dependent on age-dependent changes in anatomical and physiological parameters (alveolar
3-14
ventilation rates and metabolic rates); age-dependent differences in partition coefficient values
had minimal impact on predicted differences in uptake.

Table 3-9. Age-dependence of tissue:air partition coefficients in rats

Age
a
Liver Kidney Fat Muscle Brain
PND 10 male 22.1 2.3
b
15.2 1.3 398.7 89.2 43.9 11.0 11.0 0.6
PND 10 female 21.2 1.7 15.0 1.1 424.5 67.5 48.6 17.3 11.6 1.2
Adult male 20.5 4.0 17.6 3.9
c
631.4 43.1
c
12.6 4.3 17.4 2.6
Aged male 34.8 8.7
c,d
19.9 3.4
c
757.5 48.3
c,d
26.4 10.3
c,d
25.0 2.0
c,d


a
n = 10, adult male and pooled male and female litters; n = 11, aged males.
b
Data are mean SD.
c
Statistically significant (p 0.05) difference between either the adult or aged partition coefficient and the PND 10
male partition coefficient.
d
Statistically significant (p 0.05) difference between aged and adult partition coefficient.

Source: Mahle et al. (2007).

Table 3-10. Predicted maximal concentrations of TCE in rat blood following
a 6-hour inhalation exposure

Age
Exposure concentration
50 ppm 500 ppm
Predicted peak
concentration (mg/L) in:
a

Predicted time to
reach 90% of
steady state (hr)
b

Predicted peak
concentration (mg/L) in:
a

Predicted time to
reach 90% of
steady state (hr)
b

Venous
blood Brain
Venous
blood Brain
PND 10 3.0 2.6 4.1 33 28 4.2
Adult 0.8 1.0 3.5 22 23 11.9
Aged 0.8 1.2 6.7 21 26 23.3

a
During a 6-hour exposure.
b
Under continuous exposure.

Source: Rodriguez et al. (2007).

Finally, TCE binding to tissues or cellular components within tissues can affect overall
pharmacokinetics. The binding of a chemical to plasma proteins, for example, affects the
availability of the chemical to other organs and the calculation of the total half-life. However,
most studies have evaluated binding using [
14
C]-TCE, from which one cannot distinguish
covalent binding of TCE from that of TCE metabolites. Nonetheless, several studies have
demonstrated binding of TCE-derived radiolabel to cellular components (Mazzullo et al., 1992;
Moslen et al., 1977). Bolt and Filser (1977) examined the total amount irreversibly bound to
tissues following 9-, 100-, and 1,000-ppm exposures via inhalation in closed-chambers. The
largest percent of in vivo radioactivity taken up occurred in the liver; albumin is the protein
3-15
favored for binding (see Table 3-11). Banerjee and van Duuren (1978) evaluated the in vitro
binding of TCE to microsomal proteins from the liver, lung, kidney, and stomachs in rats and
mice. In both rats and mice, radioactivity was similar in stomach and lung, but about 30% lower
in kidney and liver.

Table 3-11. Tissue distribution of TCE metabolites following inhalation exposure

Tissue
a

Percent of radioactivity taken up/g tissue
TCE = 9 ppm,
n = 4
b

TCE = 100 ppm,
n = 4
TCE = 1,000 ppm,
n = 3
Total
metabolites
Irreversibly
bound
Total
metabolites
Irreversibly
bound
Total
metabolites
Irreversibly
bound
Lung 0.23 0.026
c
0.06 0.002 0.24 0.025 0.06 0.006 0.22 0.055 0.1 0.003
Liver 0.77 0.059 0.28 0.027 0.68 0.073 0.27 0.019 0.88 0.046 0.48 0.020
Spleen 0.14 0.015 0.05 0.002 0.15 0.001 0.05 0.004 0.15 0.006 0.08 0.003
Kidney 0.37 0.005 0.09 0.007 0.40 0.029 0.09 0.007 0.39 0.045 0.14 0.016
Small
intestine
0.41 0.058 0.05 0.010 0.38 0.062 0.07 0.008 0.28 0.015 0.09 0.015
Muscle 0.11 0.005 0.014 0.001 0.11 0.013 0.012 0.001 0.10 0.011 0.027 0.003

a
Male Wistar rats, 250 g.
b
n = number of animals.
c
Values shown are means SD.

Source: Bolt and Filser (1977).

Based on studies of the effects of metabolizing enzyme induction on binding, there is
some evidence that a major contributor to the observed binding is from TCE metabolites rather
than from TCE itself. Dekant et al. (1986b) studied the effect of enzyme modulation on the
binding of radiolabel from [
14
C]-TCE by comparing tissue binding after administration of
200 mg/kg via gavage in corn oil between control (nave) rats and rats pretreated with
phenobarbital (a known inducer of CYP2B family) or Aroclor 1254 (a known inducer of both
CYP1A and CYP2B families of isoenzymes) (see Table 3-12). The results indicate that
induction of total CYP content by 34-fold resulted in nearly 10-fold increase in radioactivity
(disintegrations per minute; [DPM]) bound in liver and kidney. By contrast, Mazzullo et al.
(1992) reported that phenobarbital pretreatment did not result in consistent or marked alterations
of in vivo binding of radiolabel to deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or
protein in rats and mice at 22 hours after an intraperitoneal (i.p.) injection of [
14
C]-TCE. On the
other hand, in vitro experiments by Mazzullo et al. (1992) reported reduction of TCE-radiolabel
binding to calf thymus DNA with introduction of a CYP inhibitor into incubations containing rat
liver microsomal protein. Moreover, increase/decrease of GSH levels in incubations containing
lung cytosolic protein led to a parallel increase/decrease in TCE-radiolabel binding to calf
thymus DNA.
3-16

Table 3-12. Binding of [
14
C] from [
14
C]-TCE in rat liver and kidney at
72 hours after oral administration of 200 mg/kg [
14
C]-TCE

Tissue
DPM/g tissue
Untreated Phenobarbital Arochlor 1254
Liver 850 100 9,300 1,100 8,700 1,000
Kidney 680 100 5,700 900 7,300 800

Source: Dekant et al. (1986b).

3.3. METABOLISM
This section focuses on both in vivo and in vitro studies of the biotransformation of TCE,
identifying metabolites that are deemed significant for assessing toxicity and carcinogenicity. In
addition, metabolism studies may be used to evaluate the flux of parent compound through the
known metabolic pathways. Sex-, species-, and interindividual differences in the metabolism of
TCE are discussed, as are factors that possibly contribute to this variability. Additional
discussion of variability and susceptibility is presented in Section 4.10.

3.3.1. Introduction
The metabolism of TCE has been studied mostly in mice, rats, and humans and has been
extensively reviewed (Lash et al., 2000a; Lash et al., 2000b; IARC, 1995b; US EPA, 1985). It is
now well accepted that TCE is metabolized in laboratory animals and in humans through at least
two distinct pathways: (1) oxidative metabolism via the CYP mixed-function oxidase system
and (2) GSH conjugation followed by subsequent further biotransformation and processing,
either through the cysteine conjugate beta lyase pathway or by other enzymes (Lash et al., 2000a;
Lash et al., 2000b). While the flux through the conjugative pathway is less, quantitatively, than
the flux through oxidation (Bloemen et al., 2001), GSH conjugation is an important route
toxicologically, giving rise to relatively potent toxic biotransformation products (Elfarra et al.,
1987; Elfarra et al., 1986).
Information about metabolism is important because, as discussed extensively in
Chapter 4, certain metabolites are thought to cause one or more of the same acute and chronic
toxic effects, including carcinogenicity, as TCE. Thus, in many of these cases, the toxicity of
TCE is generally considered to reside primarily in its metabolites rather than in the parent
compound itself.

3.3.2. Extent of Metabolism
TCE is extensively metabolized in animals and humans. The most comprehensive mass-
balance studies are in mice and rats (Dekant et al., 1986a; Dekant et al., 1986b; Green and Prout,
1985; Prout et al., 1985; Dekant et al., 1984) in which [
14
C]-TCE is administered by gavage at
3-17
doses of 22,000 mg/kg, the data from which are summarized in Figures 3-2 and 3-3. In both
mice and rats, regardless of sex and strain, there is a general trend of increasing exhalation of
unchanged TCE with dose, suggesting saturation of a metabolic pathway. The increase is
smaller in mice (from 16 to 1018%) than in rats (from 13 to 4378%), suggesting greater
overall metabolic capacity in mice. The dose at which apparent saturation occurs appears to be
more sex- or strain-dependent in mice than in rats. In particular, the marked increase in exhaled
TCE occurred between 20 and 200 mg/kg in female NMRI mice, between 500 and 1,000 mg/kg
in B6C3F
1
mice, and between 10 and 500 mg/kg in male Swiss-Webster mice. However,
because only one study is available in each strain, interlot or interindividual variability might
also contribute to the observed differences. In rats, all three strains tested showed marked
increase in unchanged TCE exhaled between 20 and 200 mg/kg or between 10 and 500 mg/kg.
Recovered urine, the other major source of excretion, had mainly TCA, TCOH, and
trichloroethanol-glucuronide conjugate (TCOG), but revealed no detectable TCE. The source of
radioactivity in feces was not analyzed, but it is presumed not to include substantial TCE given
the complete absorption expected from the corn oil vehicle. Therefore, at all doses tested in
mice, and at doses <200 mg/kg in rats, the majority of orally administered TCE is metabolized.
Pretreatment of rats with P450 inducers prior to a 200 mg/kg dose did not change the pattern of
recovery, but it did increase the amount recovered in urine by 1015%, with a corresponding
decrease in the amount of exhaled unchanged TCE (Dekant et al., 1986b).

3-18


0
10
20
30
40
50
60
70
80
90
100
2 mg/kg 20
mg/kg
200
mg/kg
10
mg/kg
500
mg/kg
1000
mg/kg
2000
mg/kg
10
mg/kg
500
mg/kg
1000
mg/kg
Mouse Sex/Strain and Dose
%

r
a
d
i
o
a
c
t
i
v
i
t
y

r
e
c
o
v
e
r
e
d
Cage
wash
Carcass
CO2
Exhaled
Feces
Urine
TCE
Exhaled
F/NMRI M/B6C3F1 M/Swiss-Webster

Sources: Dekant et al. (1986b; 1984); Green and Prout (1985); Prout et al. (1985).

Figure 3-2. Disposition of [
14
C]-TCE administered by gavage in mice.
3-19




0
10
20
30
40
50
60
70
80
90
100
2 mg/kg 20
mg/kg
200
mg/kg
10
mg/kg
500
mg/kg
1000
mg/kg
2000
mg/kg
10
mg/kg
500
mg/kg
1000
mg/kg
Rat Sex/Strain and Dose
%

r
a
d
i
o
a
c
t
i
v
i
t
y

r
e
c
o
v
e
r
e
d
Cage
wash
Carcass
CO2
Exhaled
Feces
Urine
TCE
Exhaled
F/Wistar M/Osborne-Mendel M/Alderley-Park Wistar
Sources: Dekant et al. (1986b; 1984); Green and Prout (1985); Prout et al. (1985).

Figure 3-3. Disposition of [
14
C]-TCE administered by gavage in rats.
3-20
The differences among these studies may reflect a combination of interindividual
variability and errors due to the difficulty in precisely estimating dose in inhalation studies, but
in all cases, <20% of the retained dose was exhaled unchanged and >50% was excreted in urine
as TCA and TCOH. Therefore, it is clear that TCE is extensively metabolized in humans. No
saturation was evident in any of these human recovery studies at the exposure levels tested.

3.3.3. Pathways of Metabolism
As mentioned in Section 3.3.1, TCE metabolism in animals and humans has been
observed to occur via two major pathways: P450-mediated oxidation and GSH conjugation.
Products of the initial oxidation or conjugation step are further metabolized to a number of other
metabolites. For P450 oxidation, all steps of metabolism occur primarily in the liver, although
limited oxidation of TCE has been observed in the lungs of mice, as discussed below. The GSH
conjugation pathway also begins predominantly in the liver, but toxicologically significant
metabolic steps occur extrahepaticallyparticularly in the kidney (Lash et al., 2006; Lash et al.,
1999a; Lash et al., 1998b; Lash et al., 1995). The mass-balance studies cited above found that at
exposures below the onset of saturation, >50% of TCE intake is excreted in urine as oxidative
metabolites (primarily as TCA and TCOH), so TCE oxidation is generally greater than TCE
conjugation. This is discussed in detail in Section 3.3.3.3.

3.3.3.1. CYP-Dependent Oxidation
Oxidative metabolism by the CYP, or CYP-dependent, pathway is quantitatively the
major route of TCE biotransformation (Lash et al., 2000a; Lash et al., 2000b; US EPA, 1985).
The pathway is operative in humans and rodents and leads to several metabolic products, some
of which are known to cause toxicity and carcinogenicity (IARC, 1995c; US EPA, 1985).
Although several of the metabolites in this pathway have been clearly identified, others are
speculative or questionable. Figure 3-4 depicts the overall scheme of TCE P450 metabolism.

3-21


Cl Cl
H Cl (TCE)
Cl Cl
H Cl (TCE)
OH
Cl
3
C
OH
(CH)
OH
Cl
3
C
OH
(CH)
Cl
3
C
OH
(TCOH)
Cl
3
C
OH
Cl
3
C
OH
(TCOH)
O
Cl
3
C
H
O
Cl
3
C
H
(CHL)
Cl
3
C
O-gluc
(TCOG)
Cl
3
C
O-gluc
Cl
3
C
O-gluc
(TCOG)
P450
(TCA)
O
Cl
3
C
OH
(TCA)
O
Cl
3
C
OH
O
Cl
3
C
OH
Cl O
OH (OA) HO
O
OHCH
2
N
H
(CH
2
)
2
OH
O
OHCH
2
N
H
(CH
2
)
2
OH
(DCA)
O
Cl
2
CH
OH (DCA)
O
Cl
2
CH
OH
O
Cl
2
CH
OH
O
Cl CH
2
OH
(MCA)
O
Cl CH
2
OH
O
Cl CH
2
OH
(MCA)
O
Cl
2
CH
Cl
(DCAC)
O
Cl
2
CH
Cl
(DCAC)
O
Cl
2
CH
Cl
O
Cl
2
CH
Cl
(DCAC)
P450
GST-
zeta
ALDH
ADH or
P450
UGT
EHR
O
Cl Cl
H Cl
(TCE-O)
O
Cl Cl
H Cl
O
Cl Cl
H Cl
(TCE-O)
TCE-O-P450
(N-(Hydroxyacetyl)-
aminoethanol)
O O
OH
(Glyoxylic
acid)
H
CO
2

Adapted from: Clewell et al. (2000); Cummings et al. (2001); Forkert et al.
(2006); Lash et al. (2000a; 2000b); Tong et al. (1998).

Figure 3-4. Scheme for the oxidative metabolism of TCE.
In brief, TCE oxidation via P450, primarily CYP2E1 (Guengerich and Shimada, 1991),
yields an oxygenated TCE-P450 intermediate. The TCE-P450 complex is a transition state that
goes on to form chloral or TCE oxide. In the presence of water, chloral rapidly equilibrates with
chloral hydrate (CH), which undergoes reduction and oxidation by alcohol dehydrogenase and
aldehyde dehydrogenase or aldehyde oxidase to form TCOH and TCA, respectively (Dekant et
al., 1986b; Green and Prout, 1985; Miller and Guengerich, 1983). TCE oxide can rearrange to
DCAC. Table 3-13 summarizes available in vitro measurements of TCE oxidation, as assessed
by the formation of CH, TCOH, and TCA. Glucuronidation of TCOH forms TCOG, which is
readily excreted in urine. Alternatively, TCOG can be excreted in bile and passed to the small
intestine where it is hydrolyzed back to TCOH and reabsorbed (Bull, 2000). TCA is poorly
metabolized but may undergo dechlorination to form dichloroacetic acid (DCA). However, TCA
is predominantly excreted in urine, albeit at a relatively slow rate as compared to TCOG. Like
3-22
the TCE-P450 complex, TCE oxide also seems to be a transient metabolite. Recent data suggest
that it is transformed to dichloroactyl chloride, which subsequently decomposes to form DCA
(Cai and Guengerich, 1999). As shown in Figure 3-4, several other metabolites, including oxalic
acid and N-(hydroxyacetyl) aminoethanol, may form from the TCE oxide or the TCE-O-P450
intermediate and have been detected in the urine of rodents and humans following TCE
exposure. Pulmonary excretion of CO
2
has been identified in exhaled breath from rodents
exposed to [
14
C]-labeled TCE and is thought to arise from metabolism of DCA. The following
sections provide details as to pathways of TCE oxidation, including discussion of inter- and
intraspecies differences in metabolism.

Table 3-13. In vitro TCE oxidative metabolism in hepatocytes and
microsomal fractions

In vitro system
K
M
V
MAX

1,000
V
MAX
/K
M
a
Source M in medium
nmol TCE
oxidized/min/mg MSP
or 10
6
hepatocytes
Human
hepatocytes
210 159
b

(45403)
0.268 0.215
(0.1010.691)
2.45 2.28
(0.465.57)
Lipscomb et al. (1998b)
Human liver
microsomal
protein
16.7 2.45
(13.319.7)
1.246 0.805
(0.4903.309)
74.1 44.1
(38.9176)
Lipscomb et al. (1997) (low K
M
)
30.9 3.3
(27.036.3)
1.442 0.464
(0.8902.353)
47.0 16.0
(30.181.4)
Lipscomb et al. (1997) (mid K
M
)
51.1 3.77
(46.755.7)
2.773 0.577
(2.0783.455)
54.9 14.1
(37.369.1)
Lipscomb et al. (1997) (high K
M
)
24.6 1.44 58.5 Lipscomb et al. (1998c) (pooled)
12 3
(914)
0.52 0.17
(0.370.79)
48 23
(2679)
Elfarra et al. (1998) (males, high affinity)
26 17
(1345)
0.33 0.15
(0.190.48)
15 10 (1129) Elfarra et al. (1998) (females, high affinity)
Rat liver
microsomal
protein
55.5 4.826 87.0 Lipscomb et al. (1998c) (pooled)
72 82 0.96 0.65 24 21 Elfarra et al. (1998) (males, high affinity)
42 21 2.91 0.71 80 34 Elfarra et al. (1998) (females, high affinity)
Rat kidney
microsomal
protein
940 0.154 0.164 Cummings et al. (2001)
Mouse liver
microsomal
protein
35.4 5.425 153 Lipscomb et al. (1998c) (pooled)
378 414 8.6 4.5 42 29 Elfarra et al. (1998) (males)
161 29 26.06 7.29 163 37 Elfarra et al. (1998) (females)

a
K
M
for human hepatocytes converted from ppm in headspace to M in medium using reported hepatocyte:air partition
coefficient (Lipscomb et al., 1998b).
b
Results presented as mean SD (minimummaximum).

MSP = Microsomal protein.

3.3.3.1.1. Formation of TCE oxide
In previous studies of halogenated alkene metabolism, the initial step was the generation
of a reactive epoxide (Anders and Jakobson, 1985). Early studies in anesthetized human patients
3-23
(Powell, 1945), dogs (Butler, 1949), and later reviews (e.g., Goeptar et al., 1995) suggest that the
TCE epoxide may be the initial reaction product of TCE oxidation.
Epoxides can form acyl chlorides or aldehydes, which can then form aldehydes,
carboxylic acids, or alcohols, respectively. Thus, earlier studies suggesting the appearance of
CH, TCA, and TCOH as the primary metabolites of TCE were considered consistent with the
oxidation of TCE to an epoxide intermediate (Butler, 1949; Powell, 1945). Following in vivo
exposures to 1,1-DCE, a halocarbon very similar in structure to TCE, mouse liver cytosol and
microsomes and lung Clara cells exhibited extensive P450-mediated epoxide formation (Forkert,
1999b; Forkert, 1999a; Forkert et al., 1999; Dowsley et al., 1996). Indeed, TCE oxide inhibits
purified CYP2E1 activity (Cai and Guengerich, 2001b) similarly to TCE inhibition of CYP2E1
in human liver microsomes (Lipscomb et al., 1997).
Conversely, cases have been made against TCE oxide as an obligate intermediate to the
formation of chloral. Using liver microsomes and reconstituted P450 systems (Miller and
Guengerich, 1983, 1982) or isolated rat hepatocytes (Miller and Guengerich, 1983), it has been
suggested that chlorine migration and generation of a TCE-O-P450 complex (via the heme
oxygen) would better explain the observed destruction of the P450 heme, an outcome not likely
to be epoxide-mediated. Miller and Guengerich (1982) found CYP2E1 to generate an epoxide
but argued that the subsequent production of chloral was not likely related to the epoxide. Green
and Prout (1985) argued against epoxide (free form) formation in vivo in mice and rats,
suggesting that the expected predominant metabolites would be carbon monoxide, CO
2
, MCA,
and DCA, rather than the observed predominant appearance of TCA, TCOH, and TCOG.
It appears likely that both a TCE-O-P450 complex and a TCE oxide are formed, resulting
in both CH and DCAC, respectively, though it appears that the former predominates. In
particular, it has been shown that DCAC can be generated from TCE oxide, dichloracetyl
chloride can be trapped with lysine (Cai and Guengerich, 1999), and dichloracetyl-lysine adducts
are formed in vivo (Forkert et al., 2006). Together, these data strongly suggest TCE oxide as an
intermediate metabolite, albeit short-lived, from TCE oxidation in vivo.

3.3.3.1.2. Formation of CH, TCOH and TCA
CH (in equilibrium with chloral) is a major oxidative metabolite produced from TCE as
has been shown in numerous in vitro systems, including human liver microsomes and purified
P450 CYP2E1 (Guengerich et al., 1991) as well as recombinant rat, mouse, and human P450s
including CYP2E1 (Forkert et al., 2005). However, in rats and humans, in vivo circulating CH is
generally absent from blood following TCE exposure. In mice, CH is detectable in blood and
tissues but is rapidly cleared from systemic circulation (Abbas and Fisher, 1997). The low
systemic levels of CH are due to its rapid transformation to other metabolites.
CH is further metabolized predominantly to TCOH (Shultz and Weiner, 1979; Sellers et
al., 1972) and/or CYP2E1 (Ni et al., 1996). The role for alcohol dehydrogenase was suggested
3-24
by the observation that ethanol inhibited CH reduction to TCOH (Larson and Bull, 1989; Muller
et al., 1975; Sellers et al., 1972). For instance, Sellers et al. (1972) reported that co-exposure of
humans to ethanol and CH resulted in a higher percentage of urinary TCOH (24% of CH
metabolites) compared to TCA (19%). When ethanol was absent, 10 and 11% of CH was
metabolized to TCOH and TCA, respectively. However, because ethanol can be oxidized by
both alcohol dehydrogenase and CYP2E1, there is some ambiguity as to whether these
observations involve competition with one or the other of these enzymes. For instance, Ni et al.
(1996) reported that CYP2E1 expression was necessary for metabolism of CH to mutagenic
metabolites in a human lymphoblastoid cell line, suggesting a role for CYP2E1. Furthermore, Ni
et al. (1996) reported that cotreatment of mice with CH and pyrazole, a specific CYP2E1
inducer, resulted in enhanced liver microsomal lipid peroxidation, while treatment with
2,4-dichloro-6-phenoxyethylamine, an inhibitor of CYP2E1, suppressed lipid peroxidation,
suggesting CYP2E1 as a primary enzyme for CH metabolism in this system. Lipscomb et al.
(1996) suggested that two enzymes are likely responsible for CH reduction to TCOH based on
observation of biphasic metabolism for this pathway in mouse liver microsomes. This behavior
has also been observed in mouse liver cytosol, but was not observed in rat or human liver
microsomes. Moreover, CH metabolism to TCOH increased significantly both in the presence of
nicotinamide adenine dinucleotide (NADH) in the 700 g supernatant of mouse, rat, and human
liver homogenate as well as with the addition of nicotinamide adenine dinucleotide phosphate-
oxidase (NADPH) in human samples, suggesting that two enzymes may be involved (Lipscomb
et al., 1996).
TCOH formed from CH is available for oxidation to TCA (see below) or glucuronidation
via uridine 5-diphospho-glucuronyltransferase to TCOG, which is excreted in urine or in bile
(Stenner et al., 1997). Biliary TCOG is hydrolyzed in the gut and available for reabsorption to
the liver as TCOH, where it can be glucuronidated again or metabolized to TCA. This
enterohepatic circulation appears to play a significant role in the generation of TCA from TCOH
and in the observed lengthy residence time of this metabolite, compared to TCE. Using jugular-,
duodenal-, and bile duct-cannulated rats, Stenner et al. (1997) showed that enterohepatic
circulation of TCOH from the gut back to the liver and subsequent oxidation to TCA was
responsible for 76% of TCA measured in the systemic blood.
Oxidation of CH and TCOH to TCA has been demonstrated in vivo in mice (Larson and
Bull, 1992a; Dekant et al., 1986b; Green and Prout, 1985), rats (Stenner et al., 1997; Pravecek et
al., 1996; Templin et al., 1995b; Larson and Bull, 1992a; Dekant et al., 1986b; Green and Prout,
1985), dogs (Templin et al., 1995b), and humans (Sellers et al., 1978). Urinary metabolite data
in mice and rats exposed to 200 mg/kg TCE (Larson and Bull, 1992a; Dekant et al., 1986b); and
humans following oral CH exposure (Sellers et al., 1978) show greater TCOH production
relative to TCA production. However, because of the much longer urinary half-life in humans of
TCA relative to TCOH, the total amount of TCA excreted may be similar to TCOH (Fisher et al.,
3-25
1998; Monster et al., 1976). This is thought to be primarily due to conversion of TCOH to TCA,
either directly or via back-conversion of TCOH to CH, rather than due to the initial formation
of TCA from CH (Owens and Marshall, 1955).
In vitro data are also consistent with CH oxidation to TCA being much less than CH
reduction to TCOH. For instance, Lipscomb et al. (1996) reported 1,832-fold differences in K
M

values and 10195-fold differences in clearance efficiency (V
MAX
/K
M
) for TCOH and TCA in all
three species (see Table 3-14). Clearance efficiency of CH to TCA in mice is very similar to
humans but is 13-fold higher than rats. Interestingly, Bronley-DeLancey et al. (2006) recently
reported that similar amounts of TCOH and TCA were generated from CH using cryopreserved
human hepatocytes. However, the intersample variation was extremely high, with measured
V
MAX
ranging from 8-fold greater TCOH to 5-fold greater TCA and clearance (V
MAX
/K
M
)
ranging from 13-fold greater TCOH to 17-fold greater TCA. Moreover, because a comparison
with fresh hepatocytes or microsomal protein was not made, it is not clear to what extent these
differences are due to population heterogeneity or experimental procedures.

Table 3-14. In vitro kinetics of TCOH and TCA formation from CH in rat,
mouse, and human liver homogenates

Species
TCOH TCA
K
M
a

V
MAX
b

V
MAX
/K
M
c

K
M
a

V
MAX
b

V
MAX
/K
M
c

Rat 0.52 24.3 46.7 16.4 4 0.24
Mouse
d
0.19 11.3 59.5 3.5 10.6 3.0
High affinity 0.12 6.3 52.5 Not applicable Not applicable Not applicable
Low affinity 0.51 6.1 12.0 Not applicable Not applicable Not applicable
Human 1.34 34.7 25.9 23.9 65.2 2.7

a
K
M
presented as mM CH in solution.
b
V
MAX
presented as nmoles/mg supernatant protein/minute.
c
Clearance efficiency represented by V
MAX
/K
M
.
d
Mouse kinetic parameters derived for observations over the entire range of CH exposure as well as discrete, bi-
phasic regions for CH concentrations below (high affinity) and above (low affinity) 1.0 mM.

Source: Lipscomb et al. (1996).

The metabolism of CH to TCA and TCOH involves several enzymes including CYP2E1,
alcohol dehydrogenase, and aldehyde dehydrogenase enzymes (Ni et al., 1996; Wang et al.,
1993; Guengerich et al., 1991; Miller and Guengerich, 1983; Shultz and Weiner, 1979). Because
these enzymes have preferred cofactors (NADPH, NADH, and NAD
+
), cellular cofactor ratio
and redox status of the liver may have an impact on the preferred pathway (Lipscomb et al.,
1996; Kawamoto et al., 1988a).

3-26
3.3.3.1.3. Formation of DCA and other products
As discussed above, DCA could hypothetically be formed via multiple pathways. The
work reviewed by Guengerich (2004) suggested that one source of DCA may be through a TCE
oxide intermediary. Miller and Guengerich (1983) reported evidence of formation of the
epoxide, and Cai and Guengerich (1999) reported that a significant amount (about 35%) of DCA
is formed from aqueous decomposition of TCE oxide via hydrolysis in an almost pH-
independent manner. Because this reaction forming DCA from TCE oxide is a chemical process
rather than a process mediated by enzymes, and because evidence suggests that some epoxide
was formed from TCE oxidation, Guengerich (2004) notes that DCA would be an expected
product of TCE oxidation (see also Yoshioka et al., 2002). Alternatively, dechlorination of TCA
and oxidation of TCOH have been proposed as sources of DCA (Lash et al., 2000a). Merdink
et al. (2000) investigated dechlorination of TCA and reported trapping a DCA radical with the
spin-trapping agent phenyl-tert-butyl nitroxide, identified by gas chromatography/mass
spectroscopy, in both a chemical Fenton system and rodent microsomal incubations with TCA as
substrate. Dose-dependent catalysis of TCA to DCA was observed in cultured microflora from
B6C3F
1
mice (Moghaddam et al., 1996). However, while antibiotic-treated mice lost the ability
to produce DCA in the gut, plasma DCA levels were unaffected by antibiotic treatment,
suggesting that the primary site of murine DCA production is other than the gut (Moghaddam et
al., 1997).
However, direct evidence for DCA formation from TCE exposure remains equivocal. In
vitro studies in human and animal systems have demonstrated very little DCA production in the
liver (James et al., 1997). In vivo, DCA was detected in the blood of mice (Templin et al., 1993;
Larson and Bull, 1992a) and humans (Fisher et al., 1998) and in the urine of rats and mice
(Larson and Bull, 1992b) exposed to TCE by aqueous gavage. However, the use of strong acids
in the analytical methodology produces ex vivo conversion of TCA to DCA in mouse blood
(Ketcha et al., 1996). This method may have resulted in the appearance of DCA as an artifact in
human plasma (Fisher et al., 1998) and mouse blood in vivo (Templin et al., 1995b). Evidence
for the artifact is suggested by DCA AUCs that were larger than would be expected from the
available TCA (Templin et al., 1995b). After the discovery of these analytical issues, Merdink
et al. (1998) reevaluated the formation of DCA from TCE, TCOH, and TCA in mice, with
particular focus on the hypothesis that DCA is formed from dechlorination of TCA. They were
unable to detect blood DCA in naive mice after administration of TCE, TCOH, or TCA. Low
levels of DCA were detected in the blood of children administered therapeutic doses of CH
(Henderson et al., 1997), suggesting TCA or TCOH as the source of DCA. Oral TCE exposure
in rats and dogs failed to produce detectable levels of DCA (Templin et al., 1995b).
Another difficulty in assessing the formation of DCA is its rapid metabolism at low
exposure levels. Degradation of DCA is mediated by GST-zeta (Saghir and Schultz, 2002; Tong
et al., 1998), apparently occurring primarily in the hepatic cytosol. DCA metabolism results in
3-27
suicide inhibition of the enzyme, evidenced by decreased DCA metabolism in DCA-treated
animals (Gonzalez-Leon et al., 1999) and humans (Shroads et al., 2008) and loss of DCA
metabolic activity and enzymatic protein in liver samples from treated animals (Schultz et al.,
2002). This effect has been noted in young mice exposed to DCA in drinking water at doses
approximating 120 mg/kg-day (Schultz et al., 2002). The experimental data and
pharmacokinetic model simulations of several investigators (Li et al., 2008; Shroads et al., 2008;
Jia et al., 2006; Keys et al., 2004; Merdink et al., 1998) suggest that several factors prevent the
accumulation of measurable amounts of DCA: (1) its formation as a short-lived intermediate
metabolite and (2) its rapid elimination relative to its formation from TCA. While DCA
elimination rates appear approximately one order of magnitude higher in rats and mice than in
humans (James et al., 1997) (see Table 3-15), they still may be rapid enough so that even if DCA
were formed in humans, it would be metabolized too quickly to appear in detectable quantities in
blood.

Table 3-15. In vitro kinetics of DCA metabolism in hepatic cytosol of
mice, rats, and humans

Species
V
MAX

(nmol/min/mg protein)
K
M

(M) V
MAX
/K
M
Mouse 13.1 350 37.4
Rat 11.6 280 41.4
Human 0.37 71 5.2

Source: James et al. (1997).

A number of other metabolites, such as oxalic acid, MCA, glycolic acid, and glyoxylic
acid, are formed from DCA (Saghir and Schultz, 2002; Lash et al., 2000a). Unlike other
oxidative metabolites of TCE, DCA appears to be metabolized primarily via hepatic cytosolic
proteins. Since P450 activity resides almost exclusively in the microsomal and mitochondrial
cell fractions, DCA metabolism appears to be independent of P450. Rodent microsomal and
mitochondrial metabolism of DCA was measured to be 10% of cytosolic metabolism
(Lipscomb et al., 1995). DCA in the liver cytosol from rats and humans is transformed to
glyoxylic acid via a GSH-dependent pathway (James et al., 1997). In rats, the K
M
for GSH was
0.075 mM with a V
MAX
for glyoxylic acid formation of 1.7 nmol/mg protein/minute. While this
pathway may not involve GST (as evidenced by very low GST activity in this study), Tong et al.
(1998) showed GST-zeta, purified from rat liver, to be involved in metabolizing DCA to
glyoxylic acid, with a V
MAX
of 1,334 nmol/mg protein/minute and K
M
of 71.4 M for glyoxylic
acid formation and a GSH K
M
of 59 M.

3-28
3.3.3.1.4. Tissue distribution of oxidative metabolism and metabolites
Oxidative metabolism of TCE, irrespective of the route of administration, occurs
predominantly in the liver, but TCE metabolism via the P450 (CYP) system also occurs at other
sites because CYP isoforms are present to some degree in most tissues of the body. For
example, both the lung and kidneys exhibit CYP enzyme activities (Forkert et al., 2005;
Cummings et al., 2001; 1997a; Green et al., 1997b). Green et al. (1997b) detected TCE
oxidation to chloral in microsomal fractions of whole-lung homogenates from mice, rats, and
humans, with the activity in mice the greatest and in humans the least. The rates were slower
than in the liver (which also has a higher microsomal protein content as well as greater tissue
mass) by 1.8-, 10-, and >10-fold in mice, rats, and humans, respectively. While qualitatively
informative, these rates were determined at a single concentration of about 1 mM TCE. A full
kinetic analysis was not performed, so clearance and maximal rates of metabolism could not be
determined. With the kidney, Cummings et al. (2001) performed a full kinetic analysis using
kidney microsomes and found that clearance rates (V
MAX
/K
M
) for oxidation were >100-fold
smaller than average rates found in the liver (see Table 3-13). In human kidney microsomes,
Amet et al.(1997) reported that CYP2E1 activity was weak and near detection limits, with no
CYP2E1 detectable using immunoblot analysis. Cummings and Lash (2000) reported detecting
oxidation of TCE in only one of four kidney microsome samples, and only at the highest tested
concentration of 2 mM, with a rate of 0.13 nmol/minute/mg protein. This rate contrasts with the
V
MAX
values for human liver microsomal protein of 0.193.5 nmol/minute/mg protein reported
in various experiments (see Table 3-13). Extrahepatic oxidation of TCE may play an important
role for generation of toxic metabolites in situ. The roles of local metabolism in kidney and lung
toxicity are discussed in detail in Sections 4.4 and 4.7, respectively.
With respect to further metabolism beyond oxidation of TCE, CH has been shown to be
metabolized to TCA and TCOH in lysed whole blood of mice and rats and fractionated human
blood (Lipscomb et al., 1996) (see Table 3-16). TCOH production is similar in mice and rats and
is approximately twofold higher in rodents than in human blood. However, TCA formation in
human blood is two- or threefold higher than in mouse or rat blood, respectively. In human
blood, TCA is formed only in the erythrocytes. TCOH formation occurs in both plasma and
erythrocytes, but fourfold more TCOH is found in plasma than in an equal volume of packed
erythrocytes. While blood metabolism of CH may contribute further to its low circulating levels
in vivo the metabolic capacity of blood (and kidney) may be substantially lower than liver.
Regardless, any CH reaching the blood may be rapidly metabolized to TCA and TCOH. DCA
and TCA are known to bind to plasma proteins. Schultz et al. (1999) measured DCA binding in
rats at a single concentration of about 100 M and found a binding fraction of <10%. However,
these data are not greatly informative for TCE exposure in which DCA levels are significantly
lower than 100 M. In addition, the limitation to a single concentration in this experiment
precludes fitting a binding curve, as can be done for TCA with Templin et al. (1995a; 1995b;
3-29
1993), Schultz et al. (1999), Lumpkin et al. (2003), and Yu et al. (2003), all of which measured
TCA binding in various species and at various concentration ranges. Of these, Templin et al.
(1995a; 1995b) and Lumpkin et al. (2003) measured levels in humans, mice, and rats. Lumpkin
et al. (2003) studied the widest concentration range, spanning reported TCA plasma
concentrations from experimental studies. Table 3-17 shows derived binding parameters.
However, these data are not entirely consistent among researchers; two- to fivefold differences in
B
MAX
and K
d
are noted in some cases, although some differences existed in the rodent strains and
experimental protocols used. In general, however, at lower concentrations, the bound fraction
appears greater in humans than in rats and mice. Typical human TCE exposures, even in
controlled experiments with volunteers, lead to TCA blood concentrations well below the
reported K
d
(see Table 3-17, below), so the TCA binding fraction should be relatively constant.
However, in rats and mice, experimental exposures may lead to peak concentrations similar to,
or above, the reported K
d
(e.g., Yu et al., 2000; Templin et al., 1993), meaning that the bound
fraction should temporarily decrease following such exposures.

Table 3-16. TCOH and TCA formed from CH in vitro in lysed whole blood
of rats and mice or fractionated blood of humans (nmoles formed in 400 L
samples over 30 minutes)

Rat Mouse
Human
Erythrocytes Plasma
TCOH 45.4 4.9 46.7 1.0 15.7 1.4 4.48 0.2
TCA 0.14 0.2 0.21 0.3 0.42 0.0 Not detected

Source: Lipscomb et al. (1996).

Table 3-17. Reported TCA plasma binding parameters
a


A B
MAX
(M) K
d
(M)
A+
B
MAX
/K
d

Concentration
range (M
bound+free)
Human
Templin et al. (1995b) 1,020 190 5.37 31,224
Lumpkin et al. (2003) 708.9 174.6 4.06 0.063,065
Rat
Templin et al. (1995b) 540 400 1.35 31,224
Yu et al. (2000) 0.602 312 136 2.90 3.81,530
Lumpkin et al. (2003) 283.3 383.6 0.739 0.063,065
Mouse
Templin et al. (1993) 310 248 1.25 31,224
Lumpkin et al. (2003) 28.7 46.1 0.623 0.061,226

a
Binding parameters based on the equation C
bound
= A C
free
+ B
MAX
C
free
/(K
d
+ C
free
), where C
bound
is the bound
concentration, C
free
is the free concentration, and A = 0 for Templin et al. (1995b; 1993) and Lumpkin et al. (2003).
The quantity A+ B
MAX
/K
d
is the ratio of bound-to-free at low concentrations.

3-30
Limited data are available on tissue:blood partitioning of the oxidative metabolites CH,
TCA, TCOH, and DCA, as shown in Table 3-18. As these chemicals are all water soluble and
not lipophilic, it is not surprising that their partition coefficients are close to one (within about
twofold). It should be noted that the TCA tissue:blood partition coefficients reported in
Table 3-18 were measured at concentrations 1.63.3 M, over 1,000-fold higher than the reported
K
d
. Therefore, these partition coefficients should reflect the equilibrium between tissue and free
blood concentrations. In addition, only one in vitro measurement has been reported of
blood:plasma concentration ratios for TCA: Schultz et al. (1999) reported a value of 0.76 in rats.

Table 3-18. Partition coefficients for TCE oxidative metabolites

Species/tissue
a

Tissue:blood partition coefficient
CH TCA TCOH DCA
Human
b

Kidney 0.66 2.15 -
Liver 0.66 0.59 -
Lung 0.47 0.66 -
Muscle 0.52 0.91 -
Mouse
c

Kidney 0.98 0.74 1.02 0.74
Liver 1.42 1.18 1.3 1.08
Lung 1.65 0.54 0.78 1.23
Muscle 1.35 0.88 1.11 0.37

a
TCA and TCOH partition coefficients have not been reported for rats.
b
Fisher et al. (1998).
c
Abbas and Fisher (1997).

3.3.3.1.5. Species-, sex-, and age-dependent differences of oxidative metabolism
The ability to describe species- and sex-dependent variations in TCE metabolism is
important for species extrapolation of bioassay data and identification of human populations that
are particularly susceptible to TCE toxicity. In particular, information on the variation in the
initial oxidative step of CH formation from TCE is desirable, because this is the rate-limiting
step in the eventual formation and distribution of the putative toxic metabolites TCA and DCA
(Lipscomb et al., 1997).
Inter- and intraspecies differences in TCE oxidation have been investigated in vitro using
cellular or subcellular fractions, primarily of the liver. The available in vitro metabolism data on
TCE oxidation in the liver (see Table 3-13) show substantial inter- and intraspecies variability.
Across species, microsomal data show that mice apparently have greater capacity (V
MAX
) than
rat or humans, but the variability within species can be 210-fold. Part of the explanation may
be related to CYP2E1 content. Although liver P450 content is similar across species, mice and
rats exhibit higher levels of CYP2E1 content (0.85 and 0.89 nmol/mg protein, respectively)
3-31
(Davis et al., 2002; Nakajima et al., 1993) than humans (approximately 0.250.30 nmol/mg
protein) (Davis et al., 2002; Elfarra et al., 1998). Thus, the data suggest that rodents would have
a higher capacity than humans to metabolize TCE, but this is difficult to verify in vivo because
very high exposure concentrations in humans would be necessary to assess the maximum
capacity of TCE oxidation.
With respect to the K
M
of liver microsomal TCE oxidative metabolism, where K
M
is
indicative of affinity (the lower the numerical value of K
M,
the higher the affinity), the trend
appears to be that mice and rats have higher K
M
values (i.e., lower affinity) than humans, but
with substantial overlap due to interindividual variability. Note that, as shown in Table 3-13, the
ranking of rat and mouse liver microsomal K
M
values between the two reports by Lipscomb et al.
(1998c) and Elfarra et al. (1998) is not consistent. However, both studies clearly show that K
M
is
the lowest (i.e., affinity is highest) in humans. Because clearance at lower concentrations is
determined by the ratio V
MAX
to K
M
, the lower apparent K
M
in humans may partially offset the
lower human V
MAX
, and lead to similar oxidative clearances in the liver at environmentally
relevant doses. However, differences in activity measured in vitro may not translate into in vivo
differences in metabolite production, as the rate of metabolism in vivo depends also on the rate
of delivery to the tissue via blood flow (Lipscomb et al., 2003). The interaction of enzyme
activity and blood flow is best investigated using PBPK models and is discussed, along with
descriptions of in vivo data, in Section 3.5.
Data on sex- and age-dependence in oxidative TCE metabolism are limited but suggest
relatively modest differences in humans and animals. In an extensive evaluation of CYP-
dependent activities in human liver microsomal protein and cryopreserved hepatocytes,
Parkinson et al. (2004) identified no age- or gender-related differences in CYP2E1 activity. In
liver microsomes from 23 humans, the K
M
values for females was lower than males, but V
MAX

values were very similar (Lipscomb et al., 1997). Appearance of total trichloro compounds
(TTCs) in urine following i.p. dosing with TCE was 28% higher in female rats than in males
(Verma and Rana, 2003). The oxidation of TCE in male and female rat liver microsomes was
not significantly different; however, pregnancy resulted in a decrease of 2739% in the rate of
CH production in treated microsomes from females (Nakajima et al., 1992b). Formation of CH
in liver microsomes in the presence of 0.2 or 5.9 mM TCE exhibited some dependency on age of
rats, with formation rates in both sexes of 1.11.7 nmol/mg protein/minute in 3-week-old
animals and 0.51.0 nmol/mg protein/minute in 18-week-old animals (Nakajima et al., 1992b).
Fisher et al. (1991) reviewed data available at that time on urinary metabolites to
characterize species differences in the amount of urinary metabolism accounted for by TCA (see
Table 3-19). They concluded that TCA seemed to represent a higher percentage of urinary
metabolites in primates than in other mammalian species, indicating a greater proportion of
oxidation leading ultimately to TCA relative to TCOG.

3-32
Table 3-19. Urinary excretion of TCA by various species exposed to TCE
(based on data reviewed in (Fisher et al., 1991)

Species
a

Percentage of
urinary excretion of
TCA
Dose route
TCE dose
(mg TCE/kg) References Male Female
Baboon
b,c
16 Intramuscular
injection
50 Mueller et al. (1982)
Chimpanzee
b
24 22 Intramuscular
injection
50 Mueller et al. (1982)
Monkey, Rhesus
b,c
19 Intramuscular
injection
50 Mueller et al. (1982)
Mice, NMRI
d
820 Oral intubation 2200 Dekant et al. (1986b)
Mice, B6C3F
1
b
712 Oral intubation 102,000 Green and Prout (1985)
Rabbit, Japanese
White
b,c

0.5 i.p. injection 200 Nomiyama and Nomiyama
(1979)
Rat, Wistar
d
1417 Oral intubation 2200 Dekant et al. (1986b)
Rat, Osborne-Mendel
a
67 Oral intubation 102,000 Green and Prout (1985)
Rat, Holtzman
a
7 i.p. injection 10 mg TCE/rat Nomiyama and Nomiyama
(1979)

a
The human data tabulated in Fisher et al. (1991) from Nomiyama and Nomiyama (1971) were not included here
because they were relative to urinary excretion of TTCsnot as fraction of intake as was the case for the other data
included here.
b
Percentage urinary excretion determined from accumulated amounts of TCOH and TCA in urine 36 days
postexposure.
c
Sex not specified.
d
Percentage urinary excretion determined from accumulated amounts of TCOH, DCA, oxalic acid, and
N-(hydroxyacetyl)aminoethanol in urine 3 days postexposure.

3.3.3.1.6. CYP isoforms and genetic polymorphisms
A number of studies have identified multiple P450 isozymes as having a role in the
oxidative metabolism of TCE. These isozymes include CYP2E1 (Nakajima et al., 1992a;
Guengerich et al., 1991; Guengerich and Shimada, 1991; Nakajima et al., 1990; Nakajima et al.,
1988), CYP3A4 (Shimada et al., 1994), CYP1A1/2, CYP2C11/6 (Nakajima et al., 1993;
Nakajima et al., 1992a), CYP2F, and CYP2B1 (Forkert et al., 2005). Recent studies in CYP2E1-
knockout mice have shown that in the absence of CYP2E1, mice still have substantial capacity
for TCE oxidation (Forkert et al., 2006; Kim and Ghanayem, 2006). However, CYP2E1 appears
to be the predominant (i.e., higher affinity) isoform involved in oxidizing TCE (Forkert et al.,
2005; Nakajima et al., 1992a; Guengerich et al., 1991; Guengerich and Shimada, 1991). In rat
liver, CYP2E1 catalyzed TCE oxidation more than CYP2C11/6 (Nakajima et al., 1992a). In rat
recombinant-derived P450s, the CYP2E1 had a lower K
M
(higher affinity) and higher V
MAX
/K
M

ratio (intrinsic clearance) than CYP2B1 or CYP2F4 (Forkert et al., 2005). Interestingly, there
was substantial differences in K
M
between rat and human CYP2E1s and between rat CYP2F4
3-33
and mouse CYP2F2, suggesting that species-specific isoforms have different kinetic behavior
(see Table 3-20).

Table 3-20. P450 isoform kinetics for metabolism of TCE to CH in human,
rat, and mouse recombinant P450s

Experiment
K
M

M
V
MAX

pmol/min/pmol P450 V
MAX
/K
M

Human rCYP2E1 196 40 4 0.2 0.02
Rat rCYP2E1 14 3 11 0.3 0.79
Rat rCYP2B1 131 36 9 0.5 0.07
Rat rCYP2F4 64 9 17 0.5 0.27
Mouse rCYP2F2 114 17 13 0.4 0.11

Source: Forkert et al. (2005).

The presence of multiple P450 isoforms in human populations affects the variability in
individuals ability to metabolize TCE. Studies using microsomes from human liver or from
human lymphoblastoid cell lines expressing CYP2E1, CYP1A1, CYP1A2, or CYP3A4 have
shown that CYP2E1 is responsible for >60% of oxidative TCE metabolism (Lipscomb et al.,
1997). Similarities between metabolism of chlorzoxazone (a CYP2E1 substrate) in liver
microsomes from 28 individuals (Peter et al., 1990) and TCE metabolism helped identify
CYP2E1 as the predominant (high affinity) isoform for TCE oxidation. Additionally, Lash et al.
(2000a) suggested that, at concentrations above the K
M
value for CYP2E1, CYP1A2, and
CYP2A4 may also metabolize TCE in humans; however, their contribution to the overall TCE
metabolism was considered low compared to that of CYP2E1. Given the difference in
expression of known TCE-metabolizing P450 isoforms (see Table 3-21) and the variability in
P450-mediated TCE oxidation (Lipscomb et al., 1997), significant variability may exist in
individual human susceptibility to TCE toxicity.

3-34
Table 3-21. P450 isoform activities in human liver microsomes exhibiting
different affinities for TCE

Affinity group
CYP isoform activity (pmol/min/mg protein)
a

CYP2E1 CYP1A2 CYP3A4
Low K
M
520 295 241 146 2.7 2.7
Mid K
M
820 372 545 200 2.9 2.8
High K
M
1,317 592 806 442 1.8 1.1

a
Activities of CYP1A2, CYP2E1, and CYP3A4 were measured with phenacetin, chlorzoxazone, and testosterone as
substrates, respectively. Data are means SD from 10, 9, and 4 samples for the low-, mid-, and high-K
M
groups,
respectively. Only CYP3A4 activities are not significantly different (p < 0.05) from one another by Kruskal-Wallis
one-way analysis of variance.

Source: Lash et al. (2000a).

Differences in content and/or intrinsic catalytic properties (K
M
, V
MAX
) of specific
enzymes among species, strains, and individuals may play an important role in the observed
differences in TCE metabolism and resulting toxicities. Lipscomb et al. (1997) reported
observing three statistically distinct groups of K
M
values for TCE oxidation using human
microsomes. The mean standard deviation (SD) (M TCE) for each of the three groups was
16.7 2.5 (n = 10), 30.9 3.3 (n = 9), and 51.1 3.8 (n = 4). Within each group, there were no
significant differences in sex or ethnicity. However, the overall observed K
M
values in female
microsomes (21.9 3.5 M, n = 10) were significantly lower than males (33.1 3.5 M,
n = 13). Interestingly, in human liver microsomes, different groups of individuals with different
affinities for TCE oxidation appeared to also have different activities for other substrates not
only with respect to CYP2E1 but also CYP1A2 (Lash et al., 2000a) (see Table 3-21). Genetic
polymorphisms in humans have been identified in the CYP isozymes thought to be responsible
for TCE metabolism (Pastino et al., 2000), but no data exist correlating these polymorphisms
with enzyme activity. It is relevant to note that repeat polymorphism (Hu et al., 1999) or
polymorphism in the regulatory sequence (McCarver et al., 1998) were not involved in the
constitutive expression of human CYP2E1; however, it is unknown if these types of
polymorphisms may play a role in the inducibility of the respective gene.
Individual susceptibilities to TCE toxicity may also result from variations in enzyme
content, either at baseline or due to enzyme induction/inhibition, which can lead to alterations in
the amounts of metabolites formed. Certain physiological and pathological conditions or
exposure to other chemicals (e.g., ethanol and acetominophen) can induce, inhibit, or compete
for enzymatic activity. Given the well-established (or well-characterized) role of the liver to
oxidatively metabolize TCE (by CYP2E1), increasing the CYP2E1 content or activity (e.g., by
enzyme induction) may not result in further increases in TCE oxidation. Indeed, Kaneko et al.
(1994) reported that enzyme induction by ethanol consumption in humans increased TCE
metabolism only at high concentrations (500 ppm, 2,687 mg/m
3
) in inspired air. However, other
3-35
interactions between ethanol and the enzymes that oxidatively metabolize TCE metabolites can
result in altered metabolic fate of TCE metabolites. In addition, enzyme inhibition or
competition can decrease TCE oxidation and subsequently alter the TCE toxic response via, for
instance, increasing the proportion undergoing GSH conjugation Lash et al. (2000a). TCE itself
is a competitive inhibitor of CYP2E1 activity (Lipscomb et al., 1997), as shown by reduced
p-nitrophenol hydroxylase activity in human liver microsomes, and may therefore alter the
toxicity of other chemicals metabolized through that pathway. On the other hand, suicidal CYP
heme destruction by the TCE-oxygenated CYP intermediate has also been shown (Miller and
Guengerich, 1983).

3.3.3.2. GSH Conjugation Pathway
Historically, the conjugative metabolic pathways have been associated with xenobiotic
detoxification. This is true for GSH conjugation of many compounds. However, several
halogenated alkanes and alkenes, including TCE, are bioactivated to cytotoxic metabolites by the
GSH conjugate processing pathway (mercapturic acid) pathways (Elfarra et al., 1987; Elfarra et
al., 1986). In the case of TCE, production of reactive species several steps downstream from the
initial GSH conjugation is believed to cause cytotoxicity and carcinogenicity, particularly in the
kidney. Since the GSH conjugation pathway is in competition with the P450 oxidative pathway
for TCE biotransformation, it is important to understand the role of various factors in
determining the flux of TCE through each pathway. Figure 3-5 depicts the present
understanding of TCE metabolism via GSH conjugation.

3-36



Cl SG
H Cl
Cl SG
H Cl
S
Cl
2
C
2
H
O
O-
NH
3
+
S
Cl
2
C
2
H
O
O-
NH
3
+
S
Cl
2
C
2
H
O
O-
NH
O
S
Cl
2
C
2
H
O
O-
NH
O
S
Cl
2
C
2
H
O
O-
NH
3
+
O
S
Cl
2
C
2
H
O
O-
NH
3
+
O
O
S
Cl
2
C
2
H
O
O-
NH
O
O
S
Cl
2
C
2
H
O
O-
NH
O
Cl S

H Cl
Cl S

H Cl
Cl Cl
H Cl
(TCE)
Cl Cl
H Cl
(TCE)
GST
GGT
CGDP
(DCVC)
NAT
Acylase
CYP3A
(NAcDCVC)
(NAcDCVCS)
-lyase
(DCVT)
(DCVCS)
FMO-3
P450
Adapted from: Lash et al. (2000a); Cummings and Lash (2000); NRC (2006).
Figure 3-5. Scheme for GSH-dependent metabolism of TCE.

3.3.3.2.1. Formation of S-(1,2-dichlorovinyl)glutathione or S-(2,2-dichlorovinyl)-
glutathione (DCVG)
The conjugation of TCE to GSH produces S-(1,2-dichlorovinyl)glutathione or its isomer
S-(2,2-dichlorovinyl)glutathione (collectively, S-dichlorovinyl-glutathione, DCVG). There is
some uncertainty as to which GST isoforms mediate TCE conjugation. Lash and colleagues
studied TCE conjugation in renal tissue preparations, isolated renal tubule cells from male F344
rats and purified GST alpha-class isoforms 1-1, 1-2, and 2-2 (Cummings and Lash, 2000;
Cummings et al., 2000b; Lash et al., 2000b). The results demonstrated high conjugative activity
in the renal cortex and proximal tubule cells. Although the isoforms studied had similar V
MAX

3-37
values, the K
M
value for GST 2-2 was significantly lower than the other forms, indicating that
this form will catalyze TCE conjugation at lower (more physiologically relevant) substrate
concentrations. In contrast, using purified rat and human enzymes, Hissink et al. (2002) reported
in vitro activity for DCVG formation only for mu- and pi-class GST isoforms, and none towards
alpha-class isoforms; however, the rat mu-class GST 3-3 was several-fold more active than the
human mu-class GST M1-1. Although GSTs are present in tissues throughout the body, the
majority of TCE GSH conjugation is thought to occur in the liver (Lash et al., 2000a). Using in
vitro studies with renal preparations, it has been demonstrated that GST catalyzed conjugation of
TCE is increased following the inhibition of CYP-mediated oxidation (Cummings and Lash,
2000).
In F344 rats, following gavage doses of 2631,971 mg/kg TCE in 2 mL corn oil, DCVG
was observed in the liver and kidney of females only, in blood of both sexes (Lash et al., 2006),
and in bile of males (Dekant, 1990). The data from Lash et al. (2006) are difficult to interpret
because the time courses seem extremely erratic, even for the oxidative metabolites TCOH and
TCA. Moreover, a comparison of blood levels of TCA and TCOH with other studies in rats at
similar doses reveals differences of over 1,000-fold in reported concentrations. For instance, at
the lowest dose of 263 mg/kg, the peak blood levels of TCE and TCA in male F344 rats were
10.5 and 1.6 g/L, respectively (Lash et al., 2006). By contrast, Larson and Bull (1992a)
reported peak blood TCE and TCA levels in male Sprague-Dawley rats over 1,000-fold higher
around 10 and 13 mg/L, respectivelyfollowing oral doses of 197 mg/kg as a suspension in 1%
aqueous Tween 80

. The results of Larson and Bull (1992a) are similar to Lee et al. (2000b),
who reported peak blood TCE levels of 2050 mg/L after male Sprague-Dawley rats received
oral doses of 144432 mg/kg in a 5% aqueous Alkamus emulsion (polyethoxylated vegetable
oil), and to Stenner et al. (1997), who reported peak blood levels of TCA in male F344 rats of
about 5 mg/L at a slightly lower TCE oral dose of 100 mg/kg administered to fasted animals in
2% Tween 80

. Thus, while useful qualitatively as an indicator of the presence of DCVG in rats,

the quantitative reliability of reported concentrations, for metabolites of either oxidation or GSH
conjugation, may be questionable.
In humans, DCVG was readily detected at in human blood following onset of a 4-hour
TCE inhalation exposure to 50 or 100 ppm (269 or 537 mg/m
3
) (Lash et al., 1999b). At 50 ppm,
peak blood levels ranged from 2.5 to 30 M, while at 100 ppm, the mean ( SE, n = 8) peak
blood levels were 46.1 14.2 M in males and 13.4 6.6 M in females. Although on average,
male subjects had threefold higher peak blood levels of DCVG than females, DCVG blood levels
in half of the male subjects were similar to or lower than those of female subjects. This suggests
a polymorphism in GSH conjugation of TCE rather than a true gender difference (Lash et al.,
1999b) as also has been indicated by Hissink et al. (2002) for the human mu-class GST M1-1
enzyme. Interestingly, as shown in Table 3-22, the peak blood levels of DCVG are similar on a
3-38
molar basis to peak levels of TCE, TCA, and TCOH in the same subjects, as reported in
Fisher et al. (1998).

Table 3-22. Comparison of peak blood concentrations in humans exposed to
100 ppm (537 mg/m
3
) TCE for 4 hours

Chemical species
Peak blood concentration (mean SD, M)
Males Females
TCE 23 11 14 4.7
TCA 56 9.8 59 12
TCOH 21 5.0 15 5.6
DCVG 46.1 14.2 13.4 6.6

Sources: Fisher et al. (1998); Lash et al. (1999a).

Tables 3-233-25 summarize DCVG formation from TCE conjugation from in vitro
studies of liver and kidney cellular and subcellular fractions in mouse, rat, and human (tissue-
distribution and species- and gender-differences in DCVG formation are discussed below). As
shown by these tables, different investigators have reported considerably different rates for TCE
conjugation in human liver and kidney cell fractions. For instance, values in Table 3-23 from
Lash et al. (1999b) are between 2 and 5 orders of magnitude higher than those reported by Green
et al. (1997a) or Dekant et al. (1990) (see Table 3-25). In addition, Green et al. (1997a) and
Dekant et al. (1990) reported a difference in the relative importance of rat liver cytosol and rat
liver microsomes for GSH conjugation, with Green et al. (1997a) reporting activity in the cytosol
and none in the microsomes and Dekant et al. (1990) reporting the opposite.

3-39
Table 3-23. GSH conjugation of TCE (at 12 mM) in liver and kidney
cellular fractions in humans, male F344 rats, and male B6C3F
1
mice from
Lash laboratory

Species and cellular/subcellular fraction (TCE
concentration)
DCVG formation
(nmol/hr/mg protein or 10
6
cells)
a

Male Female
Human
Hepatocytes (0.9 mM) (pooled) 11 3
Liver cytosol (1 mM) (individual samples) 156 16 174 13
Liver cytosol (2 mM) (pooled) 346
Liver microsomes (1 mM) (individual samples) 108 24 83 11
Liver microsomes (1 mM) (pooled) 146
Kidney cytosol (2 mM) (pooled) 42
Kidney microsomes (1 mM) (pooled) 320
Rat
Liver cytosol (2 mM) 7.30 2.8 4.86 0.14
Liver microsomes (2 mM) 10.3 2.8 7.24 0.24
Kidney cortical cells (2 mM) 0.48 0.02 0.65 0.15
Kidney cytosol (2 mM) 0.45 0.22 0.32 0.02
Kidney microsomes (2 mM) Not detected 0.61 0.06
Mouse
Liver cytosol (2 mM) 24.5 2.4 21.7 0.9
Liver microsomes (2 mM) 40.0 3.1 25.6 0.8
Kidney cytosol (2 mM) 5.6 0.24 3.7 0.48
Kidney microsomes (2 mM) 5.47 1.41 16.7 4.7

a
Mean SE.

Sources: Lash et al. (1999a; 1998a; 1995); Cummings and Lash (2000).

3-40
Table 3-24. Kinetics of TCE metabolism via GSH conjugation in male F344
rat kidney and human liver and kidney cellular and subcellular fractions
from Lash laboratory

Tissue and cellular fraction
K
M

(M TCE)
V
MAX

(nmol
DCVG/min/mg
protein or 10
6

hepatocytes)
1,000
V
MAX
/K
M

Rat
Kidney proximal tubular cells: low affinity 2,910 0.65 0.22
Kidney proximal tubular cells: high affinity 460 0.47 1.0
Human
Liver hepatocytes
a
37~106 0.16~0.26 2.4~4.5
Liver cytosol: low affinity 333 8.77 2.6
Liver cytosol: high affinity 22.7 4.27 190
Liver microsomes: low affinity 250 3.1 12
Liver microsomes: high affinity 29.4 1.42 48
Kidney proximal tubular cells: low affinity 29,400 1.35 0.046
Kidney proximal tubular cells: high affinity 580 0.11 0.19
Kidney cytosol 26.3 0.81 31
Kidney microsomes 167 6.29 38

a
Kinetic analyses of first 69 (out of 10) data points from Figure 1 from Lash et al. (1999b) using Lineweaver-Burk
or Eadie-Hofstee plots and linear regression (R
2
= 0.50~0.95). Regression with best R
2
used first 6 data points and
Eadie-Hofstee plot, with resulting K
M
and V
MAX
of 106 and 0.26, respectively.

Sources: Lash et al. (1999b); Cummings and Lash (2000); (Cummings et al., 2000b).
3-41
Table 3-25. GSH conjugation of TCE (at 1.44 mM) in liver and kidney
cellular fractions in humans, male F344 rats, and male B6C3F
1
mice from
Green and Dekant laboratories

Species and cellular/subcellular fraction (TCE
concentration)
DCVG formation
(nmol/hr/mg protein) (substrate concentration in
mM)
a

Dekant et al. (1990) Green et al. (1997a)
Human
Liver cytosol - 0.00019 0.00014
Liver microsomes - Not determined
Kidney cytosol - Not determined
Kidney microsomes - Not determined
Rat
Liver cytosol <0.002 0.00162 0.00002
Liver microsomes 0.002 Not determined
Kidney cytosol - Not determined
Kidney microsomes - Not determined
Mouse
Liver cytosol - 0.0025
Liver microsomes - Not determined
Kidney cytosol - Not determined
Kidney microsomes - Not determined

a
Where available, mean SD.

Sources: Dekant et al. (1990), Green et al. (1997a).

The reasons for such discrepancies are unclear, but they may be related to different
analytical methods (Lash et al., 2000a). In particular, Lash et al. (1999b) employed the Reed
method, which used ion-exchange high-performance liquid chromatography (HPLC) of
derivatized analytes. This HPLC method is characterized by variability and an overall decline in
retention times over the life of the HPLC column due to derivatization of amine groups on the
column (Lash et al., 1999a). Although data are limited, the GSH pathway metabolite levels
reported by methods that utilize [
14
C]-TCE and radiochemical detection followed by mass
spectrometry (MS) identification of the metabolites are lower. In particular, Green et al. (1997a)
and Dekant et al. (1990) both used HPLC with radiochemical detection. Peak identity was
confirmed by Green et al. (1997a) using liquid chromatography (LC)/MS and by GC/MS
following hydrolysis by Dekant et al. (1990). In addition, studies using HPLC-MS/MS
techniques with stable isotope-labeled DCVG and dichlorovinyl cysteine (DCVC) standards
have also been used to detect GSH pathway metabolite levels Kim et al. (2009). Based on the in
vitro work presented in Table 3-23 using the Reed method, one would expect mouse serum
DCVG levels to be ~4-6 times lower than humans. However, using the HPLC-MS/MS
technique of Kim et al. (2009), the peak DCVG serum levels are ~1,000 times lower in mouse
3-42
serum than determined by Lash et al. (1999b) in human serum. Although advances in LC
technology, and differences in exposure routes (inhalation vs. oral, with different first pass),
exposure doses, and the degree of competition with TCE oxidation (greater in mouse than in
human) should be considered, this much-larger-than-expected difference is consistent with the
suggestion that the Reed method provides an overestimation of DCVG levels in humans. This
could occur if the Reed method identifies nonspecific derivatives as DCVG or other GSH
pathway metabolites. However, the degree of overestimation is unclear, and differing results in
humans may be attributable to true interindividual variation (especially since GSTs are known to
be polymorphic). Overall, there remains significant uncertainty in the quantitative estimation of
DCVG formation from TCE both in vivo and in vitro.

3.3.3.2.2. Formation of S-(1,2-dichlorovinyl) cysteine or S-(2,2-dichlorovinyl) cysteine
(DCVC)
The cysteine conjugate, isomers S-(1,2-dichlorovinyl) cysteine (1,2-DCVC) or
S-(2,2-dichlorovinyl) cysteine (2,2-DCVC) (collectively S-dichlorovinyl-cysteine, DCVC), is
formed from DCVG in a two-step sequence. DCVG is first converted to the cysteinylglycine
conjugate S-(1,2-dichlorovinyl)-L-cysteinylglycine or its isomer S-(2,2-dichlorovinyl)-
L-cysteinylglycine by -glutamyl transpeptidase (GGT) in the renal brush border (Lash et al.,
1988; Elfarra and Anders, 1984).
Cysteinylglycine dipeptidases in the renal brush border and basolateral membrane
convert DCVG to DCVC via glycine cleavage (Goeptar et al., 1995; Lash et al., 1995). This
reaction can also occur in the bile or gut, as DCVG excreted into the bile is converted to DCVC
and reabsorbed into the liver where it may undergo further acetylation.

3.3.3.2.3. Formation of N-Acetyl-S-(1,2-dichlorovinyl)-L-cysteine or N-Acetyl-
S-(2,2-dichlorovinyl)-L-cysteine (NAcDCVC)
N-acetylation of DCVC can either occur in the kidney, as demonstrated in rat kidney
microsomes (Duffel and Jakoby, 1982), or in the liver (Birner et al., 1997). Subsequent release
of DCVC from the liver to blood may result in distribution to the kidney resulting in increased
internal kidney exposure to the acetylated metabolite over and above what the kidney already is
capable of generating. In the kidney, N-Acetyl-S-(1,2-dichlorovinyl)-L-cysteine or N-Acetyl-
S-(2,2-dichlorovinyl)-L-cysteine (collectively N-Acetyl-S-dichlorovinyl-L-cysteine, NAcDCVC)
may undergo deacetylation, which is considered a rate-limiting-step in the production of
proximal tubule damage (Wolfgang et al., 1989a; Zhang and Stevens, 1989). As a polar
mercapturate, NAcDCVC may be excreted in the urine as evidenced by findings in mice (Birner
et al., 1993), rats (Bernauer et al., 1996; Commandeur and Vermeulen, 1990), and humans who
were exposed to TCE (Bernauer et al., 1996; Birner et al., 1993), suggesting a common
GSH-mediated metabolic pathway for DCVC among species.
3-43

3.3.3.2.4. Beta lyase metabolism of DCVC
The enzyme cysteine conjugate -lyase catalyzes the breakdown of 1,2-DCVC to reactive
nephrotoxic metabolites (Goeptar et al., 1995). This reaction involves removal of pyruvate and
ammonia and production of S-dichlorovinyl thiol (DCVT), an unstable intermediate, which
rearranges to other reactive alkylation metabolites that form covalent bonds with cellular
nucleophiles (Goeptar et al., 1995; Dekant et al., 1988). The rearrangement of DCVT to
enethiols and their acetylating agents has been described in trapping experiments (Dekant et al.,
1988) and proposed to be responsible for nucleophilic adduction and toxicity in the kidney. The
quantification of acid-labile adducts was proposed as a metric for TCE flux through the GSH
pathway. However, the presence of analytical artifacts precluded such analysis. In fact,
measurement of acid-labile adduct products resulted in higher values in mice than in rats (Eyre et
al., 1995b, a).
DCVC metabolism to reactive species via a -lyase pathway has been observed in vitro
by Green et al. (1997a), who reported greater -lyase activity in rats than in mice or humans.
However, in vitro DCVC metabolism by the competing enzyme N-acetyl transferase was also
reported to be greater in rats than mice and humans. In vivo, -lyase activity in humans and rats
(reaction rates were not reported) was demonstrated using a surrogate substrate, 2-(fluoro-
methoxy)-1,1,3,3,3-pentafluoro-1-propene (Iyer et al., 1998). -lyase-mediated reactive adducts
have been described in several extrarenal tissues, including rat and human liver and intestinal
microflora (Larsen and Stevens, 1986; Tomisawa et al., 1986; Stevens, 1985; Tomisawa et al.,
1984; Stevens and Jakoby, 1983; Dohn and Anders, 1982; Tateishi et al., 1978) and rat brain
(Alberati-Giani et al., 1995; Malherbe et al., 1995).
In the kidneys, glutamine transaminase K appears to be primarily responsible for -lyase
metabolism of DCVC (Perry et al., 1993; Lash et al., 1990; Jones et al., 1988; Stevens et al.,
1988; Lash et al., 1986; Stevens et al., 1986). -Lyase transformation of DCVC appears to be
regulated by 2-keto acids. DCVC toxicity in isolated rat proximal tubular cells was significantly
increased with the addition of -keto--methiolbutyrate or phenylpyruvate (Elfarra et al., 1986).
The presence of -keto acid cofactors is necessary to convert the inactive form of the -lyase
enzyme (containing pyridoxamine phosphate) to the active form (containing pyridoxal
phosphate) (Goeptar et al., 1995).
Both low- and high-molecular-weight enzymes with -lyase activities have been
identified in rat kidney cytosol and mitochondria (Abraham et al., 1995a; Abraham et al., 1995b;
Stevens et al., 1988; Lash et al., 1986). While glutamine transaminase K and kynureninase-
associated -lyase activities have been identified in rat liver (Alberati-Giani et al., 1995; Stevens,
1985), they are quite low compared to renal glutamine transaminase K activity and do not result
in hepatotoxicity in DCVG- or DCVC-treated rats (Elfarra and Anders, 1984). Similar isoforms
of -lyase have also been reported in mitochondrial fractions of brain tissue (Cooper, 2004).
3-44
The kidney enzyme, L--hydroxy (L-amino) acid oxidase, is capable of forming an
iminium intermediate and keto acid analogues (pyruvate or S-(1,2-dichlorovinyl)-2-oxo-
3-mercaptopropionate) of DCVC, which decomposes to dichlorovinylthiol (Lash et al., 1990;
Stevens et al., 1989). In rat kidney homogenates, this enzyme activity resulted in as much as
35% of GSH pathway-mediated bioactivation. However, this enzyme is not present in humans,
an important consideration for extrapolation of renal effects across species.

3.3.3.2.5. DCVC and NAcDCVC
A second pathway for bioactivation of TCE S-conjugates involves sulfoxidation of either
the cysteine or mercapturic acid conjugates (Krause et al., 2003; Lash et al., 2003; Birner et al.,
1998; Werner et al., 1996, 1995a; Werner et al., 1995b; Lash et al., 1994; Park et al., 1992;
Sausen and Elfarra, 1990). Sulfoxidation of DCVC was mediated mainly by flavin
monooxygenase 3 (FMO3), rather than CYP, in rabbit liver microsomes (Ripp et al., 1997) and
human liver microsomes (Krause et al., 2003). Krause et al. (2003) also reported DCVC
sulfoxidation by human cDNA-expressed FMO3, as well as detection of FMO3 protein in human
kidney samples. While Krause et al. (2003) were not able to detect sulfoxidation in human
kidney microsomes, the authors noted FMO3 expression in the kidney was lower and more
variable than that in the liver. However, sulfoxidation products in tissues or urine have not been
reported in vivo.
Sulfoxidation of NAcDCVC, by contrast, was found to be catalyzed predominantly, if not
exclusively, by CYP3A enzymes (Werner et al., 1996), whose expressions are highly
polymorphic in humans. Sulfoxidation of other haloalkyl mercapturic acid conjugates has also
been shown to be catalyzed by CYP3A (Altuntas et al., 2004; Werner et al., 1995a; Werner et al.,
1995b). While Lash et al. (2000a) suggested that this pathway would be quantitatively minor
because of the relatively low CYP3A levels in the kidney, no direct data exist to establish the
relative toxicological importance of this pathway relative to bioactivation of DCVC by -lyase or
FMO3. However, the contribution of CYP3A in S-conjugate sulfoxidation to nephrotoxicity in
vivo was recently demonstrated by Sheffels et al. (2004) with fluoromethyl-2,2-difluoro-
1-(trifluoromethyl)vinyl ether (FDVE). In particular, in vivo production and urinary excretion of
FDVE-mercapturic acid sulfoxide metabolites were unambiguously established by mass
spectrometry, and CYP inducers/inhibitors increased/decreased nephrotoxicity in vivo while
having no effect on urinary excretion of metabolites produced through -lyase (Sheffels et al.,
2004). These data suggest that, by analogy, sulfoxidation of NAcDCVC may be an important
bioactivating pathway.

3.3.3.2.6. Tissue distribution of GSH metabolism
The sites of enzymatic metabolism of TCE to the various GSH pathway-mediated
metabolites are significant in determining target tissue toxicity along this pathway. Figure 3-6
3-45
presents a schematic of interorgan transport and metabolism of TCE along the GSH pathway.
TCE is taken up either by the liver or kidney and conjugated to DCVG. The primary factors
affecting TCE flux via this pathway include high hepatic GST activity, efficient transport of
DCVG from the liver to the plasma or bile, high renal brush border and low hepatic GGT
activities, and the capability for GSH conjugate uptake into the renal basolateral membranes with
limited or no uptake into liver cell plasma membranes.


Blood flow
Bile flow
Glomerular filtration
Metabolism
Blood flow
Bile flow
Glomerular filtration
Metabolism
NAcDCVC DCVC DCVG TCE
Blood/
Plasma
Liver
Kidney
Rest of
Body
Small
Intestine
Urine
Blood/
Plasma
Liver
Kidney
Rest of
Body
Small
Intestine
Bile
Blood/
Plasma
Liver
Rest of
Body
Bile
Blood/
Plasma
Liver
Rest of
Body
Urine
DCVT DCVCS NAcDCVCS
Kidney
Small
Intestine
Small
Intestine
Kidney

See Figure 3-5 for enzymes involved in metabolic steps. Source: Lash et al.
(2000a; 2000b); NRC (2006).

Figure 3-6. Interorgan TCE transport and metabolism via the GSH
pathway.

As discussed previously, GST activity is present in many different cell types. However,
the liver is the major tissue for GSH conjugation. GST activities in rat and mouse cytosolic
fractions were measured using 1-chloro-2,4-dinitrobenzene, a GST substrate that is nonspecific
3-46
for particular isoforms (Lash et al., 1998b). Specific activities (normalized for protein content)
in whole-kidney cytosol were slightly less than those in the liver (0.64 compared to 0.52 mU/mg
protein for males and females). However, the much larger mass of the liver compared to the
kidney indicates that far more total GST activity resides in the liver. This is consistent with in
vitro data on TCE conjugation to DCVG, discussed previously (see Tables 3-23 and 3-24). For
instance, in humans, rats, and mice, liver cytosol exhibits greater DCVG production than kidney
cytosol. Distinct high- and low-affinity metabolic profiles were observed in the liver but not in
the kidney (see Table 3-24). In microsomes, human liver and kidney had similar rates of DCVG
production, while for rats and mice, the production in the liver was substantially greater.
According to studies by Lash et al. (1998a; 1998b), the activity of GGT, the first step in the
conversion of DCVG to DCVC, is much higher in the kidney than the liver of mice, rats, and
humans, with most of the activity being concentrated in the microsomal, rather than the
cytosolic, fraction of the cell (see Table 3-26). In rats, this activity is quite high in the kidney but
is below the level of detection in the liver, while the relative kidney-to-liver levels in humans and
mice were higher by 18- and up to 2,300-fold, respectively. Similar qualitative findings were
also reported in another study (Hinchman and Ballatori, 1990) when total organ GGT levels were
compared in several species (see Table 3-27). Cysteinylglycine dipeptidase was also
preferentially higher in the kidney than the liver of all tested species although the interorgan
differences in this activity (oneninefold) seemed to be less dramatic than for GGT (see
Table 3-27). High levels of both GGT and dipeptidases have also been reported in the small
intestine of rat (Kozak and Tate, 1982) and mouse (Habib et al., 1996), as well as GGT in the
human jejunum (Fairman et al., 1977). No specific human intestinal cysteinylglycine
dipeptidase has been identified; however, a related enzyme (EC 3.4.13.11) from human kidney
microsomes has been purified and studied (Adachi et al., 1989), while several human intestinal
dipeptidases have been characterized including a membrane dipeptidase (EC 3.4.13.19), which
has a wide dipeptide substrate specificity including cysteinylglycine (Ristoff and Larsson, 2007;
Hooper et al., 1994).

3-47
Table 3-26. GGT activity in liver and kidney subcellular fractions of mice,
rats, and humans

Species Sex Tissue Cellular fraction Activity (mU/mg)
Mouse Male Liver Cytosol 0.07 0.04
Microsomes 0.05 0.04
Kidney Cytosol 1.63 0.85
Microsomes 92.6 15.6
Female Liver Cytosol 0.10 0.10
Microsomes 0.03 0.03
Kidney Cytosol 0.79 0.79
Microsomes 69.3 14.0
Rat Male Liver Cytosol <0.02
Microsomes <0.02
Kidney Cytosol <0.02
Microsomes 1,570 100
Female Liver Cytosol <0.02
Microsomes <0.02
Kidney Cytosol <0.02
Microsomes 1,840 40
Human Male Liver Cytosol 8.89 3.58
Microsomes 29
Kidney Cytosol 13.2 1.0
Microsomes 960 77

Sources: Lash et al. (1999a; 1998a)

Table 3-27. Multispecies comparison of whole-organ activity levels of GGT
and dipeptidase

Species
Whole organ enzyme activity (mol substrate/organ)
Kidney Liver
GGT Dipeptidase GGT Dipeptidase
Rat 1,010 41 20.2 1.1 7.1 1.4 6.1 0.4
Mouse 60.0 4.2 3.0 0.3 0.47 0.05 1.7 0.2
Rabbit 1,119 186 112 17 71.0 9.1 12.6 1.0
Guinea pig 148 13 77 10 46.5 4.2 13.2 1.5
Pig 3,800 769 2,428 203 1,600 255 2,178 490
Macaque 988 136 181 71

Source: Hinchman and Ballatori (1990).

3.3.3.2.7. Sex- and species-dependent differences in GSH metabolism
Diverse sex and species differences appear to exist in TCE metabolism via the GSH
pathway. In rodents, rates of TCE conjugation to GSH in male rats and mice are higher than
3-48
females (see Table 3-23). Verma and Rana (2003) reported twofold higher GST activity values
in liver cytosol of female rats, compared to males, given 15 i.p. injections of TCE over 30 days
period. This effect may be due to sex-dependent variation in induction, as GST activities in male
and female controls were similar. DCVG formation rates by liver and kidney subcellular
fractions were much higher in both sexes of mice than in rats and, except for mouse kidney
microsomes, the rates were generally higher in males than in females of the same species (see
Table 3-23).
In terms of species differences, comparisons at 12 mM TCE concentrations (see
Table 3-23) suggest that, in liver and kidney cytosol, the greatest DCVG production rate was in
humans, followed by mice and then rats. However, different investigators have reported
considerably different rates for TCE conjugation in human liver and kidney cell fractions. For
instance, values in Table 3-23 from Lash et al. (1999b) are between 2 and 5 orders of magnitude
higher than those reported by Green et al. (1997a). The rates of DCVG formation by liver
cytosol from male F344 rat, male B6C3F
1
mouse, and human were 1.62, 2.5, and 0.19 pmol/
minute/mg protein, respectively, while there was no measurable activity in liver microsomes or
subcellular kidney fractions (Green et al., 1997a). The reasons for such discrepancies are unclear
but may be related to different analytical methods employed such as detection of radiolabled
substrate vs. derivatized analytes (Lash et al., 2000a).
Expression of GGT activity does not appear to be influenced by sex (see Table 3-26); but
species differences in kidney GGT activity are notable with rat subcellular fractions exhibiting
the highest levels and mice and humans exhibiting about 46 and 50%, respectively, of rat levels
(Lash et al., 1999a; Lash et al., 1998a). Table 3-27 shows measures of whole-organ GGT and
dipeptidase activities in rats, mice, guinea pigs, rabbits, pigs, and monkeys. These data show
that the whole kidney possesses higher activities than liver for these enzymes, despite the
relatively larger mass of the liver.
As discussed above, the three potential bioactivating pathways subsequent to the
formation of DCVC are catalyzed by -lyase, FMO3, or CYP3A. Lash et al. (2000a) compared
in vitro -lyase activities and kinetic constants (when available) for kidney of rats, mice, and
humans. They reported that variability of these values spans up to two orders of magnitude
depending on substrate, analytical method used, and research group. Measurements of rat,
mouse, and human -lyase activities collected by the same researchers following tetrachloro-
ethylene exposure (Green et al., 1990) resulted in higher K
M
and lower V
MAX
values for mice and
humans than rats. Further, female rats exhibited higher K
M
and lower V
MAX
values than males.
With respect to FMO3, Ripp et al. (1999) found that this enzyme appeared catalytically
similar across multiple species, including humans, rats, dogs, and rabbits, with respect to several
substrates, including DCVC, but that there were species differences in expression. Specifically,
in male liver microsomes, rabbits had 3-fold higher methionine S-oxidase activity than mice and
dogs had 1.5-fold higher activity than humans and rats. Species differences were also noted in
3-49
male and female kidney microsomes; rats exhibited two- to sixfold higher methionine S-oxidase
activity than the other species. Krause et al. (2003) detected DCVC sulfoxidation in incubations
with human liver microsomes but did not in an incubation with a single sample of human kidney
microsomes. However, FMO3 expression in the 26 human kidney samples was found to be
highly variable, with a range of five- to sixfold (Krause et al., 2003).
No data on species differences in CYP3A-mediated sulfoxidation of NAcDCVC are
available. However, Altuntas et al. (2004) examined sulfoxidation of cysteine and mercapturic
acid conjugates of fluoromethyl-2,2-difluoro-1-(trifluoromethyl)vinyl ether (FDVE) in rat and
human liver and kidney microsomes. They reported that the formation of sulfoxides from the
mercapturates N-Ac-FFVC and (Z)-N-Ac-FFVC (FFVC is (E,Z)-S-(1-fluoro-2-fluoromethoxy-
2-(trifluoromethyl)vinyl-Lcysteine) were greatest in rat liver microsomes, and 230-fold higher
than in human liver microsomes (which had high variability). Sulfoxidation of N-Ac-FFVC
could not be detected in either rat or human kidney microsomes, but sulfoxidation of
(Z)-N-Ac-FFVC was detected in both rat and human kidney microsomes at rates comparable to
human liver microsomes. Using human- and rat-expressed CYP3A, Altuntas et al. (2004)
reported that rates of sulfoxidation of (Z)-N-Ac-FFVC were comparable in human CYP3A4 and
rat CYP3A1 and CYP3A2, but that only rat CYP3A1 and A2 catalyzed sulfoxidation of
N-Ac-FFVC. As the presence or absence of the species differences in mercapturate sulfoxidation
appears to be highly chemical-specific, no clear inferences can be made as to whether species
differences exist for sulfoxidation of NAcDCVC
Also relevant to assess the flux through the various pathways are the rates of N-acety-
lation and de-acetylation of DCVC. This is demonstrated by the results of Elfarra and Hwang
(1990) using S-(2-benzothiazolyl)-L-cysteine as a marker for -lyase metabolism in rats, mice,
hamsters, and guinea pigs. Guinea pigs exhibited about twofold greater flux through the -lyase
pathway, but this was not attributable to higher -lyase activity. Rather, guinea pigs have
relatively low N-acetylation and high deacetylation activities, leading to a high level of substrate
recirculation (Lau et al., 1995). Thus, a high N-deacetylase:N-acetylase activity ratio may favor
DCVC recirculation and subsequent metabolism to reactive species. In human, Wistar rat,
Fischer rat, and mouse cytosol, deacetylation rates for NAcDCVC varied less than threefold
(0.35, 0.41, 0.61, and 0.94 nmol DCVC formed/minute/mg protein in humans, rats, and mice)
(Birner et al., 1993). However, similar experiments have not been carried out for N-acetylation
of DCVC, so the balance between its N-acetylation and de-acetylation has not been established.

3.3.3.2.8. Human variability and susceptibility in GSH conjugation
Knowledge of human variability in metabolizing TCE through the GSH pathway is
limited to in vitro comparisons of variance in GST activity rates. Unlike CYP-mediated
oxidation, quantitative differences in the polymorphic distribution or activity levels of GST
isoforms in humans are not presently known. However, the available data (Lash et al., 1999a;
3-50
Lash et al., 1999b) do suggest that significant variation in GST-mediated conjugation of TCE
exists in humans. In particular, at a single substrate concentration of 1 mM, the rate of GSH
conjugation of TCE in human liver cytosol from 9 male and 11 females spanned a range of
2.4-fold (34.783.6 nmol DCVG formed/20-minute/mg protein) (Lash et al., 1999a). In liver
microsomes from 5 males and 15 females, the variation in activity was 6.5-fold (9.964.6 nmol
DCVG formed/20 minute/mg protein). No sex-dependent variation was identified. Despite
being less pronounced than the known variability in human CYP-mediated oxidation, the impact
on risk assessment of the variability in GSH conjugation to TCE is currently unknown especially
in the absence of data on variability for N-acetylation and bioactivation via -lyase, FMO3, or
CYP3A in the human kidney.

3.3.3.3. Relative roles of the CYP and GSH pathways
In vivo mass balance studies in rats and mice, discussed above, have shown
unequivocally that in these species, CYP oxidation of TCE predominates over GSH conjugation.
In these species, at doses of 22,000 mg/kg of [
14
C]-TCE, the sum of radioactivity in exhaled
TCE, urine, and exhaled CO
2
constitutes 6994% of the dose, with the vast majority of the
radioactivity in urine (9599%) attributable to oxidative metabolites (Dekant et al., 1986b; Green
and Prout, 1985; Prout et al., 1985; Dekant et al., 1984). The rest of the radioactivity was found
mostly in feces and the carcass. More rigorous quantitative limits on the amount of GSH
conjugation based on in vivo data such as these can be obtained using PBPK models, discussed
in Section 3.5.
Comprehensive mass-balance studies are unavailable in humans. DCVG and DCVC in
urine have not been detected in any species, while the amount of urinary NAcDCVC from
human exposures is either below detection limits or very small from a total mass balance point of
view (Bloemen et al., 2001; Lash et al., 1999b; Bernauer et al., 1996; Birner et al., 1993). For
instance, the ratio of primary oxidative metabolites (TCA + TCOH) to NAcDCVC in urine of
rats and humans exposed to 40160 ppm (215860 mg/m
3
) TCE heavily favored oxidation,
resulting in ratios of 9862,562:1 in rats and 3,2927,163:1 in humans (Bernauer et al., 1996).
Bloemen et al. (2001) reported that, at most, 0.05% of an inhaled TCE dose would be excreted as
NAcDCVC, and concluded that this suggested that TCE metabolism by GSH conjugation was of
minor importance. While it is a useful biomarker of exposure and an indicator of GSH
conjugation, NAcDCVC may capture only a small fraction of TCE flux through the GSH
conjugation pathway due to the dominance of bioactivating pathways (Lash et al., 2000a).
A number of lines of evidence suggest that the amount of TCE conjugation to GSH in
humans, while likely smaller than the amount of oxidation, may be much more substantial than
analysis of urinary mercapturates would suggest. In Table 3-28, in vitro estimates of the V
MAX
,
K
M
, and clearance (V
MAX
/K
M
) for hepatic oxidation and conjugation of TCE are compared in a
manner that accounts for differences in cytosolic and microsomal partitioning and protein
3-51
content. Surprisingly, the range of in vitro kinetic estimates for oxidation and conjugation of
TCE substantially overlap, suggesting similar flux through each pathway, though with high
interindividual variation. The microsomal and cytosolic protein measurements of GSH
conjugation should be caveated by the observation by Lash et al. (1999b) that GSH conjugation
of TCE was inhibited by ~50% in the presence of oxidation. Note that this comparison cannot be
made in rats and mice because in vitro kinetic parameters for GSH conjugation in the liver are
not available in those species (only activity at 1 or 2 mM have been measured).

Table 3-28. Comparison of hepatic in vitro oxidation and conjugation of
TCE
a


Cellular or
subcellular
fraction
V
MAX
b

(nmol TCE metabolized/min/g
tissue)
K
M
c

(M in blood)
V
MAX
/K
M

(mL/min/g tissue)
Oxidation
GSH
conjugation Oxidation
GSH
conjugation Oxidation
GSH
conjugation
Hepatocytes 10.068.4 16~25 22.1198 16~47 0.0871.12 0.55~1.0
Liver
microsomes
6.1111 45 2.6611.1* 5.9* 1.7128.2* 7.6*
71.0297** 157** 0.0641.06** 0.29**
Liver cytosol 380 4.5* 84*
22.7** 16.7**

a
When biphasic metabolism was reported, only high affinity pathway is shown here.
b
Conversion assumptions for V
MAX
: hepatocellularity of 99 million cells/g liver (Barter et al., 2007); liver
microsomal protein content of 32 mg protein/g tissue (Barter et al., 2007); and liver cytosolic protein content of 89
mg protein/g tissue (based on rats: Prasanna et al. (1989); van Bree et al. (1990).
c
Conversion assumptions for K
M
:
For hepatocytes, K
M
in headspace converted to K
M
in blood using blood:air partition coefficient of 9.5 (reported
range of measured values 6.512.1, Table 3-1);
For microsomal protein, option (*) assumes K
M
in medium is equal to K
M
in tissue, and converts to K
M
in blood by
using a liver:blood partition coefficient of 5 (reported ranges of measured values 3.65.9, Table 3-8), and option
(**) converts K
M
in medium to K
M
in air using the measured microsomal protein:air partition coefficient of 1.78
(Lipscomb et al., 1997), and then converts to K
M
in blood by using the blood:air partition coefficient of 9.5; and
For cytosolic protein, option (*) assumes K
M
in medium is equal to K
M
in tissue, and converts to K
M
in blood by
using a liver:blood partition coefficient of 5 (reported ranges of measured values 3.65.9, Table 3-8), and option
(**) assumes K
M
in medium is equal to K
M
in blood, so no conversion is necessary.

Furthermore, as shown earlier in Table 3-22, the human in vivo data of Lash et al.
(1999b) show blood concentrations of DCVG similar, on a molar basis, to those of TCE, TCA,
or TCOH, suggesting substantial conjugation of TCE. In addition, these data give a lower limit
as to the amount of TCE conjugated. In particular, by multiplying the peak blood concentration
of DCVG by the blood volume, a minimum amount of DCVG in the body at that time can be
derived (i.e., assuming the minimal empirical distribution volume equal to the blood volume).
As shown in Table 3-29, this lower limit amounts to about 0.43.7% of the inhaled TCE dose.
Since this is the minimum amount of DCVG in the body at a single time point, the total amount
of DCVG formed is likely to be substantially greater, owing to possible distribution outside of
3-52
the blood as well as the metabolism and/or excretion of DCVG. Lash et al. (1999b) found that
levels of urinary mercapturates were near or below the level of detection of 0.19 M, results that
are consistent with those of Bloemen et al. (2001), who reported urinary concentrations below
0.04 M at two- to fourfold lower cumulative exposures. Taken together, these results confirm
the suggestion by Lash et al. (2000a) that NAcDCVC is a poor quantitative marker for the flux
through the GSH pathway.

Table 3-29. Estimates of DCVG in blood relative to inhaled TCE dose in
humans exposed to 50 and 100 ppm (269 and 537 mg/m
3
) (Lash et al., 1999b)

Sex exposure
Estimated inhaled TCE dose
(mmol)
a

Estimated peak amount of DCVG in blood
(mmol)
b

Males
50 ppm 4 hrs 3.53 0.11 0.08
100 ppm 4 hrs 7.07 0.26 0.08
Females
50 ppm 4 hrs 2.36 0.010 0
100 ppm 4 hrs 4.71 0.055 0.027

a
Inhaled dose estimated by (50 or 100 ppm)/(24,450 ppm/mM) (240 minutes) Q
P
, where alveolar ventilation rate
Q
P
is 7.2 L/minute for males and 4.8 L/minute for females. Q
P
is calculated as (V
T
V
D
) f
R
with the following
respiratory parameters: tidal volume V
T
(0.75 L for males, 0.46 L for females), dead space V
D
(0.15 L for males,
0.12 L for females), and respiration frequency f
R
(12 minutes
-1
for males, 14 minutes
-1
for females) [assumed sitting,
awake from The International Commission on Radiological Protection (ICRP, 2003)].
b
Peak amount of DCVG in blood estimated by multiplying the peak blood concentration by the estimated blood
volume: 5.6 L in males and 4.1 L in females (ICRP, 2003).

Sources: Fisher et al. (1998); Lash et al. (1999b).

However, as discussed in Section 3.3.3.2.1, data from other laboratories have reported
substantially lower amounts of GSH conjugation in vitro. The reasons for such discrepancies are
unclear, but they may be related to different analytical methods (Lash et al., 2000a). More recent
in vivo data from Kim et al. (2009) in mice reported ~1,000 times lower DCVG in mouse serum
as compared to the levels of DCVG reported by Lash et al. (1999b) in human blood. These data
are consistent with the suggestion that the Reed method employed by Lash et al. (1999b)
overestimated DCVG levels in humans. However, the degree of overestimation is unclear, as is
the degree to which differences may be attributable to true inter-species or inter-individual
variability.
In summary, TCE oxidation is likely to be greater quantitatively than conjugation with
GSH in mice, rats, and humans. Some evidence suggests that the flux through the GSH pathway,
particularly in humans, may be greater by an order of magnitude or more than the <0.1%
typically excreted of NAcDCVC in urine. This is evidenced both by a direct comparison of in
vitro rates of oxidation and conjugation, as well as by in vivo data on the amount of DCVG in
3-53
blood. PBPK models can be used to more quantitatively synthesize these data and put more
rigorous limits on the relative amounts of TCE oxidation and conjugation with GSH. Such
analyses are discussed in Section 3.5. However, these data are not consistent with studies in
other laboratories using different analytical methods, which report 25 orders of magnitude
lower estimates of GSH conjugation. Because the reason for these differences have not been
fully determined, substantial uncertainty remains in the degree of GSH conjugation, particularly
in humans.

3.4. TCE EXCRETION
This section discusses the major routes of excretion of TCE and its metabolites in exhaled
air, urine, and feces. Unmetabolized TCE is eliminated primarily via exhaled air. As discussed
in Section 3.3, the majority of TCE absorbed into the body is eliminated by metabolism. With
the exception of CO
2
, which is eliminated solely via exhalation, most TCE metabolites have low
volatility and, therefore, are excreted primarily in urine and feces. Although trace amounts of
TCE metabolites have also been detected in sweat and saliva (Bartonicek, 1962), these excretion
routes are likely to be relatively minor.

3.4.1. Exhaled Air
In humans, pulmonary elimination of unchanged TCE and other volatile compounds is
related to ventilation rate, cardiac output, and the solubility of the compound in blood and tissue,
which contribute to final exhaled air concentration of TCE. In their study of the impact of
workload on TCE absorption and elimination, Astrand and Ovrum (1976) characterized the
postexposure elimination of TCE in expired breath. TCE exposure (540 or 1,080 mg/m
3
; 100 or
200 ppm) was for a total of 2 hours, at workloads of 0150 watts. Elimination profiles were
roughly equivalent among groups, demonstrating a rapid decline in TCE concentrations in
expired breath postexposure (see Table 3-30).

Table 3-30. Concentrations of TCE in expired breath from inhalation-
exposed humans (Astrand, 1982)

Time postexposure
Alveolar air
I
a
II III
0 min 459 44 244 16 651 53
30 min 70 5 51 3 105 18
60 min 40 4 28 2 69 8
90 min 35 9 21 1 55 2
120 min 31 8 16 1 45 1
300 min 8 1 9 2 14 2
420 min 5 0.5 4 0.5 8 1.3
19 hrs 2 0.3 2 0.2 4 0.5

a
Roman numerals refer to groups assigned different workloads; concentrations are in mg/m
3
for expired air.
3-54

The lung clearance of TCE represents the volume of air from which all TCE can be
removed per unit time, and is a measure of the rate of excretion via the lungs. Monster et al.
(1976) reported lung clearances ranging from 3.8 to 4.9 L/minute in four adults exposed at rest to
70 and 140 ppm of TCE for 4 hours. Pulmonary ventilation rates in these individuals at rest
ranged from 7.7 to 12.3 L/minute. During exercise, when ventilation rates increased to 29
30 L/minute, lung clearance was correspondingly higher, 7.712.3 L/minute. Under single and
repeated exposure conditions, Monster et al. (1979; 1976) reported that 717% of absorbed TCE
was excreted in exhaled breath. Pulmonary elimination of unchanged TCE at the end of
exposure is a first-order diffusion process across the lungs from blood into alveolar air, and it can
be thought of as the reversed equivalent of its uptake from the lungs. Exhaled pulmonary
excretion occurs in several distinct (delayed) phases corresponding to release from different
tissue groups, at different times. Sato et al. (1977) detected three first-order phases of pulmonary
excretion in the first 10 hours after exposure to 100 ppm for 4 hours, with fitted half-times of
pulmonary elimination of 0.04, 0.67, and 5.6 hours, respectively. Opdam (1989) sampled
alveolar air up to 20310 hours after 2962-minute exposures to 638 ppm, and reported
terminal half-lives of 844 hours at rest. Chiu et al. (2007) sampled alveolar air up to 100 hours
after 6-hour exposures to 1 ppm and reported terminal half-lives of 1423 hours. The long
terminal half-time of TCE pulmonary excretion indicates that considerable time is necessary to
completely eliminate the compound, primarily due to the high partitioning to adipose tissues (see
Section 3.2).
As discussed above, several studies (Green and Prout, 1985; Prout et al., 1985; Dekant et
al., 1984) have investigated the disposition of [
14
C]-TCE in rats and mice following gavage
administrations (see Section 3.3.2). These studies have reported CO
2
as an exhalation excretion
product in addition to unchanged TCE. With low doses, the amount of TCE excreted unchanged
in exhaled breath is relatively low. With increasing dose in rats, a disproportionately increased
amount of radiolabel is expired as unchanged TCE. This may indicate saturation of metabolic
activities in rats at doses 200 mg/kg, which is perhaps only minimally apparent in the data from
mice. In addition, exhaled air TCE concentration has been measured after constant inhalation
exposure for 2 hours to 50 or 500 ppm in rats (Dallas et al., 1991), and after dermal exposure in
rats and humans (Poet et al., 2000). Exhaled TCE data from rodents and humans have been
integrated into the PBPK model presented in Section 3.5.
Finally, TCOH is also excreted in exhaled breath, though at a rate about 10,000-fold
lower than unmetabolized TCE (Monster, 1979; Monster et al., 1976).

3.4.2. Urine
Urinary excretion after TCE exposure consists predominantly of the metabolites, TCA
and TCOH, with minor contributions from other oxidative metabolites and GSH conjugates.
3-55
Measurements of unchanged TCE in urine have been at or below detection limits (e.g., Chiu et
al., 2007; Fisher et al., 1998). The recovery of urinary oxidative metabolites in mice, rats, and
humans was addressed earlier (see Section 3.3.2) and will not be discussed here. Because of
their relatively long elimination half-life, urinary oxidative metabolites have been used as an
occupational biomarker of TCE exposure for many decades (Carrieri et al., 2007; Ikeda and
Imamura, 1973). Ikeda and Imamura (1973) measured TTCs, TCOH, and TCA in urine over
3 consecutive postexposure days for four exposure groups totaling 24 adult males and one
exposure group comprising 6 adult females. The elimination half-lives for TTC were 26.1
48.8 hours in males and 50.7 hours in females. The elimination half-lives for TCOH were
15.3 hours in the only group of males studied and 42.7 hours in females. The elimination half-
lives for TCA were 39.7 hours in the only group of males studied and 57.6 hours in females.
These authors compared their results to previously published elimination half-lives for TTC,
TCOH, and TCA. Following experimental exposures of groups of twofive adults, elimination
half-lives were 3150 hours for TTC, 1929 hours for TCOH, and 3655 hours for TCA
(Nomiyama and Nomiyama, 1971; Ogata et al., 1971; Stewart et al., 1970; Bartonicek, 1962).
The urinary elimination half-lives of TCE metabolites in a subject who worked with and was
addicted to sniffing TCE for 68 years approximated 49.7 hours for TCOH, 72.6 hours for TCA,
and 72.6 hours for TTC (Ikeda et al., 1971).
The quantitative relationship between urinary concentrations of oxidative metabolites and
exposure in an occupational setting was investigated by Ikeda (1977). This study examined the
urinary elimination of TCE and metabolites in urine of 51 workers from 10 workshops. The
concentration of TCA and TCOH in urine demonstrated a marked concentration-dependence,
with concentrations of TCOH being approximately twice as high as those for TCA. Urinary
half-life values were calculated for six males and six females from five workshops; males were
intermittently exposed to 200 ppm and females were intermittently exposed to 50 ppm
(269 mg/m
3
). Urinary elimination half-lives for TTC, TCOH, and TCA were 26.1, 15.3, and
39.7 hours in males, respectively, and 50.7, 42.7 and 57.6 hours in females, respectively, which
were similar to the range of values previously reported. These authors estimated that urinary
elimination of parent TCE during exposure might account for one-third of the systemically
absorbed dose. Importantly, urinary TCA exhibited marked saturation at exposures >50 ppm.
Because neither TTC nor urinary TCOH (in the form of the glucuronide TCOG) showed such an
effect, this saturation cannot be due to TCE oxidation itself, but must rather be from one of the
metabolic processes forming TCA from TCOH. Unfortunately, since biological monitoring
programs usually measure only urinary TCA, rather than TTC, urinary TCA levels above around
150 mg/L cannot distinguish between exposures at 50 ppm and at much higher concentrations.
It is interesting to attempt to extrapolate on a cumulative exposure basis the Ikeda (1977)
results for urinary metabolites obtained after occupational exposures at 50 ppm to the controlled
exposure study by Chiu et al. (2007) at 1.2 ppm for 6 hours (the only controlled exposure study
3-56
for which urinary concentrations, rather than only cumulative excretion, are available). Ikeda
(1977) reported that measurements were made during the second half of the week, so one can
postulate a cumulative exposure duration of 20~40 hours. At 50 ppm, Ikeda (1977) report a
urinary TCOH concentration of about 290 mg/L, so that per ppm-hour, the expected urinary
concentration would be 290/(50 20 ~ 40) = 0.145 ~ 0.29 mg/L-ppm-hour. The cumulative
exposure in Chiu et al. (2007) is 1.2 6 = 7.2 ppm-hour, so the expected urinary TCOH
concentration would be 7.2 (0.145 ~ 0.29) = 1.0 ~ 2.1 mg/L. This estimate is somewhat
surprisingly consistent with the actual measurements of Chiu et al. (2007) during the first day
postexposure, which ranged from 0.8 to ~1.2 mg/L TCOH in urine.
On the other hand, extrapolation of TCA concentrations was less consistent. At 50 ppm,
Ikeda (1977) report a urinary TCA concentration of about 140 mg/L, so that per ppm-hour, the
expected urinary concentration would be 140/(50 20 ~ 40) = 0.07 ~ 0.14 mg/L-ppm-hour. The
cumulative exposure in Chiu et al. (2007) is 1.2 6 = 7.2 ppm-hour, so the expected urinary
TCA concentration would be 7.2 (0.07 ~ 0.14) = 0.5 ~ 1.0 mg/L, whereas Chiu et al. (2007)
reported urinary TCA concentrations on the first day after exposure of 0.03~0.12 mg/L.
However, as noted in Chiu et al. (2007), relative urinary excretion of TCA was 310-fold lower
in Chiu et al. (2007) than other studies at exposures of 50~140 ppm, which may explain part of
the discrepancies. However, this may be due, in part, to saturation of many urinary TCA
measurements, and, furthermore, interindividual variance, observed to be substantial in Fisher
et al. (1998), cannot be ruled out.
Urinary elimination kinetics have been reported to be much faster in rodents than in
humans. For instance, adult rats were exposed to 50, 100, or 250 ppm (269, 537, or
1,344 mg/m
3
) via inhalation for 8 hours or were administered an i.p. injection (1.47 g/kg) and the
urinary elimination of TTCs was followed for several days (Ikeda and Imamura, 1973). These
authors calculated urinary elimination half-lives of 14.315.6 hours for female rats and 15.5
16.6 hours for male rats; the route of administration did not appear to influence half-life value.
In other rodent experiments using orally administered radiolabeled TCE, urinary elimination was
complete within 1 or 2 days after exposure (Green and Prout, 1985; Prout et al., 1985; Dekant et
al., 1984).

3.4.3. Feces
Fecal elimination accounts for a small percentage of TCE as shown by limited
information in the available literature. Bartonicek (1962) exposed seven volunteers to 1.042 mg
TCE/L air for 5 hours and examined TCOH and TCA in feces on the 3
rd
and 7
th
day following
exposure. The mean amount of TCE retained during exposure was 1,107 mg, representing 51
64% (mean 58%) of administered dose. On the 3
rd
day following TCE exposure, TCOH and
TCA in feces demonstrated mean concentrations of 17.1 and 18.5 mg/100 g feces, similar to
concentrations in urine. However, because of the 10-fold smaller daily rate of excretion of feces
3-57
relative to urine, this indicates fecal excretion of these metabolites is much less significant than
urinary excretion. Neither TCOH nor TCA was detected in feces on the 7
th
day following
exposure.
In rats and mice, total radioactivity has been used to measure excretion in feces after
gavage TCE administration in corn oil, but since the radiolabel was not characterized, it is not
possible to determine whether the radiolabel in feces represented unabsorbed parent compound,
excreted parent compound, and/or excreted metabolites. Dekant et al. (1984) reported that mice
eliminated 5% of the total administered TCE, while rats eliminated 2% after gavage. Dekant
et al. (1986b) reported a dose-response-related increase in fecal elimination with dose, ranging
between 0.8 and 1.9% in rats and between 1.6 and 5% in mice after gavage in corn oil. Due to
the relevant role of CYP2E1 in the metabolism of TCE (see Section 3.3.3.1.6), Kim and
Ghanayem (2006) compared fecal elimination in both wild-type and CYP2E1 knockout mice and
reported fecal elimination ranging between 4.1 and 5.2% in wild-type and between 2.1 and 3.8%
in knockout mice exposed by gavage in aqueous solution.

3.5. PBPK MODELING OF TCE AND ITS METABOLITES
3.5.1. Introduction
PBPK models are extremely useful tools for quantifying the relationship between
external measures of exposure and internal measures of toxicologically relevant dose. In
particular, for the purposes of this assessment, PBPK models are evaluated for the following:
(1) providing additional quantitative insights into the ADME of TCE and metabolites described
in the sections above; (2) cross-species pharmacokinetic extrapolation of rodent studies of both
cancer and noncancer effects; (3) exposure-route extrapolation; and (4) characterization of
human pharmacokinetic variability. The following sections first describe and evaluate previous
and current TCE PBPK modeling efforts, then discuss the insights into ADME (1, above), and
finally present conclusions as to the utility of the model to predict internal doses for use in dose-
response assessment (24, above).

3.5.2. Previous PBPK Modeling of TCE for Risk Assessment Application
TCE has an extensive number of both in vivo pharmacokinetic and PBPK modeling
studies [see Chiu et al. (2006b) supplementary material, for a review]. Models previously
developed for occupational or industrial hygiene applications are not discussed here but are
reviewed briefly in Clewell et al. (2000). Models designed for risk assessment applications have
focused on descriptions of TCE and its major oxidative metabolites, TCA, TCOH, and TCOG.
Most of these models were extensions of the first generation of models developed by Fisher
and coworkers (Allen and Fisher, 1993; Fisher et al., 1991) in rats, mice, and humans. These
models, in turn, are based on a Ramsey and Andersen (1984) structure with flow-limited tissue
compartments and equilibrium gas exchange, saturable Michaelis-Menten kinetics for oxidative
3-58
metabolism, and lumped volumes for the major circulating oxidative metabolites TCA and
TCOH. Fisher and coworkers updated their models with new in vivo and in vitro experiments
performed in mice (Greenberg et al., 1999; Abbas and Fisher, 1997) and volunteers (Fisher et al.,
1998) and summarized their findings in Fisher (2000). Clewell et al. (2000) added enterohepatic
recirculation of TCOG and pathways for local oxidative metabolism in the lung and GST
metabolism in the liver. While Clewell et al. (2000) does not include the updated Fisher (2000)
data, they have used a wider set of in vivo and in vitro mouse, rat, and human data than previous
models. Finally, Bois (2000a, b) performed reestimations of PBPK model parameters for the
Fisher and Clewell models using a Bayesian population approach [Gelman (1996), and discussed
further below].
As discussed in Rhomberg (2000), the choice as to whether to use the Fisher, Clewell,
and/or Bois models for cross-species extrapolation of rodent cancer bioassays led to quantitative
results that differed by as much as an order of magnitude. There are a number of differences in
modeling approaches that can explain their differing results. First, the Clewell et al. (2000)
model differed structurally in its use of single-compartment volume-of-distribution models for
metabolites as opposed to the Fisher (Fisher, 2000) models, which use multiple physiologic
compartments. Also, the Clewell et al. (2000) model, but not the Fisher models, includes
enterohepatic recirculation of TCOH/TCOG (although reabsorption was set to zero in some
cases). In addition to structural differences in the models, the input parameter values for these
various models were calibrated using different subsets of the overall in vivo database [see Chiu
et al. (2006b), supplementary material, for a review]. The Clewell et al. (2000) model is based
primarily on a variety of data published before 1995; the Fisher (2000) models were based
primarily on new studies conducted by Fisher and coworkers (after 1997); and the Bois (2000a,
b) reestimations of the parameters for the Clewell et al. (2000) and Fisher (2000) models used
slightly different data sets than the original authors. The Bois (2000a, b) reanalyses also led to
somewhat different parameter estimates than the original authors, both because of the different
data sets used as well as because the methodology used by Bois allowed many more parameters
to be estimated simultaneously than were estimated in the original analyses.
Given all of these methodological differences, it is not altogether surprising that the
different models led to different quantitative results. Even among the Fisher models themselves,
Fisher (2000) noted some inconsistencies, including differing estimates for metabolic parameters
between mouse gavage and inhalation experiments. These authors included possible
explanations for these inconsistencies: the impact of corn oil vehicle use during gavage (Staats et
al., 1991) and the impact of a decrease in ventilation rate in mice due to sensory irritation during
the inhalation of solvents [e.g., Stadler and Kennedy (1996)].
As discussed in a report by the National Research Council (NRC, 2006), several
additional PBPK models relevant to TCE pharmacokinetics have been published since 2000 and
are reviewed briefly here. Poet et al. (2000) incorporated dermal exposure to TCE in PBPK
3-59
models in rats and humans, and published in vivo data in both species from dermal exposure
(Poet et al., 2000; Thrall and Poet, 2000). Albanese et al. (2002) published a series of models
with more complex descriptions of TCE distribution in adipose tissue but did not show
comparisons with experimental data. Simmons et al. (2002) developed a PBPK model for TCE
in the Long-Evans rat that focused on neurotoxicity endpoints and compared model predictions
with experimentally determined TCE concentrations in several tissues, including the brain. Keys
et al. (2003) investigated the lumping and unlumping of various tissue compartments in a series
of PBPK models in the rat and compared model predictions with TCE tissue concentrations in a
multitude of tissues. Although none of these TCE models included metabolite descriptions, the
experimental data were available for either model or evaluation. Finally, Keys et al. (2004)
developed a model for DCA in the mouse that included a description of suicide inhibition of
GST-zeta, but this model was not been linked to TCE.

3.5.3. Development and Evaluation of an Interim Harmonized TCE PBPK Model
Throughout 2004, EPA and the U.S. Air Force jointly sponsored an integration of the
Fisher, Clewell, and Bois modeling efforts (Hack et al., 2006). In brief, a single interim PBPK
model structure combining features from both the Fisher and Clewell models was developed and
used for all three species of interest (mice, rats, and humans). An effort was made to combine
structures in as simple a manner as possible; the evaluation of most alternative structures was left
for future work. The one level of increased complexity introduced was inclusion of species- and
dose-dependent TCA plasma binding, although only a single in vitro study of Lumpkin et al.
(2003) was used as parameter inputs. As part of this joint effort, a hierarchical Bayesian
population analysis using Markov chain Monte Carlo (MCMC) sampling [similar to the Bois
(2000a, b) analyses] was performed on the revised model with a cross-section of the combined
database of kinetic data to provide estimates of parameter uncertainty and variability (Hack et al.,
2006). Particular attention was given to using data from each of the different efforts, but owing
to time and resource constraints, a combined analysis of all data was not performed. The results
from this effort suggested that a single model structure could provide reasonable fits to a variety
of data evaluated for TCE and its major oxidative metabolites TCA, TCOH, and TCOG.
However, in many cases, different parameter valuesparticularly for metabolismwere
required for different studies, indicating significant interindividual or interexperimental
variability. In addition, these authors concluded that dosimetry of DCA, conjugative
metabolites, and metabolism in the lung remained highly uncertain (Hack et al., 2006).
Subsequently, EPA conducted a detailed evaluation of the Hack et al. (2006) model that
included: (1) additional model runs to improve convergence; (2) evaluation of posterior
distributions for population parameters; and (3) comparison of model predictions both with the
data used in the Hack et al. (2006) analysis as well as with additional data sets identified in the
3-60
literature. Appendix A provides the details and conclusions of this evaluation, briefly
summarized in Table 3-31, along with their pharmacokinetic implications.

3.5.4. PBPK Model for TCE and Metabolites Used for This Assessment
3.5.4.1. Introduction
Based on the recommendations of the NRC (2006) as well as additional analysis and
evaluation of the Hack et al. (2006) PBPK model, an updated PBPK model for TCE and
metabolites was developed for use in this risk assessment. The updated model is reported in
Evans et al. (2009) and Chiu et al. (2009), and the discussion below provides some details in
additional to the information in the published articles.
This updated model included modification of some aspects of the Hack et al. (2006)
PBPK model structure, incorporation of additional in vitro and in vivo data for estimating model
parameters, and an updated hierarchical Bayesian population analysis of PBPK model
uncertainty and variability. In the subsections below, the updated PBPK model and baseline
parameter values are described, as well as the approach and results of the analysis of PBPK
model uncertainty and variability. Appendix A provides more detailed descriptions of the model
and parameters, including background on hierarchical Bayesian analyses, model equations,
statistical distributions for parameter uncertainty and variability, data sources for these parameter
values, and the PBPK model code. Additional computer codes containing input files to the
MCSim program are available electronically.

3-61
Table 3-31. Conclusions from evaluation of Hack et al. (2006), and implications for PBPK model development

Conclusion from evaluation of Hack et al. (2006) model Implications for PBPK model parameters, structure, or data
For some model parameters, posterior distributions were somewhat inconsistent
with the prior distributions.
- For parameters with strongly informative priors (e.g., tissue volumes and
flows), this may indicate errors in the model.
- For many parameters, the prior distributions were based on visual fits to the
same data. If the posteriors are inconsistent, then the priors were
inappropriately informative, and, thus, the same data were used twice.
Reevaluation of all prior distributions.
- Update priors for parameters with independent data (physiological
parameters, partition coefficients, in vitro metabolism), looking across all
available data sets.
- For priors without independent data (e.g., many metabolism parameters),
use less informative priors (e.g., log-uniform distributions with wide
bounds) to prevent bias.
Evaluate modifications to the model structure, as discussed below.
A number of data sets involve TCE (i.a., portal vein), TCA (oral, i.v.), and
TCOH (oral, i.v.) dosing routes that are not currently in the model, but could be
useful for calibration.
- Additional dosing routes can be added easily.
TCE concentrations in blood, air, and tissues well-predicted only in rats, not in
mice and humans. Specifically:
- In mice, the oral uptake model could not account for the time-course of
several data sets. Blood TCE concentrations after inhalation were
consistently overpredicted.
- In rats, tissue concentrations measured in data not used for calibration were
accurately predicted.
- In humans, blood and air TCE concentrations were consistently
overpredicted in the majority of (but not all) data sets.
- In mice, uptake from the stomach compartment (currently zero), but
previously included in Abbas and Fisher (1997), may improve the model
fit.
- In mice and humans, additional extrahepatic metabolism, either
presystemic (e.g., in the lung) or postsystemic (e.g., in the kidney) and/or
a wash-in/wash-out effect may improve the model fit.
Total metabolism appears well-predicted in rats and mice based on closed-
chamber data, but required significantly different V
MAX
values between dose
groups. Total recovery in humans (6070%) is less than the model would
predict. In all three species, the ultimate disposition of metabolism is uncertain.
In particular, there are uncertainties in attributing the missing metabolism to
- GSH pathway (e.g., urinary mercapturates may only capture a fraction of the
total flux; moreover, in Bernauer et al. (1996), excretion was still on-going
at end of collection period; model does not accurately depict time-course of
mercapturate excretion).
- Other hepatic oxidation (currently attributed to DCA).
- Extrahepatic systemic metabolism (e.g., kidney).
- Presystemic metabolism in the lung.
- Additional metabolism of TCOH or TCA (see below).
- Calibration of GSH pathway may be improved by utilizing in vitro data
on liver and kidney GSH metabolism, adding a DCVG compartment to
improve the prediction of the time-course for mercapturate excretion,
and/or using the Lash et al. (1999b) blood DCVG in humans
(necessitating the addition of a DCVG compartment).
- Presystemic lung metabolism can only be evaluated if added to the
model (in vitro data exist to estimate the V
MAX
for such metabolism). In
addition, a wash-in/wash-out effect (e.g., suggested by Greenberg et al.,
(1999) can be evaluated using a continuous breathing model that
separately tracks inhaled and exhaled air, with adsorption/desporption in
the respiratory tract.
- Additional elimination pathways for TCOH and TCA can be added for
evaluation.

3-62
Table 3-31. Conclusions from evaluation of Hack et al. (2006), and implications for PBPK model development
(continued)

Conclusion from evaluation of Hack et al. (2006) model Implications for PBPK model parameters, structure, or data
TCA blood/plasma concentrations were well-predicted following TCE
exposures in all species. However, there may be inaccuracies in the total flux of
TCA production, as well as its disposition.
- In TCA dosing studies, the majority (>50%), but substantially <100%, was
recovered in urine, suggesting significant metabolism of TCA. Although
urinary TCA was well-predicted in mice and humans (but not in rats), if
TCA metabolism is significant, then the current model underestimates the
flux of TCE metabolism to TCA.
- An improved TCOH/TCOG model may also provide better estimates of
TCA kinetics (see below).
TCOH/TCOG concentrations and excretion were inconsistently predicted,
particularly after TCOH dosing.
- In mice and rats, first-order clearance for TCOH glucuronidation was
predicted to be greater than hepatic blood flow, which is consistent with a
first-pass effect that is not currently accounted for.
- In humans, the estimated clearance rate for TCOH glucuronidation was
substantially smaller than hepatic blood flow. However, the presence of
substantial TCOG in blood (as opposed to free TCOH) in the Chiu et al.
(2007) data are consistent with greater glucuronidation than predicted by
the model.
- In TCOH dosing studies, substantially <100% was recovered in urine as
TCOG and TCA, suggesting another metabolism or elimination pathway.
- Additional elimination pathways for TCOH and TCA can be added for
evaluation.
- The addition of a liver compartment for TCOH and TCOG would
permit hepatic first-pass effects to be accounted for, as appears
necessary for mice and rats.

i.v. = intravenous
3-63

3.5.4.2. Updated PBPK Model Structure
The updated TCE PBPK model is illustrated in Figure 3-7, with detailed descriptions of
the model structure, equations, and parameters found in Appendix A (see Section A.4), and the
major changes from the Hack et al. (2006) model described here. The TCE submodel was
augmented by the addition of kidney and venous blood compartments, and an updated
respiratory tract model that included both metabolism and the possibility of local storage in the
respiratory tissue. In particular, in the updated lung, separate processes describing inhalation and
exhalation allowed for adsorption and desorption from tracheobronchial epithelium (wash-in/
wash-out), with the possibility of local metabolism as well. In addition, conjugative metabolism
in the kidney was added, motivated by the in vitro data on TCE conjugation described in
Sections 3.3.3.23.3.3.3. With respect to oxidation, a portion of the lung metabolism was
assumed to produce systemically available oxidative metabolites, including TCOH and TCA,
with the remaining fraction assumed to be locally cleared. This is clearly a lumping of a
multistep process, but the lack of data precludes the development of a more sequential model.
TCE oxidation in the kidney was not included because it was not likely to constitute a substantial
flux of total TCE oxidation given the much lower CYP activity in the kidney relative to the liver
(Cummings and Lash, 2000; Cummings et al., 1999) and the greater tissue mass of the liver.
2
In
addition, liver compartments were added to the TCOH and TCOG submodels to account
properly for first-pass hepatic metabolism, which is important for consistency across routes of
exposure. Furthermore, additional clearance pathways of TCOH and TCA were added to their
respective submodels. With respect to TCE conjugation, in humans, an additional DCVG
compartment was added between TCE conjugation and production of DCVC. In addition, it
should be noted that the urinary clearance of DCVC represents a lumping of N-acetylation of
DCVC, deacetylation of NAcDCVC, and urinary excretion NAcDCVC, and that the
bioactivation of DCVC represents a lumping of thiol production from DCVC by beta-lyase,
sulfoxidation of DCVC by FMO3, and sulfoxidation of NAcDCVC by CYP3A. Such lumping
was used because these processes are not individually identifiable given the available data.

2
The extraction ratio for kidney oxidation is likely to be very low, as shown by the following calculation in rats and
humans. In rats, the in vitro kidney oxidative clearance (V
MAX
/K
M
) rate (see Table 3-13, converting units) is
1.64 10
-7
L/minutes/mg microsomal protein. Converting units using 16 mg microsomal protein to g tissue (Bong
et al., 1985) gives a clearance rate per unit tissue mass of 2.6 10
-6
L/minutes/g kidney. This is more than
1,000-fold smaller than the kidney specific blood flow rate of 6.3 10
-3
L/minutes/g kidney (Brown et al., 1997). In
humans, an in vitro clearance rate of 6.5 10
-8
L/minutes/mg microsomal protein is derived from the only detectable
in vitro oxidation rate from Cummings and Lash (2000) of 0.13 nmol/minutes/mg protein at 2 mM. Using the same
conversion from microsomal protein to tissue mass gives a clearance rate of 1.0 10
-6
L/minutes/g kidney, more
than 1,000-fold smaller than the kidney specific blood flow of 3.25 10
-3
L/minutes/g kidney (Brown et al., 1997).
No data on kidney metabolism are available in mice, but the results are likely to be similar. Therefore, even
accounting for uncertainties of up to an order of magnitude in the in-vitro-to-in-vivo conversion, kidney oxidation
should contribute negligibly to total metabolism of TCE.
3-64



Conjugative Metabolism
(rat and human only)
Oxidative Metabolism
TCA
TCOG
TCOH TCE
Respiratory
Tract Tissue
Gas Exchange
Respiratory
Tract Lumen
(Inhalation)
Respiratory
Tract Lumen
(Exhalation)
Venous
Blood
Rapidly
Perfused
Slowly
Perfused
Fat
Gut
Liver
Kidney
Oxidation &
Conjugation
Conjugation
Oxidation
Body
Liver
Blood
Body
Liver
Plasma
Body
Liver
Blood
TCA
Bile
Urine
TCOG
Other
TCE
TCOG
TCOH
TCOH
TCE
TCOH
Other
Urine
Total
Systemic
Oxidation
Lung
Oxidation
Local
Clearance
Liver
Oxidation
TCOH
TCA
Other
Liver
Conjugation
Kidney
Conjugation
DCVG
(human only)
DCVC
Urine
(NAcDCVC)
Bio-
activation
(Dead space)
Stomach
Duodenum
Oral
IV
IA
PV
Inhaled air Exhaled air
IV
Oral
IV
Oral
Input (exposure/dose)
Dynamic Compartment (solved by ODEs)
Transformation or Excretion
Static Compartment (at local steady-state)
Legend
Boxes with underlined labels are additions or modifications of the Hack et al.
(2006) model, which are discussed in Table 3-32.
Figure 3-7. Overall structure of PBPK model for TCE and metabolites used
in this assessment.

3-65
Table 3-32. Discussion of changes to the Hack et al. (2006) PBPK model
implemented for this assessment

Change to Hack et al.
(2006) PBPK model Discussion
TCE respiratory tract
compartments and
metabolism
In vitro data indicate that the lung (at least in the mouse) has a significant capacity for
oxidizing TCE. However, in the Hack et al. (2006) model, respiratory metabolism was
blood flow-limited. The model structure used was inconsistent with other PBPK
models in which the same mechanism for respiratory metabolism is assumed (e.g.,
styrene, Sarangapani et al. (2003). In these models, the main source of exposure in the
respiratory tract tissue is from the respiratory lumennot from the tracheobronchial
blood flow. In addition, a wash-in/wash-out effect has also been postulated. The
current structure, which invokes a continuous breathing model with separate
inhaled and exhaled respiratory lumens, can accommodate both respiratory
metabolism due to exposure from the respiratory lumen as well as a wash-in/wash-out
effect in which there is temporary storage in the respiratory tract tissue. Moreover,
preliminary analyses indicated that these changes to the model structure allowed for a
substantially better fit to mouse closed-chamber data under the requirement that all of
the dose levels are modeled using the same set of parameters.
TCE kidney compartment In vitro data indicate that the kidney has a significant capacity for conjugating TCE
with GSH.
TCE venous blood
compartment
Many PBPK models have used a separate blood compartment. It was believed to be
potentially important and feasible to implement here because: (1) TCE blood
concentrations were often not well-predicted by the Hack et al. (2006) model; (2) the
TCA submodel has a plasma compartment, which is a fraction of the blood volume
based on the blood volume; (3) adequate independent information on blood volume is
available; and (4) the updated model was to include the i.v. route of exposure.
TCOH and TCOG liver
compartments
In mice and rats, the Hack et al. (2006) model estimated a rate of TCOH
glucuronidation that exceeded hepatic blood flow (all glucuronidation is assumed to
occur in the liver), which indicated a significant first-pass effect. Therefore, a separate
liver compartment is necessary to account properly for hepatic first-pass.
TCOH and TCA other
elimination pathways
Mass-balance studies with TCOH and TCA dosing indicated that, although the majority
of TCOH and TCA are excreted in urine, the amount is still substantially <100%.
Therefore, additional elimination of TCOH and TCA must exist and should be
accounted for.
DCVG compartment
(human model only)
Blood DCVG data in humans exist as part of the Fisher et al. (1998) experiments,
reported in Lash et al. (1999b), and a DCVG compartment is necessary in order to
utilize those data.

3.5.4.3. Specification of Baseline PBPK Model Parameter
Point estimates for PBPK model parameters (baseline values), used as central estimates
in the prior distributions for population mean parameters in the hierarchical Bayesian statistical
model (see Appendix A), were developed using standard methodologies to ensure biological
plausibility, and were a refinement of those used in Hack et al. (2006). Because the Bayesian
parameter estimation methodology utilizes the majority of the useable in vivo data on TCE
pharmacokinetics, all baseline parameter estimates were based solely on measurements
independent of the in vivo data. This avoids using the same data in both the prior and the
likelihood. These parameters were, in turn, given truncated normal or lognormal distributions
for the uncertainty in the population mean. If no independent data were available, as is the case
for many downstream metabolism parameters, then no baseline value was specified, and a
3-66
noninformative prior was used. Section 3.5.5.4, below, discusses the updating of these
noninformative priors using interspecies scaling.
In keeping with standard practice, many of the PBPK model parameters were scaled by
body or organ weights, cardiac output, or allometrically by an assumed (fixed) power of body
weight. Metabolic capacity and cardiac output were scaled by the power of body weight and
rate coefficients were scaled by the - power of body weight, in keeping with general
expectations as to the relationship between metabolic rates and body size (West et al., 2002; U.S.
EPA, 1992). So as to ensure a consistent model structure across species as well as improve the
performance of the MCMC algorithm, parameters were further scaled to the baseline point-
estimates where available, as was done by Hack et al. (2006). For example, to obtain the actual
liver volume (VLivC) in L, a point estimate is first obtained by multiplying the fixed, species-
specific baseline point estimate for the fractional liver volume by a fixed body weight (measured
or species-specific default) with density of 1 kg/L assumed to convert from kg to L. Then, any
deviation from this point estimate is represented by multiplying by a separate scaled parameter
VLivC that has a value of 1 if there is no deviation from the point estimate. These scaled
parameters are those estimated by the MCMC algorithm, and for which population means and
variances are estimated.
Baseline physiological parameters were reestimated based on the updated tissue lumping
(e.g., separate blood and kidney compartments) using the standard references, International
Commission on Radiological Protection (ICRP, 2003) and Brown et al. (1997). For a few of
these parameters, such as hematocrit and respiratory tract volumes in rodents, additional
published sources were used as available, but no attempt was made to compile a comprehensive
review of available measurements. In addition, a few parameters, such as the slowly perfused
volume, were calculated rather than sampled in order to preserve total mass or flow balances.
For chemical-specific distribution and metabolism parameters, in vitro data from various
sources were used. Where multiple measurements had been made, as was the case for many
partition coefficients, TCA plasma protein binding parameters, and TCE metabolism, different
results were pooled together, with their uncertainty reflected appropriately in the prior
distribution. Such in vitro measurements were available for most chemical partition coefficients,
except for those for TCOG (TCOH used as a proxy) and DCVG. There were also such data to
develop baseline values for the oxidative metabolism of TCE in the liver (V
MAX
and K
M
), the
relative split in TCE oxidation between formation of TCA and TCOH, and the V
MAX
for TCE
oxidation in the lung. For GSH conjugation, the geometric means of the in vitro data from Lash
et al. (1999a)and Green et al. (1997a) were used as central estimates, with a wide enough
uncertainty range to encompass both (widely disparate) estimates. Thus, the prior distribution
for these parameters was only mildly informative, and the results are primarily determined by the
available in vivo data. All other metabolism parameters were not given baseline values and
needed to be estimated from the in vivo data.
3-67

3.5.4.4. Dose-Metric Predictions
The purpose of this PBPK model is to make predictions of internal dose in rodents used
in toxicity studies or in humans in the general population, and not in the groups or individuals for
which pharmacokinetic data exist. Therefore, to evaluate its predictive utility for risk
assessment, a number of dose-metrics were selected for simulation in a generic mouse, rat, or
human, summarized in Table 3-33. The parent dose-metric was AUC in blood. TCE
metabolism dose-metrics (i.e., related to the amount metabolized) included both total
metabolism, metabolism splits between oxidation vs. conjugation, oxidation in the liver vs. the
lung, the amount of oxidation in the liver to products other than TCOH and TCA, and the
amount of TCA produced. These metabolism rate dose-metrics are scaled by body weight in the
case of TCA produced, by the metabolizing tissue volume and by body weight to the power in
the cases of the lung and other oxidation in the liver, and by body weight to the power only
in other cases. With respect to the oxidative metabolites, liver concentrations of TCA and blood
concentrations of free TCOH were used. With respect to conjugative metabolites, the dose-
metrics considered were total GSH metabolism scaled by body weight to the power, and the
amount of DCVC bioactivated (rather than excreted in urine) per unit body weight to the
power and per unit kidney mass.

Table 3-33. PBPK model-based dose-metrics

Abbreviation Description
ABioactDCVCBW34 Amount of DCVC bioactivated in the kidney (mg) per unit body weight

(kg

)
ABioactDCVCKid Amount of DCVC bioactivated in the kidney (mg) per unit kidney mass (kg)
AMetGSHBW34 Amount of TCE conjugated with GSH (mg) per unit body weight

(kg

)
AMetLiv1BW34 Amount of TCE oxidized in the liver per unit body weight

(kg

)
AMetLivOtherBW34 Amount of TCE oxidized to metabolites other than TCA and TCOH in the liver (mg) per
unit body weight

(kg

)
AMetLivOtherLiv Amount of TCE oxidized to metabolites other than TCA and TCOH in the liver (mg) per
unit liver mass (kg)
AMetLngBW34 Amount of TCE oxidized in the respiratory tract (mg) per unit body weight

(kg

)
AMetLngResp Amount of TCE oxidized in the respiratory tract (mg) per unit respiratory tract tissue
mass (kg)
AUCCBld Area under the curve of the venous blood concentration of TCE (mg-hr/L)
AUCCTCOH Area under the curve of the blood concentration of TCOH (mg-hr/L)
AUCLivTCA Area under the curve of the liver concentration of TCA (mg-hr/L)
TotMetabBW34 Total amount of TCE metabolized (mg) per unit body weight

(kg

)
TotOxMetabBW34 Total amount of TCE oxidized (mg) per unit body weight

(kg

)
TotTCAInBW Total amount of TCA produced (mg) per unit body weight (kg)

All dose-metrics are converted to daily or weekly averages based on simulations lasting
10 weeks for rats and mice and 100 weeks for humans. These simulation times were the shortest
3-68
for which additional simulation length did not add substantially to the average (i.e., less than a
few percent change with a doubling of simulation time).

3.5.5. Bayesian Estimation of PBPK Model Parameters, and Their Uncertainty and
Variability
3.5.5.1. Updated Pharmacokinetic Database
An extensive search was made for data not previously considered in the PBPK modeling
of TCE and metabolites, with a few studies identified or published subsequent to the review by
Chiu et al. (2006b). The studies considered for analysis are listed in Tables 3-34 and 3-35, along
with an indication of whether and how they were used.
3

The least amount of data was available for mice, so an effort was made to include as
many studies as feasible for use in calibrating the PBPK model parameters. Exceptions include
mouse studies with CH or DCA dosing, since those metabolites are not included in the PBPK
model. In addition, the Birner et al. (1993) data only reported urine concentrations, not the
amount excreted in urine. Because there is uncertainty as to total volume of urine excreted, and
over what time period, these data were not used. Moreover, many other studies had urinary
excretion data, so this exclusion should have minimal impact. Several data sets not included by
Hack et al. (2006) were used here. Of particular importance was the inclusion of TCA and
TCOH dosing data from Abbas et al. (1997), Green and Prout (1985), Larson and Bull (1992a),
and Templin et al. (1993). A substantial amount of data is available in rats, so some data that
appeared to be redundant were excluded from the calibration set and saved for comparison with
posterior predictions (a validation set). In particular, those used for validation are one
closed-chamber experiment (Andersen et al., 1987b), several data sets with only TCE blood data
(Lee et al., 1996; Jakobson et al., 1986; D'Souza et al., 1985), and selected time courses from
Fisher et al. (1991) and Lee et al. (2000a; 2000b), and one unpublished data set (Bruckner et al.,
unpublished). The Andersen et al. (1987b) data were selected randomly from the available
closed-chamber data, while the other data sets were selected because they were unpublished or
because they were more limited in scope (e.g., TCE blood only) and so were not as efficient for
use in the computationally-intensive calibration stage. As with the mouse analyses, TCA and
TCOH dosing data were incorporated to better calibrate those pathways.




3
Additional in vivo data on TCE or metabolites published after the PBPK modeling was completed (Kim et al.,
2009; Liu et al., 2009; Sweeney et al., 2009) were evaluated separately, and discussed in Appendix A.
3-69
Table 3-34. Rodent studies with pharmacokinetic data considered for analysis

Reference
Species
(strain) Sex TCE exposures Other exposures Calibration Validation
Not
used Comments
Mouse studies
Abbas et al. (1996) Mouse
(B6C3F
1
)
M CH i.v. \ CH not in model.
Abbas and Fisher
(1997)
Mouse
(B6C3F
1
)
M Oral (corn oil) \
a

Abbas et al. (1997) Mouse
(B6C3F
1
)
M TCOH, TCA i.v. \
Barton et al. (1999) Mouse
(B6C3F
1
)
M DCA i.v. and oral
(aqueous)
\ DCA not in model.
Birner et al. (1993) Mouse (NMRI) M+F Gavage \ Only urine concentrations
available, not amount.
Fisher and Allen,
(1993)
Mouse
(B6C3F
1
)
M+F Gavage (corn oil) \
Fisher et al. (1991) Mouse
(B6C3F
1
)
M+F Inhalation \
a

Green and Prout
(1985)
Mouse
(B6C3F
1
)
M Gavage (corn oil) TCA i.v. \
Greenberg et al.
(1999)
Mouse
(B6C3F
1
)
M Inhalation \
a

Larson and Bull
(1992b)
Mouse
(B6C3F
1
)
M DCA, TCA oral
(aqueous)
\ Only data on TCA dosing was
used, since DCA is not in the
model.
Larson and Bull
(1992a)
Mouse
(B6C3F
1
)
M Oral (aqueous) \
Merdink et al. (1998) Mouse
(B6C3F
1
)
M i.v. CH i.v. \ Only data on TCE dosing was
used, since CH is not in the
model.

3-70
Table 3-34. Rodent studies with pharmacokinetic data considered for analysis (continued)

Reference
Species
(strain) Sex TCE exposures Other exposures Calibration Validation
Not
used Comments
Prout et al. (1985) Mouse
(B6C3F
1
,
Swiss)
M Gavage (corn oil) \
a

Templin et al. (1993) Mouse
(B6C3F
1
)
M Oral (aqueous) TCA oral \
a

Rat studies
Andersen et al.
(1997)
Rat (F344) M Inhalation \
a

Barton et al. (1995) Rat (Sprague-
Dawley)
M Inhalation \ Initial chamber concentrations
unavailable, so not used.
Bernauer et al.
(1996)
Rat (Wistar) M Inhalation \
a

Birner et al. (1993) Rat (Wistar,
F344)
M+F Gavage (ns) \ Only urine concentrations
available, not amount.
Birner et al. (1997) Rat (Wistar) M+F DCVC i.v. \ Single dose, route does not
recapitulate how DCVC is
formed from TCE, excreted
NAcDCVC ~100-fold greater
than that from relevant TCE
exposures (Bernauer et al.,
1996).
Bruckner et al.
unpublished
Rat (Sprague-
Dawley)
M Inhalation \ Not published, so not used for
calibration. Similar to Keys
et al. (2003) data.
Dallas et al. (1991) Rat (Sprague-
Dawley)
M Inhalation \
D'Souza et al. (1985) Rat (Sprague-
Dawley)
M i.v., oral (aqueous) \ Only TCE blood
measurements, and 10-fold
greater than other similar
studies.
Fisher et al. (1989) Rat (F344) F Inhalation \
Fisher et al. (1991) Rat (F344) M+F Inhalation \
a
\ Experiment with blood only
data not used for calibration.
3-71
Table 3-34. Rodent studies with pharmacokinetic data considered for analysis (continued)

Reference
Species
(strain) Sex TCE exposures Other exposures Calibration Validation
Not
used Comments
Green and Prout
(1985)
Rat (Osborne-
Mendel)
M Gavage (corn oil) TCA gavage
(aqueous)
\
Hissink et al. (2002) Rat (Wistar) M Gavage (corn oil),
i.v.
\
Jakobson et al.
(1986)
Rat (Sprague-
Dawley)
F Inhalation Various
pretreatments
(oral)
\ Pretreatments not included.
Only blood TCE data
available.
Kaneko et al. (1994) Rat (Wistar) M Inhalation Ethanol
pretreatment
(oral)
\ Pretreatments not included.
Keys et al. (2003) Rat (Sprague-
Dawley)
M Inhalation,
oral (aqueous), i.a.
\
Kimmerle and Eben
(1973b)
Rat (Wistar) M Inhalation \
Larson and Bull
(1992b)
Rat (F344) M DCA, TCA oral
(aqueous)
\ Only TCA dosing data used,
since DCA is not in the model.
Larson and Bull
(1992a)
Rat (Sprague-
Dawley)
M Oral (aqueous) \
a

Lash et al. (2006) Rat (F344) M+F Gavage (corn oil) \ Highly inconsistent with other
studies.
Lee et al. (1996) Rat (Sprague-
Dawley)
M Arterial, venous,
portal, stomach
injections
\ Only blood TCE data
available.
Lee et al. (2000a;
2000b)
Rat (Sprague-
Dawley)
M Stomach injection,
i.v., p.v.
p-Nitrophenol
pretreatment (i.a.)
\ \ Pretreatments not included.
Only experiments with blood
and liver data used for
calibration.
Merdink et al. (1999) Rat (F344) M CH, TCOH i.v. \ TCOH dosing used; CH not in
model.
Poet et al. (2000) Rat (F344) M Dermal \ Dermal exposure not in model.
Prout et al. (1985) Rat (Osborne-
Mendel,
Wistar)
M Gavage (corn oil) \
a

3-72
Table 3-34. Rodent studies with pharmacokinetic data considered for analysis (continued)

Reference
Species
(strain) Sex TCE exposures Other exposures Calibration Validation
Not
used Comments
Saghir et al. (2002) Rat (F344) M DCA i.v., oral
(aqueous)
\ DCA not in model.
Simmons et al.
(2002)
Rat (Long-
Evans)
M Inhalation \
Stenner et al. (1997) Rat (F344) M intraduodenal TCOH, TCA i.v. \
Templin et al.
(1995b)
Rat (F344) M Oral (aqueous) \
a

Thrall et al. (2000) Rat (F344) M i.v., i.p. With toluene \ Only exhaled breath data
available from i.v. study; i.p.
dosing not in model.
Yu et al. (2000) Rat (F344) M TCA i.v. \

a
Part or all of the data in the study was used for calibration in Hack et al. (2006).

p.v. = intraperivenous
3-73
Table 3-35. Human studies with pharmacokinetic data considered for analysis

Reference
Species
(number of
individuals) Sex
TCE
exposures
Other
exposures Calibration Validation
Not
used Comments
Bartonicek (1962) Human (n = 8) M+F Inhalation \ Sparse data, so not included for
calibration to conserve computational
resources.
Bernauer et al. (1996) Human M Inhalation \
a
Grouped data, but unique in that
includes NAcDCVC urine data.
Bloemen et al. (2001) Human (n = 4) M Inhalation \ Sparse data, so not included for
calibration to conserve computational
resources.
Chiu et al. (2007) Human (n = 6) M Inhalation \
Ertle et al. (1972) Human M Inhalation CH oral \ Very similar to Muller data.
Fernandez et al. (1977) Human M Inhalation \
Fisher et al. (1998) Human (n = 17) M+F Inhalation \
a

Kimmerle and Eben
(1973a)
Human (n = 12) M+F Inhalation \
Lapare et al. (1995) Human (n = 4) M+F Inhalation \
b
Complex exposure patterns, and only
grouped data available for urine, so
used for validation.
Lash et al. (1999b) Human M+F Inhalation \ Grouped only, but unique in that
DCVG blood data available (same
individuals as Fisher et al. (1998)].
Monster et al. (1976) Human (n = 4) M Inhalation \
b
Experiments with exercise not
included.
Monster et al. (1979) Human M Inhalation \
a
Grouped data only.
Muller et al. (1972) Human ns Inhalation \ Same data also included in Muller
et al. (1975).

3-74
Table 3-35. Human studies with pharmacokinetic data considered for analysis (continued)

Reference
Species
(number of
individuals) Sex
TCE
exposures
Other
exposures Calibration Validation
Not
used Comments
Muller et al. (1974) Human M Inhalation CH, TCA,
TCOH oral
\ \
a
TCA and TCOH dosing data used for
calibration, since it is rare to have
metabolite dosing data. TCE dosing
data used for validation, since only
grouped data available. CH not in
model.
Muller et al. (1975) Human M Inhalation Ethanol oral \
a
Grouped data only.
Paykoc et al. (1945) Human (n = 3) ns TCA i.v. \
Poet et al. (2000) Human M+F Dermal Dermal exposure not in model.
Sato et al. (1977) Human M Inhalation \
Stewart et al. (1970) Human ns Inhalation \
a

Treibig et al. (1976) Human ns Inhalation \
a

Vesterberg and Astrand
(1976)
Human M Inhalation \ All experiments included exercise, so
were not included.

a
Part or all of the data in the study was used for calibration in Hack et al. (2006).
b
Grouped data from this study was used for calibration in Hack et al. (2006), but individual data were used here.
3-75
The human pharmacokinetic database of controlled exposure studies is extensive, but also
more complicated. For the majority of the studies, only grouped or aggregated data were
available, and most of those data were saved for validation since there remained a large
number of studies for which individual data were available. However, some data that may be
uniquely informative are only available in grouped form, in particular DCVG blood
concentrations, NAcDCVC urinary excretion, and data from TCA and TCOH dosing. While
there are analytic uncertainties as to the DCVG blood measurements, discussed above in
Section 3.3.3.2.1, they were nonetheless included here because they are the only in vivo data
available on this measurement in humans. The uncertainty associated with their use is discussed
below (see Section 3.5.7.3.2).
In addition, several human data sets, while having individual data, involved sparse
collection at only one or a few time points per exposure (Bloemen et al., 2001; Bartonicek, 1962)
and were subsequently excluded to conserve computational resources. Lapare et al. (1995),
which involved multiple, complex exposure patterns over the course of a month and was missing
the individual urine data, was also excluded due to the relatively low amount of data given the
large computational effort required to simulate the data. Several studies also investigated the
effects of exercise during exposure on human TCE toxicokinetics. The additional parameters in
a model including exercise would include those for characterizing the changes in cardiac output,
alveolar ventilation, and regional blood flow as well as their interindividual variability, and
would have further increased the computational burden. Therefore, it was decided that such data
would be excluded from this analysis. Even with these exclusions, data on a total of
42 individuals, some involving multiple exposures, were included in the calibration.

3.5.5.2. Updated Hierarchical Population Statistical Model and Prior Distributions
While the individual animals of a common strain and sex within a study are likely to vary
to some extent, this variability was not included as part of the hierarchical population model for
several reasons. First, generally, only aggregated pharmacokinetic data (arithmetic mean and SD
or SE) are available from rodent studies. While methods exist for addressing between-animal
variability with aggregated data (e.g., Chiu and Bois, 2007), they require a higher level of
computational intensity. Second, dose-response data are generally also only separated by sex
and strain, and otherwise aggregated. Thus, in analyzing dose-response data (see Chapter 5), one
usually has no choice but to treat all of the animals in a particular study of a particular strain and
sex as identical units. In the Hack et al. (2006) model, each simulation was treated as a separate
observational unit, so different dosing levels within the same study were treated separately and
assigned different PBPK model parameters. However, the animals within a study are generally
inbred and kept under similarly controlled conditions, whereas animals in different studies
even if of the same strain and sexlikely have differences in genetic lineage, diet, and handling.
Thus, animals within a study are likely to be much more homogeneous than animals between
3-76
studies. As a consequence, in the revised model, for rodents, different animals of the same sex
and strain in the same study (or series of studies conducted simultaneously) were treated as
identical, and grouped together as a single subject. Thus, the predictions from the population
model in rodents simulate average pharmacokinetics for a particular lot of rodents of a
particular species, strain, and sex. Between-animal variability is not explicitly modeled, but it is
incorporated in a residual error term as part of the likelihood function (see Appendix A,
Section A.4.3.4). Therefore, a high degree of within-study variability would be reflected in a
high posterior value in the variance of the residual-error.
In humans, however, interindividual variability is of interest, and, furthermore,
substantial individual data are available in humans. However, in some studies, the same
individual was exposed more than once, so those data should be grouped together [in the Hack
et al. (2006) model, they were treated as different individuals]. Because the primary interest
here is chronic exposure, and because it would add substantially to the computational burden,
interoccasion variabilitychanges in pharmacokinetic parameters in a single individual over
timeis not addressed. Therefore, each individual is considered a single subject, and the
predictions from the population model in humans are the average across different occasions for
a particular individual (adult). Between-occasion variability is not explicitly modeled, but it is
incorporated in a residual error term as part of the likelihood function (see Appendix A,
Section A.4.3.4). Therefore, a high degree of between-occasion variability would be reflected in
a high posterior value in the variance of the residual-error.
As discussed in Section 3.3.3.1, sex and (in rodents) strain differences in oxidative
metabolism were modest or minimal. While some sex-differences have been noted in GSH
metabolism (see Sections 3.3.3.2.7 and 3.3.3.2.8), almost all of the available in vivo data are in
males, making it more difficult to statistically characterize that difference with PBPK modeling.
Therefore, within a species, different sexes and (in rodents) strains were considered to be drawn
from a single, species-level population. For humans, each individual was considered to be drawn
from a single (adult) human population.
Thus, from here forward, the term subject will be used to refer to both a particular lot
of a particular rodents species, strain, and sex for, and a particular human individual. The term
population will, therefore, refer to the collection of rodent lots of the same species and the
collection of human individuals.
Figure A-1 in Appendix A illustrates the hierarchical structure. Informative prior
distributions reflecting the uncertainty in the population mean and variance, detailed in
Appendix A, were updated from those used in Hack et al. (2006) based on an extensive analysis
of the available literature. The population variability of the scaling parameter across subjects is
assumed to be distributed as a truncated normal distribution, a standard assumption in the
absence of specific data suggesting otherwise. Because of the truncation of extreme values, the
sensitivity to this choice is expected to be small as long as the true underlying distribution is uni-
3-77
modal and symmetric. In addition, most scaling parameters, being strictly positive in their
original units, were log-transformedso these parameters have lognormal distributions in their
original units. The uncertainty distribution for the population parameters was assumed to be a
truncated normal distribution for population mean parameters and an inverse gamma distribution
for population variance parametersboth standard choices in hierarchical models.
Section 3.5.5.3, next, discusses specification of prior distributions in the case where no data
independent of the calibration data exist.

3.5.5.3. Use of Interspecies Scaling to Update Prior Distributions in the Absence of
Other Data
For many metabolic parameters, little or no in vitro or other prior information is available
to develop prior distributions. Initially, for such parameters, noninformative priors in the form of
log-uniform distributions with a range spanning at least 10
4
were specified. However, in the
time available for analysis (up to about 100,000 iterations), only for the mouse did all of these
parameters achieve adequate convergence. This suggests that some of these parameters are
poorly identified for the rat and human. Additional preliminary runs indicated replacing the log-
uniform priors with lognormal priors and/or requiring more consistency between species could
improve identifiability sufficiently for adequate convergence. However, an objective method of
centering the lognormal distributions that did not rely on the in vivo data (e.g., via visual
fitting or limited optimization) being calibrated against was necessary in order to minimize
potential bias.
Therefore, the approach taken was to consider three species sequentially, from mouse to
rat to human, and to use interspecies scaling to update the prior distributions across species. This
sequence was chosen because the models are essentially nested in this order, the rat model
adds to the mouse model the downstream GSH conjugation pathways, and the human model
adds to the rat model the intermediary DCVG compartment. Therefore, for those parameters
with little or no independent data only, the mouse posteriors were used to update the rat priors,
and both the mouse and rat posteriors were used to update the human priors. Table 3-36 contains
a list of the parameters for which this scaling was used to update prior distributions. The scaling
relationship is defined by the scaled parameters listed in Appendix A (see Section A.4.1,
Table A-4), and generally follows standard practice. For instance, V
MAX
and clearance rates
scale by body weight to the power, whereas K
M
values are assumed to not scale, and rate
constants (inverse time units) scale by body weight to the - power.
3-78
Table 3-36. Parameters for which scaling from mouse to rat, or from mouse and rat to human, was used to
update the prior distributions

Parameter with no or highly uncertain a priori data
Mouse
rat
Rat
human
Mouse+
rat
human Comments
Respiratory lumentissue diffusion flow rate No a priori information
TCOG body/blood partition coefficient Prior centered on TCOH data, but highly uncertain
TCOG liver/body partition coefficient Prior centered on TCOH data, but highly uncertain
Fraction of hepatic TCE oxidation not to TCA+TCOH No a priori information
V
MAX
for hepatic TCE GSH conjugation Rat data on at 1 and 2 mM. Human data at more
concentrations, so V
MAX
and K
M
can be estimated
K
M
for hepatic TCE GSH conjugation
V
MAX
for renal TCE GSH conjugation Rat data on at 1 and 2 mM. Human data at more
concentrations, so V
MAX
and K
M
can be estimated
K
M
for renal TCE GSH conjugation
V
MAX
for Tracheo-bronchial TCE oxidation Prior based on activity at a single concentration
K
M
for Tracheo-bronchial TCE oxidation No a priori information
Fraction of respiratory oxidation entering systemic circulation No a priori information
V
MAX
for hepatic TCOHTCA No a priori information
K
M
for hepatic TCOHTCA No a priori information
V
MAX
for hepatic TCOHTCOG No a priori information
K
M
for hepatic TCOHTCOG No a priori information
Rate constant for hepatic TCOHother No a priori information
Rate constant for TCA plasmaurine Prior centered at glomerular filtration rate, but highly
uncertain
Rate constant for hepatic TCAother No a priori information
Rate constant for TCOG liverbile No a priori information
Lumped rate constant for TCOG bileTCOH liver No a priori information
Rate constant for TCOGurine Prior centered at glomerular filtration rate, but highly
uncertain
Lumped rate constant for DCVCUrinary NAcDCVC Not included in mouse model
Rate constant for DCVC bioactivation Not included in mouse model

a
See Appendix A, Table A-4 for scaling relationships.
3-79
The scaling model is given explicitly as follows. If
i
are the scaled parameters
(usually also natural-log-transformed) that are actually estimated, and A is the universal
(species-independent) parameter, then
i
= A +
i
, where
i
is the species-specific departure
from the scaling relationship, assumed to be normally distributed with variance

2
. Therefore,
the mouse model gives an initial estimate of A, which is used to update the prior distribution
for
r
= A +
r
in the rat. The rat and mouse together then give a better estimate of A, which is
used to update the prior distribution for
h
= A +
h
in the human, with the assumed distribution
for
h
. The mathematical details are given in Appendix A, but three key points in this model are
worth noting here:

- It is known that interspecies scaling is not an exact relationship, and that, therefore, in
any particular case, it may either over- or underestimate. Therefore, the variance in the
new priors reflect a combination of (1) the uncertainty in the previous species
posteriors as well as (2) a prediction error that is distributed lognormally with
geometric standard deviation (GSD) of 3.16-fold, so that the 95% confidence range about
the central estimate spans 100-fold. This choice was dictated partially by practicality, as
larger values of the GSD used in preliminary runs did not lead to adequate convergence
within the time available for analysis.

- The rat posterior is a product of its prior (which is based on the mouse posterior) and its
likelihood. Therefore, using the rat and mouse posteriors together to update the human
priors would use the mouse posterior twice. Therefore, the rat posterior is
disaggregated into its prior and its likelihood using a lognormal approximation (since the
prior is lognormal), and only the (approximate) likelihood is used along with the mouse
posterior to develop the human prior.

- The model transfers the marginal distributions for each parameter across species, so
correlations between parameters are not retained. This is a restriction on the software
used for conducting MCMC analyses. However, assuming independence will lead to a
broader joint distribution, given the same marginal distributions. Therefore, this
assumption tends to reduce the weight of the interspecies scaling as compared to the
species-specific calibration data.

To summarize, in order to improve rate of the convergence of the MCMC analyses in rats
and humans, a sequential approach was used for fitting scaling parameters without strong prior
species-specific information. In particular, an additional assumption was made that across
species, these scaling parameters were, in absence of other information, expected to have a
common underlying value. These assumptions are generally based on allometric scaling
principleswith partition coefficients and concentrations scaling directly and rate constants
scaling by body weight to the - power (so clearances and maximum metabolic capacities would
scale by body weight to the power). These assumptions are used consistently throughout the
parameter calibration process. Therefore, after running the mouse model, the posterior
distribution for these parameters was used, with an additional error term, as priors for the rat
3-80
model. Subsequently, after the mouse and rat model were run, their posterior distributions were
combined, with an additional error term, to use a priors for the human model. With this
methodology for updating the prior distributions, adequate convergence was achieved for the rat
and human after 110,000~140,000 iterations (discussed further below).

3.5.5.4. Implementation
The PBPK model was coded in for use in the MCSim software (version 5.0.0), which was
developed particularly for implementing MCMC simulations. As a quality control check, results
were checked against the original Hack et al. (2006) model, with the original structures restored
and parameter values made equivalent, and the results were within the error tolerances of the
ordinary differential equation (ODE) solver after correcting an error in the Hack et al. (2006)
model for calculating the TCA liver plasma flow. In addition, the model was translated to
MatLab (version 7.2.0.232) with simulation results checked and found to be within the error
tolerances of the ODE solver used (ode15s). Mass balances were also checked using the
baseline parameters, as well as parameters from preliminary MCMC simulations, and found to
be within the error tolerances of the ODE solver. Appendix A contains the MCSim model code.

3.5.6. Evaluation of Updated PBPK Model
3.5.6.1. Convergence
As in previous similar analyses (David et al., 2006; Hack et al., 2006; Bois, 2000b, a;
Gelman et al., 1996), the potential scale reduction factor R is used to determine whether
different independent MCMC chains have converged to a common distribution. The R
diagnostic is calculated for each parameter in the model, and represents the factor by which the
SD or other measure of scale of the posterior distribution (such as a confidence interval [CI])
may potentially be reduced with additional samples (Gelman et al., 2003). This convergence
diagnostic declines to 1 as the number of simulation iterations approaches infinity, so values
close to 1 indicate approximate convergence, with values of 1.1 commonly considered adequate
(Gelman et al., 2003). However, as an additional diagnostic, the convergence of model dose-
metric predictions was also assessed. Specifically, dose-metrics for a number of generic
exposure scenarios similar to those used in long-term bioassays were generated, and their natural
log (due to their approximate lognormal posterior distributions) was assessed for convergence
using the potential scale reduction factor R. This is akin to the idea of utilizing sensitivity
analysis so that effort is concentrated on calibrating the most sensitive parameters for the purpose
of interest. In addition, predictions of interest that do not adequately converge can be flagged as
such, so that the statistical uncertainty associated with the limited sample size can be considered.
The mouse model had the most rapid reduction in potential scale reduction factors.
Initially, four chains of 42,500 iterations each were run, with the first 12,500 discarded as burn-
in iterations. The initial decision for determining burn-in was determined by visual
3-81
inspection. At this point, evaluating the 30,000 remaining iterations, all of the population
parameters except for the V
MAX
for DCVG formation had R < 1.2, with only the first-order
clearance rate for DCVG formation and the V
MAX
and K
M
for TCOH glucuronidation having
R > 1.1. For the samples used for inference, all of these initial iterations were treated as burn-
in iterations, and each chain was then restarted and run for an additional 68,700
71,400 iterations (chains were terminated at the same time, so the number of iterations per chains
was slightly different). For these iterations, all values of R were <1.03. Dose-metric predictions
calculated for exposure scenarios of 10600 ppm either continuously or 7 hours/day,
5 days/week and 103,000 mg/kg-day either continuously or by gavage 5 days/week. These
predictions were all adequately converged, with all values of R < 1.03.
As discussed above, for parameters with little or no a priori information, the posterior
distributions from the mouse model were used to update prior distributions for the rat model,
accounting for both the uncertainty reflected in the mouse posteriors as well as the uncertainty in
interspecies extrapolation. Four chains were run to 111,960128,000 iterations each (chains
were terminated at the same time and run on computers with slightly different processing speeds,
so the number of iterations per chains was slightly different). As is standard, about the
first half of the chains (i.e., the first 64,000 iterations) were discarded as burn-in iterations,
and the remaining iterations were used for inferences. For these remaining iterations, the
diagnostic R was <1.1 for all population parameters except the fraction of oxidation not
producing TCA or TCOH (R = 1.44 for population mean, R = 1.35 for population variance), the
K
M
for TCOH TCA (R = 1.19 for population mean), the V
MAX
and K
M
for TCOH
glucuronidation (R = 1.23 and 1.12, respectively for population mean, and R = 1.13 for both
population variances), and the rate of other metabolism of TCOH (R = 1.29 for population
mean and R = 1.18 for population variance). Due to resource constraints, chains needed to be
stopped at this point. However, these are similar to the degree of convergence reported in Hack
et al. (2006). Dose-metric predictions calculated for two inhalation exposure scenarios (10
600 ppm continuously or 7 hours/day, 5 days/week) and two oral exposure scenarios (10
3,000 mg/kg-day continuously or by gavage 5 days/week).
All dose-metric predictions had R < 1.04, except for the amount of other oxidative
metabolism (i.e., not producing TCA or TCOH), which had R = 1.121.16, depending on the
exposure scenario. The poorer convergence of this dose-metric is expected given that a key
determining parameter, the fraction of oxidation not producing TCA or TCOH, had the poorest
convergence among the population parameters.
For the human model, a set of four chains was run for 74,16084,690 iterations using
preliminary updated prior distributions based on the mouse posteriors and preliminary runs of
the rat model. Once the rat chains were completed, final updated prior distributions were
calculated and the last iteration of the preliminary runs were used as starting points for the final
runs. The center of the final updated priors shifted by <25% of the SD of either the preliminary
3-82
or revised priors, so that the revised median was between the 40
th
and 60
th
percentile of the
preliminary median, and vice versa. The SDs changed by <5%. Therefore, the use of the
preliminary chains as a starting point should introduce no bias, as long as an appropriate burn-in
period is used for the final runs.
The final chains were run for an additional 59,14061,780 iterations, at which point, due
to resource constraints, chains needed to be stopped. After the first 20,000 iterations, visual
inspection revealed the chains were no longer dependent on the starting point. These iterations
were therefore discarded as burn-in iterations, and for the remaining ~40,000 iterations used
for inferences. All population mean parameters had R < 1.1 except for the respiratory tract
diffusion constant (R = 1.20), the liver:blood partition coefficient for TCOG (R = 1.23), the rate
of TCE clearance in the kidney producing DCVG (R = 1.20), and the rate of elimination of
TCOG in bile (R = 1.46). All population variances also had R < 1.1 except for the variance for
the fraction of oxidation not producing TCOH or TCA (R = 1.10). Dose-metric predictions were
assessed for continuous exposure scenarios at 160 ppm in air or 1300 mg/kg-day orally. These
predictions were all adequately converged with all values of R < 1.02.

3.5.6.2. Evaluation of Posterior Parameter Distributions
Posterior distributions of the population parameters need to be checked as to whether
they appear reasonable given the prior distributions. Inconsistency between the prior and
posterior distributions may indicate insufficiently broad (i.e., due to overconfidence) or
otherwise incorrectly specified priors, a misspecification of the model structure (e.g., leading to
pathological parameter estimates), or an error in the data. As was done with the evaluation of
Hack et al. (2006) in Appendix A, parameters were flagged if the interquartile regions of their
prior and posterior distributions did not overlap.
Appendix A contains detailed tables of the sampled parameters, and their prior and
posterior distributions. Because these parameters are generally scaled one or more times to
obtain a physically meaningful parameter, they are difficult to interpret. Therefore, in
Tables 3-373-39, the prior and posterior population distributions for the PBPK model
parameters obtained after scaling are summarized. Since it is desirable to characterize the
contributions from both uncertainty in population parameters and variability within the
population, the following procedure is adopted. First, 500 sets of population parameters (i.e.,
population mean and variance for each scaling parameter) are either generated from the prior
distributions via Monte Carlo or extracted from the posterior MCMC samplesthese represent
the uncertainty in the population parameters. To minimize autocorrelation, for the posteriors, the
samples were obtained by thinning the chains to the appropriate degree. From each of these
sets of population parameters, 100 sets of subject-level parameters were generated by Monte
Carloeach of these represents the population variability, given a particular set of population
parameters. Thus, a total of 50,000 subjects, representing 100 (variability) each for 500 different
3-83
populations (uncertainty), were generated. For each of the 500 populations, the scaling
parameters are converted to PBPK model parameters, and the population median and GSD is
calculatedrepresenting the central tendency and variability for that population. Then, the
median and the 95% CIs for the population median and GSD are calculated, and presented in the
tables that follow. Thus, these tables summarize separately the uncertainty in population
distribution as well as the variability in the population, while also accounting for correlations
among the population-level parameters. Finally, Table 3-40 shows the change in the CI in the
population median for the PBPK model parameters between the prior and posterior distributions,
as well as the shift in the central estimate (median) of the population median PBPK model
parameter.
The prior and posterior distributions for most physiological parameters were similar. The
posterior distribution was substantially narrower (i.e., less uncertainty) than the prior distribution
only in the case of the diffusion rate from the respiratory lumen to the respiratory tissue, which
also was to be expected given the very wide, noninformative prior for that parameter.
For distribution parameters, there were only relatively minor changes between prior and
posterior distributions for TCE and TCOH partition coefficients. The posterior distributions for
several TCA partition coefficients and plasma binding parameters were substantially narrower
than their corresponding priors, but the central estimates were similar, meaning that values at the
high and low extremes were not likely. For TCOG as well, partition coefficient posterior
distributions were substantially narrower, which was expected given the greater uncertainty in
the prior distributions (TCOH partition coefficients were used as a proxy).
Again, posterior distributions indicated that the high and low extremes were not likely.
Finally, posterior distribution for the distribution volume for DCVG was substantially narrower
than the prior distribution, which only provided a lower bound given by the blood volume. In
this case, the upper bounds were substantially lower in the posterior.
Posterior distributions for oral absorption parameters in mice and rats (there were no oral
studies in humans) were also informed by the data, as reflected in their being substantially more
narrow than the corresponding priors. Finally, with a few exceptions, TCE and metabolite
kinetic parameters showed substantially narrower posterior distributions than prior distributions,
indicating that they were fairly well specified by the in vivo data. The exceptions were the V
MAX

for hepatic oxidation in humans (for which there was substantial in vitro data) and the V
MAX
for
respiratory metabolism in mice and rats (although the posterior distribution for the K
M
for this
pathway was substantially narrower than the corresponding prior).
3-84
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model parameters

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Cardiac output (L/hr) QC 0.84 (0.59, 1.2) 1 (0.79, 1.3) 1.17 (1.1, 1.4) 1.35 (1.15, 1.54)
Alveolar ventilation (L/hr) QP 2.1 (1.3, 3.5) 2.1 (1.5, 2.7) 1.27 (1.17, 1.54) 1.45 (1.28, 1.66)
Scaled fat blood flow QFatC 0.07 (0.03, 0.11) 0.072 (0.044, 0.1) 1.65 (1.22, 2.03) 1.64 (1.3, 1.99)
Scaled gut blood flow QGutC 0.14 (0.11, 0.17) 0.16 (0.14, 0.17) 1.15 (1.09, 1.19) 1.12 (1.07, 1.19)
Scaled liver blood flow QLivC 0.02 (0.016, 0.024) 0.021 (0.017, 0.024) 1.15 (1.09, 1.19) 1.15 (1.09, 1.19)
Scaled slowly perfused
blood flow
QSlwC 0.22 (0.14, 0.29) 0.21 (0.15, 0.28) 1.3 (1.15, 1.38) 1.3 (1.17, 1.39)
Scaled rapidly perfused
blood flow
QRapC 0.46 (0.37, 0.56) 0.45 (0.37, 0.52) 1.15 (1.11, 1.2) 1.17 (1.12, 1.2)
Scaled kidney blood flow QKidC 0.092 (0.054, 0.13) 0.091 (0.064, 0.12) 1.34 (1.14, 1.45) 1.34 (1.18, 1.44)
Respiratory lumen:tissue
diffusive clearance rate
(L/hr)
DResp 0.017 (0.000032, 15) 2.5 (1.4, 5.1) 1.37 (1.25, 1.62) 1.53 (1.37, 1.73)
Fat fractional compartment
volume
VFatC 0.071 (0.032, 0.11) 0.089 (0.061, 0.11) 1.59 (1.19, 1.93) 1.4 (1.19, 1.78)
Gut fractional compartment
volume
VGutC 0.049 (0.041, 0.057) 0.048 (0.042, 0.055) 1.11 (1.07, 1.14) 1.11 (1.08, 1.14)
Liver fractional
compartment volume
VLivC 0.054 (0.038, 0.071) 0.047 (0.037, 0.06) 1.22 (1.12, 1.29) 1.23 (1.17, 1.3)
Rapidly perfused fractional
compartment volume
VRapC 0.1 (0.087, 0.11) 0.099 (0.09, 0.11) 1.08 (1.05, 1.11) 1.09 (1.06, 1.11)
Fractional volume of
respiratory lumen
VRespLumC 0.0047 (0.004, 0.0053) 0.0047 (0.0041, 0.0052) 1.09 (1.06, 1.12) 1.09 (1.07, 1.12)
Fractional volume of
respiratory tissue
VRespEffC 0.0007 (0.0006, 0.00079) 7e-04 (0.00062, 0.00078) 1.09 (1.06, 1.12) 1.1 (1.07, 1.12)
Kidney fractional
compartment volume
VKidC 0.017 (0.015, 0.019) 0.017 (0.015, 0.019) 1.08 (1.05, 1.11) 1.09 (1.06, 1.11)
Blood fractional
compartment volume
VBldC 0.049 (0.042, 0.056) 0.048 (0.043, 0.054) 1.1 (1.06, 1.13) 1.1 (1.08, 1.13)

3-85
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Slowly perfused fractional
compartment volume
VSlwC 0.55 (0.5, 0.59) 0.54 (0.51, 0.57) 1.05 (1.04, 1.07) 1.05 (1.04, 1.07)
Plasma fractional
compartment volume
VPlasC 0.026 (0.016, 0.036) 0.022 (0.016, 0.029) 1.24 (1.15, 1.35) 1.27 (1.19, 1.36)
TCA body fractional
compartment volume [not
incl. blood+liver]
VBodC 0.79 (0.77, 0.8) 0.79 (0.78, 0.81) 1.01 (1.01, 1.02) 1.01 (1.01, 1.02)
TCOH/G body fractional
compartment volume [not
incl. liver]
VBodTCOHC 0.84 (0.82, 0.85) 0.84 (0.83, 0.85) 1.01 (1.01, 1.02) 1.01 (1.01, 1.02)
TCE blood:air partition
coefficient
PB 15 (10, 23) 14 (11, 17) 1.22 (1.12, 1.42) 1.44 (1.28, 1.53)
TCE fat:blood partition
coefficient
PFat 36 (21, 62) 36 (26, 49) 1.26 (1.14, 1.52) 1.32 (1.16, 1.56)
TCE gut:blood partition
coefficient
PGut 1.9 (0.89, 3.8) 1.5 (0.94, 2.6) 1.36 (1.2, 1.75) 1.36 (1.2, 1.79)
TCE liver:blood partition
coefficient
PLiv 1.7 (0.89, 3.5) 2.2 (1.3, 3.3) 1.37 (1.2, 1.75) 1.39 (1.21, 1.84)
TCE rapidly perfused:blood
partition coefficient
PRap 1.8 (0.98, 3.7) 1.8 (1.1, 3) 1.37 (1.2, 1.76) 1.37 (1.2, 1.77)
TCE respiratory tissue:air
partition coefficient
PResp 2.7 (1.2, 5) 2.5 (1.5, 4.2) 1.36 (1.19, 1.78) 1.37 (1.19, 1.74)
TCE kidney:blood partition
coefficient
PKid 2.2 (0.96, 4.6) 2.6 (1.7, 4) 1.36 (1.2, 1.77) 1.51 (1.25, 1.88)
TCE slowly perfused:blood
partition coefficient
PSlw 2.4 (1.2, 4.9) 2.2 (1.4, 3.5) 1.38 (1.2, 1.78) 1.39 (1.21, 1.8)
TCA blood:plasma
concentration ratio
TCAPlas 0.76 (0.4, 16) 1.1 (0.75, 1.8) 1.21 (1.09, 1.58) 1.23 (1.1, 1.73)
Free TCA body:blood
plasma partition coefficient
PBodTCA 0.77 (0.27, 17) 0.87 (0.59, 1.5) 1.41 (1.23, 1.8) 1.39 (1.24, 1.9)
Free TCA liver:blood
plasma partition coefficient
PLivTCA 1.1 (0.36, 21) 1.1 (0.64, 1.9) 1.41 (1.23, 1.8) 1.4 (1.24, 1.87)
Protein:TCA dissociation
constant (mole/L)
kDissoc 100 (13, 790) 130 (24, 520) 2.44 (1.73, 5.42) 2.64 (1.75, 5.45)
3-86
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Maximum binding
concentration (mole/L)
B
MAX
87 (9.6, 790) 140 (28, 690) 2.72 (1.92, 5.78) 2.88 (1.93, 5.89)
TCOH body:blood partition
coefficient
PBodTCOH 1.1 (0.61, 2.1) 0.89 (0.65, 1.3) 1.29 (1.16, 1.66) 1.31 (1.17, 1.61)
TCOH liver:body partition
coefficient
PLivTCOH 1.3 (0.73, 2.3) 1.9 (1.2, 2.6) 1.3 (1.16, 1.61) 1.35 (1.18, 1.68)
TCOG body:blood partition
coefficient
PBodTCOG 0.95 (0.016, 77) 0.48 (0.18, 1.1) 1.36 (1.19, 2.05) 1.41 (1.22, 2.19)
TCOG liver:body partition
coefficient
PLivTCOG 1.3 (0.019, 92) 1.3 (0.64, 2.6) 1.36 (1.18, 2.13) 1.56 (1.28, 2.52)
DCVG effective volume of
distribution
VDCVG 0.033 (0.0015, 15) 0.027 (0.0016, 4.1) 1.28 (1.08, 1.97) 1.31 (1.1, 2.19)
TCE stomach absorption
coefficient (/hr)
kAS 1.7 (0.0049, 450) 1.7 (0.37, 13) 4.74 (2.29, 23.4) 4.28 (2.39, 13.4)
TCE stomach-duodenum
transfer coefficient (/hr)
kTSD 1.4 (0.043, 51) 4.5 (0.51, 26) 3.84 (2.09, 10.6) 4.79 (2.53, 10.9)
TCE duodenum absorption
coefficient (/hr)
kAD 1.2 (0.0024, 200) 0.27 (0.067, 1.6) 4.33 (2.14, 26) 4.17 (2.34, 14.4)
TCA stomach absorption
coefficient (/hr)
kASTCA 0.63 (0.0027, 240) 4 (0.2, 74) 4.26 (2.27, 23.4) 5.15 (2.56, 22)
V
MAX
for hepatic TCE
oxidation (mg/hr)
V
MAX
3.9 (1.4, 15) 2.5 (1.6, 4.2) 2.02 (1.56, 2.85) 1.86 (1.59, 2.47)
K
M
for hepatic TCE
oxidation (mg/L)
K
M
34 (1.6, 620) 2.7 (1.4, 8) 1.25 (1.15, 1.61) 2.08 (1.48, 3.49)
Fraction of hepatic TCE
oxidation not to
TCA+TCOH
FracOther 0.43 (0.0018, 1) 0.023 (0.0037, 0.15) 1.23 (1, 2.13) 1.49 (1.25, 2.83)
Fraction of hepatic TCE
oxidation to TCA
FracTCA 0.086 (0.00022, 0.66) 0.13 (0.084, 0.21) 1.48 (1.12, 2.56) 1.4 (1.21, 1.96)
V
MAX
for hepatic TCE GSH
conjugation (mg/hr)
V
MAX
DCVG 3.7 (0.0071, 2,800) 0.6 (0.01, 480) 1.55 (1.33, 2.52) 1.61 (1.37, 2.91)
K
M
for hepatic TCE GSH
conjugation (mg/L)
K
M
DCVG 250 (0.0029, 6,500,000) 2200 (0.17, 2,300,000) 1.81 (1.47, 3.62) 1.93 (1.49, 3.68)
3-87
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
V
MAX
for renal TCE GSH
conjugation (mg/hr)
V
MAX
KidDCVG 0.34 (0.00051, 180) 0.027 (0.0012, 13) 1.49 (1.26, 2.49) 1.54 (1.28, 2.72)
K
M
for renal TCE GSH
conjugation (mg/L)
K
M
KidDCVG 150 (0.0053, 6,200,000) 160 (0.078, 280,000) 1.79 (1.43, 3.45) 1.91 (1.5, 3.91)
V
MAX
for tracheo-bronchial
TCE oxidation (mg/hr)
V
MAX
Clara 0.24 (0.03, 3.9) 0.42 (0.1, 1.5) 2.32 (1.74, 3.66) 4.13 (2.27, 6.79)
K
M
for tracheo-bronchial
TCE oxidation (mg/L)
K
M
Clara 1.5 (0.0018, 630) 0.011 (0.0024, 0.09) 1.47 (1.25, 2.58) 1.63 (1.28, 5.02)
Fraction of respiratory
metabolism to systemic circ.
FracLungSys 0.34 (0.0016, 1) 0.78 (0.18, 0.99) 1.24 (1, 2.1) 1.11 (1, 1.72)
V
MAX
for hepatic
TCOHTCA (mg/hr)
V
MAX
TCOH 0.064 (0.000014, 380) 0.12 (0.048, 0.28) 1.5 (1.24, 2.61) 1.6 (1.28, 2.92)
K
M
for hepatic
TCOHTCA (mg/L)
K
M
TCOH 1.4 (0.00018, 5,300) 0.92 (0.26, 2.7) 1.48 (1.24, 2.41) 1.49 (1.26, 2.4)
V
MAX
for hepatic
TCOHTCOG (mg/hr)
V
MAX
Gluc 0.11 (0.000013, 310) 4.6 (1.9, 16) 1.48 (1.26, 2.53) 1.47 (1.26, 2.14)
K
M
for hepatic
TCOHTCOG (mg/L)
K
M
Gluc 1.8 (0.0018, 610) 30 (5.3, 130) 1.48 (1.25, 2.48) 1.8 (1.3, 4.72)
Rate constant for hepatic
TCOHother (/hr)
kMetTCOH 0.19 (0.000039, 1,400) 8.8 (1.9, 23) 1.47 (1.25, 2.36) 1.54 (1.26, 2.92)
Rate constant for TCA
plasmaurine (/hr)
kUrnTCA 32 (0.38, 1700) 3.2 (1.2, 7.1) 1.57 (1.34, 2.61) 1.84 (1.44, 2.94)
Rate constant for hepatic
TCAother (/hr)
kMetTCA 0.12 (0.0004, 130) 1.5 (0.63, 2.9) 1.48 (1.25, 2.32) 1.51 (1.26, 2.27)
Rate constant for TCOG
liverbile (/hr)
kBile 0.3 (0.0004, 160) 2.4 (0.74, 8.4) 1.48 (1.24, 2.29) 1.51 (1.26, 2.39)
Lumped rate constant for
TCOG bileTCOH liver
(/hr)
kEHR 0.21 (0.00036, 150) 0.039 (0.0026, 0.11) 1.47 (1.23, 2.29) 1.53 (1.28, 2.94)
Rate constant for
TCOGurine (/hr)
kUrnTCOG 1 (0.00015, 6,200) 12 (2.6, 77) 1.71 (1.4, 3.13) 3.44 (1.89, 9.49)
Rate constant for hepatic
DCVGDCVC (/hr)
kDCVG 0.24 (0.0004, 160) 0.81 (0.0033, 46) 1.48 (1.25, 2.39) 1.52 (1.25, 2.5)
3-88
Table 3-37. Prior and posterior uncertainty and variability in mouse PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Lumped rate constant for
DCVCurinary NAcDCVC
(/hr)
kNAT 0.29 (0.0004, 160) 0.37 (0.0021, 34) 1.5 (1.25, 2.49) 1.53 (1.25, 2.77)
Rate constant for DCVC
bioactivation (/hr)
kKidBioact 0.18 (0.0004, 150) 0.23 (0.0024, 33) 1.48 (1.25, 2.51) 1.53 (1.25, 3.03)

3-89
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model parameters

Parameter description
PBPK
parameter
Prior population
median: median
(2.5%, 97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Cardiac output (L/hr) QC 5.3 (4.2, 6.9) 6.1 (5.2, 7.4) 1.12 (1.07, 1.28) 1.26 (1.12, 1.36)
Alveolar ventilation (L/hr) QP 10 (5.1, 18) 7.5 (5.8, 10) 1.32 (1.18, 1.71) 1.52 (1.33, 1.84)
Scaled fat blood flow QFatC 0.071 (0.032, 0.11) 0.081 (0.06, 0.1) 1.66 (1.21, 2.02) 1.5 (1.3, 1.86)
Scaled gut blood flow QGutC 0.15 (0.12, 0.18) 0.17 (0.15, 0.19) 1.15 (1.09, 1.19) 1.13 (1.08, 1.18)
Scaled liver blood flow QLivC 0.021 (0.017, 0.026) 0.022 (0.018, 0.025) 1.15 (1.09, 1.2) 1.15 (1.1, 1.19)
Scaled slowly perfused
blood flow
QSlwC 0.33 (0.21, 0.46) 0.31 (0.23, 0.4) 1.31 (1.15, 1.4) 1.32 (1.22, 1.41)
Scaled rapidly perfused
blood flow
QRapC 0.28 (0.15, 0.42) 0.28 (0.18, 0.36) 1.38 (0.0777, 1.72) 1.42 (0.0856, 1.75)
Scaled kidney blood flow QKidC 0.14 (0.12, 0.16) 0.14 (0.12, 0.16) 1.11 (1.07, 1.14) 1.11 (1.08, 1.14)
Respiratory lumen:tissue
diffusive clearance rate
(L/hr)
DResp 9.9 (0.48, 85) 21 (9.5, 46) 1.41 (1.26, 1.77) 1.59 (1.41, 1.9)
Fat fractional
compartment volume
VFatC 0.069 (0.031, 0.11) 0.069 (0.046, 0.091) 1.61 (1.2, 1.93) 1.59 (1.34, 1.88)
Gut fractional
compartment volume
VGutC 0.032 (0.027, 0.037) 0.032 (0.028, 0.036) 1.11 (1.07, 1.14) 1.11 (1.08, 1.14)
Liver fractional
compartment volume
VLivC 0.034 (0.026, 0.042) 0.033 (0.028, 0.039) 1.16 (1.09, 1.21) 1.17 (1.12, 1.21)
Rapidly perfused
fractional compartment
volume
VRapC 0.087 (0.076, 0.1) 0.088 (0.079, 0.097) 1.1 (1.06, 1.13) 1.1 (1.07, 1.13)
Fractional volume of
respiratory lumen
VRespLumC 0.0046 (0.0037, 0.0057) 0.0047 (0.0039, 0.0055) 1.16 (1.1, 1.21) 1.16 (1.11, 1.21)
Fractional volume of
respiratory tissue
VRespEffC 0.0005 (0.00039,
0.00061)
5e-04 (0.00041, 0.00058) 1.16 (1.09, 1.21) 1.16 (1.11, 1.2)
Kidney fractional
compartment volume
VKidC 0.0069 (0.0056, 0.0082) 0.007 (0.006, 0.008) 1.13 (1.08, 1.17) 1.13 (1.09, 1.17)

3-90
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Blood fractional
compartment volume
VBldC 0.073 (0.063, 0.085) 0.074 (0.066, 0.082) 1.1 (1.06, 1.13) 1.1 (1.07, 1.13)
Slowly perfused fractional
compartment volume
VSlwC 0.6 (0.55, 0.63) 0.6 (0.57, 0.62) 1.05 (1.04, 1.06) 1.05 (1.04, 1.06)
Plasma fractional
compartment volume
VPlasC 0.039 (0.025, 0.054) 0.04 (0.032, 0.049) 1.24 (1.15, 1.35) 1.22 (1.16, 1.33)
TCA body fractional
compartment volume [not
incl. blood+liver]
VBodC 0.79 (0.78, 0.81) 0.79 (0.78, 0.8) 1.01 (1.01, 1.01) 1.01 (1.01, 1.01)
TCOH/G body fractional
compartment volume [not
incl. liver]
VBodTCOHC 0.87 (0.86, 0.87) 0.87 (0.86, 0.87) 1.01 (1, 1.01) 1.01 (1, 1.01)
TCE blood:air partition
coefficient
PB 22 (14, 33) 19 (16, 24) 1.26 (1.19, 1.35) 1.3 (1.22, 1.38)
TCE fat:blood partition
coefficient
PFat 27 (16, 46) 31 (24, 42) 1.32 (1.22, 1.44) 1.32 (1.23, 1.43)
TCE gut:blood partition
coefficient
PGut 1.3 (0.69, 3) 1.1 (0.79, 1.7) 1.36 (1.21, 1.79) 1.36 (1.2, 1.68)
TCE liver:blood partition
coefficient
PLiv 1.5 (1.2, 1.9) 1.6 (1.3, 1.8) 1.15 (1.11, 1.2) 1.15 (1.11, 1.2)
TCE rapidly perfused:blood
partition coefficient
PRap 1.3 (0.66, 2.7) 1.3 (0.82, 2.1) 1.35 (1.18, 1.82) 1.37 (1.2, 1.76)
TCE respiratory tissue:air
partition coefficient
PResp 0.97 (0.48, 2.1) 1 (0.62, 1.6) 1.37 (1.19, 1.77) 1.36 (1.19, 1.78)
TCE kidney:blood partition
coefficient
PKid 1.3 (0.77, 2.2) 1.2 (0.9, 1.7) 1.31 (1.19, 1.5) 1.3 (1.2, 1.45)
TCE slowly perfused:blood
partition coefficient
PSlw 0.57 (0.35, 0.97) 0.73 (0.54, 0.97) 1.32 (1.23, 1.43) 1.33 (1.25, 1.46)
TCA blood:plasma
concentration ratio
TCAPlas 0.78 (0.6, 0.96) 0.78 (0.71, 0.86) 1.12 (1.06, 1.22) 1.11 (1.07, 1.17)
Free TCA body:blood
plasma partition coefficient
PBodTCA 0.7 (0.18, 2.2) 0.76 (0.46, 1.3) 1.72 (1.39, 2.81) 1.65 (1.4, 2.19)
Free TCA liver:blood
plasma partition coefficient
PLivTCA 0.84 (0.25, 3.3) 1.1 (0.61, 2.1) 1.71 (1.39, 2.78) 1.66 (1.38, 2.37)
3-91
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Protein:TCA dissociation
constant (mole/L)
kDissoc 270 (95, 790) 280 (140, 530) 1.62 (1.31, 2.43) 1.6 (1.31, 2.31)
Maximum binding
concentration (mole/L)
B
MAX
320 (80, 1300) 320 (130, 750) 1.89 (1.5, 2.64) 1.84 (1.49, 2.57)
TCOH body:blood partition
coefficient
PBodTCOH 1 (0.33, 4) 1.1 (0.51, 2.1) 1.71 (1.37, 2.69) 1.76 (1.38, 2.45)
TCOH liver:body partition
coefficient
PLivTCOH 1.3 (0.39, 4.5) 1.2 (0.59, 2.8) 1.71 (1.37, 2.8) 1.78 (1.37, 2.75)
TCOG body:blood partition
coefficient
PBodTCOG 0.48 (0.021, 14) 1.6 (0.091, 16) 1.39 (1.2, 1.97) 1.42 (1.21, 2.52)
TCOG liver:body partition
coefficient
PLivTCOG 1.3 (0.078, 39) 10 (2.7, 41) 1.4 (1.2, 2.14) 1.42 (1.21, 2.3)
DCVG effective volume of
distribution
VDCVG 0.27 (0.27, 0.27) 0.27 (0.27, 0.27) 1 (1, 1) 1 (1, 1)
TCE stomach absorption
coefficient (/hr)
kAS 0.73 (0.0044, 400) 2.5 (0.32, 19) 4.16 (2.21, 20) 9.3 (4.07, 31.1)
TCE stomach-duodenum
transfer coefficient (/hr)
kTSD 1.4 (0.04, 45) 3.2 (0.31, 19) 3.92 (2.13, 10.4) 5.54 (2.77, 10.7)
TCE duodenum absorption
coefficient (/hr)
kAD 0.96 (0.0023, 260) 0.17 (0.038, 1) 4.17 (2.15, 20.8) 4.07 (2.51, 11.9)
TCA stomach absorption
coefficient (/hr)
kASTCA 0.83 (0.0024, 240) 1.4 (0.13, 13) 4.15 (2.2, 18.7) 4.21 (2.4, 11.4)
V
MAX
for hepatic TCE
oxidation (mg/hr)
V
MAX
5.8 (2, 19) 5.3 (3.9, 7.7) 1.97 (1.54, 2.92) 1.69 (1.47, 2.15)
K
M
for hepatic TCE
oxidation (mg/L)
K
M
18 (1.9, 240) 0.74 (0.54, 1.4) 2.76 (1.89, 6.46) 1.84 (1.51, 2.7)
Fraction of hepatic TCE
oxidation not to
TCA+TCOH
FracOther 0.027 (0.0018, 0.59) 0.29 (0.047, 0.56) 1.42 (1.15, 2.33) 2.15 (1.32, 5.06)
Fraction of hepatic TCE
oxidation to TCA
FracTCA 0.2 (0.027, 0.76) 0.046 (0.023, 0.087) 1.35 (1.11, 2.14) 1.84 (1.36, 2.8)
V
MAX
for hepatic TCE GSH
conjugation (mg/hr)
V
MAX
DCVG 2 (0.015, 1,100) 5.8 (0.16, 340) 1.52 (1.3, 2.67) 1.57 (1.32, 2.93)
K
M
for hepatic TCE GSH
conjugation (mg/L)
K
M
DCVG 1,500 (1.2, 1,800,000) 6300 (120, 720,000) 1.83 (1.45, 3.15) 1.88 (1.48, 3.49)
3-92
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
V
MAX
for renal TCE GSH
conjugation (mg/hr)
V
MAX
KidDCVG 0.038 (0.00027, 13) 0.0024 (0.0005, 0.014) 1.52 (1.3, 2.81) 1.56 (1.29, 2.72)
K
M
for renal TCE GSH
conjugation (mg/L)
K
M
KidDCVG 470 (0.47, 530,000) 0.25 (0.038, 2.2) 1.84 (1.47, 4.27) 1.93 (1.49, 3.57)
V
MAX
for tracheo-bronchial
TCE oxidation (mg/hr)
V
MAX
Clara 0.2 (0.0077, 2.4) 0.17 (0.042, 0.69) 2.26 (1.71, 3.3) 4.35 (1.99, 6.7)
K
M
for tracheo-bronchial
TCE oxidation (mg/L)
K
M
Clara 0.016 (0.0014, 0.58) 0.025 (0.005, 0.15) 1.47 (1.26, 2.39) 1.65 (1.28, 10.5)
Fraction of respiratory
metabolism to systemic circ.
FracLungSys 0.82 (0.027, 1) 0.73 (0.06, 0.98) 1.09 (1, 1.71) 1.13 (1.01, 1.86)
V
MAX
for hepatic
TCOHTCA (mg/hr)
V
MAX
TCOH 0.75 (0.037, 20) 0.71 (0.27, 2.2) 1.51 (1.25, 2.64) 1.68 (1.3, 3.23)
K
M
for hepatic
TCOHTCA (mg/L)
K
M
TCOH 1 (0.029, 23) 19 (3.6, 94) 1.52 (1.26, 2.7) 1.72 (1.26, 3.93)
V
MAX
for hepatic
TCOHTCOG (mg/hr)
V
MAX
Gluc 27 (0.83, 620) 11 (4.1, 32) 1.5 (1.25, 2.59) 2.3 (1.41, 5.19)
K
M
for hepatic
TCOHTCOG (mg/L)
K
M
Gluc 31 (1, 570) 6.3 (1.2, 20) 1.5 (1.25, 2.74) 2.04 (1.3, 8.4)
Rate constant for hepatic
TCOHother (/hr)
kMetTCOH 4.2 (0.17, 150) 3 (0.57, 15) 1.49 (1.27, 2.67) 1.72 (1.3, 8.31)
Rate constant for TCA
plasmaurine (/hr)
kUrnTCA 1.9 (0.21, 47) 0.92 (0.51, 1.7) 1.56 (1.33, 2.81) 1.58 (1.36, 2.25)
Rate constant for hepatic
TCAother (/hr)
kMetTCA 0.76 (0.037, 19) 0.47 (0.17, 1.2) 1.5 (1.26, 2.74) 1.52 (1.27, 2.45)
Rate constant for TCOG
liverbile (/hr)
kBile 1.4 (0.052, 31) 14 (2.7, 39) 1.5 (1.25, 2.8) 1.63 (1.29, 4.1)
Lumped rate constant for
TCOG bileTCOH liver
(/hr)
kEHR 0.013 (0.00055, 0.64) 1.7 (0.34, 7.4) 1.5 (1.25, 2.49) 1.67 (1.26, 5.91)
Rate constant for
TCOGurine (/hr)
kUrnTCOG 11 (0.063, 1,000) 12 (0.45, 370) 1.74 (1.42, 2.99) 1.86 (1.43, 3.54)
Rate constant for hepatic
DCVGDCVC (/hr)
kDCVG 30,000 (30,000, 30,000) 30,000 (30,000, 30,000) 1 (1, 1) 1 (1, 1)
3-93
Table 3-38. Prior and posterior uncertainty and variability in rat PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Lumped rate constant for
DCVCurinary NAcDCVC
(/hr)
kNAT 0.15 (0.00024, 84) 0.0029 (0.00066, 0.015) 1.49 (1.24, 2.8) 1.54 (1.26, 2.45)
Rate constant for DCVC
bioactivation (/hr)
kKidBioact 0.12 (0.00023, 83) 0.0092 (0.0012, 0.043) 1.48 (1.24, 2.68) 1.52 (1.25, 2.5)

3-94
Table 3-39. Prior and posterior uncertainty and variability in human PBPK model parameters

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Cardiac output (L/hr) QC 390 (280, 560) 330 (280, 390) 1.17 (1.1, 1.39) 1.39 (1.26, 1.54)
Alveolar ventilation (L/hr) QP 380 (220, 640) 440 (360, 530) 1.27 (1.17, 1.52) 1.58 (1.44, 1.73)
Scaled fat blood flow QFatC 0.051 (0.021, 0.078) 0.043 (0.033, 0.055) 1.64 (1.23, 2) 1.92 (1.72, 2.09)
Scaled gut blood flow QGutC 0.19 (0.15, 0.23) 0.16 (0.14, 0.18) 1.16 (1.1, 1.21) 1.16 (1.12, 1.2)
Scaled liver blood flow QLivC 0.063 (0.029, 0.099) 0.039 (0.026, 0.055) 1.62 (1.22, 1.92) 1.8 (1.62, 1.98)
Scaled slowly perfused
blood flow
QSlwC 0.22 (0.13, 0.3) 0.17 (0.14, 0.21) 1.34 (1.18, 1.45) 1.39 (1.31, 1.46)
Scaled rapidly perfused
blood flow
QRapC 0.29 (0.18, 0.4) 0.39 (0.34, 0.43) 1.31 (1.14, 1.57) 1.22 (1.16, 1.3)
Scaled kidney blood flow QKidC 0.19 (0.16, 0.22) 0.19 (0.18, 0.21) 1.1 (1.07, 1.13) 1.1 (1.07, 1.12)
Respiratory lumen:tissue
diffusive clearance rate
(L/hr)
DResp 560 (44, 3300) 270 (130, 470) 1.37 (1.25, 1.61) 1.71 (1.52, 2.35)
Fat fractional compartment
volume
VFatC 0.19 (0.088, 0.31) 0.16 (0.12, 0.21) 1.66 (1.23, 1.93) 1.65 (1.4, 1.9)
Gut fractional compartment
volume
VGutC 0.02 (0.018, 0.022) 0.02 (0.019, 0.021) 1.07 (1.04, 1.08) 1.06 (1.05, 1.08)
Liver fractional
compartment volume
VLivC 0.026 (0.018, 0.032) 0.026 (0.022, 0.03) 1.21 (1.12, 1.28) 1.2 (1.13, 1.26)
Rapidly perfused fractional
compartment volume
VRapC 0.087 (0.079, 0.096) 0.088 (0.083, 0.093) 1.07 (1.05, 1.09) 1.06 (1.05, 1.08)
Fractional volume of
respiratory lumen
VRespLumC 0.0024 (0.0018, 0.003) 0.0024 (0.0021, 0.0027) 1.18 (1.1, 1.23) 1.17 (1.12, 1.22)
Fractional volume of
respiratory tissue
VRespEffC 0.00018 (0.00014,
0.00022)
0.00018 (0.00015, 0.00021) 1.18 (1.1, 1.24) 1.17 (1.13, 1.23)
Kidney fractional
compartment volume
VKidC 0.0043 (0.0034, 0.0052) 0.0043 (0.0038, 0.0048) 1.15 (1.09, 1.19) 1.14 (1.1, 1.19)

3-95
Table 3-39. Prior and posterior uncertainty and variability in human PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Blood fractional
compartment volume
VBldC 0.077 (0.066, 0.088) 0.078 (0.072, 0.084) 1.1 (1.06, 1.13) 1.1 (1.07, 1.13)
Slowly perfused fractional
compartment volume
VSlwC 0.45 (0.33, 0.55) 0.48 (0.43, 0.52) 1.18 (1.1, 1.24) 1.16 (1.12, 1.22)
Plasma fractional
compartment volume
VPlasC 0.044 (0.037, 0.051) 0.044 (0.04, 0.048) 1.11 (1.08, 1.14) 1.11 (1.08, 1.14)
TCA body fractional
compartment volume [not
incl. blood+liver]
VBodC 0.75 (0.74, 0.77) 0.75 (0.74, 0.76) 1.01 (1.01, 1.01) 1.01 (1.01, 1.01)
TCOH/G body fractional
compartment volume [not
incl. liver]
VBodTCOHC 0.83 (0.82, 0.84) 0.83 (0.83, 0.83) 1.01 (1, 1.01) 1.01 (1, 1.01)
TCE blood:air partition
coefficient
PB 9.6 (6.5, 13) 9.2 (8.2, 10) 1.18 (1.13, 1.26) 1.21 (1.16, 1.28)
TCE fat:blood partition
coefficient
PFat 68 (46, 98) 57 (49, 66) 1.18 (1.11, 1.33) 1.18 (1.11, 1.3)
TCE gut:blood partition
coefficient
PGut 2.6 (1.3, 5.3) 2.9 (1.9, 4.1) 1.37 (1.2, 1.78) 1.41 (1.21, 1.77)
TCE liver:blood partition
coefficient
PLiv 4 (1.9, 8.5) 4.1 (2.7, 5.9) 1.37 (1.22, 1.81) 1.33 (1.19, 1.6)
TCE rapidly perfused:blood
partition coefficient
PRap 2.6 (1.2, 5.7) 2.4 (1.8, 3.2) 1.37 (1.21, 1.78) 1.5 (1.25, 1.87)
TCE respiratory tissue:air
partition coefficient
PResp 1.3 (0.65, 2.7) 1.3 (0.9, 1.9) 1.36 (1.19, 1.81) 1.32 (1.2, 1.56)
TCE kidney:blood partition
coefficient
PKid 1.6 (1.1, 2.3) 1.6 (1.3, 1.9) 1.17 (1.1, 1.33) 1.15 (1.09, 1.25)
TCE slowly perfused:blood
partition coefficient
PSlw 2.1 (1.2, 3.5) 2.3 (1.9, 2.8) 1.28 (1.14, 1.53) 1.51 (1.36, 1.66)
TCA blood:plasma
concentration ratio
TCAPlas 0.78 (0.55, 15) 0.65 (0.6, 0.77) 1.08 (1.03, 1.53) 1.52 (1.23, 2.03)
Free TCA body:blood
plasma partition coefficient
PBodTCA 0.45 (0.19, 8.1) 0.44 (0.33, 0.55) 1.36 (1.19, 1.75) 1.67 (1.38, 2.2)
Free TCA liver:blood
plasma partition coefficient
PLivTCA 0.59 (0.24, 10) 0.55 (0.39, 0.77) 1.36 (1.18, 1.76) 1.65 (1.37, 2.16)
Protein:TCA dissociation
constant (mole/L)
kDissoc 180 (160, 200) 180 (170, 190) 1.05 (1.03, 1.09) 1.04 (1.03, 1.07)
3-96
Table 3-39. Prior and posterior uncertainty and variability in human PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
Maximum binding
concentration (mole/L)
B
MAX
830 (600, 1100) 740 (630, 880) 1.17 (1.1, 1.3) 1.16 (1.1, 1.28)
TCOH body:blood partition
coefficient
PBodTCOH 0.89 (0.51, 1.7) 1.5 (1.3, 1.7) 1.29 (1.16, 1.64) 1.34 (1.25, 1.47)
TCOH liver:body partition
coefficient
PLivTCOH 0.58 (0.32, 1.1) 0.63 (0.45, 0.87) 1.29 (1.16, 1.65) 1.29 (1.17, 1.5)
TCOG body:blood partition
coefficient
PBodTCOG 0.67 (0.036, 16) 0.72 (0.3, 1.8) 1.38 (1.2, 2.42) 7.83 (4.86, 12.6)
TCOG liver:body partition
coefficient
PLivTCOG 1.8 (0.11, 28) 3.1 (0.87, 8.1) 1.38 (1.19, 2.04) 4.94 (2.73, 8.58)
DCVG effective volume of
distribution
VDCVG 73 (5.2, 36000) 6.1 (5.4, 7.3) 1.27 (1.08, 1.95) 1.1 (1.07, 1.16)
TCE stomach absorption
coefficient (/hr)
kAS 1.4 (1.4, 1.4) 1.4 (1.4, 1.4) 1 (1, 1) 1 (1, 1)
TCE stomach-duodenum
transfer coefficient (/hr)
kTSD 1.4 (1.4, 1.4) 1.4 (1.4, 1.4) 1 (1, 1) 1 (1, 1)
TCE duodenum absorption
coefficient (/hr)
kAD 0.75 (0.75, 0.75) 0.75 (0.75, 0.75) 1 (1, 1) 1 (1, 1)
TCA stomach absorption
coefficient (/hr)
kASTCA 0.58 (0.0022, 210) 3 (0.061, 180) 4.26 (2.13, 17.6) 5.16 (2.57, 22.3)
TCOH stomach absorption
coefficient (/hr)
kASTCOH 0.49 (0.0024, 210) 7.6 (0.11, 150) 4.19 (2.22, 21.5) 5.02 (2.44, 18.5)
V
MAX
for hepatic TCE
oxidation (mg/hr)
V
MAX
430 (130, 1500) 190 (130, 290) 1.98 (1.69, 2.31) 2.02 (1.77, 2.38)
K
M
for hepatic TCE
oxidation (mg/L)
K
M
3.7 (0.22, 63) 0.18 (0.078, 0.4) 2.74 (2.1, 5.62) 4.02 (2.9, 5.64)
Fraction of hepatic TCE
oxidation not to
TCA+TCOH
FracOther 0.12 (0.0066, 0.7) 0.11 (0.024, 0.23) 1.4 (1.11, 2.38) 2.71 (1.37, 5.33)
Fraction of hepatic TCE
oxidation to TCA
FracTCA 0.19 (0.036, 0.56) 0.035 (0.024, 0.05) 2.55 (1.51, 3.96) 2.25 (1.89, 2.87)
V
MAX
for hepatic TCE GSH
conjugation (mg/hr)
V
MAX
DCVG 100 (0.0057, 690,000) 340 (110, 1,100) 1.91 (1.55, 3.76) 6.18 (3.35, 11.3)
K
M
for hepatic TCE GSH
conjugation (mg/L)
K
M
DCVG 3.1 (0.21, 42) 3.6 (1.2, 11) 1.52 (1.26, 2.91) 4.2 (2.48, 8.01)
V
MAX
for renal TCE GSH
conjugation (mg/hr)
V
MAX
KidDCVG 220 (0.028, 6,700,000) 2.1 (0.17, 9.3) 1.86 (1.51, 3.33) 4.02 (1.57, 33.9)
3-97
Table 3-39. Prior and posterior uncertainty and variability in human PBPK model parameters (continued)

Parameter description PBPK parameter
Prior population
median: median (2.5%,
97.5%)
Posterior population
median: median (2.5%,
97.5%)
Prior population GSD:
median (2.5%, 97.5%)
Posterior population
GSD: median (2.5%,
97.5%)
K
M
for renal TCE GSH
conjugation (mg/L)
K
M
KidDCVG 2.7 (0.14, 41) 0.76 (0.29, 5.8) 1.5 (1.27, 2.56) 1.49 (1.27, 2.32)
V
MAX
for tracheo-bronchial
TCE oxidation (mg/hr)
V
MAX
Clara 25 (1, 260) 18 (3.8, 41) 2.25 (1.85, 3.25) 2.9 (2.12, 6.49)
K
M
for tracheo-bronchial
TCE oxidation (mg/L)
K
M
Clara 0.019 (0.0017, 0.5) 0.31 (0.057, 1.4) 1.48 (1.25, 2.39) 10.8 (1.99, 37.6)
Fraction of respiratory
metabolism to systemic circ.
FracLungSys 0.75 (0.051, 0.99) 0.96 (0.86, 0.99) 1.12 (1, 1.75) 1.02 (1, 1.1)
V
MAX
for hepatic
TCOHTCA (mg/hr)
V
MAX
TCOH 42 (0.77, 2,200) 9.2 (5.5, 20) 1.83 (1.46, 3.43) 3.15 (2.3, 5.44)
K
M
for hepatic
TCOHTCA (mg/L)
K
M
TCOH 5 (0.23, 81) 2.2 (1.3, 4.5) 1.49 (1.25, 2.57) 2.58 (1.75, 4.5)
V
MAX
for hepatic
TCOHTCOG (mg/hr)
V
MAX
Gluc 720 (12, 50,000) 900 (340, 2,000) 1.83 (1.48, 3.5) 2.29 (1.84, 4.57)
K
M
for hepatic
TCOHTCOG (mg/L)
K
M
Gluc 10 (0.53, 190) 130 (47, 290) 1.5 (1.25, 2.6) 1.58 (1.26, 3.69)
Rate constant for hepatic
TCOHother (/hr)
kMetTCOH 0.83 (0.035, 10) 0.25 (0.042, 0.7) 1.5 (1.26, 3) 5.13 (2.72, 16.7)
Rate constant for TCA
plasmaurine (/hr)
kUrnTCA 0.26 (0.038, 4) 0.11 (0.083, 0.15) 1.48 (1.29, 2.29) 1.86 (1.58, 2.28)
Rate constant for hepatic
TCAother (/hr)
kMetTCA 0.19 (0.01, 2.6) 0.096 (0.038, 0.19) 1.48 (1.26, 2.57) 2.52 (1.79, 4.34)
Rate constant for TCOG
liverbile (/hr)
kBile 1.2 (0.059, 16) 2.5 (1.1, 6.9) 1.47 (1.25, 2.75) 1.56 (1.27, 3.21)
Lumped rate constant for
TCOG bileTCOH liver
(/hr)
kEHR 0.074 (0.004, 1.4) 0.053 (0.033, 0.087) 1.52 (1.26, 2.64) 1.72 (1.35, 2.51)
Rate constant for
TCOGurine (/hr)
kUrnTCOG 2.9 (0.061, 260) 2.4 (0.83, 7) 1.75 (1.4, 3.31) 18.7 (11.6, 31.8)
Rate constant for hepatic
DCVGDCVC (/hr)
kDCVG 0.044 (0.000063, 22) 2.5 (1.9, 3.4) 1.48 (1.25, 2.83) 1.51 (1.3, 1.86)
Lumped rate constant for
DCVCurinary
NAcDCVC (/hr)
kNAT 0.00085 (0.000055, 0.041) 0.0001 (0.000047, 0.0007) 1.51 (1.25, 2.34) 1.47 (1.24, 2.48)
Rate constant for DCVC
bioactivation (/hr)
kKidBioact 0.0022 (0.000095, 0.079) 0.023 (0.0062, 0.061) 1.51 (1.25, 2.57) 1.52 (1.25, 2.69)
3-98
Table 3-40. CI widths (ratio of 97.52.5% estimates) and fold-shift in median estimate for the PBPK model
population median parameters, sorted in order of decreasing CI width
a


Mouse Rat Human

Width of CI on population
median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
K
M
DCVG 2,230,000,000 13,400,000 8.8 K
M
DCVG 1,500,000 5,800 4.29 kASTCA 94,300 3,040 5.18
K
M
KidDCVG 1,170,000,000 3,540,000 1.05 V
MAX
DCVG 71,100 2,130 2.86 kASTCOH 85,900 1,420 15.6
V
MAX
DCVG 400,000 46,200 6.18 kUrnTCOG 16,700 822 1.04
V
MAX
-
KidDCVG 236,000,000 55.1 105
V
MAX
KidDCVG 357,000 11,000 12.8 PBodTCOG 666 172 3.43 K
M
Clara 289 23.9 16.2
kASTCA 89,300 374 6.3 kASTCA 98,200 95.7 1.69 K
M
KidDCVG 287 20 3.48
kTSD 1,190 51.1 3.26 kTSD 1,130 61.8 2.29 kMetTCOH 289 16.6 3.28
kEHR 412,000 42.1 5.43 kAS 91,000 60.2 3.41 kNAT 756 15.1 8.14
FracOther 567 39.5 18.5 K
M
KidDCVG 1,130,000 58.6 1880 V
MAX
Clara 255 10.6 1.41
K
M
Clara 351,000 37.5 134 kKidBioact 366,000 35.6 13.3 kKidBioact 833 9.91 10.5
kAS 91,900 35.9 1 K
M
Clara 406 29.9 1.53 V
MAX
DCVG 122,000,000 9.78 3.29
kUrnTCOG 4,0500,000 29.9 11.8 V
MAX
KidDCVG 48,500 27.5 15.6 FracOther 106 9.75 1.09
B
MAX
81.8 24.4 1.66 kMetTCOH 891 26.4 1.41 PLivTCOG 253 9.32 1.77
K
M
Gluc 344,000 24.3 16.3 kAD 115,000 26.3 5.53 K
M
DCVG 198 9.13 1.18
kAD 84,900 23.8 4.53 K
M
TCOH 781 26 18.7 kUrnTCOG 4,290 8.5 1.19
kDissoc 60.3 21.8 1.33 kNAT 351,000 22.7 50.2 kBile 274 6.54 2.01
V
MAX
Clara 131 15 1.75 kEHR 1,160 21.9 134 K
M
Gluc 365 6.07 13.4
kMetTCOH 35,500,000 12.1 47.4 K
M
Gluc 562 17.1 4.98 PBodTCOG 454 5.85 1.08
kBile 390,000 11.3 8.23 V
MAX
Clara 305 16.5 1.21 V
MAX
Gluc 4,330 5.71 1.25
K
M
TCOH 29,600,000 10.5 1.47 FracLungSys 36.7 16.3 1.12 K
M
288 5.1 20.5
V
MAX
Gluc 23,600,000 8.28 41.1 PLivTCOG 501 14.8 8.07 kMetTCA 248 4.89 1.94

3-99
Table 3-40. CI widths (ratio of 97.52.5% estimates) and fold-shift in median estimate for the PBPK model
population median parameters, sorted in order of decreasing CI width
a
(continued)

Mouse Rat Human

Width of CI on population
median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
PBodTCOG 4,770 6.27 1.95 kBile 588 14.8 9.67 DResp 74.3 3.71 2.06
V
MAX
TCOH 27,100,000 5.78 1.8 FracOther 331 11.9 10.7 V
MAX
TCOH 2,900 3.62 4.56
K
M
386 5.76 12.5 V
MAX
TCOH 550 8.25 1.06 K
M
TCOH 359 3.48 2.33
kUrnTCA 4,540 5.76 10.2 V
MAX
Gluc 740 7.79 2.4 kEHR 339 2.62 1.39
FracLungSys 608 5.55 2.27 kMetTCA 507 6.93 1.61 V
MAX
11.5 2.27 2.33
kMetTCA 316,000 4.59 12 B
MAX
16.2 5.79 1 PResp 4.1 2.16 1.01
PLivTCOG 4,860 3.99 1.04 DResp 180 4.81 2.12 PLiv 4.44 2.14 1.02
DResp 475,000 3.64 147 PLivTCOH 11.5 4.7 1.09 QLivC 3.46 2.11 1.62
PLivTCA 58.3 2.88 1 PBodTCOH 12.1 4.03 1.03 PGut 4.21 2.1 1.11
PResp 4 2.85 1.07 kDissoc 8.38 3.85 1.04 FracTCA 15.5 2.06 5.37
PRap 3.78 2.79 1.03 FracTCA 28.1 3.85 4.27 PLivTCA 42.6 1.98 1.07
PGut 4.33 2.77 1.25 PLivTCA 13.3 3.49 1.37 PLivTCOH 3.52 1.93 1.08
V
MAX
10.7 2.67 1.58 kUrnTCA 219 3.28 2 kDCVG 344,000 1.8 55.7
PBodTCA 62.6 2.55 1.14 PBodTCA 12 2.8 1.09 kUrnTCA 105 1.79 2.32
PSlw 4.04 2.54 1.06 PResp 4.32 2.6 1.04 VFatC 3.49 1.76 1.21
PLiv 3.87 2.5 1.26 K
M
123 2.56 24 PRap 4.66 1.74 1.09
FracTCA 3,060 2.49 1.49 PRap 4.01 2.53 1.01 QFatC 3.7 1.7 1.19
TCAPlas 40.6 2.38 1.46 PGut 4.35 2.16 1.17 PBodTCA 42.9 1.7 1.04
PKid 4.78 2.37 1.2 V
MAX
9.5 1.98 1.11 PSlw 2.9 1.5 1.11
QFatC 3.62 2.26 1.02 QRapC 2.77 1.97 1 PKid 2.05 1.49 1.01
PLivTCOH 3.19 2.13 1.48 VFatC 3.58 1.96 1 QP 2.97 1.48 1.16
PBodTCOH 3.41 2.01 1.27 PKid 2.89 1.85 1.11 QSlwC 2.25 1.48 1.26
QKidC 2.39 1.91 1.01 QP 3.59 1.79 1.38 QC 2.04 1.39 1.19
PFat 3.01 1.89 1.01 PSlw 2.76 1.79 1.28 B
MAX
1.92 1.38 1.12
QSlwC 2.04 1.88 1.02 PFat 2.91 1.77 1.16 VLivC 1.79 1.36 1.01
VPlasC 2.18 1.87 1.17 QSlwC 2.19 1.69 1.06 PFat 2.13 1.34 1.2
VFatC 3.49 1.83 1.25 QFatC 3.47 1.66 1.14 VDCVG 6,820 1.34 12
QP 2.75 1.82 1.02 VPlasC 2.17 1.55 1.03 VRespEffC 1.66 1.33 1.02
VLivC 1.85 1.6 1.16 PB 2.37 1.51 1.15 PBodTCOH 3.32 1.32 1.68
3-100
Table 3-40. CI widths (ratio of 97.52.5% estimates) and fold-shift in median estimate for the PBPK
a
population median parameters, sorted in order of decreasing CI width (continued)
model

Mouse Rat Human

Width of CI on population
median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median

Width of CI on
population median
Fold-shift
in
population
median
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
PBPK
parameter Prior Posterior
QC 2.1 1.59 1.2 QC 1.64 1.43 1.15 VRespLumC 1.65 1.31 1
PB 2.3 1.54 1.07 VRespEffC 1.56 1.43 1 TCAPlas 26.9 1.29 1.21
QLivC 1.55 1.42 1.02 VRespLumC 1.56 1.41 1 VKidC 1.54 1.28 1.01
QRapC 1.51 1.41 1.03 VLivC 1.57 1.4 1.05 PB 2.04 1.28 1.04
VGutC 1.38 1.3 1.01 PLiv 1.67 1.37 1.05 QRapC 2.22 1.25 1.34
VBldC 1.34 1.27 1.02 QLivC 1.53 1.34 1.04 QGutC 1.59 1.23 1.19
VRespLumC 1.32 1.26 1.01 VKidC 1.47 1.33 1.01 VSlwC 1.66 1.21 1.07
VRespEffC 1.31 1.26 1 QKidC 1.39 1.28 1 VPlasC 1.39 1.2 1.01
QGutC 1.52 1.24 1.15 VGutC 1.38 1.28 1.01 QKidC 1.36 1.17 1
VKidC 1.29 1.24 1 VBldC 1.34 1.25 1.01 VBldC 1.34 1.17 1.02
VRapC 1.3 1.23 1.01 VRapC 1.34 1.23 1 FracLungSys 19.4 1.14 1.29
VSlwC 1.19 1.11 1.01 QGutC 1.53 1.22 1.14 VRapC 1.22 1.12 1
VBodC 1.05 1.03 1.01 TCAPlas 1.6 1.21 1.01 kDissoc 1.23 1.12 1.01
VBodTCOHC 1.04 1.03 1.01 VSlwC 1.15 1.09 1 VGutC 1.22 1.11 1.01
VBodC 1.04 1.03 1 VBodC 1.04 1.02 1
VBodTCOHC 1.02 1.01 1 VBodTCOHC 1.02 1.01 1

a
Shifts in the median estimate greater than threefold are in bold to denote larger shifts between the prior and posterior distributions
3-101
However, for some parameters, the posterior distributions in the population medians had
CIs >100-fold. In mice, the absorption parameter for TCA still had posterior CI of 400-fold,
reflecting the fact that the absorption rate is poorly estimated from the few available studies with
TCA dosing. In addition, mouse metabolism parameters for GSH conjugation have posterior CIs
>10,000-fold, reflecting the lack of any direct data on GSH conjugation in mice. In rats, two
parameters related to TCOH and TCOG had CIs between 100- and 1,000-fold, reflecting the
poor identifiability of these parameters given the available data. In humans, only the oral
absorption parameters for TCA and TCOH had CIs >100-fold, reflecting the fact that the
absorption rate is poorly estimated from the few available studies with TCOH and TCA dosing.
In terms of general consistency between prior and posterior distributions, in most cases,
the central estimate of the population median shifted by less than threefold. In almost all of the
cases that the shift was greater (see bold entries in Table 3-40), the prior distribution had a wide
distribution, with CI greater (sometimes substantially greater) than 100-fold. The only exception
was the fraction of TCE oxidation directly producing TCA, which shifted by fourfold in rats and
fivefold in mice, with prior CIs of 28- and 16-fold, respectively. These shifts are still relatively
modest in comparison to the prior CI, and moreover, the posterior CI is quite narrow (fourfold in
rats, twofold in humans), suggesting that the parameter is well identified by the in vivo data.
In addition, there were only a few cases in which the interquartile regions of the prior and
posterior distributions did not overlap. In most of these cases, including the diffusion rate from
respiratory lumen to tissue, the K
M
values for renal TCE GSH conjugation and respiratory TCE
oxidation, and several metabolite kinetic parameters, the prior distributions themselves were
noninformative. For a noninformative prior, the lack of overlap would only be an issue if the
posterior distributions were affected by the truncation limit, which was not the case. The only
other parameter for which there was a lack of interquartile overlap between the prior and
posterior distribution was the K
M
for hepatic TCE oxidation in mice and in rats, though the prior
and posterior 95% CIs did overlap within each species. As discussed Section 3.3, there is some
uncertainty in the extrapolation of in vitro K
M
values to in vivo values (within the same species).
In addition, in mice, it has been known for some time that K
M
values appear to be discordant
among different studies (Greenberg et al., 1999; Abbas and Fisher, 1997; Fisher et al., 1991).
In terms of estimates of population variability, for the vast majority of parameters, the
posterior estimate of the population GSD was either twofold or less, indicating modest
variability. In some cases, while the posterior population GSD was greater than twofold, it was
similar to the prior estimate of the population GSD, indicating limited additional informative
data on variability. This was the case for oral absorption parameters, which are expected to be
highly variable because the current model lumps parameters for different oral dosing vehicles
together, and a relatively wide prior distribution was given. In addition, in some cases, this was
due to in vitro data showing a higher degree of variability. Examples of this include TCA
plasma binding parameters in the mouse, and the V
MAX
for hepatic oxidation and the fraction of
3-102
oxidation to TCA in humans. In a few other cases, the in vivo data appeared to indicate greater
than twofold variability between subjects, and these are discussed in more detail below.
In the mouse, the two parameters for which this is the case are the V
MAX
for respiratory
tract oxidation and the urinary excretion rate for TCOG. In the first case, the variability is driven
by the need for a higher respiratory tract V
MAX
for males in the Fisher et al. (1991) study as
compared to other studies. In the second case, it is driven by the relatively low estimate of
urinary excretion of TCOG in the Abbas and Fisher (1997), Abbas et al. (1997), and Greenberg
et al. (1999) studies as compared with the relatively high estimate in Green and Prout (1985) and
Prout et al. (1985).
In the rat, the two parameters for which the in vivo data suggest greater than twofold
variability are the fraction of oxidation not producing TCA or TCOH, and the V
MAX
for
respiratory tract oxidation. In the first case, this is driven by three studies that appeared to
require greater (Bernauer et al., 1996; Kimmerle and Eben, 1973b) or lower (Hissink et al., 2002)
estimates for this parameter as compared with the other studies. Nonetheless, the degree of
variability is not much greater than twofold, with a central estimate population GSD of 2.15-fold.
In the case of the V
MAX
for respiratory tract oxidation, two studies appeared to require higher
(Fisher et al., 1989) or lower (Simmons et al., 2002) values for this parameter as compared with
the other studies.
In humans, as would be expected, more parameters appeared to exhibit greater than
twofold variability. In terms of distribution, the partition coefficients for TCOG had rather large
posterior estimates for the population GSD of eightfold for the body and fivefold for the liver. In
terms of the body, a few of the subjects in Fisher et al. (1998) and all of the subjects in Monster
et al. (1976) appeared to require much higher partition coefficients for TCOG. For the liver, the
variability did not have a discernable trend across studies. In addition, almost all of the
metabolism and clearance parameters had posterior estimates for population variability of greater
than a twofold GSD. The largest of these was the urinary excretion rate for TCOG, with a GSD
of 19-fold. In this case, the variability was driven by individuals in the Chiu et al. (2007) 1 ppm
study, who were predicted to have much lower rate of urinary excretion as compared to that
estimated in the other, higher exposure studies.
In sum, the Bayesian analysis of the updated PBPK model and data exhibited no major
inconsistencies in prior and posterior parameter distributions. The most significant issue in terms
of population central estimates was the K
M
for hepatic oxidative metabolism, for which the
posterior estimates were low compared to, albeit somewhat uncertain, in vitro estimates, and it
could be argued that a wider prior distribution would have been better. However, the central
estimates were not at or near the truncation boundary, so it is unlikely that wider priors would
change the results substantially. In terms of population variability, in rodents, the estimates of
variability were generally modest, which is consistent with more homogeneous and controlled
experimental subjects and conditions, whereas the estimates of human population variability
3-103
were greaterparticularly for metabolism and clearance. Overall, there were no indications
based on this evaluation of prior and posterior distributions either that prior distributions were
overly restrictive or that model specification errors led to pathological parameter estimates.

3.5.6.3. Comparison of Model Predictions With Data
Comparisons of model predictions and data for each species are discussed in the sub-
sections below. First, as an overall summary, for each species and each output measurement, the
data and predictions generated from a random sample of the MCMC chain are scatter-plotted to
show the general degree of consistency between data and predictions. Next, as with the Hack
et al. (2006) model, the sampled subject-specific parameters were used to generate predictions
for comparison to the calibration data (see Figure 3-8). Thus, the predictions for a particular data
set are conditioned on the posterior parameter distributions for same data set. Because these
parameters were optimized for each experiment, these subject-specific predictions should be
accurate by designand, on the whole, were so. In addition, the residual-error estimate for
each measurement (see Table 3-41) provides some quantitative measure of the degree to which
there were deviations due to intrastudy variability and model misspecification, including any
difficulties fitting multiple dose levels in the same study using the same model parameters.

3-104


Posterior Posterior
Posterior subject-
specific

i
Posterior subject-
specific

i
Posterior
2
Posterior
2
MCMC outputs
Posterior population

Posterior population

Group/
Individual i
Experiment j

i
E
ij
t
ijkl
Group/
Individual i
Experiment j

i
E
ij
t
ijkl
PBPK
model
Posterior population
prediction
y
jkl
Posterior group-specific
prediction
y
ijkl

Two sets of posterior predictions were generated: population predictions
(diagonal hashing) and subject-specific predictions (vertical hashing). (Same as
Figure A-2 in Appendix A)
Figure 3-8. Schematic of how posterior predictions were generated for
comparison with experimental data.

Table 3-41. Estimates of the residual-error

Measurement
abbreviation Measurement description
GSD for "residual" error
(median estimate)
Mouse Rat Human
RetDose Retained TCE dose (mg) - - 1.13
CAlvPPM TCE concentration in alveolar air (ppm) - - 1.44~1.83
CInhPPM TCE concentration in closed-chamber (ppm) 1.18 1.11~1.12 -
CMixExh TCE concentration in mixed exhaled air (mg/L) - 1.5 -
CArt TCE concentration in arterial blood (mg/L) - 1.17~1.52 -
CVen TCE concentration in venous blood (mg/L) 2.68 1.22~4.46 1.62~2.95
CBldMix TCE concentration in mixed arterial and venous blood
(mg/L)
1.61 1.5 -
CFat TCE concentration in fat (mg/L) 2.49 1.85~2.66 -
CGut TCE concentration in gut (mg/L) - 1.86 -
CKid TCE concentration in kidney (mg/L) 2.23 1.47 -
CLiv TCE concentration in liver (mg/L) 1.71 1.67~1.78 -
CMus TCE concentration in muscle (mg/L) - 1.65 -
AExhpost Amount of TCE exhaled postexposure (mg) 1.23 1.12~1.17 -
3-105
Table 3-41. Estimates of the residual-error (continued)

Measurement
abbreviation Measurement description
GSD for "residual" error
(median estimate)
a

Mouse Rat Human
CPlasTCA TCA concentration in plasma (mg/L) 1.40 1.13~1.21 1.12~1.17
CBldTCA TCA concentration in blood (mg/L) 1.49 1.13~1.59 1.12~1.49
CLivTCA TCA concentration in liver (mg/L) 1.34 1.67 -
AUrnTCA Cumulative amount of TCA excreted in urine (mg) 1.34 1.18~1.95 1.11~1.54
AUrnTCA_collect Cumulative amount of TCA collected in urine
(noncontinuous sampling) (mg)
- - 2~2.79
CTCOH Free TCOH concentration in blood (mg/L) 1.54 1.14~1.64 1.14~2.1
CLivTCOH Free TCOH concentration in liver (mg/L) 1.59 - -
TotCTCOH Total TCOH concentration in blood (mg/L) 1.85 1.49 1.2~1.69
ABileTCOG Cumulative amount of bound TCOH excreted in bile
(mg)
- 2.13 -
CTCOG Bound TCOH concentration in blood - 2.76 -
CTCOGTCOH Bound TCOH concentration in blood in free TCOH
equivalents
1.49 - -
CLivTCOGTCOH Bound TCOH concentration in liver in free TCOH
equivalents (mg/L)
1.63 - -
AUrnTCOGTCOH Cumulative amount of total TCOH excreted in urine
(mg)
1.26 1.12~2.27 1.11~1.13
AUrnTCOGTCOH_
collect
Cumulative amount of total TCOH collected in urine
(noncontinuous sampling) (mg)
- - 1.3~1.63
AUrnTCTotMole Cumulative amount of TCA+total TCOH excreted in
urine (mmol)
- 1.12~1.54 -
CDCVGmol DCVG concentration in blood (mmol/L) - - 1.53
AUrnNDCVC Cumulative amount of NAcDCVC excreted in urine
(mg)
- 1.17 1.17

a
Values higher than twofold are in bold.

Next, only samples of the population parameters (means and variances) were used, and
new subjects were sampled from appropriate distribution using these population means and
variances (see Figure 3-8). That is, the predictions were only conditioned on the population-
level parameters distributions, representing an average over all of the data sets, and not on the
specific predictions for that data set. These new subjects then represent the predicted
population distribution, incorporating variability in the population as well as uncertainty in the
population means and variances. Because of the limited amount of mouse data, all available data
for that species were utilized for calibration, and there were no data available for out-of-
sample evaluation (often referred to as validation data, but this term is not used here due to
ambiguities as to its definition). In rats, several studies that contained primarily blood TCE data,
which were abundant, were used for out-of-sample evaluation. In humans, there were substantial
individual and aggregated (mean of individuals in a study) data that were available for out-of-
sample evaluation, as computational intensity limited the number of individuals who could be
used in the MCMC-based calibration.
3-106

3.5.6.3.1. Mouse model and data
Each panel of Figure 3-9 shows a scatter plot of the calibration data and a random
posterior prediction for each of the measured endpoint. The endpoint abbreviations are listed in
Table 3-41, as are the implied GSDs for the residual errors, which include intrastudy
variability, interindividual variability, and measurement and model errors. The residual-error
GSDs are also shown as grey dotted lines in Figure 3-9. Table 3-42 provides an evaluation of
the predictions of the mouse model for each data set, with figures showing individual time-
course data and predictions in Appendix A.



Each panel shows results for a different measurement. The solid line represents
prediction = data, and the grey dotted lines show prediction = data GSD
err
and
data GSD
err
, where GSD
err
is the median estimate of the residual-error GSD
shown in Table 3-41.

Figure 3-9. Comparison of mouse data and PBPK model predictions from a
random posterior sample.

3-107



Each panel shows results for a different measurement. The solid line represents
prediction = data, and the grey dotted lines show prediction = data GSD
err
and
data GSD
err
, where GSD
err
is the median estimate of the residual-error GSD
shown in Table 3-41.
Figure 3-9 (continued). Comparison of mouse data and PBPK model
predictions from a random posterior sample.

3-108
Table 3-42. Summary comparison of updated PBPK model predictions and
in vivo data in mice

Study Exposure(s) Discussion
Abbas and Fisher
(1997)
TCE gavage
(corn oil)
Generally, model predictions were quite good, especially with respect to
tissue concentrations of TCE, TCA, and TCOH. There were some
discrepancies in TCA and TCOG urinary excretion and TCA and TCOG
concentrations in blood due to the requirement (unlike in Hack et al., 2006)
that all experiments in the same study utilize the same parameters. Thus, for
instance, TCOG urinary excretion was accurately predicted at 300 mg/kg,
underpredicted at 600 mg/kg, overpredicted at 1,200 mg/kg, and
underpredicted again at 2,000 mg/kg, suggesting significant
intraexperimental variability (not addressed in the model). Population
predictions were quite good, with the almost all of the data within the 95%
CI of the predictions, and most within the interquartile region.
Abbas et al. (1997) TCOH, TCA i.v. Both subject-specific and population predictions were quite good. Urinary
excretion, which was overpredicted by the Hack et al. (2006) model, was
accurately predicted due to the allowance of additional untracked
clearance. In the case of population predictions, almost all of the data were
within the 95% CI of the predictions, and most within the interquartile
region.
Fisher and Allen
(1993)
TCE gavage
(corn oil)
Both subject-specific and population predictions were quite good. Some
discrepancies in the time-course of TCE blood concentrations were evidence
across doses in the subject-specific predictions, but not in the population
predictions, suggesting significant intrasubject variability (not addressed in
the model).
Fisher et al. (1991) TCE inhalation Blood TCE levels during and following inhalation exposures were still
overpredicted at the higher doses. However, there was the stringent
requirement (absent in Hack et al., 2006) that the model utilize the same
parameters for all doses and in both the closed and open-chamber
experiments. Moreover, the Hack et al. (2006) model required significant
differences in the parameters for the different closed-chamber experiments,
while predictions here were accurate utilizing the same parameters across
different initial concentrations. These conclusions were the same for
subject-specific and population predictions (e.g., TCE blood levels remained
overpredicted in the later case).
Green and Prout
(1985)
TCE gavage
(corn oil)
Both subject-specific and population predictions were adequate, though the
data collection was sparse. In the case of population predictions, almost all
of the data were within the 95% CI of the predictions, and about half within
the interquartile region.
Greenberg et al.
(1999)
TCE inhalation Model predictions were quite good across a wide variety of measures that
included tissue concentrations of TCE, TCA, and TCOH. However, as with
the Hack et al. (2006) predictions, TCE blood levels were overpredicted by
up to twofold. Population predictions were quite good, with the exception of
TCE blood levels. Almost all of the other data was within the 95% CI of the
predictions, and most within the interquartile region.
Larson and Bull
(1992a)
TCE gavage
(aqueous)
Both subject-specific and population predictions were quite good, though the
data collection was somewhat sparse. In the case of population predictions,
all of the data were within the 95% CI of the predictions.
Larson and Bull
(1992b)
TCA gavage
(aqueous)
Both subject-specific and population predictions were quite good. In the
case of population predictions, most of the data were within the interquartile
region.
Merdink et al.
(1998)
TCE i.v. Both subject-specific and population predictions were quite good, though the
data collection was somewhat sparse. In the case of population predictions,
all of the data were within the 95% CI of the predictions.
3-109
Table 3-42. Summary comparison of updated PBPK model predictions and
in vivo data in mice (continued)

Study Exposure(s) Discussion
Prout et al. (1985) TCE gavage
(corn oil)
Both subject-specific and population predictions were adequate, though
there was substantial scatter in the data due to the use of single animals at
each data point.
Templin et al.
(1993)
TCE gavage
(aqueous)
Both subject-specific and population predictions were quite good. With
respect to population predictions, almost all of the other data was within the
95% CI of the predictions, and most within the interquartile region.

In terms of total metabolism, closed-chamber data (see Figure 3-9, panel A) were fit
accurately with the updated model, with a small residual-error GSD of 1.18. While the previous
analyses of Hack et al. (2006) allowed for each chamber experiment to be fit with different
parameters, the current analysis made the more restrictive assumption that all experiments in a
single study utilize the same parameters. Furthermore, the accuracy of closed-chamber
predictions did not require the very high values for cardiac output that were used by Fisher et al.
(1991), confirming the suggestion (discussed in Appendix A) that additional respiratory
metabolism would resolve this discrepancy. The accurate model means that uncertainty with
respect to possible wash-in/wash-out, respiratory metabolism, and extrahepatic metabolism could
be well characterized. For instance, the absence of in vivo data on GSH metabolism in mice
means that this pathway remains relatively uncertain; however, the current model should be
reliable for estimating lower and upper bounds on the GSH pathway flux.
In terms of the parent compound TCE (see Figure 3-9, panels B-G), the parent PBPK
model (for TCE) appears to now be robust, with the exception of the remaining overprediction of
TCE in blood following inhalation exposure. As expected, the venous-blood TCE concentration
had the largest residual-error, with a GSD of 2.7, reflecting largely the difficulty in fitting TCE
blood levels following inhalation exposure. In addition, the fat and kidney TCE concentrations
also are somewhat uncertain, with a GSD for the residual-error of 2.5 and 2.2, respectively.
These tissues were only measured in two studies, Abbas and Fisher (1997) and Greenberg et al.
(1999), and the residual-error reflects the difficulties in simultaneously fitting the model to the
different dose levels with the same parameters. Residual-error GSDs for other TCE
measurements were less than twofold. Thus, most of the problems previously encountered with
the Abbas and Fisher (1997) gavage data were solved by allowing absorption from both the
stomach and duodenal compartments. Notably, the addition of possible wash-in/wash-out,
respiratory metabolism, and extrahepatic metabolism (i.e., kidney GSH conjugation) was
insufficient to remove the long-standing discrepancy of PBPK models overpredicting TCE blood
levels from mouse inhalation exposures, suggesting another source of model or experimental
error is the cause. However, the availability of tissue concentration levels of TCE somewhat
ameliorates this limitation.
3-110
In terms of TCA and TCOH, the overall mass balance and metabolic disposition to these
metabolites also appeared to be robust, as urinary excretion following dosing with TCE, TCOH,
and TCA could be modeled accurately (see Figure 3-9, panels K and Q). The residual GSDs for
the urinary excretions are small: 1.34 for TCA and 1.26 for total TCOH. In addition, the blood
and tissue concentrations were also accurately predicted (see Figure 3-9, panels H-J, L-P). All of
the residual GSDs were less than twofold, with those for TCA measurements <1.5-fold. This
improvement over the Hack et al. (2006) model was likely due in part to the addition of
nonurinary clearance (untracked metabolism) of TCA and TCOH. Also, the addition of a liver
compartment for TCOH and TCOG, so that first-pass metabolism could be properly accounted
for, was essential for accurate simulation of the metabolite pharmacokinetics both from
intravenous (i.v.) dosing of TCOH and from exposure to TCE.

3.5.6.3.2. Rat model and data
Each panel of Figure 3-10 shows a scatter plot of the calibration data and a random
posterior prediction for each of the measured endpoint. The endpoint abbreviations are listed in
Table 3-41, as are the implied GSDs for the residual errors, which include intrastudy
variability, interindividual variability, and measurement and model errors. The residual-error
GSDs are also shown as grey dashed or dotted lines in Figure 3-10. A summary evaluation of
the predictions of the rat model as compared to the data are provided in Tables 3-43 and 3-44,
with figures showing individual time-course data and predictions in Appendix A.

3-111



Each panel shows results for a different measurement. The solid line represents
prediction = data, and the grey lines show prediction = data GSD
err
and data
GSD
err
, where GSD
err
is the lowest (dotted) and highest (dashed) median estimate
of the residual-error GSD shown in Table 3-41.
Figure 3-10. Comparison of rat data and PBPK model predictions from a
random posterior sample.

3-112



Each panel shows results for a different measurement. The solid line represents
prediction=data, and the grey lines show prediction = data GSD
err
and data
GSD
err
, where GSD
err
is the lowest (dotted) and highest (dashed) median estimate
of the residual-error GSD shown in Table 3-41.
Figure 3-10 (continued). Comparison of rat data and PBPK model
predictions from a random posterior sample.

3-113



Each panel shows results for a different measurement. The solid line represents
prediction=data, and the grey lines show prediction = data GSD
err
and data
GSD
err
, where GSD
err
is the lowest (dotted) and highest (dashed) median estimate
of the residual-error GSD shown in Table 3-41.
Figure 3-10 (continued). Comparison of rat data and PBPK model
predictions from a random posterior sample.

3-114
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in rats

Study Exposure(s) Discussion
Bernauer et al.
(1996)
TCE inhalation Posterior fits to these data were adequate, especially with the requirement that
all predictions for dose levels utilize the same PBPK model parameters.
Predictions of TCOG and TCA urinary excretion was relatively accurate,
though the time-course of TCA excretion seemed to proceed more slowly
with increasing dose, an aspect not captured in the model. Importantly, unlike
the Hack et al. (2006) results, the time-course of NAcDCVC excretion was
quite well simulated, with the excretion rate remaining nonnegligible at the
last time point (48 hrs). It is likely that the addition of the DCVG submodel
between TCE and DCVC, along with prior distributions that accurately
reflected the lack of reliable, independent (e.g., in vitro) data on bioactivation,
allowed for the better fit.
Dallas et al.
(1991)
TCE inhalation These data, consisting of arterial blood and exhaled breath concentrations of
TCE, were accurately predicted by the model using both subject-specific and
population-sampled parameters. In the case of population predictions, most
of the data were within the 95% CI of the predictions.
Fisher et al.
(1989)
TCE inhalation These data, consisting of closed-chamber TCE concentrations, were
accurately simulated by the model using both subject-specific and population-
sampled parameters. In the case of population predictions, most of the data
were within the 95% CI of the predictions.
Fisher et al.
(1991)
TCE inhalation These data, consisting of TCE blood, and TCA blood and urine time-courses,
were accurately simulated by the model using both subject-specific and
population-sampled parameters. In the case of population predictions, most
of the data were within the 95% CI of the predictions.

Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in rats (continued)

Study Exposure(s) Discussion
Green and Prout
(1985)
TCE gavage (corn
oil)
TCA i.v.
TCA gavage
(aqueous)
For TCE treatment, these data, consisting of one time point each in urine for
TCA, TCA +TCOG, and TCOG, were accurately simulated by both subject-
specific and population predictions.
For TCA i.v. treatment, the single datum of urinary TCA+TCOG at 24 hrs
was at the lower 95% CI in the subject-specific simulations, but accurately
predicted with the population-sampled parameters, suggesting intrastudy
variability is adequately accounted for by population variability.
For TCA gavage treatment, the single datum of urinary TCA+TCOG at
24 hrs was accurately simulated by both subject-specific and population
predictions.
Hissink et al.
(2002)
TCE gavage (corn
oil)
TCE i.v.
These data, consisting of TCE blood, and TCA+TCOG urinary excretion
time-courses, were accurately simulated by the model using subject-specific
parameters. In the case of population predictions, TCA+TCOH urinary
excretion appeared to be somewhat underpredicted.
Kaneko et al.
(1994)
TCE inhalation These data, consisting of TCE blood and TCA and TCOG urinary excretion
time-courses, were accurately predicted by the model using both subject-
specific and population-sampled parameters. In the case of population
predictions, TCA+TCOH urinary excretion appeared to be somewhat
underpredicted, However, all of the data were within the 95% CI of the
predictions.
3-115
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in rats (continued)

Study Exposure(s) Discussion
Keys et al.
(2003)
TCE inhalation,
gavage (aqueous),
i.a.
These data, consisting of TCE blood, gut, kidney, liver, muscle, and fat
concentration time-courses, were accurately predicted by the model using
both subject-specific and population-sampled parameters. In the case of
population predictions, most of the data were within the 95% CI of the
predictions.
Kimmerle and
Eben (1973b)
TCE inhalation Some inaccuracies were noted in subject-specific predictions, particularly
with TCA and TCOG urinary excretion, TCE exhalation postexposure, and
TCE venous blood concentrations. In the case of TCA excretion, the rate was
underpredicted at the lowest dose (49 mg/kg) and overpredicted at 330 ppm.
In terms of TCOG urinary excretion, the rate was overpredicted at 175 ppm
and underpredicted at 330 ppm. Similarly for TCE exhaled postexposure,
there was some overprediction at 175 ppm and some underprediction at
300 ppm. Finally, venous blood concentrations were overpredicted at
3,000 ppm. However, for population predictions, most of the data were
within the 95% confidence region.
Larson and Bull
(1992b)
TCA gavage
(aqueous)
These data, consisting of TCA plasma time-courses, were accurately
predicted by the model using both subject-specific and population-sampled
parameters. In the case of population predictions, all of the data were within
the 95% CI of the predictions.
Larson and Bull
(1992a)
TCE gavage
(aqueous)
These data, consisting of TCE, TCA, and TCOH in blood, were accurately
predicted by the model using both subject-specific and population-sampled
parameters. In the case of population predictions, all of the data were within
the 95% CI of the predictions.
Lee et al.
(2000a; Lee et
al., 2000b)
TCE i.v., p.v. These data, consisting of TCE concentration time course in mixed arterial
and venous blood and liver, were predicted using both the subject specific
and population predictions. In both cases, most of the data were within the
95% CI of the predictions.
Merdink et al.
(1999)
TCOH i.v. TCOH blood concentrations were accurately predicted using subject-specific
parameters. However, population-based parameters seemed to lead to some
underprediction, though most of the data were within the 95% CI of the
predictions.
Prout et al.
(1985)
TCE gavage (corn
oil)
Most of these data were accurately predicted using both subject-specific and
population-sampled parameters. However, at the highest two doses
(1,000 and 2,000 mg/kg), there were some discrepancies in the (very sparsely
collected) urinary excretion measurements. In particular, using subject-
specific parameters, TCA+TCOH urinary excretion was underpredicted at
1,000 mg/kg and overpredicted at 2,000 mg/kg. Using population-sampled
parameters, this excretion was underpredicted in both cases, though not
entirely outside of the 95% CI.
Simmons et al.
(2002)
TCE inhalation Most of these data were accurately predicted using both subject-specific and
population-sampled parameters. In the open-chamber experiments, there was
some scatter in the data that did not seem to be accounted for in the model.
The closed-chamber data were accurately fit.
3-116
Table 3-43. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in rats (continued)

Study Exposure(s) Discussion
Stenner et al.
(1997)
TCE
intraduodenal
TCOH i.v.
TCOH i.v., bile-
cannulated
These data, consisting of TCA and TCOH in blood and TCA and TCOG in
urine, were generally accurately predicted by the model using both subject-
specific and population-sampled parameters. However, using subject-
specific parameters, the amount of TCOG in urine was overpredicted for
100 TCOH mg/kg i.v. dosing, though total TCOH in blood was accurately
simulated. In addition, in bile-cannulated rats, the TCOG excretions at 5 and
20 mg/kg i.v. were underpredicted, while the amount at 100 mg/kg was
accurately predicted. On the other hand, in the case of population
predictions, all of the data were within the 95% CI of the predictions, and
mostly within the interquartile region, even for TCOG urinary excretion.
This suggests that intrastudy variability may be a source of the poor fit in
using the subject-specific parameters.
Templin et al.
(1995b)
TCE oral
(aqueous)
These data, consisting of TCE, TCA, and TCOH in blood, were accurately
predicted by the model using both subject-specific and population-sampled
parameters. In the case of population predictions, all of the data were within
the 95% CI of the predictions.
Yu et al. (2000) TCA i.v. These data, consisting of TCA in blood, liver, plasma, and urine, were
generally accurately predicted by the model using both subject-specific and
population-sampled parameters. The only notable discrepancy was at the
highest dose of 50 mg/kg, in which the rate of urinary excretion from 0 to
6 hrs appeared to more rapid than the model predicted. However, all of the
data were within the 95% CI of the predictions based on population-sampled
parameters.

Table 3-44. Summary comparison of updated PBPK model predictions and
in vivo data used for o ut-of-sample evaluation in rats

Study Exposure(s) Discussion
Andersen et al.
(1987a)
TCE inhalation These closed-chamber data were well within the 95% CI of the predictions
based on population-sampled parameters.
Bruckner et al.
unpublished
TCE inhalation These data on TCE in blood, liver, kidney, fat, muscle, gut, and venous
blood were generally accurately predicted based on population-sampled
parameters. The only notable exception was TCE in the kidney during the
exposure period at the 500 ppm level, which was somewhat underpredicted
(though levels postexposure were accurately predicted).
Fisher et al. (1991) TCE inhalation These data on TCE in blood were well within the 95% CI of the
predictions based on population-sampled parameters.
Jakobson et al.
(1986)
TCE inhalation These data on TCE in arterial blood were well within the 95% CI of the
predictions based on population-sampled parameters.
Lee et al. (1996) TCE i.a., i.v., p.v.,
gavage
Except at some very early time-points (<0.5 hr), these data on TCE in
blood were well within the 95% CI of the predictions based on population-
sampled parameters.
Lee et al. (2000a;
2000b)
TCE gavage These data on TCE in blood were well within the 95% CI of the
predictions based on population-sampled parameters.

Similar to previous analyses (Hack et al., 2006), the TCE submodel for the rat appears to
be robust, accurately predicting blood and tissue concentrations (see Figure 3-10, panels A-K),
with residual-error GSDs generally less than twofold. The only exceptions are the predictions of
3-117
venous blood from Kimmerle and Eben (1973b), which have residual-error GSDs greater than
fourfold, and the predictions of fat concentrations from Simmons et al. (2002); with residual-
error GSD of 2.7-fold. For Kimmerle and Eben (1973b), the inaccuracy was primarily at the
3,000-ppm exposure, which might reflect other factors related to the high exposure. For
Simmons et al. (2002), the high residual-error appears to reflect scatter due to intrastudy
variability. Unlike in the mouse, some data consisting of TCE blood and tissue concentrations
were used for out-of-sample evaluation (sometimes loosely termed validation). These data
were generally well simulated (see Table 3-44); most of the data were within the 95% CI of
posterior predictions. This provides additional confidence in the predictions for the parent
compound.
In terms of TCA and TCOH, as with the mouse, the overall mass balance and metabolic
disposition to these metabolites also appeared to be robust: urinary excretion following dosing
with TCE, TCOH, and TCA could be modeled accurately (see Figure 3-10 panels O, T, and U),
with the residual-errors also indicating good predictions in most cases. Residual-error for these
measurements was larger for Green and Prout (1985), Prout et al. (1985), and Stenner et al.
(1997), ranging from a GSD of 1.8 to 2.3, reflecting largely intrastudy variability. Residual-
errors for the other studies had GSDs of 1.11.5. This improvement over the Hack et al. (2006)
model was likely due in part to the addition of nonurinary clearance (untracked metabolism) of
TCA and TCOH. In addition, adding a liver compartment for TCOH and TCOG, so that first-
pass metabolism could be properly accounted for, was essential for accurate simulation of the
metabolite pharmacokinetics both from i.v. dosing of TCOH and from TCE exposure. Blood
and plasma concentrations of TCA and free or total TCOH were also fairly well simulated (see
Figure 3-10, panels L, M, P, Q, and S), with GSDs for the residual-error of 1.11.6. A bit more
discrepancy (residual-error GSD of 1.7) was evident with TCA liver concentrations (see
Figure 3-10, panel N). However, TCA liver concentrations were only available in one study (Yu
et al., 2000), and the data show a change in the ratio of liver to blood concentrations at the last
time point, which may be the source of the added residual-error. Predictions of biliary excretion
of TCOG in bile-cannulated rats (see Figure 3-10, panel R), from Green and Prout (1985), and
TCOG in blood (see Figure 3-10, panel S), from Stenner et al. (1997), were less accurate, with
residual-error GSDs >2. However, the biliary excretion data consisted of a single measurement,
and the amount of free TCOH in the same experiment from Stenner et al. (1997) was accurately
predicted.
In terms of total metabolism, as with the mouse, closed-chamber data (see Figure 3-10,
panel A) were fit accurately with the updated model (residual-error GSD of about 1.1). In
addition, the data on NAcDCVC urinary excretion was well predicted (see Figure 3-10, panel V),
with residual-error GSD of 1.18. In particular, the fact that excretion was still ongoing at the end
of the experiment was accurately predicted (see Figure 3-11, panels A and B). Thus, there is
greater confidence in the estimate of the flux through the GSH pathway than there was from the
3-118
Hack et al. (2006) model. However, the overall flux is still estimated indirectly, and there
remains some ambiguity as to the relative contributions of respiratory wash-in/wash-out,
respiratory metabolism, extrahepatic metabolism, DCVC bioactivation vs. N-acetylation, and
oxidation in the liver producing something other than TCOH or TCA. Therefore, there remains a
large range of possible values for the flux through the GSH conjugation and other indirectly
estimated pathways that are nonetheless consistent with all of the available in vivo data. The use
of noninformative priors for the metabolism parameters for which there were no in vitro data
means that a fuller characterization of the uncertainty in these various metabolic pathways could
be achieved. Thus, the model should be reliable for estimating lower and upper bounds on
several of these pathways.

3-119





Data are from Bernauer et al. (1996) for (A and B) rats or (C and D) humans
exposed for 6 hour to 40 (), 80 (), or 160 (+) ppm in air (thick horizontal line
denotes the exposure period). Predictions from Hack et al. (2006) and the
corresponding data (A and C) are only for the 1,2 isomer, whereas those from the
updated model (B and D) are for both isomers combined. Parameter values used
for each prediction are a random sample from the subject-specific parameters
from the rat and human MCMC chains (the last iteration of the first chain was
used in each case). Note that in the Hack et al. (2006) model, each dose group
had different model parameters, whereas in the updated model, all dose groups are
required to have the same model parameters. See files linked to Appendix A for
comparisons with the full distribution of predictions.
Figure 3-11. Comparison of urinary excretion data for NAcDCVC and
predictions from the Hack et al. (2006) and the updated PBPK models.

3.5.6.3.3. Human model and data
Each panel of Figure 3-12 shows a scatter plot of the calibration data and a random
posterior prediction for each of the measured endpoint. The endpoint abbreviations are listed in
Table 3-41, as are the implied GSDs for the residual errors, which include intrastudy
3-120

variability, interindividual variability, and measurement and model errors. The residual-error
GSDs are also shown as grey dashed or dotted lines in Figure 3-12. Table 3-453-46 provide a
summary evaluation of the predictions of the model as compared to the human data, with figures
showing individual time-course data and predictions in Appendix A.


Each panel shows results for a different measurement. The solid line represents
prediction = data, and the grey lines show prediction = data GSD
err
and data
GSD
err
, where GSD
err
is the lowest (dotted) and highest (dashed) median estimate
of the residual-error GSD shown in Table 3-41.

Figure 3-12. Comparison of human data and PBPK model predictions from
a random posterior sample.

3-121



Each panel shows results for a different measurement. The solid line represents
prediction = data, and the grey lines show prediction = data GSD
err
and data
GSD
err
, where GSD
err
is the lowest (dotted) and highest (dashed) median estimate
of the residual-error GSD shown in Table 3-41.
Figure 3-12 (continued). Comparison of rat data and PBPK model
predictions from a random posterior sample.

Table 3-45. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in humans

Reference Exposure(s) Discussion
Bernauer et al.
(1996)
TCE inhalation These data, consisting of TCA, TCOG and NAcDCVC excreted in urine,
were accurately predicted by the model using both individual-specific and
population-sampled parameters. The posterior NAcDCVC predictions were
an important improvement over the predictions of Hack et al. (2006), which
predicted much more rapid excretion than observed. The fit improvement is
probably a result of the addition of the DCVG submodel between TCE and
DCVC, along with the broader priors on DCVC excretion and bioactivation.
Interestingly, in terms of population predictions, the NAcDCVC excretion
data from this study were on the low end, though still within the 95% CI.

3-122
Table 3-45. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in humans (continued)

Reference Exposure(s) Discussion
Chiu et al. (2007) TCE inhalation Overall, posterior predictions were quite accurate across most of the
individuals and exposure occasions. TCE alveolar breath concentrations
were well simulated for both individual-specific and population-generated
simulations, though there was substantial scatter (intraoccasion variability).
However, TCE blood concentrations were consistently overpredicted in most
of the experiments, both using individual-specific and population-generated
parameters. This was not unexpected, as Chiu et al. (2007) noted the TCE
blood measurements to be lower by about twofold relative to previously
published studies. As discussed in Chiu et al. (2007) wash-in/wash-out and
extrahepatic (including respiratory) metabolism were not expected to be able
to account for the difference, and indeed all of these processes were added to
the current model without substantially improving the discrepancy.
With respect to metabolite data, TCA and total TCOH in blood were
relatively accurately predicted. There was individual experimental variability
observed for both TCA and TCOH in blood at 6 hrs (end of exposure). The
population-generated simulations overpredicted TCA in blood, while they
were accurate in predicting blood TCOH. Predictions of free TCOH in blood
also showed overprediction for individual experiments, with variability at the
end of exposure timepoint. However, TCOH fits were improved for the
population-generated simulations. TCA and TCOG urinary excretion was
generally well simulated, with simulations slightly under- or overpredicting
the individual experimental data in some cases.
Fisher et al. (1998) TCE inhalation The majority of the predictions for these data were quite accurate.
Interestingly, in contrast to the predictions for Chiu et al. (2007), TCE blood
levels were somewhat underpredicted in a few cases, both from using
individual-specific and population-generated predictions. These two results
together suggest some unaccounted-for study-to-study variance, though
interindividual variability cannot be discounted as the data from Chiu et al.
(2007) were from individuals in the Netherlands and that from Fisher et al.
(1998) were from individuals in the United States. As reported by Fisher
et al. (1998), TCE in alveolar air was somewhat overpredicted in several
cases; however, the discrepancies seemed smaller than originally reported for
the Fisher et al. model.
Fisher et al.
(1998) (continued)
TCE inhalation
(continued)
With respect to metabolite data, TCOH and TCA in blood and TCOG and
TCA in urine were generally well predicted, though data for some individuals
appeared to exhibit inter- and/or intraoccasion variability. For example, in
one case in which the same individual (female) was exposed to both 50 and
100 ppm, the TCOH blood data was overpredicted at the higher one exposure.
In addition, in one individual, initial individual-specific simulations for TCA
in urine were underpredicted but shifted to overpredictions towards the end of
the simulations. The population-generated results overpredicted TCA in urine
for the same individual. Given the results from Chiu et al. (2007),
interoccasion variability is likely to be the cause, though some dose-related
effect cannot be ruled out.
Finally, DCVG data was well predicted in light of the high variability in the
data and availability of only grouped data or data from multiple individuals
who cannot be matched to the appropriate TCE and oxidative metabolite data
set. In all cases, the basic shape (plateau and then sharp decline) and order of
magnitude of the time-course were well predicted, Furthermore, the range of
the data was well-captured by the 95% CI of the population-generated
predictions.
Kimmerle and
Eben (1973a)
TCE inhalation These data were well fit by the model, using either individual-specific or
population-generated parameters.
3-123
Table 3-45. Summary comparison of updated PBPK model predictions and
in vivo data used for c alibration in humans (continued)

Reference Exposure(s) Discussion
Monster et al.
(1976)
TCE inhalation The data simulated in this case were exhaled alveolar TCE, TCE in venous
blood, TCA in blood, TCA in urine, and TCOG in urine. Both using
individual-specific and population-generated simulations, all fits are within
the 95% CI. The one exception was the retained dose for a male exposed to
65 ppm, which was outside the 95% CI for the population-generated results.
Muller et al.
(1974)
TCA,


TCOH oral
The data measured after oral TCA was timecourse TCA measured in plasma
and urine. Individual-specific predictions were accurate, but both data sets
were overpredicted in the population-generated simulations.
The data measured after oral TCOH were timecourse TCOH in blood, TCOG
in urine, TCA in plasma, and TCA in urine. Individual-specific predictions
were accurate, but the population-generated simulations overpredicted TCOH
in blood and TCOG in urine. The population-based TCA predictions were
accurate.
These results indicate that unusual parameter values were necessary in the
individual-specific simulations to give accurate predictions.
Paykoc et al.
(1945)
TCA i.v. These data were well fit by the model, using either individual-specific or
population-generated parameters.

Table 3-46. Summary comparison of updated PBPK model predictions and
in vivo data used for o ut-of-sample evaluation in humans

Reference Exposure(s) Discussion
Bartonicek (1962) TCE inhalation While these data were mostly within the 95% CI of the predictions,
they tended to be at the high end for all of the individuals in the
study.
Bloemen et al. (2001) TCE inhalation These data were all well within the 95% CI of the predictions.
Fernandez et al. (1977) TCE inhalation These data were all well within the 95% CI of the predictions.
Lapare et al. (1995) TCE inhalation These data were all well within the 95% CI of the predictions.
Monster et al. (1979a) TCE inhalation These data were all well within the 95% CI of the predictions.
Muller et al. (1975; 1974) TCE inhalation Except for TCE in alveolar air, which was overpredicted during
exposure, these data were all well within the 95% CI of the
predictions.
Sato et al. (1977) TCE inhalation These data were all well within the 95% CI of the predictions.
Stewart et al. (1970) TCE inhalation These data were all well within the 95% CI of the predictions.
Triebig et al. (1976) TCE inhalation Except for TCE in alveolar air, these data were all well within the
95% CI of the predictions.



3-124
With respect to the TCE submodel, retained dose, blood, and exhaled air measurements
(see Figure 3-12, panels A-C) appeared more robust than previously found from the Hack et al.
(2006) model. TCE blood concentrations from most studies were well predicted, with residual-
error GSD in most studies of less than twofold. However, those from Chiu et al. (2007) were
consistently overpredicted (i.e., data <0.1 mg/L in Figure 3-12, panel C), with residual-error
GSD of almost threefold, and a few of those from Fisher et al. (1989) were consistently
underpredicted. Alveolar breath concentrations and retained dose of TCE were well predicted
(residual-error GSD <1.5-fold) from all studies except Fisher et al. (1998), which had a residual-
error GSD of 1.8-fold. However, the discrepancy in alveolar breath appeared smaller than that
originally reported by Fisher et al. (1998) for their PBPK model. In addition, the majority of the
out-of-sample evaluation data consisted of TCE in blood or breath, and were generally well
predicted (see Table 3-46), lending confidence to the model predictions for the parent compound.
In terms of TCA and TCOH, as with the mouse and rat, the overall mass balance and
metabolic disposition to these metabolites also appeared to be robust, as urinary excretion
following TCE exposure could be modeled accurately (see Figure 3-12, panels F, G, J, and K).
In most cases, the residual-error GSD was less than twofold. However, TCA urinary data from
Chiu et al. (2007) (panel G in Figure 3-12) indicated greater interoccasion variability, reflected in
the residual-error GSD of 2.8. In this study, the same individual exposed to the same
concentration on different occasions sometimes had substantial differences in urinary excretion.
In addition, many TCA urine measurements in this study were saturated, and had to be omitted,
and the fact that the remaining data were sparse and possibly censored may have contributed to
the greater intrastudy variability. Blood and plasma concentrations of TCA and free TCOH (see
Figure 3-12, panels D, E, and H) were fairly well simulated, with GSD for the residual-error of
1.11.4, though total TCOH in blood (see Figure 3-12, panel I) had slightly greater residual-error
with GSD of about 1.6. This partially reflects the sharper peak concentrations of total TCOH
in the Chiu et al. (2007) data relative to the model predictions. In addition, TCA and TCOH
blood and urine data were available from several studies for out-of-sample evaluation and
were generally well predicted by the model (see Table 3-46), lending further confidence to the
model predictions for these metabolites.
In terms of total metabolism, no closed-chamber data exist in humans, but, as discussed
above, alveolar breath concentrations and retained dose (see Figure 3-12, panels A and B) were
generally well simulated, suggesting that total metabolism may be fairly robust. In addition, as
with the rat, the data on NAcDCVC urinary excretion was well predicted (see Figure 3-11,
Figure 3-12 panel M), with residual-error GSD of 1.12). In particular, the model accurately
predicted the fact that excretion was still ongoing at the end of the experiment (48 hours after the
end of exposure). Thus, there is greater confidence in the estimate of the flux through this part of
the GSH pathway than there was from the Hack et al. (2006) model, in which excretion was
completed within the first few hours after exposure (see Figure 3-11, panels C and D).
3-125
If only urinary NAcDCVC data were available, as is the case for the rat, the overall GSH
conjugation flux would still be estimated indirectly, and there would remain some ambiguity as
to the relative contributions of respiratory wash-in/wash-out, respiratory metabolism,
extrahepatic metabolism, DCVC bioactivation vs. N-acetylation, and oxidation in the liver
producing something other than TCOH or TCA. However, unlike in the rat, the blood DCVG
data, while highly variable, nonetheless provide substantial constraints (at least a strong lower
bound) on the flux of GSH conjugation, and is well fit by the model (see Figure 3-12, panel L,
and Figure 3-13). Importantly, the high residual-error GSD for blood DCVG reflects the fact
that only grouped or unmatched individual data were available, so in this case, the residual-error
includes interindividual variability, which is not included in the other residual-error estimates.
However, as discussed above in Section 3.3.3.2.1, there are uncertainties as to the accuracy of
analytical method used by Lash et al. (1999b) in the measurement of DCVG in blood. Because
these data are so determinative of the overall GSH conjugation flux, these analytical
uncertainties are important to consider in the overall evaluation of the PBPK model predictions
(see below, Section 3.5.7).

3-126


Data are mean concentrations for males () and females () reported in Lash et al.
(1999b) for humans exposed for 4 hours to 100 ppm TCE in air (thick horizontal
line denotes the exposure period). Data for oxidative metabolites from the same
individuals were reported in Fisher et al. (1998) but could not be matched with the
individual DCVG data (Lash 2007, personal communication). The vertical error
bars are SEs of the mean as reported in Lash et al. (1999b) (n = 8, so SD is
80.5-fold larger). Lines are PBPK model predictions for individual male (solid)
and female (dashed) subjects. Parameter values used for each prediction are a
random sample from the individual-specific parameters from the human MCMC
chains (the last iteration of the 1
st
chain was used). See files linked to Appendix
A for comparisons with the full distribution of predictions.

Figure 3-13. Comparison of DCVG concentrations in human blood and
predictions from the updated model.

For the other indirectly estimated pathways, there remain a large range of possible values
that are nonetheless consistent with all of the available in vivo data. The use of noninformative
priors for the metabolism parameters for which there were no in vitro data means that a fuller
characterization of the uncertainty in these various metabolic pathways could be achieved. Thus,
as with the rat, the model should be reliable for estimating lower and upper bounds on several of
these pathways.

3.5.6.4. Sensitivity Analysis With Respect to Calibration Data
To assess the informativeness of the calibration data to the parameters, local sensitivity
analysis is performed with respect to the calibration data points. For each scaling parameter, the
3-127
central difference is used to estimate the partial derivatives by centering on the sample mean of
its estimated population mean, and then increasing and decreasing by 5%. The relative change in
the model output f() is used to estimate a local sensitivity coefficient (SC) as follows:

SC = 10 {f(
+
) f(

)}/[ {f(
+
) + f(

)}]

Here, f() is one of the model predictions of the calibration data,

is the maximum
likelihood estimate (MLE) or baseline value of 5%. For log-transformed parameters, 0.05 was
added or subtracted from the baseline value, whereas for untransformed parameters, the baseline
value was multiplied by 1.05 or 0.95. The resulting values of SC are binned into five categories
according to their sensitivity coefficient: negligible (|SC| 0.01) very low (0.01 < |SC| 0.1),
low (0.1 < |SC| 0.5), medium (0.5 < |SC| 1.0), and high (|SC| > 1.0).
Note that local sensitivity analyses as typically performed in deterministic PBPK
modeling can only inform the primary effects of parameter uncertainties (i.e., the direct change
on the quantity of interest due to change in a parameter). They cannot address the propagation
of uncertainties, such as those that can arise due to parameter correlations in the parameter fitting
process. Those can only be addressed in a global sensitivity analysis, which is left for future
research.
The results of local sensitivity analyses are shown in Figures 3-143-16. For each
parameter, the number of data points (out of the entire calibration set) that have sensitivity
coefficients in the various categories are shown graphically. As summarized in Table 3-47, most
of the parameters have at least some calibration data to which they are at least moderately
sensitive (|SC| > 0.5). Across species, the cardiac output (lnQCC), ventilation-perfusion ratio
(lnVPRC), blood-air partition coefficient (lnPBC), V
MAX
for oxidation (lnV
MAX
C), and VLivC
are consistently among the most sensitive parameters, with >10% of the calibration data
exhibiting |SC| > 0.5 to these parameters. Note that the reason the liver volume is sensitive is
that it is used to scale the capacity or clearance rate for oxidation.

3-128

0 200 400 600 800 1000 1200
lnQCC
lnVPRC
QFatC
QGutC
QLivC
QSlwC
lnDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEf f C
VKidC
VBldC
lnPBC
lnPFatC
lnPGutC
lnPLivC
lnPRapC
lnPRespC
lnPKidC
lnPSlwC
lnPRBCPlasTCAC
lnPBodTCAC
lnPLivTCAC
lnkDissocC
lnBMaxkDC
lnPBodTCOHC
lnPLivTCOHC
lnPBodTCOGC
lnPLivTCOGC
lnkTSD
lnkAS
lnkAD
lnkASTCA
lnVMaxC
lnKMC
lnFracOtherC
lnFracTCAC
lnVMaxDCVGC
lnClDCVGC
lnVMaxKidDCVGC
lnClKidDCVGC
lnVMaxLungLivC
lnKMClara
lnFracLungSysC
lnVMaxTCOHC
lnKMTCOH
lnVMaxGlucC
lnKMGluc
lnkMetTCOHC
lnkUrnTCAC
lnkMetTCAC
lnkBileC
lnkEHRC
lnkUrnTCOGC
P
a
r
a
m
e
t
e
r
Number of mouse calibration data points
|SC|0.01 0.01<|SC|0.1 0.1<|SC|0.5 0.5<|SC|1 1<|SC|

Figure 3-14. Sensitivity analysis results: Number of mouse calibration data
points with SC in various categories for each scaling parameter.

3-129

0 200 400 600 800 1000 1200
lnQCC
lnVPRC
QFatC
QGutC
QLivC
QSlwC
lnDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEf f C
VKidC
VBldC
lnPBC
lnPFatC
lnPGutC
lnPLivC
lnPRapC
lnPRespC
lnPKidC
lnPSlwC
lnPRBCPlasTCAC
lnPBodTCAC
lnPLivTCAC
lnkDissocC
lnBMaxkDC
lnPBodTCOHC
lnPLivTCOHC
lnPBodTCOGC
lnPLivTCOGC
lnkTSD
lnkAS
lnkAD
lnkASTCA
lnVMaxC
lnKMC
lnFracOtherC
lnFracTCAC
lnVMaxDCVGC
lnClDCVGC
lnVMaxKidDCVGC
lnClKidDCVGC
lnVMaxLungLivC
lnKMClara
lnFracLungSysC
lnVMaxTCOHC
lnKMTCOH
lnVMaxGlucC
lnKMGluc
lnkMetTCOHC
lnkUrnTCAC
lnkMetTCAC
lnkBileC
lnkEHRC
lnkUrnTCOGC
lnkNATC
lnkKidBioactC
P
a
r
a
m
e
t
e
r
Number of rat calibration data points
|SC|0.01 0.01<|SC|0.1 0.1<|SC|0.5 0.5<|SC|1 1<|SC|

Figure 3-15. Sensitivity analysis results: Number of rat calibration data
points with SC in various categories for each scaling parameter.

3-130

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
lnQCC
lnVPRC
QFatC
QGutC
QLivC
QSlwC
lnDRespC
QKidC
FracPlasC
VFatC
VGutC
VLivC
VRapC
VRespLumC
VRespEf f C
VKidC
VBldC
lnPBC
lnPFatC
lnPGutC
lnPLivC
lnPRapC
lnPRespC
lnPKidC
lnPSlwC
lnPRBCPlasTC
lnPBodTCAC
lnPLivTCAC
lnkDissocC
lnBMaxkDC
lnPBodTCOHC
lnPLivTCOHC
lnPBodTCOGC
lnPLivTCOGC
lnPef f DCVG
lnkASTCA
lnkASTCOH
lnVMaxC
lnClC
lnFracOtherC
lnFracTCAC
lnClDCVGC
lnKMDCVGC
lnClKidDCVGC
lnKMKidDCVGC
lnVMaxLungLiv
lnKMClara
lnFracLungSys
lnClTCOHC
lnKMTCOH
lnClGlucC
lnKMGluc
lnkMetTCOHC
lnkUrnTCAC
lnkMetTCAC
lnkBileC
lnkEHRC
lnkUrnTCOGC
lnkDCVGC
lnkNATC
lnkKidBioactC
P
a
r
a
m
e
t
e
r
Number of human calibration data points
|SC|0.01 0.01<|SC|0.1 0.1<|SC|0.5 0.5<|SC|1 1<|SC|
Figure 3-16. Sensitivity analysis results: Number of human calibration data
points with SC in various categories for each scaling parameter.



3-131
Table 3-47. Summary of scaling parameters ordered by fraction of
calibration data of moderate or high sensitivity

Mouse Rat Human
Parameter
a

Fraction
with
|SC|>0.5 Parameter
a

Fraction
with
|SC|>0.5 Parameter
a

Fraction
with
|SC|>0.5
lnV
MAX
C 0.4405 VLivC 0.4213 lnQCC 0.4159
VLivC 0.428 lnQCC 0.4182 lnVPRC 0.3777
lnPBC 0.3233 lnVPRC 0.4158 lnClTCOHC 0.2871
lnQCC 0.2454 lnV
MAX
C 0.3984 QGutC 0.2137
lnkAD 0.1675 lnPBC 0.2893 lnClGlucC 0.186
lnPBodTCAC 0.1642 VFatC 0.1455 lnkUrnTCAC 0.1789
lnVPRC 0.1575 QFatC 0.1273 FracPlasC 0.1553
lnFracTCAC 0.1323 lnPBodTCAC 0.1162 lnPBodTCOHC 0.1486
lnV
MAX
GlucC 0.1147 lnPFatC 0.1154 lnV
MAX
C 0.1358
lnPFatC 0.093 lnV
MAX
GlucC 0.1083 lnPBC 0.1269
lnPLivTCAC 0.0896 QGutC 0.0885 VLivC 0.1225
lnkAS 0.0863 lnkUrnTCAC 0.0696 lnPBodTCAC 0.12
VFatC 0.0762 lnPSlwC 0.0664 lnBMaxkDC 0.0897
lnKMGluc 0.0762 lnFracTCAC 0.064 VBldC 0.0586
lnkMetTCAC 0.0762 lnKMGluc 0.0625 lnkDCVGC 0.0515
lnkUrnTCAC 0.0754 lnkBileC 0.0538 lnPLivTCOGC 0.0446
lnKMC 0.0653 lnPLivTCOGC 0.0514 lnClDCVGC 0.0435
lnkUrnTCOGC 0.0544 lnPLivC 0.0482 lnkBileC 0.0422
lnV
MAX
LungLivC 0.0511 lnkAD 0.0474 QFatC 0.0401
lnkTSD 0.0469 lnKMC 0.0427 lnPSlwC 0.0372
QGutC 0.0452 lnV
MAX
TCOHC 0.0427 QSlwC 0.0345
QFatC 0.0402 lnPKidC 0.0324 lnKMTCOH 0.0305
lnPLivC 0.0402 lnPGutC 0.03 lnPFatC 0.0292
lnPLivTCOHC 0.0377 lnFracOtherC 0.03 lnClC 0.0288
lnPKidC 0.0352 lnPLivTCAC 0.0292 lnkUrnTCOGC 0.0282
lnPLivTCOGC 0.0352 lnBMaxkDC 0.0285 lnPRBCPlasTCAC 0.0147
lnPRBCPlasTCAC 0.031 lnkMetTCAC 0.0213 lnPLivTCAC 0.0135
lnV
MAX
TCOHC 0.0235 lnV
MAX
LungLivC 0.0182 lnkMetTCAC 0.013
lnPBodTCOHC 0.0201 lnKMTCOH 0.0182 lnFracTCAC 0.0103
lnPSlwC 0.0134 lnkAS 0.0158 lnPBodTCOGC 0.0095
lnBMaxkDC 0.0134 lnPBodTCOHC 0.015 VRapC 0.0063
lnDRespC 0.0109 FracPlasC 0.0126 VKidC 0.0057
lnkBileC 0.0084 lnkTSD 0.0103 lnClKidDCVGC 0.0057
FracPlasC 0.0059 VKidC 0.0095 lnkNATC 0.0057
lnPBodTCOGC 0.005 lnV
MAX
KidDCVGC 0.0095 lnPRapC 0.005
VGutC 0.0025 lnkNATC 0.0095 lnPLivTCOHC 0.005
lnPGutC 0.0025 lnDRespC 0.0063 lnkMetTCOHC 0.005
lnKMTCOH 0.0017 QSlwC 0.0055 lnFracOtherC 0.0046
lnkMetTCOHC 0.0017 lnPLivTCOHC 0.0016 VFatC 0.0036
lnkEHRC 0.0017 lnkASTCA 0.0016 lnkEHRC 0.0036
QKidC 0.0008 lnkMetTCOHC 0.0016 lnDRespC 0.0011
VKidC 0.0008 VGutC 0.0008 lnKMDCVGC 0.0011
lnPRBCPlasTCAC 0.0008 lnkKidBioactC 0.0002
lnkUrnTCOGC 0.0008

a
Parameters not shown have no data with |SC| > 0.5.

3-132
For scaling parameters for which all of the calibration data are negligibly sensitive
(|SC| < 0.01), it is important that they either have informative prior data or are unimportant for
dose-metric predictions. For mice, these parameters are the volumes of the respiratory lumen
and tissue (VRespLumC, VRespEffC), the partition coefficient for the respiratory tissue
(lnPRespC), and the V
MAX
values for GSH conjugation in the liver and kidney. For the
respiratory tract parameters, there are prior data to identify the parameters. Moreover, none of
the dose-metric predictions are sensitive to these parameters (see Section 3.5.7.2, below). For
GSH conjugation, it should be noted that for the clearance in the liver and lung (V
MAX
/K
M
), some
data are available with sensitivity 0.01 < |SC| < 0.1. The data are not at all informative as to the
maximum capacity for GSH conjugation.
For rats, all of the scaling parameters have at least one calibration data point with
|SC| > 0.01. However, for the volumes of the respiratory lumen and tissue (VRespLumC,
VRespEffC), the partition coefficient for the respiratory tissue (lnPRespC), and the V
MAX
values
for GSH conjugation in the liver, these consist of only one or two data points. As with mice,
there are prior data to help identify the respiratory tract parameters. Moreover, none of the dose-
metric predictions are sensitive to the respiratory tract parameters (see Section 3.5.7.2, below).
The data are not very informative as to maximum capacity for GSH conjugation in the liver.
However, there are some data that have low or moderate informativeness (0.1 < |SC| < 1) as to
the maximum capacity for GSH conjugation in the kidney, and clearance via GSH conjugation
(V
MAX
/K
M
) in the liver and kidney, which have much greater impact on the dose-metric
predictions than the maximum capacity in the liver (see Section 3.5.7.2, below).
For humans, all of the scaling parameters have at least one calibration data point with
|SC| > 0.01. However, for the volumes of the respiratory lumen and tissue (VRespLumC,
VRespEffC), the partition coefficient for the respiratory tissue (lnPRespC), and the oral
absorption rate for TCA, these consist of only one or two data points. As with mice and rats,
there are prior data to help identify the respiratory tract parameters. Moreover, none of the dose-
metric predictions are sensitive to the respiratory or TCA oral absorption parameters (see
Section 3.5.7.2, below).
Therefore, the local sensitivity analysis with respect to calibration data confirms that
most of the scaling parameters are informed by at least some of the calibration data. In addition,
the parameters for which the calibration data have very little or negligible sensitivity are either
informed by prior data or have little impact on dose-metric predictions.

3.5.6.5. Summary Evaluation of Updated PBPK Model
Overall, the updated PBPK model, utilizing parameters consistent with the available
physiological and in vitro data from published literature, provides reasonable fits to an extremely
large database of in vivo pharmacokinetic data in mice, rats, and humans. Posterior parameter
distributions were obtained by MCMC sampling using a hierarchical Bayesian population
3-133
statistical model and a large fraction of this in vivo database. Convergence of the MCMC
samples for model parameters was good for mice, and adequate for rats and humans. Evaluation
of posterior parameter distributions suggests reasonable results in light of prior expectations and
the nature of the available calibration data. In addition, in rats and humans, the model produced
predictions that are consistent with in vivo data from many studies not used for calibration
(insufficient studies were available in mice for such out of sample evaluation). Finally, the
local sensitivity analysis with respect to calibration data confirms that most of the scaling
parameters are informed by at least some of the calibration data, and those that were not either
were informed by prior data or would not have great impact on dose-metric predictions.

3.5.7. PBPK Model Dose-Metric Predictions
3.5.7.1. Characterization of Uncertainty and Variability
Since it is desirable to characterize the contributions from both uncertainty in population
parameters and variability within the population, the following procedure is adopted. First,
500 sets of population parameters (i.e., population mean and variance for each parameter) are
extracted from the posterior MCMC samplesthese represent the uncertainty in the population
parameters. To minimize autocorrelation, they were obtained by thinning the chains to the
appropriate degree. From each of these sets of population parameters, 100 subject-specific
parameters were generated by Monte Carloeach of these represents the population variability,
given a particular set of population parameters. Thus, a total of 50,000 subjects, representing
100 (variability) each for 500 different populations (uncertainty), were generated.
Each set was run for a variety of generic exposure scenarios. The combined distribution
of all 50,000 individuals reflects both uncertainty and variability (i.e., the case in which one is
trying to predict the dosimetry for a single random subject). In addition, for each dose-metric,
the mean predicted internal dose was calculated from each of the 500 sets of 100 individuals,
resulting in a distribution for the uncertainty in the population mean. Comparing the combined
uncertainty and variability distribution with the uncertainty distribution in the population mean
gives a sense of how much of the overall variation is due to uncertainty vs. variability.
Figures 3-173-25 show the results of these simulations for a number of representative
dose-metrics across species continuously exposed via inhalation or orally. For display purposes,
dose-metrics have been scaled by total intake (resulting in a predicted fraction metabolized) or
exposure level (resulting in an internal dose per ppm for inhalation or per mg/kg-day for oral
exposures). In these figures, the thin error bars represent the 95% CI for overall uncertainty and
variability, and the thick error bars represent the 95% CI for the uncertainty in the population
mean. The interpretation of these figures is that if the thick error bars are much smaller (or
greater) than the thin error bars, then variability (or uncertainty) contributes the most to overall
uncertainty and variability.
3-134



0.1 1 10 100 1000
Mouse
Rat
Human
Continuous inhalation ( ppm )
0
.
0
0
.
2
0
.
4
0
.
6
0
.
8
1
.
0
Fraction Metabolized A
0.1 1 10 100 1000
Mouse
Rat
Human
Continuous oral ( mg/kg-d )
0
.
0
0
.
2
0
.
4
0
.
6
0
.
8
1
.
0
Fraction Metabolized B
Bars and thin error bars represent the median estimate and 95% CI for a random
subject, and reflect combined uncertainty and variability. Circles and thick error
bars represent the median estimate and 95% CI for the population mean, and
reflect uncertainty only.

Figure 3-17. PBPK model predictions for the fraction of intake that is
metabolized under continuous inhalation (A) and oral (B) exposure
conditions in mice (white), rats (diagonal hashing), and humans (horizontal
hashing).
3-135


0.1 1 10 100 1000
Mouse
Rat
Human
Continuous inhalation ( ppm )
0
.
0
0
.
2
0
.
4
0
.
6
0
.
8
1
.
0
Fraction Oxidized A
0.1 1 10 100 1000
Mouse
Rat
Human
Continuous oral ( mg/kg-d )
0
.
0
0
.
2
0
.
4
0
.
6
0
.
8
1
.
0
Fraction Oxidized B
Bars and thin error bars represent the median estimate and 95% CI for a random
subject, and reflect combined uncertainty and variability. Circles and thick error
bars represent the median estimate and 95% CI for the population mean, and
reflect uncertainty only.
Figure 3-18. PBPK model predictions for the fraction of intake that is
metabolized by oxidation (in the liver and lung) under continuous inhalation
(A) and oral (B) exposure conditions in mice (white), rats (diagonal hashing),
and humans (horizontal hashing).

3-136


Continuous inhalation ( ppm )
M M M M M R R R R R H H H H H
Fraction Conjugated A
10
1
1 10
1
10
2
10
3
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
Continuous oral ( mg/kg-d )
M M M M M R R R R R H H H H H
Fraction Conjugated B
10
1
1 10
1
10
2
10
3
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.
Figure 3-19. PBPK model predictions for the fraction of intake that is
metabolized by GSH conjugation (in the liver and kidney) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats
(dashed line), and humans (solid line).

3-137



Continuous inhalation ( ppm )
R R R R R H H H H H
Fraction bioactivated in kidney A
10
1
1 10
1
10
2
10
3
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
Continuous oral ( mg/kg-d )
R R R R R H H H H H
Fraction bioactivated in kidney B
10
1
1 10
1
10
2
10
3
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-20. PBPK model predictions for the fraction of intake that is
bioactivated DCVC in the kidney under continuous inhalation (A) and oral
(B) exposure conditions in rats (dashed line) and humans (solid line).
3-138

Continuous inhalation ( ppm )
M M M M M R R R R R H H H H H
Fraction lung oxidation A
10
1
1 10
1
10
2
10
3
1
0
6
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
Continuous oral ( mg/kg-d )
M M M M M R R R R R H H H H H
Fraction lung oxidation B
10
1
1 10
1
10
2
10
3
1
0
6
1
0
5
1
0
4
1
0
3
1
0
2
1
0
1
1
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-21. PBPK model predictions for fraction of intake that is oxidized
in the respiratory tract under continuous inhalation (A) and oral (B)
exposure conditions in mice (dotted line), rats (dashed line), and humans
(solid line).

3-139

Continuous inhalation ( ppm )
M M M M M R R R R R H H H H H
Fraction 'other' liver oxidation A
10
1
1 10
1
10
2
10
3
1
0
4
1
0
3
1
0
2
1
0
1
1
Continuous oral ( mg/kg-d )
M M M M M R R R R R H H H H H
Fraction 'other' liver oxidation B
10
1
1 10
1
10
2
10
3
1
0
4
1
0
3
1
0
2
1
0
1
1
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-22. PBPK model predictions for the fraction of intake that is
untracked oxidation of TCE in the liver under continuous inhalation (A)
and oral (B) exposure conditions in mice (dotted line), rats (dashed line), and
humans (solid line).

3-140

Continuous inhalation ( ppm )
m
g
-
h
/
l
-
w
k

p
e
r

p
p
m
M M M M M R R R R R H H H H H
AUC TCE in blood
per ppm A
10
1
1 10
1
10
2
10
3
1
0
2
1
0
1
1
1
0
1
1
0
2
Continuous oral ( mg/kg-d )
m
g
-
h
/
l
-
w
k

p
e
r

m
g
/
k
g
-
d
M M M M M R R R R R H H H H H
AUC TCE in blood
per mg/kg-d B
10
1
1 10
1
10
2
10
3
1
0
2
1
0
1
1
1
0
1
1
0
2
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-23. PBPK model predictions for the weekly AUC of TCE in venous
blood (mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under
continuous inhalation (A) and oral (B) exposure conditions in mice (dotted
line), rats (dashed line), and humans (solid line).
3-141

Continuous inhalation ( ppm )
m
g
-
h
/
l
-
w
k

p
e
r

p
p
m
M M M M M R R R R R H H H H H
AUC TCOH in blood
per ppm A
10
1
1 10
1
10
2
10
3
1
0
2
1
0
1
1
1
0
1
1
0
2
1
0
3
1
0
4
Continuous oral ( mg/kg-d )
m
g
-
h
/
l
-
w
k

p
e
r

m
g
/
k
g
-
d
M M M M M R R R R R H H H H H
AUC TCOH in blood
per mg/kg-d B
10
1
1 10
1
10
2
10
3
1
0
2
1
0
1
1
1
0
1
1
0
2
1
0
3
1
0
4
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-24. PBPK model predictions for the weekly AUC of TCOH in blood
(mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under continuous
inhalation (A) and oral (B) exposure conditions in mice (dotted line), rats
(dashed line), and humans (solid line).
3-142

Continuous inhalation ( ppm )
m
g
-
h
/
k
g
-
w
k

p
e
r

p
p
m
M M M M M R R R R R H H H H H
AUC TCA in liver
per ppm A
10
1
1 10
1
10
2
10
3
1
0
1
1
1
0
1
1
0
2
1
0
3
1
0
4
Continuous oral ( mg/kg-d )
m
g
-
h
/
k
g
-
w
k

p
e
r

m
g
/
k
g
-
d
M M M M M R R R R R H H H H H
AUC TCA in liver
per mg/kg-d B
10
1
1 10
1
10
2
10
3
1
0
1
1
1
0
1
1
0
2
1
0
3
1
0
4
X-values are slightly offset for clarity. Open circles (connected by lines) and thin
error bars represent the median estimate and 95% CI for a random subject, and
reflect combined uncertainty and variability. Filled circles and thick error bars
represent the median estimate and 95% CI for the population mean, and reflect
uncertainty only.

Figure 3-25. PBPK model predictions for the weekly AUC of TCA in the
liver (mg-hour/L-week) per unit exposure (ppm or mg/kg-day) under
continuous inhalation (A) and oral (B) exposure conditions in mice (dotted
line), rats (dashed line), and humans (solid line).

For application to human health risk assessment, the uncertainty in and variability among
rodent internal dose estimates both contribute to uncertainty in human risk estimates. Therefore,
it is appropriate to combine uncertainty and variability when applying rodent dose-metric
predictions to quantitative risk assessment. The median and 95% CI for each dose-metric at
some representative exposures in rodents are given in Tables 3-48 and 3-49, and the CI in these
tables includes both uncertainty in the population mean and variance as well as variability in the
population. On the other hand, for use in predicting human risk, it is often necessary to separate,
to the extent possible, interindividual variability from uncertainty, and this disaggregation is
summarized in Table 3-50.

3.5.7.2. Local Sensitivity Analysis With Respect to Dose-Metric Predictions
To assess the parameter sensitivity of dose-metric predictions, a local sensitivity analysis
is performed. The representative exposure scenarios in Tables 3-483-50 are used, but with
3-143
metabolic flux dose-metrics converted to fraction of intake (i.e., amount metabolized through a
pathway divided by total dose). Each parameter is centered on the sample mean of its estimated
population mean, and then increased and decreased by 5%. The relative change in the model
output f() is used to estimate a local SC as follows:

SC = 10 {f(
+
) f(

)}/ [ {f(
+
) + f(

)}]

Here, f() is one of dose-metric predictions,

is the MLE or baseline value of 5%. For

log-transformed parameters, 0.05 was added or subtracted from the baseline value, whereas for
untransformed parameters, the baseline value was multiplied by 1.05 or 0.95.
Note that local sensitivity analyses as typically performed in deterministic PBPK
modeling can only inform the primary effects of parameter uncertainties (i.e., the direct change
on the quantity of interest due to change in a parameter). They cannot address the propagation
of uncertainties through an analysis, such as those that can arise due to parameter correlations in
the parameter fitting process. Those can only be addressed in a global sensitivity analysis, which
is left for future research.
The results of local sensitivity analyses are shown in Figures 3-263-31. As expected,
each dose-metric is sensitive to a only a small fraction of the scaling parameters. Many of these
are well-specified a priori, either due to their being physiological parameters or partition
coefficients that can be measured in vitro. The remaining sensitive parameters are generally
related to metabolism or clearance.


3-144

Table 3-48. Posterior predictions for representative internal doses: mouse
a


Dose-metric
Posterior predictions for mouse dose-metrics: median (2.5%, 97.5%)
Units 100 ppm, 7 hr/d, 5 d/wk 600 ppm, 7 hr/d, 5 d/wk 300 mg/kg-d, 5 d/wk 1,000 mg/kg-d, 5 d/wk
ABioactDCVCBW34 0.304 (0.000534, 12.4) 2.35 (0.00603, 37) 0.676 (0.00193, 18.4) 2.81 (0.0086, 42.4) mg/wk-kg
3/4

ABioactDCVCKid 43.7 (0.0774, 1780) 336 (0.801, 5,240) 96.8 (0.281, 2,550) 393 (1.23, 6,170) mg/wk-kg tissue
AMetGSHBW34 0.684 (0.0307, 17.6) 5.15 (0.285, 44.9) 1.66 (0.0718, 24.5) 6.37 (0.567, 49.4) mg/wk-kg
3/4

AMetLiv1BW34 170 (61.2, 403) 878 (342, 2,030) 400 (125, 610) 874 (233, 1,960) mg/wk-kg
3/4

AMetLivOtherBW34 3.81 (0.372, 38.4) 20 (1.86, 192) 8.38 (0.773, 80.1) 20 (1.55, 202) mg/wk-kg
3/4

AMetLivOtherLiv 196 (19, 2,070) 1,030 (96.5, 10,100) 437 (39.5, 4,180) 1,020 (82.1, 10,400) mg/wk-kg tissue
AMetLngBW34 187 (7.75, 692) 263 (10.9, 2,240) 38.5 (3.49, 147) 127 (8.59, 484) mg/wk-kg
3/4

AMetLngResp 638,000
(26,500, 2,510,000)
918,000
(36,800, 7,980,000)
134,000
(12,500, 514,000)
433,000
(30,200, 1,690,000)
mg/wk-kg tissue
AUCCBld 96.9 (45, 211) 822 (356, 2,040) 110 (6.95, 411) 592 (56, 1,910) mg-hr/L-wk
AUCCTCOH 87.9 (9.9, 590) 480 (42.1, 4,140) 132 (14.4, 670) 389 (34, 2,600) mg-hr/L-wk
AUCLivTCA 1,880 (444, 7,190) 5,070 (1,310, 18,600) 2,260 (520, 8,750) 4,660 (939, 18,900) mg-hr/L-wk
TotMetabBW34 377 (140, 917) 1,260 (475, 3,480) 472 (165, 617) 1,110 (303, 2,010) mg/wk-kg
3/4

TotOxMetabBW34 375 (139, 916) 1,250 (451, 3,450) 465 (161, 616) 1,100 (294, 2,010) mg/wk-kg
3/4

TotTCAInBW 272 (88.9, 734) 729 (267, 1,950) 334 (106, 875) 694 (185, 1,910) mg/wk-kg

a
Mouse body weight is assumed to be 0.03 kg. Predictions are weekly averages over 10 weeks of the specified exposure protocol. CI reflects both uncertainties
in population parameters (mean, variance) as well as population variability.
3-145
Table 3-49. Posterior predictions for representative internal doses: rat
a


Dose-metric
Posterior predictions for rat dose-metrics: median (2.5%,97.5%)
Units
100 ppm, 7 hr/d,
5 d/wk 600 ppm, 7 hr/d, 5 d/wk 300 mg/kg-d, 5 d/wk 1,000 mg/kg-d, 5 d/wk
ABioactDCVCBW34 0.341 (0.0306, 2.71) 2.3 (0.175, 22.6) 2.15 (0.17, 20.2) 8.89 (0.711, 84.1) mg/wk-kg
3/4

ABioactDCVCKid 67.8 (6.03, 513) 450 (35.4, 4,350) 420 (31.6, 3,890) 1,720 (134, 15,800) mg/wk-kg tissue
AMetGSHBW34 0.331 (0.0626, 2.16) 2.27 (0.315, 19.3) 2.13 (0.293, 16) 8.84 (1.35, 69.3) mg/wk-kg
3/4

AMetLiv1BW34 176 (81.1, 344) 623 (271, 1,270) 539 (176, 1,060) 951 (273, 2,780) mg/wk-kg
3/4

AMetLivOtherBW34 45.5 (2.52, 203) 160 (7.84, 749) 134 (6.83, 659) 238 (11.3, 1390) mg/wk-kg
3/4

AMetLivOtherLiv 1,870 (92.1, 8,670) 6,660 (313, 31,200) 5,490 (280, 27,400) 9,900 (492, 59,600) mg/wk-kg tissue
AMetLngBW34 15 (0.529, 173) 24.5 (0.819, 227) 15.1 (0.527, 115) 32.1 (1.01, 311) mg/wk-kg
3/4

AMetLngResp 41,900 (1,460, 496,000) 67,900 (2,350, 677,000) 40,800 (1,500, 325,000) 85,700 (2,660, 877,000) mg/wk-kg tissue
AUCCBld 86.7 (39.2, 242) 1,160 (349, 2,450) 670 (47.8, 1,850) 3,340 (828, 8,430) mg-hr/L-wk
AUCCTCOH 83.6 (1.94, 1,560) 446 (6, 10,900) 304 (4.71, 7,590) 685 (8.14, 32,500) mg-hr/L-wk
AUCLivTCA 587 (53.7, 4,740) 2,030 (186, 13,400) 1,730 (124, 11,800) 3,130 (200, 21,000) mg-hr/L-wk
TotMetabBW34 206 (103, 414) 682 (288, 1,430) 572 (199, 1,080) 1,030 (302, 2,920) mg/wk-kg
3/4

TotOxMetabBW34 206 (103, 414) 677 (285, 1,430) 568 (191, 1,080) 1,010 (286, 2,910) mg/wk-kg
3/4

TotTCAInBW 31.7 (3.92, 174) 110 (13.8, 490) 90.1 (10.4, 417) 164 (17.3, 800) mg/wk-kg

a
Rat body weight is assumed to be 0.3 kg. Predictions are weekly averages over 10 weeks of the specified exposure protocol. CI reflects both uncertainties in
population parameters (mean, variance) as well as population variability.
3-146
Table 3-50. Posterior predictions for representative internal doses: human
a


Dose-metric
Posterior predictions for human dose-metrics:
2.5% population: median (2.5%, 97.5%)
50% population: median (2.5%, 97.5%)
97.5% population: median (2.5%, 97.5%)
Female
0.001 ppm continuous
Male
0.001 ppm continuous
Female
0.001 mg/kg-d continuous
Male
0.001 mg/kg-d continuous
ABioactDCVCBW34 0.000256 (6.97 10
-5
, 0.000872) 0.000254 (6.94 10
-5
, 0.000879) 0.000197 (6.13 10
-5
, 0.000502) 0.0002 (6.24 10
-5
, 0.000505)
0.00203 (0.00087, 0.00408) 0.00202 (0.000859, 0.00413) 0.00262 (0.0012, 0.00539) 0.00271 (0.00125, 0.00559)
0.0119 (0.00713, 0.0177) 0.012 (0.00699, 0.0182) 0.021 (0.0118, 0.0266) 0.022 (0.0124, 0.0277)
ABioactDCVCKid 0.02 (0.00549, 0.0709) 0.0207 (0.00558, 0.0743) 0.0152 (0.0048, 0.0384) 0.016 (0.00493, 0.0407)
0.16 (0.0671, 0.324) 0.163 (0.0679, 0.342) 0.207 (0.0957, 0.43) 0.22 (0.102, 0.459)
0.95 (0.56, 1.45) 0.979 (0.563, 1.51) 1.68 (0.956, 2.26) 1.81 (1.03, 2.43)
AMetGSHBW34 0.000159 (4.38 10
-5
, 0.000539) 0.000157 (4.37 10
-5
, 0.00054) 0.000121 (3.82 10
-5
, 0.000316) 0.000123 (3.82 10
-5
, 0.000323)
0.00126 (0.000536, 0.00253) 0.00125 (0.000528, 0.00254) 0.00161 (0.000748, 0.00331) 0.00167 (0.000777, 0.00343)
0.00736 (0.00442, 0.011) 0.00736 (0.00434, 0.0112) 0.013 (0.00725, 0.0164) 0.0136 (0.00759, 0.0171)
AMetLiv1BW34 0.00161 (0.000619, 0.00303) 0.00157 (0.000608, 0.00292) 0.00465 (0.00169, 0.0107) 0.00498 (0.00184, 0.0112)
0.00637 (0.00501, 0.00799) 0.00619 (0.00484, 0.00779) 0.0172 (0.0153, 0.0183) 0.018 (0.0161, 0.0191)
0.0157 (0.0118, 0.0206) 0.0152 (0.0115, 0.02) 0.0192 (0.019, 0.0193) 0.02 (0.0198, 0.0201)
AMetLivOtherBW34 4.98 10
-5
(8.59 10
-6
, 0.000222) 4.87 10
-5
(8.33 10
-6
, 0.000214) 0.000143 (2.35 10
-5
, 0.000681) 0.00015 (2.49 10
-5
, 0.000713)
0.000671 (0.000134, 0.00159) 0.000652 (0.000129, 0.00153) 0.00166 (0.00035, 0.00365) 0.00173 (0.000365, 0.00382)
0.00507 (0.00055, 0.00905) 0.00491 (0.000531, 0.00885) 0.00993 (0.00109, 0.0153) 0.0103 (0.00113, 0.0159)
AMetLivOtherLiv 0.000748 (0.000138, 0.00335) 0.00065 (0.000119, 0.00288) 0.00214 (0.000354, 0.00979) 0.00197 (0.00033, 0.00907)
0.0104 (0.00225, 0.0237) 0.00898 (0.00193, 0.0203) 0.0253 (0.00564, 0.0543) 0.0234 (0.00526, 0.0503)
0.0805 (0.00871, 0.147) 0.0691 (0.00751, 0.127) 0.157 (0.0188, 0.251) 0.146 (0.0173, 0.232)
AMetLngBW34 6.9 10
-6
(6.13 10
-7
, 7.99 10
-5
) 7.25 10
-6
(6.44 10
-7
, 8.39 10
-5
) 7.54 10
-8
(6.59 10
-9
, 7.85 10
-7
) 7.05 10
-8
(6.1 10
-9
, 7.25 10
-7
)
0.00122 (0.000309, 0.0032) 0.00127 (0.000325, 0.00329) 1.51 10
-5
(3.44 10
-6
, 4.6 10
-5
) 1.39 10
-5
(3.21 10
-6
, 4.24 10
-5
)
0.0123 (0.00563, 0.0197) 0.0124 (0.00582, 0.0199) 0.000396 (0.000104, 0.00097) 0.000366 (9.54 10
-5
, 0.000906)

3-147
Table 3-50. Posterior predictions for representative internal doses: human
a
(continued)

Dose-metric
Posterior predictions for human dose-metrics:
2.5% population: median (2.5%, 97.5%)
50% population: median (2.5%, 97.5%)
97.5% population: median (2.5%, 97.5%)
Female
0.001 ppm continuous
Male
0.001 ppm continuous
Female
0.001 mg/kg-d continuous
Male
0.001 mg/kg-d continuous
AMetLngResp 0.0144 (0.00116, 0.155) 0.0146 (0.00118, 0.157) 0.00015 (1.27 10
-5
, 0.00153) 0.000134 (1.15 10
-5
, 0.00137)
2.44 (0.613, 6.71) 2.44 (0.621, 6.65) 0.0313 (0.00725, 0.0963) 0.0279 (0.00644, 0.086)
25.8 (12.4, 42.3) 25.3 (12.2, 41.2) 0.813 (0.216, 2.13) 0.716 (0.189, 1.9)
AUCCBld 0.00151 (0.00122, 0.00186) 0.00158 (0.00127, 0.00191) 4.33 10
-5
(3.3 10
-5
, 6.23 10
-5
) 3.84 10
-5
(2.89 10
-5
, 5.61 10
-5
)
0.00285 (0.00252, 0.00315) 0.00295 (0.00262, 0.00326) 0.000229 (0.000122, 0.000436) 0.000204 (0.000109, 0.000391)
0.00444 (0.00404, 0.00496) 0.00456 (0.00416, 0.00507) 0.00167 (0.000766, 0.00324) 0.00153 (0.000693, 0.00303)
AUCCTCOH 0.00313 (0.00135, 0.00547) 0.00305 (0.00134, 0.00532) 0.00584 (0.00205, 0.0122) 0.00615 (0.00213, 0.0127)
0.0181 (0.0135, 0.0241) 0.0179 (0.0133, 0.0238) 0.0333 (0.025, 0.0423) 0.035 (0.0264, 0.0445)
0.082 (0.0586, 0.118) 0.0812 (0.0585, 0.117) 0.115 (0.0872, 0.163) 0.122 (0.0919, 0.172)
AUCLivTCA 0.0152 (0.00668, 0.0284) 0.0137 (0.00598, 0.0258) 0.029 (0.0116, 0.0524) 0.0279 (0.0114, 0.0501)
0.126 (0.0784, 0.194) 0.114 (0.0704, 0.177) 0.227 (0.138, 0.343) 0.219 (0.133, 0.33)
0.754 (0.441, 1.38) 0.699 (0.408, 1.3) 1.11 (0.661, 1.87) 1.09 (0.64, 1.88)
TotMetabBW34 0.0049 (0.00383, 0.00595) 0.00482 (0.0038, 0.00585) 0.0163 (0.0136, 0.0181) 0.0173 (0.0147, 0.019)
0.0107 (0.00893, 0.0129) 0.0105 (0.00877, 0.0127) 0.0191 (0.0188, 0.0193) 0.0199 (0.0196, 0.0201)
0.0246 (0.0185, 0.0326) 0.0244 (0.0183, 0.0324) 0.0194 (0.0194, 0.0194) 0.0202 (0.0202, 0.0202)
TotOxMetabBW34 0.00273 (0.00143, 0.00422) 0.00269 (0.00143, 0.00415) 0.0049 (0.00183, 0.0108) 0.00516 (0.00194, 0.0114)
0.00871 (0.0069, 0.0111) 0.00857 (0.00675, 0.011) 0.0173 (0.0154, 0.0183) 0.018 (0.0161, 0.0191)
0.0224 (0.0158, 0.0309) 0.0222 (0.0155, 0.0308) 0.0192 (0.019, 0.0193) 0.02 (0.0198, 0.0201)
3-148
a
Table 3-50. Posterior predictions for representative internal doses: human

(continued)
Dose-metric
Posterior predictions for human dose-metrics:
2.5% population: median (2.5%, 97.5%)